Trypsin-activated Cry1F toxin was purchased from Envirologix (Portland, OR, USA).

Susceptible (ACB-BtS) and Cry1F-resistant (ACB-FR) strains of ACB were described previously (8) and were reared on an artificial diet as reported elsewhere (26). Susceptibility estimates for Cry1F toxin (LC₅₀ value, Cry1F concentration killing 50% of ACB larvae) were determined from 7-day diet incorporated bioassays with ACB neonates as 1.2 μg/g and 1,075 μg/g (toxin/diet) for the ACB-BtS and ACB-FR strains, respectively. The results indicated more than 903-fold resistance in ACB-FR.

Hemolymph was collected from 4^th instar ACB-BtS and ACB-FR larvae after exposure to their respective Cry1F LC₅₀ values (1.2 μg/g for ACB-BtS and 1,075 μg/g for ACB-FR) for 24 and 48 hours. Larvae exposed to an untreated diet were used as controls. Hemolymph was collected after surface sterilizing the ventral surface of the abdomen using 75% ethanol and cutting a proleg with fine scissors. About 10 μL of hemolymph, containing around 1.9 × 10⁶ hemocytes/mL as determined by counting in the hemocytometer chamber, were collected per larva. Hemolymph collected from 50 larvae was pooled as a biological replicate in 200 μL of ice-cold iso-osmotic Tris-buffered saline (ITBS, 50 mM Tris-HCl, 10 mM CaCl₂, 100 mM NaCl, 100 mM dextrose, 6 mM KCl, 5 mM MgCl₂, pH 7.2, 300 mOsm) to prevent hemocyte aggregation [27]. Three biological replicates were prepared and processed per ACB strain. The hemocyte suspensions were centrifuged (400 × g, 15 min, 4°C), and the resultant hemocyte pellets were used for RNA extraction and RNA-Seq analysis. The cell-free hemolymph in the supernatants was used for iTRAQ analysis.

Total RNA was extracted from hemocytes using TRIzol^® (Ambion, Life Technologies, CA, USA), following the manufacturer’s instructions. The RNA concentration and quality were evaluated using a NanoDrop 2000 spectrophotometer (Thermoscientific, MA, USA). Complementary DNA (cDNA) was synthesized from 700 ng of total RNA using the One-Step gDNA Removal and cDNA Synthesis SuperMix kit (TransGen Biotech Co., Ltd., Beijing, China), following the manufacturer’s protocol.

Libraries were prepared from 1 μg of total RNA per sample using the NEBNext^® Ultra™ RNA Library Prep Kit for Illumina (New England Biolabs, MA, USA), according to the manufacturer’s instructions. Briefly, mRNA was purified from total RNA using magnetic beads coated with poly-T oligo, and after fragmentation was used for cDNA synthesis and ligation of adaptors. cDNA library fragments of 250-300 bp in length were purified with the AMPure XP system (Beckman Coulter, MA, United States) and then used for PCR with Universal and Index (X) PCR primers. Finally, PCR products were purified (AMPure XP system), and library quality was assessed on the 2100 Bioanalyzer (Agilent, CA, USA). The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina, CA, USA), according to the manufacturer’s instructions. After cluster generation, the libraries were sequenced on a Novaseq 6000 platform at Novogene Co. Ltd. (Beijing, China) with a 150 bp paired-end sequencing configuration.

Raw reads were processed to clean reads through custom Perl scripts by removing adapter-related sequences (796,654, 3.45%), ploy-N (23,878, 0.10%) and low-quality reads (42,988, 0.19%). A publicly available ACB genome assembly (Version: ASM419383v1, GenBank accession: GCA_004193835.1) was downloaded and used as the reference genome for assembly reads. Paired-end clean reads were aligned to the reference genome using Hisat2 v2.0.5 (27). Differential expression analysis was performed using the DESeq2 R package v1.16.1 (28). The resulting P-values (P<0.05) were adjusted using Benjamini and Hochberg’s approach for controlling the false discovery rate. KEGG statistical (P<0.05) enrichment of differentially expressed genes was performed using the cluster Profiler v3.12 R package (29). The raw data from the three biological replicates is publicly available in the Sequence Read Archive (SRA) database as BioProject ID PRJNA723187.

Primers were designed to amplify selected target genes ( Table S1 , Supporting Information), and specificity was confirmed by amplification and sequencing of amplicons ( Figure S1 ; Supplementary file ). Original cDNA samples were diluted 10-fold, five serially diluted concentrations were subjected to qRT-PCR, and mean CT values from three replicates were used to construct the standard curve. The amplification efficiency (E value) of all primers was calculated using the formula: E = 10^(−1/slope) -1 (30). A two-step PCR approach was implemented using 20 μL final reaction volumes in the TB Green™ mix. Amplification by qRT-PCR was performed on an ABI 7500 Fast (Applied Biosystems, USA) thermocycler with TB Green™ Premix Ex Taq™ (TaKaRa, Japan). The amplification program was set as follows: 1) initial denaturation 30 s at 95°C, 2) 95°C for 5 s, and 3) 60°C for 30 s; steps 2 and 3 were repeated for 40 cycles. Ribosomal protein L8 was used as a reference gene (31), and the 2^-ΔΔCT method was used to calculate the difference in expression of selected genes (32). Three independent biological replicates were performed per treatment, and three technical replicates were performed per biological sample.

The GST activity was estimated in hemolymph from larvae of the ACB-BtS and ACB-FR strains after 24 and 48 hours of treatment with Cry1F (LC₅₀ dose) or a control diet. The protein concentrations in the samples were estimated using the Easy Protein quantitative kit (TransGen Biotech Co., Ltd., Beijing, China) and diluted to 1 mg/mL using the 1 × GST sample buffer provided in the kit. The activity of GST was measured using the Glutathione S-Transferase Assay Kit (Cayman Chemical, MI, USA), according to the manufacturer’s instructions. Briefly, 20 µL of hemocyte-free hemolymph was added to individual wells in a 96-well microplate, and then 150 µL of assay buffer and 20 µL of glutathione provided with the kit were added per well. Finally, 1-chloro-2,4-dinitrobenzene (CDNB) was added (10 µL), and the plate was incubated for 10 min at RT and then read at 340 nm to measure the conjugation of CDNB with reduced glutathione.

Isobaric tag for relative and absolute quantitation (iTRAQ) proteomic analysis was conducted as described elsewhere (24) on hemolymph proteins from ACB-BtS and ACB-FR larvae treated with Cry1F (LC₅₀ dosage). Briefly, the hemolymph protein samples were precipitated with acetone and digested using sequencing-grade modified trypsin. The digested peptides were labeled with iTRAQ reagents and analyzed by LC-MS/MS for identification and quantification. The hypergeometric distribution equation analyzed the significantly enriched proteins in the KEGG pathway (P< 0.05). Gene Ontology classification analysis was performed using OmicsBox v1.2.4 integrated with the Blast2GO function annotation module (33). The enrichment of the differentially expressed proteins was represented in the -LOG₁₀ P-value.

Proteins selected for PPI were macrophage mannose receptor 1-like (MMR1), C-type mannose receptor 2-like (C-MR2), CD209 antigen-like protein E (CD209), C-type lectin lectoxin-Phi1-like (C-lectoxin), hemolymph lipopolysaccharide-binding protein (LBP), and CD63 antigen-like (CD63). The predicted full-length amino acid sequences of activated Cry1F and the hemocyte proteins selected for further studies were subjected to structure homology modeling ( Supplementary file ) (Swiss-Model server, https://swissmodel.expasy.org/) (34). The TMHMM Server v2.0 (http://www.cbs.dtu.dk/services/TMHMM/) (35) was employed for transmembrane prediction analysis. The protein domain prediction was achieved by analyzing all the sequences using the Pfam or InterProScan database (http://pfam.xfam.org/) (36) (https://www.ebi.ac.uk/interpro/) (37). The DOG 2.0 software was used to illustrate the domains of selected hemocyte proteins and Cry1F (38). The in silico protein-protein interaction (PPI) was analyzed using carbohydrate-binding domains of C-type lectins and the extracellular domain (EC 2) of tetraspanin (CD63) with the central β-sheet domain of Cry1F. The solvent-accessible surface-exposed residues in the interacting domains were predicted by RaptorX-Property Prediction tool (http://raptorx.uchicago.edu/) (39). The domain-based PPIs were carried out using the High Ambiguity Driven protein-protein DOCKing approach (HADDOCK 2.4 server) (40). The HADDOCK 2.4 default settings were used, and no special constraints were input during the PPI studies. The UCSF ChimeraX v1.1 was used to visualize the PPI complex after the in silico analysis (41).

The full length cDNAs encoding selected C-type lectins and tetraspanin and including a N-6 His-Tag were synthesized using a PCR-based Accurate Synthesis (PAS) system. These synthesized protein-coding ORFs were constructed into the pCZN1 vector at EcoRv-XbaI sites and expressed in the ArcticExpress Competent Cells E. coli expression system (Agilent Technologies, CA, USA). Transformants were grown in liquid LB medium to OD 0.6-0.8 and then induced by the addition of 1 mM isopropyl-b-d-thiogalactoside (IPTG, Coollaber Science and Technology, Beijing, China) and placed in a shaking incubator at 37°C for 6 h. The induced competent cells were harvested by centrifugation (8,000 × g, 4°C, 15 min) and the pellets were resuspended in 20 mL of lysis solution (20 mM Tris-HCL, 1 mM phenylmethylsulfonyl fluoride, bacterial protease inhibitor cocktail, pH 8.0). The induced competent cells resuspension was subjected to ultra-sonication (400 W, operating 4 sec, paused 8 sec, total duration 20 min). The ultra-sonicated cell lysate solution was centrifuged (10,000 × g, 4°C, 15 min) and the pelleted inclusion bodies washed in washing solution (20 mM Tris-HCL, 1 mM EDTA, 2 M carbamide, 1 M NaCl, 1% Triton X-100, pH 8.0). These centrifugation and washing steps were performed a total of three times and then inclusion bodies in washing solution were solubilized in a 1:2 (v/v) ratio of lysis buffer (20 mM Tris-HCL, 5 mM DTT, 8 mM carbamide, pH 8.0) overnight at 4°C. After the overnight incubation, the His-tag protein precipitate was recovered by centrifugation (10,000 × g, 4°C, 15 min), resuspended and dialyzed overnight in Tris buffer (20 mM Tris-HCL, 0. 15 mM NaCl, pH 8.0).

The recombinant proteins were purified in a Ni-IDA-Sepharose Cl-6B affinity column coupled to a low-pressure chromatography system (BioLogic LP System with BioFrac Fraction Collector, Bio-Rad, CA, USA) at a flow rate of 1 mL/min, using 20 mM Tris-HCL with 250 mM imidazole (pH 7.4) for elution. Proteins in eluted fractions were resolved by 12% SDS-PAGE and transferred (100 V, 200 mA, 1 h) to a polyvinylidene fluoride (PVDF) filter. After transfer, the filter was washed four times in phosphate-buffered saline with Tween-20 (PBST) and then blocked for 1 hour in PBST plus 5% skim milk. The PVDF filter was then incubated in primary rabbit antisera against the His-tag (1:1,000 dilution) on a shaker overnight at 4°C. After washing in PBST, the filter was probed with sheep anti-rabbit secondary antibody conjugated to horseradish peroxidase (HRP, 1:10,000 dilution) at 37°C for 1 h. The filter was then washed once before developing using enhanced chemiluminescence (ECL Western Blotting Substrate, Pierce, USA) in an ImageQuant LAS 4000 (GE Healthcare, PA, USA) imager.

Purified MMR1, C-MR2, C-lectoxin, LBP and CD63 proteins (1 mg) to be used for SPR analysis were digested using 5U of SUMO Protease (Beijing Solarbio Science & Technology Co., Ltd, China) in PBS (pH 7.2) at 4°C overnight to remove the N-6xHis tag, as confirmed by Western blotting. Samples of purified MMR1, C-MR2, C-lectoxin, LBP and CD63 proteins non-digested with SUMO Protease were subjected to immobilized-metal affinity chromatography (IMAC) using a nickel Ni-NTA matrix (Beijing Solarbio Science & Technology Co., Ltd, China). The purified proteins containing the N-6xHis tag were used as bait proteins in the pull-down assays, while proteins purified without the His-tag were used in surface plasmon resonance analysis to avoid interference with PPI analyses.

Macromolecular interactions between Cry1F protein and C-type lectins or tetraspanin were analyzed on a Biacore™ 8K surface plasmon resonance system (GE Healthcare, IL, USA). The activated Cry1F activated toxin (0.5 mg/mL) was immobilized on a CM5 (GE Healthcare, IL, USA) sensor chip using sodium acetate (10 mM, pH 4.0) as immobilization buffer with a flow rate of 10 μL/min. Running and regeneration buffers were HEPES (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 0.005% Tween-20, pH 7.4) and 3M magnesium chloride (pH 7.4), respectively. The association (120 sec) and disassociation (280 sec) steps were maintained at the flow rate of 30 μL/min. The analytes (MMR1, C-MR2, C-lectoxin, LBP, CD209 and CD63 without His-Tag) were desalted using the running buffer before SPR analysis. To calculate the equilibrium dissociation constant, the analytes that could interact with Cry1F were used to perform multiple analyte concentration injections. The sensogram data sets were fit to 1:1 binding model to determine binding parameters.

Pull-down assays were performed using the Pierce™ Pull-Down PolyHis Protein : Protein Interaction Kit (Thermo Scientific, USA), according to manufacturer’s instructions. The MMR1, C-MR2, C-lectoxin, LBP, and CD63 proteins containing a N-6 His-Tag were immobilized on HisPur™ Cobalt Resin as bait proteins. Initially, 50 μL of HisPur™ Cobalt Resin was used to separately immobilize the MMR1, C-MR2, C-lectoxin, LBP, and CD63 proteins. About 300 μL of protein sample (0.4 mg/mL) were added to each of the columns containing cobalt resin, and the mixtures were incubated for 1 hour at 4°C on a rocking platform. After incubation, the columns were placed on a new collection tube and spun down (1,200 × g for 1 min at RT) to eliminate the flowthrough. The columns were subsequently washed with buffer and centrifuged five times to eliminate unbound bait proteins. After washing, 600 μL of activated Cry1F toxin (prey protein, 0.3 mg/mL) were added to each column, and after sealing the columns were incubated for 2 h at 4°C on a rocking platform. After incubation, the columns were placed on a new collection tube and spun down (1,200 × g for 1 min at RT) to collect and discard the flowthrough, The columns were washed five times to eliminate the residue of prey protein that did not interact with bait protein. The bait and prey column flowthroughs were later used for comparison by SDS-PAGE analysis. For elution of interacting prey and bait proteins, 200 μL of elution solution (290 mM imidazole) was added to the columns, which were then sealed and incubated for 20 min at RT on a rocking platform. The final eluate from each column was collected by centrifugation (1,200 × g for 1 min at RT) in a new collection tube. A non-treated control with cobalt resin without immobilized bait proteins was also run to identify and eliminate false positives caused by nonspecific binding of proteins to the HisPur™ cobalt resin. The final elution was analyzed in a 4-12% ExpressPlus™ PAGE gel (GenScript, USA) using 15 μL of eluate. Proteins were detected by staining with the FastBlue Protein stain (Coolaber, Beijing, China) and ProteoSilver silver staining kit (Sigma-Aldrich, MA, USA), according to the manufacturer’s instructions.

Activated Cry1F toxin at the LC₅₀ concentration (1.2 μg/g for ACB-BtS and 1,075 μg/g for ACB-FR) was incorporated in an artificial diet and used to feed 4^th instar susceptible and Cry1F-resistant ACB larvae. After 48 hours, the hemolymph was collected, and hemocytes isolated as described above. The pelleted hemocytes were washed three times with iso-osmotic tris buffered saline (ITBS, pH 7.2; 300 mOsm). The hemocyte lysate sample and MAPK family assays were prepared using the InstantOne™ MAPK Family Activation Multispecies ELISA Kit (Thermofisher Scientific, MA, USA), according to the manufacturer’s instructions. The kit consists of antibody cocktails for detecting three important types of protein kinase; extracellular signal-regulated kinases 1/2 (ERK 1/2), MAPK p38α and c-Jun N-terminal kinase 1/2 (JNK 1/2). The protein concentrations in the lysate samples were estimated using the Easy Protein quantitation kit (TransGen Biotech Co., Ltd., Beijing, China). Before performing the assays, the final concentration of cell lysate samples was set to 0.5 mg/mL. Three biological replications (hemolymph from 30 pooled larvae/replicate) were performed per sample.

We prepared double stranded RNA (dsRNA) targeting the CD63 gene (dsCD63) and orally delivered it to ACB-FR. Briefly, gene-specific primers ( Table S1 , Supporting Information) containing a T7 promoter on the 5’ end were designed using the Primer Premier 5.0 software (Premier Biosoft, Canada) and used to amplify CD63 from a plasmid containing the CD63 ORF (pCloneEZ) or the green fluorescent protein (GFP) ORF from vector pRTHSP70-GFP containing the GFP gene (GenBank accession number AB062168.1) Amplicons (572 bp for CD63 and 467 bp for GFP) were used as templates for in vitro dsRNA synthesis using the T7 RiboMAX Express RNAi System (Promega, Madison, WI, USA), following the manufacturer’s instructions.

Fourth instar ACB-FR larvae were starved for 6 hours and then fed a control diet or diet containing Cry1F at the LC₅₀ concentration and dsRNA targeting the GFP or CD63 genes (1 µL containing 700 ng per mg of diet). After the larvae ate the diet pieces, fresh artificial diet incorporated with toxin (without dsRNA) was added. Three independent replicates (24 larvae pooled/replicate) were maintained. After 8 hours, the second dose of dsGFP or dsCD63 was added to the diet in order to sustain the silencing effect. The hemolymph was collected from larvae after 24, 48, 72, 96 and 120 hours, and hemocytes recovered as described above. RNA was extracted from hemocytes collected at different time points, three biological replicates were used for each time point (hemolymph from 30 larvae pooled/replicate). The level of MAPK activity was assessed in the hemocytes of dsRNA treated and control groups after 48 hours, three biological replicates were performed for each treatment (hemolymph from 30 larvae pooled/replicate). Purified RNA was used to prepare cDNA using the cDNA Synthesis SuperMix (TransGen Biotech Co., Ltd., Beijing, China). This cDNA was used as a substrate in qPCR analyses using the TB Green™ Premix Ex Taq™ (TaKaRa, Japan) to detect transcript levels of the CD63 gene. Relative expression levels were estimated with the 2^-ΔΔCT method [32] using transcript levels for RPL8 (GenBank Acc. No. XM_028306632.1) as an internal reference gene. Control treatment with a diet not containing dsRNA or Cry1F toxin was used for normalization of CD63 gene expression.

Student’s t-test (P < 0.05) was used to analyze GST activity and qRT-PCR expression data. One-way ANOVA (P < 0.05) was used to test for significant differences among treatments for other experiments. Differences in MAPK levels among hemocytes of ACB-BtS and ACB-FR after exposure to Cry1F toxin were compared with their respective control groups. Percentage larval mortality data from the RNAi experiment was transformed using Arcsine square root and treatment means were analysed using One-way ANOVA (Tukey test LSD, P< 0.05). All analyses were performed using IBM SPSS v 20 (IBM Corporation, Armonk, NY).

Differentially expressed genes (DEGs) were identified in hemocytes from larvae of the ACB-BtS and ACB-FR strains after treatment for 48 h with their respective LC₅₀ concentrations of Cry1F when compared to expression in hemocytes from larvae treated with control diet for each strain ( Figure S2 ; Supplementary file ). Treatment of ACB-FR larvae with Cry1F induced up-regulation of 4,289 genes, while only 890 genes were up-regulated in larvae from the ACB-BtS strain after toxin treatment ( Figure S2 ; Supplementary file ). On the other hand, Cry1F treatment induced down-regulation of 4,532 genes in ACB-FR and 975 in the ACB-BtS strains ( Figure S2 ; Supplementary file ).

Overall, the KEGG pathway analysis detected enrichment for both ACB-BtS and ACB-FR in genes involved in direct/indirect control of cellular activities such as adhesion, migration, differentiation, proliferation, and apoptosis. In ACB-BtS, significant enrichment was detected for 70 genes involved in 5 different pathways, including Ribosome, ECM-receptor interaction, Drug metabolism - cytochrome P450, Drug metabolism - other enzymes and Metabolism of xenobiotics by cytochrome P450 ( Figure 1A ). In contrast, the 332 genes significantly enriched in ACB-FR were assigned in Ribosome, Spliceosome, Ribosome biogenesis in eukaryotes, Oxidative phosphorylation and Proteasome pathways ( Figure 1A ). Gene Ontology (GO) analyses identified 863 genes significantly enriched in ACB-BtS hemocytes after Cry1F treatment, while in ACB-FR that number increased to 3,163 genes. Interestingly, the relative proportion of enriched genes in each category (biological process, molecular function and cellular component) was significantly higher in ACB-FR strains ( Figure 1B ).

Based on Cry1F receptors being membrane proteins and the potential participation of the immune system in response to intoxication and potentially resistance, we constructed a heat map of up-regulated genes encoding cellular membrane proteins and proteins involved in detoxification and immune-related pathways ( Figure 2A ). The biggest differences in expression detected after feeding on Cry1F compared to control diet in both ACB-BtS and ACB-FR strains were observed for UDP-glucuronosyltransferase 2B15 (2.54- and 2.85-fold in ACB-BtS and ACB-FR, respectively), and cadherin 89D (2.19- and 2.47-fold in ACB-BtS and ACB-FR, respectively). Genes that were highly up-regulated in ACB-FR but not in ACB-BtS included gluthathione S-transferase (2.45- versus 0.38- fold in ACB-BtS), cytochrome P450 4C1-like (4.2- versus 0.83-fold in ACB-BtS), and CD209 antigen-like protein E (3.96-fold in ACB-FR and not detected among DEGs in ACB-BtS). Other DEGs unique to ACB-FR were CD63 antigen-like, lysozyme-like and C-type lectin lectoxin-Phi1-like ( Figure 2A ).

Differential expression of selected genes involved in detoxification and immune pathways between ACB-BtS and ACB-FR was tested using qRT-PCR ( Figure 2B ). Compared to ACB-BtS, treatment with Cry1F induced significantly higher expression in ACB-FR hemocytes of GST (3.68-fold), C-type mannose receptor 2-like (2.38-fold), lipopolysaccharide-binding protein-like (1.98-fold), beta-1,3-glucan-binding protein-like (2.62-fold) and phenoloxidase-activating enzyme-like (1.09-fold). On the other hand, cytochrome P450 4C1-like, cytochrome P450 306a variant X2, CD209 antigen-like protein E and caspase 1-like gene had significantly higher expression in ACB-FR in RNAseq but had lower transcript levels when compared to ACB-BtS in qRT-PCR ( Figure 2B ).

Based on the consistent high GST expression difference detected, we performed GST enzymatic assays comparing hemolymph from ACB-BtS and ACB-FR larvae. Treatment with Cry1F for 24 and 48 hours increased GST activity in hemolymph from larvae of both strains ( Figure 3 ). In agreement with RNAseq and qRT-PCR results, GST activity was clearly higher in hemolymph from ABC-FR compared to ACB-BtS. This difference in GST activity between the strains was statistically significant after 24 but not after 48 hours of treatment with Cry1F ( Figure 3 ).

Analysis by iTRAQ identified differentially abundant proteins (DAPs) in the hemolymph of ACB larvae after Cry1F treatment compared to controls ( Figure 4A ). Comparatively, a much higher number of DAPs with increased abundance were detected in the hemolymph of ACB-FR (111 proteins) compared to ACB-BtS (9 proteins). Among DAPs involved in immune pathways, all but one had a higher abundance in hemolymph from ACB-FR compared to ACB-BtS larvae ( Figure 4B ). Importantly, DAPs included DEGs such as macrophage mannose receptor 1-like (MMR1), C-type mannose receptor 2-like (C-MR2), beta-1 3-glucan-binding-like, lysozyme-like and phenoloxidase-activating factor 2-like ( Figure 2A ). Cathepsin L was the most significantly enriched protein in hemolymph of both strains and was identified as involved in phagosome/autophagy pathways ( Figure 4B ). Gene ontology (GO) annotation of iTRAQ data showed that ACB-BtS biological processes were related to metabolism, while ACB-FR were related to antibacterial defense mechanisms ( Figure 4C ). Carbohydrate-binding or recognition protein was a molecular function commonly enriched in both ACB-BtS and ACB-FR ( Figure 4C ).

Based on commonalities in results from RNA-Seq, qRT-PCR and iTRAQ analyses, DEGs and DAPs involved in defense response to bacteria, carbohydrate-binding, and a transmembrane protein involved in cellular signaling were selected for in silico protein-protein interaction (PPI) studies. Selected proteins included members of the tetraspanin superfamily (CD63 antigen-like), a hemolymph lipopolysaccharide-binding protein (LBP), the CD209 antigen-like protein E (CD209), and C-type lectins macrophage mannose receptor 1-like (MMR1), mannose receptor 2-like (C-MR2), and lectoxin-Phi1-like (C-lectoxin). All the proteins were annotated in UniProtKB and QuickGo databases with their putative functions ( Table 1 ).

Based on the Cry1F protoxin being cleaved by midgut digestive enzymes to the activated toxin form and domain II in the activated toxin core being responsible for binding specificity, we first determined the activated Cry1F toxin structure for in silico protein-protein interaction (PPI) studies in vitro ( Figures S3 , S4 ; Supplementary file ). None of the selected C-type lectin proteins contained transmembrane regions, while CD63 had four transmembrane regions and two extracellular domains ( Figures S5 , S6 ; Supplementary file ). The carbohydrate-binding domains of C-type lectins (MMR1, C-MR2, CD209, and C-lectoxin), LBP and the longest extracellular domain in CD63 (EC 2) were used for in silico prediction of binding to domain II of Cry1F (IPR001178) ( Figure 5A ). Interactions between these proteins and Cry1F were successfully predicted as PPI models ( Figure 5B ). The HADDOCK Score and root mean square deviation (RMSD) of the protein complexes were assessed using the top first cluster ( Table 2 ). The Cry1F-MMR1 PPI had the highest HADDOCK score (-84.1 ± 7.6), while Cry1F interacting with the CD63 E2 domain had the lowest (-29.7 ± 4.8). Overall, the interaction parameters of Cry1F domain II with the tested hemocyte proteins and obtained HADDOCK scores were shown in Table 2 .

C-type lectins (MMR1, C-MR2, CD209, C-lectoxin and LBP) and tetraspanin (CD63) were successfully expressed and purified in a heterologous system ( Figure S7 ; Supplementary file ) for in vitro PPI by surface plasmon resonance (SPR) analysis. Results from SPR experiments supported interactions between activated Cry1F toxin and MMR1, C-MR2, C-lectoxin, LBP or CD63 ( Table 3 ). Multiple analyte injections and corresponding association and disassociation curves ( Figure S8 ; Supplementary file ) were used to estimate binding parameters. No interaction was detected between Cry1F and CD209 ( Figure S9A ; Supplementary file ), and this protein was excluded from further experimentation. The lowest equilibrium disassociation constant (22 nM) was observed for Cry1F-C-MR2 interactions, whereas Cry1F-CD63 had the highest equilibrium disassociation constant (294.4 nM) ( Table 3 ). In contrast, MMR1 showed very low response units, preventing estimation of binding parameters.

As a second in vitro method to test for PPI, we performed pull-down assays with activated Cry1F as a prey protein and C-type lectins and tetraspanin as immobilized bait proteins ( Figure 6 ). Eluted samples contained Cry1F independently of the bait protein used ( Figure 6 , lanes 3), denoting interactions between the toxin and the immobilized hemocyte proteins. Despite the low binding between MMR1 and Cry1F detected in SPR analyses, pull-down assays detected interactions between these proteins ( Figure 6 ). In agreement with SPR results, the pull-down assay did not detect interactions between Cry1F and CD209 ( Figure S9B ; Supplementary file ).

The activity of MAPKs was measured in hemocyte lysates after 48 h of larval treatment with a control or Cry1F diet. Exposure to the Cry1F diet increased ERK1/2, p38α, and JNK1/2 activity in hemocytes from larvae of ACB-BtS and ACB-FR strains ( Figure 7 ). In control treatments, activity of tested MAPKs was relatively lower in ACB-BtS compared to ACB-FR hemocytes. In ACB-BtS, the activity of both ERK1/2 and JNK1/2 was significantly increased after Cry1F exposure, while p38α activity was not significantly changed compared to control ( Figure 7 ). In contrast, treatment with Cry1F significantly increased p38α and JNK 1/2 activities in hemocytes from ACB-FR, while ERK1/2 activity was not significantly changed ( Figure 7 ).

Based on previous reports of linkage between resistance to Cry toxins and a tetraspanin [20], we chose CD63 for functional tests as a putative Cry1F receptor. Treatment with dsCD63 significantly silenced CD63 gene expression after 24 hours, and the effect lasted until 96 hours ( Figures 8A, B ). In contrast, control treatments with dsGFP or water had no effect on CD63 expression. Bioassays with dsRNA-treated ACB-FR larvae and a sublethal dose of Cry1F detected a significant increase in susceptibility to the toxin in larvae treated with dsCD63 compared to controls ( Figure 8C ).

Hemocytes were collected from larvae treated with dsRNA and used to detect MAPK-ERK 1/2, p38α and JNK 1/2 activity. A significant increase in activity was detected for p38α and ERK 1/2 for all treatments when compared to control ( Figure 8D ). The most drastic difference unique to treatment with dsCD63 and Cry1F was a significant reduction in JNK 1/2 activity levels ( Figure 8D ).