A confinable female-lethal population suppression system in the malaria vector, Anopheles gambiae

Malaria is among the world’s deadliest diseases, predominantly affecting Sub-Saharan Africa and killing over half a million people annually. Controlling the principal vector, the mosquito Anopheles gambiae, as well as other anophelines, is among the most effective methods to control disease spread. Here, we develop a genetic population suppression system termed Ifegenia (inherited female elimination by genetically encoded nucleases to interrupt alleles) in this deadly vector. In this bicomponent CRISPR-based approach, we disrupt a female-essential gene, femaleless (fle), demonstrating complete genetic sexing via heritable daughter gynecide. Moreover, we demonstrate that Ifegenia males remain reproductively viable and can load both fle mutations and CRISPR machinery to induce fle mutations in subsequent generations, resulting in sustained population suppression. Through modeling, we demonstrate that iterative releases of nonbiting Ifegenia males can act as an effective, confinable, controllable, and safe population suppression and elimination system.

: gFLE/Cas9 individuals have Δfle mutations under gRNA target sites. gRNA7 (top): Reads 1-5 are from a + J /+ larvae, gFLE G /Cas9 adult male, gFLE J /Cas9 larvae, a F2 gFLE I /Cas9 phenotypic female that survived to adulthood, and a F3 gFLE J /Cas9 intersex individual that died during eclosure respectively. gRNA10 (bottom): Reads 6-14 are from F2 larvae of mixed genotypes encompassing most genotypes and gFLE families. Read 15 was isolated from the only F1 gFLE G /Cas9 phenotypic female identified (Table S16), however which later PCR amplified for the presence of a Y-chromosome suggesting feminization. Reads 16 and 17 were isolated from a F2 gFLE I /Cas9 phenotypic female that survived to adulthood (Table S17), which also contained the gRNA7-derived mutation listed in Read 4. Reads 18 and 19 were identified from a F2 gFLE J /Cas9 phenotypic female that died as a pupae. Reads 20 and 21 are from a F2 gFLE J /Cas9 female that died as an adult. Reads 22 and 23 are from a F3 gFLE J /Cas9 intersex individual that died during eclosure, which also contained gRNA7-derived Read 5.    Reported as 1 day old (do) larvae counts by genotype of the offspring from a cross of gFLE/+ males to Cas9/+ females. Significant embryonic female death would be expected to be observed as an approximate halving of the gFLE/Cas9 group. A) Raw counts of 1do larvae by genotype, offspring from a +/gFLE G ♂ to +/Cas9♀cross. Two replicates shown stacked in black and grey. Both not significantly different from expected 1:1:1:1 Mendelian ratios, (χ 2 , p > 0.1 and p > 0.2 respectively). B) 1do larvae counts from a +/gFLE I ♂ to +/Cas9♀ cross. Two replicates shown stacked in black and grey. Both are not significantly different from expected 1:1:1:1 Mendelian ratios, (χ 2 , p > 0.5 and p > 0.2 respectively). C) 1do larvae counts from a +/gFLE J ♂ to +/Cas9♀ cross. Two replicates shown stacked in black and grey. Both significantly different from expected 1:1:1:1 Mendelian ratios (both χ 2 , p < 0.005) consistent with multiple insertions of transgene gFLE J. Family gFLE J was therefore omitted from the analysis in D). D) Data pooled from A) and B). Together these results demonstrate very slight levels of female embryo lethality in the gFLE/Cas9 group (χ 2 p = 0.0236) Figure S7. Femaleless mosaic Δfle mutants (gFLE/+ and gFLE/Cas9) die during larvaehood. 40 random 1 day old (do) larvae were isolated into trays by genotype and reared separately. The number, sex, and genotype of individuals is reported upon pupation. Replicates 1-4 are denoted by triangle, square, diamond, and circle respectively. Mean and SD shown A) Δfle females (gFLE G /+ and gFLE G /Cas9) were present at 1do but failed to pupate, all 4 replicates shown. B) Δfle females (gFLE I /+ and gFLE I /Cas9) were present at 1do but failed to pupate. A third replicate was not performed on this line as the line was deemed sub-optimal for release and omitted from downstream analysis. C) Δfle females (gFLE J /+ and gFLE J /Cas9) from family were present at 1do but failed to pupate, three replicates shown.

Figure S9. Ifegenia males have high mating competitiveness. A)
Offspring genotype ratio from a cross of 35 Ifegenia males to 35 WT females. Genotype ratios were used to calculate the number of WT offspring in (B) that are attributed to Ifegenia males vs WT males. From this experiment it was determined that Ifegenia males sire 34.5% wild type offspring and 65.5% transgenics due to transgene linkage on the 2nd chromosome, instead of 25%/75% as would be expected if all transgenes were unlinked. B) Male mating competition assays of 35 Ifegenia males X 35 WT males X 35 WT females. All larvae were counted and genotyped. A fraction of WT larvae were attributed to Ifegenia fathers according to the ratios shown in (A) and reported with transgenic larvae as percent Ifegenia offspring (WT adjusted). Ifegenia positive larvae were calculated as (n Ifegenia = n transgenics x 1/0.655), where 0.655 is the percent of transgenics from Ifegenia fathers as determined in (A). Mean and SD shown. p<0.0001, unpaired two-tailed t-test C) Egg numbers produced by each replicate. Mean and SD shown, no significant difference unpaired t-test D) Hatching rate of each replicate. Mean and SD shown, no significant difference, unpaired t-test.  Figure 1C, F2 offspring from family gFLE I were followed through the F2 generation, and offspring from family gFLE J were followed through the F3 generation. Phenotypic gFLE/Cas9 females were identified in both families though at significantly reduced frequencies than should be expected by Mendelian segregation. Hybrid F1 males inherited Cas9 maternally (Cas9 m ) A) To quantify approximate knockout frequencies under the gRNA7 and gRNA 10 target sites, fle was PCR-amplified with primers 1154A.S29 and 1154A.S8 (red) and subsequently digested with either restriction enzyme BseYI (gRNA7) or BstNI (gRNA10). These enzymes have semi-unique recognition sites (gray boxes) in wild type overlapping the predicted CRISPR mutagenesis sites (salmon chevron). CRISPR mutagenesis will preclude enzyme digestion of the PCR amplicons corresponding to that allele for most mutations. In diploids, PCR of wild type sequences should fully digest, PCR of heterozygotes should partially digest, and PCR of homozygotes should fail to digest. B) 20 gFLE G /Cas9 individuals for each genotype-sex are shown numbered across the top in brackets. Three consecutive wells are loaded for each sample, left to right: undigested fle PCR product, BseYI-digested PCR product, BstNI-digested PCR product. Undigested BseYI and BstNI bands indicate probable CRISPR mutations under gRNA 7 (teal arrows) and gRNA 10 (olive arrows), respectively. Biallelic Δfle mutants are noted (highlighted arrows), most of which are in the gFLE G /Cas9 group due to active mosaic mutagenesis. Probable large insertions or deletions are denoted with purple arrows. Areas of the gel where exposure and brightness were adjusted separately are noted with orange boxes. The frequency of Δfle mutant individuals (those with at least one mutation) in that genotype-sex cohort are summarized at right in yellow.

Figure S12. Sensitivity analysis of Ifegenia and pgSIT model outcomes to parameters describing the constructs, their fitness consequences and release schemes.
Sensitivity analysis of the "window of protection" model outcome (i.e., the duration for which the A. gambiae population is suppressed by ≥90%) to parameter values describing the release scheme (number of releases, and number of eggs per release), fitness of transgenic mosquitoes (male mating competitiveness), and construct attributes (allelic cutting rate, and frequency of maternal deposition of Cas in embryos of mothers expressing Cas) for Ifegenia (1-3 sites) and pgSIT. The sensitivity analysis was conducted by simulating a randomly-mixing population of 10,000 adult mosquitoes using the MGDrivE simulation framework [27] with parameters described in Table S20. For parameters varied in the sensitivity analysis, number of releases varied from 1-48, number eggs released per adult varied from 10-500, male mating competitiveness varied from 0.5-1, allelic cutting rate varied from 0.75-1, and frequency of maternal deposition of Cas varied from 0.75-1. Parameter sensitivity was calculated based on three methods: i) Delta [54], ii) Fourier Amplitude Sensitivity Test (FAST) [55], and iii) High-Dimensional Model Representation (HDMR) [56]. According to all three sensitivity metrics, the window of protection outcome is most sensitive to parameter values describing the release scheme, and relatively insensitive to those describing the construct and mating competitiveness of transgenic mosquitoes.

Text S1. Modeling generation of alleles resistant to Ifegenia.
We calculate the efficiency of Ifegenia ( -target sites) at suppressing the propagation of generated resistant alleles. To this end, we first note that mosquito survival is a necessary condition for a newly-generated resistant allele to propagate to the next generation. Given the inheritance rules of Ifegenia, the resistant allele can only be passed to the next generation if the other target sites are not cleaved on both chromosomes. Thus, we first calculate the − 1 probability, , of a single target site not being cleaved on both chromosomes. This is the case when: i) no target site alleles are cleaved, ii) only one target site allele is cleaved, or iii) two alleles are cleaved and at least one of them develops resistance. Under these conditions, we calculate the probability of survival, , as, Here, represents the probability that a target site is cleaved, given the presence of Cas and that the guide RNA that targets that site, and represents the probability that a resistant allele is generated at that target site, given it is cleaved. It follows from the above equation that, assuming cleavage of each target allele is an independent event, the probability that all target sites are not cleaved on both chromosomes is given by . This means that, beginning with a wild-type population, when a resistant allele emerges at one target site, the probability that the mosquito will survive and the resistant allele will be passed to the next generation is given by .

−1
We simulate Ifegenia with 2-4 target sites ( Figure S13) and note that, for >90% and <10%, <35%, 12% and 5%, respectively. In other words, the probability of propagating a resistant −1 allele, assuming >90% and <10%, is reduced by more than 65%, 88% and 95% for Ifegenia with two, three and four target sites, respectively. Figure S13. Mosquito survival probability given emergence of a resistant allele at one of the Ifegenia target sites. Survival probability is shown for Ifegenia having two (A), three (B) and four (C) target sites, and is relative to the case for a single target site. The model used to derive mosquito survival probability is provided in Text S1. For cutting rate probability, >90%, and resistance generation probability, <10%, the mosquito survival probability, was found to be <35%, <12% and <5%, −1 , respectively. In other words, when the cutting rate is high ( >90%) and emergence rate of resistance alleles is low ( <10%), the probability of propagating a resistant allele from one generation to the next is reduced by more than 65%, 88% and 95% for Ifegenia having two, three and four target sites, respectively, as compared to the case of having one target site. Figure S14. In vitro gRNA cleavage assays performed by CRISPR QC. Ten gRNAs predicted to be good candidates by http://crispor.tefor.net/ were assayed for cleavage efficacy using CRISPR QC's proprietary technology. Cleavage rates are reported as fold change in solution conductance normalized to KMT2D, a non-cutting gRNA with the template. gRNA7 and gRNA10 were selected for use in this work.   Fig 1D Top row).       Table S10: RNAseq differential expression analysis comparing Vasa-Cas9 /gFLE(G) transheterozygotes vs. gFLE(G) Table S11: Mendelian ratios of one day old larvae from +/gFLE X +/Cas9 -females survive embryogenesis.

SUPPLEMENTARY TABLES
Table S12: Survival of genotypes through larval stage -females die before pupation.     Figure 1E) Table S17 : Raw sex-genotype pupae count for F2 offspring from +/gFLE X +/Cas9 crosses.   Table S19. Sex-genotype ratios of the F3 generation (offspring of F2 gFLE/Cas9 males x WT females) show that female-killing persists for multiple generations.