Identifying potential drivers of haplotype frequency increase
The genomic region into which the ZZB-TE inserted does not show histone signals of regulatory activation (H3K27ac, H3K9ac) or repression (H3K9me3), and ATAC-seq data suggests it is not in an open chromatin region [35]. However we took two in silico approaches to determine whether the ZZB-TE insertion (748bp) carried putative regulatory variants. The inserted region had 98.5% sequence identity to a putative enhancer (2R:45966598-45966822) identified by homology withDrosophila melanogaster [35]. However, despite the similarity, the insertion lacked enhancer-like combination of chromatin marks identified in [35] and its potential regulatory role in nearby genes is unclear. The second approach involved screening the ZZB-TE inserted sequence for putative enhancers using iEnhancer-2L[36] and iEnhancer-EL [37]. In a windowed analysis of 200bp with a 1bp step across the entire length of ZZB-TE both predicted that some of the windows would have strong enhancer activity, however the windows were not-concordant, precluding further analysis.
Given that cytochrome P450 mediated resistance is commonly associated with differential gene expression we performed transcription studies within the Cyp6aa/Cyp6p cluster between the most contrasting haplotypes, wild-type and triple mutant, present in the BusiaUg colony. The group homozygous for the triple mutant haplotype significantly overexpressed both Cyp6aa1 (2.23-fold, 95% CI: 1.73-2.90, P=0.0003) and Cyp6p4 (2.57-fold, 95% CI 1.25-5.93, P=0.039) compared to wild-type individuals. The ratio of expression ofCyp6aa 1 broadly reflected the expected pattern based on genotype (ie 2:1.5:1 for triple mutant homozygotes: heterozygotes: wild type genotypes, respectively. Figure 4). As a control we examined a neighbouring, very commonly resistance-associated gene, Cyp6p3 , but triple mutant and wild type homozygotes did not differ significantly in expression (1.33 fold, 95% CI 0.64-2.74, P>0.05).
To investigate whether resistance may be driven at least in part by an effect of the allelic variant on metabolic activity of CYP6P4, we expressed the wild-type (236I) and mutant (236M) forms in an E. coli based recombinant protein system (Supplementary materials Appendices 2 and 4). Both alleles were shown to be capable of metabolizing class I (permethrin) and II (deltamethrin) pyrethroids but there was no evidence that the mutant (236M) or wildtype (236I) alleles had different rates of pyrethroid depletion. We also expressed the duplicated P450 CYP6AA1, in an Sf9 -baculovirus protein expression system. Again, metabolism assays demonstrated that the enzyme was capable of metabolizing both deltamethrin and permethrin (Supplementary materials Appendices 3 and 4). Depletion of deltamethrin was 36.6% greater (SE= 3.79) in the presence of NADPH than in the control (t-test: t=-9.67; d.f. = 8; P= 9.6 x10-6), demonstrating that CYP6AA1/CPR is capable of metabolizing deltamethrin in vitro . Similarly, permethrin was metabolised by CYP6AA1/CPR, with permethrin being depleted by 22.4% (SE = 0.63) compared to the control without NADPH (t-test: t= -31.08; d.f. = 14; P= 2.55 x10-14).
Given clear evidence of increased expression of both Cyp6aa1 andCyp6p4 in the triple mutant haplotype and the ability of both enzymes to metabolise pyrethroids in vitro , we investigated whether the mutations were significantly associated with resistancein vivo. Exposure of An. gambiae females from Busia, Uganda and Nord Ubangi, DRC to new LLINs in cone assays resulted in negligible mortality to the pyrethroid only LLINs, Olyset and Permanent 2.0 (Figure 5). Simultaneous exposure to pyrethroid plus the P450 inhibitor PBO in Olyset + and the top of Permanent 3.0 nets resulted in a marked reduction in resistance, demonstrating that the resistance phenotype is substantially mediated by P450s. We performed laboratory backcrosses between additional mosquitoes from Busia with the pyrethroid susceptible Mbita colony, and found that the triple mutant haplotype was significantly associated with resistance to the most commonly used type II pyrethroids in LLINs: deltamethrin (Fisher’s exact test p=3.2x10-6) and alphacypermethrin (Fisher’s exact test p=5.9x10-7) resistance although not to permethrin (Fisher’s exact test p=0.06) nor, as a control, DDT (Fisher’s exact test p=0.84) (Table 1) in WHO tube assays. Similarly, specimens collected in 2016 from the DRC showed a strong association between the triple mutant genotype and survival rate 24 hours post-exposure to either 0.05% deltamethrin for 1 hour or 3-minute exposures to deltamethrin-treated sides of a new PermaNet 3.0 net (Table 2). No association was found in samples exposed to permethrin (24 hour WHO tube assay) or permethrin-treated Olyset Plus nets (3-minute WHO cone assay) (Table 2). Complete linkage of the three mutants in the BusiaUG colony and the DRC wild caught collections precludes determination of the relative contribution of each of the three mutations to the resistance phenotype but taken together these results demonstrate a strong impact of the triple mutant on the efficacy of pyrethroid resistance.