Genome-wide transcriptional analyses in Anopheles mosquitoes reveal an unexpected association between salivary gland gene expression and insecticide resistance

Abstract

Background: To combat malaria transmission, the Ugandan government has embarked upon an ambitious programme of indoor residual spraying (IRS) with a carbamate class insecticide, bendiocarb. In preparation for this campaign, we characterized bendiocarb resistance and associated transcriptional variation among Anopheles gambiae s.s. mosquitoes from two sites in Uganda.

Results: Gene expression in two mosquito populations displaying some resistance to bendiocarb (95% and 79% An. gambiae s.l. WHO tube bioassay mortality in Nagongera and Kihihi, respectively) was investigated using whole-genome microarrays. Significant overexpression of several genes encoding salivary gland proteins, including D7r2 and D7r4, was detected in mosquitoes from Nagongera. In Kihihi, D7r4, two detoxification-associated genes (Cyp6m2 and Gstd3) and an epithelial serine protease were among the genes most highly overexpressed in resistant mosquitoes. Following the first round of IRS in Nagongera, bendiocarb-resistant mosquitoes were collected, and real-time quantitative PCR analyses detected significant overexpression of D7r2 and D7r4 in resistant mosquitoes. A single nucleotide polymorphism located in a non-coding transcript downstream of the D7 genes was found at a significantly higher frequency in resistant individuals. In silico modelling of the interaction between D7r4 and bendiocarb demonstrated similarity between the insecticide and serotonin, a known ligand of D7 proteins. A meta-analysis of published microarray studies revealed a recurring association between D7 expression and insecticide resistance across Anopheles species and locations.

Conclusions: A whole-genome microarray approach identified an association between novel insecticide resistance candidates and bendiocarb resistance in Uganda. In addition, a single nucleotide polymorphism associated with this resistance mechanism was discovered. The use of such impartial screening methods allows for discovery of resistance candidates that have no previously-ascribed function in insecticide binding or detoxification. Characterizing these novel candidates will broaden our understanding of resistance mechanisms and yield new strategies for combatting widespread insecticide resistance among malaria vectors.

Keywords: Anopheles gambiae; Bendiocarb; Indoor residual spraying; Long-lasting insecticidal nets; Malaria control; Pyrethroids; Salivary gland proteins; Whole-genome microarray.

Fig. 1

Map of Uganda showing study sites. Larval collections were completed in both Kihihi and Nagongera in January–May 2014. Larval collections were completed in Nagongera in January–May 2015. The geographical origin of the Kisumu laboratory strain of susceptible mosquito is also indicated. The figure was made by the authors using Google Maps (maps.google.com)

Fig. 2

Expression of insecticide resistance candidates in An. gambiae s.s. mosquitoes collected in 2014, measured by real-time quantitative PCR. Gene expression in mosquitoes collected from Kihihi (a) and Nagongera (b). Bendiocarb-resistant mosquitoes were selected using a standard WHO tube bioassay. Unexposed mosquitoes were placed in a tube with a control paper. Each RNA sample was extracted from pools of five mosquitoes. The y-axis depicts the level of transcript in each sample relative to the unexposed control group. The median value for each treatment is indicated by a line. Two-tailed Mann-Whitney p < 0.05 (*) or p < 0.01 (**)

Fig. 3

Expression of insecticide resistance candidates in An. gambiae s.s. mosquitoes collected in Nagongera in January–May 2015, measured by real-time quantitative PCR. Resistant and unexposed mosquitoes were selected as described in Fig. 2. Each RNA sample was extracted from an individual mosquito. The y-axis depicts the level of transcript in each sample relative to the unexposed control group. The median value for each treatment is indicated by a line. Two-tailed Mann-Whitney p < 0.01 (**)

Fig. 4

Comparison between serotonin binding to D7r4 protein and the predicted mode of binding of bendiocarb. Serotonin, as visualised in the crystal structure (PDB code 2qeh; [15]), is displayed on the left, and the ROSIE server [16, 17] predicted pose of bendiocarb for the same protein on the right, each in a stick representation. Ligand atoms are coloured white (carbon), red (oxygen) or blue (nitrogen). Solvent-accessible protein surfaces were calculated within PyMOL using the default solvent molecule radius of 1.4 Å. Surface contributed by carbon atoms is coloured green in the serotonin complex and yellow in the bendiocarb complex. In both, red and blue are used for surface contributed by oxygen or nitrogen atoms, respectively. The figure was made with PyMOL (pymol.org)