The introduction of An. stephensi mosquitos into the Horn of Africa is a significant threat to malaria control, particularly as this vector can be found in urban areas, which are at lower risk of malaria transmission compared to rural regions. As vector control strategies have been successful in reducing malaria transmission, surveillance of insecticide resistance, which is spreading globally, is essential to inform malaria control programmes. To assist surveillance activities, we evaluate an Amp-seq assay to screen for known and novel mutations associated with a range of insecticides. In the An. stephensi screened here the kdr mutation L958F (L1014F in M. domestica; L1014F kdr-west) was present at a frequency of 13.7%. This L1014F kdr-west mutation was first detected in An. gambiae in West Africa and has been reported in low frequency (5.7%) in An. stephensi collected in 2018 in other regions in Ethiopia (Bati, Degehabur, Dire Dawa, Gewane, and Semera), but was not detected in the same study in samples screened from Awash Sebat Kilo, the region where our samples were collected. Just a year after, we were able to detect the L1014F kdr-west mutation, maybe because our samples size was larger (n = 95, compared to n = 8). This mutation has also been confirmed in populations of An. stephensi in Middle East and South Asia, with both kdr-west (L1014F) and kdr-east (L1014S) being described19,30,31, 40–43.
In this study we also identified the A296S mutations in the GABA gene (rdl). This mutation has not been reported previously in An. stephensi, probably as this locus has not been screened in this mosquito species. Dieldrin, alongside other cyclodienes that target the GABA receptor, is no longer officially used due to its environmental toxicity and possible impact on human health. However, the A296S mutation may confer resistance to other insecticides, such as fipronil; and broflanilide, a newly discovered GABA-gated chloride channel allosteric modulator under evaluation as an IRS tool44,45 44,45, so it is important to document its presence. We targeted two other mutations in the rdl gene (V327I, T345S), which are often seen in tandem with the A296S SNP38,46,47. However, we found no evidence of these non-synonymous SNPs in this population. These secondary SNPs have been described in both Asian and African populations of Anopheles mosquitoes, but never without the presence of the A296S mutation38,46,47. Further, no other commonly seen mutations, such as the ace1 G119S, were identified in our analysis. However, two new mutations were detected, N177D in the ace1 gene present also in the insecticide susceptible colony, suggesting that this mutation may not be involved in insecticide resistance, and V189L in the GSTe2 gene detected only in 3 samples.
The genetic analysis using data from its2 and cox1 genes revealed a low level of genetic diversity, which is consistent with studies of invasive An. stephensi populations in Sri Lanka and India, where few variants and minimal genetic differentiation in these loci was observed 13, 48–50. For its2, the Ethiopian samples cluster together and separate from the An. stephensi from other countries. The Ethiopian samples were all collected in the same village, likely resulting in the clustering pattern observed. More haplotypes were observed for the cox1 gene, with a core haplotype shared by all samples, except those sourced from Saudi Arabia. Ethiopian samples shared haplotypes with samples of Pakistan origin, as previously reported51, but also shared haplotypes with samples from neighbouring countries (Sudan, Djibouti), supporting evidence of the spread of this vector across the Horn of Africa. It is possible that this vector is already present in other regions in Africa, as recently reported in Nigeria (year 2020). Strengthening the surveillance activities of this species, through large-scale genetic characterisation of this vector, is urgently needed to rapidly implement targeted control strategies.
Overall, the cost-effective and high throughput Amp-seq approach presented here, can be implemented by vector control programs, to screen An. stephensi mosquitoes for both known and putative novel insecticide resistance mutations. A limitation of our approach is that it currently only targets loci that are established mechanisms of insecticide resistance. However, it is easy to add other targets to the Amp-seq method as new loci are detected. This approach can be used to complement diagnostic bioassays, by providing genotyping data alongside phenotypic information, to detect new genetic variants underlying insecticide resistance. Our approach is highly flexible, with the easy addition of genomic regions of interest to the panel, for example, to include loci related to metabolic resistance or that can target pathogens such as malaria parasites, thereby enhancing surveillance activities.