A Synthetic Maternal-Effect Selfish Genetic Element Drives Population Replacement in Drosophila

See allHide authors and affiliations

Science  27 Apr 2007:
Vol. 316, Issue 5824, pp. 597-600
DOI: 10.1126/science. 1138595


One proposed strategy for controlling the transmission of insect-borne pathogens uses a drive mechanism to ensure the rapid spread of transgenes conferring disease refractoriness throughout wild populations. Here, we report the creation of maternal-effect selfish genetic elements in Drosophila that drive population replacement and are resistant to recombination-mediated dissociation of drive and disease refractoriness functions. These selfish elements use microRNA-mediated silencing of a maternally expressed gene essential for embryogenesis, which is coupled with early zygotic expression of a rescuing transgene.

Mosquitoes with a diminished capacity to transmit malaria or dengue have been identified in the wild and/or created in the laboratory, demonstrating that endogenous or engineered mosquito immunity can be harnessed to attack these pathogens (15). However, it will be necessary to replace a large percentage of the wild mosquito population with refractory insects to achieve substantial levels of disease control (68). Mosquitoes carrying genes that confer disease refractoriness are not expected to have a higher fitness than native mosquitoes, implying that Mendelian transmission is unlikely to result in an increase in the frequency of transgene-bearing individuals after their initial release into the wild (4, 9). Thus, effective population replacement will require the coupling of genes conferring disease refractoriness with a genetic mechanism for driving these genes through the wild population at greater than Mendelian frequencies (10, 11).

Maternal-effect selfish genetic elements [first described in the flour beetle Tribolium castaneum and known by the acronym Medea (maternal-effect dominant embryonic arrest)] select for their own survival by inducing maternal-effect lethality of all offspring not inheriting the element-bearing chromosome from the maternal and/or paternal genome (12) (Fig. 1A). Current models predict that if Medea elements are introduced into a population above a threshold frequency, determined by any associated fitness cost, they will spread within the population (1214) (Fig. 1, C and D). When introduced into a population at relatively high frequencies, Medea elements are predicted to rapidly convert the entire population into element-bearing heterozygotes and homozygotes (Fig. 1C). Medea in Tribolium is hypothesized to consist of a maternal lethal activity (a toxin) that kills non-Medea–bearing progeny and a zygotic rescue activity (an antidote) that protects Medea-bearing progeny from this maternal lethal effect (12, 15) (Fig. 1A).

Fig. 1.

Characteristics of a maternal-effect selfish genetic element (Medea) and a synthetic Medea element in Drosophila. (A) It is postulated that females heterozygous for Medea (Medea/+) deposit a protoxin or toxin (red dots) into all oocytes. Embryos that do not inherit a Medea-bearing chromosome die because toxin activation or activity is unimpeded (bottom left square). Embryos that inherit Medea from the maternal genome (top left square), the paternal genome (bottom right square), or both (top right square) survive because zygotic expression of a Medea-associated antidote (green background) neutralizes toxin activity. (B) (Top) Schematic of a simple molecular model that accounts for Medea behavior postulates the existence of two tightly linked loci. One locus consists of a maternal germline–specific promoter that drives the expression of RNA or protein that is toxic to the embryo. The second locus consists of a zygotic promoter that drives the expression of an antidote. (Bottom) Schematic of Medeamyd88 is shown. ORF, open reading frame; Mir 6.1-Myd88-1+2, transcript encoding two copies of Drosophila miRNA 6.1 modified to target the myd88 5′ untranslated region. (C) Frequency of genotypes lacking Medea for an element carrying the additive fitness costs indicated, over generations, with Medea introduced at a 1:1:2 ratio of homozygous Medea-bearing males, non-Medea–bearing males, and females lacking Medea. Generation zero refers to the wild population (non-Medea/non-Medea = 1) before population seeding. Generation one refers to the progeny of crosses between these individuals and homozygous Medea-bearing males. (D) Frequency of the non-Medea–bearing chromosome for the populations described in (C). (E and F) Medeamyd88-1 drives population replacement in Drosophila. Medeamyd88-1 was introduced into seven population cages at an allele frequency of ∼25% (16). (E) The frequency of genotypes lacking Medea (+/+) over generations is indicated for two separate sets of population cage experiments, involving three (green lines; 20 generations) or four (blue lines; 15 generations) population cages each. The predicted frequency of genotypes lacking Medea, fora Medea element with zero fitness cost (introduced at an allele frequency of 25%) is indicated by the red line. (F) The frequency of the non-Medeamyd88-1–bearing chromosome (+/Medea and +/+) over generations from the population cage experiments in (E) is indicated as above, as is the predicted frequency for an element with zero fitness cost.

To create a Medea-like maternal-effect selfish genetic element in Drosophila, we generated a P transposable element vector in which the maternal germline–specific bicoid (bic) promoter drives the expression of a polycistronic transcript encoding twomicroRNAs (miRNAs) designed to silence expression of myd88 (the gene producing the toxin) [Fig. 1B and (16)]. Maternal Myd88 is required for dorsal-ventral pattern formation in early embryo development. Females with germline loss-of-function mutations for myd88 give rise to embryos that lack ventral structures and thus fail to hatch, even when a wild-type (WT) paternal allele is present (17). This vector (known as Medeamyd88) also carries a maternal miRNA– insensitive myd88 transgene expressed under the control of the early embryo–specific bottleneck (bnk) promoter (the gene producing the zygotic antidote) (Fig. 1B). Our analysis focused on flies carrying a single autosomal insertion of this element, Medeamyd88-1.

Matings between heterozygous Medeamyd88-1/+ males (where + indicates a chromosome that does not carry Medeamyd88-1) and homozygous +/+ females resulted in high levels of embryo viability, similar to those for the w1118 strain used for transformation (Table 1). In addition, 50% of the adult progeny carried Medeamyd88-1, as expected for Mendelian segregation without dominance. Matings among homozygous Medeamyd88-1 flies also resulted in high levels of egg viability. In contrast, when heterozygous Medeamyd88-1/+ females were mated with homozygous +/+ males, ∼50% of progeny embryos had ventral patterning defects (fig. S1) and did not hatch (Table 1). All adult progeny (n > 12,000 flies) carried Medeamyd88-1 (Table 1). On the basis of these data and the results of several other crosses (Table 1), we inferred that a single copy of bic-driven miRNAs targeting maternal myd88 expression was sufficient to induce maternal-effect lethality and a single copy of zygotic bnk-driven myd88 expression was sufficient for rescue.

Table 1.

Medeamyd88-1 shows maternal-effect selfish behavior. Progeny of crosses between parents of several different genotypes (M refers to the Medeamyd88-1–bearing chromosome; + refers to the non–element-bearing homolog) are shown. The maternal copy number (0 to 2) of bic-driven miRNAs targeting the endogenous myd88 transcript (maternal toxin) and zygote copy number (0 to 2) and percentage of embryos inheriting bnk-driven myd88 (zygotic antidote) are indicated, as are the adult progeny genotypes predicted for Mendelian inheritance of Medeamyd88-1 and the percent embryo survival. -, not measured. The asterisk denotes that embryo survival was normalized with respect to percent survival (± SD) observed in the w1118 stock used for transgenesis (97.1 ± 0.7%).

Inherited by theEmbryo survival (%)*
Parental genotypeOocyteEmbryoAdult M progeny (%)
MaleFemaleMaternal toxinZygotic antidote (n, %)PredictedObserved
M/+ +/+ 0 0, 50 50 50 99.6 ± 1
1, 50 (n > 7000)
M/M M/M 2 2, 100 100 - 98.1 ± 0.4
+/+ M/+ 1 0, 50 50 100 48.3 ± 2
1, 50 (n > 12,000)
M/M M/+ 1 1, 50 100 - 98 ± 1
2, 50
M/+ M/+ 1 0, 25 75 - 74.3 ± 0.5
1, 50
2, 25
M/+ M/M 2 1, 50 100 - 98.3 ± 1
2, 50
+/+ M/M 2 1, 100 100 - 99.1 ± 0.4
M/M +/+ 0 1, 100 100 - 98.8 ± 0.5

The above observations, in conjunction with the lack of any obvious fitness effects (lethality) in individuals carrying one or two copies of Medeamyd88-1, suggested that Medeamyd88-1 should be able to drive population replacement. To test this prediction, we mated equal numbers of WT (+/+) and Medeamyd88-1/Medeamyd88-1 males with homozygous +/+ females, giving rise to a progeny population with Medeamyd88-1 present at an allele frequency of ∼25% (16). This level of introduction, although high, is not unreasonable, given previous insect population suppression programs (18). Replicate population cage experiments, carried out in a darkened incubator to prevent Medeamyd88-1–bearing flies (which are Pw+ and thus red-eyed) from obtaining any vision-dependent advantage over their +/+ counterparts (which are w1118 and white-eyed) (19), followed three replicates for 20 generations. A second set of four replicates, which were initiated by crossing heterozygous Medeamyd88-1/+ males with homozygous +/+ females, was followed for 15 generations. In both experiments, non-Medeamyd88-1–bearing flies permanently disappeared from the population between generations 10 and 12 (Fig. 1E), without a loss of non-Medea–bearing + chromosomes (in Medeamyd88-1/+ individuals) in the population (Fig. 1F). The observed changes in Medeamyd88-1 were not significantly different from the null hypothesis that the element had no fitness cost [(16) and fig. S2], although we cannot exclude the possibility that a Medeamyd88-1–associated cost is counterbalanced by an unknown negative effect associated with the w1118 mutation in +/+ individuals. Finally, we carried out three further replicate population cage experiments in which the Medeamyd88-1 transgene was introduced at a frequency of 25% into the Oregon-R strain, which is WT with respect to the endogenous w locus (and thus members of which are red-eyed). Evidence for population replacement by generation 12 was observed in this context as well (16), suggesting that Medeamyd88-1–associated Pw+ expression is unlikely to be a major contributor to the ability of Medeamyd88-1 to drive population replacement.

For any gene-drive mechanism to be successful in reducing parasite transmission, there must be tight linkage between the genes that mediate drive and effector functions (10). If the driver becomes separated from the effector gene through chromosome breakage and nonhomologous end joining (as in a reciprocal translocation) (Fig. 2A), and the effector gene carries a fitness cost, selection will favor and promote the spread of individuals carrying Medea elements that lack the effector (MedeaΔeff). Locating the effector gene between the toxin and antidote prevents a single chromosome breakage and end joining event from creating a MedeaΔeff-bearing chromosome (Fig. 2B). However, it does not prevent the creation of Medeains-bearing chromosomes that carry the antidote, and perhaps the effector, but not the toxin (Fig. 2B). Medeains-bearing chromosomes are insensitive to Medea-dependent killing. If the presence of the toxin and/or the effector results in a fitness cost, then Medeains-bearing chromosomes gain a fitness advantage with respect to those carrying the complete Medea element, thereby promoting spread of the former. This outcome can lead to the reappearance of pathogen-transmitting insects (14).

Fig. 2.

A strategy for enhancing the functional lifetime of Medea elements in the wild and for carrying out cycles of population replacement. (A to D) Locating Medea toxin and effector genes in an intron of the antidote prevents chromosome breakage and end joining–mediated separation of drive and effector genes and creation of Medeains-bearing chromosomes. [(A) to (C)] Medea constructs with different gene arrangements are shown. Sites of chromosome breakage and end joining with a second nonhomologous chromosome are indicated by the crossed lines. Recombinant products referred to in the text are indicated by thick lines, the color of which indicates the centromere (solid circles) involved. (A) Recombination at site 1 generates a Medeains-bearing chromosome that carries the effector. Recombination at site 2 generates a MedeaΔeff-bearing chromosome. (B) Recombination at site 1 or site 2 generates a Medeains-bearing chromosome. (C) Recombination at sites 1 to 3 generates benign chromosomes that cannot show Medeains or MedeaΔeff behavior. (D) Schematic of Medeamyd88-int. Splice donor and acceptor sites are indicated in large red letters, with the branchpoint and polypyrimidine stretch in small red letters. (E and F) A strategy for carrying out cycles of population replacement with Medea. (E) A first-generation Medea element (Medean), driven by Toxinn and Antidoten, is integrated into the chromosome [thick black line with centromere (solid circle) at the right] at a specific position (triangle). A second-generation Medea element (Medean+1), driven by Toxinn+1 and Antidoten+1, can be integrated at the same position using site-specific recombination (24). Locating both elements at the same position limits the possibility of homologous recombination creating chromosomes that carry both elements. (F) Because progeny carrying Medean are sensitive to Toxinn+1, the only progeny of females heterozygous for Medean+1 that survive are those that inherit Medean+1, regardless of their status with respect to Medean. In contrast, the progeny of Medean females that fail to inherit Medean survive if they inherit Antidoten as a part of Medean+1.

One way to prevent chromosome breakage and end joining–mediated formation of MedeaΔeff and Medeains elements is to put the toxin and effector genes into an intron of the antidote (Fig. 2C). To test this approach, we generated flies carrying Pw+Medeamyd88-int, in which the toxin, a transcript generating maternally expressed miRNAs targeting myd88, was placed in an intron of the zygotically expressed antidote, bnk-driven myd88, in the opposite orientation (Fig. 2D). We characterized the behavior of one autosomal insertion of this element Medeamyd88-int-1, which behaved as a maternal-effect selfish genetic element (Table 2). Females heterozygous for Medeamyd88-int-1 gave rise to a high frequency of Medeamyd88-int-1–carrying progeny (>99%), and the maternal-effect lethality associated with a single copy of Pw+Medeamyd88-int-1 was rescued by zygotic expression of the antidote from either the maternal or paternal genome (Table 2). However, when homozygous element-bearing females were crossed with non–element-bearing males, progeny embryo survival was very poor (∼20%), suggesting an inefficient zygotic rescue, perhaps resulting from inefficient splicing of the myd88 artificial intron. Population replacement for an element with these fitness characteristics is still expected to occur, though with some delay (fig. S3) as compared to that for an element in which the fitness costs are a simple function of copy number in either sex (Fig. 1, C and D).

Table 2.

Medeamyd88-int-1 shows maternal-effect selfish behavior. Progeny of crosses between parents of several different genotypes are shown, and notations are the same as those in Table 1.

Inherited by theEmbryo survival (%)*
Parental genotypeOocyteEmbryoAdult M progeny (%)
MaleFemaleMaternal toxinZygotic antidote (n, %)PredictedObserved
M/+ +/+ 0 0, 50 50 51 98.4 ± 0.6
1, 50 (n = 5000)
M/M M/M 2 2, 100 100 - 98.6 ± 0.8
+/+ M/+ 1 0, 50 50 99.5 48.7 ± 0.6
1, 50 (n = 5000)
M/M M/+ 1 1, 50 100 - 98.4 ± 0.7
2, 50
M/+ M/+ 1 0, 25 75 - 73.6 ± 1.2
1, 50
2, 25
M/+ M/M 2 1, 50 100 - 57.2 ± 1.5
2, 50
+/+ M/M 2 1, 100 100 - 20.2 ± 1.1
M/M +/+ 0 1, 100 100 - 98.5 ± 0.7

Medeains-bearing chromosomes can also arise if the toxin mutates to inactivity. Although toxin mutation cannot be prevented, the use of miRNAs as toxins can provide a degree of redundant protection because multiple miRNAs, each processed and functioning as an independent unit, can be linked in a polycistronic transcript [(Fig. 1B and (16)]. The use of miRNAs as toxins also provides a basis by which selfish genetic element drive can be limited to the target species. Medea elements only show drive when maternal-effect lethality creates an opportunity for zygotic rescue of progeny that inherit the element. Therefore, drive can be limited to a single species by the use of miRNAs that are species-specific in their ability to target the maternally expressed gene of interest.

Perhaps the most likely point of failure in any population-replacement strategy involves the effector. Effector genes can mutate to inactivity, creating MedeaΔeff-bearing chromosomes. In addition, parasites may undergo selection for resistance to these effectors. These events, as well as the possible appearance of Medeains-bearing chromosomes discussed above, will lead to the reappearance of permissive conditions for disease transmission. Therefore, it is important that strategies be available for removal of an element from the population, followed by its replacement with a new element. One potential strategy for achieving this goal, in which different Medea elements located at a common site in the genome compete with each other for germline transmission in transheterozygous females, is illustrated in Fig. 2, E and F.

Our data show de novo synthesis of a selfish genetic element able to drive itself into a population. This laboratory demonstration notwithstanding, several obstacles remain to the implementation of Medea-based population replacement in the wild. First, for pests such as mosquito species, there is little genetic or molecular information regarding genes and promoters used during oogenesis and early embryogenesis. This information is straightforward to generate, with the use of transcriptional profiling to identify appropriately expressed genes and transgenesis and RNA interference in adult females to identify those required for embryonic development, but it remains to be acquired. In addition, current models of the spread of Medea do not take into account important real-world variables, such as migration, nonrandom mating, and the fact that important disease vectors such as Anopheles gambiae consist of multiple partially reproductively isolated strains (20, 21). Although an understanding of the above issues is critical for the success of any population-replacement strategy, the problems are not intractable, as evidenced by past successes in controlling pests by means of sterile-male release (18) and as implied by our growing understanding of mosquito population genetics, immunity, and ecology (2023).

Supporting Online Material

Materials and Methods

Figs. S1 to S5


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article