rewriting the instruction manual of life itself
For billions of years, all life on Earth has spoken the same molecular language — twenty amino acids, no more, no less. Now, a team at Columbia University has demonstrated that bacteria can not only survive but thrive on nineteen, using artificial intelligence to quietly rewrite what was once considered an inviolable rule of biology. The work, centered on redesigning the ribosome itself, suggests that the alphabet of life may be more flexible than evolution ever required it to be — and that human ingenuity may now be capable of editing that alphabet further still.
- A cornerstone assumption of biology — that all life requires exactly twenty amino acids — has been experimentally broken for the first time.
- Earlier attempts to remove even one amino acid from bacterial cells failed catastrophically, with more than half of modified proteins collapsing, forcing researchers to abandon the project for years.
- The arrival of AI tools like AlphaFold and protein language models gave scientists the ability to find unintuitive redesigns no human researcher would have conceived, turning a dead end into a breakthrough.
- By targeting the ribosome — the cell's most essential protein-building machine — researchers created a proof of concept that could theoretically extend to every system in the cell.
- The field now stands at a threshold: if this approach scales beyond bacteria and beyond the ribosome, it could mark the beginning of engineering life with capabilities that nature never evolved.
Every living thing on Earth has always relied on the same twenty amino acids to build its proteins — a molecular constant so universal it seemed untouchable. But Harris Wang and his team at Columbia University have now shown that bacteria can function with just nineteen, removing isoleucine from the cellular vocabulary entirely and watching the cells continue to grow and divide.
The path to this result was not straightforward. Wang's first attempt years ago relied on swapping isoleucine for structurally similar amino acids — a logical but ultimately failed strategy that left more than half of the modified proteins nonfunctional. He set the project aside, waiting for better tools. They arrived in the form of AI systems like AlphaFold and newer protein language models, which can not only predict how proteins fold but suggest entirely unintuitive sequences that human researchers would never have considered.
Rather than tackle all four thousand proteins in E. coli at once, Wang made a strategic decision: redesign the ribosome first. As the molecular machine responsible for translating genetic instructions into proteins, the ribosome is both the most complex and most central system in the cell. If it could operate without isoleucine, the logic followed that the rest of the cell could be redesigned the same way.
The implications extend in two directions at once. Looking forward, the work opens the door to engineering organisms with entirely novel capabilities — cells that produce proteins or respond to chemical signals that have never existed in nature. Looking backward, it raises the possibility that the earliest life on Earth operated with an even simpler molecular toolkit, accumulating the full twenty amino acids gradually as biological complexity grew. Whether this remains a proof of concept or becomes the foundation of a new era in biological engineering now depends on how far the approach can scale.
Every living thing on Earth speaks the same molecular language: twenty amino acids, strung together in different orders to build every protein a cell needs to survive. It's been this way for billions of years. But in a laboratory at Columbia University, a synthetic biologist named Harris Wang has just shown that bacteria don't actually need all twenty. They can thrive on nineteen.
The work, published this week in Science, amounts to rewriting the instruction manual of life itself. Wang and his team used artificial intelligence to redesign the proteins in E. coli bacteria, systematically removing one amino acid—isoleucine—from the cellular vocabulary. The bacteria kept working. Their proteins folded correctly. They divided and grew. It's the kind of result that sounds simple in retrospect but required solving a problem that had stumped researchers for years: how do you remove a letter from the alphabet of life without breaking everything that depends on it?
Wang first attempted this work years ago with a straightforward approach: swap isoleucine for amino acids that looked similar in size and shape. It failed. Fewer than half of the modified proteins remained functional. The cells couldn't handle the change. He shelved the project, waiting for better tools. Those tools arrived in the form of machine learning systems like AlphaFold, which can predict how a protein will fold in three dimensions, and newer protein language models that can suggest entirely new amino acid sequences likely to work. These systems don't just predict structure—they can identify unintuitive solutions, the kinds of changes a human researcher might never think to try.
Rather than attempt to redesign all four thousand proteins in the bacterium at once, Wang made a strategic choice. He focused on the ribosome, the massive molecular machine that sits at the center of every cell, translating genetic instructions into actual proteins. The ribosome is built from more than fifty different proteins plus catalytic RNA, making it one of the most complex and essential systems in the cell. If the ribosome could operate without isoleucine, Wang reasoned, then perhaps the entire cell could be redesigned the same way.
The implications ripple outward in multiple directions. For synthetic biology, the work opens a path toward engineering organisms with capabilities that don't exist in nature—cells that could produce novel proteins or respond to entirely new chemical signals. It also offers a window into deep evolutionary history. If modern life can function with a reduced amino acid set, perhaps the earliest organisms on Earth relied on an even simpler molecular toolkit, gradually accumulating the full twenty as chemistry and biology became more sophisticated.
Julius Fredens, a synthetic biologist at the National University of Singapore who was not involved in the research, called the achievement exciting precisely because it proves the principle works at all. The question now is how far the approach can extend. Can the same technique redesign not just the ribosome but the thousands of other proteins that keep a cell alive? Can it work in organisms more complex than bacteria? The answers will determine whether this is a clever proof of concept or the beginning of a new era in biological engineering.
Notable Quotes
It's very exciting that it's possible— Julius Fredens, synthetic biologist at National University of Singapore
The Hearth Conversation Another angle on the story
Why does it matter that bacteria can work with nineteen amino acids instead of twenty? Isn't that just a small change?
It's not the number itself—it's what the number represents. Every protein in every organism is built from these twenty building blocks. Removing one means every protein that used it has to be redesigned. Most proteins break when you do that. The fact that they didn't break here suggests we can rewrite the rules of life itself.
But why would you want to remove an amino acid? What's the advantage?
The immediate advantage is proof that it's possible. But longer term, if you can remove one, you can add others that don't exist in nature. You could build organisms that produce molecules no living thing currently makes, or respond to signals that don't exist in the wild. You're expanding the palette.
How did AI actually solve the problem that humans couldn't?
Humans tried obvious solutions—swap isoleucine for something similar. AI looked at the actual three-dimensional structure of thousands of proteins and found non-obvious ways to compensate. It could see patterns humans would miss because there are too many variables to hold in your head at once.
They only redesigned the ribosome, not the whole cell. Why start there?
Because the ribosome is the most critical system. It's the machine that makes all other proteins. If you can remove an amino acid from there and the cell still works, you've proven the principle works at the deepest level. Everything else becomes possible.
What does this tell us about the past?
It suggests early life might have operated with fewer amino acids. Maybe life started simpler and accumulated more building blocks over time. This work is like finding evidence that the alphabet of life wasn't always this long.