Gene Drives‎ > ‎


An analysis of gene drive risks and safeguards

“Natural Selection... is a power incessantly ready for action,
and is as immeasurably superior to man's feeble efforts,
as the works of Nature are to those of Art.”

                                                                 -Charles Darwin

      Altering the traits of living organisms normally carries few ecological risks. Because we reshape living things for our own benefit, nearly all such changes reduce the odds that they will survive and reproduce in their ancestral environment. If they escape into the wild, the deleterious traits we have imposed will consequently be eliminated by natural selection. From an evolutionary and ecological perspective, both traditional selective breeding and modern genome engineering are unlikely to be consequential because they have little chance of spreading or even persisting in the wild. While a handful of specific alterations may confer advantages, they are exceptions to the general rule.

      In contrast, gene drive systems can spread through populations even if they reduce the fitness of each carrier organism. They accomplish this by distorting inheritance such that they are found in more than 50% of the surviving offspring of sexually reproducing organisms.

      There are many types of natural gene drives, but the most straightforward way to create a synthetic gene drive is to use the RNA-guided CRISPR/Cas9 nuclease to edit the genome while also encoding both Cas9 and the necessary guide RNAs next to the alteration. Editing will consequently re-occur in subsequent generations, converting heterozygotes into homozygotes which are guaranteed to pass the RNA-guided gene drive to all of their offspring.

      It's important to recognize that the vast majority of genome editing experiments do not risk creating a gene drive because they do not deliver DNA encoding both Cas9 and guide RNAs into germline cells. Yet it's equally important to realize that almost every molecular biology laboratory has access to Cas9 and that basic gene drives can be designed by gifted undergraduates or even high school students. The limitation is not in designing the gene drive system, but in delivering the DNA into the germline. This technology is already accessible to many researchers with expertise in model organisms such as fruit flies, but is likely to become even more widely available as new methods of transgenesis are developed.

      RNA-guided gene drives could revolutionize public health, sustainable agriculture, and ecological conservation. We could potentially use them to eradicate vector-borne diseases like malaria, dengue, and Lyme as well as schistosomiasis, control agricultural pests and weeds without using broadly toxic pesticides and herbicides, and curtail or remove environmentally damaging invasive populations of rats, cane toads, mosquitoes, and other destructive organisms.

      However, laboratory experiments involving potential gene drive constructs consequently pose a unique risk of spreading outside the laboratory. Traditional laboratory safety precautions assume that only those working in laboratory can be impacted by experiments. With gene drives, this is no longer the case. As scientists working in this area, we have a moral obligation to go above and beyond when it comes to ensuring that our experiments cannot directly affect the outside world. We elected to publish a description of RNA-guided gene drive technology and its implications in order to raise awareness of this issue and detail potential safeguards and confinement strategies before we initiated experiments that verified our findings.

      The need for widespread awareness was highlighted by a publication in March of 2015 that described a method of using RNA-guided gene drives as genome editing tools for routine laboratory use. The authors independently invented the concept and performed their experiments without prior awareness of the gene drive field or of our work on safeguards. Upon contacting them and discussing the issues, we jointly convened a diverse group of researchers from related fields to publish consensus recommendations concerning appropriate safeguards for laboratory research on gene drives.

      It's worth noting that while our team could not claim to speak for the scientific community as a whole, we included representatives from every lab that has published on endonuclease gene drives or on CRISPR technology development in Drosophila fruit flies, the directors of all major fruit fly genetic stock centers, and several leaders in fruit fly genetics and meiotic drive. As a highly mobile species famous for escaping the laboratory and now endemic worldwide, the fruit fly poses by far the greatest risk of accidental release, although other species are also of concern. Our recommendations are intended to inform the experimental designs of scientists who are considering entering the field and to assist institutional biosafety committees and other authorities evaluating proposed experiments.

      If there was a major flaw in our effort – apart from the inherent difficulty of 27 scientists reaching a firm consensus – it was that our efforts were constrained by wordcount. A more detailed analysis is not only in order, but it may also inform the efforts of the panel convened by the National Academy of Sciences to evaluate gene drive technology and produce formal guidelines.

Confinement strategies

      Gene drive research has the potential to directly impact the lives of people outside the laboratory if a drive construct unintentionally spreads into the wild. Fortunately, multiple confinement strategies are available and should be used in combination for greater safety. Any exceptions should require exceptional scrutiny and explicit authorization from a higher-level (not local) authority.

Extrinsic confinement aims to prevent organisms carrying gene drives from mating with wild counterparts.

Barrier confinement can prevent gene drive organisms from escaping the laboratory. Barriers are the traditional first line of defense and can be very effective when accounting for the idiosyncrasies of each organism and attempting to account for the inevitability of human error. For example, flying organisms should be kept in facilities with as many doors as possible between the laboratory and the outside world. At least three of them should be equipped with “air curtains” to blow organisms back inside when the doors are opened. Additional precautions might make the transition areas inhospitable to the organism; few insects will traverse a -20C freezer room. All holes for wiring, lights, drains, and other potential exit points must be sealed with mesh or other fillers. The organisms should be kept in multiple nested containers that are only ever opened when their occupants have been anesthetized. Ideally, experiments should be performed by only a single investigator to reduce the risk of miscommunication and error. While precautions for less mobile organisms need not be so extreme, all barrier protocols should aim to completely prevent organismal escape. Laboratories considering barrier confinement of gene drive organisms - especially flying insects - should refer to the expertise developed by ecologists employing biocontrol and researchers who work with insects carrying human pathogens.

Weaknesses: Barriers are known to be vulnerable to human error, as has been extensively documented for work with dangerous pathogens. Eventually, a researcher will forget to close a door or fail to notice an organism on their clothing. The more experiments performed and researchers involved, the greater the chance of a mistake. With a gene drive, it may take only one mistake. The reliability of barriers also depends on the escape routes available to the organism. While flying insects are generally a greater hazard than fish, one can readily imagine live fish or even their embryos escaping if washed down a drain. Barrier confinement is also vulnerable to natural disasters such as earthquakes that could break the walls and containers. Finally, barriers are especially susceptible to deliberate human action: should people successfully break into the laboratory and release all the organisms, barrier confinement will certainly fail.

Ecological confinement is achieved by performing experiments in geographic areas lacking populations of the organism in question. Ideally, the organism will not be able to survive outdoors: tropical Anopheles malarial mosquitoes would not survive winters in Boston or London.

Weaknesses: The strength of ecological confinement is defined by the difficulty of travelling to a native population. With respect to Anopheles mosquitoes in Boston or London, any mosquitoes that escaped in the summer could theoretically survive for long enough to catch a ship or more likely a plane to Africa and subsequently mate with native populations there. While difficult to quantify, the likelihood of such an event is clearly thousands or millions of times lower than if the research were performed in a laboratory where potential mates were just outside the window. Researchers travelling to the native habitat could also inadvertently transport organisms in their clothing or luggage; refraining from visiting the laboratory for several days prior to such a trip may be advisable. Finally, ecological confinement is vulnerable to deliberate transport of the organisms by humans.

Intrinsic confinement strategies seek to prevent the drive from spreading even if organisms carrying gene drives succeed in mating with wild counterparts. They must be inherent to the organism or the design or the gene drive construct itself.

Reproductive confinement involves the use of laboratory organisms that cannot reproduce with wild counterparts and consequently cannot pass on the gene drive element. For example, some laboratory strains of Drosophila melanogaster have a “compound autosome” formed by conjoining both copies of a large autosome centromere. These strains are fertile when crossed amongst themselves, but are sterile when outcrossed to any normal or wild-type strain because all progeny are monosomic or trisomic and die early in development.

Weaknesses: Reproductive confinement is not vulnerable to human error or deliberate misuse. It will fail whenever hybrid offspring survive, which will presumably occur at a fixed rate. This rate can be easily quantified for any given wild-type strain, though it could differ considerably depending on the origin of the reproductive incompatibility and the available genetic diversity of wild-type strains. Because reproductively incompatible strains are available for only a handful of species, reproductive confinement is not an option for most organisms absent future technological developments.

Molecular confinement ensures that the gene drive element is constructed such that it cannot sustainably bias inheritance in a wild population. There are two orthogonal approaches to molecular confinement that can be used separately or together.

      First, separating the Cas9 and guide RNAs creates a “split drive.” An sgRNA-only element will spread readily in transgenic organisms that already express Cas9 from an unlinked locus, but cannot bias inheritance without it. Should organisms accidentally escape the laboratory, the population frequency of the Cas9 gene will be determined by normal Mendelian dynamics and consequently limit the spread of the sgRNA cassette. In yeast, sgRNA-only systems are copied as efficiently as Cas9+sgRNA systems.

   Second, gene drive elements can target synthetic sequences not found in wild populations. They will only be capable of cutting and copying in transgenic laboratory populations encoding the synthetic sequence. In yeast, “synthetic site targeting” prevented the drive from copying when mated with wild-type yeast lacking the synthetic site.

Weaknesses: Molecular confinement will fail at a certain rate independent of human error or misuse. Of the two methods, the failure rate is almost certainly greater for the split drive approach because a single recombination event that moves Cas9-encoding DNA adjacent to or within the sgRNA-only element would create an autonomous Cas9+sgRNA gene drive. Nonetheless, this should be a very rare event. The frequency will likely depend on the degree of homology between the elements and their chromosomal locations. In contrast, synthetic site targeting would require multiple mutations within the guide RNAs to allow the drive to cut a wild-type sequence in the same chromosomal location. The requirement for a specific combination of individually rare mutations makes such an event extraordinarily unlikely. As long as the synthetic sites are many mutations away from the closest wild-type sequences, the risk reduction should be at least a billion- or trillion-fold, if not considerably more. While empirical tests will be needed to quantify the exact level, this is the one confinement strategy that may suffice if used by itself. If carefully designed, it could be more reliable on its own than any combination of the other methods. This level of reliability is important because there are already laboratories making virus-resistant transgenic organisms that express Cas9 and guide RNAs targeting the viral genome; such DNA constructs are no different from a synthetic site targeting gene drive.

When to use safeguards

      It is essential to emphasize that the existence of these safeguards strongly suggests that any experiments that do not employ them whenever possible are unethical. That is to say, even if experiments studying gene drives in the laboratory could be performed exclusively using conventional barrier confinement with what is deemed an acceptable level of risk, it would be unethical to do so if molecular confinement were also an option. 

   This suggests a bright line when it comes to safeguards: use multiple confinement methods whenever possible, including at least one intrinsic method that is not vulnerable to human error. What does this mean for different types of gene drive experiments?

Basic research studies of gene drive in large and fast-reproducing laboratory populations can always employ intrinsic confinement methods. I cannot think of a valid scientific reason why native sequences must be targeted, nor intact Cas9+sgRNA cassettes be used. Using both forms of molecular confinement would certainly reduce the risk of altering wild populations to negligible levels. Ideally, the species chosen for these studies would also permit ecological confinement to be used instead of or in addition to one of these methods. With two forms of molecular and one form of ecological confinement, the organisms could be safely distributed to schoolchildren for education and citizen science projects.

Applied research involves creating gene drive systems that are candidates for eventual release. While clever genome engineering may eventually provide a way, at present it is not possible to build and test a gene drive construct using intrinsic confinement and then allow it to escape. Hence, intrinsic confinement can't be used when developing candidates for eventual release; extrinsic confinement strategies are the only options available. This presents a problem, as many species that we may wish to target are endemic to much of the world – or at least in those regions where laboratories work with them.

      For example, the invasive southern house mosquito Culex quinquefasciatus will need to be curtailed or removed from Hawaii within the next 10-50 years or it will drive most species of native honeycreeper birds extinct. But southern and northern house mosquitoes (the two can hybridize and permit gene drives to transfer) are now endemic just about everywhere.

      Perhaps even more salient are black and brown rats, the most environmentally destructive (and economically damaging) invasive species in the world. Rats do not require deliberate human action to travel thousands of miles; there is continual gene flow between rat populations around the world thanks to stowaways on cargo ships.

      Thus, if we decide to move forwards in these specific cases, there will have to be exceptions to the general rule of multiple confinement. Even so, any project proposing to use barriers alone should always require approval from a higher authority than the local institutional biosafety committee.

      There is one more very important safeguard available to gene drive systems that are candidates for eventual release: the immunizing reversal drive. This is a gene drive system designed to spread through wild populations and those altered with the candidate drive. Wild populations will undergo no phenotypic change, but will be rendered immune to the candidate. Those altered by the candidate will be overwritten to undo the phenotypic change. Reversal is not perfect, as residual Cas9 and guide RNAs will still be present. But since the primary effect of any gene drive will be due to the changes it spreads - not the drive itself - immunizing reversal drives represent a powerful safeguard. As a consequence, laboratories should build an appropriate immunizing reversal drive when they build the candidate gene drive itself.

      In summary, all gene drive experiments should use multiple confinement strategies, including at least one intrinsic method, whenever possible. All basic research studies can and should use at least two strategies, one of which should always be molecular or reproductive. Applied research should take place in locations permitting ecological confinement to be used if at all possible; those cases in which it is not should require strict oversight and formal authorization. All applied research projects should build an appropriate immunizing reversal drive capable of reversing the relevant alteration that could be used in the event of an accidental release.