Discussion
Insecticide resistance monitoring is the key to sustain insecticide-mediated control efficiency. Molecular detecting assays can be used to detect resistant markers accurately at early stages to avoid resistance evolution (Network, 2016). Target-site resistance, which is mainly caused by target insensitive mutations, and metabolic resistance, which is mainly caused by overexpressed detoxification genes, are the two main mechanisms of insecticide resistance (Ffrench-Constant, 2013). The detection of these two kinds of resistance can well reveal the mechanism of resistance of insect pest populations. PCR-based target-site mutation detection assays rely on genotyping individuals one by one within an insect pest population and are not only time-consuming, but also result in a high false-positive rate (Hirayama et al., 2010; Blais et al., 2015). The DNA microarray which used to detect differentially expressed detoxification genes are inefficient and complex, because of the demand for prerequisite knowledge of the reference sequences, low resolution of expression level, and background signals (Kogenaru, Qing, Guo, & Wang, 2012; Mantione et al., 2014). RNA-Seq sequences the transcription products of pooled samples of insect pest populations and can obtain the SNP information in gene expressed regions as well as provide gene expression level comparison (De Wit et al., 2015). More and more researchers have adopted RNA-Seq as a method to study resistance mechanisms and detect resistant markers (David et al., 2014; Faucon et al., 2017; Mamidala et al., 2012).
Here, we developed FastD to detect the target insensitive mutations and overexpressed detoxification genes. By collecting insensitive mutations on four kinds of insecticide targets and resistance-associated gene sequences of 82 insect species, the FastD program can be applied to detect resistant markers of a wide-range of species. The webserver of the FastD program uses SAM files as input and can analyze the samples more quickly than traditional methods such as PCR or microarrays. With these characteristics, FastD program offers a wide range of applications and great value.
As a proof of concept, FastD program was used to detect the resistance-associated markers of two insects, P. xylostella andA. gossypii . The resistance of insect populations can be well estimated by these resistant markers via FastD program. The RyRmutation G4946E and CYP6BG1 gene overexpression have also been reported to be associated with resistance to chlorantraniliprole (Guo, Liang, Zhou, & Gao, 2014; X. Li et al., 2018). Interestingly, The resistance level of CHR population with higher G4946E frequency (94.55%) is higher than ZZ population with lower G4946E frequency (66.1%) and six overexpressed detoxification genes. We speculated that G4946E may play a dominant role in resistance or there are other mechanisms conferring resistance in these resistant populations. In addition, 40 resistant allele reads among 575 all allele reads were detected in susceptible CHS population. We speculated that there may be few resistant individuals in CHS population. The discrepancy need further investigation. The nAChR beta1 subunit mutation R81T (Koichi Hirata et al., 2015) and overexpression of CYP6CY22 andCYP6CY13 genes have been reported to be associated with resistance to neonicotinoids. Moreover, four genes overexpressed in theP. xylostella ZZ population and seven genes overexpressed in theA. gossypii KR population which were not reported before are worth further study, indicating the value of FastD as a tool for both confirmation of resistance and discovery of new resistance mechanisms.
As a tool to detect resistant markers to monitor the emergence and development of insecticide resistance from RNA-Seq data, there are still some limitations. We plan to improve the following areas in the future. First, insecticide resistance with the polygene inheritance model is also associated with other important mechanisms, especially the detoxification gene amplification. Due to the limitation of RNA-Seq technique, gene amplification can’t be identified by FastD-MR. We plan to add new function to identify gene amplification based on genome resequencing data. Second, the accuracy of mutation frequency calculated by FastD-TR is limited by the fact that RNA-Seq reads from pooled sample have potentially different levels of contribution from each insect sample and allele. Therefore, we recommend users to use larger number of individuals sampled in pool to get more accurate result. Third, the resistance level is determined empirically based on detected resistant markers by the FastD program. More quantitative relationships between the resistant markers and resistance are critical and could be established with machine learning methods. Fourth, aside from insecticide resistance, resistance in other pests (herbicide resistance and fungicide resistance) are also associated with target insensitive mutations and overexpressed detoxification genes (Bohnert et al., 2019; Q. Li et al., 2013). Estimating the resistance to herbicide and fungicide will be added in the next version of FastD program.