Will Yale or the researchers have to pay compensation for the next deaths?

Failed GM mosquito control experiment may have strengthened wild mozzies

A trial to control mosquito populations using genetic engineering has gone wrong. People not only in or around Jacobina / Brazil must be alerted.

By Michael Irving (*) - 12. September 2019

Mosquitoes are more than just a pest – they can be downright dangerous carriers of disease. One of the most innovative ideas to control populations of the insects has been to release genetically modified male mosquitoes that produce unviable offspring. But unfortunately a test of this in Brazil appears to have failed, with genes from the mutant mosquitoes now mixing with the native population.

The idea sounded solid. Male Aedes aegypti mosquitoes were genetically engineered to have a dominant lethal gene. When they mated with wild female mozzies, this gene would drastically cut down the number of offspring they produced, and the few that were born should be too weak to survive long.

Ultimately, this program should cut down the population of mosquitoes in an area – up to 85 percent, in some early tests. This of course means fewer bug-borne diseases, such as dengue, yellow fever, zika, and malaria, in humans. And since the offspring don’t live long enough to breed themselves, genes from the engineered bugs should stay neatly out of the gene pool of the wild population. The only visible effect should be the reduction of mosquito populations.

Unfortunately, that hasn’t been the case. Researchers from Yale University have now examined mosquitoes around the city of Jacobina, Brazil, where the largest test of this technique has taken place over the last few years. Not only did numbers bounce back up in the months after the test, but some of the native bugs, they found, had retained genes from the engineered mosquitoes.

“The claim was that genes from the release strain would not get into the general population because offspring would die,’’ says Jeffrey Powell, senior author of a study describing the discovery. “That obviously was not what happened.”

The GM mosquito strain was developed by a company called Oxitec, and it had previously been given FDA approval for these kinds of tests. In the Brazilian case, around 450,000 modified males were released in Jacobina every week for 27 months, totaling tens of millions of bugs. To keep tabs on them, the Yale team studied the genomes of both the GM strain and the wild species before the release, then again six, 12 and 27 to 30 months after the release began.

Sure enough, by the end of the test there was clear evidence that genes from the transgenic insects had been incorporated into the wild population. Although the GM mosquitoes only produce offspring about three to four percent of the time, it seems that those that are born aren’t as weak as expected. Some appear to make it to adulthood and breed themselves.

While populations did drop initially, numbers did bounce back after about 18 months. The researchers suggest that female mosquitoes may have learned and begun avoiding mating with the modified males.

Worse still, the genetic experiment may have had the opposite effect and made mosquitoes even more resilient. The bugs in the area are now made up of three strains mixed together: the original Brazilian locals, plus strains from Cuba and Mexico – the two strains crossed to make the GM insects. This wider gene pool could make the mozzies more robust as a whole.

The scientists assure the public that the mixed mosquitoes pose no extra health risk, but there is still cause for concern. It’s unclear exactly what effect this will have on disease transmission or other control methods.

“It is the unanticipated outcome that is concerning,” says Powell. “Based largely on laboratory studies, one can predict what the likely outcome of the release of transgenic mosquitoes will be, but genetic studies of the sort we did should be done during and after such releases to determine if something different from the predicted occurred.”

The research was published in the journal Scientific Reports.

Source: Yale University

(*) Author: 

Michael Irving has always been fascinated by space, technology, dinosaurs, and the weirder mysteries of physics and the universe. With a Bachelor of Arts in Professional Writing under his belt, he’s been writing for various online outlets and print publications for eight years, and New Atlas for the last three years.


Transgenic Aedes aegypti Mosquitoes Transfer Genes into a Natural Population

Scientific Reportsvolume 9, Article number: 13047 (2019) | Download Citation


In an attempt to control the mosquito-borne diseases yellow fever, dengue, chikungunya, and Zika fevers, a strain of transgenically modified Aedes aegypti mosquitoes containing a dominant lethal gene has been developed by a commercial company, Oxitec Ltd. If lethality is complete, releasing this strain should only reduce population size and not affect the genetics of the target populations. Approximately 450 thousand males of this strain were released each week for 27 months in Jacobina, Bahia, Brazil. We genotyped the release strain and the target Jacobina population before releases began for >21,000 single nucleotide polymorphisms (SNPs). Genetic sampling from the target population six, 12, and 27–30 months after releases commenced provides clear evidence that portions of the transgenic strain genome have been incorporated into the target population. Evidently, rare viable hybrid offspring between the release strain and the Jacobina population are sufficiently robust to be able to reproduce in nature. The release strain was developed using a strain originally from Cuba, then outcrossed to a Mexican population. Thus, Jacobina Ae. aegypti are now a mix of three populations. It is unclear how this may affect disease transmission or affect other efforts to control these dangerous vectors. These results highlight the importance of having in place a genetic monitoring program during such releases to detect un-anticipated outcomes.


Mosquito-borne diseases take a tremendous toll on human health and economies especially in Third World countries. Effective vaccines and drugs are available for only a few so the major means of controlling these diseases is to control the mosquitoes that transmit them. As traditional methods of control, such as insecticides, have become less effective and acceptable, alternative methods have been sought1. Methods based on genetic manipulations are among the most appealing and actively pursued2. One such genetic-based program has involved releasing a strain of Aedes aegypti (OX513A) that was transgenically modified to be homozygous for a conditional dominant lethal3,4. This strain also carries a fluorescent protein gene that allows detection of OX513A X wild type F1 offspring. Release of this strain in large numbers has been effective in reducing populations of Ae. aegypti by up to 85%5. The largest such releases to date have been carried out in the city of Jacobina in Bahia, Brazil6. We monitored the Jacobina Ae. aegypti population to determine if the releases have affected the genetics of the natural population by transferring genes, introgressing. If lethality is complete, such releases should result only in population reduction and not affect the genetics of the target population. However, it is known that, under laboratory conditions, 3–4% of the offspring from matings of OX513A with wild type do survive to adulthood although they are weak and it is not known if they are fertile4.

Materials and Methods

Release and rearing sites

Jacobina, in the state of Bahia, Brazil, is a moderately sized city of ~75,000 inhabitants located at coordinates 11°10′51″S, 40°31′04″W (Fig. 1). Jacobina is surrounded for several kilometers in all directions by caatinga, a dry ecological biome in which Ae. aegypti cannot breed, making Jacobina an island for this mosquito.

Figure 1



Map of Jacobina. Ovitraps where samples were collected are indicated with colored dots, coded by neighborhood. Releases were made in the neighborhoods of Pedra Branca, Catuaba, and Inocoop but never in the Centro area. © OpenStreetMap contributors.

Full size image

The rearing facility for the release strain is located at the Biofabrica Moscamed Brasil in Juazeiro, some 200 kilometers north of Jacobina. Mass rearing and sexing are described in Harris et al.7. Weekly, male pupae were transported to Jacobina and held in a local facility for one week to allow eclosion before release; approximately 450 thousand OX513A males were released each week beginning in June 2013 and continued through September 20156. Releases were made in the Pedra Branca, Catuaba, and Inocoop neighborhoods, but never in Centro. Oviposition traps were sampled weekly in the localities indicated in Fig. 1. Eggs were hatched and the frequencies of fluorescent and wild type larvae recorded; see Garzeira et al.6 for details of proportion fluorescent and wildtype at each time point. Fourth instar larvae of each type were placed in ~80% ethanol and brought to Yale University of genotyping. Further data on the effect of releases in Jacobina can be found in Graziera et al.6.

Genetic analyses

We used a custom developed Affymetrix SNP chip for genotyping8. Approximately 200 ng of genomic DNA from individual mosquitoes were placed in 95 wells of a 96 well plate, with one distilled water control. Plates were sent to the Functional Genomics Core at the University of North Carolina, Chapel Hill, for hybridization and production of data files sent to Yale University. We used the R package SNPolisher v1.4 (Afffymetrix, Santa Clara, CA) to generate and process genotype calls. While the SNP chip contains probes for about 27,000 well-validated biallelic SNPs passing tests for Mendelian inheritance and genotyping >98% of all samples8, 21,770 were polymorphic in our samples from Jacobina and genotyped in >98% of all individuals.

We genotyped samples taken from Centro and a combined Catuaba/Pedra Branca sample before releases began. Then, while the releases were continuing, we sampled all neighborhoods six, 12 and 27–30 months after releases began. The last sample at 27–30 months was a combined sample for three months included after the releases ceased at 27 months. Sample sizes are in Table 1. Except for the final combined 27–30 month sample, each sample analyzed after releases began were from egg traps exposed for a single week and larvae sampled from at least five traps in each neighborhood. The position of the traps remained the same throughout the study.

Population (sample size) Hybrid index (h) Range (mean) Number of samples with h-index >0.02 Number of samples with h-index >0.04
OX513A strain (25) 0.99–1.00 (0.999)    
F1 hybrids 6 months (57) 0.40–0.53 (0.47)    
Centro (64) 0–0.02 (0.0006) 0 0
Catuaba/Pedra (88) 0–0.01 (0.0002) 0 0
Catuaba 6 months (93) 0.001–0.134 (0.023) 29 (31.2%) 10 (10.8%)
Catuaba 12 months (35) 0.002–0.123 (0.033) 21 (60.0%) 11 (31.4%)
Catuaba 27 months (21) 0–0.120 (0.016) 5 (23.8%) 1 (4.8%)
Inocoop 12 months (44) 0.002–0.11 (0.027) 23 (52.3%) 8 (18.2%)
Inocoop 27 months (26) 0–0.123 (0.018) 5 (19.2%) 4 (15.4%)
Pedra Branca 6 months (6) 0.008–0.016 (0.013) 0 0
Pedra Branca 12 months (56) 0.002–0.134 (0.03) 25 (44.6%) 11 (19.6%)
Pedra Branca 27 months (22) 0–0.110 (0.016) 5 (20.8%) 4 (14.8%)
Centro 6 months (16) 0.003–0.010 (0.009) 0 0
Centro 12 (14) 0.0–0.040 (0.014) 4 (28.6%) 0
Centro 27 months (7) 0.0–0.007 (0.004) 0 0
  1. Two cutoffs for introgressed individuals: h = 0.02, the maximum observed in prerelease samples and h = 0.04, the maximum observed in Centro after releases.

Table 1 Results of the “INTROGRESS” analysis as performed using the R package10Full size table

To confirm our genetic analyses were accurate in detecting hybrids, we also genotyped 57 fluorescent larvae collected six months into the releases representing F1 offspring between the release strain and the natural population.


We performed three types of analyses. First, to confirm that our panel of SNPs could discriminate between the release strain OX513A and the natural population before release, we performed a Principal Components Analysis (PCA) using the R package in LEA9. Second, the R package “introgress”10 was implemented designating OX513A and Jacobina before release (combined Centro, Catuaba, and Pedra Branca neighborhoods) as the two parental populations. Third, we performed an ADMIXTURE analysis as describe in11 and shown Fig. 2C. For this analysis we filtered to exclude tightly linked SNPs using the –indep option of PLINK12 resulting in a panel of 14,252 SNPs. Then, an ANOVA analysis followed by a post-hoc TukeyHSD test was used to test for statistical differences (confidence level 0.95) in the mean Q values between the populations and most importantly between the pre- and the post-release populations.

Figure 2



(A) Principal Components Analysis (PCA) on the OX513A release strain and three neighborhoods Jacobina (Centro and Catuaba/Pedra Branca) before releases began. (B) Hybrid index (h-index) as performed in INTROGRESS10. An index of 1.0 indicates the “pure” OX513A individuals, 0.0 indicates the “pure” Jacobina pre-release individuals. Individuals are organized by neighborhood indicated at bottom of the figure, then by collection date: pre-release, 6, 12 or 27–30 months post release. Fluorescence verified F1 hybrids are grouped and labeled as F1. The horizontal dashed line represents cutoff (h-index = 0.02) the maximum observed pre-release. (C) ADMIXTURE11 analysis of all individual genotypes. Proportion of each color for each individual represents the proportion of that individual’s ancestry attributable to the red (OX513A) or blue (Jacobina pre-release) cluster.

Full size image

Virus infections

The dengue virus serotype 2 (DENV-2) strain tested was isolated during an epidemic in Brazil in 2010 from a patient in Santos, Brazil. The strain, designated ACS4613, was described in Cugola et al.14 and was kindly provided by the Evandro Chagas Institute in Belém, Pará.

The mosquito infection procedures are described in detail in Cost-da-Silva et al.15. Briefly, pre-mated five to seven day old females were blood-fed artificially using Glytube feeder (22). DENV-2 of ninth subculture (T9) or ZIKVBR of fourth subculture (T4) were mixed with human concentrated erythrocyte and inactivated blood serum to feed the females. DENV-2 and ZIKVBR final concentrations in the feeding solution were 1.7 × 1010 genome copies/ml and 2.2 × 106 plaque forming unit (pfu)/mL, respectively.

Virus assays

Engorged females from ROCK, OX513A and Jacobina strains were separated from non-engorged mosquitoes and maintained on 10% sucrose. Fourteen days post-blood meal (14 PBM), females were CO2 anaesthetized and kept on ice. Individual mosquito bodies were separated from heads and frozen separately immediately on dry ice and stored at −80 °C. Total RNA was extracted using QIAamp Viral RNA Mini Kit (Qiagen). DENV-2 or ZIKVBR genomic copies were measured using one-step qRT-PCR method as described in (22). To generate DENV-2 standard curve, a 119-bp fragment from the ACS46 strain was amplified with D1-TS2 primers15 and was cloned into the pCR2.1 vector (Invitrogen). This plasmid was used to estimate the number of DENV copies for each sample. The thermocycler conditions for DENV-2 amplification were 48 °C for 30 min and 95 °C for 10 min; 45 cycles of 95 °C for 30 sec, 55 °C for 30 sec and 60 °C for 30 sec, and a melting curve step of 95 °C for 1 min, 60 °C for 30 sec and 95 °C for 1 min, with temperature ramping from 60 °C to 95 °C at 0.02 °C/sec.

Statistical analyses were performed to assess significant differences in viral levels (Kruskal-Wallis test followed by Dunn’s Multiple Comparison Test) or infection rates of heads or bodies (Fisher’s exact test) between the three mosquito strains. The program and procedures to perform the analyses were previously described15.


Figure 2A shows that our 21,770 SNPs clearly distinguish OX513A and the natural Jacobina population. In Fig. 2B,C it is clear that the three neighborhoods before releases, Pedra Branca, Catuaba, and Centro, are genetically quite homogeneous; that is, there is no indication of genetic heterogeneity in Ae. aegypti samples across the ~6 km length of the city (Fig. 1) before releases began. Figure 2B,C also indicate we can identify F1 offspring between the release strain and natural population in Jacobina.

To detect introgression we genotyped a total of 347 wild type (non-fluorescent) Ae. aegypti in Jacobina sampled at 6 months, 12 months, and 27–30 months after releases began. Figure 2B,C clearly indicate individual mosquitoes with mixed genomes in neighborhoods where the releases were done. Even in the neighborhood where no releases were performed, Centro, some degree of introgression may be detectable, possibly due to migration from the release neighborhoods about four kilometers distant (Fig. 1). In Table 1, we present numerical data including sample sizes (in parentheses) for each sample in each locality at each time point. We use two cutoff points indicating unambiguous introgressed individuals: h = 0.02 the maximum observed before releases (also the dotted line in Fig. 2B) and h = 0.04 the maximum found after releases in Centro where no releases were performed. In the less stringent criterion, between about 20 and 60% of the sampled mosquitoes were introgressed; for the more stringent criterion, about 5 to 30% are introgressed. The maximum introgression (h value) possible is 25%, first generation backcross. The maximum we observed was 0.13 indicative of second generation backcross; our fist sample at six months is sufficient time to produce multiple backcross generations given a generation time of about one month. It is expected that the earliest backcross progeny to be rarer than later generations so it is not surprising that only a single second generation backcross progeny was observed with the majority advanced backcrossed.

The data in Fig. 2 and Table 1 are for all mosquitoes sampled. We also pruned the data to control for unequal sample sizes and the results are similar with, in fact, more individuals over the cutoff points due likely to a more homogeneous parental groups (Extended Data, Table E1). The frequency of sampling introgressed individuals increased between samples at six months and 12 months, but decreases somewhat at 27 months (Table 1 and Extended Data Table E2).

It is difficult to perform statistical tests on the h-index (Fig. 2B) but the STRUCTURE plots with Q values (Fig. 2C) allow statistical testing. ANOVA followed by a TukeyHSD tests confirmed significant (p < 0.05) differences on the mean Q values of pre-release in Catuaba at six and 12 months, and at 12 months in Inocoop and Pedra Branca (Extended Data, Fig. E1).

The results of our tests of the infectivity of one strain each of the dengue and Zika viruses in females of the OX513A strain and the Jacobina natural population (before releases) indicate no significant differences (Fig. 3).

Figure 3



Levels of DENV-2 (left) and ZIKV (right) genomic copies detected in heads and bodies without heads of Rockefeller, OX513 and Jacobina mosquitoes challenged via oral infection. None of the strains differ by Kruskal Wallis tests followed by Dunn’s post-test (p > 0.05). Experimental details in Supplementary Materials.

Full size image


Our data clearly show that release of the OX513A has led to significant transfer of its genome (introgression) into the natural Jacobina population of Ae. aegypti. The degree of introgression is not trivial. Depending on sample and criterion used to define unambiguous introgression, from about 10% to 60% of all individuals have some OX513A genome (Tables 1 and E1).

One seeming anomaly in the data is the apparent decrease in frequency of introgressed individuals between the 12 month sample and the 27–30 month sample. However, it is clear from the data in Garziera et al.6 that the effectiveness of the release program began to break down after about 18 months, i.e., the population which had been greatly suppressed rebounded to nearly pre-release levels. This has been speculated to have been due to mating discrimination against OX513A males, a phenomenon known to occur in sterile male release programs16. This observation also implies that introgressed individuals may be at a selective disadvantage causing their apparent decrease after release ceased, although much more data would be needed to confirm this.

It is not known what impacts introgression from a transgenic strain of Ae. aegypti has on traits of importance to disease control and transmission. We tested OX513A and Jacobina before releases for infection rates by one strain each of the dengue and Zika viruses and found no significant differences (Fig. 3). However, this is for just one strain of each virus under laboratory conditions; under field conditions for other viruses the effects may be different. Also, introgression may introduce other relevant genes such as for insecticide resistance. The release strain, OX513A, was derived from a laboratory strain originally from Cuba, then outcrossed to a Mexican population7. The three populations forming the tri-hybrid population now in Jacobina (Cuba/Mexico/Brazil) are genetically quite distinct (Extended Data Fig. E2), very likely resulting in a more robust population than the pre-release population due to hybrid vigor.

These results demonstrate the importance of having in place a genetic monitoring program during releases of transgenic organisms to detect un-anticipated consequences.

Change history

  • 17 September 2019

    Editor's Note: readers are alerted that the conclusions of this paper are subject to criticisms that are being considered by editors. A further editorial response will follow the resolution of these issues.


  1. 1.

    Ferguson, N. M. Challenges and opportunities in controlling mosquito-borne infections. Nature 559, 490 (2018).

  2. 2.

    McGraw, E. A. & O’Neill, S. L. Beyond insecticides: new thinking on an ancient problem. Nature Rev Microbiol 11, 181–193 (2013).

  3. 3.

    Thomas, D. D., Donnell, C. A., Wood, J. J. & Alphey, L. S. Insect population control using a dominant, repressible, lethal genetic system. Science 287, 2474–2476 (2000).

  4. 4.

    Phuc, H. K. et al. Late-acting dominant lethal genetic systems and mosquito control. BMC Biol 5, 11 (2007).

  5. 5.

    Carvalho, D. O. et al. Suppression of a field population of Aedes aegypti in Brazil by sustained release of transgenic male mosquitoes. PLoS Negl Trop Dis 9(7), e0003864 (2015).

  6. 6.

    Garziera, L. et al. Effect of interruption of over-flooding releses of transgenic mosquitões over wild populatio of Aedes aegypi: two case studies in Brazil. Entomol. Eperiment. Appl. 164, 327–339 (2017).

  7. 7.

    Harris, A. et al. Successful suppression of a field mosquito population by sustained release of engineered male mosquitões. Nature Biotch. 30, 828–830 (2002).

  8. 8.

    Evans, B. R. et al. A multipurpose high-throughput SNP chip for the dengue and yellow fever mosquito. Aedes aegypti. G3(5), 711–718 (2015).

  9. 9.

    Frichot, E. & Francois, O. LEA: an R package for Landscape and Ecological Association studies. Methods Ecol. Evol. 6, 925–929 (2015).

  10. 10.

    Gompert, G. & Buerkle, C. A. INTROGRESS: a software package for mapping components of isolation in hybrids. Mol Ecol Res 10, 378–384 (2010).

  11. 11.

    Alexander, D. H. & Lange, K. Enhancements to the ADMIXTURE algorithm for individual ancestry estimation. BMC Bioinformatics 12, 246 (2011).

  12. 12.

    Purcell, S. et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81, 559–575 (2007).

  13. 13.

    Romano, C. M. et al. Inter- and intra-host viral diversity in a large seasonal DENV2 outbreak. PLoS One 8, e70318 (2013).

  14. 14.

    Cugola, F. R. et al. The Brazilian Zika virus strain causes birth defects in experimental models. Nature 534, 267–271 (2016).

  15. 15.

    Costa-da-Silva, A. L. Laboratory strains of Aedes aegypti are competent to Brazilian Zika virus. PLoS One 12, e0171951 (2017).

  16. 16.

    Powell, J. R. Genetic variation in insect vectors: death of typology? Insects 9, 139 (2018).

Download references


Financial support was provided by grants from the U. S. National Institute of Allergies and Infectious Diseases, UO1 AI115595, JRP, Principal Investigator. BRE was a training fellow on NIH 5T32AI007404. JRP was a recipient of a Science Without Borders Fellowship from the Brazilian National Government.

Author information


  1. Yale University, 21 Sachem Street, New Haven, CT, 06520-8105, USA

    • Benjamin R. Evans
    • , Panayiota Kotsakiozi
    •  & Jeffrey R. Powell
  2. Departamento de Parasitologia, Instituto de Ciências Biomédicas, Universidade de São Paulo, Av. Prof. Lineu Prestes, 1374, São Paulo, SP, 05508-000, Brazil

    • Andre Luis Costa-da-Silva
    • , Rafaella Sayuri Ioshino
    • , Michele C. Pedrosa
    •  & Margareth L. Capurro
  3. Instituto Nacional de Ciência e Tecnologia em Entomologia Molecular, INCT-EM, Rio de Janeiro, Rio de Janeiro, Brazil

    • Andre Luis Costa-da-Silva
    • , Rafaella Sayuri Ioshino
    • , Luiza Garziera
    • , Michele C. Pedrosa
    •  & Margareth L. Capurro
  4. Moscamed Brasil, Loteamento Centro Industrial São Francisco 9 - lt 15, Juazeiro, BA, 48908-000, Brazil

    • Michele C. Pedrosa
    • , Aldo Malavasi
    •  & Jair F. Virginio


B.E. and J.P. developed technologies and performed genotyping; B.E. and P.K. performed analyses and prepared figures; A.L.C., R.S.I. and M.L.C. performed vector competence tests; L.G., M.P., A.M. and J.V. performed the releases and collected samples post release; J.P. conceived and directed the project and wrote the manuscript.

Corresponding author

Correspondence to Jeffrey R. Powell.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Transgenic Aedes aegypti Mosquitoes Transfer Genes into a Natural Population

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Share this article



Zika Virus is Patented ...