Uridine diphosphate (UDP)-glycosyltransferases (UGTs) are associated with insecticide resistance in the major malaria vectors Anopheles gambiae s.l. and Anopheles funestus

Malaria remains one of the highest causes of morbidity and mortality, with 249 million cases and over 608,000 deaths in 2022. Insecticides, which target the Anopheles mosquito vector, are the primary method to control malaria. The widespread nature of resistance to the most important insecticide class, the pyrethroids, threatens the control of this disease. To reverse the stall in malaria control there is urgent need for new vector control tools, which necessitates understanding the molecular basis of pyrethroid resistance. In this study we utilised multi-omics data to identify uridine-diphosphate (UDP)-glycosyltransferases (UGTs) potentially involved in resistance across multiple Anopheles species. Phylogenetic analysis identifies sequence similarities between Anopheline UGTs and those involved in agricultural pesticide resistance to pyrethroids, pyrroles and spinosyns. Expression of five UGTs was characterised in An. gambiae and An. coluzzii to determine constitutive over-expression, induction, and tissue specificity. Furthermore, a UGT inhibitor, sulfinpyrazone, restored susceptibility to pyrethroids and DDT in An. gambiae, An. coluzzii, An. arabiensis and An. funestus, the major African malaria vectors. Taken together, this study provides clear association of UGTs with pyrethroid resistance as well as highlighting the potential use of sulfinpyrazone as a novel synergist for vector control.

pyrethroids and organochlorines.In addition, the Ace-1 mutation (G119S) confers resistance to both carbamate and organophosphate classes 9 .Increased detoxification of insecticides through the upregulation of protein families that metabolise these compounds or aid their clearance are reported ubiquitously across malaria endemic countries [10][11][12] .The detoxification system is composed of three 'phases': phase I includes the direct oxidation, reduction or hydrolysis of compounds; phase II is the conjugation of a moiety; and phase III is excretion.The most important and well-studied detoxifiers for insecticide resistance are cytochrome P450 monooxygenases (CYP450s) which directly detoxify multiple insecticides through hydroxylation resulting in a product that is less hydrophobic 13,14 .Although these enzymes are the best studied, other phase I enzymes, such as carboxylesterases 15 , phase II enzymes including Glutathione-S-Transferases (GSTs) 16 and phase III transporters such as ABC-transporters 17 have been linked to IR.
In the last decade the increased availability of transcriptomic and whole genome sequence data is progressively allowing identification of novel transcripts and genomic regions driving IR in Anopheles spp.These studies have repeatedly identified the overexpression of uridine diphosphate (UDP)-glycosyltransferases (UGTs) in IR populations across Africa including: An. coluzzii in Nigeria, Niger, Chad, and Burkina Faso 18,19 ; An. gambiae in Burkina Faso 20 ; An. arabiensis in Tanzania (UGT308D1) 21 ; and Anopheles funestus in Malawi, Cameroon, Uganda and Kenya 22,23 .In addition to constitutive overexpression, UGTs have been shown to be induced following permethrin exposure in Kenyan populations 24 , following deltamethrin exposure in An. coluzzii 19 and are associated with bendiocarb and DDT-resistance in Cameroon 25,26 .Furthermore, there is evidence of selective sweeps in genomic regions containing UGTs which may confer an increase in transcript expression 12,27 .Additionally, UGT overexpression has been repeatedly reported in other insecticide resistant mosquito species, such as Aedes spp.9][30][31][32][33] indicating that their role may be important across multiple vector species.Interestingly, UGTs are known phase II detoxifying enzymes, and act through increasing excretion of xenobiotic compounds by conjugating a UDP-donated glucose molecule which makes the product more hydrophilic 34 .
Unlike in mosquito control, UGTs involved in agricultural pesticide resistance have been extensively validated for their role in the detoxification of a plethora of compounds, including several utilised in malaria control.Distinct families of UGTs have been characterised in resistance towards spinosad 35 , which is recommended by the World Health Organisation (WHO) for larvicidal control of Anopheles spp.Furthermore, neonicotinoids, namely clothianidin and imidacloprid, both used in IRS formulations, have been shown to be detoxified by numerous UGT families in multiple species [36][37][38][39] .Strikingly, many of these UGT families confer cross-resistance between imidacloprid and pyrethroids 40,41 .Importantly for vector control, UGT-mediated resistance to chlorfenapyr, a pyrrole now widely used in ITNs 35,[42][43][44] , and pyrethroids have been demonstrated in several agricultural pest species, with UGT families linked to lambda-cyhalothrin, bifenthrin, alpha-cypermethrin and deltamethrin resistance 40,[45][46][47][48][49][50] .Additionally, there is evidence of UGTs aiding resistance towards dichlorodiphenyltrichloroethane (DDT) 45 .
With the wealth of evidence of UGT-mediated resistance to insecticides across both mosquito species and agricultural pests, this study adopts multiple methods to investigate the role of UGTs in insecticide resistance in the four major African malaria vectors: the An.gambiae complex and An.funestus.Here we show that UGT308G1, 306A2 and 302A1 are over-expressed in multi-resistant An. gambiae and An.coluzzii.We demonstrate tissue specificity of these UGTs and explore their expression post-pyrethroid exposure.Finally, we use a UGT inhibitor, sulfinpyrazone (SULF), to demonstrate that UGTs confer resistance to different pyrethroids and DDT in An. gambiae, An. coluzzii, An. arabiensis and An.funestus.Taken together, we demonstrate that UGTs play a key role in insecticide resistance in four major African malaria vectors and could be used as a target for vector control.

Phylogenetic analysis of An. gambiae UGTs
UGTs are a conserved protein family found ubiquitously across arthropods; however, the number of UGTs is highly variable with order-specific gene diversification and cross-species conservation 51,52 .For example, Aedes albopictus has 46 UGTs and Drosophila melanogaster has 35 UGTs, whilst just 25 UGTs are annotated in the An.gambiae genome and 20 in An. sinensis.To explore the evolutionary relationship of these, a phylogenetic tree encompassing multiple mosquito species, D. melanogaster and M. domestica was constructed (Supplementary Fig. S1).Two large clades of M. domestica have expanded alongside the UGT304, 303 and 35 families in D. melanogaster relative to the mosquito species, indicating that this expansion likely occurred after the last common ancestor (LCA) of mosquitoes, D. melanogaster and M. domestica (~ 200MYA) 53 and the LCA of D. melanogaster and M. domestica (~ 100MYA) 54 .There is also a substantial expansion of UGTs in mosquito species compared to M. domestica and D. melanogaster, including the UGT308, 309, and 310 families in An. gambiae; this expansion could indicate a derived role for these UGT families in mosquitoes.
Next, induction of UGTs post-pyrethroid exposure was explored.UGT expression in resistant An. coluzzii following 1-h exposure to 0.05% deltamethrin WHO tubes 55 showed the induction of UGT302H2 transcripts immediately after exposure (1.5-fold, p < 0.05), as well as 12 h (1.8-fold, p < 0.01) and 24 h (1.8-fold, p < 0.01) post-exposure (Fig. 3b).UGT306A2 induction is delayed in comparison with an increase in expression observed at 4 h (2.2-fold, p < 0.05), 12 h (twofold, p < 0.05) and 24 h (1.7-fold, p < 0.01) post-exposure.Interestingly, this population showed no induction of UGT308G1.Due to the influence of the mosquito's circadian rhythm, the expression patterns of the UGTs were investigated to identify their 12-h cycle.In parallel with the exposed mosquitoes, unexposed mRNA was isolated immediately after the deltamethrin exposure and 12 h-post exposure (Supplementary Fig. S2).Comparisons of these cohorts displayed a significant difference in expression for UGT302A1, UGT302H2 and UGT308G1 inferring that any increase in expression at 12 h post-exposure is potentially obscured by the natural decrease in expression due to circadian rhythms.
Finally, tissue specific expression of the UGTs of interest was explored.The highest expression of UGTs was observed in the legs and the antennae (Fig. 3c) which are tissues that most commonly interact with insecticidetreated surfaces.In the legs, UGT306A2 (19.1-fold, p < 0.0001) and 302A1 (3.4-fold, p < 0.001) expression levels were significantly escalated, and in the antennae UGTs 302A1 (45.5-fold, p < 0.0001), 302H2 (17.3-fold, p < 0.0001) and 306A1 (6.7-fold, p < 0.01) showed the most dramatic increases displaying a potential role of UGTs in olfaction.An important tissue for insecticide resistance is likely the Malpighian tubules 56 , and here, UGT302A1 transcript levels are significantly elevated (3.5-fold, p < 0.001).Furthermore, UGT302H2 (3.4-fold, p < 0.01) is elevated in the abdomen integument where metabolic homeostasis is maintained by the fat body 57 .The head of the mosquito, consisting of the proboscis, palps, and brain, displayed a significant decrease in UGT302H2 (0.0065-fold, p < 0.0001).

RNAi of UGTS of interest did not restore susceptibility to pyrethroids
Validation of the functional role of the three most overexpressed UGTs-302A1, 306A2 and 308G1-was investigated using RNAi to determine whether targeted knockdown of these transcripts can restore susceptibility to deltamethrin in resistant An. coluzzii (Fig. 4).The efficiency of the dsRNA was determined using qPCR; transcript levels were successfully reduced by 87% (302A1, p < 0.001), 88.5% (306A2, p < 0.001), and 80.7% (308G1, p < 0.001) (Fig. 4a).Despite the silencing of UGT transcripts there was no impact on the sensitivity of this population to deltamethrin (Fig. 4c).Mortality induced by deltamethrin exposure was 7.7% in the dsGFP non-target control, 6.9% in the ds302A1 group, 7.7% in the ds306A2 group, and 6% in the ds308G1 group (Fig. 4c).As the UGT family is large it is possible that targeting individual UGTs is having little effect due to functional  www.nature.com/scientificreports/redundancy; therefore, dsRNA for all three candidates (ds306A2 + 302A1 + 308G1) was pooled and injected.Despite the combined dsRNA efficiently silencing each of the targeted UGTs (Fig. 4b)-302A1 (reduced 89.8%, p < 0.001), 306A2 (reduced 89.4%, p < 0.001), and 308G1 (reduced 98.5%, p < 0.0001)-there was no increase in the sensitivity of these resistant mosquitoes to deltamethrin at only 6.4% mortality (Fig. 4c; Supplementary Table S2).

Sensitivity of resistant Anophelines to insecticides due to UGT inhibition
As targeting individual UGTs had no effect, the topical application of the UGT inhibitor sulfinpyrazone (SULF) 49,58 was used to inhibit the activity of all UGTs simultaneously in four IR African species prior to insecticide exposure: An. gambiae, An. coluzzii, An. arabiensis and An.funestus.These populations all have high intensity resistance against the tested insecticides (pyrethroids: alpha-cypermethrin, deltamethrin, bifenthrin and permethrin and the organochloride DDT with 3.9-40% mortality), with the exception of An. funestus which is completely susceptible to DDT (Fig. 5; Supplementary Table S3).As a preliminary method of testing the impact of the inhibitor, increasing doses of SULF was topically applied [59][60][61] with the resistant populations with and without WHO-tube 0.05% deltamethrin exposure to determine a concentration whereby there was minimal intrinsic mortality but clear increase in pyrethroid-induced mortality; thus, a concentration of 1% SULF was used with other insecticides (Supplementary Fig. S3, Table S3).At the 1% concentration used, there is varying levels of intrinsic mortality in the different species when SULF is applied alone (Fig. 5; Fig. S4): An. gambiae-18.1%,An. coluzzii-7.8%,An. arabiensis-20.9%, and An.funestus-31.3%.
Next, the resistant mosquito populations were treated with a topical application of 1% SULF 59 followed by WHO-tube assays for the previously mentioned insecticides.Inhibiting UGTs in An. gambiae produces significant increases in mortality to alpha-cypermethrin (49.8% increase, p < 0.05), bifenthrin (16% increase, p < 0.01), permethrin (38.6% increase, p < 0.01), and DDT (28.1% increase, p < 0.05) (Fig. 5a).The increases observed for bifenthrin and DDT, however, are likely due to the additive mortality of sulfinpyrazone instead of the inhibition of UGTs, as demonstrated by the synergy ratio of close to or equal to one: bifenthrin-1.01;DDT-0.9 (Supplementary Table S5).This ratio computes the additive or synergistic impact of SULF by dividing the combined insecticide and SULF survival with a calculated survival rate, a ratio of less than one indicates a synergistic affect, equal to one is additive and more than one is antagonistic (Supplementary Table S5) 62 .An. coluzzii demonstrates almost complete restoration of deltamethrin susceptibility (71.8% increase, p < 0.01) and partial restoration of mortality towards permethrin (29.9% increase, p < 0.05), although the latter increase is likely due to the additive lethality of sulfinpyrazone (synergy ratio-0.81).There is a trend of increasing mortality with bifenthrin; however, this is not statistically significant and thus likely due to the intrinsic activity of sulfinpyrazone (synergy ratio-0.85)(Fig. 5b, Supplementary Table S5).The An. arabiensis population tested here demonstrates increasing mortality for alpha-cypermethrin (42.2% increase, p < 0.001), deltamethrin (50.2% increase, p < 0.01), and DDT (18.9% increase, p < 0.05) (Fig. 5c).As with An. gambiae it seems that the increase in mortality when exposed DDT is the additive effect of the insecticide with sulfinpyrazone (synergy ratio-1.02)(Supplementary Table S5).There is a trend of increasing permethrin susceptibility (21.2% increase; synergy ratio-0.91);however, this is not statistically significant (Fig. 5c, Supplementary Table S5).Mortality to alpha-cypermethrin (synergy ratio-0.41)and deltamethrin (synergy ratio-0.65) is again restored when inhibiting UGTs, with increases of 58% (p < 0.001) and 49% (p < 0.01) respectively in An. funestus (Fig. 5d).A smaller, but significant increase in mortality is also demonstrated with permethrin mortality (28.7%, p < 0.05; synergy ratio-0.14)after the addition of sulfinpyrazone.There is some restoration of bifenthrin susceptibility (25.6% increase; synergy ratio-0.58);however, due to the variability between replicates this is not statistically significant.
To ensure that SULF was not inhibiting the known pyrethroid metabolisers of the CYP450s, competitive binding assays were carried out with CYP6P3, CYP9K1, CYP6M2 and CYP6P4 63 .The lack of affinity of CYP450s for sulfinpyrazone ensures that the observed mortality was not induced by the inhibition of these major pyrethroid metabolisers (Supplementary Fig. S5; Table S4); however, we cannot rule out other off target effects.

Discussion
Resistance to insecticides used in malaria control is complex, with continued reports of novel mechanisms which overcome insecticide-induced mortality 8,11,12,64,65 .Understanding such mechanisms is necessary to protect the current control tools and to innovate new chemicals and methods for implementing mosquito control.Here, we show that multi-resistant Anopheles species display elevated expression of multiple UGTs over time and from widespread geographical areas.We then characterise the relationship of UGTs linked to insecticide resistance in Anopheles with those from agricultural pests with proven roles in resistance.We further establish that key UGTs identified in transcriptomics data have increased expression of these transcripts in highly resistant lab-reared mosquitoes.Expression analysis indicates that these transcripts are enriched in tissues important for insecticide contact such as the legs and head and are induced after pyrethroid exposure.Finally, and crucially, we show that inhibition of these enzymes restores pyrethroid and DDT susceptibility in all major African vector species: An. gambiae, An. coluzzii, An. arabiensis and An.funestus.
Mining of transcriptomic datasets collected over disparate time periods demonstrates consistent overexpression of UGTs in IR populations representing all the major African malaria vectors 12,19 .UGTs found overexpressed across multiple subfamilies in An. gambiae cluster with UGTs from agricultural pests linked to resistance to compounds used in malaria vector control 35,47,48 .Interestingly, the UGT308 (UGT308G1 studied here), 309 and 310 family, which show expansion in mosquito species relative to D. melanogaster and M. domestica, cluster with A. gossypii AgosUGT344B4 which has been previously linked with neonicotinoid resistance 38 and B. dorsalis BdUGT50B5 linked to neonicotinoid and pyrethroid cross-resistance 48 .Furthermore, UGT302H2, UGT306D1 and UGT306A2, three other UGTs identified in this study, cluster with D. melanogaster DmelUGT35B1 and BdUGT35F2, both of which have been identified in resistance to DDT, pyrethroids, and neonicotinoids, respectively 48,66 .As silencing of these individual genes did not restore susceptibility to pyrethroid insecticides whilst inhibition with SULF does, it would be of further interest to explore other UGTs clustering with resistancerelated UGTs of other pest species.For example, AgosUGT344B4 and BdUGT50B2 provide neonicotinoid and pyrethroid resistance 38,48 and cluster closely with An. gambiae UGT50B5.Moreover, there is an abundance of data to support the role of UGTs in pyrethroid resistance in Chrysodeixis includens, Spodoptera exigua, Tetranychus urticae, A. gossypii, Sp. litura, and Sp.littoralis 41,[45][46][47]49,50 which were not explored here due to lack of availability of the sequences.
Here, UGT306A2 was shown to be overexpressed in three resistant populations compared to the control and further induced upon pyrethroid exposure; this UGT has previously been shown to be up-regulated across all populations in Côte D'Ivoire and Burkina Faso 67 and across temporally and geographically disparate An. funestus and An.gambiae s.l. 18,26,68,69.The UGT302 family is associated with pyrethroid resistance in B. dorsalis as well as being enriched in the antennae and maxillary palp in this species, suggesting a role in odour transduction and detoxification 48 .The UGT302 family has also been identified as over expressed in pyrethroid-resistant An. sinensis 70 , and in a recently published study on An. funestus 22 .The UGT308 family was also differentially expressed in pyrethroid resistant An. funestus compared to insecticide susceptible mosquitoes 23 .UGT306A2 shows overexpression in the Tiefora population as expected 71 , as well as 308G1 and 302A1, but UGT302H2, observed as one of the mostly highly upregulated genes in the study by Williams et al. 71 , is not significantly overexpressed at a basal level in this study.The latter, however, is up-regulated post-deltamethrin exposure alongside UGT306A2 and conversely 308G1 is down-regulated.Surprisingly, UGT308G1 was previously shown to have sustained www.nature.com/scientificreports/overexpression following deltamethrin exposure in An. coluzzii 19 , indicating population-specific differences in response to insecticide challenge.The UGTs explored here show mixed tissue localisations; however, UGT302A1, UGT306A2 and UGT306A2 are all enriched in the antennae and legs.The localisation to these tissues is also seen in Sp. littoralis and B. dorsalis displaying resistance to various insecticides including deltamethrin, lambda-cyhalothrin, and imidacloprid 48,50 .These studies outlined a dual role of UGTs in odorant metabolism and insecticide resistance hinting that they may play a similar role in Anopheles.Intriguingly in situ hybridisation has highlighted UGT expression at the site of olfactory neurons in B. dorsalis antennal sensilla 50 , as the nervous system is the target of multiple malaria control insecticides 72 .As these tissues are likely the most important for insecticide uptake and have recently been shown to be the site of potential sequestration 64 , it may be that UGT detoxification is important in these tissues for insecticide metabolism.
In addition to leg and antennal expression, UGT302H2 and UGT302A1 are enriched in the abdomen integument and the Malpighian Tubule, respectively.Expression of UGTs in the Malpighian tubules and the abdomen, where the fat body is located, further points to the role of UGTs in insecticide resistance as these tissues help mediate the impact of xenobiotics throughout the insect body 56,57,73 .CYP450s and GSTs, known Anopheline insecticide detoxifiers, as well as UGTs, are overexpressed in Malpighian tubules of resistant D. melanogaster, and silencing CYP6G1, which is specific to this tissue, improves insecticide sensitivity [74][75][76][77] .A study in Sp. exigua found UGTs and CYP450s co-localise in the fat body of these resistant insects, and when treating fat body cells in vitro with a range of insecticides, including pyrethroids, these same enzyme families were then induced as a response 46 .These co-localisations provide some evidence of a sequential detoxification pathway of insecticides with UGTs (Phase II) further detoxifying the metabolites of CYP450s (Phase I) 78,79 .Interestingly, UGT308G1 and CYP6M2 follow the same induction patterns observed over 24 h post-deltamethrin exposure in An. coluzzii 19 which could also indicate a link between these enzyme groups.
To assess the role of UGTs in resistance on a phenotypic level two methods were adopted: (i) RNAi was used to silence the UGTs, and (ii) sulfinpyrazone was used to inhibit them 39,43,49,58,[80][81][82] .Characterising UGTs in vitro was not possible by silencing individual UGTs despite the dsRNA efficiently reducing the UGT transcripts to lower levels observed in resistant agricultural pests (Myzus persicae, Diaphorina citri, Plutella xylostella, A. gossypii) 39,58,80,82,83 .The lack of phenotype observed here could be due to three primary reasons.Firstly, there could be functional redundancy within the family, as observed with CYP450s upon RNAi.Secondly, the metabolites that are processed by the UGT enzymes may not be directly toxic; however, this is refuted by the SULF inhibition and finally, the UGTs targeted in this paper may not be directly involved in pyrethroid metabolism.
Despite a lack of phenotype when silencing the UGTs identified in this paper, the chemical inhibition of the enzyme superfamily restored susceptibility to pyrethroids and DDT in the An.gambiae s.l. and An.funestus strains tested, painting a potential role for UGTs in resistance to these insecticides.The consistent phenotype seen in both An.gambie s.l. and An.funestus, separated by 85 MY of evolution 12 , indicates that these enzymes are involved in insecticide metabolism.Although UGTs are unlikely to directly metabolise pyrethroids, the products of CYP450 metabolism (phase I) are reported to cause mortality in An. funestus 78 and thus inhibiting this pathway with sulfinpyrazone could lead to a build-up of lethal insecticide metabolites.In addition to increased mortality after insecticide exposure, SULF resulted in varying levels of intrinsic mortality when applied alone to the different Anopheline species, and this could be due to the disruption of critical pathways such as tissue homeostasis as seen in mammals, fungi, and plants 84,85 .Furthermore, UGTs are involved in the metabolism of many endogenous compounds in insects that aid mechanisms such as steroid regulation, UV shielding, and cuticle formation 51 , and again, the interruption of these pathways could be having fatal consequences.The ability of sulfinpyrazone to produce these lethal affects, as well as providing synergistic results when alongside currently used insecticides, is integral evidence of how this compound could be used as a new vector control tool.
Although SULF restored susceptibility to pyrethroids and DDT, it did so differentially across species and across pyrethroid chemistries.The varying profiles of restored insecticide susceptibility with this inhibitor demonstrates population-specific and insecticide-specific mechanisms of resistance, as well as highlighting how mosquitoes have evolved multiple resistance mechanisms 6 .In line with results observed here, a pattern of differential response to varying pyrethroids across populations is also observed in field populations with the CYP450 inhibitor, piperonyl butoxide (PBO).Indeed, populations of An. gambiae s.l.from different countries, and even different ecological zones within a country, respond contrastingly to deltamethrin and permethrin, with PBO exposure restoring varying levels of partial susceptibility [86][87][88][89][90] .Analogous to SULF, PBO specifically binds and inhibits CYP450s and thus restores mortality in mosquitoes through blocking phase I detoxification 91 and is currently incorporated on pyrethroid-PBO bed nets being used in Africa 1,4 .Taken together, this suggests that CYP450-mediated resistance, as with UGT-associated resistance, is population-and insecticide-dependant.Despite these differing profiles, PBO bed nets are efficiently reducing malaria burden through increased mortality in mosquito vectors, indicating that SULF could be an additional tool for vector control.Given the reduced efficacy of PBO in areas without CYP450 resistance 86,88,92 , it could be interesting to investigate the additive effects of PBO and sulfinpyrazone together, especially given the intrinsic mortality caused by sulfinpyrazone.
Inhibition of UGTs restores susceptibility to both pyrethroids and DDT, thus demonstrating that UGTmediated resistance to these compounds is an overlooked mechanism in current studies.Future investigations to directly link enzymatic activity with pyrethroids or pyrethroid metabolites in Anopheles mosquitoes would be key in understanding the exact molecular basis of this resistance mechanism.The overexpression of UGTs in olfactory and detoxification tissues observed in this study hints at a potential role in olfaction and a synergy with cytochrome P450s, though further studies are needed to confirm this.Taken together, this study shows a concrete link between UGT enzymes and pyrethroid resistance, highlighting a new avenue for malaria control.

Mosquito husbandry and strains
Mosquitoes were reared at Heidelberg University in standard insectary conditions; 27 °C, 70-80% relative humidity, 12 h light cycle with 1 h dawn:dusk.Larvae are fed on ground fish food (Tetramin, Germany) and the adults on 10% sucrose solution.The mosquito colonies used in the experiments were the insecticide susceptible Kisumu from Kenya (An.gambiae s.s), and insecticide resistant populations: Tiassalé from Côte D'Ivoire (An.gambiae s.l), Tiefora and Banfora from Burkina Faso (An.coluzzii), Gaoua from Burkina Faso (An.arabiensis), and FuMOZ from Mozambique (An.funestus) 71,93 .Resistant colonies are selected to maintain resistance every fourth generation on 0.05% deltamethrin and 0.75% permethrin.All mosquitoes used for testing were presumed mated.

Identification of candidate UGTs
Published transcriptomics data was analysed to investigate UGT expression throughout Africa and identify candidates for characterisation and differentially expressed transcripts were extracted from AnoExpress (https://t.ly/ bXKfF) for the Anopheles gambiae species complex; this included both RNAseq and microarray data 11,12 .R (ggplot2) was then used to create a map displaying the fractions of the 25 UGTs quantifiable across all datasets and those showing differential expression.Five UGTs were regularly overexpressed throughout this data or shown to be induced across multiple time points post-insecticide exposure 97 and were selected for characterisation-UGT302A1 (Gene ID: AGAP006222), UGT302H2 (AGAP007029), UGT306A2 (AGAP007589), UGT306D1 (AGAP011564), UGT308G1 (AGAP007990).

RNA extraction and cDNA synthesis
For whole body RNA extractions, 3-5-day old female mosquitoes were collected in triplicate containing seven adults each.RNA extraction from specific tissues was also completed in triplicate, 10-100 individual tissues were dissected per sample from 3 to 5-day old females.Dissections of the following tissues were completed with tweezers and pins in iced PBS or extraction buffer: legs, ovaries, abdomen integument including fat body, malpighian tubules, midguts, antennae, thorax, head, and reproductive organs.All samples were homogenised in 100 μl Extraction Buffer from the PicoPure™ RNA Isolation Kit, the manufacturer's protocol was then followed (ThermoFisher Scientific, Germany).Subsequently, cDNA was synthesised per RNA sample following the SuperScript™ III First-Strand Synthesis System protocol using Oligo(dT) 20 primers to select for messenger RNA (mRNA) (ThermoFisher Scientific, Germany).cDNA was purified using the QIAquick PCR purification kit following the manufacturer protocol (QIAGEN, Germany).Quality and quantity of RNA and cDNA was measured using a NanoDrop One spectrophotometer (Thermo Scientific, Germany).

Silencing of UGTs with RNA interference
Primers were designed for targets using Primer Blast (NCBI) for a 300-600 bp product length and 20-50% GC, T7 promoter sequences were added to 5ʹ end of forward and reverse primers (Supplementary Table S6).PCR products were amplified under the following conditions: 98 °C for 30 s, then 35 cycles of 98 °C for 7 s, 55 °C for 10 s and 72 °C for 60 s, and a final extension for 5 min at 72 °C.Amplicons were either PCR/gel purified using the QIAquick Gel Extraction Kit and QIAquick PCR & Gel Cleanup Kit following the manufacturer's manual (QIAGEN, Germany).Double-stranded RNA (dsRNA) was synthesised using MEGAscript™ T7 Transcription www.nature.com/scientificreports/Kit (Thermo Fisher Scientific, Germany) following the user manual with a 16 h incubation at 37 °C.Samples were purified using the MEGAclear™ Transcription Clean-Up Kit (Thermo Fisher Scientific, Germany) with a twice-heated elution step at 65 °C for 10 min in 100 μl final elution volume.The dsRNA quality and quantity were measured on a NanoDrop One spectrophotometer (Thermo Scientific, Germany) and concentrated to 3 μg/ μl in a vacuum centrifuge.3-5-day old female mosquitoes were anaesthetised on CO 2 and 69 nl of dsRNA was injected into the thorax between cuticle plates for RNA interference (RNAi).In parallel, dsGFP was injected as a non-target control at the same concentration and volume.72 h post-injection, the dsRNA was tested for efficiency using qPCR with transcript expression compared to the dsGFP control.Subsequent injections were followed by exposure to insecticides after 72 h.Insecticide tube assays were performed using standard WHO protocol 98 with deltamethrin (0.05%) for 1 h, 24 h mortality was scored.

Cytochrome P450 inhibition assays
The affinity of key cytochrome P450s-CYP6P3, 6P4, 6M2 and 9K1-to sulfinpyrazone was determined by inhibiting the metabolism of the fluorescent diethoxyfluorescein (DEF) substrate.Sulfinpyrazone (1000, 200, 40, 8, 1.6, 0.32, 0.064, 0.0128 μM) and DEF (5 μM) was prepared in DMSO, with a final solvent concentration of 3%.Reactions were 200 μl containing 0.1 μM CYP450 and an NADPH regenerating system of 50 mM K 3 PO 4 at pH 7.4 containing 1 mM glucose-6-phosphate (G6P), 1 U/ml G6P dehydrogenase, 0.1 mM NADP+, 0.25 mM MgCl 2 .Negative controls were carried out in the absence of the NADPH regenerating system.Each reaction occurred in triplicate in a black, flat-based, opaque 96-well plate on a fluorescence plate-reader (Ex ¼ 485 nm, Em ¼ 520 nm), assays were monitored for 20 min after the addition of the NADPH regenerating system.Relative fluorescence units per nmol CYP450 per second (RFU/nmol/s) was calculated by linear regression of the difference in RFU between 10 and 13 min after assays were started.

Statistical analysis
Results are presented as mean with standard deviation.All statistical analysis and graphs were produced by GraphPad Prism version 9 for Windows (GraphPad Software, La Jolla California USA, https://t.ly/ SPQ3X).All data passed Shapiro-Wilk's test for normality, qPCR, RNAi, and topical data were analysed using a one-way ANOVA and Dunnett's multiple comparison test.dsRNA efficiency data was analysed using unpaired t-tests.

Fig. 2 .
Fig. 2. Differential expression of UGTs in published transcriptomics datasets throughout Africa.Each bar is a single dataset divided into differential expression (blue), no-significant expression (grey) and missing data (purple) for the 25 UGTs present in An. gambiae.Within each bar, the number of significantly upregulated UGTs are displayed above downregulated UGTs.Countries highlighted in black have available transcriptomic data: Côte D'Ivoire, Burkina Faso, Togo, Niger.Nigeria, Equatorial Guinea, Cameroon, Chad, Sudan, Ethiopia, Uganda, Tanzania, and Zambia.Data taken from Nagi and Ingham, and Ingham et al. 11,12 .

Fig. 3 .
Fig. 3. Differential UGT transcript expression using qPCR.(A) Three insecticide resistant populations were investigated-Banfora (dark turquoise), Tiefora (dark blue) and Tiassalé (yellow)-and compared to Kisumu (dark peach), the susceptible counterpart.The y-axis is the relative transcript expression fold change compared to the control, and the x-axis is each UGT.(B) Induction of UGT-UGT302A1 (purple), UGT302H2 (pale blue), UGT306A2 (pale turquoise), UGT308G1 (pale peach)-expression in Anopheles coluzzii at varying time points following deltamethrin exposure, each compared to the unexposed control.The y-axis is the relative transcript expression fold change compared to the control, and the x-axis is each time point that the samples were stored after exposure to the insecticide.(C) Tissue specificity of UGT-UGT302A1 (purple), UGT302H2 (pale blue), UGT306A2 (pale turquoise), UGT308G1 (pale peach)-expression in insecticide resistant Anopheles gambiae antennae, head, thorax, abdomen integument, midgut, malpighian tubules, reproductive organs, and legs, each in comparison to the whole body.The y-axis is the relative transcript expression fold change compared to control, and the x-axis is each of the mosquito tissues tested.Significance determined by one-way ANOVA and Dunnett's multiple comparison test, thorax data was analysed by unpaired t-test, only statistically significant overexpression presented, dotted line represents average fold change of the control, the data are mean ± SD.

Fig. 4 .
Fig. 4. Silencing of candidate UGTs using RNAi in insecticide resistant Anopheles coluzzii.(A) dsRNA targeting individual candidate UGTs-302A1 (turquoise), 306A2 (blue), 308G1 (purple)-efficiently reduced transcript levels when tested using RT-qPCR compared to a dsGFP (green) non-target control.(B) Combined dsRNA targeting the all three candidate UGTs (dsUGTs) efficiently reduced transcript levels of each when tested using RT-qPCR compared to a dsGFP non-target control.In both graphs the y-axes are the relative transcript expression fold change compared to dsGFP control, and the x-axes are the targets of the RNAi.(C) Silencing individual and combined candidate UGTs (dsUGTs) (pink) does not increase mortality when exposed to deltamethrin compared to a dsGFP non-target control.The y-axis is the 24 h mortality post-exposure, and the x-axis are the target of the RNAi.Statistical significance determined by unpaired t-test for RT-qPCR data (A,B) or one-way ANOVA and Dunnett's multiple comparison test (C), non-significant statistics not presented, dotted line in (A,B) represents fold change of the control, in all cases the data are mean ± SD.