The BmN cell line was cultured at 27°C in SF900 II SFM (Gibco) supplemented with 3% fetal bovine serum (FBS; Gibco). The T3 strain of BmNPV was maintained in our laboratory and used as the wild-type (WT) virus. The antibodies used in this study including anti-GP64 (Sigma, A2980), anti-GFP (TransGene, HT801), anti-Flag (HuaBio, 0912-1), anti-HA (Yeasen, 30702ES60), and anti-α-tubulin (Beyotime, AF5012) were commercially available.

The expression plasmids used in this study including pIZ-EGFP, pIZ-p35, pIZ-p35shift, pIZ-p35Flag, and pIZ-BmAgo2-HA were generated from the pIZ/V5-His plasmid. The pIZ/V5-His was preserved in our laboratory, and its multiple cloning sites (MCSs) had been optimized to be the same as the MCS of pFastBacHTB plasmid. All insertion open reading frames (ORFs) of the plasmids were confirmed by sequencing in both directions.

For transfection of the expression plasmids, BmN cells were resuspended and equally added to individual wells of a 6-well plate. Once monolayers had formed, the medium was replaced by 1ml TC-100 medium (Sangon, E60032) before transfection medium consisting of 125 μl TC-100 medium, 5 μl Lipo8000 (Beyotime, C0533), and 3 μg plasmid was added. The cells were incubated for 5 h after addition of the mixture and then replenished with 1.5 ml of SF900 medium supplemented with 3% FBS. For transfection of BmNPV bacmid, the method was similar. In brief, the medium of BmN cells was replaced, and a mixture containing 250 μl TC-100 medium, 8 μl LipoInsect (Beyotime, C0551), and 16 μg bacmid was added. The medium was rechanged after 4 h.

To generate the p35KO mutant, a p35 knockout BmNPV bacmid was constructed via λ-Red homologous recombination as described previously (35–37). A linear chloramphenicol acetyltransferase (cat) gene cassette flanked by 50-nt fragments homologous with the adjacent regions of the p35 ORF that mediate the insertion of the cat gene into the p35 locus while concomitantly deleting the p35 ORF (Figure S1A) was amplified from plasmid pKD3. The PCR fragment was gel purified and electroporated into Escherichia coli BW25113 competent cells containing the BmNPV bacmid and the temperature-sensitive plasmid pKD46 (38). Transformants were screened with kanamycin and chloramphenicol and confirmed by PCR ( Figures S1B , E–H ). For visualization of the viral infection, polyhedrin (polh) of BmNPV with its own promoter was inserted into the polh locus of the resulting p35KO bacmid using pFastBacDual plasmid via Tn7-mediated transposition, as previously described (39–41) ( Figure S1A ).

As for the p35V71P mutant, the mutated p35 gene whose nucleotides coding for the valine 71 were changed to those for proline was amplified via overlap extension PCR ( Figure S1C ), and mutation was confirmed by sequencing in both directions ( Figure S1D ). This mutated gene was then inserted into the polh locus together with the polh gene via Tn7-transpositon to generate the p35V71P bacmid ( Figure S1A ). The two mutant viruses were amplified and purified from BmN cells transfected with the corresponding bacmids.

To silence the genes, we used RNAi by generating dsRNA synthesized in vitro using the T7 RNAi transcription kit (Vazyme, TR102) according to the instructions from the manufacturer. T7 promoter sequences were incorporated in both forward and reverse primers designed to amplify the DNA fragments of target genes as transcription templates for dsRNA. The control template in this kit was used to generate dsCtr as a negative control.

For transfection of the dsRNA, BmN cells were resuspended and equally added to individual wells of a 6-well plate. Once monolayers had formed, the medium was replaced by 1ml TC-100 medium before transfection medium consisting of 125 μl TC-100 medium, 5 μl LipoRNAi (Beyotime, C0535), and 5 μg dsRNA was added. The cells were incubated for 5 h after addition of the mixture and then replenished with 1.5 ml of SF900 medium supplemented with 3% FBS.

BmN cells (1 × 10⁶) infected with indicated viruses at a multiplicity of infection (MOI) of 10 or transfected with dsRNAs were harvested at the given time points. Total RNA of the cells was extracted using RNAiso Plus (Takara, 9109) to synthesize the first-strand cDNAs with the PrimeScript™ RT reagent Kit (Takara, RR047). Reactions were performed using Hieff^® qPCR SYBR Green Master Mix (Yeasen, 11203) and run on a Light Cycler 480 (Roche). Bmrpl32 (Gene ID 778453) was used for normalizing the data. Three technical and three biological replicates per treatment were analyzed using the 2^-△△Ct method to calculate the relative expression levels of selected genes. Student’s two-tailed t test was used for significant difference analysis between two samples. The statistical analysis and figure generation were completed by GraphPad Prism 8.

Viral growth curve analysis was conducted as described previously (42). In brief, BmN cells were infected in triplicate with indicated viruses at an MOI of 10. After 1 h of incubation at 27°C, fresh medium was added to cells, and this time point was defined as 0 h post infection (hpi). At the given time points, cells were centrifuged and supernatants containing the BV were harvested. The titers of BV were determined by TCID₅₀ end-point dilution assay in BmN cells (43).

To quantify viral genomic DNA (gDNA) copies, total DNA of sediments obtained from the centrifugation was extracted using Universal Genomic DNA Extraction Kit (Takara, 9765). Viral gDNA copies were determined by quantification of a viral gene gp41 (Gene ID 1488698) by qPCR. DNA concentrations were measured by a NanoDrop instrument (Thermo), and 30 ng total genomic DNA was used for each qPCR reaction. Bmrpl27 (Gene ID 692703) served as an internal normalization control, and the 0 h sample was used as input to normalize the data. The differences in means between two or three sets of data at a certain time point were compared by t test or one-way ANOVA. GraphPad Prism 8 was used for statistical analysis and figure generation.

BmN cells (1 × 10⁶) of different treatment conditions were harvested at the designated time points and lysed in cell lysis buffer (Beyotime, P0013) for 30 min on ice. Protein concentrations in lysates were determined using Bicinchoninic Acid (BCA) Protein Assay Kit (Takara, T9300A). For each sample, 20 μg of protein extract was separated by a 12% Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS PAGE) gel and transferred onto a polyvinylidene fluoride (PVDF) membrane. The membrane was blocked with 5% non-fat milk in Tris Buffered Saline with Tween (TBST) for 2 h and incubated with corresponding antibodies for 1.5 h at room temperature or overnight at 4°C. Thereafter, Horseradish Peroxidase (HRP)-conjugated antibodies were used for enhanced chemiluminescence. α-Tubulin served as an internal reference.

BmN cells (1 × 10⁷) were infected with BmNPV WT or p35KO mutant virus at an MOI of 10. After 48 h of infection, cells were harvested and total RNA of the cells was extracted. The small RNAs were size selected by PAGE gel, and the small RNAs ranging from 15 to 35 nt were excised. Approximately 3 μg of RNA per sample was used as the input material for library preparation, the libraries were generated using NEB Next Multiplex Small RNA Library Prep Set for Illumina (NEB), and index codes were added to attribute sequences to each sample. The prepared libraries were sequenced on an Illumina Hiseq 2500/2000 platform (Allwegene), and 50-bp single-end reads were generated.

For the analysis of viral small RNAs, raw reads generated from 3 independent libraries were cleaned using Trim Galore (Version 0.6.6) and mapped to the BmNPV genome (GenBank accession NC_001962) using Bowtie2 (Version 2.4.2) (44). Mapped reads were normalized to library size and expressed as reads per million (RPM). For the genome distribution, the number of 5’ ends of all vsiRNAs at each position was plotted. miRNA levels were analyzed by mapping the cleaned reads to the B. mori genome (SilkBase) (45). The union of differentially expressed miRNAs between two groups was used for miRNA heat map. Differential expression analysis of two groups was performed using R package edgeR (Version 3.30.3), and P value of 0.05 was set as the threshold for significant differential expression. Figures were generated using R package ggplot2 (Version 3.3.2) or pheatmap (Version 1.0.12) and embellished using Adobe Illustrator.

BmN cells were resuspended and equally added to individual wells of a 96-well plate. At designated time points, BmNPV p35KO or p35V71P mutant was added at an MOI of 10. Apoptosis degree of the cells was determined by the activity of caspase-3 and caspase-7 using Caspase-Glo 3/7 Assay System (Promega, G8090) (46). Caspase-Glo 3/7 substrate and buffer in the kit were mixed, and each well was added by 100 μl of the reagent simultaneously. The luminescent signal was measured by Microplate Reader (Bio Tek) after incubation for 1 h. Caspase 3/7 activity was represented by relative light unit (RLU). Three replications were set for each time point, and the fresh medium served as the blank control.

BmN cells (1 × 10⁶) transfected with siRNA of Enhanced Green Fluorescent Protein (siEGFP) for 48 h were harvested, and small RNA of the cells was extracted using RNAiso for Small RNA (Takara, 9753). β-Elimination of the RNA was performed as the protocol described before (47).

In this study, 40 µg of RNA in 47.5-µl nuclease-free water was oxidized by addition of 12.5 µl 200 mM NaIO₄ and 40 µl 5× borate buffer and incubation at room temperature for 30 min. In the control group, NaIO₄ was replaced by RNase-free water. Unreacted NaIO₄ was quenched by addition of 10 µl glycerol and incubation at room temperature for 10 min. To induce β-elimination, 10 µl of 500 mM NaOH was added and the reaction was kept at 45°C for 90 min. The treated RNA was purified by ethanol precipitation.

Northern blot analysis was performed as the protocol described before (48). Here, 10 μg of small RNA were separated on a 7-M urea-denaturing 15% polyacrylamide gel and transferred onto a nylon membrane. siRNA Ladder Marker (TaKaRa, 3430) was used as a molecular size marker. UV-cross-linked membrane was prehybridized for 3 h and hybridized with biotin-labeled DNA probe. Viral probe was synthesized using Biotin Random Prime DNA Labeling Kit (Beyotime, D3118). The signal was detected using Chemiluminescent Biotin-labeled Nucleic Acid Detection Kit (Beyotime, D3308). Ethidium bromide–stained 5S rRNA served as an internal reference.

Cell viability was determined using Cell Counting Kit-8 (Beyotime, C0038) and Calcein-AM/PI Live-Dead Cell Staining Kit (Solarbio, CA1630) according to the protocol provided by the manufacturers. In brief, BmN cells in 96-well plate were infected with BmNPV at an MOI of 10 at designed time points. And the cells were added with corresponding reagent. The absorbance of each well was measured at a test wavelength of 450 nm.

For RNA immunoprecipitation, BmN cells (1 × 10⁶) co-transfected with siEGFP and pIZ-BmAgo2-HA for 48 h were harvested and washed twice by PBS and then lysed using the lysis buffer with freshly added 1 mM Dithiothreitol (DTT), 200 U/ml RNase inhibitor (Promega), and protease inhibitor cocktail (Bimake, B14001) for 30 min at 4°C. Here, 10% of lysates were used as protein inputs and 20% of lysates were used for small RNA isolation as RNA inputs. The remnant lysates were prepared for immunoprecipitation overnight at 4°C using anti-HA magnetic beads (Bimake, B26201). Immunoprecipitation beads were washed with lysis buffer five times. Then, 25% of the washed beads were used for Western blotting, and the remnant beads were used for RNA extraction.

The co-immunoprecipitation is similar. In brief, cells cotransfected with pIZ-p35Flag and pIZ-BmAgo2-HA were lysed using the lysis buffer with freshly added cocktail for 30 min at 4°C. Here, 20% of lysates were used as inputs and remnant lysates were used for immunoprecipitation overnight at 4°C using Anti-Flag magnetic beads (Bimake, B26101). Immunoprecipitation beads were washed with PBS five times and used for Western blotting.

Baculovirus-derived siRNAs from Helicoverpa armigera single nucleopolyhedrovirus (HaSNPV)-infected HzFB cells and Autographa californica multiple nucleopolyhedrovirus (AcMNPV)-infected Sf9 cells have been reported previously (14, 34). Nonuniformly distributed vsiRNAs on viral genomes resulted in hot spots with a large number of mapped vsiRNAs and cold spot with a low number of mapped vsiRNAs. Also, the expression of genes in hot spots instead of cold spots was affected when Dcr-2 was silenced (18). To confirm the significance of cellular siRNA pathway in resistance of BmNPV, the key factors, Dcr-2 and Ago2, were knocked down respectively via dsRNAs, and the silencing efficacy was examined by qPCR. After 48 h of dsRNA transfection, the gene-silenced cells were infected with BmNPV, and the effect on viral amplification was assessed by quantifying viral titers and gDNA copies at various time points as well as the expression level of GP64, a key envelope protein of BmNPV.

After 48 h of dsRNA transfection, the expression of Dcr-2 and Ago2 was dramatically decreased ( Figure 1A ). As a consequence, the expression of GP64 was enhanced ( Figure 1B ), and viral replication was also increased. During 12–48 h, viral titers in Dcr-2- and Ago2-silenced groups showed a modest but significant increase compared with the negative control ( Figure 1C ). Also, viral gDNA copy number in RNAi-defective groups was significantly higher than the control until 96 h postinfection ( Figure 1D ). These results illustrated that BmNPV is also a target of RNAi like the viruses mentioned above, and its infection progress can be restricted by the siRNA pathway of the host. However, the siRNA pathway seems to make no difference on viral invasion at a relatively late stage, and it may be due to the overwhelming viral load that is unaffordable for the RNAi-mediated antiviral immunity.

In a previous study, p35 of AcMNPV was found to be responsible for the suppression of RNAi in diverse insect and mammalian cells, while the molecular mechanism of this suppression was not elucidated (34). Before exploring this mechanism, we firstly verified the RNAi suppression effect of BmNPV p35. To visualize the RNAi efficiency, the constructed pIZ-EGFP plasmid was cotransfected with dsRNA of EGFP (dsEGFP) into BmN cells. Subsequently, the cells were infected with wild type and p35-deleted BmNPV ( Figure 2A ). After 24 h of infection, the cells were harvested to detect the expression level of EGFP. In cells treated by dsEGFP without the existence of p35, namely, mock and p35KO virus-infected cells, the expression level of EGFP was much lower than that in the dsCtr-treated cells, while no obvious difference was observed between the dsCtr and dsEGFP treatments of WT virus-infected cells ( Figure 2C ). These results indicated that the p35 gene of BmNPV was responsible for the suppression of cellular siRNA pathway.

To investigate whether the suppression effect is independent of viral infection, we constructed a pIZ-p35 plasmid to express p35 of BmNPV in vitro and a negative control pIZ-p35shift, in which a stop codon was introduced after the p35 ATG start codon by frameshift mutation to exclude the influence of p35 transcript ( Figure 2B ). As expected, after 48 h of cotransfection, the nullification of dsEGFP was only observed in pIZ-p35-transfected cells ( Figure 2D ), indicating that it is an individual function of p35.

In siRNA pathway, there are two major parts for gene silencing, namely, Dcr-2 cleavage and its downstream steps. Considering that p35 is able to restrict cellular siRNA pathway, we wonder which step is primarily responsible for this effect. In contrast to dsRNA, transfected siRNAs do not require processing by Dcr-2 and directly participate in the next steps. Therefore, it is suitable to estimate the relative importance of Dcr-2 cleavage and its downstream steps by detecting siRNA-induced gene silencing efficiency of EGFP. If inhibition of Dcr-2 function is the predominant mechanism, p35 will fail to restrict the function of directly introduced siRNA, and vice versa.

In siEGFP-transfected cells, the expression of EGFP was obviously inhibited. Such an inhibitory effect could also be observed in p35KO BmNPV-infected and pIZ-p35shift-transfected BmN cells, but not in WT BmNPV or pIZ-p35-treated cells ( Figures 2E, F ). These results indicated that the skip of dsRNA cleavage cannot avoid the suppression of siRNA pathway, and the steps downstream of Dcr-2 cleavage are primarily responsible for this p35-mediated suppression.

Previous results showed that the silencing efficiency of introduced siRNA could also be effectively restricted by p35 ( Figures 2E, F ), implying that the steps downstream of Dcr-2-induced dsRNA slicing were affected. After being processed by Dcr-2, resulted siRNAs must be loaded onto Ago2 to execute their function (6, 49, 50). Therefore, we hypothesized that this key process of RISC formation was blocked by p35 therein. To directly test this hypothesis, we analyzed sensitivity to β-elimination of introduced siEGFP. In previous studies, once siRNAs were incorporated in Ago2, a 2′-O-methyl group would be added on their 3′ termini by the protein Hen1 (51–53). Such a methylation modification endows siRNAs with the resistance to β-elimination. Unmodified siRNAs are sensitive to the reaction, being converted into RNAs with 3′-monophosphate by the reaction. The resulting RNAs migrate faster in polyacrylamide gel electrophoresis (47, 54), making it visible to determine whether the siRNA is loaded onto Ago2.

We conducted Northern blot to detect the effect of the β-elimination reactions and found that in the pIZ-p35-transfected cells, an appreciable fraction of siEGFP was sensitive to β-elimination, as evident from the increased mobility on gel, indicating that it had not been loaded into Ago2. In contrast, transfected siEGFP in mock-transfected and pIZ-p35shift-transfected cells was not affected by β-elimination, indicating that they were incorporated with Ago2 ( Figure 3A ).

To further confirm that p35 prevents siRNA from loading on Ago2, we performed RNA immunoprecipitation assay with anti-HA magnetic beads by using pIZ-BmAgo2-HA- and siEGFP-transfected BmN cells in the absence or presence of p35. As expected, the binding between Ago2 and siRNA was significantly suppressed ( Figure 3B ). Together, these results proved that p35 can obstruct the steps downstream of Dcr-2 cleavage by preventing siRNAs from associating with RISC.

The functional mechanisms of VSRs are diverse. Some of them can restrict both Dcr-2 cleavage to alter vsiRNA generation and the formation of RISC to block the function of siRNA, as is described in the Introduction. Although p35 mainly affects the downstream steps of siRNA pathway, we also would like to make clear its role in vsiRNA accumulation. To this end, small RNA-seq of BmN cells infected with WT or p35KO BmNPV was conducted and the small RNAs mapped to BmNPV genome were analyzed. As observed before (14), small RNAs in both treatments were predominantly 20 nt in length ( Figure 4D ), which coincides with the typical size of Dcr-2 products. However, to our surprise, the distribution of mapped vsiRNAs across the viral genome was not similar in WT and p35KO BmNPV infection. Instead, in the context of WT BmNPV infection, the count of vsiRNAs mapped to approximately 50k–95k nt was evidently lower than that in the context of p35KO BmNPV infection ( Figures 4A, B ). Correspondingly, most of the genes in this region were mapped to more vsiRNAs in the absence of p35 ( Figure S3A ). To validate this phenomenon, Northern blot analysis was performed and the probes were generated from the viral fragments in the genome coordinates of 50k–95k nt. Consistent with the sequencing result, the band of approximately 20-nt vsiRNA complementary with probes appears more pronounced in the p35KO BmNPV-infected cells ( Figures 4C and S3B ), indicating a disequilibrium induced by BmNPV p35 in the accumulation of vsiRNA.

In plants, Ago1 is a shared factor of antiviral RNAi and miRNA pathways that can load with both vsiRNAs and miRNAs (55). Due to this convergence of functions, VSRs of plants can affect cellular miRNAs (56). In contrast, Dicer and Argonaute proteins are exclusive in miRNA and siRNA pathways of insects (57). We thus analyzed miRNA profiles of uninfected and WT or p35KO BmNPV-infected cells to evaluate the role of p35 in miRNA pathway. The length distribution was not affected during viral infection, and most of the detected miRNAs were 20–22 nt as expected ( Figure S3C ). In both virus-infected cells, a reduction in the level was observed for the majority of changed cellular miRNAs, while the difference between the two virus-infected libraries was not remarkable. Generally, there appeared to be an identical trend of miRNAs in the infection of WT and p35KO BmNPV. However, two small sets of miRNAs were worthy of attention. One was represented by miR-2780d and miR-2744, which were only induced by p35KO BmNPV, and another was represented by miR-3275, which was only upregulated by p35 therein and might be related to some biological functions of p35 ( Figures 4E and S3D ). Together, these data suggest that p35 has little effect on the expression of cellular miRNA and the length distribution of small RNAs but alters the vsiRNA accumulation in the viral genome coordinates of 50k–95k nt.

Since the existence of p35 reduced the vsiRNA accumulation of the 50k–95k nt in the BmNPV genome coordinates, we focused on the specific property of this region. To this end, we analyzed the distribution of genes important for viral replication and spread and found that the distribution density of essential genes (virus lacking these genes will not be able to survive) and important genes (viral infection and spread will be severely restricted without these genes) in the 50k–95k region was higher than that of other regions in the BmNPV genome ( Figure 5A ) (7, 58). Therefore, we examined how the decrease of vsiRNA affects the expression of these genes in the 50k–95k region. Considering that p35 is also well known as an inhibitor of apoptosis besides its VSR function (33), to eliminate the influence of apoptosis induced by p35KO BmNPV infection, we mutated valine 71 of BmNPV p35 to proline (p35V71P), which was previously shown to disrupt the spatial conformation of the reactive loop structure in p35 and to completely abolish its antiapoptotic activity by failing to inhibit the caspase activity (59, 60). The mutated p35 gene was confirmed by sequencing and recombined into p35KO BmNPV to generate the p35V71P-mutated virus.

The functional effect of BmNPV p35V71P was examined first. As expected, after destruction of the crucial site of apoptosis resistance, the mutated virus failed to restrict cellular caspase activity as the WT BmNPV did ( Figure S2C ), and apoptosis induced by p35V71P BmNPV in BmN cells was as severe as that in p35KO BmNPV-infected cells ( Figure S4 ). Nevertheless, the function of p35V71P as a VSR was not affected by this single amino acid mutation. In p35V71P mutated virus-infected cells, gene silencing mediated by dsEGFP was obviously inhibited, and the inhibition effect was the same as the WT BmNPV ( Figure 5B ), demonstrating it an ideal strategy to avoid the discrepancy produced by distinct apoptosis rate with no impact on the suppression of cellular RNAi. Eight essential genes and 1 important gene (lef-7) in 50k–95k region were selected to analyze whether the altered vsiRNA level affects the gene expression of this region. In the p35V71P-infected BmN cells, mRNA levels of the tested genes showed an increase of 3~4-fold ( Figure 5C ). Taken together, our results indicated that the p35-induced decline of vsiRNA mapped to the 50k–95k of the BmNPV genome coordinates promoted the expression of critical genes in this region, which may create a favorable condition for viral infection.

We have observed an enhanced viral infection in cells with RNAi defects caused by Dcr-2 or Ago2 knockdown and an increased gene expression in the region with reduced vsiRNA yield caused by the existence of BmNPV p35. Therefore, we hypothesized that the p35-mediated inhibitory effect on host siRNA pathway will contribute to an improved viral infection. In order to avoid the diversity in the level of cell apoptosis, we also used the mutant virus p35V71P as the control of p35KO mutant virus.

To assess the impact of the impaired siRNA pathway caused by p35, we infected BmN cells with the two mutant viruses respectively and monitored viral titers and DNA levels over time. A consistent infection and replication defect were observed in p35KO virus, with 4.2–5.1-fold lower viral titers and 2.4–3.3-fold lower DNA levels compared with the control ( Figures 6A, B ). To vindicate that such differences was due to the p35-induced inhibition on cellular RNAi instead of other variables, we generated RNAi-defective cells by knocking down the key factors of siRNA pathway, Dcr-2 or Ago2 as mentioned above, and infected these cells with the two mutant viruses, respectively. As expected, the differences in viral titers and DNA levels were not observed. Mutant viruses p35KO and p35V71P replicated with similar kinetics and accumulated to similar levels in RNAi mutant cells ( Figures 6C–F ). It is because that p35-induced suppression cannot lead to discrepancies in these RNAi-impaired cells. However, it is noteworthy that in the case of viral infection, cellular siRNA pathway was not completely blocked by p35, based on the fact that in the presence of p35, RNAi defects caused by Dcr-2 or Ago2 knockdown still led to significantly enhanced viral infection and accumulation ( Figure S5 ). Overall, these data indicated that the VSR function of BmNPV p35 can create a favorable condition for the establishment of viral infection and the replication of viral genome during a relatively early stage, even though the antiviral activity of cellular siRNA pathway cannot be fully masked.