Main

As a reversible post-translational modification, phosphory-lation is indispensable for many fundamental cellular processes. It has become increasingly clear that dynamic phosphorylation also serves to regulate the replicative cycle of many viruses1,2,3. Advances in quantitative mass spectrometry-based phosphoproteomic technology enable the interrogation of complex signalling programmes, including viral phosphorylation networks4,5,6,7. Yet, for most, a systems-level understanding of viral phospho-networks and how they drive infection phenotypes and assure virion infectivity is lacking. Filling this gap in our knowledge of viral signalling networks provides an opportunity to bridge our current understanding of genotype–phenotype relationships using proteotype information.

Vaccinia virus (VACV) is the prototypic member of the Poxviridae, a family of large double-stranded DNA viruses that include variola virus, the causative agent of smallpox8. Poxviruses, the largest, most complex mammalian viruses produce infectious mature virions containing ~80 different viral proteins9,10. Assembly of these virions occurs exclusively in the cytoplasm of infected cells proceeding through several morphologically distinct stages reviewed elsewhere8,11.

VACV encodes a set of enzymes to assure that assembly of these complex particles occurs in a tightly coordinated spatio-temporal fashion. These include the F10 kinase, two proteases (I7 and G1), a virus-encoded redox-system (E10, A2.5, G4) and the H1 phosphatase11. Classical genetic studies of inducible or temperature-sensitive (ts) viruses show that each is essential for formation of infectious mature virions12,13,14,15,16,17,18,19,20,21. The F10 kinase is required for the earliest stage of morphogenesis, diversion of cellular membranes, I7 protease for the transition from immature virions to mature virions and the H1 phosphatase assures virion transcriptional competence14,18,22. Although F10 and H1 share viral substrates important for virion assembly11,23,24,25, the dynamic phospho-signalling network through which they regulate phosphorylation of viral proteins to drive production of infectious virions has not been evaluated.

Using quantitative mass spectrometry-based proteomics we dissected the abundance and phosphorylation status of proteins within VACV virions and the phospho-signalling network in cells infected with wild-type, F10 kinase- and H1 phosphatase-deficient viruses. Comparison of these viral phospho-proteomes revealed 105 phosphosites in 43 viral proteins, 33 of which relied on F10 and/or H1. Analysis of I7 protease, a new F10/H1 substrate, indicated that dynamic phosphorylation at S134 drives virion structural protein cleavage. Relative quantitative comparison of phosphosites within wild-type and H1-deficient virions revealed that phosphorylation of the viral transcription factor A7 at Y367 directly contributes to the transcriptional incompetence of H1-deficient virions. These results establish a key role for dynamic phosphorylation in the regulation of infectious poxvirus particle assembly and highlight the utility of combining quantitative proteomic screens with mutant viruses to uncover proteotype–phenotype–genotype relationships.

Results

Defining the F10/H1 phosphoproteome

To define the viral phospho-signalling network regulated by the F10 kinase and H1 phosphatase we infected HeLa cells with either VACV wild-type, inducible F10 (vF10V5i referred to as F10(−)) or H1 (vindH1 referred to as H1(−)) recombinants under non-permissive conditions22,26 (Fig. 1a). Cells were harvested 12 h post infection, when intermediate and late viral genes are maximally expressed8,27. Proteins were subjected to tryptic digestion and phosphopeptide enrichment on TiO2 beads, resulting in phosphopeptide enrichments ranging between 71% and 84% (Supplementary Fig. 1a).

Fig. 1: Proteotype-based decoding of the viral signalling network.
figure 1

a, Conceptual approach and analytical strategies adopted in this study to uncover the viral signalling network and its function. b, Heat map showing the relative abundance of viral phosphorylation sites (S, T and Y) between VACV wild-type (WT), F10(−), H1(−) infected cells and uninfected (uninf.) cells. The last column on the right (reg.) indicates phosphosites regulated in an F10 (green square) and/or H1 (purple squares) -dependent fashion by at least twofold relative to wild-type infections. Experiments were performed in biological triplicate and significance scored as a minimum of twofold with an adjusted P value ≤ 0.05. c, Relative abundance of viral proteins in wild-type versus H1(−) mature virions. Proteins are colour-coded as more abundant (red), less abundant (blue) or displaying no change in abundance (black) in H1(−) mature virions relative to wild-type mature virions. Experiments were performed in biological triplicate and proteins displaying at least a twofold change in abundance ( +/− ) with an adjusted P value ≤ 0.01 were considered significant. d, Relative abundance of viral phosphorylation sites between H1(−) and wild-type mature virions. Each phosphorylation site is labelled as the viral protein name and its amino acid position. Phosphosites are colour-coded as more abundant (red), less abundant (blue) or displaying no change in abundance (black) in H1(−) mature virions relative to wild-type mature virions. Experiments were performed in biological triplicate. Only phosphorylation sites that change in abundance at least twofold (+/−) with an adjusted P value ≤ 0.05 between H1(−) and wild-type mature virions are considered significant. MV, mature virion; EV, extracellular virion.

Samples were analysed by data-dependent acquisition (DDA) liquid chromatography–tandem mass spectrometry (LC-MS/MS) and quantified across conditions. We identified 105 non-redundant serine (S), threonine (T) and tyrosine (Y) phosphorylation sites in 43 viral proteins (Supplementary Table 1). Relative quantification of the wild-type, H1(−) and F10(−) phospho-enriched samples revealed 28 unique F10-dependent phosphosites in 14 viral proteins, 21 unique H1-dependent phosphosites in 13 viral proteins and 16 unique phosphorylation sites in 10 viral proteins regulated in an F10/H1-dependent fashion (Fig. 1b; green and purple boxes). Amongst the ten shared substrates, eight phosphosites were within four known shared substrates: S85 of A14, S235 of A4, S32, S167 and S197 of A17, and S229 and S261 of G711,22,23,28,29, and eight phosphosites within six unknown shared substrates: A12, A19, G3, H5, F17 and I7. Amongst the F10/H1 substrates, two membrane proteins (A14, A17), five core proteins (A19, A4, A12, G7, H5), the lateral body protein F17 and the I7 protease are essential for VACV assembly11.

Proteotype profiling of phosphatase-deficient VACV virions

Often, phenotypic outcomes are best described by the underlying proteotype; the acute state of the proteome under given constraints30. We detected many quantitative changes in the VACV phosphoproteome in infected cells lacking F10 or H1 (Fig. 1b). To determine if these changes affected the virion proteotype at the level of protein abundance wild-type and H1(−) mature virions were subjected to comparative proteomics. As F10 is essential for VACV morphogenesis, virions cannot be produced in its absence13. We identified and quantified 139 viral proteins between wild-type and H1(−) mature virions (Fig. 1c, Supplementary Table 2). Forty-eight were less abundant by more than twofold in H1(−) mature virions, including H1 which was down 43-fold. In line with previous reports22, mature virion-specific proteins A25 and A26 did not change abundance between wild-type and H1(−) mature virions (Fig. 1c, dashed underlined label)31, indicating no defect in mature virion production. As the majority of virions produced are mature virions8 and no striking changes in protein abundance between wild type and H1(−) were seen, we reasoned that the F10/H1 phospho-signalling network does not control protein copy number for progeny virion assembly.

We next compared the phosphoproteotype of wild-type and H1(−) mature virions. Quantitative phosphoproteomics analysis after TiO2 phosphopeptide enrichment (Supplementary Fig. 1a) yielded 179 phosphosites (78.6% serine and 21.4% threonine) on 47 viral proteins (Fig. 1d). Of these 179, 62 were hyperphosphorylated and 37 hypophosphorylated in H1(−) mature virions (Fig. 1d, Supplementary Table 3). Of the 17 viral phosphoproteins comprising the F10/H1 phospho-signalling network in infected cells (Fig. 1b, Supplementary Table 1), only G3L was non-phosphorylated in mature virions (Fig. 1d, Supplementary Table 3). Among the 21 H1 phosphosites identified in cells (Fig. 1b, purple squares), 15 were hyperphosphorylated in H1(−) mature virions. Of note, F10 and I7 protease, enzymes essential for mature virion formation, were identified as H1 substrates (Fig. 1b,d).

Dynamic phosphorylation of I7 is required for proteolytic processing and virus production

That F10 and I7 are H1 substrates suggested that in addition to assuring transcriptional competence of newly formed virions, H1 may facilitate their formation. Phosphorylation of F10 S7/8 or Y11 was recently reported, yet any link to H1 and the importance of these modifications was not investigated32. For this, we complemented an F10 mutant virus (Cts28) with F10–HA or F10–HA phosphomimetic and phosphodeletion mutants. These proteins rescued virus yield equally, suggesting that dynamic phosphorylation of F10 plays no role in virus assembly (Supplementary Fig. 2a,b).

Next, we investigated the I7 protease, the enzyme essential for virion maturation via cleavage of core and membrane proteins14,18,33. Phosphoproteomics indicated that I7 S134 is phospho-regulated in an F10/H1-dependent fashion (Fig. 1b). To test if dynamic phosphorylation of S134 regulates I7 activity we used transient complementation of a VACV temperature-sensitive (ts) mutant of I7, Cts1615,34. This virus displays a 4,000-fold reduction in 24 h virus yield when grown at non-permissive temperature (40.0 °C). While virus production was complemented by 2-logs using an HA-tagged version of wild-type I7 (I7-HA), neither the phosphodeletion (I7-HA S134A) nor the phosphomimetic (I7-HA S134E) mutant could rescue virus yield (Fig. 2a). Immunoblot analysis assured that comparable amounts of wild-type and mutant proteins were expressed (Fig. 2a; inset). Next, cleavage of I7 core (p4a) and membrane (A17) substrates was analysed15,18. While expression of wild-type I7-HA rescued cleavage of both p4a and A17, I7-HA S134A and S134E mutants were unable to complement (Fig. 2b), strongly suggesting that F10/H1 dynamic phosphorylation of I7 at S134 is required for its protease activity.

Fig. 2: F10-dependent phosphorylation of I7 modulates cleavage of viral structural proteins.
figure 2

a, BSC40 cells were infected with Cts16 (MOI = 4), transfected with empty plasmid (pBSIIK) or the various I7-HA constructs, and infectious yields were determined at 24 h. Mean ± s.e.m. is shown for three biological replicates. Immunoblot analysis using anti-HA confirmed the expression of the various I7 proteins (inset). b, Proteolytic processing of the core protein P4a and the membrane protein A17 by transiently expressed I7 proteins at 40 °C was assessed by immunoblot analysis with immunoblot against tubulin serving as a loading control. c, BSC40 cells were infected with VACV wild-type, Cts15 or Cts28 (MOI = 4) for 12 h in the presence of Rif at 31 °C. Cells were then washed and released into medium without Rif for 12 h at 31 °C or 40 °C. d, Following the protocol outlined in c, cell lysates were analysed for cleavage of the core structural protein P4a. Immunoblots directed against tubulin were used as loading control. All immunoblots are representative of three independent biological replicates. pfu, plaque-forming unit; T, temperature.

F10 regulates I7 protease activity during the immature virion to mature virion transition

F10 is required for two stages of virus assembly; first for the diversion/formation of viral membranes13,35, and second during the formation of immature virions36. As we identified I7 as an F10 substrate (Fig. 1b), we reasoned that F10 may be required for I7-mediated proteolytic processing of viral proteins during immature virion formation. To assess if loss of F10 impacts I7-mediated proteolytic processing of viral proteins we used a rifampicin (Rif)-release assay to bypass the first F10-dependent stage of infection. Cells were infected with wild-type or tsF10 viruses (Cts15 or Cts28) in the presence of Rif at permissive temperature (31.0 °C) for 12 h, then released from Rif and kept at 31.0 °C or shifted to non-permissive temperature (40.0 °C) for 12 h (Fig. 2c). When released at 40.0 °C, Cts15 and Cts28 displayed a 48- and 400-fold decrease in virus yield, confirming the second F10-dependent block (Supplementary Fig. 2c). Analysis of p4a in Cts15 and Cts28 infected cells after Rif release showed that at 31.0 °C p4a proteolytic processing occurred normally, while at 40.0 °C p4a remained unprocessed (Fig. 2d). These results suggest that I7 protease activity during immature virion formation and/or immature virion to mature virion maturation requires F10 kinase.

To gain spatial-temporal information of I7 S134 phosphorylation during morphogenesis we coupled the Rif release with phosphoproteomics. Wild-type infections were synchronized for 12 h in Rif and harvested or released from Rif for 10 min, 90 min or 180 min. Representative transmission electron microscopy (TEM) images confirmed the Rif block, many virosomes surrounded by flaccid membranes and the synchronous resumption of viral morphogenesis upon its release (Fig. 3a, Rif). As reported37,38, 10 min post release viral crescents appeared at the edge of the virosomes (Fig. 3a, T10). While at later times synchronization was less apparent, by 90 min immature virions and immature virions containing nucleoids were seen (Fig. 3a, T90). At 180 min post release vast numbers of mature virions had formed (Fig. 3a, T180).

Fig. 3: Phosphorylation of I7 S134 is temporally regulated during virus assembly.
figure 3

a, HeLa cells were infected with wild-type VACV (MOI = 5) for 12 h in the presence of Rif and subsequently released into medium without Rif for 10 min (T10), 90 min (T90) or 180 min (T180). Samples were prepared for Epon embedding and conventional TEM. Morphogenesis intermediates formed in the presence of Rif or upon Rif release are shown. Rifampicin bodies (RB), crescents (C), immature virions (IV), immature virions with DNA nucleoids (IVN) and mature virions (MVs) are labelled. Scale bars, 1 µm. Representative images of from two biological replicates are shown. b, Cells prepared as in a were subjected to HRM/DIA mass spectrometry with heavy labelled spike-in peptides for the relative quantitative assessment of both I7 and I7 S134 phosphorylation. The relative abundances of I7 and I7 S134 (normalized to 0 min) over time are displayed. Mean ± s.e.m. is shown for three biological replicates. Statistical significance at P < 0.001 (Student’s t-test) of I7 S134 phosphorylation changes between 0 min and 10 min and between 10 min and 180 min is displayed with an asterisk. c, Cell lysates as prepared in a were analysed for cleavage of the membrane protein A17 and the core structural protein p4a. Immunoblots directed against tubulin were used as loading control. Representative blots from three independent biological replicates.

Phosphoproteomics were employed to monitor I7 S134 phosphorylation. Prior to phosphoenrichment, crude heavy labelled tryptic peptides representing both phosphorylated and non-phosphorylated forms of I7 S134 were added to the trypsin-digested samples at a constant concentration. Input samples and phospho-enriched samples were measured by LC-MS/MS using data-independent acquisition (DIA). Phosphoenrichment was reproducible across samples (91.1% to 96.8%; Supplementary Fig. 1b), enabling unbiased parallel monitoring of endogenous (light) and spiked-in (heavy) peptides. Co-elution of light and heavy forms of the AIDFpSQMDLK peptide at precursor and fragment ion levels confirmed phosphorylation of I7 S134 (Supplementary Fig. 3a,b). In the presence of Rif significant phosphorylation of I7 was detected, as seen with other F10 substrates23 (Fig. 3b). Within 10 min of Rif release I7 was rapidly dephosphorylated. At 90 and 180 min post release I7 S134 phosphorylation increased over time reaching the level observed in the presence of Rif (Fig. 3b, T90 and T180).

This suggests that I7 may be active during two stages of VACV morphogenesis—early during initial viral membrane formation and late during the immature virion to mature virion transition. Consistent with this, in the presence of Rif I7-mediated cleavage of membrane, but not core, proteins occurs18,39. To determine if there was a temporal correlation between I7 S134 phosphorylation and I7 activity, proteolytic processing of A17 and p4a during Rif release was assessed. In the presence of Rif A17 was cleaved and p4a was not (Fig. 3c, Rif). While A17 cleavage remained stable over the time course of Rif release, increasing only at T180, p4a cleavage was only observed at T180 post release (Fig. 3c, T180). As cleavage of A17 occurs on the virion surface and p4a cleavage within virions, we reasoned that these results indicate that F10-mediated phosphorylation of I7 S134 regulates its proteolytic activity both temporally and spatially during VACV infection.

H1 regulates virion-associated I7 proteolytic activity

As our data suggest that F10 phosphorylation activates I7, we reasoned that H1-mediated dephosphorylation may lead to I7 inactivation. To test if H1 regulates I7 activity, cells infected with H1(−) in the absence or presence of inducer (−/+ isopropyl β-D-1-thiogalactopyranoside (IPTG)) were assessed for p4a cleavage over time (Fig. 4a). No major difference in p4a cleavage kinetics or efficiency was detected between the two conditions.

Fig. 4: H1 regulates virion-associated I7 proteolytic activity.
figure 4

a, Time course of P4a cleavage in vindH1 infected cell lysates in the absence (−) and presence (+) of inducer. b, The relative changes in abundance of tryptic peptides measured for structural proteins between H1(−) and wild-type mature virions are displayed on the proteins’ linear sequences from the N to the C terminus. The protein name and per cent sequence coverage obtained are displayed. Each bar represents a tryptic peptide, its width the amino acids covered, its height the relative abundance (up being more abundant and down being less abundant) between H1(−) and wild-type mature virions. Experiments were performed in biological triplicate. Peptides showing significant changes in abundance ( ≥1-fold up/down with an adjusted P value ≤ 0.0001) are red, peptides showing no significant change in abundance are blue. Red arrows mark the positions of I7 cleavage sites at AG↓X consensus sequences. c, Immunoblot analysis showing the proteolytic processing of the core proteins P4a, P4b and the membrane protein A17 in purified mature virions isolated from cells infected by either VACV wild-type or vindH1 without addition of inducer with immunoblot against the core protein A4 serving as a loading control. All immunoblots are representative of three independent biological experiments.

However, our Rif release DIA data (Fig. 3c,d) indicated that I7-mediated p4a cleavage occurs within virions during the immature virion to mature virion transition14,18. In addition, the abundance of I7 and its core substrates remained unchanged between wild-type and H1(−) mature virions (Fig. 1c), while phosphorylation of I7 S134 within H1(−) virions increased by 35-fold (Fig. 1d). Thus, we performed a quantitative comparison of I7 substrate peptide distribution between wild-type and H1(−) mature virions. In addition to p4a and A17, four core proteins: p4b (A3), G7, A12 and L4, undergo I7-mediated processing at conserved AG↓X motifs18,23,40,41,42,43,44. Single and consecutive tryptic peptides in the majority of I7 substrates showed significant abundance changes between wild-type and H1(−) mature virions (Fig. 4b). Each of these localized to the AG↓X cleavage sites in the various proteins (Fig. 4b, red arrows). With the exception of G7, all polypeptides from regions of expected cleavage were less abundant in H1(−) mature virions: For p4a the polypeptide between cleavage sites at positions 614 and 697 was down 2.5-fold, for p4b three consecutive polypeptides covering the N terminus cleavage site at position 61 were 8.5-fold down and for A12 the polypeptide containing the cleavage site at residue 56 was threefold down. For G7, one polypeptide harbouring the second cleavage site at position 238 was 2.3-fold down while four consecutive polypeptides including the first cleavage site at position 187 were 1.7-fold more abundant. We could not assess L4 cleavage as the N-terminal cleavage site is upstream of the first tryptic cleavage site. For A17, two of three polypeptides harbouring the cleavage sites at positions 16 and 185 were fourfold and threefold less abundant in H1(−) mature virions. Cleavage of A17 at position 16 releases the immature virion scaffolding protein D13 during the immature virion to mature virion transition45,46,47. Consistent with N-terminal hypercleavage of A17 in H1(−) virions, D13 was 5.8-fold less abundant in these mature virions (Fig. 1c). Immunoblot analyses of p4a, p4b and A17 in purified wild-type and H1(−) virions showed that cleavage is driven to completion in the absence of H1 (Fig. 4c). To assess if hypercleavage correlated with virion structural changes, cytoplasmic H1(−) and H1(+) virions were compared by TEM. No obvious morphological differences were detected between the two (Supplementary Fig. 4). Thus, we show that in the absence of H1 phosphatase, virion-associated I7 is hyperphosphorylated resulting in complete processing of I7 substrates. These results further support an essential role for F10/H1 dynamic phosphorylation in the spatial/temporal control of I7 proteolytic activity.

Viral early transcription proteins are hyperphosphorylated in H1(−) virions

We next asked if we could use virion proteotype profiling to define the underlying cause of a mutant phenotype. While the phenotype of H1(−) virus is the production of transcriptionally incompetent virions22, molecular understanding of this phenotype remains undefined. Comparison of wild-type and H1(−) mature virion proteomes showed no quantitative differences in the composition or abundance of viral transcription machinery (Fig. 1c, green and purple dots). In addition, no transcription proteins containing AG↓X sites were cleaved in H1(−) mature virions (Supplementary Fig. 5), few were phosphorylated on serine/threonine and none subject to H1 regulation (Fig. 1d, Supplementary Table 3). As H1 is a dual-specificity enzyme22,24, we analysed tyrosine phosphorylation in wild-type and H1(−) virions24,25,32. For this, 5 mg of virions were purified and phosphotyrosine-containing peptides enriched by immunoprecipitation followed by LC-MS/MS. We identified and quantified 29 phosphotyrosines on 18 viral proteins within wild-type and H1(−) mature virions (Fig. 5a). Amongst these, three components of the viral early transcription machinery were more phosphorylated in H1(−) virions: L3 on Y89/Y90, the RNA helicase NPH-II (I8) on Y634 and the viral early transcription factor subunit A7 on Y367 (Fig. 5a, red bars; Supplementary Table 4)48,49,50,51,52.

Fig. 5: Tyrosine phosphorylation of viral early transcription factor A7 controls virion transcription but not assembly.
figure 5

a, Band-purified wild-type and H1(−) mature virions were lysed in 8 M urea, digested with trypsin, phosphotyrosine-containing peptides purified using anti-phosphotyrosine antibodies and peptide abundance quantified by LC-MS/MS. The relative abundance of phosphorylated Y residues is shown for H1(−) versus wild-type mature virions. Hyperphosphorylated early transcription proteins are highlighted in red. Data acquired on a single biological replicate. b, For transient complementation of viA7 infection, BSC40 cells were infected with viA7 (MOI = 4) in the absence (−Dox) or presence (+Dox) of inducer for 4 h prior to transfection with empty plasmid (pBSIIK) or A7-HA constructs. At 24 hours post infection samples were harvested and cell lysates analysed by immunoblot for A7 expression (anti-HA) and anti-tubulin as a control (inset). The infectious yields of the various transient complementation assays were determined by plaque assay. The means ± s.e.m. of three biological replicates are displayed. c, Plaque morphology seen upon transient complementation with the wild-type, phosphodeletion and phosphomimetic A7 mutants in the presence of Dox and at distinct dilution factor (DF). Representative images from three biological replicates are shown. d, Transient complementation of A7 virion morphogenesis defect with the wild-type, phosphodeletion and phosphomimetic A7 mutants. VACV viA7L mCherry-A4 mature virions were bound to BSC40 cells on glass coverslips at room temperature (MOI = 4). Samples were incubated for 4 h at 37 °C and subsequently transfected with empty plasmid (pBSIIKS) or the various A7-HA constructs in presence or absence of Dox, as indicated. At 24 h post infection the coverslips were fixed, permeabilized and stained with anti-HA antibodies to detect A7-expressing cells. Cells were then imaged for A7 expression (anti-HA), infection (EL gpt-EGFP) and A4-positive virions (mCherry-A4). Representative confocal images from three biological replicates are shown. Gpt, xanthine-guanine phosphoribosyl transferase. Scale bar, 10 µm.

Dynamic phosphorylation of A7 is required for infectious virus formation

The identification of H1-regulated phosphotyrosines on viral transcription proteins provided a possible link between the proteotype and transcription-deficient phenotype of H1(−) mature virions22. To address their functional relevance wild-type HA-tagged versions of L3, I8 and A7, and their corresponding phosphodeletion and phosphomimetic mutants, were tested for complementation of non-permissive infections with inducible or ts viruses of each protein. For L3, the L3-HA was expressed but could not complement infectious virus production in non-permissive vL3Li-infected cells36 (Supplementary Fig. 6a,b). For I8, the wild-type (I8-HA), phosphodeletion (I8-HA Y634A) and phosphomimetic (I8-HA Y367E) each expressed and rescued non-permissive Cts10 infection in correlation with the expression level of the complementing protein (Supplementary Fig. 6c,d). Transient complementation of non-permissive viA7 infections53 showed that wild-type A7-HA was expressed and could rescue infection by 2-logs (Fig. 5b, pBSIIK A7-HA − Dox). However, both phosphodeletion (A7-HA Y367A) and phosphomimetic (A7-HA Y367E) mutants failed to complement despite being expressed as well as wild-type A7-HA (Fig. 5b). These results suggest that dynamic phosphorylation of A7 Y367 is required for A7 function and infectious virus production.

A7 phosphomutants phenocopy transcription-deficient H1(−) virions

Assuming the transcriptional defect in H1(−) virions is due to unregulated dynamic phosphorylation of A7, then A7 phosphomutants should phenocopy the H1(−) small plaque phenotype linked to their transcription defect22. Plaques formed by virions produced during transient complementation by A7-HA were indistinguishable from wild-type plaques (Fig. 5c, pBSIIKS A7-HA). Virions produced during transient complementation with the phosphodeletion (A7-HA Y367A) or phosphomimetic (A7-HA Y367E) mutants produced few wild-type size plaques (Fig. 5b), and large numbers of tiny plaques reminiscent of those seen upon repression of H1 (Fig. 5c). We reasoned that the altered phosphorylation of A7 in these virions results in a transcriptional defect, rather than the assembly defect seen upon total loss of A750,53,54. To differentiate these phenotypes we constructed viA7L mCherry-A4, an inducible A7 virus containing a fluorescent core and expressing gpt-EGFP from an early-late promoter53. viA7L mCherry-A4-infected cells were transfected with A7-HA or the Y367A or Y367E A7 mutants and analysed for mature virion formation (Fig. 5d). In the absence of inducer no discernable virion structures were formed (Fig. 5d, pBSIIKS − Dox). When A7 was expressed, distinct mCherry-A4 punctae were seen (Fig. 5d, pBSIIKS + Dox). Transient expression of A7-HA, A7-HA Y367A or A7-HA Y367E resulted in the formation of mCherry-A4 punctae indistinguishable from those in the presence of inducer (Fig. 5d). This suggested that the A7 phosphomutants could bypass the morphogenesis block seen upon A7 repression. To assure that these punctae correspond to mature virions, A7 transient complementation experiments were analysed by TEM. (Fig. 6a). During viA7L infections in the absence of inducer crescents and empty immature virions were formed, while in the presence of inducer mature virions were detected (Fig. 6a, pBSIIKS + Dox). In cells transiently expressing A7-HA, A7-HA Y367A or A7-HA Y367E abundant mature virions were observed (Fig. 6a). Quantification showed that mature virion formation in cells expressing wild-type or phosphomutant versions of A7 was comparable to that seen in the presence of inducer (Fig. 6b).

Fig. 6: Intact mature virions are formed in the absence of A7 Tyr phosphorylation exemplifying phosphodynamic regulation of poxvirus assembly by F10 and H1.
figure 6

a, Transient complementation infection assays using viA7L were performed as in Fig. 5. Samples were prepared for Epon embedding and conventional TEM. Representative images (n = 2) of the morphogenesis intermediates formed in the absence (pBSIIKS − Dox) and intact virions in the presence (pBSIIKS + Dox) of A7 expression, or transient complementation with A7 constructs. Regions of interest are boxed and magnified to the right of the micrographs. b, For each sample the average number of mature virions per section was quantified. Non-transfected cells formed no mature virions in the absence of Dox and were excluded from analysis (grey triangles). A minimum of 24 cells were quantified per condition (n = 2). Scale bar, 1.5 µm. c, Model of poxvirus assembly driven by F10 kinase and H1 phosphatase. Virion assembly begins with formation of viral crescents, which relies on F10 and membrane proteins A14 and A17, which are F10/H1 substrates. During this process, A17 is proteolytically processed by phosphorylated I7. As crescents grow to form immature virions they encapsidate virosomal material including virus enzymes (F10, H1 and I7), core structural proteins (A4, A12, A19, H5, 4A and 4B) and LB components (F17). During immature virion formation I7 phosphorylation is low, perhaps to prevent full cleavage of A17 and allow for recruitment of the D13 scaffold. Packaging of the viral genome and early transcriptional machinery, including A7, completes the formation of IVNs. During the IVN to mature virion transition, I7 phosphorylation increases and the protease acts outside virions to cleave A17 and release D13, and within virions to process core components. Finally, to assure that the mature virions formed are infectious, H1 dephosphorylates the early transcription factor A7, assuring mature virion transcriptional competence. Collectively, our data suggest that the formation of intact infectious mature virions requires a delicate balance of spatial and temporal F10/H1-mediated phosphoregulation.

Discussion

Recent phosphoproteomic studies suggest that viruses, similar to eukaryotic cells, use phosphorylation to modulate protein function32,55,56. Their relative simplicity, lack of redundancy and genetic manipulability make viruses outstanding tools to probe proteotype–phenotype–genotype relationships.

Revealing the F10/H1 viral signalling network

In vitro experiments on intravirion phosphorylation of VACV during activation/uncoating have been described32,57. These studies reported 29 phosphosites on 13 viral proteins modified by ‘intravirion kinases’. Of the 28 sites in 14 viral proteins we identified as F10 substrates (Fig. 1b), A14 (S85 and S88), A19 (S36 and S42), F10 (S7/S8/Y10) and G7 (T304/S305/T308/T312) were identified in these studies32,57. The modest overlap between phosphorylation events seen during mature virion uncoating/activation and virion morphogenesis supports the concept that dynamic phosphorylation regulates the virus lifecycle.

Focusing on the F10/H1 signalling network using proteomic profiling and mutant viruses we uncovered the largest intrinsic virus signalling network to date. Consistent with other studies32,57, sequence alignment of 15 unambiguous sites provided no consensus for F10/H1 substrate recognition (Supplementary Fig. 7), suggesting that other features are required. Nonetheless, we uncovered 21 H1-dependent phosphosites within 13 substrates required for virion membrane assembly, maturation and proteolytic processing (Supplementary Table 1). Highlighting the power of proteotype analyses, these results show that the F10/H1 signalling network regulates aspects of the VACV lifecycle beyond those revealed by mutant viruses.

Dynamic phosphorylation of I7 and VACV maturation

The phenotype of recombinant virus mutants always reflects the first temporal block in the virus lifecycle. This makes it nearly impossible to define additional phenotypes that underlie the loss of a particular protein. Using quantitative phosphoproteomics of wild-type, F10(−) and H1(−) infected cell lysates we show that the viral protease I7 is an F10/H1 substrate and demonstrate a genetic link between F10 and I7 proteolytic activity. Using phosphoproteomics with spike-in of heavy labelled reference peptides in combination with DIA mass spectrometry, we further show that F10-mediated phosphorylation of I7 S134 occurs during two independent stages of infection: early during viral membrane diversion and late during immature virion formation/immature virion to mature virion transition. That early I7 proteolytic activity correlated with extravirion cleavage of A17 and late activity with cleavage of p4a within the virion suggested that F10-dependent I7 activity was also spatially regulated. Corroborating this notion, proteotype profiling uncovered an I7 substrate hypercleavage phenotype within H1(−) virions. That this phenotype was detected in purified virions but not infected cell lysates dictates that regulation of I7-mediated core protein processing occurs within virions. Together, these results indicate that I7 proteolytic activity is controlled by F10/H1 dynamic phosphorylation in a spatial/temporal fashion. Perhaps the differential subvirion localization of F1016 and H158 reflects the need to partition their activities for successful I7-mediated virion maturation.

H1(−) mature virion proteotype and transcriptional defect

Phosphoproteotype analysis of virion-associated transcription machinery revealed that three early transcription proteins displayed increased tyrosine phosphorylation in H1(−) mature virions (Fig. 5a and Supplementary Table 4). Of these, dynamic phosphorylation of A7 Y367 was required for productive virus infection (Fig. 5b). While complete repression of A7 results in an assembly defect53,59,60, targeted mutation of Y367 allowed for the formation of mature virions. That these virions phenocopy H1(−) virions22 suggests that the transcription deficiency of H1(−) is due, at least in part, to hyperphosphorylation of A7. Within mature virions A7 is located inside the core and H1 relegated to lateral bodies (LBs)58. This would suggest that virion transcriptional competence must be assured in immature virions when H1, A7 and other early transcription factors are within a single virion compartment (Fig. 6c).

Our results reveal the underlying complexity and implicit spatial/temporal regulation of the F10/H1 signalling network required to assure the formation of infectious poxvirus particles. By combining mutant viruses and proteotype profiling we have demonstrated that one can go beyond virus genotype–phenotype relationships to define viral signalling networks, uncover masked phenotypes and define new links between mutant virus phenotypes and their causal proteotypes.

Methods

Antibodies and drugs

Mouse anti-HA MAb HA.11 (16B12) was purchased from Covance. Mouse anti-α-tubulin MAb DM1A and mouse anti-γ-tubulin MAb GTU-88 were purchased from Sigma-Aldrich. Rabbit anti-α-tubulin MAb (11H10) was purchased from Cell Signaling Technologies. Anti-4a was generously provided by Dr Dennis E. Hruby (Oregon State University). Anti-A17 and anti-A4 were generously provided by Dr Jacomine Krijnse-Locker (Institute Pasteur). Goat anti-mouse AlexaFluor-647 conjugated secondary antibody was purchased from Life Technologies. Rif, doxycycline hyclate and IPTG were purchased from Sigma-Aldrich.

Cell and viruses

African green monkey BSC40 cells (ATCC CRL-2761) and HeLa cells (ATCC CCL2) used throughout this study were not independently authenticated in our laboratories. Both cell lines were cultivated in DMEM (Life Technologies) supplemented with 10% heat-inactivated FCS, sodium-pyruvate, non-essential amino acids, glutamax and penicillin-streptomycin. Regular testing assured that cell lines used throughout this study were negative for mycoplasma contamination.

Temperature-sensitive VACV strains Cts15 (F10L), Cts28 (F10L), Cts16 (I7L) and Cts10 (I8R) were kindly provided by Dr Richard Condit34 (University of Florida, Gainesville, FL, USA) and grown at 31 °C. A doxycycline-inducible VACV mutant of the A7L gene (viA7L) was a kind gift of Dr Paulo H. Verardi53 (University of Connecticut, Mansfield, CT, USA). IPTG-inducible VACV strains of the F10L and L3R genes were kindly provided by Dr Bernard Moss26,51 (National Institutes of Health, Bethesda, MD, USA) and grown in the presence of 50 μM and 25 μM IPTG, respectively. The IPTG-inducible VACV strain vindH1 was a kind gift of Paula Traktman22 (Medical University of South Carolina, SC, USA) and grown in the presence of 5 mM IPTG. Recombinant VACV strain viA7L mCherry-A4 was generated based on strain viA7L by replacing the endogenous A4L gene with the mCherry-A4L fusion as described before61,62. Recombinant viA7 mCherry-A4 was identified based on plaque fluorescence and purified by at least two rounds of plaque purification. Mature virion particles were produced in BSC40 cells and purified from cytoplasmic lysates as described elsewhere63.

Lysis and tryptic digest

Mature virions or infected cells were vortexed for 10 min at full speed in 8 M urea containing 100 mM ammonium bicarbonate pH 8.2, one tablet of phosphatase inhibitors cocktail (PhosStop, Roche) per 10 ml buffer followed by three times sonication of 30 seconds each at 80% amplitude and 0.8 cycle (Vialtweeter, Hielscher). The insoluble fraction was pelleted at 15,000 × g for 10 min and the supernatant was collected. The proteins were reduced with 5 mM tris(2-carboxyethyl)phosphine for 20 min at room temperature. Free cysteines were alkylated with 10 mM iodoacetamide for 30 min at room temperature in the dark. For tryptic digest, the solution was diluted eightfold with 100 mM ammonium bicarbonate pH 8.2 and sequence grade trypsin (Promega) was added to a protein/enzyme ratio of 50:1 and incubated for 16 hours at 37 °C under constant agitation. Trypsin was inactivated by addition of trifluoroacetic acid (TFA) to a final concentration of 0.2% (v/v). Resulting peptides were desalted on a reverse phase C18 column (Waters) and eluted with 50% acetonitrile (ACN), 0.5% TFA. The solvents were evaporated using a centrifuge evaporator device. Peptides from mature virion origin (H1(−) or wild-type) were either resuspended in 2% ACN, 0.1% formic acid (FA) for direct LC-MS/MS analysis for whole proteome comparison or further processed as described below.

Heavy labelled peptide synthesis

The internal standards used for relative quantification of I7 and I7 S134 phosphorylation were obtained from JPT Peptide Technology. The peptide sequences AIDFSQMDLK and AIDFpSQMDLK were synthesized with a stable isotope standard label on the C-terminal lysine K (U-13C6;U-15N2). Crude synthetic peptides were resuspended in 0.2:0.8:0.1 ACN/H2O/TFA and further diluted with 1:0.1 H2O/TFA. A constant amount of AIDFSQMDLK peptide was spiked into the protein-digested samples before C18 desalting and, in the case of the AIDFpSQMDLK peptide, before phosphopeptide enrichment.

Phosphopeptides enrichment on titanium dioxide beads

Two milligrams mature virions or 3 mg proteins from infected HeLa CCL2 cells (~15 × 106 cells) were lysed and digested as described above. Biological triplicates were used for each condition.

Phosphopeptide enrichment was adapted from64. TiO2 resin was used at 1.25 mg per 1 mg desalted peptides (Titansphere, 5 micron, GL Sciences). The resin was washed with 2-propanol and incubated in saturated phtalic acid solution containing 80% ACN and 3.5% TFA. The peptides were solubilized in the same phtalic acid solution. The resin was added to the peptide solution, incubated for 1 hour at room temperature and pelleted. The supernatant was discarded, the resin washed twice with phtalic acid solution and three times with 50% ACN, 0.1% FA. Bound peptides were eluted with 0.3 M ammonium hydroxide and acidified with TFA to pH < 3. Subsequently, the peptides were desalted using a C18 column (UltraMicroSpin, The Nest Group) following the manufacturer’s instructions. Desalted peptides were dried in a centrifuge evaporator device and resuspended in 2% ACN, 0.1% FA for LC-MS/MS analysis.

Immunoprecipitation of peptides containing phosphorylated tyrosine

Peptides containing phosphorylated tyrosine residue(s) were enriched following the instructions of the PTMScan Phospho-Tyrosine P-Tyr-100 kit (Cell Signalling Technology). Briefly, 5 mg of desalted peptides from mature virion lysate were resuspended in IAP buffer plus detergent and incubated with p-Tyr-100 antibodies coated beads. The beads were washed and eluted with 0.15% TFA according to the manufacturer instructions. Enriched peptides were desalted using C18 containing tips (ZipTip, Millipore). Desalted peptides were dried in a centrifuge evaporator and resuspended in 2% ACN, 0.1% FA for LC-MS/MS analysis. Due to the high protein amount needed for pY enrichment, we restricted our analysis to one unique replicate for each condition (H1(−) or wild-type mature virions).

LC-MS/MS

DDA

Phosphopeptide enriched samples from mature virions and HeLa cells were separated by reversed-phase chromatography on a high-pressure liquid chromatography (HPLC) column (75 μm inner diameter, New Objective) that was packed in-house with a 10 cm stationary phase (Magic C18AQ, 200 Å, 3 Michrom Bioresources) and connected to a nano-flow HPLC combined with an autosampler (EASY-nLC II, Proxeon). The HPLC was coupled to an LTQ-Orbitrap XL mass spectrometer (Thermo Scientific) equipped with a nanoelectrospray ion source (Thermo Scientific). Peptides were loaded onto the column with 100% buffer A (99.9% H2O, 0.1% FA) and eluted at a constant flow rate of 300 nl per min over a 60 min (mature virions) or a 90 min (HeLa) linear gradient from 7% to 35% buffer B (99.9% ACN, 0.1% FA). After the gradient, the column was washed with 80% buffer B and re-equilibrated with buffer A. Mass spectra were acquired in a data-dependent manner, with an automatic switch between MS1 and MS2. High-resolution mass spectrometry scans were acquired in the Orbitrap (resolution 60,000 at 400 m/z, automatic gain control (AGC) target value 106) to monitor peptide ions in the mass range of 300–1,600 m/z, followed by collision-induced dissociation MS2 scans in the ion trap (minimum signal threshold 250, AGC target value 104, isolation width 2 m/z) of the five most intense precursor ions. To avoid multiple scans of dominant ions, the precursor ion masses of scanned ions were dynamically excluded from MS2 analysis for 30 s. Singly charged ions and ions with unassigned charge states were excluded from MS2 fragmentation.

Peptide samples from whole mature virion lysis were separated by reversed-phase chromatography on an ultra high-pressure liquid chromatography (uHPLC) column (75 μm inner diameter, 15 cm, C18, 100 Å, 1.9 μm, Dr Maisch, packed in-house) and connected to a nano-flow uHPLC combined with an autosampler (EASY-nLC 1000, Thermo Scientific). The uHPLC was coupled to a Q-Exactive Plus mass spectrometer (Thermo Scientific) equipped with a nanoelectrospray ion source (NanoFlex, Thermo Scientific). Peptides were loaded onto the column with buffer A (99.9% H2O, 0.1% FA) and eluted at a constant flow rate of 300 nl per min over a 90 min linear gradient from 7% to 35% buffer B (99.9% ACN, 0.1% FA). After the gradient, the column was washed with 80% buffer B and re-equilibrated with buffer A. Mass spectra were acquired in a data-dependent manner, with an automatic switch between mass spectrometry and MS/MS scans. Survey scans were acquired (70,000 resolution at 200 m/z, AGC target value 106) to monitor peptide ions in the mass range of 350–1,500 m/z, followed by higher energy collisional dissociation MS2 scans (17,500 resolution at 200 m/z, minimum signal threshold 420, AGC target value 5 × 104, isolation width 1.4 m/z) of the ten most intense precursor ions. To avoid multiple scans of dominant ions, the precursor ion masses of scanned ions were dynamically excluded from MS/MS analysis for 10 s. Singly charged ions and ions with unassigned charge states were excluded from MS/MS fragmentation.

DIA

Retention time peptides (iRT peptides, Biognosys AG) were added to the input (non-phospho-enriched peptide) and phosphopeptide samples from the Rif release time course experiment performed in HeLa cells. One microgram of peptide of each sample was loaded on the analytical column and separated by reversed-phase chromatography on a uHPLC column (75 μm inner diameter, 15 cm, C18, 100 Å, 1.9 μm, Dr Maisch, packed in-house) and connected to a nano-flow uHPLC combined with an autosampler (EASY-nLC 1000, Thermo Scientific). The uHPLC was coupled to either a Q-Exactive Plus (input sample) or a Fusion Tribrid mass spectrometer (phosphosamples) (both Thermo Scientific) equipped with a nanoelectrospray ion source (NanoFlex, Thermo Scientific). Peptides were loaded onto the column with buffer A (99.9% H2O, 0.1% FA) and eluted at a constant flow rate of 300 nl per min over a 120 min linear gradient from 7% to 35% buffer B (99.9% ACN, 0.1% FA). After the gradient, the column was washed with 80% buffer B and re-equilibrated with buffer A. Survey scan covering the 400–1220 m/z range (MS1, 60,000 resolution, AGC target 1E6, 64 ms) was followed by higher energy collisional dissociation mass spectra acquired in a data-independent manner using 19 overlapping variable windows (MS2, 30,000 resolution, AGC target 1E6, injection time 54 ms).

VACV H1(−) and wild-type mature virion relative protein quantification

SEQUEST (v27.0)65 was used to search fragment ion spectra for a match to fully tryptic peptides without missed cleavage sites from a protein database, which was composed of human proteins (SwissProt, v57.15), VACV proteins (UniProt, strain Western Reserve, v57.15) and various common contaminants, as well as sequence-reversed decoy proteins. The precursor ion mass tolerance was set to 20 ppm. Carbamidomethylation was set as a fixed modification on all cysteines. The PeptideProphet and the ProteinProphet tools of the Trans-Proteomic Pipeline (TPP v4.6.2)66 were used for probabilistic scoring of peptide-spectrum matches (PSMs) and protein inference. Protein identifications were filtered to reach an estimated false discovery rate of ≤ 1%. Peptide feature intensities were extracted using the Progenesis LC-MS software v2.0 (Nonlinear Dynamics). Protein fold changes and their statistical significance between paired conditions were tested using at least two fully tryptic peptides per protein with the MSstats library (v1.0)67. Resulting P values were corrected for multiple testing with the Benjamini–Hochberg method68.

Phosphorylation site identification and localization

SEQUEST (v27.0)65 was used to search fragment ion spectra for a match to semitryptic peptides with up to two missed cleavage sites from a protein database, which was composed of human proteins (SwissProt, v57.15), VACV proteins (UniProt, strain Western Reserve, v57.15)69 and various common contaminants, as well as sequence-reversed decoy proteins. The precursor ion mass tolerance was set to 0.05 Da. Carbamidomethylation was set as a fixed modification on all cysteines and phosphorylation of serines, threonines and tyrosines as well as oxidation of methionines were considered optional modifications. Resulting PSMs were statistically validated and filtered for a minimum probability of 0.9 using PeptideProphet (TPP v4.6.2)66. Phosphorylated PSMs were further assessed using PTMprophet (TPP v4.6.2)70 computing a localization probability for any of the serines, threonines and tyrosines within a peptide sequence. Based on these probabilities, we pinned down phosphorylations to actual sites as follows. For a PSM where a phosphorylation was localized to a site with a probability of 0.9 or higher, this site was considered phosphorylated. Phosphorylation sites with a probability of 0.1 or lower were discarded. For a PSM with phosphorylation localization probabilities between 0.1 and 0.9, the following heuristic was used to derive the phosphorylation site: For each phosphorylation site its maximum localization probability across all samples was calculated. If there was only one site with a sample-wide probability of 0.9 or higher, this site was considered phosphorylated. If there were multiple sites with sample-wide probabilities of 0.9 or higher, those sites were combined to a ‘shared phosphorylation’. Also, if none of the sites had sample-wide probabilities of 0.9 or higher, the sites were combined to a ‘shared phosphorylation’ and reported with the separator ‘|’. This procedure was repeatedly applied for multiply phosphorylated PSMs.

Phosphorylation site quantification and differential expression analysis

Peptides were quantified on MS1 level using Skyline (version 1.3)71. The integrated areas of a peptide’s isotopic peaks were summed and peptides with ambiguous PTM sites were merged if they had a retention time overlap of more than 50%. Phosphorylation site localization was determined as described above. MSstats (version 1.0)67 was used to determine statistically significant differentially expressed phosphorylation sites by building an analysis of variance model for each site, based on all quantified peptides featuring this site. Shared phosphorylation sites as well as protein groups were used in each site/group repeatedly. Tests for differential expression were performed for each site and each pair of conditions and resulting P values were corrected for multiple testing with the Benjamini–Hochberg method68.

Data analysis of DIA LC-MS/MS experiment

LC-MS/MS DIA runs were analysed with the software Spectronaut Pulsar version 11.0.18206.14.29112 (Biognosys AG)72,73 using default settings. Briefly, a spectral library for the input samples (non-phospho-enriched) was generated from four input samples measured in DDA on a Q-Exactive HF (Thermo Scientific). DDA data were searched with MaxQuant74,75 using a combined protein database human (SwissProt, v57.15) and VACV (UniProt, strain Western Reserve, v57.15), full tryptic allowing two missed cleavages including carbamidomethylation as a fixed modification on all cysteines and oxidation of methionines and protein N-terminal acetylation as optional modifications. For the phospho-enriched samples, DIA runs were directly searched using the DirectDIA function of Spectronaut Pulsar, full tryptic allowing two missed cleavages including carbamidomethylation as a fixed modification on all cysteines, oxidation of methionines, protein N-terminal acetylation and phosphorylation of serines, threonines and tyrosines as optional modifications. Both spectral libraries were imputed with the mass spectrometric assays of the heavy labelled peptides spike-in which were derived in silico in Spectronaut Pulsar from the identified endogenous peptides. Spectral libraries were then used for targeted data extraction of DIA runs in Spectronaut Pulsar. Extracted peptide intensities were used for plotting in R.

Immunofluorescence and electron microscopy

BSC40 cells were seeded onto 13 mm glass coverslips in 24-well plates, infected with strain viA7 mCherry-A4 at a multiplicity of infection (MOI) of 4 as described in the transient complementation section and transfected with 800 ng plasmid and 3 ul Lipofectamine2000. For immunofluorescence staining, cells were fixed in 4% formaldehyde, permeabilized with 0.5% Triton X-100 and stained with primary anti-HA (1:1,000) and AlexaFluor-647-coupled secondary antibody (1:1,000). Images were acquired using a Leica DM2500 confocal microscope. For conventional TEM, coverslips were fixed in 2.5% glutaraldehyde (0.05 M sodium cacodylate adjusted to pH 7.2, 50 mM KCl, 2.5 mM CaCl2) for 45 min at room temperature. After several washes with 50 mM sodium cacodylate buffer, they were postfixed in OsO4 (2% OsO4 in water) for 1 h on ice, followed by blockstaining in 0.5% aqueous uranyl acetate overnight. The specimens were then dehydrated in graded ethanol series and propylene oxide, followed by embedding in Epon. Ultrathin sections (50–60 nm) were obtained using an FC7/UC7-ultramicrotome (Leica). Sections were examined with a CM10 Philips transmission electron microscope with an Olympus ‘Veleta’ 2k x 2k side-mounted TEM charge-coupled device (CCD) camera.

Transient complementation assay

VACV genes F10L, I7L, L3R, I8R and A7L and their endogenous promoters were amplified from VACV genomic DNA. C-terminal HA-tags were introduced by two consecutive rounds of PCR and the products cloned into pBSIIKS using the primers and enzymes listed in Supplementary Table 5. Amino acid changes were introduced using QuickChange site-directed mutagenesis (Agilent Technologies).

For transient complementations, confluent six-well dishes of BSC40 cells were infected at an MOI of 4 as for the Rif release experiment and then maintained either at permissive/induced or non-permissive/non-induced condition, as indicated. At 4 h post infection, 4 ug of plasmid DNA encoding either the wild-type protein sequence or phosphodeletion and phosphomimetic mutations thereof were introduced by transfection using 15 ul Lipofectamine2000 reagent. Empty vector (pBSIIK) was used as control. At 24 h post infection, produced infectious virions were isolated as described and titered on BSC40 cells at the permissive condition. For I8R transient complementations, cells were maintained at permissive condition until 5 h post infection to allow early transcription of the thermolabile particle76. Expression of transfected constructs was verified by western blot analysis using anti-HA antibodies. Full immunoblot images of all cropped immunoblots presented throughout the manuscript are presented in Supplementary Fig. 8.

Reporting Summary

Further information on experimental design is available in the Nature Research Reporting Summary linked to this article.

Data availability

The datasets generated during and/or analysed during the current study are available in the MassIVE repository under MassIVE ID MSV000081854.

Data can be downloaded via FTP: ftp://massive.ucsd.edu/MSV000081854. The DIA runs and spectral libraries can be reviewed with the Spectronaut Viewer software (www.biognosys.com/spectronaut-viewer).