GRIM-19-mediated translocation of STAT3 to mitochondria is necessary for TNF-induced necroptosis

Tumor necrosis factor α (TNF) is a cytokine with a myriad of effects on varied cell types including adipocytes, hepatocytes and immune cells. However, its most studied and dramatic effect is in cell death. TNF does this by engaging TNF receptor 1 (TNFR1), which is the only TNF receptor capable of initiating apoptosis or necroptosis. TNF binding to TNFR1 causes receptor trimerization and the formation of an intracellular death-inducing signaling complex (DISC) (Ganten et al., 2004; Kim et al., 2000). The intracellular signaling network initiated by TNF is remarkably diverse and can result in the stimulation of survival pathways such as that mediated by NFκB or in the initiation of caspase activity. Pro-caspase-8 is recruited to the DISC where it exhibits proximity-induced activation (Ganten et al., 2004; Kuwana et al., 1998; Zhuang et al., 1999). In turn, caspase-8 can cleave and activate effector caspases such as caspase-3 that promulgate the apoptotic cascade. Inhibition of caspase activity prevents TNF-induced apoptosis. However, in some cell types, inhibition of caspase-8 leads to an alternative pathway that terminates in necrotic cell death (Vercammen et al., 1998). The programmed necrosis brought about by TNF has been termed necroptosis. The enzymatic activity of receptor-interacting serine/threonine-protein kinase 1 (RIPK-1) is required for necroptosis. The internalization of the TNFR leads to the assembly of a complex that includes TRADD, FADD, caspase- 8, RIPK-3 and RIPK-1, termed complex II. Remarkably, activated caspase-8 cleaves and inactivates RIPK-1 and 3, eventuating in apoptosis (Oberst et al., 2011). However, when caspase-8 is absent or inactivated, RIPK-1 and RIPK-3 become active in what has been termed a necrosome (Galluzzi et al., 2011; Vandenabeele et al., 2010; Vanlangenakker et al., 2011; Wallach et al., 2011). RIPK-3 phosphorylates and regulates the activity of RIPK-1, stabilizing the pro-necrotic RIPK-1–RIPK-3 complex (Cho et al., 2009). In turn, RIPK-1 activity is required for RIPK-3 phosphorylation and the necrotic activity of the necrosome, with inhibition of RIPK-1 by necrostatin-1 preventing necroptosis.

Inhibition of reactive oxygen species (ROS) production with antioxidants prevents TNF-induced necroptotic cell death (Goossens et al., 1995; Schulze-Osthoff et al., 1992). The RIPK-1–RIPK-3 necrosome triggers the increased production of ROS that characterizes and precedes TNF-induced cytotoxicity (Kasof et al., 2000; Lin et al., 2004; Zhang et al., 2009). There is evidence that mitochondrial energy metabolism is the source of necrosome-induced ROS formation. The necrosome stimulates the activities of glycogen phosphorylase and glutamate dehydrogenase, the products of which feed into the Krebs cycle and ultimately the mitochondrial respiratory chain. Components of the mitochondrial respiratory chain have been shown to be a source of ROS, with complex I being a particularly active producer of ROS under some circumstances. Indeed, site-specific inhibitors of complex I such as rotenone and amytal prevent TNF-induced cytotoxicity (Schulze-Osthoff et al., 1992).

STAT3 is localized to the cytosol and can translocate to the nucleus, where it is a transcriptional regulator (Demaria et al., 2010; Demaria and Poli, 2011). However, recent studies indicate that STAT3 is also localized to the mitochondria (Gough et al., 2009; Wegrzyn et al., 2009). In particular, phosphorylation of STAT3 on tyrosine 705 promotes its translocation to the nucleus, whereas phosphorylation on serine 727 is associated with a mitochondrial localization. Mitochondrial STAT3 increases the activity of complexes I and II of the respiratory chain in a transcription-independent manner. However, without a mitochondrial localization sequence, the mechanism by which STAT3 transits from the cytosol to the mitochondria is unclear. GRIM-19 (also known as NADH dehydrogenase [ubiquinone] 1 alpha subcomplex subunit 13; encoded by Ndufa13) is a component of complex I of the mitochondrial respiratory chain (Fearnley et al., 2001). Suppression of GRIM-19 reduced ROS formation and cell death induced by interferon-β and retinoic acid treatment (Huang et al., 2007). GRIM-19 interacts with STAT3, with serine727 of STAT3 being necessary for the GRIM-19–STAT3 interaction (Zhang et al., 2003). Like STAT3, GRIM-19 is also found in the mitochondria, cytosol and nucleus and is capable of translocating from one site to another upon stimulation by cytokines. In the study reported here we demonstrate that during TNF-induced necroptosis, STAT3 undergoes phosphorylation on serine727 that is dependent on RIPK-1 expression and activity. Upon phosphorylation of serine727, STAT3 interacts with GRIM-19, which brings about accumulation of STAT3 and GRIM-19 in the mitochondria, where they induce an increase in ROS production and cell death by necroptosis.

STAT3 expression was suppressed in L929 mouse fibrosarcoma cells using siRNA targeting STAT3. As seen in Fig. 1A, transfection of Stat3 siRNA greatly suppressed the expression of STAT3 while having little effect on the levels of STAT1. By contrast, siRNA targeting STAT1 had no effect on STAT3 expression but suppressed expression of STAT1. TNFα in the presence of the pan caspase inhibitor, Z-VAD-FMK (ZVAD), is known to induce necroptosis in L929 cells. As shown in Fig. 1B (left panel), L929 cells transfected with non-targeting siRNA displayed an extensive loss of cell viability when exposed to TNFα in the presence of ZVAD, with only 12% of the cells viable after 18 hours of exposure. By contrast, suppression of STAT3 expression prevented TNFα-induced necroptosis, with cell viability remaining at 88% after 18 hours of exposure (Fig. 1B, left panel). Notably, suppression of STAT1 expression did not prevent TNF+ZVAD-induced necrosis, with only 9% of the cells remaining viable after 18 hours of exposure (Fig. 1B, left panel).

TNF-induced necroptosis is mediated in part by ROS generated by mitochondria. MitoSOX is a potentiomeric dye that localizes to the mitochondria and exhibits an increase in fluorescence when oxidized by superoxide anions generated by the mitochondrial electron transport chain. As shown in Fig. 1B, right panel, treatment with TNF+ZVAD induced a marked increase in MitoSOX fluorescence that was not prevented by transfection with non-targeting siRNA or siRNA against STAT1. Notably, the peak in ROS production occurred 4 hours after TNF+ZVAD addition, a time that precedes any appreciable loss of cell viability. By contrast, suppression of STAT3 expression completely prevented TNF+ZVAD-induced ROS formation, with MitoSOX fluorescence the same as that of control non-treated cells (Fig. 1B, right panel).

STAT3 is phosphorylated on many residues with tyrosine 705 and serine 727 being the best characterized. As shown in Fig. 2A, in untreated cells, there was a low level of STAT3 phosphorylation on both tyrosine 705 and serine 727. Treatment of the cells with TNF alone for 4 hours did not stimulate phosphorylation of STAT3 on tyrosine 705 or serine 727 (Fig. 2A, lane 2). Likewise, treatment with ZVAD alone for 4 hours did not stimulate STAT3 phosphorylation on either residue (Fig. 2A, lane 3). However, exposure of the cells to TNF in the presence of ZVAD for 4 hours induced a marked increase in STAT3 phosphorylation on serine 727, while having little effect on the phosphorylation of tyrosine 705 (Fig. 2A, lane 4). As shown in Fig. 2B, the TNF+ZVAD-induced phosphorylation of STAT3 on serine 727 was first detectable at 30 minutes and became maximal by 4 hours of exposure, with no detectable change in STAT3 expression (less than 10% variation, according to densitometry).

We next wanted to determine the importance of TNF+ZVAD-induced phosphorylation of STAT3 for ROS production and cytotoxicity. L929 cells were generated with doxycycline inducible expression of non-phosphorylatable forms of STAT3 mutated at serine (S) 727 or tyrosine (Y) 705 to determine their effects on TNF-induced necrosis. Doxycycline induced the expression of FLAG-tagged STAT3 Y705A and STAT3 S727A at 24 hours (supplementary material Fig. S1). As shown in Fig. 2C, left panel, inducible expression of STAT3 Y705A provided no protection against TNF+ZVAD-induced necrosis, with only 14% of the cells viable after 18 hours of exposure. By contrast, cells with inducible expression of STAT3 S727A were refractory to TNF+ZVAD-induced necrosis, with 94% of the cells still viable after 18 hours of exposure. Similarly, expression of STAT3 S727A prevented the TNF+ZVAD-induced spike in ROS production, whereas expression of STAT3 Y705A had no inhibitory effect on TNF+ZVAD-induced ROS production (Fig. 2C, right panel).

We next wanted to determine the pathway responsible for the phosphorylation of STAT3 on serine 727 during TNF+ZVAD exposure. During necroptosis, engagement of the TNF receptor activates RIPK-1. As shown in Fig. 3A, siRNAs targeting RIPK-1 suppressed its expression in L929 cells, but not that of TNFR1. Suppression of RIPK-1 expression prevented both the cytotoxicity and generation of ROS provoked by exposure to TNF+ZVAD. Cells transfected with a non-targeting control siRNA exhibited only 15% viability following 18 hours of exposure to TNF+ZVAD (Fig. 3B, left panel). By contrast, cells transfected with siRNA targeting RIPK-1 displayed a marked resistance to TNF+ZVAD-induced necrosis, with 90% of the cells still viable after 18 hours of TNF+ZVAD treatment (Fig. 3B, left panel). Additionally, suppression of RIPK-1 expression also prevented the antecedent surge of ROS production provoked by exposure to TNF+ZVAD (Fig. 3B, right panel). Necrostatin-1 binds to and inhibits the activity of RIPK-1. As with suppression of RIPK-1 expression, a 1-hour pretreatment with necrostatin-1 prevented the necrosis and surge in ROS production brought about by TNF+ZVAD treatment, whereas an inactive analog of necrostatin, lacking a methyl group, failed to prevent TNF+ZVAD-induced cytotoxicity or ROS generation (Fig. 3C, left and right panels, respectively).

The phosphorylation of STAT3 on serine 727 brought about by TNF+ZVAD exposure for 4 hours was largely prevented by the suppression of RIPK-1 expression (Fig. 4A, lane 3). However, there was slight phosphorylation of STAT3 in cells where RIPK-1 expression was suppressed; despite this, cell death was completely negated, possibly indicating a threshold effect for cell killing. Similarly, pretreatment for 1 hour with necrostatin-1 also prevented the induction of STAT3 phosphorylation on serine 727 brought about by TNF+ZVAD treatment for 4 hours (Fig. 4B, lane 3), whereas pretreatment with the inactive analog of necrostatin-1 failed to inhibit TNF+ZVAD-induced STAT3 phosphorylation (Fig. 4B, lane 4). Importantly, suppression of RIPK-1 or treatment with necrostatin-1 had no effects on total STAT3 expression (Fig. 4A,B).

STAT3 is localized to the mitochondria (Gough et al., 2009; Wegrzyn et al., 2009). Because TNF+ZVAD exposure induced mitochondrial ROS production, we wanted to determine whether TNF+ZVAD had any effect on mitochondrial STAT3 levels. Mitochondria were isolated from L929 cells and mitochondrial lysates probed by western blotting for the levels of STAT3 and serine-727-phosphorylated STAT3. As shown in Fig. 4C (panel 1, lane 1), mitochondria isolated from control cells exhibited a low basal level of STAT3. After 4 hours of TNF+ZVAD exposure, there was a marked accumulation of STAT3 in mitochondrial lysates (Fig. 4C, panel 1, lane 2). Importantly, most of the STAT3 that accumulated in the mitochondria was phosphorylated on serine 727 (Fig. 4C, panel 2, lane 2). Moreover, pretreatment with necrostatin-1 prevented the TNF+ZVAD-induced accumulation of STAT3 in the mitochondria (Fig. 4C, panels 1 and 2, lane 3), whereas the inactive analog of necrostatin-1 did not (Fig. 4C, panels 1 and 2, lane 4). Moreover, as shown in Fig. 5A, lane 4, induction of STAT3 S727A expression prevented TNF+ZVAD-induced phosphorylation of endogenous STAT3, whereas expression of STAT3 Y705A had no effect (Fig. 5A, lane 3). Additionally, as shown in Fig. 5B, lane 2, immunoprecipitation of RIPK-1 revealed that a 2-hour exposure to TNF+ZVAD dramatically increased the interaction between STAT3 and RIPK-1, which was prevented by overexpression of STAT3 S727A (Fig. 5B, lane 4) but was not prevented by STAT3 Y705A (Fig. 5B, lane 3).

GRIM-19 is a component of complex I of the mitochondrial respiratory chain and has been implicated in the induction of cell death. Like STAT3, GRIM-19 also localizes to the cytosol, mitochondria and nucleus. Furthermore, GRIM-19 is known to bind to STAT3 with serine 727 being crucial for the interaction. Because STAT3 does not possess a mitochondrial targeting sequence, we explored the possibility that GRIM-19 helps to chaperone STAT3 to the mitochondria. As shown in Fig. 6A, lane 1, as expected for a component of mitochondrial complex I, mitochondria isolated from control cells contained a notable level of GRIM-19. However, treatment with TNF+ZVAD for 4 hours caused a striking increase in the level of mitochondrial GRIM-19 (Fig. 6A, lane 2). The stimulation in the level of mitochondrial GRIM-19 brought about by exposure to TNF+ZVAD was dependent on RIPK-1 activity. The RIPK-1 inhibitor, necrostatin-1, prevented the accumulation of mitochondrial GRIM-19 induced by TNF+ZVAD (Fig. 6A, lane 3), whereas the inactive analog of necrostatin-1 failed to prevent the increase of mitochondrial GRIM-19 levels in TNF+ZVAD-treated cells (Fig. 6A, lane 4).

GRIM-19 has been implicated in the induction of ROS generation and cell death. As shown in Fig. 6B, transfection with siRNA targeting GRIM-19 suppressed its expression in L929 cells while having no effect on the levels of STAT3. Similarly, transfection with siRNA targeting STAT3 did not diminish expression of GRIM-19 (Fig. 6B). As shown in Fig. 6C, left panel, suppression of GRIM-19 expression prevented TNF+ZVAD-induced necrosis. In cells transfected with non-targeting siRNAs, TNF+ZVAD reduced cell viability to 12% following 18 hours of treatment. By contrast, suppression of GRIM-19 levels greatly ameliorated TNF+ZVAD-induced necrosis, with 87% of the cells still viable after 18 hours of exposure to TNF+ZVAD. Similarly, suppression of GRIM-19 expression also prevented the stimulation of ROS production incited by TNF+ZVAD (Fig. 6C, right panel).

Given the known association between STAT3 and GRIM-19, we wanted to determine whether this interaction is modulated during TNF+ZVAD-induced necroptosis. As shown in Fig. 7A, lane 1, immunoprecipitation of GRIM-19 from lysates prepared from mitochondria isolated from control cells detected some interaction between GRIM-19 and STAT3. However, there was a marked stimulation in the interaction of GRIM-19 with STAT3 in lysates prepared from mitochondria isolated after 4 hours of TNF+ZVAD treatment (Fig. 7A, lane 2). Importantly, the enhanced interaction between STAT3 and GRIM-19 induced by TNF+ZVAD was dependent on RIPK-1 activity. Pretreatment with the RIPK-1 inhibitor necrostatin-1, prevented the increased interaction between STAT3 and GRIM-19 brought about by TNF+ZVAD exposure (Fig. 7A, lane 3). By contrast, the stimulation of GRIM-19–STAT3 interaction induced by TNF+ZVAD treatment was not prevented by the inactive necrostatin-1 analog (Fig. 7A, lane 4). We next determined the functional significance of the augmented interaction between GRIM-19 and STAT3 with regards to the increased levels of STAT3 and GRIM-19 in the mitochondria of TNF+ZVAD-treated cells. As shown in Fig. 7B, left panel, suppression of GRIM-19 expression prevented the mitochondrial accumulation of STAT3 upon treatment with TNF+ZVAD. Similarly, suppression of STAT3 levels prevented the accumulation of GRIM-19 in the mitochondria that is triggered by TNF+ZVAD treatment (Fig. 7B, right panel). Interestingly, although suppression of GRIM-19 expression prevented TNF+ZVAD-induced accumulation of STAT3 in the mitochondria, it did not prevent the RIPK-1-dependent phosphorylation of STAT3 in the cytosol (Fig. 7C, lane 3). These data suggest that the interaction between GRIM-19 and STAT3 is required for their mutual translocation to the mitochondria upon TNF+ZVAD exposure, but not for STAT3 phosphorylation.

STAT3 S727D was used as a constitutively serine-727-phosphorylated mimetic of STAT3. L929 cells were generated in which the expression of STAT3 S727D was under the control of a tetracycline-inducible promoter. As shown in Fig. 8A, STAT3 S727D expression induced a 62% loss of viability in L929 cells within 48 hours of doxycycline induction, which was not prevented by non-targeting siRNA. However, suppression of GRIM-19 expression prevented the loss of cell viability brought about by STAT3 S727D expression, with 92% of the cells still viable after 48 hours after induction of STAT3 S727D expression with 1 µg/ml doxycycline. These data indicate that GRIM-19 is required for STAT3 S727D to exert a cytotoxic effect (Fig. 8A). As shown in Fig. 8B, lane 2, mitochondria isolated from cells expressing STAT3 S727D exhibited an elevation of GRIM-19 levels in mitochondrial lysates over that of controls (45% increase by densitometry), whereas cells overexpressing STAT3 S727A exhibited a decreased level of GRIM-19 in the mitochondria (Fig. 8B, lane 3, 69% decrease by densitometry).

The present study presents data indicating that GRIM-19 and STAT3 are crucial components of the TNF-induced necroptosis pathway. STAT3 is phosphorylated on serine 727 when the TNF receptor is engaged in the absence of caspase activation. The phosphorylation of STAT3 is dependent on RIPK-1 expression and activity. The phosphorylation of STAT3 on serine 727 increases its interaction with GRIM-19, leading to their accumulation in the mitochondria where they incite an increase in ROS generation resulting in cell death by necroptosis (Fig. 9).

STAT3 is a transcription factor stimulated by the interleukin-6 family of cytokines and growth factors. In the nucleus, STAT3 induces the expression of a number of genes crucial for cell survival and growth (Bernier et al., 2011; Demaria et al., 2010; Demaria and Poli, 2011). In tumor cells, STAT3 phosphorylation on tyrosine 705 induces dimerization and nuclear localization. However, recent reports also indicate a mitochondrial localization for STAT3. In the mitochondria, STAT3 seems to be necessary for optimal functioning of the mitochondrial respiratory chain (Bernier et al., 2011; Qiu et al., 2011). Mitochondria isolated from cardiomyocytes of Stat3^−/− mice showed a decrease of complex I and II activities (Wegrzyn et al., 2009). Moreover, expression of STAT3 targeted exclusively to the mitochondria restored mitochondrial respiration. Similarly, Ras-dependent transformation was reconstituted in Stat3^−/− cells with STAT3 mutants that were transcriptionally inactive by mutation of tyrosine 705 but able to localize to mitochondria (Gough et al., 2009). By contrast, mutation of serine 727 to alanine negated the ability of STAT3 to restore Ras-dependent transformation and mitochondrial activity, whereas substitution with the phosphorylation mimetic, STAT3 S727D, enhanced both Ras transformation and mitochondrial function. Similarly, in neuronal cells, nerve growth factor (NGF) induced phosphorylation of STAT3 on serine 727, resulting in an increased mitochondrial localization of STAT3 and elevation of mitochondrial ROS production, both necessary for NGF-stimulated neurite outgrowth (Zhou and Too, 2011). It has also been demonstrated that STAT3 targeted to the mitochondria protects against ischemia-induced changes in the activity of mitochondrial complex I and II and binds to cyclophilin-D (Boengler et al., 2010; Szczepanek et al., 2011). How mitochondrial STAT3 induces alterations of mitochondrial function is unclear, because a recent study indicated that, under basal conditions, STAT3 is not present in the mitochondria sufficiently to interact stoichiometrically with components of complex I or II (Phillips et al., 2010).

Similar to STAT3, GRIM-19 is also found in many cellular compartments, with localization being modified by cytokines and growth factors. GRIM-19 is a component of complex I of the mitochondrial respiratory chain and is required for maintenance of mitochondrial membrane potential (Fearnley et al., 2001; Lu and Cao, 2008). GRIM-19 binds to and suppresses the transcriptional activity of STAT3, with serine 727 being required for the STAT3–GRIM-19 interaction (Lufei et al., 2003; Nallar et al., 2010; Zhang et al., 2003). GRIM-19 is selective for STAT3 and is not known to bind the other STAT family members. By binding to nuclear STAT3, GRIM-19 suppresses the expression of proteins required for cell metabolism, cell cycle progression and the Bcl-2 family of anti-apoptotic proteins, leaving the cell vulnerable to injury. In this way GRIM-19 behaves as a tumor suppressor (Kalakonda et al., 2007; Okamoto et al., 2010; Wang et al., 2011). Moreover, there is evidence that the mitochondrial localization of GRIM-19 in and of itself enhances cytokine-induced cytotoxicity. Interferon-β in combination with retinoic acid potently induces cellular demise through a ROS-dependent mechanism. Indeed, treatment with interferon-β and retinoic acid induces an increase in GRIM-19 in the mitochondria that colocalizes with elevated levels of STAT3. Lowering GRIM-19 levels ameliorated interferon-β–retinoic-acid-induced cell death and ROS generation (Huang et al., 2007). Interestingly a 2.7 kb virally encoded RNA from human cytomegalovirus binds to GRIM-19 and inhibits cytokine-induced cell death by preventing GRIM-19 accumulation in the mitochondria (Reeves et al., 2007). Such observations indicate that GRIM-19 has dual functions that work in tandem in mediating cell injury. In the nucleus, GRIM-19 binds to STAT3 and suppresses expression of proteins that enhance cell survival, whereas in the mitochondria, GRIM-19 induces ROS generation, resulting in cell injury and death.

The RIPK-1–RIPK-3 necrosome activates enzymes involved in energy metabolism, such as glutamate dehydrogenase and glycogen phosphorylase (Zhang et al., 2009). The activation of such metabolic enzymes would feed substrates into the Krebs cycle, producing NADH that stimulates mitochondrial respiratory chain activity. The phosphorylation of STAT3 and its accumulation in the mitochondria with GRIM-19 would further enhance mitochondrial respiration. In some instances, the stimulation of respiratory activity triggered by mitochondrial STAT3 is protective, especially against ischemic damage (Szczepanek et al., 2011). So by an uncharacterized mechanism, during the induction of necroptosis, the physiological and protective response of increased mitochondrial respiration induced by STAT3 undergoes metamorphosis into a damaging event. Presently, the data demonstrate that RIPK-1 expression and activity are required for phosphorylation of STAT3 during TNF-induced necroptosis, but do not definitively identify RIPK-1 as the kinase responsible for the phosphorylation of STAT3. It is conceivable that RIPK-1 activates downstream kinases that in turn phosphorylate STAT3. Indeed, STAT3 is phosphorylated on serine 727 by a number of kinases including mTOR that do not necessarily result in necroptosis (Ge and Ren, 2012; Yokogami et al., 2000; Yokoyama et al., 2007). Whether phosphorylation of STAT3 on serine 727 induces interaction with GRIM-19 and their mitochondrial relocalization in other cellular contexts is unknown. The degree to which GRIM-19 and STAT3 accumulate in the mitochondria could influence the cells fate. Currently, however, the mechanism(s) that regulate the influx and egress of GRIM-19 and STAT3 to and from mitochondria is not characterized.

L929 mouse fibrosarcoma cells were maintained in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum, 100 units/ml penicillin and 100 µg/ml streptomycin under an atmosphere of 95% air and 5% CO₂ at 37°C. Cells were subcultured 1:5 once a week.

L929 cells were plated in either 24-well plates at 50,000 cells/well or in 25 cm² flask at 1.0×10⁶ cells. Following overnight incubation, cells were treated with 20 ng/ml TNFα (Santa Cruz Biotechnology) in the presence or absence of 20 µM Z-VAD-FMK (Santa Cruz Biotechnology). Where indicated, L929 cells were transfected with the indicated siRNAs targeting STAT3, STAT1, GRIM-19 or with a non-targeting control siRNA using a lipid-based method supplied from a commercial vendor (Gene Therapy Systems, San Diego, CA) at a final siRNA concentration of 50 nM. After formation of the siRNA–liposome complexes, the mixture was added to the L929 cells for 24 hours. Afterward, the medium was aspirated, and complete medium was added back for a further 24 hours, after which time the cells were used for experiments. Tetracycline-inducible L929 cells were generated using the Tet-ON/Tet-Off system (Promega). Where indicated, expression of STAT3 mutants was induced by incubating with 1 µg/ml doxycycline. Expression of STAT3 mutants was maximal at 24 hours after doxycycline addition, at which time, where indicated, the cells were treated with TNF and ZVAD.

Following treatments, L929 cells were harvested and centrifuged at 700 g. The cell pellet was resuspended in phosphate-buffered saline to which was added 5 µM propidium iodide. After 5 minutes incubation, the cells were pelleted and resuspended in PBS. The percentage of viable cells was determined, using a Cellometer (Nexelom, Lawrence, MA), as the ratio of the number of cells in the fluorescent images (propidium iodide positive) to the bright-field images. For measurement of ROS production, 5 µM MitoSOX (Invitrogen) was added to cells 10 minutes before harvesting. The cells were pelleted and resuspended in PBS. In cells with active production of ROS, MitoSOX is oxidized to a fluorescent species. The percentage MitoSOX-positive cells was determined using a Cellometer, which calculated the ratio of the number of MitoSOX-positive cells in the fluorescence images to the number of cells in the bright-field images.

Following treatments, approximately 1.0×10⁶ cells from a 25 cm² flask were harvested by trypsinization and centrifuged at 700 g for 10 minutes at 4°C as described previously (Shulga et al., 2010). The cell pellets were washed once in PBS and then resuspended in 3 volumes of isolation buffer (20 mM Hepes, pH 7.4, 10 mM KCl, 1.5 mM MgCl₂, 1 mM sodium EDTA, 1 mM dithiothreitol, 10 mM phenylmethylsulfonyl fluoride, 10 µM leupeptin and 10 µM aprotinin) in 250 mM sucrose. After chilling on ice for 3 minutes, the cells were disrupted by 40 strokes of a glass homogenizer. The homogenate was centrifuged twice at 1500 g at 4°C to remove unbroken cells and nuclei. The mitochondria-enriched fraction (heavy membrane fraction) was then pelleted by centrifugation at 12,000 g for 30 minutes. The supernatant was removed and then filtered through a 0.2 µm and then a 0.1 µm Ultrafree-MC filter to obtain the cytosolic fraction.

GRIM-19 or RIPK-1 was immunoprecipitated from mitochondrial extracts or whole-cell extracts, respectively (GRIM-19 antibody from Novus Biologicals and RIPK-1 antibody from Cell Signaling). The immunoprecipitates were then run on SDS-PAGE gels and blotted onto PVDF membranes. The western blots were developed with antibodies against STAT3 phosphorylated on serine 727 (Cell Signaling).

Results are expressed as means ± s.d. of at least three independent experiments. Statistical significance was defined at P<0.05.