Human gene editing sounds like a big deal. Is it?

These genetic tools pose fraught biological and ethical questions—but how much call for them is there anyway?

October 29, 2019
Photo: TNS/SIPA USA/PA Images
Photo: TNS/SIPA USA/PA Images

Gene editing of human embryos for reproductive purposes—a medical technology often said to promise (or threaten) “designer babies”—is more or less upon us already. It has arrived far sooner than many researchers imagined, and sooner than many wanted. And recent developments show that there may be more just around the corner.

To edit the genome of an embryo, biochemical tools snip out a section of its DNA (typically a gene or part thereof) before implantation, and an altered segment is inserted in its place. This was first done to human embryos in 2015, but not for reproductive purposes via IVF. Most scientists in the field had assumed the latter was still a long way off—until Chinese biologist He Jiankui announced in November last year that he had used a gene-editing technique called Crispr to modify the genes of twin baby girls born by IVF, in an attempt to give them resistance to infection by HIV.

Notwithstanding the shock and condemnation that greeted this announcement by most other researchers in the field, Russian biologist Denis Rebrikov now says he hopes soon to get approval to use Crispr to modify embryos for IVF, to change a gene mutation that would otherwise cause deafness.

Meanwhile, a new gene editing technique has just been unveiled by a team from the Broad Institute of Harvard and the Massachusetts Institute of Technology that seems to overcome much of the imprecision incurred by Crispr—a key reason why Crispr is widely considered still too risky for use in human reproduction. With better gene-editing tools potentially on hand, the safety objections to using it in this context might lose some of their force.


The main premise of gene editing for human embryos is that it could correct genetic mutations that cause disease. Rebrikov is planning to use Crispr to target one of these in a gene called GJB2, where the mutant form leads to severe hearing impairment. As with many impairment- or disease-linked genes, the defective form of GJB2 is recessive, meaning that only if a person has it in both copies of their genome (inherited from the biological mother and father) will they suffer the consequences—in this case, deafness that requires a cochlear implant or hearing aid. If at least one parent has a “healthy” copy of the gene, it is possible in principle to select an IVF embryo for implantation that also possesses it, an approach called pre-implantation genetic diagnosis. Then no gene editing would be required.

But Rebrikov says he is in discussion with several hearing-impaired couples where both partners have double copies of the mutant form of the gene. They would then certainly pass the condition to their offspring—unless the defective gene is repaired by editing.

Despite that benefit, many researchers consider Rebrikov’s plans irresponsible, because not enough is known about the safety of Crispr to offset its use for a condition that is not life-threatening. Because the editing would be done in the very early stages of embryo development, any changes to the genome—including possible harmful ones—will be passed on to future offspring of children born this way: they are in the “germline.” (In contrast, gene therapies involving editing in the tissues of fully-developed children or adults would induce non-inheritable changes.) Besides, some people in the deaf community challenge the notion that their condition needs “correcting,” regarding that as a form of eugenics that could harm the status of hearing-impaired people in society.

Given our currently deficient knowledge of the potential consequences, germline editing of the human genome for reproduction is prohibited in some countries (including the UK) and generally disavowed by scientists. Jennifer Doudna, a biologist at the University of California at Berkeley who pioneered the Crispr methodan improvement on older techniques of editing genes, for which she will almost certainly win a Nobel prize at some point—told Nature that Rebrikov’s plans are “recklessly opportunistic [and] clearly unethical.” Others have called him a "cowboy."

Doudna had similar harsh words for He’s work. The Chinese scientist modified a gene called CCR5 in the two female embryos to a form that could make them less susceptible to HIV, so that it would not be passed from parent to child. The work seems now to have been alarmingly shoddy in motivation and execution. For one thing, it’s not clear how effective the strategy is—and in any case drugs already exist to block HIV infection, so the high-risk procedure wasn’t necessary. What’s more, He’s use of Crispr seems likely to have produced many off-target edits of the children’s genomes, with unknown long-term effects. In one of the twins, only one of the two copies of CCR5 seems to have been altered, so any protection would be compromised. To make matters worse, CCR5 has various roles in the cell, and it’s not clear what other consequences could arise from making the changes.

These problems exemplify the dangers of Crispr gene editing. The technique uses a bacterial enzyme called Cas9 to target and snip out a chosen segment of DNA, allowing a replacement strand to be inserted. The cell’s molecular machinery then patches up the join. In principle the method can be more accurate at making the edit than older approaches to gene editing but it is still far from perfect. As well as sometimes hitting the wrong stretch of DNA, it also leaves the genome susceptible to the random insertion or removal of other DNA “letters” at the place where the strand is cut. And if Crispr is used on embryos with several cells, it might not work in all of them, creating an embryo that is a mosaic of altered and unaltered cells: a genetic patchwork that would persist as the embryo develops.

Problems like this lead most researchers to conclude that it is too early to use the technique for human reproduction, and some have called for an international moratorium until the risks are better known. He’s work may have opened the floodgates too soon by showing that Crispr will not necessarily induce catastrophically uncontrolled genetic changes: the two Chinese babies seem healthy so far, although it’s still early days. The work appears to have been conducted without proper ethical clearance, and he has been fired from his position at Southern University of Science and Technology in Shenzhen, China and is said to be under house arrest while under investigation. He might eventually face prosecution.

Rebrikov, a geneticist at a major government-run IVF clinic, is undeterred by all of this. He insists that he knows how to check embryos for off-target edits and dangerous levels of mosaicism. He seems to be striking a defiant pose: when asked by Nature what he feels about the position of the World Health Organisation that it is still too early for such experiments, he responded, “What does it mean, too soon? Lenin said, ‘yesterday was too early, tomorrow it will be too late.’” Still, he says that he won’t go ahead with implanting GJB2-edited embryos for IVF until he receives approval from the Ministry of Health of the Russian Federation—which has indicated that such permission would at this point be "premature and irresponsible." He also wants to see international guidelines drawn up soon, but others say this is not something that can or should be rushed.


How the new, seemingly improved gene-editing technique will play out in this context is anyone’s guess. Called “prime editing,” it was developed by a team at the Broad led by chemical biologist David Liu. It also uses the Cas9 enzyme to cut DNA at the target position—but the enzyme is modified so that it only makes one nick at a time, opening up a “flap” on the DNA where a second enzyme can go to work to rebuild the strand according to instructions it carries with it. Once that is done on one strand of the DNA’s double helix, the editing system goes to work on the other. Doing things this way creates far less chance of random errors creeping in at the nick positions, as well as of off-target edits. The Broad team has used prime editing to replace disease-inducing mutants of genes in human cells responsible for sickle-cell and Tay-Sachs disease, which cause anaemia and a fatal nerve condition respectively. They say that in principle it should work for about 90 per cent of all the gene variants known to cause disease.

The work shows that there is still plenty of scope for improvement in how Cas9 is used. With each such advance, the prospects of avoiding some of these often nasty or fatal genetic conditions will look rosier.

Even researchers cautious about the immediate viability or desirability of human reproductive gene-editing are generally agreed that in the long term there could be a strong medical case for it. But would such use stop at avoiding disease? And what (as the arguments over GJB2 and CCR5 show) counts as “disease” anyway? When does cure become prevention—or even enhancement?

This is where the ethics of gene-editing become a quagmire. But hyperbolic headlines about “designer babies” can distort perceptions of what is truly achievable. There’s no reason to think it will become possible in the foreseeable future to order up an IVF child with à la carte traits. That, indeed, may never happen, or at least not in a manner that will look attractive to those inclined towards the idea. Even though a genetic—and thus inheritable—component has been identified for pretty much every human trait for which it has been sought, no amount of gene-editing skill is going to let you tailor the outcome for most of the behavioural traits we might care most about.

That’s because the genetic contribution is often very thinly spread across the 23,000 genes or so of the human genome. For intelligence, say (commonly at the top of any wish-list for a hypothetical designer baby), there may be hundreds or even thousands of genes that have some influence. (Only around 50 per cent of intelligence seems to be inheritable anyway, the rest coming from unknown influences that might or might not be primarily environmental.) None of those genes will be in any meaningful sense an “intelligence gene”—like CCR5 and GJB2, they are likely to have a variety of roles deeply embedded in the biochemistry of the cell, out of which observable traits like cognitive skills somehow emerge. If you tried to tweak even a dozen or so of the genes (feebly) linked to intelligence, you would surely create all kinds of other, unpredictable knock-on effects in the child, as well as incurring some off-target edits even with the best editing tools. Such tinkering might have no discernible effect on intelligence anyway, but simply wreak genetic havoc. It would be a perilous, foolhardy enterprise.

Given its limited application, then (mostly to single-gene diseases), and uncertainties about its safety, it may yet be that human germline gene editing might not be too big a deal after all. Stanford University bioethicist Henry Greely suspects this will be so at least for the next several decades. “It has few potential benefits,” he says, “particularly when compared with other, more conventional interventions, such as embryo selection.” Despite the noise about brave new worlds, we need to keep in perspective where this is all likely to be heading.