Multiple Autonomous Cell Death Suppression Strategies Ensure Cytomegalovirus Fitness

Programmed cell death pathways eliminate infected cells and regulate infection-associated inflammation during pathogen invasion. Cytomegaloviruses encode several distinct suppressors that block intrinsic apoptosis, extrinsic apoptosis, and necroptosis, pathways that impact pathogenesis of this ubiquitous herpesvirus. Here, we expanded the understanding of three cell autonomous suppression mechanisms on which murine cytomegalovirus relies: (i) M38.5-encoded viral mitochondrial inhibitor of apoptosis (vMIA), a BAX suppressor that functions in concert with M41.1-encoded viral inhibitor of BAK oligomerization (vIBO), (ii) M36-encoded viral inhibitor of caspase-8 activation (vICA), and (iii) M45-encoded viral inhibitor of RIP/RHIM activation (vIRA). Following infection of bone marrow-derived macrophages, the virus initially deflected receptor-interacting protein kinase (RIPK)3-dependent necroptosis, the most potent of the three cell death pathways. This process remained independent of caspase-8, although suppression of this apoptotic protease enhances necroptosis in most cell types. Second, the virus deflected TNF-mediated extrinsic apoptosis, a pathway dependent on autocrine TNF production by macrophages that proceeds independently of mitochondrial death machinery or RIPK3. Third, cytomegalovirus deflected BCL-2 family protein-dependent mitochondrial cell death through combined TNF-dependent and -independent signaling even in the absence of RIPK1, RIPK3, and caspase-8. Furthermore, each of these cell death pathways dictated a distinct pattern of cytokine and chemokine activation. Therefore, cytomegalovirus employs sequential, non-redundant suppression strategies to specifically modulate the timing and execution of necroptosis, extrinsic apoptosis, and intrinsic apoptosis within infected cells to orchestrate virus control and infection-dependent inflammation. Virus-encoded death suppressors together hold control over an intricate network that upends host defense and supports pathogenesis in the intact mammalian host.


Introduction
In the arms race between eukaryotic hosts and pathogens, programmed cell death (PCD) mechanisms evolved as a network of pathways primarily intended to sustain health by resisting infection [1][2][3][4][5]. PCD pathways play roles in host defense against intracellular pathogens by eliminating cells before the production of progeny occurs. A primary form of PCD is apoptosis, a caspase (CASP)-dependent death pathway initiated either by infection-associated external danger signals (extrinsic) or by intracellular stress sensed via mitochondria (intrinsic) [5][6][7]. An alternate form of cell death, necroptosis, unleashes in the face of virus-encoded CASP inhibitors and has been characterized as a highly potent host defense pathway known to limit the ability of a pathogen to invade the host [8][9][10]. Cytomegalovirus (CMV) has evolved to be a ubiquitous, highly successful mammalian pathogen by employing an array of conserved immunomodulatory strategies that subvert anti-viral defenses, particularly through the effective suppression of both extrinsic and intrinsic PCD pathways [10][11][12][13][14][15][16][17][18][19]. These suppression strategies sustain human (H) and murine (M) CMV replication in cells and, most importantly, ensure fitness in vivo [10,[12][13][14][15][16][17][18].
Here we showed that combined BAX/BAK suppression by vMIA/vIBO prevents intrinsic PCD, limits HtrA2/Omi impact, and regulates innate inflammation. vMIA/vIBO remain essential for sustained MCMV titers in the salivary glands of mice, the tissue known to mediate horizontal transmission of CMV. Macrophages represent a relevant, natural setting where MCMV-encoded PCD suppressors show functions that correspond to behavior in infected hosts [30]. Macrophages and their monocyte precursors play central roles during CMV pathogenesis by sustaining viral replication to support reservoirs that underly life-long persistence and latency [64][65][66][67][68][69][70][71]. We showed that MCMV faces receptor interacting protein (RIP) kinase (RIPK)3-dependent necroptosis within~6 to 9 h, TNFR1dependent CASP8 activation within~12 h, and mitochondrial cell death within~24 h of infection. Thus, PCD pathways are temporally staged and the suppression of each PCD pathway proceeds independently based on studies that rely on mutant macrophages that lack specific cell death components. MCMV gene products, M45-encoded viral inhibitor of RHIM activation (vIRA), M36-encoded vICA, and M38.5/M41.1 encoded vMIA and vIBO work in a concerted manner during infection with each suppressor playing a largely non-redundant role in sustaining viability, facilitating virus replication, and orchestrating infection-associated inflammation. vMIA/vIBO also suppresses the amplification of TNF signaling via mitochondria and likely underlies observations showing M36 mutant MCMV does not activate this sub-pathway [72]. The interfacing of the mitochondrial suppressors with the TNF signaling brings to light new insights regarding the role of this inflammatory cytokine in CMV biology, revealing impacts on replication in a variety of cell types as well as pathogenesis in vivo [73][74][75][76].
Given that vMIA/vIBO function plays out independently of extrinsic apoptotic or necroptotic players, MCMV employs sequential, specific suppression mechanisms with non-overlapping functions to establish a protective shield against host immunity. This prevents viral clearance and ensures fitness. Our work sheds light on the potential for targeting the CMV-encoded PCD suppressors as well as the associated host pathways as intervention strategies to reduce pathogen load and disease pathogenesis.
A second double mutant virus was generated to confirm the results produced by ∆M38.5/m41.1-BAC and to identify differences that may be due to the insertion cassette disrupting M38.5 as well as neighboring genes. The second double mutant virus, M38.5.StopFS/M41.1.∆Start-BAC, was also generated from M41.1.∆Start-BAC with the more efficient En Passant mutagenesis system [85,86]. A point mutation was engineered into M38.5, creating a stop codon and frame shift mutation just eight amino acids from the start of the protein. Furthermore, the point mutation created an XbaI restriction site used for screening and characterization of the mutant virus. The following primers were used to insert this point mutation: M38.5.StopFS-F(5 gcgccccagaggcgaagagcagcgctggtcgttcgcttacaaac cctctagaagggtcggcgcacaggatgacgacgataagtaggg) and M38.5.StopFS-R(5 gtagttgtggaggggaca gcgatggagagtgtgcgccgacccttctagagggtttgtaagcgaaccaaccaattaaccaattctgattag). Finally, a single M38.5 mutant virus, M38.5.StopFS-BAC, was generated from the parental K181-BAC with the same point mutations using the primers described above.
All BAC-derived clones were screened by sequencing PCR products over the M41/M41.1 locus as previously described [38] as well as over the M38.5 locus. The following primers were used to amplify and screen the M38.5 locus: M38.5-F(5 gcagaagtcacgtcggatccag); m38.5-R(5 ggctgctacgagaacgtgac). Furthermore, BAC-derived full-length clones were analyzed by RFLP with five restriction enzymes to ensure the integrity of the viral genome. Following genetic characterization, mutant and parental viruses were reconstituted and plaque-purified as previously described [38]. Growth curves and viral yields were performed by infecting cells in a 48-well plate in 0.25 mL of viral inoculum at the indicated multiplicity of infection (MOI) for 1 h at 37 • C with 5% CO 2 . Following adsorption, the inoculum was replaced with complete DMEM. Samples were harvested at the indicated times, sonicated, and titered by plaque assay on 3T3-SA fibroblasts.

Mice, Infections, and Organ Harvests
BALB/c, B6 × 129 P2, and C57BL6/J mice were obtained from Jackson Laboratory (Bar Harbor, ME USA). B6 × 129-Bax tm2Sjk Bak1 tm1Thsn/J (also referred to as Bak −/− Bax fl/fl ) were obtained and used as previously described [38]. Casp8 −/− RIPK3 −/− were bred and maintained as previously described [87]. Six-week-old BALB/c mice were infected via the intraperitoneal (IP) route of inoculation with 2 × 10 5 PFU of tissue culture-derived virus. Other strains of mice (6-18 weeks old) were infected via the IP route or via footpad route with 1 × 10 6 PFU or 5 × 10 6 PFU of virus respectively. At the time of sacrifice, organs were placed in 1 mL of complete DMEM, disrupted by sonication, and virus titer was determined by plaque assay as described [30]. IE1 positive cells were identified as described [30].

Cell Viability Assay
At the indicated times, cell viability was determined by measuring the ATP levels using the Cell Titer-Glo Luminescent Cell Viability Assay kit (Promega Corporation, Madison, WI, USA) according to manufacturer's protocol. Cells (~50,000 cells/well) were seeded into 96-well plates in triplicate. Approximately 18 h post seeding, medium was replaced with 100 µL of virus inoculum containing 10 PFU/cell (MOI = 10). At 24 hpi, and DNA fragmentation was visualized by the DeadEnd Fluorometric TUNEL System (Promega Corporation, Madison, WI, USA). Cells (60,000/well) were seeded into 48-well plates in triplicate. Approximately 18 h post seeding, cells were infected at MOI = 10. The number of cells with detectable fragmented DNA were expressed as a percent of cells expressing IE1. To assess cell morphology, cells were plated in 48-well plates as indicated above and infected with virus at MOI = 20 at 18 h post plating. Phase-contrast microscopy was performed 24 hpi. To assess the inclusion of cell permeable SYTOX, Green dye (S7020, Invitrogen, Waltham, MA, USA) uptake was used in real time using an IncuCyte instrument (Essen BioScience, Ann Arbor, MI, USA) as described [82]. BMDM were plated at 5 × 10 4 cells per well in 24-well tissue culture plates. Cells were infected as described above at MOI = 10.

Cytokine Array
Released cytokines and chemokines were detected by cytokine array (Cytokine Array Panel A, ARY006, R&D Systems, Minneapolis, MN, USA). Supernatants were collected from cultures of WT BMDM infected with MCMV (K181 or mutant) for 24 h. Following centrifugation at 500 g for 5 min at 4 • C, cell-free supernatants were pooled from two independent experiments for each condition. Array strips were exposed to chemiluminescent chemicals (provided with the kit) and raw images were developed on X-ray film. Films were analyzed using Adobe Photoshop 8 to generate pixel intensity of dots corresponding to individual cytokines as well as reference dots. Pixel intensities of control dots were averaged for each array strip. For comparing four array strips (heat map in Figure 3), reference spots were normalized across the dot blots.

Statistics
All statistical analyses were performed using GraphPad Prism 8 Software (GraphPad Software Inc. La Jolla, CA, USA). Two groups of data were compared for significance using non-parametric, Wilcoxon matched-pairs signed rank test; three or more groups of data were compared using ANOVA. All graphs show standard error and mean for datasets, except viral titers in organs where mean is indicated for each data set. p-values (p) of <0.05 were considered significant and indicated with * <0.05, ** <0.01, *** < 0.001, **** <0.0001.

Software
Chemiluminescent images of immunoblots were captured using the Kwik Quant system (Kindle Biosciences, Greenwich, CT, USA) and analyzed on Adobe Photoshop 8 (Adobe Corporation Inc., San Jose, CA, USA). All cell death as well as ELISA data were assembled using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) and graphed using GraphPad Prism 8 (GraphPad Software Inc. San Diego, CA, USA). For cytokine array, pixel intensities were determined using Adobe Photoshop 8. All figures were assembled using Adobe Illustrator 8. Software Inc. San Diego, CA, USA). Model cartoons were constructed using BioRender software (BioRender, Toronto, ON, Canada).

Data Availability
Further information and requests for resources and reagents should be directed to and will be fulfilled by mocarski@emory.edu and pratyusha.mandal@emory.edu. All data are included in manuscript. All raw files are available upon request. This manuscript generated a new MCMV mutant (∆M38.5/M41.1 double mutant virus) that is available upon request. All reagents utilized are described in detail in these methods.

Study Approval
All experiments were conducted with approval (4 February 2019, Proto 201700351) from the Emory University Biohazard, Chemical Hazard Review and Animal Use (IACUC) Committees.

M38.5-Encoded vMIA and M41.1-Encoded vIBO Support Viral Fitness
To determine the impact of combined BAX/BAK suppression by M38.5-encoded vMIA and M41.1-encoded vIBO, we compared the replication of mutant viruses to parental K181 ( Figure 1A-C). The genomic integrity of double mutants and control viruses were confirmed by direct evaluation ( Figure S1A), as previously shown for single mutant viruses [29,30,38]. MCMV was shed in saliva mediates horizontal transmission [88], so viral load in these glands may therefore be considered an indicator of viral fitness. MCMV detection in salivary glands requires entry followed by replication in cells at the origin of infection, myeloid-cell mediated dissemination to different tissues including the salivary glands, and ultimate replication salivary glands [88][89][90][91][92][93]. Virus-encoded modulators of adaptive and innate immunity (including cell death suppressors) suppress host defense to influence infection at different stages in dissemination [30,93,94]. We determined the titers of mutant viruses described in the salivary glands of BALB/c mice inoculated intraperitoneally and C57BL6/J mice inoculated via footpad at 14 days post infection (dpi) ( Figure 1B,C). At 14 dpi, vMIA/vIBO-deficient viruses (∆M38.5/M41.1 and M38.5.StopFS/M41.1.∆Start) showed titers that averaged~1000-fold lower than K181 in the salivary glands of highly susceptible BALB/c mice ( Figure 1B). Viral titers were markedly more variable and lower in most animals than observed with single mutant viruses (M38.5.StopFS or M41.1.∆Start), which exhibited a modest compromise compared with K181 [30,38]. The variability of double mutant viruses contrasts the invariant titers of K181, or single mutants, reflecting the importance of BAX and BAK suppression to sustaining MCMV fitness. All mutants (single or double) remained severely compromised when titers were assessed in salivary glands of C57BL6/J mice following footpad inoculation ( Figure 1C). Together, these observations revealed that the vMIA/vIBO double mutant MCMV is attenuated for dissemination to or replication in salivary glands, highlighting the importance of BAX/BAK suppression for viral fitness in vivo. Horizontal lines indicate data compared by Wilcoxon matched-pairs signed rank test. For panels D-G, error bars indicate the standard error and mean for each dataset. * is p < 0.05; ** is p < 0.01; **** is p < 0.0001, n.s. is non-significant.
To better understand the infection pattern in different cell types when BAX/BAK were not suppressed, we evaluated the replication of double-deficient viruses in cultured primary macrophages, and immortalized fibroblasts as well as endothelial cells, all of which support parental K181 replication but show differential susceptibility to various MCMV mutants [30] consistent with cell type-specific function of virus-encoded cell death suppressors (Figures 1D-F and S1B-E). ∆M38.5/M41.1 and M38.5.StopFS/M41.1.∆Start viruses showed severe attenuation in BMDM from wild type (WT) C57BL/6J mice ( Figure 1D), a pattern that was confirmed for ∆M38.5/M41.1 over a nine-day time course ( Figure S1B). Double-mutant virus did not increase in titer over this time course. Surprisingly, the infection of ∆M38.5/M41.1 in BMDM exhibited reduced IE1+ cells over MOI range from 0.626 through 10 ( Figure S1C) at 24 h post infection (hpi), even though no defect in IE1 expression was observed for either M38.5 or M41.1 single-mutant viruses [31,32]. Importantly, the M38.5/M41.1 double-mutant viruses were constructed utilizing the original K181 background strain ( Figure 1A) that displays no defect in entry for myeloid cells, unlike other pSM3fr-derived MCMV [95]. The reason vMIA/vIBO mutant MCMV shows this defect will require further investigation. Immortal 3T3-SA fibroblasts exhibited modestly lower vMIA/vIBO mutant viral titers at 24 through 72 hpi even though titers increased by 5 dpi to match K181 ( Figures 1E and S1D). SVEC4-10 cells exhibited comparable levels of K181 and mutant virus replication ( Figures 1F and S1E). Thus, in culture settings, mitochondrial suppression by vMIA and vIBO contributes most dramatically to viral replication in macrophages.

vMIA and vIBO Interface with vIRA and vICA
To evaluate the temporal aspects of PCD suppression by vIRA, vICA, vMIA and vIBO, we compared the parental virus to M45mutRHIM (inducer of RIPK3-dependent necroptosis), ∆M36 (inducer of CASP8-dependent apoptosis), and ∆M38.5/M41.1 viruses during infection of WT macrophages (Figures 3E and S3C). Cell death was measured by integration of specific fluorescent signal of cell-impermeable dye Sytox Green as the execution of PCD signaling increases membrane permeability. Neither mock treatment nor K181 infection induced permeability. M45mutRHIM-infected cells exhibited a rapid increase in cell permeability within 6 to 9 hpi. Necroptosis was evident by 12 hpi (Figures 3E and  S3C, third row). ∆M36 and ∆M38.5/M41.1 mutant viruses revealed comparable timing of cell death, first becoming evident by 12 to 15 hpi ( Figure 3E). ∆M38.5/M41.1-induced cell death remained overall less prominent compared with that induced by ∆M36 infection ( Figure S3C fourth and fifth rows). This relatively slow development of PCD with vMIA/vIBO double-deficient viruses was expected for infection-triggered mitochondrial stress as observed with HCMV vMIA mutant or MCMV single mutant disrupting M38.5 or M41.1 expression [13,17,31,36,38]. Therefore, M45mutRHIM-induced necroptosis represents the most robust and immediate form of cell death in MCMV-infected BMDM, followed by ∆M36-induced extrinsic apoptosis and ∆M38.5/M41.1-induced mitochondrial cell death. Elaboration of PCD suppression during MCMV infection of macrophages controls this temporal network. Each suppressor functions independently of the others with a feed-forward crosstalk between pathways as indicated by the contributions of TNF to the mitochondrial signaling ( Figure 3A,B) or BAX and BAK to necroptotic signaling ( Figure S3B), as well as the role of CASP8 suppression in the induction of necroptosis [30].

vMIA and vIBO Interface with vIRA and vICA
To evaluate the temporal aspects of PCD suppression by vIRA, vICA, vMIA and vIBO, we compared the parental virus to M45mutRHIM (inducer of RIPK3-dependent  Figure S4). For each mutant infection setting, references are normalized with respect to K181 infection.

Ripoptosome Components Are Dispensable in vMIA/vIBO-Dependent Signaling
To identify crosstalk between extrinsic and intrinsic cell death components during MCMV infection, we assessed cytosolic death players RIPK1 and CASP8 because of their key roles in PCD signaling [102][103][104][105][106]. Extracellular or intracellular stress often induces the association of adaptor FADD with RIPK1 and CASP8, leading to the formation a multi-functional, pro-death platform known as the ripoptosome [106][107][108][109]. RIPK1 engagement has been observed downstream to both extrinsic TNF-dependent and intrinsic BAX/BAK-mediated signaling [110][111][112]. MCMV infection engages RIPK1, CASP8, and RIPK3 in BMDM [29,30], but how these proteins play out in the context of vMIA/vIBO-dependent signaling remains unresolved. RIPK3 and ZBP1 may contribute to ripoptosome function. We infected WT,  [29] or mitochondrial cell death (CASP9-CASP3-CASP7) [113,114] at 24 hpi ( Figure 4C). WT, but not Ripk1 −/− Casp8 −/− Ripk3 −/− cells infected with K181 or M45mutRHIM, differed from mock infected cells by a detectable increase in levels of cleaved CASP7 (~19 and 17 kDa). Thus, MCMV infection activates CASP7 independently of vIRA, vICA, and vMIA/vIBO, but is dependent on ripoptosome components. However, this activated CASP7 did not result in the death of WT cells during infection with K181. Infection with ∆M36 resulted in the expected appearance of cleavage-activated CASP8 (~43 and 18 kDa), CASP7 (17 kDa), and CASP3 (17 kDa) cleavage products. Activated CASP9 (~39-37 kDa) levels were also detected, indicating that this mitochondrial cell-death-associated caspase was engaged during ∆M36 infection, even though mitochondrial BCL2-family proteins were dispensable for the death [29]. These data clearly showed that removal of vICA-dependent CASP8 suppression has a cascade impact on activation of multiple caspases beyond CASP8. As expected, Ripk1 −/− Casp8 −/− Ripk3 −/− cells that remained completely resistant to ∆M36-induced apoptosis [30] resisted the activation of CASP8 and CASP3, but surprisingly continued to exhibit activation of CASP9 and CASP7. It remains surprising that Ripk1 −/− Casp8 −/− Ripk3 −/− cells showed full activation of proapoptotic caspases despite the absence of core ripoptosome components. However, CASP9/CASP7 activation in these cells did not drive death in the absence of these critical ripoptosome components. With ∆M38.5/M41.1 infection, WT cells exhibited activation of CASP9, CASP7, and CASP3. CASP8 was activated over K181 but in reduced levels compared to ∆M36, consistent with the requirement for vICA in CASP8 suppression. Ripk1 −/− Casp8 −/− Ripk3 −/− cells also activated CASP9, CASP7, and CASP3, but at levels modestly reduced compared with WT cells. Death did not differ between the genotypes ( Figure 4A,B), demonstrating that the lower levels of the CASPactivation pattern expressed in Ripk1 −/− Casp8 −/− Ripk3 −/− cells were sufficient to induce mitochondria-dependent death signaling. The lower levels of CASP8 activation may reflect the contribution from TNFR1 ( Figure 3B), which is known to signal through RIPK1 and CASP8 [109]. Therefore, extrinsic death signaling does not impact double mutant-induced mitochondrial death signaling. vMIA/vIBO double-deficient virus maintained attenuation in Casp8 −/− Ripk3 −/− BMDM ( Figure S5A-C). In salivary glands from Casp8 +/− Ripk3 −/− mice, this attenuation was evident at 14 dpi ( Figure S5D). However, in Casp8 −/− Ripk3 −/− mice, double-deficient MCMV exhibited a spread in viral titer, indicating TNF-CASP8dependent signaling may amplify the innate immune response to restrict virus infection in vivo. Therefore, vMIA and vIBO target the mitochondria to prevent cell death and benefit fitness independently of known extrinsic PCD signaling components and pathways. Other PCD signaling pathways potentially influence outcomes in vivo. In summary, we established that the mitochondria are central sensors of MCMV infection via multiple processes. Mitochondrial serine protease HtrA2/Omi restrains replication ( Figure 5) while HtrA2/Omi-independent signaling engages PCD signaling. M38.5/M41.1-encoded vMIA and vIBO suppress mitochondrial signaling by: a) modestly interfering with HtrA2/Omi and b) preventing cell death through suppression of BAX/BAK-dependent signaling. Overall, MCMV suppresses PCD in the follow ways: M45 (vIRA) suppresses RIPK3-dependent necroptosis, with ZBP1 contributing in non-myeloid cells but not in macrophages; M36 (vICA) suppresses the TNF signaling-dependent CASP8-CASP3 activation axis, as well as the CASP8-independent cleavage-activation of CASP9 and CASP3; vMIA and vIBO together suppress mitochondrial signaling. vIRA, vICA, or vMIA/vIBO target distinct host proteins blocking non-redundant PCD pathways. Our data reveal crosstalk between different PCD pathways in macrophages. TNFR1 contributes to the mitochondrial cell death observed in absence of vMIA and vIBO; BAX and BAK restrained necroptosis in the absence of vIRA, suggesting cell death initiated by In summary, we established that the mitochondria are central sensors of MCMV infection via multiple processes. Mitochondrial serine protease HtrA2/Omi restrains replication ( Figure 5) while HtrA2/Omi-independent signaling engages PCD signaling. M38.5/M41.1-encoded vMIA and vIBO suppress mitochondrial signaling by: (a) modestly interfering with HtrA2/Omi and (b) preventing cell death through suppression of BAX/BAK-dependent signaling. Overall, MCMV suppresses PCD in the follow ways: M45 (vIRA) suppresses RIPK3-dependent necroptosis, with ZBP1 contributing in non-myeloid cells but not in macrophages; M36 (vICA) suppresses the TNF signaling-dependent CASP8-CASP3 activation axis, as well as the CASP8-independent cleavage-activation of CASP9 and CASP3; vMIA and vIBO together suppress mitochondrial signaling. vIRA, vICA, or vMIA/vIBO target distinct host proteins blocking non-redundant PCD pathways. Our data reveal crosstalk between different PCD pathways in macrophages. TNFR1 contributes to the mitochondrial cell death observed in absence of vMIA and vIBO; BAX and BAK restrained necroptosis in the absence of vIRA, suggesting cell death initiated by mutant viruses benefits from host PCD proteins involved in other forms of cell death as a feedforward loop. Therefore, MCMV orchestrates an efficient arsenal of sequential, independent, and non-overlapping suppression strategies against the interconnected mammalian PCD benefitting fitness during natural infection of cells or intact hosts.

021, 13, x FOR PEER REVIEW 16 of 25
feed-forward loop. Therefore, MCMV orchestrates an efficient arsenal of sequential, independent, and non-overlapping suppression strategies against the interconnected mammalian PCD benefitting fitness during natural infection of cells or intact hosts.

Discussion
Here, we reveal an intricate interplay between virus-encoded PCD suppressors, mammalian PCD machinery, and TNF-dependent signaling. Cellular pathways may collaborate to undermine cell viability, cut short virus replication, and drive infection-associated inflammation in the absence of dedicated MCMV-encoded PCD suppressors. We introduce new concepts regarding the involvement of mitochondria in this infected cell deathinflammation matrix. Mitochondrial HtrA2/Omi restrains MCMV replication, reminiscent of its role in restricting HCMV infection in fibroblasts via the induction of PCD [13,49]. However, in MCMV-infected BMDM, this serine protease limits infection independently of PCD. Recently, it has been demonstrated that HCMV vMIA suppresses BAX/BAKdependent signaling to facilitate the infection of mast cells [37]. Future investigations with HCMV and MCMV in myeloid and non-myeloid lineages will be necessary to resolve potential differences between HCMV studies and our current observations with MCMV. Tissue-specific contributions from HtrA2/Omi or differences between the human and murine viruses might certainly be expected given the ninety million years of evolutionary divergence in these two biologically similar viruses. Overall, we add MCMV vMIA/vIBO as a modest suppressor of HtrA2/Omi to the known herpesvirus-encoded suppressors, HCMV-encoded vMIA [13] and HHV-8-encoded vIRF1 [115]. Studies with MCMV single mutant for either vMIA or vIBO have established BAX and BAK as specific targets for the viral suppressors [31,38]. We expanded that understanding by implicating HtrA2/Omi and TNF-dependent signaling as additional targets. vMIA/vIBO double-deficient MCMV failed to efficiently express the immediate early viral protein IE1 in macrophages ( Figure S1C). The other cell death suppressor, mutant MCMV generated on this K181-BAC parental background, as well as single mutants for vMIA or vIBO did not show a similar defect [30][31][32], indicating vMIA/vIBO may potentially regulate the initial steps controlling entry prior to immediate early gene expression. Future studies in primary and immortalized murine cells will be necessary to determine the full scope of contribution by these suppressors during infection. During infection, vMIA/vIBO suppress an early involvement of TNF beyond BAX/BAK such that the double-mutant virus drives TNFR1-dependent death by 24 hpi (Figure 3B). In Casp8 −/− Ripk3 −/− but not Casp8 +/− Ripk3 −/− mice, vMIA/vIBO mutant MCMV exhibited a spread in replication ( Figure S5D). TNF signaling showed an intricate association with both HCMV and MCMV infection [12,13,17,18,29,75,[116][117][118], although never before implicated in mitochondrial signaling. Our observations indicated that TNF-dependent signaling augments vMIA/vIBO mutant MCMV-driven death in macrophages and amplifies response in mice via CASP8 activation.
Surprisingly, TNF-driven death during vMIA/vIBO-deficient mutant infection occurs independently of CASP8 and despite the presence of virus-encoded vICA. Thus, vICA successfully restricts the known TNFR1-CASP8 death signaling. TNFR1 plays a role early in infection, but BAX/BAK activity bypasses any contribution from this cytokine later in infection. Bax −/− Bak −/− BMDM exhibit a partial rescue from double mutant-induced death. Elimination of TNFR1-dependent signaling along with BAX/BAK must be evaluated for impact on cellular viability and in vivo replication during double-mutant virus infection. From the data shown here, it appears that overlapping intrinsic and extrinsic PCD signaling pathways may converge in the mitochondria late during MCMV infection where vMIA/vIBO mitigates impacts from both pathways. BAX/BAK-interacting partners BID, BIM, and PUMA are individually dispensable but contribute together, consistent with the potential diversity of BAX and BAK activators as infection progresses. Whether HCMV vMIA elicits a similar suppression strategy against both extrinsic and intrinsic death signaling will come from future assessment. TNF is known to induce extrinsic death via cytosolic FADD-RIPK1-CASP8 association [119], with mitochondrial amplification in BID-dependent as well as BID-independent signaling [57,58,[120][121][122][123]. RIPK1 and CASP8 are elevated by 14 hpi in infected macrophages [29] and possibly contribute to the early TNFR1 impact but remain ultimately uninvolved in the pathway suppressed by vMIA/vIBO. The role of BID, the recognized bridge between TNF and mitochondrial amplification [44,57,58,120,122,123], remains unclear. The picture that emerges regarding vMIA/vIBO suppression seems reminiscent of adenovirus-encoded viral BCL-2 antagonist E1B 19K, a function that suppresses analogous TNF-mitochondria synergy during infection [124]. There may be a common pattern during DNA virus infections. Our observations demand future experiments to elucidate how the TNF-mitochondria synergy plays out in cells as well as in vivo, especially during MCMV pathogenesis settings where TNF signaling contributes to outcomes [74,75,125].
Even though the host players for extrinsic apoptosis, necroptosis and mitochondrial death, exhibit limited crosstalk, a remarkable partitioning exists between the pathways MCMV-encoded PCD suppressors vIRA, vICA, and vMIA/vIBO hold under control. MCMV has evolved to antagonize amplification of the antiviral response as infection progresses such that each suppressor functions non-redundantly in particular subcellular location. Immediately following the infection of macrophages, MCMV faces RIPK3 in the cytoplasm. vIRA, a tegument-associated protein, is expressed early to usurp RIPK3 RHIM engagement [17,126]. vIRA also disrupt RIPK1, ZBP1, and TRIF RHIM-dependent signaling [14,72,82]. However, all these players, as well as TNF and types I and II interferons, remain dispensable for vIRA function during MCMV infection of macrophages ( Figures 3A and S3B and [30]). In fibroblasts or endothelial cells, the nucleic acid sensor ZBP1 initiates this signaling [14,15,30]. Therefore, even though vIRA-dependent RIPK3 suppression is critical for MCMV infection and pathogenesis independent of cell type [30], the requirements to trigger this signaling in macrophages are less stringent than in nonmyeloid cells. Restricting virus in macrophages may be of critical importance for the host such that additional requirements beyond RIPK3-MLKL are bypassed. BAX/BAK surprisingly restrain necroptosis unleashed in the absence of vIRA such that vIRA-deficient MCMV drives enhanced death in BAX/BAK-deficient cells ( Figure S3B). This feedback loop in combination with our identified TNF-mitochondria crosstalk reveal that extrinsic apoptosis, necroptosis, and mitochondrial cell death overlap during MCMV infection. The suppressors function in a non-overlapping fashion to subvert the integrated network of host PCD pathways. vIRA functions are followed by requirements for vICA in cytoplasm and vMIA/vIBO in mitochondria as host PCD pathways become activated. Pyroptosis remains dispensable for all these processes ( Figure S3A and [29]). Thus, MCMV subverts distinct PCD pathways by sequential activation of PCD suppressors that target distinct PCD pathways in different cellular localizations.
The suppressor-PCD interface during MCMV infection influences innate inflammation such that by 24 hpi, distinct patterns of cytokine/chemokine release became evident in the absence cell death suppression ( Figures 3F and S4). vIRA-deficient virus exhibited an overall hypo-inflammatory phenotype. In mice, this mutant virus associates with elevated innate and adaptive responses [30]. MCMV-infected cells undergoing necroptosis may be inherently inflammatory by releasing danger signals, but it remains to be determined whether the balance of signaling undermines viability or protects surrounding cells. Remarkably, vICA-deficient virus infection also exhibited a hyper-inflammatory phenotype. This could be due to either of two reasons. This apoptotic signal may itself be inflammatory. By suppressing CASP8, vICA blocks both death-dependent as well as death-independent inflammation that appear to be unleashed during ∆M36 infection. We previously suggested this possibility because vICA-deficient virus infection induced higher levels of TNF early during infection [29]. At the time of assessment (24 hpi), vMIA/vIBO double-deficient virus infection produced a hypo-inflammatory state, suggesting that unleashed BAX and BAK reduce the amplification of inflammatory signaling. Thus vMIA/vIBO appear to be potentiators of inflammation late during MCMV infection, presumably because this benefits the virus in some way. The consequences of each suppression mechanism on pathogenesis require further assessment.
In summary, we showed that PCD pathways restrict MCMV infection in macrophages. Sequential engagement of vIRA, vICA, and vMIA/vIBO, suppressors of RIPK3-dependent necroptosis, TNFR1-CASP8-dependent apoptosis, and mitochondrial signaling, respec-tively, regulate levels of inflammation in cultured macrophages and in vivo. Combined functions of all suppressors establish a protective shield for CMV infection, thereby establishing these suppressor pathways as potential therapeutic targets during viral pathogenesis. Deactivation of one or more PCD suppressors will restrict virus infection. Elimination of the mitochondrial suppressors (vMIA/vIBO) reduces viral fitness and overall inflammation. Given that the functions of these suppressors are conserved between MCMV and HCMV, we reveal that targeting the mitochondrial suppressors of CMV may be a safe, viable way to target this virus. Horizontal lines indicate the two groups being compared in each setting utilizing Wilcoxon matched-pairs signed rank test. Error bars indicate standard error and mean for each dataset. ** is p < 0.01; *** is p < 0.001; **** is p < 0.0001; n.s. is non-significant. (C) Representative light microscopy images (lower left scale bar = 200 µM) of WT BMDM infected with indicated viruses at indicated times. Time stamp from microscope is included in lower right corner of each image. Cell death is measured by inclusion of cell-impermeable dye (Sytox Green) as cellular membrane loses integrity due to execution of death signaling. Diffused circular green signal (detected in middle of lanes of fields without signal from cells) is a nonspecific consequence of the microscope light filter and camera. Figure S4: Dot blot proteome profiler array of supernatant from infected cells (related to Figure 3F). Cell-free supernatants from WT BMDM, infected collected 24 hpi with indicated viruses, were assessed by protein profiler array. A-F represent 6 rows where every two dots are replicates of distinct proteins. The names of the proteins are reflected in the map provide at lowest panel. Figure S5