The haplolethal gene wupA of Drosophila exhibits potential as a target for an X-poisoning gene drive

Abstract

A synthetic gene drive that targets haplolethal genes on the X chromosome can skew the sex ratio toward males. Like an “X-shredder,” it does not involve “homing,” and that has advantages including the reduction of gene drive resistance allele formation. We examine this “X-poisoning” strategy by targeting 4 of the 11 known X-linked haplolethal/haplosterile genes of Drosophila melanogaster with CRISPR/Cas9. We find that targeting the wupA gene during spermatogenesis skews the sex ratio so fewer than 14% of progeny are daughters. That is unless we cross the mutagenic males to X^{^}XY female flies that bear attached-X chromosomes, which reverses the inheritance of the poisoned X chromosome so that sons inherit it from their father, in which case only 2% of the progeny are sons. These sex ratio biases suggest that most of the CRISPR/Cas9 mutants we induced in the wupA gene are haplolethal but some are recessive lethal. The males generating wupA mutants do not suffer from reduced fertility; rather, the haplolethal mutants arrest development in the late stages of embryogenesis well after fertilized eggs have been laid. This provides a distinct advantage over genetic manipulation strategies involving sterility which can be countered by the remating of females. We also find that wupA mutants that destroy the nuclear localization signal of shorter isoforms are not haplolethal as long as the open reading frame remains intact. Like D. melanogaster, wupA orthologs of Drosophila suzukii and Anopheles mosquitos are found on X chromosomes making wupA a viable X-poisoning target in multiple species.

Introduction

For many decades, it has been envisioned that insect populations that vector diseases or that are pests of crops could be controlled using genetic methods (Knipling 1955; Curtis 1968; Serebrovsky 1969; Whitten 1985; Burt 2003). Indeed, sterile insect techniques have successfully controlled fly species (Bushland et al. 1955; Hendrichs et al. 1995; Scott et al. 2017) and field trials relying on self-limiting transgenes are currently being deployed and assessed (Waltz 2021; Yao et al. 2022). Recently, the prospect of synthetic gene drives a distinct class of population manipulation, where transgenes can be engineered to spread through otherwise natural populations of insect pests has received much attention (Champer et al. 2016; Alphey et al. 2020; Hay et al. 2021). The effectiveness of various gene drive designs, many of which use CRISPR/Cas9 machinery, has been assessed in laboratory populations (Chen et al. 2007; Windbichler et al. 2007; Akbari et al. 2014; Kyrou et al. 2018; Oberhofer et al. 2019; Webster et al. 2020; Yang et al. 2022). Many of these require “homing” where double-stranded DNA breaks must be repaired by homologous recombination (HR) so that the transgenic selfish elements are copied and thereby spread. A drawback of such homing designs has been that some double-stranded breaks can be repaired using the nonhomologous end-joining (NHEJ) pathway. In these instances, not only does homing fail but mutations that are resistant to subsequent CRISPR/Cas9 cleavages can be introduced into the population (Windbichler et al. 2011; Gratz et al. 2014; Champer et al. 2017, 2018; Hammond et al. 2017).

One genetic control design that does not require homing is referred to as the X-shredder (Galizi et al. 2014). It targets highly repeated sequences on the X chromosome during spermatogenesis such that X-bearing sperm are inviable, and the drive skews the sex ratio toward males. If the X-shredding transgenes could be placed on the Y chromosome, then an inheritance can be established such that fathers only have sons. Originally, the X-shredder was attempted in lab populations of Anopheles gambiae mosquito using the endonuclease PpoI, but subsequent implementations in A. gambiae mosquitos, Drosophila melanogaster, and Ceratitis capitata fruit flies have deployed CRISPR/Cas9 machinery (Galizi et al. 2014; Fasulo et al. 2020; Meccariello et al. 2021; Haber et al. 2023).

A related approach, dubbed “X-poisoning,” also targets loci on the X chromosome, not so that the chromosome is physically destroyed but so that mutations are introduced such that progeny receiving the X chromosome will not be viable (Burt and Deredec 2018; Fasulo et al. 2020; Haber et al. 2023). This strategy uses CRISPR/Cas9 to target X-linked haplolethal genes. Although haplolethal genes are typically thought of as genes where both alleles of a diploid organism need to be functional, X-linked haplolethal genes can occur in many dipterans because their dosage compensation mechanism elevates the X chromosome-encoded output in males (that is in contrast to eutherian mammals where there is silencing of X loci in females, and therefore, haplolethals on the X would be incompatible with maleness) (Rose et al. 2016; Cook et al. 2012). There may be a few potential advantages of X-poisoning over X-shredding. Firstly, X-poisoning does not rely on natural repeats occurring exclusively on the X chromosome and theoretically this may increase its applicability to some species. Secondly, CRISPR cutting in 1 or a few sites on the X chromosome might occur more readily than cutting the X chromosome at an overwhelming number of places, as is required for X-shredding. Thirdly, X-shredding has been shown to be ameliorated by the NHEJ pathway, thus potentially limiting its utility (Fasulo et al. 2020). Fourthly, it may be that postzygotic effects of X-poisoning may change the spread dynamics so it has a higher threshold than X-shredding and that may lead to more controllable population manipulation.

Here, we describe the targeting of haplolethal genes on the X chromosome of D. melanogaster with CRISPR/Cas9. The loci differ from the 2 targeted in the X-poisoning study of Fasulo et al. (2020) as we examine 3 other ribosomal protein encoding genes and the haplolethal gene Wings-up A (wupA). In contrast to the ribosomal protein genes which might be required for the viability of most cell types including those in spermatogenesis, wupA encodes a troponin that is expressed in muscles from mid embryogenesis. We therefore sought experimental affirmation that these genes are indeed haplolethal genes and whether they could be valuable targets for X-poisoning gene drives.

Materials and methods

gRNA design and cloning

A combination of ChopChop v3 (Labun et al. 2019) and CRISPR Optimal Target Finder (Gratz et al. 2014) was used to identify the gRNA target sites with the highest efficiency scores with a low likelihood of off-target activity. All gRNA sequences were checked against the Drosophila Genetic Reference Panel (DGRP) using FlyBase's JBrowse (Larkin et al. 2021; Gramates et al. 2022) to avoid standing variation at the target site that may give natural resistance.

The design protocol of Port and Bullock (2016) was used to clone gRNA sequences into the pCFD5_w plasmid (Addgene #112645). Three separate plasmids were generated that targeted haplolethal genes on the X chromosome; pRpS19a_RpL35, pRpL36_RpL35, and pwupA. pRpS19a_RpL35 encoded 4 gRNAs, 2 targeting RpS19a (RpS19_RNA1: GAAGGATATTGACCAGCACG, RpS19_RNA2: AACCGACTCCAGCGGGACTG) and 2 targeting RpL35 (RpL35_gRNA1: GTGCTCCGAGCTGAGGATCA, RpL35_gRNA2: CACAATGTAGACGCGAGCGA). pRpL36_RpL35 encodes 4 gRNA, 2 targeting RpL36 (RpL36_gRNA1: GCTGGCTATTGGCCTGAACA, RpL36_gRNA2: CATGCGCGACTTGGTCCGCG) and 2 targeting RpL35 (as aforementioned). Each of the 3 sgRNAs used to target wupA (wupA_gRNA1: ACCAAAAACACAAATCAAAA, wupA_gRNA2: TGAGGTGCGCAAGCGCATGG, wupA_gRNA3: CGCATCATCGAAGAACGTTG) would affect all 13 alternate transcripts annotated in FlyBase. The PAM site for wupA_sgRNA1 corresponds to the most common start codon and is in the 5′ UTR of transcripts using the alternate start codon, the PAM site for wupA_sgRNA2 corresponds to the alternate wupA start methionine which is a codon used by in all transcripts, and that for wupA_sgRNA3 is in a coding exon common to all transcripts (Fig. 1). All primers used and sgRNA sequences are tabulated in Supplementary Table 1.

Fig. 1.

Targeting haplolethal genes on the X chromosome. a) Premeiotic cells expressing CRISPR/Cas9 (from the autosomes) that effectively target haplolethal loci on the X chromosome will result in male only offspring. If the machinery were put on the Y chromosome, then inheritance of the elements will be limited to the male germ line and a gene drive would be created. b) Eleven haplolethal/haplosteriles were reported by Cook et al. (2012), 2 (RpS6 and RpS5a) were targeted by Fasulo et al. (2020), and the location of the 4 loci targeted in this manuscript and the number of sgRNA used (rounded blue rectangles) are shown. c) The wupA gene encodes for 13 transcript isoforms. SgRNA1–3 target 3 conserved regions of the gene’s coding sequence.

Microinjections

A total of 250–500 ng/µL of the pRpS19a + RpL35 and pwupA plasmids were microinjected into the 09w fly line, while the pRpL36 + RpL35 was injected into the 10w fly line. These 2 fly lines were obtained from Trent Perry, Bio21 Unimelb who had generated them using the Bloomington stocks #24749 and #25709 (y[1] v[1] P{y[+t7.7]=nos-phiC31\int.NLS}X; P{y[+t7.7]=CaryP} attP40) and #25710 (P{y[+t7.7]=nos-phiC31\int.NLS}X, y[1] sc[1] v[1] sev[21]; P{y[+t7.7]=CaryP}attP2) to generate second chromosome landing site flies (09W) and third chromosome landing site flies (10W) in a white null background. Insertion events were validated by sequencing and single integration lines were maintained.

Assessing sex ratio biasing capabilities of sgRNA constructs

To assess the sex ratio biasing rates of each of the sgRNA expressing constructs, we first generated male flies that expressed both Cas9 and sgRNAs in their germ cells. Individual females homozygous for each of the 3 sgRNA constructs (RpS19a + RpL35, RpL36 + RpL35 and wupA) were crossed to males containing (1) Cas9 under the control of the nanos promoter (either NIG CAS-0001; y² cho² v¹; attP40{nos-Cas9}/CyO which carries a recessive lethal allele on the second chromosome and cannot be made homozygous) or (1) a line we generated that has nos-Cas9 (from Addgene plasmid 62208 courtesy of Simon Bullock which has a single NLS and a 3′ UTR from nanos) placed in the 09w second chromosome attP40 landing site (referred to herein as nos-Cas9(II)_NH) or (3) a vasa-Cas9 on the Y chromosome from BL91386 (Gamez et al. 2021). The paternal inheritance of Cas9 prevented maternal deposition and somatic activity of the drive (Champer et al. 2019) and is consistent with the ultimate aim of placing X-poisoning transgenes on the Y chromosome. Initially, single males heterozygous for Cas9 and one of the sgRNA constructs were crossed to 3 09w 3–4-day-old virgins (14 replicates) and allowed to mate and lay in a vial for 7 days before being cleared and all offspring that emerged counted. Subsequent crosses (14 replicates) included single nos-Cas9/wupA-sgRNA males left to mate with single females for 4 h or alternatively crossed to 3-day-old virgin females with the genotype C(1)DX,y¹,f¹/Y (Bloomington Stock number 4559; 7 replicates), which is an attached-X” line that forces paternal X chromosome inheritance.

Analysis of life stage associated with wupA lethality

Next, we assessed the lifestage in which sex skewing occurred with wupA X-poisoning crosses. Single males with a GFP marked X chromosome (y[1] M{RFP[3xP3.PB] GFP[E.3xP3]=vas-int.Dm}Zh-2A w[*]/Y; sg(wupA)/nos-Cas9 were crossed to 3 wild-type virgin females (DGRP line 897) and allowed to mate for 48 h. The control cross was the same except there was no Cas9. All flies were then transferred to a vial with grape juice agar and allowed to lay eggs for 14 h, after which the flies were cleared. At the time of clearing, the total number of eggs laid was counted. Twenty-four hours later, the eggs were sexed and the female eggs (GFP) were separated from the male eggs (non-GFP). We then tracked how many of the female eggs successfully emerged into larvae, pupae, and adults. We performed 20 replicates for each cross.

Competitive mating assays

We assessed mate choice by placing 1 ebony (dark body color) homozygous female in a vial with 1 ebony homozygous male and 1 test male of the following genotypes: (1) sg(wupA)/nos-Cas9, (2) sg(wupA)/+, and (3) +/nos-Cas9, D) 09w. The flies were allowed to mate for 24 h, after which the parents were cleared and the progeny was phenotyped for body color once they reached adulthood. All flies were 2–3-day-old virgins at the time of mating.

Sequence characterization of wupA survivors

Sons of crosses between X-poisoning males and attached-X females had their DNA extracted and were characterized with Sanger sequencing using standard protocols. PCR primers were used to amplify the DNA around all sgRNA-targeted sites of a single gene; these were purified with a spin column and Sanger sequenced by Macrogen (see Supplementary Table 1 for sequencing primers).

Two hundred of the surviving female progeny from the cross 09w × sg(wupA)/nos-Cas9 were collected and pooled into batches of 10. Their DNA was extracted following a standard phenol:chloroform DNA extraction protocol. PCR reactions were then performed using the primers labeled “wupA sg# Illumina sequencing” in Supplementary Table 1. Each set of primers amplifies a specific sg(wupA) target site. The PCR products of each female sample were pooled together and purified with Bioneer AccuPrep PCR/Gel Purification Kit. Samples were sent for MiSeq 300 sequencing with AGRF. The sequencing results were input into CRISPResso2 for allele-specific quantification of each sg(wupA) target site (Clement et al. 2019).

Statistical analyses

All statistical analyses were undertaken in R (R Core team 2023). Sex ratio biasing in the Drosophila X-poisoning gene drive crosses were analyzed using an unpaired 2-tail t-test with a Bonferroni correction for multiple comparisons where applicable.

Results

X-poisoning targeting ribosomal protein genes

Four of the haplolethal genes described on the X chromosome of D. melanogaster (Cook et al. 2012) were targeted using 3 multi-sgRNA constructs and Cas9 expressed in the male germline via a nanos promoter (Fig. 1). The expectation was that males carrying the Cas9 and sgRNA transgenes would have a reduced number of viable daughters relative to controls yet the same number of sons (Fig. 2a). All the crosses targeting ribosomal protein encoding genes (RpS19a + RpL35, RpL36 + RpL35) yielded significantly fewer offspring when compared with the controls (P < 0.0001, t-test with Bonferroni correction; Fig. 2b; Table 1). The crosses involving the RpS19a + RpL35 sgRNA construct followed expectations with significantly fewer daughters than sons (P < 0.0001, t-test with Bonferroni correction) and no significant difference in the number of sons compared with the control. In contrast, the RpL36 + RpL35 crosses yielded significantly fewer sons and daughters compared with control (P < 0.05 and P < 0.01 respectively, t-test with Bonferroni correction) without a significant sex ratio bias (Table 1).

Fig. 2.

Sex biases with X-poisoning. a) A symbolic representation of what is expected in our crosses. A skew against daughters (red; shown on left) is expected when sgRNAs target haplolethal loci on the X chromosome during gametogenesis. The number of sons is expected to remain unchanged as they would arise from sperm carrying the Y chromosome. b) The observed amount of sex biasing skew ranged from nothing (RpL36 + RpL35), to modest (RpL19a + RpL35), to promising (wupA). Note that the total number of offspring was also variously affected. c) By excluding multiple mating, the total number of offspring in the wupA cross was reduced and there was a strong skew toward males. d) CRISPR/Cas9 fathers crossed to attached-X mothers reverse the inheritance of the sex chromosomes so that sons inherited the CRISPR modified X chromosome and daughters inherited the Y chromosome. (*P < 0.05, **P < 0.01, **P < 0.001, ****P < 0.0001, unpaired 2-tail t-test with a Bonferroni correction for multiple comparisons).

Table 1.

Sex skews in X-poisoning crosses and controls.

Gene/s targeted with sgRNA	nos-cas9 used	Number of fathers used per vial	Number of mothers per vial	Number of replicate vials used	Stock origin of mothers	Number of sons	Number of daughters	Proportion of rarer sex	Drive efficiency (%)a
None	CAS-0001	1	3	14	09w	1,014	1,081	0.48	−3
None	CAS-0001	1	1	14	09w	603	544	0.47	5
None	CAS-0001	1	3	4	C(1)DX, y[1] f[1]/Y	182	151	0.45	9
wupA	CAS-0001	1	3	14	09w	1,409	176	0.11	78
wupA	CAS-0001	1	1	14	09w	698	94	0.12	76
wupA	NH-nos-Cas9	3	10	18	09w	1,569	257	0.14	72
wupA	CAS-0001	1	3	6	C(1)DX, y[1] f[1]/Y	7	290	0.02	95
wupA	NH-nos-Cas9	3	10	18	C(1)DX, y[1] f[1]/Y	14	780	0.02	96
RpS19a + RpL35	CAS-0001	1	3	14	09w	789	195	0.20	60
RpS19a + RpL35	CAS-0001	1	3	7	C(1)DX, y[1] f[1]/Y	55	143	0.28	44
RpL36 + RpL35	CAS-0001	1	3	14	10w	678	717	0.49	−3
RpL36 + RpL35	CAS-0001	1	3	6	C(1)DX, y[1] f[1]/Y	234	237	0.50	1

^aTwo times the difference between the expected frequency of 50% and the observed % of rarer sex.

Wings-up A poisoning skews the sex ratio

When the CRISPR machinery was targeted to wupA, significant skews in sex ratio toward maleness was observed (P < 0.0001, t-test with Bonferroni correction; Fig. 2b). However, we initially found the expression of sgRNAs targeting wupA did not result in a significant reduction in the total number of offspring produced when compared with controls (Fig. 2b). The highly significant reduction in the number of female offspring (P < 0.0001, t-test with Bonferroni correction) was offset by the significantly higher number of male offspring compared with control (P < 0.0001, t-test with Bonferroni correction). We hypothesized that the lack of reduction in reproductive output may be due to a carrying capacity within the vials with an excess of embryos being laid during the 7-day period where remating was allowed. To address this possibility, single males heterozygous for both X-poisoning drive components (i.e. the wupA targeting sgRNAs and nos-Cas9) were placed in vials with single females for 4 h only. In these single pair crosses, a significant reduction in the total number of adult offspring (P < 0.01, t-test with Bonferroni correction) as well as a highly significant change in sex ratio was observed (P < 0.0001, t-test with Bonferroni correction; Fig. 2c) and there was no significant difference in the number of males produced when compared with the control.

As wupA has very low expression in the early stages of embryos (first 8 h; modENCODE Temporal Expression Data FlyBase), it contrasts to many ribosomal proteins because it is not required for the viability of all cells and particularly not those in early development. The gene wupA encodes troponin I, a protein that inhibits the interaction of myosin and actin in muscles although it is also thought to have nonmuscle functions that relate to the nuclear localization of at least 1 of its 9 isoforms (Casas-Tintó and Ferrús 2021; FlyBase). To explicitly test where the sex skewing of wupA CRISPR/nos-Cas9 elicited its effect, we tracked the fate of X chromosomes labeled with GFP. In the crosses we conducted, female embryos glowed green (after the 3XP3 GFP transgene is turned on), whereas male embryos and unfertilized eggs did not. Twenty single pair crosses were set up for the haplolethal cross (males bearing wupA sgRNAs and Cas9) and for controls (the males had wupA sgRNAs but lacked Cas9), and these revealed that the number of female embryos did not differ significantly between the haplolethal cross and the control cross, but the number of larvae, pupae, and adults did (Fig. 3). We could determine that females of the haplolethal cross typically died in late embryogenesis (dorsal closure was completed, and we observed mouthparts and denticle belts; however, segmental furrows appeared unevenly spaced which may be attributable to abnormal muscle contraction; see Supplementary Fig. 1). Only ∼13% of GFP eggs became 1st instar larvae, whereas ∼72% of GFP eggs from the control cross were observed to develop into larvae (Fig. 3, Supplementary File 1). Consistent with earlier results, the ratio of females among adults in these experiments was ∼11% in the haplolethal cross and ∼52% in the control. We also examined 15 adult females emerging from this cross and all were fertile.

Fig. 3.

Developmental stage of lethality from wupA-poisoned fathers. a) The average number of daughters tracked from 20 vials of the haplolethal (wupA-poisoned) cross (right bar of each pair) and 20 vials from a control cross (left of each pair). Error bars represent the standard error of the mean, and asterisks represent significance (P < 0.05). b) The total number of daughters from 20 vials from the haplolethal (red) and control cross (blue) at each life stage.

No fitness cost detected in nos-Cas9 wupA-poisoning males

Transcriptome data sets (e.g. FlyAtlas2—anatomy RNA-Seq) indicate that wupA is expressed in the testis, and so, we also sought to determine whether CRISPR/Cas9 of wupA during spermatogenesis affected the fertility of the bearer. Males generating wupA alleles during spermatogenesis could have a phenotype that makes them less successful in mating, perhaps because of female choice, or they could produce lower quality sperm. To test for these possibilities, we examined the fitness of males in a competitive assay. Four classes of test males (those with CRISPR/nos-Cas9 targeting wupA and 3 control lines: 1 that carried the Cas9 without the sgRNA, another that had the sgRNA but no Cas9, and a third that had neither CRISPR component) individually competed for a female homozygous for the recessive ebony body color mutation against an ebony male. Eighty-five vials containing 2 males and 1 female were set up (∼20 vials for each of the 4 tester male genotypes; Table 2), and the number of ebony and wild-type progeny was counted. In the crosses involving males with both the sgRNAs and Cas9 transgenes (cross A, Table 2), 84 of the 112 progeny were wild type with 10 of the former being females. So these test males were much more competitive than ebony males, even though there was a massive skew in the sex ratio so that only 12% of the progeny (10/84) were daughters (Table 2, Supplementary File 1). In contrast, ebony males were more successful than the males of the 3 control crosses although only significantly so in cross D (Table 2).

Table 2.

Competitive mating assays.

Cross	A	B	C	D
Test male genotype	sg(wupA)/nos-Cas9	sg(wupA)/+	+/nos-Cas9	+/+
Total vials assayed	21	19	23	22
Vials with only ebony offspring	5	9	14	17
Vials with only wild-type offspring	16	9	9	2
Total ebony progeny (% female)	28 (64%)	98 (52%)	151 (50%)	213 (51%)
Total wild-type progeny (% female)	84 (12%)	57 (65%)	106 (49%)	76 (47%)

If competition is assessed at the level of vials, then 81 of the 85 vials produce either ebony progeny or wild-type progeny. At this level, the males with the full CRISPR/Cas9 machinery were also seen to be more successful than ebony males as they were the only sires in 16 vials, whereas ebony were the only sires in 5 of the vials (P = 0.03). Again, this was in contrast to the crosses involving the 3 control genotypes where the number of ebony either did not differ significantly from wild type (sgRNA-only control P = 0.82; Cas9-only control P = 0.21) or in the case of the background control line were more fit (P = 7.6 × 10⁻⁵). So, we found no evidence of a pre-laying fitness deficit for males that have CRISPR machinery targeting wupA during spermatogenesis in the nanos expression domain.

In contrast, we attempted to create wupA-poisoning males that would have Cas9 expressed under the vasa promoter, but this failed. We crossed males which had vasa-Cas9 on the Y chromosome and females that were homozygous for the wupA-sgRNA construct. This yielded 264 daughters (that did not have Cas-9 because they did not inherit the Y chromosome) and only 2 small and infertile males. This is consistent with earlier reports that vasa expresses outside the germline (Gamez et al. 2021) and shows that X-poisoning with wupA could have drastic fitness costs if the CRISPR machinery is active in broader expression domains.

The nos-Cas9 wupA sex biasing is not absolute

To understand why the sex biasing was not absolute, X chromosomes that had survived being passed through the paternal germ line were characterized. If the targeted loci had been cut and reannealed, indels or single-nucleotide mutations may have been introduced, and they would be revealed by sequence analysis at the sites targeted by the sgRNAs. Initially, to simplify the sequence analysis, we used X^{^}XY females (hereafter “attached-X”) which reverse the inheritance of the paternal X's so that they are passed to sons. Consistent with expectations, males expressing nanos-Cas9 and sgRNAs targeting wupA sired significantly more daughters than sons (the sex ratio bias being flipped due to sons paternally inheriting the X chromosome; P < 0.001, t-test with Bonferroni correction). The RpS19a + RpL35 cross showed no significant sex ratio bias but had significantly fewer male offspring than the control and no significant difference in the number of females (Fig. 2d; P < 0.0001, t-test with Bonferroni correction). There was no reduction in the number of offspring produced by males expressing gRNA targeting RpL36 + RpL35.

The X chromosomes of a total of 60 surviving sons (14 for wupA, 22 for RpS19a + RpL35, and 24 for RpL36 + RpL35) from 15 independent crosses were Sanger sequenced, and only the offspring of 1 male showed any sign of CRISPR cutting. A complex indel resulting in a net 9 nucleotide insertion that did not introduce a premature stop codon was recovered from a cross involving the RpL36 + RpL35 sgRNAs (Supplementary Fig. 2). The fact that this individual survived suggests that gene drive resistance alleles could arise against RpL35 gRNA1. The data also concurs with the idea that these genes are indeed lethal genes as no frameshifting indels or nonsense mutations were recovered. More generally, these data are consistent with the notion that most of the surviving progeny carrying a paternal X never had their X chromosome cut by the CRISPR machinery.

Not all wupA mutations are haplolethal

Curiously, we found that the skew in sex ratio was more extreme when the wupA-poisoned X chromosome was passed to sons (2%; via attached-X females) rather than to daughters (14%; Fig. 2d vs Fig. 2b and c; Table 1). As the fathers in these crosses are the same, this difference is not due to the generation of X-poisoned alleles. Instead, heterozygosity in daughters may protect against lethality more than the exposed hemizygosity of sons, which suggests that some of the CRISPR generated alleles may encode recessive lethals. If true, then we reasoned it might be possible to find classes of mutations in daughters of males that have wupA targeted during spermatogenesis that do not occur in their sons (from crosses with attached-X). Three PCR amplicons surrounding the 3 sgRNA sites in wupA were sequenced from 20 pools of 10 daughters using Illumina's short-read MiSeq sequencing technology. Strikingly, each of the 20 pools had a high frequency of deletion alleles (between 5% and 18%) among the reads from the amplicon that amplified across the site targeted by sgRNA2 (Fig. 4, Supplementary File 1). Given that each of the 20 pools was composed of 10 daughters and we saw no such indels among the 14 sons from the attached-X cross, a significant number of the mutations generated by CRISPR-editing seem to be lethal (purged in sons) but not haplolethal (and so present in sequences from daughters). Notably, all of the high-frequency indels observed had a length divisible by 3 and therefore did not change the ORF. So it would seem that frameshift alleles are haplolethal but some indels that maintain the ORF are lethal but not haplolethal.

Fig. 4.

Recessive lethal wupA mutations revealed in females. a) The 3 sites within wupA that are targeted by sgRNAs (see Fig. 1c) vary in the number of CRISPR mutations uncovered by sequencing 200 daughters of an X-poisoning cross. Here, a CRISPResso2 output shows the total number of modified and unmodified reads from Illumina sequencing at each target site. b) A nucleotide alignment of the modified reads of target site 2 showing their frequencies. Amino acid sequences are shown at the top, the sgRNA sequence is highlighted in gray, the PAM site is underlined, the triangle is the cut site, and the black blocks represent insertions (shown more to detail in Supplementary Fig. 3). c) An overlay of Isoform G of troponin I without any mutations (blue) and with the largest deletion found in our sequencing results (a 21 bp deletion shown in red). Protein folding predicted with AlphaFold. The deletion location is highlighted in gray. No mutations in target site 2 will affect the acting-binding and troponin-interacting areas.

Discussion

For an X-poisoning strategy to elicit an effect in a single generation, the targeted genes need to be haplo-insufficient. Such is the knowledge associated with D. melanogaster that it is known there are 11 loci on the X chromosome of D. melanogaster that are haplolethal or haplosterile (Cook et al. 2012). Haplolethal genes can only exist on the X chromosome in species such as Diptera where dosage compensation between the homogametic and heterogametic sex is due to upregulation—doubling the output of the X chromosome in males, rather than the inactivation of the X as seen in humans.

Eight of the X-linked haplolethals of D. melanogaster encode ribosomal proteins, which contribute to the translational machinery of cells, and an insufficient quantity of them leads to lethality or to the minute syndrome where flies are tiny and sterile (Marygold et al. 2007). Previously, Fasulo et al. (2020) pioneered the “X-poisoning” strategy by targeting ribosomal protein genes S6 and S5a. They found that when 1 particular sgRNA targeting RpS6 was combined with Cas9 driven by a sperm-specific beta tubulin 85D promoter, the sex ratio was biased so that >92% of progeny were males. Our targeting of RpL36, RpL35, and RpL19a did not produce such extreme biasing. However, Fasulo et al. (2020) also showed heterogeneity in the success of their sgRNAs with different guide RNAs targeting the same genes skewing to various extents. Therefore, the failure to see skewing in our combination of RpL36 and RpL35 sgRNAs or the skew to only 80% males in the RpL35 and RpL19a combination may well have more to do with the particularities of guide RNA efficacy than being a property of the genes they target, and in our case by subtle design, features such as the position of particular guide RNAs in the array may be important.

A surprising finding of Fasulo et al. (2020) was that some alleles among the survivors contained frameshift mutations. This led the authors to suggest that the targeted ribosomal protein genes may not be haplolethal genes after all and that previously studied alleles may have harbored dominant negative mutations instead. In our analyses, we did not find such frameshifting alleles among individuals carrying X chromosomes that had passed through X-poisoning males. Instead, the loci showed no signs of being cut at all or exhibited mutations that maintained the frame (Fig. 4b for sgRNA2 of wupA and Supplementary Fig. 3 for sgRNA1 and sgRNA3 of wupA).

The wupA locus of Drosophila was the only nonribosomal protein haplolethal gene that has been molecularly characterized on the X chromosome, and it showed the greatest skewing in our X-poisoning experiments. It encodes troponin I, a protein that inhibits the interaction of myosin and actin in muscles. It is also thought to have nonmuscle functions that relate to the nuclear localization of at least 1 of its isoforms (Casas-Tintó and Ferrús 2021; FlyBase). The gene has 12 exons that are variously combined to produce 13 alternate transcriptional splice forms. Two of the 3 sgRNAs used in our study target exons common to all splice forms (sgRNA2 and sgRNA3), and the third targets a PAM site that is found in the first 2 codons of isoforms with the most common start methionine. The latter site was previously targeted with CRISPR by Casas-Tintó and Ferrús (2021), and the resulting haplolethality is consistent with our findings.

Our results provide an insight into the distinct functions of wupA because indels were enriched around the sgRNA2 target site which lies on top of the methionine codon used by wupA isoforms that translocate into the nucleus. We infer that such mutations are recessive lethals that are exposed in the hemizygous condition of males. However, as heterozygotes in females, they produce troponin I with an intact actin-binding site and an intact troponin T interaction site and so they are not haplolethal. The haplolethality of wupA seems to relate to the delicate stoichiometry required for troponin I and F-actin. Without sufficient troponin I, calcium is not required to allow tropomyosin to move and allow actin to interact with myosin. Thus, lethality arises from the irreversible hypercontraction of muscles. Such a model is consistent with observations that mild wupA alleles (yielding the wings-up phenotype) can be rescued by a mild allele of tropomyosin (Naimi et al. 2001).

There is much about wupA that makes it an attractive target for future X-poisoning strategies. As a haplolethal gene not expressed until late embryogenesis, the progeny of a poisoned cross are unaffected at fertilization and egg-lay. It thus contrasts with X-poisoning targeting ribosomal proteins which may have reduced fertility because spermatogenesis may be affected. It also contrasts with genetic systems causing sterility (such as pgSIT (Kandul et al. 2019)) which can be counteracted by females mating multiple times. We also did not find males that poison wupA to have reduced fitness, although we acknowledge that our competitive assays are likely be confounded with genetic background effects (especially eye color attributable to the number of transgenes) but argue that if there is any fitness cost attributable to wupA-poisoning using a nanos-driven Cas9 it is less than the loss of a second transgenic copy of mini-white and less than that of ebony males.

The wupA gene is a highly conserved gene that is found on the X chromosome on numerous dipteran species, including some important pests (Supplementary Fig. 4). Drosophila suzukii, an invasive pest of soft-skinned fruits, appears to share all 12 exons with D. melanogaster and all the predicted proteins have >98% amino acid identity with D. melanogaster homologs. In fact, 2 of the 3 sgRNAs used here in D. melanogaster match D. suzukii targets perfectly, and the third has only 1 nucleotide different. wupA of Anopheles mosquitos has a slightly different gene structure (e.g. Exons 1 and 2 are fused, e.g. XM_053811545); however, it is also found on the X chromosome in these species suggesting it could be a target of X-poisoning strategies in the many pest species of this genus. Unfortunately, wupA orthologs are not on the X chromosome of all economically important dipterans, such as the tephritids which have Muller element F (chromosome 4 of Drosophila) as an X chromosome.

Taken together, the above results suggest that wupA could be a good target for an X-poisoning gene drive (Burt and Deredec 2018; Fasulo et al. 2020; Haber et al. 2023). To create such a drive, transgenes encoding a germline-restricted Cas9 and the sgRNAs targeting wupA would ideally be placed on the Y chromosome. Placing transgenes on the Y chromosome has been technically challenging although it has been done (Gamez et al. 2021). Indeed, we examined the 1 D. melanogaster line currently available that has a Cas9 transgene on the Y chromosome; however, because it uses a vasa promoter that expresses outside, the germline males that poisoned the X during spermatogenesis could not be generated. An important next step in this research program will be to overcome the challenge of placing germline-limited drive components onto the Y chromosome.

Finally, we note that where an X-poisoning strategy can take the form of a Y-linked editor (Burt and Deredec 2018) it may not necessarily drive through a population. While fathers may only have sons, the Y chromosome bearing the X-poison may not produce more sons than any other Y chromosome in a population and so it may not be considered a gene drive. However, if X-poisoned males produce more sons than other Y chromosomes, perhaps as a result of reduced larval competition with sisters (such as seen in the original wupA cross), then a Y-linked editor could indeed be a gene drive (Friberg and Rice 2015; Burt and Deredec 2018). In this larval competition scenario, as the frequency of the Y-linked editor increases, then so too will the strength of the gene drive.

Data availability

Strains and plasmids are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.

Supplemental material available att G3 online.

Funding

The research was funded by an Australian Research Council (DP190102512).