Abstract

   Strategies to improve activity and selectivity are major goals in
   oncological drug development. Medical gas plasma therapy has been
   subject to intense research in dermatooncology recently. Based on
   partial gas ionization, this approach is exceptional in generating a
   variety of reactive oxygen species simultaneously that can be applied
   locally at the tumor side. It is hypothesized that combined gas plasma
   treatment can potentiate drug responses in the treatment of melanoma.
   Using a plasma jet approved as medical device in Europe, a systematic
   screening of 46 mitochondria‐targeted drugs identifies five agents
   synergizing in vitro and in vivo. Increased intratumoral leucocyte
   infiltration points to immunomodulatory aspects of the treatment,
   motivating to investigate responses to immune checkpoint blockade in
   combination with plasma. Tumor growth is monitored based on
   bioluminescent imaging, and single‐cell suspensions are retrieved from
   each tumor to characterize tumor‐infiltrating leucocytes using
   multicolor flow cytometry. Gene expression profiling is done using a
   validated NanoString panel targeting 770 genes specifically designed
   for immuno‐oncological research. Cell type abundancies are
   characterized from bulk RNA samples using the CIBERSORT computational
   framework. Collectively, the results indicate that local application of
   medical gas plasma technology synergizes with mitochondria‐targeted
   drugs and anti‐PD1 checkpoint therapy in treating melanoma.

   Keywords: B16F10, CAP, immuno‐oncology, reactive oxygen species
     __________________________________________________________________

   Medical gas plasma is a partially ionized gas operated at body
   temperature and generating various reactive species simultaneously.
   Using in vitro and in vivo melanoma models, gas plasma‐derived reactive
   species potently combine with novel mitochondria‐targeting drugs as
   well as antibody immune‐checkpoint therapy to increase tumor
   immunogenicity, antitumor immune responses, and tumor eradication.

   graphic file with name ADVS-10-2303183-g005.jpg

1. Introduction

   Our scientific understanding of the pathogenesis of diseases evolves
   with technological and diagnostical advances, which, at the same time,
   drives the development of refined therapies interfering with
   deregulated signaling pathways tailored to the patient's situation. In
   oncology, the identification of cancer‐associated (and specific)
   microenvironmental and molecular signatures initiated a therapeutical
   shift from indiscriminate, cytotoxic agents to novel genome‐ and
   immune‐targeted therapies that improved outcomes of many malignancies
   in recent years. Yet, precision medicine is regularly challenged by
   on‐treatment tumor regrowth caused by secondary resistance mechanisms
   that limit therapeutic efficacies. In melanoma, a highly aggressive
   skin cancer often considered a model tumor in oncological research,
   therapeutic efficacy of standard‐of‐care BRAF (v‐raf murine sarcoma
   viral oncogene homolog B1)‐ and MEK (mitogen‐activated protein kinase
   kinase)‐inhibitors is often limited by extensively branched evolution
   of melanoma subclones.^[ [34]^1 , [35]^2 ^] Likewise, immune checkpoint
   blockade (ICB), the shining example of therapeutical advances in modern
   oncology, shows limited efficacy in 40% to 65% of patients, while
   further 43% of patients develop secondary resistance within three
   years.^[ [36]^3 ^] Considering tumor heterogeneity and the bottleneck
   effects of targeted therapies, growing evidence suggests that the
   precision medicine paradigm of cancer therapy requires the
   identification of eligible combination regimes rather than matching
   monotherapeutical approaches.^[ [37]^4 , [38]^5 ^]

   In the early 1990s, the unexpected finding of reactive oxygen species
   (ROS) limitation to promote cancer development,^[ [39]^6 , [40]^7 ^]
   followed by the discovery of ROS‐dependent cell death signaling
   elicited by an array of conventional chemotherapeutics and
   radiotherapy,^[ [41]^8 , [42]^9 ^] fostered the development of
   ROS‐generating therapy approaches such as photodynamic therapy (PDT)^[
   [43]^10 ^] and medical gas plasma technology.^[ [44]^11 ^] Exploiting
   the concept of hormesis, such approaches aim at eliciting irreversible
   oxidative distress in cancer cells upon exposure to ROS at
   supraphysiological levels. While PDT is characterized by the local
   formation of singlet delta oxygen upon photoactivation of a
   systemically administered photosensitive agent, gas plasmas are
   exceptional in generating a multitude of highly reactive species
   simultaneously upon partial ionization of an operating gas. Plasma
   treatment of cancer cells has been subject of intense research in
   recent years showing antiproliferative, antimetabolic, senescence‐, and
   cell death‐elevating effects in diverse tumor models in vitro and in
   vivo.^[ [45]^12 , [46]^13 , [47]^14 , [48]^15 , [49]^16 , [50]^17 ^]
   Clinically, medical gas plasma therapy has been applied in the
   palliation of head and neck cancer patients in a series of clinical
   case reports conducted at the University Medical Center in Greifswald.
   Here, partial tumor remission (reduction in tumor size) could be
   observed in one‐third of patients, paralleled by an increase in
   patients’ life quality due to reduced odor, pain, and microbial load.^[
   [51]^18 , [52]^19 ^] Despite direct tumor toxicity, gas plasma‐mediated
   tumor oxidation has been shown to create an inflammatory‐like
   environment able to trigger antitumor immune responses, e.g., through
   induction of immunogenic cell death (ICD) via ROS‐induced endoplasmatic
   reticulum (ER)‐stress^[ [53]^20 , [54]^21 ^] or formation of
   neoantigens via oxidation of biomolecules.^[ [55]^22 , [56]^23 ^]

   We hypothesized that local tumor oxidation using medical gas plasma
   technology could serve as a therapeutical sensitizer in oncological
   treatment regimens, improving response rates. Starting from a library
   of 46 mitochondria‐targeted drugs, five combination regimes were
   identified showing strong synergistic effects and subjected to further
   functional analysis in vitro and in vivo. In light of observed
   immune‐modulatory effects, medical gas plasma technology was combined
   with immune checkpoint blockade using pembrolizumab (anti‐PD1 antibody)
   in a syngeneic model of melanoma. Tumor growth was monitored based on
   bioluminescent imaging, and single‐cell suspensions were retrieved from
   each tumor to characterize tumor‐infiltrating leucocytes using
   multicolor flow cytometry. Gene expression profiling was done using
   NanoString technology and the PanCancer IO 360 Panel targeting 770
   genes specifically designed for immuno‐oncological research. Cell type
   abundancies were characterized from bulk RNA samples using the
   CIBERSORT computational framework.

2. Results

2.1. Mitochondria‐Targeted Drugs Synergize with Medical Gas Plasma in the
Treatment of Melanoma In Vitro and In Vivo

   A library of 46 mitochondria‐targeted drugs was screened to identify
   promising candidates synergizing with medical gas plasma technology in
   the treatment of melanoma. A three‐step screening procedure was
   employed to identify substances with maximal combinational efficacy in
   murine and human melanoma cells while exhibiting low toxicity on
   non‐neoplastic human keratinocytes. Selected drug combination regimes
   were further investigated in vitro and in a syngeneic model of melanoma
   in vivo (Figure [57] 1a). Placed in the heart of applied redox biology,
   medical gas plasma technology is exceptional in generating a variety of
   ROS and RNS simultaneously (Figure [58]1b), eliciting oxidative stress
   in cells (Figure [59]1c) followed by induction of irreversible cell
   death signaling when applied at supraphysiological levels. Increased
   basal intracellular ROS levels occurring upon malignant transformation
   suggest a lowered effective threshold in cancer cells, entailing a
   potential and often described selectivity of medical gas plasmas and a
   high therapeutic index (Figure [60]S1, Supporting Information). Based
   on the cellular metabolic activity 24 h after treatment, fourfold
   dose–response curves were generated for murine B16F10 and human
   SK‐MEL‐28 melanoma cells to identify substances without therapeutic
   efficacy (no effect), without synergistic effects (drug alone), and
   significant combination effects following exposure to gas
   plasma‐derived ROS (combination). Drugs synergizing with medical gas
   plasmas were chosen for further analysis (n = 32 drugs). Inhibitory
   concentration (IC) 50 ratios were calculated to identify drugs showing
   a leastwise fivefold decreased IC50 in combination versus
   monotreatments in human SK‐MEL‐28 (Figure [61]1d) and murine B16F10
   melanoma cells (Figure [62]S2a, Supporting Information), contributing
   to a set of 11 drug candidates. Combination regimes with a narrow
   therapeutic index were further excluded based on calculated IC50 ratios
   of combination treatments in non‐neoplastic HaCaT keratinocytes versus
   SK‐MEL‐28 melanoma cells (Figure [63]S2b, Supporting Information).
   Thus, five compounds, A‐1210477 (A12), Carvedilol (Car), Cozymasei
   (Coz), SBI‐0206965 (Coz), and Navitoclax (Nav), were finally identified
   as promising drug candidates and further characterized in vitro and in
   vivo.

Figure 1.

   Figure 1
   [64]Open in a new tab

   Mitochondria‐targeted drugs synergize with medical gas plasma treatment
   of melanoma cells in vitro. a) Study protocol. b) Plasma jet kINPen MED
   used in this study. c) Representative H[2]DCFDA and DAPI fluorescence
   images of control and gas plasma‐treated cells (scale bar = 50 µm). d)
   Representative dose–response curves and waterfall plot showing
   inhibitory concentration (IC) 50 ratios of monotreatment versus
   combination regimes in SK‐MEL‐28 human melanoma cells. Drugs were
   selected for significant combination effects (p < 0.05; n = 32) and
   IC50 fold change (FC) > 5 (n = 11). Graphs show mean. Statistical
   analysis was performed using one‐way analysis of variances (ANOVA). e)
   Representative digital phase contrast (DPC) images (scale bar =
   100 µm). f) Cell death 24 h after treatment. Bar graphs show mean ±
   standard error of the mean (SEM). Statistical analysis was performed
   using two‐way ANOVA against vehicle (plasma) or drug monotreatments
   (combination regimes) (*p < 0.05, **p < 0.01, ***p < 0.001). g)
   Intracellular levels of reduced glutathione (GSH) 6 h after treatment
   (scale bar = 100 µm). Heat map shows mean. Statistical analysis was
   performed using two‐way ANOVA against vehicle (plasma) or drug
   monotreatments (combination regimes) (*p < 0.05, **p < 0.01, ***p <
   0.001). h) Representative tetramethylrhodamine (TMRE) fluorescence
   images indicative of mitochondrial membrane potential (ΔΨm) 24 h after
   treatment. i) Cell death characterization. j) Representative
   brightfield and sytox orange fluorescence images (scale bar = 50 µm).
   k) Cell death 24 h after treatment. Graph shows mean. Statistical
   analysis was performed using two‐way ANOVA against vehicle (plasma) or
   drug monotreatments (combination regimes). l) Representative flow
   cytometry intensity histogram showing caspase 3 and 7 activity. m)
   Representative Hoechst and sytox green fluorescence images of tumor
   spheroids. n) 5–95 percentile box plots showing sytox green intensity
   in tumor spheroids 24 h after treatment. Mean is indicated as (+).
   Statistical analysis was performed using two‐way ANOVA against plasma
   monotreatment (*p < 0.05). ns = nonsignificant. ROS = reactive oxygen
   species. nec‐1 = necrostatin‐1. w/o = without. MFI = mean fluorescence
   intensity.

   Combined drug–plasma therapy reduced cancer cell proliferation
   (Figure [65]1e) and enhanced tumor toxicity 6 h (Figure [66]S2c,
   Supporting Information) and 24 h (Figure [67]1f) after treatment
   compared to drug monotreatments. Oxidative distress is characterized by
   an overwhelming consumption of cellular redox capacities, observed for
   the antioxidant tripeptide glutathione (GSH) predominantly after
   exposure to gas plasma‐derived ROS in our study (Figure [68]1g). Loss
   in mitochondrial membrane potential (ΔΨm) as a surrogate marker for
   mitochondrial dysfunction and irreversible oxidative damage was
   evaluated using tetramethylrhodamine (TMRE; Figure [69]1h). TMRE is
   readily sequestered by active mitochondria due to the dye's positive
   charge and indicated significant mitochondrial damage upon
   combinational treatment (Figure [70]S1d, Supporting Information).
   Mitochondria are implicated in an expanding array of cellular processes
   and considered multifaceted regulators of cell death. Cell death
   modalities predominantly induced in combination regimes were further
   evaluated using the pan‐caspase, hence apoptosis, inhibitor Z‐VAD‐FMK,
   and the RIP‐1 kinase inhibitor necrostatin‐1 (nec‐1) for inhibition of
   necroptosis (Figure [71]1i). Algorithm‐based image analysis for
   segmentation of sytox orange^+ melanoma cells (Figure [72]1j) revealed
   significant cell death inhibition in the presence of Z‐VAD‐FMK
   indicating apoptosis as the preferred mode of execution upon combined
   drug–plasma therapy (Figure [73]1k). This was underlined by increased
   activation of effector caspases 3 and 7 in combination regimes
   (Figure [74]1l). A superior efficacy was further confirmed in 3D
   melanoma spheroids (Figure [75]1m) outlining significant synergistic
   effects (Figure [76]1n).

   Promising drug candidates were next investigated in a syngeneic model
   of melanoma in vivo. Tumor growth was induced by subcutaneous injection
   of 1 × 10^6 B16F10 melanoma cells in the left flank of C57/Bl6 mice.
   Animals were randomly divided into ten groups after reaching a tumor
   volume of ≈50 mm^2. Melanoma‐bearing mice received either (a)
   intraperitoneal drug monotreatment, (b) local gas plasma treatment, or
   (c) combination of both. Treatments were performed every other day in
   four treatment cycles until animals were sacrificed on day 7
   (Figure [77] 2a). Strikingly, combined tumor toxic effects were
   observed in all treatment groups (Figure [78]2b). Histopathological
   analysis of cryosectioned tumor nodes further revealed a reduced
   proliferative index (Figure [79]2c) and increased activation of
   effector caspase 3 in line with previous experiments (Figure [80]2d).
   Moreover, combinational treatment increased the number of
   tumor‐infiltrating CD45^+ leucocytes in all groups indicating immune
   modulatory aspects of the treatment (Figure [81]2e).

Figure 2.

   Figure 2
   [82]Open in a new tab

   Mitochondria‐targeted drugs synergize with medical gas plasma treatment
   in a syngeneic melanoma model in vivo. a) Study protocol. b) Tumor
   weight. Bar graphs show mean ± standard error of the mean (SEM).
   Statistical analysis was performed using two‐way analysis of variance
   (ANOVA) against vehicle (plasma) or drug monotreatments (combination
   regimes), respectively (*p < 0.05, ***p < 0.001). c–e) Representative
   immunohistochemistry images of Ki‐67 (c), cleaved caspase 3 (d), and
   CD45 (e) staining and quantification thereof (scale bar = 200 µm).
   Graphs show mean and individual values. Statistical analysis was
   performed using two‐way ANOVA against vehicle (plasma) or drug
   monotreatments (combination regimes), respectively (*p < 0.05, **p <
   0.01, ***p < 0.001). ns = nonsignificant.

2.2. Gas Plasma Treatment Reduces Myeloid and Lymphoid Checkpoint Expression
Levels on Melanoma Cells In Vitro and In Vivo

   The previous findings highlighted the potential of combined drug–plasma
   therapy in the treatment of melanoma and indicated immune‐modulatory
   aspects of the treatment in vivo. Despite direct tumor‐toxic effects
   outlined above, gas plasma‐derived ROS have likewise been implicated to
   act as bona fide inducers of ICD and may facilitate formation of
   neoantigens via oxidation of biomolecules. In this light, we
   hypothesized that gas plasma treatment could potentiate responses to
   ICB in the treatment of melanoma. Following successful
   cross‐presentation and T cell priming, an inhospitable tumor
   microenvironment (TME) may preclude proper function of the expanded T
   cell repertoire, limiting the efficacy of ICB. This might include the
   aberrant expression of alternate immune checkpoints or coinhibitory
   receptors, immunosuppressive cytokines, and immune‐inhibitory
   metabolites. As the ICD‐inducing activity of gas plasmas is well
   described in the literature, we aimed to characterize the immune
   checkpoint expression profile in response to plasma‐mediated tumor
   oxidation (Figure [83] 3a). Flow cytometric screening (Figure [84]3b)
   of an array of 23 surface‐expressed checkpoint molecules
   (Figure [85]3c) revealed a significant downregulation for eight markers
   upon plasma treatment, while a significant upregulation was only found
   for CD70 (Figure [86]3d). Checkpoint expression signatures were further
   evaluated after gas plasma monotreatment and combined gas plasma–ICB in
   vivo (Figure [87]3e). Single cell suspensions were retrieved from
   excised B16F10 tumors and stained with monoclonal antibodies targeting
   CD47, CD80/CD86, and CD278 followed by flow cytometric analysis
   (Figure [88]3f). WPGMA‐weighted hierarchical clustering of checkpoint
   marker expression signatures on individual tumors revealed lower
   expression levels in groups receiving plasma monotreatment or the
   combination with anti‐PD1 ICB (Figure [89]3g), which was underlined by
   notable differences in the principal component (PC) 1 (Figure [90]3h).
   Propagation of an effector T cell response in draining lymph nodes
   requires successful antigen‐processing and presentation by
   antigen‐presenting cells (APCs). As outlined above, medical gas plasmas
   are considered bona fide ICD inducers, indicated by expression and
   release of damage‐associated molecular patterns (DAMPs) after
   ROS‐induced cell death that may attract APCs to the tumor side. In
   order to investigate cancer–immune cell interactions after plasma
   treatment, monocyte‐derived CD11c^+ dendritic cells were challenged
   with cell trace violet (CTV)‐labeled plasma‐oxidized melanoma cells
   (oxmel; Figure [91]3i). Flow cytometric evaluation (Figure [92]3j) of
   CTV fluorescence in CD11c^+ dendritic cells (Figure [93]3k), indicative
   of tumor uptake, indicated increased phagocytosis of oxmel versus
   heat‐inactivated melanoma cells (mel; Figure [94]3l).

Figure 3.

   Figure 3
   [95]Open in a new tab

   Gas plasma treatment reduces myeloid and lymphoid checkpoint expression
   levels on melanoma cells in vitro and in vivo. a) Study protocol. b)
   Representative flow cytometry intensity histograms showing tumoral
   expression of immunosuppressive surface markers. c) Waterfall plot
   showing fold change (FC) expression levels of immunosuppressive surface
   markers on human melanoma cells after gas plasma treatment. Bar graphs
   show mean. d) Volcano plot showing differentially (p > 0.05) up‐ (red)
   or downregulated (green) markers. Statistical analysis was performed
   using one‐way analysis of variances (ANOVA). e) Study protocol. f)
   Representative flow cytometry intensity histogram showing CD80/CD86
   expression B16F10 melanoma tumors. g) WPGMA‐weighted hierarchical
   clustering of z‐scored surface expression of CD47, CD80/CD86, and CD278
   on B16F10 melanoma tumors of individual animals after treatment with
   gas plasma, anti‐PD1 or the combination of both. h) Principal component
   analysis (PCA) calculated from z‐scored mean surface expression levels
   of CD47, CD80/CD86, and CD278 on B16F10 melanoma tumors of individual
   animals after treatment with gas plasma, anti‐PD1 or the combination of
   both. i) Study protocol. j) Representative flow cytometry cell trace
   violet (CTV) versus CD11c contour plots of CD11c^+ dendritic cells
   (DCs) challenged with CTV‐labeled melanoma cells. k) Representative
   flow cytometry intensity histograms showing CTV fluorescence in CD11c^+
   DCs challenged with CTV‐labeled heat‐killed (mel) or oxidized melanoma
   cells (oxmel). l) Basline‐corrected CTV mean fluorescence intensity
   (MFI) in CD11c^+ DCs challenged with heat‐killed (mel) or oxidized
   melanoma cells (oxmel). Bar graphs show mean ± standard error of the
   mean (SEM). Statistical analysis was performed using t‐test (**p <
   0.01). norm = normalized. FDR = false discovery rate. PC = principal
   component. ROS = reactive oxygen species. αPD1 = anti‐PD1.

2.3. Medical Gas Plasma Technology Synergizes with Anti‐PD1 Immune Checkpoint
Blockade in a Syngeneic Model of Melanoma In Vivo

   Before investigating the therapeutic effects of combined gas
   plasma–ICB, the efficiency of respective monotreatments was first
   confirmed in vivo. For tumor induction, 5 × 10^4 syngeneic B16F10
   melanoma cells were injected in both flanks of C57/Bl6 mice. Animals
   were left untreated, received intraperitoneal injection of anti‐PD1
   checkpoint antibodies or local application of medical gas plasma in
   three treatment cycles starting from day 2 postinoculation (Figure [96]
   4a). Tumor growth was monitored based on bioluminescent imaging showing
   that both medical gas plasma and anti‐PD1 ICB successfully reduced
   melanoma growth progression (Figure [97]4b). Next, monotherapies were
   tested against the combination of both (Figure [98]4c). Strikingly,
   combined gas plasma–ICB increased remission rates compared to anti‐PD1
   and gas plasma monotreatments (Figure [99]4d) and reduced melanoma
   growth progression in tumor‐bearing animals (Figure [100]4e). Moreover,
   a tending contralateral reduction in growth of the non‐plasma‐treated
   tumor indicated an augmented abscopal effect (Figure [101]S3). We
   hypothesized that the augmented efficacy of combined gas plasma–ICB
   would also correlate with an altered immune landscape in the TME. To
   this end, single‐cell suspensions were retrieved from each tumor and
   levels of tumor‐infiltrating leucocytes were evaluated (Figure [102]
   5a). Flow cytometric analysis of individual tumors revealed gas plasma
   treatment to increase infiltration of CD11c^+ dendritic cells and
   F4/80^+ macrophages in the tumor bed, which was also observed upon
   combined ICB, but not in animals receiving ICB monotreatment
   (Figure [103]5b). By contrast, significant differences in CD4^+ and
   CD8^+ T cell infiltration could not be observed (Figure [104]5c).
   Regulation of CD62L (l‐selectin) expression plays a pivotal role in T
   cell trafficking to and from peripheral lymph nodes. Shedding of CD62L
   (Figure [105]5d), as a surrogate marker for T cell activation, was
   significantly increased on CD8^+ cytotoxic T cells upon combined ICB
   and monotreatment (Figure [106]5e). Moreover, ICB was a major predictor
   for CD25 expression (Figure [107]5f) on CD4^+ and CD8^+ T cells
   (Figure [108]5g). Similar observations were made for CD69
   (Figure [109]5h) underlining increased activation of CD4^+ and CD8^+ T
   cells upon exposure to ICB (Figure [110]5i). T cell exhaustion as a
   state of T cell dysfunction is characterized by increased expression of
   alternate checkpoints on T lymphocytes upon chronic activation and can
   impair response to ICB. To this end, surface expression of CD134
   (Ox40), CD278 (ICOS), and CD279 (PD1) was evaluated on
   tumor‐infiltrating T cells using flow cytometry (Figure [111]5j).
   Interestingly, ICB reduced relative expression of Ox40 and (unbound)
   PD1 on intratumoral T cells, while ICOS expression was only reduced
   upon plasma monotreatment (Figure [112]5k). By mapping intratumoral
   cytokine release profiles using a bead‐based multiplex ELISA, we aimed
   to understand further the immunological consequences of combined gas
   plasma–ICB (Figure [113]5l). Antibodies targeting interferon gamma
   (IFNγ), tumor necrosis factor alpha (TNFα), and interleukin (IL) 2
   indicated only minor differences in antitumoral T[H]1 responses when
   comparing the different treatment regimes. A slight increase was found
   for T[H]2‐ and T[H]17‐related cytokines, including IL4, IL6, IL17A, and
   IL17F in animals receiving ICB (Figure [114]5m).

Figure 4.

   Figure 4
   [115]Open in a new tab

   Medical gas plasma technology synergizes with anti‐PD1 immune
   checkpoint blockade in a syngeneic model of melanoma in vivo. a) Study
   protocol. b) Tumor bioluminescence 2, 6, 10, and 14 days after
   subcutaneous (s.c.) tumor inoculation and calculated area under the
   curve (AUC). Graphs show mean ± standard error of the mean (SEM).
   Statistical analysis was performed using one‐way analysis of variances
   (ANOVA) (*p < 0.05). c) Representative bioluminescent images. d)
   Remission rates based on bioluminescence. Each circle refers to an
   individual animal. e) Tumor bioluminescence on day 11. Bar graphs show
   mean ± SEM. Statistical analysis was performed using one‐way ANOVA
   (*p < 0.05). ns = nonsignificant. αPD1 = anti‐PD1. p.i. =
   postinoculation.

Figure 5.

   Figure 5
   [116]Open in a new tab

   Medical gas plasma and anti‐PD1 immune checkpoint blockade synergize in
   remodeling the intratumoral melanoma immune landscape. a) Study
   protocol. b,c) Flow cytometric quantification of myeloid (b) and
   lymphoid (c) leucocytes in tumors retrieved from melanoma‐bearing
   animals. Bar graphs show mean and individual data points. Statistical
   analysis was performed using one‐way analysis of variances (ANOVA)
   (*p < 0.05, **p < 0.01). d,e) Representative optSNE plots of CD62L
   expression (d) and quantification thereof on CD4^+ and CD8^+
   lymphocytes (e). Graphs show mean (line) and individual data points.
   Statistical analysis was performed using one‐way ANOVA (**p < 0.01,
   ***p < 0.001). f,g) Representative flow cytometry intensity histograms
   of CD25 expression (f) and quantification thereof on CD4^+ and CD8^+
   lymphocytes (g). Graphs show mean (line) and individual data points.
   Statistical analysis was performed using one‐way ANOVA (***p < 0.001;
   g). h,i) Representative flow cytometry intensity histograms of CD69
   expression (h) and quantification thereof on CD4^+ and CD8^+
   lymphocytes (i). Graphs show mean (line) and individual data points.
   Statistical analysis was performed using one‐way ANOVA (*p < 0.05,
   **p < 0.01, ***p < 0.001; i). j,k) Representative flow cytometry
   intensity histograms of CD134 (Ox40), CD278 (ICOS), and CD279 (PD1)
   expression (j) and quantification thereof on CD4^+ and CD8^+
   lymphocytes (k). Heat map shows mean. Statistical analysis was
   performed using one‐way ANOVA (*p < 0.05, ***p < 0.001; k). l)
   Intratumoral cytokines. Graph shows mean ± standard error of the mean
   (SEM). m) Volcano plots showing differentially expressed cytokines (p <
   0.05). Statistical analysis was performed using one‐way ANOVA. ns =
   nonsignificant. αPD1 = anti‐PD1. ROS = reactive oxygen species. MFI =
   mean fluorescence intensity. FDR = false discovery rate. FC = fold
   change.

2.4. Combinational Gas Plasma and Anti‐PD1 Checkpoint Therapy Alter
Intratumoral Immune‐Related Gene Expression Signatures

   Immune‐related gene expression signatures were analyzed in bulk RNA
   samples derived from excised tumors of melanoma‐bearing mice using the
   NanoString murine PanCancer IO 360° Panel. WPGMA‐weighted hierarchical
   clustering outlined dissimilar expression patterns among the different
   treatment groups (Figure [117] 6a). Differentially expressed genes were
   subjected to further analysis (Figure [118]6b). In the Venn diagram, a
   concordant downregulation compared to ICB or plasma monotreatments is
   indicated for 26 targets. By contrast, an overlapping upregulation is
   shown for 3 targets, HES1, SOX10, and MLANA (Figure [119]6c). Gene set
   enrichment analysis (GSEA) revealed a strong interconnection of
   regulated processes (Figure [120]6d). Compared to ICB monotreatment,
   top‐regulated categories included processes related to innate immune
   responses, antigen processing and presentation, and IFN signaling upon
   plasma–ICB combination. By contrast, alterations in categories related
   to response to oxygen compounds, inflammatory responses, TNFα
   signaling, chemotaxis, and migration of myeloid cells were found
   compared to gas plasma monotreatment (Figure [121]6e). Notably, process
   category priority was the same for the different comparisons. Last,
   cell type abundancies were calculated from bulk RNA expression data
   using the CIBERSORTx computational framework (Figure [122]6f).
   Comparison of relative abundancies revealed a significant increase in B
   cell signatures compared to plasma monotreatment in the tumors. Along
   similar lines, eosinophil signatures were increased compared to both
   monotherapeutic approaches. In line with previous flow cytometric
   results, CD4^+ and CD8^+ T cell signatures were increased upon combined
   ICB and monotreatment (Figure [123]6g). Differences in intratumoral
   immune landscapes were underlined by a calculated PCA (Figure [124]6h).

Figure 6.

   Figure 6
   [125]Open in a new tab

   Combinational gas plasma and checkpoint therapy alter intratumoral
   immune‐related gene expression signatures. a) WPGMA‐weighted
   hierarchical clustering of gene expression levels in bulk RNA samples
   retrieved from melanoma‐bearing mice. b) Volcano plot showing
   differentially (p < 0.05) up‐ (orange) and downregulated genes (blue).
   c) Venn diagram showing overlapping differentially expressed genes
   (DEG). d) Cytoscape enrichment network of DEG. e) Pathway enrichment
   dot plots of gene ontology (GO) terms. f) log[2] expression values of
   genes contributing to the GO term innate immune response. Graph shows
   mean ± standard error of the mean (SEM). g) Cell type abundancies were
   calculated from bulk RNA samples using the CIBERSORTx computational
   framework. Graph shows mean and individual data points. Statistical
   analysis was performed using one‐way analysis of variances (ANOVA)
   (*p < 0.05, **p < 0.01, ***p < 0.001). h) Principal component analysis
   (PCA) calculated from cell type abundances. ns = nonsignificant. αPD1 =
   anti‐PD1. FDR = false discovery rate. immune syst. = immune system.
   antigen proc./pres. = antigen procession and presentation. resp. O[2]
   comp. = response to oxygen‐containing compound. rel. = relative. PC =
   principal component.

3. Discussion

   Strategies to improve therapeutic activity and selectivity are a major
   goal in oncological drug development. Driven by technological and
   diagnostical innovations, recent advances have been made by exploiting
   cancer‐associated molecular and environmental signatures using targeted
   drugs. Yet, acquired drug resistance and genome instability of cancer
   cells remain a major challenge of targeted therapies. Considering the
   genetic heterogeneity of cancers, it becomes evident that a combination
   of several agents might be required to effectively eliminate these
   cells. Here, the implementation of ROS‐based therapy approaches has
   been suggested.^[ [126]^9 ^] We investigated the ability of a novel
   anticancer treatment modality, medical gas plasma, to potentiate
   responses in therapeutic strategies targeting melanoma.

   Starting from a library of 46 mitochondria‐targeted drugs, 11 drugs
   were identified showing strong synergistic effects upon combination
   with gas plasma‐mediated melanoma oxidation in two cell lines. Hereof,
   six drug combination partners were excluded due to superior toxicity in
   nonmalignant human keratinocytes, indicating an unfavorable
   risk‐benefit profile and a low therapeutic index. The remaining five
   combination regimes (A‐1210477, Carvedilol, Cozymasei, SBI‐0206965, and
   Navitoclax) were subjected to further analysis. Increased consumption
   of the antioxidant tripeptide glutathione indicated oxidative distress
   in melanoma cells, which was expected upon exposure to medical gas
   plasma as a redox‐based treatment modality but was not increased in
   drug monotreatment and combination regimes. Cell death modalities
   predominantly induced in combination regimes were evaluated using the
   pan‐caspase, hence apoptosis, inhibitor Z‐VAD‐FMK, and the RIP‐1 kinase
   inhibitor necrostatin‐1 (nec‐1) for inhibition of necroptosis. In the
   presence of Z‐VAD‐FMK but not nec‐1, a significant cell death
   inhibition was found in B16F10 and SK‐MEL‐28 melanoma cell lines,
   indicating apoptosis as the preferred mode of execution. Attributing
   ROS‐induced cell death to a certain modality is complicated not only by
   the heterogeneous and cell‐type‐specific death responses but further by
   the fact that exposure to exogenous (plasma‐derived) ROS can lead to
   endogenous ROS amplification, making it difficult to distinguish
   between responses to either. The notion that medical gas plasma
   triggers proapoptotic signaling in many cell types is emphasized by
   previous studies that observed an increase of proapoptotic Bax over
   Bcl‐2,^[ [127]^24 , [128]^25 ^] phosphatidylserine flipping,^[ [129]^26
   ^] and DNA fragmentation.^[ [130]^27 ^] Yet, other pathways, including
   ferroptosis,^[ [131]^28 , [132]^29 ^] lysosome‐dependent cell death,^[
   [133]^30 , [134]^31 ^] or autophagy,^[ [135]^32 ^] have been
   convincingly reported in the literature. In accordance with the
   Nomenclature Committee on Cell Death, apoptosis is defined as a form of
   regulated cell death initiated by perturbations of the intracellular or
   extracellular microenvironment, which is demarcated by mitochondrial
   outer membrane permeabilization (MOMP) and precipitated by executioner
   caspases, mainly caspase 3.^[ [136]^33 ^] Both characteristics, loss of
   mitochondrial membrane potential as a sign of MOMP paralleled by
   activation of two executioner caspases 3 and 7, were exhibited upon
   combined drug–plasma treatment. The superior toxicity of combined
   drug–plasma treatment was confirmed for all regimens in a syngeneic
   model of melanoma in vivo. Here, reduced tumor growth was accompanied
   by increased activation of effector caspase 3 and a reduced
   proliferative index. Cancer preventive and therapeutic effects have
   been prescribed for A‐1210477, Carvedilol, SBI‐0206965, and Navitoclax
   before. Hereof, a number of phase 1 and 2 clinical studies^[ [137]^34 ,
   [138]^35 , [139]^36 ^] have been enrolled in evaluating the use of
   Navitoclax, a targeted high‐affinity inhibitor of Bcl‐2, in the
   treatment of lymphoid malignancies based on previous studies indicating
   a broad activity against a panel of human tumor cell lines in vitro and
   durable tumor regression in vivo.^[ [140]^37 , [141]^38 , [142]^39 ^]
   In preclinical studies, targeting antiapoptotic Bcl‐2 family proteins
   has likewise succeeded in using A‐1210477 in treating BRAF‐mutant
   melanoma. Our study indicates that ROS‐induced proapoptotic signaling
   synergizes with inhibitors of antiapoptotic protein families,
   potentiating their effects. SBI‐0206965 is a highly selective ULK1
   kinase inhibitor that blocks autophagy.^[ [143]^40 ^] Elimination of
   damaged proteins, protein complexes, and organelles by autophagy is a
   central cellular mechanism and plays a crucial physiological role in
   response to endo‐ or exogenous stressors. However, in cancer, where
   cells experience ongoing metabolic stress, autophagy can promote tumor
   initiation, progression, and resistance to treatments, motivating its
   pharmacological inhibition.^[ [144]^41 ^] Hence, despite direct
   initiation of proapoptotic signaling partners, inhibition of cellular
   stress responses could potentiate ROS‐induced oxidative distress alike.
   Among the five selected compounds, carvedilol is the only clinically
   applied drug. Although FDA‐approved for treating cardiovascular
   diseases, recent evidence suggests that carvedilol use is associated
   with reduced cancer risk^[ [145]^42 ^] and has been implicated as a
   coadjuvant therapy.^[ [146]^43 ^] However, the mechanisms are not yet
   fully understood. Notwithstanding, additive or synergistic effects of
   combined drug–plasma treatment have been described before. For
   instance, plasma treatment has been shown to restore the
   chemosensitivity of TRAIL‐resistant colorectal cancer and
   chemoresistant multiple myeloma and glioblastoma.^[ [147]^44 , [148]^45
   , [149]^46 , [150]^47 ^] In melanoma, responses to doxorubicin,^[
   [151]^48 , [152]^49 ^] epirubicin, oxaliplatin, vorinostat,^[ [153]^50
   ^] and dacarbazine have been observed.^[ [154]^51 ^] Moreover, the
   combination with radiotherapy^[ [155]^52 ^] and nanoparticles have
   improved responses in vitro and in vivo.^[ [156]^53 , [157]^54 ,
   [158]^55 ^] However, this is the first study systematically
   investigating the synergistic effects of combined gas‐plasma treatment
   to identify eligible therapeutic targets for clinical application in
   dermato‐oncology.

   Notably, novel therapies have to be efficient and safe. To this end,
   our screening approach encompassed exclusion of treatment regimens
   showing high toxicity in HaCaT keratinocytes, indicating a low
   therapeutic index. The safety of gas plasma treatment has been
   convincingly demonstrated in previous studies. Following kINPen plasma
   exposure, anti‐ but not prometastatic effects were observed in vitro.^[
   [159]^46 ^] Genotoxic events were found to be absent using the
   WHO‐accredited hen's egg test for micronucleus induction model.^[
   [160]^56 ^] A 1‐year follow‐up in mice receiving repeated plasma
   treatment of skin wounds showed a lack of gas plasma‐induced
   preneoplastic or neoplastic lesions after a 1‐year follow‐up.^[
   [161]^57 ^] Similar observations have been made in a 5‐year follow‐up
   in a wound patient cohort lacking malignant transformation,
   inflammatory reactions, or pathological modifications after plasma
   treatment.^[ [162]^58 ^] Following repeated plasma exposure of the oral
   mucosae over 12 months, histopathological analysis of 406 animals did
   not indicate increased formation of noninvasive lesions or squamous
   cell carcinomas.^[ [163]^59 ^]

   Despite direct tumor toxic effects, our findings indicated
   immune‐modulatory aspects of combined drug–plasma treatment as seen by
   increased levels of tumor‐infiltrating leucocytes. Immune‐targeted
   therapies have been a major breakthrough in many malignancies in recent
   years.^[ [164]^60 , [165]^61 ^] While the majority of conventional
   chemotherapeutics elicit cytotoxicity by interfering with proteins
   affecting DNA synthesis and replication, immunotherapies aim to disrupt
   local and systemic immune perturbations occurring during malignant
   transformation to spark a self‐sustaining cycle of cancer immunity.
   However, despite valuable excitement dedicated to advances in antibody
   and cell‐based therapy approaches, immunotherapy still remains
   ineffective for the majority of patients with cancer.^[ [166]^3 ,
   [167]^62 ^] Efficient priming of an antitumor immune response is often
   impaired by primary and secondary resistance mechanisms causing
   insufficient initiation of the cancer immunity cycle involving (a)
   antigen presentation and T cell activation, (b) infiltration of T cells
   in the tumor microenvironment, and (c) efficient T cell killing
   activity.^[ [168]^63 ^] Checkpoint inhibitors prove their advantages in
   “hot” tumors, given that the immune system has successfully passed the
   finely tuned process of antigen‐cross presentation and critical T cell
   priming in lymphoid tissues but will fail their effect without previous
   generation of antigen‐specific effector T cells. The limitations of ICB
   have pushed research in immuno‐oncology toward combinational treatments
   lately. Strikingly, shortly after the succession of nivolumab
   (anti‐PD1)–ipilimumab (anti‐CTLA4) as the first ICB combination
   treatment in advanced melanoma,^[ [169]^64 ^] pairing of pembrolizumab
   (anti‐PD1) and platinum‐based chemotherapy in nonsmall lung cancer was
   granted FDA‐approval in 2017.^[ [170]^65 ^] By eliciting an
   inflammatory response to dying cancer cells, a so‐called ICD,
   coadministered chemotherapeutics turned tumors from “cold” to “hot”,
   synergizing with ICB in unleashing a potent antitumor immune response.
   Furthermore, several studies demonstrated increased efficacy of
   combined radiotherapy and ICB, synergizing via activation of
   inflammatory pathways, induction of ICD, release of antigens from
   irradiated cells, and increased tumor sensitization to T cell
   responses.^[ [171]^66 ^] Experimental strategies for the detection of
   ICD encompass providing evidence of (a) expression and release of DAMPs
   on dying cancer cells followed by (b) successful activation of
   antigen‐presenting cells in vitro and (c) dying cancer cells to
   initiate adaptive immunity in a syngeneic, immunocompetent host in
   vivo.^[ [172]^67 ^] In preclinical plasma research, a range of
   surrogate ICD‐related DAMPs, including ecto‐expression of calreticulin
   and heat shock proteins,^[ [173]^68 , [174]^69 ^] release of ATP^[
   [175]^70 ^] and nuclear translocation of HMGB‐1^[ [176]^71 ^] has been
   convincingly shown upon oxidation in many different cell lines.
   Likewise, chemokine detection in culture supernatants of
   plasma‐oxidized tumor cells revealed an increased release of several
   proimmunogenic and chemotactic agents.^[ [177]^72 ^] Functional
   analysis of APCs after challenging with plasma‐oxidized tumor cells
   increased phagocyte activation,^[ [178]^73 ^] followed by enhanced
   tumor material uptake^[ [179]^74 ^] as seen in our study and
   interaction with T cells.^[ [180]^69 ^] Moreover, a recent screening on
   a protein library pointed out to an inherent DAMP activity of
   oxidatively modified proteins generated by plasma treatment^[ [181]^75
   , [182]^76 ^] comparable to advanced glycosylation end‐products on
   naturally occurring carbohydrates.^[ [183]^77 ^] Following successful
   cross‐presentation and T cell priming, an inhospitable TME may preclude
   proper function of the expanded T cell repertoire. This might include
   the aberrant expression of immune checkpoints or coinhibitory
   receptors, immunosuppressive cytokines, and immune‐inhibitory
   metabolites.^[ [184]^78 ^] Strikingly, evaluation of checkpoint
   expression signatures in vitro and in vivo in our study revealed an
   overall downregulation of investigated markers after plasma treatment.

   In vivo, the ability of dying cancer cells to drive adaptive immunity
   depends both on their adjuvanticity and antigenicity. While the former
   is an inherent trait of most ICD inducers, stimulating the coordinated
   release and expression of danger signals necessary for the recruitment
   of APCs, this does not account for tumor cell antigenicity. Malignant
   transformation and cancer progression are frequently accompanied by a
   high mutation rate in cancer cells, culminating in MHC I‐mediated
   expression of tumor neoantigens. Representing a poor structural
   homology of physiologically occurring epitopes, neoantigens fail to
   induce clonal deletion in the context of central tolerance and can be
   recognized by T cells.^[ [185]^79 , [186]^80 ^] However, genetic and
   epigenetic defects, including antigen loss and subclonal evolution
   combined with impaired antigen presentation, often comprise adaptive
   immune responses.^[ [187]^81 ^] Therapeutic regimens that boost tumor
   antigenicity are hence considered to potentiate the ability of dying
   cancer cells to initiate an adaptive immune response.^[ [188]^82 ,
   [189]^83 ^] Using gas plasma technology, post‐translational, oxidative
   modification of the model protein ovalbumin (oxOva) enhanced T cell
   activation compared to native ovalbumin (Ova) in genetically engineered
   mice recently. Moreover, oxOva vaccination of wild‐type mice provided
   enhanced protection from B16F10‐Ova melanoma growth, indicating a
   pronounced antitumor immune response in terms of an amplified and/or
   broadened T cell response.^[ [190]^22 ^] From those findings, it is
   hypothesized that plasma‐derived ROS create an inflammatory‐like
   environment able to trigger antitumor immune responses not only by
   increasing tumor cell adjuvanticity but also their antigenicity by
   formation of neoantigens via oxidation of biomolecules. Yet, the exact
   determinants are not fully understood. Concomitant with enhanced immune
   cell infiltration and leucocyte activation in melanoma‐bearing mice, we
   recently provided evidence of local plasma treatment to induce
   ICD‐mediated adaptive immunity and protect mice from tumor growth after
   preventive vaccination.^[ [191]^84 ^] Evidence of medical gas plasma
   technology to drive adaptive immunity in vivo as the ultimate proof of
   ICD has further been supported by reports on abscopal effects after
   plasma‐mediated tumor oxidation in murine tumor models recently.^[
   [192]^85 , [193]^86 , [194]^87 ^] In this light, we hypothesized that
   combined gas plasma treatment could potentiate responses to anti‐PD1
   ICB, which was confirmed in a syngeneic model of melanoma in vivo.
   Interestingly, flow cytometric analysis of the intratumoral immune
   compartment revealed a major influence of plasma treatment on the
   innate immune compartment, while effects on the lymphoid immune
   compartment were dominated upon checkpoint administration. This was
   emphasized by evaluation of gene expression profiles in bulk RNA tumor
   samples based on the NanoString PanCancer IO 360 Panel. Compared to ICB
   monotreatment, GSEA indicated alterations in innate immune responses,
   antigen presentation and processing, and IFN response. In line with
   previous flow cytometric results, cell type abundancies characterized
   from bulk RNA samples using the CIBERSORT computational framework
   indicated an increase in CD4^+ and CD8^+ T cell signatures in animals
   receiving checkpoint administration. An additional increase in
   expression signatures associated with B cell infiltration was observed
   upon combined plasma–ICB compared to plasma monotreatment. Along
   similar lines, eosinophil signatures were increased compared to both
   monotherapeutic approaches. Our data indicate that local tumor
   oxidation using gas plasma technology primarily acted on APCs which may
   facilitate the generation of antigen‐specific T cell responses.
   Combining medical gas plasma treatment with anti‐PD1 checkpoint therapy
   suppressed tumor growth more effectively than either agent alone. We
   hypothesize that by promoting the successful initiation of the cancer
   immunity cycle, as discussed above, plasma treatment provides the
   fundamental basis for efficient ICB, unleashing efficient T cell
   killing activity only after the previous generation of a pool of
   antigen‐specific effector T cells. Moreover, we found plasma treatment
   to reduce the aberrant expression of coinhibitory surface markers,
   supporting the proper function of the expanded T cell repertoire.
   Clinically, the dual application of gas plasma and ICB in cancer
   therapies could broaden the range of patients profiting from ICB. As a
   major advantage, plasma treatment is an easy‐to‐implement,
   cost‐effective technology with good tolerability, already applied in
   many dermatological centers in Europe. However, it should be mentioned
   that the treatment is limited to superficially growing skin cancers for
   now, as eligible plasma sources for treating internal neoplasms have
   not yet been developed.

   In summary, this study systematically demonstrates the ability of gas
   plasma‐mediated tumor oxidation to potentiate drug responses in the
   treatment of melanoma. From our findings, it can be concluded that
   agents targeting stress responses (e.g., autophagy inhibitors),
   antiapoptotic proteins (e.g., bcl‐2), or immune‐targeted therapies may
   particularly profit from such an approach.

4. Conclusion

   Implementation of ROS‐based therapy approaches in therapeutical
   combination regimes has been suggested to potentiate therapeutic
   responses in oncology. We hypothesized that combined gas
   plasma‐mediated tumor oxidation could potentiate drug responses in the
   treatment of melanoma and aimed to identify eligible agents in a
   systematic screening approach. Starting from a library of 46
   mitochondria‐targeted drugs, five drugs were identified to synergize
   with plasma treatment in vitro and in vivo. The systematic screening
   indicated that plasma treatment might potentiate responses to
   inhibitors of antiapoptotic protein families and inhibitors of cellular
   stress responses, outlining eligible target mechanisms to identify
   further combination partners in future studies. In addition,
   immunomodulatory aspects of the treatment were found in vivo,
   motivating us to investigate if combined gas plasma treatment could
   further potentiate responses to anti‐PD1 ICB. Strikingly, combining gas
   plasma treatment with anti‐PD1 ICB suppressed tumor growth more
   effectively than either agent alone. Evaluation of underlying
   mechanisms indicated that local tumor oxidation using gas plasma
   technology primarily acted on APCs. We hypothesize that plasma
   treatment provides the fundamental basis for efficient ICB by promoting
   the successful initiation of the cancer immunity cycle. Collectively,
   our results indicate that gas plasma treatment could be a powerful tool
   to sensitize cancer cells to a broad range of oncological treatment
   approaches in the future.

5. Experimental Section

Cell Culture

   Human SK‐MEL‐28 (CLS: 300337; CLS cell line service, Germany),
   MaMel‐86a (RRID: CVCL_A221), A375 (ATCC: CRL‐1619; ATCC, Germany), and
   murine B16F10 (ATCC: CRL‐6475; ATCC, Germany) melanoma cells were
   cultured in Dulbecco's modified Eagle or Roswell Park Memorial
   Institute (both from PanBiotech, Germany) medium supplemented with 10%
   fetal bovine serum, 1% l‐glutamine, and 1% streptomycin (all Corning,
   Germany) according to the supplier's instructions. Human HaCaT (CLS:
   300493; CLS cell line service, Germany) keratinocytes were cultured in
   Roswell Park Memorial Institute medium supplemented with 10% fetal
   bovine serum, 1% l‐glutamine, and 1% streptomycin. Cells were kept
   under standard culture conditions at 37 °C, 5% CO[2], and 95% humidity
   in a specified cell culture incubator (CB212; Binder, Germany).

Medical Gas Plasma Jet Technology

   Gas plasma treatment in this study was performed using an atmospheric
   pressure plasma jet (kINPen; neoplas, Germany). The device is
   technically similar to the kINPen Med, which has been accredited as
   medical class IIa/IIb in Europe since 2013 and frequently applied in
   dermatology.^[ [195]^88 ^] In standard mode, the jet is operated using
   a flow of argon gas (purity 99.9999%; Air Liquide, France) at three
   standard liters per minute (slm). The noble gas was excited at the
   electrode within the head of the plasma jet at a frequency of 1 MHz at
   a generated power of ≈1 W. The jet was installed on an xyz motorized
   table (CNC step) controlled via software to maximize the
   reproducibility of the plasma treatment.

Intracellular ROS/RNS

   Cells were stained with 2 µm 2′,7′‐dichlorodihydrofluorescein diacetate
   (DCF‐DA; Thermo Fisher, Germany) for 30 min at 37 °C prior to plasma
   treatment. After washing, cells were exposed to plasma and 1 µm
   4′,6‐Diamidin‐2‐phenylindol (DAPI; Sigma‐Aldrich, Germany) was added
   immediately to stain nuclei of terminally dead cells. Images were
   acquired in brightfield and fluorescence channels at λ [ex] 365 nm and
   λ [em] 465 nm for DAPI and λ [ex] 475 nm and λ [em] 525 nm for DCF
   using High Content Imaging (Operetta CLS; PerkinElmer, Germany) and a
   20× air objective (NA = 0.4; Zeiss, Germany).

In Vitro Screening Study

   A library of 46 mitochondria‐targeted compounds (Cat# L5300; TargetMol,
   USA; Table [196]S1, Supporting Information) was screened for
   combination effects with medical gas plasma technology. For in vitro
   experiments, 5 × 10^3 cells were seeded in 96‐well flat bottom plates
   (Thermo Fisher, Germany) 24 h prior to experiments. The cell culture
   medium was refreshed on the day of experiments. A three‐step screening
   strategy (Figure [197]1a) was employed to identify promising drug
   candidates. First, cells were exposed to 0.01, 0.1, 1, and 10 µm of
   each compound to generate dose–response curves based on cellular
   metabolic activity. Cells exposed to plasma only received dimethyl
   sulfoxide (DMSO) instead of drug treatment. Gas plasma treatment was
   done at 2 slm for 30 s 16 h after exposure to the respective compounds
   or DMSO in plasma monotreatment and combination groups, respectively.
   Drugs showing significant combination effects were selected. The
   coefficient of drug interaction (CDI) was calculated to evaluate if the
   observed effect was additive or synergistic based on the formula: CDI =
   AB/(A × B). Second, IC50 values were calculated for drug monotreatment
   and combination groups. Drugs showing a fivefold (FC > 5) decreased
   IC50 value in combination regimes were tested on nonmalignant HaCaT
   keratinocytes in a similar fashion to determine the selectivity of the
   approach. Functional analysis was performed on drugs showing a twofold
   (FC > 2) decreased IC50 in SK‐MEL‐28 human melanoma cells compared to
   nonmalignant HaCaT keratinocytes.

Metabolic Activity

   The Almar Blue assay was used to assess the metabolic activity of
   cells. Briefly, 100 µm 7‐hydroxy‐3H‐phenoxazin‐3‐on‐10‐oxid (resazurin;
   Alfa Aesar, USA) was added to the cells 24 h after treatment, followed
   by 3 h of incubation under standard culture conditions. Viable cells
   metabolize nonfluorescent resazurin to fluorescent resorufin in an
   NADPH/H^+‐dependent reaction. Resorufin fluorescence was measured at λ
   [ex] 535nm and λ [em] 590 nm using a multimode plate reader (F200;
   Tecan, Switzerland).

Cellular Viability

   Cellular viability was assessed using high‐content imaging 6 and 24 h
   after exposure. Briefly, cells were stained with sytox green (SG) for
   30 min at 37 °C and cell culture media was exchanged immediately after
   to reduce autofluorescence. Images were acquired in brightfield and
   fluorescence channels at λ [ex] 490 nm and λ [em] 520 nm using a 20×
   air objective (NA = 0.4; Zeiss, Germany). Algorithm‐driven image
   analysis was performed using Harmony 4.9 image analysis software
   (PerkinElmer; Germany). The analysis sequence included cellular
   segmentation based on digital phase contrast (DPC), and cellular
   viability was assessed based on SG fluorescence.

Intracellular Glutathione

   Intracellular levels of reduced GSH as a surrogate marker for cellular
   redox capacities were determined using High Content Imaging (Operetta
   CLS; PerkinElmer, Germany). Briefly, cells were stained with 5 µm
   GSH‐tracer (Tocris, UK) for 90 min at 37 °C 6 h after treatment. The
   excitation/emission maxima of the cell‐permeable, ratiometric probe
   shift from λ [ex]/λ [em] 520/580 nm to λ [ex]/λ [em] 430/510 nm upon
   binding to reduced GSH. The F[510]/F[580] ratio correlates with the
   intracellular GSH concentration. After washing, images were acquired in
   brightfield, DPC (pseudocytosolic signal) and fluorescence channels for
   bound (λ [ex] 390–420 nm and λ [em] 500–550 nm) and unbound (λ [ex]
   460–790 nm and λ [em] 570–650 nm) using a 20× air objective (NA = 0.4;
   Zeiss Germany). The experimental setup and the software‐based
   quantification algorithms were done using Harmony 4.9 image analysis
   software (PerkinElmer, Germany). Cells were detected via DPC signal for
   image segmentation, and the mean fluorescence intensity (MFI) of bound
   and unbound GSH tracer was quantified. Intracellular levels of reduced
   GSH were calculated according to the formula: GSH level = (MFI [bound
   tracer])/(MFI [unbound tracer]).

Mitochondrial Dysfunction

   Mitochondrial dysfunction was evaluated 24 h after treatment using High
   Content Imaging (Operetta CLS; PerkinElmer, Germany). Briefly, cells
   were loaded with 100 nm tetramethylrhodamine ethyl ester (TMRE) for
   30 min at 37 °C. TMRE is a cell‐permeable, cationic fluorescent dye
   readily sequestered by active mitochondria. After washing, images were
   acquired in brightfield, DPC and fluorescence channels at λ [ex] 550 nm
   and λ [em] 580 nm using a 20× air objective (NA = 0.4; Zeiss, Germany).
   Algorithm‐driven image analysis was performed using Harmony 4.9 image
   analysis software (PerkinElmer, Germany). Cells were detected via DPC
   signal for image segmentation, and TMRE mean fluorescence intensity was
   quantified.

Cell Death Inhibition and Activation of Effector Caspases

   Cell death modalities were investigated using respective inhibitors.
   For inhibition of apoptosis, cells were loaded with 50 µm of the
   cell‐permeable, pan‐caspase inhibitor Z‐VAD‐FMK 1 h prior to plasma
   treatment. Likewise, Necrostatin‐1 (nec‐1) was used for necroptosis
   inhibition at a final concentration of 40 µm. Cellular viability was
   evaluated using High Content Imaging (Operetta CLS; PerkinElmer,
   Germany) as described above. Therefore, cells were stained with sytox
   orange for 30 min at 37 °C and cell culture media was exchanged
   immediately after to reduce autofluorescence. Initiation of apoptosis
   was further evaluated based on activation of the effector caspases 3
   and 7. Briefly, cells were stained with Cell Event caspase 3/7 reagent
   (Thermo Fisher, Germany) at a concentration of 400 nm for 15 min at
   37 °C 24 h after treatment. After washing, cells were acquired using
   flow cytometry (CytoFLEX S; Beckman‐Coulter, Germany). Analysis was
   done using Kaluza 2.2 software (Beckman‐Coulter, Germany).

5.1. 3D Melanoma Spheroids

   Human SK‐MEL‐28 cells were seeded at a density of 1 × 10^4 cells per
   well in 96‐well ultralow attachment plates (PerkinElmer, Germany) and
   centrifuged at 1000 × g for 10 min. Consistent 3D melanoma spheroids
   formed after 48 h followed by drug, plasma, or combined treatment as
   described before. Spheroid cytotoxicity was evaluated using High
   Content Imaging (Operetta CLS; PerkinElmer, Germany) 24 h after
   exposure. Briefly, spheroids were stained with 5 µm SG for
   discrimination of terminally dead cells and 40 µm Hoechst (both Thermo
   Scientific, Germany), for nuclear staining, for 1 h at 37 °C.
   Immediately after, images were acquired in brightfield and fluorescence
   channels at λ [ex] 490 nm and λ [em] 520 nm for SG and λ [ex] 365 nm
   and λ [em] 465 nm for Hoechst from a z‐stack of 30 images at a 10 µm
   distance using a 5× air objective (NA = 0.16; Zeiss, Germany).
   Algorithm‐driven image analysis was performed using Harmony 4.9 image
   analysis software (PerkinElmer, Germany).

In Vivo Antimelanoma Drug Combination Therapy

   The combinational efficacy of promising drug candidates identified in
   vitro drug screenings was further investigated in a syngeneic model of
   melanoma in vivo. The ethical implications for the experiment were
   reviewed by the local ethical authority Comissão de Ética no Uso de
   Animais (CEUA/UEL) of the University of Londrina, Brazil (approval
   number: 3185.2019.53). Wildtype C57BL/6 mice were shaved on the left
   flank and subcutaneously inoculated with 1 × 10^6 B16F10 syngeneic
   murine melanoma cells in PBS. Animal randomization was done once the
   tumor volume reached ≈50 mm^3. Animals were divided into ten groups (7
   animals per group), including one vehicle group (DMSO), one group
   exposed to gas plasma (3 slm of argon, 5 min), four groups receiving
   intraperitoneal injection of respective drugs (A‐1210477, Carvedilol,
   SBI‐0206965 or Navitoclax; all at 10 mg kg^−1), and four groups
   receiving combined drug–plasma treatment. In combination groups,
   intraperitoneal drug injection was done 1 h prior to gas plasma
   treatment. Animals received treatments in four treatment cycles every
   second day. Animals were euthanized 24 h after the last treatment
   cycle. Tumors were carefully excised, and tumor weight was assessed
   using a precision balance. Excised tumors were embedded in optimal
   cutting temperature (Thermo Scientific, USA) compound and frozen in
   liquid nitrogen before storage at −80 °C.

Immunohistochemistry

   For immunohistochemistry, 5 µm sections were mounted on silanized
   slides, deparaffinized, rehydrated, immersed in 15 mmol L^−1 citrate
   buffer (pH 6.0), and subjected to head‐induced epitope retrieval using
   a vapor lock for 40 min. The slides were rinsed with PBS, and
   peroxidase blockade was done using methanol/H[2]O[2]. Nonspecific
   protein binding was blocked with 3% bovine serum albumin. The sections
   were incubated overnight with primary antibodies targeting Ki‐67 (Santa
   Cruz, USA), cleaved caspase 3 (Cell Signaling, Germany), and CD45
   (Santa Cruz, USA). After incubation, peroxidase‐conjugated secondary
   anti‐rabbit IgG antibodies (Santa Cruz, USA) were added for 1 h after
   washing. For chromogenic detection, slides were incubated with
   3,3′‐diaminobenzidine for 5 min after washing and counterstained with
   hematoxylin. Negative controls were done by omitting the primary
   antibody. Images were acquired (Observer Z.1; Carl Zeiss, Germany), and
   quantification was done using ImageJ software.

Flow Cytometry Immune Checkpoint Analysis

   For in vitro immune checkpoint expression analysis, MaMel‐86a human
   melanoma cells were gas plasma exposed, cultured for 24 h, collected
   using accutase, and stained with several fluorescently labeled
   antibodies targeting immuno‐stimulating and suppressing receptors and
   ligands. After washing, cells were acquired using flow cytometry
   (CytoFLEX S; Beckman‐Coulter, Germany). Checkpoint expression
   signatures were further investigated on B16F10 tumors after in vivo
   exposure to plasma, anti‐PD1 ICB and combined plasma–ICB. Single‐cells
   suspensions retrieved from excised tumors were stained with antibodies
   targeting (conjugate) MHCI (BUV661), CD47 (PerCp‐Cy5.5), CD80/CD86
   (APC), and CD278 (PE‐Cy7). DAPI (Sigma‐Aldrich, Germany) and iFluor
   maleimide 860 (AAT Bioquest, USA) were added for live–dead
   discrimination. After washing, cells were acquired using flow cytometry
   (CytoFLEX LX; Beckman‐Coulter, Germany), analysis was performed using
   Kaluza 2.2 software (Beckman‐Coulter, Germany).

Peripheral Blood Mononuclear Cell (PBMC) Isolation and Generation of
Monocyte‐Derived Dendritic Cells (DCs)

   Isolation of PBMCs in this study was approved by the local ethics
   committee at the University Medical Center (approval number: BB
   014‐14). PBMCs were isolated from buffy coats of healthy donors
   provided by the Institute of Transfusion Medicine Greifswald by
   density‐gradient centrifugation using lymphocyte separation medium
   (VWR, Germany). Residual erythrocytes were lysed using a red blood cell
   lysis buffer (BioLegend, Germany). For monocyte isolation, CD14
   microbeads (Miltenyi Biotec, Germany) were used according to the
   manufacturer's protocol. Flow cytometric (Attune NxT; Applied
   Biosystems, USA) verification of purity was performed immediately after
   isolation and was always greater than 85%. For differentiation of
   monocytes into DCs, 1 × 10^5 monocytes per well were seeded in 24‐well
   plates (Sarstedt, Germany). The cell culture medium was supplemented
   with 800 IU granulocyte‐macrophage colony‐stimulating factor (GM‐CSF)
   and 500 IU interleukin (IL) 4 (all PeproTech, Germany) to initiate
   monocyte‐derived dendritic cells (DCs). After two days, cells were
   restimulated by adding the same amount of GM‐CSF and IL‐4 to the cell
   culture medium. Immature DCs were used for coculture experiments on day
   5.

Phagocytosis of Oxidized Melanoma Cells

   Human A375 melanoma cells were labeled with 1 µm CTV (Thermo Fisher,
   Germany) and seeded at a density of 1 × 10^5 cells per well in 24‐well
   plates (Sarstedt, Germany). Immediately after, cells were exposed to
   plasma for 30 s. Oxidized melanoma cells were collected and added to
   immature DCs prepared from PBMCs, as described before. After incubation
   for 24 h, cocultured cells were harvested and stained with monoclonal
   antibodies (conjugate) targeting human CD11c (phycoerythrin‐cyanine 7;
   BioLegend, The Netherlands). For live–dead discrimination, iFluor
   maleimide 860 (AAT Bioquest, USA) was added. After washing, cells were
   acquired using flow cytometry (CytoFLEX LX; Beckman‐Coulter, Germany),
   analysis was performed using Kaluza 2.2 analysis software
   (Beckman‐Coulter, Germany). Tumor cells were identified by CTV
   staining, while DCs were distinguished based on CD11c expression.

In Vivo Antimelanoma ICB and Plasma Treatment

   The ethical implications for the experiment were reviewed by the local
   ethical authority Landesamt für Landwirtschaft, Lebensmittelsicherheit
   und Fischerei (LALLF) Mecklenburg Vorpommern (approval number: M‐V
   7221.3‐1‐023/17). Wildtype C57BL/6 mice were shaved on the flank and
   subcutaneously inoculated with 5 × 10^4 B16F10‐luc (PerkinElmer:
   [198]BW124734; PerkinElmer, Germany) syngeneic murine melanoma cells in
   100 µL of matrigel (Corning, Germany). First, treatment efficacies of
   monotreatments were confirmed. Animals received three treatment cycles
   of either (a) intraperitoneal injection of anti‐PD1 antibodies (clone:
   29F.1A12; BioLegend, Germany), (b) 5 min local gas plasma treatment
   (kINPen Med; neoplas, Germany; 3 slm argon) or (c) were left untreated
   (control). Tumor growth was monitored using bioluminescence imaging
   (IVIS; PerkinElmer, Germany) over a time period of 20 min after
   intraperitoneal injection of 100 µL 30 mg mL^−1 d‐luciferin
   (PerkinElmer, Germany). Animals were sacrificed on day 14 after tumor
   inoculation. Next, the combination regime was compared to monotreatment
   of anti‐PD1 and local gas plasma treatment. Animals received three
   treatment cycles of either (a) anti‐PD1 (clone: 29F.1A12), (b) 5 min
   local gas plasma treatment (kINPen Med; 3 slm argon) or (c) the
   combination of both. Again, tumor growth was monitored using
   bioluminescence imaging as described before. Single‐cell suspensions of
   excised tumors were retrieved using the GentleMacs tumor dissociation
   kit mouse and the OctaMacs device (both Miltenyi, Germany). Single‐cell
   suspensions were subjected to further analysis by flow cytometry to
   evaluate tumor‐infiltrating leucocytes. In addition, supernatants were
   collected after tumor dissociation and stored at −80 °C for
   intratumoral chemokine and cytokine analysis.

Tumor‐Infiltrating Leucocytes

   Abundancies and expression signatures of tumor‐infiltrating leucocytes
   were assessed using flow cytometry. Single‐cell suspensions retrieved
   from excised tumors were stained with two core panels. Tumor
   infiltrating CD4^+ and CD8^+ T cells were analyzed using antibodies
   targeting (conjugate) B220/45R, IA/IE, Ly6G, CD11c, F4/80 (DUMP; all
   APC‐Cy7), CD134 (APC), CD4 (AF700), CD62L (PE/Dazzle594), CD278 (PC7),
   CD279 (PE), CD45 (PerCP‐Cy5.5), CD3 (BV421), CD8 (BV510), CD69 (BV650),
   CD25 (BV785), and MHCI (BUV661). For live–dead discrimination, iFluor
   maleimide 860 (AAT Bioquest, USA) was added. After washing, cells were
   acquired using flow cytometry (CytoFLEX LX; Beckman‐Coulter, Germany),
   analysis was performed using Kaluza 2.2 software (Beckman‐Coulter,
   Germany). In a second core panel, the number of tumor‐infiltrating
   antigen‐presenting cells was assessed using monoclonal antibodies
   targeting (conjugate) CD206 (BV421), Ly6G (BV510), CD11b (BV605), IA/IE
   (PerCP‐Cy5.5), CD80/CD86 (PE), Ly6C (PE/Dazzle594), F4/80 (PE‐Cy7),
   Cd274 (APC), CD11c (AF700), and CD45 (APC‐Cy7). For live–dead
   discrimination, SG (Thermo Fisher, Germany) was added. After washing,
   cells were acquired using flow cytometry (CytoFLEX S and LX;
   Beckman‐Coulter, Germany).

Chemokine and Cytokine Analysis

   Intratumoral chemokines and cytokines were analyzed using a bead‐based
   sandwich multianalyte assay (LegendPLEX; BioLegend, The Netherlands)
   according to the manufacturer's instructions. The assay panel contained
   beads targeted against IFNγ, IL2, IL4, IL5, IL6, IL9, IL10, IL13,
   IL17A, IL17F, IL22, and TNFα. Beads were labeled with fluorescent
   detection antibodies, and samples were acquired using flow cytometry
   (CytoFLEX LX; Beckman‐Coulter, Germany). Absolute concentrations of
   respective analytes were calculated against a standard curve using the
   LEGENDplex data analysis software (Vigene Tech, USA).

RNA Isolation and NanoString Gene Expression Analysis

   RNA was isolated from excised tumors after generation of single‐cell
   suspensions using an RNA isolation kit (Bio & Sell, Germany) according
   to the supplier's protocol. The RNA concentration of each sample was
   determined using fluorescent RNA labeling (Quant‐it RiboGreen RNA
   Reagent; Thermo Fisher, Germany) and a fluorometer (DS‐11 FX; DeNovix,
   Germany). Absolute concentrations were calculated against a standard
   curve. Subsequent gene expression analysis was performed using the
   NanoString nCounter technology, which directly evaluates RNA levels
   without prior complementary DNA synthesis. Briefly, 50 ng RNA per
   sample was hybridized for 24 h at 65 °C with reporter and capture
   ProbeSets of the nCounter PanCancer IO 360 Panel targeting 770 genes
   specifically designed for research in immuno‐oncology. Immediately
   after, samples were loaded into the nCounter SPRINT cartridge and
   processed by the NanoString nCounter system. Differential gene
   expression analysis was performed using the nSolver 4.0 analysis
   software (all NanoString Technologies, USA). Pathway enrichment
   analysis was evaluated using GSEA Software^[ [199]^89 , [200]^90 ^] and
   processed with Cytoscape.^[ [201]^91 ^] Cell fractions were quantified
   from bulk tissue gene expression profiles using CIBERSORTx.^[ [202]^92
   ^]

Statistical Analysis

   Graphing and statistical analysis were done using prism 9.5.1 (GraphPad
   Software, USA). Statistical analysis was performed using t‐test or
   analysis of variances (ANOVA) where appropriate. Data show mean ±
   standard error of the mean if not indicated otherwise in the figure
   legends.

Conflict of Interest

   The authors declare no conflict of interest.

Supporting information

   Supporting Information
   [203]Click here for additional data file.^ (363.8KB, pdf)

Acknowledgements