Host BAG3 Is Degraded by Pseudorabies Virus pUL56 C-Terminal 181L-185L and Plays a Negative Regulation Role during Viral Lytic Infection

Bcl2-associated athanogene (BAG) 3, which is a chaperone-mediated selective autophagy protein, plays a pivotal role in modulating the life cycle of a wide variety of viruses. Both positive and negative modulations of viruses by BAG3 were reported. However, the effects of BAG3 on pseudorabies virus (PRV) remain unknown. To investigate whether BAG3 could modulate the PRV life cycle during a lytic infection, we first identified PRV protein UL56 (pUL56) as a novel BAG3 interactor by co-immunoprecipitation and co-localization analyses. The overexpression of pUL56 induced a significant degradation of BAG3 at protein level via the lysosome pathway. The C-terminal mutations of 181L/A, 185L/A, or 181L/A-185L/A in pUL56 resulted in a deficiency in pUL56-induced BAG3 degradation. In addition, the pUL56 C-terminal mutants that lost Golgi retention abrogated pUL56-induced BAG3 degradation, which indicates a Golgi retention-dependent manner. Strikingly, BAG3 was not observed to be degraded in either wild-type or UL56-deleted PRV infected cells as compared to mock infected ones, whereas the additional two adjacent BAG3 cleaved products were found in the infected cells in a species-specific manner. Overexpression of BAG3 significantly suppressed PRV proliferation, while knockdown of BAG3 resulted in increased viral yields in HEK293T cells. Thus, these data indicated a negative regulation role of BAG3 during PRV lytic infection. Collectively, our findings revealed a novel molecular mechanism on host protein degradation induced by PRV pUL56. Moreover, we identified BAG3 as a host restricted protein during PRV lytic infection in cells.


Introduction
Pseudorabies virus (PRV), which is one of the most important swine pathogens, belongs to the Alphaherpesvirinae subfamily and contains a large (approximately 150 kb) double-stranded DNA genome. The virion is composed of a complex structure including a linear genome surrounded by nucleocapsid, proteinaceous tegument, and a lipid bilayer envelope from the inner side to the outer side, in addition to some host proteins (e.g., Rab GTPases) [1,2]. During the lytic infection of alpha-herpesviruses, the interplay between virus and host proteins contributes to a diversity of influences on the viral life cycle (e.g., viral entry, replication, envelope, and egress) [2,3].
Host BCL-2-associated athanogene (BAG) 3 is a co-chaperone protein of the BAG protein family, and was first isolated and identified as a Bcl-2 binding protein [4,5]. The family members share a highly dependence of a specific Golgi retention of pUL56, which indicates a novel molecular mechanism. Unexpectedly, this degradation was not observed in PRV infection assays using either wild-type (WT) or UL56-deleted (∆UL56) PRV. Two BAG3 cleaved products were found in WT and ∆UL56 PRV infected cells in a species-specific manner. Overexpression and knockdown assays indicated that BAG3 played a negative regulation role during the PRV lytic infection in HEK293T cells.
We first examined whether pUL56 could interact with BAG3. The plasmids expressing Flag-pUL56 and HA-mouse BAG3 (mBAG3) were separately transfected into human embryonic kidney 293T (HEK293T) cells. At 48 h post transfection (hpt), the whole cell lysates were collected and mixed with FLAG M2 beads at 4 • C for 6 h. The co-immunoprecipitation (co-IP) result showed that HA-mBAG3 was specifically immunoprecipitated with Flag-pUL56, which indicates an interaction between pUL56 and BAG3 ( Figure 1A). We continued to examine the co-localization between pUL56 and endogenous (endo-) BAG3 using a specific anti-BAG3 polyclonal antibody ( Figure 1B). An evenly cytoplasmic dot-like distribution of BAG3 was observed, which was consistent with previous observations regarding the BAG3 distribution pattern in both cell cultures and tissues ( Figure 1C) [23][24][25]. The representative confocal micrographs showed that endo-BAG3 intensively co-localized with GFP-pUL56, but not with green fluorescent protein (GFP) in the perinuclear region of transfected Vero cells ( Figure 1C). Notably, the co-localization was nearly 100% observed in the GFP-pUL56 expressed cells. 293T (HEK293T) cells. At 48 h post transfection (hpt), the whole cell lysates were collected and mixed with FLAG M2 beads at 4 °C for 6 h. The co-immunoprecipitation (co-IP) result showed that HA-mBAG3 was specifically immunoprecipitated with Flag-pUL56, which indicates an interaction between pUL56 and BAG3 ( Figure 1A). We continued to examine the co-localization between pUL56 and endogenous (endo-) BAG3 using a specific anti-BAG3 polyclonal antibody ( Figure 1B). An evenly cytoplasmic dot-like distribution of BAG3 was observed, which was consistent with previous observations regarding the BAG3 distribution pattern in both cell cultures and tissues ( Figure 1C) [23][24][25]. The representative confocal micrographs showed that endo-BAG3 intensively co-localized with GFP-pUL56, but not with green fluorescent protein (GFP) in the perinuclear region of transfected Vero cells ( Figure 1C). Notably, the co-localization was nearly 100% observed in the GFP-pUL56 expressed cells. Furthermore, the expression of BAG3 was detected in the mouse nervous tissues dorsal root ganglia (DRGs), spinal cord, and brain, which are the target organs infected by PRV ( Figure 1D). This suggested a possible interplay between BAG3 and PRV in the nervous system. and BAG3. Each of the indicated plasmids is transfected into HEK293T cells, respectively. At 48 hpt, the whole cell lysates are obtained and mixed with the indicated combinations and immunoprecipitated with FLAG M2 beads. A 20% aliquot of whole cell lysates is used to confirm protein expression. The asterisk indicates a IgG light chain in the co-IP assay.(B) HEK293T cells in a 12-well plate are transfected with 120 pmol of the indicated siRNAs targeting to BAG3 for 48 h. The specificity of the anti-BAG3 polyclonal antibody is confirmed by a Western blot. (C) A co-localization analysis of endogenous BAG3 with GFP-pUL56 in the co-transfected Vero cells. BAG3 distributes evenly in the cytoplasm (asterisk, upper panel) in non-transfected Vero cells. The distribution of BAG3 is not affected by GFP (arrowheads, upper panel), whereas it is intensively restricted and colocalized with GFP-pUL56 in the perinuclear region (arrowheads, lower panel). (D) BAG3 can be detected in both the peripheral (dorsal root ganglion, DRG) and central (spinal cord and brain) nervous system in C57/BL6 mice. The scale bar indicates 10 μm (C). and BAG3. Each of the indicated plasmids is transfected into HEK293T cells, respectively. At 48 hpt, the whole cell lysates are obtained and mixed with the indicated combinations and immunoprecipitated with FLAG M2 beads. A 20% aliquot of whole cell lysates is used to confirm protein expression. The asterisk indicates a IgG light chain in the co-IP assay. (B) HEK293T cells in a 12-well plate are transfected with 120 pmol of the indicated siRNAs targeting to BAG3 for 48 h. The specificity of the anti-BAG3 polyclonal antibody is confirmed by a Western blot. (C) A co-localization analysis of endogenous BAG3 with GFP-pUL56 in the co-transfected Vero cells. BAG3 distributes evenly in the cytoplasm (asterisk, upper panel) in non-transfected Vero cells. The distribution of BAG3 is not affected by GFP (arrowheads, upper panel), whereas it is intensively restricted and co-localized with GFP-pUL56 in the perinuclear region (arrowheads, lower panel). (D) BAG3 can be detected in both the peripheral (dorsal root ganglion, DRG) and central (spinal cord and brain) nervous system in C57/BL6 mice. The scale bar indicates 10 µm (C). Furthermore, the expression of BAG3 was detected in the mouse nervous tissues dorsal root ganglia (DRGs), spinal cord, and brain, which are the target organs infected by PRV ( Figure 1D). This suggested a possible interplay between BAG3 and PRV in the nervous system.

PRV pUL56 Re-Localized BAG3 to Golgi
We recently indicated that PRV pUL56 primarily localized to the Golgi complex and tans-Golgi network (TGN) [13]. Thus, we attempted to examine whether the interaction of pUL56 and BAG3 could specifically target BAG3 to the Golgi-TGN. HEK293T cells were co-transfected with plasmids pCMV-HA-mBAG3 and pAcGFP-pUL56 or pAcGFP. At 24 hpt, the transfected cells were fixed and stained with an anti-GM130 antibody. In the pCMV-HA-mBAG3 transfected cells, HA-mBAG3 was observed to have a cytoplasmic distribution without specific organelle localization ( Figure 3A). However, HA-mBAG3 intensively localized to the Golgi upon co-expression with GFP-pUL56. In addition, some HA-mBAG3 was also anchored to the nuclear membrane by GFP-pUL56 ( Figure 3B). Notably, the co-localization of HA-mBAG3, GFP-pUL56, and GM130 was nearly 100% observed in the co-expressed cells. The co-localization analysis shows that pUL56 PYM1−3 co-localize with BAG3 in the transfected Vero cells, respectively. The pUL56 PYM4 has no apparent co-localization with BAG3 in the transfected Vero cells. The inset enlargements show the co-localized fluorescence in the cells. The arrowheads indicate co-localization. (C) Co-IP of BAG3 and pUL56 PYM1−4, respectively. A 20% aliquot of the whole cell lysates is used to confirm protein expression. The scale bar indicates 20 µm (B).

PRV pUL56 Re-Localized BAG3 to Golgi
We recently indicated that PRV pUL56 primarily localized to the Golgi complex and tans-Golgi network (TGN) [13]. Thus, we attempted to examine whether the interaction of pUL56 and BAG3 could specifically target BAG3 to the Golgi-TGN. HEK293T cells were co-transfected with plasmids pCMV-HA-mBAG3 and pAcGFP-pUL56 or pAcGFP. At 24 hpt, the transfected cells were fixed and stained with an anti-GM130 antibody. In the pCMV-HA-mBAG3 transfected cells, HA-mBAG3 was observed to have a cytoplasmic distribution without specific organelle localization ( Figure 3A). However, HA-mBAG3 intensively localized to the Golgi upon co-expression with GFP-pUL56. In addition, some HA-mBAG3 was also anchored to the nuclear membrane by GFP-pUL56 ( Figure 3B). Notably, the co-localization of HA-mBAG3, GFP-pUL56, and GM130 was nearly 100% observed in the co-expressed cells.

BAG3 was Downregulated by the pUL56 C-Terminus
To investigate whether the expression of BAG3 could be regulated by pUL56, 1 μg pAcGFP and pAcGFP-pUL56 was transfected into HEK293T cells, respectively. At 48 hpt, the protein levels of endo-BAG3 were determined by a Western blot. The relative level of endo-BAG3/β-actin was significantly downregulated in GFP-pUL56 as compared to GFP or non-transfected cells ( Figure 4A).
Given an important role of the pUL56 C-terminus in Golgi−TGN targeting and retention [13], it was essential to clarify the relationship between specific Golgi localization and BAG3 degradation induced by pUL56. Subsequently, the pCMV-HA-mBAG3 was co-transfected with pAcGFP,

BAG3 Was Downregulated by the pUL56 C-Terminus
To investigate whether the expression of BAG3 could be regulated by pUL56, 1 µg pAcGFP and pAcGFP-pUL56 was transfected into HEK293T cells, respectively. At 48 hpt, the protein levels of endo-BAG3 were determined by a Western blot. The relative level of endo-BAG3/β-actin was significantly downregulated in GFP-pUL56 as compared to GFP or non-transfected cells ( Figure 4A).
Given an important role of the pUL56 C-terminus in Golgi−TGN targeting and retention [13], it was essential to clarify the relationship between specific Golgi localization and BAG3 degradation induced by pUL56. Subsequently, the pCMV-HA-mBAG3 was co-transfected with pAcGFP, pAcGFP-pUL56, or pAcGFP-pUL56 M13 (encoding a pUL56 C-terminal mutant that has lost Golgi localization and retention), respectively ( Figure 4D) [13]. At 48 hpt, the relative levels of HA-mBAG3/β-actin were determined in each group. Strikingly, the mutant GFP-pUL56 M13 could not mediate BAG3 downregulation. This result suggested that the mutant amino acids or/and Golgi retention could be the key factors affecting pUL56-induced BAG3 downregulation ( Figure 4E). pAcGFP-pUL56, or pAcGFP-pUL56 M13 (encoding a pUL56 C-terminal mutant that has lost Golgi localization and retention), respectively ( Figure 4D) [13]. At 48 hpt, the relative levels of HA-mBAG3/β-actin were determined in each group. Strikingly, the mutant GFP-pUL56 M13 could not mediate BAG3 downregulation. This result suggested that the mutant amino acids or/and Golgi retention could be the key factors affecting pUL56-induced BAG3 downregulation ( Figure 4E). . Co-transfection of pCMV-HA-mBAG3 with pAcGFP-UL56 S1, S2 and S3 into HEK293T cells, respectively. At 48 hpt, HA-mBAG3 is significantly downregulated in pAcGFP-UL56 S3, but not in S1 or S2 transfected cells as compared to the empty vector. (D) A pUL56 C-terminal mutant (M13) that loses Golgi localization in Vero cells. (E) Co-transfection of pCMV-HA-mBAG3 with pAcGFP-UL56 M13 does not affect relative HA-mBAG3/β-actin levels as compared to empty vector (right panel). The statistical analyses are performed using an ordinary one-way ANOVA post-Tukey's multiple comparisons test (**: p < 0.01, ***: p < 0.001, ns: no statistical significance). Each statistical data represents three independent experiments. Scale bar indicates 10 μm (D).

BAG3 Degradation Was Induced by pUL56 C-Terminal 181 L-185 L via the Lysosome Pathway in a Golgi Retention-Dependent Manner
The pUL56 C-terminal mutants (M1−12) were employed for further investigations of a precise molecular mechanism underlying the disrupted BAG3 degradation induced by pUL56 M13 ( Figure 5A) [13]. Western blot results showed that HA-mBAG3 was significantly downregulated by WT, M1−2 and M5−9, but not by M3−4 or M10−12 ( Figure 5B). Thus, these data indicated that the 174 P/A-177 L/A-181 L/A-185 L/A mutations were crucial in pUL56-induced BAG3 degradation. In addition, Golgi retention was also required because pUL56-induced BAG3 degradation was fully interrupted upon the loss of Golgi localization in pUL56 mutants M10−13. An unexpected finding was that HA-mBAG3 degradation was restored in pUL56 M5−9 as compared to M3 and M4 ( Figure 5B).
To clarify the cellular degradation pathway involved in pUL56-induced BAG3 degradation, the lysosome and proteasome pathways were inhibited by MG132 and Chloroquine (CQ). Western blot results indicated that the lysosome pathway, but not the proteasome pathway, was responsible for pUL56-induced BAG3 degradation ( Figure 5F). Taken together, we revealed a novel molecular mechanism in pUL56-induced host protein degradation, which occurred at the Golgi prior to proteins being transported into the cytoplasm ( Figure 5G).

Endo-BAG3 Was Cleaved by PRV Infection in a Species-Specific Manner
Furthermore, we tested whether pUL56-induced BAG3 degradation also occurred in PRV infected cells. The cells were infected with WT or ∆UL56 PRV at the multiple of infection (MOI) of 0.05, respectively. At 24 h post infection (hpi), the relative endo-BAG3/β-actin levels were found to have no differences between mock and infection groups ( Figure 6). Two adjacent BAG3 cleaved products with molecular weight around 60 kD were observed in PRV infected Vero and BHK21 cells. However, the BAG3 cleaved products were not found in HEK293T, PC12, or mouse embryonic fibroblast (MEF) cells ( Figure 6). Moreover, this phenotype was not affected by pUL56 deficiency. Taken together, these data suggested that PRV-induced BAG3 cleavage had a species-dependent character, which was independent of pUL56.
To clarify the cellular degradation pathway involved in pUL56-induced BAG3 degradation, the lysosome and proteasome pathways were inhibited by MG132 and Chloroquine (CQ). Western blot results indicated that the lysosome pathway, but not the proteasome pathway, was responsible for pUL56-induced BAG3 degradation ( Figure 5F). Taken together, we revealed a novel molecular mechanism in pUL56-induced host protein degradation, which occurred at the Golgi prior to proteins being transported into the cytoplasm ( Figure 5G). have no differences between mock and infection groups ( Figure 6). Two adjacent BAG3 cleaved products with molecular weight around 60 kD were observed in PRV infected Vero and BHK21 cells. However, the BAG3 cleaved products were not found in HEK293T, PC12, or mouse embryonic fibroblast (MEF) cells ( Figure 6). Moreover, this phenotype was not affected by pUL56 deficiency. Taken together, these data suggested that PRV-induced BAG3 cleavage had a species-dependent character, which was independent of pUL56.

BAG3 Played a Negative Regulation Role during PRV Lytic Infection
To examine whether BAG3 overexpression could influence PRV proliferation, we transfected HEK293T cells with plasmid expressing HA-mBAG3 at different amounts (0.5, 1, and 2 µg). This was followed by infection with 0.05 MOI of WT and ∆UL56 PRV, respectively. At 24 hpi, the BAG3 and PRV gB proteins were detected by using a Western blot. The results showed overexpression of HA-mBAG3 at higher levels significantly suppressed the PRV gB protein level as compared to the non-transfection group in both WT and ∆UL56 infected cells ( Figure 7A). The intracellular viral titers of WT were examined in HA-mBAG3 overexpressed and mock transfected groups at 12 and 24 hpi, respectively. The results indicated significantly lower viral titers were observed in 2 µg than 0.5 µg pCMV-HA-mBAG3 transfected or mock transfected cells at 24 hpi, which indicated a negative regulation role of BAG3 during PRV infection ( Figure 7B). mBAG3 at higher levels significantly suppressed the PRV gB protein level as compared to the nontransfection group in both WT and ΔUL56 infected cells ( Figure 7A). The intracellular viral titers of WT were examined in HA-mBAG3 overexpressed and mock transfected groups at 12 and 24 hpi, respectively. The results indicated significantly lower viral titers were observed in 2 μg than 0.5 μg pCMV-HA-mBAG3 transfected or mock transfected cells at 24 hpi, which indicated a negative regulation role of BAG3 during PRV infection ( Figure 7B).  To further confirm this finding, we performed an RNAi experiment to examine if the downregulation of BAG3 could promote PRV proliferation. The siRNA 1659 specific for BAG3 was transfected into HEK293T cells for 48 h, which is followed by infection with WT and ∆UL56 PRV at an MOI of 0.05, respectively. At 24 hpi, the BAG3 and PRV gB proteins were detected by a Western blot. As expected, the BAG3 levels were significantly lower in siBAG3 than siNC transfected groups. The gB protein levels were significantly higher in siBAG3 than siNC transfected groups ( Figure 8A). The intracellular viral titers of WT were significantly increased in siBAG3 than siNC transfected groups ( Figure 8B). However, no significant differences were observed in viral titers at 12 hpi in either overexpression or knockdown assays (Figures 7 and 8). Collectively, these results supported a negative role of BAG3 during PRV lytic infection. To further confirm this finding, we performed an RNAi experiment to examine if the downregulation of BAG3 could promote PRV proliferation. The siRNA 1659 specific for BAG3 was transfected into HEK293T cells for 48 h, which is followed by infection with WT and ΔUL56 PRV at an MOI of 0.05, respectively. At 24 hpi, the BAG3 and PRV gB proteins were detected by a Western blot. As expected, the BAG3 levels were significantly lower in siBAG3 than siNC transfected groups. The gB protein levels were significantly higher in siBAG3 than siNC transfected groups ( Figure 8A). The intracellular viral titers of WT were significantly increased in siBAG3 than siNC transfected groups ( Figure 8B). However, no significant differences were observed in viral titers at 12 hpi in either overexpression or knockdown assays (Figures 7 and 8). Collectively, these results supported a negative role of BAG3 during PRV lytic infection.

Discussion
Viruses have evolved various strategies to modulate and further utilize host protein machineries for their efficient replication and assembly. The modulations generally take place via the interplay between virus and host proteins, and, thus, become potential targets for designing antiviral tools.
BAG3 is a stressor-response gene that has been reported to be modulated by a series of stressinducers including high temperature, heavy metals, oxidative stress, and viral infection [12,[26][27][28]. Host co-chaperone protein BAG3 can be regulated by a variety of viruses at the gene or posttranscriptional levels, which produces significant effects on the viral life cycle. For example, the expression of BAG3 can be inhibited by polyomavirus JC (JCV), which is a human neurotropic virus, as a result of suppressed transcription of the BAG3 promoter through physical interaction with the T antigen in the infected glial cells or transfected human glioblastoma cells [29]. In turn, an overexpression of BAG3 suppresses the replication of JCV [30]. Conversely, the efficient replication in some viruses, e.g., VZV and SARS-CoV, relies on the presence of BAG3 [9,12]. During viral (B) HEK293T cells grown in a 12-well plate are transfected with siBAG3 or siNC at a final amount of 120 pmol for 48 h, which is followed by infection with WT PRV and MOI of 0.05. The intracellular viral titers are determined by TCID 50 in Vero cells at 12 and 24 hpi, respectively. The statistical analysis is performed using two-way ANOVA post Sidak's multiple comparison test (***: p < 0.001, ns: no statistical significance) and it represents three independent experiments for titer determination.

Discussion
Viruses have evolved various strategies to modulate and further utilize host protein machineries for their efficient replication and assembly. The modulations generally take place via the interplay between virus and host proteins, and, thus, become potential targets for designing antiviral tools.
BAG3 is a stressor-response gene that has been reported to be modulated by a series of stressinducers including high temperature, heavy metals, oxidative stress, and viral infection [12,[26][27][28]. Host co-chaperone protein BAG3 can be regulated by a variety of viruses at the gene or post-transcriptional levels, which produces significant effects on the viral life cycle. For example, the expression of BAG3 can be inhibited by polyomavirus JC (JCV), which is a human neurotropic virus, as a result of suppressed transcription of the BAG3 promoter through physical interaction with the T antigen in the infected glial cells or transfected human glioblastoma cells [29]. In turn, an overexpression of BAG3 suppresses the replication of JCV [30]. Conversely, the efficient replication in some viruses, e.g., VZV and SARS-CoV, relies on the presence of BAG3 [9,12]. During viral infection, the SARS-CoV significantly upregulates BAG3, but HSV1 does not substantially alter BAG3 within 12 h at the protein level [9,10]. Thus, these findings suggest that BAG3 can be either a positive or a negative regulator during viral infection.
Previous studies showed that the WW-domain bearing BAG3 is an interactor of the PPxY motif containing viral proteins, which provides a molecular basis for screening for BAG3 interacting viral proteins [7,11,31]. Based on this protein−protein interaction module, pUL56 was first identified to be a BAG3 interacting PRV protein in this study.
Although significant amino acid differences are present among pUL56 homologs in alphaherpesviruses (e.g., PRV vs. HSV1/2), they share many similar biological characteristics, such as Golgi-TGN localization, C-terminal transmembrane helix, and being a viral neurovirulence-associated factor [13,15,17,32]. In addition, the alpha-herpes viral pUL56 can induce a range of host protein (e.g., Nedd4, Itch, SLA1, and MHC-I) degradation through conserved PPxY motifs [16,[18][19][20]33]. Differently, our findings indicated that pUL56-induced BAG3 degradation was independent of the PPxY motif ( Figure 4B). Further evidence indicated that the C-terminus was responsible for pUL56-induced BAG3 degradation, which suggests a novel molecular mechanism. These results combined with our previous finding showing pUL56-induced non-interacted Rab6a degradation strongly supported that the interaction between pUL56 and its degraded protein was not required for pUL56 C-terminal-induced protein degradation [13]. Thus, it is impossible that the degraded BAG3 was transported as a cargo by interacting with pUL56 through vesicles to the degradation pathway. Given an important role of the pUL56 C-terminal leucine-rich transmembrane helix in Golgi retention [13], we, therefore, asked if pUL56-induced BAG3 degradation was in a specific Golgi localization and retention dependent manner. Using WT and sixteen pUL56 C-terminal mutants that retained or lost Golgi localization [13], we found that pUL56-induced BAG3 degradation was strongly correlated with the Golgi retention of pUL56. Of importance, the 181 L-185 L was responsible for the Golgi-resident pUL56-induced BAG3 degradation. In addition, the revertible BAG3 degradation induced by pUL56 M5-9 as compared to M3 and 4 remains to be clarified by using additional methodological approaches in the future work. Throughout this study, we used a series of pUL56 and its mutant overexpression assays to confirm the key residues contributing to the pUL56-induced BAG3 degradation. To our knowledge, there have been no papers showing an overexpression of alpha-herpes viral pUL56 could lead to cell apoptosis, cyctoxicity, or other unwarranted effects yet. Based on our data, we hypothesize that the abundant leucine at the C-terminus of pUL56 mutants could re-form the degradation-related sorting signal, which likely resembles 181 L-185 L in WT, via conformational contact.
An unexpected observation was that BAG3 degradation could not be observed in the PRV infected cells. This suggests that potentially undefined mechanism does exist in antagonizing pUL56-induced BAG3 degradation during viral infection. However, two BAG3 cleaved products with similar molecular weight were observed in PRV infected cells, and this phenotype occurred in Vero (monkey) and BHK-21 (hamster) cells, but not in HEK293T (human), PC12 (rat), or MEF (mouse) cells, which suggests a species-specific manner. The cleavage of BAG3 was reported in a previous study, which showed a correlation with the apoptosis process in pancreatic cells [34]. The authors identified the C-terminal Asp 347 (KEVD347) as the cleaved site. The cleaved BAG3 has a molecular weight that seems to be much smaller than the ones observed in this study. Thus, the observed BAG3 cleaved products are likely derived from the N terminus in this study. In further work, the function of BAG3 cleaved products on PRV life cycle still needs to be clarified.
In the performed infection assays, HEK293T cells were employed to investigate the effects of BAG3 overexpression or knockdown on lytic infection of PRV. We used HEK293T cells because: (i) Human cell lines are susceptible to PRV, and support the efficient replication of PRV [35,36], (ii) HEK293T cells are highly efficient for overexpression of foreign protein by the transfected plasmid, and (iii) PRV has been reported in several recent case reports to infect humans, which results in encephalitis [37][38][39][40][41][42]. Both overexpression and knockdown assays indicated that BAG3 played a negative regulation role during PRV lytic infection. This conclusion was drawn by two findings: (i) Overexpression of BAG3 at an original transfection of 2 µg pCMV-HA-mBAG3 significantly decreased viral gB level and titer as compared to non-transfected cells at 24 hpi, (ii) knockdown of BAG3 significantly increased viral gB level and titer as compared to siNC transfected cells at 24 hpi. In an earlier study, the BAG3 was found to be a positive regulator during HSV1 infection through enhanced ICP0 activity [10]. This may also be true in PRV because the original transfection of 0.5 µg pCMV-HA-mBAG3 did not affect viral gB level ( Figure 7A and B) and viral titer ( Figure 7C) as compared to non-transfection cells. However, the suppression role of BAG3 could be observed in the original transfection of 1 or 2 µg pCMV-HA-mBAG3 as compared to the non-transfection group. It is also interesting to use an immediate early gene IE180 deficient mutant to investigate if the positive regulation of BAG3 presents during PRV infection.

Cells and Viruses
HEK293T, PC12, BHK-21, and MEF cells were obtained from the American Type Culture Collection (ATCC). Vero cells were described in our previous study [43]. PC12 cells were differentiated in Dulbecco's Modified Eagle's Medium (DMEM; Thermo Fisher Scientific, Beijing, ChinaH supplemented with 1% horse serum and nerve growth factor (PEPROTECH, USA) at a concentration of 100 ng/mL. Differentiation medium was replaced every third day for 11 days before viral infection [44]. All cell lines were cultured in DMEM supplemented with 10% (v/v) heat-inactivated fetal bovine serum (Thermo Fisher Scientific, Beijing, China) and 1% penicillin-streptomycin, and cultured at 37 • C in a 5% CO 2 environment. The WT and ∆UL56 PRV HeN1 strains (GenBank Accession No. KP098534) were propagated in Vero cells.
The UL56 mutants, PYM1−4, were amplified by overlapping extension PCR using mutant site-induced primers shown in Table 2, and subsequently cloned into vectors pAcGFP-C1 or p3×Flag, respectively. All constructs were validated by sequencing.

Generation of ∆UL56 PRV Using the CRISPR/Cas9 System
The use of the CRISPR/Cas9 system for the PRV double-stranded DNA genome editing was originally described in an earlier study [45]. To remove the UL56 gene from the PRV genome, double sgRNAs targeting to 3 -and 5 -ends of the UL56 gene were designed using an online CRISPR design tool (http://crispr.mit.edu/) (Supplementary Figure S2A). The plasmids expressing sgRNAs were constructed by inducing synthesized oligonucleotides into a Bbs I digested vector pX330 (Addgene) ( Table 3). The recombinant plasmids pX330-sgRNA1 and 2 were validated by sequencing. PRV gene editing induced by the CRISPR/Cas9 system requires the co-existence of the viral genome and CRISPR/Cas9 components [45]. Despite swine being the natural host of PRV, the human receptor nectin-1 recognized by PRV gD for cellular entry was found to have comparable efficiency with that of swine in PRV-mediated cell fusion [35]. Thus, the human cell line HEK293T was found to be susceptible for PRV, and was, therefore, used for pX330-sgRNA transfection to elevate the levels of intracellular sgRNAs. HEK293T cells grown to approximately 80% confluence in a 12-well plate were co-transfected with pX330-sgRNA1 and 2 (1 µg for each). At 24 hpt, the cells were infected with 225 plaque forming units (PFUs) of PRV. The supernatant was collected when approximately 50% of a cytopathic effect was observed. The obtained cell supernatant was serially diluted from 1 × 10 −1 to 1 × 10 −4 . The 1 × 10 −4 cell supernatant was inoculated into 100% confluent Vero cells in a 10-cm cell culture dish for viral plaque purification. The genome of the purified virus clone was extracted using an EasyPure Viral DNA/RNA Kit (TransGen Biotech, Beijing, China), according to the manufacturer's instructions. Lastly, a pair of detection primers (UL56 KO F&R, Table 1) was used to verify the removal of the UL56 gene by PCR ( Figure S2B,C). The obtained ∆UL56 clone was subjected to three rounds of plaque purification to ensure that WT viruses were excluded. The deletion sites were identified by sequencing (Supplementary Figure S2D).

Co-IP
HEK293T cells grown to approximately 80% confluence in six-well plates were transiently transfected with the indicated plasmids (2 µg for each), using the X-tremeGENE HP DNA Transfection Reagent (06366236001, Sigma-Aldrich), according to the manufacturer's instructions. At 48 hpt, the transfected cells were harvested and lysed in RIPA lysis buffer (50 mM Tris, 150 mM NaCl, 1% NP40, 0.25% sodium deoxycholate, Beyotime Biotechnology, Shanghai, China) containing 1 mM PMSF and a complete protease inhibitor cocktail (Roche Diagnostics Gmb H, Mannheim, Germany) for 1 h on ice. Cell lysates were centrifuged at 12,000× g for 10 min at 4 • C, and the obtained supernatants were used for immunoprecipitation with FLAG M2 beads (Sigma-Aldrich), according to the manufacturer's instructions. Briefly, the supernatants were incubated with lysis buffer pre-washed FLAG M2 beads and incubated for 6 h under continuous shaking at 4 • C. The beads were rinsed three times with cold phosphate-buffered saline (PBS) (pH 7.4), and the immunoprecipitated proteins were separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) followed by a Western blot analysis.

Viral Stocks and Titers
Viral titers were determined by a PFU assay [46] or TCID 50 , according to the Reed-Muench method [47]. The PFU assay was performed as described in our previous study [46]. The HEK293T cells pretreated with the overexpression or knockdown of BAG3 were infected with 0.05 MOI WT and ∆UL56 PRV, respectively. At the indicated time points, the infected intracellular components were collected, and then subjected to one freeze-thaw cycle and centrifuged at 3000× g for 5 min to obtain the viral stock.

Animals and Tissue Preparation
Three six-week-old C57BL/6 mice were sacrificed, and the DRGs, spinal cord, and brain were dissected, and immediately put on dry ice. The samples were separately pooled and placed in lysis buffer (R0010, RIPA, Solarbio) containing 1 mM PMSF and a complete protease inhibitor cocktail, and were then properly sonicated on ice. The lysates were centrifuged at 12,000× g for 10 min at 4 • C. The supernatants were collected and the protein concentration was determined by a BCA protein assay kit (Thermo Fisher Scientific, Beijing, China). Laemmli samples containing approximate 20 µg protein were loaded in each lane and then subjected to a Western blot analysis. The animal experiment was approved by The Animal Ethics Committee of the Harbin Veterinary Research Institute (HVRI), Chinese Academy of Agricultural Sciences (CAAS) (Heilongjiang-SYXK-2006-032, 17 December 2017) approved the animal experiment. The animal experiment was performed in accordance with animal use ethical guidelines (American Psychological Association, APA).