‘Gene Drives’ Are Too Risky for Field Trials, Scientists Say

In 2013, scientists discovered a new way to precisely edit genes — technology called Crispr that raised all sorts of enticing possibilities. Scientists wondered if it might be used to fix hereditary diseases, for example, or to develop new crops.

One of the more intriguing ideas came from Kevin M. Esvelt and his colleagues at Harvard University: Crispr, they suggested, could be used to save endangered wildlife from extinction by implanting a fertility-reducing gene in invasive animals — a so-called gene drive.

When the genetically altered animals were released back into the wild, the fertility-reducing gene would spread through the population, eradicating the pests.

The idea appealed to conservation biologists who had spent decades fighting a losing battle against exotic species. Some labs began running preliminary experiments. But now, three years later, Dr. Esvelt wishes he hadn’t broached the idea.

“I feel like I’ve blown it,” Dr. Esvelt, now an assistant professor at M.I.T., said in an interview. Championing the notion was “an embarrassing mistake.”

His regret arises from a study that he and his colleagues published on Thursday on the preprint bioRxiv server.

They created a detailed mathematical model describing what happens following the release of Crispr-altered organisms. And they discovered an unacceptable risk: Altered genes might spread to places where the species isn’t invasive at all, but a well-established part of the ecosystem.

Dr. Esvelt and other researchers have also been investigating the possibility of using gene drives to eradicate diseases. The most advanced of these projects seeks to wipe out malaria-carrying mosquitoes. These projects are still viable but, Dr. Esvelt warned, scientists now must be mindful of just how powerful gene drives may become.

“It’s an important contribution,” said John M. Marshall, a mathematical biologist at the University of California, Berkeley, said of the new research. “A study like this is the beginning of a formal analysis we need.”

Crispr makes it possible to build molecules that can find a particular sequence of DNA inside a cell. The molecules then snip out the sequence, allowing it to be replaced by a different one.

The technique might make it possible to introduce not just a gene engineered to reduce fertility in, say, an invasive weasel, but also the genes for the Crispr molecules themselves. Then the weasel would gene-edit itself.

Weasels inheriting just one copy of the low-fertility gene would end up with two copies, which they’d pass down to offspring. Soon the whole population of invasive weasels would be producing fewer young, until eventually the population collapsed.

Researchers at the University of California, San Diego, showed that the idea could really work by spreading a gene in fruit flies reared in the lab. Soon afterward, Dr. Esvelt’s own team showed that the process could make certain genes more common in yeast.

The National Academy of Sciences released a report on gene drives in 2016. While experts recognized a number of potential risks, they endorsed more research — possibly including “highly controlled field trials.”

So what exactly would happen if a gene drive were set loose in the wild? Dr. Esvelt collaborated with Charleston Noble, a graduate student at Harvard, and other colleagues to make an informed guess.

The researchers created a detailed mathematical model that took into account how often Crispr fails to do its job and how often mutations arise that protect a target gene from editing, among many other factors.

The model revealed that a gene drive would be remarkably aggressive. It would take relatively few engineered organisms to spread a new gene through much of a population. “It only takes a handful,” said Dr. Esvelt.

That aggressiveness might be good for eradicating an invasive weasel that couldn’t be stopped by poison baits or hunting. But if a few engineered weasels managed to escape the local environment — or were intentionally taken somewhere else — they could easily spread the gene drive throughout the weasel’s native habitat.

That may well mean that experiments in the real world are just too risky right now.

“The very idea of a field trial is that it’s a trial that’s confined to an area,” Dr. Esvelt said. “Our model indicates that this is not the case.”

“The kind of gene drive that is invasive and self-propagating is in many ways the equivalent of an invasive species,” he added.

But safer forms of the technology might be able to attack species where they’re invasive and not harm them elsewhere. In his own lab, Dr. Esvelt is investigating a gene drive that can self-destruct after several generations.

Other researchers are trying to build gene drives that are tailored to invasive populations on islands but can’t harm mainland relatives.

“I would buy into that,” said James P. Collins, an evolutionary ecologist at Arizona State University and co-chairman of the N.A.S. committee on gene drives. “Universal gene drives do have the downsides that these guys talk about.”

But when it comes to attempts to wipe out malaria, Dr. Esvelt draws a different conclusion from his data.

While self-limiting gene drives might be easier to control, they may be too weak to affect vast mosquito populations. It might well be necessary to deploy a quickly spreading gene drive.

Dr. Esvelt’s study suggests that if one nation decided to release such genetically engineered mosquitoes, neighboring countries quickly would become part of the experiment — whether they liked it or not.

International negotiations might be required before such genetically modified mosquitoes were set loose. “That’s not a question for scientists to answer on their own,” said Jason A. Delborne, a social scientist at North Carolina State University and a member of the N.A.S. gene drive committee.

Yet Dr. Esvelt would be willing to take that leap. “I have two kids,” he said. “If they lived in Africa, I would say do it.”