Impact of naturally occurring target-site polymorphisms on an autonomous Cas9/guide RNA-based gene drive system for population modification of Anopheles gambiae s.l.

Mosquito lines and maintenance

Seven mosquito lines were utilized for these experiments, with 3 lines derived from an An. gambiae Ndokayo strain (referred to as AgNdo), 1 transgenic line, AgTP13, derived from the An. gambiae X1 line (a derivative of the G3 strain, Malaria Research and Reference Reagent Resource Center [MR4] strain MRA-112, BEI Resources), 2 lines (hemi- and homozygous) produced from hybridizations of an AgNdo line with the AgTP13 line, and the Agcd^Δ11 line derived from the G3 strain that contains an 11-nucleotide deletion within the cardinal target site (strain MRA-1278, MR4, Volohonsky et al. 2015; Carballar-Lejarazú et al. 2022, 2023; BEI Resources 2024). The TP13 drive system has 2 engineered anti-Plasmodium falciparum effector genes, m1C3 and m2A10, encoding single-chain antibodies (scFvs) that target the ookinete and sporozoites stages, respectively, of the P. falciparum life cycle, regulated, respectively, by the bloodmeal-inducible, tissue-specific control DNA of the An. gambiae Carboxypeptidase A (AgCPA, [Cp]) and An. stephensi Vitellogenin 1 (AsVg1, [Vg]) genes (Isaacs et al. 2011, 2012; Carballar-Lejarazú et al. 2023). An An. gambiae strain from Burkina Faso was used in the survey of SNP variants in geographically diverse colonies.

The 3 AgNdo-derived lines include 1 line, AgNdo-WT, homozygous for a wild-type cardinal target site identical to the one targeted by the AgTP13 gRNA, 1 line, AgNdo-H1, homozygous for a haplotype containing an A→T transversion within the target site (2R:38748898) along with 3 additional SNPs (G→A at 2R:38748902, G→A at 2R:38748910, and C→T at 2R:38748916) distal to the 3′-end of the target site and another line, AgNdo-H2, homozygous for the 2R:38748898 transversion along with an additional SNP (A→C at 2R:38748865) 14 base pairs (bp) from the 5′-end of the target site (Fig. 1). The AgTP13 line is homozygous for a Cas9/gRNA drive system targeting the cardinal gene (gRNA Sequence: 5′-GGTTAGCGACGATGCCAAGGCGG-3′), a cyan fluorescent protein (CFP) dominant marker gene and the m1C3 and m2A10 scFvs.

Generation of AgNdo-WT, AgNdo-H1, and AgNdo-H2 isolines

The AgNdo-WT, AgNdo-H1, and AgNdo-H2 lines were derived from a colony of the AgNdo mosquitoes, a mixed-allele strain carrying various frequencies of the wild-type, haplotype 1, and haplotype 2 alleles. At the time the AgNdo colony was used to generate the isolines, the frequency of alleles was 57.4% wild-type with 14.7% haplotype 1 and 27.9% haplotype 2 (n = 68) (Supplementary Table 1). Homozygous isolines were produced using family pools created by crossing 1 AgNdo male with ∼10 AgNdo females in a 16-oz (∼473 mL) cup. The families were allowed to mate for 3 to 4 d and assigned a number to keep track of the contributing male parent. After mating, males were removed from each family and sequenced individually to determine their genotype. Mated females from male parents homozygous for either the wild-type, haplotype 1, or haplotype 2 alleles were allowed to blood-feed for 2 consecutive days, and engorged females were added to a small cup to lay eggs. Ovipositing females were assigned a second number in addition to their male family number to track their contributing female parent. Females that laid eggs were subsequently sequenced to determine their genotypes. Any progeny that had 2 parents homozygous for the same allele were reared from larvae to adulthood, and ≥50 F1 mosquitoes were sequenced to verify that they were homozygous for the same target-site sequence as their male and female parents. This protocol was repeated on 2 separate occasions for the AgNdo-H1 line, with the first pooled mating generating 2 separate isofamilies that were homozygous for haplotype 1. These 2 isofamilies were subsequently combined, but sequencing of their progeny revealed that the combined families had some haplotype 2 alleles. The mating protocol was performed once more to produce the final homozygous haplotype 1 isofamily. The mating protocol for the AgNdo-H2 line was repeated multiple times to generate a homozygous line. The first pooled mating generated an isofamily that was heterozygous for haplotype 1 and haplotype 2. This isofamily mating protocol was repeated and produced 5 isofamilies that were homozygous for haplotype 2. These 5 isofamilies were combined to create a haplotype 2 line. However, sequencing of the progeny revealed some haplotype 1 alleles. Therefore, the mating protocol was repeated a third time to produce an isofamily homozygous for haplotype 2. Lastly, the mating protocol was performed only once for the AgNdo-WT line, with 1 isofamily being generated that was homozygous for the wild-type target site. All progeny from this isofamily were verified as homozygous wild-type by sequencing the target site.

Molecular analysis of individual mosquitoes

Genomic DNA from the various lines was extracted and the target site amplified using the Extract-N-Amp Tissue PCR Kit (Sigma-Aldrich, USA) and the primer pair Ak5/Ak21 to yield a predicted product amplicon 526 bp in length (Supplementary Table 2; Carballar-Lejarazú et al. 2023). The amplified DNA was resolved in a 1% agarose gel, and the DNA was extracted from excised samples recovered using the DNA Clean & Concentrator kit (Zymo Research, USA). Cleaned DNA was sent to Azenta for Sanger sequencing. The DNA sequence of each sample was compared to a reference sequence (AGAP003502, An. gambiae PEST [AgamP4]; https://vectorbase.org/) using Snapgene bioinformatics software. Genotypes of each sample were determined by observing the chromatogram data that displays a different color for each unique nucleotide. DNA samples from mosquitoes containing the drive construct were validated using the same process with the primer pair Ak5/TP217, which amplify the drive allele with a predicted product amplicon 1,021 bp in length (Supplementary Table 2; Carballar-Lejarazú et al 2023).

SNP frequency analysis of wild An. gambiae field samples and recently colonized laboratory strains

DNA sequence data from wild An. gambiae samples were accessed using the MalariaGen Ag3 cloud data via Google Cloud. Precise information on the cardinal SNP frequency, 2R:38748898, in wild An. gambiae was obtained using the “snp_allele_freqs_df=ag3.snp_allele_frequencies(transcript=, cohorts=, drop_invariant=, and sample_sets=)” function, specifying SNP frequencies in transcript: “AGAP003502-RA,” cohorts: “admin2_year,” drop_invariant: “false,” and sample_sets: “1315-VO-NG-OMITOLA-OMOL-NG-2008, 1288-VO-UG-DONNELLY-VMF00219, and 1281-VO-CM-CHRISTOPHE-VMF00208.” The cohort: admin2_year parameter groups samples based on region, collection year, and taxon. The sample sets: “1315-VO-NG-OMITOLA-OMOL-NG-2008,” “1288-VO-UG-DONNELLY-VMF00219,” and “1281-VO-CM-CHRISTOPHE-VMF00208” parameters contain An. gambiae sequence data from Nigeria, Uganda, and Cameroon, respectively. The data frame was organized to display only the SNP frequency by using the function: “frequency_columns = [col for col in snp_allele_freqs_df.columns if col.startswith(“frq_”)].” The data frame then was generated to display SNP frequency only at the desired position within the cardinal target site by using the function “snp_allele_freqs_df.loc[(“2R”, 38_748_898), frequency_columns].”

The 2R:38748898 cardinal SNP frequency data for each year and region were displayed on a data frame and compared to the overall sample metadata that displays detailed information on the total number of samples collected in each year and the exact collection spot for each sample set. The sample metadata were displayed using the function ag3.sample_metadata(sample_sets=“ ”) with sample sets: “3.8” for 1315-VO-NG-OMITOLA-OMOL-NG-2008 and 1288-VO-UG-DONNELLY-VMF00219 and sample sets: “3.7” for 1281-VO-CM-CHRISTOPHE-VMF00208. A weighted average SNP frequency for a specified time period was calculated as follows: (i) a weighted yearly SNP frequency was calculated for each location by the following equation (yearly SNP frequency) × (the number of samples from that year/the total number of samples from every year included in analysis) and (ii) the weighted SNP frequency for each year was summed and divided by the number of input values to generate a weighted average SNP frequency. The SNP frequencies in the recently-colonized laboratory strains were determined using the same Sanger sequencing protocol described above with the allele type data of >50 mosquitoes from the first-generation cohorts received by the James laboratory. These mosquitoes included the An. gambiae Ndokayo strain received from (MR4, BEI Resources) and colonized from “Ndokayo” Cameroon (2008), the An. gambiae Burkina Faso strain received from Dr. Abdoulaye Diabaté (Professor Roch K. Dabire and the Institut de Recherche en Sciences de la Santé (IRSS), Direction Régionale de l’Ouest) and colonized from “Valleé de Kou” Burkina Faso (2021), and the An. coluzzii STP Ana strain received from the UC Malaria Initiative and colonized from São Tomé and Príncipe (2023).

Cas9 in vitro cleavage assay

A total of 5 plasmids with different target-site sequences were created for use in the in vitro cleavage assay involving Streptococcus pyogenes Cas9 endonuclease (Supplementary Table 3). These plasmids were generated using amplified sequences from the AgNdo-WT, AgNdo-H1, AgNdo-H2, and Agcd^Δ11 lines, all of which include the cardinal target site plus a 36-bp flanking sequence adjacent to the 5′-end from the target site and a 329-bp flanking sequence at the 3′-end of the target site. The Agcd^Δ11 line target site serves as a cleavage-negative control (Carballar-Lejarazú et al. 2022). Insert sequences used for all plasmids were extracted from mosquito samples using Extract-N-Amp Tissue PCR Kit (Sigma-Aldrich). All insert fragments were amplified using CloneAmp HiFi PCR premix (Takara Bio). The insert sequence for plasmids pBM1 to pBM4 were amplified as 1 fragment using the CloneAmp HiFi PCR premix (Takara Bio), with primer pair BM1/BM2 (Supplementary Table 2). The pBM5 insert sequence was amplified as 2 fragments using DNA from a AgNdo-WT mosquito sample. The first fragment includes the 94-bp 5′-end upstream flanking sequence and the target site with no PAM amplified with primer pair BM4/BM2. The second fragment (368 bp), amplified with primer pair BM1/BM3, also contains the target site with no PAM along with the 3′-end downstream flanking sequence. This fragment also was amplified using the CloneAmp HiFi PCR premix (Takara Bio), forward primer BM1, and reverse primer BM3 (368 bp). Amplified inserts were resolved in 1% agarose gels and cleaned using the Zymo Gel DNA Recovery Kit. Following the amplification and isolation of the inserts, all fragments were cloned into linearized pUC19 vectors via the In-Fusion cloning method (Takara Bio) at a molar ratio of 3:1 insert-to-vector for pBM1 to pBM4 or 3:3:1 inserts-to-vector for pBM5. Successfully transformed plasmids were extracted and cleaned using QIAprep Spin Miniprep Kit (QIAGEN) and verified via Sanger sequencing by Azenta. A total of 1.5 μg of each plasmid was treated with the ScaI-HF restriction enzyme (New England BioLabs) for 15 min at 37 °C to linearize them. The plasmids then were mixed with preincubated Cas9 (New England BioLabs) and wild-type cardinal gRNA (Synthego) at a molar ratio of 10:10:1 Cas9–gRNA–plasmid. This mix was incubated at 37 °C for 15 min and then for 10 min at room temperature after the addition of 1 μL proteinase K. A 2% agarose gel was utilized to resolve the reaction products for subsequent analysis.

AgTP13 introgression into AgNdo-WT, AgNdo-H1, and AgNdo-H2 lines

A total of 3 mating experiments were utilized to test the TP13 introgression drive dynamics into the AgNdo-WT, AgNdo-H1, and AgNdo-H2 lines. Three replicates each of 75 AgTP13 males crossed to 75 females from either AgNdo-WT, AgNdo-H1, or AgNdo-H2 represent the parental generation of each experiment as described in Tushar et al. (2024). All resulting F1 progeny from the parental cross were hatched in plastic containers (11″ [7.94 cm] × 6″ [15.24 cm] × 4″ [10.16 cm]) with ∼175 larvae per container. If >1,000 progeny were generated from the parents, an equal proportion of randomly-selected second instar (L2) larvae were taken from each container and consolidated in 5 new containers that contained 175 larvae each. The larvae from 4 of these containers were reared and a representative sample of ∼600 pupae resulting from them screened for phenotypes. Mosquitoes from the fifth container were screened as pupae, reared to adults, and allowed to mate freely and contribute to the next generation. If <1,000 larvae were hatched from the parents, then the total population was screened with 1 container created containing the 175 randomly selected L2 larvae that contributed to the next generation. The pupae were screened based on 3 parameters: sex, presence/absence of a dominant CFP marker, and eye phenotype (black-eye [wild-type], red-eye [cardinal], or tear [mosaic]). All F1 mosquitoes that had the ability to contribute to the next generation, along with an additional sample of ∼50 F1 mosquitoes from each unique eye phenotype, were frozen to preserve for future sequencing. A total of 20 to 30 F1 individuals (or all if less than 20 produced) with the black-eye/CFP⁺ phenotype per experiment and replicate were sequenced in the F1 generation. This same procedure was repeated for the F2 generation, where a minimum of 600 randomly selected pupae were screened and sorted and a subset (20 to 30 mosquitoes per phenotype, experiment, and replicate, or all if less than 20 produced) of the following phenotypes sequenced: black-eye/CFP⁺, black-eye/CFP⁻, red-eye/CFP⁺.

Introduction of the TP13 gene-drive system into AgNdo-H2

A TP13/Ndokayo hybrid line, designated AgNdoH2-TP13, was created by crossing males from the homozygous AgTP13 line with females from the homozygous AgNdo-H2 line. The black-eye/CFP⁺ hemizygous progeny with 1 copy of an AgTP13 allele and 1 copy of an AgNdo-H2 allele were maintained as a hemizygous line by continuously backcrossing black-eye/CFP⁺ drive hemizygous males with AgNdo-H2 females. The hemizygous line also was converted into a homozygous line by allowing AgNdoH2-TP13 hemizygous males to backcross with AgNdo-H2 females for 5 generations, followed by an intercross of the fifth generation of AgNdoH2-TP13 hemizygous individuals in which subsequent progeny displaying a drive homozygous phenotype (red-eye/CFP⁺) were selected and allowed to continuously intercross.

AgNdoH2-TP13 hemizygous male and female gene drive dynamics

AgNdoH2-TP13 drive efficiency was compared between males and females through 2 different sets of hemizygous outcrosses to AgNdo-H2 for 2 generations. For male drive dynamic experiments, 3 replicates of 50 AgNdoH2-TP13 hemizygous male parents (black-eye/CFP⁺) with 1 copy of the drive system and 1 copy of the haplotype 2 allele were crossed to 50 AgNdo-H2 female parents (black-eye/CFP⁻) to establish the F1 generation. For female drive dynamic experiments, the parents included 3 replicates of 50 AgNdoH2-TP13 hemizygous female parents (black-eye/CFP⁺) with 1 copy of the drive system and 1 copy of the haplotype 2 allele crossed to 50 AgNdo-H2 male parents (black-eye/CFP⁻) to establish the F1 generation. F1 mosquitoes from male and female lineages were screened at the pupal stage based on eye phenotype (black-eye, red-eye or tear) and presence or absence of the CFP marker. A random sample of 50 male and female progeny from each lineage displaying a hemizygous phenotype (black-eye/CFP⁺) were separated from the F1 generation after screening as pupae, reared to adults as virgins, and outcrossed once more to 50 AgNdo-H2 from the opposite sex. F2 progeny from these crosses were screened in the same manner as the F1 generation. F1 and F2 progeny of each phenotype and a subset of 20 to 30 wild-type individuals (or all if less than 20 produced) from every male and female lineage replicate were preserved for DNA sequencing. The experiments described above were performed at 2 different time points, once with the hemizygous line prior to initiating the cage trials as described below and a second time when a sample of mosquitoes was taken from the population of each replicate of the cage trial after 18 continuous generations (∼1 yr).

AgNdoH2-TP13 cage trials

Triplicate experimental cages, A, B, and C, each with a volume of 5,000 cm³, were established with a 1:1 initial release ratio of 75 homozygous drive males AgNdoH2-TP13 to 75 wild-type AgNdo-H2 males. A total of 150 AgNdo-H2 females were added to each cage to achieve a balanced male-to-female ratio with a total population of 300 adults per cage. Subsequent generations were screened and analyzed following the methodology described by Carballar-Lejarazú et al. (2023). Every generation, 300 randomly-selected L2 larvae from each cage were removed to a new cage for the subsequent generations, and an additional 800 randomly-selected L2 larvae were reared separately to pupae for sexing and phenotypic screening. Pupae were screened for the CFP marker to determine the presence or absence of the drive system and for eye color phenotype (wild-type black-eye, cardinal mutant red-eye, or mosaic tear-eye). The remaining larvae were counted and added to determine the total population size.

Molecular validation of gene-drive cassette integration and effector molecule integrity in AgNdoH2-TP13 cage trials

Genomic integration was verified by gene amplification analyses using genomic DNA of cage trial samples at selected generations as templates and gene-specific oligonucleotide primers. Genomic DNA extraction, PCR reactions and sequencing were carried as described in Carballar-Lejarazú et al. (2023). The right-hand junction between the TP13 drive cargo and the cardinal gene left homology arm was verified by the primer pair VgProm-seqF4/TP179 (Supplementary Table 2). Primers upHm1cdn/TP179 were used to amplify a fragment of the Agcd gene spanning the insertion location of the drive cassette between the left and right homology arms to verify the presence of a WT allele in the hemizygous mosquitoes. The primer pair upHM1cdn/TP217 was used for the left-hand junction between the right homology arm and the cargo of TP13. The integrity of the effector molecules was validated using primer pairs TP218/TP219 and Vg3UTR-seqF2/TP220 to amplify fragments from the 5′-end untranslated region (UTR) to the 3′-end UTR of m1C3 and m2A10, respectively.

RT-PCR validation of antimalarial effector and Cas9 gene expression

Females from each cage trial at selective generations were bloodfed and collected after 24 h for RT-PCR analysis of antimalarial effector gene transcripts and abundance and Cas9 expression. Total RNA extractions, cDNA synthesis, and PCR reactions were carried out as described in Carballar-Lejarazú et al. (2023). Synthesized cDNA was amplified with gene-specific primer pairs, TP236/TP237, TP240/TP241, and RPS7/RPS7-R, to determine the expression of m1C3, m2A10, and the ribosomal protein, RPS7, respectively.

AgNdo fitness experiments

Life-table parameters for AgNdo, AgNdo-H2, and AgNdoH2-TP13 hemizygous and homozygous lines were evaluated following the methodology described previously for An. gambiae and An. coluzzii gene drive lines (Carballar-Lejarazú et al. 2020, 2023). All assays were conducted in biological triplicates except for the longevity experiments where survival curves were based on 100 males and 100 females per line. Data from hemi- and homozygous AgNdoH2-TP13 and AgNdo were compared with the triplicate biological replicate AgNdo-H2 data. The combined data from all 3 replicates of each line were used in the analyses. All statistical tests were conducted in R (version 4.2.2) using the same analytical methods as reported in Carballar-Lejarazú et al. (2023).

Male likelihood experiments and molecular analysis

The ability of haplotype-bearing and gene drive-carrying males to mate with females and contribute to the next generation in the presence of competing males with other genotypes is important for drive dynamics. Male progeny from outcrosses of the AgNdoH2-TP13 line to the AgNdo-H2 line, male progeny from AgNdoH2-TP13 intercrosses, and males from the AgNdo colony were tested in competition with AgNdo-H2 males. A total of 150 3- to 5-d-old virgin AgNdo-H2 females were mated with 75 3- to 5-d-old virgin test males (outcross or intercross AgNdoH2-TP13 or AgNdo) and 75 AgNdo-H2 control virgin males with all males from each respective line and strain age-matched to the day in each replicate. Mating took place in triplicate 5,000-cm³ cages over the course of 3 d, and females then were provided with blood meals on 2 successive days. After feeding, females were allowed 2 d to complete the gonotrophic cycle and then provided an egg cup for oviposition for 3 d. All eggs and larvae were subsequently hatched from the egg cup and reared under standard insectary conditions. A minimum of 1,000 randomly selected larvae were screened per replicate for presence/absence of the dominant fluorescent marker at the L3-L4 stage, and the total number of positive and negative mosquitoes was recorded.

The haplotype 2 SNP frequency in the AgNdo strain at the time of the competitive crosses was determined with PCR and Sanger sequencing using a total of 153 individual siblings of the males used in the male-likelihood experiment. For each replicate of the AgNdo and AgNdo-H2 male likelihood experiment, 25 pools of 50 L4 larvae progeny were extracted for genomic DNA (gDNA) using the Blood and Tissue DNeasy kit (QIAGEN). The gDNA from each pool was normalized to 100 ng/μL and used as template for PCR amplification with the primer pair TP624 and TP635 (Supplementary Table 2) and CloneAmp Hifi PCR Mix (Takara Bio) following the manufacturer’s instructions to produce a fragment 271 bp in length spanning the cardinal gRNA target site. The amplicons were subjected to deep sequencing with Amplicon-EZ service from Genewiz (Azenta). Deep sequencing data were analyzed with CRISPResso2 (Clement et al. 2019) for haplotype 2 SNP frequency in the progeny resulting from the crosses.

Parasite challenge assays

Adult hemi- and homozygous AgNdoH2-TP13 and AgTP13 adult females were used in parasite membrane feeding assays along with their respective AgNdo-WT and AgX1 control lines. All mosquitoes were maintained under standard insectary conditions as described (Carballar-Lejarazú et al. 2023). Briefly, mosquitoes were kept at 27 ± 1 °C, 70% to 80% relative humidity, with a 12:12-h light:dark photoperiod. Females were provided a 10% sucrose solution ad libitum until just prior to the infectious blood meal. The P. falciparum NF54 strain was maintained in human O⁺ erythrocytes until mature stage V gametocytes were obtained. Cultures were diluted to achieve target gametocytemia levels for experimental challenges and designated “low” (0.08%) gametocytemia, representative of many natural transmission levels, and “medium” (0.1%), and “high” (0.3%) gametocytemia levels, used to increase statistical power in laboratory assays. Infectious blood meals were prepared by combining cultured gametocytes with fresh human serum and erythrocytes. Blood meals were maintained at 37 °C and offered to mosquitoes for 60 min using a membrane-feeding apparatus fitted with parafilm following previously established protocols (Tripathi et al. 2020). Fully engorged females were separated within 1-h post-feeding and maintained on 10% sucrose for the duration of the parasite development period. Additional uninfected blood meals were provided at 5- and 10-d post-infection to boost the blood meal-inducible effector gene expression.

Oocyst counts (intensities of infection) were obtained from midguts dissected at 7 to 8 d post-infection, stained with 0.1% mercurochrome, and examined under a light microscope at 100× or 200× magnification. The number of oocysts per midgut was recorded for each mosquito. Sporozoite counts were obtained from salivary glands dissected 14 to 16 d post-infection. Salivary glands were homogenized in 30 µL of phosphate-buffered saline (PBS), and 10 µL of the homogenate was loaded onto a hemocytometer for counting using light microscopy (Leica). Two fields were counted for the low-infection samples, and the average sporozoite number calculated from the largest grid (0.1 µL) was multiplied by 300. The high-infection samples were counted using the smallest grid (0.004 µL), or the next appropriate scale depending on the density, and multiplied accordingly.

Dot plots of parasite counts (oocysts and sporozoites) were generated in GraphPad Prism version 10. Mean and median intensities of infection were calculated for each group. Prevalence (percentage of infected mosquitoes) and intensity data from independent biological replicates were pooled prior to analysis to maintain balanced sample sizes across treatments. Infection intensity was compared between experimental and control groups using the 2-tailed Mann–Whitney U test. Prevalence was compared using Fisher's exact test. Differences were considered statistically significant at P < 0.05.

Ethics statement

All studies were performed in compliance with the requirements of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, and the Institutional Animal Care and Use Committee at Johns Hopkins University approved the protocol (permit no.: M006H300). Mice were used only as a blood source for mosquito rearing. Commercial anonymous human blood was used for Plasmodium gametocyte cultures and infection assays and informed consent was not required.

Statistics

P-values <0.05 were considered significant in all tests.

Generation of An. gambiae Ndokayo wild-type and SNP haplotype target-site isolines

Twelve of the 23 nucleotides in the Agcd target-site DNA sequence used in the TP13 and AgNosCd-1 drive systems have naturally-occurring SNP variations, and these are relatively uncommon (0.06% to 5.8%) except at 1 position (2R:38748898) distal to the PAM site (refer to Supplementary Table 18 in Carballar-Lejarazú et al. 2020). This position has an A→T transversion located 17 bp to the 5′-end of the PAM and is found in as high as 43% of the various wild A. gambiae s.l. populations surveyed by the 1,000-genome consortium and among recently colonized laboratory strains (Fig. 1a; Ag1000G Consortium 2023). Live colonies of variant-carrying strains are generally unavailable, but the Ndokayo strain (AgNdo) was obtained from a colony kept at the MR4 (strain MRA-1278, MR4, BEI Resources) and we obtained a colony from Burkina Faso (BF) as a gift of Dr. Abdoulaye Diabaté.

Preliminary molecular characterizations of the gene-drive target sites in these strains showed that the A→T SNP transversion (2R:38748898) was present in multiple haplotypes with variable frequencies in each strain (Fig. 1a). Haplotype 1 (H1) has the 2R:38748898 transversion along with 3 additional SNPs (G→A at 2R:38748902, G→A at 2R:38748910, and C→T at 2R:38748916) distal to the 3′-end of the target site and is represented in 13% of the alleles in the Ndokayo and BF strains (Fig. 1b; Table 1). Haplotype 2 (H2) represents 16% of the alleles seen in the Ndokayo strain and has the 2R:38748898 SNP along with an additional SNP (A→C at 2R:38748865) 14 bp from the 5′-end of the target site. Haplotype 3 (H3) represents 27% of the alleles found in the Burkina Faso strain and has the 2R:38748898 SNP along with a C→T transversion distal to the 3′-end of the target site. Continued long-term maintenance of the colonies in our laboratory resulted in apparent heterozygosity of haplotypes 2 and 3 with canonical wild-type alleles in their respective strains accompanied by the loss of haplotype 1, most likely as a result of founder effects eliminating the least frequent variant. We chose the Ndokayo strain for further analysis because it contained the greatest diversity of target-site haplotypes, and generating all experimental isolines from a single genetic background would minimize other, potentially confounding, genetic effects.

Single AgNdo males and females were used to generate 3 isolines to assess the impact of the high-frequency polymorphisms on the TP13 drive system. The 3 derived lines include 1 line, AgNdo-WT, homozygous for the wild-type cardinal target site identical to the one targeted by the AgTP13 gRNA and lines homozygous for haplotypes 1, AgNdo-H1, and 2, AgNdo-H2 (Fig. 1c).

Cleavage in vitro of AgNdo variant target sites

The Cas9/gRNA complex cleavage efficiency for the target-site mutations was determined using an in vitro cleavage assay and 5 plasmids containing the different target-site sequences. The 5 plasmids (pBM1 to pBM5) contained the AgNdo-WT, AgNdo-H1, AgNdo-H2, Agcd^Δ11 (negative control), and AgNdo-WT-PAM deleted (negative control) target sites, respectively, along with a 36-bp homology arm adjacent to the 5′-end of the target site and a 329-bp homology arm adjacent the 3′-end (Supplementary Table 3; Carballar-Lejarazú et al. 2022). The plasmids were incubated with Cas9/wild-type gRNA complexes and cleavage observed by the presence of specific DNA fragments resolved in agarose gels. The results show that all plasmids with the natural SNPs, pBM1 to pBM3, were cleaved, while the negative controls, pBM4 and pBM5, remained uncut (Fig. 1d). There were no visible qualitative differences in the amounts of cleavage with plasmids containing either haplotype 1 or haplotype 2 (pBM2 or pBM3) in comparison with the wild-type sequence (pBM1) as evidenced by the similar staining intensities for the cut fragments in pBM1 to pBM3. However, in the absence of densitometry, these differences are not definitive. While these findings show that the AgTP13 Cas9/gRNA complex can efficiently cleave target sites containing haplotype 1 or haplotype 2 sequences in vitro, they do not reveal what impact the haplotypes may have on cleavage in vivo. It is possible that the haplotypes may impact post-cleavage HDR or other end-joining repair outcomes, or the abilities of the Cas9/gRNA complexes to cleave the target site in a time- and site-specific manner. Thus, we performed a number of experiments to assess them in vivo.

TP13 gene-drive system dynamics following outcrosses to Ndokayo isolines

Three different mating experiments utilizing 75 homozygous AgTP13 males crossed to 75 Ndokayo females from either the AgNdo-WT, AgNdo-H1, or AgNdo-H2 lines were performed in triplicate to determine the impact of haplotypes on the ability of TP13 to drive in the presence of target-site SNPs. Briefly, F1 progeny from these crosses were screened for eye phenotypes and to ensure the presence of the CFP marker, and 150 randomly-selected F1 progeny were subsequently allowed to intercross to produce an F2 generation (Table 2). Drive inheritance and drive conversion rates were determined in the F2 generation based on the proportions of progeny with (CFP⁺) or without (CFP⁻) the dominant marker gene and by determining the corresponding genotypes of each unique eye phenotype resulting from the F1 hemizygous intercrosses. Possible F2 generation eye phenotypes include black-eye/CFP⁺, red-eye/CFP⁺, tear eye/CFP⁺, black-eye/CFP⁻, and red-eye/CFP⁻ (Carballar-Lejarazú et al. 2022, 2023; Tushar et al. 2024). The tear-eye phenotype was shown previously to result from maternal- and paternal-effect target-site cleavage in somatic tissues that generated mosaic genotypes of functional and nonfunctional mutant alleles in small clones of cells that produced mosquitoes with some but not all ommatidia lacking pigment (Carballar-Lejarazú et al. 2020, 2022, 2023). The expected genotypes possible from these phenotypes based on previous observations of the AgNosCd-1 and AgTP13 drive systems are black-eye/CFP⁺ mosquitoes with 1 copy of the TP13 drive allele and either 1 copy of a wild-type cardinal allele (cd⁺) or functional mutant allele; mosquitoes with a red-eye/CFP⁺ phenotype that have either 2 copies of the TP13 drive allele or 1 copy of the TP13 drive allele and a nonfunctional mutant allele (cd⁻); mosquitoes with a tear-eye/CFP⁺ phenotype having 1 copy of the TP13 drive and a mixture of wild-type and mutant alleles indicative of mosaicism at the cardinal gene; black-eye/CFP⁻ mosquitoes either heteroallelic for a wild-type cardinal allele and functional mutant allele or homozygous for either allele, and mosquitoes with a red-eye/CFP⁻ phenotype conferred by 1 (homozygous) or 2 (heteroallelic) nonfunctional mutant alleles.

As expected of outcomes of crossing homozygous TP13 male parents, all progeny in the F1 generation were confirmed to be phenotypically CFP⁺ in all 3 experiments (Table 2 and Supplementary Tables 4 to 6). The majority of progeny had the black-eye/CFP⁺ phenotype resulting from a wild-type cardinal allele and 1 drive allele (Supplementary Tables 1 and 7). However, there were small numbers of progeny in each experiment with the tear-eye/CFP⁺ phenotype, 0.4% (AgNdo-WT), 0.2% (AgNdo-H1), and 0.74% (AgNdo-H2) (Supplementary Tables 4 to 6). The tear-eye/CFP⁺ F1 individuals were not sequenced in these experiments due to their allelic complexity (Carballar-Lejarazú et al. 2025). There were no significant differences among all 3 parental crosses in the likelihood to produce tear-eye progeny (pairwise independent samples t-tests with Bonferroni correction, adjusted P-values ≥ 0.12). A single white-eye/CFP⁺ mosquito was observed in the AgNdo-H2 parental cross and sequencing of the known genes involved in eye pigmentation in this individual revealed that the phenotype resulted from an insertion of 667 bp in length into exon 4 of the An. gambiae white AGAP000553 gene (Supplementary Fig. 1, Supplementary Methods). This spontaneous white-eyed phenotype was observed on occasion in the AgNdo colony and is unrelated to the drive activity.

All 3 experiments showed high drive inheritance in F2 generation progeny from subsequent haplotype-specific intercrosses regardless of their target-site genotype (Table 2 and Supplementary Tables 4 to 6). Over 99% of F2 progeny had at least 1 copy of the TP13 drive, which is much higher than the 75% that would be expected to result from Mendelian random segregation (AgNdo-WT: χ² = 576.91, P < 2.2⁻¹⁶; AgNdo-H1: χ² = 603.81, P < 2.2⁻¹⁶; AgNdo-H2: χ² = 634.01, P < 2.2⁻¹⁶). Most individuals (average of 85.2% across all 3 experiments) exhibited the red-eye/CFP⁺ phenotype indicative of either 2 copies of the drive allele or 1 copy of the drive allele and 1 copy of a nonfunctional mutant allele. Molecular analysis confirmed that every red-eye/CFP⁺ phenotype mosquito sequenced (n = 199) from all 3 replicates of every experiment had only TP13 drive alleles, and no nonfunctional mutant alleles were observed (Supplementary Tables 1 and 8). The second most common phenotype (averaging 14.4%) was black-eye/CFP⁺ individuals, which could have either an unaltered AgNdo-WT, AgNdo-H1, or AgNdo-H2 allele or functional mutant allele. The majority (>98% of the 206 sequenced) had unaltered AgNdo-WT, AgNdo-H1, or AgNdo-H2 alleles. However, there was evidence of functional mutant alleles in black-eye/CFP⁺ individuals sequenced from the AgNdo-WT (1.4%; 1/73) and AgNdo-H1 (1.5%; 1/67) intercrosses, while AgNdo-H2 intercrosses only produced black-eye/CFP⁺ individuals with AgNdo-H2 alleles (66/66) (Supplementary Tables 1, 8, and 9). Interestingly, no evidence of mutant alleles was observed in previous male lineage introduction crosses involving the AgNdo strain (containing a mix of haplotypes 1, 2, and wild-type target sites) and the AgNosCd-1 gene drive line (Tushar et al. 2024). The black-eye/CFP⁻ individuals sequenced from the AgNdo-WT and AgNdo-H1 intercrosses also contained mutant alleles, 14.3% (1/7) and 28.6% (2/7), respectively. One mutant allele from the AgNdo-H1 intercross was shared with its black-eye/CFP⁻ and black-eye/CFP⁺ siblings from the same replicate (Supplementary Table 9). These data are consistent with the early formation of this allele in the germline with subsequent clonal copies being passed along in multiple gametes to progeny (“clustering” effect; Carballar-Lejarazú et al. 2025). However, these data do not provide evidence for which sex the germline mutation event may have occurred since both hemizygous drive-carrying males and females contributed to the F2 generation.

Despite the small rates of formation of mutant alleles, haplotype 1 or haplotype 2 target sites did not have a significant effect on the drive dynamics of the TP13 drive system. One way this can be quantified is by calculating the drive conversion rate (“homing”) for each experiment, which measures the frequency of the number of target alleles converted to drive alleles divided by the total number of potential target alleles and expressed as a percentage (Sánchez et al. 2020; Carballar-Lejarazú et al. 2023, 2025). There is no significant difference between AgNdo-WT (83.7%), AgNdo-H1 (85.4%), and AgNdo-H2 (85.4%) when looking at the drive conversion rates of all 3 experiments (pairwise independent samples t-tests with Bonferroni correction, adjusted P-values ≥ 1.0; Supplementary Table 10). This supports the conclusion that Cas9/gRNA complexes are able to cleave target sites with the haplotype 1 and haplotype 2 SNPs as efficiently as canonical wild-type target sites.

AgNdoH2-TP13 male and female lineage gene-drive dynamics

The drive dynamics of the TP13 system in a population with a SNP background was determined by introgressing it into the AgNdo-H2 line by outcrossing homozygous AgTP13 males to AgNdo-H2 females. This was followed by backcrossing resulting AgTP13/AgNdo-H2 males to AgNdo-H2 females for an additional 4 generations to yield the AgNdoH2-TP13 gene drive line theoretically carrying ∼96.9% of the AgNdo-H2 genome except for the Y chromosome (discussed in Carballar-Lejarazú et al. 2025).

Sex-specific differences in the drive dynamics and mutant allele formation rates in the AgNdoH2-TP13 line were examined in outcrosses of hemizygous AgNdoH2-TP13 males or females containing 1 TP13 allele and 1 haplotype 2 allele to AgNdo-H2 mosquitoes of the opposite sex. The F1 progeny from these crosses were screened based on eye phenotypes and a sample of hemizygous black-eye/CFP⁺ male and female progeny were removed, kept as virgins, and outcrossed once more to AgNdo-H2 of the opposite sex to observe drive dynamics over 2 generations. For both male and female lineages, the majority of progeny generated from the hemizygous outcrosses had a black-eye/CFP⁺ phenotype in the F1 and F2 generations (Table 3). However, it was observed that F1 and F2 progeny from AgNdoH2-TP13 hemizygous males had a significantly higher average drive inheritance of 97.2% compared to female founder lineages, which averaged 80.7% (Table 3, Supplementary Tables 11 and 12) (χ² = 758, P < 2.2⁻¹⁶). Previous outcrosses with hemizygous AgTP13 (X1 background) also have shown that hemizygous males have a higher drive inheritance although the difference is much less pronounced, ranging from 99.5% to 100% in hemizygous AgTP13 males to 85.5%to 96.9% in hemizygous AgTP13 females (Carballar-Lejarazú et al 2023).

There were no red-eye/CFP⁺ progeny in the male lineages and small percentages, 0.1% to 0.4%, of the tear-eye/CFP⁺ phenotypes (Table 3 and Supplementary Table 11) There were a relatively small numbers of red-eye/CFP⁺ and tear-eye/CFP⁺ progeny generated from AgNdoH2-TP13 hemizygous female outcrosses when compared to the previous outcrosses with hemizygous AgTP13 females (X1 background) (Table 3 and Supplementary Table 12). There was only 1 individual in the F1 progeny exhibiting the red-eye/CFP⁺ phenotype (0.02%) in the AgNdoH2-TP13 hemizygous female outcross, while every AgTP13 hemizygous female (X1 background) outcross in both generations had at least 2 red-eye/CFP⁺ progeny, which ranged from 0.15% to 0.5% of the total (Carballar-Lejarazú et al 2023). The tear-eye progeny ranged from 0.3% to 1.2% in AgNdoH2-TP13 hemizygous female outcrosses compared to 4.2% to 13.9% in hemizygous AgTP13 female outcrosses (X1 background) (Table 3; Carballar-Lejarazú et al 2023). The observed reductions in the CFP⁺, tear-eye, and red-eye phenotypes in female lineages may be due to lower Cas9/gRNA activity and/or reduced susceptibility to maternal effects in the AgNdo-H2 genetic background.

Sequencing of the hemizygous drive female outcross progeny revealed that despite having low drive inheritance, the CFP⁻ progeny had few mutant alleles (Supplementary Tables 1 and 13). Only 1 mutant allele of 206 sequenced (∼0.5%) was recovered from the F1 and F2 black-eye/CFP⁻ progeny of female parents of the AgNdoH2-TP13 hemizygous female outcrosses. In contrast, mutant alleles were found to be much more common in the black-eye/CFP⁻ progeny generated from male parents of AgNdoH2-TP13 hemizygous male outcrosses. Every hemizygous male outcross had some black-eye/CFP⁻ progeny with mutant alleles present in both the F1 and F2 generations (range of 8.8% to 31.7% of total black-eye/CFP⁻ progeny). Many of these alleles were shared among multiple progeny from the same replicate, with as many as 9/16 black-eye/CFP⁻ progeny from one of the replicates having the same mutant allele (Supplementary Tables 1 and 14). This supports the previous observations from the introduction experiments that the AgNdo isolines may have a greater incidence of germline mutation events and resulting cluster effects. In this case, the hemizygous drive males generate a greater proportion of mutant alleles in their germline, which is in sharp contrast to previous experiments with AgTP13 (X1 background), that found drive hemizygous and homozygous females produce the majority of mutant alleles via maternal effects while germline mutation events were rare (Carballar-Lejarazú et al. 2022, 2023).

AgNdoH2-TP13 cage trials

The efficacy of the TP13 drive system in a population with the target-site SNP was assessed using triplicate, discrete, nonoverlapping cages, A, B, and C, with a founder generation (G0) release ratio of 1:1 AgNdoH2-TP13 gene-drive to AgNdo-H2 wild-type males. Three hundred randomly-selected first generation post introduction, G1, L2 larvae were put aside to start the next generation, 800 randomly selected L2 larvae were reared to pupae for sex and phenotypic screening, and the remaining larvae were counted and the numbers were added to the selected progeny to get the total population size. This was repeated to produce the subsequent G2 to G21 generations. A total of 403,044 mosquitoes, excluding the G0 founders, were scored from the 3 cages throughout the experiment (Supplementary Tables 15 to 17).

The G1 yielded a high portion of CFP⁺ progeny with the 69% (412/596, cage A), 80% (514/640, cage B), and 78% (471/602, cage C) being significantly higher than the expected 50% (1/2 chance of parental females mating with homozygous transgenic males) (cage A: χ² = 87.2, P < 0.0001; cage B: χ² = 235.2, P < 0.0001; cage C: χ² = 192.0, P < 0.0001) (Fig. 2a). The percentage of CFP⁺ mosquitoes continued to increase in the caged population and reached full introduction (100% CFP⁺, every mosquito carrying at least 1 copy of the gene-drive system) at G3 for all 3 cages. The drive allele remained fixed in the cages until the end of the experiment at G21 without any drive-negative or unexpected phenotype mosquitoes recovered from any cage. Population sizes fluctuated between 3,000 and 11,000 mosquitoes/reproductive cycle/cage and were generally higher toward the end of the experiment (Fig. 2b).

Molecular assessments carried out to ensure drive system integrity during the cage trials showed no mutations at the gRNA target site following sequencing of the few black-eye individuals remaining in the cages before the cage population reached fixation (G3 to G4) (Supplementary Tables 1 and 18) and all mosquitoes in the cages were red-eye/CFP⁺ at G5 (Supplementary Tables 15–17). Assessment of CFP⁺/cd⁻ mosquitoes at ∼1 yr (G18) after drive release continued to validate the insertion of the drive cassette, and amplification of the drive junction between the drive cassette and the cardinal gene homology arm showed proper drive integration into the genome (Supplementary Fig. 2, a and b). Effector scFv gene sequences and expression were assessed to ensure the activity of the effector genes. Amplification with primers spanning the promoters, 5′-UTR and 3′-UTR control DNA of the Cp and Vg DNA captured the whole sequences of the respective m1C3 and m2A10 effectors and no mutations were detected in the DNA sequences (Supplementary Fig. 2, c and d). RT-PCR result showed expression of m1C3 and m2A10 in caged CFP⁺/cd⁻ mosquitoes at 24 h post blood meal as expected (Isaacs et al. 2011, 2012; Supplementary Fig. 2e).

A random selection of AgNdoH2-TP13 homozygous males was removed from each cage replicate 1 yr after drive release (G18), and outcrossed to AgNdo-H2 colony females to generate AgNdoH2-TP13 hemizygous individuals, representative of each cage, for male and female lineage gene-drive dynamic assessments. The subsequent gene drive inheritance patterns and drive associated phenotypes/genotypes observed at the 1-yr cage trial assessment then were compared to those observed at the initial assessment of the drive system (Supplementary Tables 11, 12, 19, and 20). No substantial differences were observed in gene-drive inheritance in the F1 and F2 between the initial assessment and 1 yr after cage trial initiation with hemizygous females transmitting the drive to an average of 80.7% of progeny in the initial assessment and 80.4% of progeny 1 yr later (t = −0.1, P-value = 0.92) and hemizygous males transmitted the drive to an average of 97.2% of progeny in the initial assessment and 98.2% of progeny at the 1-yr mark (t = 1.6, P-value = 0.13). The proportions of parental effect-generated phenotypes, tear-eyes and red-eyes (cardinal mutant phenotype), were low in both assessments with tear eyes being observed in less than 2% of progeny irrespective of parental lineage and cardinal red eyes only being observed in 1 or 2 rare events of a female lineage cross.

Molecular analyses of selected phenotypes from the 1-yr drive assessments showed similar results with the initial AgNdoH2-TP13 founder experiments (Supplementary Tables 1, 21, and 22). There was only 1 mutant allele recovered from the red-eye/CFP⁺ progeny of hemizygous female outcross; no other mutant alleles were recovered from 360 progeny sequenced (both black-eye/CFP⁻ or black-eye/CFP⁺) of hemizygous AgNdoH2-TP13 female outcrosses and the black-eye/CFP⁺ of hemizygous AgNdoH2-TP13 male outcrosses. Mutant alleles were found in the black-eye/CFP⁻ progeny of all hemizygous male outcrosses in both F1 and F2 1-yr assessment generations ranging from 17.4% to 17.9% (83 sequenced). This percentage range is similar to the range in the initial founder experiment, indicating that the drive system and Cas9/gRNA cleavage activity remain unchanged during the cage trial.

AgNdoH2-TP13 fitness parameters

The original AgTP13 line in the X1 genetic background had a genetic load in homozygous males that resulted in significant differences in their likelihood to contribute to the next generation when in direct 1:1 competition with wild-type males and displayed a shorter lifespan as well (Carballar-Lejarazú et al. 2023). However, no fitness load was observed in the same study with AcTP13 homozygous males after introduction into the An. coluzzii genetic background. The impact of the Ndokayo genetic background on genetic load and fitness was assessed using the same life-stage parameters measured for the AgTP13 and AcTP13 lines (Supplementary Table 23). The lines examined included AgNdo-H2, hemi- and homozygous AgNdoH2-TP13 and the AgNdo line, the latter of which has been maintained as a free-mating colony in our insectary since receipt from MR4 in 2020. The AgNdo line was used to distinguish differences in AgNdo-H2 fitness parameters that might be conflated with reductions in fitness known to occur in highly in-bred mosquito strains (Baeshen et al. 2014).

The AgNdoH2-TP13 homozygous line had slight advantages over the AgNdo-H2 line at the immature stages. Homozygous AgNdoH2-TP13 had greater larval viability (78%) and larvae developed into adults slightly faster (9 d) than AgNdo-H2 with a 59% viability and 9.4 d average development time to adulthood (Supplementary Table 23). Both hemizygous and homozygous AgNdoH2-TP13 progeny were female-biased (χ² = 4.1, P = 0.04) and (χ² = 5.8, P = 0.02), respectively, while neither AgNdo-H2 nor AgNdo showed a sex bias. Similar to the original AgTP13 observations, AgNdoH2-TP13 homozygotes had shorter lifespans (16 d for males and 17 d for females) when compared to AgNdo-H2 (19 d for males and 21 d for females) although this difference was more pronounced in female longevity (P = 0.0007) as compared to male longevity (P = 0.2) (Carballar-Lejarazú et al. 2023). Also similar to the original AgTP13 observations, AgNdoH2-TP13 homozygous females had greater fecundity at 86 eggs per single female egg batch compared to AgNdo-H2 females with 72 eggs per batch (P = 0.003) and their fertility was lower with a hatch rate of 67% compared to AgNdo-H2 at 75%, although the difference is not statistically significant (P-values >0.05).

AgNdoH2-TP13 progeny males from both outcrosses and intercrosses were able to outcompete equal numbers of AgNdo-H2 males and contribute ∼68% and ∼60%, respectively, of the progeny to the next generation (P-values < 0.0001; Supplementary Table 23). Assuming an equal reproductive contribution of males from the AgNdo-H2 line and AgNdo strain, progeny from a competitive cross are expected to have 14 of 16, 87.5%, of the possible haplotype 2 alleles. This expected frequency is based on a 50% frequency of both canonical wild-type and haplotype 2 alleles in samples of the AgNdo strain sequenced at the time of the experiment and a 100% frequency of haplotype 2 alleles contributed by the homozygous AgNdo-H2 males and females (Supplementary Table 1). The observed 79.3% frequency of haplotype 2 alleles in the progeny indicates that AgNdo-H2 males did not compete equally with AgNdo males in contributing to the next generation (P-values < 0.0001; Supplementary Table 23). These results, along with those of the AgNdoH2-TP13 competitive crosses, may result from poorer male reproductive health in the highly selected AgNdo-H2 colony.

AgNdoH2-TP13 parasite challenge assays

Parasite challenge assays were conducted by feeding infectious, P. falciparum (NF54 strain) gametocyte-containing blood meals to hemizygous and homozygous AgNdoH2-TP13 and AgTP13 adult females with AgNdo-WT and AgX1 as respective controls. The protocol was adapted from previous work and provides 2 additional, noninfectious blood meals at intervals after the initial infection (Carballar-Lejarazú et al. 2023, 2026a). Midguts from bloodfed females were dissected at 7 to 8 d post-infection (dpi) to determine oocyst infection prevalence and median and mean intensities of infection, and salivary glands of another cohort from the same feeding were dissected 14 to 16 dpi to obtain the same information for sporozoites.

Hemi- and homozygous AgNdoH2-TP13 mosquitoes challenged with low (0.08%) gametocytemias (percentage of gametes in infected blood preparations) showed significant reductions in oocyst prevalence, 30.6% and 56.1%, respectively, and mean intensities of infection, 33.3% and 75%, respectively (P-values ranging from 0.009 to <0.0001) (Fig. 3, Tables 4 and Supplementary 24). Median oocyst counts were reduced by half and to zero in the hemi- and homozygotes, respectively. Similar significant reductions of 44.6% and 54.4% in prevalence, and 61.1% and 66.6% in mean intensities, were seen in the AgTP13 mosquitoes challenged at the same time compared to controls (P-values from 0.0017 to <0.0001).

Not surprisingly, hemi- and homozygous females challenged with a medium level (0.1%) gametocytemia were not as robust in suppressing oocysts (Fig. 3, Table 4, and Supplementary Table 24). Hemizygous AgNdoH2-TP13 and AgTP13 mosquitoes showed no significant reductions in prevalence (P = 0.4353 and 0.5634, respectively), however, homozygotes were significant with reductions of 10.6% and 21% (P-values from 0.01 to 0.040). Significant reductions of mean intensities were observed in hemi- (32.6%, P = 0.0007) and homozygous (41%, P = <0.0001) AgNdoH2-TP13 and in AgTP13 homozygotes (57.1%, P = <0.0001).

The numbers of mosquitoes sampled for sporozoites in the low and medium gametocytemia regimens were low, so the results were pooled for analysis (Fig. 3, Table 4, and Supplementary Table 24). Despite this, significant reductions in prevalence were seen in hemi- (23.4%; P = 0.0011) and homozygous (43.9%; P < 0.0001) AgNdoH2-TP13 samples. Similar prevalence reductions were seen in AgTP13 samples (hemizygotes: 34.1% reduction, P = 0.0002; homozygotes: 46.6% reduction, P < 0.0001). Reductions in the mean intensities of sporozoite infection also were seen with those in AgNdoH2-TP13 homozygotes (67.4%; P = 0.0008) and AgTP13 hemi- (71.2%; P = 0.0239) and homozygotes (88%; P = 0.0002) being statistically significant. A proposed benchmark to impact parasite transmission is the proportion of mosquitoes with >10,000 sporozoites in their salivary glands (Aleshnick et al. 2020; Graumans et al. 2020). Relatively few mosquitoes exceeded this threshold in the pooled low/medium gametocytemia samples with 5/28, 5/38, 4/30, and 0/37 in AgNdoH2-TP13 hemi-, AgNdoH2-TP13 homo-, AgTP13 hemi-, and AgTP13 homozygotes, respectively (Table 4 and Supplementary Table 24).

With 1 exception, similar challenges of all lines using a high (0.3%) gametocytemia showed lowered or no statistically significant reductions in oocyst prevalence (AgNdoH2-TP13 hemizygotes, 0.5%, P > 0.999; AgNdoH2-TP13 homozygotes, 5.9%, P = 0.0649; AgTP13 homozygotes, 2.4%, P = 0.6827) and sporozoite (AgNdoH2-TP13 hemizygotes, 4.2%, P = 0.3727; AgNdoH2-TP13 homozygotes, 12.5%, P = 0.0081; AgTP13 hemizygotes 23.5, P = 0.0002; AgTP13 homozygotes, 21.9%, P = 0.0004) (Fig. 3, Table 5, and Supplementary Table 24). AgTP13 hemizygotes showed a slight increase of 4.25 (P > 0.999). Oocyst and sporozoite mean intensities reflected the higher gametocytemia loads with significant reductions seen in oocyst (AgNdoH2-TP13 hemi, 29.2%, P = 0.0002; AgNdoH2-TP13 homozygotes, 55.6%, P < 0.0001; AgTP13 hemizygotes 3.1%; P = 0.6747; AgTP13 homozygotes, 11%, P = 0.1058) and sporozoite (AgNdoH2-TP13 hemizygotes, 44.5%, P = 0.0132%; AgNdoH2-TP13 homozygotes, 86.7%, P = <0.0001; AgTP13 hemizygotes 42.5%, P = 0.0193; AgTP13 homozygotes, 73.6.0%, P = <0.0001). More samples had >10,000 sporozoites in their salivary glands (AgNdoH2-TP13 hemizygotes, 21/25; AgNdoH2-TP13, 21/25; AgTP13 hemizygotes, 24/34, and AgTP13 homozygotes, 13/25; Table 5 and Supplementary Table 24). These combined data show that the effector molecules suppress parasite oocysts and sporozoites in the AgNdoH2-TP13 lines with expected impacts greater in homozygotes than hemizygotes (Carballar-Lejarazú et al. 2023, 2026a). Furthermore, the impact is generally numerically higher in the AgNdoH2-TP13 females than in the corresponding AgTP13 mosquitoes tested at the same time.