Vaccinia-induced epidermal growth factor receptor-MEK signalling and the anti-apoptotic protein F1L synergize to suppress cell death during infection

F1L is a functional Bcl-2 homologue that inhibits apoptosis at the mitochondria during vaccinia infection. However, the extent and timing of cell death during ΔF1L virus infection suggest that additional viral effectors cooperate with F1L to limit apoptosis. Here we report that vaccinia growth factor (VGF), a secreted virulence factor, promotes cell survival independently of its role in virus multiplication. Analysis of single and double knockout viruses reveals that VGF acts synergistically with F1L to protect against cell death during infection. Cell survival in the absence of F1L is dependent on VGF activation of the epidermal growth factor receptor. Furthermore, signalling through MEK kinases is necessary and sufficient for VGF-dependent survival. We conclude that VGF stimulates an epidermal growth factor receptor-MEK-dependent pro-survival pathway that synergizes with F1L to counteract an infection-induced apoptotic pathway that predominantly involves the BH3-only protein Bad.


Introduction
Upon virus infection most cells attempt to undergo cell death as a means to limit viral spread. To circumvent this, viruses have evolved numerous mechanisms to inhibit apoptosis, thereby prolonging the survival of infected cells to facilitate productive replication (Roulston et al., 1999). Consequently, many viruses encode sequence homo-logues of anti-apoptotic members of the Bcl-2 family (Cuconati and White, 2002). Like their cellular counterparts, these viral proteins inhibit the pro-apoptotic Bak and Bax from executing mitochondrial-dependent cell death. Bcl-2 sequence homologues are noticeably absent from most poxvirus genomes. However, the Lepripox virus myxoma encodes a protein, M11L, which prevents apoptosis by inhibiting Bak and Bax (Wang et al., 2004;Su et al., 2006). Similarly, vaccinia virus encodes a mitochondrially targeted protein, F1L, which is able to inhibit apoptosis by binding and preventing activation of Bak and Bim (Wasilenko et al., 2005;Postigo et al., 2006;. This anti-apoptotic activity is dependent on a BH3-like, Bak-binding domain, suggesting that F1L acts as a functional Bcl-2 homologue (Postigo et al., 2006).
In the absence of F1L cells succumb to cell death much earlier than during infection with Western Reserve (WR), even though replication and spread appear unaffected (Wasilenko et al., 2005;Postigo et al., 2006). Furthermore, it has been reported that infection triggers a cellular suicide response shortly after virus entry (Ramsey-Ewing and Moss, 1998) and before F1L is detected at 2-3 h after infection (Postigo et al., 2006). Given this, we wondered whether an additional early viral protein cooperates with F1L to prevent cell death. One possible candidate expressed early during infection is the highly conserved vaccinia growth factor (VGF), an epidermal growth factor (EGF)-like growth factor (Brown et al., 1985). EGF belongs to a family of ligands that is conserved from invertebrates to mammals (Yarden and Sliwkowski, 2001). All these ligands induce ligand-induced homoand heterodimerization of the ErbB1-4 receptor tyrosine kinases leading to activation of Ras-MEK-ERK, phospholipase C and STAT pathways (Yarden and Sliwkowski, 2001). EGF receptor (EGFR) (ErbB1)-mediated signalling stimulates cell proliferation and differentiation, but it can also act to inhibit apoptosis (Sibilia et al., 2000;Jost et al., 2001a;Yarden and Sliwkowski, 2001).
Vaccinia growth factor is secreted early during infection. It binds EGFR homodimers with reduced affinity, a feature that enhances signalling by attenuating receptor inactivation (Tzahar et al., 1998). Deletion of VGF from the vaccinia genome (DVGF) leads to reduced plaque size in vitro and reduced virulence in vivo, consistent with a role in virus replication (Buller et al., 1988a;. Furthermore, secreted VGF acts as a mitogen that primes noninfected cells for subsequent infection (Buller et al., 1988a;. In addition to this non-cell autonomous effect, VGF activates ERK in infected cells, which is required for efficient viral replication and multiplication (Andrade et al., 2004).
In this study we report that VGF promotes survival of vaccinia-infected cells by inducing an EGFR-MEKdependent pathway that mimics host survival signals in epithelial cells. Furthermore, by using the DF1L-sensitized genetic background we show that VGF and F1L act synergistically to counteract a cell death pathway predominantly dependent on the BH3-only protein Bad.

VGF and F1L act synergistically to prevent infection-induced cell death
In order to test whether growth factor signalling by VGF contributes to cell survival during infection, we deleted the F1L gene from the DVGF virus. Infection with the resulting double mutant virus (DF1L/VGF) leads to inhibition of phosphorylation of the EGFR, indicating that the virus, like DVGF, was unable to induce growth factor signalling (Fig. 1A). Consistent with previous observations using the single-deletion strains, we found that WR and DF1L infection produces large plaques, while both the DVGF and DF1L/VGF viruses form small plaques (Fig. 1B). Re-infection assays with WR, DVGF, DF1L or DF1L/VGF viruses confirmed that the absence of VGF results in a reduction in infectious particle production consistent with the plaque phenotype and the replication defect caused by the absence of VGF (Fig. 1B). Thus, absence of F1L in the double mutant does not exacerbate the replication defect reported for the DVGF virus.
To determine if VGF also promotes cell survival, we examined whether infection with the DVGF virus resulted in increased apoptosis using quantitative immunofluorescence assays that detect increases in fragmented nuclei, accumulation of activated Bax and loss of mitochondrial cytochrome C ( Fig. 2A). All three assays revealed a small but significant increase in apoptosis after DVGF infection (Fig. 2B). A more pronounced level of cell death was observed in DF1L infection cells (Fig. 2B). Moreover, infection with the DF1L/VGF virus resulted in a synergistic increase in cell death in all three assays (Fig. 2B). The increase in cell death in the absence of VGF or F1L also correlated with classical biochemical hallmarks of apoptosis (Fig. 2C). Infection with the WR did not induce poly (ADP-ribose) polymerase (PARP) cleavage or activation of caspase-3. Loss of VGF resulted in a small but detectable cleavage of PARP and caspase-3 activation, while deletion of F1L induced pronounced PARP cleavage and activation of caspase-3. Cleavage of these apoptotic markers was significantly enhanced in the absence of both F1L and VGF during DF1L/VGF infections (Fig. 2C). Taken together, our results show that infection with the DF1L/VGF virus leads to a synergistic increase in cell death, consistent with a role for VGF in promoting cell survival during infection.

VGF mediates cell survival in an EGFR-dependent manner
It has been reported that VGF preferentially binds EGFR-containing hetero-and homodimers. To establish whether VGF stimulates cell survival solely via the EGFR receptor, we examined the effect of AG1478, a specific EGFR inhibitor, and DAPH, an inhibitor of ErbB1/2, on the

ΔF1L
ΔVGF/F1L  Fig. 3A and B). Western blot analysis revealed that AG1478 inhibited activation of the ERK pathway due to WR infection (Fig. 3B). Both AG1478 and DAPH treatment substantially increased cell death during DF1L infection to levels that were similar to those observed with the DF1L/VGF virus (Fig. 3A). This corre-lates with AG1478 blockade of vaccinia-induced EGFR signalling and downstream ERK activation. In addition, activation of ERK by phosphorylation is severely reduced in the absence of VGF (Fig. 3B). Furthermore, similar to the corresponding cell death counts, analysis of apoptosis markers revealed an increase in PARP and caspase-3 cleavage in lysates from DF1L-infected cells to the levels  A. Quantitative analysis of DNA fragmentation in HeLa cells infected for 20 h with the indicated virus strains in the presence or absence of the EGFR inhibitors AG1478 and DAPH. B. Western blot analysis of cells infected for 20 h with the indicated virus in the presence (+) or absence (-) of AG1478 reveals that loss of EGFR signalling reduces ERK phosphorylation and increases PARP processing as well as caspase-3 cleavage. C. Quantitative analysis of cell death in HeLa cells infected with indicated virus strains for 20 h in the presence or absence of the VGF peptide. Error bars represent standard error of the mean, *P < 0.05 and ***P < 0.001. D. Western blot analysis of infected cells in the presence (+) or absence (-) of the VGF peptide reveals that activation of EGFR signalling promotes ERK phoshorylation and reduces PARP processing and caspase-3 cleavage during DVGF and DF1L/VGF infection. observed with DF1L/VGF (Fig. 3B). AG1478 had no additional effects on PARP and caspase-3 cleavage in cells infected with the DF1L/VGF virus (Fig. 3B).
Having established that VGF-mediated survival depends on the EGFR, we next asked whether exogenously supplied VGF is sufficient to rescue F1Lindependent cell survival during DF1L/VGF infection. Addition of VGF led to an increase in cell survival in DVGF-infected cells, but had no effect on DF1L-infected cells (Fig. 3C). Moreover, VGF decreased cell death in DF1L/VGF-infected cells to levels similar to DF1L (Fig. 3C). Addition of the synthetic VGF peptide also stimulated ERK phosphorylation for the duration of the experiment (Fig. 3D). Furthermore, VGF application inhibited PARP and caspase-3 cleavage in DF1L/VGFinfected cells (Fig. 3D). Together, our results demonstrate that VGF signalling via the EGFR is sufficient to promote F1L-independent cell survival during infection.

Late gene expression is dispensable for VGF-mediated cell survival
Expression of VGF is required for efficient replication and late gene expression. It is possible therefore that VGF mediates cell survival indirectly, for instance, if it is required for expression of a late gene that promotes cell survival. To exclude whether this is the case, we infected cells with the four different viruses in the presence of cytosine arabinoside (AraC), an inhibitor of DNA replication and late gene expression. AraC would be expected to exacerbate cell death if VGF survival depends on late gene expression. However, Western blot analysis of infected cell lysates treated with AraC showed no further increase in PARP cleavage or caspase-3 processing in any of the virus strains (Fig. 4A). This suggests that VGF, as well as F1L exert their effects on cell survival independently of late gene expression.
Addition of exogenous VGF peptide is sufficient to limit cell death during DF1L/VGF infection to the levels observed with DF1L (Fig. 3C). Addition of AraC did not increase PARP and caspase-3 cleavage in DF1L/VGFinfected cells in the presence of the VGF peptide (Fig. 4B). These experiments using an exogenously added peptide demonstrate that VGF promotes survival independently of late gene expression.

MEK is necessary and sufficient to mediate VGF-dependent survival
Activation of the EGFR induces cell survival by stimulating the Ras-MEK-ERK and phosphoinositol 3-phosphate kinase (PI3K)-Akt pathways. ERK is activated by phosphorylation in response to vaccinia virus infection, and this increase in phosphorylation is severely reduced in the absence of VGF (Figs 3B and 5A). In contrast, we failed to see changes in Akt phosphorylation throughout infection irrespective of the viral strain (Fig. 5A). We therefore investigated whether the Ras-MAPK pathway contributes to VGF-dependent survival.
To determine if VGF-mediated survival is dependent on activation of the Ras-MAPK pathway, we examined the effect of the specific MEK1/2 inhibitors UO126 and PD98059 during infection with WR and the deletion strains. Treatment with either drug had no effect on DVGF or DF1L/VGF-induced cell death, but increased DF1Linduced cell death to similar levels as observed with the DF1L/VGF virus (Fig. 5B). WR and DF1L-mediated ERK activation was blocked in the presence of UO126 (Fig. 5C). Western blot analysis of the extent of PARP and caspase-3 cleavage confirmed that inhibition of MEK1/2 by UO126 resulted in increased apoptosis in DF1L infected cells (Fig. 5C). Conversely, addition of VGF peptide to DF1L/VGF-infected cells activated ERK and at the same time reduced PARP and caspase-3 cleavage (Fig. 5D). The addition of U0126 blocked ERK phosphorylation by VGF peptide and abrogated the VGF-mediated decrease in DF1L/VGF-induced PARP and caspase-3 cleavage (Fig. 5D). We next examined if constitutively active MEK (MEK-EE) would be sufficient to promote F1L-independent cell survival during infection in the absence of an upstream VGF signal. To ensure that cell death was only quantified in infected HeLa cells expressing MEK-EE we used GFP expression as a co-reporter for cell transfection. This analysis reveals that expression of MEK-EE did not impact on DF1L-induced death (Fig. 6A). In contrast, cell death in DF1L/VGF-infected cells was reduced to similar levels as observed in DF1L infections (Fig. 6A). Western blot analysis revealed that there was a corresponding decrease in infection-induced PARP and caspase-3 cleavage in DF1L/VGF cells expressing MEK-EE (Fig. 6B)  necessary and sufficient for the VGF-mediated cell survival during infection.

VGF and F1L cooperatively block infection-induced cell death via Bad
It is at present unclear what molecular trigger induces vaccinia-induced cell death when either F1L or VGF are absent. If VGF and F1L synergize to prevent cell death, the loss of a crucial host component mediating cell death should abrogate cell death in DF1L/VGF cells. Bad and Bim are two BH3-only pro-apoptotic members of the Bcl-2 family that are known to be regulated by growth factor signalling (Puthalakath and Strasser, 2002). We therefore tested whether Bad and Bim might be responsible for mediating infection-induced cell death by examining the effect of their downregulation through RNAi-mediated ablation (Fig. 7). Remarkably, for the three deletion viruses, we found that loss of Bad resulted in a dramatic reduction in cell death to similar levels to that observed in WR infections (Fig. 7A). In contrast, knockdown of Bim only resulted in a significant reduction in cell death in DF1L/VGF-infected cells. These results were consistent with Western blot analysis of apoptosis markers (Fig. 7B). Loss of Bad significantly reduced PARP and caspase-3 cleavage in both DF1L and DVGF-infected cells. In DF1L/ VGF-infected cells loss of Bim resulted in a small reduction of cell death, whereas lack of Bad resulted in a large reduction in PARP and caspase-3 cleavage (Fig. 7). Our data demonstrate that VGF and F1L cooperate to suppress vaccinia-induced cell death through a pathway that primarily involves Bad.

Discussion
The EGF stimulates complex signalling networks to promote proliferation, differentiation, migration or survival in a context-dependent manner (Yarden, 2001;Yarden and Sliwkowski, 2001). Consistent with its critical role in cell proliferation and survival components of the EGF signalling networks are frequently constitutively activated within tumours (Olayioye et al., 2000). It is not surprising given that viruses are also dependent on cell survival for productive replication that they have evolved elegant mechanisms to inhibit infection-induced cell death (Roulston et al., 1999). Our data now clearly show that vaccinia virus stimulates the EGF signalling pathway to enhance cell survival during infection. Previous studies have shown that VGF, which is secreted early during infection, induces continuous activation of the EGFR signalling throughout the virus infection cycle as a consequence of its lower affinity for the receptor (Tzahar et al., 1998). This prolonged EGFR activation and signalling would ensure that the infected cell receives a sustained survival signal that antagonizes infection-induced cell death. In accordance with previous studies (Andrade et al., 2004), we have shown that triggering signalling pathways by VGF involves a sustained activation of the ERK pathway during infection. Furthermore, our data indicate that activation of the EGFR/MEK pathway is the downstream signalling cascade stimulated by VGF sufficient to elicit a cell survival signal during infection (Figs 5 and 6). Survival pathways stimulated by growth factors in many biological systems are frequently PI3K-Akt-dependent (Cantley, 2002). However, phosphorylation of EGFR is known to only weakly activate the PI3K-Akt pathway (Soltoff et al., 1994). Consistent with this we did not observe any changes in activation of the PI3K-Akt pathway during vaccinia infection (Fig. 5A). This is in contrast to the situation observed with several RNA viruses, as well as herpes viruses, which stimulate PI3K-Akt signalling to inhibit apoptosis via intracellular adaptors and decoys (Cooray, 2004).
The role of MEK in EGFR-dependent cell survival has previously been described in keratinocytes, with inhibition of MEK resulting in apoptosis (Jost et al., 2001b). In these cells, survival in the absence of matrix-derived signals is dependent on EGFR signalling, which is required to maintain a high-level expression of the anti-apoptotic protein Bcl-xL (Jost et al., 2001a). In other epithelial cell types, EGFR-dependent ERK activation is thought to promote cell survival by preventing upregulation of the proapoptotic protein Bim (Reginato et al., 2003).
Our data reveal that VGF, in concert with F1L, inhibits apoptosis by activating the EGFR/MEK pathway to counteract a cell death pathway principally through the BH3only protein Bad (Fig. 7). It has been previously reported that during vaccinia infection Bim is a significant inducer of cell death in the absence of F1L . We found that Bim plays only a modest or redundant role in the induction of cell death by DF1L/VGF infection (Fig. 7). Whether these results are attributable to assay/ cell type differences or caused by differences between acute siRNA-mediated knockdown versus knockout cells, which may have adapted, is unclear. Therefore, we cannot formally exclude that incomplete Bim knockdown allowed apoptosis to progress as normal, although our findings are consistent with the observation that cell death induced by MVA-DF1L infection is not reduced in Bim -/-MEF cells (Fischer et al., 2005).
In contrast to Bim, loss of Bad significantly reduced infection-induced cell death by all our virus strains (Fig. 7). It is thought that vaccinia induces apoptosis via several routes (Fischer et al., 2005;. For instance, vaccinia dsRNA-induced apoptosis depends exclusively on the BH3-only protein Noxa and is blocked by E3L (Fischer et al., 2005). Notably, in the context of DE3L infection cell death is not inhibited by the presence Cell death (%) *** of F1L (Fischer et al., 2005). Our data indicate that Vaccinia infection, in addition, triggers a Bad-dependent cell death pathway that is antagonized by VGF and F1L. It would be of interest whether VGF and F1L directly inhibit Bad or at some point downstream of Bad to inhibit cell death. Interestingly, no role for Bad in vacciniainduced cell death has so far been reported. Bad binds to the pro-survival proteins Bcl-2, Bcl-w and Bcl-xL, thus preventing their inhibition of Bak and/or Bax . F1L may inhibit Bad by direct sequestration as observed for Bim , yet no Bad binding to F1L has been reported thus far. Alternatively, F1L could represent a Bcl-xL-like molecule that retains a high Bakbinding affinity, but unlike host Bcl-xL, does not interact with BH3-only proteins, including Bad. Thus, while activated Bad could sequester Bcl-2 pro-survival proteins to prevent them from neutralizing Bak, F1L would ensure the continued inactivation of Bak. A similar mechanism has been proposed for the myxoma virus protein M11L based on its structure (Kvansakul et al., 2007). Interestingly, M11L binding to Bak and Bax is required for its survival promoting activity, yet despite its interaction with Bim, the latter is not involved in the anti-apoptotic function of M11L (Kvansakul et al., 2007). While the structure of F1L has been solved (Kvansakul et al., 2008), it remains to be established whether a similar mechanism is operating during vaccinia infection. In contrast to M11L, F1L binds Bak but not Bax (Wasilenko et al., 2005;Postigo et al., 2006). It was recently reported that vaccinia N1L, which structurally resembles Bcl-2 proteins, can interact with Bax and Bad (Cooray et al., 2007). It is an intriguing possibility that N1L and F1L target complimentary sets of Bcl-2 proteins, although functionally N1L is not sufficient to inhibit the extensive Bad-dependent cell death induced by the DF1L/VGF virus.
Our study has revealed VGF to be an important inhibitor of cell death during infection. It will be interesting to test what targets of the VGF pathway promote survival, including potential phosphorylation of viral proteins required for survival, although our work suggests these would need to be early proteins. In conclusion, we have shown for the first time that a virus is capable of using two cooperative strategies to suppress a Bad-dependent cell death pathway: VGF stimulates EGFR/MEK kinase survival signalling, while F1L inhibits mitochondrial-dependent cell death.

Viruses and cells
We deleted the F1L locus in the DVGF WR strain that lacks both copies of VGF (open reading frames WR009 and WR210) (Buller et al., 1988a) using the guanine phosphoribosyl transferase cassette replacement strategy previously used to generate the DF1L virus (Postigo et al., 2006). The resulting DF1L/VGF virus was verified by sequencing and Western blot analysis. Infections of HeLa cells with vaccinia virus and processing for immunoblot analysis were as described previously (Postigo et al., 2006).

Infections and cell survival quantification
HeLa cells plated on dishes coated with fibronectin were infected with WR or the recombinant vaccinia virus strains at a multiplicity of infection of 2 in Opti-MEM reduced serum media and processed for microscopy or immunoblot analysis 20 h post infection. All cell death quantification experiments were performed at least three times, each consisting of a minimum of three counts of over 200 cells each. For western analysis, detached apoptotic cells were collected and added to the adherent cells before cell lysis.

Drug treatments, transfections and siRNA
HeLa cells were incubated 30 min prior to infection and during infection with the following as appropriate: UO126 (10 mM), PD98059 (36 mM), AG1478 (20 mM), DAPH (15 mM), VGF peptide (10 nM), AraC (50 mM). HeLa cells were transfected with pE/L-MEK1.EE together with pE/L-GFP as a transfection reporter using Effectene according to the manufacturer's instructions (Qiagen) before being infected with WR or the deletion virus strains 2 h later. Cells were processed for immunoblot analysis 20 h after treatments and transfections. HeLa cells were transfected with Bad, Bim or a control non-targeting siRNA (20 nM) according to the manufacturer's instructions using Hiperfect (QIAGEN). One day post transfection, cells were infected with vaccinia virus and incubated for a further 20 h before being processed for Western blotting.