The recent escalation in strength of pyrethroid resistance in Anopheles coluzzi in West Africa is linked to increased expression of multiple gene families

Abstract

Background: Since 2011, the level of pyrethroid resistance in the major malaria mosquito, Anopheles coluzzi, has increased to such an extent in Burkina Faso that none of the long lasting insecticide treated nets (LLINs) currently in use throughout the country kill the local mosquito vectors. We investigated whether this observed increase was associated with transcriptional changes in field-caught Anopheles coluzzi using two independent whole-genome microarray studies, performed in 2011 and 2012.

Results: Mosquitoes were collected from south-west Burkina Faso in 2011 and 2012 and insecticide exposed or non-exposed insects were compared to laboratory susceptible colonies using whole-genome microarrays. Using a stringent filtering process we identified 136 genes, including the well-studied detoxification enzymes (p450 monoxygenases and esterases) and non-detoxification genes (e.g. cell transporters and cuticular components), associated with pyrethroid resistance, whose basal expression level increased during the timeframe of the study. A subset of these were validated by qPCR using samples from two study sites, collected over 3 years and marked increases in expression were observed each year. We hypothesise that these genes are contributing to this rapidly increasing resistance phenotype in An. coluzzi. A comprehensive analysis of the knockdown resistance (kdr) mutations (L1014S, L1014F and N1575Y) revealed that the majority of the resistance phenotype is not explained by target-site modifications.

Conclusions: Our data indicate that the recent and rapid increase in pyrethroid resistance observed in south-west Burkina Faso is associated with gene expression profiles described here. Over a third of these candidates are also overexpressed in multiple pyrethroid resistant populations of An. coluzzi from neighbouring Côte d'Ivoire. This suite of molecular markers can be used to track the spread of the extreme pyrethroid resistance phenotype that is sweeping through West Africa and to determine the functional basis of this trait.

Figure 1

Interwoven loop designs for microarray experiments performed in 2011 and 2012. In 2011, deltamethrin selected mosquitoes from VK7 (VKR; LT₅₀ = 254 min), unexposed mosquitoes from VK7 (VKC) and the Mali susceptible lab strain (MAL) were compared. In 2012, four mosquitoes populations were compared including deltamethrin selected mosquitoes from VK7 (VKR; 600 mins exposure), unexposed mosquitoes from a deltamethrin resistant population approximately 120 km from VK7, Tengrela (TEN), the susceptible MAL strain and a second fully susceptible lab strain from N’Gousso, Cameroon (NG). The direction of the arrows represents a cy3 to cy5 hybridisation and three biological replicates were used for each comparison.

Figure 2

Microarray data analysis schema showing the different steps and the number of probes obtained after each filtering step. Each filtering step is based on our hypothesis of over-expression in field resistant populations compared to laboratory susceptible strains (steps A, B and C), over-expression in VK7 compared to TEN (step D) and up-regulation in 2012 compared to 2011 according to the VK7-MAL comparison (step E). The green and red arrows represent dye swaps between the samples.

Figure 3

Relative mRNA levels of candidate resistance genes measured by qPCR. The level of expression was measured from An. coluzzi from Vallée du Kou (2011, 2012, and 2013) and Tengrela (2012 and 2013) in samples independent to those used in the microarray experiments and not exposed to insecticide. The relative values are presented according to the ddCt method [18] ± SEM of six biological replicates. A two-tailed Student’s t-test was performed between the yearly values within population (log-scale). *= p < 0.05; n.s. = not significant.

Figure 4

Hierarchial clustering analysis of the candidate probes for deltamethrin resistance over-expressed in An. coluzzi from VK7. Clustering was performed using the Euclidean distance method on the 157 probes which were over-expressed in VK7 versus MAL in 2011 and 2012 as well as those over-expressed against NG. The colour scale represents the log fold change of VK7 against the susceptible strain.

Figure 5

Haplotypic association tests for the three kdr haplotypes ( 1014L-1575N , 1014F-1575N and 1014 F-1575Y ) with survival of VK7 An. coluzzi to deltamethrin LT ₅₀ . Odds ratios (OR) are represented with the level of significance and the arrows within the triangles show the direction of the OR calculation. *p < 0.05, **p < 0.01.