DeSUMOylation of Apoptosis Inhibitor 5 by Avibirnavirus VP3 Supports Virus Replication

ABSTRACT SUMOylation is a reversible posttranslational modification involved in the regulation of diverse biological processes. Growing evidence suggests that virus infection can interfere with the SUMOylation system. In the present study, we discovered that apoptosis inhibitor 5 (API5) is a SUMOylated protein. Amino acid substitution further identified that Lys404 of API5 was the critical residue for SUMO3 conjugation. Moreover, we found that Avibirnavirus infectious bursal disease virus (IBDV) infection significantly decreased SUMOylation of API5. In addition, our results further revealed that viral protein VP3 inhibited the SUMOylation of API5 by targeting API5 and promoting UBC9 proteasome-dependent degradation through binding to the ubiquitin E3 ligase TRAF3. Furthermore, we revealed that wild-type but not K404R mutant API5 inhibited IBDV replication by enhancing MDA5-dependent IFN-β production. Taken together, our data demonstrate that API5 is a UBC9-dependent SUMOylated protein and deSUMOylation of API5 by viral protein VP3 aids in viral replication.

vivo. UBC9, a unique SUMO-conjugating enzyme, is the key protein of the SUMO machinery and functions as a hub for protein SUMOylation and transfers SUMO to targets (13). Recently, growing evidence indicates that viruses can interfere with the SUMOylation of proteins (14)(15)(16). In particular, UBC9 is frequently targeted by numerous viral proteins to create a conducive environment for their propagation, i.e., human papillomavirus E6/E7 oncoprotein (17), avian adenovirus early protein Gam1 (18), and human adenovirus oncoprotein E1A (19).
Apoptosis inhibitor 5 (API5), also known as antiapoptosis clone 11, was first reported to support cell viability under serum and growth factor deprivation conditions, potentially by inhibiting caspase-3 or E2F transcription factor 1 (E2F1)-induced apoptosis (20)(21)(22). It was then observed that API5 is overexpressed by many types of tumors to promote their progression and serves as a marker of poor prognosis for these cancers (23)(24)(25)(26)(27)(28). Recently, API5 was reported to induce dendritic cell and immune response activation via Toll like receptor 4 (TLR4) and to target the nucleoprotein (NP) of influenza A virus to regulate virus replication (29,30). Study in our laboratory indicated that API5 is upregulated in Avibirnavirus-infected DF-1 cells and translocates to the cytoplasm from the nucleus (31). However, the role of API5 in Avibirnavirus infection is poorly understood.
Avibirnavirus infectious bursal disease virus (IBDV), a member of the Birnaviridae family, is a nonenveloped RNA virus containing a bi-segmented double-stranded RNA (dsRNA) genome comprising segment A and segment B (32). Segment B encodes the putative RNA-dependent RNA polymerase VP1 (33). Segment A encodes nonstructural protein VP5 and the precursor polyprotein NH3-pVP2-VP4-VP3-COOH, which can be self-cleaved to form pVP2, VP3, and VP4 (34). Recently, we found that Avibirnavirus VP1 can employ the host posttranslational modification system to support viral replication (35,36). In addition, reports demonstrated that VP3, a scaffolding protein with multiple functions, can inhibit the phosphorylation of dsRNA-dependent protein kinase R (PKR) to promote IBDV replication (37). However, whether viral protein VP3 can affect viral replication by modulating small ubiquitin-like modification of cellular proteins is unknown.
The objective of this study was to explore whether API5 undergoes SUMOylation and, if so, how to regulate its function and whether Avibirnavirus VP3 can regulate self-replication through interfering with API5 SUMOylation. Here, we revealed that API5 K404 can be conjugated by SUMO3. Meanwhile, we demonstrated that Avibirnavirus infection inhibited API5 SUMOylation. Further investigation of the molecular mechanism demonstrated that VP3 directly inhibited SUMO3 modification of API5 via their interaction and facilitated UBC9 degradation. Finally, we revealed that the deSUMOylation of API5 at K404 supports IBDV replication by blocking MDA5-dependent IFN-b induction.

RESULTS
API5 is a SUMOylated protein.
To explore the chemical modification of API5, lysates from DF-1 cells were analyzed. In immunoblotting assays, we observed the expected band with an approximate molecular weight of 55 kDa (API5) and also higher molecular weight bands (termed M-API5) (Fig. 1A). To determine whether the posttranslational modification of API5 involves SUMOylation, we screened highly efficient short hairpin RNA (shRNA) against UBC9 in DF-1 cells (Fig. 1B) and observed that M-API5 levels were significantly reduced in UBC9-silenced DF-1 cells compared with control cells (Fig. 1C). In addition, Fig. 1D and E show that both exogenous and endogenous API5 interacted with UBC9. These data indicated that M-API5 might represent SUMOylated API5. Meanwhile, in vivo and in vitro SUMOylation assays indicated that SUMO3-conjugated API5 could be detected ( Fig. 1F and G), suggesting that API5 can be effectively modified by SUMO3.
Residue 93 (cysteine [C]) of UBC9 and residue 92 (glycine [G]) of SUMO3 were reported to be the active sites for the SUMOylation of substrates (38). We constructed Myc-Ubc9 C93S and HA-SUMO3 G92A mutants. SUMOylation assays revealed that SUMO3-API5 levels were reduced significantly in Ubc9 C93S -and SUMO3 G97A -transfected cells ( Fig. 1H and I), indicating that SUMOylation of API5 is UBC9-and SUMO3-dependent. SUMOylation of substrates is commonly controlled by SUMO E3 ligases, such as protein inhibitor of activated STAT (PIAS), and deSUMOylation enzymes (sentrin-specific proteases, SENPs) (39,40). To determine the SUMO3 E3 ligases and deSUMOylateases of API5, we constructed multiple PIAS and SENP expression plasmids. Finally, the results showed that PIAS2 increased SUMO3 modification of API5 (Fig. 1J), and SENP1 and SENP2 exhibited the most efficient deconjugation effect ( Fig. 1K and L). Amino acids C603 of SENP1 and C548 of SENP2 function as the active sites of SUMO covalent attachment (41,42). We constructed SENP1 CS and SENP2 CS mutants, and further confirmed that SENP1 and SENP2 were responsible for reversing API5 SUMO3 modification ( Fig.1K and L). Taken together, these data indicated that SUMO3 modification of API5 is regulated by PIAS2, SENP1, and SENP2.
Lysine 404 of API5 is the SUMO3 modification site. To determine the SUMO3 modification site of API5, we generated a series of API5 mutants according to the results of online SUMOylation prediction (http://sumosp.biocuckoo.org/). Figure 2A shows that SUMO3 greatly increased the SUMOylation of API5-WT (wild type), K36R, K126R, and K474R mutants, but not the K404R mutant, suggesting that the residue K404 is critical for the SUMO3 modification of API5. To explore if SUMO modification of API5 (API5-s) affected its nuclear transport, cytoplasmic and nuclear components from DF-1 cells were analyzed. Immunoblotting results showed that API5-s was mainly located in the nucleus (Fig. 2B). Moreover, confocal microscopy revealed that API5 K404R failed to localize to the nucleus (Fig. 2C) and this mutant localized in the nucleus in cells treated with leptomycin B, an inhibitor of nuclear export (Fig. 2D), indicating that SUMO3 modification at K404 contributes to the nuclear retention of API5. Collectively, these results demonstrated that K404 of API5 is important for its SUMO3 modification and nuclear localization.
Residue L211 on the CC3 domain of VP3 is critical for direct interaction with API5. Avibirnavirus VP3 protein is involved in interactions with cellular proteins and the modulation of numerous key cellular pathways (37,(43)(44)(45). Previous study in our laboratory showed that endogenous API5 colocalizes with VP3 in cytoplasm during infection (31). In the present study, we observed that overexpressed API5 colocalized with VP3 in the cytoplasm of IBDV-infected cells ( Fig. 3A to C). To further explore the relationship between VP3 and API5, coimmunoprecipitation (co-IP) assays were conducted separately during infection and transfection. We found that VP3 interacted with API5 in IBDV-infected cells (Fig. 3D) and in cotransfected cells ( Fig. 3E and F). In addition, glutathione S-transferase (GST) pulldown assays further confirmed that API5 interacted directly with VP3 (Fig. 3G). VP3 contains three coiled-coil (CC) domains, CC1, CC2, and CC3 (43). In co-IP assays, we further found that the residues 205 to 219 of the CC3 domain of VP3, but not CC1 and CC2, were required for the interaction with API5 ( Fig. 3H and I). Substitution mutation assays further showed that the residue leucine (L) 211 on VP3 was essential for the association with API5 (Fig. 3J). These data revealed that the residue L211 on the CC3 domain of VP3 was responsible for binding to API5.
VP3 inhibits SUMOylation of API5 via its CC3 domain with function similar to deSUMOylation enzyme. To explore if Avibirnavirus infection could interfere with the SUMOylation of API5, lysates from mock-infected and IBDV-infected DF-1 cells or chicken's bursa of fabricius (BF, the IBDV-infected target organ) were subjected separately to SUMOylation assays. Figure 4A and B show that SUMOylated API5 in IBDV-infected cells and BF decreased markedly compared with that in mock-infected cells and BF. To further investigate whether VP3 was directly related to the Avibirnavirus-induced decrease of API5 SUMOylation, small interfering RNA (siRNA) against VP3 was used. Figure 4C shows that VP3 knockdown significantly inhibited IBDV-induced SUMO2/3-API5 reduction. Moreover, we observed that levels of SUMO3-API5 were reduced significantly in VP3-overexpressing HEK293T cells (Fig. 4D). These data clearly suggested that VP3 plays an important role for IBDV-induced reduction of API5 SUMOylation. Figure 1K and L demonstrate that SENP1 and SENP2 could reverse the SUMO3 modification of API5, while Fig. 4E shows that VP3 was not detected in SENP1-or SENP2-precipitations, indicating that the VP3-induced decrease in API5 SUMOylation is independent of SENP1 and SENP2.
Given that the SUMOylated API5 locates in the nucleus, we hypothesized that VP3 could translocate into the nucleus and deSUMOylate SUMO3-API5 by direct interaction. A potential nuclear localization signal (NLS) is predicted at residues 76 to 105 of the VP3 protein (R 76 AKYGTAGYGVEARGPTPEEAQRGKDTRIS 105 ) by an online tool (NLS Mapper, http://nls-mapper.iab.keio.ac.jp/cgi-bin/NLS Mapper form.cgi). Subsequently, the cytoplasmic elements (CE) and nuclear elements (NE) of the cells transfected with wild-type Myc-VP3 or Myc-VP3DNLS were subjected to immunoblotting analysis. As  DF-1 cells were transfected with Flag-API5 for 24 h and then infected for another 12 h with IBDV. Cells were stained with anti-VP3 mouse MAb and anti-Flag rabbit pAb, followed by FITC-labeled goat anti-mouse IgG and Alexa Fluor 546-conjugated donkey anti-rabbit IgG antibodies, and finally with DAPI, before being observed by confocal microscopy. Scale bar, 10mm. (B) Scatterplot. The Image J Just Another Colocalization Plugin (JACoP) was applied for quantitative colocalization analysis. In the scatterplot, the symbol r represents the Pearson's correlation coefficient. (C) Relative intensity of the three channels along the white line in A. (D) Viral protein VP3 interacts with endogenous API5. DF-1 cells were infected for 12 h with IBDV. The resulting cells were lysed for co-IP assays using mouse MAb against API5. The interaction between VP3 and API5 was detected using Western blotting with anti-VP3 mouse MAb and anti-API5 rabbit pAb. (E and F) VP3 associates with API5 during transfection. Flag-API5 and Myc-VP3 were cotransfected into HEK293T cells for 48 h and then cellular lysates were separately subjected to co-IP with anti-Flag or anti-Myc mouse MAb, followed by Western blotting with anti-Flag and anti-Myc rabbit pAbs.   Fig. 4F, VP3, but not VP3DNLS, could be detected in the NE fraction. Moreover, VP3 could translocate into the nucleus during IBDV infection (Fig. 4G to I) and bind directly SUMO3-API5 in a GST pulldown assay (Fig. 4J). Furthermore, in vitro deSUMOylation assays demonstrated that the protein bands with high molecular weight detected in the lane corresponding to API5-SUMO3 were greatly reduced after incubation with VP3 (Fig. 4K). In cells, SUMOylation assays further confirmed that VP3 CC3, but not CC1 and CC2, was essential for decreasing SUMO3-conjugated API5 (Fig. 4L), suggesting that the VP3-mediated reduction of API5 SUMOylation depended on their interaction. Altogether, the viral protein VP3 CC3 domain might have an activity similar to deSUMOylation enzymes for directly inhibiting the SUMO3 modification of API5.
VP3 inhibits API5 SUMOylation by facilitating UBC9 degradation. To investigate if Avibirnavirus VP3 can bind to UBC9, co-IP assays were performed. Results showed that VP3 associated with UBC9 during transfection ( Fig. 5A and B) and infection (Fig. 5C). Moreover, endogenous UBC9 in IBDV-infected cells colocalized with VP3 ( Fig. 5D to F). In addition, immunoblotting results showed that the UBC9 protein level decreased markedly in comparison with that in mock-infected control cells and BF during virus infection (Fig. 5G). Meanwhile, we observed that the level of SUMOylated API5 decreased markedly, although IBDV infection upregulated the level of total API5 (Fig. 5H). These data indicated that IBDV infection might inhibit the SUMOylation of API5 by downregulating UBC9. To further investigate if the decrease of IBDV-induced UBC9 is relevant to VP3, the lysates of IBDV-infected DF-1 cells with siRNA against VP3 were detected by immunoblotting. Figure 5I shows that VP3 knockdown blocked the IBDV-induced UBC9 reduction, indicating that IBDV-induced decrease of UBC9 was viral protein VP3 dependent. However, in quantitative real-time reverse transcriptase PCR (qRT-PCR) assays, we observed no obvious changes in UBC9 mRNA levels, although the API5 mRNA level was upregulated markedly following IBDV infection (Fig. 5J). These findings revealed that IBDV infection facilitates the degradation of UBC9 protein. Figure 5A to F show the interaction and colocalization of VP3 and UBC9. Thus, to further explore whether VP3 was a direct regulator that accelerated UBC9 degradation, lysates of DF-1 cells transfected with expression plasmid encoding VP3 or empty vector were subjected to immunoblotting. Figure 5K and L show that VP3 overexpression enhanced UBC9 degradation in a dose-dependent manner, and shortened the half-life of UBC9. These data demonstrated that accelerating UBC9 degradation might be another mechanism by which VP3 decreases API5 SUMOylation.
VP3 promotes UBC9 proteasome-dependent degradation by targeting ubiquitin E3 ligase TNF receptor-associated factor 3. K48-linked ubiquitination of substrate proteins commonly regulates protein stability (46). To explore the detailed molecular mechanism by which VP3 targets UBC9 for degradation, ubiquitination assays of UBC9  were carried out in HEK293T cells transfected with the indicated plasmids. We observed that VP3 overexpression greatly enhanced K48-linked ubiquitination of UBC9 (Fig. 6A), suggesting that VP3 might facilitate UBC9 degradation via the ubiquitin proteasomal system.
Tumor necrosis factor (TNF) receptor-associated factor 3 (TRAF3) is an important regulator of nuclear transcription factor-kappa B (NF-k B) and type I interferon activation (47), and it has been reported that the severe acute respiratory syndrome coronavirus (SARS-CoV) protein ORF3a activates the NLRP3 inflammasome by promoting TRAF3-mediated caspase recruitment domain ubiquitination (48). To confirm whether UBC9 degradation driven by IBDV was related to TRAF3, TRAF3 knockdown DF-1 cells (TRAF3-KD) were constructed using Cas9 enzyme (Fig. 6B to D) and infected with IBDV. Immunoblotting results showed that IBDV-induced UBC9 degradation was decreased in TRAF3-KD DF-1 cells compared with that in control cells (Fig. 6E). Moreover, in cells, SUMOylation assays showed that in comparison with control HEK293T cells, TRAF3 knockdown not only markedly decreased the ubiquitination of UBC9 but also inhibited the increase of VP3-mediated UBC9 ubiquitination (Fig. 6F and G), suggesting that TRAF3 was required for VP3 to promote UBC9 proteasomal degradation. To further confirm that the VP3-induced UBC9 degradation was proteasome dependent, we assessed the effect of the proteasome inhibitor MG132 on VP3-mediated UBC9 degradation in DF-1 cells. Figure 6H shows that MG132 treatment inhibited VP3-mediated degradation of UBC9. Collectively, these findings revealed that TRAF3 acts as a ubiquitin E3 ligase of UBC9 ubiquitination and that VP3 facilitates UBC9 proteasome-dependent degradation by recruiting TRAF3.
Inhibition of API5 on IBDV replication depends on its K404 SUMOylation. Previous study demonstrated that replication of influenza A virus can be suppressed by API5 (30). To evaluate the impacts of API5 on IBDV replication, API5 knockdown DF-1 cells (API5-KD) were generated ( Fig. 7A and B). The levels of viral protein and virus titers in WT and API5-KD cells with IBDV were determined by immunoblotting and 50% tissue culture infectious dose (TCID 50 ) assays, respectively. As shown in Fig. 7C and D, virus titer and viral protein level were increased in IBDV-infected API5-KD DF-1 cells compared with that in WT cells, indicating that API5 plays a role as a negative regulator in IBDV replication. Subsequently, to investigate if the inhibition of API5 on IBDV proliferation depended on SUMO3 modification of API5 at the K404 site, API5-KD DF-1 cells restored with API5-WT, API5-K404R, or empty vector were infected with IBDV. Results showed that levels of viral protein and virus titer were greatly decreased in API5-KD DF-1 cells restored with API5-WT compared with that in cells containing mutant API5-K404R or empty vector ( Fig. 7E and F), indicating that API5 K404 SUMO3 modification is essential for resisting IBDV infection. Altogether, these data indicated that the de-SUMOylation of API5 at K404 plays an important role in supporting IBDV replication.  (IFN-b) was reported to play a critical role in inhibiting viral replication (49). IBDV dsRNA could be sensed by MDA5 (cytoplasmic RIG-I-like receptor melanoma differentiation gene 5) to induce IFN-b expression (44,50,51). Therefore, we speculated that API5 might play an important role in regulating MDA5-induced IFN-b production. As shown in Fig. 8A, API5 significantly enhanced IFN-b activation triggered by MDA5 instead of MAVS (mitochondrial antiviral signaling protein) or TBK1 (TANK-binding kinase 1). Consistent with this, the mRNA transcript of IFNB1 was downregulated in IBDV-infected API5 knockdown cells compared with IBDV-infected wildtype cells (Fig. 8B), demonstrating that API5 significantly activated the MDA5-mediated IFN-b signaling pathway. Additionally, given that SUMOylation of substrates is involved in antiviral innate immunity (52-54) and API5 is a SUMOylated protein (Fig. 1), we speculated that the role of API5 on IBDV-driven IFN-b production might be relevant to its SUMOylation. As shown in Fig. 8C to E, in IBDV-infected API5 knockdown cells, the activity of IFN-b promoter and the levels of IFNB1 mRNA and phosphorylated TBK1 were significantly enhanced by restoring wild-type API5, but not API5K404R, suggesting that the SUMOylation of residue K404 of API5 is critical for upregulation of IFN-b expression. Overall, these results demonstrated that SUMOylation of API5 K404 contributed to IBDV-driven MDA5-dependent IFN-b production.

DISCUSSION
SUMOylation has been reported to have an important role in modulating protein function under conditions of physical or other types of stress (55). Growing evidence reveals that SUMOylation of substrates affects their subcellular distribution and activity (56). In addition, nuclear targeting proteins prefer to be attached to SUMO moieties (57,58). Antiapoptotic protein API5 was initially identified as a nuclear protein that protects cells from apoptosis and, although its lysine 251 acetylation is related to this antiapoptotic function, it was observed to be irrelevant to API5's subcellular localization (20,59,60). However, to date, SUMOylation of API5 has not been reported. In this study, we first identified API5 as a SUMOylated protein that is localized mainly in the nucleus and showed that the K404 residue can be conjugated to SUMO3. Interestingly, a SUMO3 modification-deficient API5 mutant failed to locate to the nucleus, implying that SUMO3 modification at K404 contributes to API5 nuclear localization. Similarly, phosphatase and tensin homolog and extracellular-signal-regulated kinase 5 proteins could be efficiently modified by SUMO, contributing to their nuclear localization, and their SUMOylation-deficient mutant variants were not retained in nuclei (61)(62)(63).
Previous studies have shown that viral proteins can interfere with the SUMO system by different mechanisms, including interacting with SUMO pathway proteins (15), mimicking SUMO E3 ligases (64)(65)(66), inhibiting SUMO modification of substrate proteins (67), targeting UBC9 (16), and exhibiting SUMO protease activity (68). Our findings demonstrated that Avibirnavirus infection reduces the SUMO3-conjugation of API5 by facilitating degradation of UBC9. In parallel to its deSUMOylation, Avibirnavirus infection induces API5 to colocalize with the viral protein VP3 in the cytoplasm. Further investigation of the molecular mechanism indicated that viral protein VP3 directly inhibits the SUMOylation of API5 by targeting API5 and promoting UBC9 proteasomedependent degradation through binding to the ubiquitin E3 ligase TRAF3. Thus, we propose that Avibirnavirus VP3 protein has a similar SUMO protease activity. As to whether viral protein VP3 is a new member of the SUMO-specific protease family, further investigation is required.
SUMOylation is a vital regulator of antiviral innate immunity (69,70). Increasing evidence demonstrates that many host proteins involved innate immunity can be SUMOylated (71). During infection, viruses ensure their replication and infection through inhibition or induction of SUMOylation of cellular proteins implicated in antiviral defense. Ebola Zaire virus VP35 protein increases SUMOylation of interferon regulatory factor 7 and represses type I IFN transcription, leading to the enhancement of virus infection (52). Vesicular stomatitis virus infection reduces SUMOylation of DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide 39A (DDX39A) to repress nuclear export of antiviral transcripts for limiting IFN-b production and promoting viral replication (53). In this study, we revealed that SUMOylated API5 played a role as a positive regulator in antiviral innate immunity, and that IBDV VP3 deSUMOylated API5 to block MDA5-dependent IFN-b production and thereby support viral proliferation (Fig. 4, Fig. 8). This is the first reveal of the mechanism by which IBDV VP3 negatively regulates MDA5-dependent antiviral innate immunity by API5 deSUMOylation.
Additionally, our experiments demonstrated that SUMOylated API5 mainly localizes in the nucleus (Fig. 2B and C) and the deSUMOylation of API5 by viral protein VP3 caused translocation to the cytoplasm (Fig. 3A). Further, IBDV infection upregulated the expression of API5 mRNA (Fig. 5J). Avibirnavirus VP3 protein has been reported to inhibit antiviral innate immunity by blocking viral double-stranded RNA binding to MDA5 (44), and API5 is a RNA-binding protein and promotes mRNA nuclear export (72). Considering that deSUMOylated API5 promotes IBDV replication, it needs further investigation as to whether the deSUMOylated API5 blocks the viral dsRNA binding to MDA5 for inhibiting IFN-b induction.
In summary, we first identified that API5 is a SUMOylated protein and that Lys404 is an important SUMO3 modification site. In addition, we showed that IBDV infection decreases the SUMOylation of API5 and that viral protein VP3 directly inhibits the SUMOylation of API5 to block IFN-b induction for supporting virus replication by targeting API5 and promoting UBC9 proteasome-dependent degradation via binding to TRAF3 (Fig. 9). Taken together, using Avibirnavirus IBDV as a model of a nonenveloped RNA virus, we revealed the fine-tuned regulation mechanism of the conversion between SUMOylation and deSUMOylation of API5 upon virus infection and provided new ideas for the design of anti-IBDV agents.
Co-IP and GST pulldown assays. For co-IP, cells were lysed in NP-40 buffer (P0013F; Beyotime, Shanghai, China) containing 1 mM PMSF for 4 h or overnight at 4°C. After centrifugation, the supernatant was incubated with the indicated antibodies and protein A/G PLUS-agarose at 4°C for 4 h or overnight. For the GST pulldown assay, purified GST and GST-VP3 protein were incubated separately with His-API5 at 4°C for 4 h and the GST resin was added. After centrifugation, the pellets were washed five times with NP-40 buffer at 4°C, and then lysed for immunoblotting analysis.
In vitro deSUMOylation. API5-SUMO3 obtained in an in vitro SUMOylation reaction was incubated with recombinant VP3 in reaction buffer (50 mM Tris [pH 7.5], 2 mM MgCl 2 , and 5 mM b-mercaptoethanol) at 37°C for 1 h. Finally, the reaction was stopped by the addition of 20 ml of SDS sample buffer containing b-mercaptoethanol. The prepared samples were subjected to SDS-PAGE and immunoblotting.
Ubiquitination assay. HEK293T cells transfected with the indicated plasmids were treated with MG132 (S1748-5mg; Beyotime) (10 mM) for 12 h before harvesting the samples. The supernatant of the cellular lysates was subjected to co-IP assays.
Virus titer detection. The harvested cells were freeze-thawed three times and, after centrifugation (12 000 Â g), the supernatants were subjected to virus titer detection as described previously (35).
Quantitative real-time reverse transcription PCR (qRT-PCR). Total RNA was isolated from DF-1 cells or BF using the TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Then, 1 mg RNA was reverse transcribed into cDNA using the PrimeScript RT reagent kit with gDNA Eraser (RR047A; TaKaRa, Kusatsu, Japan) according to the manufacturer's instructions. The mRNAs of indicated genes were amplified using TB Green Premix Ex Taq (Tli RNaseH Plus) (RR420A; TaKaRa). GAPDH was amplified to normalize the relative abundance of the indicated mRNAs. The primers for qRT-PCR are listed in Table 2.
Statistical analysis. Statistical differences were assessed using one-way analysis of variance (ANOVA) using SPSS Statistics 20 (IBM Corp., Armonk NY, USA). The results of all statistical analysis are shown as the mean 6 SD (standard deviation) of at least two independent experiments. For all experiments, P , 0.05 was considered statistically significant. *, P , 0.05; **, P , 0.01; ***, P , 0.001; ns, P . 0.05.

ACKNOWLEDGMENTS
This study was supported by grants from the National Natural Science Foundation of China (grant no. 31630077) and the China Agriculture Research System of MOF and MARA (grant no. CARS-40-K13).