A review of the safety & efficacy of genetically engineered mosquitoes

In early 2015, researchers at Imperial College London made headlines for engineering a special breed of female mosquitoes that produced predominantly male offspring [1]. The researchers used the CRISPR/​Cas9 gene editing system to encode a homing endonuclease—a special type of gene that will cut the female’s X chromosome—in order to shift offspring sex ratios, which is hypothesized to cause a population crash. [1]. In another study, researchers at University of California, Irvine used CRISPR/​Cas9 to create a strain of mosquitoes with malarial-resistance genes [2]. Although this is not the first time that genetically modified mosquitoes (“GMMs”) have been produced in the lab, it does present an interesting opportunity to revisit the question of whether we should advocate for the release of GMMs to suppress vector borne illnesses.

There have been a handful of science articles about this subject, arguing everything from advocating for the release of these mosquitoes immediately in order to combat the Zika virus outbreaks, to articles that advocate caution so that we may be able to more carefully evaluate the unknowns of GMMs. As a result, I wanted to dig deeper into the scientific literature, and really try to understand what sorts of conclusions researchers had currently come to, and where they believe their knowledge gap lies.

Therefore, in order to address the issue of releasing GMMs into the wild, I want to start by looking at three main questions: first, what are the risks and benefits associated with these modified genetic elements, and how effective are they at achieving their goals? Second, what are the ecological implications of releasing mutant mosquitos into wild populations? And finally, what effect can this truly have on vector-borne disease?

Using “Gene Drives” to Suppress Mosquito-Borne Illnesses

The CRISPR/​Cas9 system used to engineer these mosquitoes are part of a bigger concept known as a ‘gene drive’. Researchers have ben talking about gene drives from as early as 2003, where it was suggested that gene drives could alter the genetic makeup of populations by rerouting traditional methods of Mendelian genetic inheritance [3].

Mendelian genetic inheritance, named after botanist Gregor Mendel, is what we classically think of in genetics: if you cross-breed a red rose and a white rose, the offspring will be pink because it inherits one ‘red’ gene from one parent and inherits a ‘white’ gene from another parent. Of course, genetics is not quite so simple as that; there are many different factors involved, including gene dominance and epigenetics; even environmental factors can have an effect on whether a gene is expressed or not. But for the most part, the Mendelian inheritance model holds true. If you thought about engineering mosquitoes with specific genes and released them into the wild, each offspring would only inherit 50% of these genes, and with subsequent generations, the genes would be diluted in pretty quickly.

Gene drives are a way to shift that inheritance. A gene drive is comprised of a ‘selfish’ gene element that cuts its sister gene and substitutes a copy of itself. If we had a gene drive for ‘red’ gene, the ‘red’ gene would cut the ‘white’ gene, and produce another ‘red’ gene. This greatly increases the odds that future offspring will be red roses.

Figure 1: Normal Mendelian Inheritance vs Gene Drive Inheritance [4]

In order to engineer gene drives, researchers used the CRISPR/​Cas9 gene-editing system. CRISPR gene-editing technology has made headlines in the past year as being one of the greatest breakthroughs in genetic engineering. First published by Jennifer Doudna’s research team at UC Berkeley, the CRISPR/​Cas9 gene editing system allows researchers to target virtually any position on a genome, and perform a whole host of functions, including gene cutting, mutation, or activation. Although gene drives need not necessarily be made from CRISPR technology, there is strong evidence to suggest that, going forward, most gene-editing technologies will involve CRISPR, due to its efficiency, versatility, and ease of use [4]. As a result, the majority of this review will concern CRISPR-based technologies.

Nevertheless, since it is a relatively new technology, it might first be worth asking: how safe is it to release CRISPR/​Cas9-mediated gene drives into the wild? Will it mutate? Can it be controlled?

Safety and Efficacy of CRISPR/​Cas9 Gene Drives

According to a paper released by Harvard molecular geneticist and Future of Life Institute Board Member George Church, CRISPR-based gene drive systems exhibit great specificity, versatility, and programmability [4].

First, CRISPR/​Cas9 based gene drives are highly specific. The CRISPR system consists of a guide RNA that targets a specific gene. The guide RNA complexes the enzyme Cas9, and together, these two associate with the DNA sequence that is complementary to the guide RNA sequence. The size of the guide RNA is approximately 21 bases; since nucleotides can be one of four unique bases, the chance of having two identical genetic sequences is extremely low [4]. Furthermore, CRISPR’s off-target effects can be further reduced by engineering varieties of Cas9 that minimize non-specific interactions with its DNA target site, as demonstrated in a study by Kleinstiven et al. at Harvard Medical School [5]. Other strategies, such as using unique genes with few similar sequences, truncated guide RNAs, or Cas9-fusion proteins often has reduced off-target effects to “undetectable” levels in model organisms such as the fruit fly[4]. As a result, we can expect to use similar strategies, and see similar effects with the mosquito.

Second, CRISPR/​Cas9 systems are reversible. Suppose the gene drive that is sent out in a population has unexpected effects. To “reverse” this, we can simply engineer another gene drive that silences the effect of the first gene drive. However, it should be noted that this is only a solution on the genetic scale—even if the gene drive trait were reversed, the potential ecological effects of the first gene drive may not be reversed [4].

Third, CRISPR/​Cas9 gene drives are highly versatile. We can use them for anything from sensitization (for instance, encoding a gene in mosquitoes that would make them highly sensitive to certain chemicals or insecticides), to immunization (for instance, encoding a gene that would prevent a mosquito from being sensitized to a particular molecule) [4]. As a result, CRISPR editing can give us combinatorial control over how to engineer mosquito populations.

The main drawback with CRISPR/​Cas9 gene drives is that they do not exhibit evolutionary stability. Since many of the engineered traits from gene drives are deleterious to the organism, mutations could override gene drives and make it less susceptible to spreading. For instance, if we released mosquitoes that cut the X chromosome, and this killed 99% of the mosquito population, the remaining 1% of the population may have developed a mechanism of immunity to the drive. As a result, there may need to be several releases of mosquitoes with modified gene drives in order to out-compete and continually whittle down the remaining population [4].

Arguments for Mosquito-Knockout Gene Drives

If our first question is “are gene drives safe”, our next question becomes “are gene drives necessary”?

Initially, I considered the idea that the malaria-resistant gene drives, as engineered by researchers at UC Irvine, might be the more desirable strategy. It doesn’t kill the entire mosquito population (thereby reducing unwanted ecological impacts) and it also induces antibody production in mosquitoes to prevent malaria transmission. Overall, it seems like a win-win. However, there are a few limiting factors when we are engineering mosquitoes with specific resistances to specific disease.

First, a single species of mosquito can carry multiple diseases. The anopheles mosquito—the genus primarily responsible for malarial transmission—is also known for carrying pathogens such as filarial worms (the WHO estimates there are 1.2 billion people at risk for infection), arboviruses and the onyong-nyong virus [6], [7]. Other breeds of mosquitoes, such as Aedes aegypti, are also known for spreading diseases such as dengue fever, yellow fever, Japanese encephalitis, Rift Valley fever, Chikunguya virus and West Nile virus. The Zika Virus pandemic has been largely spread by the A. aegypti mosquitoes [8].

Second, genetic engineering is limited by genetic load. It would be impossible to create gene drives that encode antibodies to all of these diseases. If a cell has to make additional non-essential proteins (such as antibodies for diseases), this siphons away resources that could otherwise be used for cell growth and replication. As a result, these cells are less likely to survive, and would be selected against—other cells with less “load” would grow more successfully instead[9].

Third, multiple mosquitoes can carry the same disease. For instance, over 30 different species of mosquitoes are responsible for the spread of malaria [10]. Moreover, it is entirely possible, for instance, that if we engineered A. gambiae to be malaria-resistant, a closely related species might pick up the parasite instead [11]. This can very quickly lead to a large combinatorial problem, where we must ensure that all possible species of mosquitoes that are responsible for carrying a certain disease are producing antibodies against that disease.

It is for this reason that we must seriously consider using gene drives for total mosquito wipeout strategies. Drastically suppressing mosquito populations may be the most effective, and indeed the only effective, way to limit mosquito-borne illnesses.

The Ecological Impact of Mosquito Eradication

One of the main deterrents to the use of GMMs is the possible environmental impact. Mosquitoes are thought to provide two significant ecological functions: the first is to pollinate certain types of flowering plants, and the second is to act as a food source for a variety of birds, fish, and amphibians. If we release GMMs into certain areas, how will the resulting wildlife be affected?

In the first case, there are very few documented studies of mosquitoes acting as pollinators. Some reviews state that mosquitoes primarily function as nectarophages—that they visit flowers to consume nectar, and provide little function for pollination [12]. The most well-documented case of mosquito pollination occurs in the arctic, where the Aedes mosquitoes pollinate certain types of orchids in Alaska [13]. Another species of mosquitoes also pollinate the Silene(catchfly) plants in the Netherlands. In fact, the anatomy of a mosquito—they haeve suction-based mouthparts and a short proboscis—makes them less important as pollinators than other species of insects (such as the short-horned flies). However, some studies suggest that mosquitoes may make up for this by their large numbers [12].

In the issue of mosquitos as food sources, scientists are currently divided on how bird, fish and amphibian populations would be impacted if we removed mosquitoes as a food source. For instance, in the Arctic, two species of mosquitoes act as food sources for migratory birds that rest in the tundra. Some scientists estimate that there could be a reduction of the bird population by 50% without enough mosquitoes to eat, while others believe this effect would be less severe, due to the fact that mosquitoes do not show up in bird stomach samples in high numbers. Another study found that birds that regularly eat mosquitoes only produced 2 chicks per nest, after mosquitoes were removed, as compared to 3 chicks per nest in the control group [14].

David et al. in the Department of Ecology, Evolution, and Behaviour at the University of Minnesota performed ecological modeling to determine the impact of mosquito gene drives.’s study claims that the initial transitory phase that follows immediately after GMM release may look very different from the steady state phase. For instance, if we had to release large numbers of mosquitoes in order for gene drives to successfully propagate though the population, we would have a spike in mosquito population initially, but a crash in the steady-state phase. The difference between the steady state and transitory phases also correlates to the fish and insect populations: although there may be a short-term transitory impact of reduced bird, fish, and amphibian populations, these populations may increase again as mosquitoes get substituted for other insect-based food sources [11].

Overall, the ecological implications are probably the most poorly understood facet of GMMs. Although literature studies reveal little, it is unclear whether this is because there would be few significant ecological impacts from releasing GMMs, or whether this is because scientific literature is biased towards impacts of mosquitoes on human populations, and less on bird, fish and amphibian populations.

Effects on Disease from Mosquito Population Reduction

A study by Deredec et al. at Imperial College London claims that a gene-drive based mosquito population introduction at even 0.1% size of the total mosquito population could cause the Anopheles species to crash, even though this is within the range of typical population fluctuation during seasonal variation [15]. The studies by Hammond et al. at Imperial College London, and Gantz et al. at UC Irvine reported that their trials had led to gene transmission rates ranging from 94% to 99% of mosquito offspring. Furthermore, Oxitec, a company in the UK that has been performing field trials with mosquitoes using traditional genetic elements—i.e., not gene drives—reported seeing mosquito population reductions between 80% and 99% (more about Oxitec will be discussed below). Therefore, it seems fairly likely that releasing GMMs will, in fact, cause significant reductions in insect populations [16].

The main area of concern with this strategy is the impact that the transitory mosquito population would have on acquired immunity and herd immunity.

For instance, some parts of the human population have built up “acquired immunity” to vector-borne illnesses, such as malaria. The mechanism of this immunity for malaria is still largely unknown; nevertheless, a study by Doolan et al. states that it should be appreciated to be “100% effective” [17]. Similarly, a large number of such individuals with acquired immunity can also collectively enable what is known as “herd immunity”, which further reduces spread of the infection by reducing the frequency with which non-immune people are exposed to the pathogen (this is the same effect that vaccinations have; they don’t just protect the vaccinated individual from getting sick, they also reduce the amount of exposure non-vaccinated people have to the disease, thereby lowering disease transmission rates).

Although reducing the vector population rate should also reduce the disease transmission rate, the three factors (reducing vector population, reduction in acquired immunity, and reduction in herd immunity) must be balanced so that the overall disease transmission rate is lower than it was prior to releasing genetically modified mosquitoes. Loss in herd immunity was responsible for the resurgence of dengue fever in Singapore in 2013, even though the dengue virus had been suppressed for the past 15 years [11]. This would be especially problematic if the genetically modified insect interventions failed, and the insect population rebounded.

Oxitec’s GM Trials

Recently, Oxitec, a UK-based company that has been running field trials of genetically modified insects since 2010, came to public attention after its work in releasing GMMs in Brazil, in order to combat the Zika virus outbreak.

Oxitec’s work does not use gene drives; rather they use a system called release of insects carrying a dominant lethal gene (“RIDL”), which they use to engineer sterile males. In this case, the lethal gene is a tetracycline transactivator, which is lethal at high doses for larvae, unless the antibiotic tetracycline is present. Oxitec first genetically modifies male mosquitoes to carry the lethal lethal gene and includes tetracycline in their feed, which allows them to grow to adulthood[18]. Since male mosquitoes do not bite humans, they also do not spread disease. Furthermore, when these males mate with wild-type female mosquitoes, all offspring will also have a copy of the lethal gene and will die before reaching adulthood [16].

Oxitec’s performed its first field trial with A. aegypti mosquitoes (RIDL strain OX513A) in the Grand Cayman Islands, which reduced the target mosquito population by 80%. Since then they’ve also done field trials in Brazil, in three different cities which reduced in the mosquito populations by 92%, 94%, and 99%. After a field trial in Panama which concluded in 2014, Oxitec reported that a 90% reduction in A. aegypti mosquitoes, six months post release [16].

The main disadvantage of Oxitec’s work is that RIDL is only effective for the first generation of mosquito offspring; in order to continually whittle down the mosquito population, more and more RIDL male mosquitoes must be produced. This also leads to issues of rebounding if the mosquito population is not fully suppressed.

Final Thoughts

To summarize briefly, using CRISPR/​Cas9 mediated gene drives to engineer genetically modified mosquitoes display significant robustness at the cellular level; the CRISPR/​Cas9 system can be tailored with high specificity, reversibility, and versatility to reduce unwanted effects. Furthermore, using gene drives to suppress the mosquito population also seems to have much higher impact than using gene drives to merely engineer disease resistance to mosquito populations.

The main areas that require additional research would be ecological impacts, and the issue of disease transmission. Oxitec’s trials look promising, though it is still unclear what the long-term impacts of mosquito eradication will be, and how long the GMM release needs to be sustained to prevent disease rebounding. Oye et al., wrote an expansive piece in Science in August 2014 advocating for increased regulation of gene drives, and investigating possibilities such as reversal drives and other fail-safes—though, of course, it’s important to balance risk management studies with the many millions of lives that could potentially be saved if we were to release GMMs sooner [19].

I am cautiously optimistic about the strategy of using GMMs to eradicate mosquito-borne illnesses. If we can hammer out some of the uncertainty in this approach, and achieve more accurate models of the interaction between GMMs and disease progression, we can quickly approach the beginning of the end of mosquito-borne illnesses.

References

[1] A. Hammond, R. Galizi, K. Kyrou, A. Simoni, C. Siniscalchi, D. Katsanos, M. Gribble, D. Baker, E. Marois, S. Russell, A. Burt, N. Windbichler, A. Crisanti, and T. Nolan, “A CRISPR-Cas9 gene drive system targeting female reproduction in the malaria mosquito vector Anopheles gambiae.,” Nat. Biotechnol., vol. 34, no. 1, pp. 78–83, Dec. 2015.

[2] V. M. Gantz, N. Jasinskiene, O. Tatarenkova, A. Fazekas, V. M. Macias, E. Bier, and A. A. James, “Highly efficient Cas9-mediated gene drive for population modification of the malaria vector mosquito Anopheles stephensi,” Proc. Natl. Acad. Sci., vol. 112, no. 49, p. 201521077, Nov. 2015.

[3] A. Burt, “Site-specific selfish genes as tools for the control and genetic engineering of natural populations.,” Proc. Biol. Sci., vol. 270, no. 1518, pp. 921–928, 2003.

[4] K. M. Esvelt, A. L. Smidler, F. Catteruccia, and G. M. Church, “Concerning RNA-guided gene drives for the alteration of wild populations,” Elife, vol. 3, p. e03401, Jul. 2014.

[5] B. P. Kleinstiver, V. Pattanayak, M. S. Prew, S. Q. Tsai, N. T. Nguyen, Z. Zheng, and J. Keith Joung, “High-fidelity CRISPR–Cas9 nucleases with no detectable genome-wide off-target effects,” Nature, vol. 529, no. 7587, pp. 490–495, Jan. 2016.

[6] “WHO | Lymphatic filariasis.”

[7] M. G. and G. K. Christophides, Anopheles mosquitoes—New insights into malaria vectors. InTech, 2013.

[8] L. A. M. Carneiro and L. H. Travassos, “Autophagy and viral diseases transmitted by Aedes aegypti and Aedes albopictus.,” Microbes Infect., Jan. 2016.

[9] M. C. Whitlock and D. Bourguet, “Factors Affecting the Genetic Load in Drosophila: Synergistic Epistasis and Correlations Among Fitness Components,”Evolution (N. Y)., vol. 54, no. 5, pp. 1654–1660, Oct. 2000.

[10] G. F. Killeen, U. Fillinger, I. Kiche, L. C. Gouagna, and B. G. Knols, “Eradication of Anopheles gambiae from Brazil: lessons for malaria control in Africa?,”Lancet Infect. Dis., vol. 2, no. 10, pp. 618–627, Oct. 2002.

[11] A. S. David, J. M. Kaser, A. C. Morey, A. M. Roth, and D. A. Andow, “Release of genetically engineered insects: a framework to identify potential ecological effects,” Ecol. Evol., vol. 3, no. 11, pp. 4000–4015, 2013.

[12] B. Larson, P. Kevan, and D. Inouye, “Flies and flowers: taxonomic diversity of anthophiles and pollinators,” Can. Entomol., vol. 133, no. 4, pp. 439–465, Jul. 2011.

[13] L. B. Thien, “Mosquito Pollination of Habenaria obtusata (Orchidaceae),” Am. J. Bot., vol. 56, no. 2, p. 232, 1969.

[14] J. Fang, “Ecology: A world without mosquitoes.,” Nature, vol. 466, no. 7305, pp. 432–4, Jul. 2010.

[15] A. Deredec, H. C. Godfray, and A. Burt, “Requirements for effective malaria control with homing endonuclease genes,” Proc Natl Acad Sci U S A, vol. 108, no. 43, pp. E874–80, 2011.

[16] L. Versteeg, Q. Wang, and C. M. Beaumier, “Invited Commentary on Genetically Modified Mosquitoes for Population Control of Pathogen-Transmitting Wild-Type Mosquitoes,” Curr. Trop. Med. Reports, pp. 16–18, 2016.

[17] D. L. Doolan, C. Dobano, and J. K. Baird, “Acquired immunity to Malaria,” Clin. Microbiol. Rev., vol. 22, no. 1, pp. 13–36, 2009.

[18] P. Gabrieli, A. Smidler, and F. Catteruccia, “Engineering the control of mosquito-borne infectious diseases.,” Genome Biol., vol. 15, no. 11, p. 535, Jan. 2014.

[19] K. A. Oye, K. Esvelt, E. Appleton, F. Catteruccia, G. Church, T. Kuiken, S. B.-Y. Lightfoot, J. McNamara, A. Smidler, and J. P. Collins, “Biotechnology. Regulating gene drives.,” Science, vol. 345, no. 6197, pp. 626–8, Aug. 2014.