Gene Drives‎ > ‎


Gene Drive FAQ

CRISPR-based gene drive

What is a gene drive system?

    It's a stretch of DNA that is inherited more frequently than normal. In sexually reproducing organisms, most DNA sequences have a 50% chance of being inherited by each offspring. This is called “Mendelian inheritance”. Gene drive systems manage to rig the game so that they are inherited more frequently - up to 100% of the time.

Do they occur in nature?
    Yes, many different kinds of gene drive occur frequently in nature. Almost every species has either an active drive or the broken remnants of one in its genome.

What kinds of gene drive are there?
    Local drive systems can spread alterations through a local population of organisms, then stop and can't go any further. Examples include daisy drive and underdominance. Global drive systems are self-sustaining. They consequently will spread to most of the local population, and quite possibly to every population of that species in the world.

What is the advantage of using a gene drive?
    When we engineer an organism, our alterations almost always harm its ability to reproduce in the wild. With very few exceptions, evolution is simply better at doing this than we are. But if we embed the same change within a gene drive, the inheritance advantage conferred by the drive can counterbalance the harm from our alteration, “driving” our new trait through a wild population over many generations. Given enough generations, nearly all organisms of the target species will have the same alteration as the first ones released.

Why would that be useful?
    Right now, we have very few ways of addressing ecological problems because we can't alter the traits of wild populations. We can't stop mosquitoes from carrying diseases like malaria, dengue, yellow fever, or West Nile. We can't eradicate the trematode worms that causes diseases like schistosomiasis. If a weed has evolved resistance to an herbicide or a crop pest to an insecticide, we can't do anything about it except switch to using new pesticides and herbicides. If an invasive species damages an ecosystem and causes native species to go extinct, we just don't have many good options for controlling it.

How could gene drives solve these problems?
   We might be able to alter the mosquitoes so they can't carry disease or suppress their populations until the disease is permanently eradicated. Gene drives could directly reverse the many of the mutations conferring pesticide and herbicide resistance, supporting sustainable no-till agriculture and keeping us from having to switch to potentially more toxic pesticides and herbicides. Eventually, we might even be able to alter pests so they don't like the taste of crops or humans. Local populations of invasive species might be specifically suppressed or even eliminated by spreading traits that reduce the reproductive capability of each individual organism.

So why haven't we done this before?
    Austin Burt first proposed the idea of building gene drives based on cutting DNA more than ten years ago, but we didn't have the molecular tools to cut and drive alterations in useful genes and species - especially not in an evolutionarily stable manner.

What changed?
    In 2013, laboratories all over the world started using an enzyme from bacteria, called Cas9 or CRISPR, to alter the genomes of different species. Because Cas9 can recognize almost any DNA sequence using complementary RNA sequences, it's very easy to use it to edit any gene. We realized that we could use Cas9 to make an “RNA-guided gene drive” that recognizes competing genes on the other chromosome and copies itself - and whichever genes it is driving - in their place.

RNA-guided gene drives could edit almost any gene in any population?
    Not every gene, but most of the ones we'd want to.  It only works in sexually reproducing populations and may be more difficult in some species than in others.

That sounds pretty powerful. Should we be worried?
    It's very reasonable to be concerned about new and potentially powerful technologies. This one could be used to alter ecosystems, which we're still learning about, so we'll have to be very careful not to cause damage accidentally. At the same time, there are a lot of existing problems hurting people and ecosystems that we might be able to fix.

Could gene drives be used to alter human populations?
    Not without taking many centuries or millennia. Gene drives take generations to spread, and our generations are quite long. Driving a trait would only increase the number of people carrying it by fourfold after a hundred years - and that assumes subsequent generations wouldn't decide to remove it with their doubtless superior technology.

What about crops?
    It would be quite difficult and highly impractical. Most crops and domesticated animals are the products of careful cross-breeding programs. Annual crops are grown from seeds obtained from central suppliers, while animals are often bred using artificial insemination guided by genetic records and screening. You can't really drive a trait when we already control reproduction so carefully. Non-annual crops such as fruit and nut trees have long enough generation times that drives would take too long to spread. Agriculture would be much better served by controlling the weeds and insect pests rather than altering the crops.

Why did you decide to make this public before actually doing it?
    If we're right about this, it's a powerful advance that could make the world a much better place, but only if we use it wisely. We think that's most likely to happen if we make sure that people have a chance to learn what we're thinking of well before we could actually put it in practice. That will let us collectively explore possible ways of developing and using this technological responsibly. Before going public, we consulted experts in a lot of different areas to be sure they agreed with our assessment.

Whom did you check with and what did they think?
    We consulted with a wide variety of people, from ecologists to molecular biologists to government regulators to security experts to representatives from environmental organizations. They had varying concerns about it, but everyone agreed it was better to tell the world so we can collectively start deciding whether, when, and how to use gene drives wisely. We chose to do so before performing any experiments in the laboratory to set an example for the field: public notification and discussion should precede and inform experiments.

How can we prepare to use gene drives wisely?
    It's tremendously important to realize that the gene drive system is just the vehicle. The effects will depend on the passengers - the particular genes or alterations that are driven through the population. So it doesn't really make sense to ask whether we should use gene drives. Rather, we'll have to evaluate whether it's a good idea to consider driving this particular change through this particular population. Gene drives could prevent a great deal of human suffering and fix a lot of the damage we've already done to the environment, but each time we'll need to carefully weigh the benefits of each proposed gene drive against the possible risks. This is an enterprise best done collectively, in the open light of day.

What else can we do?
    We can be sure to develop safeguards as we develop the technology. For example, we outlined possible ways of controlling and reversing the effects of alterations driven through populations.

The effects of gene drives can be reversed?
    Yes, genomic changes can be reversed by releasing a “reversal drive”, which we have demonstrated in yeast. It's also possible to protect a population from being affected by a specific gene drive using an “immunizing reversal drive”.  It's important to note that reversal drives aren't perfect because they require gene drive components of their own in order to spread fast enough, but these components shouldn't affect the traits of the organism if we build them correctly.

Could gene drive systems be misused or weaponized?
    Gene drive systems make poor bioweapons because they are readily detected by sequencing, are slow to spread, and can be easily countered by building and releasing an immunizing reversal drive.

Detection - Sequencing will always detect human-made CRISPR gene drive systems as unnatural. They cannot be disguised as a natural CRISPR bacterial system because bacterial signals don't work in sexual organisms. The cost of sequencing is dropping faster than any other technology, which will make monitoring even cheaper with time.

Speed - the abundance of gene drive systems will increase by a factor of only 1.2-1.9 per generation in the wild. Altering entire populations takes many generations.

Reversal - Building a working immunizing reversal drive system is trivially done by altering the unwanted gene drive system. Populations carrying this construct can expand far more rapidly in the laboratory and factory than the gene drive will spread in the wild. For example, mosquitoes multiply by a factor of 200 per generation in the laboratory, more than a hundred times faster than the drive system in the wild.

In summary, while gene drive systems could be misused in theory, attempting to do so in practice would be difficult, costly, and comparatively ineffective.

What other precautions could we take?
    When we're developing gene drives in the laboratory, we can ensure there's negligible risk that they will spread into wild populations by using appropriate confinement. There are two main types.

Extrinsic confinement is based on ensuring that the organisms never escape to mate with wild counterparts:

Barrier confinement physically prevents organisms carrying gene drives via containers, biosafety hoods, doors, and so forth.

Ecological confinement prevents them from surviving or finding mates if they do escape by performing experiments in geographic areas that don't harbor wild populations.  If we're building gene drives in organisms native to Australia, we should do the work in the United States, and vice versa.

    The problem with extrinsic confinement is that it's vulnerable to human error. We know from the long history of pathogen research that even the barrier protocols used by the highest-rated labs periodically suffer accidents that let pathogens out. With gene drives, a single escaped organism mating with wild counterparts could suffice. Even ecological confinement is vulnerable to someone deliberately transporting organisms carrying gene drive systems to their native habitat.

Intrinsic confinement isn't similarly vulnerable to human error and deliberate misuse because it builds the gene drives in organisms such that they couldn't spread the drive even if they could mate with wild counterparts.

Reproductive confinement involves building gene drive systems in organisms that can't produce viable offspring if they mate with wild counterparts. For example, some laboratory fruit fly strains have their chromosomes rearranged so they can reproduce with one another, but not with wild flies.

Molecular confinement involves building the gene drives such that they won't function in wild populations. There are two approaches:

Split drives encode only one of the two necessary components and rely on the organism to provide the other. Since wild organisms won't provide it, they can't spread in the wild. There are many strains of fruit flies and zebrafish that express Cas9 and consequently could be used to build split drives that only encode the guide RNAs instructing Cas9 where to cut.

Synthetic target drives will only cut and spread through synthetic DNA sequences not found in wild populations. If a drive-carrying organism were to mate with a wild counterpart, the drive couldn't cut and replace the wild-type chromosome because it wouldn't have the necessary sequence.

All gene drive experiments should use multiple confinement strategies. Since they have independent failure modes, the risk reductions should be multiplicative. 

Basic research experiments should use at least one form of intrinsic confinement plus one other strategy.

Applied research - that is, experiments attempting to create gene drives that are candidates for eventual release - can't use intrinsic confinement because the drive would consequently be unable to spread through the wild population once approved. They should consequently use both ecological and barrier confinement whenever possible. Any rare cases that can't use more than one form of confinement should require explicit regulatory approval from a national regulatory body - not just a local institutional biosafety committee.

Finally, laboratories building gene drives that make a particular change should simultaneously build reversal drives capable of “undoing” the change just in case something goes wrong.

Who will be able to make gene drives and how quickly?
   Making a drive in a new species will probably take expert laboratories at least months to years. Just learning to edit the genomes of cells that will produce eggs or sperm can be quite difficult in some organisms. And while being able to cut the target with Cas9 is critical, there are many other factors that have to be considered and optimized to make a good gene drive - and those will differ with the species.

Doesn't informed consent require researchers to inform everyone who might be affected?
   We believe that all gene drive research should be conducted transparently from the proposal stage onwards. That is, all grant proposals, reports and data from laboratory meetings, preprints, and formal publications should be made available to the public.

What are you doing to ensure that will happen?
   We are currently launching a project, Responsive Science, devoted to building a platform for this type of disclosure and encouraging the growth of a diverse community dedicated to guiding the development of this inherently collective technology. More information will be available soon.

We have verified the effectiveness of intrinsic safeguards and reversal drives in yeast. Our ongoing experiments in nematode worms are currently focused on "licensing" Cas9 expression in the germline in order to enable split-drive confinement. Plans for the first self-sustaining gene drive experiments in any new organism will be made available before experiments begin and all results will be posted within two weeks of the date they are obtained.