Two Arctic Fish Are Breaking the Rules of Genetics

“I’m still very much puzzled about how this is possible."

A school of fish
All Canada Photos / Alamy Stock Photo

To survive in the frigid ocean waters around the Arctic and Antarctica, marine life evolved many defenses against the lethal cold. One common adaptation is the ability to make antifreezing proteins (AFPs) that prevent ice crystals from growing in blood, tissues, and cells. It’s a solution that has evolved repeatedly and independently, not just in fish but in plants, fungi, and bacteria.

It isn’t surprising, then, that herrings and smelts, two groups of fish that commonly roam the northernmost reaches of the Atlantic and Pacific Oceans, both make AFPs. But it is very surprising, even weird, that both fish do so with the same AFP gene—particularly because their ancestors diverged more than 250 million years ago and the gene is absent from all the other fish species related to them.

A March 2021 paper in Trends in Genetics holds the unorthodox explanation: The gene became part of the smelt genome through a direct horizontal transfer from a herring. It wasn’t through hybridization, because herring and smelt can’t crossbreed, as many failed attempts have shown. The herring gene made its way into the smelt genome outside the normal sexual channels.

Laurie Graham, a molecular biologist at Queen’s University, in Ontario, and the lead author on the paper, knows she’s making a bold claim in arguing for the direct transfer of a gene from one fish to another. That kind of horizontal DNA movement once wasn’t imagined to happen in any animals, let alone vertebrates. Still, the more she and her colleagues study the smelt, the clearer the evidence becomes.

Nor are the smelt unique. Recent studies of a range of animals—other fish, reptiles, birds, and mammals—point to a similar conclusion: The lateral inheritance of DNA, once thought to be exclusive to microbes, occurs on branches throughout the tree of life.

Sarah Schaack, an evolutionary genomicist at Reed College, in Portland, Oregon, believes these cases of horizontal transfer still have “a pretty big wow factor” even among scientists, “because the conventional wisdom for so long was that it was less likely or impossible in eukaryotes.” But the smelt discovery and other recent examples all point to horizontal transfers playing an influential role in evolution.


Back in the early 2000s, Graham was studying AFPs under the direction of her lab leader, Peter Davies, when she was struck by the uncanny resemblance of the smelt protein’s gene to one of the antifreeze genes produced by herring. The genes’ introns—stretches of noncoding DNA, which generally mutate faster than coding regions—are more than 95 percent identical. “The only conclusion we could come up with was that the gene was transferred horizontally,” she says.

But when the team tried to publish its findings, the researchers faced significant pushback. “We sent it to journal after journal, and we really had to fight to get it in,” Graham says. Their timing was perhaps unfortunate. In 2001, one of the major papers describing the newly sequenced human genome made extraordinary claims of horizontal gene transfers from bacteria based on comparisons with some animal genomes. These claims were quickly refuted by work showing that the lineages of species between bacteria and humans had once had the genes but lost them.

“Then we came along and tried to publish this, and it was in a climate where people were saying, ‘These claims are bunk,’” Graham says. Even after the paper was eventually published in 2008 by Plos One, the conclusions were doubted. “Probably about half the people in our field said, ‘Oh! This is really cool!’ and the other half said, ‘No. We don’t believe you,’” she says.

The disbelief in the team’s findings was understandable because the barriers to horizontal transfer in eukaryotes looked insurmountable. Horizontal transfers are common and easy in bacteria, whose DNA is just within their cytoplasm. If a DNA fragment can make its way through a bacterium’s cell wall and membrane, there’s not much to bar its integration into the genome. But eukaryotes keep their genome cloistered inside a second barrier, the nucleus, and most of the time, their DNA is tightly condensed into chromosomes that limit the opportunities for splicing into the genome. Moreover, for a horizontal transfer to establish itself in a eukaryotic species, it can’t integrate into just any cell’s DNA; it needs to end up in a germ cell, be passed on to offspring, and persist in the general population. That chain of events seemed wildly unlikely to many scientists.

Not to be deterred, Graham and her colleagues probed deeper. By cloning large chunks of the smelt genome in bacteria, they determined that smelt have only one AFP gene. When they looked at the corresponding genomic regions of other fish with published genome sequences, they found no traces of an AFP gene, not even a defunct one, although the array of genes that flanked the smelt’s AFP gene was there. “That suggested that this was something that came in; it was a new arrival in the smelt,” Graham explains. “And yet, we still had people who said, ‘No, we don’t buy it.’”

Finally, in 2019, a full sequence for the herring genome was published. It let the team better examine the sequences surrounding the AFP gene, some of which appeared to be “transposable elements” (TEs, or transposons), mobile chunks of DNA that can copy and paste themselves in a genome. The herring genome holds many copies of these TEs, but they are absent from other fish—with one telling exception. Three of them flank the rainbow smelt’s AFP gene, in the same order seen around the herring AFP gene.

Graham thinks that these sequences are “definitive proof” that a small chunk of a herring chromosome made its way into a smelt’s. “If anybody wants to dispute this,” she says, “you know, I don’t see how they possibly could.”

Cédric Feschotte, a genome biologist at Cornell University who was not involved in the study, agrees. “It seems unmistakable when you look at the data,” he says. What really intrigues him, though, is how well this finding lines up with work that he and others are doing on TEs and the rise of new genes.

For instance, in a 2008 study published in the Proceedings of the National Academy of Sciences, he and his colleagues identified a new kind of TE found in a disparate group of vertebrates, including a few species of mammals, a reptile, and an amphibian. These TEs were more than 96 percent identical in these species but were strangely absent from other examined genomes. Because the elements seemed to have appeared suddenly, Feschotte and his colleagues dubbed them “Space Invader elements” (“SPIN elements” for short) and concluded that they must have recently moved horizontally between the sundry lineages. These TEs weren’t merely genetic noise in their new hosts, either: Mice, for example, had obtained a whole new functional gene by co-opting a SPIN-element enzyme.

Since the 2008 SPIN paper, thousands of other horizontal TE transfers between animals have been reported. While these putative horizontal transfers were initially met with surprise, much as Graham’s AFP gene was, the evidence is now undeniable.

For context, it’s worth noting that horizontal transfers can be hard to detect: Over time, ever more mutations accumulate in both the original and the recipient lineages, obscuring similarities in a shared gene. Proving that a gene has been horizontally transferred also depends on demonstrating that it wasn’t once present in other related species and then lost through evolution, which can be difficult when some of those species are extinct.

“The rate of actual horizontal transfer is probably much, much higher than we realize,” Schaack says.


Although no one knows how often DNA jumps between vertebrate cells, Clément Gilbert, an evolutionary biologist at Paris-Saclay University, in France, and his colleagues found at least 975 transfers when they screened the 307 vertebrate genomes that were publicly available on GenBank at the end of 2017. What stands out in that data is that these transfers occur overwhelmingly between fishes. Almost 94 percent of the transfers involved ray-finned fishes; less than 3 percent involved birds or mammals.

“I’m still very much puzzled about how this is possible,” Gilbert says.

One explanation could hinge on herring’s famously exuberant spawning efforts. “If you were flying in a plane, and you look down at the coastline where they’re spawning, the water is milky-white in color because there’s so much sperm being released during the mating process,” Graham says. The vast majority of those sperm fail to find eggs, degrade, and release their DNA.

She thinks the DNA could stick to the gametes from other species spawning in the same area, and then get dragged into an egg cell during fertilization. For decades, genetic engineers have used a similar technique called sperm-mediated gene transfer to make transgenic organisms. There’s no guarantee that this will result in a successful DNA insertion, but sometimes it does—“and then all of a sudden, bam! There you go. You’ve got a transgenic organism,” Graham says. This could be how the herring AFP gene got into the smelt, and how many of the other horizontal gene transfers that Gilbert saw in fish occurred too.

But it can’t be the only explanation: Horizontal genetic transfers also turn up in animals whose gametes don’t have the same opportunity to pick up errant DNA.


One day, David Adelson, who studies genome evolution at the University of Adelaide, in Australia, was approached for help by his graduate student James Galbraith, who was assisting a colleague in annotating the genome of the olive sea snake (Aipysurus laevis). Galbraith was finding that some of the sequences did not relate to anything else found in reptiles. So Adelson and Galbraith dug deeper.

In the end, they discovered evidence of seven horizontal transfers of TEs into the sea-snake genome. It was hard to say exactly what species the donors had been, but Adelson said the best matches were found in fish and, in one case, corals.

Graham’s external-fertilization hypothesis doesn’t fit for these snakes, which fertilize internally and bear live young. Adelson is betting instead on the involvement of parasites. “Parasites go from species to species, they have interesting life cycles, and they can be internal,” he explains. Studies of horizontal transfers in terrestrial species have implicated parasites too.

While Atma Ivancevic, now a postdoctoral researcher at the University of Colorado at Boulder, was a doctoral student of Adelson’s, she traced the evolutionary history of TEs across vertebrates and other species. Of particular interest to her were the TEs called BovBs—small sequences of roughly 3,000 base pairs with a patchy distribution among animals. They are in cows and some marsupials, but in few mammals in between, and they also show up in snakes and other reptiles. Ivancevic searched for evidence of the loss of these transposons in the intermediate lineages, but found none.

Then the team found something even more intriguing: BovBs very similar to the ones in cows and snakes were also in ticks and bedbugs. Studies have shown that these biting parasites can pass viruses into their hosts through secreted vesicles called exosomes. If TEs can also end up in exosomes, then Ivancevic thinks that the biting parasites and their exosomes might jointly act as vectors for transferring TEs from one host to a second host.

Differences in how often various species are parasitized might therefore contribute to the frequency with which they pick up horizontal transfers. The differences might also reflect something more basic to their physiology or biochemistry: Certain organisms may be less discriminating about preventing wandering chunks of DNA from settling into their genome. Perhaps the increased abundance of transfers in fish has more to do with the animals themselves, rather than their habitat.

In reality, “there’s not a single answer,” Feschotte says. “It’s a little bit of everything.” Still, testing these hypotheses would be helpful. Not only might the research reveal potential environmental factors involved; it could help elucidate exactly how these transfers happen.


“How often am I actually taking up a piece of DNA from my environment and I don’t know it?” Schaack asks. “Maybe a lot; maybe more than I realize.” Such transfers could go unnoticed unless they happened in germ cells. The detected transfer events must therefore be only a tiny proportion of the full number of transfers that occur.

The unsettling possibility is that horizontal DNA transfers could happen all the time. For instance, a 2020 Plos Genetics paper found that TEs not only jump into mosquito genomes but can be transmitted into other species through filarial worms (nematodes) that the mosquitoes carry. “How many times have we all been bitten by mosquitoes?” Adelson asks. “My view is this happens at an astounding rate … I bet we’re constantly being bombarded.”

Whatever the rate of horizontal transfers, their cumulative impact on evolutionary history is undeniable. Take the effect of the transferred AFP gene in smelt. “We think that as soon as the smelt got this gene, it had an immediate selective advantage,” Graham says. “That selective advantage is not freezing to death in icy waters.” It meant the smelt could spread northward and enjoy free rein in the Arctic, shifting the composition of that ecosystem.

Even if transmissible elements are sometimes dismissed as “junk” DNA, they can have dramatic impacts. TEs are “the most exciting, dynamic, and potentially influential sector of the genome,” Schaack says, especially because they represent “an internal source of mutagenesis in every genome.” Not only do they alter DNA when they’re pasted in, but because they consist of repetitive sequences, their very presence increases the likelihood of genetic recombination.

“Anything you want, I can give you an example of what a transposon can do and has done,” Feschotte says. “Bringing in new genes, new regulatory sequences, rearranging chromosomes—you name it.” He points out that the immune system’s ability to generate an astronomical diversity of antibodies seems to have come from a TE that suddenly entered the genome of the ancestor of all jawed vertebrates 400 million years ago.

Gilbert agreed that TEs can be an immense source of genomic novelty. “The fact that these elements are transferred a lot between taxa through horizontal transfer, and not only through vertical evolution, means that horizontal transfer has an important impact,” he says.

Even though transfers into somatic cells are evolutionary dead ends, they could affect the health and fitness of individuals, Adelson notes. Altering any cell’s genome could have physiological ramifications, such as triggering a cancerous growth. “We just don’t really have any idea of what the impact of these things might be,” he says.

Future studies may quantify the rate of DNA transfer in our everyday lives. Recent advances in single-cell and long-read sequencing put the experiments needed to see transfers in real time within reach, Adelson says. “I think the first time someone reports that there’s a transfer of DNA from a mosquito bite to a human … that’ll get some attention,” he says.

Developments in evolutionary modeling and genetics may also finally allow us to look for ancient gene transfers that were too hard to spot just a decade ago. Gilbert believes enough whole genomes have now been sequenced that he could scan the vertebrate tree for whole-gene transfers, much as he and his colleagues detected hundreds of TE transfers. “We’ve been thinking of doing it,” he says. “We just need to do it.”