For billions of years, the machinery of life has assembled proteins from the same twenty molecular building blocks, a constraint so ancient it seemed immutable. Researchers at Harvard Medical School have now quietly dismantled that boundary, developing a cell-free system called AGENTEX that works with 34 amino acids simultaneously — not by rewriting the genomes of living organisms, but by reconstructing the essential protein-building machinery in a test tube. In doing so, they overturned a foundational belief held for more than two decades: that the molecular couriers responsible for protein a
Harvard scientists expand protein-building alphabet to 34 amino acids without genetic engineering
We can alter the most fundamental system found in nature and it's functional.
Mark
Why does it matter that they can now use 34 amino acids instead of 20? What do the extra 14 actually do?
Mimi
Nature settled on 20 amino acids billions of years ago, and that was enough to build every protein in every living thing. But those 20 have specific chemical properties. The extra 14 that researchers can now add have different properties—they might be stickier, or more rigid, or reactive in ways nature never needed. That means you can design proteins that do things natural proteins can't.
Mark
And they did this without touching DNA at all. How is that possible?
Mimi
They worked in test tubes with just the molecular machinery cells use to make proteins—no actual cells. They engineered the transfer RNAs and ribosomes to work with new genetic codes, then added them to a cell-free soup. The proteins assemble themselves without any living organism involved.
Mark
What was the big discovery about transfer RNAs?
Mimi
For decades, everyone believed transfer RNAs had to end with the sequence CCA or they wouldn't work. It was treated as biological law. Radford's team found that wasn't true—they can end with other sequences like CGA and still function perfectly. It's like discovering a fundamental rule of nature was actually just a habit.
Mark
Why did it take so long to figure that out?
Mimi
Because no one had a reason to test it. The dogma was so strong that alternative sequences were never seriously investigated. You don't look for exceptions to rules you're certain about.
Mark
So what happens next? Can they actually use this in medicine?
Mimi
Not immediately. The system works in test tubes, but moving it into living cells or organisms requires more work. But now they have a platform to design and test thousands of protein variants quickly. That's the real acceleration—instead of spending a decade engineering one new amino acid into cells, they can design and test thousands of possibilities in weeks.
The Pulse
- Progress in expanding the protein alphabet had stalled at a punishing pace — one new amino acid per decade, each requiring years of dangerous, painstaking genome rewriting inside living organisms.
- The discovery that transfer RNAs can function with alternative tail sequences shattered a 20-year-old dogma, unlocking a door that the entire field had assumed was permanently sealed.
- AGENTEX bypasses living cells entirely, running its protein-building experiments in test tubes with robotic software — eliminating containment risks and collapsing the timeline from years to parallel batch tests.
- The system now allows researchers to customize 34 codons simultaneously and screen thousands of protein variants at once, a leap so dramatic it redefines what protein engineering can realistically attempt.
- The platform is already being aimed at medicine, agriculture, materials science, and environmental cleanup, with AI integration planned to push its optimization capabilities even further.
For billions of years, the machinery of life has assembled proteins from the same twenty molecular building blocks, a constraint so ancient it seemed immutable. Researchers at Harvard Medical School have now quietly dismantled that boundary, developing a cell-free system called AGENTEX that works with 34 amino acids simultaneously — not by rewriting the genomes of living organisms, but by reconstructing the essential protein-building machinery in a test tube. In doing so, they overturned a foundational belief held for more than two decades: that the molecular couriers responsible for protein assembly require a specific genetic tail to function at all. What was once a decade-long endeavor per amino acid has become a platform capable of generating thousands of novel proteins in parallel, opening a new chapter in humanity's long effort to engineer life's most versatile molecules.
For more than two decades, scientists have chased a deceptively simple goal: persuade living cells to build proteins that nature never intended them to make. The potential rewards — new medicines, novel materials, tools for environmental repair — were clear. The path was not. Rewriting an organism's DNA to reprogram its protein-making machinery proved so slow and difficult that progress arrived in the smallest possible increments: one new amino acid per decade.
Researchers at Harvard Medical School and the Wyss Institute have now bypassed that obstacle entirely. Their system, called AGENTEX, abandons living cells altogether. Instead, it reconstructs only the molecular components cells need to build proteins — and runs the entire process in a test tube. The result is a protein-building alphabet expanded from nature's 20 amino acids to 34, achieved without touching a single genome.
The breakthrough hinged on overturning a belief that had stood unchallenged for over twenty years. Scientists had long held that transfer RNAs — the molecules that ferry amino acids into growing protein chains — must end with a specific three-letter sequence, CCA, or they simply would not work. Felix Radford and colleagues in George Church's lab proved otherwise, demonstrating that alternative tail sequences, including CGA, are fully functional. That single discovery unlocked the entire platform.
The practical shift in scale is difficult to overstate. In 2013, after nine years of effort, Church's team freed one codon in E. coli. A decade later, they had freed a second. AGENTEX allows researchers to customize 34 codons simultaneously, design the molecular machinery to use them, and test thousands of protein variants in parallel — all in a beaker, in a fraction of the time. Because the engineered components use alternative genetic sequences, they would not disrupt natural protein synthesis if eventually introduced into living cells.
Radford described the discovery as a fundamental shift in understanding one of biology's most ancient and conserved processes — protein synthesis that has remained essentially unchanged across all organisms for billions of years. The team also uncovered previously unknown details about how DNA is translated into proteins, a process so well-studied it had seemed fully mapped.
Applications are envisioned across medicine, agriculture, materials science, and environmental remediation. The platform also offers new tools for studying transfer RNA mutations linked to diseases such as diabetes and hearing loss. Church's team is already planning to integrate artificial intelligence to further refine protein design. The bottleneck that once consumed a decade per amino acid has been replaced by a system that works with 34 at once — and the central question now is simply how quickly researchers can learn to use it.
For more than two decades, scientists have pursued a deceptively simple goal: convince living cells to manufacture proteins that nature never designed them to make. The payoff could reshape medicine, agriculture, materials science, and environmental cleanup. But the path has been brutally slow. The conventional approach—rewriting an organism's DNA to reprogram its protein-making machinery—proved so difficult and time-consuming that it yielded only marginal results.
Now researchers at Harvard Medical School and the Wyss Institute for Biologically Inspired Engineering have sidestepped the problem entirely. Rather than editing genes inside living cells, they built a system that works in test tubes, using only the molecular components cells need to build proteins, without the cells themselves. The tool, called AGENTEX, expands the protein-building alphabet from nature's 20 amino acids to 34—and it does so without touching an organism's genome at all.
The breakthrough rests on a discovery that upends decades of molecular biology doctrine. Scientists have long believed that transfer RNAs, the molecules responsible for assembling amino acids into protein chains, must end with a specific three-letter genetic sequence: CCA. Any other ending, the thinking went, would render the molecule useless—enzymes wouldn't load it with amino acids, and ribosomes wouldn't let it into the protein factory. Felix Radford and colleagues in George Church's lab found that assumption was wrong. They discovered that transfer RNAs can function with alternative tail sequences, including CGA. This flexibility opened a door that had seemed permanently locked.
The implications are staggering. In 2013, after nine years of work, Church's team freed up a single codon in E. coli bacteria, allowing the microbes to incorporate one new amino acid. A decade later, they had managed to free up a second codon. Progress measured in single amino acids per decade. AGENTEX collapses that timeline entirely. The system allows researchers to customize 34 codons simultaneously, design the molecular machinery needed to use them, and test thousands of protein variants in parallel—all in a beaker, without the years of genome rewriting that previously accompanied each new amino acid.
The workflow is elegant. Researchers use AGENTEX software, which runs on an open-source robot, to design transfer RNAs with different non-CCA endings and any of the 34 amino acids. The system batch-tests them in cell-component solutions to identify which variations work best. Once successful variants are identified, AGENTEX designs matching ribosomes and adds everything to more solutions, where the new proteins assemble themselves. Because these engineered transfer RNAs and ribosomes use alternative genetic sequences, they won't interfere with natural protein-making machinery if the system is eventually moved into living cells.
Radford emphasized that the discovery itself—that the CCA tail is far more flexible than anyone had assumed—represents a fundamental shift in understanding one of life's most ancient and conserved processes. "We've shown that we can alter one of the most fundamental portions of one of the most fundamental systems found in nature, the protein synthesis system that has existed pretty much unchanged for billions of years across all organisms, and it's functional," he said. The work also revealed unexpected science: the team discovered something previously unknown about how DNA gets translated into proteins, a process so basic that it seemed fully mapped.
The applications stretch across multiple industries. In medicine, the ability to rapidly design and test novel proteins could accelerate development of new therapeutics. The work also offers a new lens for studying transfer RNA mutations linked to diseases like diabetes and hearing loss. Beyond health, researchers envision using AGENTEX to engineer proteins for agriculture, create new materials, and develop solutions for environmental remediation. Church's team is already planning to integrate artificial intelligence into the platform to further optimize protein design.
What makes AGENTEX transformative is not just the expanded amino acid alphabet, but the speed and safety it introduces. Previous methods required handling living organisms and altering their genomes—work that is slow, risky, and demands careful containment. AGENTEX operates in test tubes with no living cells and no genome modification. Thousands of unique molecules can be built, tested, and evolved in parallel. The bottleneck that consumed a decade per amino acid has been replaced by a platform that can work with 34 at once. The question now is not whether this approach works, but how quickly researchers can learn to use it.
Notable Quotes
This is more rapid and safe than existing methods, as it does not rely on handling living cells or altering their genomes.— Felix Radford, HMS research fellow in genetics
The breakthrough turns protein engineering into something closer to a molecular design and discovery platform. Thousands of unique molecules can be built, tested and evolved in parallel.— Felix Radford
Want the full story?
Read the original at Phys.org ↗
Education