Catch Me If You Can: A Spatial Model for a Brake-Driven Gene Drive Reversal

Population management using artificial gene drives (alleles biasing inheritance, increasing their own transmission to offspring) is becoming a realistic possibility with the development of CRISPR-Cas genetic engineering. A gene drive may however have to be stopped. ``Antidotes'' (brakes) have been suggested, but have been so far only studied in well-mixed populations. Here, we consider a reaction--diffusion system modeling the release of a gene drive (of fitness $1-a$) and a brake (fitness $1-b$, $b\leq a$) in a wild-type population (fitness $1$). We prove that, whenever the drive fitness is at most $1/2$ while the brake fitness is close to $1$, coextinction of the brake and the drive occurs in the long run. On the contrary, if the drive fitness is greater than $1/2$, then coextinction is impossible: the drive and the brake keep spreading spatially, leaving in the invasion wake a complicated spatio-temporally heterogeneous genetic pattern. Based on numerical experiments, we argue in favor of a global coextinction conjecture provided the drive fitness is at most $1/2$, irrespective of the brake fitness. The proof relies upon the study of a related predator--prey system with strong Allee effect on the prey. Our results indicate that some drives may be unstoppable, and that, if gene drives are ever deployed in nature, threshold drives, that only spread if introduced in high enough frequencies, should be preferred.