Released to Repeat

The guest woken on purpose and designed to keep acting without its keepers — and the honest catch: it has not been let go yet

R856 left Sleeping Beauty leashed by absence: the enzyme arrives separately, does its one jump in a dish, and is gone, so the act cannot repeat. Its closing turn asked for the deliberate inverse — a kept thing woken on purpose and then released to keep acting on its own, out in the world, generation after generation, where the keepers’ whole design is that the act does repeat without them. That design exists. It is called a gene drive. And the first honest thing to say about it is that, for mosquitoes, the release at its heart is still prospective: the repeating has been shown in cages, not yet in the wild.

A cutter that carries its own instructions for copying itself. A CRISPR gene drive is a cassette inserted at a chosen site in a chromosome, carrying both the Cas9 cutter and a guide RNA aimed at that same site on the other, wild-type chromosome. In an animal that inherits the drive from one parent only, the cassette cuts the partner chromosome and the cell repairs the break using the drive chromosome as its template — copying the whole cassette across. A heterozygote becomes, in effect, a homozygote, and instead of passing the trait to about half its offspring, as ordinary inheritance would, it passes it to nearly all of them. That is the whole trick, and it inverts R856’s arrangement exactly: Sleeping Beauty’s cargo was stripped of its enzyme so it could never move again; a drive’s cargo is its enzyme, packaged to move itself, forever, in every generation that inherits it.

In cages, it finished the job. Kyrou and colleagues reported in Nature Biotechnology in 2018 a drive in the malaria mosquito Anopheles gambiae aimed at the boundary between intron 4 and exon 5 of doublesex, a sex-determination gene whose female-specific exon is highly conserved across Anopheles. Males carrying the disruption developed normally and stayed fertile; females with two disrupted copies were intersex and completely sterile. Released into caged populations, the drive spread to full prevalence within seven to eleven generations while egg production fell to total population collapse. Resistant variants at the cut site did arise, but in those experiments none that also preserved the gene’s function was selected — the target’s own constraint was part of the design. This is a keeping whose endpoint is not keeping at all: a population that carries the guest in every individual, briefly, and then is gone.

Not yet released — and that pause is part of the story. Target Malaria, the consortium developing this line of work, has taken a staged path. Its actual field release in Burkina Faso in July 2019 was deliberately not a drive: several thousand genetically modified sterile males, a modification expected to disappear, released to learn how such mosquitoes fare and how a release is governed. The self-sustaining drive strains remain, by the consortium’s own account, at the laboratory stage. So the opening’s premise needs its tense corrected: nobody has yet decided to let this guest remember in the wild. What has been decided, repeatedly and in public, is the argument about whether anyone should — because a drive that works as designed will not stay where it was released, will not respect a border, and, critics and many developers agree, could not simply be recalled once it is spreading on its own.

The proposed recall is another guest of the same kind. The countermeasures on offer do not reach in from outside; they ride the drive’s own machinery. Xu, Bulger, Gantz, Bier and colleagues reported in Molecular Cell in 2020, in fruit flies, two neutralising elements: one that uses the drive’s Cas9 to copy itself while inactivating that Cas9, chasing the drive down over several generations, and a second, placed at the drive’s own site, that uses the same Cas9 to cut the drive out and replace it. In other words, the only proposed way to call the guest back is to release a second guest built to hunt the first. The senior author’s own caution, in summaries of the work, is the sober note to end on: such elements should not be treated as a safety guarantee that licenses a release, because unforeseen scenarios remain. A leash made of absence can be checked in a dish. A recall made of another drive can only be checked the same way the first one would be — by letting it go.

So the keeping run reaches the shape where keeping and releasing become the same act. RAG was kept by being disarmed in place. Sleeping Beauty was summoned, used once, and disarmed by protocol. A gene drive is kept only by being designed to dispense with its keepers: its success is measured by how completely they become unnecessary. And the argument its makers are actually having — staged releases, non-drive first, years of consultation before any drive release is even sought — is whether a design whose virtue is irreversibility can ever be tested at less than full consequence.

Three stages. One cassette, one cage, one unopened door. Walk them.

Toy, schematic only: the three stages below are a way of walking the arrangement in order, not a model of any actual population, release, or outcome.