Suppression of RIP3-dependent Necroptosis by Human Cytomegalovirus

Background: Viral suppressors of RHIM-dependent activation of pro-necrotic RIP3 kinase are crucial for successful infection in mice. Results: Human CMV blocks TNF-induced and murine CMV-induced necroptosis after RIP3 activation. Conclusion: Necrotic membrane leakage is blocked in infected cells despite the activation of MLKL. Significance: Viral inhibition of necroptosis will facilitate understanding of the final steps in this pathway. Necroptosis is an alternate programmed cell death pathway that is unleashed by caspase-8 compromise and mediated by receptor-interacting protein kinase 3 (RIP3). Murine cytomegalovirus (CMV) and herpes simplex virus (HSV) encode caspase-8 inhibitors that prevent apoptosis together with competitors of RIP homotypic interaction motif (RHIM)-dependent signal transduction to interrupt the necroptosis. Here, we show that pro-necrotic murine CMV M45 mutant virus drives virus-induced necroptosis during nonproductive infection of RIP3-expressing human fibroblasts, whereas WT virus does not. Thus, M45-encoded RHIM competitor, viral inhibitor of RIP activation, sustains viability of human cells like it is known to function in infected mouse cells. Importantly, human CMV is shown to block necroptosis induced by either TNF or M45 mutant murine CMV in RIP3-expressing human cells. Human CMV blocks TNF-induced necroptosis after RIP3 activation and phosphorylation of the mixed lineage kinase domain-like (MLKL) pseudokinase. An early, IE1-regulated viral gene product acts on a necroptosis step that follows MLKL phosphorylation prior to membrane leakage. This suppression strategy is distinct from RHIM signaling competition by murine CMV or HSV and interrupts an execution process that has not yet been fully elaborated.

Regulated cell death is important for eliminating cells during development and to guard the host from infection with microbial or viral pathogens (1)(2)(3). The mitochondrial pathway of apoptosis, mediated by the Bcl2 family proteins, contributes to both development and host defense. In humans, Casp8 4 and Casp10 initiate apoptosis triggered through TNF family death receptors, although this pathway may switch to an alternate receptor-interacting protein kinase 3 (RIP3)-dependent programmed necrosis outcome, called necroptosis, in cells with adequate levels of this pro-necrotic kinase (2, 4 -7). In mice, Casp8-dependent apoptosis and RIP3-dependent necroptosis contribute directly to host defense (6) and are together dispensable for development (8). Programmed necrosis was initially demonstrated in cultured cells subjected to Casp8 compromise after death receptor activation (9,10). The RIP3-mediated pathway was implicated in the innate immune response to vaccinia infection (11) and was shown to be physiologically relevant against the natural mouse pathogen murine cytomegalovirus (CMV). This betaherpesvirus specifically blocks necroptosis to sustain infection (12,13). In addition to death receptors and virus infection, Toll-like receptor (TLR) activation and interferon-dependent innate immune signaling unleash necroptosis (14 -16). Although there are strong parallels in all of these settings (17), most of the mechanistic understanding of this pathway derives from studies on TNF receptor 1 (TNFR1) signaling (8, 11, 18 -23).
Murine CMV M45-encoded vIRA also suppresses RHIM-dependent apoptosis (27,43,44) and regulates NF-B activation (26) independent of virus infection; however, the physiological importance of vIRA during infection is directed at necroptosis (12,13), which would otherwise eliminate infected host cells and halt infection (45,46). Thus, M45mutRHIM virus infection fails to infect either immunocompetent or immunodeficient mice (12,47) but proceeds in Rip3 Ϫ/Ϫ or Dai Ϫ/Ϫ (also called Zbp1 Ϫ/Ϫ ) mice where other immune control mechanisms remain operational. M45 carries out activities that are independent of RHIM function, increasing RIP1-dependent NF-B activation immediately following infection and preventing NEMO-mediated activation of NF-B later in infection (43,48,49). Recently, the RHIM-targeted suppressor strategy has been extended to HSV, where the large subunit of ribonucleotide reductase (R1) suppresses necroptosis in human cells (41) in addition to its role in suppressing Casp8-dependent apoptosis (50). Surprisingly, HSV triggers necroptosis in mouse cells, and this vulnerability is influenced by R1 RHIM signaling (51,52). In contrast, murine CMV vIRA suppresses cell death in both mouse and human cells (41). Expression of the HSV R1 during infection blocks Casp8 (50) and exposes cells to necroptosis (41), a dramatic interplay between one viral gene product and cognate-regulated cell death pathways (6,47). Human CMV UL36 (53) and murine CMV M36 (54,55) encode the viral inhibitor of Casp8 activation (vICA) (56,57). Casp8 suppression prevents apoptosis and is necessary for successful infection of monocyte-derived cell lineages (58) that control viral dissemination (59). In addition, both murine and human CMV encode evolutionarily conserved suppressors that prevent acti-vation of pro-apoptotic Bcl2 family members Bax and Bak (60 -66). These viral countermeasures reinforce the central role of cell death pathways in host defense against infection (6,58,67,68). In addition, the human CMV major immediate early 1 (IE1) gene product has been associated with cell death suppression (69) and plays a crucial role regulating levels of viral and host cell gene expression (70). An evolutionary dialogue seems to have played out over hundreds of millions of years between successful herpesviruses and their hosts such that perturbation of Casp8 and RIP3 may contribute to inflammatory disease driven by infectious and genetic insults (71,72).
Here, we demonstrate that human CMV blocks necroptosis induced by either TNFR1 activation or murine CMV infection in permissive human fibroblasts transduced with human RIP3 (hRIP3). RIP3 transduction confers sensitivity to necroptosis that is lost during cell propagation in culture, a condition that is reminiscent of cultured mouse fibroblasts (11,12). Human CMV confers IE1-dependent suppression of necroptosis at a step that follows RIP3 phosphorylation and activation of MLKL. This death is independent of the viral UL138 modulation of TNF signaling (73,74). Thus, human CMV prevents necroptosis through a mechanism distinct from either murine CMV or HSV. We establish the core parameters through which this alternate form of extrinsic cell death contributes to host defense in humans and add to the list of evolutionarily ancient, common cell signaling pathways against which human CMV deploys countermeasures.
Cells and Viruses-HFs and ihf-ie1.3, 293T, and 3T3-SA cells were cultured as described (12,70,77). HT-29, IMR-90, and human microvascular endothelial (HMEC-1) cells were maintained in DMEM containing 4.5 g/ml glucose, 10% fetal bovine serum (Atlanta Biologicals), 2 mM L-glutamine, 100 units/ml penicillin, and 100 units/ml streptomycin (Invitrogen). THP-1 cells from A. Kowalczyk (Emory University) were cultured and induced differentiation to macrophages as described previously (57). Human CMV strains were chosen to represent known genetic variants that arise during propagation in cell culture. Laboratory strain Towne-BAC, Towne⌬UL36, IE1-null virus (CR208), and its repaired virus (CRQ208) were described previously, as was the parental virus TownevarRIT 3 used in their preparation (57,70,78). Laboratory strain AD169varATCC and low passage strain Merlin strain were obtained from American Type Culture Collection, variant AD169-BAC was from J. Munger (Rochester University, New York, USA), variant AD169varDE was from M. Mach (Institute for Virologie, Erlangen, Germany), low passage strain Toledo was from S. Plotkin (Philadelphia, Pennsylvania, USA), and low passage strain endotheliotropic TB40E-BAC4 was from C. Sinzger (Institute for Virology, Tubingen, Germany). In experiments utilizing UV-irradiated virus, the viral suspension was exposed to 254-nm light at 360 mJ/cm 2 with a model XL-1500 Spectrolinker UV cross-linker (Spectronics Corp.) prior to infection. Virus titers were determined by a plaque assay on HFs or ihf-ie1.3 cells. BAC-derived parental murine CMV K181-BAC and M45mutRHIM viruses were described previously (12). All human or murine CMV infections were performed at an MOI of 3 or 10, and cells were incubated for 1 h with virus before changing the medium.
Cell Viability Assay-Cells (5,000 cells/well) were seeded into Corning 96-well tissue culture plates. 16 -24 h postseeding, medium was replaced with 50 l of viral inoculum. Alternatively, cells were treated with the indicated reagents, and solvent, DMSO, was kept constant for all experiments. Unless otherwise indicated, 30 ng/ml TNF, 5 M BV6, and/or 25 M Z-VAD-fmk were used. The concentration of each reagent was optimized to exhibit specific efficacy without detectable cytotoxicity. Cell viability was assessed by measuring the intracellular levels of ATP using the Cell Titer-Glo luminescent cell viability assay kit (Promega) according to the manufacturer's instructions. Luminescence was measured on a Synergy HT multidetection microplate reader (BioTek) (12). For real-time imaging of cell permeability with IncuCyte Zoom (Essen Bioscience), cells (15,000 cells/well) were seeded into Corning 48-well tissue culture plates and cultured in medium containing 50 nM SYTOX Green (Molecular Probes) or 5 M propidium iodide (PI) (Sigma) (15).
Immunoblot and Immunoprecipitation-IB and IP were performed as described previously (77,79). Clarified cell lysates were incubated overnight with anti-FLAG M2 affinity gel and washed four times prior to analysis. -Phosphatase treatment was done according to the manufacturer's instructions. Cell lysates and IP samples were separated on an SDS-polyacrylamide gel, followed by transfer to a polyvinylidene difluoride membrane (Immobilon, Millipore), probed with primary antibodies, incubated with the HRP-conjugated secondary antibody, and detected with ECL Western blotting detection reagent (GE Healthcare).
Microscopy-IFA and transmission electron microscopy (TEM) were performed as described previously (77,79). For IFA, cells were fixed in 3.7% formaldehyde and incubated in blocking buffer containing 0.5% bovine serum albumin and 5% goat serum. After incubation with rabbit anti-RIP3 or mouse anti-murine CMV IE1, goat anti-rabbit IgG conjugated to Alexa Fluor 488 or goat anti-mouse IgG conjugated to Alexa Fluor 568 was added, respectively. The cells were incubated with DAPI (Roche Applied Science) to stain nuclei. Images were acquired on an LSM 510 Meta confocal fluorescence microscope (Carl Zeiss). For TEM, cells were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.2). The cells were washed in 0.1 M cacodylate buffer, postfixed with the same buffer with 1% osmium tetroxide, washed, and dehydrated through a graded series of ethanol to 100% and embedded in epoxy resin. The sections were counterstained with uranyl acetate and lead citrate. Images were acquired on a JEOL JEM-1210 transmission electron microscope operated at 75 kV.
Statistical Analyses-Statistical comparisons employed parametric evaluation using Student's t test (GraphPad Prism software). All experiments were repeated at least three times with similar results, and data are represented as the mean Ϯ S.D.

Necroptosis Sensitivity of Human CMV-susceptible Cells-
We initially sought to identify human CMV-susceptible cells capable of supporting RIP3-dependent death. Human colon carcinoma HT-29 cells showed the expected pattern of apoptosis when treated with TNF (T) together with BV6 (S), an IAP antagonist (also called a SMAC mimetic) that reduces polyubiquitination of RIP1 and increases sensitivity to cell death (80), and the death pattern switched to necroptosis, as expected, when the broad caspase inhibitor Z-VAD-fmk was employed (V) (22,41). When evaluated under similar conditions, CMVsusceptible HFs, HMEC-1 cells, and THP-1-derived macro-phage-like cells were insensitive to T alone, T ϩ S, or T ϩ S ϩ V (Fig. 1A). Like HFs, CMV-susceptible, commercially available IMR-90 fetal lung fibroblasts were also insensitive to treatment with T, T ϩ V, T ϩ S, or the combination of all three (data not shown). Undifferentiated THP-1 cells, which are not susceptible to CMV, were sensitive to necroptosis (Fig. 1A), although these cells resisted S alone, S ϩ V, or T ϩ S across a broad range of IAP antagonist concentrations from 0.01 to 3 M (data not shown). THP-1-derived macrophages were insensitive to death elicited by any treatment combination (Fig. 1A). The pattern of cell death in undifferentiated and differentiated THP-1 cells was similar to the behavior of primary macrophages (81). HMEC-1 cells were sensitive to T ϩ S apoptosis, and death was blocked by caspase inhibition. IB analyses were used to assess levels of hRIP1, hRIP3, and hMLKL in the different cell types (Fig. 1B). HFs or IMR-90 fibroblasts had readily detectable hRIP1 but low levels of hRIP3 and hMLKL compared with HT-29 cells, suggesting that resistance may have been dictated by inadequate levels of key mediators (11,12). hRIP3 and hMLKL levels were low in HMEC-1 cells, consistent with the inability of these cells to support necroptosis. Undifferentiated, necroptosis-sensitive THP-1 cells had readily detectable hRIP3 but low levels of hRIP1 and hMLKL. The observation that THP-1 monocytes support necroptosis even though levels of hRIP1 and MLKL are low suggested that hRIP3 may be limiting in HFs. MLKL levels increased as THP-1 monocytes were differentiated into macrophages, where a slower migrating hRIP3 species was observed but was not investigated further. Although THP-1-derived macrophages support human CMV replication (57), these cells resist necroptosis. Although both HT-29 and undifferentiated THP-1 cells are sensitive to necroptosis, neither supports human CMV replication. These data show that available necrosis-sensitive cells are unsuitable for human CMV studies, leading us to modify CMV-permissive HFs using a strategy applied in previous investigations of the necroptosis pathway in mouse cells (12,13).  (12), HFs transduced with a tetra-Ala RHIM substitution mutant (mutRHIM) exhibited 10 -20-fold higher levels of mutant protein in a heterogeneous pattern distinct from WT hRIP3. Cells transduced with kinase-inactive K50A mutant or a dual K50A/ mutRHIM mutant hRIP3 showed reduced levels of hRIP1 (Fig.  2C), indicating that RIP1 levels were influenced by kinase-inactive RIP3 independent of RHIM interactions. Furthermore, levels of hRIP1 did not increase markedly in the presence of the proteasome inhibitor MG132, although the RIP3 kinase-inactive mutant itself became modified as a result of this treatment (Fig. 2D). From this evaluation, proteasome degradation does not appear to contribute to the reduced RIP1 levels in cells carrying kinase-inactive RIP3.

Transduction of HFs with hRIP3 Confers Sensitivity to TNFR1dependent Necroptosis-HFs
RIP3 levels are known to be limiting in cultured cells (11,12,22,23), although cells in mouse tissues are susceptible to necroptosis (12,13). Once transduced with WT hRIP3, HFs became sensitive to death induced by treatment with T alone, T ϩ V, T ϩ S, or T ϩ S ϩ V (Fig. 2, E and F). The sensitivity of cells to TNFR1-dependent death aligns with the host defense contribution of signal transduction through this death receptor. Cells resisted S alone or S ϩ V across a broad range of IAP antagonist concentrations from 0.01 to 3 M (Fig. 2E). Treatment with either T ϩ S or T ϩ S ϩ V resulted in the greatest levels of death, indicating that sensitization to T was markedly enhanced by inhibition of polyubiquitination even in the absence of caspase inhibition. Parenthetically, although RIP1 and RIP3 cleavage may be the mechanism through which Casp8 short circuits necroptotic machinery (82-84), cleavage prod-   2G). In addition to the requirement for pro-necrotic kinases (17,22), knockdown of hMLKL prevented death (Fig. 2, H and I), demonstrating the key contribution (31,32) of this recognized RIP3 kinase target to execution (31,32). These data indicate that HFs support necroptosis as long as RIP3 levels are increased above a critical threshold level. hRIP3-transduced HFs, like control cells transduced with EV, exhibited modest sensitivity to apoptosis induced by cycloheximide (C) alone or T ϩ C (60). Under these conditions, the addition of caspase inhibitor potentiated necroptosis only in hRIP3-transduced HFs and not in cells expressing K50A mutant (data not shown). In a pattern that was distinct from TNFR1 activation, anti-Fas antibody (F) activation of death receptor CD95/Fas alone did not sensitize to death. F ϩ C induced apoptosis in EV-or K50A mutant-transduced cells and this death was reversed by the addition of caspase inhibitor; however, in hRIP3-transduced HFs, F ϩ C ϩ V induced the expected necroptosis pattern of death (data not shown). In all, HFs became sensitive to death receptor-initiated hRIP3-dependent necroptosis once engineered to express sufficient hRIP3, as observed previously with other human as well as mouse cells (11,12,22,23). Cell Morphology, hRIP3 Localization, and hRIP3 Phosphorylation during TNFR1-dependent Necroptosis in HFs-Because necroptosis proceeds via cell swelling and plasma membrane permeabilization (2, 4), we employed time lapse analysis with cell-impermeable dye to interrogate plasma membrane integrity in hRIP3-expressing HFs (Fig. 3, A and B). Cells treated with T ϩ S ϩ V developed increased SYTOX Green fluorescence by 2 h post-treatment (hpt) and affected nearly 80% of cells by 24 hpt. Although fewer than 20% of EV-transduced HFs stained positive with SYTOX Green over a similar time course, this level of death was eliminated in K50A mutant-expressing HFs, most likely due to the reduction in RIP1 as well as dominantnegative inhibition of endogenous hRIP3. When death of WT hRIP3-expressing cells was evaluated by TEM, membrane disruption and other features characteristic of necroptosis were observed (Fig. 3C). By IFA, hRIP3 showed a diffuse pattern in the absence of treatment that developed into discrete foci within 4 h of treatment (Fig. 3D). The proportion of the cells with a disrupted membrane and discrete foci of hRIP3 was similar to the number of SYTOX Green-positive cells.  Next, we employed IB analysis to interrogate the impact of treatment on hRIP3. hRIP3-expressing HFs undergoing necroptosis showed the expected slower migrating pro-necrotic hRIP3 forms within 2 hpt (Fig. 3E). These species increased by 4 hpt as RIP1 levels declined, changes that did not occur in hRIP3-K50A-transduced HFs, where low levels of RIP1 remained constant. Consistent with the behavior of necroptosis-sensitive HT-29 cells (22,41), treatment with T ϩ S was sufficient to induce modification of hRIP3 that was modestly enhanced in the presence of caspase inhibitor (Fig. 3F), and this slow migrating hRIP3 band was eliminated by -phosphatase treatment (Fig. 3G), affirming a modification of hRIP3 in HFs as observed previously in HT-29 cells (22,41). Taken together, these data demonstrate that hRIP3-expressing HFs are highly sensitive to TNFR1-dependent necroptosis as long as IAPs and/or Casp8 are compromised.
Murine CMV M45mutRHIM-induced Necroptosis in HFs-Murine CMV M45mutRIM virus induces rapid DAI-RIP3-dependent necroptosis in mouse cells independent of death receptor signaling and RIP1 (12,13). Although the virus-induced pathway is distinct from TNFR1-dependent death (6, 7), M45-encoded vIRA blocks RHIM signaling in either pathway (12,13). Like all CMVs, murine CMV replication is speciesrestricted but is able to efficiently enter HFs and express early genes without progressing into viral DNA replication (86 -88). hRIP3-transduced HFs became sensitive to M45mutRHIM-induced death, assessed by either ATP levels (Fig. 4A) or permeability to SYTOX Green (Fig. 4, B and C), following a course that developed more slowly than in permissive mouse cells (12,13). The elaboration of murine CMV vIRA by parental K181 blocked necroptosis at times when M45mutRHIM virus induced death. Similar to mouse cells, the addition of IAP antagonist (Fig. 4D) or caspase inhibitor (Fig. 4E) failed to increase the sensitivity of infected cells to death. Murine CMV encodes the Casp8 inhibitor, vICA, an immediate early product expressed in HFs that may sensitize cells to necroptosis (8). hRIP3 RHIM, kinase-inactive and hMLKL interaction site mutants all failed to support virus-induced necroptosis (Fig.  4F), consistent with the expected role of hRIP3 in phosphorylation and activation of hMLKL (17,31,32). When WT and mutant M45 protein were subjected to IB, levels were found to be comparable in virus-infected HFs (Fig. 4G). Unexpectedly, protein levels in infected human cells were similar to mouse fibroblasts despite the fact that HFs do not support murine CMV replication. Importantly, M45-encoded vIRA interacted with hRIP3 in a pattern that was similar to mouse cells (Fig. 4H) shown in previous studies (12,13). The interaction depended on the RHIM and kinase activity of RIP3. Thus, M45mutRHIMinfected HFs are sensitive to necroptosis, and WT murine CMV prevents this pathway through the elaboration of vIRA although HFs are nonpermissive for this mouse virus. Human CMV Inhibits TNFR1-dependent Necroptosis-Next, we investigated the impact of human CMV on necroptosis induced by treatment with T ϩ S ϩ V. Following infection for 24 h with Towne-BAC, TownevarRIT 3 , TB40E-BAC4, Toledo, Merlin, AD169-BAC, AD169varDE, or AD169varATCC (MOI of 3), hRIP3-expressing HFs were treated for an additional 24 h to trigger TNFR1-dependent necroptosis. All tested human CMV strains inhibited cell death (Fig. 5A). Towne-BAC, TownevarRIT 3 , TB40E-BAC4, Toledo, and Merlin strains exhibited the greatest resistance, and AD169-BAC, AD169varDE, and AD169varATCC showed intermediate levels of resistance (p Ͻ 0.05 compared with mock). TownevarRIT 3 and AD169varATCC protected to a different extent, although neither encodes a functional vICA (53), so we employed Towne-BAC-derived Towne⌬UL36 virus to assess the contribution of Casp8 inhibition to death patterns. RIP3-expressing HFs infected with Towne⌬UL36 virus (57) were more sensitive than Towne-BAC to T alone or T ϩ S. The addition of caspase inhibitor blocked death (Fig. 5B). Thus, elaboration of vICA did not impact necroptosis but revealed its expected contribution to suppression of apoptosis (53). In order to exclude any impact of CMV on generation of ATP by modifying mitochondrial respiration (89), we assessed uptake of PI as direct measures of membrane permeability. RIP3-expressing HFs infected with Towne-BAC and treated with T ϩ S ϩ V resisted PI uptake (Fig. 5, C and D) compared with mockinfected cells. Surprisingly, a slower migrating, modified form of RIP3 previously associated with necroptosis (22) remained detectable, although cells remained viable and expressed IE1-p72 at levels similar to untreated cells (Fig. 5E). Interestingly, phosphorylated MLKL levels were readily detected, suggesting that the virus-imposed block occurs downstream of this executioner. Treatment with -phosphatase affirmed the phosphorylation of hRIP3 (see Fig. 3G), consistent with the characterization in HT-29 cells (22). Thus, inhibition of necroptosis by human CMV influences steps that follow RHIM signal transduction as well as RIP3 autophosphorylation and phosphorylation-dependent activation of MLKL, a pattern that is distinct from either murine CMV (12,13) or HSV (41). We evaluated the impact of RIP3 expression on viral replication levels and found that both hRIP3-WT and hRIP3-K50Aexpressing HFs supported Towne-BAC, AD169-BAC, or TB40E-BAC4 (MOI of 3) replication to levels observed in control HFs (Fig. 5F). In addition, treatment with T, T ϩ S, or T ϩ S ϩ V failed to compromise viral replication when assessed at 4 dpi, although overall titers were modestly reduced in cells with combined treatment (Fig. 5G). These data indicate that viral suppression of TNFR1-mediated, hRIP3-dependent necroptosis sustains cell viability to support production of viral progeny.
Human CMV Infection Inhibits M45mutRHIM Virus-induced Programmed Necrosis-We employed superinfection experiments with murine CMV in order to determine whether human CMV blocks virus-induced necroptosis. hRIP3-ex-pressing HFs were infected with Towne, AD169, or Toledo strains (MOI of 3) followed 24 h later by murine CMV K181-BAC or M45mutRHIM virus (MOI of 10). After an additional 48 h, cell viability was assessed. Whereas only 50% of HFs exposed to M45mutRHIM virus survived these conditions, 80 -90% of M45mutRHIM-infected cells survived when HFs were infected with human CMV independent of strain (Fig.  6A). The proportion of murine CMV-derived IE1-expressing cells was similar in Towne-BAC virus and mock-infected hRIP3-WT-expressing HFs, indicating that murine CMV was not compromised by prior human CMV infection (Fig. 6B). Furthermore, murine 3T3-SA cells exposed to Towne or AD169 infection resisted M45mutRHIM-induced necroptosis to equivalent levels (Fig. 6C). Although human CMV is speciesrestricted, both of these strains efficiently enter and express early gene products in mouse fibroblasts (90). The ability of human CMV to block virus-induced necroptosis as well as TNFR1-dependent necroptosis was most consistent with the function of a virion component or early gene product acting on necroptosis.
Human CMV IE1 Function Is Required for Suppression of Necroptosis-In order to further evaluate the contribution of human CMV functions to suppression of necroptosis, we first conducted a time-of-addition experiment to determine when during infection HFs became resistant to necroptosis. The viability of HFs transduced with WT, K50A, and EV was determined by treating cells for 24 h with T ϩ S ϩ V starting at 6, 24, or 48 hpi with Towne-BAC virus (MOI of 3) (Fig. 7A). Cells remained sensitive to necrosis at 6 hpi but resisted death initiated later, and, as expected, neither EV control nor K50A mutant showed sensitivity. Inhibition of viral DNA synthesis with the polymerase inhibitor phosphonoformate (300 g/ml) did not sensitize to necroptosis (Fig. 7B), indicating that suppression develops independent of viral DNA synthesis. To determine the impact of input virion or newly synthesized proteins, we tested UV-inactivated virus. Virus particles failed to confer resistance to cells at a time when viable virus suppressed necroptosis (Fig. 7C). As expected, IB analysis revealed that UV-irradiated virus delivered virion tegument protein (pp65) but blocked expression of IE1 (Fig. 7D). Together with the time  3). B, human CMV-infected hRIP3-WT-expressing HFs exposed to K181-BAC or M45mutRHIM virus as described in A and assessed by IFA with murine CMV IE1 (mIE1) stained (red) at 48 hpi with human CMV. The merged images include Towne-BAC-encoded GFP fluorescence (green) and DAPI staining to identify nuclei (blue). Original magnification was ϫ200. C, relative viability of the indicated human CMV-infected 3T3-SA cells (MOI of 3) exposed to M45mutRHIM virus (MOI of 10) at 24 hpi assessed at 48 hpi with human CMV and expressed as a percentage relative to a parallel set of cultures exposed to K181-BAC murine CMV (n ϭ 3). Error bars, S.D.; *, p Ͻ 0.05; **, p Ͻ 0.001; #, not significant (p Ͼ 0.05). course (Fig. 7A), these data implicate a newly synthesized viral or cellular gene product in viral suppression of TNFR1-dependent necroptosis.
We next evaluated the contribution of the viral regulatory protein IE1 (91), which dictates gene expression levels and modulates innate immune signaling early after infection (92). IE1-null virus (CR208) and repaired virus (CRQ208) (70) were employed to test whether IE1 influenced suppression. CR208infected cells were insensitive to T ϩ S ϩ V, whereas IE1-expressing CRQ208-infected cells resisted necroptosis (Fig. 7, E and F) as efficiently as parental TownevarRIT 3 (see Fig. 5A). As expected, EV-and hRIP3-K50A-transduced HFs resisted necroptosis. The levels of replication, assessed in supernatants of CRQ208-or CR208-infected (MOI of 3) cells, were not influenced by the higher levels of hRIP3-WT or K50A mutant expression (Fig. 7G), although induction of necroptosis at 1 dpi significantly reduced CR208 virus yields in the supernatant at 4 dpi (Fig. 7H). Thus, human CMV-encoded IE1 protein is necessary for the virus to block necroptosis.
cell death pathway is completely suppressed by M45-dependent disruption of RHIM signaling (12,13). It has remained a mystery whether suppression of necroptosis plays an important role in human CMV pathogenesis as it does for the biologically similar rodent virus. Neither human CMV UL45 nor any other viral protein has a RHIM (93), and UL45 mutant virus fails to exhibit any phenotypic similarity to M45 mutants (94,95). We have shown here that human CMV blocks necroptosis but employs a different strategy than murine CMV. By making use of primary human fibroblasts transduced with RIP3 to increase levels of this key pro-necrotic kinase, we generated human CMV permissive primary cells that support necroptosis. Like transduced NIH3T3 cells (11,12), the resultant hRIP3-expressing HFs became susceptible to TNFR1dependent necroptosis. Sensitivity to either TNFR1-of M45mutRHIM-induced necroptosis was eliminated during infection with human CMV, providing direct evidence of virusspecific inhibition of this alternate death pathway. The mechanism of inhibition was distinct from MCMV and appeared to target a step downstream of RHIM signaling, after RIP3 kinasedependent phosphorylation of MLKL (29,72). This execution phase of necroptosis, which drives the disruption of the plasma membrane, remains to be completely dissected. The mechanism of human CMV suppression is likely to provide additional insights into the precise steps driving execution of human cells. Therefore, this work, together with our previous identification of M45 and UL39 as RHIM-dependent inhibitors of necroptosis (12,13,27,41), establishes the core parameters through which cell death pathways function in pathogenesis as well as the diverse ways herpesviruses infecting rodents and humans successfully counter these host defense pathways.
Human CMV IE1 expression is necessary for the suppression of TNFR1-dependent necroptosis. The IE1 gene product, IE1-p72 is primarily involved in regulating viral transcription during productive replication (70) as an accessory to IE2-p86 (91). IE1-p72 also impacts the host cell response to infection and modulates the activation of interferon (92). IE1-p72 is a suppressor of TNFR1-dependent apoptosis in HeLa cells (69), although phenotypic evaluation of IE1 mutant viruses has not revealed any impact on cell survival before our current study. In addition to the work shown, IE1-p72-expressing ihf-ie1.3 cells were tested and failed to resist necroptosis, suggesting that the requirement here during infection is for optimal expression of an early gene (70). Using permissive cells engineered to carry adequate RIP3, IE1 mutant virus fails to protect from TNFR1dependent signal transduction, whereas parental virus effectively protects cells from insult.
Death receptor signaling contributes to CMV biology, during both productive infection and reactivation from latency. Human CMV strains down-modulate expression of TNFR1 starting between 18 and 24 hpi (96 -98). Up to that time, the UL138-encoded viral promoter of TNFR1 signaling sustains the stability of TNFR1 and confers increased signal transduction that may also contribute to reactivation from latency (73,74). In our hands, viral strains that lack UL138 (Towne and AD169 variants) exhibit varying sensitivity to TNFR1-dependent signaling, indicating that the viral promoter of TNFR1 signaling does not impact necroptosis in the assays we employed here. Furthermore, strains that encode UL138 (TB40E-BAC4, Toledo, and Merlin) all showed resistance to necroptosis similar to UL138-deficient Towne. These results demonstrate a crucial contribution of IE1-p72 in establishing a cellular environment that prevents necroptosis during viral infection, acting either directly (92) or indirectly (91). Such countermeasures certainly may be important during productive infection as well as during latency.
Observations that murine CMV M45 protects and M45mutRHIM fails to protect human fibroblasts extend findings made in necroptosis-sensitive HT-29 cells (41) and are consistent with other work showing vIRA suppression of RHIM signal transduction in both mouse and human cells (12,17,26,27). It is important to note that the cellular processes leading to M45mutRHIM activation of necroptosis in human cells has not been the focus of this study, leaving the investigation of similarities between human and mouse cells (13) for future investigation. In mouse cells infected with M45mutRHIM, the cytosolic dsDNA sensor, DAI, recruits RIP3 to trigger RHIM-dependent necroptosis. Human CMV infection is already known to be sensed by DAI, and this contributes to the stimulation of an innate immune response (99). The activation of interferon by DAI depends on RHIM signal transduction (100). The apparent mechanism of IE1-dependent suppression of necroptosis downstream of RHIM signal transduction as well as downstream of RIP3 kinase-dependent activation of MLKL aligns with the ability of infected cells to sustain DAI-dependent interferon-like responses (99). Further characterization of the triggers and outcomes affecting necroptosis will require the use of transduced cells similar to those described here to unveil collaborating viral and host factors. Future studies will seek to identify the precise viral or host cell inhibitors as well as to characterize the behavior of potential sensitizers, such as the Casp8 inhibitor, vICA, as well as the mitochondrial cell death suppressors. It is important to recognize that the detection of virus-induced necroptosis emerged from the finding that murine CMV M45 is a powerful suppressor of RHIM signal transduction (12,17,26,27). Whereas mouse studies have supported the broad importance of necrotic death as an alternate pathway of host defense, our findings here bring to light the central role that RIP3 plays in human cells and reveals the existence of machinery that may contribute to viral pathogenesis as well as inflammation and associated tissue damage.