Generation of a gRNA-only homing element mosquito line targeting white
To test whether a CRISPR-based homing gene drive is feasible in Cx. quinquefasciatus, we employed a split, gRNA-only gene-drive strategy which takes advantage of a vasa-Cas9 line that we previously built26 as a source of Cas9 protein. As this Cas9 transgene was inserted at the cardinal locus, we decided to utilize another gene causing a visible phenotype, the white gene (CPIJ005542), as a homing target. Both white and cardinal are located on chromosome I of Cx. quinquefasciatus, but are genetically unlinked, being located on opposite chromosomal arms. Additionally, we note that white is known to be tightly linked to the sex-determining region in both Culex and Aedes (hereon, we use ‘M’ to represent the allele carrying the Male-determining sequences and ‘m’ to represent the corresponding locus lacking the male-determining ability; therefore males are indicated as M/m, and females are indicated as m/m) (Fig. 1a)30,31. The vasa-Cas9 transgene used to test the split gene-drive arrangement was previously generated by integrating a Cas9 transgene expressed under the control of regulatory sequences from the Cx. quinquefasciatus vasa gene, and marked by a DsRed gene controlled by the Opie2 promoter (Fig. 1b). We then built the second component, the ‘homing’ gRNA-only drive by inserting it into exon 5 of the white locus at the exact cut site specified by the white-gRNA6, a construct comprising: 1) the white-gRNA6 under the control of the Cx. quinquefasciatus U6:1 promoter, which carries the “Loop” modified scaffold version for increased efficiency26,32; 2) an eGFP fluorescent marker under the control of the Hr5IE1 promoter to track the gRNA transgene; and 3) an in-frame re-coded portion of the white gene that would produce a functional white protein thereby restoring the endogenous gene’s activity (Fig. 1c). The targeted transgene delivery was achieved by adding, to the plasmid, two ~1kb homology arms (HAs) matching the genomic sequence of the white locus abutting the white-gRNA6 cutting site (Fig. 1c).
To generate transformant mosquitoes carrying this white-gRNA6-only drive construct, the plasmid depicted in Fig. 1c was injected into eggs collected from the vasa-Cas9 line. The surviving G0 mosquitoes were then separated into male and female pools and outcrossed to our wildtype line (see methods for details). The resulting 693 G1 progeny were phenotypically screened for the presence of the eGFP marker, and 1 transformant male was recovered (Supplementary Data 1). This eGFP+/DsRed+ male, carrying both the white-integrated gRNA-drive and cardinal-integrated vasa-Cas9 transgenes, was then mated to 20 virgin wildtype females to establish a transgenic line (Supplementary Data 2). From this cross, nearly all of the recovered male offspring displayed eGFP fluorescence, suggesting that the white-gRNA6 transgene had likely inserted at the white locus in tight linkage with the sex-determining region (Fig. 1a, Supplementary Data 2), and specifically, linked to the M allele. From the same cross, we also recovered a single eGFP+ female, providing a first indication that the white-gRNA6 element was potentially capable of homing onto the opposing, m-allele-carrying chromosome in the presence of the Cas9 transgene (Supplementary Data 2), although, from these results, we cannot rule out a recombination event occurring between the two loci. This single eGFP+ female was then crossed to 8 eGFP+ males in order to isolate a second m-linked white-gRNA6 transgenic line (Supplementary Data 2). The insertion of the transgene was confirmed by PCR amplification and Sanger sequencing, indicating the integration of all the functional portions of transgenes along with the backbone (Supplementary Figure 2). Since the two insertion junctions were seamless with the surrounding genomic flanking region, as confirmed by Sanger sequencing, the presence of the backbone indicates a tandem integration of the construct, which, if limited in the amount of repeats, should not impact its functionality and ability to home onto the opposing chromosome.
Assessing gene drive at the white locus using an inheritance-bias approach
In our transgenic recovery strategy described above, we were able to obtain two separate lines: the first carrying the white-gRNA6 homing element tightly linked to the M-locus and homozygous for the vasa-Cas9 transgene, and a second line carrying the white-gRNA6 homing element tightly linked to either the M-locus or the m-locus, in an otherwise wildtype background. These two lines allowed us to assess whether local sequence differences around the sex locus impair chromosomal pairing and efficient homing onto the homologous chromosome. We utilized each of these lines to assess the homing of the white-gRNA6 drive and evaluate the conversion efficiency in either males, with the white-gRNA6 insertion homing from an M-linked to an m-linked locus (M-to-m homing), or in females with the white-gRNA6 insertion homing from an m-linked to another m-linked locus (m-to-m homing).
To assess M-to-m homing, we mated 10 G0 males from the white-gRNA6; vasa-Cas9 line to 10 wildtype G0 females. From their progeny, we isolated trans-heterozygous G1 males that carried both the vasa-Cas9 and white-gRNA6 transgenes so as to ensure that the only white-gRNA6 element present was M-linked. We then mated these trans-heterozygous males to virgin wildtype females in single pairs to obtain and analyze each of their G2 progeny independently (Fig. 2a). Separately, to assess m-to-m homing, we mated 10 white-gRNA6 G0 females to 10 vasa-Cas9 G0 males, and from their offspring we isolated G1 trans-heterozygous females carrying both transgenes and crossed them to wildtype males in single pairs to obtain and analyze their G2 progeny (Fig. 2b). For both crossing strategies, the phenotypic analysis of the fluorescence ratios in the G2 generation allowed us to evaluate the homing efficiency of the white-gRNA6 element (Fig. 2a-b).
For the M-to-m homing assessment, the G2 progeny of 7 G1 crosses were scored and an average inheritance of 51.2% [95% CI: 45.2% – 57.1%] was observed for the white-gRNA6 transgene, showing no significant difference (β=0.05±0.12, Wald statistic z=0.4, p=0.67) from the expected 50% inheritance ratio of Mendelian inheritance (Fig. 2c, Supplementary Data 3). For the m-to-m homing assessment, the G2 progeny derived from 19 trans-heterozygous female crosses were evaluated and a modest but significant inheritance bias of 54.8% [95% CI:51.6% – 57.9%] was observed (β=0.19±0.07, z=2.75, p=0.006) (Fig. 2d, Supplementary Data 3). When analyzing the inheritance of the vasa-Cas9 transgene, we found no evidence for inheritance bias of this transgene (Males: 53.7% [95% CI:47.7% – 59.5%]; Females: 50.2% [95% CI:46.9% – 53.5%]). In summary, we seem to observe a slight inheritance bias only when the split drive homes in the m-to-m condition, and not in the M-to-m, suggesting that homing could be potentially impaired at this locus when homing occurs between chromosomes with slight differences in the local sequences. Additionally, since it seem that the homing process may occur at very low levels (Fig. 2d), homing events may also be happening in the M-to-m condition, although the smaller sample size and the limited resolution of this assay may be hiding this (Fig. 2c).
Assessing gene drive at the white locus using a marked chromosome approach
During our initial experiment, we noticed a modest yet significant bias in the inheritance of the white-gRNA6 transgene. To better understand the underlying chromosomal conversion events, we designed a follow-up experiment to visualise these events more effectively, that would allow us to confirm homing events independently from other mechanisms of biased inheritance.
In order to investigate the inheritance of the white-gRNA6 transgene, we designed a more sensitive assay. This assay involved using a marked-chromosome approach by using a specific mutation closely linked to the homing site on the receiver “m” chromosome. This mutation is not present on the donor, white-gRNA6-element-containing “M” chromosome (Fig. 3a). If a successful homing event occurred, it would be identified by the linkage of the white-gRNA6 element with this unique marker mutation on the receiver “m” chromosome. This marker mutation, hereon referred to as w4-, was repurposed from a white mutant line that we previously built 27, which carries a disruption of the white coding sequence under the action the white-gRNA4, targeting exon 3 (w4 mutation, Fig. 1c, Fig. 3a), and sitting at a 5.6 kbp distance from the white-gRNA6 element insertion site. Contrary to the sex-determining locus, which while being tightly linked with white allows for recombination in between the markers, the w4- mutation is too close to the white-gRNA6 insertion site to generate meaningful recombination in the intervening sequence.
To conduct the analysis we first crossed homozygous vasa-Cas9 females with homozygous w4-/w4- mutant males to obtain trans-heterozygous w4+/w4-; cd-,vasa-Cas9+/cd+ offspring (Fig. 3b). Subsequently, we intercrossed these individuals to generate and isolate progeny with a DsRed+/w4- phenotype, which would be homozygous for the w4- mutation (white eyes) and carrying at least one copy of the vasa-Cas9 transgene (DsRed+) (Fig. 3b). To set up the marked-chromosomes homing analysis, we took 30 of these DsRed+/w4- G0 males and crossed them in a pool with 30 G0 virgin females containing either one or two copies of the white-gRNA6 transgene (Fig. 3b). From the resulting offspring we chose trans-heterozygous G1 virgin females carrying both the vasa-Cas9 (DsRed) and the white-gRNA6 (eGFP) transgenes, along with the w4- mutation in heterozygotes (w4+/w4-). These G1 females were then pool mated to w4-/w4- males, and the resulting G2 offspring was collected at the egg-raft stage and allowed to hatch in separate containers. This setup enabled us to assess the inheritance of the white-gRNA6 in the germline of each G1 females by examining the eye phenotype and marker presence of their G2 offspring (Fig. 3b).
Since the w4- precedes the white-gRNA6 insertion site on the white coding sequence (Fig.1c), by scoring the eye color of the G2 offspring we were able to distinguish the original white-gRNA6 which had dark eyes (w4+,white-gRNA6/w4-,w6+, where w6+ indicates wildtype sequence for the white coding sequence at the white-gRNA6 location), from the ‘homed’ copies which instead displayed a white-eye phenotype (w4-,white-gRNA6/w4-,w6+) (Fig. 3b). Consistent with our previous experiment (Fig. 2d), in the G2s, we observed an average inheritance rate for the eGFP+ marker of 56.3% [95% CI:52.3% – 60.2%], significantly above the null hypothesis of Mendelian inheritance (β=0.25±0.08, z=3.18, p=0.001) (Fig. 3c). Furthemore, by implementing this multi-step procedure involving the w4- allele and the eGFP marker, we could assess both homing events and biased inheritance of the donor chromosome, providing us with a comprehensive understanding of transgene inheritance as this locus. Homing events occurred in 4 out of 15 of the analyzed rafts, and in 4.1% (95% CI: 2.2 – 7.3%) of the total receiver chromosomes we observed homing, which caused a mean bias in inheritance of 1.9% [95% CI:1.0% – 3.3%] (Fig. 3d). Separately, due to our ability of tracking donor and receiver chromosomes, we also found evidence of a significant bias in the inheritance of the donor chromosome at a mean rate of 4.4% [95% CI: 0.4 - 8.4%]; higher than predicted by mendelian inheritance. Taken together, homing-generated bias, and increased donor-chromosome inheritance account for the total bias in the eGFP+ marker observed (Supplementary Data 3).
Assessing gene drive at the kmo locus using an inheritance bias approach
Previous work on split-drive systems in Ae. aegypti identified a strong influence of the target site on homing efficiency, even when using the same Cas9 expressing line (e.g. comparing the inheritance rates of homing elements integrated into the kmo33 and white25 genes in Aedes aegypti when paired with the same SDS3-Cas9 line). To assess whether the homing rate might be higher at a target site different to white, we paired the vasa-Cas9 line with our previously-established gRNA homing element inserted into exon 5 of the kynurenine 3-monooxygenase (kmo) gene (kmo-gRNA line)29,34. Unlike the white-gRNA6 line used above, this homing element did not contain a kmo rescue fragment and thus represents a null allele for the gene function, resulting in eye lacking pigment. In order to test homing of the kmo-gRNA drive, male and female pupae from the homozygous vasa-Cas9 line were sexed and reciprocally mated en masse to heterozygous individuals from the kmo-gRNA line (the G0 generation). Progeny from each of the two crosses were screened as pupae for the presence of the two transgenic markers as well as for any eye color mosaic phenotypes resulting from Cas9-based disruption of kmo (Supplementary Fig. 4). Unlike for the white-gRNA6 experiment, where the two transgenes were differentiated using different fluorescent markers, in this analysis we used transgenic lines both marked with DsRed; fortunately we were able to identify differences in the DsRed expression pattern driven by either the Hr5IE1 or the Opie2 promoters that allowed us to clearly identify animals carrying each of the transgenes (Supplementary Fig. 3). It was noted in both crosses that all trans-heterozygous pupae (those that had inherited both transgenes) displayed a strong kmo mosaic phenotype, while no such phenotype was observed in progeny inheriting only the vasa-Cas9 transgene (Supplementary Fig. 4). Trans-heterozygous G1 animals from each cross were sexed and independently crossed en masse to wildtype individuals of the opposite sex (giving four pooled-cross cages). One week after crossing, cages were blood-fed and egg rafts collected and allowed to hatch and recorded individually. Larval progeny from each egg raft were screened for the presence of the two transgenes (the G2 generation). For each egg raft replicate, kmo-gRNA and wildtype G2 progeny were isolated and allowed to continue developing till pupation, at which point they were screened for the eye color phenotype, indicative of kmo activity.
Overall, we observed a kmo-gRNA inheritance rate of 57.5% [95% CI:55.5 - 59.6%] equating to approximately 15% of wildtype alleles being converted (homed) to kmo-gRNA alleles in the G1 germline, a highly significant difference from the null hypothesis of Mendelian inheritance (β= 0.24±0.03, z = 7, p<0.001). No significant effect was observed regarding the Cas9-bearing sex of the G0 generation, however when this treatment was collapsed, sex of the transheterozygous parent in the G1 cross had a marginally significant effect on kmo-gRNA inheritance (progeny of female G1 transheterozygotes = 55.0% [95% CI:53.2 - 56.7%] for kmo-gRNA inheritance, progeny of male G1 transheterozygotes = 60.0% [95% CI:56.3 - 63.7%] for kmo-gRNA inheritance. Akaiake Information Criteria (AIC) selection confirmed this was the minimal significant model.
Transgenerational deposition of Cas9/kmo gRNA ribonucleoprotein (RNP)
By screening the kmo-gRNA-expressing and wildtype G2 progeny of the above experiment for their kmo phenotype we were able to assess whether RNP was being transferred either maternally or paternally between generations. In these cohorts, we observed a substantial number of mosaic-eye color individuals (Fig.5), indicating transgenerational deposition of Cas9 and the kmo-targeting gRNA from the G1 generation into G2 embryos. These mosaics occurred along a spectrum of severity – from ‘faint’ to ‘strong’, alongside fully white-eyed individuals (Fig. 5a-e). In cohorts where these mosaics occurred, they were generally observed in less than 50% of individuals screened, with the highest proportion of mosaics being grouped into the ‘faint’ category and fewer ‘strong’ category or completely white-eyed individuals observed. While substantial levels of mosaicism were observed in the progeny of G1 transheterozygous females, three out of the four G2 cohorts deriving from G1 transheterozygous males displayed no mosaicism or white eyes at all, with the fourth showing only two faint mosaic individuals (c. 5% of individuals in that cohort). Interestingly, while we expected to observe higher levels of mosaicism in the kmo-gRNA G2 progeny than their wildtype siblings, the opposite was the case, although these differences were not statistically assessed.