Scientists uncover an energy paradox in the brain during REM sleep

A recent study published in Communications Biology provides evidence that the brain manages its energy resources in surprisingly different ways depending on whether it is in a deep sleep or a dreaming state. By monitoring live brains, scientists observed that during rapid eye movement sleep, the brain receives a surge of blood and energy nutrients, yet the actual energy levels inside neurons drop. This indicates that the sleeping brain might prioritize complex memory processing or biological protection over maintaining steady energy reserves.

The brain requires a constant supply of energy to process information and maintain basic biological functions. When a neuron fires an electrical signal, it alters the balance of charged ions across its cellular membrane. Restoring this balance requires massive amounts of energy to power specialized molecular pumps. This energy is primarily delivered through the bloodstream in the form of glucose.

Once inside the brain, glucose is taken up directly by neurons or by star-shaped support cells called astrocytes. Astrocytes convert glucose into a molecule called pyruvate, which acts as a stepping stone to produce adenosine triphosphate, commonly known as ATP. ATP is the primary chemical fuel that powers cells. Under normal waking conditions, the supply of these energy substrates usually matches the brain’s activity levels.

During sleep, the brain remains highly active, sorting memories and restoring biological systems. Sleep is broadly divided into non-rapid eye movement (NREM) sleep, characterized by deep, slow brain waves, and rapid eye movement (REM) sleep, a state marked by faster brain waves and vivid dreaming. The authors wanted to understand exactly how the brain distributes its limited metabolic resources across these distinct physiological states.

“REM sleep is a paradoxical state: the body is deeply relaxed, but the brain is remarkably active,” explains Ko Matsui, a professor of super-network brain physiology at the Graduate School of Life Sciences at Tohoku University in Sendai, Japan. “That made us wonder what happens to the brain’s energy system during this unusual state.”

Matsui notes that because REM sleep involves vivid dreaming and intense electrical activity, the team suspected its metabolic demands would reflect that intensity. “With new fluorescence imaging approaches, we were able to observe blood volume and energy-related molecules across large areas of the brain and ask how they change as the brain shifts between non-REM and REM sleep,” Matsui said.

To track these metabolic changes, the researchers observed the brains of 15 live male mice. They used genetically modified animals that expressed fluorescent sensors, allowing the team to visually measure ATP levels inside neurons and pyruvate levels inside astrocytes. The scientists applied a transparent UV-curable resin to the animals’ intact skulls to maintain visibility. This technique avoided invasive surgeries that might disrupt natural blood flow or alter intracranial pressure.

The researchers then placed the animals under wide-field fluorescence microscopes to simultaneously monitor brain blood volume, pyruvate, and ATP. They recorded the animals for several hours as the mice naturally transitioned through different sleep and wake states. Electrical monitors attached to the skull and neck muscles allowed the team to precisely identify when the mice were awake, in NREM sleep, or in REM sleep based on their brain waves and muscle relaxation.

During NREM sleep, the researchers observed a tight coordination between electrical brain activity and energy delivery. Specific electrical patterns, known as theta-band brain waves, reliably predicted changes in brain blood volume about four to five seconds in advance. In this deep sleep phase, blood volume fluctuated in fast waves that swept from the front of the brain to the back in about one second.

By dividing the images of the cortex into a grid, the researchers mapped how different brain regions synchronized their blood flow. During NREM sleep, the blood volume changes were somewhat localized. The cortex partitioned into multiple distinct functional clusters, suggesting an organized system that adjusts local blood flow to anticipate the energetic needs of specific resting brain circuits.

The metabolic environment shifted drastically as the mice transitioned into REM sleep. About 50 seconds before the official onset of REM sleep, brain blood volume began to surge. This large-scale increase originated in the posterior regions of the brain and slowly propagated forward over 15 seconds. During REM sleep, the localized, fast fluctuations diminished, and the brain experienced a massive wave of blood volume that synchronized broad swaths of the cerebral cortex.

Once REM sleep fully commenced, blood volume remained elevated, and the levels of pyruvate in astrocytes increased accordingly. Under normal circumstances, an increase in blood flow and intermediate nutrients like pyruvate would be expected to boost cellular energy. But the researchers found that neuronal ATP levels dropped sharply during REM sleep.

“The brain’s energy system is much more dynamic than simply ‘more blood flow means more energy for neurons,’” Matsui told PsyPost. “During REM sleep, we found what we call an ‘energy paradox’: cerebral blood volume increased and pyruvate increased in astrocytes, yet ATP in neurons decreased.”

Astrocytes belong to a broader category of support cells known as glial cells. As Matsui explained, this disconnect between nutrient delivery and final energy production hints at a highly complex support system. “This tells us that energy supply, energy transfer, energy production, and energy consumption can behave very differently from one another. It also suggests that blood vessels, glial cells, and metabolism may be more actively involved in brain function than we traditionally assumed.”

To verify that this energy drop was unique to REM sleep, the team conducted a separate test on three of the mice, encompassing nine total trials. They used a drug called sodium nitroprusside, a chemical vasodilator, to artificially widen the blood vessels. When this medication was administered, the resulting increase in blood volume was accompanied by expected increases in both astrocytic pyruvate and neuronal ATP.

Reflecting on these differences, Matsui points out that the vasodilator test behaved as anticipated, making the natural sleep findings even more unexpected. “The biggest surprise was that neuronal ATP decreased during REM sleep even though blood volume increased,” Matsui says. “When we artificially increased blood volume with a vasodilator, astrocytic pyruvate increased and neuronal ATP eventually increased as well, which was what we expected. REM sleep behaved very differently.”

The physical movement of the blood also stood out to the researchers. “We were also surprised by how complex the vascular activity itself was: blood-volume changes formed spatial waves, changed direction, and reorganized dramatically depending on the sleep state,” Matsui says.

Interpreting these energy fluctuations requires acknowledging a few biological contexts. The study relies on animal models, and mouse brain metabolism might differ from human brain function. The imaging techniques measure fluorescent signals as proxies for blood volume and chemical concentrations, meaning they do not provide absolute numerical values of molecules in the brain. The researchers also did not directly measure local oxygen levels or other metabolites like lactate, which could offer a more complete picture of the chemical environment.

Matsui cautions against drawing overly simplistic conclusions from the ATP drop. “Our results do not mean that REM sleep is simply an ‘energy-deficient’ state, nor do they show that changes in blood vessels cause the transition into REM sleep,” Matsui says. Instead, the findings reflect a complex internal economy, involving specialized cellular components like mitochondria, which generate most of a cell’s ATP.

“ATP concentration reflects the balance between ATP production and consumption,” Matsui explained. “A decrease could therefore result from increased energy use, altered transfer of metabolic substrates from astrocytes to neurons, changes in mitochondrial ATP production, or some combination of these mechanisms.” He adds, “Our experiments were also performed in mice, so further work will be needed to determine how directly these findings translate to the human brain.”

Moving forward, the research team hopes to untangle exactly why this energy drop occurs. “In the short term, we would like to understand why neuronal ATP falls during REM sleep, including possible changes in astrocyte-to-neuron metabolic transfer and mitochondrial function,” Matsui said. “Our long-term goal is to understand how the brain’s metabolic network interacts with its neuronal information-processing network.”

Matsui emphasized a growing interest in how support systems might dictate brain capabilities. “More broadly, however, we are interested in whether vascular, glial, and metabolic states can actively influence what neuronal circuits are able to do, rather than simply responding to neuronal activity,” Matsui says.

“One message I find particularly exciting is that there may be a great deal of biological information hidden in signals that neuroscience has often treated mainly as supporting signals,” Matsui said. “Blood vessels, astrocytes, and metabolic molecules showed rich spatial and temporal dynamics that could not be predicted simply from neuronal electrical activity.”

This indicates that intelligence and brain function rely on a much wider array of biological players than neurons alone. “We may therefore be seeing an additional layer of brain information processing – one created by interactions among neuronal, glial, vascular, and metabolic networks,” Matsui said.

The study, “Energy paradox in REM sleep: balancing supply and consumption in brain metabolism,” was authored by Yusuke Takahashi, Yoko Ikoma, and Ko Matsui.