The impact of agrochemical pollutant mixtures on the selection of insecticide resistance in the malaria vector Anopheles gambiae: insights from experimental evolution and transcriptomics

Abstract

Background: There are several indications that pesticides used in agriculture contribute to the emergence and spread of resistance of mosquitoes to vector control insecticides. However, the impact of such an indirect selection pressure has rarely been quantified and the molecular mechanisms involved are still poorly characterized. In this context, experimental selection with different agrochemical mixtures was conducted in Anopheles gambiae. The multi-generational impact of agrochemicals on insecticide resistance was evaluated by phenotypic and molecular approaches.

Methods: Mosquito larvae were selected for 30 generations with three different agrochemical mixtures containing (i) insecticides, (ii) non-insecticides compounds, and (iii) both insecticide and non-insecticide compounds. Every five generations, the resistance of adults to deltamethrin and bendiocarb was monitored using bioassays. The frequencies of the kdr (L995F) and ace1 (G119S) target-site mutations were monitored every 10 generations. RNAseq was performed on all lines at generation 30 in order to identify gene transcription level variations and polymorphisms associated with each selection regime.

Results: Larval selection with agrochemical mixtures did not affect bendiocarb resistance and did not select for ace1 mutation. Contrastingly, an increased deltamethrin resistance was observed in the three selected lines. Such increased resistance was not majorly associated with the presence of kdr L995F mutation in selected lines. RNA-seq identified 63 candidate resistance genes over-transcribed in at least one selected line. These include genes coding for detoxification enzymes or cuticular proteins previously associated with insecticide resistance, and other genes potentially associated with chemical stress response. Combining an allele frequency filtering with a Bayesian FST-based genome scan allowed to identify genes under selection across multiple genomic loci, supporting a multigenic adaptive response to agrochemical mixtures.

Conclusion: This study supports the role of agrochemical contaminants as a significant larval selection pressure favouring insecticide resistance in malaria vectors. Such selection pressures likely impact kdr mutations and detoxification enzymes, but also more generalist mechanisms such as cuticle resistance, which could potentially lead to cross-tolerance to unrelated insecticide compounds. Such indirect effect of global landscape pollution on mosquito resistance to public health insecticides deserves further attention since it can affect the nature and dynamics of resistance alleles circulating in malaria vectors and impact the efficacy of control vector strategies.

Keywords: Anopheles gambiae; Agrochemical pesticides; Metabolic resistance; Resistance selection; Transcriptomics.

Fig. 1

Evolution of adult resistance to bendiocarb and deltamethrin in each selected line compared to the control line. For each line, insecticide resistance levels are shown as % mortality to 0.1% bendiocarb and to 0.05% deltamethrin. * Indicate significantly distinct mortalities using Fisher’s exact test (P < 0.05), and error bars show SD in means 80 < n > 100 computed from each replica tube

Fig. 2

Evolution of the kdr L995F mutation frequency during the selection process. Coloured bars show the genotype frequencies, as assayed from 30 individuals. Generation 0 corresponds to the parental line. Blue: 995FF kdr (resistant) homozygotes; Orange: 995FL heterozygotes; Grey: 995LL (wildtype) homozygotes. Black line: 995F allele frequency

Fig. 3

Evolution of the Ace1 mutation frequency during the selection process. Coloured bars show the genotype frequencies, as assayed from 30 individuals. Generation 0 corresponds to the parental line. Blue: 119SS (resistant) homozygotes; Orange: 119GS heterozygotes; Grey: 119GG (wildtype) homozygotes. Black line: 119S allele frequency

Fig. 4

Expression profiles of candidate resistance genes in each selected line. Gene transcription levels were quantified by RNA-seq after 30 generations of selection. Only genes showing a significant differential transcription level between at least one selected line and the control line are shown (*: FC ≥ 1.5-fold in either direction and corrected P value ≤ 0.005). Red dots indicate genes known as contributing to insecticide resistance in malaria vectors. Black dots indicate genes affected by differential or outlier SNPs

Fig. 5

Selection signatures observed in each selected line. SNPs diverging between the control line and each selected line were identified using a frequency-based approach (Diff SNPs) and a FST-based approach (outlier SNPs) and then averaged by gene (see methods). The upper Y axis shows the mean Diff SNP score per gene. The lower Y axis shows the proportion of outliers per gene. Symbol size increases with the number of polymorphic SNPs per gene. Triangles and circles denote candidate and non-candidate genes, respectively. Filled symbols indicate the presence of at least one differential or outlier SNP affecting the protein sequence. Blue and red symbols indicate candidate genes with a mean differential score > 0.4 or > 20% outliers, respectively; the corresponding gene names are indicated. Loci commonly associated with insecticide resistance in An. gambiae are indicated by dashed lines. The genomic scale shows chromosome arms with ticks every 10 Mb