An OpIE2-DsRed marker disrupts female blood-feeding and shortens lifespan in the malaria vector Anopheles gambiae

Abstract

Anopheles gambiae is one of the principal vectors of human malaria. Over the past 2 decades, transgenic mosquito strains have been essential tools for studying mosquito biology and developing genetic control strategies such as gene drives. Mosquito transformants are typically identified using fluorescent markers, which are assumed to be phenotypically neutral. While generating CRISPR-based gene drive strains carrying an OpIE2-DsRed marker, we unexpectedly found that transgenic females were unable to blood-feed and were consequently sterile, whereas males initially appeared normal and fertile. Given the potential utility of dominant, female-specific sterility for mosquito control, we established additional strains controlling for transgene content and integration site, confirming that the OpIE2-DsRed cassette caused the defect. Behavioral assays showed that females exhibited normal attraction to a membrane feeder but failed to initiate blood-feeding, performing repeated cycles of probing and proboscis grooming in rapid succession before ultimately leaving the feeder unfed. Microscopy showed that both sexes possessed a distally curved proboscis, providing a morphological explanation for the blood-feeding defect of females and the reduced male lifespan. A second promoter variant (OpIE2b), differing in flanking sequences at the IE-2 junction, drove strong marker expression without impairing blood-feeding or longevity. These findings demonstrate that minor differences in promoter architecture can produce major, unexpected phenotypic effects. OpIE2b provides a robust, phenotypically neutral marker for An. gambiae research, while OpIE2a highlights the need for rigorous validation of transgenic components intended for research and applied releases.

Introduction

The African malaria mosquito Anopheles gambiae is among the most efficient vectors of Plasmodium parasites and remains a major public-health challenge in sub-Saharan Africa. Transmission occurs exclusively through the blood-feeding of adult females, which digest blood to obtain nutrients required for oocyte development and egg production. This link between blood-feeding and reproduction makes female feeding behavior a critical determinant of both the vectorial capacity and mosquito population dynamics.

Since stable germline transformation of Anopheles mosquitoes was first demonstrated in the early 2000s (Catteruccia et al. 2000; Grossman et al. 2001), transgenic strains have become important tools for mosquito research. They now underpin a wide range of studies in mosquito biology, from malaria transmission biology (Pascini et al. 2022) to olfaction and sensory behavior (Riabinina et al. 2016; Maguire et al. 2022; Georgiades et al. 2023; Giraldo et al. 2024) and insecticide resistance (Adolfi et al. 2019). Transgenic approaches are also integral to the development of genetic control strategies, including gene drives designed for population suppression and replacement (Kyrou et al. 2018; Carballar-Lejarazú et al. 2020; Haber et al. 2024; Li et al. 2025; Strampelli et al. 2025; Tolosana et al. 2025; Verkuijl et al. 2025).

These advances depend not only on the ability to introduce and stably integrate transgenes into the mosquito genome but also on reliably identifying transformed individuals. In practice, this is achieved through fluorescent reporters driven by 5′ and 3′ regulatory sequences (hereafter, 5′ regulatory sequences are referred to as promoters for simplicity) that ensure strong and easily detectable expression, most often during larval stages when screening is performed under a fluorescent microscope. An ideal reporter is bright throughout development, spectrally distinct from other markers (eg eGFP vs DsRed), and compatible with multiple constructs, for instance through tissue-specific expression patterns (eg eye- vs muscle-specific). Differences in brightness between homozygotes and heterozygotes can sometimes aid husbandry and genetics, though these are highly dependent on the genome integration site and are more reliably distinguished using automated optical sorters (eg COPAS) than by eye (Marois 2024). Above all, reporters should be phenotypically neutral, imposing no measurable fitness costs that might confound experiments or limit their application. Several promoters have been used to drive fluorescent marker expression in Anopheles gambiae, including the synthetic 3xP3 promoter, which drives strong eye- and neuron-specific fluorescence (Berghammer, Klingler, and Wimmer 1999) and the heterologous Actin5C promoter from Drosophila melanogaster, which provides broad somatic expression (Volohonsky et al. 2015). Endogenous promoters have also been developed, including the polyubiquitin promoter (Adolfi et al. 2018) and the α-tubulin-1b promoter (Lycett et al. 2012), which drive multitissue expression. Viral regulatory elements like hr5-IE1 have also been used to drive ubiquitous expression (Grossman et al. 2001; Meredith et al. 2013). The OpIE2 promoter, derived from the immediate early gene 2 (IE-2) of the Orgyia pseudotsugata multicapsid nucleopolyhedrovirus (OpMNPV) (Theilmann and Stewart 1992), has proven highly effective in diverse insect systems. The OpIE2 promoter works well in lepidopteran and dipteran cell line transfections (Pfeifer et al. 1997; Xu et al. 2013) and in transgenic strains of several insect species, including in Drosophila melanogaster (Buchman et al. 2018; Kandul et al. 2019), Drosophila suzuki (Liu et al. 2025; Witherbee and Gamez 2025), Bombyx mori (Li et al. 2015; Tang et al. 2018), Pectinophora gossypiella (Simmons et al. 2011), Plutella xylostella (Martins et al. 2012), Ceratitis Capitata (Davydova et al. 2023), Aedes aegypti (Li et al. 2017; Estevez-Castro et al. 2024; Rouyar et al. 2024; Vainer et al. 2024), Aedes albopictus (Lutrat et al. 2022; Zaada et al. 2025), and Culex quinquefasciatus (Feng et al. 2021). However, its performance in An. gambiae has not been systematically examined.

Here, we report that expression of DsRed in An. gambiae under a specific variant of the OpIE2 promoter severely impairs female blood-feeding and reduces male longevity. To investigate this effect, we generated a number of strains controlling for genomic integration site and construct elements, characterized female feeding behavior, and tested an alternative OpIE2 promoter sequence. The alternative promoter maintained marker visibility while restoring normal feeding, underscoring that promoter choice can have profound effects on mosquito biology and is critical for both transgenic tool design and genetic control strategies.

Materials and methods

Mosquito rearing

An. gambiae mosquitoes from the G3 (wild-type) strain, and all transgenic mosquito lines were reared under standard conditions of 28 °C ± 2 °C and 80% ± 10% relative humidity. Larvae were fed Essence fish food (Alltech Coppens), while adult mosquitoes were provided with a 10% (wt/vol) sucrose solution. For egg production, adult mosquitoes were allowed to mate for 5–7 d and then fed bovine blood using an artificial feeder (Hemotek, Ltd). An overnight egg-bowl was placed in the mosquito cages 48 h post-blood-feeding for oviposition. Eggs were allowed 48 h to hatch, and first instar larvae (L1) were transferred from the egg-bowl to plastic trays, each containing approximately 250 larvae. The larvae were provided with ground fish food daily until they reached the pupal stage.

Plasmid construction

The OpIE2-DsRed transformation constructs were assembled using Golden-Gate assembly. Both plasmids contained an OpIE2 promoter driving DsRed expression, differing only in the promoter variant used. The OpIE2a promoter fragment (541 bp) and the OpIE2b promoter fragment (652 bp) were PCR-amplified with primers incorporating BsaI recognition sites and compatible overhangs for directional assembly. Each promoter fragment was assembled upstream of the DsRed coding sequence to generate an OpIE2-DsRed cassette. In both constructs, the cassette was flanked by piggyBac inverted terminal repeats and a single attB site, enabling either random piggyBac-mediated or phiC31 site-specific integration. Final plasmids were validated by Oxford Nanopore sequencing (Plasmidsaurus) to confirm promoter sequence identity and integrity of the assembled cassette. Plasmid sequences are available in Supplementary File 1.

Embryo microinjections

Microinjections were performed into eggs of the attP-E strain, a homozygous transgenic line carrying an attP docking site on chromosome 3 and a 3xP3-CFP fluorescence marker. An. gambiae attP-E eggs were gently aligned on a glass slide against a strip of thin nitrocellulose membrane, with moisture maintained using UltraPure Distilled Water (Invitrogen) applied to dampened filter paper. Using a fine brush, eggs were oriented with the posterior end facing upwards to maximize the chances of germline integration. A mixture of 300 ng/µl of the transformation construct and 100 ng/µl of a helper construct encoding vasa-driven phiC31 integrase was injected. Hatched larvae were screened for transient expression of the OpIE2-DsRed marker, and all transient-positive and transgenic individuals were backcrossed to G3 wild-type mosquitoes. Progeny were maintained under standard rearing conditions.

Blood-feeding assay and recording of female behavior

Adult males and females were allowed to mate for 5 to 7 d prior to blood-feeding. For each strain, 3 biological replicates of 14 to 20 females were transferred to Bugdorm minicages (Model BD4S1515) for feeding. Females were offered blood for 25 min using the Hemotek Ltd membrane feeding system, with reservoirs filled with thawed bovine blood and sealed with stretched Parafilm. Feeding sessions were recorded using an Arducam camera system with underlighting to document female behavior for subsequent analysis. Reservoirs were replaced with freshly thawed blood and new Parafilm membranes between assays performed on different strains. If no feeding was observed, a 15 min rest period was followed by a stimulation feeding session of 25 min, using a fresh Hemotek reservoir with fresh blood and a new Parafilm membrane. During stimulation, air was exhaled intermittently onto the reservoir surface to increase female attraction. At the end of each assay, females were anesthetized at 4 °C, and those with visibly blood-filled abdomens were aspirated and counted as successfully fed. Blood-feeding recordings were analyzed using the YOLito detection and tracking framework, a pretrained YOLOv11l-based mosquito detector developed from diverse mosquito imaging datasets collected across multiple laboratories (Sar-Shalom et al. 2025). For inference, detection parameters were set to a confidence threshold of 0.2 and an intersection-over-union (IoU) threshold of 0.5 for bounding-box predictions. Feeder occupancy was analyzed using a linear mixed model with the different strains and time as fixed effects and replicate as a random intercept. Because cages contained different numbers of females, visits counts were divided by the number of females in each cage to obtain a normalized visit rate (Supplementary File 3).

Blood-feeding videos recorded during the assay were additionally analyzed using the open-source event-logging software BORIS—Behavioral Observation Research Interactive Software (Friard and Gamba 2016). For each strain, one video with the highest number of females was selected for analysis (WT: 20 females, replicate R1; OpIE2b: 20 females, replicate R3; OpIE2a: 20 females, replicate R1; see Supplementary file 2). Female behavior was tracked for the first 5 min following placement of the Hemotek feeder in the cage (time = 0). Because individual females could not be distinguished, behaviors were scored per visit to the feeder area. A visit was defined as entry into the area of interest over the Hemotek feeder, with multiple visits by the same female possible. For each visit, arrival and departure times, the number of probing events, and, where applicable, the start and end times of blood-feeding were recorded.

Female fertility assays

Females that were blood-fed during the assay were individually isolated for oviposition. The number of females per strain was determined by the number of OpIE2a-DsRed females that successfully fed after stimulation. Wild-type and OpIE2b-DsRed females were randomly selected to match this number. 48 h post blood-feeding, an egg-bowl was placed in each cage and removed the following day. Egg papers were photographed for counting using Clickmaster2000 (https://github.com/EuracBiomedicalResearch/clickmaster2000). Larvae were provided food 24 h after egg collection, and the number of hatched larvae was recorded 24 h later to calculate hatching rates per female. Progeny from these assays were subsequently reared to adulthood and tested for blood-feeding behavior as described above (filming not performed due to technical limitations).

Fitness and survival assays

To evaluate potential fitness costs of the OpIE2-DsRed cassette, paternal crosses were established between transgenic males and wild-type females, with wild-type intercrosses as controls. Eggs were collected and counted to determine hatching rates. Hatchlings were distributed into trays at a density of 75 larvae per tray and reared under standard conditions. At pupation, male and female pupae were collected and counted. Pupae were allowed to emerge in Bugdorm minicages (Model BD4S1515), and the number of adults was determined by counting remaining pupal cases. These data were used to calculate larval-to-pupal and pupal-to-adult survival rates for each sex and strain, as well as the sex ratio of emerging adults.

Adult longevity was assessed in Bugdorm minicages (Model BD4S1515) containing 30 to 35 adults (24 to 72 h postemergence) per strain, maintained on 10% sucrose solution. Dead adults were recorded and removed daily until all individuals had died. If sex or age at death could not be determined, the individual was censored on the day of discovery. Sucrose solution was replaced with a new, clean feeder every 14 d. Survival data were analyzed using Kaplan–Meier survival curves with log-rank tests.

Mosquito imaging

Photos of larvae, pupae, and adults were captured under standardized conditions to ensure consistent exposure across strains. Larvae at the fourth instar and pupae were imaged using an Olympus SZX16 binocular equipped with a U-HGLGPS fluorescence illumination system. All fluorescence images were acquired using identical exposure settings for each strain. Adult mosquitoes were imaged using a Nikon SMZ25 stereo microscope. Two to three-day-old adults were first anesthetized with CO₂, after which legs and wings were removed to expose the body and proboscis for imaging. Dissected adults were then transferred to 50 mL plastic cups sealed with Parafilm containing silica gel desiccant to prevent condensation and placed at −20 °C for 30 min to achieve full immobilization. Following freezing, cups were thawed briefly at room temperature immediately before imaging. Fluorescence parameters and exposure settings were kept identical between strains to allow direct comparison of DsRed signal patterns.

Statistical analysis

Statistical tests were performed in Python using standard packages (3.10.9; scipy 1.15.3, statsmodels 0.14.4) and JMP Pro 18.0.2 software. All experiments were independently replicated a minimum of 3 times unless stated otherwise, and each replicate represented an independent biological group (eg cage or clutch). The number of individuals or replicates used for each analysis is indicated in the figure panels and methods sections. A summary of statistical tests and P-values is provided in Supplementary File 4.

Results

Generation of OpIE2-DsRed transgenic strains

In previous work, we used an OpIE2-DsRed fluorescent marker in D. melanogaster strains containing split Cas9 and gRNA sex ratio distorters (Haber et al. 2024). Seeking to develop similar capacities in An. gambiae, we cloned this OpIE2-DsRed to label sgRNA transgenic constructs, complementing available Cas9 strains marked by fluorescent proteins driven by the neuron-specific 3xP3 promoter. Four transformation constructs were initially generated, each containing OpIE2-DsRed and a single U6-driven sgRNA targeting different sex chromosome sequences (Fig. 1a). The constructs included piggyBac repeats for transposase-mediated integration and an attB site for phiC31-mediated site-specific recombination. Constructs were injected either into wild-type eggs with a piggyBac transposase helper or into eggs of the attP-E strain, which carries a well-characterized, 3xP3-CFP-labeled attP docking site on chromosome 3 (Meredith et al. 2011). In total, 8 independent strains were generated, 5 with random piggyBac insertion (construct #1; Fig. 1a; Supplementary Table 1), and 3 in the attP-E line (constructs #2–4; Fig. 1a; Supplementary Table 1).

Fig. 1.

Transgenic constructs and marker expression in OpIE2a and OpIE2b promoters driving DsRed. a) Transgenic construct design carrying either OpIE2a, OpIE2b, or Actin5C promoters driving DsRed, with or without a U6-sgRNA cassette. All were flanked by piggyBac inverted repeats and attB recombination sites enabling phiC31-mediated site-specific integration. The genomic location and female blood-feeding phenotype are summarized on the right. b) Brightfield and fluorescence of fourth-instar larvae and pupae expressing DsRed under the OpIE2a promoter. c) Alignment of the OpIE2a and OpIE2b promoter sequences relative to the Orgyia pseudotsugata immediate early-2 (IE-2) gene. Both promoters share a common 461 bp core region, but OpIE2a extends 66 bp into the IE-2 coding sequence, while OpIE2b includes an additional 90 bp upstream region and terminates before the start codon. d) Brightfield and fluorescence of fourth-instar larvae and pupae expressing DsRed under the OpIE2b promoter.

Larvae of all strains displayed bright DsRed fluorescence in thoracic and abdominal segments, and weaker expression in the larval head (Fig. 1b). In pupae, DsRed fluorescence was bright throughout the entire body (Fig. 1b). However, while establishing the strains, we observed that adult transgenic females from all 8 lines failed to blood-feed (Fig. 1a), and oviposition bowls provided 48 h later contained no eggs—a phenotype described in detail below. In contrast, transgenic males appeared physiologically normal and fertile, producing offspring when crossed to wild-type females and enabling maintenance of the strains as heterozygotes. Because the defect was consistent across both piggyBac and phiC31 integrations, we concluded that it likely resulted from the OpIE2-DsRed cassette itself rather than the insertion site or sgRNA sequence, which differed between constructs.

To test this, we generated a new transformation construct in which DsRed was driven by the Actin5C promoter instead of OpIE2. When integrated at the same attP-E site, this construct restored female blood-feeding (construct #5; Fig. 1a; Supplementary Table 1), suggesting that the phenotype was specific to OpIE2-DsRed expression. We confirmed this by generating a new construct containing only the OpIE2-DsRed cassette and integrating it into the attP-E site (construct #6; Fig. 1a; Supplementary Table 1). As this strain exhibited the same nonblood-feeding phenotype, we concluded that OpIE2-DsRed expression alone was sufficient to interrupt female blood-feeding.

When we analyzed the cloned OpIE2 promoter sequence, we noticed that the promoter derived from the Drosophila constructs differed from another OpIE2 promoter that we recently used in Aedes albopictus (Zaada et al. 2025). To better understand these differences, we aligned both promoters, which we called OpIE2a (for D. melanogaster constructs) and OpIE2b (for A. albopictus constructs), to the OpMNPV IE-2 gene region. Both promoters shared a 461 bp core sequence, but there were several differences in the cloned flanking sequences (Fig. 1c). The OpIE2a promoter extended 66 bp into the viral IE-2 coding sequence, including the start codon and fusing 22 amino acids (aa) of IE-2 to the N-terminus of DsRed. The OpIE2b promoter included an additional 90 bp upstream sequence, but terminated prior to the IE-2 start codon in its 3′ end, where it connects to DsRed. The 3′ end of the OpIE2b promoter also contained T7 and EM7 expression elements and restriction sites retained from previous cloning steps.

To test whether these sequence differences accounted for the phenotype, we generated a new strain carrying the OpIE2b-DsRed cassette integrated into the same attP-E site (construct #7; Fig. 1a; Supplementary Table 1). DsRed expression driven by OpIE2b was slightly weaker in larvae, pupae, and adults compared to OpIE2a but followed a similar spatial pattern (Fig. 1d). Importantly, OpIE2b-DsRed females blood-fed normally (Fig. 1a), demonstrating that OpIE2a-DsRed caused the feeding defect.

Blood-feeding capacity is impaired in OpIE2a-DsRed females

To characterize and quantify the blood-feeding defect, we established triplicate cages of 14–20 OpIE2a-DsRed or OpIE2b-DsRed females, previously mated to wild-type males, alongside wild-type control females. Females were offered a membrane feeder containing bovine blood for 25 min, and a camera inside each cage recorded female behavior (Fig. 2a). At the end of the assay, all females were aspirated from their cages to record the number of blood-fed females. OpIE2b-DsRed females fed normally, with feeding rates indistinguishable from wild type females (P = 0.254, unpaired two-sided t-test; Fig. 2b). In contrast, none of the OpIE2a-DsRed females fed during the 25-min window (P = 0.0002, unpaired two-sided t-test; Fig. 2b). To test whether additional stimulation could elicit feeding, OpIE2a-DsRed females were given a 15-minute rest and re-offered fresh blood for an additional 25 min, accompanied by intermittent human breath stimulation to mimic host-associated cues. Eventually, a small fraction of OpIE2a-DsRed females managed to feed (median = 6.25%; 4 females across all replicates; Fig. 2b).

Fig. 2.

Blood-feeding behavior of OpIE2-DsRed females. a) Illustration of the experimental setup for the blood-feeding assay. Females were placed in minicages containing a camera and light source. A membrane feeder filled with bovine blood was placed on top of the cage at the beginning of the experiment. b) Feeding success of wild type, OpIE2b and OpIE2a females. OpIE2b-DsRed females were fed normally, whereas OpIE2a-DsRed females failed to feed. Re-exposure of OpIE2a-DsRed females to a new feeder following 15 min of resting with human-breath stimulation induced only occasional feeding. Boxplots show medians ± IQR; points represent independent replicate cages. N indicates the total number of females tested per treatment. Statistical significance was derived from unpaired two-sided t-tests, assuming equal or unequal variances, comparing the feeding rates between the different strains to the wild-type control (α = 0.05). c) Feeder occupancy over time quantified by a YOLO-based behavioral tracking network. Solid lines show Gaussian-smoothed mean visits per minute ± SEM across (shaded ribbon) replicate cages. OpIE2a-DsRed females spent significantly less time on the membrane-feeder, and their visits declined rapidly compared to WT and OpIE2b-DsRed controls. d) Manually annotated behavioral ethograms during the first 5 min of the blood-feeding assay: Bars represent sequential visits by individual females, with segments indicating periods spent on the feeder, probing events, and blood-feeding. OpIE2a-DsRed females displayed repeated short probing attempts without transition to sustained feeding.

Recordings of female blood-feeding behavior were analyzed using YOLito, a generalized mosquito detection model built on the YOLOv11l architecture and trained on a globally assembled dataset of mosquito images (Sar-Shalom et al. 2025). At the beginning of the assay, OpIE2a-DsRed females exhibited significantly fewer visits per female than wild type (β = −0.312 ± 0.098 SE, P = 0.001), whereas OpIE2b-DsRed females showed a modest but significant increase in visit rate (β = 0.236 ± 0.098 SE, P = 0.016) (Fig. 2c). Over time, OpIE2a-DsRed females showed a markedly steeper decline in feeder visits relative to wild-types (interaction β = −0.010 ± 0.002 SE, P < 0.001), while OpIE2b-DsRed females did not differ significantly from controls (β = −0.003 ± 0.002 SE, P = 0.153) (Fig. 2c). Although OpIE2b-DsRed females began at a slightly higher baseline, their overall visit trajectory closely resembled that of wild-type females.

Behavioral ethograms for the first 5 min of each assay were also generated, annotating arrival, probing, blood-feeding, and departure events for each female. Wild-type and OpIE2b-DsRed females began probing shortly after arrival and remained stationary once feeding commenced, spending significantly more time on the feeder overall (Fig. 2c; Supplementary Fig. 1). OpIE2b-DsRed females showed a modest reduction in probing frequency and slightly longer feeding durations relative to wild-type (Supplementary Fig. 1), although these differences occurred in the context of otherwise normal feeding behavior. In contrast, OpIE2a-DsRed females displayed normal attraction to the feeder, probing the membrane at a similar frequency to wild-type females, but all failed to initiate blood-feeding. Instead, they performed repeated cycles of unsuccessful probing followed by proboscis grooming, ultimately leaving the feeder unfed after brief visits (Fig. 2c; Supplementary Fig. 1). These observations indicate that the OpIE2a-DsRed defect impairs feeding initiation rather than attraction.

OpIE2a-DsRed adults exhibit a curved proboscis

To investigate whether the behavioral defect corresponded to morphological abnormalities, we examined the anterior feeding and sensory appendages of adult mosquitoes by stereo-fluorescence microscopy. All OpIE2a-DsRed females displayed a conspicuous distal curvature of the proboscis and maxillary palps, in contrast to the straight, parallel alignment characteristic of wild-type and OpIE2b-DsRed females (Fig. 3a–c; Supplementary Fig. 2). The curvature averaged 21.3° along the distal end of the proboscis, forming an abrupt ventral deflection. The paired maxillary palps also appeared bent at the same position as the proboscis.

Fig. 3.

Proboscis morphology and DsRed fluorescence in Anopheles gambiae adults carrying OpIE2-driven markers. a–c) Females and d–f) males from wild-type (WT), OpIE2b-DsRed, and OpIE2a-DsRed strains imaged under bright-field (left) and fluorescence (middle, grayscale intensity; right, single-channel DsRed fluorescence) illumination. a, d) WT adults display a straight proboscis with no fluorescence. b and e) OpIE2b-DsRed adults show strong, widespread DsRed expression but retain normal proboscis morphology. c and f) OpIE2a-DsRed adults exhibit a pronounced distal curvature of the proboscis and a strong DsRed signal localized on the proboscis and maxillary palps. Images were acquired under identical exposure settings to permit direct comparison of fluorescence intensity and anatomical deformation across strains and sexes.

Because the feeding defect initially appeared female-specific, we also examined males as controls (Fig. 3d–f). Unexpectedly, OpIE2a-DsRed males exhibited the same deformity, with curvature of both the proboscis and palps even more pronounced than in females (mean of 39.2°). The male proboscis often appeared twisted or bent laterally at the tip, and in severe cases, the palps crossed over one another, partially covering the proboscis. In contrast, OpIE2b-DsRed males retained normal morphology, with straight, evenly aligned proboscis and palps indistinguishable from wild-type.

Fluorescence imaging revealed notable DsRed signals within the proboscis and maxillary palps of OpIE2a-DsRed adults, whereas OpIE2b-DsRed mosquitoes exhibited faint fluorescence in the maxillary palps under identical exposure settings. The co-occurrence of a strong DsRed signal and abnormal curved morphology in the mouthparts of OpIE2a-DsRed individuals suggests that differences in promoter sequences, including most notably the 22 aa fusion, likely alter the accumulation, stability, and impact of DsRed in the physiology of anterior appendages.

OpIE2a-DsRed males exhibit reduced adult longevity

To determine whether the OpIE2a-DsRed construct was associated with broader fitness costs, we assessed reproductive output and survival in both sexes. Among the few OpIE2a-DsRed females that successfully blood-fed after stimulation (N= 4), only 2 laid eggs. The number of eggs produced by these females was markedly lower than in controls, with a median of 28 eggs compared to 97.5 in wild-type and 74.5 in OpIE2b-DsRed females (unpaired two-sided t-test, P = 0.042; Fig. 4a). In contrast, egg hatching rates did not differ significantly among strains (Fig. 4b), indicating that fertilization and embryonic development were unaffected. When the blood-feeding assay was repeated with daughters of these females, none successfully blood-fed, confirming that the phenotype was heritable and fully penetrant (Supplementary Table 2).

Fig. 4.

Reproductive output and fitness of OpIE2-DsRed strains. a and b) Reproductive performance of females that are successfully blood-fed. a) Number of eggs laid per female and b) corresponding hatching rate (% L1 per egg batch). Each point represents an individual female; Boxplots show medians ± IQR; N indicates the total number of eggs and hatched larvae tested per female. Statistical significance was derived from unpaired two-sided t-tests, assuming equal or unequal variances, comparing the feeding rates between the different strains to the wild-type control (α = 0.05). OpIE2a-DsRed females produced significantly fewer eggs than wild-type (WT) or OpIE2b-DsRed females, while hatching rates did not differ significantly. c–f) Fitness parameters from crosses in which transgenic males were mated to WT females. c) Hatching rate (% L1 per egg batch), d) larval-to-pupal survival (%), and e) pupal-to-adult survival (%) were comparable across strains. Boxplots show medians ± IQR; points represent replicate cages. f) Adult sex ratios of progeny from the same crosses. Stacked bars show the mean proportion of males and females. Statistical comparisons were calculated relative to WT control using unpaired two-sided t-tests; no significant differences were detected (all P > 0.05). g and h) Adult longevity of males g) and females h). OpIE2a-DsRed males exhibited a markedly reduced lifespan compared to WT and OpIE2b-DsRed males, while females showed a moderate but significant reduction in survival. Survival curves were analyzed using Kaplan–Meier estimates and compared using log-rank statistics. Diamonds denote median survival per strain.

To evaluate potential effects on male fitness, we crossed transgenic males from each strain to wild-type females, with wild-type intercrosses as controls. Egg hatching, larval-to-pupal, and pupal-to-adult survival rates were comparable across all groups (unpaired two-sided t-tests, all P > 0.05; Fig. 4c–e), and adult sex ratios did not differ significantly (Fig. 4f). However, adult longevity was severely reduced in OpIE2a-DsRed males, which had a median lifespan of only 6 d compared to 30 d in wild-type males (log-rank test, P < 0.001; Fig. 4g). OpIE2b-DsRed males showed no difference in survival relative to controls (median 25 d vs 30 d, P = 0.355). Among females, OpIE2a-DsRed also exhibited a modest reduction in survival (median 20 d vs 25 d in wild type, P = 0.035), whereas OpIE2b-DsRed females survived comparably to controls (median 28 d vs 25 d, P = 0.165, Fig. 4h). Together, these results overturned our initial conclusion that the OpIE2a-DsRed marker negatively affected fitness in a female-specific manner, by being detrimental to adult longevity, particularly among males, whereas the OpIE2b-DsRed variant maintains normal fecundity, survival, and developmental progression.

Discussion

In this study, we demonstrate that expression of the fluorescent marker DsRed under the control of a specific variant of the OpIE2 promoter (OpIE2a) causes profound and unexpected phenotypic effects in An. gambiae. Females carrying the OpIE2a-DsRed marker were unable to blood-feed and were therefore sterile, while males exhibited markedly reduced adult lifespan. In contrast, an alternative promoter variant, which we called OpIE2b, drove robust marker expression without measurable effects on feeding, development, or longevity. These results reveal that even minor differences in promoter architecture can produce major biological consequences, emphasizing the need to validate commonly used regulatory elements when deployed in new species or experimental contexts.

The female blood-feeding impairment associated with OpIE2a was fully penetrant and heritable, persisting across generations—some lines are now over 30 generations old. Behavioral analyses showed that females oriented toward the feeder and initiated probing, but left soon after failing to transition to blood-feeding, spending less time on the feeder overall. Microscopic examination identified a fully penetrant curvature of the proboscis and maxillary palps in OpIE2a adults of both sexes, suggesting a structural rather than sensory or motivational basis for the defect. This distal bend likely prevents proper fascicle unsheathing and membrane penetration of the proboscis (Jones 1978; Kong and Wu 2010), explaining the repeated probing-grooming cycles and failure to feed. In males, the same curvature is more pronounced and may interfere with sugar uptake, consistent with their severely shortened lifespan.

The difference between the OpIE2a and OpIE2b promoters lies in a 66 bp extension of OpIE2a into the viral IE-2 coding region, which adds 22 aa fused to the N-terminus of DsRed. Although both variants contain the core regulatory motifs required for promoter activity (Theilmann and Stewart 1992), this fusion likely alters the transcriptional or translation context, potentially modifying expression strength, tissue specificity, or protein localization. The presence of DsRed fluorescence within the proboscis and maxillary palps of OpIE2a mosquitoes supports this hypothesis and suggests that expression strength, or toxicity of the OpIE2a-DsRed marker, underlies the morphological phenotype. While we did not directly measure promoter activity in dissected tissues or test other reporters under the same sequence, the data suggest that even seemingly small differences in cloned promoter sequences can markedly alter the impact of marker expression in transgenic insects, as we have previously observed a native germline promoter in transgenic An. gambiae strains (Papathanos et al. 2009). A recent study also reported hr5-IE1-associated fitness effects in An. stephensi (Larrosa-Godall et al. 2025), indicating that promoters can exhibit species-specific physiological impacts.

Beyond its mechanistic implications, this finding carries practical significance for transgenic mosquito research and genetic control tool development. First, transgenic reporters are typically assumed to be phenotypically neutral (Catteruccia et al. 2003; McArthur et al. 2014); however, as we have shown here, this assumption needs to be confirmed each time independently, since even standard markers can impose unexpected fitness costs when newly introduced into an engineered insect (Irvin et al. 2004; Marrelli et al. 2006).

In the case of OpIE2a-DsRed, the apparent female-specific sterility initially seemed attractive for engineering dominant female lethality, eg in self-limiting or split gene-drive systems (Champer et al. 2022; Verkuijl et al. 2025). However, the severe reduction in male longevity makes this configuration unsuitable for practical applications of population-suppression gene drives in its current form. If expression could be restricted to females, eg through the use of sex-specific splicing (Scali et al. 2005; Fu et al. 2010; Spinner et al. 2022; Weng et al. 2023, 2024; Liu et al. 2024), the dominant appendages phenotype could be repurposed as a female-specific lethal trait in future designs.

Finally, the identification of OpIE2b-DsRed as a functionally neutral marker with bright, broad expression provides a valuable addition to the transgenesis toolkit available for An. gambiae. Because its expression pattern is distinct from commonly used markers such as 3xP3 or Actin5C, OpIE2b is well suited for multiplexed or combinatorial transgenic designs, including split gene drives and multicolor labeling systems.

In conclusion, our work reveals that promoter sequences alone can profoundly influence expression and phenotype in transgenic insects, transforming an otherwise inert fluorescent reporter into a dominant morphological impairment of mosquito physiology. This work offers both a cautionary example for construct design and a validated promoter variant (OpIE2b) suitable for future An. gambiae genetic engineering and gene-drive applications.

Data availability

All data necessary for confirming the conclusions of the article are present within the article, figures, and tables. Strains and plasmids are available upon request.

Supplemental material available at GENETICS online.

Author contributions

AS and PAP conceived and designed the study. AS, ESY, ESMK, and LBL generated transgenic mosquito lines. AS performed behavioral, imaging, and fitness assays and analyzed the data. ARL helped with figure illustrations. ES-S and JB developed the trained behavioral detection and classification neural network. ESMK, ESY, and LBL assisted with mosquito genetic engineering. TP and GP contributed to adult proboscis imaging. PAP supervised the project, secured funding, and wrote the manuscript together with AS with input from all authors. All authors read and approved the final manuscript.

Funding

This work was supported by research grants from the Gates Foundation (INV-004363 and INV-075374 to PAP) and the Israel Science Foundation (2388/19 to PAP). Additional support was provided by the Israel Science Foundation (719/21 to JB).

Ethics

All animals were handled in accordance and under the supervision of the ARO Institutional Animal Care and Use Committee approval number 2307-118-2-VOL-IL. All insect work was performed in facilities maintaining Arthropod Containment Level 2. This work received Institutional Approval and relevant authorizations from the Israel Ministry of Environmental Protection and Ministry of Agriculture (#31/2019).