Proteomic Comparison of Three Wild-Type Pseudorabies Virus Strains and the Attenuated Bartha Strain Reveals Reduced Incorporation of Several Tegument Proteins in Bartha Virions

To generate and collect virions for mass spectrometry, ST cells were infected with either Bartha, Becker, Kaplan, or NIA3 PRV strains at a multiplicity of infection (MOI) of 10, and supernatants were harvested at 24 h postinfection (hpi). To confirm efficient virus replication in ST cells, infected cells were harvested and analyzed by Western blotting. Figure 1A demonstrates expression of the major viral capsid protein VP5 in the infected cells for all PRV strains. Moreover, as expected due to the large US deletion in the Bartha genome, Bartha-infected cells did not express gE. Extracellular virions were collected from supernatants of infected cells by ultracentrifugation through a Ficol-400 cushion, as described before to isolate mass spectrometry (MS)-grade PRV or HSV-1 virions. Based on these prior studies, it is reasonable to assume that these preparations are devoid of noninfectious light particles (L-particles) (40, 49). Virus preparations were subsequently subjected to Western blot analysis, revealing similar quantities of virions based on VP5 levels and again confirming the absence of gE in Bartha virions. Tubulin was not detected in the purified virions, indicating that any possible contamination of cellular material is minimal or nonexistent (Fig. 1B).

For MS analysis, virus preparations were lysed, separated by SDS-PAGE, and visualized by Coomassie blue staining (Fig. 1C). Distinct bands were noted for all four virus preparations similar to what has been observed by others, making it possible to distinguish several abundant viral proteins with a high degree of certainty (see arrows in Fig. 1C) (40, 50). For each replicate, all the proteins of the entire lane were then in-gel digested, extracted from the gels, and processed for MS analysis.

Since we aimed to compare the proteome of the four viral strains in this study, we needed to correct the peptide data set for strain-specific differences, i.e., point mutations. Therefore, we deployed a custom analysis leveraging the Progenesis QIP match-between-runs label-free quantification strategy. More specifically, all runs were aligned in the MS1 space, and all fragmentation spectra acquired on a specific peptide over the experiment were aligned to a single “precursor feature” in a single project. This way, nonidentical peptides between strains can be discarded by serially performing the peptide annotation against every strain, creating a data set with only those peptides that are conserved in all four strains. Unfortunately, this approach also inadvertently discarded all the peptides of pUS2, gE, gI, and pUS9 due to the absence of the corresponding protein sequences in the Bartha proteome. Therefore, another Progenesis QIP project was created for which the peptides were not annotated against the Bartha strain, allowing the identification of pUS2, gE, gI, and pUS9 in the WT strains.

In most proteome analyses, at least two unique peptides are required to claim protein identification. Although the aforementioned filtering step sacrificed several nonconserved viral peptides, this only resulted in a minor reduction in unique peptides. In fact, pUL34 was the only protein that fell below the identification threshold due to this filtering process. Using this approach, we identified a total of 37 structural viral proteins (Table 1) with at least two unique peptides by data-dependent acquisition (DDA). In addition, we detected six proteins that were presumed to be nonstructural (Table 2). Notably, these six proteins were also observed in Becker virions in a proteomic study by Kramer et al. (40), whereas 3 of these proteins were observed in Bartha virions in the proteomic study by Yu et al. (41). The only established structural proteins that were not identified in our study were gL, gK, gG, and EP0 (Table 1).

To increase the sensitivity of the analysis, we reanalyzed all the samples in the same randomized order, this time operating the MS instrument in data-independent acquisition (DIA) mode (SWATH). As recently shown, using predicted spectral libraries increases the depth of the detectable proteome (47). Here, predicted libraries were used to search for peptides using the neural network-based DIA-NN search engine, and peptides were considered detected upon a minimum posterior error probability value below 0.01 (51). This approach allowed us to confirm the presence of the structural proteins EP0, gL, and gK. However, for gG, detection remained uncertain as only 1 unique peptide was detected (Table 3). Moreover, using this SWATH data set, we additionally identified the presumed nonstructural proteins pUL2, pUL5, pUL8, pUL14, pUL23, pUL29, pUL31, and pUL34 (Table 3). Similar unique peptides were found for all four strains for most of these proteins, except for pUL29, which had less peptides detected in Becker and Kaplan virions. Curiously, we identified both the proteins pUL31 and pUL34, albeit in very small amounts. Both proteins are thought to serve as markers for immature, primary virions (52), yet pUL34 was previously also detected in mature PRV virions, while both pUL31 and pUL34 have been found in HSV-1 virions (40, 53, 54).

We next performed a differential proteome analysis of Becker, Kaplan, NIA3, and Bartha virions. Therefore, the viral proteins for each strain were quantified based on the data set of conserved peptides in the DDA background (Tables 1 and 2) and protein abundances were normalized to the major capsid protein VP5 (4). Figure 2 shows the normalized protein abundance organized by virion localization for each of the PRV strains. Of note, the normalized abundance of gE, gI, pUS2, and pUS9 (which are absent in Bartha) were superimposed based on the approach mentioned in the previous section. It should also be pointed out that these normalized abundances do not allow strong claims about the absolute abundance of individual proteins in virions because of potential differences in detectability of individual peptides. It is therefore not strictly possible to compare abundance of different proteins within the same virus preparation. However, since the same peptides were used to quantify viral proteins in each strain, it is possible to compare relative abundance of individual proteins in-between strains. Even though label-free liquid chromatography-mass spectrometry (LC-MS) analysis does not allow for absolute quantification, VP5 was by far the most abundantly detected protein (Fig. 2), which is in line with its dominant abundance in virions as the major capsid protein and correlates with the strong VP5 signal observed in the Coomassie blue staining in Fig. 1C. Similarly, gB and tegument proteins VP13/14 (pUL47), VP22 (pUL49), and VP1/2 (pUL36), which are known to be present in high copy numbers in the viral particle, also stand out as highly abundant in the proteome analysis (Fig. 2). At the other end of the spectrum, pUL6, of which only 12 copies are present in each capsid (55), shows the lowest abundance of all capsid proteins (Fig. 2). When comparing the different strains, all of the capsid proteins depict a near-identical virion abundance across strains, which is in accordance with the stringent position of these proteins in the icosahedral capsid structure (4). Similar as observed for the capsid proteins, all envelope proteins display fairly similar distributions among the four PRV strains, with the obvious exception of gE, gI, and pUS9, which are absent in Bartha as the corresponding genes are (partially) deleted from its genome. The same holds true for the tegument protein pUS2, which is also absent in Bartha virions. Interestingly, tegument proteins VP11/12 (pUL46) and VP22 are both drastically less incorporated in Bartha virions, which corresponds with previous observations by Western blotting and mass spectrometry (42, 56). Remarkably, the tegument protein IE180 that represents the only immediate early protein of PRV is almost completely absent in Bartha virions, even though it shows substantial and well-conserved abundance in all the WT strains. This virtual lack of IE180 in Bartha virions is also noticeable on the Coomassie blue staining, as shown in Fig. 1C.

To better evaluate the observed differences in-between strains, a data set was created showing the normalized protein abundance of each strain set relative to the Kaplan strain (Fig. 3). This further supports the notion that all capsid proteins have a highly similar abundance in all the virus strains (Fig. 3, upper graph). However, for the tegument proteins, we noticed that several proteins were prone to high interstrain variation, whereas others showed very little interstrain variation. Moreover, of those proteins that show the highest variance, almost all were drastically reduced in Bartha (Fig. 3, middle graph). Indeed, the analysis not only pointed out that the relative amounts of the VP11/12, pUS3, VP22, and IE180 proteins are severely compromised in Bartha virions, in line with the analysis in Fig. 2 and published literature (42, 56), but also that pUL41 shows a reduced abundance in Bartha virions compared to all the WT strains (Fig. 3, middle graph). This demonstrates that the number of tegument proteins that display reduced incorporation in Bartha virus particles stretches substantially beyond the previously described pUS3, VP11/12, and VP22 proteins (50, 56). Becker and NIA3 virions, on the other hand, show an overall increased incorporation of these tegument proteins compared to Kaplan virions. Of interest, many of the tegument proteins that have a highly similar abundance in between PRV strains (VP13/14, VP1/2, pUL37, VP16, pUL21, and pUL16) have been described as inner tegument proteins or closely associated with the capsid, whereas the more variable tegument proteins mainly represent outer tegument proteins (6). The envelope proteins on the other hand show a fairly similar incorporation in-between strains, with the exception of pUL43 and pUL56 (Fig. 3, lower left graph). Whereas the former is substantially more abundant in Becker and NIA3 virions, the latter is distinctly less abundant in Bartha virions. Of interest, gM and gN, which form a disulfide-linked heterodimeric complex (57), appear to show the least variability with regard to virion incorporation. Lastly, with regard to the presumed nonstructural proteins, pUL50 displays an almost uniquely high abundance in Becker virions (Fig. 3, lower right graph). In conclusion, despite overall strong similarities in viral protein incorporation in virions of different WT PRV strains, we observed some drastic strain-specific differences with regard to the incorporation of selected viral proteins into the PRV virion. Most notably, we found that Bartha PRV not only shows reduced (and in some cases virtually absent) virion incorporation of pUS3, VP22, and VP11/12 but also of additional tegument proteins (IE180 and pUL41) and an envelope protein (pUL56).

Proper protein quantification based on peptide measurements is a challenging aspect of proteomics. Therefore, to further confirm our observations and apply statistical analysis, viral peptide measurements were statistically analyzed in six pairwise comparisons using MSqRob and represented as volcano plots (Fig. 4) (58). Note that gE, gI, pUS9, and pUS2 were not included in the pairwise comparison, since these proteins are not present in the Bartha strain, and superimposition of the measurements would falsify the statistical analysis. Clearly, the reduced incorporation of the proteins IE180, VP11/12, pUS3, VP22, pUL41, and pUL56 in Bartha virions compared to any of the WT strains is apparent. In addition, also the presumed nonstructural proteins pUS1 and pUL40 were less incorporated in Bartha virions compared to the WT virions. This was statistically confirmed, as -log₁₀ q values above 1.3 are considered statistically significant (58, 59). Similarly, this analysis confirms the increased incorporation of pUL50 in Becker virions. (Fig. 4).

The observation that Bartha PRV incorporates less pUS3, VP11/12, and VP22 in virions has been linked to mutations in the UL21 gene of Bartha, which encodes the pUL21 tegument protein (42). Indeed, the Bartha-derived strain Ba43/25aB4, in which the mutated UL21 locus as well as the large US deletion in the Bartha genome were repaired, shows a substantially increased incorporation of pUS3, VP11/12, and VP22 in its virions which was not observed when only the US region was repaired (42). Therefore, we checked whether the virion incorporation of IE180 was restored in virions of the Ba43/25aB4 strain. Western blot analysis revealed, as expected, equal amounts of the major capsid protein VP5 and the tegument protein VP16 in all virus preparations and an absence of gE in Bartha, but not in the Bartha partial-rescue strain Ba43/25aB4 in which the US deletion and UL21 mutation of Bartha are repaired (Fig. 5A). Nevertheless, we only noticed a marginal increase in IE180 protein levels in Ba43/25aB4 virions compared to Bartha virions (Fig. 5A). Interestingly, the incorporation of pUS3 in Ba43/25aB4 virions was increased compared to Bartha, as expected (42), yet remained substantially lower compared to WT strains. Hence, although repair of the US region and the UL21 locus in the Bartha genome partially restored packaging of pUS3 and, to a lesser extent, IE180, the current data indicate that additional mechanisms are involved in the packaging defect of Bartha.

Comparison of WT PRV and Bartha virions was done at 24 hpi. Previous research showed that PRV virions are assembled and/or released earlier in cells infected with PRV Bartha compared to cells infected with WT PRV (60). Hence, to assess whether or not the reduced incorporation of pUS3 and IE180 in Bartha virions may be time dependent, we analyzed Bartha virions harvested at 12 hpi. Figure 5B shows that also at 12 hpi, Bartha virions contain almost no IE180 and substantially reduced amounts of pUS3, similar to the observations at 24 hpi. This indicates that also early during infection, the incorporation of tegument proteins is altered in Bartha-infected cells.

An alternative explanation for the largely pUL21-independent reduced incorporation of IE180 in Bartha virions could be an altered subcellular distribution of IE180 in Bartha-infected cells versus WT PRV-infected cells, leading to reduced virion incorporation. Indeed, for HSV, it was shown that ICP4 (the IE180 ortholog in HSV) needs to interact with ICP27 (pUL54) to induce its cytoplasmic translocation and subsequent virion incorporation (61). However, fractionation assays revealed similar amounts of IE180 in the cytoplasm and nucleus of Bartha- versus Kaplan-infected cells at different time points post inoculation (Fig. 6). As a control for successful fractionation, the nuclear marker histone H3 was nearly undetectable in the cytoplasmic fraction, whereas no tubulin could be detected in the nuclear fractions (Fig. 6). In line with this, immunofluorescence assays did not show obvious differences in subcellular IE180 localization in Bartha- versus Kaplan-infected ST cells (data not shown).

It is remarkable that several abundant tegument proteins, often proteins that make part of the outer tegument layer, display a substantially reduced incorporation in Bartha virions, seemingly without compensation by other viral proteins (Fig. 3). We therefore wondered whether this difference would correlate with a reduced size of Bartha virions compared to that of WT virions. Figure 7A shows that the average particle area of Bartha virions is indeed significantly smaller compared to that of WT virions (respectively, 16,507 ± 1,521 nm² and 17,141 ± 1,634 nm²). Moreover, we also observed that the average circularity of Bartha virions is slightly but significantly decreased (Fig. 7B) (respectively, 0.935 ± 0.016 and 0.944 ± 0.021). A possible explanation for these observations could be that the reduced abundance of tegument proteins results in an overall compacted virion. In line with this, Bartha virions often display an icosahedral shape that more closely matches the icosahedral shape of the capsid, compared to the rounder, more flexible shape of WT virions (Fig. 7C). We hypothesize that, as a consequence of the reduced incorporation of tegument proteins, the rigid icosahedral shape of the capsid has a greater influence on the overall shape of the membrane envelope of Bartha virions. This appears to be in line with the observation that the inner tegument layer in HSV-1 exhibits an icosahedral structure (62).

A total of 128 host proteins with at least two unique peptides associated with the PRV virion preparations were identified in the DDA analysis (Table 4). Unfortunately, the current experimental setup does not allow a clear distinction between proteins that are truly packaged inside the virions or proteins that are associated with the virions. In other studies, proteinase K treatment was used to remove proteins that were not incorporated into PRV virions (40, 41). Upon comparison with these reports assessing packaged host proteins, we identified a total of 28 host proteins in our PRV virion preparations that were discovered in at least one of these studies (Table 4). We next investigated how the DIA approach (SWATH) could increase the detection sensitivity of the host proteins. Similar as for the viral proteins, predicted libraries were used to search for peptides using the neural network-based DIA-NN search engine, and peptides were considered detected upon a minimum posterior error probability value below 0.01 (57). By these standards, we identified 678 host proteins with at least two unique peptides in at least one of the virus strain preparations. (see Data Set S1).

As a subsequent step, we wanted to explore whether the acquisition of host proteins in viral particles exhibits a certain strain specificity. Therefore, it was investigated which host proteins showed a statistically significant difference between strains (indicated in Table 4). The 54 host proteins that did show significant differences were then compared to the VP5-normalized abundance in Kaplan (Fig. 8). However, given that most of these host peptide measurements reside in virion preparations near the limit of detection of the DDA acquisition, interpretation of these relative abundances should be done with caution. Due to these low abundances, quantification and comparison of several host proteins were more susceptible to batch effects compared to the analyses of viral proteins, resulting in several large error bars in Fig. 8 and making it difficult to draw robust conclusions. Even so, it appears that, on average, fewer host proteins were detected in Becker virion preparations. Furthermore, specific host proteins appeared to show a strain-specific abundance. In particular, virion preparations of the Bartha strain display reduced abundance of the mitogen activated kinase 3 (MK03 or ERK1), the protein phosphatase 1 catalytic subunit α (PP1A), the adapter molecule crk (CRK), and the casein kinase II subunit α (CSK23), whereas an increased abundance of the carboxypeptidase D (CBPD) and the growth factor receptor-bound protein 2 (GRB2) was observed (Fig. 8). However, additional research is required to further clarify the suspected differences and elucidate whether and how these may affect virus biology.

Epithelial swine testicle (ST) cells were cultured in Earle’s minimal essential medium (MEM) supplemented with 10% fetal calf serum (FCS), 1 mM sodium pyruvate, and antibiotics (100 U/mL penicillin, 0.1 mg/mL streptomycin, and 0.05 mg/mL gentamicin) (Life Technologies) (= ST medium). All viruses used in study this have been described before. The WT PRV Kaplan strain (27) and the Bartha rescue strain Ba43/25aB4 (US region + UL21) (105) were kindly provided by T. Mettenleiter (Friedrich-Loeffler Institute, Germany). The WT Becker strain (28) was kindly provided by L. Enquist (Princeton University, USA). The WT NIA3 strain (29) was kindly donated by the ID-DLO (The Netherlands). The Bartha vaccine strain (30) was kindly provided by H. Nauwynck (Ghent University, Ghent, Belgium). All viral stocks were titrated on monolayers of ST cells.

The antibody directed against the PRV gE (18E8 and mIgG1) was described before (106). Antibodies directed against VP5 (3C10) and pUS3 (8F86, mIgG1) (107) were kindly provided by L. Enquist (Princeton University, USA). Antibodies raised against PRV VP16 (polyclonal rabbit) (10) were kindly donated by W. Fuchs (Friedrich-Loeffler Institute, Germany), and those raised against IE180 (polyclonal rabbit) (108) were kindly provided by E. Tabares (Autónoma University of Madrid, Madrid, Spain). Antibodies against histone H3 were purchased from Proteintech (polyclonal rabbit, 17168-1-AP) and horseradish peroxidase (HRP)-labeled alpha-tubulin antibodies were obtained from Abcam (Ab40742). HRP-labeled goat anti-mouse and goat anti-rabbit (Dako, catalog numbers P0447 and P0448, respectively) were used as secondary antibodies in Western blot analyses.

The isolation of PRV virions (H-particles) was performed as previously described with some minor adjustments (40). Briefly, two 175-cm² flasks with confluent ST cells were infected with Becker, Kaplan, NIA3, or Bartha PRV at an MOI of 10 in MEM at 37°C for 2 h; the cells were then washed twice and overlaid with ST medium without FCS. The supernatant was collected at 24 hpi, unless mentioned otherwise, and cell debris was removed by centrifugation for 10 min at 1,000 × g and 0.45-μm-pore size filtration. In contrast to the earlier described protocol (40), no DNase treatment was performed. The virions were pelleted for 1 h at 20,000 × g using a Type-35 rotor (Beckman Coulter), resuspended in phosphate-buffered saline (PBS), briefly sonicated, carefully layered onto a 10% Ficoll 400 solution, and subsequently centrifuged for 2 h at 116,000 × g in a SW41-Ti rotor (Beckman Coulter). Pelleted PRV virions were resuspended in HEPES buffer (20 mM HEPES [pH 7.4]) and pelleted for 1 h at 20,000 × g in the same rotor. Pelleted virions were resuspended in HEPES buffer, aliquoted, and stored at −80°C until further use.

ST cells were inoculated with the indicated PRV strains at an MOI 10 in MEM at 37°C for 2 h; the inoculum was then removed, and the cells were washed twice and overlaid with prewarmed ST medium. At the indicated time points, the cells were scraped into the medium and pelleted by centrifugation for 10 min at 700 × g (4°C). The cell pellet was washed with ice cold PBS and lysed for 1 h at 4°C in TNE lysis buffer (50 mM Tris, 150 mM NaCl, 1 mM EDTA, 10% NP-40 [Roche]) and protease inhibitor cocktail (Sigma-Aldrich). Nuclei were pelleted by spinning 10 min at 10,000 × g, and the supernatant was collected as the cytoplasmic lysate. If appropriate, the pelleted nuclei were washed twice in PBS and lysed with radioimmunoprecipitation assay buffer supplemented with protease inhibitor cocktail (Abcam ab156034) for 1 h at 4°C. Purified PRV virions were lysed and heated for 5 min at 95°C in SDS-PAGE loading buffer without β-mercaptoethanol or bromophenol blue. Protein concentrations of the cell and viral lysates were determined using a Pierce BCA kit according to the manufacturer’s instructions (Thermo Fisher), and the cell lysates were then mixed with SDS-PAGE loading buffer and heated for 5 min at 95°C. β-Mercaptoethanol and bromophenol blue were added to the viral lysates used for Coomassie staining and Western blotting. A total of 20 μg (cell lysates) or 5 μg (virion lysates) of protein was loaded and run on a polyacrylamide gel (10%) via SDS-PAGE and either blotted on a Hybond-P polyvinylidene difluoride membrane (GE Healthcare) or directly visualized using Coomassie brilliant blue staining (Thermo Fisher). For SDS-PAGE gels, samples were further processed for MS analysis as described below. The membranes for Western blotting were blocked in blocking buffer (PBS with 5% milk powder [Nestlé] and 0.1% Tween 20 [Sigma-Aldrich]) for 1 h at room temperature. Next, blots were incubated with primary antibodies at 4°C overnight, washed, and subsequently incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Primary and secondary antibodies were diluted in blocking buffer. Finally, the blots were developed using chemiluminescence.

Virions of each strain were purified as described above in two independent runs, and three replicates of each run were included, creating a total of six replicates that were divided over six different SDS-PAGE gels. For each gel, the entire lane containing the fractionated proteins was cut into small pieces of about 1 × 1 mm and transferred to protein low-bind tubes (Eppendorf). Gel pieces were washed three times for 10 min with 25 mM triethylammonium bicarbonate buffer (TEABC; Sigma-Aldrich) in 50% acetonitrile (ACN; Sigma-Aldrich) under mild agitation. The proteins were then reduced in 10 mM dithiothreitol (Sigma-Aldrich) in 25 mM TEABC for 1 h at 56°C. Next, the proteins were alkylated by incubation in 200 mM methyl methanethiosulfonate in isopropanol (both from Sigma-Aldrich) for 1 h at room temperature, and then the samples were washed in 25 mM TEABC in 50% ACN for 10 min under mild agitation. The samples were next dehydrated by a twice-repeated suspension in 100% ACN until the gel pieces turned opaque and then dried. Next, the gels were resuspended in a trypsin/Lys-C-solution containing 10 ng/μL trypsin/Lys-C mix (Promega; V507A) in 1 mM CaCl₂–5% ACN in 50 mM TEABC, followed by incubation overnight at 37°C. The peptides were sequentially extracted from the gel with 50 and 100% ACN in 25 mM TEABC for 30 min, and the samples were vacuum dried and prepared for mass spectrometry.

For LC-MS analysis, the previously described SCIEX Triple-TOF 6600+ coupled with capillary flow LC setup was used with minor adaptions (109). In brief, the digested samples were resuspended in 20 μL of 0.1% formic acid (FA), of which 8 μL was loaded onto the LC system. The samples were trapped on a YMC capillary guard column (5 × 0.5 mm, 3-μm particle size) for 5 min with a mobile phase A at a 10-μL/min flow rate prior to LC separation using a gradient elution from 3 to 40% mobile phase B over 30 min. Mobile phase A consisted of 0.1% FA in water, and mobile phase B consisted of 0.1% FA in ACN. For DDA acquisition, MS1 spectra were acquired between 400 and 1,200 m/z for 250 ms, followed by selection and fragmentation of the 30 most intense precursor ions to acquire MS2 fragment spectra between 100 and 1,500 m/z for 50 ms. Selected precursor ions were dynamically excluded for 10 s. For SWATH acquisition (DIA), a 99-isolation window acquisition scheme was used to acquire MS2 spectra from 100 to 1,500 m/z for 37.5 ms. Before each 4-s cycle, an MS1 survey scan was acquired for 250 ms. The samples were randomized prior to LC-MS analysis, and a mixture of all samples was analyzed every six samples to monitor LC-MS performance over time. In addition, 40 fmol of PepCalMix (Sciex, catalog no. 5045751) was analyzed every five samples to calibrate the instrument in between analyses.

The obtained DDA raw data were imported into the Progenesis QIP software. Next, the features—clusters of signal with a clear isotope distribution—were aligned along the retention time and m/z dimension, and the obtained peak list was exported as an MGF file after peak picking. This file was used to identify the features using the Mascot Daemon database search engine. The databases used were the pig reference proteome (UniProt ID UP000008227 at 23FEB2021) and the FASTA protein sequence proteomes of Becker (JF797219.1 at 10JUN2020), Kaplan (JF797218.1 at 10JUN2020), Bartha (JF797217.1 at 10JUN2020), and NIA3 (KU900059.1 at 10JUN2020). In addition, the common repository of adventitious proteins was also added to control for lab contaminants. One outlier sample of the Becker strain was omitted.

Several Progenesis QIP projects were created during data analysis, each serving a separate research question. First, to obtain only those peptides that were identical in all PRV strains, we performed a sequential search with all strains separately, while each time exporting and searching an MGF file from Progenesis QIP that only contained identified features. A second project was created to include the four proteins, pUS2, gI (pUS7), gE (pUS8), and pUS9, which are (partially) deleted in the Bartha genome. To this end, the Bartha search was omitted from the sequential searches described above.

For all searches, at least two unique peptide identifications were required to confidently identify the protein.

Samples were prepared for TEM as described previously (110). Virion area and circularity were calculated using ImageJ. Circularity was defined as follows:Statistical analysis was performed using a two-tailed Student t test.