Gene editing still has a few bugs in the system

Gene editing has been in the news lately due to an ethically reckless experiment in which human embryos were subjected to an inefficient form of gene editing. The subjects, now born, gained uncertain protection from HIV in exchange for a big collection of potential risks. A large number of ethicists and scientists agreed that this isn't the sort of thing we should be using gene editing for.

That response contains an implicit corollary: there are some things that might justify the use of gene editing in humans. Now, a series of papers looks at some reasonable use cases in mice and collectively finds that the technology really isn't ready for use yet.

Use cases

Gene editing will likely always come with a bit of risk; when you're cutting and pasting DNA in millions of cells, extremely rare events can't be avoided. So the ethical questions come down to how we can minimize those risks and what conditions make them worth taking.

In the new work, spread across three papers, researchers tackle two genetic diseases that seem to tick most of the boxes for "worth the risk." The disease are Duchenne muscular dystrophy (DMD) and Hutchinson–Gilford progeria syndrome (HGPS). There are no effective treatments for either disease, and affected individuals die in their 20s or teens. Neither can be screened for effectively. For DMD, a third of the cases are due to new mutations; for HGPS, all of them are.

Both can be addressed by gene editing in mature cells rather than by editing in germ-line cells or early embryos. And the editing that is needed to return a cell to normal is relatively simple. So all in all, a fatal disease that is difficult to screen for and has no treatments can be addressed by relatively simple gene editing. These are exactly the use cases that many would consider ethical.

If these are the sorts of disorders we could treat with gene editing, how close are we to doing so? To find out, researchers decided to test out the systems in mice that had been engineered to carry mutations that cause these diseases in humans.

Tackling progeria

HGPS is caused by a mutation in a gene that encodes two proteins that help structure the cell's nucleus and the DNA it contains. The damaged form of one of those protein gets stuck immediately after it's sent into the nucleus, distorting the structure and altering gene activity. The result is a set of symptoms that resemble early and accelerated aging.

It's possible to delete a stretch of DNA that's specific to the damaged protein, while leaving the other protein untouched. So, two different research teams made CRISPR/Cas-9 constructs that would target this stretch of DNA and engineered them into a virus that can infect cells but can't duplicate once it's inside. These were tested in both human and mouse cells and found to work. So the researchers turned to actual mice.

It's unrealistic to think these viruses would infect every cell in the body. But past research found that mice with a mix of healthy and mutant cells could develop normally—the problems are apparently dependent in part on interactions among cells. The question is whether the virus can infect enough cells to make a difference. The answer is... sort of?

After a systemic infection, editing varied based on tissue. Over 13 percent of liver cells were edited, while only one percent of lung cells were; heart and muscle were somewhere in between. And these levels were somewhat helpful. The mice saw a slight improvement in body weight, and they lived a bit longer. As a percentage change, the difference was pretty large (over 25% in one study, even longer in the other), but that's in part because the life span was pretty dramatically shortened by the mutation.

Somewhat disturbingly, one of the two papers suggest that the mice die due to severe constipation.

Restoring muscle

DMD is another great candidate because the disease-causing mutations tend to occur in one of a large series of repeated units that encode an extended structure of the protein. It's possible to delete the damaged repeat and make a slightly shorter but still functional protein. Which is what the authors of the new research did. This wasn't the first time it had been tried, but this was the first report that followed the mice for a year after the gene editing took place.

That year turned out to be significant because editing continued to happen in the background at low levels throughout the year, meaning that the percentage of edited cells gradually went up. This happened even though the gene-editing virus didn't infect new cells, which suggests that it remained in place for long periods.

While the researchers also used a systemic infection of the virus, they tried to restrict the gene editing to muscles by engineering the virus with a stretch of DNA that ensured the editing genes were only active there. Except that didn't work, as some level of editing was found in many other organs. This included the testes, suggesting that a gene-edited version of the DNA might accidentally be passed on to the next generation.

Other potential problems included an immune response to both the virus that carried the editing constructs, and to the protein that did the editing. Either of these could prevent us from using multiple rounds of infections to make up for cases like HGPS, where editing in more cells appears to be important. The other thing is that a small fraction of the cells had large stretches near the targeted site that were deleted entirely. While this happened in far less than one percent of the infected cells, the number of cells we'll want to infect is extremely large.

Overall, the three papers present a consistent picture. The technology works, it's possible to fix genetic defects through gene editing, and the fixes can have an effect on health. But the editing isn't currently efficient enough to provide anything more than some partial help for people with these diseases. And that help comes with some significant risks: errors in editing, activity in the wrong cells, and an immune response that limits future interventions. We've got a lot of work ahead of us before this is ready for the clinic.

Nature Medicine, 2019. DOI: 10.1038/s41591-018-0338-6, 10.1038/s41591-019-0343-4, 10.1038/s41591-019-0344-3 (About DOIs).