American Scientist

Malaria is a tropical illness that blood-feeding mosquitoes spread from person to person. In 2018, approximately 228 million malaria cases emerged, with an estimated 405,000 deaths recorded in the same period. More than half of those deaths were young children.

Mosquito nets have limited effectiveness in preventing infection, as do pesticides, and antimalarial drugs also have limited effectiveness in preventing and treating infection (see figure below). Compounding this problem is the lack of resources and infrastructure in regions where the risk for malaria is greatest.

Biologists combatting malaria face some choices. They can intervene in human beings, the malaria parasite, or the mosquito. And, instead of traditional methods of fighting malaria, they can use new genetic tools to better reduce the prevalence of the disease among humans, without the toxicity of pesticides and drugs or the challenge of convincing humans to modify their behavior. Now, rather than targeting humans or the malaria parasite, biologists are assessing the use of new tools to edit the genes of malaria-transmitting mosquitoes. The main tool is a gene drive: When scientists insert a gene drive into an organism, the organism passes that new gene to all its offspring, and the offspring do the same.

The gene-drive method of altering populations of mosquitoes is a result of the discovery in the past decade of clustered regularly interspaced short palindromic repeats (CRISPR)—an immunity system in bacteria that has led to a revolution in precision genetic engineering. A CRISPR gene drive inserted on one chromosome locates a gene sequence on the chromosome next to it. Like a pair of scissors, it “cuts” that sequence and inserts a copy of the pair of “scissors” in the DNA wound. The organism then has two copies of the CRISPR driving gene, one on each of its two chromosomes, so that gene is present in all the ova or sperm that the organism produces. When that organism mates with an individual in which the CRISPR gene is absent, all offspring initially have one copy of the CRISPR gene and one of the wild-type gene that the species had before the CRISPR system was introduced. Then CRISPR goes to work in those offspring, changing them from having just one copy of CRISPR to having two.

In 2015, Valentino Gantz, a graduate student in the laboratory of one of the authors, Ethan Bier, developed the first CRISPR gene drive for sexually reproducing animals that have different tissue types and organs. Then, in collaboration with the laboratory of Anthony James at the University of California, Irvine, Bier and Gantz developed the first CRISPR gene-drive system for mosquitoes.

CRISPR biologists copy nature, just as plant and animal breeders mimic natural selection when they decide which organisms in a population will be parents of the next generation.

• Facebook
• Twitter

The novelty of CRISPR goes beyond enabling biologists to insert into an organism a new gene that will be passed from one generation to the next. These new genes are designed to circumvent the “laws” of Mendelian genetics and thereby spread in the population, generation after generation, with increased rapidity. These CRISPR-powered self-copying gene drives must be understood in the context of evolutionary and epidemiological processes before the ethics of designing and using them to fight malaria can be adequately considered. One of the authors, Elliott Sober, a philosopher of science at University of Wisconsin–Madison, has published on the evolution side of the story, in collaboration with David Sloan Wilson, a biologist at Binghamton University.

Evolutionary Principles

Although CRISPR gene drives may sound like an exotic technological intervention that has no counterpart in natural processes, they work by means of a process that has affected evolution for as long as organisms have had genes. That process is meiotic drive; it was discovered by biologists almost 80 years ago. CRISPR gene drives are engineered meiotic drives. CRISPR biologists copy nature, just as plant and animal breeders mimic natural selection when they decide which organisms in a population will be parents of the next generation.

Understanding CRISPR gene drives requires understanding meiosis in nature. Sexual organisms are diploid; they inherit pairs of chromosomes, one from each parent. Each chromosome contains genes arrayed in a line, like beads on a string. The chromosomes in a pair lie side-by-side, so a gene on one chromosome has another gene next to it on the other chromosome. When an organism has two copies of the same gene variant, its genotype at that chromosomal locus is said to be homozygous. When the organism has different variants of the gene at a locus, the genotype is heterozygous.

Sexual reproduction proceeds by organisms’ forming gametes (ova and sperm). Gametes have singleton chromosomes instead of the pairs found in other cells in the body; gametes are haploid. Meiosis is the process by which diploid parents form haploid gametes that come together in reproduction to form a diploid offspring.

Gregor Mendel in the 1860s, and the early geneticists whom he inspired around 1900, thought that meiosis was “fair,” meaning that heterozygotes have 50 percent of their gametes with one gene and 50 percent with the other (see figure on below). Geneticists called this even distribution “the law of independent segregation.” This law stood unchallenged until 1942, when Barbara McClintock discovered a gene in maize that breaks the law. Such segregator distorter genes garner more than 50 percent representation in the set of gametes that a heterozygote produces. These “driving genes” have since been discovered in multiple plant and animal species.

If driving gene D and wild-type gene W are in the population, then one can discover that D is a driving gene by looking at the gametes that heterozygotes produce. But this situation presents a puzzle. Why do copies of W remain in the population even though the W gene competes so poorly with the D gene? The presence of W suggests that a second, countervailing cause of trait frequency must be at work.

Discerning this second cause was part of the 1950s discovery of a house mouse driving gene. DW heterozygotes produce 85 percent D gametes and 15 percent W gametes. Male DD homozygotes are sterile, however, whereas WW and DW mice are fertile. Consequently, in heterozygotes, D outcompetes W, but WW homozygotes outcompete DD homozygotes (see sidebar below). Selection within heterozygote individuals therefore coexists with selection between homozygote individuals. Gene D wins the first competition but loses the second, with the result that neither D nor W achieves 100 percent representation in the population. The takeaway lesson is that understanding the evolution of a driving gene involves two considerations: how the gene works in DW heterozygotes, and how DD and WW homozygotes compare with each other.

Gene-Editing Targets

In the war against malaria, biologists can consider three possible targets: humans, mosquitoes, and the malaria parasite. Intervention in humans is limited to developing effective vaccines and antimalarial drugs, but vaccine development has proven challenging. To understand why, consider two possible vaccine types. The first type is the familiar protective vaccine that inoculates the vaccinated person against infection. The second, more promising type of vaccine blocks transmission of malarial parasites to others. This type of vaccine is called altruistic because it protects others, not the person who is vaccinated. Protective vaccines have been more difficult to develop because they act on parasite stages active in humans, whereas transmission-blocking agents counteract parasite stages present in the mosquito.

Why is it easier to take out the parasite in mosquitoes than in humans? Human beings, like all vertebrates, have two immune systems: the sophisticated and highly specific adaptive immune system, and a more basic innate system. Mosquitoes have only the simpler innate immune system. The malaria parasite has evolved the ability to live successfully in both hosts, but the evolutionary challenge of coping with the two immune systems was more severe than the challenge of coping with just the one. Consequently, the parasite now has numerous flexible devices for defending itself against the human immune system, but fewer and less flexible devices for defending against the simpler, less challenging mosquito immune system.

Another consideration strengthens this conclusion. Producing a vaccine for human beings that protects them against parasites that have cell nuclei (eukaryotes) has always been problematic. Eukaryote parasites, malaria included, are much more complex than viruses and bacteria. They have evolved many adaptations to undermine a host’s immunity. Developing a vaccine that renders human beings immune from malaria has therefore proven challenging.

Alternatively, why not genetically modify the malaria parasite itself? In principle, this strategy sounds sensible, but in practice scientists have been unable to determine how genetic information can be inserted into the parasite. One possibility would be to insert a virus that would either kill the parasite, or make it less virulent. Unfortunately, no such virus has been described in the literature. And even if such a virus could be found, what fraction of parasites would it infect, and would the parasite respond by simply latching onto another molecule produced by the host mosquito?

The bottom line, therefore, is that modifying mosquitoes is the most plausible strategy for protecting humans.

Option 1: Mosquito Suppression

Having settled on mosquitoes as the most feasible organism to genetically modify, scientists face the question of what the gene drive should do. One option is to drive some mosquito species to extinction. In the late 1960s, Christopher Curtis suggested using chromosomal variants to drive a beneficial trait into a population. This idea, although serving as a guiding concept in the field, was technically impracticable. Then, as “selfish” genetic elements came to light, Austin Burt of Imperial College London did the mathematical modeling that pioneered the idea of using them to suppress a population by linking them to a trait that imposes a great fitness cost (often sterility). Sterility drives could be built to exterminate a population of mosquitoes.

Instead of eliminating mosquitoes, scientists have proposed modifying them so that they would be less able to transmit malaria to human beings.

• Facebook
• Twitter

Across the ocean in Southern California, in research spanning the past three decades, Anthony James from the University of California, Irvine (UCI), has considered a different approach. Instead of eliminating mosquitoes, he proposed modifying them so that they would be less able to transmit malaria to human beings.

Different labs around the world have invested in one or the other of these two strategies. The Bill and Melinda Gates Foundation is supporting research at Imperial College London, in Andrea Crisanti’s lab, which is pursuing the extinction strategy; at the University of California, San Diego, in the lab of Ethan Bier, coauthor of this article, and in collaboration with Anthony James at UCI, the Tata Foundation and the UCI Malaria Initiative are supporting the modification strategy.

The extinction strategy is well along in its development, although it has not been tried in the wild. In its laboratory, the Imperial College group conducted a small-scale exploratory experiment involving two cages of mosquitoes. Each began with 450 WW mosquitoes and 150 DW individuals (so, counting two genes for each diploid organism, the initial frequency of D was 12.5 percent). No DD homozygotes existed initially, but when the first generation had offspring, DD individuals appeared, and they were sterile. Drive gene D increased in frequency in every generation, and the first cage went extinct after 8 generations, whereas the second went to zero after 12. A generation spans about 14 days. The gene drive at work here did better than the segregation ratio of 85 percent to 15 percent that biologists observe in the house mouse (see sidebar above, "Driving Genes To Three Different Destinations"). The Imperial College machinery produces the most extreme segregation ratio possible: 100 percent D to 0 percent W. With this extreme degree of segregation distortion, what evolves is not a balance point in which D and W are both maintained in the population. Instead, D goes all the way to 100 percent, at which point every mosquito is sterile, and the population goes extinct.

Inspired by the results of these small-cage experiments, the Crisanti group recently initiated a new set of tests. In a high-security facility built in Terni, Italy, the group is using large cages that computers control to mimic natural environmental conditions, such as light, temperature, and humidity. The next step, pending approval, will be an open field test. Gene-drive research is moving fast.

So far we have described what happens in a single population that contains driving gene D and wild-type W. The analysis becomes more complicated when we consider a country or an entire continent where many local populations of mosquitoes exist and the following conditions apply: The populations have W at 100 percent, biologists introduce D into a few of those populations, and migrants move from one population to another. In this multigroup setting, if extinction occurs in a local population, those denuded locales become repopulated by wild-type mosquitoes. The result is that D increases in frequency in groups that contain both D and W, but groups in which W is common avoid extinction and send out migrants more successfully than groups in which D is common. Using estimated rates of between-group migration, modeling by Burt and colleagues suggests that mosquitoes do not entirely disappear from locales in which D has been introduced, but their numbers are substantially reduced (by about 90 percent).

Because the extinction approach is biologically feasible, ethical questions must be considered. If the gene drive is inserted into mosquitoes in the wild, will the ecosystem crash or suffer a major disruption? For example, more than 400 mosquito species exist in India, but only 4 to 6 of those species are significant in the spread of malaria to humans. Although extinguishing a species can sometimes cause a major ecosystem disruption, extinguishing these 4 to 6 species is unlikely to do so. True, some bats eat blood-feeding mosquitoes, but they eat numerous other organisms as well, including mosquitoes that do not feed on humans. This property of the food chain suggests that the probability of the extinction strategy’s crashing the ecosystem is very small, although unforeseen consequences are always possible.

Another problem with the mosquito suppression approach is less speculative: If the mosquitoes in one valley are driven to extinction, nothing prevents mosquitoes in a neighboring valley from flying in and repopulating. Extinction by gene editing is therefore subject to the same problem as old-fashioned insecticide: Repeated applications of gene drives will be required, just as is true for the insecticide DDT. How many months or years would lapse between the extinction of mosquitoes in one locale and a migration that repopulates it? The answer depends greatly on the specific context, so it is hard to provide a blanket estimate. Although even a brief respite from malaria may provide real human benefits, this problem attending the extinction approach explains the appeal of an alternative approach.

Option 2: Mosquito Modification

Instead of exterminating mosquitoes, a second option is to make them unable to spread malaria—which could benefit mosquitoes and humans alike. Mosquitoes stop eating for two weeks after they are infected with malaria. Once an infected mosquito starts to feed, the parasite manipulates the mosquito’s behavior, leading it to eat many small blood meals rather than a few large ones. A drive gene that causes the mosquito to kill the malaria inside its body, or to expel the malaria without injecting it into human beings, may have a double evolutionary advantage: It would be favored in the competition between genes that occurs inside of DW heterozygotes, and DD organisms would be fitter than WW organisms (see sidebar above, "Driving Genes To Three Different Destinations").

This gene drive avoids problems that the suppression strategy faces. The mosquito suppression strategy creates vacant locales, which mosquitoes from other locales can colonize, whereas the modification strategy creates no such unoccupied domains. The modified population continues to exist and carries the drive gene at 100 percent. The wild-type gene that it drove to zero in one population may be present elsewhere, but that outsider cannot enter the modified population and displace the drive gene. The WW migrant will be at a selective disadvantage when it enters a population of organisms that are DD.

When biologists talk about “eradicating malaria,” nonscientists often assume that their goal is to rid the whole planet of the disease in one fell swoop. Realistically, the nations of the world probably lack both the resources and the will to carry out this coordinated global undertaking. If any such project is to succeed, it must proceed piecemeal, with malaria eliminated in one locale, then another, and so on. When malaria is finally eliminated everywhere, eradication will have been achieved. If the modification strategy is used, this stepwise program is plausible; if the suppression strategy is used alone, it is not.

An ethics question remains: If the malaria parasite is removed from a mosquito population and the population continues to exist, will that modified population create a vacancy that a new parasite will fill, perhaps making matters worse for human beings? The probable answer is no. Only about 1 to 5 percent of female mosquitoes, at any given time, carry malaria. This statistic indicates that an already ample opportunity has existed for other parasites to move in, but so far none have.

The mosquito suppression strategy creates vacant locales, which mosquitoes from other locales can colonize, whereas the modification strategy creates no such unoccupied domains.

• Facebook
• Twitter

Another ethical concern pertains to both the suppression and modification strategies. Perhaps a driving gene will jump from one species to another that is not closely related, such as humans, propelled by a virus that facilitates horizontal gene transfer. This pathway for gene flow exists alongside the vertical transmission from parent to offspring. The worry here is that a gene inserted into mosquitoes would find its way into the human genome, with disastrous consequences for us. This scenario has virtually no chance of happening in the foreseeable future, however, because CRISPR drives rely on two highly specific constraints: the sequence of the targeting component of the “scissors,” which tells it where to cut, and a nearly perfect alignment of DNA sequences between donor and recipient sites into which the genetic element is to be copied. Scientists can choose both these factors to be unique to the genetic element inserted at its intended site, with the result that the insertion occurs only in the species of choice. A jump to even a closely related species that lacks these DNA sequences would be virtually impossible.

A more plausible scenario would be that, if the drive gene were not designed carefully, an organism carrying that gene in the target species might mate with an individual in a closely related species and produce viable, fertile hybrids. In plants, the drive gene might indeed jump relatively far, because plants apparently hybridize more readily than animals. For insects, however, the gene drive would be limited to spreading within the same genus. In any case, designing CRISPR gene drives that confine themselves to the targeted species should be possible. Laboratory tests will show whether spread to related species is likely, in which case the effect would be simply that the additional species are more resistant to spreading malaria.

If the modification strategy eliminates malaria in a locale, the game may not be over. True, gene-drive D displaces wild-type W, and the population persists with all the individuals now having two copies of gene D. True, W individuals from elsewhere cannot invade this population. The sticking point is that new mutations may arise that disable D. Natural selection is a process in which fitter traits replace traits that are less fit. D’s being fitter than W does not prevent a new trait from arising that outcompetes D. Evolution does not stop. This is a problem in eradicating any infectious disease, however, not only malaria. It is not a cause for despair, but it does mean that biological solutions may have limited shelf lives.

Because of the shelf-life problem, the modification strategy may face an invadability problem akin to the one that the suppression strategy faces. The problem with the suppression strategy is that a vacant locale may be repopulated by migrants with the old gene W. The problem with the modification strategy, by contrast, is that the modified population may be invaded by a gene that arises by mutation and exhibits some advantage over D. That said, successful invasion would likely occur more quickly with the suppression strategy than with the modification strategy.

Prospects and Risks Versus Benefits

The Crisanti group’s experiments provide evidence that the suppression strategy is feasible. Does evidence exist that the modification strategy is feasible too? The answer is yes. Remember the transmission-blocking antibodies that were developed to provide an altruistic vaccine? As published in 2012, Anthony James’s lab at UCI developed a set of altruistic genes that encode antibodies that are made in female mosquitoes only after they draw blood. (Only female mosquitoes bite humans, and these females need a blood meal to reproduce.) The lab’s experiments showed that these antibodies are 100 percent effective in blocking the transmission of the most lethal form of malaria.

After the Bier lab devised the first CRISPR gene drive in insects, the two labs collaborated, combining the gene drive from the Bier lab with the antimalarial genes from the James group. The result was a mosquito that delivers transmission-blocking antibodies to almost 100 percent of its progeny. Experiments provided proof-in-principle for the feasibility of this gene drive, which spreads to the mosquitoes’ antibodies, doing the mosquitoes no harm while also preventing them from spreading the malaria parasite from one human host to another. As in the experiments conducted by the Crisanti group, experiments with laboratory cages of mosquitoes were constructed in which all individuals were initially either WW homozygotes or DW heterozygotes. The results suggest that it would take only one to two seasons (10 to 20 mosquito generations) from the time of introducing the new gene into a wild population to its reaching 100 percent.

Unintended consequences are possible in any undertaking, but to allow that possibility to govern action is to choose inactivity, which extracts a price of its own. The price is enormous—some 2 million people will suffer from malaria each year, and some 500,000 will die. Those of us who live in countries that are free of malaria may feel no impetus to accept a risk if that risk is the price for saving lives elsewhere. But those of us who live in societies immersed in the malaria plague may be more inclined to accept some risk. Would malaria-free countries reject a gene drive that could eradicate Alzheimer’s disease or cancer?