March 30, 2025

Scientists sawed a human brain into 703 cubes to map its energy system for the first time

Credit: Midjourney.

The human brain runs on a tight energy budget. Despite making up only 2% of our body weight, it devours around 20% of our energy. Every time we think, feel, move, or remember, billions of brain cells draw on power from microscopic engines inside them — tiny organelles called mitochondria.

But until now, no one had mapped where those powerhouses are, how many there are, or how well they work across the vast landscape of the brain.

Now, for the first time, scientists have created a comprehensive atlas of these essential energy-makers. The project, known as MitoBrainMap, is now showing how the brain powers itself, region by region. In doing so, it may open new pathways for understanding — and eventually treating — conditions ranging from depression to Alzheimer’s disease.

A Hidden Architecture of Neural Power

To build the map, researchers at Columbia University and the University of Bordeaux took a slice of a frozen human brain and, using a woodworking saw (yikes!), chopped it into 703 sugar-cube-sized pieces — each three millimeters on a side.

“The most challenging part was having so many samples,” said Martin Picard, a psychobiologist at Columbia and one of the study’s leaders.

Scientists Sawed A Human Brain Into 703 Cubes To Map Its Energy System For The First Time
Analysis of mitochondria within 703 3x3x3mm cubes of a human brain created the first energy map of the brain. Image provided by Martin Picard / Columbia University Vagelos College of Physicians and Surgeons.

From there, the team analyzed the density and energy efficiency of mitochondria in each cube using biochemical and molecular techniques. They looked not just at how many mitochondria were packed into the tissue, but how well those mitochondria could churn out energy.

The data formed the basis of a computational model that predicted how mitochondria are distributed and function across the entire brain. The result was a striking new map of the brain’s energy landscape.

“It’s both technically impressive and conceptually groundbreaking,” said Valentin Riedl, a neurobiologist at the Technical University of Munich (not involved in the study).

<!– Tag ID: zmescience_300x250_InContent_3

[jeg_zmescience_ad_auto size=”__300x250″ id=”zmescience_300x250_InContent_3″]

–>

Newer Brain Regions Need More Juice

Such a high-resolution novel map was bound to reveal new things about the brain. For instance, the map shows that the brain’s power supply isn’t evenly distributed. Some areas demand more energy — and are built to get it.

Grey matter, the dense regions where brain cells process information, had more than 50% more mitochondria than white matter, which carries messages between cells. Not only that, but mitochondria in grey matter were more efficient at making energy, thanks to their specialized enzyme systems.

Additionally, the team developed a metric called “mitochondrial respiratory capacity” (MRC), which measures how much energy each mitochondrion can produce. Grey matter consistently showed higher MRC values, suggesting its mitochondria are molecularly fine-tuned for energy-demanding cognitive work.

The most advanced part of our brain — the wrinkled outer layer known as the cortex — stood out.

“These newer brain regions not only contained more mitochondria,” said Anna Monzel, a computational research scientist on the team, “but these mitochondria were specialized for more efficient energy production.”

That fits with what we know about brain evolution.

Brain energy evolution

The cortex, which evolved more recently (relatively speaking — about 300 to 500 million years ago), supports complex functions like planning, abstract thought, and language. These processes are energy-intensive, and now we know the mitochondria there have adapted to meet those demands. Older brain structures, like those in the brainstem and basal ganglia, tended to show lower values for respiratory capacities.

“The alignment of mitochondria molecularly specialized for energy transformation with evolutionary patterns sheds light on the underlying subcellular bioenergetic infrastructure evolved to sustain the elevated energy costs unique to humans,” the authors write.

This gradient could help explain why complex cognitive functions are especially vulnerable in mitochondrial disorders. Such disorders have been linked to diseases from Parkinson’s to schizophrenia

“Energy is the missing dimension of biomedicine,” said Picard. “If you think of health as energy, it inspires you to ask different questions.”

Questions like: How much energy does it take to heal a brain injury? Can we detect early changes in mitochondria that signal the onset of disease? Do the foods we eat influence how our brain produces energy?

And perhaps the most intriguing: Can we peer into this hidden mitochondrial world using standard brain scans? The team believes that’s possible.

What This Means for the Future

Using brain scans from nearly 2,000 healthy adults, the researchers trained a statistical model to predict mitochondrial features based on standard MRI measures, like tissue density, functional activation, and neurite complexity.

Astonishingly, the model could predict mitochondrial characteristics — such as enzyme activity or density — across brain regions not included in the initial tissue slice.

Maps generated from MRI alone closely matched those derived from direct biochemical analysis. In essence, the team had built a tool to estimate mitochondrial activity from noninvasive imaging.

“Our neuroimaging-based estimates of mitochondrial specialization across the brain showed a significant correlation with maps estimating brain evolution,” they note.

This paves the way for personalized “mitochondrial maps” in living humans — potentially transforming how we study brain development, aging, and disease.

But the study has limits. It’s based on a single brain. Biological variation between individuals, ages, and sexes remains unexplored.

Still, the researchers are optimistic. Their method — freezing, cutting, testing, mapping — can be scaled up. And their model can already be applied to existing brain scans.

The findings appeared in the journal Nature.