Genetically engineered insects with sex-selection and genetic incompatibility enable population suppression.

Engineered Genetic Incompatibility (EGI) is a method to create species-like barriers to sexual reproduction. It has applications in pest control that mimic Sterile Insect Technique when only EGI males are released. This can be facilitated by introducing conditional female-lethality to EGI strains to generate a sex-sorting incompatible male system (SSIMS). Here, we demonstrate a proof of concept by combining tetracycline-controlled female lethality constructs with a pyramus-targeting EGI line in the model insect Drosophila melanogaster. We show that both functions (incompatibility and sex-sorting) are robustly maintained in the SSIMS line and that this approach is effective for population suppression in cage experiments. Further we show that SSIMS males remain competitive with wild-type males for reproduction with wild-type females, including at the level of sperm competition.

Introduction

Arguably the most successful large-scale insect control approach to date is Sterile Insect Technique (SIT) (Dyck et al., 2005; Barsanti et al., 1979). SIT uses irradiation at sub-lethal doses to sterilize mass-reared insects prior to their targeted environmental release. Such an approach was successfully applied to eradicate the New World Screwworm (Cochliomyia hominivorax) from North and Central America over a 50-year period from the mid 1950s to early 2000s (Scott et al., 2017). SIT has been successfully applied for the broad-scale control of tephritid fruit flies (e.g. Ceratitis capitata, Zeugodacus cucurbitae, Bactrocera tryoni, and Anastrepha ludens), onion maggots (Delia antiqua), tse-tse flies (Glossina spp.), and several coleoptera and lepidoptera species (Dyck et al., 2005).

Several factors limit the widespread adoption of SIT for more insect pests. Exposing insects to sterilizing doses of irradiation can impact their longevity and mating competitiveness (Lance et al., 2000). Sometimes the lethal dose of radiation is sufficiently close to the sterilizing dose of radiation such that achieving 100% sterilization means that the vast majority of the males will not survive (Ricardo Machi et al., 2019). Many batch irradiation approaches do not separate males from females prior to release, and this can both decrease the effectiveness (Rendón et al., 2004), and lead to public health concerns when the female insect is a disease vector (Benedict, 2021).

Recently, biotechnology-enabled approaches have been developed that have the potential to complement or replace SIT programs, or allow SIT-like programs that target a wider range of species. Integrating an early acting female lethality system to New World Screwworm without impacting male fitness and fecundity has the potential to improve SIT effectiveness by several fold (Concha et al., 2020). Release of insects with dominant lethality (RIDL), has been successfully demonstrated to suppress target mosquito populations in field trials in the Caribbean islands and South America (Harris et al., 2012). Precision-guided SIT (pgSIT) utilizes expression of Cas9 nuclease to knock out female survival and male fertility genes prior to generate potential biocontrol agents (Li et al., 2021; Kandul et al., 2019).

Engineered Genetic Incompatibility (EGI) is a strategy for creating bi-directional (male-to-female and female-to-male) mating incompatibility that mirrors the behavior of post-zygotic speciation mechanisms. (Maselko et al., 2020; Maselko et al., 2017; Buchman et al., 2021; Waters et al., 2018). We created EGI lines by engineering flies to express a dCas9-based programmable transcription activator (PTA) that targets the promoter of a gene which is lethal with expressed too high or in the tissues. This lethal over- or ectopic-expression is avoided in the EGI line by making it homozygous for a mutation that prevents binding of the PTA. Because the EGI line is homozygous for both the PTA and the promoter resistance mutation, it is true-breeding and has normal fecundity. However, outcrossing with wild-type produces hybrid offspring that inherit one copy of the PTA and only one copy of the resistance allele. Because the PTA is haplosufficient for lethality (i.e. one copy is lethal) and the resistance allele is haploinsufficient (i.e. strains must be homozygous for the resistance allele to survive in the pressence of the PTA), all hybrid offspring are inviable. Importantly, multiple mutually incompatible EGI lines can be created that are incompatible with wild-type and with each other, providing a way to rationally engineer negatively correlated cross-resistance for the first time (Maselko et al., 2020).

As with other underdominant biocontrol methods, EGI could be used as a threshold-dependent gene drive (Magori and Gould, 2006; Davis et al., 2001; Buchman et al., 2021). Alternatively, it could be coupled to a conditional female-lethal construct, such as the tetracycline (Tet)-repressible positive feedback circuit (Fu et al., 2007; Li et al., 2014), to produce a strain that more closely replicates SIT biocontrol agents. Here, we demonstrate the successful engineering of such a system in Drosophila melanogaster. We quantify the behavior of individual genetic components, perform male mating competition assays, and test the ability of SSIMS flies to suppress a wild-type population in multi-generational laboratory cage trials.

Results

Generating and validating SSIMS lines

To generate the SSIMS line, we made a stock that contains two female lethal (FL) constructs (Das et al., 2020) on the X chromosome and an EGI construct on the third chromosome (Figure 1). The female lethal construct contains 21 copies of the tet operator upstream of the Hsp70 promoter (Li et al., 2014). An intron from the C. hominivorax tra gene that exhibits sex-specific splicing (alternative 5’ splice sites) is present immediately downstream of the start codon and upstream of a coding DNA sequence for the tet transactivator protein (Figure 1c). In females, this construct is expected to produce the tTA transgene, while the alternative splicing in males is expected to produce mature mRNA with several stop codons upstream of the tTA CDS. In comparing the behavior of strains containing one- or two-copies of this construct on the X chromosome, the two-copy strains showed a stronger female lethal phenotype, so we used a two-copy strain to construct SSIMS flies (Das et al., 2020). The EGI construct comprised a dCas9-VPR programmable transcriptional activator (PTA) targeting the promoter of a critical developmental morphogen gene (Figure 1d). In this study, we use a previously described EGI genotype that leverages lethal ectopic expression of the pyramus gene for hybrid incompatibility, but other genotypes are expected to work equally well. Throughout the strain-construction process, a mutation in the promoter of the pyramus gene was maintained to prevent binding of the PTA to the pyramus promoter, and therefore lethal ectopic expression (Appendix 1—figure 1). The SSIMS line used the foxo promoter to drive the PTA, as this genotype provided strong genetic incompatibility in previous studies (Maselko et al., 2020). Following the mating scheme outlined in Supplementary Note 1, we were able to generate viable SSIMS flies, which contained both of the visual markers of the FL genotype (GFP expression) and the EGI genotype (red eyes) (Figure 2a). Other EGI genetic designs (Maselko et al., 2020; Buchman et al., 2021) or FL designs (Yan and Scott, 2015) are predicted to work equally well for engineering SSIMS.

Figure 1.

Overall design and implementation of Sex-sorting Incompatible Male System (SSIMS).

(a) Illustration of desired behavior of SSIMS insects (green-yellow) when grown in the absence of Tet. Brown flies represent wild-type, and red crosses represent inviable offspring. (b) Genome-scale location of engineered loci in D. melanogaster SSIMS line. (c,d) Genetic cassette diagrams for the female lethal (FL) locus. The orange oval represents 21 repeats of the tet operator. The blue box represents the start codon. The red box denotes the exon sequence containing multiple stop codons that is retained when an alternative 5’ splice site is utilized in males. Female splicing fuses the start codon to the tTA coding DNA sequence. (d) Genetic cassette diagram for the engineered genetic incompatibility (EGI) loci, respectively. SBOL iconography is used for genetic part representation. FL, X-linked female lethal construct; Mbp, megabasepairs; kb, kilobases; Chr, chromosome; Mit, mitochondrial; PTA, Programmable Transcriptional Activator.

Appendix 1—figure 1.

Cross strategy to generate SSIMS line.

Female lethal (FL) and EGI (pyr337; PTA, gRNA-pyr) were crossed to balancer chromosomes, and subsequent males and females were isolated indicated by the black arrow. FL chromosomes contains 2 insertions on the X chromosomes, homozygous females contain 4 copies of the transgene. PTA+gRNA-pyr is a single insertion site containing foxo driven dead-Cas9::VPR and 2 guide RNAs driven by U6:3 and U6:1 promoters. Green shading indicates crosses done in the presence of 100 g/ml Tet. FL was tracked by GFP, PTA+gRNA was tracked by red eyes (mini white), and the pyr337 promoter mutation was tracked using balancers.

Figure 2.

Phenotype of SSIMS flies.

(a) Visual markers present in homozygous FL lines (GFP, left) and EGI lines (red eyes, center) are present in SSIMS flies generated via the mating scheme in Appendix 1—figure 1. (b) Counts of male (pink) and female (blue) offspring of SSIMS strain raised in food with 10 μg/ml Tetracyline (Tet) and without Tet. No females offspring were observed when larvae are raised in the absence of Tet. Chi-squared test shows significant difference from expected 50:50 ratio of males:female, n = 10 per group, *** denotes < 0.0001. (c) Counts of egg-lay (light gray) and surviving adults (dark gray) of wild-type females crossed with either wild-type or SSIMS males. SSIMS male mate successfully with wild-type females, producing slightly more eggs, but no adult offspring emerge, n = 5 per group, *** denotes p < 0.0001. (d) Percent female offspring when SSIMS stocks are reared in food with increasing concentrations of Tet, n = 10 per group. (e) Percent female offspring when respective progeny from (d) are reared in food without Tet, n = 7–10 per group. Line behind data shows Hill function that fits data with statistics described in the main text. (f) Percent female offspring when progeny from the 100 μg/ml group in (e) are reared in food without Tet, n = 30. FL, X-linked Female Lethal; EGI , engineered genetic incompatibility targeting pyramus; WT, wild-type; SSIMS, sex-sorting incompatible male system; Tet, Tetracycline.

Efficacy of genetic components in SSIMS line

To validate the SSIMS line we first tested whether the two underlying behaviors (FL and EGI) were maintained. We confirmed that the SSIMS line bred true in the presence of Tet at 100 μg/ml and contained both visual markers for FL (GFP) and EGI (mini-white) (Figure 2a). In the presence of Tet at (10 μg/ml) the SSIMS line shows a slight sex-ratio bias with fewer females (40% female) surviving to adulthood than expected 50% (Figure 2b). This suggests that the FL transgene is partially leaky in females at 10 μg/ml Tet. In the absence of Tet no females survive to adulthood and either die as larvae or pupae (Figure 2b). While we did not empirically characterize sex-specific splicing, the phenotype of the SSIMS line suggests that sex-specific splicing occurs as previously reported for this genetic construct (Li et al., 2014). To confirm the behavior of the EGI components, 20 wild-type females were mated with either four wild-type males or four SSIMS males, then we quantified number of eggs laid per vial in 24 hr and total number of offspring produced per vial (24 hr egg-lay). Wild-type females did not produce any viable adults when mated with SSIMS male (Figure 2c). There was no difference in fecundity of the wild-type and SSIMS strains, when mated to their own type, in a head-to-head comparison (Appendix 1—figure 2).

Appendix 1—figure 2.

Fecundity of SSIMS flies.

Experimental design schema and results of fecundity assay. Bar graphs show the number of offspring produced by one WT male crossed to one WT female (gray, n = 25) or with one SSIMS male crossed to one SSIMS female (green, n = 20). Welch two-sample t test was performed between the two groups. ns = not significant.

We tested the behavior of the female lethality construct at various concentrations of Tet (Figure 2d–f). When used in high concentrations, Tet has been reported to carryover between generations by maternal deposition in oocytes (Schetelig et al., 2009). In the presence of Tet at concentrations between 10 and 100 μg/ml there is a consistent but slight sex-ratio bias with 40–45% female offspring (Figure 2d). There is no statistically significant difference in the sex-ratio of offspring across these concentrations. When these offspring are transferred to food without Tet, the number of females that survive in the next generation varies from 0% to 35% in a dose-dependent manner (R2 = 0.86 to best-fit Hill function, t-value = 24.3, p < 0.00001). Parents raised on 10 μg/ml had no surviving female offspring when transferred to Tet-free medium, but parents raised on 100 μg/ml had 35% female offspring on Tet-free medium (Figure 2e). We confirmed that this likely arises from maternal deposition of Tet and not genetic instability of the female lethal construct by rearing these flies for one additional generation in Tet-free medium. We observed 100% female lethality in the second generation of rearing on Tet-free media (Figure 2f).

Male mating competitiveness and evidence for sperm displacement

We tested the ability of SSIMS males to compete with wild-type males for mating to wild-type females using two assays (Figure 3a and b). In the first, one male SSIMS fly, one male wild-type, and one virgin female were co-housed in a vial for 8 hr for mating. The female was then isolated to count how many offspring she had in the next four days (Figure 3a). Females that had not mated during the 8 hr period (eggs laid by virgin females neither hatch nor necrose as do eggs laid by non-virgins) were not included in the final counts. In control experiments lacking the SSIMS male, the females had an average of 76 offspring surviving to adulthood. In the competitive experiments, there was a clear bimodal distribution. Nine females produced an average of 82 females for one mode, and fifteen females produced zero adult offspring. This latter group had clearly mated, because the females laid eggs that hatched into larva that died soon after. There were two females that produced an intermediate number of offspring, suggesting they may have mated with with both wild-type and SSIMS males. We performed a second competition experiment with an excess of females, similar to that reported previously (Kandul et al., 2019). Twenty virgin females were mated with four males at a 4:0, 3:1, 2:2, 1:3, or 0:4 ratio of wild-type to SSIMS. Control matings were also performed with twenty wild-type females and three, two, or one wild-type males and no SSIMS males to account for the decrease in offspring that is due to limited wild-type male availability when competing with SSIMS males. Fewer offspring survive when SSIMS males are included in the mating vials (Figure 3b), and this cannot be explained by the decrease in the number of males that could sire surviving offspring (p < 0.05 for each comparison).

Figure 3.

Male mating competition with wild-type D. melanogaster.

(a) Experimental design schema and results of individual male mating competition assay. Violin plots show the number of offspring produced by one WT female crossed to one WT male (gray, n = 15) or one WT female crossed to one WT male and one SSIMS male (orange, n = 26). A Welch two-sample t test was performed between the two groups. (b) Experimental design schema and results of mixed-male mating assay. Box-whisker plots show the median (center line), 25th and 75th percentile boundaries (box), and min/max (whiskers). Student’s t tests were performed between experiments with identical numbers of wild-type males. n = 7 per group. (c) Experimental design schema and results of sperm displacement assay. Box-whisker plots colored to show the addition of no males (gray), wild-type males (blue), or SSIMS males (green) to the non-virgin wild-type females. Student’s t tests were performed for each egg-lay period comparing+ SSIMS male vs both no-male and WT-male groups. n = 10 per group. (d) Experimental design schema and results of offensive and defensive sperm displacement assay. Bar graphs show the number of offspring produced by one WT female mated to a WT male and remated with a second WT male (blue, n = 60), one WT female mated to a WT male and remated with a SSIMS male (green, n = 53), or one WT female mated to a SSIMS male and remated with a WT male (maroon, n = 54). Welch two-sample t tests were performed pairwise between each of the three groups. Statistical significance: ns = not significant, *=p < 0.01, **=p < 0.001, ***=p < 0.0001.

In Drosophila spp., females can store the sperm from multiple males in their spermathecae which allows for sperm to compete to fertilize the eggs. We tested the hypothesis that sperm from SSIMS males can displace wild-type sperm in the spermathecae of previously mated wild-type females. We placed three 5–6 day old non-virgin wild-type females (whose previous mating was with wild-type males) either with no males, three wild-type males, or three SSIMS males and measured total offspring of each group in 4-day windows for a total of 12 days (Figure 3c). Non-virgin wild-type females with either no males or with wild-type males for re-mating, produced an average of 70–90 offspring that survived to adulthood from each vial. While we did not empirically determine that females had mated during the initial 5–6 days, the assumption that females mated is supported by the negative control group (i.e. females moved to vials lacking males) which all produced offspring. When SSIMS males were available for re-mating, the number of surviving offspring dropped by 36% within the first egg lay period, and by 98–99% for the next two egg laying periods (Figure 3c). In the presence of SSIMS males, we observed post-zygotic lethality of eggs and larvae during all three egg-laying periods, which suggests that SSIMS males mated with the females to replace wild-type sperm with SSIMS sperm in the spermathecae. We performed a similar sperm displacement assay using shorter mating periods in both an offensive and defensive format, following methods described in previous studies (Clark et al., 1995). In both cases, the average number of surviving offspring following the second mating event decreased to one third the level of the control experiment (Figure 3d). There was not a significant difference in the number of offspring in offensive or defensive matings. This demonstrates that wild-type D. melanogaster will re-mate with SSIMS males in laboratory settings, and that the sperm from SSIMS males is competitive for fertilizing female eggs even in the presence of sperm from wild-type males in the spermathecae.

Population suppression cage trials

We assessed whether SSIMS could be applied to suppress wild-type populations by conducting a cage trial with multi-generational populations of wild-type D. melanogaster with and without additions of SSIMS animals (Figure 4a and Supplementary Note 4). In cages, adults were sustained by feeding a liquid solution of 5% yeast extract and 5% sucrose solution via a cotton wick. This method of feeding adults was used to overcome the problem of adults getting stuck in semi-solid food containers. To allow the population to expand, a 250-ml bottle (reproduction bottle) with 30-ml standard cornmeal food was introduced to each cage once a week for 24 hr. After 24 hr, adults were removed from the bottle and released back into the cage. The bottles were capped to prevent additional egg-laying, larvae overcrowding, and adults getting stuck in the semi-solid medium. Reproduction bottles were uncapped 10 days after egg-laying and left open for 3 days to allow for the next generation to eclose and integrate with the multi-generational caged population. To assess population size and genotype, we placed an active-yeast/sucrose trap inside each cage once per week for 24 hr. We did not measure whether the fraction of total flies caught in the yeast/sucrose trap was density dependent, but the trapped counts did seem to correlate with total population size in the cage upon visual inspection. We also measured the fecundity of 10 randomly captured wild-type females from each cage once per week.

Figure 4.

Laboratory cage trial of population suppression of wild-type D. melanogaster.

(a) Cage trial design showing Bioquip cage with adult feeding apparatus (1), yeast/sugar trap (2), reproduction bottle (3), fecundity assay vials (4), and weekly schematic of the trial, with tasks done each day highlighted below. Two control and two SSIMS treated cages were initiated, see methods for details. (b) Results of the fecundity assay from each week. Each colored dot represents total number of offspring from a single vial with one wild-type female from control cages (blue) and SSIMS treated cages (orange). Sample size of each group per week is illustrated above groups. Statistical tests of significance (Mann-Whitney test) are performed for each week comparing control and +SSIMS treated cages, n = 11–20 per group. (c–e) Total (c), female only (d), and male only (e) counts from the yeast/sugar trap. Light green shading indicates weekly additional SSIMS releases until week 8. Wild-type population in the control cages (blue lines) rise gradually and reach plateau by week 6. Wild-type population in the SSIMS-treated cages (orange line) decreases gradually until wild-type females became undetectable by week 11. SSIMS populations in the treatment cages (green line) rise during weekly release periods, but fall rapidly after week 8 when releases cease. Statistical significance: *=p < 0.01, **=p < 0.001, ***=p < 0.0001.

At the start of the experiment and for each of the first 8 weeks, we released 600 SSIMS flies to the two experimental cages (approximately 45:55 female:male ratio). The decision to stop at week 8 was arbitrary in this experiment. The released SSIMS flies had been reared on 100 μg/ml Tet, so we expected the female lethality phenotype from conspecific matings to be delayed by one generation (Figure 2d–f). On week 1, all four cages were seeded with 300 wild-type virgin females and 300 wild-type males. Weekly trapping, egg-laying, and fecundity assays were performed until week 16.

The SSIMS-treatment cages saw an immediate reduction of fecundity in wild-type females. Female fecundity did not follow a normal distribution, but rather was multimodal with a mode at 0 and another mode at about 50, likely representing females that had mated with SSIMS or wild-type males, respectively (Figure 4b). Several females gave low numbers of offspring, possibly due to re-mating with both wild-type and SSIMS males. By week 7, none of the captured wild-type females were fecund. The number of female fecundity assays that we could perform each week dropped as wild-type females could no longer be isolated from treatment cages.

Figure 4c–e show the results of the weekly trap counts. In the control cages (blue lines), counts of wild-type flies reached a steady-state of approximately 500 flies captured per week (Figure 4c). While cages were too large to allow for counting of all of the individuals, we estimate by visual inspection that the total cage population was approximately three times larger than the number of trapped/counted flies.

In the SSIMS treatment cages, the counts of wild-type flies dropped continuously throughout the experiment. While wild-type flies remained in each treatment cage by the time the SSIMS releases stopped in week eight, the high numbers of SSIMS males in the cages at that point prevented the wild-type population from recovering. By week 10, there were no wild-type females identified in the weekly traps.

The number of SSIMS flies, both males and females, remained high in the trap counts during weeks 1 to 8 when weekly releases occurred. This was expected as weekly releases of SSIMS adults occurred for the first 8 weeks. After SSIMS releases ended, counts of male and female SSIMS flies decreased, with female SSIMS flies decreasing rapidly to the point of disappearing as early as week 12. Male SSIMS flies disappeared from the cages a few weeks later. This is expected due to the FL construct allowing for one last generation of male-only offspring. By week 16, only a single male SSIMS fly was trapped across the two cages.

Agent-based model of SSIMS releases

To investigate how SSIMS compares to other genetic biocontrol technologies, we developed an agent-based model for Spotted Wing Drosophila (SWD), an agricultural pest that is the target of genetic biocontrol research and development projects (Asplen et al., 2015; Schetelig et al., 2021). Details of the model are described in Supplementary Note 6.

We simulated genetic biocontrol by RIDL, FL, and three different rearing/release scenarios for SSIMS. The first scenario is a male-only release, which could be achieved by mass rearing SSIMS agents on 10 μg/ml Tet and hatching the released agents on Tet-free food. The second scenario is a single-amplification release, which could be achieved by mass rearing SSIMS agents on 10 μg/ml Tet and hatching the released agents on 10 μg/ml Tet food. This would be a bi-sex release in which one additional generation of male-only SSIMS biocontrol agents could be produced in the field if the released SSIMS males and females mate with each other. The third scenario is a double-amplification release, which could be achieved by mass rearing SSIMS agents on 100 μg/ml Tet and hatching the released agents on 100 μg/ml Tet food. This would be a bi-sex release that could mate to produce another bi-sex generation, but the subsequent generation would be male-only (Figure 2d–f).

We simulated a single growing season (March 1 - September 1) using historical temperature data for St. Paul, MN (Figure 5). Untreated populations expand to tens of thousands by early July before diminishing due to unfavorable temperatures. Each biocontrol agent was released across a combinatorial regime of release numbers and release frequencies and the total WT SWD that were hatched across the duration of the season were compared (Figure 5b–e). We also report days to eradication for each release strategy in Figure 5f.

Figure 5.

Agent-based modeling.

(a) Schematic of model progression for an individual SWD agent. (b–d) Traces of total adult female populations over a growing season with different biocontrol agents. Shaded area represents one standard deviation among 10 simulations. Numbers inset to bottom left of plots show the insect release numbers and frequency (10–7 for (b), 40–7 for (c), and 80–50 for (d)) and correspond to grid cells identified with colored boxes in (e). (e) Heatmaps representing the cumulative total of WT adults over the course of the season at different release strategies. (f) Heatmaps representing the the average time to Wt eradication. (g) GM agent release was limited to the month of April. Traces of total WT and GM populations were tracked over the season from 10 simulations. (h–i) Total counts of WT and GM populations at June one and July one time points for the April limited release. Error bars show standard deviation of the mean from 10 replicates.

There are several notable results from this modeling. First, FL and SSIMS generally outperform RIDL by providing more population suppression for the same number of release agents. The closest comparison to RIDL is the male only release of SSIMS. Because sex separation is automatic for SSIMS, we simulated 100% of the released biocontrol agents as male for this technique and 50% for RIDL. If manual sex separation were used to release only male RIDL agents, the behavior should mirror male-only SSIMS releases.

FL outperformed double-amplification SSIMS (100 mg/ml Tet) and single-amplification SSIMS (10 mg/ml Tet) at the lowest release numbers. This is due to the persistence of the FL genotype in the simulated population, even many generations after release. However, at moderate release numbers SSIMS outperformed FL. The largest advantage of SSIMS over FL comes when analyzing the time to eradication (Figure 5f), where SSIMS achieves population suppression faster than FL.

The simulations in Figure 5b–f include regular release of biocontrol agents throughout the entire growing season. A more realistic release scenario is an aggressive, but time-limited release early in the season. We simulated a weekly release of 80 biocontrol agents for four successive weeks during the month of April (Figure 5g). Bar graphs comparing the numbers of genetically modified and wild-type agents at June one and July one are shown in Figure 5h and i, respectively. For each release we plot the number of genetically modified and wild-type SWD throughout the growing season. Both RIDL and male only SSIMS persist only a short period after the releases end, as expected. In these simulations, wild-type SWD never reach as high of population levels as the non-treatment controls due to the lag time in population expansion caused by the early season suppression.

FL, single-amplification SSIMS, and double-amplification SSIMS all persisted longer than RIDL and male-only SSIMS and acheived better population suppression. Of these, FL persisted the longest and was not able to suppress the wild-type population completetly before the mid-summer peak. Both single- and double-amplification SSIMS eradicated the wild-type population by the end of June. The double-amplification SSIMS release agents persisted for an additional two weeks after the wild-type were eradicated. For single-amplificaiton SSIMS, both the GM and wild-type populations were eradicated by the end of June. It is worth noting that for both SSIMS strategies the last generation of GM agents is exclusively male, which do not damage fruit. While we did not simulate migration, the additional persistence of SSIMS males could potentially help combat small introductions later in the season.

Discussion

Several new techniques for genetic biocontrol have been shown effective in proof-of-concept experiments (Alphey and Bonsall, 2014; Maselko et al., 2020; Buchman et al., 2021; Buchman et al., 2018; Akbari et al., 2014; Champer et al., 2020; Gantz and Bier, 2015; Terradas et al., 2021; Windbichler et al., 2011). Combination approaches that bring together two or more distinct techniques can exhibit synergistic behaviors that are attractive for genetic biocontrol applications. Here, we combine conditional female lethality with EGI to create the SSIMS strain.

The final genotype of the SSIMS flies generated in this study contains 23 synthetic genetic elements (operators, promoters, CDSs, terminators, etc.) spread across four chromosomal loci in the haploid genome. In total, 42,654 bases of synthetic DNA are integrated to the genome of engineered flies. This makes SSIMS one of the most complex engineered systems demonstrated in insects. This is exciting because it parallels the improvements in the complexity of engineered systems in other organisms and suggests that even more complex engineering will be possible with continued improvements of engineering methods. At the same time, the genetic complexity of SSIMS is a hurdle for its translation to pest organisms. Without the tools that we exploited in engineering SSIMS, such as multiple genetic markers and balancer chromosomes, engineering the SSIMS genotype in non-model insects will require more tedious molecular characterization of offspring produced during strain construction. However, we are encouraged by recent publications highlighting new genetic tools for non-model organisms (Kandul et al., 2021). Also, the molecular components needed to engineer SSIMS have been shown to work in diverse organisms.

Stacking more genetic components into a single organism can lead to fitness defects due to problems of resource allocation or cross-talk (Cardinale and Arkin, 2012; Segall-Shapiro et al., 2014). We observe a sex-ratio bias in favor of males when the SSIMS flies are reared with Tet, suggesting that the two X-linked female lethal constructs confer a slight fitness penalty in females even in the repressed state, presumably due to leaky or basal expression of tTa. This same bias was seen previously in the absence of the EGI constructs (Das et al., 2020). While slight sex-ratio biases like this might slow the expansion of a colony during mass rearing efforts, this is likely to be offset by the cost savings attained by automatic sex-separation or by the field amplification following bi-sex release. Based on previous observations with dosage effect of the female lethal construct, we predict that fine-tuning the expression strength of the tTA-regulated promoter will provide control over the dynamic range of the on/off state change. Here we favored strong penetrance at the expense of genetic burden, but this should be examined more carefully in species appropriate for field applications.

The competitiveness of biocontrol males with wild-type males is a critical factor in the efficacy and economy of SIT-like approaches. We show that the SSIMS males are competitive with wild-type males in laboratory mating experiments (Figure 3a), and that this extends to the level of sperm competition. This latter point is important for polyandrous insects. For example, Drosophila suzukii (Spotted Wing Drosophila, SWD) is closely related to D. melanogaster and is a global pest of fruit production. Several groups have observed female SWD re-mating in observational studies (Revadi et al., 2015; Krüger et al., 2019; Clark et al., 2020; Lanouette et al., 2020), and we have observed high levels of remating using a genetic assay (Supplementary Note 3). If a wild female mates with both a wild male and a SSIMS male, she will have a decrease in fecundity due to sperm displacement/competition. This would not occur for SIT-mimicking approaches that prevent sperm development in males. The laboratory competition assays performed here are promising but do not necessarily predict competition in the field. Suboptimal rearing conditions (Dyck et al., 2005), microbiome differences between wild and reared flies (Sharon et al., 2010), or behavioral differences like assortative mating (Reisen et al., 1982) could impact male mating competition. These factors should all be considered during the translation of an approach like SSIMS to field settings.

In its simplest application, SSIMS could be used to facilitate the sorting of male from female biocontrol agents prior to environmental release. In a male-only release, no biocontrol agents would persist beyond the released generation. Such an approach would be highly self-limiting, but would require larger release numbers to achieve suppression. Alternatively, males and females could be released together to provide limited field amplification. In a bi-sex release, if either sex of SSIMS insect mates with wild-type, no offspring will survive. If two SSIMS insects mate with each other, more biocontrol agents will be produced to potentiate the population suppression.

Interestingly, we can control the persistence of the SSIMS genotype to a degree by tuning the amount of Tet present in the food during the stock maintenance phase (Figure 2c–e). When the SSIMS line is reared from eggs to adults in 10 μg/ml Tet and then the adults are transferred to food without Tet, 100% female lethality occurs in the F1 generation. However, higher concentrations of Tet (100 μg/ml) in the stock maintenance phase delays the female lethality by a generation, (i.e. we observe 100% female lethality in the F2 generation). This feature of the SSIMS line could be useful for field applications, where simply varying the amount of Tet during the stock maintenance phase and/or perhaps spraying Tet in the field could oscillate the state of SSIMS between persistence (high Tet, both males and females emerge to sustain SSIMS population) and self-limiting (no/low Tet, females die after two generations and SSIMS population collapses). Tet is already approved for use as a foliar spray during fruit cultivation.

Another benefit to the SSIMS approach is redundancy. Both EGI and conditional female lethality are biocontrol techniques in their own respect. If one component were to mutate to become inactive in a released SSIMS strain, the other component would still prevent long-term persistence in most scenarios. However, this redundancy is partially reduced by the fact that the components are not linked on the chromosome. Also, high levels of genetic resistance to tTA female lethality has been described in wild Drosophila populations (Knudsen et al., 2020).

There are several shortcomings of this study that will need to be addressed to determine the suitability of the SSIMS approach to real-world applications. First, our cage trial was not large enough to reveal low frequency resistance mutations that would undermine the population suppression effect. Such resistance mutations could (i) inactivate or silence expression of the PTA, (ii) introduce SNPs at the target site to prevent PTA binding, (iii) suppress the negative impact of ectopically expressing pyramus or an equivalent developmental morphogen, or (iv) produce assortative mating phenotypes that prevent wild-type females from mating with released males. We discuss and demonstrate methods to mitigate some of these resistance mechanisms in previous publications (Maselko et al., 2017; Maselko et al., 2020). Recently, published work describes instances where such resistance mutations can arise reproducibly at very low frequencies (Zhao et al., 2020).

This study also lacks a techno-economic analysis or feasibility study to determine whether insects can be mass-reared to sufficient numbers to allow applications of SSIMS. Such studies will be more appropriate when SSIMS has been demonstrated in a pest organism that has potential applications in biocontrol.

In conclusion, we present a proof of concept of SSIMS in a model insect, D. melanogaster. SSIMS represents a new genetic biocontrol approach that provides a unique set of strengths and weaknesses compared to other demonstrated or proposed strategies (Alphey and Bonsall, 2014). In species where release of genetically engineered females is acceptable, SSIMS affords a tunable field-amplification, while still displaying low persistence. The male biocontrol agents are competitive with wild-type males for copulation and their sperm is competitive within the spermatheca. This makes SSIMS an especially attractive self-limiting genetic control approach for polyandrous pest organisms.

Materials and methods

Fly stocks and rearing conditions

D. melanogaster strains were maintained at 25 °C and 12 hr light/dark cycle and 60–70% relative humidity. All experiments were conducted in standard cornmeal/agar food Bloomington Formula (Genesee Scientific, 66–121), 0.05 M propionic acid (Sigma, 402907), 0.1% Tegosept (Genesee Scientific, 20–258). Tetracycline was added to 70 °C cool fly food (100 μg/ml, unless specified) before pouring into vials. White1118 strain was used as a wild-type. All crosses were performed with 2- to 5-day-old virgin females and males, unless specified.

Generation of SSIMS line

SSIMS is generated by combining the sex-sorting female-lethal strain with the Engineered Genetic Incompatibility strains published previously (Maselko et al., 2020; Das et al., 2020). We made a compound stock that contains two copies of the female-lethal construct on the X-chromosome, a refactored pyramus promoter on the 2nd-chromosome, and a programmable transcriptional activator targeting the wild-type pyramus promoter (dCas9-VPR)PTA+(pyr)gRNA expression cassette on the 3rd-chromosome. The combination of the refactored promoter and PTA + gRNA garners the EGI capacity of the SSIMS line. The cross strategy used to combine the two transgenes is illustrated in Appendix 1—figure 1.

Offspring lethality assays

Efficacy of female-lethality and EGI was measured by crossing 2- to 5-day-old SSIMS male and virgin females (SSIMS and wild-type) with or without Tet (10 μg/ml). Five virgin females and males were were allowed to mate in vials for 3 days, then removed to prevent overcrowding of larvae. Male and female offspring were counted 10 days post egg lay.

Male competition assays

For the 1:1:1 competition assays (Figure 3a), 2- to 5-day-old flies (one wild-type male, one SSIMS male, and one virgin wild-type female) were placed in a vial with fresh media for 8 hr to mate. Females were removed to a new vial for a 4-day egg lay period. Females were confirmed to have mated by looking for eggs that hatched to larva. The number of offspring that survived to adulthood was counted 14 days post egg lay. To assess SSIMS male competitiveness various combinations of 0–4 SSIMS and wild-type males were placed together with 20 wild-type virgin females, and allowed to mate for 24 hr. After the mating period, males were discarded and females were transferred into a new vial for 24 hr egg lay period. Total offspring were counted 11 days post egg lay.

Sperm displacement assay

To assess the ability of SSIMS sperm to displace wild-type sperm in utero (Figure 3c), three (5–6 day old) non-virgin wild-type females, which were previously mated with wild-type males, were placed in a vial either alone, with three wild-type male, or three SSIMS male. These males were left in the vials to remate with females for the remainder of the experiment. Females (with or without males) were allowed to lay eggs in 4-day periods then flipped into new vials for a total of 12 days. Total offspring were counted 11 days post egg lay for each period.

For the offense/defense sperm displacement assay (Figure 3d), virgin wild-type females were placed in a mating vial with males (wild-type or SSIMS) for 8 hr and then moved to individual tubes for 2 days. Females that showed evidence of mating (eggs hatching to larva) were then mated with the other type of male for 8 hr. Females were moved to a fresh vial to lay eggs for 2 days and then transferred to a final vial for an additional 7 days of egg laying. We report the number of surviving adult flies from the final 2 day and 7 day egg lay vials. The number of offspring that survived to adulthood was counted 14 days post egg lay.

To investigate polyandry and sperm competition in D. suzukii, two inbred SWD lines, with unique fixed SNPs in the 5’ promoter region of the wg gene, were crossed. Mating pairs of 3 non-virgin ’SWD-A’ females with 0 or 2 males of ’SWD-A’ or ’SWD-B’ were placed in vials. Genomic DNA of adult offspring were isolated as previously described (Das et al., 2020). PCR of the wg promoter region was done using primers wgF (5’-AGATTGCGCAAATAATCCGGC-3’) and wgR (5’-ATTCGAGCGGAGGAGTGAAG-3’), and Q5 polymerase following manufacturer recommendations(New England Biolabs). PCR products were Sanger sequenced by ACGT (Wheeling, IL), and the results were analyzed by Synthego’s ICE software (ice.synthego.com).

Cage trial

Cage trial experiments were designed to simulate multi-generational and generational-overlapping populations dynamics of wild-type Drosophila. On week 1, the cage trial was initiated with 600 (50% female) wild-type virgin males and females into each cage. A total of 600 SSIMS adults (40–50% female) were released weekly in the+ SSIMS treatment cage. Adult populations were housed in 0.027 m3 cages (Bioquip, 1452) and were fed by supplying 50 ml of 5% yeast-extract, 5% sucrose (YES), 0.1% Tegosept solution via a cotton wick in a 250 ml standard fly bottle and wrapped at the rim to prevent flies from getting stuck in the bottle. YES solution was replaced weekly. To allow the population to grow, 250 ml bottle (reproduction bottle) with 30 ml standard CMF were introduced once a week for a 24 hr egg lay period, after which the adults were removed. Reproduction bottle were uncapped 10 days after egg-lay and left open for 3 days to allow for the next generation to eclose and mix/mate with the preexisting population in the cage. To assess population size and genotype, an adult fly trap with yeast/sugar (YS) solution was placed in the cage once a week for 24 hr. The fly trap was baited with 30 ml of YS solution consisting of 0.5% active-yeast, 2.5% sucrose, 0.5% Micro90 (Cole Parmer). To measure fecundity of wild-type females, once a week, random adults from each cage were trapped in CMF bottle and wild-type females were isolated individually into 10 separate vials, and the remaining adults were released back in the cage. Wild-type females were allowed to lay eggs in vials for 3 days then released back in the cage. Total offsprings from each vial was quantified 11 days post egg-lay.

Data analysis, statistics, and visualization

All experiments were performed at least twice except the cage trial. Each batch of experiments contained at least five replicates, see caption for each Figure. Chi-squared test was performed to test difference between observed and expected sex ratio for Figure 2b and student’s t-test for Figure 2c. For Figure 3a/d, a Welch’s t-test was applied. For Figure 3b and a one-way ANOVA with Tukey post hoc test for multiple comparisons was applied. For Figure 3c and a Student’s t-test was applied. For Figure 4b, Mann-Whitney non parametric test was applied. Data was analyzed in Python using pandas, scipy, and statsmodels libraries. Statistical significance: *=p < 0.01, **=p < 0.001, ***=p < 0.0001. Graphs were generated using the Seaborn and Matplotlib libraries. Graphs were modified in Illustrator to fit publisher requirements.

This paper is of interest to entomologists caring for genetic pest control or molecular biologists following synthetic biology. The authors describe a fruit fly strain that combines constructs that establish both repressible female-lethality and genetic incompatibility based on CRISPR transactivation. They show that this strain has high penetrance for these two traits and that it can suppress wild-type flies when released into cycling cage populations.

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Genetically engineered insects with sex-selection and genetic incompatibility enable population suppression" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Patricia Wittkopp as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Ying Yan (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

Both reviewers as well as the reviewing editor agree that this paper provides an interesting proof-of-concept for a genetic control strategy potentially applicable to insect pests and disease vectors. The cage suppression experiment (with the caveats mentioned in the reviews) is a nice set of data but the steps towards it have some issues. This study should warrant publication in eLife if the following essential revisions can be addressed.

1) We would like to see a better assessment of both male competitiveness and female fecundity, as laid out in the recommendations below.

2) These parameters should then be plugged into the model the authors developed previously to get a sense of how the strategy would compare.

3) A more balanced discussion should be provided, especially regarding potential resistances that directly link to the developed constructs/strains.

Reviewer #1 (Recommendations for the authors):

The "mating competition" experiment featured an excess of females rather than an excess of males. It is thus more a "mating capability" experiment as there is less need for males to compete for females. Can the authors explain why this was done like that and how it connects to the literature? Given what this paper tries to achieve, I would have liked to see a true competition experiment.

I'm missing the female fecundity data of the SSIMS strain. It is mentioned that the "slight sex-ratio biases […] might slow the expansion of a colony during mass rearing efforts" but female fecundity is also a major factor here. It can possibly extracted from the data underlying Figures 2b and c but it is far from clear.

Figure 3b. How was it determined that the females were non-virgin? Is this based simply on the expectation that they should all have mated after 24 hours?

Reviewer #2 (Recommendations for the authors):

Line 23: change "laboratory" into "mass". The scale of field SIT program is way beyond laboratory level.

Line 26: Cochliomyia hominivorax was also eradicated from the Central American until 2006. Please find the proper reference for that and I think it's better to give a whole picture here.

Line 32: delete "(cite Lance 2000)".

Figure 1c: please add copy number to tetO in the image; the blue and red box are not explained in the legend, and I suppose that they are the first exon and M exon in the Chtra gene, respectively, which are not applicable here since the tTA instead of Chtra is used.

A few words about the alternative splicing are necessary (either in the legend or text) as this is the key to female lethality. Currently it's not straightforward in the figure if male or female-specific splicing is designed in the construct, therefore I would delete the male intron in the diagram which is not used.

Line 42: the cited paper (Scott et al., 2020) is mainly genomic analyses, and it's "Concha, 2020, BMC Genetics" that actually did the work "Integrating an early acting female lethality system to New World Screwworm" as described in the text.

Line 45-46: this sentence gave impression that pgSIT insects were actually released into environment, which is not the case. Please consider to rephrase it.

Line 62: none of the cited papers developed the "the tetracycline (Tet)-repressible

positive feedback circuit". Instead, the authors may want to cite "Fu, 2007, Nat. Biotechnol." which is the first demonstration in insects and "Li, 2014, Insect Biochem and Mol Biol" for the development in D. melanogaster.

Line 68-79: although this study is a follow-up of previous reports from the same group, it should be able to stand on its own. Therefore, some more explanation about the previously characterized strains or tools are needed for general readers. For example, what's the rational for constructing two FL constructs into X chromosome (since most transgenic sexing strains of different insect species used one FL construct)? What does pyramus gene do? And "foxo promoter… performed especially well" in terms of what?

Line 92-94: this sentence is confusing: based on the figure 2c I would guess it's SSIMA X WT compared to WT X WT, but in the text it firstly mentioned "SSIMS males" then again "when mated with SSIMS males", so why SSIMA X WT compared to SSIMA X WT? In addition, the statistics suggested the difference is significant (p=0.02) so "a slight increase" is not a proper description. Also, for figure 2c it would be better to switch total offspring (to the right position) and egg-lay (to the left position as the 1st bar), since the eggs were produced/counted first before the adult offspring.

Figure 2: (b) shows 10 ;;g/ml Tet was not enough to supress the female lethality, then the authors increased Tet concentration all the way to 100 ;;g/ml "to further optimize rearing of SSIMS stock". But no statistics or conclusion was provided to show if these concentrations made difference. In other words, statistics should be included in (d) and (e) or result section.

Figure 2 legend: "(e) Percent female offspring when respective progeny from (c) are reared in food without Tet", (c) produce either WT progeny or no progeny so it can't be right here. More likely the (c) should be (d) and similarly the (d) afterwards should be (e). To make it clearer, better add the y and x axis title to each figure and use "F0, F1,.." to indicate the exact generation for certain Tet treatment or number scoring.

Line 105-113: the male mating competitiveness assay is problematic as it used 20 virgin females and 4 males, resulting 5:1 sex ratio so the males are not really competing for females. In addition, the controls with 1,2,3 WT males were directly compared to the mix tests using the same number of WT male, but they are not really comparable since the sex ratio is different (so the chance of encounter and other factors that affecting mating will be different in a given time and space). A more meaningful comparison/statistical analysis should be among the mix-males groups (since they have the same sex ratio) which I don't see. Nevertheless, I wouldn't call this mating competitiveness assay at all when a single male paired with five virgin females. This should be repeat with a more rational design (such as WT females: WT males: SSIMA males at 1:1:1 ratio).

Line 114-127: the sperm competition is usually carried out in two directions (the offense and defence assays) and the duration of the second mating is usually limited to a few hours to avoid multiple mating (Clark, 1995, Genetics; Yeh, 2012, PNAS; Jayaswal, 2018, Evolution; Lüpold, 2020, Evolution letter). This study performed offense assay but not defence assay, which can show the ability of SSIMA sperm to resist being displaced by subsequent sperm. Additional defence assay would provide a better assessment about the competitiveness of SSIMA sperm. While this could be optional, what is mind-troubling is that the mated females were housed with SSIMA males for a total of 12 days! Such period allows the females to mate with same or different males several times. In addition, it is known that seminal-fluid proteins mediate proper sperm storage and fertilization (such as the sex peptide binds to sperm) and the related network proteins quick degrade post mating. Figure 3 shows 5-8 days old females (in the text it says 5-6 days) were used, but it didn't specify the duration of the first crossing and importantly when the WT females were separated from the WT males before the second crossing. If those females were sired by WT males multiple days before the competition test, the long-stored WT sperm may be naturally become less competitive than SSIMA sperm which are freshly produced with seminal-fluid proteins. Therefore, it's necessary to give such critical information. Anyway, the lack of defence test and the long duration of the second crossing make this a sperm displacement assay rather than sperm competition assay. The authors should either repeat it with a strict time control, or soften their statements (in my opinion no conclusion can be made for sperm competitiveness).

Line 128-137: the female remating of D. suzukii has been reported from different groups (Revadi, 2015, Insects; Krüger, 2018, J. Appl. Entomol; Clark, 2020, J Insect Behav; Lanouette, 2020, The Canadian Entomologist), and almost all of them showed the low remating rate of this species. The authors should discuss and compare the high remating rate here to the previous studies. Again, the exact experimental set ups such as duration of and lag between the first and second crossing could play huge role in such test.

Line 186-191: this whole paragraph "…..This makes SSIMS one of the most complex engineered systems in insects" is somewhat contradictory to claim "the genetic design is likely to be portable to other species for applications in pest control" afterwards. With "23 synthetic genetic elements (operators, promoters, CDSs, terminators, etc.) spread across four chromosomal loci in the haploid genome", such design would be challenge to be transferred to other Drosophila species, and unimaginable to be transferred to species if either the genome or site-specific integration is not available. In addition, the design with so many functional elements also make it more vulnerable to the spontaneous mutation (Zhao, 2020, Nat Commu).

Line 202 and other places: is the SSIMS really capable of complete penetrance? The Figure 2 and 3 shows the offspring count for a few hundred and Figure 4 counts up to a thousand. Such number can hardly be used to predict field or mass-rearing condition. It has been reported that F1 survivors of a Drosophila Tetracycline-controlled genetic lethal strain (so immediately after release) could at a one out of 10,000 frequency due to known mutation in construct and even unknown suppressors that inherited maternally (Zhao, 2020, Nat Commu), and current New World Screwworm SIT facility in Panama release 15 millions sterile flies per week. What is even more critical, is that the actual basis for tTA lethality (original tTA overexpression system that adopted in this study) is still unknown which is subject to suppression by a pre-existing inherent variation in the targeted field population (Knudsen, 2020, G3). The very phenomenon may also be true for any gene-overexpression-based lethality including EGI lines generated here. The authors should discuss such potential resistances that directly link to their constructs.

Line 218-223: male-only was compared to bi-sex SSIMS release as "require larger release numbers to achieve suppression" but no data or modeling to support this, so the comparison is not grounded nor necessary here.

Line 228-229: this is cryptic and should be re-written.

Line 266: this is more like "lethality assay" rather than "mating assay". Where is the Tet treatments between 10 – 100 ;;g/ml?

Essential revisions:

Both reviewers as well as the reviewing editor agree that this paper provides an interesting proof-of-concept for a genetic control strategy potentially applicable to insect pests and disease vectors. The cage suppression experiment (with the caveats mentioned in the reviews) is a nice set of data but the steps towards it have some issues. This study should warrant publication in eLife if the following essential revisions can be addressed.

1) We would like to see a better assessment of both male competitiveness and female fecundity, as laid out in the recommendations below.

2) These parameters should then be plugged into the model the authors developed previously to get a sense of how the strategy would compare.

3) A more balanced discussion should be provided, especially regarding potential resistances that directly link to the developed constructs/strains.

Since these summary-level revision requests are described in detail by individual reviewers, I refer the editor to those responses (below). We have rigorously responded to these suggestions in the revised manuscript.

We are grateful for the detailed criticism provided by both reviewers. We have carefully considered their critiques and have improved our study by following their suggestions.

Reviewer #1 (Recommendations for the authors):

The "mating competition" experiment featured an excess of females rather than an excess of males. It is thus more a "mating capability" experiment as there is less need for males to compete for females. Can the authors explain why this was done like that and how it connects to the literature? Given what this paper tries to achieve, I would have liked to see a true competition experiment.

We performed a mating competition experiment with an excess (i.e. non-limiting) number of adult females because this type of experiment was performed in the literature for a genetic biocontrol approach that is in the same application class as what we are reporting in this paper (Kandul, et al., 2019. https://doi.org/10.1038/s41467-018-07964-7). In that paper, male mating competitiveness is assayed at a 2 male : 10 female mating experiment, and experiments were performed with 2 wildtype, 1:1 wildtype to pgSIT, and 2 pgSIT males. In our original draft, we doubled the numbers of males and females (4 males: 20 females) to allow us to assess competition at 1:3, 2:2, and 3:1 mixed-male ratios.

We disagree with the reviewer that this is merely a ‘mating capability’ experiment, as mating capability can be assessed without males of each type. With that said, the excess number of females de-emphasizes a potentially important kinetic aspect to mating competition, wherein we might miss a competiveness phenotype that arises from GE males taking longer to initiate copulation with wild-type females. For this reason, we have performed a new mating competitiveness experiment (1:1:1) and include the data in a revised Figure 3a. This new experiment follows the literature precedent recommended by Reviewer #2. We continue to report the excess-female data as well, as this is still a good comparator to existing literature on SIT-like genetic biocontrol approaches, as mentioned above.

I’m missing the female fecundity data of the SSIMS strain. It is mentioned that the “slight sex-ratio biases […] might slow the expansion of a colony during mass rearing efforts” but female fecundity is also a major factor here. It can possibly extracted from the data underlying Figures 2b and c but it is far from clear.

This reviewer is correct that the fecundity data can be seen in Figures 2b and 2c (SSIMS male and female + Tet in Figure 2b, compared to wildtype male and female minus Tet in Figure 2c). We opted to report those data in separate subpanels since they are reporting the efficacy of separate genetic constructs (female lethality and EGI). For a more straight-forward comparison of fecundity between wild-type and the SSIMS strain, we added a supplementary figure (Supplementary Note 2) with a head-to-head comparison (new experiment). There was no statistically significant difference in fecundity.

Figure 3b. How was it determined that the females were non-virgin? Is this based simply on the expectation that they should all have mated after 24 hours?

We did not empirically test for prior mating before moving females to new tubes containing WT or SSIMS males in original Figure 3b (currently Figure 3c in the revised version). We performed a negative control experiment in which females from this same pool of putative non-virgins were moved to vials lacking males. The observation of reproduction in each of these control vials supports our assumption that the females had mated during the first 24 hours. However, this is different than direct empirical observation, so we note this weakness in our revised manuscript.

Also, we complemented this experiment with a new offense/defense sperm displacement experiment, in which the prior mating of females was confirmed empirically prior to moving to vials with new males (Figure 3d in revised manuscript). The new experiment mirror sperm displacement assays reported previously in the literature, as described in our response to Reviewer 2’s points below.

Reviewer #2 (Recommendations for the authors):

Line 23: change "laboratory" into "mass". The scale of field SIT program is way beyond laboratory level.

Fixed

Line 26: Cochliomyia hominivorax was also eradicated from the Central American until 2006. Please find the proper reference for that and I think it's better to give a whole picture here.

We have revised this sentence for accuracy. We have replaced the two citations with a separate comprehensive review from Scott MJ et al., from 2017.

Line 32: delete "(cite Lance 2000)".

Fixed.

Figure 1c: please add copy number to tetO in the image; the blue and red box are not explained in the legend, and I suppose that they are the first exon and M exon in the Chtra gene, respectively, which are not applicable here since the tTA instead of Chtra is used.

A few words about the alternative splicing are necessary (either in the legend or text) as this is the key to female lethality. Currently it's not straightforward in the figure if male or female-specific splicing is designed in the construct, therefore I would delete the male intron in the diagram which is not used.

We have added a more detailed description of the genetic design of both the female lethal and EGI cassettes in the figure, legend, and/or main text. We also describe the alternative splicing in the revised text. We have retained the male intron in the diagram, because previously published papers using this genetic design have shown that the mechanism of alternative splicing involves unique 5’ splice sites.

Line 42: the cited paper (Scott et al., 2020) is mainly genomic analyses, and it's "Concha, 2020, BMC Genetics" that actually did the work "Integrating an early acting female lethality system to New World Screwworm" as described in the text.

Thanks for catching this. We have revised with the appropriate citation.

Line 45-46: this sentence gave impression that pgSIT insects were actually released into environment, which is not the case. Please consider to rephrase it.

Fixed.

Line 62: none of the cited papers developed the "the tetracycline (Tet)-repressible

positive feedback circuit". Instead, the authors may want to cite "Fu, 2007, Nat. Biotechnol." which is the first demonstration in insects and "Li, 2014, Insect Biochem and Mol Biol" for the development in D. melanogaster.

Fixed.

Line 68-79: although this study is a follow-up of previous reports from the same group, it should be able to stand on its own. Therefore, some more explanation about the previously characterized strains or tools are needed for general readers. For example, what's the rational for constructing two FL constructs into X chromosome (since most transgenic sexing strains of different insect species used one FL construct)? What does pyramus gene do? And "foxo promoter… performed especially well" in terms of what?

We clarified the text in this section to indicate that the two-copy FL strain produced a stronger female-lethal phenotype in a previous study, and we have cited that study. We also state that pyramus is a developmental morphogen. We have clarified that the EGI design integrated into SSIMS was selected because it provided strong genetic incompatibility in a previous study. We also note that SSIMS is expected to work with other specific genetic designs for the FL and EGI components as well.

Line 92-94: this sentence is confusing: based on the figure 2c I would guess it's SSIMA X WT compared to WT X WT, but in the text it firstly mentioned "SSIMS males" then again "when mated with SSIMS males", so why SSIMA X WT compared to SSIMA X WT? In addition, the statistics suggested the difference is significant (p=0.02) so "a slight increase" is not a proper description. Also, for figure 2c it would be better to switch total offspring (to the righ position) and egg-lay (to the left position as the 1st bar), since the eggs were produced/counted first before the adult offspring.

We have rearranged the bars in the figure as requested and clarified the description of these results (i.e. that we are comparing WT(f) x WT (m) to WT(f) x SSIMS(m)). We did not remove the descriptor ‘slightly’, because we see this change as modest. We do not think that the statistical significance (p=.02) makes it improper to describe the change as slight.

Figure 2: (b) shows 10 ;;g/ml Tet was not enough to supress the female lethality, then the authors increased Tet concentration all the way to 100 ;;g/ml "to further optimize rearing of SSIMS stock". But no statistics or conclusion was provided to show if these concentrations made difference. In other words, statistics should be included in (d) and (e) or result section.

We did not intend to communicate that the concentration of Tet was optimized in propagate greater numbers of SSIMS flies, but that it could be optimized to control the persistence of the genotype in biocontrol scenarios. Specifically, rearing parents in 10 ug/ml leads to 100% female lethality (within our limit of detection) in the F1 generation. Rearing parents in 100 ug/ml leads to 35% female survival, which in a field release scenario would allow for a subsequent generation (F2) that would be male-only. Rearing the original parents at concentrations between 10 ug/ml and 100 ug/ml does not impact female survival in the F1 generation.

We performed students t-test to confirm that there is no statistically significant difference between tetracycline concentration and the percentage of female offspring in figure 2d. We fit data in 2e (concentration dependent female survival in F1 reared without Tet) to a rectangular hyperbola (i.e., Hill function), which explains 86% of the variance in the data (R-squared = 0.86). This corresponds to a t-value of 24.3 and a p-value < 0.00001.

Figure 2 legend: "(e) Percent female offspring when respective progeny from (c) are reared in food without Tet", (c) produce either WT progeny or no progeny so it can't be right here. More likely the (c) should be (d) and similarly the (d) afterwards should be (e). To make it clearer, better add the y and x axis title to each figure and use "F0, F1,.." to indicate the exact generation for certain Tet treatment or number scoring.

Fixed.

Line 105-113: the male mating competitiveness assay is problematic as it used 20 virgin females and 4 males, resulting 5:1 sex ratio so the males are not really competing for females. In addition, the controls with 1,2,3 WT males were directly compared to the mix tests using the same number of WT male, but they are not really comparable since the sex ratio is different (so the chance of encounter and other factors that affecting mating will be different in a given time and space). A more meaningful comparison/statistical analysis should be among the mix-males groups (since they have the same sex ratio) which I don't see. Nevertheless, I wouldn't call this mating competitiveness assay at all when a single male paired with five virgin females. This should be repeat with a more rational design (such as WT females: WT males: SSIMA males at 1:1:1 ratio).

As described in our response to Reviewer #1, the mating competition assay in which females are not limiting was performed based on literature precedent. We have elected to keep this dataset in the revised manuscript, as it allows comparison of the SSIMS system to previously published genetic biocontrol approaches (that used a similar mating assay). To address this reviewer’s concern, we added a conventional 1:1:1 male mating competition experiment and show that there is a statistically significant decrease in the amount of offspring compared to a control treatment (new Figure 3a).

Line 114-127: the sperm competition is usually carried out in two directions (the offense and defence assays) and the duration of the second mating is usually limited to a few hours to avoid multiple mating (Clark, 1995, Genetics; Yeh, 2012, PNAS; Jayaswal, 2018, Evolution; Lüpold, 2020, Evolution letter). This study performed offense assay but not defence assay, which can show the ability of SSIMA sperm to resist being displaced by subsequent sperm. Additional defence assay would provide a better assessment about the competitiveness of SSIMA sperm. While this could be optional, what is mind-troubling is that the mated females were housed with SSIMA males for a total of 12 days! Such period allows the females to mate with same or different males several times. In addition, it is known that seminal-fluid proteins mediate proper sperm storage and fertilization (such as the sex peptide binds to sperm) and the related network proteins quick degrade post mating. Figure 3 shows 5-8 days old females (in the text it says 5-6 days) were used, but it didn't specify the duration of the first crossing and importantly when the WT females were separated from the WT males before the second crossing. If those females were sired by WT males multiple days before the competition test, the long-stored WT sperm may be naturally become less competitive than SSIMA sperm which are freshly produced with seminal-fluid proteins. Therefore, it's necessary to give such critical information. Anyway, the lack of defence test and the long duration of the second crossing make this a sperm displacement assay rather than sperm competition assay. The authors should either repeat it with a strict time control, or soften their statements (in my opinion no conclusion can be made for sperm competitiveness).

We have changed the language in our paper to call this a sperm displacement assay. We have clarified the figure and text to state that the females were mated for 5-6 days before immediately transferring to a new vial. We have performed additional experiments that replicate as closely as possible the sperm displacement experiments described in the (Clark 1995) paper cited by Reviewer 2. We observed a statistically significant decrease in the number of offspring on both the offense and defense assays compared to wt-mating controls. In both cases, the number of viable offspring decreased to approximately 1/3 that of the control. There was no difference between the offense or defense assays. We have updated Figure 3 with these results. We elected to retain both the original and new experiment in the revised manuscript. Both experiments contain appropriate controls, sample sizes, and statistical analysis. We describe in the revised Results section the unique insights that both experiments provide.

Line 128-137: the female remating of D. suzukii has been reported from different groups (Revadi, 2015, Insects; Krüger, 2018, J. Appl. Entomol; Clark, 2020, J Insect Behav; Lanouette, 2020, The Canadian Entomologist), and almost all of them showed the low remating rate of this species. The authors should discuss and compare the high remating rate here to the previous studies. Again, the exact experimental set ups such as duration of and lag between the first and second crossing could play huge role in such test.

We are thankful for this reviewer alerting us to these relevant publications, as there are important lessons/comparisons to be made between those works and our manuscript. Revadi et al., (2015) observed instances of female remating but did not quantify it in that publication. Kruger et al., (2018) observed low rates of re-mating when fertile females were co-housed with fertile virgin males and observed for a three-hour period each day. The opportunity for re-mating was limited to this three hour period, which had previously been established as a preferred mating window. Clark et al., (2020) report a decline in female mating frequency post copulation, from 50% to 16%. This was also measured in a short window (1 hour long) on intermittent days after the observed copulation event. The objective of Lanouette et al., (2020) was to compare remating frequencies of females that mated with wild-type or sterilize males, and they report low remating frequencies between 5 and 10% for both groups. Like the previous studies, females that mated with the first male were kept in isolation, and then reintroduced to a single male for a short window of two hours for remating.

As Reviewer 2 mentions, the experimental set ups are different in important ways. The previous reports on female remating employed direct observation of females that were kept in isolation outside of relatively short windows of potential mating. Our experiment did not rely on direct observation, but on unambiguous genetic data to determine the fraction of offspring that result from a remating event(s). This allowed us to continuously co-house the previously mated females with new males. Because (i) we do not need to manipulate the female flies after the first mating event (e.g., move them into a petri dish for direct observation), (ii) we can measure remating events that occur 24 hours per day for several days, and (iii) our flies are kept in social communities throughout the duration of the experiment, we feel that our experimental set-up more closely represents the opportunities to remate that females would experience in nature.

We understand that this SWD experiment seemed like a non-sequitur given our focus on the model system in D. melanogaster for the rest of the paper. Our intention was to illustrate that putative targets of genetic biocontrol are polyandrous, which has implications on the outcomes of various control approaches. In other words, we were more interested in a binary assessment of whether SWD females remate, which all of the papers cited above demonstrate. We have revised our manuscript to make this point using the cited literature, and we move our assay to the Supplemental Information section to preserve the focus of the paper on our model system.

Line 186-191: this whole paragraph "…..This makes SSIMS one of the most complex engineered systems in insects" is somewhat contradictory to claim "the genetic design is likely to be portable to other species for applications in pest control" afterwards. With "23 synthetic genetic elements (operators, promoters, CDSs, terminators, etc.) spread across four chromosomal loci in the haploid genome", such design would be challenge to be transferred to other Drosophila species, and unimaginable to be transferred to species if either the genome or site-specific integration is not available. In addition, the design with so many functional elements also make it more vulnerable to the spontaneous mutation (Zhao, 2020, Nat Commu).

This is a good point and is something that we have been thinking about in depth. We will add to this paragraph the downsides to the complexity of this system. Discussion of vulnerability to spontaneous mutation is discussed alongside other resistance mechanisms later in the Discussion section.

Line 202 and other places: is the SSIMS really capable of complete penetrance? The Figure 2 and 3 shows the offspring count for a few hundred and Figure 4 counts up to a thousand. Such number can hardly be used to predict field or mass-rearing condition. It has been reported that F1 survivors of a Drosophila Tetracycline-controlled genetic lethal strain (so immediately after release) could at a one out of 10,000 frequency due to known mutation in construct and even unknown suppressors that inherited maternally (Zhao, 2020, Nat Commu), and current New World Screwworm SIT facility in Panama release 15 millions sterile flies per week. What is even more critical, is that the actual basis for tTA lethality (original tTA overexpression system that adopted in this study) is still unknown which is subject to suppression by a pre-existing inherent variation in the targeted field population (Knudsen, 2020, G3). The very phenomenon may also be true for any gene-overexpression-based lethality including EGI lines generated here. The authors should discuss such potential resistances that directly link to their constructs.

We have revised the text to state that we favored ‘strong penetrance’ instead of ‘complete penetrance’. The sentence in line 202 was only meant to highlight or prioritization of penetrance over genetic burden. We address the limits of our study (including the inability to observe low frequency mutations) in new paragraphs added near the end of the discussion.

Line 218-223: male-only was compared to bi-sex SSIMS release as "require larger release numbers to achieve suppression" but no data or modeling to support this, so the comparison is not grounded nor necessary here.

We have added agent-based modeling to the paper to support this claim.

Line 228-229: this is cryptic and should be re-written.

We have re-written this paragraph to improve the clarity.

Line 266: this is more like "lethality assay" rather than "mating assay". Where is the Tet treatments between 10 – 100 ;;g/ml?

We have changed to “Offspring lethality assay”.

Appendix 1—table 1.

Key parameters in SWD model.

Parameter	Value	Reference
Maximum Egg/Female	150	Tochen
Egg/Larval Degree Days - mean	140.785	Asplen
Egg/Larval Degree Days - SD	20.49	Asplen
Pupal Degree Days - mean	93.22	Asplen
Pupal Degree Days - SD	6.18	Asplen
Adult Degree Days - mean	1,050	Asplen
Adult Degree Days - SD	40.41	Asplen

Appendix 1—table 2.

Key to genotype nomenclature in the model.

Allele	Definition
b	wildtype promoter for EGI target locus
B	resistant promoter for EGI target locus
d	wildtype allele at location of PTA integration
D	PTA targetting wildtype promoter b
p	wildtype promoter for EGI target locus
P	resistant promoter for EGI target locus
t	wildtype promoter for EGI target locus
T	PTA targetting wildtype promoter p
l	wildtype locus at the site of Female Lethal transgene integration
L	Female Lethality transgenic cassette
W	wildtype allele at locus of Gene Drive integration
G	Homing gene-drive allele
R	Gene Drive resistance allele with functional target locus
I	Gene Drive resistance allele that inactivates target locus
X	female sex chromosome
Y	male sex chromosome
q	promoter conversion of p (which behaves like P but is tracked independently)
r	natural resistance mutation in p that provides protection to T
z	promoter conversion of b (which behaves like B but is tracked independently)
c	natural resistance mutation in b that provides protection to D

Appendix 1—table 3.

Genotypes of wildtype or biocontrol agents released in simulations described in this study.

Genotype in model	Definition
bbddppttllWWXX	wildtype female
bbddppttllWWXY	wildtype male
bbddPPTTllWWXX	EGI female
bbddPPTTllWWXY	EGI male
bbddppttLLWWXX	female-lethal female
bbddppttLLWWXY	female-lethal male
bbddppttLLWWXX*	RIDL female
bbddppttLLWWXY*	RIDL male
bbddPPTTLLWWXX	SSIMS female
bbddPPTTLLWWXY	SSIMS male

Removed "and self.sex==’female’" from line 236.