Both apoptosis and necroptosis are forms of programmed cell death and are initiated by tumor necrosis factor (TNF)-α stimulation. A growing number of studies described molecular switches between the two programmed cell death types, such as receptor-interacting protein 3 (RIP3), a key molecule that determines specific necrosome formation in the case of TNF-α-induced necroptosis in a receptor-interaction protein 1 (RIP1)-dependent manner (Cho et al., 2009; He et al., 2009; Zhang et al., 2009). Upon TNF-α binding, the TNF receptor (TNFR) trimerizes and mediates either apoptosis via caspase 8 activation or necroptosis via RIP1-RIP3 interaction. During necroptosis, TNFR-mediated RIP1activation and phosphorylation leads to subsequent RIP3 phosphorylation through the RHIM domain. The RIP1-RIP3 complexes recruit and phosphorylate the downstream pseudo-kinase mixed lineage kinase domain-like to form a necrosome complex. Finally, the necrosome complexes translocate to the plasma membrane and cause membrane swelling and rupture (Tang et al., 2019; Vanden Berghe et al., 2014; Weinlich et al., 2017).

Reactive oxygen species (ROS) are highly reactive molecules including superoxide anion, hydrogen peroxide (H₂O₂), and hydroxyl radical. Two major ROS sources related to the TNF-α-induced cell death are NADPH oxidase (NOX) and mitochondria (Blaser et al., 2016; Morgan and Liu, 2010). It has long been thought that ROS play pivotal roles in TNF-α-induced cell death in various cell types. Nonetheless, the precise role of ROS in necroptosis is still debated.

In this study, we investigated this question using specific probes that distinguish specific ROS types. Unexpectedly, the cytosolic H₂O₂ was a major ROS type produced related to apoptosis, whereas mitochondrial superoxide accumulation was a key phenomenon in necroptosis. More interestingly, NOX was involved in mitochondrial superoxide production in necroptosis.

Antibodies and reagents

Anti-human RIP3 (Cat. No. 13526) and anti-human SOD2 (Cat. No. 13194S) antibodies were purchased from Cell Signaling Technology (USA). Anti-mouse RIP3 antibody (Cat. No. NBP1-77299) was purchased from NOVUS Biologicals (USA). Anti-human PrxIII (Cat. No. LF-MA0045), anti-Prx-SO_2/3 (Cat. No. LF-PA0004), and anti-α-tubulin (Cat. No. LF-MA0117) antibodies were purchased from AbFrontier (Korea). HRP-conjugated rabbit IgG and mouse IgG antibodies were purchased from Bio-Rad (USA). Human TNF-α (Cat. No. PHC3016) was purchased from Gibco (USA). Hydrogen peroxide (Cat. No. H1009) and cycloheximide (Cat. No. C7698) were purchased from Sigma-Aldrich (USA). zVAD (Cat. No. 627610), BV6 (Cat. No. S7597), and MitoPY1 (Cat. No. 4428) were purchased from Calbiochem (USA), Selleckchem (USA), and TOCRIS (UK), respectively. MitoSOX Red (Cat. No. M36008) and dihydroethidium (Cat. No. D11347) were purchased from Invitrogen (USA).

Cell culture

HeLa, HT-29, NIH-3T3, and L929 cells were purchased from the American Type Culture Collection (USA). HeLa and L929 cells were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (Gibco), 100 units/ml of penicillin, 100 units/ml of streptomycin (Lonza, Switzerland). HT-29 cells were cultured in McCoy’s 5A medium supplemented with 10% fetal bovine serum (Gibco), 100 units/ml of penicillin, 100 units/ml of streptomycin (Lonza). NIH-3T3 cells were cultured in DMEM supplemented with 10% calf bovine serum (Gibco), 100 units/ml of penicillin, 100 units/ml of streptomycin (Lonza). These cells were maintained at 37°C under an atmosphere of 5% CO₂.

RNA interference

The human and mouse RIP3-specific siRNA duplex sequences (5’-GCA GUU GUA UAU GUU AAC GAG CGG UCG-3’ and 5’-AAG AUU AAC CAU AGC CUU CAC CUC CCA-3’) were selected according to a previous study (Cho et al., 2009). The sequence-verified human SOD2-specific siRNA was purchased from Santa Cruz Biotechnology (USA). A control siRNA against firefly luciferase was purchased from Dharmacon (USA). The siRNA was transfected for 48 h using the LipoJet™ reagent (SignaGen, USA).

Measurement of intracellular ROS

Mitochondrial and cytosolic superoxide anion levels were determined using MitoSOX Red and dihydroethidium (DHE) (Molecular Probes, USA), respectively. The mitochondrial H₂O₂ level was assessed using MitoPY1 (Molecular Probes). Cells were cultured on 35-mm glass-bottom dishes and treated with either TNF-α (20 ng/ml) plus cycloheximide (10 µg/ml) or TNF-α plus BV6 (1 µM) in the presence of a pan-caspase inhibitor zVAD-fmk (20 µM) for apoptosis or necroptosis induction, repectively. After the stimulation, cells were rinsed once with Hanks’ Balanced Salt Solution and incubated for 15 min with the indicated dyes. Fluorescence images were taken using a Zeiss LSM 880 confocal laser scanning microscope and analyzed using the Image J software.

Ratiometric imaging of cytosolic H₂O₂ was assessed using an H₂O₂-sensing and cytosol-targeted fluorescence protein called HyPer-Cyto (Evrogen, Russia). Cells were transfected with the plasmid vector encoding HyPer-Cyto for 16 h. After the transfection, the cells were treated with indicated stimuli. The fluorescence images were obtained with two excitation wavelengths at 405 nm and 488 nm. The fluorescence intensity ratios (F₄₈₈/F₄₀₅) represent the cytosolic H₂O₂ level.

Immunoblotting

Cells were lysed in a chilled lysis buffer containing 20 mM HEPES (pH 7.0), 150 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EGTA, 1 mM EDTA, 5 mM Na₃VO₄, 5 mM NaF, 1 mM AEBSF, aprotinin (5 µg/ml), and leupeptin (5 µg/ml). The lysates were centrifuged for removing cell debris and nuclei. The protein concentration in the supernatant was measured using the Bradford assay. Proteins were resolved by SDS-PAGE and transferred onto nitrocellulose membranes by electroblotting. Immunoblot analysis was performed with specific antibodies following the standard protocol. The immune complexes with horseradish peroxidase-conjugated secondary antibodies were visualized by enhanced chemiluminescence on ChemiDoc imaging system (Bio-Rad).

Necroptotic cell death assay

Cells (2 × 10⁵) were seeded onto 6-well plates and cultured for 24 h. Cells were treated with indicated stimuli for 6 h and then collected into a 5 ml FACS tube by pipetting. Then, the cells were centrifuged for 3 min, washed with cold phosphate-buffered saline (PBS). The final cell pellets were stained using Annexin V-FITC apoptosis detection kit I (BD Biosciences, USA). Cells were stained with Annexin V–FITC (BD Biosciences; Cat. No. 556419) for 20 min, then propidium iodide (Molecular Probes; Cat. No. P1304) for 5 min. The stained cells were analyzed using the FACS Calibur system (BD Biosciences).

SOD activity assay

Intracellular SOD activity was measured with a Colorimetric Assay Kit (BioVision, USA) using WST-1 according to the manufacturer’s instructions. The washed cells were lysed in ice-cold 0.1 M Tris/HCl, pH 7.4 containing 0.5% Triton X-100, 5 mM β-ME, 0.1 mg/ml PMSF. Whole cell lysates (20 µl) were incubated with 200 µl of WST-1 working solution together with 20 µl of Enzyme Solution for 20 min at 37°C. WST-1 would be converted to WTS-1 formazan by remaining superoxide in assay solutions. The amount of formazan was measured at 450 nm using an Epoch™ 2 Microplate Spectrophotometer (BioTek, USA).

NADH/NAD assay

Cells (2 × 10⁵) were seeded onto 6-well plates and cultured for 24 h. After the indicated stimulation, the cells were washed thrice with PBS and lysed with 50 μl of 0.2 N NaOH with 1% dodecyltrimethylammonium bromide (DTAB) solution. The ratio of NAD⁺/NADH was measured by the NAD/NADH-Glo assay kit (Promega, USA) according to the manufacturer’s instructions. To measure the oxidized forms (NAD⁺), 50 μl of lysate was incubated with 25 μl of 0.4 N HCl at 60°C for 15 min. The samples were then incubated at room temperature for 10 min and supplemented with 25 μl of 0.5 M Tris base for neutralization. To measure the reduced forms (NADH), 50 μl of lysate was heated at 60°C for 15 min and incubated at room temperature for 10 min. Then, 50 μl of 0.25 M Tris in 0.2 N HCl was added for neutralzation. The NAD⁺ and NADH levels were measured individually using a luminometer (PerkinElmer, USA).

Statistics

All experiments were repeated at least three times. Data were analyzed with the Student’s t-test for comparisons between two groups (IBM SPSS Statistics ver. 25; IBM, USA) to determine the statistically significance. P values < 0.05 were considered statistically significant.

Characterization of TNFα-induced ROS generation in apoptosis and necroptosis

To determine the type of ROS generated during necroptosis, we selected two RIP3-positive cell lines, HT-29 and L929, for necroptosis studies (Fig. 1A). We also used HeLa cells that undergo apoptosis upon TNF-α stimulation. Briefly, TNF-α plus cycloheximide (CHX) treatment induces apoptosis in HeLa cells. Treatment of either TNF-α plus a pan-caspase inhibitor z-VAD-fmk (T/Z combination) or TNF-α/z-VAD plus a SMAC mimetic BV6 (T/B/Z combination) induces necroptosis in murine L929 cells or human HT-29 cells, respectively. The stimulated cells were labeled with various ROS probes and subjected to time-lapse imaging on a confocal microscope. First, we measured the mitochondrial ROS level in two ways. Labeling with MitoSOX showed that the mitochondrial superoxide level was persistently increased in both HT-29 and L929 cells, not in the HeLa cells (Fig. 1B). In contrast, labeling with MitoPY1, monitoring the mitochondrial H₂O₂ level, showed no positive fluorescence signal in either of HT-29 and L929 cells, but a strong positive fluorescence signal increase in HeLa cells that underwent apoptosis (Fig. 1C). Then, we measured the cytosolic ROS level similarly in two ways. Labeling with DHE monitoring the cytosolic superoxide level showed a broad superoxide level increase in both the HT-29 and L929, not HeLa cells (Fig. 1D). Inversely, HyPer-cyto protein expression, a cytosolic-targeted green fluorescent protein variant that monitors the cytosolic H₂O₂ level (Belousov et al., 2006), indicated a strong H₂O₂ level increase in the cytosol of HeLa, but not HT-29 and L929 cells that underwent necroptosis (Fig. 1E, Supplementary Fig. S1A). Taken together, we concluded that necroptosis and apoptosis involve the differential production of superoxide and H₂O₂ in the mitochondria.

Mitochondrial superoxide production is RIP3-dependent

Since we reported the role of cytosolic H₂O₂ in the apoptosis (Lee et al., 2019), we were interested in the role of mitochondrial ROS during necroptosis. Several studies have previously shown that RIP3 affects in the TNF-induced necroptosis and ROS generation (He et al., 2009; Zhang et al., 2009). Therefore, we examined the mitochondrial superoxide level in the RIP3-depleted HT-29 cells (Fig. 2A). RIP3 knockdown reduced indeed necroptosis in HT-29 cells stimulated with T/B/Z combination (Fig. 2B) as well as in L929 cells stimulated with T/Z combination (Supplementary Fig. S1B). Subsequently, the RIP3 knockdown also suppressed the TNF-α-induced mitochondrial superoxide production, measured by MitoSOX (Figs. 2C and 2D). Therefore, the data indicated that RIP3 is essential for necroptosis-related mitochondrial superoxide production.

NADPH oxidase produces the mitochondrial superoxide anion

A recent study showed that the RIP3-dependent aerobic respiration enhancement contributed to cellular ROS levels in T/Z combination-stimulated cancer cells (Yang et al., 2018). Therefore, we measured the NADH/NAD⁺ ratio, to assess the mitochondrial respiration state (Stein and Imai, 2012), in the HT-29 cells that underwent necroptosis. As shown in Fig. 3A, the NADH/NAD⁺ ratio was unchanged during necroptosis, suggests that mitochondrial energy metabolism was not responsible for TNF-α-induced necroptosis. Accordingly, we decided to test another ROS source called NOX, catalyzing the electron transfer from NADP`H to molecular oxygen for superoxide anion generation. We used a pan-NOX inhibitor named APX-115 (Cha et al., 2017; Joo et al., 2016) for this experiment. Indeed, APX-115 completely abrogated the mitochondrial superoxide production in T/B/Z combination-stimulated HT-29 cells (Figs. 3B and 3C). As NOX4 is a mitochondrial isoform (Block et al., 2009), APX-115 might target NOX4 in the mitochondria of HT-29 cells.

We further addressed whether the superoxide-metabolizing enzyme present in mitochondria could be involved in mitochondrial superoxide accumulation. Firstly, we examined the activity or status of manganese-superoxide dismutase (Mn-SOD, also called SOD2) that catalyzes the dismutation of superoxide anion to H₂O₂ (Wang et al., 2018). We observed neither cellular SOD2 level nor total SOD activity alteration in HT-29 cells upon T/B/Z combination (Supplementary Figs. S2A and S2B). However, the SOD2 knockdown increased the mitochondrial superoxide level (Supplementary Figs. S2C and S2D), confirming that the mitochondrial ROS in necroptosis is indeed the superoxide anion.

The superoxide anion is a short-lived free radical and it is spontaneously or enzymatically converted to H₂O₂ in the cytosol as well as in the mitochondria. Since SOD2 was intact, the accumulated superoxide anions are constantly converted to H₂O₂ molecules in the mitochondria, which are in turn reduced to water by the mitochondrial peroxidase peroxiredoxin (Prx) type III (Chang et al., 2004). Since Prx enzymes are inactivated by hyperoxidation during repeated reaction cycles, we then examined the Prx enzymes status using a specific antibody recognizing the hyperoxidized form of the Prx enzymes, designated as the Prx sulfinic/sulfonic form (Prx-SO_2/3) (Woo et al., 2003). As a result, the hyperoxidized Prx enzymes were not present in the HT-29 and L929 cells that underwent necroptosis, although they were detected in the same cells exogenously treated with a bolus of H₂O₂ solution at a micromolar concentration (Supplementary Fig. S2E). Taken together, the results showed that the mitochondrial ROS-metabolizing systems function well.

Mitochondrial superoxide accumulation is a result of necroptosis

A major question was whether mitochondrial superoxide accumulation is a cause or result of necroptosis. To better understand the role of mitochondrial superoxide in necroptosis, we pre-treated the HT-29 cells with various ROS inhibitors and performed a necroptosis assay. First, we applied chemical ROS scavengers, such as Necrox-2 and Mito-Tempo, which specifically eliminate the mitochondrial superoxide anion (Dikalova et al., 2010; Kim et al., 2010). Necrox-2 and or Mito-Tempo treatments indeed completely eliminated the mitochondrial superoxide level in T/B/Z combination-stimulated HT-29 cells (Figs. 4A and 4B). Nonetheless, neither chemicals affected the TNF-α-induced necroptosis in the HT-29 and L929 cells (Fig. 4C). To further confirm that, we used a pan-NOX inhibitor APX-115, which still did not affect TNF-α-induced necroptosis in either HT-29 or L929 cells (Figs. 4D and 4E). Finally, we examined necroptosis in SOD2-depleted HT-29 cells. SOD2 knockdown did not affect the TNF-α-induced necroptosis in the HT-29 cells (Fig. 4F). Overall, we concluded that the mitochondrial superoxide accumulation is a consequence of necroptosis.

Accumulating data indicate that ROS levels correlate with the caspase-independent cell death pathway (Dixon and Stockwell, 2014; Kim et al., 2007; Yang et al., 2018; Zhang et al., 2017; Zhao et al., 2012). In particular, TNF-α-induced necroptosis involves excessive ROS production. However, where and how ROS regulates necroptosis remain mostly uncharacterized. Recently, we demonstrated that the cytosolic H₂O₂ acts a signaling messenger role in TNF-α-induced apoptosis. In that study, we risen the cytosolic H₂O₂ level by cellular thiol peroxidases, Prx and glutathione peroxidase (Gpx), depletion. These two peroxidase enzymes eliminate cellular H₂O₂ by utilizing the NADPH-derived reducing power (Kang and Kang, 2013; Veal et al., 2007). Specifically, Prx type I depletion increased DNA damage-dependent apoptosis, whereas that of Prx type II enhanced c-IAP/caspase-8 activation (Lee et al., 2019). More interestingly, Gpx1 depletion promoted apoptosis-inducing kinase 1 (ASK1)/c-Jun N-terminal kinase (JNK) pathways (Lee et al., 2021). Therefore, we interpreted that cytosolic H₂O₂ is an essential apoptosis signaling messenger. Since then, we decided to characterize necrosis-associated ROS level and type. For the first time, our data revealed mitochondrial superoxide anions accumulation in RIP3-positive cancer cells that underwent the TNF-α-induced necroptosis. Unlike the cytosolic H₂O₂ executing apoptosis, the mitochondrial superoxide did not execute necroptosis. Our results are consistent with those of previous studies showing that ROS production is dispensable in necroptosis of several cell types (He et al., 2009; Tait et al., 2013). However, it is potentially conceivable that the increased mitochondrial superoxide is involved in the release of damage-associated molecular pattern (DAMP), such as TNF, interleukin, or mtDNA, from damaged cells. Recent studies indicated that necroptosis could be a physiological endogenous inducer of inflammatory response via RIP3-dependent release of DAMPs during necroptotic cell death, and mitochondrial oxidative stress liberates mitochondrial DNA into the cytosol in which the interferon-dependent inflammatory response is triggered via the cGAS-STING pathway (Kim et al., 2019; Seong et al., 2020; Silke et al., 2015). Therefore, the accumulated superoxide anions within the mitochondria might be necessary for the release of pro-inflammatory factors such as mtDNA.

Our data also provide a key evidence of the fact that ROS types are different between apoptosis and necroptosis. The superoxide anion is a short-lived free radical, immediately converted to H₂O₂ by SOD (Imlay, 2008; Wang et al., 2018). Therefore, it was surprising that superoxide anions accumulated in the mitochondria of necroptotic cancer cells, where SOD2 and Prx type III are intact and abundant. However, we showed that mitochondrial superoxide production is mediated by the NOX enzyme, not the NADH-dependent respiratory energy metabolism, during TNF-induced necroptosis. Although NOX4 is known as a constitutively-active isoform in mitochondria (Block et al., 2009), we could not exclude the existence of a regulatory mechanism underlying the mitochondrial NOX activity enhancement by necroptosis.

In summary, the uniqueness of our study is live-cell imaging using various ROS-sensing probes to analyze the nature and localization of ROS in the necroptotic cells. As a result, our data demonstrated that superoxide anion, not H₂O₂, levels robustly increased in the mitochondria of RIP3-positive cancer cells undergoing necroptosis. This is a significant contrast to the robust increase of cytosolic H₂O₂ in TNF-α-induced apoptosis. Hence, the mitochondrial superoxide accumulation can be a useful marker of necroptosis.

We thank Prof. Yun Soo Bae (Ewha Womans University) for sharing APX-115 reagent. This study was supported by grants from the National Research Foundation of Korea (2018R1A2B3006323 and 2017M3A9B6073098). S.M. was also supported by grants from the National Research Foundation of Korea (2019R1I1A1A01057557) and the RP-Grant (2019-2020) of Ewha Womans University.

J.L. and S.L. performed experiments. J.L., S.L., S.M., and S.W.K. designed experiments and analyzed the data. S.W.K. conceived the study and interpreted data. J.L., S.M., and S.W.K. wrote the manuscript.

The authors have no potential conflicts of interest to disclose.