In November 2018, the world was shocked to learn that two babies had been born in China with DNA edited while they were embryos — as dramatic a development in genetics as the 1996 cloning of Dolly the sheep. The official story told by the Chinese authorities is that a scientist named He Jiankui conducted a rogue experiment that resulted in the birth of the babies, non-identical twin girls, who were the first “CRISPR’d” people ever born, with genes inserted to supposedly provide them with immunity from HIV.
While much about this story remains unconfirmed and dubious, it is clear that He himself (who was subsequently jailed by the Chinese authorities) was stunned to find himself not greeted as a scientific hero, but rather excoriated around the world for having conducted a grossly reckless, irresponsible, immoral and illegal experiment.
CRISPR — an acronym for “clustered regularly interspaced short palindromic repeats” — is a powerful gene-editing method that was discovered by Spanish molecular biologist Francisco Mojica in 1993 and then turned into a workable gene-editing technology in 2012 by the biochemist Jennifer Doudna and her colleague Emmanuelle Charpentier. Doudna and Charpentier’s method uses a protein named Cas9 to enable different RNAs to cut and edit different DNAs. This discovery, which would earn Doudna and Charpentier the 2020 Nobel Prize in chemistry, enabled dramatically more precise gene-editing than had ever before been possible.
While CRISPR was immediately hailed as a technology capable of revolutionizing whole industries from chemistry to agriculture to medicine, it was also recognized as having the potential to open a new and fateful chapter in humanity’s genetic history. Even as many scientists and bioethicists called for a moratorium on all editing of the human germline — the egg and sperm cells that pass DNA from parent to child — many others regarded it as inevitable that such editing eventually would take place. And in 2018, the prediction came true.
Human germline genome editing could be regulated without being banned. Some limits on what it could be used for, in what settings and by what kinds of intending parents, are likely to be suggested and in some places adopted. But there are also questions about who should regulate it — and how.
One way to limit human germline genome editing is to allow it to be used for specific purposes only. The likely purposes can be divided into two main classes: disease-related and non-disease-related. With autosomal recessive diseases — sickle cell anemia, beta thalassemia or cystic fibrosis — people can have two copies of the disease-linked DNA variation or one disease-linked copy and one healthy copy. People having two copies are affected by the disease, while those having only one are carriers, able to pass on the gene to their offspring. There is a strong argument to be made that if the embryo inherits two copies of the disease-linked DNA, and thus will be affected by the disease, editing its genome would be beneficial.
With autosomal dominant diseases — Huntington disease or neurofibromatosis, for example — only one disease-linked copy of the gene is needed to cause the disease. If a prospective parent had two disease-linked versions, any child would necessarily receive one of that parent’s disease-causing genes and hence would have the disease. Or when a woman has a mitochondrial disease, impairing her cells’ ability to use energy, her mitochondria with the disease-causing DNA variant would be inherited by all of her children.
What unites these cases is that preimplantation genetic diagnosis (PGD) or prenatal genetic testing cannot let the parents choose to have a healthy child; no healthy children can be born with those parents’ DNA. This is the most compelling use of human germline genome editing.
But from there, things get complicated. What if the genetic variations do not always cause the disease, or are not always “fully penetrant”? Would human germline genome editing be allowed if the children who got disease “predisposing” genes only got sick 80% of the time? Fifty percent? Twenty percent?
What if the disease would not affect all of a couple’s children but would affect all boys — a disease carried on the Y chromosome, for example, or one on the X chromosome where the mother has two disease-causing copies. Or all girls, as where the man’s X chromosome bears a dominant disease variation? What if the disease is not particularly serious, such as a mild version of red/green color blindness? What if the disease is treatable after birth, by drugs or by gene therapy? What if the treatments are not always completely effective, or completely safe?
So far we have been talking about parents whose children must inherit some disease-related genetic variations. There will also be parents whose DNA variations mean their children might or might not inherit such variations. PGD would tell those parents what variations their embryos have. And they could pick one without any disease-causing genetic variations — but only if they have enough embryos and good luck. It could be that the embryos they made do not include any with their preferred choice. If they are unable to make more — if they have a shortage of viable eggs or sperm — they could be allowed to use editing to change the DNA in one of their already created embryos to turn it from disease-bearing to healthy, or from carrier to noncarrier.
These are the “easy” cases where a disease is involved. Then there are the possible uses of human germline genome editing not related to disease. It is tempting to call these “enhancement uses,” but some of them, including the most technically plausible today, do not produce “enhanced” babies. Rather, the babies would have traits that their parents prefer — sex, eye color, hair color, skin color. Because they are not related to preventing diseases or disabling conditions, in most places they will have much less ethical and political support. A country might regulate or ban any or all such non-disease-related uses of human germline genome editing.
As with disease-related traits, PGD would be able to provide parents with what they want in some cases, but not all. Two parents with light-colored eyes probably could not use PGD to select an embryo that would become a child with dark eyes because the parents will not carry the “dark-eyed” DNA variations. They would have to use editing to get that result.
No matter what, as governments move to regulate or ban genomic editing of unborn humans, there will be uncertainties about the likelihood of diseases or traits, of seriousness or importance, of treatment or other methods of coping or adapting.
There is a semantic issue here too that could, in practice, be important. Regulation of certain uses of human germline genome editing could be framed as permitting some or forbidding others. This could set a burden of proof. If only specific uses are permitted, any use that is ambiguous or unclear would have to prove that it, in fact, fits inside the allowed scope.
But if only specific uses are prohibited, someone would have to prove that an ambiguous or unclear use fits inside the definition of prohibited uses. Does uncertainty or ambiguity mean that the use can go forward, or not?
This essay is excerpted and adapted from the author’s recent book, “CRISPR People: The Science and Ethics of Editing Humans” (MIT Press, 2021).