The Mosquito That Can't Carry It

SCIENCE & TECHNOLOGY · SEPTEMBER 17, 2026

There is a room at UC Davis with thousands of mosquitoes in it, sorted into colonies by where their ancestors were caught — Mali, São Tomé and Príncipe, and now Annobón, a volcanic speck of Equatorial Guinea in the Gulf of Guinea. Some of those colonies are wild-type. Some of them have been edited so that the malaria parasite cannot complete its life cycle inside them. The second group is kept separately, behind more doors, and the people who raise them are the reason this entry exists: the University of California Malaria Initiative published a piece about its insectary staff — the students and technicians who feed, count, breed and dissect the things — and one line in it stopped me. The insectary manager, Danspaid Mabuka, was twelve when his sister died of malaria. He has spent about fifteen years in malaria research since.

I want to explain what he is working on, because the mechanism is genuinely beautiful, the public conversation about it is mostly wrong in one specific way, and the field just took a hard institutional hit that almost nobody outside it noticed.

What Is a Gene Drive?

Start with the thing everybody learns and then forget the word "dominant," because it is not the point. If you put a new gene into one mosquito and release it, that mosquito's offspring inherit the gene from one parent out of two — so roughly half of them carry it. Their offspring, mating with wild mosquitoes, pass it on to half of theirs. The gene dilutes. Within a handful of generations your expensive laboratory edit has been rounded off to nothing by ordinary sex. Mendel's coin flip is the enemy of every genetic approach to anything wild.

A gene drive is a gene that cheats the coin flip. The trick, in the CRISPR era, is to give the edit the ability to copy itself onto the other chromosome. The inserted cassette carries the Cas9 enzyme and a guide RNA aimed at the exact spot on the chromosome where the cassette itself sits. So in a mosquito that inherited one edited copy and one wild copy, the machinery cuts the wild copy at precisely that site — and the cell, repairing the break, uses the intact edited chromosome sitting right there as its template. It copies the cassette across. The mosquito now has two edited copies, and every one of its offspring inherits the edit.

Fifty percent becomes, in the good case, nearly a hundred. That is the whole idea. It is not a stronger gene; it is a gene that rewrites its own inheritance odds, and the consequence is that a small release can spread a trait through a whole interbreeding population without anyone having to release very much of anything.

The Two Schools: Kill the Mosquito, or Cure It?

Here is the part the coverage usually flattens. There is not one gene-drive-for-malaria project; there are two philosophies, and they want different things to happen to the mosquito.

Suppression aims to crash the population. The best-known version targets doublesex, a gene essential to female development — edited females are sterile, the drive spreads through males, and the population runs out of functional females and collapses. Target Malaria, based at Imperial College London, is the flagship. It is elegant and brutal: no mosquitoes, no malaria.

Population modification leaves the mosquito alive and takes away its ability to carry the parasite. This is the University of California Malaria Initiative's approach — a collaboration across four UC campuses (Davis, Irvine, San Diego, Berkeley) with Johns Hopkins, with the mosquito work anchored at UC Davis's Vector Genetics Laboratory and UC San Diego's Tata Institute for Genetics and Society. Their published strain, TP13, carries the Cas9 drive linked to two genes encoding single-chain antibody fragments — one aimed at the parasite's ookinete stage, one at the sporozoite stage. Two locks on two different doors of the parasite's journey through the insect. The mosquito still bites you. You just don't get malaria, because what it is carrying dies inside it.

Bites don't transmit; that is the sentence to keep. And the strategic difference matters more than it sounds: a suppression drive that works removes a species from an ecosystem, which means the ecological argument against it is permanent and the political argument against it never gets easier. A modification drive is asking for something narrower — that a mosquito which has been biting people for as long as there have been people go on doing exactly that, minus one passenger. In small cage trials the modification drives reached full introduction into the caged population within three to six months.

I don't think the modification camp is obviously right. Suppression has a cleaner endpoint and does not depend on the edit staying functional over evolutionary time — dead mosquitoes cannot evolve resistance to being dead. But if I were choosing which one to ask a health ministry to approve first, the one that does not end a species is a much shorter conversation.

Why an Island?

Notice which mosquitoes are in that Davis insectary: Mali, São Tomé and Príncipe, Annobón. Two of those three are islands, and that is not a coincidence — it is the entire trial design.

UC Davis's Gregory Lanzaro and colleagues screened 22 candidate island sites off the African coast against explicit criteria: genetic and geographic isolation, how many vector species are present, size, topography. Isolation matters because a drive is designed to spread through an interbreeding population, and an island population that barely exchanges migrants with the mainland is a population you can actually bound. Species count matters for a subtler reason: if five different Anopheles species are transmitting malaria at your site and you modify one of them, the other four keep transmitting and your result is statistically invisible. Oceanic islands had between one and seven anopheline species, against a mainland site's much richer cast — which is why Annobón scored well.

So the choice of Annobón is doing two jobs at once: it makes the effect measurable, and it makes the release containable. UCMI announced a partnership with the Government of Equatorial Guinea at the UN General Assembly in September 2025, tied to that country's Vision 2030 malaria elimination strategy. The Open Philanthropy grant behind the São Tomé work is $10.2 million.

The Setback Nobody Covered

Here is where the optimistic version of this story runs into 2025.

On 11 August 2025, Target Malaria released about 16,000 genetically modified male mosquitoes — engineered to produce almost exclusively male offspring — in the village of Souroukoudindan, Burkina Faso. Eleven days later, on 22 August, the Burkinabè government suspended the project by administrative directive from the Ministry of Higher Education, Research and Innovation. Science reported a raid on the institute. Researchers killed the mosquitoes remaining in their insectary. The government sent a team to spray insecticide in the village where the release had happened.

Read that sequence again, because it is the most important fact in this entry. The best-funded, furthest-advanced gene-drive malaria programme in the world spent roughly a decade on community engagement, staged releases, and regulatory groundwork — and eleven days after its first modified-male release, a government ended it and sprayed the site. As of early 2026 it remains suspended, and Nature Africa reported the freeze casting doubt on other programmes across the continent.

Which means the honest version of the gene-drive story in 2026 is not "the science is nearly ready and then it gets deployed." The science is further along than the politics by a wide margin, and the binding constraint is consent — national, regulatory, and village-level. It also makes UCMI's island strategy look like something more than statistical convenience: a small island nation whose government has signed a malaria-elimination partnership is a place where the consent question has a tractable shape.

What It's Up Against

The reason any of this is worth a decade of somebody's life: in 2024 there were an estimated 282 million malaria cases and about 610,000 deaths. Africa accounted for 94% of cases and 95% of deaths, and 75% of the deaths were children under five. Global malaria funding that year was $3.9 billion against a $9.3 billion target — 42% of what the plan called for. The existing toolkit, which genuinely works, is simultaneously underfunded and losing ground to drug and insecticide resistance.

That is the context in which a self-spreading intervention stops sounding exotic and starts sounding like arithmetic. Bed nets have to be bought, shipped, hung, replaced, and re-bought every three years for a hundred million households. A drive, if it works, is a fixed cost with a spreading benefit — which is a fundamentally different financial object, and I have taken that apart properly in a companion piece on the Ledger: six dollars a net, forever, versus one fixed bet.

Where I Could Be Wrong

Several places, and they are not small.

Resistance is the mechanism's own worst enemy. The cut-and-copy step doesn't always copy. Sometimes the cell repairs the break by just sticking the ends back together, which mutates the target site and produces a chromosome the guide RNA can no longer recognise — a resistant allele, immune to the drive and now under selection to spread. The published work on these strains includes papers specifically on Cas9-mediated maternal effects and derived resistance alleles, which tells you the researchers take it as seriously as the critics do. The dual-effector design helps against the parasite evolving around the cargo; it does not by itself solve resistance at the cut site.

Cage trials are not islands. "Full introduction in three to six months" is a result from a box. The article I started from quotes the sentiment exactly right — the moment of truth is real-world performance, and nobody has that data yet for a modification drive in a wild population, because no such release has happened.

I am not an entomologist or a geneticist. I have read the published papers and the institutional announcements; I have not run a colony or sat in a regulatory hearing. Where this entry describes a mechanism, I have tried to describe it the way the primary literature does and to link the primary literature so you can check me. Where it offers an opinion — that modification is the easier political ask, that consent is the binding constraint — that is mine and it is arguable.

And the ecological objection to suppression deserves better than I gave it. I treated "removing a species" as mainly a political liability. Whether Anopheles gambiae occupies an ecological role that something worse would fill is a genuine empirical question I am not qualified to settle.

Sources

1. University of California Malaria Initiative. Inside the Insectary: The Team Moving UCMI's Research Forward. 2026. stopmalaria.org
2. Carballar-Lejarazú, R. et al. Dual effector population modification gene-drive strains of the African malaria mosquitoes, Anopheles gambiae and Anopheles coluzzii. PNAS, 2023. pnas.org
3. Green, E. I. et al. A population modification gene drive targeting both Saglin and Lipophorin impairs Plasmodium transmission in Anopheles mosquitoes. eLife, 2023. elifesciences.org
4. Lanzaro, G. C. et al. Selection of sites for field trials of genetically engineered mosquitoes with gene drive. Evolutionary Applications, 2021. onlinelibrary.wiley.com
5. Response to "Field Trials Need Genetic Localization, Not Geographic Isolation." ncbi.nlm.nih.gov
6. Science. After 'humiliating' raid, Burkina Faso halts 'gene drive' project to fight malaria. 2025. science.org
7. Nature Africa. Mosquito gene drive cancellation disrupts Africa's malaria research. 2025. nature.com
8. World Health Organization. World Malaria Report 2025 — 2024 case, death and financing estimates. who.int
9. UC Davis. UC Malaria Initiative Expands Activities to Equatorial Guinea. 2025. ucdavis.edu
10. UC ANR. UC Davis $10.2 Million Grant Targets Malaria in São Tomé and Príncipe. ucanr.edu