Two M. usitatus Bagnall strains (one acetamiprid-resistant (AcR) and one susceptible (SS)) were used in this study. The SS strain was collected from Nanbin Farm, Sanya City, Hainan Province, in 2008 and was reared for 60 generations in the laboratory without exposure to any insecticide. The AcR strain was established from the SS population through continuous exposure to acetamiprid for 40 generations, using the leaf dip bioassay method (Rueda and Shelton, 2003). Both strains were reared on fresh cowpea pods in the laboratory at 26°C ± 1°C, under a photoperiod of 14:10 h (light:dark).

The median lethal concentration was determined using the leaf tube film method. Acetamiprid (25% water dispersible granules) was used in a commercially available formulation. Five graded concentrations of acetamiprid were used. Cowpea pods were dipped for 15 s in the designated concentration of insecticide or distilled water (the latter as a control) and placed in the shade until air dried. Emerging adult insects were transferred to the beans. Bioassays were performed in the laboratory at 26°C ± 1°C, under a photoperiod of 12:12 h (light:dark). Each concentration was performed in triplicate, and mortality was assessed after 2 days. LC₅₀ (lethal concentration 50%) values were calculated via probit analysis using the SPSS software package (LeOra Software Inc., Berkeley, CA, United States). Resistance factor (RF) was estimated at LC₅₀ as RR = LC₅₀ of the AcR strain/LC₅₀ of the SS strain; the 95% CI for RR was calculated following Preisler and Robertson (1989).

A total of 500 adult females of both the susceptible and resistant strains were collected for three biological replicates. Sample processing, extraction, and metabolite detection for transcriptome analysis were performed by Suzhou PANOMIX Biomedical Tech Co., Ltd. (Suzhou, China), following standard procedures, and the fragments per kilobase of transcript per million mapped reads (FPKM) of the assembled transcripts was calculated. Transcript expression abundance was calculated using the FPKM method (Mortazavi et al., 2008). Benjamini–Hochberg correction of the p-value for multiple tests was applied using the false discovery rate (FDR). FDR ≤0.001 and absolute value of the log2 ratio ≥2 were the thresholds for determining significance of differences in gene expression (Itai et al., 2001). Given the large number of P450s in the M. usitatus transcriptome, we first characterized gene expression using real-time quantitative polymerase chain reaction (RT-qPCR). Our goal was to identify genes with consistent differential expression.

The SteadyPure Agarose Gel DNA Purification Kit (Accurate Biotechnology, China) was used to purify and recycle the cloned MusiDN2722 PCR products. The primers used are listed in Supplementary Table S1. The following procedure was adopted to combine the vector components in a 5-μL reaction system. First, the reaction mixture was blended gently and placed in a PCR instrument at 25°C for 5 min. A measure of 5 ml of the transformation product was centrifuged for 1 min, and the entire bacterial solution was subsequently used to cover the Luria–Bertani solid culture. Next, 100 µl of the PCR product was mixed with Trans1-T1 phage-resistant chemically competent cells and aseptically coated on a Luria–Bertani (LB)/ampicillin plate. After growth, preparation, and plasmid DNA analysis, the plate was incubated at 37°C for 12 h; subsequently, a colony was harvested. The colony was cultured in 5 mL of LB liquid medium with ampicillin; the cells were cultured overnight at 37°C. Thereafter, white monoclonals were collected in 10 μL of sterile water and vortexed. A measure of 1 µl of the mixture was mixed with 20 μL of PCR mixture, and positive clones were identified with M13 forward and M13 reverse primers. After these steps, the recombinant plasmid from the positive bacterial fluid was sequenced by Shanghai Sangon Biological Company to determine the clone. The GenBank nucleotide sequence was subtyped and homologously analyzed.

Total RNA was extracted from 300 adults and nymphs of the AcR and SS strains of M. usitatus using the Total RNA TRIzol Extractor (Sangon Biotech, China, Shanghai). cDNA was constructed from the total RNA using the PrimeScript™ RT Reagent Kit with gDNA Eraser (TaKaRa, Tokyo, Japan). RT-qPCR was performed in CFX96 TOUCH (Bio-Rad) using TB Green^® Premix Ex Taq™ II (Tli RNaseH Plus; TaKaRa, Japan). Gene-specific primers were designed using Premier 5.0 and synthesized by Sangon Biotech Co., Ltd. (Shanghai, China). Experiments were performed thrice with different RNA preparations for each strain. The following cycling conditions were used: 95°C for 30 s; 40 cycles of 95°C for 5 s, 60°C for 30 s, 60°C for 30 s; and 95°C for 1 s for plate reading. Reaction fluorescence was continuously monitored after the cycling protocol using the dissociation temperature of the PCR products at a temperature transition rate of 0.1°C/s to generate a melting curve. Relative gene expression was calculated using the 2^−ΔΔCT method (Pfaffl, 2001). The RT-qPCR product was resolved via 1.0% agarose gel electrophoresis, and a DNA fragment of approximately 500 bp was obtained. These results indicated that P450 was expressed in both SS and AcR strains of M. usitatus.

The sequencing results were submitted to the NCBI (National Center for Biotechnology Information), and the target gene sequences were predicted using the open reading frame (ORF) and conserved domain. The amino acid sequence of the protein encoded by MusiDN2722 was predicted and analyzed using bioinformatics software applications. The physicochemical properties of the target protein were predicted using ProtParam software (SIB, Swiss Institute of Bioinformatics). The transmembrane region of the target protein was predicted using the TMHMM-2.0 online tool (Department of Health Technology). The signal peptide of the target protein was predicted using the Signal P5.0 server. Finally, the phylogenetic relationships of the target proteins were predicted using MEGA 7.

Specific primers for dsRNA synthesis were designed based on the cDNA sequence of MusiDN2722 and the fragment sequence of a green fluorescent protein (GFP) containing T7 polymerase promoter sequences at both ends. Approximately 50 newly emerged 3-day-old adult females were collected and placed in a specially designed device. Ten biological replicates were set up. Two approaches were employed for RNA interference: membrane feeding and microinjection of dsRNA. In the membrane-feeding approach, the mouth of a tube was covered with a thin film of BuddyTape (Aglis, Japan). Thereafter, 30 μL of dsRNA was added to the membrane and sealed with Parafilm (Sangon Biotech). After 2 consecutive days of feeding, live adults were collected and half of them were subjected to fluorescence qPCR to verify the silencing efficiency. The remaining half of the live adults were treated with acetamiprid at LC₅₀ for bioassays. In the microinjection approach, dsRNA was injected using a microinjector between the mesothoracic shield plate and abdominal segment shield plate into female adult worms.

All results are expressed in the form mean ± standard error; IBM SPSS Statistics 20 software (SPSS, Chicago, IL, United States) was used for statistical analyses. A Student's t-test or one-way analysis of variance (ANOVA) was used to compare the differences between samples or among multiple samples in RT-qPCR and bioassays, respectively. Differences were considered significant at p > 0.05.

After 40 generations of selection, the screening acetamiprid concentration was determined to be 4,700 mg/L, and the LC₅₀ values of acetamiprid for the SS and AcR strains were 85.676 and 1439.425 mg/L, respectively. The AcR strain developed a moderate level of resistance to acetamiprid, with a 16.78-fold resistance ratio.

The RT-qPCR analysis showed that the expression of MusiDN2722 was 4.17 times higher in the AcR strain than in the SS strain (Figures 1, 2). Transcriptomic sequencing analysis of the susceptible and resistant strains (SS and AcR) showed that there were more than 40 million original sequences. The Trinity software tool was used to concatenate the filtered sequences, and 21,740 unigenes were obtained (Supplementary Figure S1); the total length was 38343948 bp. The results of differential expression analysis based on the FPKM values showed that there were 167 upregulated and 139 downregulated genes in AcR strains compared with those in the SS strain (Supplementary Figure S2). The results of GO (Gene Ontology) term enrichment analysis indicated enrichment of genes under the following terms: binding, catalytic activity, cellular process, metabolic process, single-organism process, cell, cell junction, and organelle (Supplementary Figure S3). The results of KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis indicated enrichment of genes under the following terms: carbohydrate metabolism, translation, folding, sorting and degradation, signal transduction, transport and catabolism, and the endocrine system (Supplementary Figure S4). The P450 superfamily genes were identified in the transcriptome analysis of the two strains of M. usitatus. We list the top 10 CYP450 genes with a fold difference greater than 2 in Tables 1 and 2. We selected a gene of interest (TRINITY_DN2722_c0_g1) for full characterization in this study; this gene was named MusiDN2722.

RT-qPCR analysis of M. usitatus at different developmental stages indicated the expression of MusiDN2722 in different instars. Expression was significantly higher during instars from the third larval stage to the adult stages than during the 1st and 2nd instar stages, and peaked during the 4th instar and adult stages. Expression was 1.67–3.23-fold higher at the 4th instar and female adult stages than at the other stages (Figures 5, 6A, 7). Expression in the head was used as a baseline (relative gene expression = 1) for comparison of gene expression in different tissues. The results of a relative gene expression test using RT-qPCR revealed that MusiDN2722 was predominantly expressed in the thorax and abdomen, with levels 16.73- and 12.57-fold higher than the level in the head (Figure 6B).

After 24 h, insects administered dsMusiDN2722, dsGFP, and 10% hydromel exhibited survival rates of 87.70%, 89.49%, and 88.54%, respectively. After 48 h, the survival rates were 78.53%, 81.19%, and 79.73%, respectively (Figure 7A1). At 24 and 48 h after microinjection for RNA interference, the survival rate of insects administered air (no injection solution), PBS, dsGFP, and dsMusiDN2722 was assessed. As shown in Figure 6B, after 24 h, the survival rates of these insects were 80.00%, 81.48%, 68.00%, and 62.07%, respectively; after 48 h, the survival rates were 52.00%, 51.85%, 40.00%, and 41.38%, respectively. Analysis using SPSS revealed no significant difference between insects administered air and PBS or between those administered dsGFP and dsMusiDN2722, but the survival rate of the latter was considerably lower than that of the former (Figure 7B1).

The control group was fed and injected with dsGFP. RT-qPCR analysis showed that MusiDN2722 expression in the membrane-fed insects, compared with that occurring in the control group fed dsGFP, was 21.76% and 39.21% after 24 and 48 h, respectively (Figure 7A2). This finding suggests that interference via the membrane-feeding method was effective to some degree. dsMusiDN2722 was microinjected into the cavity of M. usitatus. At 24 and 48 h after the injection, gene expression was 29.35% and 43.20%, respectively, of that occurring in the control group, and the difference between the groups was significant (p < 0.05) (Figure 7B2).

The knockdown of MusiDN2722 substantially increased the mortality of adults of the ACR strain relative to that of control adults upon exposure to 1878.99 mg/L acetamiprid. Toxicity bioassay showed that membrane-fed M. usitatus individuals were more susceptible to acetamiprid than the controls at 24 and 48 h (Figure 8). As shown in the figure, the mortality rate of the dsGFP group at 24 and 48 h was 30.38% and 52.31%, respectively, and that of the dsMusiDN2722 group at 24 and 48 h was 46.98% and 68.29%, respectively. Compared with the dsGFP group, mortality increased by 23.4%.

Based on the results of prediction analysis using the STRING database, the corresponding genes were selected from the transcriptome for expression analysis. After silencing of MusiDN2722, there was a significant decrease in gene expression. CYP18A1, CYP12A4, and CYP4G15 correspond to transcripts C47928.graph, C51472.graph, and C50145.graph, respectively. According to the qPCR results, expression of C47928.graph and C51472.graph significantly decreased after silencing of MusiDN2722. C50145.graph exhibited no difference in expression before and after treatment (Figure 9).