Vaccinia Virus Proteome: Identification of Proteins in Vaccinia Virus Intracellular Mature Virion Particles

Cells, virus, and reagents.

HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Vaccinia virus (Western Reserve strain) was propagated in HeLa cells and purified as described previously (80, 86). In brief, the infected cells were harvested when the cytotoxic pathological effect was complete. All stages of purification were performed at 4°C. The cells were centrifuged at 850 × g for 15 min to remove the medium and washed three times with phosphate-buffered saline. The pelleted cells were resuspended in TM buffer (10 mM Tris, pH 7.4, 5 mM MgCl₂), and the suspension was passed 20 times in a Dounce homogenizer to break the cells. Nuclei were removed by centrifugation at 850 × g for 10 min, and the supernatant, containing virions, was laid on top of 36% sucrose solution and centrifuged at 45,000 × g for 80 min in a Beckman SW28 rotor. The virus pellet was resuspended in TM buffer, sonicated, and further purified by centrifugation through a continuous 25 to 40% sucrose gradient at 27,500 × g for 40 min in a Beckman SW28 rotor. Fractions were taken, and the virion infectivity in each fraction was analyzed by a plaque formation assay. Fractions containing virions were diluted fourfold in TM buffer and centrifuged to pellet the IMV. The purity of the IMV was confirmed by electron microscopy of negatively stained preparations. No detectable cellular debris was found in these preparations.

Tryptic and lysine C digestion of IMV particles.

Purified IMV (9 to 50 μg) were incubated at 25°C overnight with 0.5 N cyanogen bromide (Fluka) in 90% formic acid (Fluka) to cleave the proteins into smaller polypeptides, and excess reagents were removed by vacuum drying. The polypeptides were denatured and reduced at 37°C for 1 h in 6 M urea (Sigma), 2 M thiourea (Riedel-de Haën), and 10 mM dithiothreitol (Pharmacia Biotech), and then iodoacetamide (Sigma) was added to a final concentration of 20 mM and the sample was incubated at 37°C for 1 h in the dark. The peptide mixture was then diluted eightfold with 100 mM ammonium bicarbonate (Sigma), and sequencing-grade porcine trypsin (Promega) or Lys-C (Wako) was added at a substrate-to-enzyme ratio of 20:1 (wt/wt), and the mixture was incubated at 37°C overnight. The peptides formed were desalted using a C₁₈ trap, lyophilized on a SpeedVac, and stored at −80°C.

Two-dimensional chromatography.

The peptide mixture from 50 μg of IMV was fractionated by two-dimensional chromatography (strong cation-exchange chromatography [SCX]-reverse-phase liquid chromatography). The first dimension, SCX, was eluted with a linear gradient of 0 to 300 mM KCl in 5 mM ammonium formate, pH 3.0. The peptides eluted from the SCX column were trapped in two C₁₈ reverse-phase traps operating alternately; the bound peptides were eluted at 2-min intervals and collected on a fraction collector; a total of 22 fractions were collected. The amount of peptide in each fraction was estimated from the UV absorption at 214 nm.

Mass spectrometry.

Mass spectrometric analysis was performed on a nanoscale LC-tandem mass spectrometry (quadrupole time-of-flight mass spectrometer; QStar XL; Applied Biosystems). The instrument setup was as follows. The flow (150 μl/min) from the binary pump (Agilent 1100, with solvent A [100% deionized water] and solvent B [90% acetonitrile] [J. T. Baker], both solvents containing 0.1% formic acid) was split with two T-shaped connectors connected to a self-packed precolumn with an appropriate flow restrictor to give a column flow rate of 10 μl/min for sample loading and 200 to 300 nl/min for sample elution from the analytical column. The sample of 2 to 5 μg was injected onto the precolumn (15 mm long; 150-μm internal diameter; C₁₈) via a 20-μl sample loop. The analytical column (C₁₈; 15 cm long; 75-μm internal diameter) was connected to a 15-mm electrospray emitter (10-μm tip opening) by a 1-cm Teflon sleeve. The chromatographic separation was performed with a 120-min gradient profile as follows: 2% B (0 to 4.5 min), linear gradient of 2 to 10% B (4.5 to 5 min), 10 to 40% B (5 to 80 min), 40 to 50% B (80 to 100 min), 50 to 80% B (100 to 105 min), 80 to 2% B (105 to 106 min), and 2% B (106 to 120 min). The spectra of the eluted peptides were acquired in data-dependent mode by first acquiring a full MS scan from m/z 400 to 1900 for 1 second to determine the three most intense peptide peaks with charge states above 2, and then three MS/MS scans between m/z 100 and 2000 (1.5 s each) were performed for the MS-scanned parent ions with a threshold above 20 counts. Once sampled, each MS/MS precursor mass was excluded from further tandem experiments for 2 min.

The data files completed from the LC-MS runs were converted to Mascot generic-format files using the Mascot.dll script supplied with the Analyst QS software. The Mascot software package (Matrix Science) was used for database searching and protein identification using the modified vaccinia virus protein database. Peptide mass tolerance and fragment mass tolerance were set at 100 ppm and 0.25 Da, respectively, for the initial search. An alternative calibration algorithm based on Mascot protein identifications was applied to the raw data file to give mass accuracies within 20 ppm.

Proteomic analysis of IMV-derived peptides and database generation.

Individual MS/MS spectra were submitted to analysis using MASCOT version 2.0 (Matrix Science Inc.) and searched against the Human International Protein Index protein sequence database (version 3.10; 57,478 protein sequences; European Bioinformatics Institute [http://www.ebi.ac.uk/IPI/]) or against a modified vaccinia virus protein database. The modified viral-protein database combines 218 protein sequences of the WR strain of vaccinia virus obtained from http://poxvirus.org, 64 nonredundant Copenhagen strain orthologs, and 12 additional peptide sequences predicted from WR viral genome sequences using two programs, GeneMarkS (http://opal.biology.gatech.edu/GeneMark/) and fgenesV (http://softberry.com; 17). These databases allowed the identification of novel vaccinia virus proteins by an exhaustive search of all possible virus-derived peptide sequences. Proteins were scored using a probability-based MOWSE algorithm, and the scores were reported in the form −10 × log(P), where P is the probability that the observed match is a random event (123). Matches with scores higher than the 95% confidence level were regarded as significant. In all searches, phosphorylation, methionine oxidation, homoserine lactone, and carbamidomethylation of cysteine residues were considered as possible modifications. In order to increase the data confidence, the same data sets were submitted for a search against a “reverse database,” in which all peptide sequences were put in reverse order in silico; the percentage of peptides identified in the reverse database was less than 3.5%, confirming the specificity of the data search. Individual peptide interpretations were classed as significant if the MASCOT MOWSE score for the MS/MS spectra was greater than the cutoff value of 25.

Relative quantification of MS.

Quantification of protein abundance in IMV was performed based on the criteria and equations defined previously (85). In brief, the relative abundances of the proteins were estimated using an exponentially modified protein abundance index (emPAI) approach, which is based on the correlation of peptides identified in MS/MS experiments with observable peptides obtained by in silico digestion, which allows the relative abundances of proteins in a complex protein mixture to be determined. The abundance of each protein is presented as the molar and weight percentages of the total protein molecules in the IMV particles.

Identification of IMV proteins.

Vaccinia IMV particles were purified from the lysates of HeLa cells infected with the vaccinia virus WR strain, using two consecutive sucrose purification procedures to achieve maximal purity. The purified IMV particles were examined by electron microscopy to ensure that the virions had normal virion morphology and that cellular organelles and debris were undetectable. Proteins in purified virions were separated on 12% SDS-PAGE and silver stained (Fig. 1A). Gel electrophoresis of the purified virions followed by in-gel digestion often causes loss of proteolytic fragments generated from low-abundance proteins, and the gel matrix is inherently not optimal for resolving proteins of low molecular weight. Thus, a gel-free scheme with two-stage proteolysis was adopted to facilitate the efficient extraction of peptides from purified IMV particles. The purified IMV were completely dissolved in 90% formic acid and subjected to cyanogen bromide cleavage, followed by tryptic digestion in solution to sequentially yield a mixture of peptides. Alternatively, proteolytic digestion with Lys-C was performed in order to preserve longer peptides for better quality in data acquisition. These peptides were then analyzed by MS/MS using a modified vaccinia virus (WR) protein database constructed (see Materials and Methods) to include all predicted vaccinia virus translation products, as the commonly used vaccinia virus (WR) strain protein database does not include all predicted viral proteins or small proteins with fewer than 50 amino acids. As shown in Fig. 1B, the MS/MS spectrum of a tryptic peptide was interpreted as HAFDAPTLYVK, one of 30 unique peptides identified in a putative protein, WR062/E6R, resulting in 53% sequence coverage (Fig. 1C). Another tryptic peptide, ADEDDNEETLK, was identified in an envelope protein, A27L (Fig. 1D), that contains 71% sequence coverage, as shown in Fig. 1E. A total of three different IMV preparations were used in five independent mass spectrometry analyses. Combining the peptide data generated from trypsin and Lys-C digests, we identified 75 viral proteins in the IMV (Table 1). Because the gene names in the vaccinia virus WR strain are not as widely used as their orthologs in the Copenhagen strain, the latter gene names are used to refer to the viral translation products we identified in this work. Of these, 69 were identified by at least two trypsin- or Lys-C-generated peptides, while the other 6 (C6L, I5L, A9L, A21L, A22R, and A31R) were identified by single peptide hits in the trypsin- or Lys-C-digested samples. A further study using two-dimensional chromatographic fractionation to separate peptides prior to MS/MS was performed (SCX-LC/MS/MS), and at least one extra peptide (each) was identified for C6L, A9L, A21L, and A31R (Table 2). All the extra peptides identified in the SCX-LC/MS/MS study had MASCOT scores of at least >28. No extra peptide was identified for I5L and A22R by SCX-LC/MS/MS, although the same peptides as in one-dimensional LC/MS/MS were reproducibly detected. Both proteins are included in Table 1, because they were previously detected in the IMV (63, 84, 160).

Of the 75 viral proteins identified, 65 were consistent with previous reports of their presence in IMV particles (references in Table 1); these were 22 membrane proteins (H3L, D8L, H2L, L1R, L5R, A2.5, A9L, A13L, A14L, A17L, A21L, A25L, A26L, A27L, A28L, E8R, E10R, F9L, I2L, I5L, J1R, and D13L), 12 core proteins (A3L, A4L, A10L, A30L, F17R, E11L, I1L, I3L, G7L, L4R, D2L, and D3R), 24 enzymes [kinase (F10L), phosphatase (H1L), glutaredoxins (G4L and O2L), topoisomerase (H6R), proteinases (G1L and I7L), ATPase (D11L), poly(A) polymerase subunits (E1L and J3R), capping enzymes (D1R and D12L), DNA-dependent RNA polymerase complexes (A5R, A24R, A29L, J4R, J6R, H4L, E4L, G5.5R, and D7R), helicases (I8R and A18R), and endonuclease (A22R)], 3 viral transcription factors (A7L, H5R, and D6R), and 4 proteins (A42R, A45R, A46R, and F8L) interacting with host proteins. This shows that our proteomic data are reliable and consistent with previous results. More importantly, the proteomic data revealed 10 new viral proteins present in IMV, consisting of the products of seven former putative ORFs (E6R, G3L, L3L, A6L, A15L, A31R, and C6L) and three proteins (K4L, G9R, and A16L) previously with unknown locations and functions. K4L shows homology with phospholipase D but is nonessential for viral growth in cell cultures (21, 24, 122), while both G9R and A16L have been identified as myristylated proteins in virus-infected cells (102).

Six viral proteins, I6L, G5R, A12L, A14.5, A32L, and B1R, have been cited in the literature as being IMV associated but were not identified in our analyses (Table 3) (19, 34, 43, 48, 68, 99, 160). All six proteins were not detected in LC/MS/MS using different IMV preparations, although three proteins (A12L, A32L, and B1R) were detected in SCX-LC/MS/MS, suggesting that the abundances of these three proteins were below the threshold of LC/MS/MS and that SCX chromatography was necessary to reduce the sample complexity in order to enhance the detection sensitivity (Table 3). Consistent with our interpretation, A32L was previously shown to be difficult to detect in purified IMV (34). The other three proteins (I6L, G5R, and A14.5L) were not identified in either LC/MS/MS or SCX-LC/MS/MS analyses. Several possible reasons may explain why they were not detected in both MS/MS analyses. First, MS is biased against small and hydrophobic proteins. A14.5L contains 53 amino acids, producing at most two tryptic peptides (19). It is also very hydrophobic, with a grand average of hydropathicity index of 1.438 (96). These features of A14.5L may have jeopardized its detection in MS analyses. Second, MS, despite great sensitivity, can only detect peptides at femtomole range. Third, peptides that are too close to be resolved in MS will also be ignored by data acquisition software. We speculate that these technical reasons might explain the lack of detection of G5R and I6L.

Relative abundances of viral proteins in the IMV.

The quantity of each viral component within the IMV particle is an important issue, but it has been difficult to address in the past. Crude estimation of the quantity of each protein by gel staining intensity is not suitable for complex structure analysis. While recent advances in differential protein analyses have provided a means for comparing relative protein expression between two populations, there is a need for a method giving the relative quantification of a single protein in a complex population. Although quantification of particular proteins of interest using isotope-labeled synthetic peptides, termed Protein-AQUA, is in principle applicable to comprehensive analyses (12, 64), it is hampered by the high cost of isotope-labeled peptides and the difficulty of quantitative digestion of proteins in gel (73).

Recently, Ishihama et al. demonstrated that it is feasible to quantify label-free protein components using an emPAI approach (85), which fully utilizes the advantages of the wide dynamic detection range provided by LC/MS analysis and reliable protein search algorithms. This approach to determining the relative abundances of proteins in a complex protein mixture is based on the correlation of peptides identified by tandem MS with observable peptides obtained by in silico digestion. The abundance of each protein determined in this study is presented as the molar and weight percentages of the total protein molecules in IMV particles (Table 4). The most abundant viral-protein group (weight percentage, >5) consisted of four core proteins (A4L, A10L, F17R, and A3L). The relatively abundant protein group (5 > weight percentage > 1) consisted of seven envelope proteins (A27L, A25L, A14L, H3L, A26L, D8L, and A13L,), two core proteins (L4R and G7L), four proteins involved in transcription (D11L, J6R, A24R, and D1R), and one novel protein (E6R). Eleven of the above 18 abundant proteins have been previously identified as abundant proteins in IMV particles (86, 160). The intermediate-abundance protein group (1% > weight percentage > 0.5%) contained 15 viral proteins, and the less abundant group (weight percentage, <0.5%) contained 41 viral proteins. The distribution of viral proteins based on different molar or weight abundances in IMV revealed a similar distribution (Fig. 2); the low abundances of many proteins explain why they have escaped previous proteomic detection.

Identification of host proteins associated with the IMV.

The host protein content of vaccinia virus IMV was determined by comparing the detected peptides with a Human International Protein Index protein sequence database, and 23 host proteins associated with vaccinia virus IMV particles were identified (Table 5). All were present at low abundance (intermediate and less abundant categories) in virions and could be divided into several categories, namely, calcium binding proteins (annexin A1 and A2), chaperon proteins (cyclophilin A, Hsc71, and Hsp90), cytoskeleton proteins (actin, tubulin, and myosin), chromosome architecture proteins (histone and HMG1), translation components (eIF4A-1, eIF1α-1-1, and 60S ribosomal protein), protein transport/vesicular-trafficking proteins (ADP ribosylation factors1/3 and 4, Rab7, Rab10, and ubiquitin), and proteins involved in redox regulation (thioredoxin and peroxiredoxin 1). Of these host proteins, cyclophilin A, β-actin, and ubiquitin have been previously reported as components of vaccinia virus IMV particles (35, 86, 176), and another 13 have been reported in other viruses (Table 5) (22, 87, 92, 172, 191). Thus, seven host proteins (60S acidic ribosomal proteins, 66-kDa protein, ADP ribosylation factor 4, HMG1, peroxiredoxin 1, Rab10, and thioredoxin) were found to be associated with vaccinia virus IMV and have not yet been detected in other viruses.