Candidate genes for Bti resistance were identified using a keyword search in the VectorBase database http://aaegypti.vectorbase.org/index.php. A total of 162 genes belonging to families of toxin receptors (14 alkaline phosphatases, 23 aminopeptidases, 17 cadherins and 20 alpha-amylases) and enzymes involved in toxin activation (six chymotrypsins and 82 trypsins) previously shown to be potentially involved in Bti resistance were considered as candidates. These candidate genes were dispersed over 94 supercontigs in the Ae. aegypti genome.
The three aminopeptidases (VectorBase Gene ID AAEL007892, AAEL004738 and AAEL008155) identified in a region showing a signature of selection were sequenced for at least nine susceptible and LiTOX individuals. We also sequenced as a negative control three candidate genes described in literature as binding Cry toxins (the aminopeptidase AAEL012778, the cadherin AAEL007488 and the alkaline phosphatase AAEL009077; Chen et al. 2009; Likitvivatanavong et al. 2011b) but located in genomic regions with low Fst between the susceptible and resistant strains, and the aminopeptidase AAEL004226 located more than 1 million bp from an outlier. The complete sequence for each candidate gene was downloaded from VectorBase, and primers amplifying about 1000 bp (range 669–1036) were designed in each gene using the software package Lasergene 7.2 (DNASTAR Inc., Madison, WI, USA) (Table S2). DNA amplification was performed in 25 μL total volume with 2 mm MgCl2, 0.1 mm of each dNTP (Roche Diagnostics, Basel, Switzerland), 0.2 μm of each primer, 5 μg BSA, 0.6 U AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA, USA) and 50 ng DNA. The PCR program was initial 10 min denaturation step at 95°C; 40 cycles of denaturation at 95°C for 45 s, annealing at 58°C for 45 s and elongation at 72°C for 60 s; and a final extension step at 72°C for 5 min. Sequences performed by Genoscreen (www.genoscreen.fr) were aligned and corrected using the program Bioedit. Haplotype phase was inferred, and genetic diversity and differentiation analyses (nucleotide diversity p, haplotype diversity Hd, number of segregating sites S, number of singletons Si, Watterson's mutation parameter θW, average number of nucleotide differences K within and between strains) were performed using the software dnasp 5.0 (Librado and Rozas 2009). Deviation from neutral equilibrium expectations was tested by applying Tajima's D (Tajima 1989), Fu and Li's D* and F* (Fu and Li 1993) tests. To assess whether these statistics significantly departed from a neutral scenario of evolution given the resistant strain's known demographic history, we performed coalescent simulations using the ms program (Hudson 2002). This program generates random independent samples according to a Wright–Fisher neutral model allowing for population size changes at each generation. For each gene, the mutation rate μ was estimated from the per-locus mutation parameter observed for the susceptible strain (θ = 4Neμ and Ne = 6000) and used as the starting value for the simulations. Then, 1000 neutral samples consisting of 20–24 haplotypes, depending on the number of individuals sequenced in the LiTOX strain, were simulated based on the resistant strain's known demographic history (Bonin et al. 2009, Table S1).