The relationship between insecticide resistance, mosquito age and malaria prevalence in Anopheles gambiae s.l. from Guinea

Insecticide resistance across sub-Saharan Africa may impact the continued effectiveness of malaria vector control. We investigated the association between carbamate and pyrethroid resistance with Anopheles gambiae s.l. parity, Plasmodium falciparum infection, and molecular insecticide resistance mechanisms in Guinea. Pyrethroid resistance was intense, with field populations surviving ten times the insecticidal concentration required to kill susceptible individuals. The L1014F kdr-N1575Y haplotype and I1527T mutation were significantly associated with mosquito survival following permethrin exposure (Prevalence Ratio; PR = 1.92, CI = 1.09–3.37 and PR = 2.80, CI = 1.03–7.64, respectively). Partial restoration of pyrethroid susceptibility following synergist pre-exposure suggests a role for mixed-function oxidases. Carbamate resistance was lower and significantly associated with the G119S Ace-1 mutation. Oocyst rates were 6.8% and 4.2% among resistant and susceptible mosquitoes, respectively; survivors of bendiocarb exposure were significantly more likely to be infected. Pyrethroid resistant mosquitoes had significantly lower parity rates than their susceptible counterparts (PR = 1.15, CI = 1.10–1.21). Our findings emphasize the need for additional studies directly assessing the influence of insecticide resistance on mosquito fitness.

The G119S Ace-1 mutation was detected in 10% of samples (7/67) (Fig. 1C). All individuals with the resistant allele were heterozygous for this mutation and were identified as An. gambiae s.s. Frequencies of the G119S Ace-1 mutation varied significantly among study districts (Fisher's Exact test; p = 0.029) ( Table 4 and Fig. 1C). There was a significant association between presence of the G119S Ace-1 mutation and ability of mosquitoes to survive 30 minute and two hour carbamate exposure (Fisher's Exact test; p = 0.0047 and Fisher's Exact test; p = 0.0015, respectively).
Across Forecariah Prefecture, significant deviations from Hardy-Weinberg equilibrium were observed for L1014F kdr in Madinagbe, Maferinyah Centre I, Senguelen and Yindi (p < 0.005 for all) ( Table 3) and for N1575Y in Yindi (p = 0.040) but not for any other resistance loci.

Discussion
By 2016, resistance to at least one insecticide has been reported from over 80% of malaria endemic countries 1 , representing a threat to the continued efficacy of key malaria control strategies. However, the relative impact of decreased mosquito susceptibility on vectorial capacity remains unknown, largely due to a paucity of field data. To begin to address this deficit, in an area of high malaria transmission in Guinea, we characterized levels of insecticide resistance and age of local vector populations, in combination with molecular identification of resistance markers and detection of malaria infection.
In Forecariah Prefecture, intense pyrethroid resistance was abundant, evidenced by vector populations that were not only resistant to ten times the insecticide concentration required to kill susceptible individuals, but were also capable of surviving these doses for up to two hours. These observations are of concern given that pyrethroid LLINs are the only malaria vector control intervention deployed in this area. The restoration of mosquito susceptibility following pre-exposure to PBO and the association of mutations in the voltage-gated sodium channel (VGSC; L1014F kdr-N1575Y haplotype and I1527T) with lower mortality to permethrin suggests that both target site mutations and increased activity of P450 monooxygenases are contributing to pyrethroid resistance. This finding aligns with reports from Burkina Faso where upregulated detoxification enzymes were responsible for extreme pyrethroid resistance in An. coluzzii, with N1575Y associated with more limited tolerance to deltamethrin 29 . It has been proposed that N1575Y may compensate for fitness costs incurred by the L1014F kdr mutation and provide additional pyrethroid resistance 30 . The I1527T mutation is located within the III S6 helix of the VGSC, adjacent to a predicted pyrethroid/DDT binding site; nearby residues have already been implicated in resistance in other medically-important vector species 31,32 . However, to date, a role for I1527T in phenotypic resistance in An. gambiae s.l. has not been confirmed, nor has this mutation been detected in Guinea previously.
Of the three pyrethroid insecticides evaluated, resistance levels were often highest to permethrin, despite all mass LLIN campaigns in the country having exclusively distributed deltamethrin-treated products 25 . This may be explained by L1014F kdr playing a larger contributing role to resistance to type I (permethrin) versus type II (alpha-cypermethrin and deltamethrin) pyrethroids 33 . Levels of bendiocarb resistance were comparatively lower, which was not unexpected considering IRS is not routinely conducted by the Guinean NMCP. In the case of carbamate resistance, the presence of the G119S Ace-1 mutation was highly predictive of bioassay survivorship and tolerance to increased exposure times. Previous studies have demonstrated that the G119S Ace-1 substitution imposes a high fitness cost 34,35 by decreasing affinity of the resistant enzyme for its substrate by more than 60% 36 ; heterogeneous or homogeneous duplication of this locus has been proposed to restore activity 37 . In Forecariah Prefecture, frequency of the G119S Ace-1 resistant allele (of undetermined copy number) was low and could be a result of selective pressure imposed by unregulated agricultural use of carbamates and organophosphates, which were readily available at the local market (S. Irish, personal communication).
Our results demonstrated that pyrethroid resistance was not associated with malaria prevalence in the mosquitoes that were tested, with no significant differences observed between Plasmodium infection in susceptible or resistant individuals. Our findings contrast with a number of laboratory studies ascribing potential fitness costs to insecticide resistance which may have the collateral benefit of reducing malaria transmission 16  www.nature.com/scientificreports www.nature.com/scientificreports/ in our study, survivors of bendiocarb exposure were significantly more likely to be infected with P. falciparum. Pyrethroid exposure has previously been shown to adversely affect P. falciparum development in L1014F kdr resistant An. gambiae s.s. in Uganda 21 , which may explain part of our data, if we assume resistant individuals are able to survive contact with LLINs and susceptible vectors are not. Alternative observations have also been reported whereby kdr has been shown to potentiate the vector competence of An. gambiae s.s. 23 , and Plasmodium infection has partially restored DDT susceptibility in An. gambiae carrying the kdr mutation 38 but also reduced the survival of resistant vectors in the absence of insecticide exposure 17 . The discrepancies between in vitro studies and our field data may reflect more generalised fitness variation between laboratory mosquito strains and wild populations and/or different underlying resistance mechanisms, e.g. target site mutations vs. increased activity of metabolic enzymes, and strongly support the need for additional studies in areas of differing resistance mechanisms and disease transmission intensities.
We examined the impact of resistance on mosquito age as increasing age has been proposed to restore insecticide susceptibility [39][40][41][42][43][44][45][46] and insecticide exposure may still reduce resistant vector life-span through delayed mortality effects 16 . In Forecariah Prefecture, we observed a similar phenomenon, with resistant mosquitoes having significantly lower parity rates than their susceptible counterparts.
Elucidating the interaction between insecticide resistance and vectorial capacity is complex and challenging in field conditions and a number of limitations were encountered during this study. Adult female mosquitoes were sampled using HLCs and via manual aspiration from house walls to maximise the number of individuals available for bioassay testing. However, these strategies may have introduced a bias in species composition collected; previous studies have suggested that proportions of An. gambiae s.s. and An. coluzzii may differ significantly between larval and adult spray catches 47 . We also encountered issues with individuals that did not amplify with two sub-species PCR assays 48,49 , potentially reflecting polymorphisms at the primer binding sites and/or the presence of a cryptic sub-species; previously a cryptic subgroup GOUNDRY has been reported, with close genetic affinities to An. coluzzii and enhanced susceptibility to P. falciparum infection [50][51][52] . While use of F 1 progeny mosquitoes is recommended for resistance testing 53,54 , to investigate the impact of resistance on Plasmodium infection in the vector, it was necessary to sample wild caught adults of undetermined, mixed age and physiological status; the latter parameter may have varied between sampling methods, with more fed mosquitoes collected from inside houses, compared to unfed, host-seeking mosquitoes in HLCs. As a consequence, mortality in our study may have been under-estimated, considering the proposed inverse relationship between vector resistance and age [39][40][41][42][43][44][45] . Similarly, blood-feeding among resistant mosquitoes has been suggested to increase insecticide tolerance 55 , however, no association between physiological status and resistance was observed in our dataset. Our study would have also benefitted from larger sample sizes of mosquitoes tested in bioassays, to statistically power comparisons between different resistance intensities. Furthermore, abdomen positive mosquitoes cannot all be assumed to become infective sporozoite-transmitters, particularly if vector life-span is reduced; a positive abdomen can also reflect a recent feed on an infected individual or the passage of sporozoites to the salivary glands. Finally, manual parity dissections have a number of known constraints, including insensitivity, especially in low endemicity areas, and inter-operator subjectivity. While every effort was made to consistently bisect individuals, and our reported sporozoite rate was low, other studies have demonstrated higher number of sporozoite false positives by PCR when abdomens were removed posterior to the junction of the abdomen and thorax 56 . We also cannot entirely discount natural variation in parity, which may have increased over the sampling period at the beginning of the rainy season.

Conclusions
These findings present a comprehensive overview of the current levels of insecticide resistance and underlying target site mutations present in Maferinyah, Guinea, an area of high malaria transmission. Local mosquito populations were intensely resistant to pyrethroids (alpha-cypermethrin, deltamethrin and permethrin), associated with high frequencies of the L1014F kdr allele. N1575Y and I1527T mutations in the VGSC gene were present at lower levels and may warrant increased surveillance efforts, particularly as L1014F kdr approaches fixation. Restoration of mosquito susceptibility following pre-exposure to PBO indicates increased activity of detoxification enzymes is also contributing to pyrethroid resistance in this area and requires additional characterization. Despite no ongoing vector control activities using carbamates, bendiocarb resistance was observed, and the G119S Ace-1 allele was detected in a subset of tolerant individuals. Malaria infection was not associated with pyrethroid resistance but it was associated with bendiocarb resistance. In general, resistant vectors were younger than their susceptible counterparts. Further studies are necessary to investigate the impact of insecticide resistance on vector fitness, including mosquito fecundity, egg viability, hatchability and parasite development following an infected blood meal.

Methods study area and mosquito collections. Mosquito collections were undertaken in six villages in the
Maferinyah sub-prefecture, Forecariah Prefecture (Fandie, Madinagbe, Maferinyah Centre I, Moribayah, Senguelen and Yindi), in Southwest Guinea. Deltamethrin-treated LLINs had been distributed as part of a national mass campaign in Maferinyah in 2016. Sampling was conducted between 22 nd June and 17 th July 2017, coinciding with the beginning of the long rainy season. Following consent from the household owner, indoor resting, female Anopheles mosquitoes were collected from house walls by manual aspiration between 7:00 h and 12:00 h. In the same villages, mosquitoes were also sampled using HLCs, carried out between 22:00 h and 3:00 h. Mosquitoes were stored in cages with access to 10% sugar solution, prior to transport to the Centre de Formation et de Recherche en Santé rurale de Maferinyah (CNFRSR) for analysis. www.nature.com/scientificreports www.nature.com/scientificreports/ CDC resistance intensity and synergist bioassays. All bioassays were performed using mosquitoes identified morphologically as An. gambiae s.l. 57 ; wild caught females were held for a maximum of 48 hours before testing. Centers for Disease Control and Prevention (CDC) resistance intensity bioassays for three pyrethroid insecticides (alpha-cypermethrin, deltamethrin and permethrin) were conducted according to published guidelines 53 . Stock solutions of 1, 2, 5 and 10 times the diagnostic dose of insecticide (alpha-cypermethrin: 12.5 μg/ bottle; deltamethrin: 12.5 μg/bottle; and permethrin: 21.5 μg/bottle), were prepared by diluting technical grade insecticide in 50 ml of acetone. Bioassays for bendiocarb were conducted using the diagnostic dose (1X: 12.5 μg/bottle). The inside of each Wheaton 250 ml bottle along with its cap was coated with 1 ml of stock solution by rolling and inverting the bottles. In each test, a control bottle was coated with 1 ml of acetone. Following coating, bottles were left to dry in a dark box. Approximately 15-25 field-caught adult female An. gambiae s.l. of unknown age and mixed physiological status were introduced into each bottle using a mouth aspirator and mortality was recorded at 15 minute intervals until all were dead or up to two hours; dead individuals at 30 minutes were removed from bottles by aspiration and again after two hours. In each bioassay, a control bottle, coated with acetone, was run in parallel. Mortality in control bottles never exceeded 5%. In select sites with significant pyrethroid resistance, synergist assays were also conducted by pre-exposing mosquitoes to piperonyl butoxide (PBO) for 1 hour (100 µg/bottle) prior to performing pyrethroid bioassays. Multiple replicates were performed per insecticide and study village, depending on mosquito availability, and individual surviving (resistant) and dead (susceptible) mosquitoes were preserved in RNAlater ® (Thermo Fisher Scientific, UK) at −20 °C at the London School of Hygiene and Tropical Medicine (LSHTM) and CDC. Prior to molecular analysis, mosquito heads/thoraxes were separated from abdomens under a dissecting microscope and stored separately. parity dissection. Ovarian dissection to determine mosquito parity was performed on ten mosquitoes (including both survivors and those that died) selected randomly from each bottle after bioassay completion 58 . The ovaries of each mosquito were dissected on a sterile microscope slide in distilled water, using a binocular dissection microscope and physiological status was determined. The ovaries were then examined under a light microscope (10X magnification) for the presence of tightly coiled skeins or loose coils, indicative of a nulliparous or parous ovary, respectively. On completion of ovarian dissection, head/thoraxes and abdomens of each mosquito were stored separately in RNAlater ® (Thermo Fisher Scientific, UK) at −20 °C.

Molecular species identification. A subset of susceptible and resistant
An. gambiae s.l. mosquitoes from all six villages containing both nulliparous and parous individuals were selected for molecular analysis at the LSHTM and CDC. Genomic DNA from dissected body parts was extracted per protocol using Qiagen DNeasy 96 Blood and Tissue kits (Qiagen, UK) at LSHTM or Extracta TM DNA Prep for PCR-Tissue kits (QuantaBio, USA) at CDC.
At LSHTM, molecular species identification was performed using a multiplex TaqMan real time PCR assay to detect and discriminate An. gambiae and An. arabiensis 59  www.nature.com/scientificreports www.nature.com/scientificreports/ cycles of 95˚C for 30 sec, 59˚C for 30 sec and 72˚C for 30 sec and a final extension of 72˚C for 5 min. PCR products were separated and visualised using 2% Egel EX agarose gels (Invitrogen) with SYBR safe and an Invitrogen E-gel iBase Real-Time Transilluminator. A PCR product of 214 bp indicated the susceptible wild type allele and a PCR product of 156 bp indicated the resistant allele. An. coluzzii AKDR and An. gambiae s.s. RSP-ST were used as positive and negative controls for L1014F kdr, respectively.
At LSHTM, a larger sub-sample of individuals from all six districts was chosen to be screened for the N1575Y mutation (previously shown to have low prevalence in West African populations) using the TaqMan real time PCR assay developed by Jones et al. 30 . PCR reactions were prepared using Qiagen Quantitect Probes Master mix (Qiagen) with each reaction containing 10 µl of master mix, a final concentration of 1 µM of each primer and 0.5 µM of each probe, 5 µl of PCR grade water and 2 µl of template DNA, for a final reaction volume of 20 µl. Prepared reactions were run on a Roche LightCycler 96 System for 15 min at 95˚C, followed by 35 cycles of 94˚C for 15 sec and 60˚C for 60 sec. Positive controls from gDNA extracted from known An. gambiae s.s. with or without the N1575Y mutation were included on each run in addition to no template controls (NTCs). PCR results were analysed using the LightCycler 96 software (Roche Diagnostics).
At CDC, to detect the N1575Y mutation, a 218 bp fragment of the VGSC channel spanning domains III-IV was sequenced. PCR reactions were prepared using 25 µl of 2X AccuStart TM II PCR SuperMix, a final concentration of 20 pmol/μl of primers Exon_29_F (5′-AAATGCTCAGGTCGGTAAACA-3′) and Exon_29_R (5′-GCCACTGGAAAGAATGGAAA-3′) 30 and 2 μl of template DNA, for a final reaction volume of 50 μl. PCR reaction conditions were 95 °C for 3 min, followed by 35 amplification cycles (95 °C for 30 sec, 58 °C for 30 sec, 72 °C for 30 sec) and a final elongation step at 72 °C for 5 min. Amplified PCR products were visualized on 2% agarose gels, stained with GelRed TM (Biotium, USA) and purified using the 96-well Millipore TM MultiScreen TM HTS vacuum manifold system. Bi-directional sequencing was performed using primers Exon_29_F and Exon_29_R with the BigDye ® Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems, USA) according to the manufacturer's protocol. Sequencing reactions were purified using the Big Dye ® Xterminator TM Purification Kit (Applied Biosystems, USA), according to the manufacturer's protocol and data were generated using a 3500xL Genetic Analyzer (Applied Biosystems, USA). Sequences were assembled manually in BioEdit v7.0.9.0 sequence alignment editor software (Ibis Biosciences, USA) and unambiguous consensus sequences were produced for each individual. Consensus sequences are available from GenBank under the accession numbers MH929325 -MH929433.
The presence of the G119S Ace-1 mutation was determined using PCR restriction fragment length polymorphism analysis 61 . At both LSHTM and CDC, PCR reactions were prepared using Hot Start Taq 2X Master Mix (New England Biolabs) with each reaction containing 12.5 µl of master mix, a final concentration of 1 µM of each primer, 2 µl template DNA, to a final reaction volume of 25 µl. Prepared reactions were run on a BIO-RAD QPCR system for 5 min at 95˚C, followed by 30 cycles of 95˚C for 30 sec, 52˚C for 30 sec and 72˚C for 1 min and a final extension of 72˚C for 5 min. PCR products were digested using the AluI restriction enzyme through incubation at 37˚C for 16 hr followed by 65˚C for 20 min. DNA fragments were visualised on 2% Egel EX agarose gels (Invitrogen) with SYBR safe and an Invitrogen E-gel iBase Real-Time Transilluminator. 194 bp undigested PCR products indicated the susceptible allele and 74 bp and 120 bp digested fragments indicated the presence of the resistant allele. Presence of all three product sizes indicated that the sample was heterozygous. plasmodium falciparum detection. Bisected mosquito head/thoraxes and abdomens from bioassay individuals were screened separately for P. falciparum infection, by targeting a 120 bp sequence of the P. falciparum cytochrome c oxidase subunit 1 (cox1) mitochondrial gene 62 . At LSHTM real-time PCR reactions were prepared using FastStart SYBR Green Master mix (Roche Diagnostics) with each reaction containing 5 µl of master mix, a final concentration of 1 µM of each primer, 1 µl of PCR grade water and 2 µl template DNA, to a final reaction volume of 10 µl. Prepared reactions were run on a Roche LightCycler 96 System for 15 min at 95˚C, followed by 35 cycles of 95˚C for 15 sec and 58˚C for 30 sec. Amplification was followed by a dissociation curve (95˚C for 10 sec, 65˚C for 60 sec and 97˚C for 1 sec) to ensure the correct target sequence was amplified. Positive controls from gDNA extracted from a cultured P. falciparum-infected blood sample (parasitaemia of ~10%) were included on each run, in addition to no template controls (NTCs). PCR results were analysed using the LightCycler ® 96 software (Roche Diagnostics).
At CDC, a conventional PCR was used for cox1 detection 63 . PCR reactions were prepared using 6.25 µl of2X AccuStart TM II PCR SuperMix, a final concentration of 5 µM of primers COX-IF (5′-AGAACGAACGCTTTTAACGCCTG-3′) and COX-IR (5′-ACWGGATGGACTTTATATCCACCATTAAGT-3′) 63 and 2 μl of template DNA, for a final reaction volume of 12.5 μl. PCR reaction conditions were 94 °C for 5 min, followed by 40 amplification cycles (94 °C for 1 min, 62 °C for 1 min, 72 °C for 1.5 min) and a final elongation step at 72 °C for 10 min. Amplified PCR products were visualized on 2% agarose gels, stained with GelRed TM (Biotium, USA), alongside the same positive and negative controls used in the real-time assay. A positive head/thorax or abdomen was assumed to indicate the presence of sporozoites or oocysts, respectively. Data analysis. Percent mosquito mortality for each insecticide dose was interpreted using the WHO criteria: 98-100% mortality at 30 minutes of exposure indicates 'susceptibility' , 90-97% mortality suggests 'possible resistance' and <90% indicates the presence of 'resistance' 54 . Where molecular species ID data were available, results are reported for An. coluzzii and An. gambiae s.s. separately; otherwise individuals are referred to as An. gambiae s.l. Pearson's Chi squared tests and Fisher's exact tests (when sample sizes were small) were used to investigate the statistical association between survival by site, allele frequencies, and deviations from Hardy-Weinberg equilibrium. Other analyses utilized quasi-Poisson regression to estimate prevalence ratios (PRs) of mosquito survival, except the model in Fig. 2, which used a binomial model to ensure confidence intervals did not extend beyond 100%. Outcomes from pyrethroid bioassays were evaluated separately to carbamate assays, unless otherwise