Mosquito species of the Anopheles gambiae complex are the main vectors of human malaria and pose an enormous burden on global health and economies (World malaria report, 2020). The progressive spread of insecticide resistant mosquitoes (Hyde, 2005; Sinha et al., 2014) has prompted the development of new methods to control these mosquitoes (Windbichler et al., 2008; Kyrou et al., 2018; Carballar-Lejarazú et al., 2020). One of the most promising is genetic control, which is based on the release of laboratory-modified insects into the environment. Released individuals mate with wild insects and transmit control traits that can suppress or modify the targeted population (Hamilton, 1967; Curtis, 1968). Among these, the most commonly used approach to genetically control insects has been the mass release of sterile males—the so-called Sterile Insect Technique (SIT) (Knipling, 1959; Wyss, 2000). When wild monandrous females mate with released sterile males, their eggs are fertilized by sperm carrying mutations that abort embryo development. If sufficient numbers of sterile males are released over a long enough period, the wild population can be effectively suppressed or even eradicated. However, the economic costs of an SIT program that aims for mosquito suppression in very large areas and the need to maintain sterile releases indefinitely, have restricted the implementation of this method to date [but see; (Zheng et al., 2019, Hendrichs et al., 2021; Balatsos et al., 2021)].

One way to improve the efficiency of such approaches is through the release of fertile males that are daughterless. Since male mosquitoes do not contribute to disease transmission, releasing males that have viable and fertile sons can help to temporarily maintain the frequency of the allele or transgene in the population, which in turn helps to reduce the abundance of females. Two strategies based on such fertile males have been developed in mosquitoes thus far: fs-RIDL (for female-specific Release of Insects carrying Dominant Lethals) and sex ratio distorters based on X-chromosome shredding (Thomas et al., 2000; Burt, 2003; Phuc et al., 2007; Windbichler et al., 2007; Galizi et al., 2014). fs-RIDL is based on a construct that is lethal to females that inherit it, so that daughters of released transgenic males born in the field and inheriting the transgene die before maturing or are unable to fly (flightless), but sons survive and pass the transgene to their offspring (Thomas et al., 2000; Phuc et al., 2007). Sex-ratio distortion based on X-chromosome shredding instead, relies on the expression of a sequence-specific endonuclease during male spermatogenesis that recognizes and cleaves sequences that are both specific and abundant on the X-chromosome (Windbichler et al., 2007; Galizi et al., 2014). As a result, X-chromosome-bearing gametes are eliminated from the viable sperm population, thus biasing offspring sex-ratios towards males (Burt, 2003; Deredec et al., 2008; Papathanos and Windbichler, 2018; Haghighat-Khah et al., 2020). Mathematical models predict that both approaches are more efficient than SIT in terms of the number of modified males that need to be released to achieve a similar level of population suppression (Schliekelman et al., 2005; Burt and Deredec, 2018). Despite being more efficient, both fs-RIDL and autosomal X-shredders (where the transgene is located on an autosome) are self-limiting. The transgenic constructs underlying the phenotype will therefore not spread in the population, because they are inherited in a Mendelian fashion and do not provide any fitness advantage over the wild type. This is different for self-sustaining approaches such as those incorporating gene drive constructs (Alphey, 2014; Hammond and Galizi, 2017). The fact that in X-shredding, the X-chromosome-bearing gametes are eliminated pre-zygotically can be used for self-sustaining genetic control applications, in the form of Y-chromosome drive as originally proposed by Hamilton (Hamilton, 1967). This could be done by linking a functional X-shredder to the Y-chromosome, in which case both the Y-chromosome and the X-shredder gain a transmission advantage through preferential inheritance of male-forming gametes (Deredec et al., 2011).

A X-shredding sex-distorter was first developed in An. gambiae by Galizi et al. (2014). They used variants of the I-PpoI endonuclease that cut a specific DNA target sequence within the 28S ribosomal DNA locus, which in An. gambiae is located exclusively on the X chromosome in approximately 200–400 copies (Collins et al., 1989). These I-PpoI variants were fused to eGFP and driven by the An. gambiae beta-2 tubulin regulatory regions, which become active in primary spermatocytes entering male meiosis (Catteruccia et al., 2005). The resulting transformation constructs also included the DsRed transformation marker driven by the neuron-specific 3xP3 promoter, and the entire cassette was flanked by piggyBac-specific left and right arms containing the inverted terminal repeat sequences (ITRs) (Figure 1A). Of all the transgenic strains examined, ^gfp124L-2, since renamed by the Target Malaria Research consortium as Ag(PMB)1 (for An. gambiae Paternal Male Bias strain 1) expressing the I-PpoI structural variant W124L, showed high sex ratio distortion among progeny of transgenic males (approximately 95% males), without significantly impairing male fertility and fitness and is thus being currently evaluated for field testing by the Consortium (Galizi et al., 2014). Inverse PCRs produced as part of that study showed an autosomal location of the transgene, from where the sex-distortion phenotype was stably inherited over consecutive generations. In large cage experiments, weekly inoculative releases of transgenic Ag(PMB)1 males led to a reduction both in the egg productivity of the population and the frequency of females over successive generations consistent with model predictions (Facchinelli et al., 2019).

Since Ag(PMB)1 males are fertile, their release should result in viable offspring in the field, unlike sterile males. This would provide invaluable information about how transgenic, laboratory-reared males of An. gambiae disperse spatially once released and can be achieved even from small-scale releases aimed at capacity-building and methodology development. Unlike fs-RIDL strains that were directly developed for deployment, the lack of a conditional expression system or some other design to control activity of the X-shredder makes this strain unsuitable for large-scale programs aiming directly for population suppression, since rearing at large numbers is logistically difficult because this strain can only be maintained through females to avoid strain loss. Moreover, the Ag(PMB)1 X-shredder was not designed for gene drive in its current form - for example it is not able to home into targeted sequences. It also does not display any fitness advantage over wild type mosquitoes. Consistent with this, models predict that the Ag(PMB)1 transgene would disappear over time when releases are discontinued (Burt and Deredec, 2018) and recent large cage experiments confirmed the loss of the transgene from the population over time (Pollegioni et al., 2020).

One feature that is unique to an autosomal X-shredder compared to other self-limiting strategies, is the possibility that it could move to the Y-chromosome after release, potentially becoming self-sustaining in the form of a Y-chromosome drive as first coined by Hamilton (1967). If active, a Y-chromosome linked X-shredder could directly benefit from the increased transmission of the Y-chromosome due to preferential inheritance of male-forming gametes, thus increasing in frequency, persisting longer and dispersing further than initially planned. The sequential events required for such a driving Y to occur can be mapped to “pathways to harm” using a problem formulation approach adopted widely in environmental risk assessments (Supplementary Figures S1, S2). Three requirements must be fulfilled for a driving Y to occur: 1) the autosomal X-shredder must first move from its original autosomal position and become physically linked to the Y-chromosome; 2) the X-shredder would need to be expressed from its new position on the Y-chromosome during late spermatogenesis in a spatiotemporal manner that is similar to its original expression from the autosome; and 3) it should impart no significant cost to male fertility or male viability as a result of its new Y-chromosome-linkage (Supplementary Figures S1, S2).

With regard to requirement 1), excluding any DNA repair mechanisms as this would require unlikely pairing of the autosome and the Y chromosome, there are two possible mechanisms that could result in an autosomal transgene moving to the Y-chromosome: 1) a transposase-mediated transposition to the Y-chromosome of the piggyBac (PB) transposable element that was used to create the transgenic strain, or 2) a recombination-mediated reciprocal translocation resulting in large chromosomal rearrangements between the autosome and the Y-chromosome. Of the two mechanisms, translocation is the less likely route, because translocations between autosomal segments and the Y-chromosome occur very rarely in nature (see Discussion). On the other hand, transposition from the autosome to the Y through the remobilization of the PB transposon could be possible, if the X-shredder transgene co-occurs in a genome containing an active PB transposase.

In this paper, we addressed the possibility of transgene remobilization by examining mosquito genomes for evidence of the PB transposase which recognize the inverted terminal repeats of the transgene and remobilize it. With regard to requirement 2), expression from the Y–we generated two independent transgenic strains containing the eGFP:I-PpoI-124L X-shredder construct on the Y-chromosome and evaluated the level of expression and sex-ratio distortion. Finally, we discuss implications for male fertility depending on the route of movement to the Y-chromosome.

Genetic control strategies that aim to suppress wild populations of mosquito disease vectors have garnered significant interest, and field trials of a number of these systems, including classical SIT, IIT (Wolbachia-based sterility), and RIDL are now underway (Harris et al., 2012; Zheng et al., 2019). Synthetic sex-ratio distorters based on X-chromosome shredding have now been developed in An. gambiae (Galizi et al., 2014; Galizi et al., 2016) and more recently in Drosophila melanogaster and Ceratitis capitata (Fasulo et al., 2020; Meccariello et al., 2021). This system has been shown both theoretically and experimentally to be more efficient than classical SIT, in terms of the number of insects that need to be released (Schliekelman et al., 2005; Galizi et al., 2014; Burt and Deredec, 2018). In their most basic form, autosomal X-shredder constructs are self-limiting, and their release can potentially result in local and limited suppression if sufficient males are released over a long enough period. Nonetheless, field releases of fertile autosomal X-shredder males have not yet been conducted.

Here, we have evaluated the theoretical possibility whereby an autosomal X-shredder could convert into a self-sustaining, driving Y-chromosome after release. The requirements for such an event to occur include: 1) movement of the X-shredder to the Y-chromosome; 2) its subsequent expression from the Y during late stages of spermatogenesis at a level that result in X-chromosome shredding; and, 3) have a low enough fitness cost to male carriers such that the X-shredding effect is net beneficial to males carrying such a Y-chromosome. We reason that there are two possible mechanisms that could result in the linkage of the autosomal Ag(PMB)1 to the Y-chromosome: 1) a transposase-mediated remobilization of the transgene and 2) a large chromosomal rearrangement resulting in the reciprocal translocation between the region of the autosome containing the transgene and the Y-chromosome.

We have evaluated the potential of remobilization of the Ag(PMB)1 transgenic construct through transposition, mediated by the intact PB inverted terminal repeats (ITRs) on either side of the transgene cassettes, which were used for the initial generation of the strain. Their presence makes it at least theoretically possible that the Ag(PMB)1 transgene could remobilize from its autosomal position, if a source of the PB transposase occurs in trans. We therefore evaluated whether the PB transposase is present in the genomes of the laboratory Ag(PMB)1 and G3 colonies, and also in field-derived samples from Burkina Faso. We found no evidence of the complete PB transposase coding sequence in any of the samples we sequenced, suggesting that PB is not present at appreciable frequencies in An. gambiae mosquitoes sampled from nature. This result is supported by the long-term stability of the Ag(PMB)1 insertion site over 100 generations since its original construction, a result that suggests that other naturally occurring repetitive elements in the genome of the Ag(PMB)1 strain, including those that appear as PB-like by genome-wide translated nucleotide searches, are not functionally capable of PB transgene remobilization. Long-term stability of transgenic insertions in An. gambiae laboratory strains in the absence of experimentally provided PB transposase is well known, including through directed efforts of enhancer trapping through transgene mobilization (O'Brochta et al., 2011). The computational methods we developed to screen for the presence of PB transposase in genome sequencing data from wild type, transgenic and field samples could be adapted in the future for high-throughput screening of sequenced samples collected from the field to identify and quantify the presence of transgenic alleles without the need for fluorescence microscopy or complicated molecular genotyping protocols.

We were not able to test the second possibility of remobilization by chromosomal translocation, as these occur very rarely during meiosis. In over 7 years of both standard laboratory rearing (Galizi et al., 2014) and large-scale multi-generational cage studies of the Ag(PMB)1 strain (Facchinelli et al., 2019; Pollegioni et al., 2020), translocations involving the autosomal transgene and the Y-chromosome have never been detected–an event that would be noticeable in defined crosses of individuals since fluorescent transgenic individuals would only be male. This extends to scaled experimental conditions, in which large numbers of Ag(PMB)1 individuals were screened by high-throughput sorting of individuals using the COPAS sorter based on the 3xP3-DsRed marker and then subsequently separated by sex at the pupal stage (Burt and Deredec, 2018). This is, in part, expected because of how rarely such events occur, meaning that experimentally verified rates for such events are not readily available for the Ag(PMB)1 strain. When translocations between autosomes and the Y are desired, they can be artificially induced in the laboratory, for example with ionizing radiation, chemical agents, or UV radiation. This is commonly done for insects to link selectable phenotypes to the Y-chromosome, in so-called genetic sexing strains (GSSs). GSSs are developed so that males can be separated from females on a large scale in insect bio-factories that produce animals for genetic control programs, such as SIT (Gilles et al., 2014). This is done by linking selectable traits, for example insect color or high temperature tolerance, to the Y-chromosome using induced reciprocal translocations of mutant alleles located on autosomes. Once generated, these GSSs are then maintained in large numbers, with billions of insects being produced weekly. The large colony size makes it possible to detect rare events that result in the breakdown of linkage between maleness and selectable trait. One of two ways this can happen is through a “reverse” reciprocal translocation involving the previously modified Y-chromosome and an autosome (known as a type 2 recombination event) (Franz et al., 2005). Because such events lead to a breakdown of the genetic sexing system and restore male fertility of the semi-sterile males (arising from the translocation itself) and leading to their accumulation, their occurrence is tightly monitored in large-scale rearing operations. In the only report that quantified the rate of type-2 recombination and distinguished it from type-1 (which does not involve the translocated Y- chromosome and is more common among the two) in a GSS of the Mediterranean fruitfly Ceratitis capitata, the rate was estimated to be 10⁻⁵ or less, i.e., occurring in less than 1 out of 100,000 male individuals (Franz et al., 2005). In this case however, there are two significant factors that would indicate that the rate of an initial, uninduced autosome-Y translocation would be much lower. First, the rate of recombination in the GSS describes an event reversing a previously induced autosome-Y translocation, which is likely to be largely mediated by homologous sequences that are now present on the two translocated Y fragments; a homology that does not normally exist between autosomes and the Y chromosome. Therefore, the expected recombination rate resulting in a reciprocal translocation between an autosome and Y would be lower. Second, the reversion of the previously translocated autosomal fragment on the Y restores male fertility, that was first compromised by the translocation to the Y (because of gamete chromosomal imbalance–discussed below) (Franz et al., 2005). This means that type-2 recombinant males will have more viable offspring increasing their rate of occurrence in the population. Together, these factors suggests that the probability of a translocation event involving the autosomal Ag(PMB)1 transgene and the Y-chromosome in progeny of released males born in the field is expected to be much lower than 10⁻⁵. This rate would depend on the size of the Y-chromosome and the relative rate of recombination in the male germline, which in An. gambiae is approximately 1.6 cM Mb⁻¹ for autosomes (Pombi et al., 2006) and similar between males and females (Benedict et al., 2003).

In the unlikely event that the transgene was to move to the Y-chromosome, we provide data regarding the expression of the X-shredder from this chromosome, and conclude that MSCI during spermatogenesis does affect the Y-chromosome of An. gambiae. Our results from two transgenic strains harboring the Ag(PMB)1 X-shredder transgene in two different positions on the Y-chromosome, reveal transcriptional suppression during late spermatogenesis from the beta2-tubulin promoter, complementing our previous work which confirmed this for the An. gambiae X-chromosome (Magnusson et al., 2011; Galizi et al., 2014; Papa et al., 2017). We found no evidence of X-shredder expression by quantitative RT-PCR, nor by fluorescence microscopy of transgenic testis. Offspring of transgenic males from both Y-linked strains therefore had sex-ratios similar to wild-type males. Hence, even if the Ag(PMB)1 transgene successfully moved to the Y-chromosome by transposition (first requirement for a driving Y) it is unlikely that the X-shedder would be active. Since MSCI-factors regulating transcriptional suppression physically spread across the sex chromosomes after becoming localized on their unsynapsed axes (Ichijima et al., 2011; Ichijima et al., 2012), it is also expected that translocated autosomal fragments would become suppressed by MSCI during meiotic stages of spermatogenesis. Therefore, the weight of evidence argues strongly against the likelihood of movement of the Ag(PMB)1 transgene to the Y chromosome, particularly via transposition. However, the equally necessary prerequisite for a pathway to a driving Y, namely expression of the X-shredder on the Y chromosome during male meiosis, seems highly implausible based on the evidence presented here.

The final requirement for a Y-linked X-shredder to spread through populations is that its movement to the Y-chromosome and subsequent expression from it would have no significant fitness costs to males harboring it. Such fitness costs would counteract the theoretical advantage gained by the Y-linked X-shredder from increased transmission through elimination of X-bearing sperm. Among the factors determining these fitness costs, the largest contributors would likely be the mechanism leading to Y-linkage and the outcomes of this movement on each chromosome. Reciprocal translocations between an autosome and the Y-chromosome have be found to result in significant male fertility costs (Roukos and Misteli, 2014). Because of the simultaneous segregation of non-homologous centromeres (adjacent-1 segregation) during meiosis, only 50% of the offspring produced by males are genetically balanced, i.e., males are 50% sterile (Yamada et al., 2012). In certain cases, this semi-sterility can be even higher, for example in an An. arabiensis GSS showing 73.3% male sterility (Yamada et al., 2012). Therefore, a Y-linked X-shredder that arose through a translocation event would likely display sufficiently high male fertility costs that it would rapidly disappear from the population. For transposition-mediated Y-linkage male fitness costs cannot be predicted a priori, as gamete balance and genic content would depend on both the excision event (i.e., how much of the surrounding chromosome is excised) and on the integration position on the Y-chromosome (i.e., subsequent knock-out of genes essential for male fitness such as the male-determining gene).

In summary, the findings of the current study support the low probability of transgene remobilization from the autosome to the Y-chromosome. Moreover, even if such a rare event occurred, where the X-shredder would become linked to Y-chromosome, activity of the X-shredder at the required stage of spermatogenesis would likely be impeded via chromosome wide suppression of gene expression on meiotic sex chromosomes. Our results also show that prospects for the successful building of self-sustaining Y-linked X-shredders for mosquito control in the future will need to find ways to circumvent this transcriptional suppression, for example using alternative germline specific promoters (Taxiarchi et al., 2019). Finally, more studies and methods are needed to systematically explore how population dynamics of released elements could be impacted by spontaneous genomic changes, such as transgene remobilization, done in a way that is technology-specific and relevant.