Cytochrome C oxidase Inhibition and Cold Plasma-derived Oxidants Synergize in Melanoma Cell Death Induction

Despite striking advances in the treatment of metastasized melanoma, the disease is often still fatal. Attention is therefore paid towards combinational regimens. Oxidants endogenously produced in mitochondria are currently targeted in pre-clinical and clinical studies. Cytotoxic synergism of mitochondrial cytochrome c oxidase (CcO) inhibition in conjunction with addition of exogenous oxidants in 2D and 3D melanoma cell culture models were examined. Murine (B16) and human SK-MEL-28 melanoma cells exposed to low-dose CcO inhibitors (potassium cyanide or sodium azide) or exogenous oxidants alone were non-toxic. However, we identified a potent cytotoxic synergism upon CcO inhibition and plasma-derived oxidants that led to rapid onset of caspase-independent melanoma cell death. This was mediated by mitochondrial dysfunction induced by superoxide elevation and ATP depletion. This observation was validated by siRNA-mediated knockdown of COX4I1 in SK-MEL-28 cells with cytotoxicity in the presence of exogenous oxidants. Similar effects were obtained with ADDA 5, a recently identified specific inhibitor of CcO activity showing low toxicity in vivo. Human keratinocytes were not affected by this combinational treatment, suggesting selective effects on melanoma cells. Hence, targeting mitochondrial CcO activity in conjunction with exogenous pro oxidant therapies may constitute a new and effective melanoma treatment modality.

phosphorylation (OXPHOS) 20,21 . This is further confirmed by the fact that ρ0 melanoma cells do not form xenografts 22 and that non-glycolytic metabolic sources like the Krebs cycle are more active in melanoma cells compared with normal melanocytes 23 . Therefore, targeting the mitochondrial OXPHOS pathway may represent an innovative approach to sensitize melanoma cells for a better efficacy of exogenous therapies like chemotherapy.
We investigated, whether combination of mitochondrial cytochrome C oxidase (CcO) inhibitor and addition of exogenous oxidants exhibits potentiated cytotoxic responses in melanoma cells. To generate oxidants, cold physical plasma was used. Cold plasma is a partially ionized gas expelling a wealth of reactive oxygen and nitrogen species 24 that are being deposited in cell culture media 25 . Such oxidant-enriched media selectively kills cancer cells in vitro by targeting the mitochondrial network 26,27 . We assessed synergistic effects of CcO inhibition and plasma-treated medium in murine B16F0 (non-metastatic), B16F10 (metastatic) 28,29 , and human SK-MEL-28 (BRAF + ) melanoma cells as well non-malignant human HaCaT keratinocytes. Our results demonstrate a pronounced additive effect of CcO inhibition and oxidants selectively in melanoma cell killing.

Materials and Methods
Cell culture. B16F0

Plasma-Treated Media (PTM).
Plasma-Treated Media (PTM) was generated using the atmospheric pressure argon plasma jet kINPen. The jet is accredited as a medical device for wound treatment in Germany and lacks genotoxic or mutagenic action [30][31][32] . It was operated at a frequency of 1 MHz with 3 l/min argon gas (99.9999%; Air Liquid) to treat 2 ml of RPMI-1640 media with 1% fetal calf serum (FCS) for 120 s. PTM was immediately used for experiments. The total concentration of H 2 O 2 in PTM was determined using amplex ultra red reagent (Thermo scientific) according to the recommended protocol. Argon gas-treated medium (with plasma off) served as control throughout all experiments.
Metabolic activity and cell viability. 1 × 10 4 cells were challenged with ADDA5, KCN or NaN 3 in the presence or absence of PTM for 3 and 24 hours. Subsequently, wells were loaded with 100 µM of resazurin (Alfa Aesar) that is transformed to fluorescent resorufin by metabolically active cells. The plate was incubated for 2 h at 37 °C. Fluorescence was measured in multimode plate reader (Tecan) at λ ex 535 nm and λ em 590 nm and normalized to untreated control. Four hours after plasma treatment, apoptosis was assessed by staining with cell event ™ caspase 3/7 (ThermoFisher) for 30 min at 37 °C. Subsequently, cells were detached using accutase (BioLegend), and accutase containing 4′,6-diamidino-2-phenylindole (DAPI; BioLegend) was added to label terminally dead cells. Cells were subjected to flow cytometric analysis (CytoFlex; Beckman-Coulter). At least 3000 cells were acquired in the caspase − /DAPI − gating region. Data analysis was performed utilizing Kaluza 1.5a software (Beckman-Coulter).
Live cell imaging. Cells were challenged with ADDA5, KCN or NaN 3 in the presence or absence of PTM for 3 h or 24 h. Cells were loaded with either cell death indicator SYTOX Green (1 µM; Thermo scientific), mitochondrial membrane potential indicator, Tetramethylrhodamine ethyl ester (TMRE, 100 nM; AAT bioquest), or superoxide sensitive dye dihydroethidium (DHE, 500 nM, Enzo life sciences) for 30 min at 37 °C. Cells were imaged with a 20X objective using a live cell high throughput imaging system (Operetta CLS; Perkin Elmer) and cell-based quantification was performed with minimum of 300 cells for each condition using dedicated imaging software (Harmony 4.6; Perkin Elmer).
Small Interfering RNA-Mediated Knockdown of COX4. 1 × 10 4 of SK-MEL-28 cells were transfected with esiRNA against human COX4I1 (Sigma-Aldrich) or non-targeting control esiRNA (Luc) using X-tremeGENE siRNA reagent (Sigma-Aldrich) according to manufacturer's recommendation. Cells were lysed after 48 h and the knockdown efficiency of CcO confirmed by immunoblotting. The remaining cells were incubated with PTM for 6 h along with respective controls and viability was measured using sytox green staining.
Immunoblotting. Cells were harvested in ice cold PBS and lysed in RIPA buffer (Pierce) supplemented with complete protease and phosphatase inhibitors (PIM complete; Roche) for 20 min on ice. After centrifugation at 15,000 g for 15 min at 4 °C, total protein in the cell extracts was quantified using Rotiquant (Carl Roth). Forty micrograms of protein was resolved by SDS-PAGE (Invitrogen) and blotted on PVDF membranes (Invitrogen). The membranes were probed with anti-caspase 3, anti-beta actin (Cell Signaling), or anti COX4 (Santa cruz) primary antibodies followed by secondary horse-radish peroxidase-coupled antibodies (Rockland Immunochemicals), and signals were acquired in a chemiluminescence detection system (Applied Biosystems) in a linear dynamic range. ATP assay. Cells were treated with KCN or NAN 3 in the presence or absence of PTM for 2 hours. Total cellular ATP levels quantified using luminescence-based ATP determination kit (Enzo Life Sciences) according to manufactures' instructions. Spheroid assay. SK-MEL-28 cells (5 × 10 3 ) were incubated in 96 well ultra-low affinity plates (Perkin Elmer). Seventy-two hours later, cells were challenged with ADDA 5 (1 µM) in the presence or absence of PTM for 24 h. Spheroids were loaded with Sytox green (5 µM; Thermo scientific) and Hoechst 33342 (40 µM; Thermo scientific) for 1 h at 37 °C. Spheroids were imaged with a 5X objective at 50 stacks per well with a live cell high throughput imaging system (Operetta CLS; Perkin Elmer) and quantified by taking the ratio between sytox green and Hoechst 33342 mean fluorescence intensity using dedicated imaging software (Harmony 4.5; Perkin Elmer).

PTM only modestly decreased metabolic activity and cell viability. Hydrogen peroxide (H 2 O 2 ) is
the most stable oxidant derived from plasma-treated medium (PTM) in this setting and with this device. To ensure a scalability of PTM, we quantified H 2 O 2 and found a linear increase with treatment time (Fig. 1B). Following optimization on cell lines (data not shown), we chose the 120 s treatment time for subsequent experiments. There was a modest decrease in metabolic activity of all three melanoma cell lines at 3 h (Fig. 1C). However, B16F10 but not B16F0 and SK-MEL-28 cells recovered from the initial insult after 24 h (Fig. 1C). We further enumerated cytotoxic effects of PTM by live cell imaging at 3 h and 24 h post incubation. There was a minor increase in cell death (<10%) in all melanoma cells treated with PTM at both time points (Fig. 1D). Furthermore, SK-MEL-28 were more sensitive to H 2 O 2 than non-malignant human HaCaT keratinocytes (Fig. S1). Altogether, the application of PTM was of minor toxicity in melanoma cells. CcO inhibition sensitized melanoma cells but not keratinocytes to PTM-mediated cell death. KCN and NaN 3 are routinely used biochemical inhibitors of mitochondrial complex IV. We tested these compounds' ability to induce cytotoxicity in the 3 melanoma cells and non-malignant human keratinocytes (HaCaT). There was no significant reduction in metabolic activity following incubation with KCN (250 or 500 µM) and NaN 3 (100 or 500 µM) at 24 h ( Fig. 2A,B). However, there was a significant reduction of metabolic activity in melanoma cells but not keratinocytes upon co-incubation of cells with CcO inhibitors and PTM (Fig. 2C,D). The observed synergistic effect was also evident by sytox green staining revealing massive induction of cell death following 6 h of treatment (Fig. 2E) and complete cell killing at 24 h (data not shown). Treating cells with KCN (500 µM) or NaN 3 (500 µM) alone did not induce any cytotoxicity (Fig. 2E). Cell death was rescued by pretreating cells with catalase indicating H 2 O 2 as primary mediator in the observed cellular toxicity. To confirm a prime role of H 2 O 2 , we repeated the metabolic assays with 50 µM of H 2 O 2 in the presence of KCN or NaN 3 . (Fig. S2A,B). Again, the effect could be quenched by catalase.

Synergistic cell death was mediated by mitochondrial dysfunction.
To understand the contributory mechanism of the synergistic of cell death, we quantified the mitochondrial membrane potential by TMRE and superoxide levels by DHE staining in B16F10 cells using live cell imaging. We observed no significant increase in superoxide levels at 2 h incubation with cells treated with only PTM. However, there was a pronounced increase in superoxide levels with co-incubation with NaN 3 (Fig. 3A,B) or KCN (Fig. 3C,D). This was accompanied by a loss of mitochondrial membrane potential, evident by reduced uptake of TMRE, which was abrogated by pretreating cells with catalase (Fig. 3A,D). As expected, this went along with a significant drop in ATP levels at 2 h following co-incubation of NaN 3 (Fig. 3E) or KCN (Fig. 3F) with PTM. These data imply that mitochondrial CcO inhibition and exogenous oxidants accelerate cell death by inducing mitochondrial dysfunction in melanoma cells. Synergistic cell death was independent of caspase 3 or 7 activity. Since there was a pronounced mitochondrial dysfunction, we then investigated whether there was an activation of the intrinsic apoptotic pathway preceding cell death in B16F10 cells. Following treatment, cells were incubated with fluorescent indicators for active caspases 3 or 7 and a membrane-impermeable nucleic acid dye to distinguish apoptotic and necrotic populations by flow cytometry (Fig. 4A,B). Results indicated the PTM led to a minor increase in cell death comparable to Fig. 1B. However, induction of cell death was more severe upon co-incubation with KCN or NaN 3 in a caspase independent mechanism. This was supported by the lack of cleaved caspase 3 at the protein level (Fig. 4C). These data indicate a rapid onset of necrotic cell death in melanoma cells due to acquired energy crisis following incubation with CcO inhibitors and exogenous oxidants.

SiRNA mediated inhibition of CcO sensitizes melanoma cells to oxidant-induced cell death.
Biochemical and pharmacological inhibitors are known to have off target effects leading to extrapolated results. Hence to validate our hypothesis we used siRNA based gene knockdown of COX subunit 4 isoform 1 (COX4l1) a major subunit of mitochondrial CcO. We performed knockdown experiments in human SK-MEL-28 cells using siRNA against the COX4l1 with appropriate controls. Silencing CcO was determined at the protein level using anti COX4 antibody and anti β-actin antibody as loading control (Fig. 5A). Cells transfected with COX4l1 siRNA had no significant change in baseline metabolic activity when compared to non-targeting control siRNAs after 24 h (Fig. 5B). However, incubation with PTM led to a significantly higher induction of cell death in COX4 siRNA compared to non-targeting and transfection control cells (Fig. 5C,D). These results validate that inhibition of CcO sensitizes melanoma cells to oxidant induced cell death. CcO inhibition by ADDA 5 in the presence of exogenous oxidants leads to melanoma cell death. ADDA 5 was recently demonstrated to be an effective small molecule inhibitor against CcO activity in glioma cells in vitro and in vivo. We first determined the cellular metabolic activity in the presence or absence of PTM or ADDA5 over 5 log concentrations during 24 h in all 3 melanoma cell lines and HaCaT. At 10 µM, co-incubation with PTM and ADDA5 resulted in a significantly reduced metabolic activity in melanoma cells but not in non-malignant HaCaT keratinocytes when compared to controls (Fig. 6A). This reduction positively correlated with image quantification of Sytox-positive terminally dead cells after 24 h (Fig. 6B). Finally, we extended our 2D monolayer findings to a 3D tumor spheroid model using SK-MEL-28 cells. Cells within spheroids exhibited terminal cell death upon co-incubation but not monotherapy with ADDA 5 and/or PTM after 24 h at indicated concentrations. The induced toxicity was abrogated by prior addition of catalase (1000 units/ml) indicating exogenous H 2 O 2 being important to synergize with CcO inhibition in melanoma cell death (Fig. 6C).

Discussion
We demonstrated that the inhibition of CcO in conjunction with plasma-derived exogenous oxidants might constitute an innovative new treatment approach in melanoma. Consistent with previous studies, we found that melanoma cells are relatively resistant towards oxidative stress 33 . In our study, PTM generating a H 2 O 2 concentration of 61.3 µM did not have a significant cytotoxic effect on melanoma cells although it was previously shown to decrease their motility 34 . This observation may be attributed to the robust anti-oxidative capacity and pro-oxidant tumor milieu in melanoma cells 11 . In our study, biochemical inhibition of CcO activity with KCN or NaN3 (≤500 µM) did not alter the viability or metabolic activity of melanoma cells or normal keratinocytes. KCN and NaN3 are routinely used inhibitors of CcO activity in mammalian cells 35 . Previous studies employed higher concentrations (>1 mM) to inhibit CcO activity in multiple cell types [36][37][38] . However, only few studies have implicated the role of CcO in tumor progression. Genetic knockdown of CcO in human esophageal squamous cell carcinoma led to a shift in metabolic reprogramming and slower growth rates but enhanced tumor invasion 39 . On the contrary, CcO inhibition led to temozolomide-dependent apoptosis in chemoresistant glioma cells suggesting a synergistic role in tumor killing 40 .
Our observed selective and synergistic cytotoxicity of PTM plus CcO inhibition was accompanied by loss of mitochondrial membrane potential and ATP depletion. It has been previously demonstrated that CcO is crucial for stability and activity of mitochondrial complexes I, II, and III 37,41 . CcO inhibition also leads to a subtle increase in superoxide generation at complex I 42 when the NADH/NAD + ratio is high 43 . Highest superoxide fluxes are seen with reverse electron transport at complex I 44 . Inhibition of complex III leads to elevated superoxide release as well 45 . Consistent with our results, CcO inhibition with, for example, cyanide, is not increasing superoxide production, at least in complex III 46 . Superoxide has a short half-life and is efficiently converted to stable H 2 O 2 by superoxide dismutase 47 . We found a significant elevation of superoxide only upon CcO inhibition in the presence of PTM, suggesting oxidative stress and subsequent redox signaling as major effector of cellular toxicity. Interestingly, mitochondrial superoxide release is also seen under hypoxic conditions 48 . Connecting to our results, inactivation of hypoxia-inducible factor 1 alpha increases mitochondrial oxidative metabolism, which sensitizes melanoma cells to pro-oxidant killing 49 .
In the current study, there was no activation of executioner caspases 3/7 indicating necrosis as the primary mode of cell death 50 . In general, melanoma cells have been shown to be resistant to apoptotic cell death 51,52 . The absence of Receptor Interacting Protein 3 Kinase (RIP3K) in melanoma cells leads to necroptosis resistance as well 53 . Although the combination treatment led to energy crisis, the observed cell death was restricted to tumor cells. Normal and cancer cells require basal mitochondrial Ca 2+ uptake for survival, however upon mitochondrial dysfunction and energy crisis, led to activation of autophagy leading cell cycle arrest and survival in normal cells. However, tumor cells bypass cell cycle checkpoints and undergo necrotic cell death during mitosis 54 . Hence, the inability of melanoma cells to slow down proliferation during ATP depletion accounts for their vulnerability to cell killing. Similar finding was recently observed that cold plasma treated medium induced caspase-independent cell death via mitochondrial dysfunction and Ca 2+ homeostasis in TRAIL resistant human malignant cells 55 .
CcO is a multimeric enzyme composed of several metal prosthetic sites and 14 protein subunits in mammals 49 . We further confirmed our treatment approach (the proposed cellular events are summarized in Fig. 7) by genetically knocking down the subunit complex IVi1 by siRNA in human SK-MEL-28 cells. Knockdown of subunit IVi1 leads to disruption of the assembly and function of CcO in the mitochondria 56 . As expected, siRNA-mediated dysfunction of CcO led to a synergistic induction of melanoma cytotoxicity when exposed to PTM (alone and in the absence of an additional pharmacological CcO inhibitor).
Finally, we validated our above findings using a specific inhibitor of CcO, ADDA 5, which was recently identified from a large compound screening 57 . The potency of the drug is underlined by the absence of side effects in mice that received up to 80 mg/kg body weight for 7 days. Our combinational approach with ADDA5 and PTM was markedly effective in 2D and 3D melanoma cultures. Furthermore, our study demonstrates ADDA5 (10 µM) is a superior complex IV inhibitor in comparison to KCN or NaN 3 (>250 µM) and could be used for biochemical studies of electron transport chain.
In conclusion, we demonstrated that inhibition of CcO in the presence of exogenous oxidants sensitized melanoma cells towards mitochondrial dysfunction and cell death. Although targeting oxidative phosphorylation pathways in healthy cells would be detrimental, selective mitochondrial complex IV inhibitors would be not. To this end, a novel inhibitor, ADDA 5, showed no toxicity in mice and significant cytotoxicity towards melanoma cells and spheroids in combination with PTM in our study. The kINPen plasma device has already received approval for clinical applications such as wound healing 58,59 and showed promising results in palliative cancer care 60,61 . We foresee plasma-derived reactive could also be delivered to cutaneous, metastatic, malignant melanoma due its direct accessibility for plasma treatment or plasma-treated liquids for mono or combination therapy. However, further exploration of this innovative approach is warranted in pre-clinical dermato-oncology.