The Major Structural Protein of African Swine Fever Virus, p73, Is Packaged into Large Structures, Indicative of Viral Capsid or Matrix Precursors, on the Endoplasmic Reticulum

Abstract

African swine fever virus (ASFV) is a large enveloped DNA virus that shares the striking icosahedral symmetry of iridoviruses. To understand the mechanism of assembly of ASFV, we have been studying the biosynthesis and subcellular distribution of p73, the major structural protein of ASFV. Sucrose density sedimentation of lysates prepared from infected cells showed that newly synthesized p73 was incorporated into a complex with a size of 150 to 250 kDa. p73 synthesized by in vitro translation migrated at 70 kDa, suggesting that cellular and/or viral proteins are required for the formation of the 150- to 250-kDa complex. During a 2-h chase, approximately 50% of the newly synthesized pool of p73 bound to the endoplasmic reticulum (ER). During this period, the membrane-bound pool of p73, but not the cytosolic pool, formed large complexes of approximately 50,000 kDa. The complexes were formed via assembly intermediates, and the entire membrane-associated pool of p73 was incorporated into the 50,000-kDa complex within 2 h. The 50,000-kDa complexes containing p73 were also detected in virions secreted from cells. Immunoprecipitation of sucrose gradients with sera taken from hyperimmune pigs suggested that p73 was the major component of the 50,000-kDa complex. It is possible, therefore, that the complex contains between 600 and 700 copies of p73. The kinetics of complex formation and envelopment of p73 were similar, and complex formation and envelopment were both reversibly inhibited by cycloheximide, suggesting a functional link between complex assembly and ASFV envelopment. A protease protection assay detected 50,000-kDa complexes on the inside and outside of the membranes forming the viral envelope. The identification of a complex containing p73 beneath the envelope of ASFV suggests that p73 may be a component of the inner core shell or matrix of ASFV. The outer pool may represent p73 within the outer capsid layer of the virus. In summary, the data suggest that the assembly of the inner core matrix and outer capsid of ASFV takes place on the ER membrane during envelopment and that these structures are not preassembled in the cytosol.

African swine fever virus (ASFV) is a large icosahedral enveloped DNA virus that causes a lethal hemorrhagic disease in domestic pigs. The virus was first described by Montgomery in 1921 (21), but classification of ASFV remains controversial. The presence of inverted terminal repeats and covalently linked ends in the 170-kDa genome suggests similarities to poxviruses (13). Structurally, however, ASFV more closely resembles the iridoviruses. Both have striking icosahedral symmetry (10, 23, 34), and the major structural protein of ASFV, p73, shares sequence homologies with the major capsid protein of frog 3 virus (19), a prototype iridovirus. ASFV assembles in cytoplasmic foci called viral factories. Sections taken through assembly sites reveal fully assembled virons as 200-nm-diameter hexagons in cross section and an ordered series of one- to six-sided assembly intermediates (2, 3, 6, 7, 10, 11, 26). Close inspection of intracellular virions identifies multiple concentric layers of different electron densities. According to recent models, the layers of ASFV represent a central electron-dense nucleoprotein core surrounded by an inner core shell, two inner envelopes, and an outer capsid layer (2, 26). The inner envelopes are formed as virions are wrapped by endoplasmic reticulum (ER) membrane cisternae (26), and a loose external envelope is added as intracellular particles bud from the plasma membrane (3). The external envelope is often lost during purification, leaving the outer capsid as the most external structure. Removal of the outer envelope in vitro by mild detergent and mercaptoethanol reveals the outer capsid layer as an ordered series of protein subunits arranged in hexagonal arrays (10).

The recent sequencing of the ASFV genome (14, 35) has provided primary sequences of several structural proteins. Three proteins with membrane-spanning domains, J13L/p54 (24, 31), i1L/p17 (28), and p22 (8) localize to the virus. Three other structural proteins, the products of the E120R (20), K78R (22), and A104R/5AR genes (5), have DNA-binding properties, and the K78R and 5AR proteins localize to the nucleoid and may be involved in DNA packaging. Interestingly the bulk of the protein content of ASF virus is made up from the products of just two reading frames. The B646L gene encodes p73 and provides 35% of the protein mass of the virus, while a further 25% of the protein content of the virus is provided by the ordered proteolysis of polyprotein pp220 encoded by the CP247L gene (2). This produces viral proteins p150, p37, p34, and p14 (2, 27). Proteolytic processing of a second polyprotein, pp62, produces two further abundant structural proteins, p35 and p15 (29). Electron microscopy studies have localized proteins p150, p37, p34, and p14 to the inner core shell of the virus, where they possibly function as a matrix during assembly (2). The precise location of p73, which has been referred to as the major capsid protein of ASFV (18), remains unclear. Immunoelectron microscopy has localized p73 to the intermediate layers or inner core shell of the virus, where it colocalizes with p37 (11). Immunogold labelling has also identified p73 in the outer capsid layer (4, 32). The observation that the infectivity of ASFV can be neutralized by antibodies specific for p73 (4, 15) also suggests that p73 is a component of the outer capsid of the virus. Neither observation, however, excludes the possibility that p73 is in both the outer capsid and the inner core shell.

We have been studying the biosynthesis and subcellular distribution of p73 as a means of understanding the mechanism of assembly of ASFV (12). One of the first identifiable steps in assembly of ASFV is the translocation of 50% of the newly synthesized pool of p73 from the cytoplasm to the ER membrane. This occurs with a half-time of 5 min. There then follows a lag period of 1 to 2 h, after which approximately 60% of the membrane-bound p73 is enveloped by the ER. This suggests that a pool of p73 is beneath an inner envelope originating from the ER and may therefore be a component of the matrix or inner core shell of the virus. We have suggested that the remaining unenveloped 40% of p73 may constitute a membrane-bound pool of p73 forming the viral capsid on the outside of the inner envelope (12). These biochemical data are consistent with electron microscopy studies showing p73 on both sides of the inner envelope (4, 10, 11, 32). We have now extended our studies of p73 and have asked whether assembly of p73 into structures indicative of capsid or matrix precursors occurs in the cytosol prior to membrane binding or on the membrane during envelopment. The results show that p73 forms a complex of approximately 200 kDa immediately after synthesis in the cytosol. Approximately half of the cytoplasmic pool of p73 then binds to cellular membranes. The membrane-bound pool of p73, but not the cytosolic pool, forms large complexes of approximately 50,000 kDa with kinetics that closely follow the time course of envelopment. Complexes of the same size were detected in virions secreted from cells. The results show that the assembly of p73 into structures indicative of a viral capsid or matrix takes place on the ER membrane and that this is an obligate step in the packaging of p73 into virions secreted from cells.

MATERIALS AND METHODS

Cells, virus, and antibodies.

Vero cells were grown and infected with the BA71V strain of ASFV as previously described (12). Monoclonal antibody 4H3, specific for p73, and a pig polyclonal anti-ASFV serum, MI92, have been described previously (12). Monoclonal antibody 17LD3, specific for p73, was purchased from Ingenasa (Madrid, Spain). Foot-and-mouth disease virus (FMDV) was provided by Wendy Blakemore, and bluetongue virus core particles were supplied by Peter Mertens and Nick Burroughs (Department of Molecular Biology, Institute for Animal Health, Pirbright Laboratories, Surrey, England).

Metabolic labelling of virus and infected cells.

Cells infected with ASFV were preincubated with cysteine- and methionine-free Eagle’s medium for 15 min, and the medium was replaced with 1.85 MBq of [³⁵S]methionine and cysteine (³⁵S-express; New England Nuclear, Boston, Mass.) per ml in methionine-cysteine-free medium for the indicated time periods at 37°C. Cells were washed and chased in Dulbecco’s modified Eagle’s medium. For preparation of radiolabelled virus, 16 h after infection with ASFV, 2 × 10⁸ Vero cells were metabolically labelled with 1.85 MBq of ³⁵S-express per ml in methionine-cysteine-free media for 6 h at 37°C. Dulbecco’s modified Eagle’s medium-HEPES containing 2% fetal calf serum was added to cells for a further 16 h. After removal of cellular debris by centrifugation for 15 min at 3,000 rpm, the medium containing ASFV was centrifuged at 25,000 rpm for 45 min at 4°C in a Beckman SW28 rotor to pellet the virus. Extracellular virus was subsequently purified by Percoll equilibrium centrifugation as previously described (9).

Cell lysis and immunoprecipitation.

At the appropriate time intervals, cells were washed once in phosphate-buffered saline and either released from the flask with EDTA-trypsin or lysed in 1% Brij 35 in immunoprecipitation buffer (10 mM Tris [pH 7.8], 150 mM NaCl, 10 mM iodoacetamide, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 μg (each) of leupeptin, pepstatin, chymostatin, and antipain [Boehringer Mannheim, Lewes, United Kingdom] per ml). Lysates were immunoprecipitated with 4H3 or MI92 immobilized on protein G-Sepharose (Pharmacia Biotech, Uppsala, Sweden), and proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by autoradiography as previously described (33). Protein bands were quantitated with a Bio-Rad 620 video densitometer.

Preparation of cellular membrane fraction.

ASFV-infected Vero cells were stripped from flasks by incubation with EDTA-trypsin, and pelleted at 1,750 rpm for 7 min in a Beckman TJ-6 centrifuge. Cells were resuspended in buffered sucrose (250 mM sucrose, 20 mM Tris, 1 mM EDTA [pH 7.5]) and homogenized by 20 passages through a 25-gauge needle. Whole cells and nuclei were removed by pelleting at 6,000 rpm for 2 min in an Eppendorf 5415 centrifuge. Postnuclear supernatants were pelleted at 14,000 rpm for 20 min at 4°C in an Eppendorf 5402 centrifuge to separate membrane (pellet) and cytosol (supernatant) fractions.

Sucrose density sedimentation analysis.

All gradients contained sucrose dissolved in 1% Brij 35 in immunoprecipitation buffer (10 mM Tris [pH 7.8], 150 mM NaCl, 10 mM iodoacetamide, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 μg [each] of leupeptin, pepstatin, chymostatin, and antipain [Boehringer Mannheim] per ml). Step gradients contained 2-ml layers (each) of 5, 10, 15, 20, and 25% sucrose for the 5 to 25% gradients or 2 ml (each) of 10, 20, 30, 35, and 40% sucrose for the 10 to 40% gradients. Gradients were left at 4°C overnight to equilibrate. Membrane and cytosol fractions were solubilized in 1% Brij 35 in immunoprecipitation buffer as described above. Two-milliliter samples were applied to the top of the gradients layered over a 1-ml 70% cushion. After centrifugation at 40,000 rpm for 20 h at 4°C in a Beckman SW40 rotor, gradients were separated into 1.2-ml fractions. The migration of p73 was analyzed by immunoprecipitation of gradient fractions with monoclonal antibody 4H3. The gradients were calibrated by monitoring the migration of the soluble marker proteins carbonic anhydrase, bovine serum albumin (BSA), immunoglobulin G, β-amylase, and apoferritin. The distribution of the marker proteins was analyzed by SDS-PAGE (12% polyacrylamide) run under reducing conditions followed by Coomassie blue staining.

Continuous 10 to 70% sucrose gradients were prepared by making step gradients containing 2 ml each of 10, 20, 30, 40, 50, and 70% sucrose dissolved in 20 mM Tris (pH 7.5). Gradients were left at 4°C overnight to equilibrate. After centrifugation of samples for 2.5 h at 25,000 rpm in a Beckman SW40 rotor, gradients were separated into 1.2-ml fractions. The velocity gradients were calibrated by observing the migration of FMDV and bluetongue virus core particles. Briefly, 1 ml of each virus preparation diluted in 20 mM Tris (pH 7.5) was loaded on top of the gradient, fractions collected after centrifugation were adjusted to 1% Triton X-100, and the RNA concentration was determined by measuring the A₂₆₀.

Trypsin protection assay for envelopment.

The trypsin protection assay for envelopment along with associated controls has been described in detail previously (12). Briefly, Vero cells infected with the BA71 strain of ASFV were pulse-labelled with ³⁵S-express and homogenized as described above. The postnuclear membrane fraction was incubated with trypsin (0.4 mg/ml) at 37°C for 30 min. Proteolysis was terminated by dilution of the sample with 3 volumes of immunoprecipitation buffer containing 5 mM phenylmethylsulfonyl fluoride, 3% fetal calf serum, and 10 mg of hen egg white trypsin inhibitor (Boehringer Mannheim) per ml. The levels of p73 remaining were determined by immunoprecipitation as described above.

RESULTS

p73 forms a 150- to 250-kDa complex immediately after synthesis.

In the first experiment, the size of newly synthesized p73 was analyzed by sucrose density centrifugation. Vero cells infected with ASFV were pulse-labelled for 20 min and lysed in 1% Brij 35. The lysate containing solubilized membrane and cytosolic proteins was layered over a preformed continuous 5 to 25% sucrose gradient. After centrifugation to equilibrium, the gradient was calibrated by monitoring the sedimentation of marker proteins with sizes of 30 kDa (carbonic anhydrase), 66 kDa (BSA), 150 kDa (immunoglobulin), and 200 kDa (β-amylase). Gradient fractions were then analyzed for the presence of p73 by immunoprecipitation. Surprisingly, p73 was essentially absent from the 70-kDa range (fraction 6 of Fig. 1A), indicating the absence of an intracellular pool of monomeric p73 molecules. Instead, p73 migrated mainly in fractions 3 and 4, which, when interpolated from the size marker distribution, sedimented at 150 to 250 kDa. The size of the p73 complex suggested formation of a dimer or trimer. The experiment did not exclude the possibility that the migration of p73 at 150 to 250 kDa represented incorporation of p73 into a complex containing other proteins; however, candidate associated proteins were not observed as obvious bands on SDS-PAGE gels of p73 immunoprecipitates (Fig. 1A). Alternatively, the complex may contain small proteins below 10 kDa that were not resolved on the gels; however, experiments with high-concentration acrylamide gels do not detect these (data not shown). It was also possible that p73 may be synthesized as a monomer but had migrated abnormally on the gradient. To eliminate this possibility, p73 was translated in vitro and analyzed on the same gradient. Fig. 1B shows that p73 does not migrate aberrantly on sucrose gradients, because the in vitro translation product comigrated with the 66-kDa BSA marker. The next experiment determined the time course of formation of the 150- to 250-kDa complex. Cells were pulse-labelled for 5 min and then chased for 5 min in complete medium, and the size of the complex was analyzed with 10 to 40% sucrose gradients calibrated with standards of 66 kDa (BSA), 200 kDa (β-amylase), and 473 kDa (apoferritin). Figure 2 shows that p73 synthesized during a short, 5-min pulse, and during the 5-min chase, migrated in the 250-kDa range (fractions 5 and 6). Given that protein synthesis proceeds at four amino acids per second and that synthesis of p73 would take approximately 3 min, the data suggest that the 150- to 250-kDa complex was formed immediately after synthesis.

FIG. 1.

Pulse-labelled p73 forms a 150- to 200-kDa complex in infected cells. (A) p73 in infected cells migrates at 150 to 200 kDa. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 20 min and then lysed in immunoprecipitation buffer containing 1% Brij 35. Precleared lysates were loaded on top of 5 to 25% sucrose gradients and centrifuged to equilibrium. Fractions collected from the bottom of the tube were immunoprecipitated with monoclonal antibody 4H3 to detect p73. All samples were analyzed by SDS-PAGE and autoradiography. The migration ranges of marker proteins of 30 kDa (carbonic anhydrase), 66 kDa (BSA), 150 kDa (immunoglobulin), and 200 kDa (β-amylase) on the same gradient are shown. (B) p73 migrates as a 70-kDa monomer when synthesized in vitro. The B646L reading frame of ASFV was metabolically labelled during expression from a rabbit reticulocyte in vitro coupled transcription-translation reaction. Reaction products were solubilized as described above, loaded onto a 5 to 25% sucrose gradient, and centrifuged to equilibrium. Fractions collected from the bottom of the tube were immunoprecipitated with 4H3 and analyzed by SDS-PAGE followed by autoradiography.

FIG. 2.

p73 forms a 150- to 250-kDa complex immediately after synthesis. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 5 min and either lysed immediately in immunoprecipitation buffer containing 1% Brij 35 or chased for 5 min before lysis. Precleared lysates were loaded on top of 10 to 40% sucrose gradients and centrifuged to equilibrium. Fractions collected from the bottom of the tube were immunoprecipitated with monoclonal antibody 4H3 to detect p73. All samples were analyzed by SDS-PAGE and autoradiography. The migration ranges of marker proteins 66 kDa (BSA), 200 kDa (β-amylase), and 473 kDa (apoferritin) on the same gradient are shown.

Membrane-bound, but not cytosolic, p73 complexes assemble into large complexes.

The next experiments were designed to determine whether p73 formed a larger complex indicative of capsid or capsid precursor at later times. We have shown previously that within 2 h of synthesis, p73 is distributed approximately evenly between cytoplasmic and membrane fractions isolated from cells (12). This translocation of p73 to membranes is shown in Fig. 3A. Cells infected with ASFV were metabolically labelled for 5 min and either placed on ice or chased in complete medium for 2 h. Cells were then homogenized, and a postnuclear supernatant was centrifuged to produce a pellet of cellular membranes and a cytosol fraction in the supernatant. These were solubilized and immunoprecipitated with monoclonal antibody 4H3, specific for p73. The results show that p73 isolated from pulse-labelled cells was distributed mainly in the cytosolic fraction (S), whereas when cells were chased for 2 h, p73 was evenly distributed between the cytosol and membranes (M). Our previous work suggests that these are ER membranes (12, 26). In a second experiment, postnuclear membranes prepared after a 2-h chase were lysed in detergent before rather than after being pelleted by centrifugation. The supernatant and pellet were then immunoprecipitated as described above. Figure 3A shows that after lysis of membranes by detergent, p73 remained in the supernatant. This experiment verified that p73 pelleted because of association with membranes, rather than through aggregation into a large complex.

FIG. 3.

Membrane-associated pool, but not cytosolic pool, of p73 assembles into large complexes. (A) p73 binds membranes. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 5 min and then chased in complete medium for 2 h. Cell samples were homogenized, and the presence of p73 in membranes (M) and cytosol (S) was determined by immunoprecipitation with 4H3 (lanes 1 to 4). In a separate experiment (lanes 5 and 6), a postnuclear supernatant prepared from infected Vero cells chased for 2 h was lysed in 1% Brij 35 in immunoprecipitation buffer and then centrifuged at 14,000 rpm for 20 min at 4°C. The presence of p73 in the pellet (P) and supernatant (S) was determined by immunoprecipitation. (B) Membrane-bound p73 forms oligomeric complexes. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 15 min and then chased for 2 h in complete medium as indicated. Cells were homogenized, and soluble and membrane fractions were solubilized in immunoprecipitation buffer containing 1% Brij 35 and analyzed with the 10 to 40% sucrose gradients described in the legend to Fig. 2. Fractions collected from the bottom of the tube were immunoprecipitated to detect p73 (4H3) or ASFV proteins (MI92).

The sizes of complexes formed in the cytosol and on the membrane fraction were compared. Vero cells infected with ASFV were pulse-labelled for 15 min or chased for 2 h in complete growth medium to allow p73 to bind the membranes. Homogenized cells were separated into membrane and cytosol fractions, solubilized in 1% Brij 35, and analyzed with preformed continuous 10 to 40% sucrose gradients. Fig. 3B shows that after a 15-min pulse and following a 2-h chase, the cytosolic pool of p73 migrated in fractions 6 and 7 equivalent to 150 to 250 kDa. The pulse-labelled membrane-associated pool of p73 also migrated in fractions 6 and 7. Surprisingly, after a 2-h chase, the membrane-associated pool of p73 molecules migrated mainly at the bottom of the gradient in fractions 1 and 2. The migration of protein standards on the gradients showed that the membrane-associated p73 complex was larger than 473 kDa; even so, p73 was the only protein band visible after immunoprecipitation. To test for the presence of other ASFV-encoded proteins in the heavy gradient fractions, the experiment with the membrane fraction was repeated, and the gradient fractions were immunoprecipitated with hyperimmune serum (MI92) isolated from pigs recovered from infection with ASFV. The bottom panels of Fig. 3B show that the serum immunoprecipitated several different proteins from membranes isolated from pulse-labelled cells. The major protein precipitated by the antiserum migrated at 70 kDa after SDS-PAGE and between 150 and 250 kDa on the sucrose gradient. These properties suggested it was p73. Consistent with the distribution of the protein standards across the gradient, smaller proteins between 14 and 35 kDa were precipitated from the fractions at the top of the gradient, while larger proteins ranging between 90 and 200 kDa comigrated with p73 in fractions 6 to 8. The protein of approximately 200 kDa is likely to be pp220. pp220 is a virus-encoded polyprotein that is proteolytically processed to produce structural proteins p150, p37, p34, and p14 and has been shown to localize to microsomal membrane fractions (2). Importantly, with the exception of a 30-kDa protein, which was recovered in small amounts from all the fractions, virus-encoded proteins were absent from the heavier fractions of the gradient. The right panel shows a similar analysis of membranes isolated from cells chased for 2 h. Significantly, the 70-kDa (p73) protein was the predominant protein detected at the bottom of the gradient. A 150-kDa protein was also detected; significantly, this protein was not observed in the pulse-labelled membrane fraction, suggesting that it is p150, one of the proteolytic products of pp220 that is produced from pp220 approximately 1 h after synthesis and then selectively packaged into virions (2, 27). The relative intensity of the bands suggests that the levels of p150 in the complex are low. The low levels of p150 visualized by autoradiography do not reflect a low incorporation of methionine or cysteine into p150, because in isolated virions, the intensities of p73 and p150 are approximately equal (2) (Fig. 4A). Low levels could not be explained by a lack of reactivity between the MI92 antibody and p150, because in separate experiments, the MI92 antibody efficiently immunoprecipitated p150 from cell lysates (data not shown). All other membrane-associated proteins detected by MI92 remained in fractions 5 to 8. The data show that of all the proteins recognized by the hyperimmune serum, p73, and possibly a small quantity of p150, formed large complexes indicative of a viral capsid or matrix.

FIG. 4.

Membrane-associated complexes containing p73 migrate at 50,000 kDa and are also present in secreted virions. (A and B) p73 is efficiently solubilized from isolated virions. Metabolically labelled virions were isolated from Vero cells infected with the BA71 strain of ASFV and purified by Percoll centrifugation. (A) Viral proteins separated by SDS-PAGE were detected by autoradiography. The major structural proteins p150, p73, p37, p35, p17, and p12 are indicated on the 12.5% polyacrylamide (left) gel and p150 and p73 are indicated on the 5% polyacrylamide gel (right). (B) Viral samples were solubilized by incubation in immunoprecipitation buffer containing 1% Brij 35 for 2 h at 4°C. Insoluble proteins were pelleted by centrifugation and separated by SDS-PAGE (12.5% polyacrylamide), and the presence of p73 was determined by Western blotting with the monoclonal antibody 17LD3. (C) Analysis of p73 complexes present in membrane fractions and in secreted virions with 10 to 70% velocity gradients. (Top) Purified virions were solubilized by incubation in immunoprecipitation buffer containing 1% Brij 35 for 2 h at 4°C and loaded on top of the 10 to 70% sucrose velocity gradients and centrifuged at 25,000 rpm for 2.5 h. (Bottom) Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 20 min and then chased for 2 h in complete medium. Cells were homogenized, and postnuclear membrane fraction was solubilized by incubation in immunoprecipitation buffer containing 1% Brij 35 for 2 h at 4°C before centrifugation for 2.5 h at 25,000 rpm on continuous 10 to 70% sucrose gradients. For both experiments, fractions were collected from the bottom of the tubes, and the levels of p73 were determined by immunoprecipitation with 4H3, followed by SDS-PAGE and autoradiography. The migrations of molecular mass standards (in kilodaltons) are indicated. (D) Calibration of 10 to 70% sucrose velocity gradients. Purified bluetongue virus cores and FMDV particles were loaded on top of 10 to 70% continuous sucrose gradients. After centrifugation for 2.5 h at 25,000 rpm, fractions were assayed for viral RNA by reading A₂₆₀. •, FMDV virus, 76,000 kDa; ○, bluetongue virus cores, 59,000 kDa.

Oligomers containing p73 migrate at approximately 50,000 kDa and are present in purified virions.

The protein complex containing p73 could be a by-product of the infectious cycle, such as an intracellular aggregate formed from p73 molecules that fail to be packaged into virions. Alternatively, the complex could be a bona fide assembly precursor incorporated into virions secreted from cells. To test these possibilities, the sizes of the p73 complexes within virions were determined. ASFV-infected Vero cells were pulse-labelled for 6 h and then chased for 16 h with complete media. Secreted virions were pelleted from culture supernatants and purified with Percoll gradients as described previously (9). Fig. 4A shows that the virus preparation contained the major structural proteins of ASFV, the most heavily labelled being p73 and p150. Importantly, the virus preparation did not contain the pp220 precursor polyprotein, indicating the absence of cell-associated proteins. The virus preparation was then solubilized with 1% Brij 35 in immunoprecipitation buffer, and the proportion of p73 released from virions was determined by Western blotting. Figure 4B indicates that the lysis buffer removed approximately 80% of p73 molecules from the virus. The size of the p73 complex released from the virus was determined by centrifugation with 10 to 70% sucrose velocity gradients. Figure 4C shows that the p73 complex present in secreted virions migrated mainly in fractions 3 to 5 on this gradient. A small quantity of p73 was observed in fractions 7 to 10, which we interpret to be p73 that had not migrated away from the loading position. This may represent a small pool of 150- to 250-kDa complexes present in virions or molecules of p73 that had disassembled from the large complex during isolation and centrifugation. The gradient was calibrated with viral particles of known molecular size (Fig. 4D). The structures of FMDV (1) and bluetongue virus cores are known and have approximate molecular masses of 7,600 and 59,000 kDa, respectively. FMDV particles migrated mainly in fraction 7 and bluetongue virus cores migrated mainly in fraction 2 of the velocity gradient. The p73 complex released from virions migrated slightly slower than the bluetongue virus core particle, suggesting a molecular mass of approximately 50,000 kDa. When the cellular membrane-associated p73 complex was analyzed on the same gradient, it migrated with a broad distribution mainly between fractions 1 and 5 (Fig. 4C), but peak levels were found in fractions 2 to 4. Importantly, the molecular size profile of the membrane-associated pool of p73 was very similar to that observed for p73 in virions. Taken together, the results suggest that the 50,000-kDa complex of p73 formed on cellular membrane is incorporated into virions.

The 50,000-kDa complex containing p73 is formed via assembly intermediates.

There are two pathways for the formation of a large complex with a size of 50,000 kDa. The newly synthesized p73 150- to 250-kDa precursors could be transferred to a large preformed structure within cells, or the complex could be formed through the progressive incorporation of p73 precursors into assembly intermediates. To test for the presence of assembly intermediates, the size of the p73 complex was analyzed at increasing times after synthesis with 10 to 70% velocity gradients. The results in Fig. 5 show that after a 20-min pulse, p73 was observed migrating at the top of the gradient in fraction 10. The results in Fig. 2 showed that at this time point, p73 migrates at 150 to 250 kDa. After 30- and 60-min chase times, a proportion of the p73 had moved from the top of the gradient and migrated as a broad range of assembly intermediates across the gradient in fractions 2 to 9. After a 90-min chase, the levels of p73 migrating at the top of the gradient were much reduced, and there was a proportional increase in the appearance of the 50,000-kDa complex migrating in fractions 2 to 5. This size profile was maintained 2 h into the chase and was similar to the migration of p73 solubilized from purified viral particles (Fig. 4C). The data suggest that the newly synthesized 150- to 250-kDa complex of p73 does not bind to a preformed structure within cells, but instead, the 50,000-kDa complex is formed through progressive assembly of intermediate structures ranging from 150 to 50,000 kDa.

FIG. 5.

The 50,000-kDa p73 complex is formed through assembly intermediates. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 20 min and then chased for increasing times (30 to 120 min) as indicated. Cells were homogenized, and solubilized membrane fractions were loaded on top of the 10 to 70% velocity gradients. Fractions collected from the bottom of the tube were immunoprecipitated with 4H3, and samples were analyzed by SDS-PAGE followed by autoradiography. The migration of molecular mass markers FMDV and bluetongue virus cores (BTV-core) is shown.

Oligomerization of p73 on the ER membrane follows kinetics similar to those of the envelopment of virus particles.

We have described a protease protection assay for envelopment of the membrane-bound pool of p73 (12). The assay involves the addition of trypsin to membrane fractions isolated from infected cells. Membrane-bound p73 that is resistant to trypsin digestion is considered to be enveloped, while nonenveloped p73 is accessible to trypsin and is degraded. With this assay, we have observed the time course of envelopment of newly synthesized p73 and shown that the capsid protein is enveloped between 1 and 2 h after synthesis (12). The next experiments combined the protease protection assay with the sucrose density centrifugation assay for complex formation to determine if envelopment preceded complex assembly or vice versa. ASFV-infected Vero cells were pulse-labelled for 20 min and chased for increasing times in complete growth media. Membrane fractions isolated at each time point were split into two samples: half were analyzed by 10 to 40% sucrose gradients to test for p73 assembly. The other half were incubated in the presence or absence of trypsin to test for envelopment. All fractions were immunoprecipitated and analyzed by SDS-PAGE. The left-hand side of Fig. 6A shows that after the 20-min pulse, greater than 90% of p73 was present in its 150- to 250-kDa form (fraction 6); this level remained approximately the same at the 30- and 60-min time points. The right-hand panels show that the membrane-bound pool at these time points was degraded and therefore accessible to trypsin and not enveloped. After 90 min, 65% of the p73 molecules had assembled into large complexes observed at the bottom of the gradient. The extent of assembly increased to 90% by 120 min. Significantly, the appearance of p73 at the bottom of the gradients correlated with the protection of the membrane-associated p73 from trypsin. Densitometric analysis of the autoradiographs (Fig. 6B) allowed the kinetics of envelopment and oligomerization to be compared. The graph shows that the high-molecular-mass complexes containing p73 were observed at the same time as the appearance of a trypsin-protected pool of p73. Assembly of high-molecular-mass p73 complexes and envelopment of p73 therefore occur with similar kinetics. After carrying out the experiment several times, we noted that on average, the extent of oligomerization was slightly greater than that of envelopment. At maximum, between 80 and 90% of the membrane-bound pool of p73 was incorporated into an oligomer, while 60 to 80% was enveloped.

FIG. 6.

p73 oligomerization and envelopment occur with similar kinetics. (A) Time course of p73 oligomerization and envelopment. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 20 min and then chased in complete medium for the indicated time intervals. Cells were homogenized and membranes were split into two fractions: half were solubilized in 1% Brij 35 in immunoprecipitation buffer and centrifuged for 20 h on 10 to 40% continuous sucrose gradients (left), and the other half were incubated in the absence (−) or presence (+) of trypsin to assess for envelopment (right). All samples were analyzed by SDS-PAGE followed by autoradiography. Note that the newly synthesized 150- to 250-kDa complex of p73 migrates in fraction 6 of these gradients. (B) Densitometric analysis of the kinetics of oligomerization and envelopment. The relative levels of p73 present on autoradiographs were determined by densitometry. The percentages of p73 in the 150-kDa (○), oligomerized (•) and enveloped (▪) forms are shown.

Oligomerization of p73 on the ER membrane is functionally linked to envelopment.

Arzuza et al. (3) have shown by electron microscopy that the assembly of ASFV can be reversibly inhibited by cycloheximide, an inhibitor of protein synthesis. This provided us with an opportunity to establish a functional relationship between the oligomerization of p73 and the assembly of ASFV. We argued that if oligomerization of p73 was essential for assembly of ASFV, then a block in assembly induced by cycloheximide would result in a block in oligomerization; furthermore, a reversal of the block in assembly by removing the drug would allow oligomerization of p73 to proceed.

ASFV-infected Vero cells were pulse-labelled and chased in the absence or presence of cycloheximide for the times indicated in the legend to Fig. 7. The sucrose gradients on the left indicate the size of the p73 complexes formed at each time point. The trypsin protection assays for envelopment are shown on the right. Figure 7 (left panels) shows that after a 2-h chase in the absence of cycloheximide (2 hrs control), 90% of p73 migrated to the bottom of the gradient and a similar amount was enveloped. This is consistent with our previous experiment showing that oligomerization and envelopment follow similar kinetics. After a 2-h chase in the presence of cycloheximide (2 hrs + cx), large complexes containing p73 were absent from the bottom of the gradient. Cycloheximide therefore blocked the assembly of p73 into large complexes. The trypsin protection assay shows that cycloheximide also prevented envelopment of p73. In the next experiment, cells incubated for 2 h in the presence of cycloheximide (2 hrs + cx) were washed and chased for 3 h in complete medium in the absence of the drug. Figure 7 (2 hrs + cx, 3 hrs − cx) shows that after removal of the cycloheximide, the block in assembly and envelopment of p73 was reversed. Three hours after removal of the drug, approximately 70% of the p73 protein now migrated to the bottom of the gradient and 40% had become protected from trypsin. The results strongly suggest a functional link between oligomerization of p73 and assembly and envelopment of the virus.

FIG. 7.

Oligomerization and envelopment of p73 on the membrane are reversibly inhibited by cycloheximide. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 20 min and then chased in complete growth medium in the absence or presence of 50 μg of cycloheximide per ml for the time intervals indicated. Cells were homogenized, and membranes were split into two fractions: half was solubilized in 1% Brij 35 in immunoprecipitation buffer and centrifuged for 20 h on 10 to 40% continuous sucrose gradients (left); the other half was incubated in the absence (−) or presence (+) of trypsin to assess for envelopment (right). All samples were analyzed by SDS-PAGE followed by autoradiography. Cells either were chased for 2 h (2 hrs control) or 5 h (5 hrs control) in complete growth medium or were incubated with cycloheximide for 2 h (2 hrs + cx) or 5 h (5 hrs + cx). The reversibility of cycloheximide was tested by incubating cells with the drug for 2 h and then chasing for 3 h in the absence of the drug (2 hrs + cx, 3 hrs − cx).

The formation of large complexes containing p73 does not require envelopment.

We have shown above that it is possible to distinguish between enveloped and unenveloped pools of p73 by adding trypsin to the membranes isolated from cells. We have carried out the assay several times, and even when chase times are extended to 4 h, we always see that at least 20% of the membrane-associated p73 remains sensitive to trypsin. The next experiments were designed to determine the molecular size of the unenveloped pool of p73. ASFV-infected Vero cells were pulse-labelled for 20 min and chased for 2 h in complete growth media. After the chase, a postnuclear membrane fraction was prepared and split into two fractions: one was incubated with trypsin for 30 min at 37°C to remove the unenveloped p73 molecules, and the other was kept on ice. The membranes were then solubilized and centrifuged to equilibrium on 10 to 40% sucrose gradients. Figure 8A shows the results obtained for the control sample that was not incubated with trypsin; approximately 90% of p73 loaded on the gradient resolved in fractions 1 and 2, indicating formation of a large complex. When membranes were incubated in the presence of trypsin before centrifugation (Fig. 8B), the levels of p73 in fractions 1 and 2 at the bottom of the gradient were reduced by approximately 40%. This external pool digested by the protease was visualized by the quantitative recovery of 15- and 25-kDa tryptic fragments migrating between 250 and 150 kDa in lanes 6 and 7. The loss of p73 from the bottom of the gradient after addition of trypsin shows that the external pool of p73 had assembled into a large complex. The results show that envelopment is not an obligate requirement for the formation of the 50,000-kDa capsid precursor containing p73 and that these complexes can form on both sides of the viral envelope. Alternatively, it is possible that, as has been shown for vaccinia virus (25), ASFV virions are not completely sealed after envelopment, allowing access of trypsin to internal proteins.

FIG. 8.

The 50,000-kDa complexes containing p73 form on both sides of membranes forming the viral envelope. Fourteen hours after infection with ASFV, Vero cells were metabolically labelled for 20 min and then chased for 2 h in complete medium. Cells were homogenized and postnuclear membranes were incubated in the absence (A) or presence (B) of trypsin as indicated. Membranes were repelleted and solubilized in 1% Brij 35 immunoprecipitation buffer, and lysates were loaded onto 10 to 40% continuous sucrose gradients and centrifuged for 20 h. Fractions were collected from the bottom of the tubes, immunoprecipitated with 4H3, and analyzed by SDS-PAGE followed by autoradiography. Note that the 150- to 250-kDa complex migrates in fraction 6 of these gradients.

DISCUSSION

The capsid layer of ASFV has been identified as an electron-dense layer surrounding the inner viral membrane envelope. Negative staining of the virus reveals the capsid as an ordered series of protein subunits arranged in hexagonal arrays (10). Recent studies have described a second protein complex, called the viral matrix or inner core shell, beneath the concave face of the inner envelope (2, 26). We have shown previously that newly synthesized p73 molecules are distributed approximately evenly between a soluble cytoplasmic pool and a membrane-associated pool bound to the ER (12). In this study, we wished to determine if the cytoplasmic pool or the membrane-bound pool was incorporated into protein complexes indicative of capsid or matrix precursors. When infected cells were pulse-labelled for 5 min, p73 migrated at 150 to 250 kDa on sucrose gradients, suggesting that p73 forms a complex of 150 to 250 kDa immediately after synthesis. We cannot exclude the possibility that p73 assembles with other proteins to produce this complex, but candidate proteins were not visible after SDS-PAGE analysis of immunoprecipitates of gradient fractions. Assembly of p73 into a 150- to 250-kDa complex was not an innate property of the protein, because p73 synthesized by in vitro translation migrated at 70 kDa on sucrose gradients. Assembly of the complex therefore required viral and/or cellular proteins. The lack of detection of candidate proteins after SDS-PAGE suggests that interactions with p73 are weak and/or transient, and their identity remains unknown.

When the membrane fraction was analysed after a 2-h chase, p73 migrated at the bottom of the sucrose gradient, suggesting formation of a complex in excess of 473 kDa. Significantly, the cytosolic pool remained at 150 to 250 kDa, suggesting that assembly of p73 into large complexes indicative of capsid or matrix precursors occurred on cellular membrane compartments and not in the cytosol. Our previous work (12, 26) suggests that these are membrane cisternae of the ER. An approximate size for the membrane-associated complex was obtained from velocity gradients calibrated with FMDV (7,600 kDa) and core particles of bluetongue virus (59,000 kDa). The p73 complex migrated slightly slower than bluetongue virus core particles, suggesting a mass of approximately 50,000 kDa. Interestingly, p73 was the only protein observed by SDS-PAGE analysis of immunoprecipitates of gradient fractions, suggesting that p73 is the only component of the complex. It is possible, however, that other proteins were present in the complex and that immunoprecipitation of gradient fractions disrupted associations with p73, such that assembly partners were not visualized as coprecipitated proteins on SDS-PAGE gels. We attempted to identify candidate proteins by immunoprecipitating gradient fractions with a polyclonal anti-ASFV serum that recognized more than 20 ASFV-encoded structural proteins. Figure 4 shows that with the exception of small quantities of p150, p73 was the only protein observed migrating as a large complex, suggesting that the complex is formed predominantly from p73. If other viral proteins are indeed absent from the complex, then 685 molecules of p73 would have to assemble together to form a capsid or matrix precursor of 50,000 kDa. It is important to note that we cannot rule out the possibility that p73 associates with p150 to form large protein complexes.

The assembly of p73 into complexes of 50,000 kDa could occur by two basic mechanisms. The complex could be formed through the progressive oligomerization of newly synthesized 150- to 250-kDa p73 precursors, or, alternatively, newly synthesized p73 molecules could be transferred to a large preformed protein complex on the ER membrane. The analysis of membranes taken at increasing times after pulse-labelling clearly showed that structural intermediates, ranging between 150 and 50,000 kDa, were formed before the appearance of the 50,000-kDa complex. These results argued against transfer of p73 to a preformed complex and suggest that newly synthesized p73 molecules assemble together on the ER membrane to form the large capsid or matrix precursor. A model of progressive assembly on the ER membrane is further supported by the observation that oligomerization was blocked by cycloheximide. Assembly of the complex therefore required ongoing protein synthesis.

At the outset, we argued that the 50,000-kDa complexes containing p73 were structural precursors assembled into virions. It was possible, however, that they were nonproductive aggregates of p73 that remained on the ER membrane and were excluded from the mature virion. The identification of 50,000-kDa complexes, but not significant levels of the 150- to 250-kDa form of p73, in virions secreted from cells (Fig. 4) provided strong evidence that the large p73 complexes were productive intermediates in ASFV assembly. Recently, Andres et al. (2) calculated that the total mass of an ASFV particle is 550,000 kDa. The p73 complex of approximately 50,000 kDa resolved on the velocity gradients was therefore 1/11 the size of the fully assembled virion. The virus is therefore disrupted by detergent lysis and centrifugation, releasing the 50,000-kDa p73 complex. Interestingly, ASFV particles are icosahedrons with 12 vertices. It is possible that the 50,000-kDa complexes we are observing are major structural units of the ASFV icosahedron. Direct evidence for this will have to await electron microscopic examination of gradient fractions. The assembly of a complex containing p73 on the ER membrane is also consistent with our previous experiments showing that binding to the ER was an obligate step in the envelopment of p73 (12).

The observation that high-molecular-mass complexes containing p73 were first observed at the same time as the appearance of enveloped p73 provided indirect evidence that oligomerization may be functionally linked to the eventual envelopment of p73. Further evidence for the functional link between envelopment and oligomerization came from experiments with cycloheximide. Arzuza et al. (3) have shown that cycloheximide produces a reversible block in ASFV assembly. Figure 7 showed that under the same conditions, cycloheximide simultaneously inhibited both the envelopment and oligomerization of p73 and that both processes proceeded as normal, with similar kinetics, after removal of the drug.

Interestingly, the results presented in Fig. 6 and 8 showed that after a 2-h chase, between 20 and 40% of the membrane-bound pool of p73 remained accessible to trypsin and that the trypsin-sensitive (unenveloped) pool migrated as a large complex on sucrose gradients. Taken together, these observations provide biochemical evidence that a protein complex containing p73 is formed on both sides of the inner viral envelope. The outer complex of p73 may form part of the outer capsid layer observed by electron microscopy (2, 10, 26). We cannot, however, exclude the possibility that the trypsin-sensitive pool of oligomers found on membranes represents nonproductive structures that are never packaged into virions. It is also possible that, as has been reported for vaccinia virus (25), ER cisternae fail to seal after wrapping virions, and this allows access of trypsin to internal proteins. Even so, an outer pool of p73 on virus particles has been implied by the observation that the virus can be neutralized by antibodies specific for p73 (4, 15). The trypsin-resistant p73 complex detected beneath the inner envelope may form part of the inner core shell of ASFV, a viral layer that also contains the proteolytic products of the pp220 polyprotein (2). This topology would explain the recent observation that p73 binds to p14.5, a structural protein encoded by the E120R open reading frame (20). The DNA-binding properties of p14.5 have led Martinez-Pomares et al. (20) to suggest that the protein has a role in encapsidation of the ASFV genome. An interaction between p14.5 and a membrane-bound protein complex of p73 present in the inner core shell could provide a bridge between the concave face of the viral envelope and the nucleoprotein core containing DNA. These interactions would facilitate the envelopment of the nucleoprotein core by the ER.

How does the biosynthesis of ASFV relate to the assembly pathways used by other DNA viruses? Much of our information on the assembly of large DNA viruses has come from work on poxviruses and herpesviruses (reviewed in reference 16). In common with most double-stranded DNA viruses (reviewed in reference 17), assembly of herpesvirus capsids takes place in the nucleus. The capsid is then exported to the cytoplasm and enveloped by cisternae of the trans-Golgi. This differs from ASFV, where assembly of p73 into capsid and/or matrix precursors appears to take place on the ER membrane during envelopment, rather than in the cytosol or nucleus. For vaccinia virus, the first identifiable step in morphogenesis is the formation of a protein scaffold or matrix on membrane crescents originating from cisternae of the ER and/or intermediate compartment between the ER and Golgi apparatus (30). Nucleoprotein cores appear to condense on the concave face of membrane crescents during envelopment. Our data showing the progressive assembly of a protein complex on the ER membrane suggest that ASFV has adopted a similar assembly strategy. Learning how interactions between viral structural proteins on the membrane lead to the ordered bending of ER cisternae into the striking 200-nm-diameter icosahedral particles observed by electron microscopy presents a major challenge for future work on ASFV assembly.