The plasmids pOST1-NubI, pTT5, pPR3N, pET28a, pGEX-6P-1, pTSU2-APP, and pBT3STE were procured from the Laboratory Animal Center of Jinzhou Medical University. DWV VP2, purified DWV, NMY32 yeast, and Apis cerana larvae yeast cDNA libraries were prepared in our laboratory. Escherichia coli DH5α and BL21 strains were purchased from TransGen Biotech (Beijing, China). The transfection reagent Lipofectamine 2000 was purchased from GeneCopoeia (Rockville, MD, United States). GST-tagged protein purification kit was purchased from Beyotime (Shanghai, China). Anti-his tag mouse monoclonal antibody was purchased from Solarbio (Beijing, China). The worker bees is provided by the Experimental Animal Center of Jinzhou Medical University.

Total RNA was extracted from worker bees infected with DWV using Trizol reagent (TransGen, Beijing, China) and reverse-transcribed into cDNA using the first-strand cDNA synthesis kit (TransGen). Five pairs of primers were designed for pET28a-VP2, pBT3STE-VP2, pGEX-6p-1-snapin, pTT5-VP2-His, and pTT5-snapin-Flag (Supplementary Table S1). Insertion of enzyme-cutting sites was based on the DWV sequence (GenBank No. MF770715) and snapin gene (GenBank No. XM_006619914). PCR was performed under the following conditions: initial denaturation at 94°C for 5 min, 94°C for 45 s, 58°C for 45 s (pBT3STE-VP2, pET28a-VP2, and pTT5-VP2-His) or 65°C for 45 s (pTT5-snapin-Flag) or 61°C for 30 s (pGEX-6p-1-snapin), 72°C for 45 s, amplification for 30 cycles, and extension at 72°C for 7 min. The PCR products were authenticated using 1.2% agarose gel electrophoresis, purified, and then cloned into vectors. These recombinant vectors were transformed into E. coli DH5α and verified through enzyme digestion and sequencing (Genewiz Biological Technology Co., Ltd., Suzhou, China).

To test the bait plasmid pBT3STE-VP2 with respect to its self-activation and function, it was co-transformed with pPR3N into the reporter NMY32 yeast strain. The co-transformants were grown on SD/−Trp/−Leu, SD/−Trp/−Leu/-His, and SD/−Trp/−Leu/-His/Ala agar plates at 30°C for 3–5 days (Supplementary Table S2).

After ensuring that the bait is functional in the membrane protein yeast two-hybrid system, the host cell proteins that interact with VP2 can be screened. We transformed the A. cerana larvae cDNA library and the bait plasmid pBT3STE-VP2 into NMY32 cells and cultured them at 30°C for 3–4 days. Then, the NMY32 cells were plated on a series of selective agar plates, including SD/−Trp/−Leu and 60 mM 3-amino-1,2,4-triazole SD/−Trp/−Leu/−Ade/-His selective media, and cultured for 1–2 days at 30°C to select potentially positive colonies. The selected positive colonies were cultured in SD/−Trp/−Leu liquid medium at 30°C for 14 h. Plasmids were extracted from yeast and retransformed into E. coli DH5α for subsequent analysis. The plasmids were sequenced by Synbio Technologies Co. Ltd (Jiangsu, China). The resulting sequences were analyzed and aligned using BLAST in NCBI and GenBank databases.

Because there was no experimental structure available for snapin, it was modeled using the I-TASSER server² and modified using the GalaxyRefine server.³

In contrast, since an experimental structure was available for VP2, it was modeled using the SWISS-MODEL workspace.⁴

The ClusPro server⁵ is a widely used tool for protein–protein docking and can be used to predict protein interactions. To reveal the binding affinity between VP2 and snapin, we used the ClusPro server for docking analysis.⁶ Snapin acts as a receptor for antigen recognition. The ClusPro server computed the models based on desolvation energy and electrostatic interactions.

The GST pull-down assay was performed according to the manufacturer’s instructions (Bio-Works, Sweden) to confirm the interaction detected between snapin and VP2. The plasmids PGEX-6p-1-snapin and PET28-VP2 were transformed into E. coli BL21 (DE3) for induction, expression, and purification. The purified GST–snapin fusion protein was incubated with glutathione agarose beads at 4°C for 3 h, followed by overnight incubation at 4°C with 0.1 mg/ml of the input protein His-VP2. The agarose complex was eluted, separated, collected, centrifuged, and solubilized in 2× sodium dodecyl sulfate sample buffer. Next, western blotting was performed to analyze the collected proteins. As a negative control, GST was incubated with beads alone with E. coli lysates.

Co-IP assays were performed to further confirm the interaction between snapin and VP2 in cells. pTT5-VP2-His and pTT5-snapin-Flag were co-transfected into BHK cells. After 48 h, the cell lysates were harvested by centrifugation (Thermo Fisher, Waltham, MA, United States) at 12,000 rpm for 25 min at 4°C and incubated overnight with a Flag-tagged antibody (Proteintech, Wuhan, China) at 4°C. Then, Protein A/G Plus-Agarose (Santa Cruz Biotechnology, Santa Cruz, CA, United States) was added, and the mixture was incubated at 4°C for 6 h. This was followed by centrifugation at 1500 rpm for 3 min at 4°C, and the resulting precipitate was washed three times with ice-cold PBS and finally identified using western blotting.

The plasmids pTT5-VP2-His and pTT5-snapin-Flag were co-transfected into BHK cells. After 24 h, the cells were fixed in 2 ml of 4% paraformaldehyde (Solarbio, Beijing, China) at 4°C for 20 min and then washed three times with PBS. Next, the cells were permeabilized with 2 ml of 0.5% Triton X-100 (Sangon Biotech, Shanghai, China) for 30 min at room temperature, washed three times with PBS, and then blocked in Tris-buffered saline-Tween 20 (TBST) containing 2% bovine serum albumin (Solarbio) for 1 h at room temperature. The coverslips were incubated overnight together with His-tagged and Flag-tagged antibodies (1:400 dilution; Proteintech) at 4°C. The cells were washed three times with PBS. The coverslips were then incubated with goat anti-mouse and goat anti-rabbit fluorescent antibodies (diluted 1:2000 with TBST) in the dark for 1 h at 37°C. DAPI (Coolaber, Beijing, China) (1:750 dilution) was used to stain the cell nucleus in the dark for 20 min, after which the stained cells were viewed under a laser scanning confocal microscope (LEICA, Germany).

A total of 20 healthy worker bees were selected from the same colony and randomly divided into two groups, with each group containing 10 bees. The bees were placed on ice and anesthetized with CO₂ for 1 min. Group I was injected with 1 μL of a DWV suspension between the 2nd and 3rd integuments using a similar controlled-volume syringe (Huang et al., 2020), and group II was injected with 1 μL PBS as control. The inoculated bees were fed with sufficient sterilized food and water. The clinical signs in each group of bees were monitored and recorded daily until the bees died. RT-PCR was performed to detect the infection of DWV, Chinese sacbrood virus (CSBV), Chronic bee paralysis virus (CBPV), Black queen cell virus (BQCV), Kashmir bee virus (KBV), Israel acute paralysis virus (IAPV), and Acute Bee Paralysis Virus (ABPV) in dead bees (Hassanyar et al., 2019). The levels of snapin in bees after DWV infection were analyzed via SYBR Green RT-PCR.

We conducted silencing experiments to examine the function of snapin in the process of viral infection. Three pairs of siRNAs of snapin were designed and synthesized by Suzhou Jima Biotechnology Co., Ltd. A total of 40 healthy worker bees were retrieved from the same colony and randomly distributed into four groups (groups 1–4), with each group containing 10 bees. Group 1 was injected with 1 μg of siRNA1, group 2 was injected with 1 μg of siRNA2, group 3 was injected with 1 μg of siRNA3, and group 4 was injected with 1 μL PBS. After 72 h, three bees were randomly selected from each group for qPCR analysis.

After screening the best siRNA capable of inhibiting the function of snapin, 20 worker bees were retrieved from the same colony and randomly distributed into two groups (groups 5 and 6), with each group containing 10 bees. Group 5 was injected with 1 μg of siRNA and group 6 was injected with 1 μL of PBS. After 24 h, groups 5 and 6 were injected with 1 μL of DWV. After 72 h, three bees were randomly selected from each group, and the replication of DWV was analyzed via qPCR (Bradford et al., 2017).

The docking model was analyzed using the PDBePISA server (see text footnote 1) and according to the interface area Å2 and ΔIg kcal/mol to judge the docking effect. The interaction domain between VP2 and snapin was analyzed according to the interactions between the protein chains and residual interactions across the interface using the PDBsum server.⁷

The positive recombinant plasmids pBT3STE-VP2, pET28a-VP2, pGEX-6p-1-snapin, pTT5-snapin-Flag, and pTT5-VP2-His were identified through restriction digestion and sequencing analysis, which indicated that all recombinant plasmids were successfully constructed (Figure 1).

Selective plate assays revealed that white clones grew on all SD/−Trp/−Leu plates. In the positive control and functional test groups, white clones were detected on the solid plates of SD/−Trp/−Leu/-His and SD/−Trp/−Leu/−Ade/-His; however, in the self-activation detection and negative control groups, no such clones were observed (Supplementary Figure S1). These results show that the bait plasmid pBT3STE-VP2 had function and no autoactivation on Y2H. It was used for subsequent screening experiments.

The yeast two-hybrid system was used to screen for VP2-interacting host proteins from the A. cerana cDNA library. A total of 24 possible yeast proteins with a positive bait–prey interaction were screened from the membrane protein Y2H system and their reporter gene was detected. Sequence analysis of the 24 positive plasmids indicated that they represented 20 A. cerana cDNAs of potential protein genes interacting with VP2 (Supplementary Table S3). After analysis, we believe that the snapin gene is a useful host protein in these genes. Therefore, we selected the snapin gene for further investment.

Snapin was modeled using the I-TASSER server and five models were generated. Model 1 had the top Ramachandran favored and was therefore selected for optimization. The initial model 1 was optimized in the GalaxyRefine server that generated five models based on repeated structure perturbation and subsequent overall structural relaxation via molecular dynamics simulation. Based on the comprehensive analysis of TM score, root mean square deviation (RMSD), MolProbity, and Ramachandran plot, model 1 was selected for docking purposes (Figure 2A). The initial model 1 generated from the I-TASSER server and the refined model 1 generated from the GalaxyRefine server were evaluated using the SWISS-MODEL workspace. The initial model 1 and the refined model 1 had 70.33 and 89.88% of residues in the Ramachandran favored region, respectively (Figure 2B). Other favorable parameters of the refined model were as follows: GDT score of 0.8926, RMSD value of 0.554, MolProbability of 1.983, clash score of 7.9, and poor rotamers totaling 0.6 (Supplementary Figure S2).

VP2 was modeled using the SWISS-MODEL workspace and two models were generated. Model 1 had the top Global Model Quality Estimate and Ramachandran favored and was therefore selected for docking purposes (Figure 2C).

The ClusPro server yielded 30 candidate models with different binding energies, among which, model complex 1 with the lowest binding energy score of −1011.6 was chosen (Figure 2D). The interaction between VP2 and snapin was predicted.

The GST pull-down assay was performed to detect the interaction between VP2 and host snapin in vitro. In this experiment, GST–snapin was used as the bait protein, His-VP2 was used as the prey protein, and GST tag was used as the negative control. The results were analyzed via western blotting. The experimental and positive control groups exhibited an approximately 33-kDa protein band, consistent with the expected His-labeled VP2 protein sizes, whereas the negative control group exhibited no protein band (Figure 3). These results indicated the presence of an interaction between VP2 and snapin.

Co-IP assay is a common technique to identify or confirm physiologically relevant protein–protein interaction events and can be used for further confirmation of the interaction between VP2 and snapin. We performed Co-IP assay in BHK cells coexpressing snapin-Flag and VP2-His to further determine whether the interaction between snapin and VP2 occurs in vitro. As shown in Figure 4, detection with either anti-Flag or anti-His antibodies revealed the corresponding bands, demonstrating a specific interaction between snapin and VP2 in the cells.

To determine whether VP2 and snapin localize within the same cellular compartment, the recombinant plasmids pEGFP-snapin-Flag and pTT5-VP2-His were co-transfected into BHK cells. As shown in Figure 5, both snapin and VP2 localized in the cytoplasm. The results of co-localization revealed that VP2 and snapin could completely overlap in cells and were primarily localized in the cytoplasm.

We conducted SYBR Green RT-PCR to further investigate the changes in snapin expression in worker honeybees after DWV infection. Results indicated that the expression of snapin in worker honeybees was increased after DWV infection (Figure 6). Moreover, all honey bees used during the experiments were analyzed using RT-PCR, and the results showed that no honey bee viruses were detected in healthy workers and that only DWV was detected in infected workers (Supplementary Figure S3).

We have found that snapin expression was upregulated in vivo when the worker bees were infected with DWV. To examine the function of snapin in the process of DWV infection, we conducted silencing experiments. Compared with other siRNAs, siRNA1 exerted the maximum inhibitory effect on snapin expression (Figure 7A). Therefore, we selected siRNA1 for subsequent experiments. Results showed that DWV expression was significantly downregulated in group 5 (Figure 7B), indicating that snapin was involved in the replication of DWV in honeybees.

The interaction domain between VP2 and snapin was analyzed using the PDBsum server, and the results are shown in Figure 8. We detected 3 salt bridges and 16 hydrogen bonds in the interaction between VP2 and snapin at the protein–protein interface. The protein–protein interaction site of VP2 was approximately located at 56–90, 136–145, 184–190, and 239–242 aa. The protein–protein interaction site of snapin was approximately located at 31–54 and 115–136 aa.