Anti-apoptotic Bcl-XL but not Mcl-1 contributes to protection against virus-induced apoptosis

Infection of mammalian cells with viruses often induces apoptosis. How the recognition of viruses leads to apoptosis of the infected cell and which host cell factors regulate this cell death is incompletely understood. In this study, we focussed on two major anti-apoptotic proteins of the host cell, whose abundance and activity are important for cell survival, the Bcl-2-like proteins Mcl-1 and Bcl-XL. During infection of epithelial cells and fibroblasts with modified vaccinia virus Ankara (MVA), Mcl-1 protein levels dropped but the MVA Bcl-2-like protein F1L could replace Mcl-1 functionally; a similar activity was found in vaccinia virus (VACV)-infected cells. During infection with murine cytomegalovirus (MCMV), Mcl-1-levels were not reduced but a viral Mcl-1-like activity was also generated. Infection of mouse macrophages with any of these viruses, on the other hand, induced apoptosis. Virus-induced macrophage apoptosis was unaltered in the absence of Mcl-1. However, apoptosis was substantially increased in infected Bcl-XL-deficient macrophages or macrophages treated with the Bcl-2/Bcl-XL-inhibitor ABT-737. Genetic loss of Bcl-XL or treatment of macrophages with ABT-737 reduced the generation of infectious VACV. These data show that Mcl-1 is dispensable for the regulation of apoptosis during infection with different large DNA viruses, either because the viruses replace its function (in fibroblasts and epithelial cells) or because the pro-apoptotic activity generated by the infection appears not to be blocked by it (in macrophages). Bcl-XL, on the other hand, can be important to maintain survival of virus-infected cells, and its activity can determine outcome of the infection.

Infection of mammalian cells with viruses often induces apoptosis. How the recognition of viruses leads to apoptosis of the infected cell and which host cell factors regulate this cell death is incompletely understood. In this study, we focussed on two major anti-apoptotic proteins of the host cell, whose abundance and activity are important for cell survival, the Bcl-2-like proteins Mcl-1 and Bcl-X L . During infection of epithelial cells and fibroblasts with modified vaccinia virus Ankara (MVA), Mcl-1 protein levels dropped but the MVA Bcl-2-like protein F1L could replace Mcl-1 functionally; a similar activity was found in vaccinia virus (VACV)infected cells. During infection with murine cytomegalovirus (MCMV), Mcl-1-levels were not reduced but a viral Mcl-1-like activity was also generated. Infection of mouse macrophages with any of these viruses, on the other hand, induced apoptosis. Virusinduced macrophage apoptosis was unaltered in the absence of Mcl-1. However, apoptosis was substantially increased in infected Bcl-X L -deficient macrophages or macrophages treated with the Bcl-2/Bcl-X L -inhibitor ABT-737. Genetic loss of Bcl-X L or treatment of macrophages with ABT-737 reduced the generation of infectious VACV. These data show that Mcl-1 is dispensable for the regulation of apoptosis during infection with different large DNA viruses, either because the viruses replace its function (in fibroblasts and epithelial cells) or because the pro-apoptotic activity generated by the infection appears not to be blocked by it (in macrophages). Bcl-X L , on the other hand, can be important to maintain survival of virus-infected cells, and its activity can determine outcome of the infection. Apoptosis of the infected cell is a defence strategy of higher organisms against viral infection. Cells have the ability to sense the presence of the virus, and this can often initiate apoptosis. If an infected cell dies by apoptosis, then viral replication is likely precluded. This interpretation of the role of apoptosis during viral infection is supported by the presence of genes coding for anti-apoptotic proteins in many viruses, ranging from baculoviruses (infecting insect cells 1 ) to complex DNA viruses infecting human cells such as poxviruses. 2 To undergo apoptosis upon viral infection, the host cell has to be able to recognize the infection and to initiate the apoptosis system. Prominently, recognition occurs by pattern recognition receptors of innate immunity, which most often recognize viral nucleic acids differing in certain structural characteristics from host cell nucleic acids (for review, see Hornung et al. 3 ). How this recognition machinery is linked to the apoptosis apparatus is less clear. The transfection of RNA that mimics viral RNA and that activates cytosolic helicases has been shown to activate the mitochondrial apoptosis pathway, inducing several pro-apoptotic proteins of the Bcl-2family. In that study, the pro-apoptotic Bcl-2-family protein Noxa was required for apoptosis. 4 Cell survival and apoptosis are regulated by the Bcl-2 family of proteins. Within the Bcl-2 family, the class of BH3-only proteins (including Bim, Bid, Puma, Noxa and Bmf) triggers the activation of the 'executor' proteins Bax and Bak, directly or indirectly. Bax and Bak can oligomerize and release cytochrome c from the mitochondria into the cytosol, where it activates caspases, leading to apoptosis. The anti-apoptotic Bcl-2-family proteins (Bcl-2, Bcl-X L , Bcl-w, Mcl-1 and A1) inhibit this process by binding to both BH3-only proteins and Bax/Bak with varying affinity. 5,6 The decision to undergo apoptosis is taken through the balance of pro-and antiapoptotic Bcl-2-family proteins. Apoptosis may be triggered by upregulation or in some cases activation of BH3-only proteins, or by the loss of anti-apoptotic proteins.
Both Poxviruses and herpes viruses are large, doublestranded DNA viruses with very different cell biology. Both types carry genes whose products resemble mammalian antiapoptotic proteins. 7 Intriguingly, poxviruses have many genes that from their primary structure are not recognizable as Bcl-2 homologues and that do necessarily not act by inhibiting apoptosis but have a very similar spatial structure to Bcl-2 (see for instance Neidel et al. 8 ).
Although the mammalian Bcl-2-like anti-apoptotic proteins have an overall similar, conserved structural fold and all can block apoptosis by binding to pro-apoptotic Bcl-2-family proteins, they substantially differ in terms of this binding affinity as well as their biological function. Deletion of individual anti-apoptotic Bcl-2-proteins in mice produces rather different 1 Institute of Medical Microbiology and Hygiene, University Medical Center Freiburg, Freiburg, Germany; 2 Faculty of Biology, University of Freiburg, Freiburg, Germany; phenotypes, ranging from male sterility (deletion of Bcl-w) and cell death of specific populations in various organs (loss of Bcl-2) to embryonic lethality (loss of Bcl-X or Mcl-1). 9 Bcl-X L and Mcl-1 appear to have complementary functions. In molecular terms, both proteins are thought to bind to and inhibit pro-apoptotic Bak in resting cells. 10 A number of studies have compared the importance of Bcl-X L and Mcl-1 for survival of specific populations and situations. The loss of Mcl-1 leads to spontaneous apoptosis of haematopoietic stem cells as well as of all haematopoietic lineages tested, with the exception of macrophages. 11,12 Bcl-X L deficiency causes embryonic lethality very likely due to neuronal apoptosis. In a number of studies, it was found consistently that loss of Mcl-1 reduced survival of the affected cell populations; this was the case for regulatory T cells, 13 haematopoietic stem/progenitor cells, 14 plasma cells 15 and NK cells 16 as well as malignant cells, as shown for myc-driven 17 and thymic lymphoma. 18 Where Bcl-X L was deleted in the same cell populations, its loss had no effect. [13][14][15]18 Melanoma cells have been shown to undergo apoptosis upon the sole loss of Mcl-1. 19 Non-transformed skin cells, however, underwent apoptosis only upon RNAimediated loss of both Bcl-X L and Mcl-1. 20 We here tested whether Bcl-X L and Mcl-1 have different roles during the infection of mammalian cells with large DNA viruses. We found that, in fibroblasts and epithelial cells, all three viruses tested (MVA and VACV are very similar poxviruses, murine cytomegalovirus (MCMV) is a herpes virus) could replace Mcl-1 function and failed to induce apoptosis. Macrophages, on the other hand, showed substantial apoptosis when infected by any of the three viruses. Surprisingly, Mcl-1, which is critical for survival in many situations, also had no role in protecting macrophages against virus-induced apoptosis. Bcl-X L , on the other hand, did contribute to the survival of infected cells, and its activity contributed to the regulation of survival of infected macrophages.  Figure S1C and D), 22 suggesting that a molecular function of F1L is the replacement of Mcl-1. We tested this possibility by inactivating Bcl-2 and Bcl-X L with the small molecule inhibitor ABT-737. In cells expressing sufficient levels of Mcl-1, ABT-737 on its own fails to induce apoptosis, but the loss of Mcl-1 sensitizes most cells to ABT-737-induced apoptosis. 24   Survival of MVA-infected macrophages is independent of Mcl-1 but depends on Bcl-X L . Upon infection of the organism, viruses are often primarily taken up by macrophages. Upon i.v. infection of mice with MVA for instance viral gene expression in spleen is primarily found in macrophages and dendritic cells. 22 Similarly, MCMV has been found to infect monocytes and induce their differentiation to macrophages before spreading of the virus. 28 In terms of antiapoptotic proteins, macrophages are unique among the tested haematopoietic cells in that they do not depend on Mcl-1 for survival. 12 This suggested that the viral strategy of replacing Mcl-1-function may not be able to protect macrophages during infection.

Infection of epithelial cells and fibroblasts by
Mcl-1-or Bcl-X-deficient mice are not viable. To test for the relative roles of Mcl-1 and Bcl-X L in macrophages during viral infection, we used cells derived from mice carrying the respective alleles flanked by cre-lox-recognition sites together with Cre under the LysM promoter. In this system, Cre is expressed during macrophage differentiation and should cause the deletion of Mcl-1 or Bcl-X L . As cellular system, we utilized a model where mature mouse macrophages can be differentiated from committed macrophage progenitor cells that are 'conditionally transformed' with the oncogene Hoxb8, fused to the ligand-binding domain of the oestrogen receptor.
In the presence of oestrogen, Hoxb8 is active and drives the expansion of progenitor cells, and in this way lines of committed macrophage progenitors can be established. When oestrogen is removed from the culture medium cells start to differentiate and after 7-10 days express the markers of mature murine macrophages. 29 Progenitor lines of either background (LysM-Cre/ Mcl-1 flox/flox or LysM-Cre/Bcl-X L flox/flox ) were generated and expanded. Since the cells were from different genetic crosses, we used wt cells isolated from littermates of each breeding. Cells were induced to differentiate, and maturation was tested by staining for the markers CD11b (myeloid marker) and F4/80 (macrophage-specific marker) on their surface. No differences in expression of these markers between the gene-deficient cells and the corresponding wt cells were noted (Supplementary Figure S3), and there was no obvious cell death during differentiation in culture (not shown). Bcl-X L was expressed at somewhat reduced levels in progenitors carrying the targeted allele, indicating that Cre is expressed at low levels or only in some cells up to this progenitor stage. A reduction in Mcl-1 was already observed in progenitors carrying the loxP-flanked Mcl-1-allele (Supplementary Figure S4). Differentiation of macrophages decreased expression of the respective gene to levels that were very low, sometimes undetectable, indicating Cre-mediated loss during differentiation (Supplementary Figure S4). The results confirm that Mcl-1 is not essential for survival of macrophages and demonstrate that Bcl-X L is likewise dispensable for the development of macrophages from these progenitors.
Infection of mature macrophages with MVA led to a loss of Mcl-1 but not Bcl-X L , similar to the events observed in the above experiments in epithelial (HeLa) and MEF cells ( Figure 4). As we have reported earlier, 22 MVA infection induces substantial apoptosis in macrophages ( Figure 5). Mcl-1-deficient macrophages showed no enhanced apoptosis when infected with MVA, compared with the corresponding wt cells (Figures 5a and b). However, Bcl-X L deficiency clearly enhanced sensitivity of macrophages to MVA-induced apoptosis (Figures 5c and d). Mcl-1 is thus not involved in the regulation of survival and apoptosis during MVA infection in macrophages, but Bcl-X L has a role in keeping infected macrophages alive.
In epithelial cells and HeLa cells, MVA infection induces an interferon response, which through the induction of the proapoptotic Bcl-2-family protein Noxa contributes to apoptosis induction (which is then blocked by F1L 22 ). We therefore tested macrophages deficient in these genes and found a small protection against MVA infection by Noxa deficiency, but no difference in cells where type I interferon signalling is impossible (IFNAR-deficient cells) (Supplementary Figure S5). Interferon signalling therefore appears to have no role in the apoptotic response.
Bcl-X L contributes to apoptosis resistance during infection with VACV and MCMV. We next tested whether this role of Bcl-X L is limited to MVA infection or also found during infection with VACV and MCMV. We found that the response to infection with these viruses was similar: Bcl-X L -deficient macrophages showed an increased apoptotic response upon infection whereas Mcl-1-deficient macrophages did Although this observation confirmed the respective roles of Mcl-1 and Bcl-X L found in the gene-deficient cells, it was surprising since the results in HeLa and MEFs described above had shown that MVA-F1L (and an unidentified activity in VACV, very likely VACV F1L) is able to replace Mcl-1 and to protect against ABT-737. When we tested for F1L expression, we however found that infected macrophages expressed only very little, sometimes undetectable amounts of F1L during infection with MVA or VACV (Figure 7c). Although the molecular basis is unclear, this suggests that reduced expression of anti-apoptotic F1L during macrophage infection contributes to the strong pro-apoptotic activity of MVA and VACV infection.
Apoptosis is typically viewed as a defence mechanism against viral infection. If apoptosis has this role in macrophage infection, then the observation that the loss of Bcl-X L function, which enhances virus-induced macrophage apoptosis, predicts that loss of Bcl-X L in macrophages also reduces the generation of infectious virus. We tested this hypothesis by measuring the generation of infectious VACV upon infection of wt versus Bcl-X L -deficient macrophages (MVA cannot replicate efficiently in mouse cells). Indeed, Bcl-X L -deficient macrophages released only about half the number of infectious viral particles of infected wt macrophages ( Figure 8a). Treatment of wt macrophages with ABT-737 reduced the production of infectious VACV even more strongly, to about 30% (Figure 8b). Apoptosis of infected macrophages is thus a factor in reducing the production of infectious VACV, and apoptosis may help limit the initial replication of the virus. Treatment with Bcl-X L antagonists has the capacity to help reducing the generation of virus during macrophage infection.

Discussion
Our results identify a differential role of the anti-apoptotic proteins Bcl-X L and Mcl-1 during infection of mammalian cells with large DNA viruses. Mcl-1, in contrast to its essential role in many other situations, has no role in the regulation of survival during infection with MVA/VACV or MCMV. In epithelial and fibroblast cells, this is very likely due to the expression of viral anti-apoptotic proteins, in the case of MVA due to F1L. In macrophages, Mcl-1 has no critical role in steady-state survival, nor has Bcl-X L . Mcl-1 also has no role in controlling macrophage apoptosis in response to infection with these viruses. However, Bcl-X L does provide a measure of protection, and its loss or its inactivation with ABT-737 increases the apoptotic response and reduces the replication of VACV. We hasten to add that we only used one particular strain of VACV and of MCMV and can well picture that other strains may have different pro-or anti-apoptotic potential in a given cell type.
Of the Bcl-2-homologues, Mcl-1 has a particularly critical function in a number of cell types, especially in the immune system (while no role of Bcl-X L in many situations is apparent, see Introduction). Why Mcl-1 has this important role is not clear but this may be linked to the strong regulation of Mcl-1 by extrinsic stimuli (such as cytokines, see for instance Sathe et al. 16 ) and major signalling pathways (such as the GSK-3 pathway 30 ). Mcl-1 can have a very rapid turnover, and a number of enzymes are known to regulate its proteasomal degradation. 31 Mcl-1 may therefore be the main anti-apoptotic Bcl-2-family member that responds to stimuli from the outside.
Indeed, Mcl-1 does respond to viral infection. Mcl-1 protein (but not mRNA) is reduced or lost upon infection with MVA/ VACV. This is in part due to the induction of Noxa, which in itself is an interferon-response gene that is induced in antiviral defence, for instance during infection with MVA (Eitz Ferrer et al. 22 and this study).
In the case of MVA, our data show that the viral protein It has been reported earlier that MVA, 22 VACV 33 and MCMV 34 all can induce apoptosis in macrophages (although there appeared to be some strain variation with VACV derivatives 35 ), while no or little apoptosis is induced (and apoptosis is typically inhibited) in fibroblasts or epithelial cells. The reason for this is molecularly unclear. One point may be that recognition and response to viral components are stronger in macrophages as professional immune cells. During MVA infection, recognition of viral RNA and the subsequent interferon response account for part of the apoptotic response. 22 This pro-apoptotic activity of viral infection may simply be stronger in macrophages.
However, two factors are also likely to contribute to the higher sensitivity of macrophages. First, many cells are more dependent on Mcl-1 than macrophages. Many cell lines die upon RNAi against Mcl-1 (see HeLa in Figure 2c), and although this may be linked to the malignant transformation of the cells, it is remarkable that macrophages are the only major lineage of immune cells that do not require Mcl-1. In macrophages, apoptotic stimuli may therefore rely more heavily on pathways not involving Mcl-1, and viral infection may be one such stimulus. Second, viruses appear to have difficulty in establishing protection against apoptosis in macrophages. Unlike during infection of fibroblasts, MVA was unable to protect Mcl-1-deficient macrophages against ABT-737; a sensitization to ABT-737 was even seen during VACV infection. Consistently, very little F1L could be detected during infection with MVA or VACV despite productive infection (in the case of VACV). Although it is unclear how this is achieved, a low expression of anti-apoptotic proteins may contribute to apoptosis induction in macrophages.
However, Bcl-X L turned out to be an important pro-survival factor in macrophages. In most situations and cell types where this has been directly compared, Bcl-X L was much less important than Mcl-1 in keeping the cells alive (see Introduction). The very clear apoptosis-enhancing effect we observed in macrophages upon the loss of Bcl-X L was therefore surprising. The findings suggest that during viral infection of macrophages a pro-apoptotic activity is generated that is to a substantial degree blocked by Bcl-X L but not by Mcl-1. What the nature of such a stimulus could be, and how it is transmitted to the mitochondria and the Bcl-2-family of proteins is unclear. There are some pro-apoptotic proteins that can be blocked by Bcl-X L but no by Mcl-1; most notably, some of the pro-apoptotic Bcl-2-family group of BH3-only proteins bind much more strongly to Bcl-X L than to Mcl-1. Bad, for instance has an affinity for Bcl-X L that is at least 100 000 times higher than for Mcl-1; for Bmf, the difference is about 100, for Bid about 25-fold (measuring binding of the BH3 domains 36 ). It is therefore at least conceivable that such proapoptotic molecules are involved. We did test macrophages deficient in Bim and Bmf; however, while Bim-deficient macrophages showed small protection against MVA-induced apoptosis (as reported earlier 22 ) additional loss of Bmf had no effect (data not shown).
The development of small molecules that specifically target individual proteins of the apoptosis apparatus in recent years has helped understand apoptosis and may contribute to future treatment of disease. ABT-737 is a molecule that binds to and inhibits Bcl-2, Bcl-X L and Bcl-w; 37 more specific molecules targeting only individual Bcl-2-family members are also available. The enhancement of virus-induced apoptosis in macrophages by ABT-737 confirms a pro-survival role of Bcl-X L (and perhaps Bcl-2) during viral infection. Although the enhanced apoptosis by deletion of Bcl-X L or by treatment with ABT-737 can explain the reduction in progeny generation observed, it has also been proposed that such manipulation may enhance autophagy, and enhanced autophagy might add to viral clearance and therefore also contribute to the reduction in virus generation. It has to be said however that this issue cannot be considered clarified at present. An apoptosisindependent contribution of anti-apoptotic Bcl-2 proteins to autophagy has been disputed. 38 More recently, this has been challenged by demonstrating that ABT-737 can induce autophagy in the absence of Bax and Bak, 39 but this has in a very recent study been suggested to be due to effects on components other than Bcl-2-like proteins. 40 This question therefore has to await further clarification.
It may be presumptuous at this stage to suggest that such inhibitors may be used in clinical settings, but a clear understanding of the pro-and anti-apoptotic processes occurring in infectious disease and in the different types of infected cells will help understand the biology and may indeed at some stage also provide a new handle on therapeutic approaches.
Hoxb8 macrophage progenitors were derived from bone marrow of mice homozygous for a conditional (loxP-flanked) Bcl-x, 41 Mcl-1, 42 Bim or Bim and Bmf allele and heterozygous for the LysM-cre-allele, 43 or from Noxa − / − or IFNAR − / − Mx1 mice (kindly provided by Peter Stäheli, Freiburg). All mice were backcrossed on the C57Bl/6 background but since they came from different breeding programmes the corresponding wt cells were for all lines derived at the same time and from littermates. Polyclonal precursor cell lines were established by retroviral transduction of Hoxb8 and selection in the presence of granulocyte macrophage colony-stimulating factor (GM-CSF). 29 After selection macrophage precursor cell lines were cultured in RPMI-1640 medium (#F1415, Merck Millipore, Darmstadt, Germany) containing 10% heat-inactivated FCS (#10270-106, Thermo Scientific), antibiotics, 1 : 100 diluted supernatant of GM-CSF producing B16 cells and 1 μM β-estradiol. Macrophage differentiation was induced by removal of β-estradiol, followed by culture for 7 days in medium containing 1% GM-CSF. All cultures were incubated under standard culture conditions (37°C, 5% CO 2 ).
Virus infections. For Modified Virus Ankara (MVA) or Vaccinia virus (VACV WR; kindly provided by Peter Aichele, Freiburg) infections, MEF and HeLa cells were seeded at 1.5 × 10 5 cells/well in 6-well plates the day before infection. At least 30 min before infection, medium was exchanged for 600 μl of infection medium without antibiotics and 2% FCS. MVA or MVAΔF1L 22 was diluted in infection medium, vortexed and sonicated for 1 min. Cells were infected with an MOI of 10 in 600 μl infection medium for indicated times.
Macrophages differentiation was started with 8 × 10 4 progenitor cells/well in a 6-well plate 7 days before the infection. Infection was conducted as above but in 1 ml volume.
Infection efficiency of VACV was determined by infecting cells with VACV-GFP (VACV B5R-EGFP, kindly provided by Peter Aichele, Freiburg) for 5 h and performing flow-cytometry analysis of the expression of viral GFP.
Infections with MCMV were performed as above. Addition of the virus was followed by centrifugation for 30 min at 870 g. MCMV was constructed as follows. MCMV dm157luc was constructed based on the MCMV-BAC pSM3fr-MCK-2fl (Jordan et al., 44 kindly provided by B Adler, Munich, Germany). Deletion of the m157 ORF and insertion of the luc sequence into the m157 locus were performed exactly as described. 45 Infectious MCMV was reconstituted by the Superfect transfection (Qiagen, Hilden, Germany) procedure. BAC sequences were removed by serial passage in MEF before use of the recombinant virus. Infection efficiency was determined by staining infected cells with CROMA 101, an antibody against viral IE-1 (pp89) 46 protein before incubating with the secondary DyLight 488 conjugated antibody (#115-485-062, Dianova, Hamburg, Germany, 1 : 300). Staining was performed as described below.
Plaque assay. The whole plate of macrophages infected with VACV including supernatant was frozen at − 80°C to detach and lyse the cells. After thawing, 200 μl of a dilution series in DMEM/2% FCS was added in duplicates to BSC-40 cells, which had been seeded 1 × 10 5 cells/well in 24-well plates the day before. After 2 h of incubation at 37°C, 800 μl of a 1 : 1 mixture of DMEM/10% FCS and 2% methylcellulose was added and incubated for 16 h at standard culture conditions. Medium was removed and cells fixed and stained with 250 μl crystal violet for 20 min. After washing with water and drying, plaques could be counted.
Mcl-1-specific RNAi and inhibitor. HeLa cells were seeded the day before transfection. siRNA (20 nM final concentration) was mixed with Lipofectamin