Antimalarials in mosquitoes overcome Anopheles and Plasmodium resistance to malaria control strategies

The spread of insecticide resistance in Anopheles mosquitoes and drug resistance in Plasmodium parasites is contributing to a global resurgence of malaria, making the generation of control tools that can overcome these issues an urgent public health priority. We recently showed that the transmission of Plasmodium falciparum parasites can be efficiently blocked when exposing Anopheles gambiae females to antimalarials deposited on a treated surface, with no negative consequences on mosquito fitness. Here, we demonstrate this approach can overcome the hurdles of insecticide resistance in mosquitoes and drug resistant in parasites. We show that the transmission-blocking efficacy of mosquito-targeted antimalarials is maintained when field-derived, insecticide resistant Anopheles are exposed to the potent cytochrome b inhibitor atovaquone, demonstrating that this drug escapes insecticide resistance mechanisms that could potentially interfere with its function. Moreover, this approach prevents transmission of field-derived, artemisinin resistant P. falciparum parasites (Kelch13 C580Y mutant), proving that this strategy could be used to prevent the spread of parasite mutations that induce resistance to front-line antimalarials. Atovaquone is also highly effective at limiting parasite development when ingested by mosquitoes in sugar solutions, including in ongoing infections. These data support the use of mosquito-targeted antimalarials as a promising tool to complement and extend the efficacy of current malaria control interventions. Significance Statement Effective control of malaria is hampered by resistance to vector-targeted insecticides and parasite-targeted drugs. This situation is exacerbated by a critical lack of chemical diversity in both interventions and, as such, new interventions are badly needed. Recent laboratory studies have shown that an alternative approach based on treating Anopheles mosquitoes directly with antimalarial compounds can render the vector incapable of transmitting the Plasmodium parasites that cause malaria. While promising, showing that mosquito-targeted antimalarials remain effective against wild parasites and mosquitoes, including drug- and insecticide-resistant populations, respectively, is crucial to the future viability of this approach. In this study, carried out in the US and Burkina Faso, we show that antimalarial exposure is highly effective, even against extremely resistant mosquitoes, and can block transmission of drug-resistant parasites. By combining lab, and field-based studies in this way we have demonstrated that this novel approach can be effective in areas where conventional control measures are no longer as effective.

mosquito-targeted antimalarials is maintained when field-derived, insecticide resistant Anopheles are 23 exposed to the potent cytochrome b inhibitor atovaquone, demonstrating that this drug escapes 24 insecticide resistance mechanisms that could potentially interfere with its function. Moreover, this 25 approach prevents transmission of field-derived, artemisinin resistant P. falciparum parasites (Kelch13 26 C580Y mutant), proving that this strategy could be used to prevent the spread of parasite mutations 27 that induce resistance to front-line antimalarials. Atovaquone is also highly effective at limiting 28 parasite development when ingested by mosquitoes in sugar solutions, including in ongoing infections. 29 These data support the use of mosquito-targeted antimalarials as a promising tool to complement and 30 extend the efficacy of current malaria control interventions.  Here we show that targeting P. falciparum with antimalarials during its development in the Anopheles 92 female circumvents the hurdles of insecticide and drug resistance, providing a critical addition to the 93 malaria elimination toolkit. Parasite development is substantially reduced when wild, as well as 94 recently lab adapted Anopheles coluzzii (a sibling species of An. gambiae) that are highly resistant to 95 pyrethroids are exposed to ATQ prior to feeding on blood taken from P. falciparum-infected donors 96 in Burkina Faso. ATQ is also fully active against field-derived P. falciparum parasites from Cambodia 97 that are resistant to artemisinin. When using distinct drug targets in humans and mosquitoes, this 98 method is therefore capable of both overcoming insecticide resistance mechanisms and stopping 99 transmission of parasite mutations that confer resistance to frontline antimalarials. Finally, we show 100 that delivering ATQ via sugar solutions causes a striking reduction in both parasite numbers and 101 growth, proving that antimalarials could also be incorporated into interventions that target outdoor 102 malaria transmission, such as ATSB. Targeting Plasmodium parasites in the mosquito vector is 103 therefore a promising strategy that circumvents key limitations of current malaria control and 104 preventative interventions.

106
Exposure to ATQ substantially reduces infection with field P. falciparum isolates in insecticide 107 resistant An. coluzzii 108 To determine whether antimalarial exposure can maintain efficacy in insecticide-resistant mosquitoes, 109 we took An. coluzzii collected as pupae from larval breeding sites in Bama, Burkina Faso and reared 110 them to adults at the IRSS, Burkina Faso. Adult mosquitoes were infected using P. falciparum 111 gametocyte positive blood obtained from a malaria infected human donor on the day of infection. The

112
An. coluzzii mosquitoes endemic to this part of Burkina Faso -hereafter named AcVK5 -are highly 113 resistant to pyrethroids (9,11,26). AcVK5 mosquitoes were exposed to ATQ for 6 minutes (min) at 114 two concentrations (100 µmol-or 1 mmol/m 2 ) or a mock-treated blank control surface prior to feeding 115 on infectious blood samples. Infection outcomes were assayed at 7 days (d) post infectious blood meal 116 (pIBM) by dissection of the mosquito midgut to determine the prevalence and intensity of parasite 117 oocysts ( Fig. 1 a). Control-exposed AcVK5 females were robustly infected with 81.3% harboring at 118 least one P. falciparum oocyst, and median infection intensity in infected females of 19 oocysts per 119 midgut. In contrast, AcVK5 mosquitoes exposed to either dose of ATQ had significantly reduced P. 120 falciparum infection both in terms of prevalence and intensity (Fig 1b) (27)), and cuticular thickening (26) Bama-R females were exposed to the 143 maximal effective concentration for tarsal ATQ (EC99, 100 μmol/m 2 , 6 min (25)) or to a vehicle control 144 immediately preceding infection, and parasite prevalence and intensity were determined at 7 d pIBM.

145
While control females were highly infected, with a median of 12 oocysts per infected midgut and 146 81.25% overall prevalence of infection, no oocysts were observed in females exposed to ATQ,147 suggesting that insecticide-resistance mechanisms found in highly resistant, natural Anopheles 148 populations do not interfere with the transmission blocking activity of ATQ (Fig. 1c).

149
Next, we established an in vitro P. falciparum culture from a polyclonal isolate (P5) collected from a 150 gametocytemic donor from Burkina Faso (28)) and infected the laboratory standard, insecticide 151 susceptible mosquito strain An. gambiae (G3). P5 development was 100% suppressed in females 152 treated with the EC99 of ATQ (Fig. 1d) such that zero oocysts were observed in midguts, compared to 153 heavy infections-both in terms of infection intensity (median 59.5 oocysts per midgut) and infection 154 prevalence (95.7%)-in controls. Delivery of ATQ to mosquitoes is therefore fully effective against 155 field-derived P. falciparum isolates currently circulating in West Africa.

156
ATQ prevents the transmission of an artemisinin-resistant P. falciparum isolate from the GMS 157 Given the results obtained with field-derived parasites from Africa, we next tested the ability of ATQ 158 to kill artemisinin resistant parasites from the GMS, where mutations conferring artemisinin resistance 159 occur in a high proportion of P. falciparum isolates, constituting a major public health threat. We 160 reasoned that these experiments would also allow us to test the concept of directly targeting drug 161 resistant P. falciparum during mosquito development, removing resistance mutations from the parasite 162 population, and thereby "rescuing" ACT efficacy in human treatment. To this end, we used a 163 Cambodian P. falciparum patient clone (KH001_029 (5), hereafter ART29) carrying the C580Y 164 mutation in PfK13 conferring resistance to artemisinin (Fig. 2a). We used the major Asian malaria 165 vector An. stephensi for these experiments as initial tests with An. gambiae did not produce appreciable 166 infections (SFig. 2). ART29 generated robust infections in control, mock-exposed An. stephensi, 167 (median 16 oocysts per midgut, 100% prevalence of infection). Conversely, no oocysts were detected 168 in females exposed to ATQ prior to infection (100 μmol/m 2 , 6 min) (Fig. 2b). These data show that 169 mosquito exposure to antimalarials, such as by incorporation in bed nets, indoor residual sprays (or In the field, mosquitoes that contact an antimalarial compound through mosquito-targeted 176 interventions may harbor parasites from a previous blood meal that have already traversed the midgut 177 lumen and formed oocysts. We therefore investigated the effects of ATQ on parasites in which oocyst 178 development is already underway, exposing G3 mosquitoes 6d pIBM (NF54, Fig. 3a). In contrast to 179 females exposed before infection, ATQ had no effect on the prevalence or intensity of infection, as 180 measured at 10d pIBM, suggesting ATQ acts differently on oocysts compared to zygote and ookinetes 181 (Fig. 3b). However, when we measured the size of the developing oocysts, we observed a significant, 182 45% decrease in the mean oocyst cross-sectional area (Fig. 3c). Oocyst size is a good proxy for rate of 183 growth (30) and as such, when we sampled mosquitoes at a later time point when sporozoite invasion 184 of salivary glands has already occurred (14 d pIBM), we observed a 33% reduction in the prevalence 185 of sporozoites in the salivary glands of ATQ-treated females (Fig. 3d). Similar results were obtained 186 when ATQ exposure instead occurred at 3 d pIBM (SFig. 3). Taken together, these results suggest that 187 ATQ exposure after oocyst formation has a partial cytostatic effect on P. falciparum. preceding infection (Fig. 4a). We observed a striking, 85% reduction in the prevalence of wild P. 196 falciparum infection in female mosquitoes that had access to ATQ-glucose prior to infection (Fig. 4b).
197 Importantly, median mosquito survival between control and ATQ treatment groups was not 198 significantly different (SFig. 4), confirming previous findings that atovaquone is not toxic to 199 mosquitoes at parasiticidal concentrations (25). This implies that other Plasmodium-specific inhibitors 200 would therefore not impose selective pressures leading to resistance mechanisms in the mosquito.

201
Using the same conditions with in vitro cultured P. falciparum (NF54) and lab-adapted An. gambiae 202 (G3) resulted in a remarkably similar infection outcome, with a 92.5% reduction in oocyst prevalence 203 in female An. gambiae given access to ATQ-glucose solution relative to controls (Fig. 4c). When ATQ 204 concentration was reduced to 10-and 100-fold ATQ dilutions, we observed progressively reduced, 205 dose dependent effects on prevalence (Fig. 4d).

207
As mosquitoes may often visit a sugar bait after acquiring an infectious blood meal, we also 208 investigated the impact of ATQ ingestion on ongoing P. falciparum (NF54) infections, providing 209 ATQ-treated sugar to G3 females from 2 d pIBM, when ookinetes have escaped the midgut lumen and 210 formed oocysts on the midgut basal lamina (Fig. 5a). This time, as based on our previous results ( Fig.   211 3) we expected a possible cytostatic effect on oocyst growth, we performed a sampling time course to 212 capture oocyst development through mid-to late sporogony (7d, 10d and 14 d pIBM). We also counted 213 salivary gland sporozoites, the end point of parasite development in the mosquito, at 14 d pIBM. In 214 agreement with our previous results, we observed no change in oocyst prevalence and intensity because 215 of ATQ ingestion (SFig. 5). However, we observed an 80.8% decrease in oocyst cross-sectional area 216 relative to controls at 7 d pIBM, which persisted at later time points (89% and 76.3% decreases at 10-217 and 14 d pIBM, respectively) ( Fig. 5b). By 14 d pIBM, ATQ-exposed oocysts had a similar size to 7 218 d pIBM control oocysts, suggesting a remarkable suppression of growth. Inspection of DAPI-stained 219 infected midguts revealed a stark decrease in the number of nuclear foci, with a single, diffuse DNA 220 signal compared to many condensed foci in oocysts in controls (Fig. 5c). Strikingly, we detected no

233
Tarsal ATQ exposure against wild An. coluzzii (AcVK5) collected as pupae from breeding sites was 234 able to strongly suppress the transmission of P. falciparum isolates circulating in children.

235
Interestingly, in these infections, carried out in Burkina Faso, we observed a small number of 236 "breakthrough" oocysts at ATQ doses that are non-permissive in tests using, respectively, lab-adapted 237 and susceptible NF54 parasites and G3 mosquitoes (25). We reasoned that this marginal reduction in 238 efficacy could be due to extant insecticide resistance circulating in wild Anopheles gambiae s.l 239 populations in Burkina Faso, including the wild AcVK5 mosquitoes used for these experiments. While can persist at ultra-low frequency in parasite populations through mitochondrial heteroplasmy (39).

265
Thus while the Burkinabe isolate (P5) used in our lab-based studies was susceptible to ATQ in vitro, 266 and Sanger sequencing of Cytochrome B from both P5 asexual blood stages and oocysts showed that 267 these parasites are wild-type, we cannot rule out the possibility that either cryptic parasite factors, lost falciparum is therefore also vulnerable to inhibition through tarsal exposure during the oocyst stage, 312 which is by far the longest developmental stage in the mosquito, taking between 7-10 days depending 313 on the frequency of blood feeding (30), and thus is the parasite life stage most likely to encounter a 314 mosquito-targeted intervention. As such, the ability to stall or kill oocysts and sporozoites is a highly 315 desirable quality for mosquito-targeted antimalarials.  Although at an early stage, mosquito-targeted antimalarials have the potential to be an effective 328 element to drive malaria incidence down. To this end, identifying more compounds with strong 329 antiparasitic activity during the mosquito stages of P. falciparum development -and in particular 330 compounds with sporogony-specific activity -will be a crucial next step, and one that should leverage 331 the extensive libraries of known antimalarials. Indeed, one of the key strengths of this approach is the 332 potential to exploit and repurpose compounds that are otherwise unsuitable for human therapeutic use,  Except for selection of resistance in Bama-R, all mosquito colonies were maintained identically at 26 353 ⁰C ± 2 ⁰C and 80% ± 10%. relative humidity (RH). Larvae were cultured in 2-liter (l) catering pans in 354 500 ml distilled water (dH2O) under an optimized density and feeding regimen. At the onset of 355 pupation, pupae were separated from larvae using a vacuum aspirator, collected in dH2O, and placed 356 in a 30x30x30 cm cage (Bugdorm, Megaview Science Co, Ltd, Thailand). After emergence, adult 357 mosquitoes had access to separate sources of 10% glucose (Sigma Aldrich, US) and dH2O ad libitum. 358 For colony maintenance, 5-7-day-old adults were provided a blood meal of donated human blood 359 using an artificial membrane feeding system (Hemotek, UK). For mosquito colony 2) females were 360 maintained on rabbit blood by direct feeding (protocol approved by the national committee of Burkina 361 Faso; IRB registration #00004738 and FWA 00007038) and adult males and females fed with a 5% 362 glucose solution. Larvae were reared at a density of about 300 first-instar larvae in 700 ml of water in 363 plastic trays and fed with Tetramin Baby Fish Food (Tetrawerke, Melle, Germany). For infections with donor isolated gametocytes ( Fig. 1(b), Fig. 4(b)), prevalence and intensity of 431 infection were analyzed using more complex statistics to account for between-replicate effects of 432 different human gametocyte donors. For prevalence: we constructed a General Linear Model as 433 follows, independent variable/y "Infected?" (Two-level, categorical (yes/no)), with cofactors 434 "Treatment" (Two-level, categorical (Control/ATQ)) and "Gametocyte Donor" (Two-level, 435 categorical (Donor 1/Donor 2)), we also included the interaction term Treatment*Gametocyte Donor 436 to detect higher level effects. As the output was categorical, the GLM model was run with a binomial 437 distribution and logit link-function. To achieve the best model fit, we iteratively removed cofactors 438 from the model, and selected the model output with the lowest corrected Akaike information criterion 439 (AICc). In all cases, the best model fit included both cofactors, but excluded the interaction term. For The authors state that they have no conflicting interests.     ATQ-glucose from 2 d pIBM caused a significant reduction in mean oocyst cross-sectional area 534 ("size"/m 2 ) relative to control at 7 d pIBM (2-way T-test, n=69, df=67, t=13.63, p<0.0001), 10 d pIBM