Comparative analysis of response to selection with three insecticides in the dengue mosquito Aedes aegypti using mRNA sequencing

Abstract

Background: Mosquito control programmes using chemical insecticides are increasingly threatened by the development of resistance. Such resistance can be the consequence of changes in proteins targeted by insecticides (target site mediated resistance), increased insecticide biodegradation (metabolic resistance), altered transport, sequestration or other mechanisms. As opposed to target site resistance, other mechanisms are far from being fully understood. Indeed, insecticide selection often affects a large number of genes and various biological processes can hypothetically confer resistance. In this context, the aim of the present study was to use RNA sequencing (RNA-seq) for comparing transcription level and polymorphism variations associated with adaptation to chemical insecticides in the mosquito Aedes aegypti. Biological materials consisted of a parental susceptible strain together with three child strains selected across multiple generations with three insecticides from different classes: the pyrethroid permethrin, the neonicotinoid imidacloprid and the carbamate propoxur.

Results: After ten generations, insecticide-selected strains showed elevated resistance levels to the insecticides used for selection. RNA-seq data allowed detecting over 13,000 transcripts, of which 413 were differentially transcribed in insecticide-selected strains as compared to the susceptible strain. Among them, a significant enrichment of transcripts encoding cuticle proteins, transporters and enzymes was observed. Polymorphism analysis revealed over 2500 SNPs showing > 50% allele frequency variations in insecticide-selected strains as compared to the susceptible strain, affecting over 1000 transcripts. Comparing gene transcription and polymorphism patterns revealed marked differences among strains. While imidacloprid selection was linked to the over transcription of many genes, permethrin selection was rather linked to polymorphism variations. Focusing on detoxification enzymes revealed that permethrin selection strongly affected the polymorphism of several transcripts encoding cytochrome P450 monooxygenases likely involved in insecticide biodegradation.

Conclusions: The present study confirmed the power of RNA-seq for identifying concomitantly quantitative and qualitative transcriptome changes associated with insecticide resistance in mosquitoes. Our results suggest that transcriptome modifications can be selected rapidly by insecticides and affect multiple biological functions. Previously neglected by molecular screenings, polymorphism variations of detoxification enzymes may play an important role in the adaptive response of mosquitoes to insecticides.

Figure 1

Transcripts differentially expressed after insecticide selection. For each Venn diagram section, the numbers of transcripts differentially expressed in any insecticide-selected strain as compared to the susceptible strain are indicated.

Figure 2

Clustering of transcripts differentially expressed across strains. The analysis was performed on the 413 known transcripts significantly differentially expressed in any insecticide-selected strain compared to the susceptible strain. Clustering was based on Euclidean distance of fold changes as compared to the susceptible strain and complete linkage algorithm. Pie charts describe biological functions affected within main clusters based on the number of transcripts assigned to each function. Stars indicate biological functions significantly enriched compared to their representation among all detected transcripts (Fisher’s test adjusted P value: * ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001). The total number of transcripts constituting each cluster is indicated. The number of transcripts with predicted functions used for building each pie chart is shown within brackets.

Figure 3

Transcripts affected by differential SNPs. For each Venn diagram section, the number of transcripts affected by differential SNPs is shown. The number of transcripts affected by differential SNPs predicted as non-synonymous according to genome annotation are shown within brackets.

Figure 4

Clustering of differential SNPs across strains. The analysis was performed on all differential SNPs falling within coding regions. Clustering was based on Euclidean distance of allele frequency variations as compared to the susceptible strain and complete linkage algorithm. Pie charts describe biological functions affected within main clusters based on the number of differential SNPs affecting transcripts assigned to each function. Stars indicate biological functions significantly enriched compared to their representation among all detected transcripts (Fisher’s test adjusted P value: * ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001). For each cluster, the total number of differential SNPs is indicated as plain text while those affecting transcripts of known function (used for pie charts) are shown within brackets.

Figure 5

Focus on detoxification: cytochrome P450s. Clustering analyses of transcription level variations (upper panel) and differential SNPs (lower panel) affecting cytochrome P450 monooxygenases (CYP genes). Clustering was performed on all transcripts showing significant transcription level variations or differential SNPs within their coding region in any strain. The green-red color scale indicates transcription level variations as compared to the susceptible strain. Stars indicate a significant differential transcription. Blue-yellow color scale indicates allele frequency variations as compared to the susceptible strain. Amino acid position, amino-acid change, transcript number and gene names are indicated. Non-synonymous variations according to genome annotation are underlined. SNPs falling within P450 Substrate Recognition Sites (SRS) are indicated. Names ending with a question mark indicate genes with ambiguous gene name (subfamily indicated).

Figure 6

Focus on detoxification: Other enzymes and transporters. Clustering analyses of transcription level variations (upper panel) and SNPs (lower panel) affecting enzymes and transporters potentially involved in insecticide detoxification. Only transcripts or differential SNPs falling within coding regions are shown. Green-red color scale indicates transcription level variations as compared to the susceptible strain. Stars indicate a significant differential transcription. Blue-yellow color scale indicates allele frequency difference as compared to the susceptible strain. Amino acid position, amino-acid change, transcript number and gene names are indicated. Non-synonymous variations according to genome annotation are underlined.