As Arctic waters got colder, cod evolved the ability to make antifreeze.Peter Leopold

It has taken her two decades, but Chi-Hing Christina Cheng has finally solved her ultimate cold case—a fishy mystery that extends from one frigid end of the planet to the other.

Cheng, a Chinese-born biologist, moved to the United States as a teenager and began working in Antarctica in 1984. There, she and her partner, Arthur DeVries, studied the notothens—a group of fishes that swim in the continent’s subzero waters. These animals survive at temperatures that would kill other fish because they produce their own antifreeze—a protein that courses through their blood and prevents ice from forming.

The protein is incredibly simple. It comprises the same three chemical building blocks, repeated over and over—one threonine and two alanines. This repetitive unit, which I’ll call “thralala” for convenience, is perfectly shaped to stick to ice crystals, creating a barrier that stops water molecules from joining and prevents the crystals from growing. Hence: antifreeze.

In 1997, Cheng and DeVries, both at the University of Illinois Urbana-Champaign, found that the gene that makes this antifreeze protein had an unexpected origin—it arose from an ancestral gene that makes a digestive enzyme. Coincidentally, a tiny snippet in the middle of this digestive gene had exactly the right code for making the thralala unit. Over millions of years, that snippet must have duplicated itself, again and again, turning an old digestive gene into a new ice-binding one, and allowing the notothens to survive in Antarctic waters. Very cool.

While discovering all this, Cheng and DeVries learned that at the other end of the world, Arctic cod also make antifreeze proteins, and their versions are built from exactly the same thralala units as the notothens’. The two groups had evolved almost identical antifreezes independently—a stunning example of convergent evolution, where two organisms turn up to life’s party in the same outfit. But there was a big difference between them: The cod antifreeze gene did not arise from a digestive one, and for the longest time Cheng couldn’t find its ancestor. “The gene had to have come from somewhere,” she says.

“These 1997 papers are classics,” says Aoife McLysaght from Trinity College Dublin, who studies the evolution of new genes. “The case for the Antarctic one was very clear, but all they could say was that the Arctic one wasn’t the same. It was also a new gene, doing the same thing. But it had a different origin, and what that origin was wasn’t clear.”

Now, after 22 years, Cheng has finally solved the mystery, and the answer is stranger and more convoluted than she imagined. Her team, which includes her colleagues Xuan Zhuang, Chun Yang, and Katherine Murphy, compared three species of cod that make antifreeze with four that do not. They compared pieces of antifreeze genes from the former against the DNA of the latter, in the hope of finding sequences that shared a vague resemblance. They found a hit—but in a functionless stretch of cod DNA that doesn’t include any genes at all. Somehow, this region of useless junk gave rise to a new and very useful gene. And Cheng’s team has deduced how this happened, step-by-step.

First, through random chance, a short stretch of junk DNA was duplicated twice, creating four identical segments in a row. The stretches between these segments were very close to the code for the thralala unit, and through a single mutation, one of them turned into exactly the right code. This snippet then duplicated, over and over, creating the core of a new antifreeze gene.

But for genes to be useful, they need more than the right sequence. They also need to be next to switches that allow them to be activated in the right time and place. In this case, the newborn antifreeze gene was serendipitously reshuffled to a different spot in the cod genome, where it landed next to one such switch.

At this point, the ancestral cods had a gene for antifreeze and could switch that gene on. But they needed one last tweak: a small signal sequence that acts like a shipping label, allowing the antifreeze proteins to be exported from the cells that make them and into the rest of the animal’s body. Fortunately, that signal was already almost in place: It took just one mutation—the loss of a single DNA letter—to create it.

These changes—duplication, mutation, relocation, and deletion—are the kinds of genomic shenanigans that happen to DNA all the time. The first steps, Cheng thinks, were likely completely random. But once the fish acquired those first thralala units, natural selection had something to work with. Those fish could probably have made small amounts of weak antifreeze, gaining an incremental advantage. Those fish with more thralala units did better; those that brought those units under the auspices of a switch did better still; and those that evolved the shipping label did best of all. All this happened as the Arctic gradually cooled over a substantial amount of time. Slowly, junk became gold. “This goes to show how creative evolution can be,” Cheng says. “It used a sequence we thought was useless and turned it into a vital adaptive protein.”

“This protein is particularly easy to evolve because you just need this [thralala] pattern,” McLysaght says. “But it’s still a very nice example of a new gene arising from scratch, from a noncoding sequence. It’s really neat.” Other proteins are much more complicated, and Cheng suspects that many of them will also have convoluted histories that began in junk. “It’s just a matter of time to figure out how they came about,” she says.

Indeed, this “is a relatively common mechanism for evolutionary innovation,” says Li Zhao from Rockefeller University, who notes that in the past decade, scientists have found many examples of genes that arose in this way. But in most of these cases, it’s not clear what those new genes actually do. The cod antifreeze is special because it’s a new gene whose role is both very obvious and very important. We not only know that it evolved, but also why—to prevent the cod from icing over.

We want to hear what you think about this article. Submit a letter to the editor or write to