Self-limiting population suppression gene drive design in the West Nile vector mosquito, Culex quinquefasciatus

Abstract

Culex quinquefasciatus is a major vector of West Nile virus and other pathogens, yet genetic population suppression tools for this species remain limited. Here, we develop a self-limiting, CRISPR-based suppression gene drive system targeting doublesex, close to the male-determining locus, promoting male transmission. A recoded dsxM sequence converts females into sterile intersexes, preventing population-level spread. The drive achieves super-Mendelian inheritance ( ~ 71%) and generates resistance alleles that are fully or partially dominant female sterile. Single-release cage trials show extended but self-limiting population suppression. Population modeling of this RIDD (Release of Insects carrying a Dominant-sterile Drive) system further indicates that repeated releases can substantially reduce fertile female numbers at low release ratios and intrinsic growth rates, outperforming non-drive strategies under comparable conditions. Together, these results establish a self-limiting suppression gene drive platform for Culex, providing a confined and sustainable framework for vector population control.

Culex mosquitoes drive the spread of West Nile virus, yet controlling their populations remains challenging. Here, authors develop a self-limiting CRISPR-based genetic approach that converts female mosquitoes into sterile intersexes, enabling effective and confined population suppression.

Introduction

Homing gene drives are genetic elements that bias their own inheritance beyond Mendelian expectations^1–4. Among the gene drive platforms developed to date, CRISPR-based systems are the most extensively studied. These systems typically consist of Cas9 and guide RNA (gRNA) components inserted at their genomic target site, such that the drive resides opposite the wildtype allele on the homologous chromosome. Upon expression, the Cas9/gRNA complex induces a double-strand break at the corresponding wildtype allele, which may be subsequently repaired via homology-directed repair (HDR), copying the drive allele into the cleaved chromosome. This process converts heterozygotes into homozygotes in the germline, resulting in super-Mendelian (>50%) inheritance in subsequent generations⁵. CRISPR-based homing gene drives have been successfully demonstrated across diverse organisms, including flies^5–8, bacteria⁹, yeast¹⁰, mice¹¹ and mosquitoes from Anopheles^12–17, Aedes¹⁸, and Culex¹⁹ genera.

In mosquitoes, CRISPR-based gene drive architectures have primarily been engineered for two purposes: population modification, in which transgenic mosquitoes carrying anti-pathogen effectors replace wildtype populations²⁰, and population suppression, in which drives typically target genes essential for reproduction, sex determination^13,15, or sex ratio distortion^21,22. While self-sustaining homing gene drives that can spread through populations hold transformative promise for controlling vector-borne diseases, their capacity for uncontrolled spread, even from low initial introduction frequency, has raised significant ecological, ethical, and regulatory concerns^23–26. In particular, the potential for transboundary dispersal and unintended gene flow remains a major challenge for safe deployment^27–29.

These concerns have motivated increasing interest in developing temporary or self-limiting gene drives. In such systems, homing or other inheritance-biasing mechanisms are intentionally combined with strong fitness costs or design constraints that prevent long-term population invasion, thus differing fundamentally from self-sustaining gene drives. Some designs, including split drives or killer-rescue systems, can transiently increase in frequency following release before ultimately being lost from the population^30–32. In contrast, other suppression-oriented architectures have dominant fitness costs that prevent population-level increases in frequency without additional releases, even under favorable conditions²⁹. In these systems, inheritance bias only operates effectively in some individuals, and population suppression depends on repeated releases rather than autonomous spread. This gives them improved power but similar population dynamics compared to female-specific Release of Insects Carrying Dominant Lethals (fsRIDL), which has been developed in several species^33–36 and deployed in the field for mosquito control^29,37.

One such strategy is RIDD (Release of Insects carrying a Dominant-sterile Drive), an improved self-limiting population suppression strategy that integrates CRISPR-based inheritance bias with dominant, sex-specific fitness costs. In RIDD systems, inheritance bias operates during male germline transmission, while female progeny inheriting either the drive or nonfunctional resistance alleles (generated via end-joining) are rendered sterile or non-viable; therefore, despite incorporating a homing mechanism, RIDD systems are not capable of increasing in frequency within populations, even with ideal performance, because female carriers are sterile^7,38. A proof-of-concept RIDD system targeting the female-specific exon of the sex determination gene doublesex (dsx) was demonstrated in Drosophila melanogaster, where repeated releases of RIDD males led to progressive suppression and eventual collapse in laboratory cages⁷. As a terminal regulator of the sex determination cascade, alternative splicing of dsx directs male and female differentiation, making it a compelling target for genetic suppression strategies^22,39–43. In Drosophila, disruption of the female-specific exon of dsx usually induces dominant female sterility^7,44,45. In contrast, studies in Anopheles mosquitoes have reported more variable outcomes, with dsx disruptions exhibiting either recessive or partially dominant effects depending on species, target site, and genetic background^13,15,43,46. These findings highlight both the potential and the species-specific complexity of dsx-based population suppression strategies.

As a non-model insect, Culex quinquefasciatus is a major vector of West Nile virus, lymphatic filariasis, and avian malaria^47,48, spreading pathogens affecting humans, livestock, companion animals, and endangered wildlife^49,50, with increasing relevance for island conservation. The escalating prevalence of insecticide resistance in Culex populations further compromises the effectiveness of conventional control strategies^51,52. Despite substantial progress in CRISPR-based genetic control strategies for Aedes^18,53 and Anopheles^12–14,17, progress in Culex mosquitoes has lagged due to inefficient genome editing and the technical challenges of achieving precise, large-fragment integrations, leaving Culex-specific genomic control approaches comparatively underdeveloped. Although our previous work established a comprehensive Culex-optimized CRISPR toolkit and demonstrated proof-of-concept gene drives targeting marker loci (white and kmo)⁵⁴, no functional population suppression drive had yet been achieved in this genus. Moreover, limited understanding of primary sex-determination signals and downstream regulatory pathways continues to hinder the development of sex-targeting genetic manipulation strategies in Culex^55,56.

In this study, we develop a self-limiting suppression drive targeting the dsx locus in Culex quinquefasciatus. To our knowledge, this work provides a proof-of-principle demonstration of a functional genetic population suppression strategy in Culex mosquitoes, representing an advance in genetic control of this understudied genus. Our system targets a conserved exon (exon 4) of the dsx locus shared by both male and female isoforms, achieving a precise ~14-kb integration that encompasses all essential drive components along with a recoded dsxM fragment that preserves dsx function in males and interferes with normal sex determination in females. In Culex, there is no Y chromosome. Instead, sex is determined by the male-determining (M) locus, which is closely linked to dsx. As a result, the M-linked RIDD construct is predominantly transmitted to male (M/m) offspring of transgenic males through co-inheritance with the M-chromosome, while female (m/m) offspring that inherit the drive via homing in the paternal germline develop intersex phenotypes. Non-drive females inheriting end-joining (EJ)-derived resistance alleles exhibit mild intersex phenotypes, revealing partially dominant mutations in dsx affecting female development in Culex, paralleling observations from the Drosophila RIDD system⁷. Single-release cage trials show that persistence of drive-carrying males is primarily supported by tight linkage to the M-locus following release, while homing-mediated cleavage and end-joining repair generate sterile female progeny. Population modeling incorporating repeated releases shows that this RIDD system can achieve substantial suppression at low intrinsic growth rates, even with low release ratios, outperforming conventional self-limiting approaches such as SIT and fs-RIDL. Collectively, this study establishes a self-limiting gene drive suppression platform for Culex mosquitoes, bridging a long-standing technological gap and offering a promising framework for developing effective, confined, and ecologically responsible genetic control strategies for non-model disease vectors.

Results

Generation of a CRISPR-based RIDD suppression system targeting the Culex dsx locus

We engineered a homing-based RIDD system in Culex quinquefasciatus targeting exon 4 of the doublesex (dsx) gene (LOC6035520), a conserved region shared by both male and female isoforms (Fig. 1A). The drive construct incorporates a recoded sequence comprising partial exon 4 downstream of the gRNA cleavage site and a dsxM cDNA fragment, precisely inserted at the 3’ junction of the target site. The sequence is followed by a 3’UTR from the An. Gambiae kynurenine hydroxylase (kh) gene (LOC1276617). This heterologous 3’UTR, previously validated in Culex plasmid designs⁵⁴, was incorporated to reduce sequence homology and thereby minimize undesired recombination when paired with a recoded fragment. This configuration enables the formation of an in-frame transcript that preserves male-specific dsxM function while rendering the recoded allele refractory to Cas9/gRNA cleavage. Consequently, the strategy aims to maintain fertility in drive-carrying males while converting genetic females into sterile intersexes exhibiting masculinized traits.

Fig. 1. Generation and molecular validation of a homing-based RIDD suppression construct targeting the doublesex (dsx) locus in Culex quinquefasciatus.

A Schematic of the dsx gene illustrating alternative splicing isoforms: the male-specific (dsxM) and female-specific (dsxF) transcripts. Exons shared by both isoforms are shaded in gray; the male-specific exon is labeled blue (dsxM), and the female-specific exons in red (dsxF). Exons and introns are not drawn to scale. Two guide RNAs (dsx-1 and dsx-2), oriented in opposite directions, target the conserved exon with their sequences shown above. A partial amino acid sequence of exon 4 encompassing the editing sites is presented. The gene drive construct includes the following elements: nanos-Cas9, U6:1-dsx1 gRNA, U6:6-dsx2 gRNA, an Opie2-DsRed fluorescent marker, a recoded dsxM cDNA fragment, followed by the An. gambiae kh 3’UTR, and two flanking homology arms (HA1 and HA2). The recoded dsxM fragment is inserted immediately downstream of HA1. B Representative images of transgenic Culex mosquitoes under brightfield and fluorescence microscopy, showing strong, ubiquitous DsRed expression. C Gel electrophoresis of diagnostic PCR amplicons confirming precise insertion of the drive construct at the dsx locus, using specific primer sets (Supplementary Data 7). Experiments were independently repeated at least twice with similar results. Source Data are provided in Source Data 1.

To restrict Cas9 cleavage activity primarily to the germline, where edits can be transmitted to the next generation, the germline-specific nanos promoter was used to drive Cas9 expression. The nanos promoter has been shown to function effectively in gene drive systems in Drosophila and Anopheles mosquitoes, although its performance in Culex gene drives has not yet been investigated. We previously validated RNA Pol-III promoters, U6:1 and U6:6, which were employed to express a pair of closely spaced, oppositely oriented gRNAs targeting exon 4, with a 21-bp interval between their cleavage sites. The dual-gRNA configuration was designed to increase the probability of target cleavage and reduce the likelihood of functional resistance allele formation. However, one of the gRNAs (gRNA-2) exhibited no detectable editing activity, as described in detail later. The drive cassette also contained a DsRed fluorescent marker driven by the Opie2 promoter and was flanked by ~1.5-kb homology arms corresponding to the dsx locus (Fig. 1A).

To generate transgenic mosquitoes, approximately 500 Culex eggs were microinjected with the gene drive plasmid at a final concentration of 300 ng/μL. Because the plasmid contained both Cas9 and gRNA expression cassettes required for HDR-mediated integration, no additional Cas9 or gRNA sources were included during microinjection. Surviving G0 individuals were sexed and outcrossed to wildtype mosquitoes. From G1 progeny derived from the male pool, six transgenic individuals were recovered based on strong, body-wide DsRed fluorescence (Fig. 1B, Supplementary Data 1). One male carrying the correctly integrated construct and exhibiting a normal wildtype male phenotype was selected to establish a stable drive line by crossing with 10 wildtype females. The line was maintained through successive generations by selecting DsRed-positive males and outcrossing to wildtype females. Precise integration of the drive cassette at the dsx locus was confirmed by PCR and Sanger sequencing (Fig. 1C).

Super-Mendelian inheritance mediated by an M-linked homing-based RIDD construct

The RIDD construct was inserted into the doublesex (dsx) locus on Chromosome I. In Culex quinquefasciatus, dsx is located approximately 12-Mb from a candidate ~3-Mb male-determining region situated within a pericentromeric domain and containing the representative Myo-m gene (LOC119771149)⁵⁷ (Fig. 2A). Despite this physical separation, previous studies in culicine mosquitoes indicate that recombination is strongly suppressed near the male-determining locus, resulting in tight functional linkage between the nearby loci and the M chromosome^58,59. This genomic architecture is therefore expected to substantially influence inheritance patterns in the present system.

Fig. 2. Inheritance outcomes and editing profiles of an M-linked homing-based RIDD suppression system targeting the dsx gene in Culex quinquefasciatus.

A Schematic representation of the genomic locations of the candidate male-determining region on chromosome I of Culex quinquefasciatus. The approximate physical distance between dsx and Myo-m, located within the M-locus region, is indicated. The estimated position of the sex-determining (M) locus is shown by a black bar. The diagram is not drawn to scale. B Schematic of the genetic crossing strategy used to assess inheritance bias mediated by the RIDD construct. The drive cassette is inserted within the dsx locus and is tightly linked to the male-determining M-locus, resulting in predominant transmission through the male lineage via physical co-inheritance. Transgenic males carrying the drive construct were crossed with virgin wildtype females to generate G1 heterozygous males, which were then single-pair crossed with virgin wildtype females to produce G2 progeny. Expected G2 phenotypic categories are illustrated below the crossing scheme. The M-linked dsx alleles carrying drive elements are shown in red, homing-converted alleles in purple, and resistance alleles generated through end-joining repair in green. Orange inverted triangles represent potential homing events. ♂ indicates males and ♀ indicates females. C Bubble chart summarizing G2 phenotypic outcomes across 33 independent germlines. Inheritance and phenotypic proportions were estimated using a generalized linear mixed model (GLMM) fitted by maximum likelihood to account for batch effects among independent crosses. Reported values, therefore, represent model-based averaged estimates (mean ± SEM) rather than direct arithmetic proportions and may not sum exactly to 100%. Bubble colors (red, yellow, pink, purple, blue, dark green, and light green) represent the proportions of overall inheritance, drive males, drive intersexes, overall intersexes, non-drive males, non-drive females, and non-drive intersexes, respectively, within G2 progeny from each independent germline. Bubble size corresponds to the number of G2 progeny per cross. Raw phenotypic scoring data are provided in Supplementary Data 3. D Editing outcomes at the dsx-gRNA-1 target site in non-drive individuals (males, females and intersexes) based on deep sequencing. Indel types and their relative frequencies are listed alongside and provided in Supplementary Data 2. Source Data are provided in Source Data 2.

To quantify inheritance dynamics, we initiated experiments by crossing 10 drive-carrying males with 10 wildtype females to generate G1 heterozygous carriers. Approximately 20 single-pair crosses between G1 heterozygous males and wildtype females were established to assess gene drive inheritance in the G2 generation. Two independent experimental batches conducted at different times served as biological replicates, and data were pooled for analysis.

At the adult stage, G2 individuals were screened for DsRed fluorescence together with external morphological phenotypes indicative of distinct gene editing outcomes (Fig. 2B). Because validated genetic sex markers remain limited in Culex quinquefasciatus, initial classification relied primarily on morphological traits. DsRed-positive individuals were scored as drive carriers. Individuals displaying normal male morphology were classified as drive-carrying males (Fig. 2B), consistent with co-inheritance of the construct with the male-determining M chromosome and representing the principal route of the drive transmission. In contrast, DsRed-positive individuals exhibiting masculinized female or intermediate morphological features were classified as drive intersexes (Fig. 2B). These phenotypes are consistent with ectopic expression of the recoded dsxM sequence in genetic females following homing-mediated inheritance of the construct. The genetic identity of these phenotypic classes was subsequently validated by RT-PCR analysis of sex-specific dsx splice variants. Individuals lacking DsRed fluorescence but displaying intersex phenotypes were classified as non-drive intersexes (Fig. 2B), consistent with inheritance of end-joining (EJ)-derived resistance alleles generated during germline repair. Female offspring inheriting unedited alleles exhibited normal wildtype morphology (Fig. 2B). Notably, the recovery of non-drive males suggests rare recombination events took place between the dsx locus and the nearby male-determining region (Fig. 2B).

Inheritance frequencies were estimated using a generalized linear mixed model (GLMM) accounting for batch effects among independent crosses. Nearly all recovered G2 males carried the drive element, with only 1.1 ± 0.3% classified as non-drive males (Fig. 2C), consistent with low recombination rates between dsx and M locus and supporting tight functional linkage in this system. Overall drive inheritance reached 70.7 ± 2.3% measured as the proportion of DsRed-positive individuals among total G2 progeny, exceeding the Mendelian expectation of 50% (Fig. 2C). This included 52.3 ± 1.6% drive-carrying males and 14.5 ± 1.9% drive intersexes. These values represent a notable improvement over our previously reported split gene drive in Culex, which achieved inheritance rates of 55%-60%⁵⁴. A total of 26.4 ± 2.6% intersex individuals were recovered among G2 progeny, comprising both drive intersexes generated through homing-mediated inheritance in genetic females and 9.3 ± 1.6% non-drive intersexes carrying EJ-derived resistance alleles at the dsx locus (Fig. 2C), which are likely to disrupt female sexual development. In addition, 15.1 ± 2.5% non-drive females inheriting unedited alleles were recovered (Fig. 2C). Together, these results indicate that inheritance dynamics are jointly shaped by physical linkage to the M chromosome and CRISPR-mediated editing outcomes during germline repair. The drive allele is transmitted predominantly through the male lineage via co-inheritance with the M-locus, whereas females inheriting either the drive construct or EJ-derived resistance alleles develop masculinized intersex phenotypes associated with substantial fitness costs, as examined in detail below.

To further characterize editing outcomes, we pooled 30 non-drive females, males, and intersexes from G2 progeny (derived from multiple parents) for deep sequencing of the dual-gRNA target region. Unexpectedly, no editing was detected at the dsx-gRNA-2 site (driven by the U6:6 promoter), despite prior validation of U6:6 activity¹⁹, indicating that dsx-gRNA-2 was inactive (Supplementary Data 2). In contrast, variable editing was observed at the dsx-gRNA-1 site (Fig. 2D). Among non-drive intersexes, 48% of alleles were disrupted, consistent with heterozygosity at the dsx locus and supporting the hypothesis that indels at this site impair female development. These observations indicate a partially dominant mutant effect of dsx disruption in Culex, consistent with findings in Drosophila^7,42 and Anopheles⁴⁶, although the underlying mechanism (haploinsufficiency versus dominant-negative effects) cannot be distinguished from the current data. In non-drive males, 7% of alleles carried edits, with a 23-bp deletion being the most common indel. These individuals retained a normal male phenotype, indicating no detectable dominant effects in males. This is consistent with the sex-specific function of dsx, where male development is primarily governed by the dsxM isoform, and suggests that heterozygous disruption at this locus does not impair male differentiation. No detectable edits at or near the target site were observed in non-drive females (Fig. 2D), further highlighting the female-specific sensitivity to dsx disruption in this system. In drive-carrying individuals, edits at the dsx-gRNA-1 site were detected in 5.5% of drive-carrying males and 2.0% of drive-carrying intersexes (Supplementary Fig. 1), likely resulting from somatic Cas9/gRNA expression. Although upstream sex determination mechanisms remain incompletely resolved in Culex, pooled deep sequencing results from non-drive offspring suggest that the inheritance of EJ-induced resistance alleles contributed to partial masculinization in heterozygous females, highlighting the female-specific, partially dominant role of the dsx during sexual differentiation.

Phenotypic and genotypic characterization of offspring

Phenotypic analysis of G2 individuals from single-pair crosses were performed 3–4 days post-eclosion, when adult sexual morphology in Culex is fully developed, to minimize potential age-related variation in external genitalia and secondary sexual traits. Individuals inheriting the RIDD element developed as either morphologically typical males or intersexes. Intersexes carrying the dsxM-recoded drive allele exhibited masculinized features, including male-like plumose antennae, elongated maxillary palps, and malformed genitalia. Two distinct intersex phenotypes were observed: Drive-carrying intersex-I, characterized by laterally twisted claspers, and drive-carrying intersex-II, with backward-rotated claspers (Fig. 3A–c’,d’). In contrast, drive-carrying males were phenotypically indistinguishable from wildtype males. Among non-drive individuals, we observed wildtype males, wildtype females, and intersexes with varying degrees of masculinization. Wildtype individuals displayed clear sexual dimorphism (Fig. 3A-a’,b’). Non-drive intersexes were classified into two categories based on external morphology. Non-drive intersex-I individuals exhibited female-like traits, including pilose antennae, elongated maxillary palps, and cerci-like genitalia (Fig. 3A–e’). Non-drive intersex-II individuals displayed masculinized features resembling those of drive-carrying intersexes, including plumose antennae, abnormal maxillary palps, and malformed claspers (Fig. 3A-f’). Notably, non-drive intersex-II individuals were rarely recovered, suggesting that this phenotype is relatively infrequent.

Fig. 3. Phenotypic and genotypic characterization of offspring types.

A Sexual dimorphism in antenna (Ant.), maxillary palps (MP.), and genitalia (Gen.) structures across various offspring types. a’-wildtype (WT)-male; b’-WT-female; c’-Drive-carrying intersex-I with laterally twisted genital claspers; d’-Drive-carrying intersex-II with backward-rotated claspers; e’-Non-drive intersex-I exhibiting partially feminized features; f’-Non-drive intersex-II with masculinized features resembling those of drive-carrying intersexes. B Schematic representation of dsx splice variants and corresponding protein isoforms detected in drive and non-drive offspring. Exons are not drawn to scale. C Gel electrophoresis of RT-PCR products showing sex-specific dsx transcript expression across offspring types. In Culex, the female splice isoform of dsx retains sequences present in the male transcript (exons 6 and 7), resulting in PCR-2 products of different sizes in males and females. Primer locations are indicated by dashed lines in panel B. The Culex Actin5C gene (CPIJ009808) served as a positive control. Sample types and expected PCR product sizes are labeled next to the gel image. Experiments were independently repeated at least twice with similar results. Source Data are provided in Source Data 3.

Due to the absence of sex-specific genetic markers in Culex quinquefasciatus, the genetic sex of non-drive intersexes could not be determined by standard PCR. To clarify the molecular basis of these phenotypes, we designed sex-specific primers (Supplementary Data 7) targeting distinct dsx splice variants and performed RT-PCR using cDNA templates (Fig. 3B). Drive-carrying males displaying typical male morphology expressed both the endogenous dsxM transcript and the recoded dsxM transcript from the drive construct, confirming their heterozygosity (Fig. 3C). Drive-carrying intersexes expressed both endogenous dsxF and recoded dsxM transcripts, indicating that they were genetic females carrying the drive allele (Fig. 3C). The ectopic expression of dsxM likely contributed to their masculinized intersex phenotype.

Non-drive intersexes expressed only endogenous splice variants. Non-drive intersexes (type-I) expressed dsxF but not dsxM (Fig. 3C), consistent with their more female-like external morphology and suggesting that their alternative splicing remained unaffected. In contrast, non-drive-intersexes (type-II) expressed both dsxF and dsxM, correlating with their strongly masculinized phenotypes that closely resembled those of drive-carrying intersexes (Fig. 3C). Because non-drive-intersexes (type-II) were rare, it remains unclear whether these phenotypes resulted from developmental abnormalities or from resistance alleles affecting dsx splicing. Sanger sequencing of type I and type II non-drive intersexes revealed diverse indel variants at the dsx-gRNA-1 target site (Supplementary Fig. 2). Whether these indels are causally linked to the distinct phenotypes remains to be determined, particularly given the rarity of recovered type II intersexes and the limited understanding of sex-conversion mechanisms in Culex mosquitoes.

Fitness cost evaluation of intersex mosquitoes

To assess potential fitness costs associated with ectopic expression of the recoded dsxM element in drive-carrying intersexes and partially dominant dsx mutations generated via end-joining repair in non-drive intersexes, we evaluated their longevity, mating performance, fertility and egg hatchability. All test individuals were derived from sibling offspring of drive-carrying males crossed with wildtype females to minimize batch and genetic background effects. The following experimental groups were established: (i) 30 drive-carrying males x 30 wildtype (WT) females, (ii) 30 drive-carrying intersexes x 30 WT-males, (iii) 30 non-drive intersexes (type-I) x 30 WT-males, and (iv) a control group of 30 WT-females x 30 WT-males. Mating crosses for non-drive intersexes (type-II) could not be established due to their low recovery frequency. However, given their phenotypic resemblance to drive-carrying intersexes, their fitness was presumed to be comparably reduced. Survival was monitored daily, and blood feeding was performed between days 5 and 6 post-eclosion.

Survival analysis revealed that drive-carrying males exhibited significantly reduced longevity compared to WT-males (log-rank test, P < 0.001), with median survival times of 10.5 and 22 days, respectively (Fig. 4A). Cox proportional hazard analysis indicated a more than two-fold increase in mortality risk of drive-carrying males (Hazard ratio [HR] = 2.1). Drive-carrying intersexes exhibited an even more pronounced reduction in survival compared with WT-females (log-rank test, P < 0.0001), with median survival times of 8.5 versus 27.5 days and substantially elevated mortality risks (HR = 3.6) (Fig. 4A). In contrast, non-drive intersexes (type-I) did not show a significant reduction in longevity compared with WT-females (median survival 25 versus 27.5 days, HR = 0.8), indicating that end-joiningderived dsx disruption alone does not strongly affect adult survival (Fig. 4A).

Fig. 4. Fitness costs associated with drive-carrying and non-drive intersex individuals.

A Kaplan-Meier survival curve for wildtype (WT)-males, WT-females, drive-carrying males, drive-carrying intersexes and non-drive intersexes. Survival differences were evaluated using log-rank tests, with summary statistics, P-values and hazard ratios reported in the table above. Curves in red, blue, purple, yellow and green represent the survival of WT females, WT males, drive-carrying males, drive-carrying intersexes and non-drive intersexes, respectively. B Fecundity and fertility measurements across WT-females, drive-carrying intersexes and non-drive intersexes. Red bars represent the number of blood-fed versus non-blood-fed individuals, while yellow bars indicate the number of blood-fed individuals that successfully oviposited versus those that did not. Oviposition was quantified based on egg raft counts, as each raft is produced by a single fertile individual. C Representative images of dissected ovaries from a fertile WT-female and a non-drive intersex individual exhibiting oviposition failure. D Hatch rates of egg rafts laid by WT-females crossed with WT or drive-carrying males, and by non-drive intersexes crossed with WT-males. Data are presented as the mean ± SEM. Differences between groups were evaluated using one-way ANOVA with Tukey’s multiple comparison test. “ns” indicates no significant difference between WT-females x WT-males and WT-females x drive-males (P = 0.4223), whereas “****” indicates a significant difference between WT-females x WT-males and WT-males x non-drive intersexes (P < 0.0001). Red, purple, and green bubbles indicate hatch rates for WT-females x WT-males, WT-females x drive-males, and non-drive intersexes x WT-males crosses, respectively. Raw counting and analyzed data, along with detailed statistical analysis, are provided in Supplementary Data 5. Source Data are provided in Source Data 4.

To assess mating competitiveness, mixed cages were established containing equal numbers of drive-carrying and WT-males (30 each) together with 30 virgin WT-females. Because Culex females typically deposit a single egg raft after insemination and are generally monandrous, assessment of paternal mating success requires isolation and progeny scoring of individual egg rafts, which substantially limits experimental throughput. The fluorescent phenotype of the offspring from each raft provides a direct indicator of the successful mating of the male. Of 19 egg rafts collected, 9 produced DsRed-positive progeny, indicating successful fertilization by drive-carrying males (Supplementary Data 4). Although the number of egg rafts analyzed limits quantitative assessment of mating competitiveness, these results suggested that drive-carrying males remained capable of successful mating under laboratory conditions despite reduced longevity.

Fecundity and fertility were assessed by monitoring blood feeding and oviposition behavior. In control crosses (WT-females x WT-males), 28 of 30 females blood-fed, and 22 subsequently oviposited (Fig. 4B). In crosses between WT-females and drive-carrying males, all females blood-fed and 21 of 30 oviposited, indicating normal female reproductive performance in the presence of drive-carrying males (Fig. 4B). In contrast, drive-carrying intersex individuals failed to blood-feed and did not oviposit when crossed with WT-males, demonstrating complete functional sterility (Fig. 4B). Among morphologically classified non-drive intersex individuals crossed with WT-males, 26 of 30 individuals were blood-fed, and 13 of these subsequently oviposited (Fig. 4B). However, reproductive performance in these individuals was highly variable and generally reduced compared with WT females. Notably, several blood-fed non-drive intersexes died with undigested blood retained in the abdomens (Supplementary Fig. 3), suggesting impaired physiological processing associated with the intersex phenotype.

Consistent with impaired reproductive performance, dissections of morphologically classified non-drive intersexes that failed to oviposit revealed retained eggs within the abdomen. Many of these eggs appeared undeveloped or melanized, in contrast to the well-developed ovaries and mature eggs observed in fertile WT females (Fig. 4C). Hatchability assays further showed high egg viability from WT-females crossed with either WT-males or drive-carrying males (92% ± 3% and 89% ± 4%, respectively), indicating the preserved fertility of drive-carrying males (Fig. 4D). Together, these findings indicate that both drive-carrying and non-drive intersex phenotypes impose strong reproductive fitness costs, reinforcing the intrinsically self-limiting nature of this RIDD suppression system.

Single release of RIDD males shows extended but self-limiting population suppression in small laboratory cages

Our Culex population suppression system represents a RIDD-type strategy that combines CRISPR-mediated inheritance bias with dominant, sex-specific fitness costs at the dsx locus. Because the RIDD construct is integrated within the dsx region, which is linked to the male-determining M locus in Culex quinquefasciatus, transmission is shaped by the sex linkage. Under this architecture, the RIDD is preferentially transmitted through fertile males via co-inheritance with the M chromosome, whereas females inheriting the construct through the homing process develop into sterile intersexes and do not contribute to subsequent transmission. Thereby, suppression is achieved through the release of transgenic males that transiently bias transmission of the construct and generate sterile intersex individuals, rather than through sustained population-level spread. We assessed the suppression efficacy of this system under laboratory conditions using single-release, small-cage experiments. Although operational RIDD deployment would require repeated releases, a single-release design enables clearer assessment of inheritance dynamics and fitness effects in a confined setting, as all individuals complete their lifecycle within the same closed population following the initial introduction.

Each cage was initiated with 100 WT-females and 100 males, with RIDD-males introduced at release ratios of 1:1 (50 RIDD + 50 WT-males), 1:3 (25 + 75), and 1:9 (10 + 90), with three replicates per ratio. In subsequent generations, 200 pupae were randomly selected to seed the next generation. To monitor population composition, 150 pupae were randomly sampled each generation to measure (i) the proportion of RIDD-males among total males, reflecting persistence of inheritance bias, (ii) the proportion of RIDD-intersexes among total genetic females (WT-females and intersexes), generated by ectopic dsxM expression, and (iii) the proportion of non-drive intersexes carrying EJ-derived resistance alleles among total genetic females, which further reflects the self-limiting nature of this system. Total larval counts were recorded each generation as a proxy for overall population size.

Following a single release, RIDD-males persisted for several generations but were ultimately eliminated from all populations (Fig. 5A). This persistence primarily is consistent with tight linkage between the drive construct and the male-determining M locus, which ensures efficient male transmission even in the absence of population-level spread. As expected, persistence was longer at higher initial release ratios (Fig. 5A). These dynamics were broadly consistent with a discrete-generation model in which RIDD-males had no fitness costs. Notably, in one 1:3 replicate, the RIDD-male frequency fluctuated around the initial seeding level before increasing transiently from generation 4 onward, likely reflecting stochastic effects associated with pupal sampling and phenotypic scoring in small populations (Fig. 5A).

Fig. 5. Population composition dynamics in single-release cage trials over 7 generations.

A Frequency of RIDD-males across generations, calculated as the proportion of RIDD-males among all counted males in each generation. Values at generation 0 represent the initial proportion of RIDD-males at the time of release. B Frequency of RIDD-females (intersexes) generated by ectopic dsxM expression, calculated as the proportion of RIDD-intersexes among all scored genetic females (WT-females plus RIDD-intersexes and non-RIDD intersexes) in each generation. C Frequency of non-drive intersexes carrying EJ-induced resistance alleles, calculated as the proportion of non-drive intersexes among all counted genetic females. Colored curves represent the frequencies for each biological replicate across the 1:1, 1:3 and 1:9 release ratios. Blue, green, and red gradients indicate the 1:1, 1:3 and 1:9 release ratios. Different line styles denote replicates within each release ratio. Light gray curves represent individual simulated trajectories, and the thick black line shows their average. Raw data are provided in Supplementary Data 6. Source Data are provided in Source Data 5.

RIDD-intersexes generated through ectopic dsxM expression showed a rapid increase in frequency following release across all ratios, followed by a gradual decline in subsequent generations (Fig. 5B). This pattern reflects the combination of male transmission and strong female-specific sterility, which together sustain suppressive effects transiently while preventing long-term persistence. Observed frequencies of RIDD-intersexes were lower than model expectations at the 1:1 release ratio, but closely matched predictions at lower release ratios. Non-drive intersexes carrying EJ-induced resistance alleles remained consistently detectable across generations and release ratios (Fig. 5C), indicating continual formation of EJ-induced resistance alleles during cleavage and repair at the dsx locus (Fig. 5C). These alleles accumulated at a level somewhat higher than predicted by the model and could only be removed through female-specific fitness effects. Across all release ratios, total population size gradually declined over successive generations (Supplementary Fig. 4). Although multiple ecological factors inherent to small-cage conditions may contribute to these dynamics, the sustained production of sterile or partially sterile intersex individuals likely reduced reproductive output and contributed to suppression. Overall, the relatively decent consistency between our models and the cage experiments suggests that the major features of system behavior are captured by the current framework, and indicates that there are no significant unexpected fitness costs that may substantially inhibit performance in cages compared to individual crosses. Together, these results demonstrate that a single release of RIDD-males can produce extended but ultimately self-limiting suppression effects.

Modeling performance of population suppression in Culex quinquefasciatus

To evaluate the actual effectiveness of this system under repeated-release scenarios, we simulated multiple releases of RIDD-males at varying release ratios and monitored the number of fertile females (a key proxy for biting rates and disease transmission potential) on a weekly basis.

The release ratio refers to the weekly number of released mosquito males per generation (one generation = 3.17 weeks under normal circumstances), expressed as a fraction of the number of wildtype males present when the population is at carrying capacity. Modeling results show that although the RIDD system exhibits only moderate suppression strength, it remains effective in populations with low intrinsic growth rates (Fig. 6A). This contrasts with previous models and experimental demonstrations^7,30, largely due to the relatively modest total cut rate (drive conversion plus germline resistance formation) in our Culex drive, which leaves a substantial proportion of intact wildtype alleles. In addition, resistance alleles exhibit incomplete dominant sterility, further limiting suppression efficiency. Although RIDD-males can eventually dominate the male population, the persistence of wildtype and partially fertile resistance alleles imposes a significant constraint on full population elimination³⁰.

Fig. 6. Modeling the RIDD suppression gene drive in Culex quinquefasciatus.

Simulations were conducted by continuously releasing engineered males into a population containing 50,000 adult females at varying release ratios. The release ratio is defined as the number of engineered males per generation (3.17 weeks) relative to the size of the wildtype adult male population. A RIDD-males were released into populations with varying growth rates at low population density and a release ratio of 4. Purple, blue, dark green, light green, and yellow represent intrinsic growth rates of 2, 4, 6, 8, and 10 under low-density conditions, respectively. B Comparative performance of different genetic control strategies. RIDD+ is a hypothetical enhanced version of RIDD with an increased germline resistance rate (0.527 vs. 0.37). Blue, yellow, pink, and red colors represent the performance of SIT/IIT, fsRIDL, RIDD, and RIDD+ strategies, respectively. Different line styles indicate performance at release ratios of 1, 4 and 8. C Suppression dynamics following a 20-week release of engineered males at a release ratio of 4. Each data point represents the average of 10 replicates.

We next compared the performance of RIDD with other population suppression strategies, including the sterile insect technique (SIT) and female-specific RIDL (fs-RIDL)⁶⁰. We also modeled a hypothetical RIDD+ drive with an increased germline cut rate (though still below the near-100% efficiency achieved in other species^54,61,62) and complete dominant sterility conferred by resistance alleles. Our results indicate that under low release ratios, both RIDD and RIDD+ systems outperform SIT and fs-RIDL due to their greater persistence (Fig. 6B). Notably, SIT is particularly vulnerable to overcompensation at low release ratios, which actually increases the population size. At higher release ratios, the advantage of the RIDD system diminishes, and the population still declines to nearly the same intermediate level. However, this limitation can be mitigated by increasing the level of germline cutting rates (as in RIDD+), enabling full population collapse even at low release ratios. Importantly, all these systems remain self-limited. When releases are conducted for only 20 weeks with a release ratio of four, both RIDD+ and RIDD systems suppress population rapidly, like fs-RIDL (Fig. 6C). When releases cease, the population size returns to its original level more slowly compared to other control strategies. While suppression is weakened due to linkage with the M locus, it is sustained over a longer period.

Discussion

In this study, we established a proof-of-principle CRISPR-based self-limiting population suppression gene drive in the non-model mosquito Culex quinquefasciatus, a major vector of multiple human and animal diseases. The construct incorporates a homing mechanism capable of biasing inheritance through germline conversion at the individual level, yet is intentionally designed not to increase in frequency at the population level. By targeting a conserved region of the dsx gene shared by both male- and female- specific isoforms and incorporating a recoded dsxM sequence that preserves male function, our design enables the drive to be propagated exclusively by fertile males through co-inheritance with the M chromosome, while females inheriting either the construct via HDR-homing or nonfunctional alleles via EJ-repair are converted into completely or partially sterile intersexes, ensuring intrinsic self-limiting and eventual loss of construct following release cessation. A similar strategy lacking dominant resistance has been successfully developed in Anopheles stephensi¹³, and another example, termed Male Drive Female Sterile (MDFS), has recently been developed in Anopheles gambiae, showing the ability to eliminate caged populations following repeated releases⁴⁶. Together, these systems illustrate a broader design space of suppression architectures operating in highly self-limiting parameter regimes, where inheritance bias enhances suppression efficiency while fitness costs prevent invasion.

Remarkably, we found that heterozygous disruption of dsx in Culex females, either by inheritance of the recoded construct or end-joining (EJ)-derived resistance alleles, frequently produces sterile or partially sterile intersexes with substantial dominant fertility costs. This indicates that disruption of dsx results in partially dominant female sterility phenotypes in Culex, consistent with observations in Drosophila melanogaster⁷, Drosophila suzukii⁴² and Anopheles gambiae⁴⁶, but contrasting with reports from other Anopheles species where disruption of female-specific dsx exon or splice sites generally yields recessive effects^13,15. Recent work in Anopheles stephensi using a split drive targeting dsx similarly reported substantially reduced fertility in trans-heterozygous females, underscoring the complexity and species-specific nature of dsx function⁴³. These differences highlight the challenges of directly translating suppression strategies between species. A deeper mechanistic understanding of dsx alternative splicing in mosquitoes, along with exploration of additional conserved regulators within the sex determination pathway, will be essential for developing broadly applicable population suppression strategies across diverse vector species.

Compared with our previously tested split gene drive system in Culex, which achieved inheritance rates of 55%–60%⁵⁴, the present system reached up to ~71% super-Mendelian transmission. This improvement may be attributable, in part, to the use of a different germline nanos promoter, which has performed well in Anopheles gambiae gene drives. However, position effects, target-locus- and gRNA-dependent effects may also have contributed. Although a dual-gRNA strategy was employed, only one gRNA (dsx-1) proved functional. The inactivity of dsx-2 is unlikely due to the U6:6 promoter, which has demonstrated moderate editing activity in prior studies^19,54. Instead, suboptimal target site selection is the probable cause. Notably, the homology arms were designed assuming both gRNAs would be functional. Since only dsx-gRNA-1 induced cleavage, the resulting mismatch between the homology arms and actual cut site may have contributed to the reduced homing efficiency⁶³. Further optimization of gRNA design would likely enhance drive performance. Even a simplified design using only dsx-gRNA-1, with homology arms precisely matching its cut site, should yield at least somewhat higher drive conversion rates. In addition, exploration of alternative germline-specific promoters, like zpg, may further boost drive efficiency.

The limited understanding of the primary sex-determining signals and downstream regulatory pathways in Culex mosquitoes complicates efforts to genetically manipulate or interfere with sex determination⁵⁶. To investigate the molecular basis of intersex formation in our system, we performed RT-PCR analysis of dsx splice variants to determine the genetic sex and assess splicing patterns across different intersex phenotypes. Drive-carrying intersexes expressed both endogenous dsxF and recoded dsxM, consistent with ectopic male isoform expression in genetic females. In contrast, EJ-derived non-drive intersexes formed two distinct phenotypic classes: Type-I individuals expressed only dsxF and retained more female-like morphology, whereas Type-II individuals expressed both isoforms and closely resembled drive-carrying intersexes. Sanger sequencing showed that these phenotypic differences may be associated with distinct indel variants, although this remains inconclusive due to the limited number of Type-II intersexes analyzed. Whether the occurrence of Type-II intersexes results from aberrant sexual development or altered splicing events requires further investigation with a larger sample size. While Type-I intersexes retained blood-feeding ability, their substantially reduced fertility and hatchability still contributed to population suppression. This phenotype mirrors observations from a suppression gene drive in Drosophila suzukii, where hemizygous females exhibited normal mating behavior but markedly reduced fertility⁴².

Single-release cage trials showed that RIDD-males persist for several generations but are ultimately eliminated, consistent with biased inheritance and M-linkage, together with strong female-specific fitness costs primarily driven by the intended sterility effect of the drive, with additional contribution from resistance-derived mutations. Although practical RIDD deployment would rely on multiple releases, the single-release design enables a clearer assessment of transgenic mosquito fitness, as all individuals complete their lifecycle within the same confined environment following the initial release. Importantly, these dynamics confirm that the system does not spread through populations, even under favorable laboratory conditions. Population suppression instead arises from the cumulative effects of sterile drive-carrying intersexes and non-drive intersexes harboring partially dominant dsx resistance alleles, which together reduce reproductive output. Future cage trials incorporating improved total cleavage rates and repeated release strategies, similar to previous successful trials of RIDD in Drosophila⁷ and the recently developed MDFS system in Anopheles gambiae⁴⁶, deserve to be explored to fully assess the suppression potential of RIDD systems in Culex.

Modeling results suggest that complete population suppression is not achievable with the current RIDD system unless the population has a very low intrinsic growth rate, which is unlikely except perhaps in marginal habitats. However, success may be attainable with further modest improvements. The RIDD system performs with high effectiveness at low release ratios or when releases are limited to a short period, highlighting its potential to achieve substantial suppression even with a limited number of engineered individuals. To enhance efficacy at higher release ratios, improved Cas9 or gRNA expression is needed to approach near 100% cutting and ensure dominant sterility of resistance alleles. Of note, while the M-linkage allows greater persistence of the drive and thus far lower release ratios to achieve maximum suppressive power, it also limits the power compared to standard RIDD^7,45. This is because the drive system will be in males a greater fraction of the time, allowing higher drive persistence, but also reducing the fraction of female offspring that are sterile. Greater cleavage rates (either leading to drive conversion or dominant resistance allele formation) could rectify this, allowing substantial benefits to be realized from the M-linked locus in terms of reduced release sizes.

A critical consideration for safety is the potential for partial loss or attenuation of female fitness costs, which may shift the system toward invasive behavior. Although our data indicate strong and consistent female sterility, any further deployment would require testing to ensure that some mutations (such as loss of the recoded dsxM sequence) do not allow drive spread through females, which could convert it to a self-sustaining drive, depending on exact performance and fitness costs.

From an operational perspective, production of RIDD males requires an additional outcrossing step to generate appropriately configured release individuals, which may introduce extra costs compared to strategies like fs-RIDL, where engineered males can be produced more directly³⁰. Nonetheless, the outcrossing step also facilitates introgression of desired genetic backgrounds in the released males, an advantage shared with other homing-type drives that can be more easily adapted to different genetic backgrounds. Of note, the linkage between the drive and the male-determining locus contributes strongly to its performance. It allows greater persistence of the drive (and thus lower release ratios to achieve maximum effect) because the lack of recombination ensures that the original drive allele is predominantly inherited by male progeny, sustaining the drive across generations. However, it also limits the overall power of the drive, as more females inherit wildtype alleles and resistance alleles that allow partial fertility. If germline cutting rates can be increased while maintaining low levels of resistance allele formation, most female progeny would become sterile, thereby preserving the drive’s benefits. Under these conditions, the system would function similarly to Y-linked editors (YLE) targeting dominant genes, achieving effective population elimination even when drive conversion is low and enabling successful population elimination with very low release sizes^30,64.

In summary, we have developed a self-limiting, population-suppression gene drive targeting dsx in Culex quinquefasciatus. Although this system does not autonomously spread through populations as in the classical sense of self-sustaining gene drives, it demonstrates how homing-based mechanisms can be repurposed to enhance suppression efficiency while maintaining intrinsic confinement. This work expands the genetic toolkit available for Culex control and provides a conceptual framework for developing precise, sustainable, and ecologically responsible genetic strategies for mosquito and pest management.

Methods

Culex mosquito rearing and maintenance

The Culex quinquefasciatus (Beijing strain, China) strain was kindly provided by Dr. Jianying Liu at Shenzhen Bay Laboratory. Mosquitoes were reared at 27 ± 1 °C, 75% relative humidity, under a 12-h light/dark cycle in a Biosafety Level 2 (BSL-2) insectary room. Adults were maintained on a 10% sucrose solution. After mating, females were blood-fed on mice, and egg rafts were collected four days post-blood feeding. Larvae were fed with fish food pellets (Aquafin, China). Fluorescence and phenotypic assessments were performed using the SZx2-ZB16 fluorescent microscope (Olympus, Tokyo, Japan). All procedures were carried out in accordance with protocols approved by the Institutional Biosafety Committee of Shenzhen Bay Laboratory and followed relevant ethical guidelines for animal research. Wastewater and used containers were frozen at −20 °C for 48 h before disposal as biohazardous waste.

Plasmid construction

All plasmids used in this study were generated using standard molecular biology techniques. Genomic DNA was extracted from approximately 10 adult wildtype Culex quinquefasciatus mosquitoes using the Animal Tissue DNA Extraction kit (Vazyme, #DC102-01, China). The dsx gene was amplified and TA-cloned to serve as the plasmid backbone. Two dsx-targeting gRNA fragments (dsx-gRNA-1 and dsx-gRNA-2) were designed based on the dsx target region and synthesized as complementary oligonucleotides (GenScript Biotech, China). The annealed oligonucleotides were inserted into the double-BbsI sites of the Cq-U6_1_2XBbsI-gRNA plasmid (Addgene #169238) and the Cq-U6_6_2XBbsI-gRNA plasmid (Addgene #169323) plasmids, respectively. The gRNA scaffold was modified to incorporate a 5-bp loop structure and a T-C polymorphism to enhance transgenesis. These modifications are consistent with those described in our previous studies¹⁹. The Cq-U6_1_dsx-gRNA-1 and Cq-U6_6_dsx-gRNA-2 cassettes were amplified and cloned into the dsx homology arm backbones flanking the gRNA cut sites. The nanos-Cas9 component was amplified from the Cq-nanos-Cas9 plasmid (Addgene #169348). All plasmids were assembled using Gibson Assembly (NEBuilder HiFi DNA Assembly Master Mix, New England Biolabs, #E2621). Following the transformation into E. coli DH5a chemically competent cells (Sangon Biotech, #B528413, China), positive clones were verified by restriction enzyme digestion and Sanger sequencing. All oligonucleotide sequences (including primers and gRNAs) are provided in Supplementary Data 1. All plasmid sequences generated in this study are available in GenBank (see Data Availability).

Fitness cost evaluation

Various mating crosses were established among different offspring categories. Survival, blood-feeding success, oviposition, and egg hatch rates were monitored daily. For non-drive intersex individuals with oviposition difficulties, abdominal dissections and imaging were performed using a microscope (SZMN, China) to assess reproductive anatomy and detect retained or abnormal eggs.

Genotyping and RT-PCR

Genomic DNA (gDNA) was extracted from mosquito legs using STE squishing buffer (Solarbio, China) supplemented with proteinase K (50:1 ratio), followed by a PCR-based digestion cycle (65 °C for 45 min, 95 °C for 15 min, and held at 16 °C). The extracted gDNA was used as a template for PCR amplification with primers designed to cover dsx-gRNA-1 and dsx-gRNA-2 target regions. PCR products were analyzed by Sanger sequencing to detect end-joining induced resistance alleles, and indel variants were quantified using the ICE CRISPR analysis tool⁶⁵. For RT-PCR analysis, total RNA was extracted from adult mosquitoes with distinct phenotypes using the phenol-chloroform method⁶⁶. First-strand cDNA synthesis was performed using a commercial cDNA synthesis kit (Vazyme, #R312-02, China). RT-PCR was conducted with gene-specific primers, and amplification products were visualized by agarose gel electrophoresis. All primers used are listed in Supplementary Data 7.

Amplicon sequencing

Approximately 30 individuals from each G2 offspring category with distinct phenotypes were collected for analysis. Genomic DNA was extracted using the Animal Tissue DNA extraction Kit (Vazyme, #DC102-01, China). PCR amplification targeting the gRNA editing sites was performed, and products were purified using a gel DNA extraction kit (Vazyme, #DC301-01, China) in preparation for deep sequencing (conducted by Qingke Biotechnology). Sequencing data were analyzed using CRISPResso2 (version: 2.2.12) pipeline⁶⁷. Primer sequences used for amplicon sequencing are listed in Supplementary Data 7.

Single-release cage trial

Each cage was initiated with 100 wildtype females and 100 males, with RIDD-males introduced at release ratios of 1:1 (50-RIDD + 50 WT-males), 1:3 (25 + 75), and 1:9 (10 + 90), with three biological replicates per ratio. After eclosion, mosquitoes in each cage were allowed to mate freely for 5 days, followed by blood feeding. Three days post blood feeding, oviposition cups were placed inside cages, and egg rafts were collected 24 h later. Egg rafts from each cage were pooled into a large bowl for hatching, and the hatched larvae were distributed into 1–2 rearing trays. Upon pupation, 200 pupae were randomly selected to seed the next generation, while an additional 150 pupae were sampled for scoring the inheritance and phenotypic outcomes. The remaining individuals in each generation were counted to estimate the total population size.

Statistics and reproducibility

Inheritance ratio and hatch ratio analyses were performed in R (version: 4.5.0) using generalized linear mixed models (GLMMs) to account for batch effects among independent crosses. Models were fitted by maximum likelihood estimation using adaptive Gauss-Hermite quadrature (nAGQ = 25), allowing variance between cross batches to be incorporated as a random effect¹³. No statistical method was used to predetermine sample size. Sample sizes were selected based on standard practice in the field and consistency with previous gene drive studies. No data were excluded from the analyses. The experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment. For cage experiments, each condition was performed with 3 independent biological replicates as indicated in the figure legends. Survival analyses were conducted using Kaplan-Meier estimates and compared using the log-rank test. Differences in fecundity, oviposition, and hatch rates were evaluated using one-way ANOVA followed by Tukey’s multiple comparison test, as specified in the corresponding figure legends. Exact P-values are reported where applicable. Sequencing analyses were performed on targeted amplicon datasets, and all data are provided in the Source Data and Supplementary Data files.

Graphical representation

All figures were generated using GraphPad Prism 10 and Adobe Illustrator (2025). Mosquito images were captured using a VHX-7000C imaging system (KEYENCE), SZX2-ZB16 (OLYMPUS) and SZMN (SOPTOP). Inheritance ratio and hatch ratio analyses were performed in R using generalized linear mixed models (GLMMs) to account for batch effects among independent crosses. Models were fitted by maximum likelihood estimation using adaptive Gauss-Hermite quadrature (nAGQ=25), allowing variance between cross batches to be incorporated as a random effect¹³. No statistical method was used to predetermine sample size, and no data were excluded from the analyses. Hatchability comparisons among different groups were conducted using one-way ANOVA with Tukey’s multiple comparison test. Survival curves among different groups were statistically compared by the log-rank test.

Modeling of the Culex quinquefasciatus RIDD

Since most mosquito vectors share broadly similar life cycles, the effectiveness of releasing RIDD mosquitoes into wild populations was evaluated using a previously developed mosquito-specific model with weekly time steps^68,69 implemented in the forward genetic simulation software SLiM 4⁷⁰. Female mosquitoes typically mate only once and store sperm for subsequent egg production. In this model, we incorporated a 5% probability per week that a female may remate, allowing for the possibility of additional mating beyond the initial one. Females mate randomly with any fertile male in the population. Each week, each mated female has a 50% possibility of producing offspring. The number of eggs laid by a mated female follows a Poisson distribution with an average of 50. However, this average is multiplied by the female fertility fitness parameter, which depends on her genotype (Table 1).

Table 1.

Default parameters for the model

Drive system	Parameter	Default value
RIDD only	Drive conversion rate	0.414
	Germline resistance formation rate	0.37
	Recombination rate (drive and M locus)	0.01876195
	Female fertility with resistance	0.31*0.62
RIDD+	Germline resistance formation rate	0.9
	Female fertility with resistance	0
All models	Low density growth rate	6
	Number of adult females	50000

In our model, the first two weeks for an individual represent the juvenile stage of mosquito development. During this period, the newly hatched larvae experience density-dependent competition, which is modeled according to the following equations:

$$expectedcompetition=k\times \left(F+2\times 0.285714\times 5\right)$$ 1

$$competitionratio\left(\gamma \right)=\left(N0+5N1\right)÷\left(expectedcompetition\right)$$ 2

$$survivalrate=2\times 0.285714\times {\beta }^{\prime }\frac{\gamma ,\beta }{F},$$ 3

where k is the number of adult females in the population at the carrying capacity, F is the expected number of offspring produced by each adult female, N₀ and N₁ represent the number of new larva (age 0) and older larva (age 1) separately, β’(r, β) represents a linearly decreasing function of r, where β denotes the intrinsic growth rate under low-density conditions. Adult mosquitoes in the model are subject to an age-dependent survival rate. For adult males, the weekly survival probabilities are [2/3,1/2,0], indicating a maximum lifespan of three weeks. For adult females, the survival probabilities are [5/6, 4/5, 3/4, 2/3, 1/2, 0], allowing them to live for up to six weeks.

In an alternate model designed to match cage populations, a discrete-generation framework was also implemented. Here, the carrying capacity (100 by default because the effective population size in the cages is likely much lower than the census size) represents the total population of males and females. Each fertile female randomly selects a male and then produces a number of offspring drawn from a binomial distribution with a maximum of 50 and an average of 2 multiplied by her fertility fitness (Table 1) and density fitness (which is the total population / carrying capacity).

To implement drive activity, we adjusted standard Mendelian inheritance in the progeny of drive/wild-type heterozygous males. If the offspring receives a wild-type allele from their male parent, it has a chance of being converted to a drive or resistance allele (which actually takes place in the male parent’s germline) at probabilities equal to the drive conversion rate and germline resistance allele formation rate, respectively (Table 1). We also modeled the male-determination locus, which can undergo recombination with the drive allele at a fixed rate (Table 1). Released individuals have the drive on the same chromosome as the M locus.

We implemented multiple releases of drive-carrying male heterozygotes for the RIDD system, in which ideally, both nonfunctional resistance alleles and the drive allele exhibit dominant female sterility⁷. However, to match experimental observations, females with nonfunctional resistance alleles can lay eggs at a rate of 31%, and if they do lay eggs, they will average 62% as many as a wildtype mosquito. We also modeled a variant that assumed dominant sterility of nonfunctional resistance alleles. For comparison, we also simulated population suppression strategies based on the Sterile Insect Technique (SIT) and the release of males homozygous for a female-specific RIDL (fs-RIDL) construct. In the SIT strategy, females that mate with released sterile males produce no viable offspring. In the fs-RIDL strategy, the construct carries a dominant lethal gene that causes female offspring to be nonviable. This occurs at the early larval stage before competition takes place (based on an existing construct⁶⁰). No fitness costs were modeled, other than the intended costs of the drive.

Simulations were conducted for up to 317 weeks (approximately 100 generations) unless the number of fertile females reached zero earlier. To establish a baseline equilibrium, wildtype mosquitoes were allowed to reproduce randomly for 10 weeks prior to the introduction of drive-carrying mosquitoes. Given that functional resistance can potentially be avoided through the use of multiple gRNAs or by targeting a highly conserved site^15,63,71, our model included only nonfunctional resistance alleles (r2).

Simulations were conducted using the High-Performance Computing Platform at the Center for Life Sciences, Peking University. Data processing and figure generation were performed using Python. Default parameters for all models are listed in Table 1, and the corresponding SLiM scripts and raw data are available on GitHub (https://github.com/jchamper/Culex-RIDD).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

X.F. and F.L. conceived the project. X.F. and V.M.G. designed the experiments. X.F. and J.D. performed the experiments and collected and analyzed the data. Y.L. performed the modeling and data analysis. X.F., F.L., J.D., Y.L., V.L.D.A., X.C. and J.C. wrote the manuscript. All authors reviewed and approved the manuscript.

Peer review

Peer review information

Nature Communications thanks Philippos Papathanos and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

X.F. discloses support for the research of this work from the National Natural Science Foundation of China [grant number 82202559], Zhejiang Province Science Foundation [grant number MS25C140016]. F.L. discloses support for publication of this work from the National Natural Science Foundation of China [grant number 82372289]. J.C. discloses support for the research of this work from the National Natural Science Foundation of China [grant number 32270672].

Data availability

The plasmid sequence generated in this study has been deposited in the GenBank database under accession code PV864762. All data supporting the findings of this study are provided in the Source Data and Supplementary Data files. Source data are provided with this paper. Physical plasmid materials generated in this study are available from the corresponding authors (X.F. and F.L.) upon reasonable request. These materials cannot be fully represented through sequence deposition alone. Requests for academic research purposes will be fulfilled without restriction, and materials will be provided within a reasonable timeframe following the request. Source data are provided with this paper.

Code availability

The code used for modeling is publicly available on GitHub [https://github.com/jchamper/Culex-RIDD], with a permanent DOI version hosted on Zenodo (10.5281/zenodo.19563797). There are no restrictions on access to the code.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-73641-9.