Gene Drives & Evolutionary Interventions
When humans engineer evolution — the theory behind synthetic gene drives
A Brief History
What Are Gene Drives?
A gene drive is a genetic system that biases inheritance in its own favour, allowing it to spread through a population even if it carries a fitness cost to the individual carrying it. Unlike typical Mendelian genes, which segregate fairly at meiosis (50% chance to each gamete), a gene drive exploits molecular machinery to copy itself onto the homologous chromosome, achieving transmission rates far above 50%. This makes it a "selfish" genetic element — in evolutionary terms, an example of intragenomic conflict.
Natural gene drives occur widely in nature. In mice, the t-haplotype is a cluster of genes that can achieve up to 99% transmission from heterozygous males because it selectively eliminates sperm carrying the wild-type allele. Similarly, segregation distorters (or segregation distortion loci) in Drosophila and other organisms spread by meiotic drive — producing skewed sex ratios or gamete ratios that favour the drive allele. These systems have been studied for decades as models of intragenomic conflict.
Synthetic gene drives use molecular tools — primarily CRISPR-Cas9 — to engineer a similar inheritance bias. The concept was pioneered by Austin Burt (2003) and has been developed extensively in the past two decades. A typical synthetic drive comprises two components:
- A nuclease (endonuclease) gene, typically Cas9, that cuts a specific DNA target site on the wild-type chromosome.
- A homology template (often the drive gene itself or a linked cargo gene) that is used to repair the cut via homology-directed repair (HDR), copying the drive allele onto the cut chromosome.
In a heterozygous individual (drive/wild-type), Cas9 cuts the wild-type allele during gametogenesis. HDR then uses the homologous drive chromosome as a template, converting the heterozygote's offspring to drive/drive homozygotes. If this conversion is highly efficient, the drive spreads rapidly through the population. If it carries a cargo gene (e.g. for malaria resistance or insecticide resistance), that trait co-spreads with the drive mechanics.
A gene drive is a self-replicating genetic element that:
- Encodes a nuclease (Cas9) that cuts a target site on the wild-type allele.
- Includes a homology template for HDR-mediated repair, copying the drive onto the wild-type chromosome.
- Achieves super-Mendelian inheritance (transmission > 50%) in heterozygotes.
Conversion efficiency $e$ is the fraction of offspring that inherit the drive allele when both parents carry one. For a perfect drive, $e = 1$ (100% conversion); for an imperfect drive, $e < 1$. Even with $e = 0.95$, a drive can still spread if it has no fitness cost.
Reference: Burt, A. (2003). Site-specific selfish genes as tools for the control and genetic engineering of natural populations. Proc. R. Soc. Lond. B, 270, 921–928.
Types of Synthetic Gene Drives
Not all synthetic gene drives are designed the same way. They differ fundamentally in their ecological and evolutionary goals. The two major categories are:
Modification Drives
A modification drive spreads a beneficial trait through a population. The drive itself is self-replicating, and it carries a cargo gene — an additional sequence that conveys the desired phenotype. Classic examples include:
- Malaria-resistant mosquitoes: Engineering Anopheles gambiae with a drive carrying genes that either block Plasmodium development or prevent transmission, thereby reducing malaria incidence in endemic regions.
- Insecticide resistance in agricultural pests: Spreading resistance alleles through wild populations to reduce pesticide use.
- Conditional lethality: Drives that make the population dependent on a particular genetic background or environmental condition, preventing establishment in the wild.
In a modification drive, the population persists; only the genetic composition changes. The fitness effects depend on the cargo. A resistance allele typically carries little cost, but the drive machinery itself may impose a burden (e.g. expression of Cas9, potential off-target effects).
Suppression Drives
A suppression drive is designed to reduce or eliminate a target population. Instead of carrying a beneficial cargo, it targets a gene essential for viability or reproduction. Examples include:
- Female sterility drives: Inactivating genes required for female reproduction, so homozygous females are sterile. The drive spreads (because heterozygous females can reproduce), but each generation, fertile females decrease in number.
- Homing nuclease drives: Targeting the drive site itself, potentially causing mutations that disrupt reproduction.
Suppression drives require a strong coupling between drive inheritance and population suppression. If the drive is not perfectly coupled to the suppression phenotype, wild-type or partially resistant individuals can persist, potentially allowing the population to recover.
Modification drives: Self-replicating systems that carry a cargo gene conferring a desired phenotype (e.g. pathogen resistance). The population persists, but the genetic composition changes rapidly.
Suppression drives: Self-replicating systems coupled to reduced fitness (e.g. female sterility). The drive spreads, but the population declines due to reduced reproduction.
A unified framework for comparing these types was established by Gokhale, Reeves & Reed (2014), which modelled both as parameter variations of a general population transformation system. This framework was further refined by Champer et al. (2016) and standardized by Gokhale et al. (2021), who defined a common gene drive language to ease regulatory assessment and ecological extensions.
References: Gokhale, C. S., Reeves, R. G. & Reed, F. A. (2014). Dynamics of a combined medea-underdominant population transformation system. BMC Evol. Biol., 14, 98. Champer, J. et al. (2016). Cheating evolution: engineering gene drives to manipulate the fate of wild populations. Nat. Rev. Genet., 17, 146–159. Gokhale, C. S. et al. (2021). A common gene drive language eases regulatory process and eco-evolutionary extensions. BMC Ecology and Evolution, 21, 156.
Population Dynamics of Gene Drives
To understand how a gene drive spreads, we need population genetics. Consider a diploid population of size $N$ (measured in individuals). Let $p_t$ denote the frequency of the drive allele in generation $t$. Under standard Mendelian inheritance, with random mating, allele frequencies do not change (Hardy–Weinberg equilibrium). But a gene drive biases inheritance.
Suppose the drive has conversion efficiency $e$, meaning a fraction $e$ of heterozygotes' offspring receive the drive allele (via HDR-mediated copying). The remaining fraction $(1-e)$ segregate normally. Under a simplified haploid-like approximation (treating conversion as equivalent to selection on gamete frequencies), the change in allele frequency per generation is:
With conversion efficiency $e$ and fitness cost $s$ to drive carriers, the allele frequency in the next generation is:
$$p_{t+1} = \frac{p_t + e \cdot p_t(1-p_t)}{2 - s \cdot [p_t^2 + 2p_t(1-p_t)]}$$
In the limit of no fitness cost ($s = 0$) and treating conversion as a form of biased inheritance, this simplifies to:
$$\Delta p = p_{t+1} - p_t \approx \frac{e \cdot p_t(1-p_t)}{2}$$
Key observations:
- If $e > 0$, the drive allele increases whenever $0 < p_t < 1$ (i.e. it is not an ESS at either fixation boundary).
- The rate of increase is fastest at intermediate frequencies ($p_t = 0.5$), where $\Delta p_{\max} = e/8$.
- If $e = 1$ and $s = 0$ (perfect, costless drive), $p_t \to 1$ exponentially fast — typically in $\sim \log N$ generations.
- A fitness cost $s$ slows spread and may create a stable equilibrium at $p^* < 1$ if $s$ is large enough.
Threshold Drives vs Self-Sustaining Drives
Gene drives can be classified by their stability properties:
Self-sustaining (underdominant) drives: These spread from any initial frequency, given sufficient conversion efficiency. If $e$ is high enough, even a single drive allele in a wild population can eventually fix, making accidental release a serious concern.
Threshold (suppression) drives: These require a minimum frequency $p_{\text{crit}}$ to spread. If the drive frequency drops below this threshold, the population genetics favour reversion to wild-type. Threshold drives offer some biosafety advantage: released individuals must reach a critical number to establish a drive. However, calculating the critical threshold requires careful modelling of local mating dynamics and gene flow.
The threshold frequency depends on the conversion efficiency $e$ and the fitness cost $s$. For a suppression drive causing female sterility with frequency-dependent mating, the threshold can be quite high (e.g. 30–50% or more), depending on the demographic parameters.
The Effect of Mating Complexity
The simple recursion equation above assumes random mating in an infinite population. Real populations have spatial structure, mate choice, and specific mating systems. Verma, Reeves, Simon, Otto & Gokhale (2023) investigated how mating complexity alters gene drive dynamics, identifying three key dimensions:
1. Mate Choice
If individuals preferentially mate with certain partners (assortative mating, sexual selection, or mate rejection based on phenotype), the effective frequency experienced by a heterozygote can differ from the population frequency. For example, if wild-type females reject drive-carrying males, drive-carrying males encounter fewer opportunities to reproduce. This reduces the effective conversion rate and slows spread.
2. Mating System
Population mating systems vary widely:
- Monogamy: Each individual mates with at most one partner per generation. Reduces the effective population size and increases genetic drift, sometimes slowing drive spread.
- Polygamy (polygyny/polyandry): One individual mates with multiple partners. Increases the number of offspring produced per generation and can accelerate drive spread because a single drive-carrying individual contributes more converted offspring.
3. Spatial Mating Networks
If mating is not uniform (e.g. individuals only mate with spatial neighbours in a lattice), the drive must first spread locally before invading distant patches. This reduces the effective selection pressure and can slow or prevent global fixation, even for a self-sustaining drive.
Verma et al. (2023) compared distortion-based drives (those relying on meiotic drive or segregation distortion) with viability-based drives (those coupling inheritance to survival differences).
Key finding: Distortion-based drives are more robust against mate choice constraints, because the conversion mechanism acts during gametogenesis before mate selection occurs. Viability-based drives (e.g. those relying on conditional lethality) are more sensitive to mate choice, since individuals that would normally die may be eliminated before reproducing, reducing the realized drive spread rate.
Biological implication: If the goal is to engineer a drive robust against wild-type mating preferences, distortion-based mechanisms (modelled on natural systems like the t-haplotype) are preferable.
Reference: Verma, P., Reeves, R. G., Simon, S., Otto, M. & Gokhale, C. S. (2023). The effect of mating complexity on gene drive dynamics. American Naturalist, 201, E1–E22.
Explore: Gene Drive Spread
Adjust the parameters below to explore how gene drive allele frequency changes over time. The left panel shows the deterministic trajectory (infinite population), while the right panel displays stochastic simulations (finite population with genetic drift). Notice how the drive spreads from a single allele and may reach fixation or equilibrate, depending on the conversion efficiency and fitness cost.
Resistance Evolution
For a gene drive to persist in a population, it must overcome target site resistance. CRISPR-Cas9 cuts a specific DNA sequence (the target site). If mutations accumulate at or near the target site, they can prevent Cas9 from binding or cutting, rendering the site refractory to the drive. Such resistance alleles are not converted and segregate normally (Mendelian inheritance).
Unckless et al. (2017) modelled resistance evolution against CRISPR drives and found that mutations conferring resistance can rapidly fix, especially in large populations or under strong genetic draft (when the drive is spreading rapidly). Even with conversion efficiencies of 99%, if resistance mutations arise at a frequency of 10−4 per gamete, they can reach fixation within tens of generations, effectively halting the drive's spread.
This creates a fundamental tension in gene drive design:
- High conversion efficiency (e.g. $e > 0.99$) speeds the drive's spread but also accelerates selection against the drive due to resistance mutations.
- Multiple target sites (engineering a drive that cuts two or more sites simultaneously) can slow resistance evolution, since resistance requires mutations at all sites. However, this increases the molecular complexity and potential off-target effects.
- Mutable target sites (placing target sites in essential genes where resistance mutations are lethal) can reduce resistance, but add unintended fitness costs.
The evolutionary arms race between drive engineering and resistance evolution is ongoing. Future generations of synthetic drives will likely incorporate redundancy, dynamic targeting (updating target sites based on emerging resistance), or other strategies to extend the lifespan of a drive in a wild population.
Reference: Unckless, R. L., Messer, P. W., Connallon, T. & Clark, A. G. (2017). Evolution of resistance against CRISPR/Cas9 gene drive. Genetics, 205, 827–841.
Eco-Evolutionary Considerations
The spread of a gene drive has consequences that extend beyond population genetics. If the drive is designed to suppress a population (e.g. reduce mosquito abundance to control malaria), the ecological effects can be profound and cascading.
Density-Dependent Dynamics
Many populations exhibit density-dependent dynamics: reproduction and survival rates depend on population density. For example, in insects, larval mortality increases at high densities due to resource competition. A suppression drive that reduces the adult population may release density-dependent constraints, allowing surviving individuals to reproduce more successfully. This can slow the drive's impact or prevent population elimination entirely. Conversely, if the population is already near carrying capacity, the drive-induced reduction can be multiplicative.
Ecological Networks
Eliminating or drastically reducing a species has effects on its predators, parasites, and competitors. For Anopheles gambiae (the primary malaria vector), ecological consequences include:
- Release of competing mosquito species: If A. gambiae is eliminated, other mosquito species (potential vectors of other diseases) might expand.
- Predator effects: Aquatic and terrestrial predators of mosquitoes may shift to alternative prey or decline in abundance.
- Ecosystem services: Insects, including mosquitoes, are pollinators and contribute to nutrient cycling. Large-scale suppression could have indirect effects.
Evolutionary Rescue
Even if a suppression drive successfully reduces a population to very low numbers, rapid evolution could allow recovery. Populations under strong selection often experience rapid evolution of resistance and behavioral changes. If the drive-suppressed population persists at low density but has high per-capita growth rates and large mutational input, it may evolve counter-adaptations (e.g. behavioral avoidance of mating with drive-carrying individuals).
Gokhale et al. (2021) developed an extended framework that couples the genetic model of gene drives to ecological parameters. They showed that responsible deployment of suppression drives requires:
- Ecological simulations predicting population-level effects.
- Risk assessment of off-target species and ecosystem services.
- Adaptive management: monitoring the wild population post-release and adjusting strategies if unintended effects emerge.
- Reversibility: designing drives or follow-up interventions to re-establish a population if needed.
Reference: Gokhale, C. S. et al. (2021). A common gene drive language eases regulatory process and eco-evolutionary extensions. BMC Ecology and Evolution, 21, 156.
Exercises
Conceptual Questions
- Explain how a standard CRISPR/Cas9 gene drive achieves super-Mendelian inheritance. Why does cleavage of the wild-type allele followed by homology-directed repair create a 100% drive inheritance rather than the typical 50%?
- What is the threshold frequency for gene drive invasion into a population, and why does this threshold depend on the drive's fitness cost? How do heterozygous vs homozygous fitness costs differ?
- Describe three mechanisms by which wild-type populations can evolve resistance to gene drives. Which is fastest and most likely to limit long-term drive persistence?
- How does underdominance (a drive construct that is unstable when heterozygous) provide a safer containment strategy than standard gene drives? What is the cost in terms of achievable frequency?
- Why might mating complexity (e.g., assortative mating, population structure) slow or prevent gene drive spread even if the drive is advantageous within panmictic populations?
Computer Problems
- Simple Gene Drive Invasion. Implement a model with three genotypes: wild-type $WW$, heterozygote $WD$, and drive homozygote $DD$. Use Mendelian inheritance but allow $WD$ and $DD$ to convert $WW$ alleles with probability $c$ (cutting efficiency). With $c = 0.99$, simulate invasion from frequency $p_D = 0.01$ over 500 generations. Plot drive frequency and show when it reaches fixation.
- Threshold Frequency and Fitness Cost. For a drive with conversion efficiency $c = 0.95$ and heterozygous fitness cost $s = 0.01, 0.05, 0.1, 0.2$, compute the threshold frequency $p_{\text{thresh}} = s/(c-1+s)$ and verify numerically that drives below this frequency decline to loss while those above invade to fixation.
- Resistance Evolution to Gene Drives. Extend the model to include resistance mutations at rate $\mu = 10^{-5}$ that prevent CRISPR cleavage. Simulate a drive invasion with mutation, showing how resistance alleles increase in frequency over time. Compute the time to limiting resistance under various cutting efficiencies $c$.
- Underdominant Drive Containment. Implement an underdominant system where $WD$ heterozygotes have fitness $w_{WD} = 1 - s_{\text{het}}$ (lower than $WW$ and $DD$), creating a bistable equilibrium. Identify both stable fixed points and the unstable separatrix. Show that drives below the separatrix are eliminated while those above invade.
- Gene Drive in Structured Populations. Simulate a spatial model (grid or network) where populations are connected by migration. Release a gene drive into a single patch and measure invasion speed across patches. Show how migration rate and local fitness cost determine whether drives spread or remain localized.
References
- Burt, A. (2003). Site-specific selfish genes as tools for the control and genetic engineering of natural populations. Proc. R. Soc. Lond. B, 270, 921–928.
- Gokhale, C. S., Reeves, R. G. & Reed, F. A. (2014). Dynamics of a combined medea-underdominant population transformation system. BMC Evol. Biol., 14, 98.
- Champer, J., Buchman, A. & Akbari, O. S. (2016). Cheating evolution: engineering gene drives to manipulate the fate of wild populations. Nat. Rev. Genet., 17, 146–159.
- Unckless, R. L., Messer, P. W., Connallon, T. & Clark, A. G. (2017). Evolution of resistance against CRISPR/Cas9 gene drive. Genetics, 205, 827–841.
- Gokhale, C. S., Reeves, R. G., Simon, S., Otto, M. & Veller, C. (2021). A common gene drive language eases regulatory process and eco-evolutionary extensions. BMC Ecology and Evolution, 21, 156.
- Verma, P., Reeves, R. G., Simon, S., Otto, M. & Gokhale, C. S. (2023). The effect of mating complexity on gene drive dynamics. American Naturalist, 201, E1–E22.
- Deredec, A., Burt, A. & Godfray, H. C. J. (2008). The population genetics of using homing endonuclease genes in vector and pest management. Genetics, 179, 2013–2026.
- Deredec, A., Burt, A. & Godfray, H. C. J. (2008). The population genetics of using homing endonuclease genes in pest control: a step towards restoring evolution. J. Appl. Ecol., 45, 524–531.
- Esvelt, K. M., Smidler, A. L., Catteruccia, F. & Church, G. M. (2014). Concerning RNA-guided gene drives for the alteration of wild populations. eLife, 3, e03401.