Ancient proteins brought back to life through inference and synthesis
In laboratories at the University of Osaka, scientists have accomplished a quiet form of resurrection — not from amber or ice, but from logic itself. By reading the evolutionary record written across modern microbial proteins, researchers reverse-engineered ancient light-sensing molecules that last existed billions of years ago, then coaxed living bacteria to produce them. The work asks, and begins to answer, one of biology's oldest questions: how does a single ancestral design give rise to so many different forms of life?
- Conventional tools for tracing protein ancestry break down when genetic sequences have drifted too far apart — a fundamental obstacle that has long blocked evolutionary biologists from seeing clearly into the deep past.
- The Osaka team built ConsistASR, a new computational method designed to handle the messy insertions and deletions that scramble standard sequence comparisons, targeting two poorly understood rhodopsin families as their test case.
- Synthesized ancient protein sequences were inserted into living E. coli bacteria — a high-stakes experimental moment where computational theory had to meet biological reality.
- Both reconstructed proteins folded correctly, matured stably, and behaved exactly as evolutionary models predicted, validating the entire approach and producing functional ancient molecules for the first time.
- ConsistASR has been released as an open tool, meaning the same pipeline that resurrected these rhodopsins could now be turned toward any protein family whose evolutionary origins remain obscured.
There is a quieter version of resurrection happening in science right now — not dinosaurs from amber, but ancient proteins rebuilt from the logic of modern ones. Researchers at the University of Osaka have done exactly this with microbial rhodopsins, a family of proteins embedded in the membranes of countless microorganisms, where they perform jobs ranging from ion pumping to light sensing. The central mystery is how one ancestral protein diversified into so many distinct functions — and answering it requires reconstructing what that ancestor actually looked like.
The obstacle is structural. While the membrane-spanning cores of rhodopsins are conserved across species, the regions extending outside the membrane vary so wildly that standard sequence-alignment tools cannot trace them backward through evolutionary time. Haruto Ishikawa and his colleagues addressed this directly by developing ConsistASR, a method built to account for the insertions and deletions that accumulate in those variable regions. Applying it to two rhodopsin families — schizorhodopsins and heliorhodopsins — they computed ancestral sequences and then synthesized them, inserting the ancient proteins into living E. coli bacteria.
The proteins worked. The ancestral schizorhodopsin pumped protons in response to light, just as its modern descendants do. The ancestral heliorhodopsin, consistent with its living relatives, did not pump ions. Both showed the characteristic colors and spectral signatures of functional rhodopsins. Computational inference and synthetic biology had together produced molecules that behaved exactly as evolutionary theory predicted they should.
The deeper significance lies in what comes next. ConsistASR is now available as an open tool, and the same pipeline could be applied to any protein family whose modern variants have diverged enough to hide their common origins. Each reconstruction adds another legible passage to the evolutionary record written in proteins themselves — a record that, until now, remained largely unreadable.
The dream of resurrecting extinct life from preserved genetic material belongs to science fiction—mosquitoes in amber, dinosaurs walking again. But there is a quieter, more tractable version of resurrection happening in laboratories right now. Researchers at the University of Osaka have figured out how to bring ancient proteins back to life, not by finding them frozen in time, but by reverse-engineering them from the proteins that exist today.
The proteins in question are called microbial rhodopsins. They sit embedded in the cell membranes of countless microorganisms, where they perform a surprising range of jobs. Some pump ions across the membrane. Others sense light. All of them belong to the same protein family, which raises a puzzle that has long intrigued evolutionary biologists: how did a single ancestral protein diversify into so many different functions? To answer that question, you need to understand what the ancestor looked like—and that requires reconstructing it from the genetic sequences of its modern descendants.
The challenge is structural. Rhodopsins all share seven transmembrane domains—the parts that thread through the cell membrane—and these domains are remarkably similar across different species. But the regions that extend outside and inside the cell, called extramembrane domains, vary wildly. This variation makes it nearly impossible to use conventional sequence-alignment techniques to trace the evolutionary path backward. You cannot line up sequences that have drifted so far apart.
Haruto Ishikawa and his colleagues at Osaka took a different approach. They developed a new analytical method, called ConsistASR, that specifically accounts for insertions and deletions in those variable extramembrane regions. Using this technique, they analyzed two types of microbial rhodopsins—schizorhodopsins and heliorhodopsins—and reconstructed what their ancestral versions would have looked like. Then they did something remarkable: they synthesized those ancient sequences and inserted them into living bacteria, specifically Escherichia coli, to see if the proteins would actually fold and function.
They did. Both ancestral proteins matured into stable, functional molecules. The ancestral schizorhodopsin displayed the same light-driven proton-pumping activity as its modern counterparts. The ancestral heliorhodopsin, consistent with current heliorhodopsins, did not pump ions. The proteins showed the distinctive colors and spectral properties you would expect from functional rhodopsins. In other words, the reconstruction worked. The ancient proteins, brought back to life through computational inference and synthetic biology, behaved exactly as evolutionary theory predicted they should.
What makes this work significant is not just that it succeeded, but that it opens a door. The researchers have released ConsistASR as an open tool for other scientists to use. The same approach that resurrected these two rhodopsin families could be applied to other protein families entirely—to any proteins where modern variants have diverged enough to obscure their common ancestry. Each successful reconstruction adds another piece to the puzzle of how proteins evolve, how a single ancestral design can branch into dozens of specialized functions. It is a way of reading the evolutionary record written in the proteins themselves, and now that record is becoming legible.
Citações Notáveis
Rhodopsins all have seven transmembrane domains that are very similar, but their extramembrane domains vary dramatically, making standard sequence alignment techniques very challenging— Haruto Ishikawa, lead author
Both ancestral sequences produced stable, mature proteins in bacteria that had distinctive color and characteristic spectral properties, just like existing rhodopsins— Yasuhisa Mizutani, senior author
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter that these ancient proteins still work? Couldn't you just study the modern ones?
The modern ones are the end result of millions of years of change. If you want to understand how a protein family evolved—how one ancestor became many different things—you need to see the intermediate steps. The ancient protein is a waypoint.
But how do you know your reconstruction is actually correct? You're guessing at what the ancestor looked like.
That's fair. But the guess is constrained by the data. You're not inventing; you're inferring from the patterns in modern sequences. And then you test it. If the reconstructed protein folds and functions like a real rhodopsin, that's evidence the inference was sound.
So the fact that it pumps protons, just like modern schizorhodopsins do—that's the validation?
Exactly. If your reconstruction was wildly wrong, the protein wouldn't fold at all, or it would be inert. Instead it does what the evolutionary logic says it should do. That's powerful.
What happens next? Do you keep going backward in time?
You could. But the more immediate use is sideways—applying the same method to other protein families. There are thousands of them. This tool could help us understand how proteins in general acquire new functions.