Finding the cells that put our brain to sleep

Rare cells with widespread connectivity seem to be able to trigger sleep.

For a long time, sleep research has treated the cerebral cortex as a passive follower reacting to signals from the deep brain. “Usually, sleep is associated with being controlled by subcortical regions,” said Geoffrey Terral, a neuroscientist at the Albert Einstein College of Medicine in New York.

The cortex is where the slow rhythms of deep sleep can be seen, but researchers assumed the signals that triggered them originated elsewhere. In a recent Nature study, Terral and Renata Batista-Brito, who runs the lab, report a population of cortical cells that challenges that assumption.

These cortical cells make up only around one percent of the cortex’s inhibitory neurons, and switching them on in a mouse puts the animal to sleep. “What our work shows is that the cortex can not only see this rhythm but also initiate it by itself, and this is sufficient to promote sleep,” Terral said.

The 1 percent

The cells are called Sst-Chodl neurons, after two genes that are active in them. Long-range inhibitory neurons marked by the activity of these genes have been known for some time, based on studies in monkeys, but researchers couldn’t figure out how to manipulate them using these genes. Aiming at either one of these genes captures a huge, heterogeneous family of different neurons. “A single gene is not able to target these cells,” Batista-Brito said.

Inhibitory neurons account for about a fifth of all cortical neurons, so Sst-Chodl cells amount to roughly one neuron in a thousand in the cortex. “If you’re not really specific, the contaminants are going to be much more dominant than the specific cells,” Batista-Brito said.

Building a strategy that labels a cell only when both genes are on took her team years.

“Going after these cells was really a high-risk project because the likelihood of seeing anything with 0.1 percent of neurons in the cortex is really low,” Batista-Brito said. “I wrote a bunch of grants on these projects that were always rejected because it was too high-risk.”

Her first look at the anatomy of Sst-Chodl neurons proved the risk worth taking. “At first, I saw two or three cell bodies in the whole brain,” Batista-Brito told Ars. “Despite that, there was massive, massive arborization all over the visual cortex, like I never saw with any other neuron.”

That “massive arborization” is a tree-like branching that connects neurons to other parts of the brain, making them part of a wider communication network. It was also very unusual, as inhibitory neurons are almost always local. They receive inputs from different, sometimes distant regions in the brain and exert control over their immediate surroundings—their own small patch.

“Sst-Chodl neurons are kind of the opposite,” Batista-Brito said. “They are receiving inputs that are quite precise, but then they are broadcasting that information everywhere.”

A single Sst-Chodl cell branches out to the entire visual cortex, then sends axons to the areas of the brain responsible for touch, hearing, spatial memory, navigation, and voluntary movements.

The quiet hours

To see when the cells fire, the team imaged them with while tracking the mice’s pupil size, muscle tone, facial movement, running, and cortical electrical activity at the same time. Ninety-five out of 111 imaged cells lit up during slow-wave sleep and quiet, motionless wakefulness, and fell silent during running and REM sleep.

During slow-wave sleep, the cortex alternates between UP states of vigorous firing and DOWN states of near-silence; the Sst-Chodl neurons broke this pattern. “Their activity is even higher at the termination of the UP state,” Terral said. “Once all the other neurons start to decrease their activity, those neurons start to increase it even more.”

They also skip the rebound that other cells show coming out of a DOWN state. “Those cells behave totally differently to any other cells we can measure around that transition,” Terral said.

To really understand what they do, the team made the SSt-Chodl neurons fire on command using optogenetics, which involves inserting the gene for a light-activated ion channel that can stimulate the nerve cells to fire.

The long reach

Optogenetic stimulation of the Sst-Chodl cells in the visual cortex drove delta power—the slowest, highest-amplitude brain waves measured during deep, restorative sleep—up across every layer. It also tightened spike timing and made DOWN states more frequent and longer. But overall firing barely changed.

“Those neurons are not changing so much the firing rate,” Terral said. “They change just the coordination—whether the neurons fire together or not.”

Normally, driving a population at a given frequency entrains the network at that frequency. Here, it didn’t matter.

“If we did a flat stimulus, or delta, or 20 hertz, or 60 hertz, you still induced the same kind of oscillation,” Batista-Brito said. “There’s something about the intrinsic properties of these cells that, once they trigger, they go on this one frequency.”

She emphasized that we don’t currently know how they do this.

The reach of these neurons also proved unusually long. “Their cell bodies were in the visual cortex, but they could regulate the activity of most of the neurons we were recording,” Terral said.

Direct inhibitory currents appeared in a third of cells two millimeters out, and stimulating the visual cortex had measurable effects as far as frontal motor areas.

In the final step, the team focused on behavioral experiments. They checked if stimulating the Sst-Chodl neurons would put the mice to sleep. It did.

A sleep switch

Activating the cells across the cortex of freely moving mice increased slow-wave and REM sleep, cut the time taken to fall asleep, and sent the animals into their nests during the day. Then the team tried it during the dark phase, when nocturnal mice are mostly awake.

“We could have them sleep more during the time that they are awake than they usually sleep during the day,” Batista-Brito said. “I never thought this experiment was going to work. We are manipulating one percent of inhibitory neurons with local injections, so we are hitting only a fraction of those cells. And we could see an effect that was really quite striking.”

That effect, Batista-Brito argues, points toward the idea that these neurons might be the cortex’s sensors of sleep pressure, something first proposed by Thomas Kilduff, the director of the SRI International’s Center for Neuroscience and co-author of the study.

Sleep, Batista-Brito explains comes in two flavors: the circadian sort that follows the light and homeostatic sleep, the accumulated fatigue that eventually makes us sleep whether we want to or not. Kilduff, Batista-Brito said, showed that after sleep deprivation, these are the most active cells in the cortex.

One question Batist-Brito and her colleagues still don’t have the answer to, though, is what activates these cells.

What flips the switch?

“One caveat of our work is that it was done in the visual cortex,” Batista-Brito said.

The findings about the coordinating role of the Sst-Chodl neurons, the team claims, should generalize to different brain regions, but the actual wiring may not. Batista-Brito’s lab is now repeating the anatomy part of their work in the prefrontal cortex. Their prediction is that there we’ll see inputs from other areas, such as the hypothalamus and the thalamus, and that will be what triggers the network for sleep specifically.

The cells are conserved from salamanders to humans, which is why the paper argues they could be an entry point into the sleep disruption running through so many psychiatric illnesses.

The team is now focused on three remaining questions: what activates these cells, whether they actually sense sleep pressure, and why they drive delta power.

Nature, 2026. DOI: 10.1038/s41586-026-10876-y