Precise gene editing in human embryos still causes unpredictable DNA changes

Researchers at Columbia University, led by developmental cell biologist Dieter Egli, tested a genetic editing method called base editing on single-cell human embryos and tracked each one's development for six to seven days, the stage at which IVF embryos can normally be implanted. The team targeted three genes: PCSK9, which is linked to high cholesterol and cardiovascular disease, and HBG1 and HBG2, which are involved in producing hemoglobin and, when mutated, can cause blood disorders such as sickle cell anemia and beta-thalassemia. In some experiments the editing worked in 100% of the resulting embryo's cells, and the embryos appeared to develop normally.

But the same editing process also produced unpredictable side effects. The base editor sometimes altered DNA at additional sites in the genome, including areas near the intended target, leaving embryos with a patchwork of different genetic changes across their cells, a condition called mosaicism. Large deletions of chromosome sections also occurred, though far less often than with the older CRISPR method. Egli said the findings mean it is "currently not possible" to use the technique safely in a clinical setting. The work was published in Nature on September 9, and an earlier version had circulated as a preprint on bioRxiv.

Editing genes in an embryo means intentionally introducing damage into its DNA, then relying on the embryo's own repair machinery to fix that damage in the way scientists intend. Egli's lab tried this a decade ago using CRISPR, a technique that acts like a pair of scissors, cutting both strands of DNA so a new sequence can be inserted, then depending on the cell to glue the cut ends back together. Egli suspected this would go badly, because his lab had already found that human embryos are often poor at repairing this kind of double-strand break. That suspicion was confirmed: CRISPR editing mostly failed, deleting large sections of chromosomes and, in some cases, entire chromosomes as the embryos developed.

Base editing works differently. Rather than cutting both DNA strands, it removes a single letter from one strand of DNA and replaces it with another, more like erasing and rewriting one word than cutting a page in half. Because it does not create the double-strand breaks that trip up CRISPR, base editing avoided one major failure mode: it did not introduce aneuploidy, an abnormal number of chromosomes, which had been a problem when the same team tested CRISPR on embryos in earlier work. Nathan Treff, an author on the study and chief clinical officer at Nucleus Genomics, called avoiding aneuploidy "the biggest success" of the new work.

Even so, base editing is not free of collateral damage. As Stepan Jerabek, a research scientist in Egli's lab and the paper's lead author, put it, editing a gene requires damaging DNA first, and that damage is also what creates the risk of unwanted changes elsewhere. Separately, when the base editor's molecular instructions were present in the embryo at high levels, the embryos failed to develop at all, revealing another risk that would need to be understood and avoided before the method could be considered further.

Egli sees long-term value in editing embryos to help people who carry disease-causing mutations have healthy children through IVF. Correcting a mutation at the embryo stage, before it has been copied into billions of cells and before any disease has developed, would in principle be more efficient than treating the disease later in life. But he was clear that the current findings do not support that use: "given our findings, it is currently not possible to do so safely."

Alternatives already exist. Preimplantation genetic testing, known as PGT-M, screens embryos for specific mutations so that only unaffected ones are used in IVF, without editing any DNA. Treff argues this method is not a complete solution, since not all embryos are chromosomally normal and many patients using it still do not end up with a baby. R. Alta Charo, an emerita bioethics professor at the University of Wisconsin–Madison, said she considers heritable genome editing morally defensible if it is proven safe and effective and offers a better alternative than existing options, adding that IVF itself is an expensive, difficult procedure whose physical risk falls almost entirely on the woman undergoing it.

The publication has drawn direct criticism from organizations that oversee gene and cell therapy. The American Society of Gene and Cell Therapy, along with the International Society of Cell and Gene Therapy and the Alliance for Regenerative Medicine, issued a call in May 2025 for a ten-year moratorium on heritable human genome editing, meaning edits made to embryos, eggs, or sperm that could be passed to future generations. David Barrett, the American Society's chief executive, called the new work "unfortunate" and said it "flies in the face" of that moratorium, adding: "Science is not ready for heritable human genome editing." Bruce Levine, who chairs the ethics committee at the International Society, said he was less concerned about the basic research itself than about how it might be characterized or used by others without good intentions.

That caution is shaped by past history. In 2018, geneticist He Jiankui edited a gene called CCR5 in human embryos to introduce resistance to HIV, and the edited embryos were used in IVF, resulting in the birth of twin girls carrying a mosaic pattern of edits, meaning not all their cells carried the intended change. He was convicted of illegal practice of medicine in China and served a prison sentence. Egli and Treff's team has stated its work was never intended to produce embryos for implantation, and the preprint underlying the study describes any path toward heritable editing as "premature" for clinical use. Treff said his company will continue basic research on embryo editing while ethical and regulatory frameworks catch up, arguing that research should not be held back even as controls are needed to prevent unauthorized clinical use.