Microbe breaks genetic code rule, producing two proteins from one instruction

For decades, molecular biologists have taught the genetic code as a kind of universal law—three letters of DNA, read in sequence, mean one thing and only one thing. A codon either specifies which amino acid to add to a growing protein, or it signals stop. The system is elegant precisely because it is rigid. But researchers at UC Berkeley have now found a microbe that does not follow this rule, and the discovery suggests that life's most fundamental instruction set is far more flexible than anyone assumed.

The organism is Methanosarcina acetivorans, an archaeon that produces methane. What makes it unusual is how it handles a particular three-letter sequence called UAG, which in nearly every other living thing serves as a stop codon—a period at the end of a genetic sentence. In this microbe, UAG sometimes functions as a stop signal, and sometimes it does not. Instead, the cell reads it as an instruction to insert an amino acid called pyrrolysine and keep building the protein. The result is that a single stretch of genetic code can produce two entirely different proteins, depending on what the cell decides to do when it encounters that ambiguous signal.

Dipti Nayak, an assistant professor of molecular and cell biology at UC Berkeley and senior author of the study published in the Proceedings of the National Academy of Sciences, explained the counterintuitive finding this way: ambiguity in the genetic code should theoretically be catastrophic. A cell that cannot reliably read its own instructions should generate a chaotic mix of proteins and malfunction. Yet Methanosarcina acetivorans appears to thrive despite this apparent sloppiness. "Biological systems are more ambiguous than we give them credit to be," Nayak said, "and that ambiguity is actually a feature—it's not a bug."

The discovery emerged from work by Katie Shalvarjian, a former graduate student now at Lawrence Livermore National Laboratory, who was investigating how methanogens that consume methylamines use a twenty-first amino acid that most organisms lack. Scientists had long known that certain archaea could produce pyrrolysine, expanding their biochemical toolkit beyond the standard twenty amino acids. Researchers assumed these organisms had simply reassigned the UAG stop codon to code for pyrrolysine instead. But when Shalvarjian examined the actual genetic sequences, she found something unexpected: UAG was not consistently translated as pyrrolysine. It was functioning as both a stop signal and a pyrrolysine instruction, seemingly at random. "The UAG codon is like a fork in the road," Shalvarjian said, "where it can be interpreted either as a stop codon or as a pyrrolysine residue."

The team searched for hidden signals in the DNA sequence or protein structure that might explain when the cell chooses one path or the other. They found none. Instead, early evidence points to something simpler and more elegant: the availability of pyrrolysine itself inside the cell appears to tip the balance. When pyrrolysine is abundant, UAG is more likely to be read as an instruction to add that amino acid and continue protein synthesis. When pyrrolysine is scarce, the same codon functions as a stop signal. This means that environmental conditions—the cell's internal chemistry at any given moment—can influence which proteins get made. Between two hundred and three hundred genes in this organism contain the UAG codon, so the potential for variation is substantial.

The implications extend far beyond archaeal metabolism. About ten percent of inherited human diseases result from premature stop codons in critical genes, producing incomplete and nonfunctional proteins. Cystic fibrosis and Duchenne muscular dystrophy are among the conditions caused this way. Researchers have long speculated that if stop codons could be made slightly "leaky"—if cells could occasionally read through them and produce full-length protein—symptoms might be eased. The Berkeley discovery suggests that such flexibility is not merely theoretical. It exists in nature, and it works. The question now is whether scientists can harness the same principle to help human cells overcome their own genetic roadblocks.

There is also a more immediate relevance to human health. Archaea and certain bacteria that consume methylamines play a role in limiting the production of trimethylamine N-oxide, a compound associated with cardiovascular disease. When people eat red meat, the liver converts certain byproducts into this molecule. Microbes that break down methylamines before they reach the liver help reduce its formation. Understanding how these organisms regulate their protein production—how they decide which version of a protein to make—could eventually inform strategies to enhance their beneficial activity.

Nayak emphasized that the finding opens a new direction for research. "This really opens the door to finding interesting ways to control how cells interpret stop codons," she said. The genetic code, long treated as immutable and universal, turns out to be more like a language with dialects—rules that can bend, meanings that can shift, and ambiguity that, under the right circumstances, becomes an advantage rather than a flaw. What happens next depends on whether human cells can learn to speak the same dialect.