The Vaccinia Virus O1 Protein Is Required for Sustained Activation of Extracellular Signal-Regulated Kinase 1/2 and Promotes Viral Virulence

Abstract

Sustained activation of the Raf/MEK/extracellular signal-regulated kinase (ERK) pathway in infected cells has been shown to be crucial for full replication efficiency of orthopoxviruses in cell culture. In infected cells, this pathway is mainly activated by the vaccinia virus growth factor (VGF), an epidermal growth factor (EGF)-like protein. We show here that chorioallantois vaccinia virus Ankara (CVA), but not modified vaccinia virus Ankara (MVA), induced sustained activation of extracellular signal-regulated kinase 1/2 (ERK1/2) in infected human 293 cells, although both viruses direct secretion of functional VGF. A CVA mutant lacking the O1L gene (CVA-ΔO1L) demonstrated that the O1 protein was required for sustained upregulation of the ERK1/2 pathway in 293 cells as well as in other mammalian cell lines. The highly conserved orthopoxvirus O1L gene encodes a predicted 78-kDa protein with a hitherto-unknown function. CVA-ΔO1L showed reduced plaque size and an attenuated cytopathic effect (CPE) in infected cell cultures and reduced virulence and spread from lungs to ovaries in intranasally infected BALB/c mice. Reinsertion of an intact O1L gene into MVA, which in its original form harbors a fragmented O1L open reading frame (ORF), restored ERK1/2 activation in 293 cells but did not increase replication and spread of MVA in human or other mammalian cell lines. Thus, the O1 protein was crucial for sustained ERK1/2 activation in CVA- and MVA-infected human cells, complementing the autocrine function of VGF, and enhanced virulence in vivo.

INTRODUCTION

Modulation of cellular signaling processes is a key factor in the hostile takeover of cells by viruses to promote efficient production and spread of viral progeny. The extracellular signal-regulated kinase (ERK) is the effector kinase of a central cellular signaling pathway regulating cell death and survival, proliferation, and differentiation and cellular migration and motility (48, 58, 66). It belongs to the class of mitogen-activated protein kinases (MAPKs), which also includes c-Jun NH2-terminal kinase (JNK) and the p38 MAP kinases. The MAPK pathways are activated by extracellular growth factors such as epidermal growth factor (EGF) and a number of other signals. The EGF receptor (EGFR) has tyrosine kinase activity and upon ligand binding transmits the signal to the cell interior by autophosphorylation. The signal is then further relayed via a number of intermediary factors to a three-tiered module consisting of the three kinases Raf, MEK1/2, and ERK1/2, which are consecutively activated by phosphorylation. ERK1 and ERK2 are highly similar kinase isoforms of 44- and 42-kDa molecular mass, respectively, and phosphorylate more than 160 known cytoplasmic and nuclear proteins (66), promoting cell proliferation, survival, migration, and motility.

Sustained activation of ERK1/2 during infection plays an important role in the multiplication of a number of RNA and DNA viruses that replicate in both the nucleus and the cytoplasm, including influenza virus A and B (29, 42), murine coronavirus (6), Borna disease virus (41), coxsackievirus B3 (30), BK virus (47), human cytomegalovirus (23), herpes simplex virus 2 (67), and the orthopoxvirus vaccinia virus (VACV) (1). For all of these virus species, inhibition of ERK1/2 activation during infection caused a decrease in replication. Poxviruses generally appear to benefit from activation of this pathway in their natural hosts, since poxviruses from most genera have been shown to express a secreted growth factor similar to mammalian EGF, resulting in upregulation of ERK1/2 activation (40, 57). The vaccinia virus growth factor (VGF) (4, 5) represents the best-studied example. Using a VGF deletion mutant of VACV-WR, it has been demonstrated that VGF is required for most of the ERK1/2 activation observed in VACV-infected cells (1, 43). In addition, VGF expression caused cell proliferation in VACV-infected tissue (5) and enhanced viral virulence (4). The advantage for poxviruses in directing the expression of such a factor was long seen as providing a metabolically active cellular environment supporting viral replication. More recently, it has been shown that, in addition, VGF acts together with the viral F1 protein to prevent premature apoptosis of VACV-infected cells (43).

It has recently been demonstrated that ERK1/2 activation can also govern the host range of poxviruses (61, 62). Modified vaccinia virus Ankara (MVA) is a severely replication-restricted orthopoxvirus (8, 15, 52), while its parent chorioallantois vaccinia virus Ankara (CVA) is replication competent and has a very broad in vitro host range (34, 35). The replication restriction of MVA occurs at the level of virus particle morphogenesis, thus permitting highly efficient viral gene expression in MVA-infected cells (46, 53). Together with its strong attenuation and excellent safety record in mammalian hosts, these properties warranted the development of MVA as a safe and efficient smallpox vaccine (26, 60) as well as for vector vaccine purposes (54). MVA was generated by subjecting the parental CVA to over 570 passages in chicken embryo fibroblasts (CEFs). Over the course of the passages, six major deletions of 2.7 to 6.5 kbp and a large number of smaller deletions, insertions, and point mutations occurred in the genome (34). We and others have recently demonstrated that, in contrast to general belief, introduction of the six major deletions was not sufficient to create a severely host-range-restricted VACV with MVA-like properties (13, 35). Host range restriction of MVA must therefore be governed by additional or completely independent mutations. We made use of MVA and the parental CVA to identify modulators of cellular signaling pathways putatively involved in the orthopoxvirus host range. To assess the role of selected intracellular signaling pathways in MVA host range restriction and VACV biology, we generated and screened a series of deletion mutants of CVA and found that the O1 protein was a positive regulator of the ERK1/2 pathway downstream of the EGFR, complementing the function of VGF by sustaining ERK1/2 activation during the course of infection. Deletion of the O1L gene in CVA caused only transient ERK1/2 activation after infection, reduced plaque size, and decreased virulence and spread in mice. Reintroduction of a functional O1L gene into MVA did not increase the capacity of MVA to replicate in human cells but restored ERK1/2 activation by MVA. In conclusion, the highly conserved poxviral O1 protein is a modulator of the Raf/MEK/ERK signaling pathway, enhancing spread and pathogenicity of replication-competent vaccinia virus.

MATERIALS AND METHODS

Cells and viruses.

All cell lines were obtained from ATCC or the European Collection of Cell Cultures and were cultivated in Dulbecco's modified Eagle medium (DMEM; Gibco/Invitrogen, Darmstadt, Germany) supplemented with 10% fetal calf serum (FCS; Pan Biotech, Aidenbach, Germany). Primary CEF cells were prepared from 11-day-old embryonated chicken eggs and cultured in VP-SFM (Gibco) for virus stock production or DMEM supplemented with 10% FCS for replication analysis. CVA and MVA used in this study were derived from bacterial artificial chromosomes (BACs) containing the respective wild-type genomes and have been characterized previously (35). The MVA wild-type strain (MVA) and MVA mutants were propagated and titrated using secondary CEF cells and the 50% tissue culture infective dose (TCID₅₀) method. The CVA wild-type strain (CVA) and CVA mutants were titrated by the TCID₅₀ method on CV-1 cells and propagated on Vero cells. Shope fibroma virus was obtained from ATCC (VR-364) and was propagated and titrated on rabbit cornea SIRC cells. All viruses used in animal experiments were purified twice through a 36% sucrose cushion.

BAC recombineering and reactivation of infectious virus.

Construction of the CVA- and MVA-BACs has been described previously (35). Briefly, the inserted BAC cassette contains miniF plasmid sequences derived from plasmid pMBO131 (38) for maintenance in Escherichia coli and a neomycin-phosphotransferase II-enhanced green fluorescent protein (neomycin-phosphotransferase II-EGFP) marker cassette. Transcription of the complete cassette is driven by a poxviral synthetic early/late promoter (9), and translation of the EGFP cistron from the bicistronic mRNA is achieved by an internal ribosomal entry site preceding the EGFP open reading frame (ORF). The BAC cassette was inserted between the MVA and CVA orthologues, respectively, of genes I3L and I4L. CVA-BAC was modified by allelic exchange in DH10B E. coli by the use of the λ Red system for homologous recombination. Two independent CVA-BAC constructs with a deletion of the O1L gene were generated. Since the 3′ end of the O1L ORF overlaps with the 5′ end of the E11L ORF, the deletions preserved at least 40 nucleotides of the 3′ end of O1L to maintain the E11L promoter and correct amino acid terminus. In the first construct (CVA-ΔO1L-K⁺), the O1L ORF of CVA was replaced with a kanamycin cassette from 40 bp upstream of the stop codon to 9 nucleotides upstream of the start codon. In the second construct (CVA-ΔO1L-K⁻), the O1L ORF from 157 bp upstream of the stop codon to 29 nucleotides upstream of the start codon was replaced with a counterselectable rpsL/neo cassette that was subsequently deleted by allelic exchange mutagenesis as described previously (35). This resulted in traceless deletion of the O1L ORF and of part of the predicted promoter. Removal of the rpsL/neo selection cassette was confirmed by sequencing of the respective regions flanking the insertion site. Experiments were carried out with either of the two CVA-ΔO1L variants. All effects of the O1L deletion, including in vitro replication behavior, in vivo virulence, ERK1/2 activation, cytopathic effect (CPE) phenotype, and plaque size reduction, were obtained with both the CVA-ΔO1L variant still containing the E. coli kanamycin cassette in place of the O1L gene and the CVA-ΔO1L variant with the traceless deletion of the O1L gene. Therefore, we did not specify the marker variants in the experiments shown.

For replacement of the fragmented MVA version of the O1L gene in the MVA genome by the intact CVA version of O1L, both MVA O1L fragments from 39 bp downstream of the full-length O1L start codon to 67 bp upstream of the full-length O1L stop codon were replaced with an rpsL/neo counterselection cassette. A PCR fragment comprising the start and stop codons of the CVA O1L ORF flanked by 30 and 46 bp, respectively, of MVA-specific nucleotides was produced. The flanking sequences were incorporated into 50- to 65-mer oligonucleotide primers that were used to amplify the CVA O1L gene. The PCR product was used to replace the counterselection cassette by allelic exchange, resulting in MVA-BAC+O1L. The complete coding region of MVA-BAC+O1L encompassing MVA ORFs 001 to 193 was directly sequenced by Eurofins MWG Operon (Ebersberg, Germany) using next-generation sequencing with GS FLX Titanium series chemistry. The resulting nucleotide sequence confirmed that no mutation in addition to the intended insertion of the CVA version of the O1L ORF had occurred. A point mutation in the 156-amino-acid (aa)-long MVA version of the C6L ORF was noted in both MVA+O1L and the BAC-derived MVA wild type. This mutation C-terminally shortened the C6L ORF of our BAC-derived MVA clones from 156 to 154 amino acids, with amino acid 154 changed from Phe to Ser. The MVA version of C6L has six additional amino acids at the C terminus compared to the 150-aa version of C6L from CVA and all other VACV C6L genes. The N-terminal 150 amino acids of C6L in CVA and MVA are completely identical. BAC-derived MVA with the 154-aa C6L version and MVA-BN (trademark of Bavarian Nordic) possessing the bona fide MVA version of C6L with 156 amino acids both showed the typical weak-to-absent ERK1/2 phosphorylation pattern and highly similar replication behavior characteristics in CEF cells and 4 mammalian cell lines (data not shown). This confirmed that the 2-aa-shorter C6L of the BAC-derived MVA did not influence the parameters determined in this study.

For reactivation of infectious virus, 10⁶ BHK-21 cells were transfected using 3 μg of BAC DNA and Fugene HD and infected 60 min later with Shope fibroma virus to provide the helper functions. Reactivated virus was isolated and helper virus was removed as previously described (35). The absence of bacterial insertion sequence (IS) elements from the genomes of all reactivated viruses was confirmed by PCR for E. coli DH10B IS elements 1, 2, 3, 4, 5, 10, 30, 150, and 186.

RT-qPCR.

RNA was isolated from 2.5 × 10⁶ infected cells per sample using an RNeasy Plus minikit (Qiagen, Hilden, Germany) according to the manufacturer's instructions, including a DNA removal step using genomic DNA (gDNA) eliminator columns (Qiagen). RNA was eluted from the RNeasy spin columns with 30 μl of RNase-free water, and the remaining viral and cellular DNA in the sample was digested with Turbo DNase (Ambion/Applied Biosystems, Darmstadt, Germany). The samples were incubated with the DNase for 30 min at 37°C, and the DNase was subsequently inactivated by addition of 1.24 μl of 0.5 M EDTA (pH 8.0) and heating to 75°C for 10 min. Reverse transcription was performed using an Omniscript reverse transcriptase (RT) kit (Qiagen) in the presence of RNase inhibitor (human; Sigma-Aldrich) at a final concentration of 10 U/μl for 90 min at 37°C using oligo(dT)_12–18 primers (Invitrogen). A 1-μl volume of RT reaction mixture was added to 20 μl of the total PCR mixture for real-time quantitative PCR (qPCR). Master qPCR mixtures and the following primer/TaqMan probe sets obtained from Applied Biosystems were used to set up the qPCRs in duplicate per sample: CVA-E3L_F (5′-TGTACAGCTCCGACGATATTCCT-3′), CVA-E3L_R (5′-GCGTCAGCCATAACATCAGCAT-3′), and CVA-E3L (5′-6-carboxyfluorescein [FAM]-ATCCGCCTCCGTTGTC-MGBNFQ-3′); CVA-C7L_F (5′-TTTAGATTCATTATACGCCCAGATTGGT-3′), CVA-C7L_R (5′-TCACCGCATAGTTGTTTGCAAATAC-3′), and CVA-C7L_M (5′-FAM-ACCTCGTCGATTTCC-MGBNFQ-3′); CVA-O1L_F (5′-CCATGTCATTGAGATCCACTATCATCAAA-3′), CVA-O1L_R (5′-GGATACCTTGCTATTTTTTCTGGCATT-3′), and CVA-O1L_M (5′-FAM-ATGTGCGAGAATATCC-MGBNFQ-3′); CVA-F17R_F (5′-CGTTTATGAGGACGGACATGCTA-3′), CVA-F17R_R (5′-AAAAGTCTAGAAGCTACATTATCGCGATT3′), and CVA-F17R (5′-FAM-CCGCGAACATATTTTG-MGBNFQ-3′); and human β-actin (assay no. Hs99999903_m1; Applied Biosystems). After 10 min of incubation at 50°C with uracil DNA glycosylase to remove carryover amplicon contamination and 10 min of incubation at 95°C, reaction mixtures were subjected to 40 cycles of incubation for 15 s at 95°C and 1 min at 60°C. qPCRs were performed on an ABI 7500 96-well real-time PCR machine (Applied Biosystems), and results were analyzed using ABI 7500 software version 2.0.1. Identical threshold values were set for all assays to allow direct comparison of the threshold cycle (C_T) values. Fold induction of viral gene expression was calculated relative to the undetectable signals in mock-infected cells, which were arbitrarily assigned a C_T value of 36, representing undetectable results, for calculation.

Immunoblot analysis of protein steady-state levels and phosphorylation status.

Cells were seeded the day before infection in 12-well tissue culture plates and infected when the cell monolayer reached confluence (∼90%). Completely confluent monolayers were excluded. For infection, 500 μl of the total of 1 ml of DMEM–10% FCS culture medium per well was aspirated and virus dilutions in DMEM were directly added at a volume of 100 μl into the culture supernatants (SN) and mixed by thorough swaying. At the indicated times after infection, cells were washed with ice-cold phosphate-buffered saline (PBS) containing P8340 protease inhibitor cocktail [104 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride, 80 μM aprotinin, 4 mM bestatin, 1.4 mM E-64, 2 mM leupeptin, 1.5 mM pepstatin A; Sigma-Aldrich, Munich, Germany] (1:100 dilution). Subsequently, cells were lysed for 10 min in ice-cold radioimmunoprecipitation assay (RIPA) buffer (150 mM NaCl, 1.0% Igepal CA-630, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate [SDS], 50 mM Tris, pH 8.0) containing the protease inhibitor cocktail described above (1%) and 1% each of phosphatase inhibitor cocktails P2850 and P5726 (Sigma-Aldrich) and lysates were centrifuged for 5 min to remove cell debris. Alternatively, cells were directly lysed in 1× Laemmli loading buffer (65 mM Tris-HCl [pH 6.8], 10% glycerol, 2% SDS, 0.1% bromophenol blue, β-mercaptoethanol [35 μl/ml]) for 10 min at room temperature and sonicated for 3 min. Lysates were stored at −20°C. For immunoblot analysis, lysates in 1× Laemmli loading buffer were thawed and then heated for 5 min at 95°C and proteins were separated on SDS-polyacrylamide gels (Miniprotean TGX; Bio-Rad, Munich, Germany) (10% to 12%) and transferred to nitrocellulose membranes (Immobilon-P; Millipore, Schwalbach, Germany) (pore size, 0.45 μm) by the use of a tank blot system (Bio-Rad). The membranes were blocked using 5% bovine serum albumin (BSA) (Carl Roth, Karlsruhe, Germany) in Tris-buffered saline (TBS; pH 7.5) containing 0.05% Tween 20 and 0.1% NaN₃. Membrane-immobilized proteins were reacted with the specific antibodies to phosphorylated and nonphosphorylated proteins described below. Primary antibodies were diluted in 5% BSA–TBS–Tween 20–0.1% NaN₃ and incubated overnight at 4°C. Membranes were extensively washed with TBS–0.05% Tween 20–0.1% NaN₃ between steps. Bound antibodies were detected by incubation with species-specific secondary horseradish peroxidase-conjugated antibodies diluted in 5% BSA–TBS–Tween 20–0.1% NaN₃ for 1 h at room temperature. Bands were visualized by enhanced chemiluminescence using SuperSignal West Pico (Thermo Scientific, Bonn, Germany) as the substrate. For stripping of bound antibody complexes, membranes were incubated in buffer containing 62.5 mM Tris-HCl (pH 6.8), 2% SDS, and 0.1 M β-mercaptoethanol (β-ME). After incubation for 30 min at 56°C, membranes were washed extensively, blocked, and then reprobed with primary and secondary antibodies.

Antibodies and reagents.

Antibodies with the following specificities were used in immunoblot analysis: ERK1/2 (dilution, 1:2,000), IκB (1:1,000), phospho-ERK (p-Thr 202/p-Tyr 204; 1:2,000), phospho-FAK (p-Tyr 925; 1:1,000), phospho-JNK (p-Thr 183/p-Tyr 185; 1:1,000) phospho-p38 (p-Thr 180/p-Tyr 182; 1:2,000), phospho-c-Raf (p-Ser-338; 1:1,000), and human EGFR (1:1,000) (all from Cell Signaling Technology, Danvers, MA) and anti-FLAG M2 (1:2,000) (Sigma-Aldrich). The polyclonal anti-B15 antibody was produced by immunization of rabbits with a peptide comprising amino acids 113 to 125 derived from CVA-B15. The polyclonal rabbit anti-B5 antibody was a generous gift of R. Drillien, Strasbourg, France.

The following inhibitors were used: U0126 (inhibitor of MEK1/2; Promega, Mannheim, Germany) (10 μM), SB203580 (p38 inhibitor; Promega) (10 μM), cycloheximide (inhibitor of protein translation; Calbiochem, Darmstadt, Germany) (100 μg/ml), cytosine arabinoside (inhibitor of DNA synthesis; Sigma-Aldrich) (40 μg/ml), LY294002 (inhibitor of phosphoinositide 3-kinase; Promega) (50 μM), cytochalasin D (inhibitor of actin polymerization; Sigma-Aldrich) (2 μM), and AG1478 (inhibitor of epidermal growth factor receptor kinase; Calbiochem) (5 μg/ml).

Human epidermal growth factor (hEGF; 10 μM) was purchased from PeproTech and used at a final concentration of 10 μM. Cell culture supernatants (SN) for EGFR stimulation were collected from infected 293 cells at different time points after infection. Cells were either mock infected or infected with MVAs or CVAs at a multiplicity of infection (MOI) of 10. Before using SN for stimulation of uninfected cells, all SN were filtered through 0.2-μM-pore-size filters to remove cell debris and vaccinia virus particles. SN were incubated with uninfected CV-1 cells for 24 h and analyzed by flow cytometry for EGFP expression to verify that the SN did not contain infectious virus.

Flow cytometry.

Infected and uninfected cell monolayers were washed with PBS, and single-cell suspensions were prepared by trypsinization (0.05% trypsin-EDTA; Gibco/Invitrogen). After being washed with PBS, cells were resuspended in ice-cold PBS–2% FCS containing 1% propidium iodide, which served to stain dead cells. Propidium iodide-negative cells were analyzed for EGFP expression by flow cytometry using a digital LSRII flow cytometer (Becton Dickinson) and FlowJo software (Tree Star).

Plaque and comet formation assay.

Cells were seeded the day before infection in 6-well tissue culture plates and then infected at 37°C with virus dilutions calculated to result in 10 to 50 single plaques per well. After 1 h of infection, cells were washed with cell culture medium without FCS and medium was changed to semisolid overlay medium (DMEM, 7% FCS, 0.5% methyl cellulose) or liquid cell culture medium. For U0126 and AG1478 treatments, the two reagents were added to virus inocula and to liquid and semisolid overlay media in the concentrations described above. Cells were cultured for another 2 to 3 days. For analysis of comet formation and plaque size, cells were stained with crystal violet for 15 min and then washed several times with water. Stained monolayers were scanned on an Epson Perfection V700 photo scanner. Plaque sizes were measured and plaque areas were calculated using the scanned monolayers and computer software developed at Bavarian Nordic. In addition, plaque sizes and morphologies were analyzed without fixation or staining of cell monolayers by fluorescence and phase-contrast microscopy using EGFP as the marker of infection.

Viral replication analysis.

For analysis of multicycle virus replication, confluent monolayers in 6-well culture plates were infected at an MOI of 0.05 in 500 μl of DMEM without FCS. After 60 min of adsorption at 37°C, cells were washed once with DMEM. The time point of subsequent addition of 2 ml of fresh DMEM–2% FCS was assigned a value of t = 0 h. Resting 293 cells were infected 4 days after seeding, in contrast to the 1-day-old 293 monolayers that were just reaching confluence (active cells) and that were used as a standard for single- and multicycle replication analysis. For analysis of single-cycle virus replication, confluent monolayers in 6-well culture plates were infected at an MOI of 10 in 500 μl of DMEM without FCS. After 60 min of adsorption at 37°C, cells were washed once with DMEM. Subsequent addition of 2 ml of fresh DMEM–2% FCS was assigned a value of t = 0 h. Cells plus supernatant were harvested at the indicated time points, freeze-thawed, and sonicated. For the 0 h time point, cells were kept on ice after the 60-min adsorption period, washed with ice-cold DMEM, supplemented with 2 ml of ice-cold DMEM–2% FCS, and immediately frozen. MVA titers were determined on CEF cells and CVA titers were determined on CV-1 cells according to the TCID₅₀ method as described previously (51). Briefly, serial dilutions of virus suspensions were plated on CEF cell monolayers grown in 96-well plates as replicates of 8. Cells were fixed with methanol:acetone at a 50/50 (vol/vol) ratio at 5 days postinfection (p.i.), and foci of infected cells were visualized by immunostaining. Fixed and permeabilized monolayers were incubated for 30 min with rabbit polyclonal vaccinia virus antibody (Quartett Immunodiagnostika, Berlin, Germany) diluted 1:1,000 with PBS–3% FCS followed by incubation with horseradish peroxidase-conjugated polyclonal goat anti-rabbit IgG antibody (Promega) diluted 1:1,000 in PBS–3% FCS for 30 min. After being washed, cells were incubated with TMB:PBS substrate solution (Seramun Diagnostica, Heidesee, Germany) for 15 min. Infected wells were identified by purple staining of cells, and the infectious titer was calculated using the method of Kärber (24).

Mouse infection experiments.

Female BALB/c mice aged 6 to 8 weeks were purchased from Harlan Winkelmann, Germany. Mice were anesthetized by ketamine-xylazine injection prior to intranasal infection with 3 × 10⁵, 1 × 10⁶, and 1 × 10⁷ TCID₅₀ of CVA and CVA mutants diluted in PBS to a final volume of 50 μl per mouse. Animals were weighed and inspected daily for 2 weeks, and the signs of illness were scored on an arbitrary scale from 0 to 4 as follows: score 0 = healthy; score 1 = slightly sick, with moderately hunched back and ruffled fur and normal mobility and activity levels; score 2 = sick, with clearly hunched back and ruffled fur, reduced mobility and activity levels, and moderate respiratory distress; score 3 = very sick, with strongly hunched posture and ruffled fur, strongly reduced mobility and activity, hedgehog-like gait with coordination problems, and significant respiratory distress; and score 4 = moribund.

All animal experiments were approved according to the regulations of the German animal welfare law by the responsible authority, the government of Upper Bavaria (Regierung von Oberbayern), and were carried out in accordance with the regulations set forth by the responsible authority and with the guidelines for animal experiments of Bavarian Nordic GmbH.

RESULTS

Virus-specific induction of sustained ERK activation in human cells.

We analyzed various cellular signaling pathways involved in the poxviral host range and found that CVA induced sustained ERK1/2 phosphorylation in human 293 cells (Fig. 1A) at levels similar to the previously described ERK1/2 activation by VACV-WR (1). In contrast, MVA only weakly induced ERK1/2 activation (Fig. 1A) despite similar levels of infection efficiency and EGFP reporter gene expression (data not shown). Activation of ERK1/2 by CVA infection was strong at 2 to 4 h p.i. and was maintained until 24 h after infection (Fig. 1B). A decrease was observed at late times of infection from 8 h until 24 h p.i. (Fig. 1B), when the cells progressively deteriorated. In contrast, ERK1/2 phosphorylation was only transiently and moderately increased in MVA-infected 293 cells at around 2 h p.i. and returned to very low or baseline levels from 4 h p.i. onward (Fig. 1A and B). Treatment of CVA-infected 293 cells with cycloheximide inhibiting mRNA translation showed that activation of ERK1/2 at 4 h p.i. was dependent on de novo protein biosynthesis (Fig. 1C). Inhibitors of the p38 (SB203580) and the PI(3)K/Akt (LY294002) pathways did not significantly affect ERK1/2 activation during CVA infection (Fig. 1C). Viral intermediate or late gene expression was not required for ERK1/2 activation, as demonstrated by the unchanged ERK1/2 phosphorylation of infected 293 cells treated with AraC, which inhibits viral DNA replication and hence postreplicative viral gene expression (Fig. 1C). Treatment with cytochalasin D disrupting actin microfilaments and partly blocking viral entry largely abrogated ERK1/2 activation and infection (Fig. 1C), demonstrating that virus binding to the cell per se is not sufficient to trigger sustained ERK1/2 activation. Hence, early viral gene expression was necessary and sufficient to trigger sustained ERK1/2 activation in infected 293 cells, whereas virus binding and core entry were not sufficient.

Fig 1.

CVA and VACV-WR, but not MVA, induce sustained ERK1/2 activation in human cells. (A and B) 293 cells were mock infected or infected with CVA, MVA, or VACV-WR at an MOI of 10. Cells were lysed at the indicated time points after infection, and the relative amounts of phosphorylated ERK1/2 (p-ERK1/2) were determined by immunoblot analysis. Immunodetection of tubulin served as loading control. (C) 293 cells were treated with dimethyl sulfoxide (DMSO) as carrier control or the following specific inhibitors as indicated: U0126 (inhibitor of MEK1/2), SB203580 (SB; p38 inhibitor), cycloheximide (Cyclo; inhibitor of protein translation), cytosine arabinoside (AraC; inhibitor of DNA synthesis), LY294002 (LY; inhibitor of phosphoinositide 3-kinase); cytochalasin D (CytoD; inhibitor of actin polymerization). After 30 min, cells were either mock infected or CVA infected (MOI of 10), and infection was stopped 4 h later by adding lysis buffer to cells. Phosphorylation of ERK1/2 (p-ERK1/2) was then analyzed by immunoblotting; the same membrane was stripped and reprobed with antibody to ERK protein to detect total ERK1/2 protein as loading control.

Role of VGF in sustained ERK activation.

It has been reported that VGF is capable of inducing sustained ERK1/2 activation, in contrast to EGF, which activated ERK1/2 only transiently in murine cells (57). Both MVA and CVA encode full-length VGF (C11R gene), but the MVA version differs from that of CVA by four amino acid exchanges. Filtered supernatants from both MVA- and CVA-infected 293 cells induced ERK1/2 activation in uninfected 293 cells within 10 min after treatment, and the responsible factor appeared to be secreted in comparable amounts (Fig. 2A). ERK1/2 activation in uninfected 293 cells by supernatants from CVA- and MVA-infected 293 monolayers was completely abolished by treatment with the EGFR inhibitor tyrphostin AG1478 (Fig. 2B). ERK1/2 phosphorylating activity in supernatants of infected cells increased strongly over the first 4 h of infection (Fig. 2C) during the viral eclipse phase, which argues against a role of virus particles in this activation. To exclude a role of residual infectious virus in the very early ERK1/2 activation induced within 15 min by supernatants from virus-infected cells (Fig. 2A to C), all supernatants used in this study were analyzed for virus contamination. The supernatants were added to CV-1 cells, and the percentage of EGFP-expressing CV-1 cells was determined as a readout for infection. The supernatant used to induce ERK1/2 activation contained an average of 10³ TCID₅₀ and a maximum of 4 × 10³ TCID₅₀ of infectious virus. Infection of 293 cells with graded doses of CVA showed that even 10⁸ TCID₅₀ was not sufficient to activate ERK1/2 within 15 min (Fig. 2D). In addition, 10³ TCID₅₀ of CVA did not induce ERK1/2 activation after 4 h of infection, whereas the high dose of virus did (Fig. 2D). Thus, the residual infectious virus that had been present in some of the filtered supernatants from CVA- and MVA-infected cells was clearly not responsible for the ERK1/2 activation induced by these supernatants. We therefore concluded that the factor leading to ERK1/2 activation in the supernatants was indeed VGF and that CVA and MVA directed the secretion of similar amounts of functional VGF. ERK1/2 activation was detectable in the cells treated with culture supernatants harvested from 2 h p.i. onwards (Fig. 2C). This suggested that the onset of sustained ERK1/2 activation in CVA-infected 293 cells at between 1 and 2 h p.i. (Fig. 1B and 3A) was most likely determined by accumulation of sufficient amounts of VGF in the extracellular environment.

Fig 2.

Supernatants from CVA- and MVA-infected cells induce comparable ERK1/2 activation levels. For all experiments, supernatants (SN) were collected from 293 cells that had been either mock infected or infected with MVA or CVA at an MOI of 10, as indicated. Before using SN for stimulation of uninfected cells, all SN were filtered through 0.2-μM-pore-size filters to remove cell debris and infectious vaccinia virus particles. After stimulation, cells were directly lysed in 1× Laemmli loading buffer and ERK1/2 phosphorylation (p-ERK1/2) and tubulin expression were analyzed by immunoblotting. (A) SN were collected 4 h after infection and used to restimulate uninfected cells in dilutions as indicated for 10 min. (B) Cells were pretreated with an inhibitor of the EGFR tyrosine kinase AG1478 (+) or a carrier control (−) for 30 min. Subsequently, SN from cells infected for 4 h and diluted 1:10 or human epidermal growth factor (EGF) were used to stimulate cells for 10 min before lysis. (C) SN was collected 4 h after mock infection or 1 h, 2 h, and 4 h after infection with CVA and used to restimulate uninfected cells for 10 min (1:25 dilution). (D) As a control for the impact of residual infectious virus in SN, 293 cells were directly infected with 10³ to 10⁸ TCID₅₀ of CVA for 15 min and with 10³ and 10⁸ TCID₅₀ of CVA for 4 h. For comparison, uninfected 293 cells were incubated in parallel for 15 min or 4 h with SN from CVA-infected cells. (E) For analysis of the stimulation kinetics of uninfected cells, SN from cells infected with MVA or CVA for 4 h (1:20 dilution), EGF, and DMSO (as carrier control) were used. Cells were lysed after 10 min, 1 h, or 4 h of stimulation.

Fig 3.

Deletion of the O1L gene from CVA leads to lack of sustained ERK1/2 activation after virus infection. (A) 293 cells were infected at an MOI of 10 with CVA or CVA-ΔO1L or mock infected for various time periods as indicated. Cells were lysed, and phosphorylation of ERK1/2 (p-ERK1/2) and expression of total ERK1/2 and tubulin as loading controls was analyzed by immunoblotting. (B) 293 cells were mock infected or infected with MVA, CVA, or CVA-ΔO1L at an MOI of 10. After 2 h, 4 h, or 8 h of infection, cells were lysed and steady-state levels of IκBα and phosphorylation of ERK1/2 were analyzed by immunoblotting. (C) Supernatants (SN) from 293 cells that had been either mock infected or infected with CVA or CVA-ΔO1L were collected 4 h after infection and filtered through 0.2-μM-pore-size filters to remove cell debris and vaccinia virus particles. SN and human epidermal growth factor (hEGF) were used to stimulate uninfected 293 cells for 10 min, 1 h, or 4 h. After stimulation, cells were lysed in loading buffer and ERK1/2 phosphorylation and tubulin expression were analyzed by immunoblotting. (D) 293 cells were pretreated with inhibitors of MEK1/2 kinase (U0126) or EGFR tyrosine kinase (AG1478) or with a carrier control (DMSO) for 30 min. Subsequently, cells were either mock infected or infected with CVA or CVA-ΔO1L for 4 h. Cells lysates were analyzed for ERK1/2 phosphorylation and expression of EGFR and tubulin by immunoblotting. (E) 293 cells were CVA or CVA-ΔO1L infected at an MOI of 10 or mock infected for 1 h, 2 h, or 4 h. Cells were then lysed, and phosphorylation of FAK, Raf, JNK, ERK1/2, and p38 was analyzed by immunoblotting. Subsequently, membranes were stripped and total expression of ERK1/2 and tubulin was determined to control for equal protein loading.

VGF from CVA and MVA activated ERK1/2 very early (within 10 min of treatment of uninfected cells) and with somewhat lower efficacy than EGF (Fig. 2E). At later time points, ERK1/2 activation by VGF decreased to low levels similar to those seen after treatment with EGF, indicating that VGF alone did not lead to clearly more sustained ERK1/2 activation than that seen with EGF in 293 cells (Fig. 2E). Therefore, the lack of sustained ERK1/2 phosphorylation induced by VGF treatment or MVA infection of 293 cells compared to CVA infection suggested the involvement of a second factor modulating sustained ERK1/2 activation that was expressed by CVA but not by MVA.

Identification of a CVA gene product modulating ERK1/2 activation.

To identify the postulated viral factor required for sustained ERK activation, we screened a collection of 13 single and multigene deletion mutants of CVA, encompassing a total of 59 ORFs, for deficiency in sustained ERK1/2 phosphorylation in 293 cells. Only one CVA mutant that lacked the complete O1L ORF (CVA-ΔO1L) showed a clear deficiency in sustained ERK1/2 activation at 4 h p.i. (data not shown). In stark contrast to CVA, CVA-ΔO1L did not stimulate sustained ERK1/2 activation at 2 to 8 h p.i. in 293 cells (Fig. 3A). At most, CVA-ΔO1L induced very weak and transient ERK1/2 activation between 2 and 8 h p.i. in some experiments (Fig. 3A and B). These results indicated that the O1 protein was necessary for sustained ERK1/2 activation. The requirement for the O1 protein for sustained ERK1/2 activation was confirmed with the nontransformed human lung cell line MRC-5 and the monkey cell lines CV-1 and Vero (see Fig. 8B; also data not shown).

Fig 8.

O1 expression in MVA-infected 293 cells causes sustained ERK1/2 activation. (A) 293 cell monolayers in 6-well plates were infected with MVA+O1L at an MOI of 10 for the indicated times in either the absence or presence of AraC. AraC was used at a final concentration of 40 μg/ml. Total RNA from infected cells was isolated and used for RT-qPCR analysis of the relative amounts of viral transcripts from genes O1L and F17R. Data represent the fold increase of viral transcripts compared to mock-infected cells (6 h). (B) 293, Vero, and CV-1 cells were infected with MVA, MVA+O1L, CVA, and CVA-ΔO1L at an MOI of 10 or were mock infected. After 4 h of infection, cells were lysed and phosphorylation of ERK1/2 (p-ERK1/2) was analyzed by immunoblotting. (C) 293 cells were infected with MVA and MVA+ΔO1L at an MOI of 10; mock infection was included as a control. At the indicated time points, cells were lysed and phosphorylation of ERK1/2 and expression of FLAG-tagged O1 protein and tubulin were analyzed by immunoblotting. (D) 293 cells were infected with MVA and MVA+ΔO1L at an MOI of 10 or mock infected for the indicated time periods, and levels of IκBα as well as phosphorylation of ERK1/2 were analyzed by immunoblotting (IB). (E and F) For stimulation of uninfected cells, SN from 293 cells were collected after 4 h of infection with MVA, MVA+ΔO1L, CVA, or CVA-ΔO1L at an MOI of 10 or mock infection. All SN were filtered through 0.2-μM-pore-size filters to remove cell debris and vaccinia virus particles. For stimulation, the indicated volumes of SN were added to uninfected 293 cells grown in 24-well plates in 300 μl of DMEM medium. As a control, cells were treated with 50 μl of DMEM or with 50 μl of SN from mock-infected cells. After stimulation, cells were directly lysed in 1× Laemmli loading buffer and phosphorylation of ERK1/2 (p-ERK1/2) and expression of tubulin were analyzed by immunoblotting.

A link between the ERK1/2 and NF-κB activation in VACV-infected cells has been proposed (18, 33). Thus, we determined NF-κB activation by following the degradation of the NF-κB inhibitory subunit IκBα (Fig. 3B). During the time period of 2 to 8 h p.i., neither CVA nor CVA-ΔO1L induced a decrease in IκBα levels (Fig. 3B). In contrast, MVA induced a long-lasting decrease in IκBα levels of infected 293 cells (Fig. 3B) that was consistent with previous reports of NF-κB activation by MVA (39, 49).

Reduced ERK1/2 activation by CVA-ΔO1L was not due to impaired VGF secretion, since supernatants of 293 cells infected for 4 h with either CVA or CVA-ΔO1L induced similar kinetics and extents of ERK1/2 activation when applied to uninfected 293 cells (Fig. 3C; see also Fig. 8F). Infection of 293 cells in the presence of the inhibitor U0126, which selectively blocks the MEK1/2 kinase pair responsible for phosphorylation of ERK1/2, demonstrated that sustained ERK1/2 phosphorylation induced by CVA and CVA-ΔO1L was dependent on MEK1/2 (Fig. 3D). This argued against a kinase activity of O1 capable of directly phosphorylating ERK1/2. Additionally, the EGFR inhibitor AG1478 blocked activation of ERK1/2 by CVA and residual activation by CVA-ΔO1L (Fig. 3D). Thus, sustained ERK1/2 activation in CVA infection is mediated by continuing EGFR activity. Upon infection of 293 cells with CVA and CVA-ΔO1L, steady-state EGFR levels were comparably downregulated (Fig. 3D). EGFR downregulation was not affected by treatment with the MEK1/2 inhibitor U0126 but was blocked by treatment with AG1478, which inhibits EGFR activation (Fig. 3D). Notably, neither U0126 treatment nor AG1478 treatment inhibited infection of 293 cells by CVA (data not shown), indicating that CVA was not dependent on EGFR activation for entry into 293 cells. These findings suggest that O1 targets molecules downstream of the EGFR to regulate ERK1/2 activation.

Activation of the ERK1/2-phosphorylating kinase MEK1/2 is mediated by Raf kinases, in particular by c-Raf, which itself is activated by phosphorylation at serine residue 338 (S338). CVA-infected 293 cells showed enhanced c-Raf phosphorylation, whereas in CVA-ΔO1L-infected cells, c-Raf was phosphorylated only marginally (Fig. 3E). This suggests that the target of the O1 protein is located upstream of c-Raf. Another signaling transducer activated by integrin and growth factor receptors is focal adhesion kinase (FAK). FAK activation is linked to cytoskeleton rearrangement and cell migration. FAK was clearly activated by CVA infection, but this activation did not appear to be regulated by O1 (Fig. 3E). Alternative MAPK pathways such as the JNK and the p38 pathway were also not significantly affected by O1 (Fig. 3E). Thus, while O1 increased ERK1/2 activation, other signaling pathways regulated by the EGFR remained largely unaffected, suggesting a unique role of the O1 protein in regulation of the Raf/MEK/ERK pathway.

The O1L ORF is preceded by a predicted early promoter sequence, and the early gene termination signal TTTTTNT is present 3 nucleotides downstream of the stop codon. O1L mRNA first became detectable by RT-qPCR between 1 and 2 h p.i., and its level further increased over the next 4 to 6 h of infection (Fig. 4A). The O1L gene was efficiently transcribed in the presence of AraC, whereas transcription of the F17R gene was almost completely suppressed, confirming that O1L is an early gene (Fig. 4A). Robust early and sustained late transcription of the O1L gene is consistent with the observed kinetics of O1-dependent increase in ERK1/2 phosphorylation, which first became detectable around 2 h p.i. and persisted during the course of infection (Fig. 1B). The relative expression levels of the immediate-early E3L gene, the early C7L gene, and the late F17R gene generally showed a certain degree of interexperiment variation (data not shown) but were not grossly altered in CVA-ΔO1L-infected 293 cells compared to CVA results (Fig. 4B). The steady-state levels of the viral B15 and B5 proteins were also indistinguishable in CVA- and CVA-ΔO1L-infected cells (data not shown), confirming that viral gene expression was unaffected by O1L deletion.

Fig 4.

O1L is an early gene, and gene transcription by CVA is independent of O1L. (A) 293 cell monolayers in 6-well plates were infected with CVA at an MOI of 10 for the indicated times in the absence or presence of AraC. AraC was used at a final concentration of 40 μg/ml. (B) 293 cell monolayers in 6-well plates were infected with CVA and CVA-ΔO1L at an MOI of 10 for the indicated times. Total RNA from cells infected as described for panels A and B was isolated and used for RT-qPCR quantification of the relative amounts of viral transcripts from genes O1L, E3L, C7L, and F17R. Fold increases of viral transcripts compared to mock-infected cell results at 6 h are shown. −RT = reverse transcription reaction without reverse transcriptase enzyme.

Spread of CVA-ΔO1L in cell culture.

To analyze the effect of the O1L deletion on viral spread, plaque assays were performed using various cell types. CVA-ΔO1L showed a clear reduction in plaque sizes on 293 cell monolayers (Fig. 5A). The area of CVA-ΔO1L plaques was decreased by approximately 45% compared to CVA plaques (Fig. 5C; P < 0.001 by Student's t test). A reduced plaque size of CVA-ΔO1L was also observed in the human cell lines HeLa and 143B and in monkey CV-1 cells (data not shown). In the presence of the U0126 inhibitor, the size of CVA plaques was reduced by approximately 50%, which was similar to the plaque size reduction observed for CVA-ΔO1L (Fig. 5B and C). Interestingly, in the presence of U0126, CVA-ΔO1L plaques were still ∼20% smaller than those produced by CVA (Fig. 5B and C), and this was also highly statistically significant (P < 0.001). This suggests that the reduced plaque size of CVA-ΔO1L on untreated monolayers resulted from reduced ERK1/2 activation and an additional effect of the O1L deletion that was ERK1/2 independent. The EGFR inhibitor AG1478 reduced CVA plaque size more efficiently than U0126, demonstrating that additional signaling pathways triggered by the EGFR were also involved in efficient spread of CVA and CVA-ΔO1L (Fig. 5B). Importantly, AG1478 did not prevent infection of 293 cells per se (data not shown).

Fig 5.

O1 contributes to efficient viral plaque formation. (A) Appropriate dilutions of CVA and CVA-ΔO1L viral stocks to obtain discernible plaques were used to infect 293 cells. Monolayers were incubated with liquid overlay medium and analyzed by fluorescence and phase-contrast microscopy at days 3 and 4 p.i. Expression of EGFP from the BAC cassette contained in the genomes of all viruses was utilized as a marker of virus-infected cells. (B) 293 cells were pretreated with inhibitors of MEK1/2 kinase (U0126) or EGFR tyrosine kinase (AG1478) or carrier control (DMSO) for 30 min before infection. Cells were infected with CVA or CVA-ΔO1L in the presence of inhibitors at appropriate virus dilutions to obtain discernible plaques. Medium was changed to semisolid overlay medium containing inhibitors at 1 h after infection. Overlay medium containing inhibitor or carrier control (DMSO) was renewed at 24 h p.i. At day 3, viral plaques were analyzed by fluorescence microscopy. Contrast and brightness of the microphotographs were adjusted using Microsoft Powerpoint 2007. (C) Sizes of CVA or CVA-ΔO1L plaques in the presence of DMSO (carrier control) or U0126 in semisolid overlay medium were quantified in crystal violet-stained 293 cell monolayers 6 days after infection. Plaque areas were calculated using scanned monolayers and computer software developed at Bavarian Nordic. *** = P < 0.001 by Student's t test. (D) The release of extracellular enveloped virions was analyzed by a comet assay. Crystal violet staining was done 4 days after infection of 293 cell monolayers in liquid overlay medium with CVA or CVA-ΔO1L at appropriate dilutions to obtain clearly separate plaques and discernible comets.

The morphology of CVA plaques also differed from that of plaques formed by CVA-ΔO1L. Plaques formed by CVA showed a zone of a broken cell monolayer at the edges of the plaque, whereas there was still a continuous monolayer at the edges of the plaques formed by CVA-ΔO1L (Fig. 5A). This phenomenon was most apparent in CV-1 cells (data not shown). There was less cytopathic effect at the edges of CVA-ΔO1L plaques, although the edges were also clearly infected, as evidenced by EGFP expression in these zones (Fig. 5A). Thus, a lack of O1 expression altered the morphology of infected cells, which might at least in part reflect cytoskeleton rearrangements due to reduced ERK1/2 activation.

As with CVA, CVA-ΔO1L did produce comets on 293 cell monolayers, indicating that CVA-ΔO1L had no general defect in release of extracellular enveloped virions (Fig. 5D). In summary, CVA-ΔO1L caused an altered CPE and formed smaller plaques due at least in part to the lack of sustained ERK1/2 activation. In addition, lack of O1 marginally decreased activation of EGFR-triggered pathways apart from ERK1/2 that also contributed to efficient plaque formation.

Replication of CVA-ΔO1L in vitro.

The replication properties of CVA-ΔO1L in comparison to the parental CVA were examined in 293 cells. Single-cycle replication kinetics of CVA and CVA-ΔO1L were very similar, with a slight transient disadvantage for CVA-ΔO1L (Fig. 6A). Multicycle replication curves of CVA and CVA-ΔO1L were also similar except for a slight but reproducible decrease of CVA-ΔO1L titers at around 24 h p.i. (Fig. 6A). Since lower ERK1/2 activation might affect CVA-ΔO1L replication more significantly in resting cells than in active cells, we also analyzed viral yields from resting 293 cells and from the respective supernatants. Titers of cell-associated virus from resting 293 cells were also slightly reduced, but the difference was not more pronounced in this setting (Fig. 6B). Yields of extracellular CVA-ΔO1L virus released into the culture supernatant of resting 293 cells (Fig. 6B) as well as active 293 cells (data not shown) were significantly lower at 24 p.i. compared to CVA yields. The slight reduction in viral yields in the multicycle experiments shown in Fig. 6A and B corresponded to the reduced size of CVA-ΔO1L foci at 24 h after infection at a low multiplicity in both active 293 cell monolayers (Fig. 6C) and resting 293 cell monolayers (data not shown). Since reductions in CVA-ΔO1L yields were small and were detectable in both multicycle and single-cycle growth curves, we were not able to conclusively distinguish whether slightly reduced replicative capacity or impaired cell-to-cell spread or both accounted for the smaller plaque size of CVA-ΔO1L.

Fig 6.

Reduced viral yields of CVA-ΔO1L in cell culture and reduced release of extracellular CVA-ΔO1L virus. (A) Active 293 cells in 6-well plates were infected in triplicate with the indicated viruses at an MOI of 10 (single-cycle analysis; left) or at an MOI of 0.05 (multicycle analysis; right). Combined cells and supernatants from active 293 cells were analyzed. (B) Resting 293 cells were infected in triplicate with the indicated viruses at an MOI of 0.05 (multicycle analysis). Cells and supernatants of resting 293 cell monolayers were analyzed separately. In panels A and B, viral output at the indicated times is plotted; each data point represents results from three independent wells. * = P < 0.02 by Student's t test for combined data from two independent experiments; ** = P < 0.01 by Student's t test. (C) Fluorescence microscopy analysis of active 293 cells at 24 h after infection at an MOI of 0.05. Virus titers obtained at 24 h p.i. from the 293 monolayers depicted here are shown in panel A (right). Infected cells show green fluorescence due to expression of EGFP encoded in the BAC cassette present in all recombinant viruses.

Virulence of CVA-ΔO1L in BALB/c mice.

To determine the contribution of the O1L gene to the virulence of CVA in vivo, BALB/c mice were intranasally infected with graded doses of CVA and CVA-ΔO1L. Reduced weight loss and disease symptoms in CVA-ΔO1L-infected mice were most pronounced at the sublethal dose of 1 × 10⁶ TCID₅₀/mouse (Fig. 7A and B and data not shown). The differences in weight losses and disease scores were statistically significant (P < 0.05) from day 3 to 11 p.i. at this dose. The differences in disease symptoms were most distinct in the recovery phase, indicating a faster recovery of mice from CVA-ΔO1L infection. Mice intranasally infected with 1 × 10⁷ TCID₅₀ of CVA or CVA-ΔO1L showed similar weight losses (Fig. 7C). CVA-ΔO1L-infected mice presented with a slightly later onset (P < 0.05) and slightly less severe clinical symptoms at the peak of disease (P < 0.01) compared to CVA-infected mice (Fig. 7D). Importantly, all mice infected with CVA-ΔO1L survived an infective dose of 1 × 10⁷ TCID₅₀ whereas, with the exception of one animal, all animals infected with that dose of CVA died (Fig. 7E). Thus, CVA-ΔO1L was clearly attenuated in BALB/c mice.

Fig 7.

Reduced virulence and spread of CVA-ΔO1L in BALB/c mice. Groups of five 6- to 8-week-old female BALB/c mice were intranasally infected with 1 × 10⁶ TCID₅₀ (A and B; 5 mice) or 1 × 10⁷ TCID₅₀ (C to E; 10 mice) of the indicated viruses in 50 μl of PBS, and animals were individually weighed and inspected daily. Body weight data are expressed as percentages of mean weights ± standard errors of the means (SEM) of mice in the respective group compared to the initial mean weight at day 0. Mean disease scores were determined at the indicated days according to the above-described scoring system; scores ranged from 0 to 4. (F) Groups of five 6- to 8-week-old BALB/c wild-type mice were infected intranasally with a dose of 1 × 10⁷ TCID₅₀ of CVA or mutant CVA-ΔO1L. Ovaries and lungs of infected animals were recovered at days 4 and 6 p.i. Organs were homogenized, and viral titers were determined by a standard TCID₅₀ assay using CV-1 cells. Total viral titers in ovaries of mice taken at days 4 and 6 after infection and lungs taken at day 6 after infection are indicated. Data show the averages of results determined in two independent experiments performed with a total of 10 mice per time point and organ. * = P < 0.05; ** = P < 0.01.

Six days after intranasal inoculation of mice with a dose of 1 × 10⁷ TCID₅₀ of CVA or CVA-ΔO1L, viral titers in the lungs were similar (Fig. 7F). However, spread of CVA-ΔO1L from the lungs to the ovaries differed from CVA spread. At 4 and 6 days p.i., viral titers in ovaries from CVA-ΔO1L-infected mice were reduced 10- to 100-fold on average compared to the levels in ovaries from CVA-infected mice (Fig. 7F). Despite considerable individual variation, reduced spread of CVA-ΔO1L to the ovaries was statistically significant at day 6 p.i. (Fig. 7F). Thus, we concluded that O1 is a virulence factor.

Construction of an MVA with a restored O1L gene.

The MVA genome contains six major deletions of 24.7 kbp in total and a number of smaller deletions leading to fragmentation or truncation of a number of additional genes. Among the latter are genes F5L, F11L, and O1L, which are the only fragmented genes in the central region of the MVA genome (2). We replaced the two fragments of the O1L ORF in MVA by the full-length CVA version of O1L to determine whether MVA would regain the ability to activate ERK1/2 in 293 cells and whether its replication would be increased in human and other mammalian cell lines. The O1L-O2L intergenic region of MVA, which is eight nucleotides shorter than that of CVA, was retained. The reinserted full-length O1L ORF in MVA+O1L was expressed with the characteristics of an early gene, as shown by strong transcription of O1L in the presence of AraC, whereas transcription of the late gene F17R was blocked by AraC (Fig. 8A). This confirmed that the shortened intergenic region of MVA still contained a fully functional O1L promoter capable of directing early expression of the full-length O1L ORF. The kinetics of E3L (immediate early), C7L (early), and F17R (late) gene transcription were not significantly different between MVA+O1L-infected and MVA-infected 293 cells (data not shown). By immunoblot analysis, FLAG-tagged O1 protein expressed by MVA+O1L was first detectable at 2 h p.i. (Fig. 8C), in agreement with the RT-qPCR analysis.

ERK1/2 activation by MVA+O1L in 293 cells.

MVA+O1L-infected but not MVA-infected cells showed a sustained activation of ERK1/2 that was comparable to ERK1/2 activation in CVA-infected 293, Vero, or CV-1 cells (Fig. 8B). The kinetics of ERK1/2 activation by MVA+O1L were consistent with the observed kinetics of expression of O1, which was first detectable at 1 to 2 h p.i. (Fig. 8A and C). O1 expression by MVA+O1L did not decrease IκBα degradation (Fig. 8D), demonstrating that O1 is not an NF-κB inhibitor. Therefore, O1 inactivation in MVA did not contribute to the increased NF-κB activation observed for MVA in comparison to parental CVA. On the other hand, increased activation of the ERK1/2 pathway in MVA+O1L-infected cells also caused no or only a very weak increase in NF-κB activation, as indicated by unchanged or very weakly increased IκBα degradation, compared to MVA infection (Fig. 8D). Stimulation of uninfected 293 cells with graded doses of supernatants from 293 cells that had been infected with MVA, MVA+O1L, CVA, or CVA-ΔO1L induced very similar levels of ERK1/2 activation (Fig. 8E and F). This confirmed that the differences in ERK1/2 phosphorylation were not due to different VGF activities in cell culture supernatants. Thus, expression of full-length O1 restored sustained ERK1/2 activation by MVA, confirming the role of O1 in regulation of the Raf/MEK/ERK pathway.

Focus formation and replication of MVA+O1L.

Chickens solely express the 42-kDa ERK2 isoform, but not ERK1 (65). In contrast to 293 or CV-1 cells, CEF cells showed high ERK2 activation following MVA infection, and the activation was not increased by O1L expression (Fig. 9A). Likewise, ERK2 activation by CVA-ΔO1L in CEF cells was not significantly decreased compared to activation by CVA (Fig. 9A). The sizes of the viral foci formed by MVA+O1L and MVA in CEF cell monolayers were also similar (Fig. 9B), which is consistent with ERK2 activation in CEF cells being independent from O1 expression. Interestingly, however, U0126 treatment also had no effect on the size of MVA and MVA+O1L foci (Fig. 9B). Thus, focus formation of MVA in CEF cells was independent from ERK2 activation. Treatment of CEF monolayers with the EGFR inhibitor AG1478 slightly reduced MVA and MVA+O1L focus size and EGFP expression (Fig. 9B). CVA plaque size in CEF cells was also not affected by U0126 or the deletion of O1L, and AG1478 had only a minor effect (data not shown), in contrast to the significant impact of AG1478 treatment on CVA plaque size in 293 cells (Fig. 5B). These results indicate that the ERK2 activation status of CEF cells is irrelevant not only for replication or spread of CEF-adapted MVA but also for replicating vaccinia virus strains such as CVA.

Fig 9.

O1 expression by MVA does not add to ERK2 phosphorylation and does not enhance viral focus size in CEF cells. (A) CEF cells were infected with MVA, MVA+O1L, CVA, or CVA-ΔO1L at an MOI of 10 or mock infected. After 4 h of infection, cells were lysed and phosphorylation of ERK2 (p-ERK2) and expression of tubulin were analyzed by immunoblotting. Chicken cells express only the ERK2 isoform. (B) CEF cells were pretreated with inhibitors of EGFR tyrosine kinase (AG1478) or MEK1/2 kinase (U0126) or carrier control (DMSO) for 30 min before infection. Cells were infected with MVA or MVA+O1L in the presence of inhibitors at appropriate virus dilutions to obtain discernible foci. After 1 h of infection and again after 1 day, medium was changed to semisolid overlay medium containing inhibitors. At day 3, focus size was analyzed by fluorescence microscopy using GFP as a marker of virus-infected cells. The contrast and brightness of the microphotographs were adjusted using Microsoft Powerpoint 2007. (C and D) For analysis of the replication behavior of MVA and MVA+O1L, monolayers of 10⁶ secondary CEF cells were infected in triplicate at an MOI of 10 (single-cycle analysis) (C) or at an MOI of 0.05 (multicycle analysis) (D) of the indicated viruses. Viral output at the indicated times is plotted; each data point represents results from three independent wells. (E) Monolayers of 10⁶ human 293 and HeLa cells, monkey Vero and CV-1 cells, hamster BHK-21 cells, and secondary CEF cells were infected in triplicate at an MOI of 0.025 with MVA and MVA+O1L (multicycle replication analysis). The ratio of the viral output 3 day p.i. to the input of 5 × 10⁴ TCID₅₀ per well is plotted. Each data point represents a single titration result determined using three independent wells. A viral output-to-input ratio of <1 is considered to represent no replication according to the definition given in reference 8.

Single- and multicycle replication experiments in CEF cells showed that O1 expression by MVA did not detectably affect viral replication and yields (Fig. 9C and D). These results again disproved a role of O1 for spread or replication of MVA in CEF cells. Unlike MVA, MVA+O1L activated ERK1/2 in human 293 cells, but the replication of MVA+O1L was unchanged in 293 cells as well as in HeLa, hamster BHK-21, and monkey CV-1 and Vero cells (Fig. 9E). In summary, MVA+O1L had no growth advantage over MVA in CEF cells and its replication competence in mammalian cells was not increased.

DISCUSSION

Viruses have evolved a multitude of strategies to modulate cellular signaling pathways to create a favorable intracellular environment for optimal viral replication. We report here that the VACV protein O1 is able to sustain virus-induced ERK1/2 activation via the EGFR. Deletion of O1L from CVA significantly decreased ERK1/2 activation and plaque size. Slight reductions in viral yields suggested impaired intracellular replication, but the possibility of reduced efficiency of cell-to-cell spread cannot presently be excluded. In addition, O1L deletion attenuated the pathogenicity of CVA in a murine infection model associated with limited viral spread in the animals. Reinsertion of the full-length O1L gene into the genome of the highly replication-restricted MVA restored sustained ERK1/2 activation, corroborating the reciprocal results obtained with CVA and CVA-ΔO1L. Notably, furnishing MVA with functional O1 did not increase its capacity to replicate in human or other mammalian cell lines.

The major phenotypic changes resulting from O1L deletion in CVA were very similar to the reported phenotypes of a vaccinia virus deletion mutant lacking the EGF-like viral growth factor gene in the VACV-WR strain background (4, 43). These VACV-ΔVGF mutants also showed reduced plaque size but almost wild-type-like viral yields and lower lethality after intracranial infection of mice, as well as smaller skin lesions in rabbits. Moreover, ERK1/2 activation was significantly reduced in HeLa cells infected with VACV-ΔVGF (43). Thus, deletion of O1L in CVA produced a phenotype very similar to that seen with the VGF gene deletion, suggesting that the two proteins cooperate in targeting the same cellular signaling pathway. Both the VGF (C11R) and the O1L gene are highly conserved not only in orthopoxviruses but also in most other poxvirus genera. Most of the genes with high sequence similarity to O1L across all poxvirus genera have a length similar to that of O1L, whereas in some species, such as taterapox virus or camelpox virus, smaller ORFs with significant homology that might represent gene fragments are found. Remarkably, not only is an O1L-like gene conserved in all genera of the chordopoxvirus subfamily, but even the insect poxvirus of Amsacta moorei contains a 604-aa domain within a 1,238-aa protein that is similar to O1. This indicates that the O1L gene was acquired early in the evolution of poxviruses, and its conservation throughout evolution suggests that it provides a significant advantage in interactions of all these viruses with their host cells.

The importance of ERK1/2 activation for viral plaque or focus formation was clearly dependent on the host cell type, which was illustrated by the lack of an effect of pharmacological blockade of ERK2 activation on the focus size of CVA or MVA in CEF cells (Fig. 9B and data not shown). It is possible that the downstream targets of ERK2 supporting viral spread or replication were activated by other infection-induced pathways in CEF cells. Alternatively, downstream effectors of ERK2 may have been active by default in these embryonal cells, circumventing the requirement for ERK2 activation. Thus, the full-length O1L ORF might have simply been lost due to lack of positive selection over the more than 570 CEF cell passages, since O1 was not required for strong and durable ERK2 phosphorylation and activated ERK2 was not required for viral spread or replication. Moreover, although CEF cells are highly dependent on EGFR stimulation, overstimulation of the ERK2 pathway might have detrimental effects, as demonstrated by overexpression of VGF by a VACV variant resulting in smaller plaques on BSC40 cells (50). Thus, O1 expression might even overstimulate the ERK2 pathway in MVA-infected CEF cells, although negative effects were not easily detected and were minor at most. Nevertheless, expression of a functional O1 protein might have been counterselected during passage of MVA in CEF cells. In contrast, complete inhibition of all EGFR-triggered pathways by AG1478 treatment slightly reduced the plaque size of MVA and MVA+O1L as well as CVA in CEF cells (Fig. 9B and data not shown). These observations suggest that a positive selective pressure had conserved a functional VGF gene during MVA adaptation to CEF cells whereas a lack of positive selection, or even a negative selection pressure, led to the inactivation of the O1L gene in MVA.

It has been reported that MVA activated ERK1/2 at late times (2 to 8 h) of infection in 293T cells whereas VACV-WR only weakly induced ERK1/2 at 2 h p.i. in these cells (18). In contrast, we observed only weak and sometimes undetectable activation of ERK1/2 by MVA throughout infection in the 293 cells used in this study. In addition, activation of NF-κB by MVA was largely independent of the activation status of ERK1/2 in our 293 cells, since low-to-undetectable ERK1/2 phosphorylation coincided with strong NF-κB activation (Fig. 3B). The proposed stimulatory effect of ERK1/2 activation on NF-κB activation in MVA-infected 293T cells (18) was only weakly if at all detectable in our 293 cells (Fig. 8D). The reasons for these discrepancies from the results of Gedey et al. (18) remain unclear but might be related to significant differences in regulation of signaling pathways between the 293T cells used by Gedey et al. and the parental 293 cells used in our study. 293T cells express the simian virus 40 (SV40) large T antigen, which interacts with the pRb family of cell cycle regulators (7). Available evidence indicates that pRB not only represents a target of ERK1/2 but is also involved in activation of the positive-feedback loop of the MEK/ERK signaling cascade (27). Thus, ERK1/2 activation might be differentially regulated in cells with different pRb activation levels in addition to differences due to variations in the stimulus, the cell density, medium conditions, and other undefined factors. Moreover, a significant role of the EGFR and Ras/Raf in the observed upregulation of ERK1/2 phosphorylation by MVA in 293T cells was excluded (18), pointing at differential regulation of VACV-induced ERK1/2 activation in these cells.

The data presented here indicate that the O1 protein specifically targets the Raf/MEK/ERK signaling module whereas alternative MAPK pathways such as the JNK and p38 pathways remain largely unaffected by O1 expression (Fig. 3E and data not shown). The strength and duration of signals generated by the three-tiered Raf/MEK/ERK signaling module are regulated by a complex network of factors (44). These include internalization and lysosomal degradation of upstream tyrosine kinase receptors such as the EGFR (31, 69), scaffold proteins which organize the spatial arrangement, interaction, and subcellular localization of the components of the signaling module, and feedback regulators such as phosphatases and kinases. In addition, active ERK itself downmodulates the activity of upstream components such as Raf (14), MEK1/2 (16), and SOS (10) via its kinase activity. Since the extents of EGFR degradation in CVA- and CVA-ΔO1L-infected 293 cells were indistinguishable and O1 does not appear to act as a kinase itself, it is conceivable that O1 targets one or more factors involved in regulation of the Raf/MEK/ERK pathway. O1 might interfere with positive- or negative-feedback loops by interacting with the respective factors or mimicking their function or might even serve as a scaffold protein itself, thereby amplifying the ERK1/2 signal. The O1L ORF encodes a protein of 666 amino acids with a hitherto-unknown function containing a predicted leucine-zipper domain and a potential bipartite nuclear localization signal. Further analysis of the as-yet-unknown subcellular distribution of O1 would contribute to the elucidation of its mechanism of action.

The interaction of the signaling components of the ERK module and also the interaction of these components with feedback regulators are coordinated by scaffold proteins such as KSR, which bring the various components into close proximity, thereby initiating and controlling their interactions (reviewed in reference 19). These scaffold proteins are themselves targets of regulation by ERK1/2. These examples illustrate the complexity and diversity of the ERK network that might be exploited by O1 to rewire the EGFR/Raf/MEK/ERK signaling pathway for converting VGF-induced transient ERK1/2 activation into sustained activation.

The conversion of a transient ERK1/2 activation into a sustained activation can have profound and qualitatively different effects on the target selectivity of ERK1/2 and the cellular response to its activation (32). For example, it has been shown that only sustained ERK1/2 activation induced by NGF, and not transient activation by EGF, can launch differentiation and neurite outgrowth programs in PC12 pheochromocytoma cells (56). Thus, it is conceivable that O1-mediated prolongation and enhancement of VGF-induced ERK1/2 signaling does not simply amplify ERK-mediated effects but might result in qualitatively different cellular responses favoring viral replication.

Inhibition of apoptosis is a central requirement for successful replication and spread of orthopoxviruses, which is underscored by the substantial number of viral genes devoted to apoptosis inhibition, including F1L, N1L, E3L, SPI-2 (B13R), and those encoding a number of secreted tumor necrosis factor alpha (TNF-α) binding proteins (11, 55). It has recently been reported that the VACV F1 protein exerted its full antiapoptotic effect in HeLa cells by synergizing with VGF-mediated EGFR stimulation. The synergistic effect was dependent on VGF-induced ERK1/2 activation (43), demonstrating the beneficial role of a sustained activated Raf/MEK/ERK pathway for survival of VACV-infected human cells. Thus, one benefit of VACV-modulated ERK1/2 activity might reside in the antiapoptotic effects of ERK1/2 activation.

The reduction in yields of extracellular CVA-ΔO1L virus suggests that reduced cell-to-cell spread might at least contribute to the reduced plaque size and possibly also to the observed delay in spread through cell monolayers (Fig. 6C). Apart from its roles in cell proliferation and survival, ERK1/2 activation promotes cell migration and motility (59). Among the multiple targets of activated ERK1/2 that affect cell motility are the actin-cross-linking protein EPLIN (20), myosin light chain kinase (37), focal adhesion kinase (22), and the key actin cytoskeleton regulators rac and RhoA (59). Cell-to-cell spread via actin tail formation or another cytoskeleton-dependent step of viral egress might be enhanced by sustained ERK1/2 activation. Actin tails propelling surface-attached VACV particles away from infected cells toward new, uninfected cells are essential for efficient cell-to-cell spread. This is documented by the small-plaque phenotype of numerous viral mutants deficient or impaired in inducing actin tail formation (12, 17, 25, 64, 68). Apart from actin tail formation, other virus-induced changes of the cellular cytoskeleton have also been demonstrated to affect VACV spread. Wrapped and nonwrapped mature viral particles are transported along the microtubule system through the cytoplasm and the cortical actin layer to the cell membrane (21, 45, 63). Reorganization of the cortical actin network was shown to be required for efficient exit of VACV-WR particles that was independent of actin tail formation (3). The viral F11 protein played an important role in this process by inhibiting the small GTPase RhoA and thereby modulating RhoA-mediated rearrangements in cortical actin (3). Since ERK1/2 is also involved in the regulation of the small GTPases RhoA and Rac, it is thus conceivable that O1-mediated sustained activation of ERK1/2 induces reorganization of cytoskeletal actin elements via these regulators in a manner favorable for efficient translocation of wrapped virus to the cell surface. A role of ERK1/2 in spread is supported by the observation that VGF-deleted virus that induced lower levels of ERK1/2 activation was also released at 2- to 3-fold-lower levels into the culture supernatant at 24 h p.i. (43). Furthermore, reduced viral spread and altered CPE without detectable impairment of replication was observed with F11L deletion mutants of VACV-WR (36). In contrast to O1L deletion, we did not observe altered ERK1/2 activation with an F11L deletion mutant of CVA (data not shown). This is in agreement with the reported direct RhoA-inhibitory function of F11 and suggests that O1 and F11 target different pathways involved in regulation of cytoskeleton rearrangement to achieve increased viral egress and spread.

We found that VACV devotes two genes, O1L and VGF, to activation of the crucial ERK1/2 signaling pathway. The finding that O1 is required for sustained ERK1/2 signaling initiated by VGF also demonstrates not only that VGF has a paracrine role in viral spread by preparing neighboring uninfected cells for efficient replication of VACV (5, 28) but also that the autocrine action of VGF substantially contributes to efficient viral spread or replication. Reinsertion of O1L into MVA was not sufficient to increase the capacity of MVA to replicate in human cells. Thus, the lack of functional O1 does not autonomously govern the replication-restricted phenotype of MVA. However, since O1L was necessary for efficient spread or replication of VACV in human cells, the inactivation of this gene, in combination with loss or mutation of other viral factors, might be involved in replication restriction of MVA in mammalian cells and might contribute to the safety of MVA and stability of the attenuated phenotype.