On the first of June, a Columbia-led team of eighteen authors headed by Dieter Egli placed a preprint on bioRxiv with a title that reads like a grant wish: “Efficient base editing and development in human embryos without chromosomal alterations”. The claim holds up in the data. What the title leaves out is the part that keeps embryo editing out of the clinic, and it is the familiar one. In an edited embryo, not every cell gets the same edit.
The work swapped CRISPR’s scissors for adenine base editors, which chemically rewrite single letters rather than cutting both DNA strands. Two targets were chosen, PCSK9, the cholesterol gene, and HBG1/2, the fetal haemoglobin genes, and in cultured cells the guides reached editing efficiencies between 70% and over 95%. Delivered as protein at fertilization or the two-pronuclear stage, the editor produced no aneuploidy and no large deletions, embryos grew to blastocysts, and three stem-cell lines derived from them carried a normal chromosome count. Nicks and mismatches, the authors conclude, are “efficiently repaired in human embryos, allowing specific on-target changes without genotoxic consequences”.
Then the awkward cell-by-cell accounting. Of nine cleavage-stage embryos dissected into single cells, seven were mosaic, 78% by the paper’s own arithmetic, and only two carried the edit uniformly. This matters practically: clinics genotype a blastocyst by biopsy of the trophectoderm, the tissue that becomes placenta, not the inner cell mass that becomes the child. A mosaic embryo therefore defeats the very test meant to clear it. The authors’ remedy is timing, editing before DNA replication, in the first five to 12 hours after fertilization.
Notably, for all the technical noise, the salient point may be an accident of delivery. The editor worked only as a ready-made protein. Injected as messenger RNA, it stopped development at the one-to-four-cell stage, which the paper attributes to “a hitherto unknown mechanism of RNA-mediated embryo arrest in human”. Off-target editing, meanwhile, tracked the guide RNA used, so guide selection can be screened in advance.
The authors are candid about scope. The targets were picked because zygotes and tested guides were available, not for therapeutic value; neither edit would qualify for heritable use under the National Academies’ rule that only naturally occurring, common variants be considered. Standard embryo testing remains the alternative, and its low live-birth rates are the strongest argument for editing at all. Still, the paper’s last word is a brake: the work “may be a step towards heritable editing”, but “translation to a clinical context remains premature”. Mosaicism is the reason, and it is, in the group’s phrasing, “a yet unsolved obstacle for human embryo editing”.
Sources
- Study Shows Limits of “Precise” Gene Editing in Human Embryos | Columbia University Irving Medical Center
- Two Labs Just Edited Human Embryos Without Breaking Their Chromosomes | Live in the Future
- bioRxiv preprint, "Efficient base editing and development in human embryos without chromosomal alterations"
