Bexorg’s machine, BrainEx, is used to restore molecular and cellular activity in dead human brain, housed in a black bucket. (Credit: Ryan Cross for Endpoints News)
June 10, 2025 10:00 AM EDT R&DIn Focus
Exclusive: Rebooting dead human brains, a biotech startup seeks to reinvent early drug testing work
Ryan Cross
Senior Science Correspondent
NEW HAVEN, CT — In an old brick building across the street from Yale School of Medicine, a jumble of plastic tubes rhythmically pulse and sway as they pump fluids in and out of a black bucket. An opaque lid obscures the contents, but waves on a computer monitor above suggest signs of life.
A surgeon gingerly lifts the bucket’s lid to reveal a wrinkled organ — pink and shiny — upside down in a liquid bath where the tubes plunge into a network of blood vessels.
It’s a human brain, removed from its dead owner, but kept ticking on a cellular and molecular level. And Zvonimir Vrselja, one of the machine’s creators, hopes that it will revolutionize neuroscience.
For the past three years, Vrselja has been building a small company to test experimental drugs. The brains are fed artificial blood to sustain the cells for roughly a day, but the organs lack the electrical signals associated with consciousness — and are heavily sedated to keep them that way.
Vrselja believes the unorthodox approach is necessary to overcome the limitations of animal studies, where countless drugs for neurological and psychiatric diseases that looked promising were ultimately unsafe or ineffective in people. The isolated brains can help drugmakers select ideal doses for clinical studies, look for glaring safety issues, and identify and monitor biomarkers to predict if a drug is working.
“We want to deal with this translational inability to predict how things are going to behave in a human,” Vrselja, the startup’s co-founder and CEO, told Endpoints News during an exclusive visit to the company’s labs last month. “This doesn’t exist anywhere else in the world.”
The startup, called Bexorg, is based on research that Vrselja conducted in Nenad Sestan’s lab at Yale University. In 2019, they published a report showing they could restore cellular activity in the brains of decapitated pigs. The news, understandably, drew Frankensteinian comparisons in the popular press — even though Sestan and Vrselja insist their approach didn’t count as bringing the pig brain back to life.
“We restore molecular activity or metabolism,” Vrselja said. “But we don’t revive brains.”
Since hiring its first full-time employee in May 2022, the now roughly 30-person startup has quietly run 100 human brains on the machine and plans to put 500 through the system this year.
“It’s actually being done in scale,” Vrselja said. “If Bexorg is successful, I think it’s going to radically change the face of drug discovery.”
Bexorg co-founders Nenad Sestan (L) and Zvonimir Vrselja (Credit: Engine Ventures, Doug Levy)
Click on the image to see the full-sized version
When Sestan was in medical school three decades ago, he was taught — the same as doctors are today — that brain cells begin to irreversibly decay within minutes of a person’s death. But he has long harbored suspicions that this was incorrect.
After all, scientists could keep brain tissue on ice for a few hours and still keep neurons from the dead organ alive in a petri dish. “The whole concept of brain death is a little bit bogus,” Sestan said. “It’s not a binary event, it’s something that goes gradually, and that means that you can reverse that process.”
That tantalizing notion lingered in Sestan’s mind for years until the mid-2010s, when he was looking for ways to study the intricate wiring of the human brain, known as the connectome. Scientists had invented high-tech ‘dyes’ that percolated along the neurons of living animals, allowing them to map the complex circuitry after the brain was removed from the animal. But doing similar studies in people, which would require killing them to remove and study the brain, was obviously a nonstarter.
But Sestan had a wild idea: If he could keep a brain from a recently deceased human nourished and oxygenated, he might be able to keep its constituent cells alive long enough to inject those circuit-tracking tracers into it and make a high-resolution map, as they were doing in animals.
Vrselja was training as a clinical radiologist in Croatia, where Sestan is also from, when he learned about the ambitious brain project in late 2015. Vrselja left his residency and moved to the US in January 2016 to join the Yale lab, where he cleaned out a storage room to house the machine. He initially focused on pig brains, which were freely available from a pork slaughterhouse nearby. Hours after the animals were killed, Vrselja would attempt to keep the cells alive.
“We were not sure if this would work,” Vrselja said. But it did. “Brain decay is a gradual process, and it can be exploited,” he said.
The experiment, published in Nature in 2019, raised a litany of questions. Could the method be used to heal injured human brains or even resuscitate dead ones? If a brain-in-a-vat regained consciousness, what would that experience, devoid of all sensory input, be like? And if regaining consciousness was possible, would that enable people to transplant their brains to a new body, in a step towards immortality? (Sestan and Vrselja say they’ve declined offers from investors and scientists who were interested in such pursuits.)
“It is fascinating, to be clear,” Vrselja said of those questions before quickly dismissing them. “These are dead brains in which we restore molecular and cellular activity,” he said. “DNA is being transcribed. And all these housekeeping functions are present, but there is no electrical or neuronal firing in these brains. There is no organized electrical activity. So these brains are not alive.”
But they can be used for a very practical purpose: Testing drugs in the brain. Sestan and Vrselja co-founded their biotech company, where the BrainEx work could move faster and attract investment backing.
“I couldn’t believe what I was hearing,” David Beyer, an investor at Amplify Partners, who contributed to Bexorg’s $20 million seed funding, told Endpoints. “I am drawn to the weirdest things, things that just sound like science fiction. And Bexorg is honestly one of the closest things to investing in science fiction.”
Zvonimir Vrselja, Bexorg’s co-founder and CEO, standing in the startup’s production line that currently has four BrainEx machines. (Credit: Ryan Cross for Endpoints News)
Click on the image to see the full-sized version
Most biology labs look the same: Benches cluttered with test tubes, pipettes and beige, boxy instruments that assess the makeup of cells and their genetic material. Bexorg’s third-floor lab is the centerpiece of its strikingly different work.
“We have a couple flights of stairs,” Vrselja said as he showed Endpoints through the facility, “or we can take the slowest elevator.”
Standing outside the brain room, we don face masks, wrap our shoes in disposable booties and cover our heads in hairnets. A dull roar of whirring motors greets us as he opens the door. “I’m always self-conscious about this stuff,” Vrselja said. “And this is where it all comes together.”
A large monitor near the entrance displays the pressure and resistance of artificial blood in the brain, just a few of the hundreds of variables Bexorg records every second. The main attraction is inside a clear-plastic tent. “This is our production line,” Vrselja said. It’s split into six smaller rooms, four of which already house a BrainEx system. Two more machines are being built. “This is enough to support 1,000 brain runs a year,” the company’s target for 2026, he added.
Two scientists hover over one device that’s hooked up to a pig brain. Another is already running an experiment on a human brain. A third device is in standby mode, waiting for the next brain to arrive.
The startup gets its brains from the procurement organizations that collect organs for transplantation or research. “There’s a consent form,” Vrselja said. “The next of kin and everyone who is involved in the donation are informed about where these brains are going, what they are being used for.”
Bexorg is notified when a brain is on the way, and its unused devices are kept on standby, ready to be hooked up to the brain as soon as it arrives, usually less than a day after a person’s death. Employees take turns handling brains that arrive after hours. “Sometimes brains come in the middle of the night,” Vrselja said. But once they are connected to the machine, the perfusion process is fully automated.
Josip Butkovic, Bexorg’s associate director of perfusion and surgery, lifts the lid of the brain bucket to reveal two simple tubes plugged into the arterial circle at the base of the brain, normally supplied by the carotid arteries on each side of the neck. A third tube connects to the vertebral artery in the back.
Devices on the left side act as the heart and lungs, pumping and oxygenating Bexorg’s artificial blood. On the right, a yellow vat dialyzes the brain fluid like a kidney. Vrselja points to another bottle that he said contains an artificial immune system. Several sensors monitor the brain and the juices that come out of it.
“We can pick up every single molecule that the brain releases,” Vrselja said.
One device regularly collects, stores and timestamps the artificial blood and cerebrospinal fluid passing out of the brain. The company will later test the samples to study how metabolites, proteins and other molecules in the fluids change over time. Studying this so-called secretome can help identify biomarkers of how a drug affects the brain, giving clues to its efficacy. Another instrument uses a laser to monitor in real-time how much of a drug is getting into the brain tissue.
“We throw drugs at the brain. We see how the brain responds,” Vrselja says. Studying these responses in a living animal or person is much messier, since drugs often impact multiple parts of the body. If Bexorg sees a biomarker move, “it has to come from the brain,” he said.
Even with the company’s technology, it is only able to confidently sustain the brains for about a day. Once the perfusion is finished, the company will run a battery of genetic and molecular tests on the precious tissue, including DNA and RNA sequencing. Every bit of data collected during and after the experiment is then fed into an artificial intelligence model for neuroscience drug discovery that Bexorg is building, an ambitious effort that could help the company eventually develop its own drugs.
One of the most surprising aspects of seeing this, once the shock of a brain-in-a-bucket wears off, is how much Bexorg has done with so little. The company’s funding and size are modest by biotech standards.
“I just didn’t believe it,” said Paul Wes, a former Pfizer scientist who is now Bexorg’s vice president of drug discovery. “The day I came, there was a Parkinson’s disease brain on the device. It was just mind-blowing to see it in reality.”
An alternative to animals
The brain isn’t just our most complex organ. It’s also the one that’s most different from what’s in lab animals. It’s little wonder that drugs for brain conditions have one of the industry’s highest failure rates.
The reasons for those failures are manifold, but Bexorg believes its brain-in-a-bucket can address three big ones: Getting the dose just right, making sure the drug actually engages its target and looking for effects on neurons and biomarkers that suggest efficacy.
So far, much of Bexorg’s work has been proving that its approach has predictive value, including on brains from people who died with diseases such as Alzheimer’s or Parkinson’s. It’s been taking molecules that drugmakers have already tested in clinical trials and making sure they behave as expected when infused into the brain. For example, the brain Bexorg showed Endpoints was two hours into a daylong infusion of avagacestat, a failed Alzheimer’s drug from Bristol Myers Squibb, to see how different doses of the drug alter amyloid beta levels.
“We can actually answer lots of questions within 24 hours, and this helps us with scale because we can put on a new brain every day,” Wes said. “We can really plow through a lot of different doses, drugs, and conditions.”
Bexorg’s emergence from stealth comes roughly two months after the FDA unveiled a plan to replace animal testing with alternative preclinical models, such as organoids and even computer simulations.
Biohaven, a neuroscience drug company down the street, is the first to disclose its partnership with the startup. In one study, Bexorg tested an experimental Biohaven drug to untangle proteins that form problematic clumps in neurodegenerative diseases like ALS, Alzheimer’s and Parkinson’s. But too much of the drug can prevent proteins that are supposed to pair up from doing so.
“It’s a fine balance,” Bruce Car, chief scientific officer of Biohaven, told Endpoints. “And Bexorg was able to define, very sensitively, how much is too much.”
It also revealed several surprises about another of Biohaven’s experimental programs for improving brain metabolism. The company had two designs of a drug that appeared equivalent in test tube studies, but one didn’t have activity in Bexorg’s brains. And the compound that did work moved some potential biomarkers in the opposite direction that the company expected, likely due to a complex interplay of different types of brain cells shuttling a metabolite between each other that isn’t captured when studying neurons in a lab dish, Wes said.
“This is a huge implication for the clinical trial,” Vrselja said. The results helped Biohaven pivot away from a disease where the drug probably wouldn’t work and focus on others where the odds of success were backed by the brain data.
The Biohaven partnership is focused on small-molecule drugs that will be taken as pills, but Bexorg also hopes to crack the longstanding struggle of getting genetic medicines into the brain. New delivery vessels that look promising in mouse brains don’t always work well in monkey brains. And ones that look good in monkeys don’t always hold up in humans.
“There’s just a litany of failures, and delivery is at the heart of many of those challenges,” University of Oxford scientist Stephan Sanders told Endpoints. Sanders and other scientists at Oxford are working with the company on designing better gene therapies, including new AAV vectors, and testing exactly how good they are at delivering their genetic cargo to particular parts of the brain. “It has the potential to revolutionize how therapies get into the brain,” Sanders said.
Bexorg CEO and co-founder Zvonimir Vrselja shows a few of the physiological variables, including blood pressure, the startup monitors during the BrainEx experiments. (Credit: Ryan Cross for Endpoints News)
Click on the image to see the full-sized version
BrainEx won’t eliminate all animal studies, and it certainly won’t replace clinical trials in living humans. One obvious constraint of a detached brain is the ability to look for safety issues in other organs, which animal studies are better at spotting. And the startup only looks at the brains over a very short time — limiting its human brain perfusions to 24 hours. It’s unknown how well the organ will hold up beyond that, although tests are underway.
And because the brains lack consciousness and a body to communicate through, they can’t reveal how drugs impact mood, memory, movement and the many other physical manifestations of being human that emanate from within our skulls. That’s an arena that the company is likely not to touch.
“You can deploy multiple methods to quench neuronal firing, and we have done all of them,” Vrselja said. “This is a safety precaution on our end.”
What if, though, the numbing medications were removed and the brains were kept on the machine for weeks? Could the mind, as happens in some seemingly brain-dead patients, snap back?
Vrselja is unsure. “Look, these brains have suffered through a lot in terms of damage and all that stuff. So I don’t really know,” he said. “But we are taking every step that we can to make sure that even if there is a possibility, it cannot occur.”
Vrselja has recruited several academic researchers to a bioethics board that will advise the company on the thorny issues of consciousness, consent from family members, and the other hard questions its research raises.
“They are being very careful with what they are doing,” Linda Coleman, associate vice provost for research policy and integrity at Stanford University, and a member of Bexorg’s bioethics board, told Endpoints. “These protocols are in place for a reason, and they’re very stringent. They don’t want to get into situations where they’re eroding public trust.”
The company’s leaders largely declined to say how much its technologies cost, putting them in the broad band of more expensive than a mouse and cheaper than a human trial.
Car, the Biohaven executive, suggested that it “will cost others more down the line when they see how valuable it is.”
At the moment, Bexorg is essentially a contract research organization, albeit a highly unusual one. BrainEx enables important but normally mundane tests on things like the pharmacokinetic and pharmacodynamic properties of a drug, as well as its ability to penetrate the blood-brain barrier. When framed this way, BrainEx loses some of its shock factor. And some researchers think that’s okay.
“The individual projects, they’re not earth-shattering, they will not make headlines, but the sum total is big news, because it is potentially moving things along much faster, much more effectively, without the use of so many research animals, especially primates,” Insoo Hyun, director of research ethics at Harvard Medical School, who is also on Bexorg’s bioethics board, told Endpoints.
“In a couple of years, everyone is going to be like, it’s clear this could be done, and it’s going to make sense. I don’t really have the best answer to why we are the first,” Vrselja said. “We just acted on it.”
AUTHOR
Ryan Cross
Senior Science Correspondent
rcross@endpointsnews.com
