|Home | About | Journals | Submit | Contact Us | Français|
Programmed necrosis is a form of caspase-independent cell death whose molecular regulation is poorly understood. The protein kinase RIP1 is crucial for programmed necrosis. Because RIP1 also mediates activation of the pro-survival transcription factor NF-κB, we postulated that additional molecules are required to specifically activate programmed necrosis. Using a RNA interference screen, we identified RIP3 as a crucial activator for programmed necrosis induced by TNF and during virus infection. RIP3 regulates necrosis-specific RIP1 phosphorylation. The phosphorylation of RIP1 and RIP3 stabilizes their association within the pro-necrotic complex, activates the pro-necrotic kinase activity, and triggers downstream reactive oxygen species production. The pro-necrotic RIP1-RIP3 complex is induced during vaccinia virus infection. Consequently, RIP3−/− mice exhibited severely impaired virus-induced tissue necrosis, inflammation, and control of viral replication. Thus, RIP3 controls programmed necrosis by initiating the pro-necrotic kinase cascade that is essential for the innate inflammatory response against virus infections.
Cell death by programmed necrosis (also known as caspase-independent cell death or necroptosis) is characterized by rapid loss of plasma membrane integrity prior to the exposure of phagocytic signal (Golstein and Kroemer, 2007). The release of endogenous “danger signals” from necrotic cells induces inflammation and can activate immune responses, trigger inflammatory diseases and promote cancer growth (Kono and Rock, 2008). In addition, nonapoptotic or necrotic cell death has been shown to critically regulate disease pathologies in animal models of hypoxic/ischemic injury (Degterev et al., 2005), acute pancreatitis (Mareninova et al., 2006) and septic shock (Cauwels et al., 2003). Consistent with these observations, blockade of necrosis was effective in slowing or reducing cell injury in models of cardiac infarct (Smith et al., 2007) and ischemic brain injury (Degterev et al., 2005). Despite its biological importance, the molecular components regulating programmed necrosis are not well defined.
TNF-like cytokines are potent inducers of programmed necrosis. We and others have previously identified an obligate role for the protein serine/threonine kinase receptor interacting protein 1 (RIP/RIP1/RIPK1) in programmed necrosis (Chan et al., 2003; Holler et al., 2000; Lin et al., 2004). RIP1 is also a crucial adaptor that mediates activation of the pro-survival transcription factor NF-κB by TNF, TLR3 and TLR4 (Cusson-Hermance et al., 2005; Hsu et al., 1996; Meylan et al., 2004; Ting et al., 1996). The kinase function of RIP1 is essential for programmed necrosis but dispensable for NF-κB activation (Chan et al., 2003; Holler et al., 2000; Lin et al., 2004), suggesting that programmed necrosis might be regulated at the level of activation of RIP1 kinase activity. However, the molecular mechanism that activates the pronecrotic RIP1 kinase activity has remained elusive.
As we have discussed above, necrosis distinguishes itself from apoptosis by its pro-inflammatory effects. Such pro-inflammatory effects might be important for anti-viral immune responses. In addition, programmed necrosis may control the viral factory by eliminating the infected host cells. An anti-viral role for programmed necrosis is further bolstered by our previous findings that certain viral FLIPs (FLICE-like inhibitor proteins) are potent inhibitors of TNF-induced programmed necrosis (Chan et al., 2003). These results suggest that viral inhibition of programmed necrosis is an important immune evasion strategy for certain viruses. However, because genetic ablation of the only known programmed necrosis mediator RIP1 resulted in neonatal lethality (Kelliher et al., 1998), the physiological roles of programmed necrosis in inflammation and anti-viral host defense have not been tested.
In this report, we identified RIP3 as a crucial upstream activating kinase that regulates RIP1-dependent programmed necrosis. We show that RIP3 acts upstream to phosphorylate RIP1, which in turn mediates downstream RIP3 phosphorylation. Both RIP3 and the kinase activity of RIP1 are essential for stable formation of the RIP1-RIP3 pro-necrotic complex, which critically controls downstream reactive oxygen species (ROS) production. Strikingly, the pronecrotic RIP1-RIP3 complex was specifically induced in the liver upon vaccinia virus (VV) infection. RIP3−/− mice failed to initiate virus-induced tissue necrosis and inflammation, resulting in highly elevated viral replication and mortality. These results show that RIP3-dependent programmed necrosis is important for virus-induced inflammation and innate immune control of viral infections.
RIP1 is a pleiotropic adaptor that mediates both NF-κB activation and programmed necrosis by TNF (Chan et al., 2003; Holler et al., 2000; Lin et al., 2004). The kinase activity of RIP1 is essential for programmed necrosis, but dispensable for NF-κB activation (Chan et al., 2003; Holler et al., 2000). We hypothesized that induction of programmed necrosis requires additional components that specifically turn on the RIP1 kinase activity. To identify such molecules, we screened a 21-mer small interference RNA (siRNA) library consisting of 691 human kinase genes in FADD-deficient Jurkat cells (Table S1 and supplemental methods), which rapidly undergo programmed necrosis in response to TNF (Chan et al., 2003). From the screen, we identified ten clones that potently inhibited TNF-induced programmed necrosis (% survival > average survival of all clones + 3 SD) (Table S2). Significantly, RIP1 was one of the protective clones, demonstrating the fidelity of our screen. Another siRNA clone, RIP3, conferred protection against TNF-induced programmed necrosis at a level comparable to that by RIP1 siRNA (Fig. S1). The protection against TNF-induced programmed necrosis by RIP1 and RIP3 siRNAs were specific, since other RIP family kinases including RIP2, RIP4 and RIP5 that were represented in the library did not protect against programmed necrosis (data not shown). We validated these results in TNFR-2+ wild type Jurkat cells (clone 4E3) and found that individual RIP3-specific siRNAs efficiently reduced RIP3 protein expression (Fig. 1A) and inhibited TNF/zVAD-fmk-induced programmed necrosis (Fig. 1B), but had little or no effects on apoptosis induced by TNF or FasL (Fig. 1C-D). Consistent with a previous report (Newton et al., 2004), RIP3 deficiency did not alter TNF-induced NF-κB activation (Fig. 1E). Thus, unlike RIP1, RIP3 is a specific activator for programmed necrotic cell death.
RIP3 contains an N-terminal kinase domain with ~ 40% identity with the kinase domain of RIP1 (Fig. 2A). A kinase-defective (KD) RIP3 mutant failed to restore TNF-induced programmed necrosis in Jurkat cells whose endogenous RIP3 expression was silenced by siRNA (Fig. 2B, RIP3-null Jurkat cells), indicating that the RIP3 kinase activity is crucial for programmed necrosis. RIP1 and RIP3 have been reported to interact with each other via the RIP homotypic interaction motif (RHIM) (Fig. 2A) (Sun et al., 2002). Tetra-alanine substitutions in the RHIM of RIP1 or RIP3 abolished their association (Fig. S2) and the ability of RIP3 to restore TNF-induced programmed necrosis in RIP3-null Jurkat cells (Fig. 2C). Similar mutations in the kinase and RHIM domains of RIP1 also abolished the RIP1-mediated rescue of programmed necrosis in RIP1-deficient Jurkat cells (Fig. 2D). Thus, both the kinase and RHIM domains of RIP1 and RIP3 are important for programmed necrosis.
TNFR signaling involves at least two spatially and temporally distinct signaling complexes: the transient membrane-associated TNFR-1 signaling complex (Complex I) and the cytoplasmic signaling complex termed Complex II (Micheau and Tschopp, 2003). Complex II is formed upon TNFR-1 internalization, when TNFR-1 dissociates from Complex I to allow recruitment of additional factors. Because RIP1 is recruited to both Complex I and Complex II, we examined RIP3 recruitment to these signaling complexes. In TNFR-2+ Jurkat 4E3 cells, TNF triggers both apoptosis and programmed necrosis, although apoptosis predominates due to its faster kinetics unless caspases are inhibited (Zheng et al., 2006). In TNFR-2+ Jurkat 4E3 cells, RIP3 was recruited to caspase-8-associated Complex II (Fig. 2E), but not to TNFR-1 associated Complex I (Fig. 2F). On short exposure, zVAD-fmk moderately enhanced RIP1 and RIP3 binding to caspase-8-associated Complex II (Fig. 2E). However, the difference was minor and not distinguishable on longer exposure (Fig. S3). On longer exposure, RIP1 cleavage in Complex II was also visible (Fig. S3). Although RIP1 cleavage by caspase-8 at D324 might inactivate the RIP1 kinase function and contribute to the bifurcation of apoptosis and programmed necrosis (Lin et al., 1999), zVAD-fmk likely sensitizes cells to programmed necrosis via additional regulatory events because full length RIP1 was still readily detected without caspase inhibition (Fig. S3).
The recruitment of RIP1 and RIP3 to Complex II suggests that their pro-necrotic kinase activity might be activated within this complex. When caspase-8-associated Complex II was examined, we did not observe strong TNF-inducible kinase activity (data not shown). Hence, although RIP1 and RIP3 were recruited to caspase-8, their activities were not regulated in this compartment. In contrast, Complex II isolated by immunoprecipitation with FADD-specific antibody exhibited a transient induction of kinase activity as measured by phosphorylation of a ~60 kDa protein and several other species (Fig. 3A, top panel). We confirmed the identity of the ~60 kDa phospho-protein to be RIP3 when we boiled the TNF-activated FADD complex in SDS and subjected it to a second immunoprecipitation with RIP3-specific antibody (Fig. 3B, lane 5). In addition to RIP3, Complex II kinase activity could also be measured using the artificial substrate MBP (Fig. 3B, lane 3). Similar Complex II kinase activity could also be detected in caspase-8 deficient Jurkat cells undergoing programmed necrosis (Fig. S4). Interestingly, FADD complex kinase activity was detected in TNF-treated TNFR2+ Jurkat cells when apoptosis was the dominant cell death mode (Fig. 3A, lane 2). Although zVAD-fmk slightly increased Complex II kinase activity (Fig. 3A, lane 5), these results suggest that zVAD-fmk might exert its necrosis-promoting effects subsequent to the induction of Complex II kinase activity.
Consistent with the lack of RIP3 binding to Complex I, TNF did not activate kinase activity in TNFR-1 complexes (Fig. 3C). Moreover, TNFR-2− Jurkat cells that did not undergo programmed necrosis in response to TNF (Chan and Lenardo, 2000) also exhibited no inducible FADD Complex II kinase activity regardless of whether zVAD-fmk was present (Fig. 3D). Thus, the induction of Complex II kinase activity was specific to programmed necrosis. Complex II kinase activity was also not detected in TNFR2+ RIP1-deficient Jurkat cells, which undergo apoptosis in response to TNF (Chan et al., 2003) (Fig. 3E). In addition, the RIP1 kinase inhibitor necrostatin-1 (Nec-1) potently inhibited Complex II kinase activity (Fig. S4). Interestingly, while RIP1 was recruited to FADD in a TNF-dependent manner, RIP3 was constitutively associated with FADD (Fig. 3A, lower panels), although this interaction may be indirect (Fig. S4). Taken together, our results show that RIP1 recruitment to Complex II is crucial for the induction of pro-necrotic Complex II kinase activity.
Although the induction of Complex II kinase activity appears to correlate with programmed necrosis, it could not be explained entirely by RIP1 recruitment to Complex II because RIP1 was recruited to Complex II in cells that did not undergo necrotic death (Fig. 3D, lower panels). To further define the molecular regulation of programmed necrosis, we examined the recruitment of RIP1 and RIP3 to Complex II in MEFs because the role of RIP3 in the assembly of the pro-necrotic complex could be evaluated using RIP3−/− MEFs. Similar to Jurkat cells, constitutive association between RIP3 and FADD was also detected in MEFs (Fig. 4A). When programmed necrosis was induced in wild type MEFs with TNF, cycloheximide and zVAD-fmk (T/C/Z) (Fig. S5), RIP1 was recruited to FADD in a TNF-dependent manner (Fig. 4A, lane 3). Moreover, TNF treatment led to further recruitment of RIP3 to the FADD-associated Complex II and RIP3 modification that resembles polyubiquitination (Fig. 4A, lane 3). Strikingly, RIP1 recruitment to Complex II was severely impaired in RIP3−/− MEFs (Fig. 4A, lanes 5-8). The RIP1 kinase inhibitor necrostatin-1 (Nec-1) (Degterev et al., 2008) also abolished RIP1 recruitment to FADD and inhibited additional RIP3 binding to the complex (Fig. 4A, lane 4 and Fig. S6). These results indicate that both RIP1 and RIP3 are critical for the assembly of the pro-necrotic Complex II.
We then ascertained the possibility that RIP3 might regulate RIP1 recruitment to the pro-necrotic Complex II via phosphorylation by labeling MEFs with [32P]-orthophosphate. Indeed, RIP1 phosphorylation was detected when programmed necrosis was induced (Fig. 4B, lane 2). Although low level of basal RIP1 phosphorylation could sometimes be detected, TNF treatment alone, which did not result in apoptosis or programmed necrosis (Fig. S6), did not induce RIP1 phosphorylation (Fig. 4B, lanes 3-4). Strikingly, ligand-dependent RIP1 phosphorylation was completely abrogated in RIP3−/− MEFs (Fig. 4B, lanes 5-6). Similar TNF-dependent RIP1 phosphorylation was observed in FADD-deficient Jurkat cells (Fig. 4C, lane 2). Importantly, Nec-1 did not inhibit RIP1 phosphorylation (Fig. 4C, compare lanes 2-3), indicating that TNF-dependent RIP1 phosphorylation was not due to autophosphorylation. To determine whether RIP1 might directly phosphorylate RIP3, we subjected wild type or kinase defective RIP1 and/or RIP3 expressed in 293T cells to in vitro kinase assays. Indeed, WT-RIP3 phosphorylated KDRIP1 to a level comparable to that achieved via RIP1 autophoshorylation (Fig. 4D, compare lanes 6-7). In contrast, although WT-RIP1 could undergo autophosphorylation, it did not phosphorylate KD-RIP3 (Fig. 4D, lanes 6). However, the level of RIP3-mediated RIP1 phosphorylation was weak, suggesting that other kinases might be involved in phosphorylating RIP1 and stabilizing its association with the pro-necrotic Complex II.
Recent evidence indicates that signal transduction by TNFR-like receptors may involve distinct Complex IIs (Wang et al., 2008; Wilson et al., 2009). We reasoned that RIP1 and RIP3 might interact specifically within a pro-necrotic Complex II. In MEFs, TNF stimulation in the presence of cycloheximide induces apoptosis (T/C), but not programmed necrosis (Lin et al., 2004) (Fig. S6). Under these conditions, no interaction between RIP1 and RIP3 was detected (Fig. 4E, lanes 1-4). By contrast, programmed necrosis induced by T/C/Z specifically induced formation of the RIP1-RIP3 complex (Fig. 4E, lanes 5-8). Similar to the recruitment of RIP1 and RIP3 to FADD, the interaction between RIP1 and RIP3 was abolished by Nec-1 (Fig. 4F, compare lanes 7-8 and Fig. 4G, lanes 2-3), indicating that the RIP1 kinase activity was also required for stable interaction between RIP1 and RIP3. The effect of Nec-1 on RIP1-RIP3 association might be due to inhibition of RIP3 phosphorylation, since necrosis-specific RIP3 phosphorylation was abolished by Nec-1 (Fig. 4H, compare lanes 2-3). Because RIP1 did not directly phosphorylate RIP3 (Fig. 4D), another downstream kinase might be responsible for RIP3 phosphorylation. Taken together, our data strongly implicate that phosphorylation of RIP1 and RIP3 plays a crucial role in the stable assembly of the pro-necrotic Complex II.
A recent report shows that the NADPH oxidase Nox1 acts within the membrane-associated receptor complex to generate superoxide anions during programmed necrosis (Kim et al., 2007). Since RIP3 acts within the cytoplasmic signaling complex, we asked whether RIP3 might function downstream of ROS to promote programmed necrosis. Surprisingly, ROS production as detected by staining with 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) (Fig. 5A) and necrotic cell death as detected by propidium iodide (PI) uptake were completely inhibited in RIP3−/− MEFs (Fig. 5B-C). Similar results were obtained with the murine fibrosarcoma L929 transfected with RIP3-specific siRNA (Fig. S7). In contrast, silencing RIP1 or RIP3 expression in L929 cells had no effect on hydrogen peroxide-induced oxidative cell injury (Fig. 5D-E). By contrast, siRNA-mediated silencing of RIP1 or RIP3 inhibited zVAD-fmk-induced necrosis in L929 cells (Fig. 5F-G), which was recently shown to be driven by autocrine TNF production (Hitomi et al., 2008). Thus, our results indicate that RIP3 acts upstream to regulate ROS production during programmed necrosis.
One physiological situation in which caspases are inhibited and programmed necrosis might be important is during viral infections (Benedict et al., 2002). To test this hypothesis, we examined cell death responses in activated wild type and RIP3−/− T-cells infected with vaccinia virus (VV), which encodes the viral caspase inhibitor B13R/Spi2. VV-infected cells were resistant to death receptor-induced apoptosis but become sensitized to TNF-induced necrosis (Li and Beg, 2000). Wild type T-cells infected with recombinant GFP-VV exhibited normal T-cell activation-induced cell death (AICD) induced by anti-CD3 antibody (Fig. 6A, compare the white bars). In contrast, AICD in GFP-VV infected RIP3−/− T-cells was significantly reduced compared with uninfected GFP− cells in the same culture (Fig. 6A, compare the black bars). VV-infected RIP3−/− cells were also protected from TNF-induced cell death (Fig. 6B, black bars). In fact, TNF stimulated the proliferation of GFP+ VV-infected RIP3−/− cells, which accounted for the negative cell loss. VV-infected wild type T-cells also exhibited a modest decrease in TNF-induced cell death due to the inhibition of caspase-dependent apoptosis (Fig. 6B, white bars).
The effect of VV infection on T-cell death was recapitulated with zVAD-fmk. While AICD was only moderately impaired by zVAD-fmk in wild type T-cells, it was significantly reduced in RIP3−/− cells (Fig. 6C). Geldanamycin (GA), which was thought to inhibit programmed necrosis by destabilizing RIP1 (Lewis et al., 2000), also suppressed RIP3 protein expression (Fig. S8). Hence, GA and Nec-1 also synergized with zVAD-fmk to suppress T-cell death (Fig. S8). The sensitization to programmed necrosis by VV infection was also observed in MEFs. While TNF alone did not cause cell death in MEFs (Fig. S6), VV-infected MEFs were highly sensitive to TNF-induced necrosis in a RIP3-dependent manner (Fig. 6D). Electron microscopy confirmed that VV-infected underwent necrotic-like cell death in response to TNF (Fig. 6E, panel c). Importantly, while RIP1-RIP3 complex was only detected when cells were stimulated with T/C/Z (Fig. 4E-G) or TNF/zVAD (Fig. 6F, lanes 3-4), TNF alone was sufficient to trigger RIP1-RIP3 association in VV-infected MEFs (Fig. 6F, lanes 5-8). This is in stark contrast to uninfected cells, in which TNF did not trigger RIP1 and RIP3 interaction (Fig. 6F, lanes 1-2). Taken together, these results show that RIP3-dependent programmed necrosis is an important cell death mechanism when the cellular apoptosis machinery was inhibited by viral inhibitors.
TNF-TNFR interactions are known to be important for the innate immune defense against VV infection (Chan et al., 2003; Ruby et al., 1997). The requirement for RIP3 to induce cell death in VV-infected cells suggests that TNF might confer protection against VV via RIP3-dependent programmed necrosis. To test this hypothesis, we infected wild type and RIP3−/− mice with VV. Consistent with its innate immune protective role, TNF expression was detected in peritoneal exudate cells (PECs) 24 hours post-infection (Fig. 7A). Moreover, TNF expression was induced in the liver and spleen of infected wild type and RIP3−/− mice (Fig. 7B). Thus, virus-induced TNF expression was not impaired in the RIP3−/− mice.
In wild type mice, VV infection causes inflammation marked by neutrophil/macrophage infiltration in the visceral fat pads (Fig. 7C, panels a, red arrows), which was conspicuously absent in the RIP3−/− mice (Fig. 7C, panels c-d). Interestingly, the inflammatory foci were concentrated in areas of extensive fat cell necrosis in the wild type mice (Fig. 7C, panels a-b), suggesting that necrosis might have promoted the inflammatory reaction. The PECs might provide a source of TNF to the visceral fat tissues (Fig. 7A). In contrast, necrotic tissue injury was significantly reduced in VV-infected RIP3−/− mice (Fig. 7C, panels c-d). The impairment in virus-induced inflammation was observed on day 3 prior to the peak of CD8+ T-cell responses on day 7-8, consistent with defective innate immune responses in the RIP3−/− mice. Similarly, extensive inflammation and necrosis were detected in liver of infected RIP3+/+ mice (Fig. 7D, panels a-b), but were absent in RIP3−/− and TNFR2−/− mice (Fig. 7D, panels c-d). Remarkably, the pro-necrotic RIP1-RIP3 complex was detected in infected liver cell extracts 12-24 hours post-infection (Fig. 7E), indicating that liver cells underwent programmed necrosis in response to VV infection. The reduced inflammation correlated with dramatic increases in viral titers in the visceral fat pad (~120X), liver (~50X) and the spleen (~2000X) of RIP3−/− (Fig. 7F). Consequently, the RIP3−/− mice succumbed to the infection (Fig. 7G). Collectively, our results strongly support a role for RIP3 in promoting programmed necrosis and virus-induced inflammation (Fig. S9).
RIP3 was originally identified as a NF-κB and apoptosis regulator (Kasof et al., 2000; Pazdernik et al., 1999; Sun et al., 1999; Yu et al., 1999). However, RIP3−/− mice exhibit no remarkable phenotypes and responded normally to apoptosis and NF-κB activation signals (Newton et al., 2004). Thus, the biological role of RIP3 was unknown until now. Although RIP3 was not identified as a necrosis mediator in a recent genome-wide RNAi screen (Hitomi et al., 2008), we show that formation of a unique pro-necrotic Complex II composed of RIP1 and RIP3 is a crucial first step in the induction of programmed necrosis. RIP3 acts upstream to regulate necrosis-specific RIP1 phosphorylation. However, since the level of RIP1 phosphorylation by ectopically expressed RIP3 was low, it remains possible that RIP3 may activate another kinase that directly phosphorylates RIP1.
Interestingly, the RIP1 kinase activity is also required for RIP3 phosphorylation during programmed necrosis. Because ectopically expressed RIP1 did not phosphorylate RIP3, RIP1 may facilitate RIP3 phosphorylation by activating another downstream kinase. In this regard, it is tempting to speculate that the other kinases we identified in the siRNA screen might fulfill this function. Alternatively, RIP1 within the pro-necrotic Complex II might directly phosphorylate RIP3, since in vitro phosphorylation of RIP3 by Complex II was inhibited by the RIP1-specific inhibitor Nec-1 (Fig. S4). In addition, ectopically expressed RIP3, but not RIP3 present within the endogenous FADD complex, efficiently phosphorylated the artificial substrate MBP. Thus, the kinase activities of RIP1 and RIP3 are tightly controlled within the context of the pro-necrotic Complex II.
Our results implicate a crucial role for RIP1 and RIP3 phosphorylation in the stable assembly of the pro-necrotic RIP1-RIP3 complex. Intriguingly, transfection of KD-RIP1 or KD-RIP3 did not dominantly inhibit programmed necrosis in cells expressing endogenous RIP1 and RIP3, but rather enhanced TNF-induced programmed necrosis (unpublished observation). The lack of dominant inhibition by KD-RIP1 or KD-RIP3 might suggest oligomerization as a crucial first step in activating the pro-necrotic kinase complex. In such scenario, the RHIM might facilitate oligomerization of RIP1 and RIP3. Phosphorylation of RIP1 and RIP3 may stabilize the structural scaffold of the pro-necrotic complex. However, once the oligomer is formed, a single copy of kinase active RIP1 and RIP3 within the oligomer might be sufficient to activate downstream function. Consistent with this model, RIP3 RHIM mutant was kinase inactive and failed to sensitize programmed necrosis in wild type Jurkat cells (unpublished observation).
The mitochondria generate ROS as a result of oxidative respiration. Disruption of mitochondrial function can further exacerbate ROS release. Interestingly, several components of the mitochondria permeability transition pore (mPTP) are kinases substrates (Le Mellay et al., 2002; Pastorino et al., 2005). It is therefore tempting to speculate that RIP1, RIP3 or other downstream kinases may phosphorylate components of the mPTP to disrupt their functions and to trigger an increase in ROS during programmed necrosis. Alternatively, the NADPH oxidase Nox-1 has been shown to mediate ROS production (Kim et al., 2007). Although Nox-1 was implicated to signal via the TNFR-1 associated complex, it is possible that RIP3 might also interact with Nox-1 within the pro-necrotic Complex II to mediate ROS generation.
TNF has long been known to be an important innate immune effector cytokine against bacterial and viral infections. TNF exerts its anti-microbial effects through induction of apoptosis and NF-κB. Using vaccinia virus infection as a model, we have now established RIP3-dependent programmed necrosis as a third mechanism by which TNF contributes to innate immune control of pathogens. TNF expression was rapidly induced upon VV infection in multiple cell types. The expression of TNF coincided with the formation of the pro-necrotic RIP1-RIP3 complex in the liver of infected mice. In RIP3−/− mice, virus-induced tissue necrosis and inflammation was severely compromised. Strikingly, the resultant increase in viral replication was comparable to that observed in the TNFR1−/− and TNFR2−/− mice (Chan et al., 2003; Ruby et al., 1997) and much higher than that observed in MyD88−/− or TLR2−/− mice (Zhu et al., 2007). These results are consistent with the normal TLR responses of the RIP3−/− mice (Newton et al., 2004) and indicate that the viral disease observed in RIP3−/− mice was caused by defective TNF signaling rather than abnormal TLR signaling. Our results also support a role for RIP3-dependent programmed necrosis in promoting the subsequent virus-induced inflammation. An important role for RIP1/RIP3-mediated programmed necrosis in anti-viral responses is further bolstered by our previous discovery of certain viral FLIPs that potently inhibit programmed necrosis (Chan et al., 2003). More recently, the M45 viral cell death inhibitor from murine cytomegalovirus was shown to interact with RIP1 and RIP3 via the RHIM (Upton et al., 2008). These results suggest that like apoptosis, inhibition of programmed necrosis might be an emerging viral immune evasion strategy.
In addition to virus-induced inflammation, necrotic cell death may have broader roles in regulating other inflammatory processes through the release of “endogenous adjuvants” into the tissue milieu (Kono and Rock, 2008). For example, RIP3 expression was upregulated during wound healing (Adams et al., 2007), a biological process that shares some hallmarks of inflammation. In addition, programmed necrosis can directly induce cancer cell death or promote cancer growth and metastasis through its pro-inflammatory effects (Coussens and Werb, 2002). In this light, it is noteworthy that Non-Hodgkin's lymphomas-associated RIP3 SNPs have been identified (Cerhan et al., 2007). Thus, RIP3-dependent programmed necrosis may be a significant component in determining the outcome of viral diseases, trauma/injury-induced inflammation and cancers.
Details of siRNA library screen can be found in supplemental experimental procedures.
Fifty to one hundred million Jurkat cells were used for each immunoprecipitation. Cell lysates were prepared by lysis in 150 mM NaCl, 20 mM Tris-Cl [pH7.5], 0.2% NP-40, 1 mM EDTA, 3 mM NaF, 1 mM β-gylcerophosphate, 1 mM sodium orthovanadate and 10% glycerol. After pre-clearing with Sepharose 6B beads, Complex I and Complex II were immunoprecipitated with specific antibodies for 4 hours to overnight at 4°C. The resulting immune complexes were washed with lysis buffer and resolved on 4-12% NuPAGE gels (Invitrogen). The rabbit anti-human RIP3 polyclonal antibody was generated by ABR Affinity Bioreagents (supplemental experimental procedures). For in vitro kinase assays, immune complexes were incubated in kinase reaction buffer (20 mM HEPES [pH 7.5], 2 mM DTT, 1 mM NaF, 1 mM Na3VO4, 20 mM β-glycerophosphate, 20 mM MgCl2, 20 mM MnCl2, 1 mM EDTA, 300 μM ATP) supplemented with 10 μCi [32P] γ-ATP and 5 μg of MBP (Stressgen) for 30 minutes at 30°C. Samples were resolved on 4-12% NuPAGE gels (Invitrogen) and exposed to autoradiographic films. Densitometry measurements were performed using the Biorad VersaDoc MP4000 imaging station and analyzed using the Quantity One software.
Ten million FADD-deficient Jurkat cells were resuspended at 1 × 106 cells/ml of phosphate-free medium and incubated for 40 minutes at 37°C. Cells were spun down and resuspended in phosphate-free medium containing 1 mCi of [32P]-orthophosphate and incubated for 2 hours prior to stimulation with TNF for 2 hours. In some experiments, 30 μM of necrostatin-1 was added to the cells one hour prior to TNF stimulation. For MEFs, 3 million cells per IP sample were seeded one day before the labeling experiment. Cells were treated with TNF for 6 hours prior to cell lysis. Immunoprecipitations were performed with RIP1- or RIP3-specific antibody and resolved on SDS-PAGE.
For plasmid transfections, Jurkat cells were transfected with 20 μg of expression vectors using the BTX 630 Electro Cell Manipulator (262V, 725 ohms, 1050 μFd). After 16-20 hours, cells were treated with the indicated stimuli and cell death was measured by flow cytometry using PI as an indicator of cell death. Cell death was determined in the GFP positive transfected populations. Percentage cell loss is calculated using the following formula: % cell loss = (1−(number of live cells in treated sample/number of live cells in untreated sample)) × 100. For 293T cells, plasmids were transfected using Fugene 6 as per manufacturer's protocol (Roche). For siRNA transfections, cells were transfected with 150 nM siRNA oligonucleotides using HiPerfect (Qiagen) as per manufacturer's protocol. Transfected cells were tested for response to TNF 48 hours post-transfection. In all cell death assays, the mean ± SEM of triplicates was presented. Statistical significance was determined using the Student's t test.
To induce programmed necrosis in TNFR2+ Jurkat cells and L929 cells, we pretreated cells with 50 μM zVAD-fmk for one hour prior to stimulation with 100 ng/ml rhTNFα (rmTNFα for L929 cells). Programmed necrosis was induced in caspase-8 and FADD deficient Jurkat cells with 100 ng/ml rhTNFα. Cell death was determined by propidium iodide (PI) staining and flow cytometry. In some experiments, we stained cells with Annexin V and FAM-FLICA (for active caspases) to confirm that the cells were undergoing programmed necrosis, but not apoptosis. For MEFs, cells were pre-treated with 20 μM zVAD-fmk and 1 μg/ml cycloheximide for 1 hour before TNF stimulation. Cell death was determined using the CellTiter96 aqueous non-radioactive assay (MTS assay, Promega) and in some cases by PI uptake and 1 μM 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) on flow cytometry. For primary activated T-cells, splenocytes from RIP3−/− or wild type littermates were activated with 5 μg/ml Concanavalin A for 2 days, followed by another 2 days of culture in 100 U/ml IL-2. Where applicable, cells were treated with 50 μM zVAD-fmk or 1 μM geldanamycin for 1 hour prior to the induction of cell death. For the induction of zVAD-induced necrosis and oxidative stress cell death, L929 cells were transfected with siRNA and treated with different doses of zVAD-fmk (Sigma) for 24-48 hours to induce autophagic cell death, or 0.2 mM H2O2 for 6 hours to induce oxidative stress. Cell death was measured by MTS assay (Promega). Procedures for cell death assays with VV-infected cultures can be found in supplemental experimental procedures.
Details can be found in supplemental experimental procedures.
This work was supported by NIH grants AI065877 (F.K.C.) and AI17672 (R. Welsh) and departmental startup fund (to F.K.C.). Core resources supported by the Diabetes Endocrinology Research Center Grant DK32520 were also used. F.K.C. was a recipient of investigator awards from the Smith Family Foundation and the Cancer Research Institute. We thank M. Wu and M. Harrington for the RIP3 cDNAs, V. Dixit for the RIP3−/− mice, members of the Pathology department, E. Baehrecke and A. Winoto for discussion and critical reading of the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.