The disease only appeared in the fused system—a finding that underscores why assembloids matter.
In the long effort to understand how the human brain regulates itself, researchers at Stanford have crossed a meaningful threshold: they have built the first laboratory model capable of capturing the chemical conversation between serotonin-producing neurons and the developing cortex. By fusing organoids grown from reprogrammed human cells, the team not only watched serotonin shape neural synchrony in real time, but also uncovered a hidden defect in serotonin recycling among people carrying a genetic deletion linked to autism and schizophrenia. The work is a reminder that some truths about the brain only reveal themselves when its parts are allowed to speak to one another.
- For eight years, the brainstem's staggering cellular complexity defeated every attempt to isolate serotonin-producing neurons in the lab — until a genetic sensor made the invisible visible.
- Fusing serotonin-producing midbrain-hindbrain organoids with cortical organoids produced the first assembloid to model neuromodulation, revealing that external chemical input is essential for the cortex to mature and synchronize.
- When cells from patients with 22q11.2 microdeletion syndrome were used, serotonin levels collapsed in the fused system — a disease signature that was completely invisible when each organoid was tested alone.
- Fluoxetine restored normal serotonin signaling in the patient assembloids, pointing the investigation toward a broken reuptake mechanism rather than a failure of production or release.
- The platform is already being extended — toward inhibitory neuron organoids, norepinephrine sensors, and mosaic designs that could identify the precise gene driving the dysfunction.
Eight years ago, Sergiu Paşca set out to grow a simple organoid that would produce serotonin the way the brain's raphe nuclei do. The brainstem, it turned out, was anything but simple — packed with dozens of cell types in close quarters, it resisted every attempt to isolate just the serotonin-producing ones. Paşca's team was close to abandoning the effort when a new genetic sensor arrived that could detect and visualize serotonin directly. That tool changed everything.
The researchers inserted the sensor into cortical organoids, then fused those structures with their newly developed serotonin-producing midbrain-hindbrain organoids. The result was the first assembloid ever built to model neuromodulation — the process by which one population of neurons uses chemical signals to regulate another. What they observed was striking: serotonin-producing cells extended long projections into the cortical tissue, and over time the cortical neurons began firing in more synchronized patterns than those in unfused organoids. The finding offered direct evidence for something long suspected but never demonstrated — that input from outside the cortex is necessary for the cortex to develop properly.
The system's deeper power emerged when the team tested cells from people carrying a 22q11.2 microdeletion, a genetic deletion associated with both autism and schizophrenia. In the fused assembloids, serotonin levels dropped compared to controls. Crucially, when each organoid was tested in isolation, no difference appeared at all. The disease signature was emergent — visible only in the combined system, and invisible to simpler models.
Applying fluoxetine, which blocks serotonin reuptake, restored normal signaling in the patient assembloids. The problem, it seems, lies not in how serotonin is made or released, but in how it is recycled. Paşca's next step is to identify which reuptake protein is impaired and which gene in the deleted region is responsible. Mosaic assembloids — where only one component carries the deletion — could help isolate the culprit.
The model has known gaps: cortical organoids contain few inhibitory neurons, which are major serotonin targets in the living brain. But the platform's messiness may prove to be its strength. A norepinephrine sensor already exists, and Paşca envisions an entire arsenal of neuromodulatory assembloids. What began as a technical obstacle has become a new way of asking how the brain regulates itself — and how that regulation breaks down in disease.
Eight years ago, Sergiu Paşca thought he could build a simple organoid—a tiny, self-organizing structure grown from reprogrammed skin cells—that would produce serotonin the way the brain's raphe nuclei do. It seemed straightforward enough. Scientists had already mastered recipes for growing most of the cell types found in the developing brain. But the brainstem, where serotonin is made, turned out to be deceptively complex. "It's just absolutely overwhelming what happens in the brainstem," says Paşca, a psychiatry professor at Stanford. The region is packed with dozens of different cell types in close quarters, and isolating just the serotonin-producing ones proved nearly impossible.
Just as Paşca's team was accepting defeat, a technological breakthrough arrived like an unexpected gift: a genetic sensor that could detect and visualize serotonin itself. This changed everything. The researchers inserted the sensor into organoids modeled after the developing cortex, then fused those structures with their newly perfected serotonin-producing organoids—which they called midbrain-hindbrain organoids, or MHOs. The result was the first assembloid ever built to model neuromodulation: the process by which one set of neurons uses chemical signals to regulate the behavior of others. Assembloids, combinations of multiple organoids, had been used before to study how neurons migrate and form long-range connections. But none had ever captured the subtle, ongoing influence of neurotransmitters like serotonin until now.
What they saw was striking. Serotonin-producing cells in the MHOs extended long projections into the cortical organoids and released serotonin into the spaces around nerve terminals. Over time, neurons in the cortical organoids that were fused with MHOs began firing in more synchronized patterns than neurons in standalone cortical organoids. This synchronization is a hallmark of neural maturation, and it confirmed what neuroscientists had long suspected but never directly demonstrated: that input from outside the cortex is essential for the cortex to develop properly. "Certainly surprised us," Paşca says. The finding opened a new window into how the brain wires itself.
But the real power of the system emerged when the researchers tested it with cells from people carrying a 22q11.2 microdeletion—a genetic deletion associated with both autism and schizophrenia. People with this deletion often struggle with serotonin regulation, and the assembloids made from their cells showed exactly that problem. When the cortical and serotonin-producing organoids were fused together, serotonin levels dropped compared to control assembloids. Strangely, when the researchers tested each organoid separately, there was no difference. The disease signature only appeared in the fused system—a finding that underscores why assembloids matter. They reveal emergent biology that simpler models miss.
The mechanism became clearer when the team applied fluoxetine, a common antidepressant that blocks serotonin reuptake. The drug restored normal serotonin signaling in the 22q11.2 assembloids, suggesting that the problem lies not in how serotonin is made or released, but in how it is recycled back into cells. Something in the 22q11.2 deletion impairs the reuptake machinery. Paşca's next step is to identify which specific reuptake protein is affected and which gene in the deleted region is responsible. Mosaic assembloids—where one component carries the deletion and the other doesn't—could help pinpoint the culprit.
The model is not perfect. The cortical organoids contain very few inhibitory neurons, which are major targets of serotonin signaling in the real brain. Adding a third organoid containing those cells could fix that gap. And because the MHOs contain multiple cell types, they could be modified to study other neurotransmitter systems for which sensors exist. A norepinephrine sensor has already been developed. The limitation that seemed most frustrating—that the MHOs are messy and diverse—may turn out to be their greatest strength. "There could be an entire arsenal of neuromodulatory assembloids for various neuromodulators," Paşca says. What began as a technical problem to solve has become a platform for understanding how the brain regulates itself, and how that regulation goes wrong in disease.
Citas Notables
It's just absolutely overwhelming what happens in the brainstem— Sergiu Paşca, Stanford University
The phenotype only appears in the fused system, the paper makes a strong case that assembloids can reveal emergent disease biology that simpler organoid models miss— Tommaso Patriarchi, University of Zurich
La Conversación del Hearth Otra perspectiva de la historia
Why did it take eight years to build something that sounds like it should be straightforward?
Because the brainstem is nothing like a straightforward system. It's not one cell type doing one thing. It's dozens of different neurons packed together, all talking to each other. Isolating just the serotonin-producing ones turned out to be nearly impossible.
So what changed? What made it suddenly possible?
A serotonin sensor. Once they could actually see where serotonin was and what it was doing, they could work around the complexity instead of fighting it. They didn't need a pure serotonin organoid—they just needed to know what the serotonin was doing.
And when they fused the two organoids together, what surprised them?
That the cortical neurons started firing in sync. That's a sign of maturation. Everyone suspected that serotonin from outside the cortex was necessary for proper development, but they'd never actually seen it happen in a model system before.
Then they tested it with cells from people with 22q11.2 deletion. What did they find?
The serotonin signaling was broken—but only when the two organoids were fused. When they tested each piece separately, everything looked normal. The disease only showed up in the integrated system.
That's the key insight, isn't it? That you need the whole picture.
Exactly. A simpler model would have missed it entirely. You need both pieces talking to each other to see what's actually wrong.
And they found that fluoxetine—an antidepressant—fixed it?
It restored the serotonin signaling, which suggests the problem isn't in making or releasing serotonin. It's in reabsorbing it. Something in that deletion breaks the recycling machinery.