All assessable cervical cancer patients (n = 250) who had undergone surgery for treatment of cervical cancer at the Department of Gynecology and Obstetrics, Ludwig-Maximilians-University Munich, Germany, from 1993 until 2002 and whose paraffin-embedded tumor tissue was available were included in this study. Patients were not pre-selected. An experienced gynecological pathologist assigned histopathological tumor subtypes (squamous cell carcinoma, adenocarcinoma, adenosquamous carcinoma) and grading (G1: well differentiated, G2: moderately differentiated, G3: poorly differentiated). TNM classification (T = primary tumor site, N = regional lymph node involvement, M = presence of distant metastatic spread) was carried out as published by the Union for International Cancer Control (UICC). FIGO classification was evaluated according to the criteria of the Fédération Internationale de Gynécologie et d’Obstétrique (FIGO). Clinical and patient follow-up data was received from the Munich Cancer Registry ( Table 1 ). The mean age of patients at diagnosis was 49.0 ± 13.0 years with a range between 20.4 and 83.3 years. Patients who died independently of their tumor or whose cause of death is unknown were censured. Other histopathological parameters were examined and published previously, enabling correlation analysis to the parameters investigated in this study.

Directly after surgical resection, the samples were fixed in neutral-buffered formalin (3.7%) followed by standardized paraffin embedding. Tissue micro arrays (TMAs) were then prepared from formalin fixed paraffin embedded (FFPE) cervical cancer samples. The tissue sections (3µm) were dewaxed in Roticlear (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) for 20 min followed by immersion in 3% H₂O₂ in methanol at room temperature (RT) for 20 min to inactivate endogenous peroxidase. A descending ethanol gradient (S99%, 70% and 50%) was used to rehydrate the specimens. The sections were prepared for epitope retrieval in a pressure cooker for 5 min using sodium citrate buffer (pH 6.0; 0.1 mol/L citric acid/0.1 mol/L sodium citrate) followed by washing in distilled water and phosphate buffer saline (PBS) (RIPK1 and RIPK3 staining) or Tris buffered saline (TBS) (pMLKL staining). To block non-specific binding of primary antibodies, Blocking Solution (reagent 1, ZytoChem Plus HRP Polymer System (Mouse/Rabbit); Zytomed Systems GmbH, Berlin, Germany) was applied for 5min at RT for RIPK1 and RIPK3 staining. For pMLKL staining, non-specific binding of primary antibodies was blocked by applying blocking serum (= two drops of normal goat serum diluted in 10ml sterile TBS; VECTASTAIN Elite ABC-HRP Kit, Peroxidase (Rabbit IgG) – PK6101; Vector Laboratories, Inc., Burlingame, CA, USA) for 20 min at RT. Specimens were then separately incubated with the following primary antibodies: anti-RIPK1 (polyclonal rabbit IgG, HPA015257; Sigma Aldrich, St. Louis, MO, USA) for 17h at 4°C followed by 30min at RT in a 1:50 dilution in PBS, anti-RIPK3 (polyclonal rabbit IgG;, HPA055087 Sigma Aldrich, St. Louis, MO, USA) for 16h at 4°C in a 1:1500 dilution in PBS and anti-MLKL (phospho S358) (monoclonal Rabbit IgG, ab187091; Abcam, Cambridge, UK) for 16h at 4°C in a 1:100 dilution in TBS. Then, the tissue slides for RIPK1 and RIPK3 staining were incubated with PostBlock Reagent (reagent 2) for 20 min and HRP-Polymer (Mouse/Rabbit) (reagent 3) for 30 min which contain secondary antibodies (anti-mouse/-rabbit) and peroxidase according to the manufacturer’s protocol (reagent 2, reagent 3, ZytoChem Plus HRP Polymer System (Mouse/Rabbit); Zytomed Systems GmbH, Berlin, Germany). The tissue slides for pMLKL staining were incubated with the secondary antibody (= two drops of normal goat serum and one drop of biotinylated goat anti-rabbit IgG diluted in 10ml sterile TBS; VECTASTAIN Elite ABC-HRP Kit, Peroxidase (Rabbit IgG) – PK6101) for 30 min followed by incubation with ABC complex (= four drops of Reagent A and four drops of Reagent B diluted in 10ml sterile TBS; VECTASTAIN Elite ABC-HRP Kit, Peroxidase (Rabbit IgG) – PK6101) for 30 min. All slides were washed with PBS (RIPK1, RIPK3) or TBS (pMLKL) after every incubation step. Next, 3,3′-diaminobenzidine chromogen (DAB; Dako, Glostrup, Denmark) and the according substrate buffer (Liquid DAB and Substrate Chromogen System, DAKO, Munich, Germany) were added to the slides. The slides were washed in distilled water to stop the reaction followed by counterstaining in Mayer’s acidic hematoxylin. The slides were then immersed in an ascending ethanol gradient and Roticlear and finally cover slipped using ROTI^® Mount (Carl Roth GmbH + Co. KG, Karlsruhe, Germany). Appropriate tissue was used as positive control (placenta for RIPK1, RIPK3 and pMLKL) and negative controls (placenta for RIPK1 and pMLKL and colon for RIPK3) ( Figure S2 ). For the negative controls the primary antibodies were each replaced with a specific isotype control antibody (BioGenex, Fremont, CA, USA).

The immunohistochemically stained cervical cancer specimens were examined using a Leitz Diaplan photomicroscope (Leitz, Wetzlar, Germany). Quantification of RIPK1, RIPK3, and pMLKL expression was performed by applying the semiquantitative immunoreactive score (IRS) which evaluates the intensity and distribution pattern of antigen expression (37). The IRS is calculated by multiplying the number of positively stained cells (in %) (0: no staining; 1: 1% - 10% stained tumor cells; 2: 11% - 50% stained tumor cells; 3: 51% - 80% stained tumor cells; 4: > 80% stained tumor cells) with the predominant staining intensity (0: none; 1: weak; 2: moderate; 3: strong). The scale goes from 0 (no expression) to 12 (very high expression). Images were taken with Flexacam C1 (Leica Microsystems (Switzerland) Ltd., Heerbrugg, Switzerland).

The human cervical cancer cell lines CaSki (epidermoid carcinoma), HeLa (adenocarcinoma) and SiHa (squamous cell carcinoma) obtained from ATCC (Rockville, MD, USA) were used for in vitro studies. They were maintained in culture using RPMI Medium 1640 (1X) + GlutaMAX (ThermoFisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum at standard conditions (humidified incubator at 37°C and 5% CO₂ saturation).

Cell viability was analyzed using Thiazolyl Blue Tetrazolium Bromide (MTT; Sigma Aldrich, St. Louis, MO, USA; M5655). After 24h, 48h, and 72h of stimulation incubation, 20µl of Thiazolyl Blue Tetrazolium Bromide (5mg MTT/ml sterile PBS) was added to each well followed by soft shaking on a plate shaker for 5 min and incubation for 1.5h at 37°C in a 5% CO₂ humidified atmosphere. Then, the supernatant was removed and 200µl dimethyl sulfoxide (DMSO)/well was added to solubilize the blue crystals. After 5 min of soft shaking on a plate shaker, optical density was evaluated at wavelengths of 595nm by the Elx800 universal microplate reader (BioTek Instruments GmbH, Bad Friedrichshall, Germany) and Gen5 software (BioTek Instruments GmbH, Bad Friedrichshall, Germany). All tests were repeated three times with 5 wells per concentration.

For analysis of cell proliferation, the colorimetric bromodeoxyuridine (BrdU) assay was performed using the Cell Proliferation ELISA, BrdU (colorimetric) kit from Roche (F. Hoffmann-La Roche AG, Basel, Switzerland). The assay was performed according to the manufacturer’s protocol. CaSki, HeLa and SiHa cells were incubated with C2 ceramide for 24h, 48h and 72h in 96-well plates as described below. Then, BrdU was added, and the cells were incubated for another 24h. The medium was removed and 200µl FixDenat per well was added to fix the cells and denature the DNA for improvement of accessibility of the incorporated BrdU for detection by the antibody. Next, anti-BrdU-POD was added and bound to the BrdU incorporated in newly synthesized DNA followed by incubation for 90 min at RT. The supernatant was removed, and the wells were washed three times with a washing solution. 100µl substrate solution (tetramethyl-benzidine, TMB) per well was added and incubated in the dark at RT to detect the immune complexes. Visualization of the substrate reaction was an increasing deep blue coloration. After 15 min of incubation, the reaction was stopped by adding sulfuric acid. The plate was then measured at a wavelength of 450nm using the Elx800 universal microplate reader using Gen5 software. All tests were repeated three times and with 5 wells per concentration.

For MTT- and BrdU assay, 5x10³ CaSki, HeLa and SiHa cells were incubated in 100µl RPMI Medium 1640 (1X) + GlutaMAX (ThermoFisher Scientific, Waltham, MA, USA) supplemented with 10% FBS in 96-well plates (flat bottom) for four hours. Then, medium with 10% FBS was removed and 100µl medium without FBS was added followed by 24h of incubation.

Stimulation with 50µM, 100µM or 200µM C2 ceramide (Enzo Life Sciences (ELS) AG, Lausen, Switzerland; BML-SL100, Prod.Nr.: ALX-306-024) was performed by dissolving the substance to 200mM (68.3mg/ml) in DMSO followed by dilution of the ceramide/DMSO stock 200-fold into a BSA solution which was prepared using fatty acid-free BSA diluted in water to 66mg/ml (1mM). At first, a precipitate was formed but dissolved after 30-60 min of stirring at RT. Then, the ceramide/DMSO/BSA solution was diluted in 100µl medium without FBS at respective concentrations (100µM, 200µM or 400µM) and added to the wells. The 1:1 dilution in the well gave the final concentration of 50µM, 100µM or 200µM.

Necrostatin-1 (nec-1, Santa Cruz Biotechnology, Inc., Heidelberg, Germany; CAS 4311-88-0) and Z-VAD-fmk (Selleck Chemicals, Munich, Germany; Catalog No. S7023, CAS 187389-52-2) were both dissolved to 50mM and further to 5mM in DMSO. Co-stimulation of C2 ceramide and nec-1 or C2 ceramide and Z-VAD-fmk was initiated by pre-stimulation with 1µM, 5µM, 20µM or 50µM nec-1 or 1µM, 5µM, 20µM or 50µM Z-VAD-fmk for 2h by diluting the substances in 100µl medium without FBS at respective concentrations (2µM, 10µM, 40µM or 100µM) and adding it to the wells. The 1:1 dilution in the well gave the final concentration of 1µM, 5µM, 20µM or 50µM. Then, the supernatant was removed and co-stimulation of 100µM C2 ceramide and 1µM, 5µM, 20µM or 50µM nec-1 or 100µM C2 ceramide and 1µM, 5µM, 20µM or 50µM Z-VAD-fmk was performed by diluting the substances in 200µl medium without FBS at respective concentrations (100µM C2 ceramide and 1µM, 5µM, 20µM or 50µM nec-1/Z-VAD-fmk) and adding it to the wells. The concentrations of the substances were based on previous studies (36, 38, 39). Unstimulated cells and wells without any cells served as negative controls. A solvent control (DMSO) was always conducted.

For visualization of C2 ceramide’s effect on cervical carcinoma cells, live-cell imaging was performed. Therefore, 5 x10⁴ CaSki, HeLa, and SiHa cells were seeded in a glass-bottomed culture dish (µ-Dish, Ø 35 mm; ibidi, Gräfelfing, Germany) and incubated under standard conditions at 37°C and 5% CO₂. The cells were then serum starved for 6h before being stimulated with C2 ceramide by dissolving C2 ceramide as described above. 100 µM of C2 ceramide diluted in medium without FBS was added to the cells, and cells were imaged for 72 h using an Axiovert 135 microscope (Carl Zeiss). To produce a time-lapse series (Micro-Manager 1.3 Microscopy Software Ron Vale’s laboratory at UCSF, USA), images (ProgRes MF, Jenoptik, Jena, Germany) were taken every 20 minutes. iMovie 9.0.3 (Apple Inc., Cupertino, CA, USA) was used to create movie sequences.

Cells were cultured under basal conditions and harvested. The method was described before (36). In brief, cellular protein was isolated using RIPA buffer containing protease and phosphatase inhibitors (PI, Thermo Fisher Scientific, Waltham, USA) and 16 µg were loaded onto a SDS-PAGE gel. Primary antibody (anti-RIPK3, polyclonal rabbit IgG, HPA055087, Lot: R73206, Sigma Aldrich, St. Louis, MO, USA) and secondary antibody (IRDye^® 680RD Donkey anti-Rabbit IgG, Secondary Antibody) were used.

For quantitative detection of cell death, Annexin V-FITC (ALX-209-256-T100, Enzo Life Sciences, Farmingdale, NY, USA) and SYTOX™ Red Dead Cell Stain (S34859, Invitrogen, Carlsbad, CA, USA) based flow cytometry was performed. Briefly, 2 x 10⁵ CaSki, HeLa, or SiHa cells were seeded and after a 6h starvation step on medium without FBS the next day cells were treated with 100µM C2 ceramide or the solvent control for 72h, as described above. Cells were trypsinized, washed in PBS and resuspended in a buffer consisting of 10mM HEPES, 140mM NaCl and 2.5mM CaCl₂ at pH 7.4, followed by incubation with Annexin V-FITC for 10 minutes according to the manufacturer’s instructions, and addition of SYTOX Red Dead Cell Stain (1:1000). Labelled cells were sorted using the BD FACSCanto™ II (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and signals (Annexin V-FITC: 488nm, 530/30 bandpass filter; SYTOX™ Red Dead Cell Stain: 633nm excitation, 660/20 bandpass filter) were analyzed with the BD FACSDiva Software (version 8.0.1, Becton, Dickinson and Company). This was repeated two times (n = 2).

For statistical analysis, IBM SPSS Statistics Version 26.0.0.0 (IBM, Armonk, New York, NY, USA) was used. p-values of p < 0.05 were considered statistically significant. Kruskal-Wallis-H test was applied as appropriate for group comparisons of independent groups regarding clinical and pathological subgroups. Bivariate correlations between staining results in this study and other variables and clinicopathological data were determined using Spearman’s rank correlation coefficient. Differences in OS and PFS times of cervical cancer patients were tested for significance by log-rank (Mantel-Cox) test and visualized using Kaplan-Meier curves. Cox regression analysis was used to ascertain the independence of the investigated prognosticators. Variables included in the Cox regression model were patient’s age, histological subtype, tumor grading, FIGO, nodal status, RIPK1, RIPK3, co-expression of RIPK1 and RIPK3, and co-expression of RIPK1, RIPK3, and pMLKL. RIPK1, RIPK3, and pMLKL expression were divided into low and high expression for survival analysis. All in vitro analyses were statistically analyzed using Wilcoxon test and visualized with GraphPad Prism 7.00 (San Diego, CA, USA). For statistical analysis of AnnexinV-FITC apoptosis staining followed by FACS analysis the percentual number of apoptotic cells after treatment was compared to the percentual number of apoptotic cells in the control and the percentual number of necrotic and intermediate (dying) cells after treatment was compared to the percentual number of necrotic and intermediate (dying) cells in the control using Wilcoxon test.

The study was approved by the local ethics committee of Ludwig-Maximilians University of Munich, Germany (approval number 259-16, 2016). All tissue samples examined in this study were obtained from material from the archives of the Department of Gynecology and Obstetrics, Ludwig-Maximilians University of Munich. The material used in this study was fully anonymized, declared as left over after finalizing all diagnostic procedures and was used more than 10 years after surgery. All experiments involving human participants were performed according to the standards of the Declaration of Helsinki from 1964 and its later amendments and were in accordance with the ethical standards of the institutional and/or national research committee. During analyses, the observers were fully blinded for patients’ data.

Immunohistochemical staining showed that RIPK1 and pMLKL were expressed in the nucleus and cytoplasm while RIPK3 was expressed in the nucleus in cervical cancer tissue. Differences in expression of RIPK1, RIPK3, and pMLKL were examined by comparing the immunoreactive score (IRS) in the groups of histological subtypes (squamous cell carcinoma, adenocarcinoma, adenosquamous carcinoma), grading (G1-G3), TNM- and FIGO-classification.

Live-cell imaging was performed over 72h after stimulation with C2 ceramide to investigate morphological changes. Incubation of CaSki and HeLa cells with C2 ceramide showed a visible reduction of confluency, and ballooning of the cells was observed. Treatment of SiHa cells with C2 ceramide did not have any visible effect on the morphology of cervical cancer cells.

To further specify the type of cell death induced by C2 ceramide, AnnexinV-FITC apoptosis staining followed by FACS analysis was performed after 72h stimulation with C2 ceramide (100μM). A significant percentual increase in apoptotic as well as in necrotic/intermediate (dying) cells compared to the respective control was observed in CaSki cells (Figures 6A, D). In SiHa cells, a significant increase in apoptotic but not in necrotic/intermediate (dying) cells was observed (Figures 6B, E). Treatment of HeLa cells with C2 ceramide did not show a significant percentual increase in apoptotic or necrotic/intermediate (dying) cells (Figures 6C, F).

The results of Western Blot showed that RIPK3 of expected size (~57kDa) was detected in all three cell lines examined ( Figure S8 ).