Inhibition of a Large Double-Stranded DNA Virus by MxA Protein

Antisera TW34, R30, and SB11 that recognize ASFV proteins p34/polyprotein 220 (pp220), pY118L, and pE120R have been described previously (23, 28, 42). Rabbit antiserum specific for the ASFV structural protein pE183L (j13L/p54) was raised using the synthetic peptide SNELDKHTYTNRQRLNEC as described previously (52). Antiserum SB2 was raised using the peptide AKPARQGHNPATGEPI, which corresponds to amino acids 65 to 80 of pA104R. Mouse monoclonal antibody 4H3, which recognizes p73, has been described previously (12), and mouse monoclonal antibody C18, which recognizes p30, was a gift from Dan Rock (College of Veterinary Medicine, University of Illinois at Urbana-Champaign). Mouse antibody to the ASFV capsid protein p73 (clone 17LD3) was purchased from Ingenasa. Mouse monoclonal antibody M143, which recognizes MxA, has been described previously (19), and rabbit anti-Mx antibody was a gift from Peter Stäheli (Department of Virology, University of Freiburg). Mouse anti-γ-tubulin (clone GTU-88) was purchased from Sigma.

Vero cells (ECACC 84113001) were cultured in Dulbecco's modified Eagle's medium (DMEM)-HEPES supplemented with 10% (vol/vol) fetal calf serum. PK-15 porcine kidney epithelial cells (ATCC CCL-33) were cultured in minimal essential medium alpha supplemented with 5% (vol/vol) fetal calf serum. Vero cells expressing MxA (VA3 cells) or MxB (VB22 cells), as well as control cells (VN36 cells) and Vero cells expressing MxA with the E645R mutation [VA(E645R) cells], have been described previously (20) and were grown in DMEM-HEPES supplemented with 10% (vol/vol) fetal calf serum and 2 mg/ml G418 sulfate. The ASFV Badajoz 1971 Vero-adapted strain Ba71v has been described previously (17).

Cytochalasin D, cytosine arabinofuranoside, and [5-(2-thienylcarbonyl)-1H-benzimidazol-2-yl]-carbamic acid methyl ester (nocodazole) were purchased from Sigma and were used at final concentrations of 1 μg/ml, 50 μM, and 0.2 μM, respectively. Recombinant porcine IFN-α was purchased from R&D Systems.

Merged confocal images were created digitally using a Leica LCS Lite 2.5 microscope. All images and gel scans were resized and annotated using Adobe Photoshop CS 8.0.

Subconfluent cells in petri dishes were infected with 1.0 PFU/cell of Ba71v. After absorption, the inoculum was discarded and replaced with DMEM containing 2% (vol/vol) fetal calf serum. Sixteen hours postabsorption, cells were pulse-labeled with 3.7 × 10⁵ Bq/ml [methyl-³H]thymidine (GE Healthcare) for 15 min. The cells were then washed twice with ice-cold phosphate-buffered saline (PBS) and once with ice-cold thymidine buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 5 mM MgCl₂) and collected into thymidine buffer by scraping. Next, cells were resuspended in 100 μl of thymidine buffer containing 0.5% NP-40 and lysed on ice for 30 min. Finally, nuclei were removed by centrifugation, the cytoplasmic fraction was precipitated with trichloroacetic acid, and the level of incorporated [methyl-³H]thymidine was determined with a PerkinElmer 1450 Microbeta counter.

Infected cells were stained with antibody 4H3 to visualize the major capsid protein p73, which accounts for 30% of the total virion mass (5) and was therefore used to determine the extent of the viral factory. Images of factories representing 0.5-μm-thick serial sections through the z plane at a magnification of ×60 were collected. Images were deconvolved and composited together to generate an image representing the maximum cross-sectional area of the factory. Regions of interest around the p73 signal that did not correspond to cytosolic virions (28) were identified, and the cross-sectional area of a given viral factory was determined using the Openlab 3.1.7 measurement module. Statistical analyses of the data sets were run using Minitab 15.1.0.0.

Cells were fixed onto coverslips (VWR) with either 4% (wt/vol) paraformaldehyde in PBS or methanol and permeabilized with 0.2% (vol/vol) Triton X-100 (Sigma). The binding of primary antibodies was visualized using suitable secondary antibodies conjugated to Alexa 488 or Alexa 594 (Molecular Probes) as appropriate. DNA was visualized with 500 ng/ml DAPI (4′,6-diamidino-2-phenylindole; Sigma), and coverslips were mounted with Flouromount G (Southern Biotechnology). For digital deconvolution, cells were viewed at a magnification of ×60 (1.4 numerical aperture) with a Nikon E800 microscope. Images were then captured with a Hamamatsu C-4746A charge-coupled device camera and deconvolved using Improvision Openlab 3.1.7. For confocal imaging, cells were viewed at a magnification of ×63 (1.4 numerical aperture), and images were captured sequentially with a Leica TCS SP2 confocal microscope. Scale bars on merged confocal images represent 10 μm.

Confluent monolayers of cells in 24-well plates were inoculated for 1 h with serial dilutions of virus; the inoculum was then discarded, and the cells were overlaid with modified Eagle's medium containing 2% fetal calf serum and 1% type VII agarose (Sigma). Cells infected with Ba71v were fixed with 10% formaldehyde in PBS 7 days later. After fixation, overlays were removed and cell sheets were stained with 0.5% (wt/vol) crystal violet in a 20:80 methanol-water mixture for 30 min, rinsed with water, and allowed to dry prior to plaque counting.

Vero cells were grown on coverslips and then transfected with plasmids encoding human MxA or porcine Mx1, which were kindly provided by Daniel Desmecht (44), by using TransFast reagent (Promega). Cells were cultured for 48 h in DMEM-HEPES supplemented with 2% (vol/vol) fetal calf serum before infection with Ba71v.

Transmission electron microscopy was carried out as described previously (22). Briefly, cells were grown on Thermanox coverslips and fixed 16 h postinfection for 1 h in phosphate-buffered 2% (wt/vol) glutaraldehyde solution before being fixed for a further 2 h in aqueous 1% (wt/vol) osmium tetroxide at room temperature. Following a dehydration series in ethanol, coverslips were washed with propylene oxide before infiltration with 50% epoxy resin (Agar Scientific) in propylene oxide for 30 min. After a further 60 min in 100% epoxy resin, coverslips were placed (cell side up) in 18-mm-diameter plastic cups (TAAB Laboratories Equipment Ltd.) and covered in fresh epoxy resin, and the resin was polymerized overnight at 60°C. The coverslips were then removed, and the resin was incubated at 60°C for a further 24 h. Finally, 60-μm sections were cut and grid stained using EMStain (Leica Microsystems) before being imaged by an FEI Tecnai 12 transmission electron microscope with a TVIPS F214 digital camera.

Cell lysates in immunoprecipitation buffer (10 mM Tris [pH 7.5], 150 mM NaCl, 1% [vol/vol] IGEPAL CA-630, 10 mM iodoacetamide, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mg/ml leupeptin, 1 mg/ml pepstatin, 1 mg/ml chymostatin, 1 mg/ml antipain) were prepared. The protein content was determined using a bicinchoninic acid protein assay kit (Pierce Biotechnology). Lysates were resolved on sodium dodecyl sulfate-polyacrylamide gels and transferred onto Hybond-C nitrocellulose membranes (Amersham). Membrane blocking and secondary incubations were carried out in a solution of 5% (wt/vol) skim milk powder in Tris-buffered saline containing 0.2% Tween 20 at room temperature. Primary-antibody incubations were carried out in a solution of 5% (wt/vol) bovine serum albumin in Tris-buffered saline containing 0.2% Tween 20 at 4°C overnight. Secondary antibodies conjugated to horseradish peroxidase were purchased from Promega. Bands were revealed by using a SuperSignal West Pico chemiluminescent substrate (Pierce Biotechnology).

Stably transfected Vero cell lines expressing human MxA or MxB were used to investigate whether Mx gene products inhibited the replication of ASFV. VN36 (control), VA3 (MxA-expressing), or VB22 (MxB-expressing) cells (20) were infected with the ASFV strain Ba71v and assayed for their abilities to support plaque formation (Fig. 1A). Ba71v plaques were apparent in monolayers of VN36 and VB22 cells, but not in those of VA3 cells. This result showed that MxA, but not MxB, inhibited ASFV replication. At high multiplicities of infection (MOI) with ASFV, there was extensive cell death in the MxA-expressing population (data not shown), and close inspection of VA3 cells infected at lower MOI revealed a small number of rounded cells (Fig. 1B). The detection of a cytopathic effect (CPE) on MxA-expressing cells in the absence of plaque growth suggested that virus could infect but not produce infectious progeny in the presence of the protein. This explanation was tested by analyzing the replication of Ba71v in MxA-expressing and control cells. The two cell types were infected with 0.5 PFU per cell, and infection was allowed to progress until a complete CPE on control cells was observed. The determination of progeny virus titers revealed that titers of ASFV in VA3 cells were approximately 2 logs lower than those in control VN36 cells (Fig. 1C). These observations confirmed that MxA expression inhibited the replication of ASFV, a large double-stranded DNA virus. VA3 cells provide a useful model for analyzing the effect of Mx proteins on viral replication, but the level of MxA expression in VA3 cells is 50 to 100 times higher than that induced in human hepatoma cells by IFN (51). It was therefore important to compare MxA expression in VA3 cells to swine Mx1 expression in pig cells. PK-15 porcine endothelial cells were incubated with or without 200 U/ml recombinant porcine IFN-α for 24 h. Lysates were prepared from these cells, as well as VN36 and VA3 cells, and the levels of Mx proteins were determined by immunoblotting using samples with known protein concentrations (Fig. 1D). This analysis suggested that VA3 cells expressed MxA at approximately twice the level at which PK-15 cells expressed swine Mx1.

Mutations in the C terminus of MxA lead to proteins with altered antiviral specificities (60). Amino acid E645 of MxA is critical for activity against influenza A virus, but not vesicular stomatitis virus or bunyaviruses, and was therefore tested for its ability to interfere with ASFV replication. VA(E645R) cells, which stably express the E645R mutant form of MxA, were infected with Ba71v and assayed for their ability to support plaque formation (Fig. 1A) and virus growth (Fig. 1C). The plaque phenotype of cells expressing mutant MxA differed from those of control cells and cells expressing nonmutant MxA; however, VA(E645R) cells did permit plaque formation at levels similar to those seen in control cells. Furthermore, the replication of ASFV in VA(E645R) cells was similar to that in control cells (Fig. 1C). These observations indicated that amino acid E645 of MxA is necessary for activity against ASFV, like influenza A virus.

In order to determine the stage in the life cycle of ASFV inhibited by MxA, control and MxA-expressing cells were infected overnight with Ba71v and the steady-state levels of a number of ASFV proteins were determined by immunoblotting (Fig. 2A). The first panel of Fig. 2A shows the results of immunoblotting with antibody C18, which recognizes the major early protein p30. There was little difference in the amounts of p30 in ASFV-infected control and MxA-expressing cells. Immunoblotting with a panel of antibodies against late ASFV gene products, including the major capsid protein p73 (Fig. 2A, second panel), revealed that late protein expression was reduced in MxA-expressing cells. The steady-state levels of all of the late ASFV proteins tested showed significant reductions in MxA-expressing cells compared to those in control cells. Total protein synthesis was not affected because levels of γ-tubulin were not significantly altered. Immunoblotting with anti-MxA antibody showed that ASFV infection did not induce endogenous Mx genes. These results suggest that MxA does not inhibit ASFV infection or the early stages of viral replication, because levels of p30 were not affected, but rather that MxA blocks the expression of late genes that encode essential viral structural proteins.

In order to determine if there was an overall decrease in viral replication in MxA-expressing cells or if individual cells were replicating virus normally, the average cross-sectional area of viral factories was analyzed. The results of a two-sample t test evaluating the mean cross-sectional areas of virus factories measured in VN36 and VA3 cells (t = 7.43; P < 0.0005) showed that there was a twofold reduction in the size of viral factories in MxA-expressing cells compared to that in controls (Fig. 2B). This finding suggests that MxA expression restricts the growth of viral factories in cells in which late gene expression does occur.

The transcription of late ASFV genes is dependent on the synthesis of viral DNA (48), and the inhibition of viral DNA synthesis may explain the low levels of late proteins in MxA-expressing cells. To explore this possibility, cytoplasmic incorporation of tritiated thymidine into VN36 and VA3 cells that had been mock infected or infected for 16 h with Ba71v (33) was analyzed. The amounts of [³H]thymidine recovered from the cytoplasm of infected cells, both control and MxA-expressing cells, were greater than that recovered from the cytoplasm of mock-infected cells (Fig. 2C). This finding indicates that MxA does not prevent cytoplasmic incorporation of [³H]thymidine and therefore does not inhibit viral DNA synthesis.

MxA protein binds the nucleoprotein of La Crosse virus and sequesters the viral protein at perinuclear sites in the cell (31, 46). Since ASFV assembly occurs in perinuclear viral factories (23), the localization of MxA in cells replicating ASFV was investigated. Virus factories in control cells (Fig. 3A to D) and MxA-expressing cells (Fig. 3E to H) were compared by staining for the major capsid protein p73. A number of large perinuclear viral factories and associated intracellular virions were detected in ASFV-infected control cells, which did not stain with the anti-MxA antibody. Four MxA-expressing cells are shown in Fig. 3G; the two on the right side of the image are uninfected, as they lack labeling for viral protein (Fig. 3F) and do not have extranuclear viral DNA (Fig. 3E). The spotty granular MxA staining of the uninfected cells is similar to that described previously (31). Strikingly, in the two infected cells on the left side of the image, the MxA protein has lost some of its granular staining and is localized to an open ring structure next to the nucleus (Fig. 3F). The merged image clearly shows that MxA protein is surrounding the viral factory (Fig. 3H). To determine if antiviral activity was linked to the perinuclear movement of MxA to viral factories, the localization of MxA in ASFV-infected cells expressing the E645R mutant protein was examined. The labeling of ASFV-infected cells expressing E645R mutant MxA showed that the mutant protein was not recruited to viral factories (Fig. 3I to L). These observations suggested that there might be a causal link between the recruitment of MxA to virus factories and the ability of the protein to exert an antiviral effect.

The recruitment of MxA to viral factories was studied in greater detail using antibody TW34, which recognizes both the polypeptide precursor pp220 and mature structural protein p34 (23). Figure 4A to D show a group of MxA-expressing cells infected with Ba71v and stained with TW34 and anti-MxA antibodies. The cells highlighted by box 1 in Fig. 4D and shown in greater detail in Fig. 4E revealed that MxA protein was recruited not only to viral factories that contained viral structural proteins and DNA but also to factories that contained only DNA. These observations suggest that MxA recruitment to viral factories does not depend on the presence of late viral proteins. The cell in the center of box 2 was analyzed in greater detail to examine the relationship between MxA and the virus factory through the z axis of an infected cell. Sections 1.1 μm apart are presented in Fig. 4F to H and show that MxA is associated with the viral factory throughout the z plane of the cell.

The dependence of MxA recruitment on viral DNA synthesis was tested by infecting cells in the presence of the deoxycytidine analog cytosine arabinofuranoside (AraC), which inhibits cellular and viral DNA replication (48). Figure 5A to D show that viral DNA synthesis is necessary for the movement of MxA protein to perinuclear sites. Note the lack of colocalization between MxA and the diminished p30 staining in perinuclear areas indicated by arrows that show the beginnings of the viral factory (50). This result suggests that early events in factory formation do not induce MxA recruitment and that the movement of MxA to perinuclear sites is dependent on virus structural proteins or DNA synthesis.

The interaction between MxA and La Crosse virus nucleocapsid protein is not dependent on the microtubule or actin networks (46). To determine if MxA recruitment to ASFV factories was dependent on elements of the cytoskeleton, cells were infected with Ba71v and then treated with either nocodazole or cytochalasin D. Nocodazole and cytochalasin D depolymerize microtubules and actin, respectively, and so prevent microtubule- or actin-dependent intracellular transport. Virus factories were visualized using DAPI labeling (Fig. 5E and I), and anti-MxA labeling revealed that the MxA protein was recruited to factories in the presence of both nocodazole (Fig. 5G) and cytochalasin D (Fig. 5K). Anti-α-tubulin labeling showed that nocodazole had disrupted the microtubule network (Fig. 5F), and the use of a phalloidin conjugate showed that the actin network was disrupted in the presence of cytochalasin D (Fig. 5J). These data demonstrate that the microtubule and actin networks are not required for the recruitment of MxA to ASFV factories. This pattern is similar to that seen in the interaction between MxA and La Crosse virus nucleocapsid protein.

Mx proteins inhibit RNA viruses by interacting with the virus nucleocapsid proteins, and ASFV encodes three structural proteins with predicted DNA binding activity, pA104R, pE120R, and pK78R (9, 34, 40). Suitable reagents to detect pK78R expression are lacking, but the localization of pA104R and pE120R in MxA-expressing cells can be analyzed. Control and MxA-expressing cells infected with Ba71v were stained with antisera SB2 and SB11, which recognize pA104R and pE120R, respectively (Fig. 6). There was little difference in the localization of pE120R between VN36 and VA3 cells (Fig. 6B and F), although the number of pE120R-expressing cells among the MxA-positive cells compared to that among control cells was greatly diminished, as predicted by the immunblotting results (Fig. 2). The distribution of pA104R in control cells was quite different from that of the other viral structural proteins tested, as pA104R localized predominately to the nucleus but could also appear in the cytoplasm and the virus factory (Fig. 6J). Strikingly, in the MxA-expressing cells, there was an absence of nuclear pA104R staining and most of the viral protein encircled the viral factory (Fig. 6N), colocalizing with MxA (Fig. 6P). These observations suggest that MxA protein may interact with the ASFV DNA binding protein pA104R.

The ring structures formed by MxA in response to ASFV were investigated at the ultrastructural level to help provide an understanding of their relationship to the virus factory. MxA-expressing cells infected with ASFV were imaged with the electron microscope, and the results are displayed in Fig. 7. Virus factories seen in MxA-expressing cells (Fig. 7B to D) appeared to have fewer mature and immature virus particles and lower levels of viroplasm, viral membranes, and zipper structures than virus factories seen in VN36 cells infected with ASFV (Fig. 7A). This finding is consistent with the reduction in factory size seen by immunofluorescence (Fig. 2B). Also noticeable in MxA cells were unusual filaments that appeared to surround the virus factory (Fig. 7). Similar structures in VA3 cells infected with La Crosse virus have been identified previously (31). These filaments are shown in detail in comparison to a zipper structure (Fig. 7D) and a pair of immature virions lacking nucleoprotein cores (Fig. 7C).

The gene product of the porcine Mx1 gene shows 77.8% identity to the human MxA protein. To test if porcine Mx1 was recruited to ASFV factories, Vero cells were transfected with plasmids encoding human MxA or porcine Mx1 and infected 48 h later with Ba71v. Cells were stained with DAPI and TW34 antisera to detect virus factories and monoclonal antibody M143 to detect Mx protein expression. Human MxA in transiently transfected ASFV-infected cells was distributed to the cytoplasm and to a single distinct ring close to the nucleus (Fig. 8C). This pattern was similar to that seen previously in stably transfected VA3 cells (Fig. 3G and 6O). In the merged image (Fig. 8D), the MxA ring appeared to encircle the viral protein and DNA (Fig. 8A and B). Similar results were obtained for the cells transfected with porcine Mx1. M143 staining revealed a perinuclear ring structure (Fig. 8G) that surrounded the viral replication site (Fig. 8E, F, and H). This pattern showed that like human MxA, porcine Mx1 was recruited to viral factories after infection with ASFV.