Host Vesicle Fusion Protein VAPB Contributes to the Nuclear Egress Stage of Herpes Simplex Virus Type-1 (HSV-1) Replication

The primary envelopment/de-envelopment of Herpes viruses during nuclear exit is poorly understood. In Herpes simplex virus type-1 (HSV-1), proteins pUL31 and pUL34 are critical, while pUS3 and some others contribute; however, efficient membrane fusion may require additional host proteins. We postulated that vesicle fusion proteins present in the nuclear envelope might facilitate primary envelopment and/or de-envelopment fusion with the outer nuclear membrane. Indeed, a subpopulation of vesicle-associated membrane protein-associated protein B (VAPB), a known vesicle trafficking protein, was present in the nuclear membrane co-locating with pUL34. VAPB knockdown significantly reduced both cell-associated and supernatant virus titers. Moreover, VAPB depletion reduced cytoplasmic accumulation of virus particles and increased levels of nuclear encapsidated viral DNA. These results suggest that VAPB is an important player in the exit of primary enveloped HSV-1 virions from the nucleus. Importantly, VAPB knockdown did not alter pUL34, calnexin or GM-130 localization during infection, arguing against an indirect effect of VAPB on cellular vesicles and trafficking. Immunogold-labelling electron microscopy confirmed VAPB presence in nuclear membranes and moreover associated with primary enveloped HSV-1 particles. These data suggest that VAPB could be a cellular component of a complex that facilitates UL31/UL34/US3-mediated HSV-1 nuclear egress.

. Optimization of conditions to isolate HSV-1 infected MMs. (A) An expression timecourse of US3 and gC viral proteins was analyzed by Western blot to determine the optimal time to isolate HSV-1 infected MMs before significant secondary envelopment had occurred. Cell lysates from Hela cells infected at MOI = 10 were prepared at indicated times post-infection. α-tubulin was used as a loading control. (B) Fluorescence microscopy images of an HSV-1 strain with VP26 capsid protein fused with RFP in Hela cells after 8, 9 and 10 hpi. At 8 hpi, most of the VP26 signal was located inside the nucleus of infected cells. At 9 hpi, cells started to show perinuclear localization of VP26. Scale bar, 10 µm. (C) Coomassie-stained gel of mock-infected NE, mock-infected MM, and HSV-1 infected MM fractions showed a clear difference in protein composition between NEs and MMs, but no notable visible differences between infected and mock-infected MMs. (D) Western blot of fractions from (C) stained with ER and NE markers to determine fraction purity. Infected MMs are shown. The ER marker calnexin was present in both NE and MM fractions as expected, because the ONM is continuous with the ER and many proteins are shared. In contrast, the NE markers lamin A/C and Lap2β were absent from MMs. Similar amounts of total protein were loaded. (E) To test for possible Golgi contamination in the preps, the NE and MM fractions were blotted for Golgi marker GM-130. Equal protein loading from HeLa cells confirmed that the antibody was working. (F) Ultrastructure of isolated MMs from HSV-1 infected HeLa cells. Electron micrographs showed the characteristic single-membrane structure of the MMs. Arrows point to electron-dense symmetrical structures of around 100 nm diameter that probably represent primary viral particles with the bottom one likely lacking a packaged genome. Scale bars, 100 nm.

Preparation of Fractions
Nuclear envelopes (NEs) and microsomal membranes (MMs) were isolated from HeLa cells using established procedures [29]. For microsomes, we started with ~2 × 10 9 cells from 15 roller bottle cultures. The post-nuclear supernatant was treated with 0.5 mM EDTA to inhibit metalloproteinases. Mitochondria and other debris were removed by pelleting 15 min at 10,000× g. Microsomes were floated through 1.86 M and 0.25 M sucrose layers by centrifugation in the SW28 rotor 4 h at 57,000× g, Figure 1. Optimization of conditions to isolate HSV-1 infected MMs. (A) An expression timecourse of US3 and gC viral proteins was analyzed by Western blot to determine the optimal time to isolate HSV-1 infected MMs before significant secondary envelopment had occurred. Cell lysates from Hela cells infected at MOI = 10 were prepared at indicated times post-infection. α-tubulin was used as a loading control. (B) Fluorescence microscopy images of an HSV-1 strain with VP26 capsid protein fused with RFP in Hela cells after 8, 9 and 10 hpi. At 8 hpi, most of the VP26 signal was located inside the nucleus of infected cells. At 9 hpi, cells started to show perinuclear localization of VP26. Scale bar, 10 µm. (C) Coomassie-stained gel of mock-infected NE, mock-infected MM, and HSV-1 infected MM fractions showed a clear difference in protein composition between NEs and MMs, but no notable visible differences between infected and mock-infected MMs. (D) Western blot of fractions from (C) stained with ER and NE markers to determine fraction purity. Infected MMs are shown. The ER marker calnexin was present in both NE and MM fractions as expected, because the ONM is continuous with the ER and many proteins are shared. In contrast, the NE markers lamin A/C and Lap2β were absent from MMs. Similar amounts of total protein were loaded. (E) To test for possible Golgi contamination in the preps, the NE and MM fractions were blotted for Golgi marker GM-130. Equal protein loading from HeLa cells confirmed that the antibody was working. (F) Ultrastructure of isolated MMs from HSV-1 infected HeLa cells. Electron micrographs showed the characteristic single-membrane structure of the MMs. Arrows point to electron-dense symmetrical structures of around 100 nm diameter that probably represent primary viral particles with the bottom one likely lacking a packaged genome. Scale bars, 100 nm.

Preparation of Fractions
Nuclear envelopes (NEs) and microsomal membranes (MMs) were isolated from HeLa cells using established procedures [29]. For microsomes, we started with~2 × 10 9 cells from 15 roller bottle cultures. The post-nuclear supernatant was treated with 0.5 mM EDTA to inhibit metalloproteinases. Mitochondria and other debris were removed by pelleting 15 min at 10,000× g.
The samples were analyzed by Multidimensional Protein Identification Technology (MudPIT) as previously described [30,31] with pressure-loading onto microcapillary columns packed with 3 cm of 5-µm Strong Cation Exchange (Luna; Phenomenex, Torrance, CA, USA), followed by 1 cm of 5 µm C18 reverse phase (Aqua; Phenomenex, Macclesfield, UK). These were connected to 100 µm columns pulled to a 5 µm tip containing 9 cm of reverse phase material. Peptides were separated on a Quaternary Agilent 1100 HPLC using a 10-step chromatography run over 20 h at 200-300 nL/min. Eluting peptides electrosprayed at 2.5 kV distal voltage into a LTQ linear ion trap mass spectrometer (Thermo Scientific, Waltham, MA, USA) with a custom-made nano-LC electrospray-ionization source. Full MS spectra were recorded on the peptides over 400 m/z to 1,600 m/z, followed by five tandem mass (MS/MS) events, sequentially generated in a data-dependent manner on the first to fifth most intense ions selected from the full MS spectrum (at 35% collision energy). Dynamic exclusion was enabled for 120 s.
RAW files extracted into ms2 files [32] using RawDistiller v1.0 [33] were queried for peptide sequences using SEQUEST v.27 (rev.9) [34] against 55,691 human proteins (non-redundant NCBI 2014-02-04 release), plus 162 usual contaminants (e.g., keratin) and 77 NCBI RefSeq HSV-1 proteins. To estimate false discovery rates (FDRs), each non-redundant sequence supplemented the database after randomization, bringing the search space to 111,524 sequences. MS/MS spectra were searched without specifying differential modifications, but +57 Da were added statically to cysteines to account for carboxamidomethylation. No enzyme specificity was imposed and mass tolerance was set at 3 amu for precursor ions and ±0.5 amu for fragment ions.
Results from different runs were compared using DTASelect and CONTRAST [35] with criterion of DeltCn ≥ 0.08, XCorr ≥ 1.8 for singly-, 2.5 for doubly-, and 3.5 for triply-charged spectra, and a maximum Sp rank of 10. Merging peptide hits from all analyses established a master protein list (Table S1). Identifications mapping to randomized peptides estimated FDRs that averaged 0.72% for proteins and 0.24% for peptides. Spectral-count based label free quantitation was used to estimate the relative levels of the proteins detected by mass spectrometry in each sample. In shotgun proteomics, the frequency of peptides being fragmented by the mass spectrometer (spectral counts) correlates with the abundance of the proteins these peptides derive from. Because longer proteins tend to generate more tryptic peptides, spectral counts are normalized by the protein molecular weight or length, defining a "spectral abundance factor" (SAF), which is further normalized against the sum of SAFs calculated for each protein/protein group detected in a sample. To deal with peptides shared between multiple proteins, distributed Normalized Abundance Factors (dNSAFs) are calculated for each non-redundant protein/protein group (Tables 1, 2 and S1), as described in [36] in which shared spectral counts (sSpC) are distributed based on spectral counts unique to each protein i (uSpC), divided by the sum of all unique spectral counts for the M protein isoforms that shared peptide j with protein i.

Bioinformatics
Ratios of HSV-1 infected MMs:mock-infected MMs and mock-infected NE:mock-infected MM were calculated from dNSAF values. Only proteins with an HSV-1 infected MMs:mock-infected MMs ratio higher than 1.3 and detected in the mock infected NEs were selected. Gene ontology (GO) analysis was performed in Ensembl BioMart. GO-terms and their corresponding child-terms were retrieved from the mySQL database http://amigo.geneontology.org [37]: "vesicle-mediated transport" (GO:0016192) "nucleocytoplasmic transport" (GO:0006913), "membrane organization" (GO:0061024), "regulation of protein phosphorylation" (GO:0001932). Relative proportions of these classes were calculated from the relative abundances of genes identified from each category, then compared to the proportions of all human-encoded proteins considering the same categories.

siRNA Knockdown of VFPs
HeLa cells were transfected for 48 h with 50 nM final concentration siRNAs to VAPB, Rab1a and Rab24 using JetPrime (Polyplus, New York, NY, USA). All siRNAs were Dharmacon (Lafayette, CO, USA) SMARTpools, except VAPB where a proven single siRNA oligo (Sigma, St. Louis, MO, USA) was used [38]. Control non-target siRNA was a scrambled firefly luciferase sequence (5 -CGUACGCGGAAUACUUCGA-3 ). Following transfection, total protein was extracted for Western blot to determine knockdown or cells were infected with HSV-1 at MOI = 10 for the times indicated.

Rescue Experiments
VAPB was cloned from keratinocyte cDNA into pCDNA3.1 and desensitized to the siRNA by changing CTTGGCTCTGGTGGTT to GTTTGCACTTGTCGTG using CloneAmp HiFi PCR Premix (Clontech, Mountain View, CA, USA). Mutations were verified by Sanger sequencing. Other siRNAs targeted 3 untranslated regions, so mutagenesis was not required for their rescue constructs. HeLa cells expressing rescue constructs were transfected with siRNA for 48 h, then infected with HSV-1 at MOI = 10 for 16 h. As indicated in the figure legends, either released virus was collected by pelleting from the supernatant or cell-associated virus was collected after extensive washing followed by pelleting cells. Titrations were carried out on U2OS cells as described [39]. U2OS cells were used for titrations because they are the most optimized for virus entry and replication and thus enable better measurement of the number of infectious particles.

Isolation of Nuclei and DNA Preparation for qPCR
Cells were scraped into 1 mL PBS then pelleted by centrifugation in an Eppendorf microfuge 1 min at 8000× g, 4 • C and resuspended in 400 µL hypotonic Buffer A (10 mM Hepes, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT, pH 7.9). Cells were lysed by 15 syringe passages through a 21-gauge needle following 15 min incubation on ice. The nuclei were pelleted by centrifugation then broken by sonication. Nuclei were resuspended in DNase digestion buffer and incubated for 1 h, 37 • C using 10 units RQ DNAse 1 (Promega, Madison, WI, USA) followed by enzyme inactivation. DNase-resistant nuclear DNA was prepared using a QiAamp DNA mini kit (Qiagen, Germantown, MD, USA) incorporating RNase and proteinase K digestion.

Western Blotting
Proteins were separated by SDS-PAGE, transferred to nitrocellulose membranes, and blocked with 5% milk powder in PBS containing 0.1% Triton-X-100 (PBST) for 1 h, RT. Primary antibodies were applied for 1-2 h, RT or overnight, 4 • C, washed in PBST, then incubated with anti-mouse IgG or anti-rabbit IgG conjugated to horseradish peroxidase for 1 h. Following washing in PBST, membranes were visualized either using ECL (GE Healthcare) and exposed to Kodak X-Omat S film or analyzed directly on a LICOR Odyssey imager (LI-COR Biosciences, Lincoln, Ne, USA) using antibodies conjugated to fluorescent markers. Antibodies
This provided a pixel size of 0.0645 µm 2 . For general widefield images shown in Figures 1B, 3B, 6B and 7A, an image was taken from a focus point at the midplane of the nucleus as this generally excludes signal from any other staining above or below the nucleus and generally affords the widest view of the ER as well outside the nucleus. For Figure 3E images, image stacks (0.2 µm steps) were deconvolved using AutoquantX (Media Cybernetics, Rockville, MD, USA). Images in Figure 6A were taken using a Zeiss LSM510 Meta confocal microscope. Micrographs were saved from source programs as 12-bit.tif files and analyzed with Image Pro Plus software and/or prepared for figures using Photoshop CS6. Images stained with the same antibodies were taken using identical settings and exposure times. For FISH experiments, DAPI images were used as a mask in ImageJ to determine nuclear area, and the total fluorescence intensity in the nucleus versus the whole cell was determined for >100 cells for each condition. For NE:ER ratio quantification, original images were analyzed in ImageJ using the edge of a DAPI image mask and taking 5 pixels in either direction to define NE fluorescence intensity. Given the 100-200 nm limit of resolution for standard (non-super resolution) fluorescence microscopy and the architecture of the NE, this 300 nm distance from the edge of the DNA signal should capture the full NE signal. ER fluorescence intensity was determined from pixels outside the NE ring and mean pixel intensities were determined by dividing total intensities by area. The NE/ER value for >50 cells per condition was determined and means plotted.

Electron Microscopy
MM pellets were washed in sterile H 2 O and fixed with 2.5% glutaraldehyde and 1% osmium tetroxide, then dehydrated through a graded alcohol series and embedded in Epon 812. siRNA-treated and HSV-infected HeLa cells were fixed with 2.5% glutaraldehyde in 2% sucrose, 0.05M Cacodylate buffer, pH 7.2 overnight at 4 • C. Cells pelleted by centrifugation were fixed with 1% osmium tetroxide (TAAB Labs, Aldermaston, UK) and stained for 1 h, RT with 2% aqueous uranyl acetate. Cells were pelleted through 3% low melting temperature agarose (Geneflow, Lichfield, UK) at 45 • C. The agarose was set at 4 • C and cell pellets were cut into~1 mm cubes, then dehydrated through a graded alcohol series (30-100%) and embedded in Epon 812 resin (TAAB Labs, Aldermaston, UK) followed by polymerization for 3 days at 65 • C. 120 nm sections were cut with a UC6 ultramicrotome (Leica Microsystems, Wetzlar, Germany) and examined in a JEOL 1200 EX II electron microscope, recording images on a Gatan Orius CCD camera. Immunoelectron microscopy was performed using the Tokuyasu method [41]. For imaging, a Hitachi H7600 TEM was used at 100 kV.

Data Availability
The mass spectrometry dataset [raw, peak, search, and DTASelect result files) can be obtained from the MassIVE database via MSV000079886, from the ProteomeXchange via PXD004519, and from the Stowers Original Data Repository at LIBPB-1083.

Identification of Cellular Proteins Potentially Involved in Viral Egress
Cellular proteins involved in nuclear egress must be present and might even accumulate in the NE during infection. Moreover, if they become part of the primary envelope and/or are involved in de-envelopment fusion with the ONM, they may specifically accumulate in the ONM as nuclear egress progresses and possibly also in the contiguous ER after fusion with the ONM. Although their residence in both membranes could be brief due to recycling back to the INM, we reasoned that a proteomics analysis of NE and ER membrane fractions may provide some clues as to the types of cellular proteins that might take part in HSV nuclear egress. We analyzed HeLa nuclear and ER (microsomal) membranes around the time of viral nuclear egress but before secondary envelopment. pUS3, a serine/threonine viral kinase that contributes to primary envelope fusion with the ONM, is expressed before primary envelopment/de-envelopment and peaks roughly when secondary enveloped particles are first detected [13,14]. By contrast, surface glycoprotein gC is expressed later in infection. Western blotting showed that pUS3 began to be highly expressed at 9 h post-infection (hpi), reaching maximal levels by 12 hpi (Figure 1A). At this time, gC was only becoming detectable. A second analysis found that in infected HeLa cells, RFP-labeled capsid protein VP26 was in punctate intranuclear spots at 8 hpi, but had accumulated in larger spots at the NE by 9 hpi that persisted at 10 hpi ( Figure 1B). Therefore, the isolation of NE and microsomal membrane (MM) fractions was performed between 8 and 9 hpi from mock-infected and HSV-1 infected (MOI = 10) HeLa cells. The fractions displayed a number of common and unique components ( Figure 1C). Western blotting revealed that Lamin A/C and LAP2β were only found in the NE fraction as expected, while ER marker calnexin was predominantly in the MM fraction with some in the NE due to ONM-ER continuity ( Figure 1D). Neither NE nor MM fractions contained the Golgi marker GM-130, suggesting low contamination with Golgi membranes involved in secondary envelopment ( Figure 1E). Electron microscopy of the isolated HSV-1 MMs revealed the expected membrane vesicles. Enveloped virus particles were also captured in some vesicles as expected due to the contiguous nature of the NE and ER ( Figure 1F).

Host and Viral Proteins Identified in HSV-1 Infected MMs
MMs from HSV-1 infected cells and NEs and MMs from mock-infected cells were analyzed by mass spectrometry using Multi-dimensional Protein Identification Technology (MudPIT) LC/LC/MS/MS [30,31]. Due to the presence of virions ( Figure 1F), HSV-1 proteins were identified in the infected MMs. All but pUL17 were detected out of 8 viral proteins previously found in a study of primary enveloped virions [42]. We identified pUL31 and its partner pUL34 (Tabls 1 and S1), key players in primary envelopment. Other identified tegument proteins and glycoproteins included gD, gB, gC all reported to associate with HSV-1 primary virions in different studies [43,44] with gB being specifically implicated in ONM fusion during egress [15,16].
Gene ontology (GO) analysis of the proteins identified revealed many proteins with functions in vesicle-mediated transport. Although this category reflects <4% of genes encoded in the human genome, it represented 10% of all proteins detected in the MMs and 12% of all proteins detected in the HSV-1 infected MMs. Moreover, when increasing selection stringency to look at only proteins enriched in the HSV-1 infected MMs, the category jumped to 20% of total proteins. Because it also makes sense scientifically that VFPs could have a potential role in HSV nuclear egress, we chose to pursue this category. We increased selection stringency, only considering VFPs with at least 5 spectra in the NE fraction to suggest that they are reasonably abundant in the NE before infection. We then plotted the abundance ratio of these proteins in the HSV-infected:mock-infected MMs using dNSAF values, a measure of abundance based on both spectral counts and percentage of total mass in the protein fraction [36] (Figure 2A, Table 2). The ratios of NE:mock-infected MM dNSAF values were also plotted ( Figure 2B). It is noteworthy that, while the relative ratios are relevant here, it is not possible to compare absolute amounts in the NE and MMs because of the differences in their volumes and because there are much greater losses of membranes when isolating MMs. Though many VFPs identified would potentially make sense to contribute to nuclear egress, VAPB was selected as a candidate for further analysis because it was enriched in the NE compared to MMs, the most enriched VFP in HSV-1 infected versus mock-infected MMs, and recently indicated to interact with the inner nuclear membrane protein emerin in the IntAct Molecular Interaction Database "https://www.ebi.ac.uk/intact/".

VAPB Expression and Localization During HSV-1 Infection
VAPB is a type II integral membrane protein previously characterized in the ER. Together with family member VAPA, it is thought to function with cytoplasmic vesicle transport proteins and cytoskeletal elements to maintain membrane structure and facilitate lipid transport, membrane trafficking and membrane fusion [45]. VAPB is a C-terminally anchored protein with its primary mass facing the cytoplasm in the ER (nucleoplasm for the inner nuclear membrane population) [46]. The N-terminal region has a major sperm protein (MSP) homology domain that interacts with FFAT motif (two phenylalanines in an acidic track) proteins [47] followed by a coiled-coil domain before the transmembrane segment ( Figure 2C). A mutation (P56S) has been reported in the VAPB MSP domain causing an autosomal dominant form of amyotrophic lateral sclerosis (ALS8) that results in VAPB aggregation and neurotoxicity [48].
VAPB-increased abundance in the infected MMs could be due to HSV-1-induced upregulation of protein expression, virus-induced ER recruitment, or to increased abundance in the membranes due to HSV-1 envelopment/de-envelopment. Total VAPB protein levels in the cell were constant during a time course of infection ( Figure 2D). VAPB has not previously been directly tested for a function in the NE, but its initially reported restriction to the ER has been challenged [46,[49][50][51][52]. To confirm a NE location, a very specific monoclonal antibody was used ( Figure 3A). In the mock-infected cells, considerable ER distribution of VAPB was observed as well as a weak, but distinct rim around the nucleus (defined by the DNA staining) and this staining was largely similar in infected cells ( Figure 3B). To identify infected cells, mock and HSV-1 infected cells were co-stained for VAPB and the herpesvirus US3 protein. Next, an algorithm was employed that measured NE versus ER fluorescence and the mean VAPB fluorescence signal intensities were calculated to identify any redistribution of VAPB ( Figure 3C). A small, but statistically significant increase in the NE pool was observed over the early stages of infection. Over 50 cells were quantified for each condition and no difference was observed in the NE:ER ratio of calnexin during HSV-1 infection. It is important to note, however, that the increase in the NE covered a wide distribution, which might reflect different cells at different stages of infection or might reflect shuttling of VAPB if it is being recycled for repeated use at the NE. Nonetheless, even with the wide distribution, the data indicated the NE:ER ratio increasing from 1.18 to 1.80 between the mock and 16 hpi conditions, respectively ( Figure 3D). Finally, we tested for co-localization between VAPB and NE-located pUL34. The majority of VAPB signal was in the ER while pUL34 concentrated in the nucleus, but both gave clear nuclear rim staining in addition to staining in the ER with the NE pool exhibiting partial overlap between the signals ( Figure 3E). Lower left panels are blowups from the above images, better showing the concentration of VAPB and pUL34 at the NE and that there are both areas with a complete distinction between VAPB (red) and pUL34 (green) signals and yellow areas where the signal overlaps. To quantify distribution, in the middle graph, each signal was quantified in the NE as defined by the perimeter of the DNA staining and signal for the rest of the cell. For both proteins, more signal was observed in the NE than in the rest of the cell. To quantify co-localization, we used Pearson's Correlation Coefficient to quantify overlap VAPB with pUL34 in the NE or the rest of the cell. A much stronger correlation was observed for the NE signal than that in the rest of the cell. The data suggest a partial recruitment of VAPB to the NE during HSV-1 infection.

Knockdown of VAPB Yields Significant Reduction of HSV-1 Viral Titers
To test the role of VAPB in virus replication, siRNA depletion followed by HSV-1 infection was carried out. Rab24, a regulator of intracellular trafficking, was used as a negative control because it did not increase in HSV-1 infected MMs compared to mock-infected MMs. Rab1a was used as a positive control because this protein, required for ER-to Golgi complex transport [53], is involved in HSV-1 mature particle assembly (secondary envelopment) and its knockdown reduces viral growth by 60% [54]. Western blotting showed that VAPB was depleted to nearly undetectable levels and both controls were knocked down by roughly 80% ( Figure 4A). As expected [55], Rab24 knockdown had little effect on viral titers while Rab1a knockdown reduced cell-released viral titers by 62%: VAPB exhibited an even stronger reduction in cell-released virus titers of >90% ( Figure 4B). These were specific effects of the siRNAs as rescue experiments yielded nearly full recovery of virus titers ( Figure 4C). Similar to Rab1a, VAPB depletion resulted in a marked reduction in cell-associated virus titers (<80%) compared to control Rab24 ( Figure 4D); yet, viral genomes were still abundantly produced ( Figure 4E). Moreover, these effects were not likely due to inhibition of production of viral proteins as pUL34 and gC levels were not reduced in VAPB knockdown cells ( Figure 4F). A multistep growth curve (starting MOI = 0.1) revealed around a two log-fold difference between HSV-1 replication in mock versus VAPB siRNA-treated cells ( Figure 4G).

VAPB Knockdown Yields Reduced Cytoplasmic Virus Particles and Nuclear Particle Accumulation
Viral titer reduction could reflect a function of VAPB in nuclear egress or secondary envelopment or cytoplasmic transport. Examination by electron microscopy of HSV-infected, and control-or Rab24-depleted cells revealed the expected distribution with many virus particles present both in the nucleus and in the cytoplasm at 16 hpi ( Figure 5A, upper panels). However, VAPB knockdown resulted in more nuclear and less cytoplasmic virus particles compared to control siRNA-treated cells ( Figure 5A, lower panels). Images focused on the cytoplasm revealed far fewer cytoplasmic particles in the VAPB-depleted cells ( Figure 5B). Quantification of particles in the nucleoplasm, the NE and the cytoplasm revealed an increase in nucleoplasmic and NE virions with a corresponding reduction in cytoplasmic virions in VAPB knockdown cells ( Figure 5C,D). Several images revealed enveloped particles in lumenal extensions of the NE and the presence of the virus envelope could clearly be distinguished by size from the un-enveloped particles in the nucleoplasm ( Figure 5A, lower right panel). Note that the NE lumen is similarly enlarged by accumulating virus particles in US3 deletions [12,13].   (A) In control and Rab24 knockdown cells, some non-enveloped virus particles could be observed in the nucleoplasm (one example shown for each upper image with downward facing arrowheads), but many enveloped and non-enveloped particles could also be observed in the cytoplasm (one example shown for each upper image with upward facing arrowheads) as well as released mature particles just outside the cell (a large accumulation of these is highlighted with the angled arrow in the upper right image). In contrast, for the VAPB knockdown, few particles could be seen in the cytoplasm and visibly more nucleoplasmic non-enveloped particles were observed. In a zoomed image, several particles that had acquired a primary envelope were observed for the knockdowns in the NE lumen (two sets of multiple virus particles indicated by arrows). A single arrowhead highlights an assembled virus nucleocapsid prior to envelopment to contrast for particle size at the different stages. (B) Images focused more on the cytoplasm reveal virus particles in the cytoplasm for the siRNA control but a relative dearth of viral particles in the cytoplasm for the VAPB knockdown. Scale bars, 1 µm. (C) Over 1400 virus particles were counted for each condition and the percentages of nuclear, peri-nuclear (NE lumen), and cytoplasmic particles is given. Virus particles specifically accumulated in the nucleus and peri-nuclear lumen with VAPB knockdown. (D) Plotting the log2 ratio of nuclear:cytoplasmic virus particles for each condition revealed an inversion in the nuclear:cytoplasmic ratio where from particles being slightly more in the cytoplasm for the siRNA and Rab24 knockdown controls they became predominantly nuclear for the VAPB knockdown. Use of the Fisher's exact test reveals very significant differences from the control (**** p < 0.0001; ***** p < 1 × 10 −50 ).
Next, we tested for capsid protein VP5, which in a normal infection generally appears predominantly at the nuclear periphery and aggregates in the cytoplasm because of the rapid trafficking of assembled virions from the nucleus [56]. This is the pattern observed for control siRNA cells, while the VAPB knockdown cells show a strong accumulation of the VP5 in the nucleus ( Figure 6A). We also used fluorescence in situ hybridization (FISH) and viral DNA qPCR to quantify viral genomes. Infected cells with control and VAPB knockdowns were fixed and hybridized with a probe for the HSV-1 ICP27 gene. Pre-treatment with RNase ensured only viral genomes would be recognized. A proportion of nuclear virus genomes will not be encapsidated and this approach will also detect them; however, the aim was to determine egress to the cytoplasm by measuring fluorescent viral genome signals in both the nucleus and cytoplasm. Co-staining with DAPI identified the nuclear boundaries and imaging revealed virus accumulating in the cytoplasm in non-target siRNA control cells, the Rab24 knockdown, and in cells depleted of Rab1a that affects HSV-1 egress in the cytoplasm ( Figure 6B). By contrast, in the VAPB knockdown, the FISH signal for viral genomes was visually restricted to the nucleus ( Figure 6B). Images for this analysis were taken from the nuclear midplane where nuclear area is greatest. While generally the ER sectional area is also greater at the nuclear midplane, it is possible that changes induced by the HSV-1 infection that alter both ER and Golgi membrane distribution might yield a skewed distribution for the positioning of virus particles. Nonetheless, all samples were treated equally and so the differences between the VAPB knockdown and control samples should still be relevant. The intensity of total FISH signal in each cell, and also that just in the nucleus (using the DAPI staining as a mask) was determined. The total signal divided by the nuclear signal was plotted so that the amount of signal outside the nucleus is reflected in values above 1 ( Figure 6C). All three controls exhibited a clear cytoplasmic signal while VAPB knockdown had a value close to 1, indicating mainly nuclear signal. Interestingly, some accumulation of virus genomes at the NE could be observed in the VAPB knock-down ( Figure 6D). Next we quantified DNase-resistant viral DNA (encapsidated genomes) in the nucleus of HSV-infected cells with or without siRNA depletion. Figure 6E shows a significant increase in DNase-resistant viral genome copies (4.5-fold) when VAPB was depleted compared to control Rab24. Rab1a depletion resulted in more DNase-resistant nuclear genome copies than the control, but this difference was not statistically significant.
notable difference was observed in pUL34 distribution between control (siCntl) and the VAPB knockdowns. NE rim staining for pUL34 was observed in individual cells with strong VAPB knockdown. These data show that VAPB contributions to nuclear egress are not indirect effects on the distribution of the most critical viral proteins involved in this step.  . FISH and immunofluorescence also indicate accumulation of nuclear and peri-nuclear virus particles in VAPB knockdown cells. (A) Immunofluorescence staining for capsid protein VP5 at 16 hpi with an MOI = 10 reveals an altered distribution to acccumulate in the nucleus with VAPB knockdown. In the lower panels, the edge of the nucleus defined by DAPI staining of the DNA is marked by a red line so that a portion of the VP5 in the nucleus can be observed in the control knockdown cells, as expected. (B) The virus gene ICP27 was used as a FISH probe and labeled with biotin. Cells were knocked down for VAPB and controls Rab24, Rab1A and non-target siRNA, infected with HSV-1 and at  16 hpi fixed and processed for FISH. The hybridized virus ICP27 gene was visualized with streptavidin conjugated to Alexa488 dye and imaged by immunofluorescence microscopy. Cells were co-stained with DAPI to identify the nucleus. Images were taken at the midplane of the nucleus to minimize any signal from above or below the nucleus. This tends to also maximize the sectioned area of the ER/Golgi. Scale bar, 10 µm. (C) Using the DAPI nuclear staining to generate a mask of the nuclear area, the nuclear pools of hybridized virus ICP27 DNA were quantified from roughly 50 cells for each condition. Our previous studies on genome organization have shown that this 2D analysis is sufficient to see differences revealed by more intensive 3D imaging. The total hybridized ICP27 DNA in the same cell was also quantified and plotted divided by the nuclear signal so that values above 1 reflect the cytoplasmic pool of viral genomes. A clear increase in cytoplasmic viral genomes can be seen for the non-target siRNA control, the Rab24 and the Rab1A knockdowns, while no notable increase in cytoplasmic viral genomes was observed for VAPB knockdown. Statistical measurements were performed using a 2-tailed ANOVA analysis: *** p < 0.001. (D) The same images of cells analyzed for FISH were also counted for the visible accumulation of viral genome signals at the NE. The percentage is plotted and statistical significance from Fisher's Exact test is given as: * p < 0.05. (E) In a separate experiment, DNase-resistant viral genomes were isolated by preparing nuclei from HSV-infected cells (MOI = 10, 16 hpi) following siRNA knock down. Nuclei were disrupted and incubated with DNase to remove unencapsidated DNA. DNA was prepared using a QIAamp DNA purification kit with RNase treatment to remove viral mRNAs. Remaining viral DNA was quantified by qPCR. Increases in encapsidated viral DNA of greater than 4-fold were observed upon siRNA depletion of VAPB, but not the control, Rab24. The data are from a single experiment and standard deviations and statistics are generated from three technical replicates with an ANOVA of p = 5.6 × 10 −05 : pair-wise Tukey test ** p < 0.01.

VAPB Knockdown Does Not Interfere with Virus Protein Accumulation at the NE
VAPB knockdown was previously reported to interfere with NE targeting of NPC proteins gp210 and Nup214 and the INM protein emerin [38]; thus, its knockdown could potentially interfere with INM accumulation of pUL34 that is essential for primary envelopment. To test if VAPB indirectly alters egress, we stained for pUL34 in HSV-1 infected VAPB-depleted cells ( Figure 7A). No notable difference was observed in pUL34 distribution between control (siCntl) and the VAPB knockdowns. NE rim staining for pUL34 was observed in individual cells with strong VAPB knockdown. These data show that VAPB contributions to nuclear egress are not indirect effects on the distribution of the most critical viral proteins involved in this step.
Confocal and EM images did not indicate major changes in the NE upon VAPB depletion. In case VAPB depletion resulted in gross changes to other cellular membranes that might affect HSV cellular egress, we determined the location of calnexin, an ER marker, and GM-130, a Golgi marker ( Figure 7B). The pattern of staining with antibodies against either protein was generally similar between control and VAPB depleted cells, suggesting that VAPB depletion does not affect the gross structure of other membranous compartments though minor effects on Golgi structure and function remain possible.

VAPB Is Present on the NE and in Association with Viral Particles
Immunogold labelling electron microscopy was used to visualize VAPB in the NE. Although typically only 3-5 particles were observed in a NE region captured in a particular section, nearly all images examined had particles in both INM and ONM in both the mock-infected and HSV-1 infected cells ( Figure 8A,B). As expected, particles were also observed in association with the ER (Figure 8A, ER membrane delineated by arrowheads). Importantly, some virus particles inside nuclear invaginations that appear to be luminal extrusions contained VAPB-labelled gold particles ( Figure  8C). These generally displayed tightly packaged genomes, an apparent envelope, and were proximal to other cellular membranes and DNA, suggesting that they are primary enveloped particles captured

VAPB Is Present on the NE and in Association with Viral Particles
Immunogold labelling electron microscopy was used to visualize VAPB in the NE. Although typically only 3-5 particles were observed in a NE region captured in a particular section, nearly all images examined had particles in both INM and ONM in both the mock-infected and HSV-1 infected cells ( Figure 8A,B). As expected, particles were also observed in association with the ER ( Figure 8A, ER membrane delineated by arrowheads). Importantly, some virus particles inside nuclear invaginations that appear to be luminal extrusions contained VAPB-labelled gold particles ( Figure 8C). These generally displayed tightly packaged genomes, an apparent envelope, and were proximal to other cellular membranes and DNA, suggesting that they are primary enveloped particles captured in the NE lumen. The image in Figure 8D shows a particle that is clearly in the NE lumen (L). Other sections that appear to be glancing sections at the nuclear surface often contained many virus particles with less densely packaged genomes associating with VAPB-labelled gold particles ( Figure 8E). The staining for VAPB on nuclear virus particles suggests the possibility that VAPB participates in the process of primary envelopment. in the NE lumen. The image in Figure 8D shows a particle that is clearly in the NE lumen (L). Other sections that appear to be glancing sections at the nuclear surface often contained many virus particles with less densely packaged genomes associating with VAPB-labelled gold particles ( Figure  8E). The staining for VAPB on nuclear virus particles suggests the possibility that VAPB participates in the process of primary envelopment.  Note that counted luminal particles may reflect sectioning angles where the gold particle artificially appears to be on the other side of the membrane, but as this cannot be ascertained for certain, they were counted separately. (C) Left: high magnification images of primary enveloped HSV-1 particles (PEPs) with VAPB-labelled gold particles that appear to be in membrane-bound invaginations of the NE. In the left image, the arrow points to the viral envelope and the arrowheads point to areas where the membrane of the invagination can be distinguished with N and C demarcating the nuclear and cytoplasmic compartments. The invaginated membrane also is associated with gold particles. (D) An enveloped HSV particle with VAPB-labelled gold particles clearly inside the NE lumen. In addition to the N and C demarcating the nuclear and cytoplasmic compartments, the INM and ONM are also highlighted by arrowheads facing one another. (E) Nuclear virus particles that appear to lack an envelope in association with VAPB-labelled particles. The sections have the appearance of being glancing sections at the surface of the nucleus. The two arrows in the left panel indicate gold particles associating with virus particles. All scale bars are 100 nm.

Discussion
It has been proposed that host cell proteins could play a role in primary envelopment and nuclear egress [21] and such a role for several host proteins has already been indicated. During HSV-1 infection activated protein, kinase C is recruited to the nuclear membrane, where it phosphorylates and remodels the nuclear lamina to facilitate primary envelopment [57]. p32 is a cellular protein that can interact with a number of herpesvirus proteins and in HSV-1 infection, it is recruited to the nucleus/nuclear rim by ICP27 and ICP34.5 [23,58], but may play a role in nuclear egress via interaction with tegument protein UL47 [24]. Two proteins, CD98 and β-integrin, can act as regulatory proteins in the case of fusion of enveloped viruses with the plasma membrane [59][60][61][62] but during HSV-1 infection, they can also be recruited to the nuclear membrane where, through interaction with key viral proteins including UL31 and US3, they may indirectly control de-envelopment [22].
Our approach to identifying additional cellular proteins was to examine changes in the proteome of NE and ER membranes comparing HSV-1 infected and mock infected HeLa cells. While there are many proteins that might change in levels due to HSV-1 infection or relocate to these membrane compartments for other reasons as a consequence of HSV-1 infection, it is likely that within the set of changing proteins would also be those involved in egress. This is because participation in egress requires first a physical presence in the NE and any INM proteins captured in the primary envelope would enter the ONM upon fusion, from whence they could diffuse into the ER. Thus, while proteins fulfilling these criteria would not necessarily be involved in egress, it is likely that a subset would. Therefore, this approach was used to identify candidates to be further tested for possible functions in egress. Interestingly, VFPs comprised of a key protein group whose presence in the NE/ER was significantly increased during HSV-1 infection. In particular, among roughly 50 VFPs identified in the NE, only a few exhibited the expected characteristics, thus identifying as a candidate VAPB, an ER resident protein whose levels do not change during HSV infection and which we found to be largely required for HSV-1 replication. It facilitates cytoplasmic viral particle accumulation, localizes to the NE in association with primary enveloped virions and co-locates with viral nuclear egress protein pUL34. These data suggest a role for VAPB in HSV-1 nuclear egress. Although exogenous overexpression of viral pUL31/34 proteins promotes formation of NE vesicles and their in vitro expression can bend membrane vesicles [10,11], it seems likely that during infection vesicle fusion, proteins could be used to make egress more efficient. Herpesviruses frequently co-opt host cell machinery to support various aspects of their life cycle [17,18] and ESCRT pathway proteins were recently shown to be involved in nuclear egress of Epstein-Barr virus (EBV) [63]. In this case, the Chmp4b protein important for scission complex assembly co-localized in perinuclear aggregates with EBV protein BFRF1 and inhibition of the Alix bridging protein yielded capsid protein accumulation in the nucleus [63]. Notably these proteins were not required for HSV-1 egress, suggesting that other host cell vesicle trafficking proteins function instead of the ESCRTs for this virus.
VAPB is a member of the Vesicle-associated membrane protein (VAMP)-Associated Protein family of C-terminal tail anchored proteins. It has been shown to function as an adaptor to recruit target proteins to the ER and execute cellular functions such as lipid transport, membrane trafficking and membrane fusion [45]. It also functions in calcium homeostasis and has an additional mitochondrial function reported [49,52]. An indirect NE link was reported for VAPB, where its knockdown resulted in reduced NE accumulation of NPC proteins gp210 and Nup214 and the INM protein emerin [38]. However, it was thought that this reflected an indirect consequence of VAPB knockdown, inducing Golgi fragmentation and disruption of ER to ERGIC (ER-Golgi Intermediate Compartment) to Golgi trafficking [64,65]. As we found clear VAPB accumulation in the NE, these data might be re-interpreted to consider a nuclear membrane trafficking role and/or a role in fusion events during NE assembly at the end of mitosis. The latter is consistent with the ESCRT proteins shown to be involved in EBV nuclear egress also functioning in NE reformation after mitosis [66]. Our data are also consistent with a potential direct function in nuclear egress as VAPB knockdown did not block NE accumulation of pUL34. This does not discount, however, the possibility of VAPB having an additional role in secondary envelopment.
A critical outstanding question is whether VAPB contributes to primary envelopment or de-envelopment or both during nuclear egress. Immunogold EM labelling revealed VAPB in association with nucleoplasmic/luminal virions in a membranous vesicle and, together with the nucleoplasmic accumulation of virus particles, this suggests a role in primary envelopment. As VAPB is in both the INM and ONM according to the immunogold EM data, it could potentially contribute to both primary envelopment and de-envelopment; however, the topology of VAPB is such that its principal functional mass should be facing the nucleoplasm in the INM and the cytoplasm in the ONM. Thus, it is perfectly poised to potentially interact with virus nucleocapsids during primary envelopment, but it should not be positioned to interact with the primary enveloped particles in the nuclear envelope lumen in order to initiate fusion with the ONM for de-envelopment unless, as has never been tested, it assumes a different topology during HSV-1 infection. Nonetheless, once the initial fusion has begun, VAPB either in the primary envelope or in the ONM could potentially be positioned to contribute to stabilizing and/or completing the process of nuclear egress. Likewise, and perhaps more likely, the separate role reported for VAPB in calcium homeostasis could lead to indirect effects on the fusion step [49,52]. In considering a potential de-envelopment function for VAPB, it is noteworthy that previous studies of HSV-1 US3, gB and gH depletion or mutants revealed an accumulation of primary enveloped particles in both the nucleoplasm and in the NE lumen, the expected phenotype for a role in the de-envelopment phase of nuclear egress [12,13,15,16] and, similarly, both nucleoplasmic and luminal viral particle accumulation were observed with VAPB knockdown. However, further experiments are required to identify the precise stage at which VAPB acts, and these are ongoing.
A potential function for VAPB in nuclear egress does not contradict recent papers arguing that pUL31 and pUL34 are sufficient for primary envelopment, as these studies showed induction of membrane invaginations in vitro in one study [10] and in context of a cell still containing VAPB in the other [11], while not addressing the issue of efficiency compared to a wild-type infection. Though with greatly reduced titers, HSV-1 has been shown to still get out of the nucleus in the absence of pUL31/34 [12][13][14]67,68]. Nonetheless, as pUL31/34 are central proteins for nuclear egress, it will be important to determine if VAPB contributions to nuclear egress result from participating in functions with or independent of these critical viral proteins. Thus far, our efforts to test for an interaction between VAPB and pUL34 have been negative; however, VAPB could also interact with HSV-1 capsid/tegument proteins or contribute in as yet undetermined ways. The lack of clarity in the composition of primary enveloped particles allows for a very wide range of potential partners, so this may be a significant undertaking in the future. If VAPB interacts with virus proteins, it could also provide a mechanism for specific recruitment of VAPB to egress complexes; however, it is also possible that HSV-1 disruption of ER and Golgi membranes unintentionally results in a redistribution of VAPB where it can indirectly contribute to nuclear egress and it remains possible that the trafficking of some important player in nuclear egress besides pUL34 is not able to gain access to the NE in the VAPB knockdown cells.
The demonstration of significant defects in virus nuclear egress for the VAPB knockdown suggests that other of the VFPs similarly highlighted by our study might also be involved in HSV-1 nuclear egress. This idea is more compelling by the general functioning of VFPs in larger complexes and observations that VAPB itself interacts with other proteins identified or closely related proteins. For example, Rab11, which was also highlighted by our mass spectrometery analysis, functions in a complex with VAPB and several other proteins [69,70]. Separately VAPB has been shown to interact with VAMP1 and 2 [51], both members of the same family, as VAMP7 that was similarly highlighted in our mass spectrometry analysis. Moreover, the IntAct Molecular Interaction Database [71] reveals that VAMP7 is involved in several different vesicle membrane trafficking complexes including multiple SNARE complexes, one of which includes Stx7 that was also highlighted by our mass spectrometry analysis. Compellingly, IntAct also reveals VAPB interactions with the inner nuclear membrane protein emerin and multiple virus proteins from Hepatitis C Virus [72].
A study reporting co-localization of gB, gH, pUL31 and pUL34 at the NE together with cellular proteins CD98hc and β1-intergrin argued for formation of a complex with a role in HSV nuclear de-envelopment [22]. Just as the viral proteins in that study might co-opt host cellular proteins, we postulate that VAPB forms part of a larger egress complex together with viral proteins [73]. The reduction in cytoplasmic particles and virus titers with VAPB knockdown and the labelling of virus particles with VAPB antibodies strongly argues that VAPB is directly involved in HSV-1 nuclear egress.