Background
Programmed cell death (PCD) is an important cellular mechanism whose dysregulation is involved in many human pathologies, especially tumor formation. Induction of PCD through the activation of caspases (apoptosis) is the best characterized route to death in most cell types [
1]. Independent from apoptosis, programmed necrosis represents an alternative form of PCD that operates without detectable caspase activity [
2,
3]. Yet incompletely understood, the mechanisms of programmed necrosis need to be intensively investigated, because a better knowledge of these pathways may directly translate into improved therapies for cancers resistant to apoptosis [
4,
5]. Our own group has previously identified the sphingolipid ceramide as one of the pivotal mediators in death receptor-mediated programmed necrosis [
3,
6,
7]. Although the induction of (caspase-dependent) apoptosis via manipulation of intracellular ceramide levels is increasingly recognized as an option in tumor therapy [
8], detailed information on the induction of (caspase-independent) programmed necrosis by ceramide in clinically relevant tumor cell systems is currently unavailable. Programmed necrosis induced by tumor necrosis factor (TNF)-receptor 1 (TNF-R1) currently represents the most comprehensively studied system. Yet, the potential usefulness of TNF in clinical oncology is severely limited by its strong systemic toxic side effects. As an alternative, TNF-related apoptosis inducing ligand (TRAIL) can selectively induce apoptosis in tumor cells while leaving non-transformed cells mostly unaffected [
9]. However, many tumor cells are intrinsically resistant against TRAIL-induced apoptosis and, even when combined with chemo- or radiotherapy, a resounding breakthrough in the therapy of cancer patients has not yet been achieved. As a potential alternative, we and others have previously demonstrated the ability of human and murine TRAIL receptors to induce programmed necrosis independently from their apoptotic capabilities when induction of apoptosis fails or is actively inhibited [
7,
10]. In consequence, the induction of programmed necrosis by TRAIL may represent a novel and additional, but still largely unexplored option for the elimination of tumor cells, in addition to the well-established strategies aimed at the induction of apoptosis.
In this study, we have therefore investigated the effects of TRAIL-induced programmed necrosis on a panel of 14 distinct human cancer cell lines of diverse origin (i.e. leukemia (U-937, CCRF-CEM), gall bladder adenocarcinoma (Mz-ChA-1), pancreatic adenocarcinoma (BxPC-3, Colo357, Panc89, PancTu-I, A818-4, Pt45P1), colorectal adenocarcinoma (HT-29), gastric adenocarcinoma (MKN-28), ovary adenocarcinoma (SK-OV-3), non-small cell lung carcinoma (KNS-62) and malignant melanoma (SK-Mel-28)). We show that TRAIL-induced programmed necrosis causes death of a wide range of these cell lines, impairs their clonogenic survival and acts in synergy with chemotherapeutic agents. Our findings also suggest that susceptibility/resistance of tumor cells to programmed necrosis is primarily determined by expression of the kinase RIPK3 (which indicates its potential usefulness as a predictive marker) and that ceramide represents a pivotal factor downstream of RIPK3 in the execution of programmed necrosis not only in the previously studied common laboratory cell lines, but also in the clinically more relevant tumor cell systems employed here.
Methods
Reagents
The Smac mimetic birinapant was provided by ChemieTek, Indianapolis, IN, USA. Necrostatin-1 and necrosulfonamide were obtained from Calbiochem, Darmstadt, Germany. Arc39 has been previously described [
11,
12]. Cisplatin, etoposide, trichostatin A, 5-fluorouracil, irinotecan, doxorubicin, camptothecin and paclitaxel were ordered from Sigma-Aldrich, Munich, Germany.
Cell lines and culture conditions
Mz-ChA-1, Colo357, PancTu-I, Panc89, A818-4, Pt45P1, MKN-28 and KNS-62 cells have been described [
13‐
16]. U-937, BxPC-3, HT-29, CCRF-CEM, SK-OV-3 and SK-MEL-28 cells were originally obtained from the American Type Culture collection. The identity of all cell lines was validated by STR profiling. The cell lines were cultured in RPMI 1640 (Life Technologies, Darmstadt, Germany) supplemented with 10% v/v FCS and 1 mM sodium pyruvate or (U-937 cells) 10% v/v FCS and 50 μg/ml penicillin/streptomycin. Wildtype and RIP3-deficient mouse embryonic fibroblasts (MEF) have been described [
17] and were cultured in DMEM (Life Technologies) supplemented with 10% v/v FCS and 50 μg/ml penicillin/streptomycin. Programmed necrosis was induced by addition of human recombinant TRAIL (Super
KillerTRAIL™, Enzo, Lausen, Germany) or highly purified human recombinant TNF (BASF Bioresearch, Ludwigshafen, Germany), in combination with benzyloxycarbonyl-Val-Ala-Asp(OMe)-fluoromethylketone (zVAD-fmk; Bachem, Heidelberg, Germany) and cycloheximide (CHX; Sigma-Aldrich). In experiments with necrostatin-1, cells were preincubated with 50 μM necrostatin-1 for 2 h before addition of TRAIL/zVAD/CHX or TNF/zVAD/CHX.
Cytotoxicity assays, viability assays
For flow cytometric analysis of cell death (i.e. loss of membrane integrity), cells were seeded onto 12-well plates at 70% confluence. After treatment, adherent and detached cells were collected, followed by one washing step in PBS/5 mM EDTA. The cells were resuspended in PBS/5 mM EDTA containing 2 μg/ml propidium iodide (PI), and analyzed in a FACSCalibur flow cytometer (BD Biosciences, San Diego, CA, USA) at red fluorescence. Alternatively (when measuring ceramide levels), loss of membrane integrity was determined by trypan blue staining. For this, cells were collected and resuspended in PBS. An aliquot of the cell suspension was added to the same volume of 0.4% v/v trypan blue staining solution (Life Technologies) and applied onto a Neubauer counting chamber. Live cells with an intact cell membrane did not absorb trypan blue and were scored separately from dead (blue) cells. For determination of cell viability by crystal violet staining, cells were seeded in flat-bottom 96-well plates. After stimulation, adherent cells were washed twice with PBS and incubated for 10 min at 37°C in 50 μl of staining solution (0.5% w/v crystal violet, 4% w/v formaldehyde, 30% v/v ethanol, and 0.17% w/v NaCl). The staining solution was washed away with tap water and cells were dried for 1 h at 50°C. Stained cell were dissolved in 33% v/v acetic acid and the absorbance of the staining was measured at 570 nm in a microplate reader (Tecan, Crailsheim, Germany). Suspension cells were alternatively analyzed by metabolic activity measurements with the XTT cell proliferation kit II (Roche, Mannheim, Germany). The intracellular ATP content of cells was determined with the Cell Titer Glo Assay Kit (Promega, Mannheim, Germany) following the instructions of the manufacturer.
Flow cytometric detection of TRAIL receptors 1 and 2 (TRAIL-R1 and TRAIL-R2)
For detection of cell-surface expression of TRAIL receptors, a total of 1.5 × 105 detached cells were incubated with anti-TRAIL-R1 or anti-TRAIL-R2 mouse monoclonal antibodies (Alexis) in PBS/1% w/v BSA for 1 h at 4°C, washed twice in PBS/1% w/v BSA and incubated with anti-mouse biotin-conjugated secondary antibodies for additional 1 h at 4°C. After two washing steps, cells were incubated with phycoerythrin-conjugated streptavidin for further 15 min at 4°C, washed twice and re-suspended in 150 μl PBS/1% w/v paraformaldehyde and analyzed in a FACSCalibur flow cytometer. Controls were incubated with appropriate isotype matched antibodies and labeled with the corresponding secondary antibodies.
Western blot
Whole cell lysates were prepared with TNE lysis buffer (50 mM Tris pH 8.0, 150 mM NaCl, 1% v/v NP-40, 3 mM EDTA, supplemented with Complete protease inhibitor mixture (Roche)). The protein lysates were separated by SDS-PAGE, transferred onto nitrocellulose membranes and reactive proteins were detected with antibodies for RIPK1 (BD Biosciences), RIPK3 (Abnova, Heidelberg, Germany), MLKL or actin (Sigma-Aldrich) via chemiluminescence (Lumiglo, Cell Signaling, Danvers, MA, USA). RIPK3 expression was quantified using the program ImageJ (Wayne Rasband, National Institutes of Health, Bethesda, MD, USA). To compare expression levels of RIPK1 and RIPK3 in tumor cell lines, identical amounts of protein (20 μg) were loaded, using lysates from PancTu-I cells as a standard on each gel, and identical exposure times were taken to allow a direct comparison of expression levels. For the quantitative analysis of the relationship between the levels of RIPK3 and the specific sensitivity of the respective tumor cell line to TRAIL/zVAD/CHX-induced programmed necrosis, values for RIPK3 expression were normalized between the gels and calculated relative to CCRF-CEM cells. In all Western blots, detection of actin served as a loading control.
RNA interference
The predesigned siRNA specific for human RIPK3 (ID # s21741), human MLKL (ID # s47087) as well as the negative control siRNA (ID # AM4611) were obtained from Life Technologies. A second, distinct siRNA specific for human RIPK3 (siGENOME human RIPK3, D-003534-01) was obtained from Thermo Scientific, Schwerte, Germany. U-937 cells were transfected with 150 pmol siRNA by Amaxa nucleofection (Lonza, Cologne, Germany), using solution V and program X-001. HT-29 cells were transfected with 20 pmol siRNA and siPortAmine transfection reagent (Life Technologies).
Ceramide quantification
Lipids were extracted according to the method of Bligh and Dyer [
18] and separated by high-performance thin layer chromatography (TLC) as described [
3]. After charring, thin layer chromatography plates were scanned and analyzed using the Molecular Dynamic Personal Densitometer SI Scanner control software (GE Healthcare, Munich, Germany).
Clonogenic survival assays
Assays for clonogenic survival of cells were essentially carried out as described by Franken and coworkers [
19]. Briefly, following treatment, 1,000 viable cells (as determined by trypan blue staining) were plated into six-well plates in complete medium without zVAD-fmk, CHX, TRAIL or TNF, cultured for 7 days at 37°C and stained with crystal violet as described above under “viability assays”, except that all steps subsequent to washing with tap water were omitted. Non-adherent U-937 cells were alternatively analyzed for their ability to form colonies in soft agarose by overlaying them onto 2 ml of 0.4% w/v Sea Plaque agarose (Cambrex, East Rutherford, NJ, USA) on top of 3 ml of a 1% w/v peqGOLD agarose underlayer (PeqLab, Erlangen, Germany), both in complete medium. After incubation for 7 days at 37°C, U-937 cells were stained with 1 ml of 3-[4,5-dimethylthiazol-2yl]-2,5-diphenylterazolium bromide (MTT, Sigma, 2.5 mg/ml in PBS) for 2 h at 37°C to allow metabolization of MTT to blue MTT-formazan. Colony formation (>10 cells) was determined from pictures taken with a Lumix DMC-FS10 digital camera (Panasonic, Wiesbaden, Germany).
Statistical analysis
For all figures, representative data from one out of at least two or more experiments with similar results are shown (n ≥ 2) and error bars indicate the standard deviations (SD) from at least triplicate determinations (n ≥ 3). P values were calculated using Student’s t-test. Statistical significance is denoted by *P < 0.05, **P < 0.01, ***P < 0.001.
Discussion and conclusion
In this study, we have investigated whether induction of TRAIL/zVAD/CHX-induced programmed necrosis represents a viable strategy for the elimination of tumor cells. Necrosis has long been regarded as an accidental, non-physiologic form of cell death, whereas caspase-dependent apoptosis was considered to be the only form of programmed and thus physiologically occurring cell death. This view has however been challenged by numerous studies which have provided evidence for the existence of programmed forms of necrosis that do not depend on caspases but nevertheless follow defined molecular steps [
20]. While caspase-dependent apoptosis is the major pathway leading to PCD, programmed necrosis can act as a backup system when the apoptotic machinery fails or is inactivated (e.g. by mutations in apoptosis-resistant cancer cells) [
5,
28]. It has been shown that programmed necrosis exerts critical functions in multiple patho-physiological settings, e.g. the regulation of bone growth, ovulation, negative selection of lymphocytes [
28], pancreatitis [
22,
29], epilepsy, ischemia–reperfusion injury, Parkinson’s, Huntington’s and Alzheimer's disease, and cell destruction by Salmonella, Shigella, HIV and vaccinia virus [
4,
28,
30,
31]. In contrast to apoptosis, a comprehensive picture of the signaling pathways of programmed necrosis is not yet available. In the most extensively studied model, TNF-R1 elicits programmed necrosis via activation of RIPK1 and RIPK3, a step which is stimulated by the deubiquitinase CYLD and the deacetylase SIRT2, but negatively regulated by the proteins FADD, FLIP, caspase-8 and members of the cIAP family. Downstream of RIPK3, the proteins MLKL and PGAM5 contribute to programmed necrosis by promoting mitochondrial fragmentation [
20]. We have previously demonstrated that ceramide acts as an additional key molecule in death receptor-mediated programmed necrosis [
3,
6,
7]. Furthermore, enzymes of the energy metabolism, the Bcl-2-family member Bmf and production of reactive oxygen species have been implicated as additional factors in programmed necrosis [
4].
The capacity to elicit programmed necrosis appears to be an intrinsic feature of death receptors and has been reported not only for TNF-R1 [
2,
3,
6], but also for Fas/CD95 [
4] and ectodermal dysplasia receptor [
32]. Independently, we and others have demonstrated the ability to trigger programmed necrosis for human and murine TRAIL receptors [
7,
10]. In contrast to programmed necrosis, the efficacy of TRAIL in the apoptotic elimination of tumor cells has been extensively demonstrated in clinical trials employing mono- or combination therapies [
9]. Consistent with the finding that TRAIL elicits apoptosis selectively in tumor but not primary cells, TRAIL was well tolerated in preclinical models at serum concentrations that were shown to be effective against cancer cells, as were agonistic TRAIL receptor antibodies applied to patients in clinical trials using mono- or combination therapies [
9]. Nevertheless, intrinsic resistance against TRAIL-induced apoptosis, even when combined with chemo- or radiotherapy, limit the therapeutic success and necessitate the search for additional, yet unexplored options for the treatment of patients.
As such a potential option, the induction of programmed necrosis by TRAIL has however been investigated only in a very limited number of studies. Our own study presented here provides strong evidence for the suitability of TRAIL/zVAD/CHX-induced programmed necrosis as a tool to eliminate tumor cells from a wide range of sources. Raising the expectation that TRAIL/zVAD/CHX-induced programmed necrosis may be even more effective under conditions that more closely resemble the
in vivo situation than mere cell culture, it clearly interfered with the capacity of all tested tumor cell lines for unlimited proliferation in clonogenic survival assays (even in a tumor cell line that had shown resistance in conventional cytotoxicity/viability assays). Furthermore, our data demonstrate that cisplatin, etoposide, trichostatin A, 5-fluorouracil, irinotecan, doxorubicin, camptothecin and paclitaxel can exert cytotoxicity not only via apoptosis, but also via programmed necrosis. Providing additional encouragement for the development of future combination therapies, TRAIL/zVAD/CHX-induced programmed necrosis synergized with chemotherapeutic agents and enhanced the cytotoxic response in eight out of 10 tested tumor cell lines as well as 41 out of 80 chemotherapeutic/TRAIL/zVAD/CHX combinations. With regard to potential predictive markers, our results identify expression of RIPK3 as a primary determinant of susceptibility or resistance of tumor cells to TRAIL/zVAD/CHX-induced programmed necrosis. However, our data also show that in future screenings, it should be kept in mind that secondary factors may additionally confer resistance downstream or independent from RIPK3. Finally, our study has confirmed and extended the role of ceramide as one of the key mediators of programmed necrosis downstream of RIPK1 and RIPK3 to the clinically more relevant tumor cell systems investigated here, with the A-SMase inhibitor Arc39 additionally validating A-SMase (rather than neutral sphingomyelinase or ceramide synthase) as the main enzyme responsible for ceramide generation. Our findings are not only fully consistent with our previous data from the initially studied laboratory cell lines [
3,
6,
7], but may also prove valuable for a future manipulation of intracellular ceramide levels to induce programmed necrosis in tumor therapy.
As pointed out above, only very few other studies have focused on the induction of programmed necrosis by TRAIL. One of those studies has recently reported that TRAIL induces necroptosis (i.e. a subset of programmed necrosis depending on RIPK1/RIPK3 [
20]) in the tumor cell lines HT-29 (which was also used in this study) and Hep G2 [
33], at first glance consistent with our results. However, unlike in our study, necroptosis was only observed under acidified (but not physiologic) conditions. Moreover, the same group had previously reported that in this very system, caspase activity is required for cell death [
34], being inconsistent with the molecular mechanisms described for necroptosis [
20] and thus suggesting a certain caution when interpreting the results of this study. More encouraging, Hunter and coworkers have reported that TRAIL can kill small cell lung cancer cells by inducing caspase-independent mechanisms of cell death in synergy with paclitaxel [
35]. Independently, platinum compounds in combination with TRAIL were found to be effective against breast cancer cells by inducing programmed necrosis (although to a lesser extent) in addition to apoptosis [
36]. Finally, Katz and colleagues have described that malignant pleural mesothelioma cells are killed by caspase-independent mechanisms after application of TRAIL in combination with sorafenib, and even find promising evidence of therapeutic efficacy in a xenograft mouse model [
37], in summary corroborating our data on the synergistic action of TRAIL/zVAD/CHX and chemotherapy in the programmed necrosis of tumor cell lines.
With regard to a future therapeutic application of TRAIL/zVAD/CHX-induced programmed necrosis, further work is required. At present, it is unknown whether RIPK3-proficient tumor cells can be stimulated to undergo programmed necrosis and thus circumvent apoptosis resistance in patients. For this purpose, strategies for the induction of programmed necrosis (e.g. as used in our study) need to be adapted to the
in vivo situation. As an example, the sensitizer CHX used here is cytotoxic also to healthy tissue. Therefore, therapies based on the treatment of patients with CHX are not an option. As a possible alternative (and in line with our own data presented in Figure
1d), Smac mimetics can similarly sensitize tumor cells for TRAIL- and TNF-induced programmed necrosis in cell culture models [
22]. Yet, their efficacy or toxicity under conditions of programmed necrosis has not been evaluated
in vivo so far.
Since this study focuses on TRAIL-induced programmed necrosis as a novel approach to eliminate tumor cells, we explicitly want to point out that TRAIL-induced programmed necrosis in principle occurs under the condition that the normal apoptotic pathway is inhibited. It has recently become clear that caspase-8 suppresses programmed necrosis under normal conditions and that it needs to be actively inhibited (e.g. by zVAD-fmk) for programmed necrosis to be executed. Notably, even the basal activity of non-stimulated caspase-8 is already sufficient for the suppression of programmed necrosis [
20]. Therefore, the induction of programmed necrosis in apoptosis-resistant cell lines in the absence of caspase inhibitors would only be effective in tumors that carry a mutation that directly inactivates caspase-8. In all other cases (i.e. in cells that harbor apoptosis-inhibiting mutations affecting other proteins) the residual activity of caspase-8 would still be sufficient to suppress programmed necrosis. Most likely, this is the reason why the application of TRAIL alone has so far not been effective against apoptosis-resistant tumors in clinical trials. Therefore, we consider the inhibition of caspase-8 as an essential prerequisite for the successful elimination of tumor cells by TRAIL-induced programmed necrosis. In future treatment regimens this could be most conveniently achieved by combining TRAIL with a caspase inhibitor such as zVAD-fmk. With regard to its physiological and clinical relevance, zVAD-fmk has so far proven to be a non-toxic substance that has no adverse effects and which is well tolerated when administered for prolonged periods of time [
3,
38,
39]. However, although TRAIL and zVAD-fmk by themselves have not shown toxicity
in vivo[
9,
40], it must be clarified whether their joint application (and additionally in combination with chemotherapeutic agents) is equally non-toxic
in vivo.
As another topic to be addressed with regard to future therapies, cells dying by programmed necrosis can release a broad range of damage-associated molecular patterns (DAMPs) which in turn can trigger inflammatory responses. Accordingly, programmed necrosis has been associated with inflammation in several
in vivo models [
20]. Therefore, it will be of high interest to clarify whether the death of tumor cells via programmed necrosis is immunogenic and may thus elicit a highly desirable anticancer immune response that would eliminate residual tumor (stem) cells [
4]. Such a beneficial inflammation elicited by tumor cells undergoing TRAIL-induced programmed necrosis could thus contribute to an even more effective treatment for cancer patients in the future.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SV, SP, SS, HK and DA designed research; SV, SP, PD and SWM performed research; SV, SP, SWM, CR, CA, AT, DK, SS, HK and DA analyzed data, SV and DA wrote the paper. All authors read and approved the final manuscript.