Designing gene drives to limit spillover to non-target populations

Abstract

Author summary CRISPR-based gene drive is an emerging genetic engineering technology that enables engineered genetic variants, which are usually designed to be harmful to the organism carrying them, to be spread rapidly in populations. Although this technology is promising for controlling disease vectors and invasive species, there is a considerable risk that the gene drive could unintentionally spillover from the target population, where it was deployed, to other non-target populations. We develop mathematical models of gene-drive dynamics that incorporate migration between a target and non-target populations to investigate the possibility of effectively applying a gene drive in the target population while limiting its spillovers to the non-target population ('differential targeting'). We observe that the feasibility of differential targeting depends on the gene-drive design specification, as well as on the migration rates between the populations. Even when differential targeting is possible, as migration increases, the possibility for differential targeting disappears. We find that differential targeting can be effective for low migration rates, and that it is sensitive to the design of the gene drive under high migration rates. We suggest that differential targeting could be used, in combination with other mitigation measures, as an additional safeguard to limit gene drive spillovers. The prospect of utilizing CRISPR-based gene-drive technology for controlling populations has generated much excitement. However, the potential for spillovers of gene-drive alleles from the target population to non-target populations has raised concerns. Here, using mathematical models, we investigate the possibility of limiting spillovers to non-target populations by designing differential-targeting gene drives, in which the expected equilibrium gene-drive allele frequencies are high in the target population but low in the non-target population. We find that achieving differential targeting is possible with certain configurations of gene drive parameters, but, in most cases, only under relatively low migration rates between populations. Under high migration, differential targeting is possible only in a narrow region of the parameter space. Because fixation of the gene drive in the non-target population could severely disrupt ecosystems, we outline possible ways to avoid this outcome. We apply our model to two potential applications of gene drives-field trials for malaria-vector gene drives and control of invasive species on islands. We discuss theoretical predictions of key requirements for differential targeting and their practical implications.