Tralopyril, the active metabolite of chlorfenapyr, was used directly to minimise confounding effects associated with metabolic conversion of the parent compound. ⁷ To determine a physiologically relevant dose for in vivo exposure assays, a topical dose-response assay was conducted on An. stephensi mosquitoes with concentrations from 6-215 μM, alongside an acetone-only control. Mortality was monitored for 10 days post-exposure to capture survival across key Plasmodium developmental stages. Most mortality occurred within the first 72 h and stabilised thereafter ( Extended Data Figure 1A), yielding LC ₃₀ = 31 μM, LC ₅₀ = 44 μM and LC ₇₀ = 63 μM at 72 h.

To ensure these concentrations were physiologically relevant, uptake was measured by LC-MS across five timepoints (30 min, 4 h, 24 h, 48 h, 72 h; Supplementary Table 1) and the LC ₃₀ was chosen for subsequent experiments as it mirrored published data on expected pyrethroid uptake. ²⁵ Next, LC-MS was performed on the LC ₃₀ dose at 21°C and 28°C, the temperatures used for P. berghei and P. falciparum infection . Interestingly, temperature had no impact on uptake though higher temperature accelerated clearance. At 21°C no appreciable reduction in internalised tralopyril over the 10-day period was found ( Figure 1B). Topical application of a 33 μM chlorfenapyr, the parent compound, resulted in level of tralopyril internalisation similar to that observed when tralopyril, was applied as a single component, again reinforcing biological relevance of the selected dose ( Extended Data Figure 1B).

We assessed all energy-dependent stages in parasite development, commencing with early sporogenic development. To avoid feeding impairment due to acute intoxication, An. stephensi mosquitoes were topically exposed to tralopyril 48 h prior to the P. berghei infectious blood meal, ensuring that the insecticide remained present in the mosquito at the time of parasite exposure ( Figure 1B, Extended Data Figure 1C). Parasite development was then assessed at six, nine and 12-days post-infection (dpi), measuring oocyst prevalence, infection intensity and size ( Figure 1C). Mosquito survival was monitored throughout the infection period. Mortality was higher in the tralopyril group, consistent with LC ₃₀ exposure, but sufficient to ensure relevant sample sizes for parasite development analyses ( Extended Data Figure 1D).

Plasmodium infection prevalence was 92% (101/110) in control mosquitoes and 84% (90/107) in tralopyril-exposed mosquitoes, pooled across 6, 9 and 12 dpi (p _{Fisher’s exact test} = 0.1). Oocyst numbers were quantified per midgut and analysed using a binomial generalised linear model (GLM). On day six post-infection, median oocyst counts were significantly lower in tralopyril-exposed mosquitoes (median = 32, IQR: 2 – 75, n = 37) compared to controls (median = 73, IQR: 15 – 146, n = 37; p _GLM = 0.02); however, no significant differences were observed at later timepoints: day nine (control: median = 69, IQR: 9 –120; tralopyril: 35, IQR: 6–83; p _GLM = 0.8) and day 12 (control: 70, IQR: 54 –106; tralopyril: 40, IQR: 13 – 121; p _GLM = 0.2). When data was pooled across all days, a modest but significant decrease in oocyst burden was observed in the tralopyril-exposed mosquitoes (control: median = 70, IQR: = 16-121; tralopyril: median = 35, IQR: 6-81; P=0.04) ( Figure 1D, Extended Data Figure 1E). These findings suggest that tralopyril may partially impair oocyst formation.

To investigate whether the observed reduction in oocyst numbers in vivo could be attributed to impaired ookinete invasion, we assessed the direct effect of tralopyril exposure of P. berghei ookinete motility in vitro.

Purified ookinetes were exposed in vitro to tralopyril or a DMSO control and motility was assessed after one and two hours post-exposure. ²⁶ A significant reduction in the proportion of motile ookinetes was observed at two-hours post-exposure (p _{Fisher’s exact test} = 0.01), whilst no significant difference was observed at the one-hour timepoint (p _{Fisher’s exact test} = 0.4) ( Figure 1E). Tralopyril-exposed ookinetes exhibited significantly increased speed compared to controls at one-hour (p _{unpaired t-Test with Welch’s correction} < 0.0001) but not at two-hours (p _Mann-Whitney = 0.8; Figure 1F).

Together, these findings suggest that short-term exposure to tralopyril in vitro might alter ookinete motility dynamics in a time-dependent manner, characterised by transient hypermotility followed by reduced motility, which may partially impair midgut invasion and oocyst establishment ( Figure 1E).

To determine whether tralopyril exposure influences oocyst development, we next analysed oocyst size ( Figure 1G). At six dpi, oocysts were smaller in the tralopyril-exposed group compared to controls (mean area ± SD: control 500 ± 110 μm ² vs. 430 ± 91 μm ²; p _{Mann–Whitney test} = 0.004). By nine dpi, no significant difference in oocyst size was observed (1470 ± 260 μm ² vs.1390 ± 360 μm ²; p _{unpaired t-Test done with Welch’s correction} = 0.3). However, by 12 dpi, oocysts in the tralopyril group were markedly larger than controls (2200 ± 450 μm ² vs. 3570 ± 1300 μm ²; p _{Mann-Whitney test on transformed data} < 0.001). These findings suggest that although tralopyril exposure reduced oocyst numbers, those that developed could grow to a larger size ( Figure 1G). Interestingly, larger oocysts can be indicative of disruption to sporozoite formation. ²⁷ Taken together, we show that indeed tralopyril disrupts parasite development, reducing ookinete motility, decreasing oocyst numbers, and altering oocyst growth dynamics.

To assess downstream effects on sporozoite development, we used a P. berghei line expressing cytosolic GFP under the control of the circumsporozoite protein (CSP) promoter, which becomes active around seven-nine dpi and thus serves as a marker of early sporozoite formation. ²⁸ Mosquitoes were dissected at the relevant timepoints and GFP positivity was recorded based on the presence of detectable fluorescence in whole midguts or salivary glands ( Figure 2A-B). Samples with no visible fluorescence above background levels were classified as GFP-negative. In mosquitoes exposed to tralopyril, no GFP fluorescence was detected in midguts on nine- or 12- dpi. Fluorescence was observed in only three percent of midguts (1/38) and four percent (2/46) at days 18 and 20 dpi, respectively, in stark contrast to controls (77-92%) across all timepoints ( Figure 2A). Similarly, no fluorescent signal was detected in salivary glands at 18 dpi, with two percent (1/45) fluorescent at 20 dpi, versus 79-84% in controls ( Figure 2B). GFP prevalence was significantly reduced in tralopyril-exposed mosquitoes relevant to controls at all timepoints (p _{Fisher’s exact test}< 0.0001). These data indicate an almost complete block in sporozoite formation despite apparent larger oocysts after exposure to tralopyril.

To investigate the presumed developmental arrest observed in sporozoite formation following tralopyril exposure we applied ultrastructure expansion microscopy (U-ExM) ²⁹ to midguts dissected at 12 dpi ( Figure 2C, Extended Data Figure 2). We stained the samples with an antibody against the major surface protein CSP, Hoechst as a nuclear DNA marker and NHS ester staining as a general protein stain.

In control oocysts, sporozoites were clearly visible indicating that the oocyst had undergone successful sporulation ³⁰ ( Figure 2C). CSP signal was predominantly localised to the sporozoite surface membrane, as expected. ³¹ Individual nuclei could be observed in the sporozoites, consistent with normal differentiation. In contrast, tralopyril-exposed midguts revealed oocysts with arrested development ( Figure 2C), indicating an early abortion of development prior to membrane invagination, a first step in sporogony. ³¹ Further, CSP signal was distributed only around the periphery and only a few distinct nuclei could be seen.

These findings were corroborated by transmission electron microscopy ( Figure 2D, Extended Data Figure 3), which revealed a comparable arrest in oocyst development on day 12 dpi. No sporulated oocysts were observed in the tralopyril group, representing 0% (0/8), compared to 83% (15/18) in controls. In control midguts, sporozoites were surrounding the sporoblast, and unsporulated oocysts had nuclei, mitochondria and spindle-like microtubules ( Extended Data Figure 3). Tralopyril-exposed oocysts, however, exhibited a growth-phase arrest, although spindle-like structures were apparent, indicative of DNA replication despite the developmental block.

Given tralopyril’s mitochondrial target, we examined parasite mitochondria, comparing them with unsporulated oocysts in controls ( Figure 2D, Extended Data Figure 4). Whilst overall mitochondrial morphology appeared broadly similar, tralopyril-exposed parasites exhibited clear alterations in the appearance of cristae, the site of respiration, with cristae appearing swollen and less circular in shape. Quantification of mitochondrial cristae circularity confirmed a significant reduction in circularity in tralopyril-exposed parasites (p _{Kolmogorov-Smirnov test} = 0.02, n =230; Figure 2E). Together, these findings demonstrate that tralopyril exposure causes a profound developmental arrest in early oocyst development, prior to plasma membrane invagination and sporoblast formation, and that a direct effect of the compound can be observed on parasite mitochondrial cristae.

To test whether tralopyril-induced developmental arrest was reversible, perhaps due to restoration of energetic resource, mosquitoes received two additional non-infectious blood meals on three and six dpi. GFP positivity was again scored in midguts on 12 and 20 dpi, and salivary glands on 20 dpi ( Figure 3A). On 12 dpi, GFP prevalence was detected in 10% (6/63) of tralopyril-exposed mosquito midguts, increasing to 30% (12/41) by 20 dpi ( Figure 3B), higher than in previously exposed mosquitoes without additional feeds ( Figure 2A). In contrast, GFP prevalence in controls ranged from 73-93%. Despite this partial increase in midgut GFP signal, no fluorescence was observed in salivary glands of exposed mosquitoes at 20 dpi, compared to 86% in controls ( Figure 3C). Moreover, GFP intensity in midguts of tralopyril mosquitoes was lower than in controls, suggesting defective oocyst development and impaired sporulation ( Figure 3D). GFP prevalence was significantly reduced in tralopyril mosquitoes relevant to controls at all timepoints (p _{Fisher’s exact test}, P < 0.0001). These data indicate that additional food resources can allow the parasite to recover to an extent but not enough to develop sporozoites in the salivary glands; this lack of recovery will likely be more pronounced in field settings where mosquitoes are competing for resources and thus have less energetic reserves. ³²

Whilst tralopyril treatment results in an almost complete loss of sporozoite formation, egressed sporozoites from prior infection may still encounter the compound post-formation. Thus, potential direct effects on sporozoites were explored through impacts on motility. Plasmodium sporozoites migrate in vitro by gliding rapidly in circular patterns, an energy-intensive process powered by an ATP-dependent actin-myosin motor, ^{33 , 34} essential for salivary gland invasion and transmission. To test an effect on sporozoites, mature P. berghei sporozoites were incubated in vitro with a range of tralopyril concentrations (0.003-64 μM), and gliding motility was assessed at 15- and 60-min post-exposure and compared to a DMSO control (Supplementary Tables 2 - 4). Productive motility, defined as continuous circular movement on a glass surface, was significantly reduced in a dose (p _{Linear regression} = ns at 0 min; p < 0.0001 at 15 and 60 min) and time-dependent (Non-linear regression, pairwise Z-tests on x ₀: all comparisons p < 0.0001) manner ( Figure 3E, Extended Data Figure 5A, Supplementary Tables 2-4), including at concentrations which mirrored published data on pyrethroid uptake ²⁵ and at the LC ₃₀ dose (31 μM) used for topical applications ( Figure 3E). At 15 min, a significant reduction was seen at 1 μM (p _{2-way ANOVA and multiple comparison} < 0.01), whilst by 60 min, impairment was evident at concentrations as low as 0.3 μM (p _{2-way ANOVA and multiple comparison} < 0.01). Although reductions at lower concentrations (<0.3 μM) were not statistically significant at 60 min (p _{2-way ANOVA and multiple comparison} > 0.05), a downward trend in motility was apparent across the full dose range. Similar impairments were induced by carbonyl-cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), a known mitochondrial uncoupler, supporting the hypothesis that tralopyril-mediated mitochondrial dysfunction underlies the motility defect ( Extended Figure 5B, Supplementary Table 5). As sporozoites glide for a limited period of time, ³⁵ it was not possible to assess the long-term impacts of tralopyril as seen in Figure 1B and thus the impact is likely to be more pronounced in vivo. Confocal scanning laser microscopy revealed that a direct incubation of sporozoites with tralopyril showed a dispersal of mitochondrial activity marker CMXRos throughout the sporozoite, consistent with loss of mitochondrial membrane potential; this phenotype was comparable to that observed with FCCP ( Extended Data Figure 5C). These findings show that additional blood meals cannot rescue oocyst maturation after tralopyril exposure and that mature sporozoites exposed to the compound are unlikely to be transmissible.

Given the potent transmission-blocking effect of tralopyril in P. berghei, we next investigated whether a similar block could be observed in the human malaria parasite Plasmodium falciparum. The experimental design mirrored that of the P. berghei assays: An. stephensi mosquitoes were exposed topically to 31 μM tralopyril 48 h prior to a blood meal containing a standardised number of stage V gametocytes. Midguts were dissected on 10 dpi to quantify oocyst burden, and salivary glands on 18 dpi to assess sporozoite development ( Figure 4A). Strikingly, no oocysts were observed in tralopyril-exposed mosquitoes, representing a significant reduction in oocyst burden (p _Mann-Whitney < 0.00001; Figure 4B). Infection prevalence was 65% (35/54) in the control group and 0% (0/45) in tralopyril-exposed mosquitoes (p _{Fisher’s exact test} < 0.00001), indicating a complete block in oocyst formation following tralopyril exposure, in contrast to P. berghei. Accordingly, no sporozoites were recovered from any tralopyril-exposed mosquitoes (n = 55, 11 pools; Figure 4C), indicating a complete block in parasite transmission (p _{Mann-Whitney test} < 0.0001). In contrast, control mosquitoes exhibited productive infections, with a median of 9700 sporozoites per mosquito (IQR 7000-12550, n = 45, 9 pools). Together, these data reveal a potent inhibitory effect of tralopyril on P. falciparum development in the mosquito, with the complete absence of transmissible sporozoites and indicates a severe impairment of ookinetes or early oocyst formation.

To determine whether the parent compound chlorfenapyr, can similarly impact P. falciparum development, we next assessed its effect on oocyst burden. Chlorfenapyr was applied at the LC ₅₀ concentration of 33 μM ( Extended Data Figure 5D). LC-MS revealed that internalisation and conversion of chlorfenapyr to its active metabolite tralopyril, peaked between 48 – 72 h post-exposure. Chlorfenapyr was not detectable at 72 h, consistent with complete metabolic conversion ( Extended Data Figure 1B).

An. stephensi mosquitoes were topically exposed to chlorfenapyr (33 μM), 72 h prior to an infectious meal with P. falciparum, and mosquito midguts were dissected at 10 dpi to determine the oocyst burden ( Figure 4A). Infection prevalence was reduced from 72% (36/50) in the control to 12% (6/49) following chlorfenapyr exposure (p _{Fisher’s exact test} < 0.00001; Figure 4B). Oocyst number per midgut were significantly decreased in chlorfenapyr-exposed mosquitoes (median = 0, IQR: 0 – 0, n = 49) compared to controls (median = 5, IQR: 0 – 19, n = 50; p _GLM < 0.0001; Figure 4B).

To assess whether chlorfenapyr also influenced oocyst development, we compared oocyst size in the six mosquito midguts where oocysts were detected with those of control mosquitoes ( Figure 4D). Interestingly, oocyst size per midgut was significantly larger in the chlorfenapyr- exposed mosquitoes compared to the controls (mean area ± SD: control 4920 ± 1268 vs. 10267 ± 4725, p _{Mann-Whitney test on transformed data} = 0.0003; Figure 4D). Together, these data indicate that chlorfenapyr exposure substantially reduces P. falciparum infection prevalence and oocyst number, whilst those that developed grew to a larger size. Notably, this phenotype mirrors that observed in P. berghei following tralopyril exposure, with enlarged oocysts and arrested oocyst development and sporozoite production. Collectively, we show that chlorfenapyr and active-metabolite tralopyril disrupts parasite development by decreasing oocyst numbers, and altering oocyst growth dynamics.

Presumed mated, adult female mosquitoes of the species An. stephensi were used for all experiments; this strain is fully susceptible to insecticides (SDA 500, MRA-1326, contributed by Peter F Billingsley, Sindh Province, Pakistan). Mosquitoes were reared at Heidelberg University in standard insectary conditions (26-28°C), 80% relative humidity, light:dark cycles of 12 hours each, one hour dawn dusk). Larvae were reared in 0.1% salt solution in large trays and fed on ground fish food (Tetramin, Germany) and adults were fed 10% sucrose solution ad libitum. Colony cages were maintained by regular blood-feeding using a Hemotek feeder (Hemotek Ltd, UK) with reconstituted human blood.

Insecticide stock solutions were prepared at 10 mg/mL by dissolving in acetone and dilutions were made from this solution to achieve the desired test concentrations. Tralopyril and chlorfenapyr were provided by BASF. Mosquitoes were reared as described above, and two-to-five-day old An. stephensi mosquitoes utilised. Nine concentrations of tralopyril ranging from 6 μM to 215 μM were tested alongside an acetone-only (control) group. Mosquitoes were anesthetised on ice, and the insecticide was applied directly to the cuticle of the mosquito using a gastight Syringe (Hamilton Company, USA) as previously described. ⁶⁷ Mosquitoes were stored in cups and mortality was scored every 24 h over a ten-day period. For each condition 25 mosquitoes were exposed and the experiment was performed in four biological replicates. For chlorfenapyr, 12 concentrations ranging from 1 μM – 86 μM were tested alongside an acetone-only control group, using the same experimental procedures. These experiments were conducted in two – three biological replicates. Dose-response data were fitted to a sigmoidal four-parameter logistic (4PL) model in GraphPad Prism 10 (GraphPad Software, La Jolla, California USA) to calculate lethal concentrations and 95% confidence intervals.

Insecticide stock solutions of the 31 μM (LC ₃₀), 44 μM (LC ₅₀) and 63 μM (LC ₇₀), were prepared as above. Twenty two-to-five-day-old An. stephensi mosquitoes per concentration and timepoint were topically exposed, ⁶⁷ and an acetone-only group was exposed in parallel. Mosquitoes remained in an incubator set at 21°C and 80% humidity. Samples were collected from five different time points: 30 minutes, four hours, 24, 48 & 72 h. The dead or alive phenotype of each mosquito was recorded, and all sample were snap-frozen in liquid nitrogen to arrest metabolic activity at the designated timepoints. These samples were pooled into groups of five, forming four technical replicates per condition for each timepoint. Samples were shipped to BASF (Chicago, USA) to undergo mass spectrometry analysis using LC-MS mass spectrometry.

Based on initial LC-MS results, the LC ₃₀ dose of tralopyril and LC ₅₀ dose of chlorfenapyr was selected for longitudinal studies to assess uptake and retention over 10 d. Four experimental groups were included: mosquitoes exposed to tralopyril and maintained at 21°C, mosquitoes exposed to tralopyril and maintained at 28°C, mosquitoes exposed to chlorfenapyr (Sigma) and maintained at 21°C and an acetone-only group. Insecticide solutions were made as previously described. For each insecticide-exposed group, fifteen two–to-five-day-old mosquitoes were included per timepoint. Samples were collected at 30 min, four hours and every 24 h post-exposure up to 10 d. Mosquitoes were pooled in groups of five, providing three technical replicates per condition. Pooled samples were shipped to BASF (Wyandotte, USA) for LC-MS analysis.

Mosquitoes were stored at -80°C immediately upon arrival from at BASF (Wyandotte, USA) from Heidelberg University. To prepare mosquitos samples for LC-MS quantification, a two-phase extract protocol was used to separate the amount of insecticide found on the exterior of the mosquito body vs the interior. The extraction protocol proceeded as follows: 400 μL of acetonitrile was added to rinse exterior of mosquito body, then decanted into a clean, labelled safe-lock Eppendorf tube. This 400 μL solution represents the “rinse” extraction and was saved to later be transferred to an High-Performance Liquid Chromatography (HPLC) vial for LC-MS analysis. Next, the mosquito pool was rinsed two more times with 100 μL of acetonitrile to ensure no residual insecticide remained on the exterior of the mosquito bodies. These two rinse solutions were discarded. To generate the internalised insecticide extract, representing, the amount of insecticide that is found in the interior of the mosquito body, 400 μL of acetonitrile was added into the now fully rinsed, pooled mosquito sample. Sample pool was homogenized using the Bullet Blender ^® for three minutes at a speed of 10 and then sonicated for 10 minutes. This sample solution is referred to as the “homogenized” extract, representing the internalised amount of the insecticide. Samples where then centrifuged for 30 minutes at 14.5K rpm before a 100 uL aliquot of the supernatant from the homogenized mosquito extract into a labelled HPLC vial.

For each pooled mosquito sample, two LC-MS samples were analysed (homogenised & rinsed) using optimised liquid chromatography and multiple reaction monitoring (MRM) methods on a SciEx Triple Quad 6500+. A series of standards for chlorfenapyr and tralopyril were prepared in acetonitrile ranging from 5-800 ng and 0.05-50 ng, respectively. These standard solutions were run first, followed by the mosquito samples. All samples were run on the same Agilent SB-C18 Zorbax column (3.0 × 100 mm, 3.5 μm). For quantification, all raw data was uploaded into SciEx MultiQuant 3.0.2 for calibration curve and quantification calculations. Each raw spectrum was manually checked to confirm peak shape and retention time was correct. Quantification output was exported into excel files and data was organised as reported in this manuscript.

Infectious feeds and topical exposure

Anopheles stephensi mosquitoes were reared as described above. 24 hours prior to topical exposure mosquitoes were acclimated to 21°C, which was maintained throughout the experiment. For all experiments, two-to-four-day-old mosquitoes were used, with 300 mosquitoes exposed in total (150 per group) and placed in cages (17.5 × 17.5 × 17.5 cm). The tralopyril group was topically exposed to 31 μM tralopyril, while the control group was exposed to acetone only, following the protocol described in Figure 1C.

Mouse infections with P. berghei

Female CD1 Swiss mice were infected via intraperitoneal injection with a cryostock of P.berghei csGFP (2-5% parasitaemia). ²⁸ Five days post-infection, a full blood transfer (FBT) was performed and the parasitaemia of the mouse was counted. The mouse was bled via cardiac puncture from anesthetised mice (120mg/kg ketamine, 16 mg/kg xylazine) where 10 million parasites were transferred to a second naïve mouse via intraperitoneal injection. Four days after the FBT, male gametocyte exflagellation was assessed from a blood smear at 21°C. If exflagellation was sufficient, the mouse was anesthetised with 100 mg/kg ketamine and 3 mg/k xylazine, and used for the infectious blood. During feeding, the mouse was rotated between control and tralopyril-exposed cages. Temperature and humidity were maintained at 21°C and 80%, optimal for P. berghei development. Unfed mosquitoes were removed, and mortality was scored daily; dead mosquitoes were removed from the cages. Survivorship curves were generated and analysed using the Mantel-Cox log-rank test in GraphPad Prism 10 (GraphPad Software, La Jolla, California USA).

Parasite development analysis

Mosquito midguts were dissected on day six, nine and 12 dpi in phosphate-buffered saline (PBS) (Thermo Fisher Scientific, Germany). Midguts were permeabilized in two percent Nonidet-P40 for 20 minutes, followed by 45 min in 0.1% mercurochrome/PBS staining (Sigma Aldrich, Germany) and washes. Midguts were imaged on a Zeiss Axio Lab.A1 microscope at 10 × magnification and oocysts were quantified using Fiji (v 2.9.0) ⁶⁸ using scaled images. ⁶⁸ For oocyst size measurements only fully visible oocysts were measured, excluding partially visible or burst oocysts.

In addition to oocyst quantification, GFP prevalence was scored on nine and 12 dpi to monitor early sporozoite development. For later-stage sporozoites mosquitoes were dissected at 18 and 20 dpi, and GFP positivity was recorded based on the presence of detectable fluorescence in entire midguts or salivary glands, whilst those showing no fluorescence above background were classified as GFP-negative. Fluorescence imaging was performed using a Nikon SMZ1500 microscope with a GFP filter set. Representative images were captured on a Zeiss Cell Observer Microscope on a × 10 objective in GFP and analysed using Fiji (v 2.9.0). ⁶⁸

All experiments were performed with three – four biological replicates. Graphs and statistical analysis were done via Graphpad prism 10 (GraphPad Software, La Jolla, California USA), unless otherwise noted. Oocyst size per midgut, which was not normally distributed, was analysed using a Mann-Whitney test. Infection and GFP prevalence were assessed using Fisher’s exact test. Oocyst number per midgut was analysed using a binomial generalised linear model (GLM) using RStudio (v4.5.0) ⁶⁹ (github, https://github.com/NataliePortwood/Portwood_et-al_2026).

In vitro ookinete motility assays were conducted as previously described, ²⁶ with minor modifications. Female Swiss mice were infected via intraperitoneal injection with a cryostock of Pb820 (parasitaemia 2-5%), a P. berghei line with fluorescent (RFP) ookinetes. ⁷⁰ A FBT would occur at five dpi via cardiac puncture from anesthetised mice (120 mg/kg ketamine, 16 mg/kg xylazine), parasitaemia was assessed, and 20 million parasites were injected intraperitoneally into a naïve mouse. To obtain high parasitaemia for in vitro cultures, reticulocytosis was induced in the recipient mouse by intraperitoneal administration of 200 μL phenylhydrazine (six mg/mL in PBS) two days prior to the FBT. When parasitaemia reached ~2% (typically 3 days post-transfer), sulfadiazine (30 g/L) was added to the drinking water to clear the asexual stages of the parasite. Two days later, the mouse was bled by cardiac puncture (120 mg/kg ketamine, 16 mg/kg xylazine), to obtain the largest amount of blood possible from the mouse (~500-900 μL) and immediately placed into ookinete medium (RPMI, supplemented with 20% FCS, 100 μM xanthurenic acid, and 50 μg/mL hypoxanthine, adjusted to a pH of 7.8-8.0). Parasites were incubated for 20-22 hours at 19°C and ookinetes were purified using a 63% Nycodenz/PBS density gradient (25 minutes at 200 g).

Purified ookinetes were resuspended in a small volume of medium containing either Dimethyl Sulfoxide (DMSO)-only (control) or tralopyril (38 μM) for two hours. Tralopyril concentrations were prepared by making 100 × stocks in DMSO and then two × stocks in ookinete media. At one- and two-hours post-exposure, ookinetes were vortexed to remove clumps and imaged on a Zeiss CellObserver microscope using a 25 × magnification in the DIC and RFP filter for 10 min with an image every 10 s.

Motility was scored, ookinetes were considered motile if they moved a distance greater than their body length. Experiments were performed with four - five biological replicates and statistical significance for motility was assessed using Fisher’s exact test. Of the motile ookinetes speed analysis was conducted using Fiji (v 2.9.0), ⁶⁸ with the manual tracking plugin. At one-hour post-exposure, data was normally distributed and analysed using an unpaired t-test with Welch’s correction. At two-hours, transformed data were non-normally distributed and analysed using a Mann-Whitney test. Graphs and statistical analysis were completed using GraphPad prism 10 (GraphPad Software, La Jolla, California USA).

Ultrastructure expansion microscopy (U-ExM) was performed following established methods ²⁹ with minor modifications. Midguts were dissected into 200 μL PBS in 1.5 mL Eppendorf tubes, and an equal volume of four percent paraformaldehyde (PFA) was added to achieve a final concentration of two percent PFA. Tissues were fixed for one hour at 37°C. Fixed midguts were transferred to a 24-well plate, rinsed three times with PBS and incubated in freshly prepared 2% formaldehyde/1.2% acrylamide in PBS. Plates were sealed and gently agitated (80 rpm) overnight at 37°C. Meanwhile, 12 mm coverslips were coated with 0.1 mg/mL poly-D-lysine overnight at 37°C.

For embedding, 35 μL of monomer solution (19% sodium acrylate, 10% acrylamide, 0.1% BIS in 10 × PBS) was supplemented with 5 μL 10% TEMED and 5 μL 10% APS, placed on Parafilm, and the tissue-bearing coverslip inverted onto the drop. After five min on ice, gels were polymerised for one hour at 37°C. Polymerised gels were then transferred to denaturation buffer (200 mM SDS, 200 mM NaCl, 50 mM Tris, pH 9), equilibrated at room temperature for 15 min, and denatured at 95°C for 90 min. Partially expanded gels were immersed in Milli-Q water which was exchanged once after 30 min, and allowed to expand overnight at room temperature.

Expanded midguts were excised from the gels as one cm ² pieces and washed twice for 15 min in PBS. Samples were blocked in five percent BSA/PBS for 30 m at room temperature, then incubated with human anti-CSP IgG (1:500) in 250 μL blocking buffer for 72 h at 4°C with gentle agitation. Following five 10 min washes in 0.5% Tween 20/PBS, tissues were incubated with goat anti-human IgG AlexaFluor 488 (1: 200) and Hoeschst 33342 (1: 100) in 250 μL blocking buffer for 72 h at 4°C. After five additional Tween washes, gels were incubated in 125 μL PBS containing NHS-ester 647 (1:200) for 1.5 h at room temperature, followed by a final series of five washes.

Gels were equilibrated for 30 min in water and allowed to expand overnight in 0.02% sodium azide/ddH ₂O. The next day, samples were mounted in eight-well Ibidi chambers pre-coated with poly-D-lysine (1 h, 37°C) and imaged using Leica Sp8 confocal microscope with a 40 × objective, with images processed in Fiji (v 2.9.0). ⁶⁸

Mosquito midguts were dissected on 12 dpi and fixed in one percent glutaraldehyde and four percent paraformaldehyde in 100 mM PHEM (60 mM PIPES, 25 mM HEPES, 10 mM EGTA and 2 mM MgSO ₄) buffer using microwave-assisted processing (BioWave Pro + PELCO, Fresno CA) with intermittent cooling on ice. After fixation, midguts were washed in PHEM buffer, post-fixed in one percent Osmium tetroxide, rinsed in distilled water, and contrasted with 1% aqueous uranyl acetate. Dehydration was carried out through a graded acetone series (30%, 50%, 70%, 90% and twice at 100%), followed by infiltration with increasing concentrations (25%, 50%, 75% in acetone) of Spurr’s resin (23.6% epoxycyclohexyl-methyl-3,4-epoxycyclohexylcarboxylate (ERL), 14.2% ERL- 4206 plasticizer, 61.3% nonenylsuccinic anhydride, 0.9% dimethylethanolamine). Samples were embedded in 100% resin and polymerised at 60°C for 48 h.

Sections of one micrometre thickness were prepared on a LEICA EM UC7 ultramicrotome and stained with one percent methylethene blue to identify oocysts in the cutting plane. Ultrathin sections (80-100 nm) were then collected and imaged on a transmission electron microscope (Technai F20, FEI/Thermo Fisher Scientific) operating at 200 kV, equipped with an Eagle 4k × 4k CCD camera (FEI). Images were acquired using the SerialEM software ⁷¹ and analysed using Fiji (v2.9.0). ⁶⁸ Mitochondrial morphology, cristae structure, and other parasite organelles were assessed qualitatively. Cristae circularity was quantified by measuring the ratio of the long axis to the short axis of individual cristae from all EM images (control images n = 37, treatment n = 25) these values were then plotted for comparative analysis, with higher values indicating more elongated structures. Kernel density plot and Kolmogorov test was done using R studio (v4.5.0) ⁶⁹ (with code available on github, https://github.com/NataliePortwood/Portwood_et-al_2026).

The experimental design was similar to that described in Figure 1C, except this study design incorporated two additional blood meals with a naïve mouse, on three and six dpi ( Figure 3A). Two-to-four-day-old An. stephensi mosquitoes were acclimated at 21°C for 24 h prior to exposure. 48 hours before the infectious feed, mosquitoes were exposed topically to 31 μM tralopyril or acetone-only (control). Fifty mosquitoes were used per group and put in cups. Unfed mosquitoes and dead mosquitoes were removed from the cups. The infectious blood meal was completed, as previously described. Additional blood meals were provided at three and six dpi, unfed mosquitoes were removed after each feed. Prevalence of GFP signal was recorded, as before, from midguts on 12 dpi and for midguts and salivary glands on day 20 dpi. Experiment was performed with four biological replicates. Pie charts were created using Graphpad Prism, and differences in prevalence were evaluated using Fisher’s exact test. Representative images were captured on a Zeiss Cell Observer Microscope on a 10× objective in GFP and analysed using Fiji (v 2.9.0). ⁶⁸

Mouse infections with P. berghei csGFP were completed as previously described, ²⁸ to infect two-to-five-day-old mosquitoes, for use on 17 to 21 dpi for in vitro sporozoite motility assays. Drug concentrations of tralopyril and FCCP were prepared as 100 × stocks in DMSO and diluted to two x stocks in RPMI for a final concentration of one percent DMSO. 10 concentrations of tralopyril were prepared between 0.003 μM and 64 μM. The control group was prepared in parallel, containing DMSO-only. In vitro motility assays were performed as per described previously ³⁵ with modifications. Sporozoites were extracted from An. stephensi 17 to 21 dpi, through salivary gland dissections in RPMI. A sporozoite density and motility check was done, where sporozoites were activated with six percent Bovine Serum Albumin/Roswell Park Memorial Institute 1640 medium (BSA/RPMI) concentration (1:1). Sporozoites were centrifuged at 200 × g for three minutes at room temperature before imaging. Sporozoites were imaged on a Zeiss CellObserver microscope (25× magnification, DIC and GFP filters) for three minutes at 20 frames per minute. If the density of sporozoites was too high, RPMI was used to dilute them to a density of around 150-200 sporozoites per field of view. For tralopyril exposure, 12% BSA was added in a 1:1:1 ratio with the sporozoites and the desired insecticide concentration, then samples were centrifuged and imaged at 0 min, 15min and 60min. Three biological replicates were performed per concentration. FCCP was tested the same way, where three concentrations (0.26 μM-26 μM) were completed, with two biological replicates. Z projections of each video were produced using Fiji (v 2.9.0), ⁶⁸ and the motility patterns were scored as: productive (circular) or non-productive motility. A productive motility ratio was calculated by dividing productive by non-productive motility, and normalised to the control of the replicate.

Sporozoite motility at each timepoint (0, 15, and 60 min) was analysed as a function of concentration. Simple linear regression was performed using the lm() function in RStudio (v4.5.0), ^{69 , 72} to evaluate the relationship between log ₁₀ transformed concentration and motility at each timepoint (github, https://github.com/NataliePortwood/Portwood_et-al_2026). To test for differences in motility across concentrations, a one-way analysis of variance (ANOVA) was performed at each timepoint using the aov() function in RStudio (v4.5.0). ⁶⁹ Significant effects of concentration were followed by pairwise comparisons with Tukey’s post hoc test. Motility at each timepoint was modelled using a nonlinear regression (github, https://github.com/NataliePortwood/Portwood_et-al_2026). The model parameters estimated were: L, the maximal motility; k, the slope of the concentration response; and x ₀, the inflection point corresponding to the concentration at which motility reached half of its maximum. Nonlinear regression was performed using nlsLM() from the minpack.lm package ⁷² and model fit was evaluated by the residual sum of squares. All statistical analyses and figure generation were conducted in RStudio (v4.5.0), ⁶⁹ using ggplot2. ⁷³

Salivary gland sporozoites were extracted from two-to-five-day-old mosquitoes 17 to 21 dpi as in the in vitro assays. Insecticide solutions were prepared from a 10 mg/mL stock, with 100 × stocks in DMSO and six × working stocks in RPMI, yielding a final concentration of 26 μM. A DMSO-only control (negative control) and a positive control (FCCP, 200 nM, Sigma) were prepared in parallel. LabTek chambers (ThermoFisher Scientific, Germany) were seeded with 3.8% BSA, Hoechst (33342), sporozoites, and treatment (tralopyril or control), and incubated for 30 min. Following incubation, supernatants were removed, samples washed with PBS, and three percent RPMI containing Mitotracker™ (200 nM) was added. After a 15 min incubation, samples were washed and fixed in four percent paraformaldehyde for 30 min. Images were acquired using a spinning disc confocal microscope (PerkinElmer) with a 63× objective, using lightning deconvolution. in the following channels: DIC, DAPI (350 nm), RFP (579 nm), and GFP (490 nm). Image processing was performed in Fiji (v 2.9.0). ⁶⁸ Images were exported as maximum intensity for visualisation.

Culturing of P. falciparum gametocytes

Asexual NF54 parasites were maintained at 0.5–3% parasitaemia in human O ⁺ erythrocytes at 2.5% hematocrit (Blood Bank, University Hospital Heidelberg, Germany) at 37°C. Cultures were grown in RPMI 1640 medium (Corning, Manassas, VA) supplemented with 25 mM HEPES, 10 mg/L hypoxanthine, 0.3% sodium bicarbonate, and 10% heat-inactivated human serum (Haema AG, Leipzig, Germany). Parasites were maintained under a gas mixture of 5% O ₂, 5% CO ₂, and N ₂ balance following established protocols. ^{74 , 75} Gametocyte differentiation was induced by diluting asexual cultures to 0.3% parasitaemia at five percent hematocrit and incubating for 16–18 days with daily medium replacement, without fresh erythrocyte supplementation, whilst avoiding disturbance of the erythrocyte layer, to generate mature stage V male and female gametocytes.

Infectious feeds with P. falciparum and topical exposure

Anopheles stephensi mosquitoes were reared as described above and maintained at insectary conditions throughout the experiment. For all experiments, two-to-four-day-old mosquitoes were used and placed in cups of 50 mosquitoes. The tralopyril group was topically pre-exposed 48 h (31 μM) prior to the infectious meal, while the chlorfenapyr group was topically pre-exposed 72 h (33 μM) prior to the infectious meal and the respective control group was exposed to acetone only, following the protocol described in Figure 4A. Mosquitoes were fed NF54 gametocyte cultures via prewarmed glass membrane feeders, adjusted to 0.5% stage V gametocytaemia with reconstituted blood. Approximately 24 h post-feed, dead, unfed or partially engorged mosquitoes were removed. Remaining mosquitoes were maintained on 10% sucrose solution. All treatment groups were analysed at 10 dpi, with only tralopyril-exposed and control mosquitoes processed for salivary glands at 18 dpi. For mosquito collection on timepoints,mosquitoes were anesthetized on ice, briefly immersed in 70% ethanol, snap-frozen at −80 °C for 10 min and transferred into phosphate-buffered saline (PBS) for dissection of midguts or salivary glands.

Parasite development analysis

Oocyst counts and size measurement were performed as described above for P. berghei, except timepoints for P. falciparum were day 10 dpi for midgut dissections (oocyst quantification) and day 18 dpi for salivary gland dissections (sporozoite quantification). Midguts were prepared the same as previously described for P. berghei and imaged on a Zeiss Axio Lab.A1 microscope at 10 × magnification. For oocyst size measurements, only fully visible oocysts were measured, and burst or partially visible oocysts were excluded. Analyses were performed using Fiji (v 2.9.0). ⁶⁸ Three-four biological replicates were completed and differences in oocyst counts were evaluated using a Mann-Whitney test (control vs tralopyril) and binomial GLM (control vs chlorfenapyr). Graphs and statistical analysis were done via Graphpad prism 10 (GraphPad Software, La Jolla, California USA), unless otherwise noted. Binomial GLM was done using RStudio (v4.5.0) ⁶⁹ (github, https://github.com/NataliePortwood/Portwood_et-al_2026). Salivary glands were dissected on day 18 dpi and pooled in groups of five in 1.5 mL tubes containing 10 μL PBS. Glands were homogenised with a pestle, and residual tissues were washed into a tube with an additional 10 μL PBS. Sporozoites were counted using a haemacytometer chamber (Neubauer, 0.1 mm) using a Zeiss Axio Lab.A1 light microscope. Six biological replicates were performed and differences in sporozoite number were evaluated using a Mann-Whitney test.