Physical plasma-treated saline promotes an immunogenic phenotype in CT26 colon cancer cells in vitro and in vivo

Metastatic colorectal cancer is the fourth most common cause of cancer death. Current options in palliation such as hyperthermic intraperitoneal chemotherapy (HIPEC) present severe side effects. Recent research efforts suggested the therapeutic use of oxidant-enriched liquid using cold physical plasma. To investigate a clinically accepted treatment regimen, we assessed the antitumor capacity of plasma-treated saline solution. In response to such liquid, CT26 murine colon cancer cells were readily oxidized and showed cell growth with subsequent apoptosis, cell cycle arrest, and upregulation of immunogenic cell death (ICD) markers in vitro. This was accompanied by marked morphological changes with re-arrangement of actin fibers and reduced motility. Induction of an epithelial-to-mesenchymal transition phenotype was not observed. Key results were confirmed in MC38 colon and PDA6606 pancreatic cancer cells. Compared to plasma-treated saline, hydrogen peroxide was inferiorly toxic in 3D tumor spheroids but of similar efficacy in 2D models. In vivo, plasma-treated saline decreased tumor burden in Balb/C mice. This was concomitant with elevated numbers of intratumoral macrophages and increased T cell activation following incubation with CT26 cells ex vivo. Being a potential adjuvant for HIPEC therapy, our results suggest oxidizing saline solutions to inactivate colon cancer cells while potentially stimulating antitumor immune responses.

Despite continuous advances in medicine, therapy and palliation of peritoneal colon cancer metastasis remains challenging 1 . This owes to the high number of sometimes difficult to visualize tumor nodes in the peritoneum that hinder surgical resection 2 . Early therapeutic options are limited because peritoneal metastasis often occurs before primary diagnosis of colon cancer 3 . As consequence, rapid tumor spread and infiltration into multiple organs are main factors constituting to the bad prognosis in many patients 4 . Current therapeutic choices for peritoneal metastasis from colorectal cancer involve classical chemotherapy, irradiation, cytoreductive surgery, and hyperthermic intraperitoneal chemotherapy (HIPEC) 5 . Yet, many cancers acquire chemoresistance and numerous patients ultimately do not benefit from therapy 6 . If successful, therapeutic profit often comes with substantial side effects 7 . This clinical situation calls for new roads for (adjuvant and palliative) therapeutic targeting of metastatic colorectal cancer. Among others, one route is sensitizing colorectal cancer via modulation of the antioxidant defense mechanisms important for handling reactive oxygen and nitrogen species (ROS/RNS) [8][9][10] .
A recent innovation in prospective cancer therapy is cold physical plasma, a partially ionized gas that produces a plethora of ROS/RNS [11][12][13] . Previous studies have provided evidence for tumor-toxic effects of cold plasma in colorectal cancers in vitro [14][15][16] . The proof-of-concept was further validated in a non-orthotopic murine model of human colon cancer in vivo, and effects were delineated to generation of reactive species 17 . The utilization of liquid previously exposed to plasmas surprisingly paralleled many of the antitumor effects seen with direct plasma treatment [18][19][20] . As carrier liquid, cell culture medium is often used to target tumor cells [21][22][23] . This approach has also successfully decreased tumor-burden in mice with widespread peritoneal pancreatic or gastric cancer metastasis 24,25 . While these studies are promising, the multi-component cell culture medium is unlikely to receive accreditation for use in clinics due to its complex formulation and unknown immunogenicity. By contrast, saline solutions are clinically accepted and utilized worldwide. Therefore, the aim of this study was to investigate the tumor-toxic and pro-immunogenic properties of physical plasma-treated saline on murine colorectal cancer cells in vitro. During microevolution in hosts, various cancer cell types with high turnover rates develop mechanisms to escape immunosurveillance [26][27][28][29] . One approach of cancer therapy is to utilize the immunogenic cancer cell death (ICD) triggering the immune system as defender against tumors. In this concept, antigen-presenting cells phagocytosing tumor material become activated through sensing of tumor cell surface marker or the release of soluble factors that act as damage-associated molecular patterns (DAMPs) [30][31][32][33] . This may generate an immune-response directed against these cancer cells. Hence, the emission of such DAMPs is accompanied with prognostic benefit in cancer patients 34 . We here evaluated the potential of plasma-treated saline that may serve as basis for a future solution used for peritoneal lavage to induce toxicity and immunogenicity in colon cancer cells. We found that such solution was storable over weeks, mediated not only cell death in tumor cells but also altered their phenotype in disadvantage of metastatic spread, and acted as a bone fide inducer of pro-immunogenic markers.

Results
Plasma-treated saline was toxic to 2D and 3D colon cancer cultures. In view of a possible application of a solution that can be used for in vitro experiments as well as for future peritoneal lavage, we treated 50 ml of phosphate-buffered saline (PBS) solution with an argon plasma jet to deposit plasma-derived oxidants (Fig. 1a). Exposure times were either 20 min (P20) or 60 min (P60) with subsequent compensation for evaporation with double distilled water. Deposition of hydrogen peroxide (H 2 O 2 ) is common with argon plasma jets, and P60 correlated to a final concentration of 100 µM of H 2 O 2 . This concentration was measured constantly even after repeated freeze-thaw cycles of aliquots of this solution (Fig. 1b). In addition, the treatment regimen generated nitrate, nitrite, and superoxide (Supplementary Fig. S1) but not hypochlorous acid (data not shown) in saline solution. For cell experiments, four regimens were used for treatment of CT26 colorectal cancer cells: P0 (control PBS), P20, P60, and H100 (100 µM of experimentally added H 2 O 2 into 50 ml of PBS that corresponds to the concentration of H 2 O 2 generated with the P60 condition). Cell cultures need specialized media to meet their energy needs, which PBS does not. To test the optimal incubation time with our saline solutions, metabolic activity was assessed 24 h after treatment (Fig. 1c). Thirty minutes of incubation with P20 and P60 saline were more toxic compared to 1 min of incubation but similar efficient than 60 min. Therefore, the 30 min exposure time was chosen for subsequent experiments. Using the H 2 O 2 scavenging enzyme catalase, we confirmed that H 2 O 2 was mainly responsible for the cytotoxic effect of the P20 and P60 as well as the H100 treatment ( Supplementary Fig. S2). The cytotoxic effect was confirmed and even more pronounced in MC38 colorectal cancer cells, and less pronounced in PDA6606 pancreatic cancer cells and HaCat keratinocytes ( Supplementary Fig. S2). To test the tumor-toxic efficacy of plasma-treated saline in a physiologically more relevant model, cancer cell death was followed over 12 h post-exposure in a 3D tumor spheroids model (Fig. 1d). Quantitative image analysis from over 5,000 images revealed a significant increase in cell death with P60 and H100 exposure (Fig. 1e). Remarkably, plasma-treated saline (P60) was significantly more effective compared to H 2 O 2 saline (H100). To confirm this finding, spheroids were collected 12 h after treatment, digested to single cell suspensions, and quantified for the percentage of dead cells using flow cytometry (Fig. 1f). Results confirmed plasma-treated but H 2 O 2 -supplemented saline had a significantly higher cytotoxic effect in 3D tumor spheroids compared to the control condition (Fig. 1g). To validate that this finding was related to oxidants deposited via plasma treatment and accumulating within cells, CT26 cells were labeled with chloromethyl 2′,7′-dichlorodihydrofluorescein diacetate (CM-H 2 DCF-DA), a redox-sensitive probe fluorescing upon intracellular oxidation 35 with help of intracellular oxidases (Fig. 1h). High content imaging analysis of intracellular CM-H 2 DCF-DA mean fluorescence intensities (MFI) retrieved from several thousand cells per conditions revealed a significant increase in fluorescence for P20, P60, and H100 (Fig. 1i). Corroborating findings with 3D tumor spheroids, plasma-treated saline (P60) gave a significantly stronger increase in fluorescence compared to the hydrogen peroxide-matched control condition H100. Despite the prime role of hydrogen peroxide in cytotoxicity as seen with catalase controls ( Supplementary Fig. S2), this suggests plasma-derived oxidants other than H 2 O 2 to play in role in oxidation and cytotoxicity in tumor cells.
Oxidizing saline solutions reduced colon cancer cell growth and induced apoptosis. After 30 min of incubation with oxidizing saline solutions, a similar reduction in metabolic activity was observed after 4 h, 24 h, and 48 h when compared to individual P0 controls of each incubation time (Fig. 2a). Despite the fact that our treatments were of moderate cytotoxicity (about 25%), this suggests that incubation with oxidizing saline solutions had a profound and long-lasting effect in CT26 colon cancer cells. To resolve in vitro growth kinetics in more depths, time-lapse video microscopy experiments were performed over 48 h. Digital phase contrast (DPC) images (Fig. 2b) were used to quantify the total number of object (Fig. 2c) and to calculate cytosolic area (Fig. 2d) in these kinetic experiments. With both parameter, a decrease was observed after exposure to plasma-treated as well as H 2 O 2 saline. A decreased object count was also seen in MC38 cells to a higher extent, and in PDA6606 and HaCat cells to a lower extent ( Supplementary Fig. 2) compared to CT26 cells (Fig. 2c). The next question was whether cell proliferation was reduced or whether cell death occurred. To investigate this, propidium iodide (PI) was used (Fig. 2e) to quantify terminal cell death (Fig. 2f). The relative increase of dead cells upon treatment to plasma-treated or H 2 O 2 saline was consistent up to 48 h post exposure. To classify the mode of cell death, cells were stained with annexin V and 4′,6-diamidin-2-phenylindol (DAPI), and flow cytometry was utilized to distinguish between apoptotic and necrotic cells (Fig. 3a). Although small in amplitude, oxidizing saline solutions predominantly induced apoptosis with only few residual necrotic cells present (Fig. 3b). To confirm results with another apoptosis assays, samples were stained with a probe indicating caspase 3/7 activity in cells. Compared to annexin V experiments, similar rates of apoptosis were identified when staining for caspase activity (Fig. 3c).   Cytotoxic conditions not only induce cell death but also may reduce proliferation. Hence, we stained for the cell proliferation marker Ki-67 (Fig. 3d). A modest, non-significant decrease in cell proliferation was observed (Fig. 3e). Oxidizing saline solutions promoted a morphologically distinct cellular phenotype. Having identified the cytotoxic effects of our treatment, the next question was whether the saline solutions affected the phenotype of CT26 cells. Exposure to plasma-treated or H 2 O 2 -supplemented saline altered the cells' nuclear morphology (Fig. 4a). In particular, the mean nuclear area was significantly enhanced from 4 h up to 48 h after treatment (Fig. 4b). Nuclear swelling was similar in both the plasma and peroxide group. Flow cytometry of DNA content showed that in both groups the proportion of cells containing twice the amount of chromatin was significantly increased (Fig. 4c,d). These results argued for a cell cycle arrest in CT26 cells following incubation with oxidizing saline solutions, and the findings were confirmed in MC38, PDA6606, and HaCat cells ( Supplementary  Fig. S3). It should be mentioned that cell cycle arrest responses in response to toxic stimuli are not unique to malignant cells, as seen with side-effects of many chemotherapeutics in cancer patients on e.g. hair growth and hematopoiesis. We next thought to utilize high content image quantification to discern effects of oxidation on life cell morphology. Similar to cell culture medium controls without exposure to PBS (data not shown), PBS-treated cells had rather round cell bodies with few bipolar cells (Fig. 4e). By contrast, cells exposed to oxidants showed distinctly elongated and spindle-shaped cell bodies with partially branched extensions. The mean length of these extensions was significantly increased in cells 24 h and 48 h after exposure to oxidizing liquids (Fig. 4f).
Hourly assessment of mean cellular roundness (normalized to mean roundness of each condition at t = 0) was performed over two days to identify the kinetic of morphological alterations (Fig. 4g). Initially, the roundness of cells decreased gradually regardless of the treatment condition.  Fig. S3). The systematic of these changes in cellular phenotype following exposure to oxidants solutions could also be shown in two the tumor cell lines and HaCaT keratinocytes ( Supplementary Fig. S3), not suggesting a tumor-specific effects of oxidizing saline solutions to cell morphology. In principle, morphological alterations are often linked to a different profile in actin cytoskeleton distribution. We therefore quantified total phalloidin fluorescence (a stain for actin) directly (data not shown) and 4 h following treatment but did not notice any difference in total actin content between treatment regimens (Fig. 5a). However, spinning-disc confocal microscopy suggested a transformation in actin distribution patterns. Specifically, a difference in small pericellular signatures in cells exposed to oxidizing saline solutions was identified 4 h post treatment (Fig. 5b). The total length ( Fig. 5c) as well as actin ( Fig. 5d) content of extrusions directed radially away from the cellular body into the surrounding region were significantly increased. However, the functional consequences of these findings were not clear. One possibility was the induction of epithelial-mesenchymal-transition (EMT), a cellular phenotype involved in supporting tumor cell metastasis 36 . As this would be an unfavorable effect of oxidizing solutions for cancer treatment, we next sought to investigate CT26 cellular function and phenotype with regard to EMT.

Oxidizing saline solutions decreased cell motility and did not induce an EMT phenotype.
Upregulation of β-catenin and the transcription factor LEF1 are key events in EMT 36,37 . Furthermore, upregulation of vimentin, smooth-muscle actin (SMA), and S100A4, or loss of E-cadherin are accompanied with a disparaging mesenchymal tumor phenotype 38 . Using 7-color flow cytometry, expression levels of all mentioned markers were investigated in following exposure to oxidizing saline solutions. Neither an up nor down-regulation of any marker was identified in CT26 colon cancer cells (Fig. 6a). The small "shoulder" on the right of some histograms was equally present in unstained cells (which were exposed to P60 PBS) and stained, P60 treated cells. The shoulder in unstained but P60-treated cells was due to a relative increase of larger cells in these treatment regimens, which gave more auto-fluorescence as seen in mean forward scatter size (data not shown). To understand functional consequences of morphological alterations observed with our treatment, time-lapse video microscopy was utilized to investigate cellular dynamics in viable (PI − ) cells at different time points post exposure to oxidizing saline solutions. Especially for plasma-treated and partially for H 2 O 2 saline, total cellular displacement was decreased over 4 h intervals at 4 h, 24 h, and 48 h (Fig. 6b). More strikingly, average cell speed of viable CT26 cells was markedly decreased with all treatments (Fig. 6c). This was most evident 4 h after treatment and with the P60 condition. Interestingly, cellular motility (in tendency) and displacement (significantly) was reduced even 48 h after initial exposure to oxidizing saline solutions. Together with a lack of upregulation of key EMT markers, these results suggest that exposure to the oxidizing solutions does not promote a more motile and thus increasingly metastatic EMT-phenotype of CT26 colon cancer cells, despite profound morphological alterations.

Exposure to oxidizing saline solutions induces immunogenicity in colon cancer cells. Research
suggested that not necessarily the extent but rather the mode of cell death can be decisive in antitumor responses, especially with involvement of cells of the immune system 33 . It is known that tumor cells have the ability to evade immune response and therefore decrease the expression of molecules triggering immune activation. Therefore, we thought to determine the release and/or expression of several markers of immunogenic cell death (ICD), which can be sufficient to induce antitumor-immunity. Following exposure to oxidizing saline solutions, we identified an upregulation of key ICD-markers, such as calreticulin (CRT), heat shock protein 70 (HSP70), and high-mobility-group-protein B1 (HMGB1) on the cell surface of CT26 colon-cancer cells (Fig. 7a,c,e) using flow cytometry. The H 2 O 2 -degrading enzyme catalase abolished this increase in all groups ( Supplementary Fig. S4). Strikingly, our treatment induced an upregulated of ICD-markers also in MC38 and PDA6606 tumor cells, but not in non-malignant HaCat keratinocytes ( Supplementary Fig. S4). For CRT, we confirmed its significantly increased accumulation on PI − (live) CT26 cells exposed to oxidizing saline solutions using high content imaging ( Supplementary Fig. S4).In general for CT26 cells, ICD marker expression peaked at 24 h (Fig. 7b,d,f). To confirm a role of HMGB1, quantitative microscopy revealed a significantly increased translocation of the nuclear protein to the cytosol ( Fig. 7g) with plasma and H 2 O 2 treatment conditions (Fig. 7h). In addition, tumor cell-derived HMGB1 was secreted in the cell culture supernatants to a significantly higher extent compared to PBS controls ( Supplementary Fig. S4). For another DAMP, ATP, we could not detect an increase but a decreased in cells supernatant after 10 min of incubation in oxidizing saline solutions ( Supplementary Fig. S4). Not only CRT, heat-shock proteins, and HMGB1 but also cytokines and chemokines are major determinants for efficient antitumor immunity. Parallel quantification of 12 soluble targets in CT26 cell culture supernatants (sampled at 24 h) revealed an increase in tendency for IL2, CXCL9, MCP-1, and IFNγ as well as a significant increase for IL1β, IL6, IL12p70, CCL4, and TNFα with oxidizing saline treatment conditions ( Supplementary Fig. S5). At the same time, also levels of two rather anti-inflammatory factors (IL4, and IL10) were elevated with another one (TGFβ, a major immunosuppressive mediator) being decreased. In summary, plasma or H 2 O 2 -treated PBS induced a pro-immunogenic phenotype in several cancer lines concomitant with an overall pro-inflammatory chemokine/cytokine profile.
Oxidized saline decreased tumor burden in vivo and increased macrophage count and T cell activation. In order to evaluate the effect of plasma-treated saline in vivo, mice with CT26 peritoneal carcinomatosis (Fig. 8a) were treated either with untreated or plasma-treated saline. Repetitive application of plasma-treated saline significantly decreased the tumor mass to one third of that of the control group (Fig. 8b). Tumors are often infiltrated with different types of immune cells that together with stroma cells form the tumor microenvironment (TME), which can either spur or halt cancer growth 39 . Macrophages, positive for the surface markers CD45 and F4/80 in murine systems (Fig. 8c), play an important role in the TME, and an increase of these cells was observed in plasma-treated animals (Fig. 8d). These innate immune cells, capable of antigen-presentation to antigen-presentation to adaptive immune cells such as T cells, could have been attracted by dying cancer cells in response to repetitive exposure to plasma-treated saline. To address whether our treatment could have elicited antitumor T cells responses, splenocytes of the control and plasma group were incubated with heat-killed CT26 cells ex vivo. Assessment of expression levels of CD25 revealed an increase in CD4 + T cells (but not CD8 + , data not shown) in the plasma over the control group (Fig. 8e). These results suggested that

Discussion
The aim of this study was to investigate a physical plasma-treated phosphate-buffered saline solution in principle suitable for peritoneal lavage of tumor patients with colon cancer metastasis. The solution was not only cytotoxic in the murine cancer cell lines CT26, MC38, and PDA6606 as well as in CT26-bearing mice, but also evoked a pro-immunogenic phenotype in these cells.
Our treatment led to a significant upregulation of calreticulin (CRT) on colon cancer cells. CRT is a leading marker for ICD and a chaperon derived from ER that translocates to the cell surface to serve as eat-me signal for phagocytosis of tumor material [40][41][42][43] . Recent studies suggested CRT to serve as marker for positive therapeutic outcome in tumor-patients [44][45][46] . We also observed other damage-associated molecular patterns (DAMPs 32 ) such as heat-shock protein 70 (HSP70) and high-mobility-group-protein B1 (HMGB1) to be increased on the tumor cell surface with oxidizing treatment. Surface HSP70 is a chemoattractant for monocytes and neutrophil granulocytes and can lead to maturation of dendritic cells 32 . The nuclear protein HMGB1 can bind to several receptors like TLR2, TLR4, RAGE, and Tim3. These are known for their immunogenic signaling cascades 32 and induce maturation of dendritic cells (DCs), DC-dependent priming of T-cells 47 , and suppression of regulatory T-cells (Tregs) 48 . Hence, by upregulation of these surface markers, our treatment could have the potential to activate antitumor immunity. This was illustrated by the T cell activation observed in our study, which is in line with previous  findings 49,50 Such notion was further supported by analysis of several immunomodulatory mediators in the supernatant of cancer cells that were exposed to saline solutions. This included increased secretion of interleukin (IL) 1β, IL6, IL12p70, cc-chemokine ligand 4 (CCL4), and tumor necrosis factor α (TNFα). Moreover, IL2, c-x-c motif ligand 9 (CXCL9), monocyte chemotactic protein 1 (MCP-1), and interferon γ (IFNγ) were increased in tendency. Indeed, we observed an increase of intratumoral macrophages in our murine model, which supports the notion of improved 5-year survival observed in patients with colorectal cancer 51 . In general, chemokines/cytokines are potent immune-effectors in cancers 52 , crucial for shaping the tumor microenvironment in vivo 39,53,54 . The pyrogenic interleukin 1β is a major inducer of inflammation and macrophage activation, and mediates apoptosis. IL6 enhances proliferation and function of T and B cells 55,56 . IL12p70 is a prominent inducer of anti-tumor immunity mostly through the stimulation of natural killer (NK) cells and cytotoxic T lymphocytes (CTLs) and their subsequent secretion of IFNγ. IL6 diminished tumor growth of colorectal cancer in pre-clinical studies 57,58 The chemokine CCL4 works as attractant for a wide spectrum of immune cells 59 whose presence positively correlates with survival in colorectal cancer patients [60][61][62][63] . We also found increased levels of TNFα that is an inducer of apoptosis 64 . Immuno-suppressive TGFβ, decreased in our study, suppresses effector T cell function and enhances growth of stromal cells and extracellular matrix components, which in turn can promote unwanted mesenchymal transmission [65][66][67] . By contrast, increased levels of IL10 are not associated with favorable prognosis, because it can decrease cross-presentation of tumor antigens by dendritic cells to T cells, for example 68,69 . In general, chemokines and cytokines modulate inflammation, which in turn affects tumor cell behavior and phenotype 70 . For another DAMP, adenosine triphosphate (ATP), a decrease was observed in supernatants of CT26 cells exposed to oxidizing saline solutions. However, original studies identifying the immune-stimulating effects of ICD-induced CT26 colorectal cancer cells did not report on ATP but rather CRT being crucial for immuno-protection in mice 41,43 . Hence, despite the lack of the "find-me" signal ATP, we propose that oxidized saline treatment could increase the in vivo immunogenicity of three tumor cell lines (but not non-malignant HaCaT keratinocytes) due to a significant increase in the ICD-markers CRT, HMGB1, and HSP70/90 71 . Exposure to plasma-treated PBS led to phenotypical changes in CT26 cells, which were visible even 48 h after initial contact. The consistently enlarged nuclei and cell area as well as the decelerated cell growth and cell cycle are hallmarks of cellular senescence 72 . Senescence was previously observed in melanoma cells following exposure to cold physical plasma 73 . Interestingly, the authors found cell senescence only for treatment conditions with moderate toxicity, which corroborates the findings of our study. Many tumors have the potential to bypass cell cycle arrest in order to divide and grow dysregulated 74 . Cellular senescence was previously found to be associated with increased intracellular oxidized protein content 75 , an observation also made after plasma treatment 76 . Moreover, we found changes in actin distribution in CT26 cells. It is known that ROS act as signaling molecules for several downstream pathways that lead to remodeling of the cytoskeleton in seconds to minutes 77,78 . The unwanted mesenchymal tumor phenotype shows increased migration, proliferation, and expression of molecules responsible for rapid cytoskeletal rearrangement 79 allowing them to dispatch from primary tumors 36 . Our results with plasma-treated PBS contradict these characteristics. We found a sustained weakened cell growth, decreased cell motility, and a lack of increased expression of important EMT markers. Recent studies suggested senescence and its phenotypic alterations to repress malign, epithelial tumor phenotypes 80 .
For the in vitro effects of plasma jets in biological systems, H 2 O 2 was suggested to be an important molecule 81 . For this reason, we mimicked its lone effect by using a plasma treatment concentration-matched amount of H 2 O 2 as positive control. The plasma gas phase is rich in hundreds of different reactive species 82 but in liquids they quickly decompose to a few stable ones 83 . Superoxide, for example, is abundant 84 but rapidly dismutates to H 2 O 2 in solutions. Therefore, plasma-derived superoxide identified in our work unlikely contributed to biological effects directly. Along similar lines, nitric oxide and related reactive nitrogen species yield nitrite that further reacts to nitrate in plasma-treated liquids 85 . Although both are of low reactivity, there is a hypothetical framework attributing them a role in specifically mediating plasma-derived antitumor effects 86 . The assumption is that tumor cell membrane-resident NAPDH oxidases (NOX) may generate cell-derived superoxide, which may react with nitric oxide to cytotoxic peroxynitrite; the latter, however, may also derive from H 2 O 2 and nitrite (both identified in plasma-treated PBS in this study) 87 . In our assays, immediate oxidation of cells was significantly enhanced in plasma-treated compared to H 2 O 2 saline, despite identical concentrations of H 2 O 2 (but absence of hypochlorous acid) 88 . This points to H 2 O 2 -independent reactions or additional/unknown reaction intermediates 89 . More evidence of the importance of this chemistry was provided by pharmacological inhibition of NOX that strongly attenuated the effect of plasma on cancer cells 90 . We have previously seen the inability of a (to the plasma treatment time concentration-matched) amount of H 2 O 2 to replicate functional consequences in human neutrophils seen after direct plasma treatment 91 . Of note, neutrophils carry large numbers of NOX molecules on their membranes 92 . In the present work, we found rather small differences between plasma-treated and H 2 O 2 -supplemented phosphate-buffered saline. As major exception, killing of colorectal cancer cells in 3D tumor spheroids, a more realistic model in oncological research 93 , performed significantly better with plasma-treated PBS compared to H 2 O 2.
Tumor-bearing mice benefited from direct exposure of bulk tumors to the short-lived reactive species generated by cold plasmas [94][95][96] . Recently, several groups suggested to expose liquids to plasmas for tumor therapy. We previously observed toxic effects of such liquids in vitro and in vivo, including diminished cell growth, proliferation, and metabolic activity as well as induction of apoptosis but not necrosis in a model of pancreatic cancer 24 , which supports earlier work [97][98][99] . We could confirm this findings and translate it to another tumor model using plasma-treated saline. However, many studies used plasma-treated cell culture medium to inactivate cancer cells [100][101][102] . While this is an approach proven to be effective also in tumor-bearing mice 24,25,100 , the clinical translation of this concept may be limited.
Double-distilled plasma-treated water has been shown to be effective against cancer cells 103,104 but its use is discouraged in the clinical setting due to unfavorable osmolality. Sodium chloride (0.9%) solution is a clinically accepted solution, and phosphate-buffered saline (PBS) is very similar to that. Even following three weeks of storage we did not find any loss of the major oxidant H 2 O 2 deposited by this argon plasma jet into solution 89 . Previous studies on plasma-treated medium stored up to 7 days showed a drastic H 2 O 2 loss over time 105,106 , presumably due to reaction with protein thiols 107 . Yet, the immediate H 2 O 2 concentration of a plasma jet-treated liquid is independent from its protein content 108 , suggesting rather slow reaction kinetics for decay. Therefore, plasma-treated saline solutions that can be frozen and stored for longer periods of time should further be tested for its use as adjuvant for intraperitoneal cancer treatment in vivo experiments and potentially also regarding its clinical applicability.
Our study has limitations. First, we did not investigate the molecular machinery responsible for immunogenic cancer cell death observed. Especially caspase 8 and other signaling events upstream of the ER-stress sensor PERK may be important for the effects found in our study 109  of e.g. apoptosis, ROS generation, and translocation of ICD-relevant molecules may further context the presented findings within the vast body of literature available on CT26 cell death [110][111][112] . Third, also other liquids in clinical use may have the potential to elicit the effects observed in this study, e.g. Ringer-Lactate solutions studied previously with regard to plasma treatment 97 . Fourth, despite our willingness to perform the animal experiments including an H 2 O 2 control group, the local ethics committee denied this request and only approved the two groups presented. Notwithstanding, the here presented findings pose a promising and safe new therapeutic avenue for the treatment of colorectal cancer that warrants further investigations.

Materials and Methods
Cell culture. Murine colon carcinoma cells (CT26 and MC38) as well as the murine pancreatic carcinoma cells (PDA6606) and human non-malignant keratinocytes (HaCat) were maintained in Roswell Park Memorial Medium (RPMI) 1640 (Pan Biotech), containing 10% fetal bovine serum, 2% glutamine, and 1% penicillin-streptomycin (all Sigma). Cells were incubated at 37 °C, 95% humidity, and 5% CO 2 . Sub-culturing was performed twice a week. For experiments, 1 × 10 4 cells were seeded in flat-bottom 96-well plates (Eppendorf). These plates have a rim, which can be filled with sterile water to prevent excessive evaporation during incubation and subsequent alterations in the outer wells of the plate (edge effects).

Plasma and treatment.
To produce a sufficient amount of treatment solution, 50 ml of phosphate-buffered saline solution (PBS) without Ca 2+ and Mg 2+ (PAN Biotech) was exposed to cold physical plasma in a 150 ml glass beaker for 20 or 60 minutes. The atmospheric pressure argon plasma jet kINPen 113 (neoplas tools) was used working with five standard liters of argon gas (99.9999% purity; Air Liquid) per minute. The physical characterization of this plasma jet were thoroughly investigated in previous studies and are summarized in 114  the 96-well plate was filled with sterile water to protect cells from excessive evaporation. The imaging system is equipped with a laser-based autofocus to maximize image focus for subsequent software segmentation of cells and their properties using Harmony software. Alternatively, the cells were fixed and imaged at several time points.
To determine mean nuclear area after treatment, cells were fixed with 70% ethanol, washed, and incubated with 4′,6-diamidin-2-phenylindol (DAPI; BioLegend) before imaging and analysis with the high content system using the above mentioned conditions (except for the fluorescence setup, which was λ ex 365 nm and λ em 405 nm). For analysis of actin distribution, cells were cultured and treated in glass-bottom 96-well plates (Cellvis) and exposed to PBS as described. For imaging experiments, cells were fixed and incubated with flash phalloidin NIR 647 (BioLegend). For quantification of total actin levels, imaging conditions were as stated above (except for the fluorescence setup, which was λ ex 630 nm and λ em 660 nm). For this, cells were segmented using actin fluorescence, and the MFI for each cell per well was quantified. To resolve and quantify actin extrusions at the cell membrane, a 40x water immersion objective (NA = 1.1) was utilized for acquisition of several true-point confocal image stacks per field of view. To determine the nuclear and cytosolic distribution of high-mobility-group-protein B1 (HMGB1), cells were fixed with paraformaldehyde (PFA) and exposed to permeabilization buffer (both BioLegend) before overnight incubation with an unconjugated anti-HMGB1 antibody (Abcam). The next day, a secondary antibody conjugated with Alexa Fluor (AF) 647 Plus (Thermo Scientific) was added, and imaging was carried out following the procedures described above (except for the fluorescence setup, which was λ ex 650 nm and λ em 665 nm). To image calreticulin (CRT), an unconjugated CRT antibody (Abcam) was added for 1 h before labeling with a secondary AF647 Plus antibody (Thermo Scientific). PI was used to exclude dead cells for analysis. The high content imaging system acquires images with a sCMOS camera (4.7MP) at 16 bit; binning was 1. High performance 30-core computers were utilized for data analysis of about 200,000 single images in total across all imaging experiments performed in this study. To determine the expression of proteins involved in immunogenic cell death, we followed the same protocol except the fixation and permeabilization steps. Cells were labeled with antibodies targeting calreticulin AF647 (CRT; Novus), heat shock protein 70 AF488 (HSP70; Abcam), heat shock protein 90 AF700 (HSP90; Novus Biologicals), high-mobility-group-protein B1 PE (HMGB1; BioLegend), and DAPI. Only live cells with intact membranes (DAPI − ) were used for quantification of these markers via each MFI. Data analysis was performed using Kaluza analysis software 2.1 (Beckman-Coulter). For analysis of tumor lysates from animal experiments, cells were fixed with 4% PFA and stained with antibodies targeted against F4/80 conjugated with PE-Dazzle (BioLegend) and CD45 conjugated with AF700 (BioLegend) as well as DAPI. To assess T cell activation among splenocytes harvested from mice of the control and plasma group, 10 6 splenocytes (each sample was pooled from two animals of the same group sacrificed at the same time) were co-cultured with 10 5 heat-killed CT26 cells. After 24 h of incubation, cells were harvested and stained with conjugated antibodies targeted against CD3 (AF700), CD4 (PE-Cy7), CD8 (BV510), and CD25 (APC) to discriminate T cells and IAIE, CD45.R, F4/80, Ly6C (all APC-Cy7), and Zombie NIR (all BioLegend) to discriminate dead and non-T cells. T cell activation was assessed by quantifying the percentage of CD25 + T cells and multiplication with the MFI of CD25 + T cells in all CD4 + T cells, respectively (MFI x gate). The assay was repeated three times with pooled samples from two mice per sample and group.
Tumor spheroids. Three thousand CT26 cells were seeded per well of a 96-well spheroid cell culture plate (PerkinElmer) that prevents the attachment of cells to support spheroid formation. Plates were centrifuged at 1000xG for 10 minutes and incubated for 72 h. Subsequently, culture medium was replaced with saline solutions. After 30 min, fully supplemented cell culture medium containing sytox green (life technologies) was added as nutrient source and to detect terminally dead cells via microscopy. Hourly over 12 hours, images were acquired using the high content imaging device with environmental control (37 °C/5% CO 2 ), a 5x air objective (NA = 0.16), nine z-stack per spheroid ranging from −70 µm to 70 µm, and for each well and fluorescence channel. Filter setup was as described as above for CM-H 2 DCF-DA experiments. For analysis, z-stacks of each well were merged in Harmony software using maximum intensity projection. Spheroids were segmented to distinguish them from background, and sytox green mean fluorescence intensity (MFI) within that area was quantified and normalized to t = 0 of each individual spheroid and well. Immediately after imaging was finished, spheroids were collected, digested with accutase, washed, and subjected to flow cytometry to very imaging results.

Supernatant analysis.
Analysis of soluble analytes in cell-culture supernatants was carried out using the Legendplex (BioLegend) multi-analyte assay, which is a bead-based sandwich immunoassay assayed with flow cytometry. The experimental procedure was performed following the suppliers' instruction. The panel consisted ScienTific REPORTS | (2019) 9:634 | DOI:10.1038/s41598-018-37169-3 of the following fluorescence encoded beads directed against: interleukin (IL) 1β, IL2, IL4, IL6, IL10, IL12p70, chemokine (C-X-C motif) ligand (CXCL) 9, monocyte chemoattractant protein (MCP) 1, CC-chemokine ligand (CCL) 4, interferon (IFN) γ, tumor necrosis factor (TNF) α, and tumor growth factor (TGF) β. For quantification of the flow cytometry results, specialized data analysis software (Vigene Tech) was utilized. For each analyte, a separate standard curve was calculated, using fifth degree polynomials, with attention to each analyte's specific detection limits. To quantify HMGB1 in cell culture supernatant, we utilized an enzyme-linked immuno absorbance assay (ELISA; Hoelzel) following the manufacturer's protocol. ATP content in PBS 10 min after addition of control and oxidizing saline solutions to cells was quantified following the supplier's manual using an ATP Detection Assay Kit (Abcam). Luminescence was then measured utilizing a multiplate reader (Tecan).
Animal experiments. Mouse maintenance and experimental procedure were thoroughly reviewed and received approval by the State agency for agriculture, food safety, fishery Mecklenburg-Vorpommern (LALLF-MV, application number 7221.3-1-048/18). All animal procedures were performed in accordance with the relevant guidelines and all efforts were made to minimize suffering. Two million CT26 colorectal cancer cells were injected intraperitoneally (i.p.) to initiate disseminated tumor growth. Starting at day two, mice received at total of five i.p. injections of 300 µl of untreated or plasma-treated 0.9% sodium chloride solution (60 min of treatment kINPen treatment of 50 ml) every other day. Tumor nodules were carefully explanted and weighted prior to digestion with a commercially available tumor dissociation kit (GentleMACS; Miltenyi BioTec, Germany). A similar kit was used to make single cell suspensions of explanted spleens.

Conclusion
Plasma-treated and H 2 O 2 -supplemented saline not only induced growth retardation and apoptotic tumor cell death, but also altered the cells' phenotype in disadvantage to metastatic spread. Plasma treatment was superior to H 2 O 2 treatment in inducing several morphological alterations and in killing cancer cells grown in 3D tumor-spheroids. In several tumor cell lines, major mediators of anticancer immunity were identified using our treatment regimens, all above calreticulin. Importantly, plasma-treated saline was effectively reducing tumor burden in vivo with a potential role of increased immunogenicity as shown by increased amount in macrophages and T cell activation. These results suggest that such a solution should be tested further as possibly promising adjuvant in colorectal cancer therapy to tackle peritoneal carcinomatosis.

Data Availability Statement
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.