Introduction
Insecticide resistance in disease vectors has become an influential model for understanding rapid contemporary evolution but, more importantly, identifying how resistance arises and spreads is crucial for disease control. Resistance to pyrethroid insecticides in African malaria vector mosquitoes has spread to near ubiquity [1, 2] and, though it is often difficult to demonstrate its impact on malaria infections [3], in some cases it has reached levels that threaten the effectiveness of vector control programmes [4, 5]. A better understanding of resistance distribution and mechanisms will permit a more informed selection and deployment of insecticides to combat evolving mosquito populations. Whilst our understanding of the genetic basis of insecticide resistance in mosquitoes has advanced substantially [6], especially in the important vector Anopheles funestus[7], molecular diagnostics for the major vectors in the An. gambiae complex remain limited to a handful of mutations[8] which explain a relatively small fraction of the variance in phenotype [9, 10] or which are now at such high frequency as to provide limited diagnostic resolution [11].
Long-lasting insecticidal nets (LLINs) are the principal tool for vector control to combat malaria, especially in sub-Saharan Africa [12]. The majority of LLINs are treated only with pyrethroid insecticides, to which resistance is now widespread [13]. Though behavioural variation and physical or physiological modifications affecting insecticide uptake may sometimes play a role, pyrethroid resistance is caused predominantly by two distinct mechanisms. The first is resistance via point mutations in the target-site of the insecticide, for pyrethroids the Voltage-gated sodium channel (Vgsc ), which results in decreased sensitivity to the insecticide [14]; the second is metabolic resistance due to over-expression or altered activity of detoxification enzymes, of which the cytochrome P450 family is commonly considered most important [15, 16]. Cytochrome P450 activity is inhibited by the synergist piperonyl-butoxide (PBO), and bed nets incorporating PBO are effective against P450-mediated resistance, as demonstrated by large-scale field trials [4, 17]. Given the continued operational use of pyrethroid-containing nets, it is vital that we understand the genetic mechanisms that may impact their efficacy, to optimise bednet deployment, preferably using information from rapidly-applied DNA markers. In advance of a randomized control trial of PBO-LLINs [4], we sought to characterise pyrethroid resistance mechanisms in the primary malaria vector An. gambiae s.s. [18] in Uganda and Kenya.
The recent development of the Anopheles gambiae 1000 genomes project (Ag1000g) has led to a step change in our ability to identify DNA variation driven by selection pressure. We have been able to perform genome-wide searches for regions under recent natural selection in insecticide resistant populations across Africa, and work has shown that the strongest selective sweeps in the genome are all found around genes known to be important for resistance [6, 19]. Furthermore, a whole-genome scan of copy number variants (CNVs) in the Ag1000g data revealed that increases in gene copy number were highly enriched in clusters of detoxification genes, pointing to a potentially widespread mechanism for increased gene expression [20], which, in some cases, may elevate resistance. A number of gene duplications were observed around the Cyp6aa /Cyp6p gene family cluster on chromosome 2R, and the majority of these duplications included the geneCyp6aa1 (Figure 1). Cyp6aa1 has been found to be overexpressed in pyrethroid resistant populations in congeneric species [21-23], but it has received very little attention compared to known insecticide-metabolizing genes such as Cyp6m2 [24, 25],Cyp6p3 [15] and Cyp9k1 [26] and its importance in resistance in An. gambiae remains unknown.
In this study, we examine a strong selective sweep detected in theCyp6aa/Cyp6p genomic region in samples of An. gambiae s.s.from Uganda and Western Kenya. We find that the sweep is closely associated with three mutations (a SNP in Cyp6p4 , a duplication of Cyp6aa1 and a partial transposable element insertion termed ZZB-TE) in tight physical and statistical linkage. The triple-mutant haplotype is associated with a high-level of pyrethroid resistance, most notably to deltamethrin. The three mutations appeared sequentially, leading to successive selective sweeps, with the triple-mutant haplotype replacing earlier variants and then spreading rapidly across East and Central Africa. We show that this haplotype is under positive selection and causes increased expression of key cytochrome P450s and through recombinant protein expression using both an E. coli and anSf9 -baculovirus system we show that both CYP6AA1 and CYP6P4 are capable of metabolizing pyrethroid insecticides.