To better understand the importance of ICP0 in HSV-1 neuronal replication, we first examined replication of an ICP0-null virus, 7134, relative to its rescued derivative, 7134R in cultured primary neurons isolated from mouse TG. The isolation was performed with density gradient centrifugation following dissociation of the tissues to obtain relatively pure neurons (Katzenell et al., 2017). Viral titers were determined by plaque assays in human bone osteosarcoma (U2OS) cells, which are permissive for ICP0-null virus replication. After infection of the primary neurons, the supernatant titers of 7134 virus were ∼2 logs lower than those of 7134R virus at both MOIs of 1 and 10 from 24 to 72 h post-infection (hpi) (Figure 1A). For proteomic analysis, which requires a large number of cells, we considered using Neuro-2a cells derived from mouse brain neuroblastoma cells because results from these cells could often be recapitulated in mouse models (Pan et al., 2014; Sun et al., 2021). After infection of Neuro-2a cells, 7134 virus replicated to ∼2 logs lower titers than 7134R at MOIs of 0.1 and 1 at 18 hpi (Figure 1B). However, the defect of 7134 virus was modest at the high MOI of 10 in agreement with the MOI dependence of ICP0 effects commonly observed in cell lines (Cai and Schaffer, 1992; Everett et al., 2004). Under the same conditions, we also compared 7134 and 7134R in HFF (human foreskin fibroblast) and 293T (human embryonic kidney) cells. The defects of 7134 virus were more severe in HFF cells but moderate in 293T cells consistent with the known cell type–specific role of ICP0. These results substantiated the importance of ICP0 and suggested that Neuro-2a cells could be used to investigate the important interactions with ICP0 in neuronal cells.

To facilitate interactome analysis in infected cells, we constructed Flag-ICP0 virus that expresses N-terminal Flag-tagged ICP0 using bacterial artificial chromosome (BAC) technology on the basis of WT BAC virus described previously (here referred to as WT virus) (Figure 2A; Pan and Coen, 2012). The Flag-tag expressing sequence was inserted into both copies of the ICP0 gene. The resulting Flag-ICP0 virus replicated with WT kinetics in Vero and Neuro-2a cells (Figure 2B). A Flag antibody readily detected ICP0 expression from the virus in 293T, HFF, and Neuro-2a cells (Figure 2C). ICP0 expression reached plateaus in all the three cell lines at around 6 hpi, so we decided to use 6 h as the time point to collect samples for the following interactomic experiments.

To analyze the ICP0 interactome, we infected Neuro-2a, HFF, and 293T cells with Flag-ICP0 or WT virus for 6 h before co-immunoprecipitation (Co-IP) with a Flag antibody followed by protein identification by liquid chromatography with tandem mass spectrometry (LC–MS/MS) (Figure 2D). After applying selection criteria (see section “Materials and Methods”), we identified 37, 68, and 43 potential ICP0-interacting proteins in Neuro-2a, HFF, and 293T cells, respectively, amounting to a total of 108 candidate proteins (Figure 2E and Supplementary Table S1). Eight proteins were found in all three cell lines: viral proteins UL50, UL39, RS1, and UL34 and host proteins USP7, USP9X, vimentin (VIM), and DBN1. The interactome overlaps very well between 293T and HFF cells in that 67% of proteins identified in 293T cells were also identified in HFF cells and 43% of proteins identified in HFFs were also identified in 293T cells. However, the overlap between mouse Neuro-2a cells and any of the human cell lines was poor. This might imply that the differences in ICP0 interactome between host species might be greater than those between cell types. Therefore, we consider the proteins identified in all three cell lines as the ICP0 interactome group for subsequent analyses. Functional enrichment analysis of this group revealed that ICP0 interacted with a host protein network related to several biological function clusters including deubiquitination, proteasome, cytoskeleton, translation elongation, mRNA processing, and protein transport (Supplementary Figure S1). The group also included many previously reported ICP0 targets such as viral proteins UL50 (Conwell et al., 2015), UL39 (Conwell et al., 2015), and RS1 (ICP4) (Yao and Schaffer, 1994; Kalamvoki et al., 2008) and host proteins USP7 (Everett et al., 1997), USP9X (Sato et al., 2016a), PRKDC (Parkinson et al., 1999), DDX17 (Mallon et al., 2012), and WDR11 (Taylor and Mossman, 2015).

To validate the proteomics data, we selected some top novel host candidates according to Flag-ICP0/WT ratios. With priorities given to the candidates identified in at least two cell lines, we selected a total of 17 top candidates and cloned their human genes into expressing plasmids with Flag tags. After co-transfection of plasmids with an ICP0 expressing plasmid into Neuro-2a and 293T cells, we performed IP in a reverse way relative to the interactome analysis. We confirmed that untagged ICP0 could be pulled down by Flag-tagged complement C1q binding protein (C1QBP), VIM, sorting nexin 9 (SNX9), acyl-coenzyme A thioesterase 8 (ACOT8), and ovarian tumour domain-containing protein 4 (OTUD4) in both Neuro-2a and 293T cells (Figure 2F). We note that although SNX9, ACOT8, and OTUD4 were not identified in the interactome analysis in Neuro-2a cells possibly due to limited sensitivity of MS or the human-mouse differences, the reverse Co-IP showed that their human forms were all able to interact with ICP0 in Neuro-2a cells. Therefore, these proteins were included in the following functional studies. Hist1H2BB co-precipitated with ICP0 in Neuro-2a but not 293T cells. MCM3 and MCM5 co-precipitated with ICP0 in 293T but not Neuro-2a cells. Therefore, about half of the candidates tested (and 7% of the interactome group) were able to coimmunoprecipitate with ICP0 in at least one cell line. Other proteins that we examined, including heterogeneous nuclear ribonucleoprotein M (HNRNPM), prolyl 4-hydroxylase subunit alpha 2 (P4HA2), ribosomal protein L10a (RPL10A), myosin light chain 12A (MYL12A), protein phosphatase 1 catalytic subunit beta (PPP1CB), caveolae associated protein 1 (CAVIN1), drebrin 1 (DBN1), proteasome 20S subunit alpha 3 (PSMA3), and proteasome 20S subunit alpha 6 (PSMA6), could not reproducibly co-immunoprecipitate with ICP0 in either cell line (data not shown).

To explore the functions of the ICP0-interacting proteins, we focused on the five proteins that co-immunoprecipitated with ICP0 in both 293T and Neuro-2a cells, including C1QBP, VIM, SNX9, ACOT8, and OTUD4. Besides using the plasmids to overexpress the proteins, we designed at least two siRNAs to knock down each protein. After evaluating the knockdown efficiencies, two siRNAs with the highest efficiencies for each protein was selected along with two negative control siRNAs. Efficient knockdown by the selected siRNAs and overexpression by the plasmids in Neuro-2a cells were confirmed in Western blots (Supplementary Figure S2). We then infected the cells after transfection of the plasmids or siRNAs. Overexpression of none of these proteins had a significant effect on replication of 7134 or KOS virus (Figure 3A). However, both siRNAs against SNX9 resulted in significantly increased 7134 virus titers relative to both control siRNAs in Neuro-2a cells, although only one of them (SNX9-si1) showed modestly increased KOS titers (Figure 3B). We then performed co-transfection of the siRNAs together with plasmids. Although both siRNAs against SNX9 significantly increased 7134 virus titers in the presence of the control plasmid, they could not do so in the presence of the SNX9 expressing plasmid (Figure 3C), suggesting that the increased replication was caused by specific knockdown of SNX9. Growth curve analysis showed that even the less effective SNX9 siRNA (SNX9-si2) significantly increased replication kinetics of 7134 virus at an MOI of 0.2 but it had no effect at an MOI of 5 (Figure 3D). Therefore, endogenous SNX9 restricted HSV-1 replication at a low MOI.

To identify ubiquitination substrates of ICP0, we compared the ubiquitinome during HSV-1 infection in the presence versus absence of ICP0. Absence of ICP0 was achieved using the ICP0-null virus 7134, which was compared with 7134R. To increase the stringency of the experiment, we also considered rescuing the absence of ICP0 by transfection because real targets should show differences in comparisons both between 7134 and 7134R viruses and between ICP0 transfected and untransfected cells. Therefore, we designed the following three groups: group A, control transfection + 7134 virus infection; group B, ICP0 transfection + 7134 virus infection; group C, control transfection + 7134R virus infection. Only ubiquitination events that were induced both by transfected ICP0 (comparing B and A) and by ICP0 expressed from the virus (comparing C and A) were considered, which should greatly reduce frequencies of false-positive discoveries. Neuro-2a cells were transfected for 24 h and then infected for 6 h at an MOI of 20 before total protein was digested by trypsin and immunoprecipitated with ubiquitin remnant K-ε-GG antibody followed by TMT labeling and LC-MS/MS analysis (Figure 4A). Overall, ratios of protein quantities in group B versus A (B/A) and group C versus A (C/A) comparisons correlated well with each other (Figure 4B). After applying selection criteria for the differences and mapping these sites to proteins, we identified 1,022 proteins in the B/A comparison and 522 proteins in the C/A comparison. Importantly, a large fraction of these proteins (351 proteins) was commonly identified by both comparisons. These proteins were considered as potential ICP0 substrates (Figure 4C and Supplementary Table S1). Functional enrichment analysis showed that the potential host substrates are enriched in multiple pathways, including metabolism of RNA/proteins, rRNA processing, cell cycle, and deubiquitination (Figure 4C). Nine proteins were previously reported to be potential host substrates of ICP0 including USP7 (Boutell et al., 2005), PML (Cuchet-Lourenco et al., 2012), OPTN (Waisner and Kalamvoki, 2019), SUMO2 (Sloan et al., 2015; Drayman et al., 2017), UBE2E1 (Vanni et al., 2012), ZBTB10 (Sloan et al., 2015), MORC3 (Sloan et al., 2016), RNF168 (Lilley et al., 2010), and ETV6 (Sloan et al., 2015). Their potential ubiquitination sites determined by our MS data are listed in Figure 4C. In addition, the candidate substrate proteins were clustered in several superfamilies classified by conserved domains such as nucleoside triphosphate hydrolase, RNA helicase, zinc finger, RNA binding, and ubiquitin (Supplementary Table S2). Regarding viral proteins, out of 55 viral proteins with detected ubiquitination, seven proteins, namely, UL12, UL21, UL5, UL51, UL54, UL9, and US3, showed increased ubiquitination in the presence of ICP0 (Supplementary Figure S3), indicating that they might be potential viral substrates of ICP0.

Comparison of the interactome and ubiquitinome data identified three viral proteins and eight host proteins, which interacted with ICP0 and exhibited elevated ubiquitination in the presence of ICP0, indicating that they were highly likely to be substrates of ICP0 (Figure 5A). All of these are novel potential substrates of ICP0 except for the deubiquitinase USP7, whose ubiquitination by ICP0 has been reported elsewhere (Canning et al., 2004; Boutell et al., 2005). Because host proteins VIM and OTUD4 were confirmed to robustly interact with ICP0 (Figure 2F), we focused on these two proteins for ubiquitination analysis. VIM was previously documented to be decorated with 11 SUMO2 modifications by another large-scale MS study (Hendriks et al., 2017). Our MS data detected 17 potential ubiquitination sites in VIM (Figure 5B). We note that the trypsin digestion resulted in different reminants in ubiquitinated or SUMOylated lysine, so the current LC-MSMS analysis was specific to ubiquitin modification. Interestingly, the majority of the newly identified ubiquitination sites overlapped with the SUMOylation sites, indicating possible cross-talk between ubiquitination and SUMOylation in VIM. Almost all ubiquitination and SUMOylation sites are located in the functional coil regions of VIM. Notably, the only site (K282) showing increased ubiquitination due to ICP0 presence has not been reported to be SUMOylated. The MS data also detected four potential ubiquitination sites in OTUD4, all of which are near the OTU domain and not far from the previously annotated phosphorylation sites. One of the sites (K264) showed increased ubiquitination due to ICP0 presence and, therefore, could be an ICP0 target site. To examine whether VIM and OTUD4 were ubiquitinated by ICP0, we co-transfected ubiquitin, VIM or OTUD4, and ICP0 or its control (empty vector or ring finger mutant) into Neuro-2a cells before immunoprecipitation of VIM and OTUD4 and analysis of ubiquitination levels by Western blots. Both VIM and OTUD4 exhibited substantial enhancement of ubiqutination in the presence of ICP0 relative to the respective controls, suggesting that they could be ubiquitinated by ICP0 (Figure 5C). However, transfected ICP0 did not cause decreases in VIM and OTUD4 levels indicating that ubiquitination of these proteins may not lead to degradation.

To understand the roles of VIM and OTUD4 during HSV-1 infection, we first assessed their levels in infected Neuro-2a cells. Interestingly, we observed substantial upregulation of OTUD4 but not VIM starting from as early as 3 hpi (Figure 6A). This upregulation was independent of the E3 ligase activity of ICP0 because it was also observed during infection with an ICP0 ring finger mutant (RFm) virus. The increases in OTUD4 protein levels correlated with similar increases in its mRNA levels indicating stimulation of Otud4 gene transcription during infection (Figure 6B). Comparison of WT and RFm viruses showed no indication of ICP0-induced degradation of OTUD4 or VIM at later times (Figure 6A). Thus, OTUD4 was upregulated during HSV-1 infection without being degraded by ICP0.

To learn more about the roles of OTUD4 and VIM, although they showed no effect on HSV-1 replication (Figures 3A,B), we wondered whether they could regulate innate immune response to infection. To test this, we first transfected Neuro-2a cells with an IFN-β promoter luciferase reporter construct together with a plasmid expressing OTUD4 or VIM before infection at an MOI of 2 for 16 h. 7134 virus only modestly stimulated luciferase expression, whereas KOS virus showed much stronger stimulation. In 7134 infected cells, overexpression of OTUD4 significantly increased luciferase activity, whereas knocking down OTUD4 or overexpressing or knocking down VIM had no significant effect (Figures 6C,D). However, in KOS-infected cells, neither overexpressing nor knocking down OTUD4 had a significant effect indicating an ICP0-dependent block of OTUD4 activity. Consistent with this, overexpressed OTUD4 increased luciferase activity during infection with an ICP0 RFm virus but had no effect during infection with 7134R virus (Figure 6E). Unexpectedly, OTUD4 also increased luciferase activity during mock infection suggesting that OTUD4 could raise the baseline expression of IFN-β in unstimulated cells. In concordance with the luciferase results, quantification of endogenous IFN-α mRNA showed that induction of IFN-α expression by KOS but not 7134 virus was obvious at an MOI of 2, and overexpressed OTUD4 significantly increased IFN-α mRNA levels in 7134 but not KOS infected cells (Figure 6F). At a lower MOI of 0.5, where induction of IFN-α expression was no longer detectable even in KOS infected cells, overexpressed OTUD4 significantly increased IFN-α mRNA levels in 7134 but not KOS virus infected cells. Therefore, OTUD4 could enhance type I IFN expression during HSV-1 infection, and this function appeared to be blocked by ICP0-dependent processes.

Vero, 293T, Neuro-2a, and HFF cells were obtained from American Type Culture Collection and cultured as previously reported (Pan et al., 2019). The method for culturing primary mouse TG neurons was described previously (Katzenell et al., 2017; Sun et al., 2021). HSV-1 strains KOS and its derivatives including 7134, 7134R (Cai and Schaffer, 1989), and WT-BAC (Pan and Coen, 2012) were described previously. HSV-1 Flag-ICP0 virus was prepared according to a previously described protocol using BAC technology (Tischer et al., 2006; Pan et al., 2014). Briefly, PCR-amplified zeocin-resistance (Zeo) cassette containing the DYKDDDDK (Flag)-coding sequence was inserted into one of the two copies of ICP0 gene just downstream of the start codon. Then, the PCR-amplified kanamycin-resistance (Kan) cassette containing the Flag-coding sequence and a homologous sequence was inserted into the other copy of ICP0. The Kan cassette was then removed by recombination via the homologous sequence, resulting in markerless insertion of the Flag-coding sequence into the second ICP0 gene copy. At the location of the first ICP0 gene copy, the Zeo cassette was then replaced with the Kan cassette, which was subsequently removed as for the other ICP0 copy. In each step, correct colonies were selected by PCR and sequencing as well as restriction fragment length polymorphism following digestion with EcoRI and HindIII. BAC DNA was purified and transfected into Vero cells for Flag-ICP0 virus production. Primers used are listed in Supplementary Table S3. HSV-1 infection of cells and plaque assays for viral titer measurements were performed as previously described (Pan et al., 2019).

FLAG-HA-pcDNA3.1 (Addgene, #52535) was used to construct plasmids expressing hemoglobin, beta adult s chain (HBB-BS), VIM, unique long 50 (UL50), ubiquitin conjugating enzyme E2 L3 (UBE2L3), C1QBP, histone H2B type 1-B (Hist1H2BB), DEAD box RNA helicase 17 (DDX17), HNRNPM, DBN1, PSMA6, P4HA2, minichromosome maintenance complex component 3 (MCM3), minichromosome maintenance complex component 5 (MCM5), RAS-related protein 1a (RAP1A), OTUD4, SNX9, ACOT8, RPL10A, MYL12A, PPP1CB, CAVIN1, and PSMA3 with Flag-HA tags. HA-ubiquitin was constructed on the basis of HA-pcDNA3.1, which was modified from FLAG-HA-pcDNA3.1. The firefly luciferase plasmid with IFN-β promoter (pIFN-β) and Renilla luciferase plasmid pRL-TK were kind gifts of Professor Chunfu Zheng at Fujian Medical University. To construct pICP0, the complete coding sequence of ICP0 was PCR-amplified from ICP0-WT plasmid (pRS-1) (Sandri-Goldin et al., 1987) and inserted into pcDNA3.1 without Flag or HA tags. Primers used for PCR were listed in Supplementary Table S3.

All transfections were conducted using Lipofectamine 3000 (Thermo Fisher, Waltham, MA, United States) according to the manufacturer’s instructions. The siRNAs were synthesized by RiboBio. Their target sequences are as follows: ACO8-si1, GACCCTAACCTTCACAAGA; ACOT8 -si2, CCAAACAGATGTTCTGGGT; C1QBP-si1, CCACCTAAT GGATTTCCTT; C1QBP-si2, CATTTGATGGTGAGGAGGA; OTUD4-si1, GCTTCTTCATGCTGAATAT; OTUD4-si2, CCAG CAGAACATATACCTT; SNX9-si1, GTGGGTTTATCCTACCT CT; SNX9-si2, CGGATCTATGATTACAACA; VIM-si1, CAGA CAGGATGTTGACAAT; VIM-si2, CTTCTCAGCATCACGA TGA.

Cells were scraped into sodium dodecyl sulfate (SDS) loading buffer (40% glycerol, 0.24 M Tris-HCl, pH 6.8, 8% SDS, 0.04% bromophenol blue, 5% β-mercaptoethanol) and heated at 95°C for 5 min before being loaded onto SDS polyacrylamide gels for Western blot analysis. The following antibodies and dilutions were used: Flag antibody (MBL, Tokyo, Japan, M185-3L), 1:5,000; ICP0 antibody (Santa Cruz, 11060), 1:2,000; β-tubulin antibody (Sungene Biotech, Tianjin, China, KM9003), 1:5,000; rabbit HA antibody (Sangon, Shanghai, China, D110004), 1:1,000; β-actin antibody (Abclonal, Wuhan, China, Ac026), 1:5,000; ICP4 antibody (Abcam, Cambridge, England, ab6514), 1:5,000; gC antibody (Fitzgerald, North Acton, MA, United States, 10-H25A), 1:1,000; ICP27 antibody (Virusys, Taneytown, MD, United States, 1113), 1:5,000; ACOT8 rabbit polyclonal antibody (Sangon, D221492), 1:500; C1QBP rabbit polyclonal antibody (Sangon, D152901), 1:1,000; OTUD4 rabbit polyclonal antibody (Abclonal, A15229): 1:1,000; SNX9 rabbit polyclonal antibody (Sangon, D154017), 1:1,000; VIM rabbit polyclonal antibody (Sangon, D220268), 1:3,000; horseradish peroxidase (HRP)-conjugated goat anti-mouse (SouthernBiotech, Birmingham, AL, United States, 1030-05), 1:2,000; goat anti-rabbit antibodies (SouthernBiotech, 4030-05), 1:2,000.

Neuro-2a cells (2 × 10⁷) in a T150 flask were infected with Flag-ICP0 or WT virus at an MOI of 20 and harvested at 6 hpi. Cells were first washed twice in ice-cooled phosphate buffered saline (PBS) and then lysed in 1 ml of lysis buffer [50 mM HEPES-KOH (pH 7.4), 1% Triton X-100, 150 mM NaCl, 10% glycerol, and 2 mM EDTA plus one Complete EDTA-free protease inhibitor tablet (Roche) per 50 ml] for 1 h. Supernatants were first incubated with 25 μl of mouse immunoglobulin G (IgG) for 1 h at 4°C to remove non-specific binding proteins and then incubated with 50 μl of Pierce Anti-DYKDDDDK Magnetic Agarose (ThermoFisher, A36797) for 2 h. The agarose was washed four times with lysis buffer for a total of 1 h and then washed four times with PBS to remove Triton X-100. Proteins were eluted in 300 μl of elution buffer (0.1 M glycine, pH 2.8) at room temperature for 7 min and neutralized with 30 μl of Neutralization Buffer: (1 M Tris, pH 8.5). Experiments in HFF and 293T cells were conducted in the same way except that an MOI of 10 was used for infection. Immunoprecipitated proteins were reduced with 10 mM dithiothreitol (DTT) for 45 min at 30°C and alkylated with 30 mM iodoacetamide (IAA) for 30 min at room temperature in the dark. Proteins were then cleaned up by acetone precipitation and resuspended in 50 mM ammonium bicarbonate. Proteins were digested by trypsin (Promega, Madison, WI, United States) at a ratio of 1:50 (trypsin:protein) at 37°C overnight. Tryptic peptides were acidified by trifluoroacetic acid (TFA) and were desalted using a hydrophilic-lipophilic-balanced (HLB) microplate (Waters, Milford, CT, United States). The microplate was conditioned with 200 μl of acetonitrile (ACN), followed by 200 μl of 60% ACN and then by 200 μl of 0.1% TFA twice. The sample was loaded to the microplate and then washed three times with 200 μl of 0.1% TFA. Desalted peptides were eluted with 60% ACN and lyophilized in a vacuum lyophilizer (LABCONCO) before LC-MS/MS analysis.

The following three groups were compared to determine the effects of ICP0 on ubiquitinome: Group A, pcDNA3.1 transfection + ICP0 null virus infection; Group B, ICP0 plasmid transfection + ICP0 null virus infection and group; and Group C, pcDNA3.1 transfection + ICP0-rescued virus infection. Neuro-2a cells in a 100-mm dish were transfected with ICP0 or pcDNA3.1 plasmid for 24 h before infection with 7134 virus (ICP0-null) or 7134R virus (ICP0-rescued) at an MOI of 20. After the 1-h incubation for virus absorption, fresh medium was added with 10 μM MG-132 to inhibit proteasome-dependent degradation. The cells were harvested at 6 hpi in urea lysis buffer (20 mM Hepes, pH 8.0, 9 M urea). After sonication and centrifugation, the supernatants were kept.

Proteins (10 mg) were reduced with 10 mM DTT for 45 min at 30°C and subsequently alkylated with 30 mM IAA for 30 min at room temperature in the dark. Lysis buffer was then replaced by 50 mM triethylammonium bicarbonate (TEAB) using Zeba Spin Desalting Columns. Proteins were digested by L-1-tosylamido-2-phenylethyl chloromethyl ketone (TPCK)-treated trypsin (Sigma, Saint Louis, Mo, United States) at a ratio of 1:40 (trypsin:protein) at 37°C overnight. Tryptic peptides were acidified by TFA and desalted using a 500 mg C18 Sep-Pak SPE cartridge (Waters). C18 cartridges were conditioned with 4 ml of ACN, followed by 4 ml of 50% ACN + 0.1% TFA and then by 4 ml of 0.1% TFA. The sample was loaded to the C18 cartridge and then washed three times with 4 ml of 0.1% TFA. Desalted peptides were eluted with 50% ACN + 0.1% formic acid (FA) and lyophilized in a vacuum lyophilizer (LABCONCO).

Di-Gly–modified peptides derived from ubiquitinated proteins were enriched by PTMScan ubiquitin remnant K-ε-GG motif kit [cell signaling technology (CST)] according to the manufacturer’s instruction. Briefly, antibody-loaded beads were washed three times with 1 ml of pre-cooled 100 mM sodium borate (pH 9.0). Antibody was then cross-linked by 20 mM dimethyl pimelimidate for 30 min. Cross-linked antibody beads (50 μg) were mixed with tryptic peptides in cold 1 × IAP buffer for 2 h at 4°C with end-over-end rotation, followed by three washes with ice-cold IAP buffer and one wash with ice-cold water. Di-Gly–modified peptides were eluted by 50 μl of 0.15% TFA twice and dried completely by vacuum centrifugation.

Di-Gly–modified peptides enriched from each sample were dissolved in 20 μl of 100 mM TEAB and labeled with 0.8 mg 10plex TMT Label Reagent (ThermoFisher) in 50 μl of ACN. The reaction was proceeded at room temperature for 1 h and stopped by 2 μl of 5% hydroxylamine. Supernatant was desalted by homemade C18 stage tip and dried by vacuum centrifugation.

Home-packed strong cation exchange (SCX) chromatography StageTip with Empore Cation-SR Extraction material (3M) was sequentially conditioned by 100 μl of ACN, 80% ACN + 0.1% TFA, and 5% ACN + 0.1% TFA. The combined TMT-labeled K-ε-GG peptides were resuspended in 80 μl of 5% ACN + 0.1% TFA and loaded to SCX Stagetip. The flow-through was labeled as SCX fraction 1. Peptides were then sequentially eluted with 100 μl of 5% ACN + 0.1% TFA supplemented with increasing concentration of ammonium acetate (40, 100, 200, 350, and 500 mM). The corresponding eluates were designated as SCX fractions 2, 3, 4, 5, and 6, respectively. All peptide fractions were dried by vacuum centrifugation.

All dried peptides were resuspended in 2% ACN and 0.1% FA and separated by nanoLC-MS/MS using an UltiMate 3000 RSLCnano system (Thermo Scientific) at the flow rate of 400 nl/min. Solvent A is 2% ACN and 0.1% FA, and solvent B is 98% ACN and 0.1% FA. Gradient elution was performed at 50°C using linear gradients of 120 min: 1–4 min with 3% (v/v) of s B, 4–6 min from 3% to 5% (v/v) B, 6–70 min from 5% to 15% (v/v) B, 70–90 min from 15% to 30% (v/v) B, 90–100 min from 30% to 80% (v/v) B, 100–110 min with 80% (v/v) B, and 110–120 min with 3% (v/v) A. The eluted peptides were analyzed by Q Exactive HF-X (Thermo Scientific) acquiring MS spectra at the resolution of 120,000 full width at half maximum (FWHM) with a mass range of 300–1,500 m/z and an AGC target of 3E6. The top 20 precursors were then fragmented by higher-energy C-trap dissociation (HCD) with collision energy approximately 32% normalized collision energy (NCE) and tandem mass spectrometry (MS2) spectra were acquired at the resolution of 45,000 FWHM.

All MS raw data were loaded into MaxQuant (version 1.6.2.10) and searched against human or Mouse UniProtKB database (November 2019) supplemented with HSV1 sequences, with the automatic reverse database on target-decoy search mode. MS2-based isobaric labeling using 10plex TMT tags were configured only for the ubiquitinome analysis. Variable modifications including oxidation (M) (+15.99491 Da) and acetyl (protein N-term) (+42.01056 Da) were specified for ubiquitinome, proteome, and ICP0-interactome analyses, whereas GlyGly(K)_10plex_TMT (+ 343.20586 Da) was specified only for the ubiquitinome analysis. Carbamidomethyl (C) (+ 57.02146 Da) was set as the fixed modification. Trypsin was set as digestion mode with two maximum missed cleavage sites. We used 20 ppm in the first search ion tolerance and 4.5 ppm in the main search ion tolerance. MaxQuant default setting was used for all other parameters. Peptide and protein identification were both filtered by false discovery rate (FDR) < 1%.

For the interactome data sets from 293T and Neuro-2a cells, potential ICP0-interacting proteins were selected by either fold change ≥ 5 or fold changes ≥ 1.2 with p < 0.05 (t-test) between Flag-ICP0 and WT comparison. For HFF cells, which identified a larger number of proteins, we used a more stringent criterion of fold changes > 5 and p < 0.1 (t-test). Proteins with < 2 unique peptides were discarded. The ubiquitinome data were normalized by protein levels determined by peptides without di-glycine modification. After normalization, modified sites with > 1.25 higher di-glycine levels in both B/A and C/A comparisons or with increasing di-glycine levels with FDR < 0.05 (t-tests) were considered as potential ICP0 target sites. To identify host pathways modulated by HSV-1 ICP0, potential ubiquitination targets of ICP0 were then submitted for functional annotation enrichment analysis in KOBAS 3 (Xie et al., 2011)¹ against Reactome database (Jassal et al., 2020). Fisher’s exact test and Benjamini and Hochberg’s FDR < 0.01 method were chosen for significant enriched terms. Enriched term with >1000 genes in database or <10 mapped genes were discarded from analyses. Similar or redundant terms were manually removed to keep the most significantly enriched ones. All potential ICP0-interacting proteins were used to generate interaction network based on String database² (Szklarczyk et al., 2019), whereas all the potential ubiquitination targets were submitted to search conserved domains by InterPro of String database (Blum et al., 2020).

Five hours after 5 × 10⁶ Neuro-2a cells were plated in 100-mm dishes, the cells were transfected with the ICP0 expressing plasmid together with a plasmid expressing a potential interacting protein with a Flag tag. Cells were lysed in 1 ml of lysis buffer (see above) at 30 h post-transfection (hpt). The Flag-tagged proteins were immunoprecipitated as described for the ICP0 interactome analysis except that 40 μl of anti-DYKDDDDK Magnetic Agarose was used and 80 μl of 1 × SDS loading buffer was added to eluted proteins. Western blotting was conducted to detect the proteins.

About 9 × 10⁶ Neuro-2a cells in 100-mm dishes were co-transfected with the following plasmids: 5 μg of pOTUD4 or pVIM (Flag-HA tag), 2 μg of HA-Ubiquitin and 3 μg of ICP0 or ICP0-RFm as control. Cells were lysed in lysis buffer at 48 hpt. IP was performed as described above using the anti-Flag agarose.

To quantify mRNA levels in cells, total RNA was purified using a RNA Isolation Kit (catalog # RC112, Vazyme Biotech, Nanjing, China) according to the manufacturer’s instructions. Reverse transcription and PCR were conducted using a HiScript II Q Select RT SuperMix and ChamQ Universal SYBR qPCR Kit (catalog # R233-01 and Q711-02/03, Vazyme Biotech). Data were normalized to GAPDH mRNA levels. The following primers for mouse transcripts were used: GAPDH-F: GAAGGTCGGTGTGAACGGATT; GAPDH-R: GCCTTGACTGTGCCGTTGAA; OTUD4-F: CCTC CATCTCAGGTGTCTGAAGGTCA; OTUD4-R: GGTTAGGC CCAAAAGACTGTTGTGG; IFN-α-F: CTCATTCTGCAATG ACCTCCACC, IFN-α-R: ACTTCTGCTCTGACCACCTCCC.

Neuro-2a cells were co-transfected with the indicated plasmid (400 ng/ml) or the indicated siRNA (80 nM), pIFN-β (80 ng/ml), and pRL-TK (40 ng/ml) for 48 h and then infected with 7134 or KOS virus for the indicated times. The cells were then lysed, and promoter activities were measured by a dual luciferase kit according to the manufacturer’s protocol (Yeasen, Shanghai, China, 11402ES60).