Role of Ambient Gas Composition on Cold Physical Plasma-Elicited Cell Signaling in Keratinocytes

Abstract

A particularly promising medical application of cold physical plasma is the support of wound healing. This is presumably achieved by modulating inflammation as well as skin cell signaling and migration. Plasma-derived reactive oxygen and nitrogen species (ROS/RNS) are assumed the central biologically active plasma components. We hypothesized that modulating the environmental plasma conditions from pure nitrogen (N₂) to pure oxygen (O₂) in an atmospheric pressure argon plasma jet (kINPen) will change type and concentration of ROS/RNS and effectively tune the behavior of human skin cells. To investigate this, HaCaT keratinocytes were studied in vitro with regard to cell metabolism, viability, growth, gene expression signature, and cytokine secretion. Flow cytometry demonstrated only slight effects on cytotoxicity. O₂ shielding provided stronger apoptotic effects trough caspase-3 activation compared to N₂ shielding. Gene array technology revealed induction of signaling and communication proteins such as immunomodulatory interleukin 6 as well as antioxidative and proproliferative molecules (HMOX1, VEGFA, HBEGF, CSF2, and MAPK) in response to different plasma shielding gas compositions. Cell response was correlated to reactive species: oxygen-shielding plasma induces a cell response more efficiently despite an apparent decrease of hydrogen peroxide (H₂O₂), which was previously shown to be a major player in plasma-cell regulation, emphasizing the role of non-H₂O₂ ROS like singlet oxygen. Our results suggest differential effects of ROS- and RNS-rich plasma, and may have a role in optimizing clinical plasma applications in chronic wounds.

Introduction

In the last few years, cold physical plasma gained attention as a promising tool for medical applications in wound management and care. It is well known that plasma is highly antimicrobial against prokaryotes and biofilms in vitro (1, 2, 3, 4, 5) or antiseptic regarding wounds in vivo (6, 7, 8). Plasma therapy seems promising because it is low in temperature, nonmutagenic, and well tolerated by tissues, and with a wide range of sources, easy to use (9, 10, 11). The first preclinical (12) as well as clinical studies (13, 14) fostered the in vivo potential of plasma treatment of poorly or nonhealing wounds in patients.

Cold plasma is defined as a partially ionized gas producing reactive oxygen (ROS) and nitrogen species (RNS), electrons, ions, an electric field, and ultraviolet, thermal, and infrared radiation (15). The kINPen plasma jet is generally operated with pure argon. By admixing molecular gases, the composition of reactive species in the plasma effluent can be modulated. Addition of N₂ leads to increased generation of RNS (16), whereas addition of O₂ shifts the balance more toward the ROS (17). Feed gas humidity is an additional source of ROS production (18). The effluents species composition is also affected by the plasma jet’s surroundings as components of the ambient air diffuse into the plasma effluent and alter yields of ROS and RNS (19).

Biological systems such as cells and tissues have established various mechanisms, enabling them to control the amount of reactive species to stabilize a baseline level of reactive oxygen intermediates that is tolerable and partly even beneficial to the cell (20, 21). Therefore, a more balanced concept of the quantity and the effects of potentially dangerous particles such as ROS/RNS on living cells is accepted (22) and is called “hormesis” (23, 24). Subtle increase in ROS/RNS has even been associated with increased lifespan (25), as opposed to the implications of the free-radical theory of aging. Plasma-generated reactive species are believed to play a central role in the medical utilizations of plasma (26), as they govern a variety of processes such as distinct cytotoxicity (27, 28) and cell signaling (29, 30). Consequently, redox-related signaling plays an important role in cellular reactions to plasma exposure.

Keratinocytes constitute the uppermost layers of the skin and as such are central players of the reepithelialization during wound repair by migrating and proliferating onto the provisional matrix of the underlying granulation tissue (31). Using HaCaT keratinocytes as an in vitro cell culture system, we here investigated plasma-derived activation of skin cells under different controlled conditions. With regard to this, shielding of an unchanged argon gas plasma created by the kINPen was varied from pure nitrogen to pure oxygen. After exposure of plasma-treated medium to HaCaT cells, mapping of the gene expression signature was conducted to obtain deeper insights into the underlying mechanisms and molecular biological pathways. We identified alterations in cell viability, gene and protein expression, and cytokine release.

Materials and Methods

Cell culture and plasma treatment

All experiments were performed using human adult low-calcium high-temperature keratinocytes HaCaT (DKFZ, Heidelberg, Germany) cultivated in Roswell Park Memorial Institute 1640 medium supplemented with 8% fetal bovine serum (Sigma-Aldrich, Munich, Germany), 2 mM L-glutamine, 0.1 mg/L streptomycin, and 100 U/mL penicillin (Lonza, Visp, Switzerland). Plasma treatment was performed using the atmospheric pressure plasma jet kINPen 09 (voltage of 2–6 kV_pp, frequency ∼1 MHz; Neoplas, Greifswald, Germany), which was surrounded by a gas-shielding device, separating the plasma effluent from the ambient air. The plasma was ignited 1 h before the treatment to remove humidity from the tubing. The kINPen was operated with argon (three standard liters per min, slm), and the shielding gas consisted of different ratios of O₂ and N₂ with a total volumetric flow rate of 5 slm. The oxygen/nitrogen (in %) in the shielding gas was varied in five steps (0:100; 25:75; 50:50; 75:25; 100:0). The parameters for each shielding gas composition were adjusted 3 min before treatment to guarantee constant conditions and homogeneous gas mixtures. The plasma treatment of 1,000,000 cells was performed in an indirect way: 5 mL of cell culture medium were treated in a 60-mm petri dish in a meandering way for desired time at a fixed distance of 12 mm. Subsequently, the cell medium was replaced by plasma-treated medium. Positive control cells received 100 μM H₂O₂. In previous work, we described the effects of pure argon plasma without shielding on the cell behavior to exclude effects of the carrier gas in HaCaT keratinocytes (32, 33, 34).

Cell cytotoxicity and apoptosis assay

Cell viability after plasma treatment was performed with the CellTox Green Cytotoxicity Assay (Promega, Mannheim, Germany). For this, 15,000 keratinocytes were seeded per well in a 96-well plate 24 h before replacing the culture medium by 75 μL plasma-treated medium indirect treatment. After 3 h, medium was replaced. After 24 h, the CellTox Green Dye was added and incubated for 15 min. The CellTox Green Dye becomes fluorescent after binding to DNA in dying cells with leaky cell membranes. Total fluorescence at λ_EX 485/λ_EM 525 nm was measured using a microplate reader (Tecan, Männedorf, Switzerland). At least six replicates were measured for all samples. Apoptosis was measured by Green Caspase 3 Staining Kit (Promo-Kine, Würzburg, Germany) according to manufacturer’s instructions. Eighteen hours after indirect plasma treatment, flow cytometry analysis was performed and data were analyzed using the software Kaluza (Beckman-Coulter, Brea, CA). This assay was repeated three times in independent experiments.

Gene expression analysis

Transcriptomic signatures of untreated and plasma-treated HaCaT cells were recorded 3 h after treatment using microarray technology as described in Schmidt et al. (34). Briefly, total RNA from four independent biological experiments was isolated using an RNA Mini Kit (Bio & Sell, Feucht, Germany), and cDNA was synthesized by reverse transcriptase reaction (SuperScript Double-stranded cDNA Synthesis Kit; Invitrogen, Karlsruhe, Germany). At this step, samples from technical replicates and independent experiments were pooled, and cDNA was labeled with Cy3. Hybridization of the Cy3-labeled cDNA (Hybridization and Tracking Control Kit) was done on a 4 × 72,000-microarray chip for 20 h at 42°C. Afterwards, the chip was washed, dried in a centrifuge for 2 min (all from the software NimbleGen; Roche Diagnostics, Basel, Switzerland), and scanned with a resolution of 5 μm by the MS 200 Microarray Scanner (Tecan). With the NimbleScan v2.6 software, a background correction of the signals was performed via a robust multichip average algorithm and the signal intensities of the raw data were converted into gene expression levels, which were then used for the analysis with the Partek Genomic Suite software (Partek, Chesterfield, MO). The Protein Analysis Through Evolutionary Relationships (Panther 11; http://pantherdb.org/) classification system was used to classify the changed genes and sort them into biological function groups (35). Molecular interaction pathways and networks were generated with the software Ingenuity Pathway Analysis (Qiagen, Hilden, Germany).

Quantitative real-time PCR (qPCR) using a human wound-healing RT² Profiler PCR Array (Qiagen) was conducted, analyzing 84 genes of interest plus housekeeping genes. After RNA isolation as described above, cDNA was synthesized with the RT² First Strand Kit. qPCR was performed using the RT² SYBR Green Mastermix and a LightCycler 480 (Roche Diagnostics). Data analysis was done at the free PCR Array Data Analysis Web Portal of Qiagen. Single qPCR reactions were performed with RealTime ready catalog assays for VEGFA, IL6, HBEGF, CSF2, prostaglandin-endoperoxide synthase 2 (PTGS2), and RPL13A (housekeeping gene) using appropriate settings (Roche Diagnostics). Each sample was measured in duplicate and in three independent experiments. The ΔΔC_t method was used to quantify gene expression for each condition in fold change (36). Gene expression was considered significantly changed if a twofold up- or downregulation compared to the untreated control was detected.

Protein expression analysis

VEGFA (Thermo Fisher Scientific, Darmstadt, Germany) and the human interleukins (IL) 6 and 8 and GM-CSF (all BioLegend, San Diego, CA) were quantified in cell culture supernatants using the ELISA technique according to the manufacturer’s instructions. Briefly, one day before sample addition, each well of the 96-well plate was coated with a capture antibody at 4°C overnight. A horseradish peroxidase-conjugated antibody specific for the target protein was added. Subsequently, a substrate solution was added and the reaction was terminated by adding sulfuric acid. The absorbance was measured utilizing a microplate reader (Tecan) at 450 nm and a reference wavelength of 570 nm, and protein concentration was calculated from respective standard curves.

Measurement of ROS and RNS

ROS and RNS were measured either with colorimetric methods (nitrate, nitrite, and hydrogen peroxide) or by spin-trap enhanced electron paramagnetic resonance (EPR) spectroscopy (hydroxyl and superoxide anion radicals) (37, 38). For the evaluation of the EPR measurements, the spin-trap efficacy of the used spin trap (5,5-dimethyl-1-pyrroline-N-oxide) was taken into account. All given concentrations were determined after 180 s plasma treatment of Dulbecco’s phosphate-buffered saline (DPBS) solution.

Statistical analysis

For comparison of experimental groups, two-way analysis of variances (ANOVA) was utilized. For comparison of results obtained from flow cytometry, one-way ANOVA with Dunnett’s postcorrection was utilized. For gene expression quantification and comparison between groups, unpaired Student’s t-test was used. Statistical significance levels were indicated as follows: ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, and ^∗∗∗p ≤ 0.001. Calculations were done using GraphPad Prism (GraphPad Software, San Diego, CA).

Results

Ambient plasma conditions slightly affect cell viability and apoptosis

To study the influence of oxygen and nitrogen, the ambience of the kINPen effluent was controlled using a shielding gas device (Fig. 1 a). Five different shielding gas mixtures from pure oxygen to pure nitrogen (from 0% O₂ and 100% N₂ to 100% O₂ and 0% N₂) were used whereas the core plasma remained unchanged. First, it was assessed whether the composition of shielding gas had any effect on cytotoxicity in HaCaT keratinocytes. It was found that the general tolerance toward the treatment was good: only a minor fraction of the cells responded with necrotic or apoptotic cell death. The mean fluorescence intensity corresponding to necrotic cell death did not significantly increase for exposure to 20 s plasma-treated medium for each shielding gas composition (data not shown). In case of 180 s plasma treatment, a significant increase in necrotic cytotoxicity was observed with increasing amount of oxygen in the shielding gas. A maximum of 10% of the cells was found to be necrotic (Fig. 1 b). Caspase 3/7 activity—reflecting apoptotic cell death—was absent for 20-s plasma treatment (data not shown). However, a treatment of 180 s induced caspase activity in ∼20% of cells for each shielding gas containing O₂. Strikingly, using shielding with pure N₂, a significantly lower percentage of cells (12.5%) showed this apoptotic marker (Fig. 1 c). The CellToxGreen assay (Promega), via DNA staining, monitors cell death caused by necrotic processes that are accompanied by a compromised cellular membrane integrity. It is accumulation based and does not allow real-time detection of cell death. Together, these two facts result in the described differences to caspase 3/7 staining.

Figure 1.

Impact of shielding gas on cell viability in human keratinocytes. Scheme (left) and image (right) of the kINPen 09 plasma jet equipped with a shielding gas device. A 60-mm petri dish containing 5 mL Roswell Park Memorial Institute 1640 medium was plasma treated for 180 s, and the liquid was immediately added to HaCaT cells for 3 h (a). Afterwards, the cell culture medium was replaced. The shielding gas mixture during plasma treatment was varied in five steps, ranging from pure nitrogen to pure oxygen. The x axis indicates the O₂ amount of the O₂ to N₂ shielding gas mixture in percentage. Quantitative analysis of percentages of dead cells (mean fluorescence intensity) shows either cytotoxicity measured by a CellTox Green assay (b) or caspase 3 activity analyzed by a Green Caspase 3 Staining Kit (c) after 24 h. Un- or H₂O₂-treated cells (100 μM) were used as controls. Data are mean ± SD of three independent experimental repetitions and technical triplicates. Asterisk (^∗) presents a significant difference (α = 0.05) measured using Tukey’s multiple comparisons posttest in one-way ANOVA.

Plasma modulates the transcriptome signature of HaCaT keratinocytes

To elucidate the fundaments of plasma-triggered cell response, the global gene expression signature of HaCaT keratinocytes was determined by detecting the specific mRNA content as a measure for the activity of each gene (transcriptome). Plasma treatment triggered a distinct impact on the cellular mRNA profile, depending on the ambient conditions created by the shielding gas. The heat map represents all genes that were up- (blue), downregulated (red), or unchanged (white). Each row represents one condition (Fig. 2 a). The number of significantly regulated genes varied with treatment time and shielding gas composition. Mainly but not absolutely, the longer the treatment duration, the higher the number of genes regulated. E.g., for the 50:50 composition of the shielding gas, nearly the same amount of differentially expressed genes was detected for 20 s (207 genes) and 180 s (162 genes). Regulated genes do not necessarily overlap; for example, plasma-treated medium at 0% and 50% O₂ in the shielding gas mixture induced a differentially gene expression of a similar number (166 vs. 162 at 180 s) but not the type of genes. A pure oxygen gas surrounding induced the highest change in the transcriptome: 538 genes were expressed greater than or equal to twofold after 180 s plasma treatment, in contrast to 25% O₂/75% N₂ that changed only the expression of 42 genes after the same treatment duration. Interestingly, shielding gas with 25% oxygen or 25% nitrogen induced the smallest changes in the transcriptome.

Figure 2.

Shielding gas composition affects gene expression signature of HaCaT cells detected by microarray. HaCaT cells were exposed to plasma-treated medium (either for 20 or 180 s) for 3 h, and microarray analysis was conducted. Heat map illustrates the fold regulation of genes, which were significantly upregulated (plus twofold; blue), downregulated (minus twofold; red), or not regulated (white) by cold plasma. The amount of O₂ was given in percent. The numbers represent significantly changed genes in comparison to untreated cells (a). Fold regulation of plasma-treated transcripts were detected by microarray, which were found to be differentially expressed in all groups (mean level) (b). Three independent experimental repetitions and technical triplicates were performed.

Plasma shielding modulates cell-signaling molecules such as growth factors and cytokines

Subsequently, gene ontology analysis was performed to characterize the observed changes in gene transcription. Similar gene ontology class profiles were found for all shielding gas conditions; the highest number of changes due to plasma treatment were in metabolic processes and cellular communication (data not shown). Furthermore, heme oxygenase 1 (HMOX1), ChaC glutathione-specific γ-glutamyl cyclotransferase 1 (CHAC1), G protein-coupled receptor 34 (GPR34), oxidative stress-induced growth inhibitor 1 (OSGIN1), RUSC1 antisense RNA 1 (C1orf104), sclerostin (SOST), and solute carrier family 22 member 13 (SLC22A13) were found to be differentially expressed for all investigated treatment conditions. Furthermore, several cell signaling molecules, i.e., growth factors such as the heparin binding epithelial (HBEGF), the macrophage-stimulating growth factor 2 (CSF2), the vascular endothelial growth factor (VEGF), cytokines (IL-6 and IL-8 and cytokine receptor IL-6R), and a signal transducer (PTGS2), were upregulated after plasma treatment. All genes showed an approximately twofold regulation independent of the shielding gas composition, especially pronounced after 180 s treatment and 0% O₂/100% N₂ shielding (Fig. 2 b). A more detailed overview over all expressed genes is given in Table 1. We found a strong upregulation of HMOX-1 and OSGIN 1. Moreover, pathways analysis used IPA-identified genes of mitogen-activated kinase (MAPK) pathways in plasma-treated cells, including those of MAPK 15.

Table 1.

Transcript Levels of Genes Detected in All Sample Groups by Microarray

Groups	20 s					180 s
% O2/N2	100/0	75/25	50/50	25/75	0/100	100/0	75/25	50/50	25/75	0/100
ID / number	131	4	192	45	19	145	40	147	88	461
C1orf104	1.53	1.96	2.94∗	1.97	1.92	2.70∗	2.20∗	2.82∗	2.10∗	2.94∗
CHAC1	−1.06	−1.08	1.12	1.46	1.03	3.95∗	3.47∗	4.25∗	4.23∗	4.96∗
GPR34	2.23∗	1.53	1.45	2.44∗	2.04∗	3.54∗	1.61	2.48∗	1.73	2.00∗
HMOX1	1.12	1.13	1.11	1.71	2.15∗	2.79∗	3.71∗	4.63∗	3.39∗	4.17∗
MAPK15	1.41	1.55	2.64∗	1.63	1.80	1.95	1.96	2.25∗	2.37∗	3.00∗
OSGIN1	1.27	1.56	2.04∗	1.80	1.24	2.36∗	2.90∗	3.34∗	3.61∗	4.21∗
SLC22A13	1.83	2.63∗	3.31∗	2.02∗	2.02∗	3.03∗	2.44∗	3.48∗	2.92∗	4.85∗
SOST	1.75	1.69	1.96	1.69	2.44∗	2.15∗	2.25∗	2.28∗	2.10∗	2.42∗
HBEGF	1.06	−1.04	1.04	1.55	1.00	2.94∗	1.78	2.21∗	2.55∗	2.36∗
PTGS2	−1.10	−1.03	−1.60	−1.35	−1.36	2.21∗	1.18	1.34	1.38	1.31
IL-6	1.05	1.20	−1.07	1.27	1.20	2.33∗	1.71	1.55	2.02∗	1.68
IL-6R	1.29	1.24	1.03	1.25	1.10	3.02∗	2.57∗	2.58∗	2.23∗	2.74∗
Wnt9a	1.10	1.67	1.69	1.17	1.41	1.47	1.96	1.89	1.69	2.80∗

The asterisk (^∗) indicates significant differences in mRNA expression.

Regulation of gene expression obtained from global transcriptomic profiling of plasma-treated HaCaT cells (20 and 180 s) 3 h after treatment in comparison to controls. The human gene expression assay detects several differentially expressed genes. Microarrays were performed with Cy3-labeled cDNA. Table shows an induction of proliferative (HBEGF), immune-modulatory (IL-6, IL6R), and signaling (HMOX1, PTGS2, WNT9a) molecules. Data are presented as mean of at least three independent experiments.

Via quantitative PCR, it was measured that plasma treatment changed the gene expression of the growth factors VEGFA, HBEGF, CSF2, PTGS2, and IL-6 significantly, and expression levels were modulated by the shielding gas composition. The cellular response was dependent on the oxygen content in the shielding gas resulting in an S-shape pattern of the expression levels. The minimum always occurred at a treatment with 25% O₂ in the gas mixture, whereas the maximum varied, but was always observed at an O₂-dominated shielding gas mixture. In VEGFA, pure nitrogen or pure oxygen shielding gas resulted in a 2.7-fold upregulation, whereas a mixture of 25% oxygen and 75% nitrogen were not as potent (2.1-fold). The highest increase of 3.6-fold was reached with the oxygen-dominated shielding gas mixture 75% O₂/25% N₂, followed by a 3.1-fold increase for 50% O₂/50% N₂ (Fig. 3 a). Similar changes were also observed for HBEGF (Fig. 3 b) and CSF2 (Fig. 3 c). A 2.2-fold upregulation of PTGS2 could be observed for 75:25% shielding gas composition (Fig. 3 d). Additionally, IL-6 mRNA level was significantly influenced by the shielding gas composition. The higher the amount of O₂ in the shielding gas, the higher the expression of IL-6 (Fig. 3 e). However, IL-8 mRNA expression was not changed by plasma treatments (data not shown).

Figure 3.

Plasma shielding gas variations induce differential gene expression signatures in HaCaT keratinocytes. Quantitative gene expression levels of VEGFA (a), HBEGF (b), CSF2 (c), PTGS2 (d), and IL-6 (e) 3 h after exposure to plasma-treated medium (180 s) are depicted. Values in the area between lines were not significantly changed (plus twofold upregulated; downregulated below 0.5). Analysis was done with Tukey’s multiple comparisons test and one-way ANOVA with a significance level of α = 0.001 (^∗∗∗). Bars and error bars are mean ± SD.

Shielding gas variations around the effluent induce an altered secretion profile in HaCaT cells

The secretion of the proteins VEGFA, GM-CSF (product of gene CSF2), and IL-6/8 was measured 6, 12, 18, or 24 h posttreatment. VEGFA was significantly released after 12 and 24 h (Fig. 4 a for 24 h), whereas CSF2 was more strongly released after 6 h (Fig. 4 b for 6 h). For both proteins, the short treatment duration did not influence the release compared to the untreated cells, independent of the shielding gas used (data not shown). Twenty-seconds plasma-treated medium showed slight increase in IL-6 secretion compared to control in oxygen dominated shielding gases (75%, 100%, data not shown) whereas a 180 s treatment released elevated amount of IL-6 (Fig. 4 c for 18 h). The secretion of IL-8 was not changed with no or low but only for higher (75 and 100%, respectively) oxygen concentrations (Fig. 4 d for 18 h). Finally, a significant protein release of HBEGF and PTGS2 was not detectable in any samples or time point (data not shown).

Figure 4.

Extracellular concentration of cytokine and chemokine signaling proteins. Secretion of VEGFA 24 h (a); GM-CSF 6 h (b); IL-6 18 h (c), and IL-8 18 h (d). Keratinocytes were incubated with plasma-treated medium (using a shielding device and varying concentrations of O₂ and N₂) or H₂O₂ (100 μM). Analysis was done with Tukey’s multiple comparisons test and one-way ANOVA with a significance level of α = 0.05 (^∗), α = 0.01 (^∗∗), and α = 0.001 (^∗∗∗). Four experimental repetitions and technical triplicates were measured. Bars and error bars are mean ± SD.

Plasma shielding gas alters reactive species compositions in liquids

The plasma treatment of a liquid with different shielding gas compositions yielded a change in ROS and RNS concentration as summarized in Fig. 5 a for DPBS measured in previous works (37, 38). Total oxidant levels were highest for pure nitrogen in the shielding gas. Contrary to the expected behavior, the amount of H₂O₂ was not maximal for 100% O₂; instead, the highest concentrations were determined for 100% N₂. For all oxygen containing shielding gas compositions, H₂O₂ concentrations were lower and no remarkable differences between these conditions were observable. In contrast to H₂O₂, free oxygen radicals (^⋅OH and O₂^⋅) were produced in higher amounts when 75% O₂ or more was present in the shielding gas. Nitrite and nitrate were not detectable after plasma treatment with 100% oxygen shielding gas, whereas for 100% nitrogen shielding gas, small amounts were observed. Both RNS show the highest concentration in the cases of 25% O₂ and 75% N₂ shielding gas composition.

Figure 5.

Shown here is the concentration of reactive oxygen and nitrogen species in plasma-treated liquids (a) and the proposed signaling pathways in this study (b). Concentrations of ROS and RNS were measured either with colorimetric methods (as for nitrate, nitrite, and hydrogen peroxide) or by spin-trap-enhanced EPR spectroscopy (as for hydroxyl and superoxide anion radicals). Samples were taken of 5 mL of DPBS solution subjected to 180 s of plasma treatment as measured and summarized (a) (37, 38). We identified an activation of several transcription factors, e.g., Nrf2 and p53 via MAPK signaling pathways, and a VEGF release in human HaCaT cells (12, 33, 40). Cold plasma plays a crucial role in the cellular response to oxidative stress via Nrf2 pathway as well as in the process of proliferation and migration ex vivo by upregulation of growth factor expression like VEGF, CSF2, etc. Furthermore, plasma-induced upregulation of CHAC1 may play a role in the unfolded protein response, and in regulation of glutathione levels and oxidative balance in the cell. Moreover, OSGIN1 is a key regulator of both inflammatory and antiinflammatory molecules (b).

Discussion

Due to their high reactivity, ROS and RNS as well as oxidants interact with cellular (bio)molecules in both specific and unspecific ways, thereby facilitating the activation of signaling events (39). Reactive species have been determined to be the central players of kINPen-generated plasmas (40), linking medical utilization of plasma to a therapeutic, and moreover, beneficial application in wound healing. Multiple studies demonstrated a certain degree of selectivity of cold plasma (41, 42, 43). Whereas low treatment intensities are already sufficient to inactivate or kill bacteria, a similar treatment is harmless for human skin cells probably due to differences in cell metabolism (44, 45). In wound healing, a plethora of signaling pathways and secondary effects such as immune cell recruitment are under redox control (46, 47, 48). Additionally, differences in the induction of the antioxidative response (33, 40), reversible loss of cell adhesion (49, 50), inhibition of migration (51, 52), or stimulation of angiogenesis (53) are some of the specific effects of human skin cells after plasma treatment. Clearly, effects of reactive species are overlapping, making explicit statements at this point of research difficult, but implicate that modulation of plasma gas parameters during treatment is an option to be tested. Thus, effective sensing of species levels is very important to cells, particularly in wound healing processes (54, 55, 56).

As shown by comparison between direct and indirect treatment, the effect of charged particles on plasma and synergy with longer living active molecules and plasma-generated radiation (e.g., electromagnetic, ultraviolet, and thermal radiation) play the essential role in interaction with, and inactivation of, bacteria (57). Any type of radiation as well as short-lived species like atomic oxygen and singlet oxygen is excluded when exposing the cells to the plasma-treated medium only. However, we observed similar effects in immune cells (58) after being exposed to either direct or indirect plasma treatment, which was largely attributed to the deposition of hydrogen peroxide into the medium and subsequent secondary reactions of this compound (59). Moreover, the drying effect of the argon gas immediately necrotizes adherent cells and causes mechanical stress, suggesting a preferential use of an indirect treatment strategy (60). Additionally, it allows a quick and homogenous stimulation of mammalian cells, regardless of model and plasma source dimensions.

In jet plasmas, feed gas admixtures can influence the presence of ROS/RNS in the effluent. Addition of N₂ leads to increased generation of RNS, whereas addition of O₂ shifts the balance more toward the ROS (61, 62). Humidification of the feed gas has a large impact on species production with high levels of ROS (e.g., H₂O₂) created (18, 59, 63). On the other hand, the ambient conditions into which the plasma plume extends also considerably impact on the species composition (19). By active or passive transport processes, atmospheric species enter the effluent, modulating gas phase chemistry and the amount of reactive species like O₃, ¹O₂, ^⋅OH, ^⋅O₂⁻, H₂O₂, NO₂^⋅, HNO₂/HNO₃, and others as detected by Fourier transformed infrared spectroscopy measurements and by optical emission spectroscopy (64, 65). Using a shielding device on the kINPen, the effluents ambience can be controlled, allowing focusing the species production toward ROS or RNS, respectively. Besides the discussed impact on the gas phase chemistry, first experiments also suggested that biological effects in liquids can be regulated (19).

Here, a more detailed investigation elucidates the potential to modulate the cellular fate by ambient control of a plasma jet. While keeping the feed gas constant (pure argon), the effluents ambience was adjusted between pure nitrogen to pure oxygen in five steps. This is reflected by changes in tolerance toward the treatment as detected in cell toxicity assays, measuring an increase in toxicity with increasing oxygen fraction in the shielding gas. Hydrogen peroxide is also effective in this assay and as has been reported previously, the feed gas composition largely determines its deposition rate (18, 64). Yet when using the gas shielding, the amount of H₂O₂ deposited decreased with oxygen content, unexpectedly creating an inverse correlation between H₂O₂ and observed cell toxicity. A similar correlation was found for the caspase activity indicating apoptotic processes in the HaCaT cells, which escalate with decreased nitrogen fraction in the gas shield. Yet, the majority of cells tolerated the treatment and responded more specifically. The amount of RNS deposited in the different treatment conditions was comparably low and showed a bell-shaped behavior with minima at 0% oxygen and 0% nitrogen, respectively, illustrating the necessity of both components to form RNS (43). Reactive nitrogen species exhibit strong biological effects, as is well documented for nitric oxide (66). More recently, nitrite and even nitrate were shown to serve as ^⋅NO storage in humans (67, 68), obviously connecting clinical or experimental plasma effects to species composition. On the other hand, with an increase in oxygen in the shielding gas, a higher fraction of this gas will appear in the effluent. By interaction with argon, metastable, atomic oxygen is created. Beside exerting liquid chemistry and biological impact of its own (61), it favors the formation of ozone via a three-body collision including molecular oxygen. Ozone is a strong oxidant, yet with a very limited solubility when compared to hydrogen peroxide (69) and its unambiguous detection in a liquid remains elusive so far for the lack of specificity. Atomic oxygen and/or ozone may also modulate the deposition or creation of other reactive species, e.g., via UV light absorption (O₃) or quenching of ^⋅OH and ^⋅HO₂ (atomic O) (65).

For this complex situation, the role of ROS in plasma-induced effects on biological systems is still under debate (22, 70). Yet, in the data presented here, indications of their impact (at least in vitro) are present as discussed for the cytotoxicity assay. Further suggestions could be taken from the transcriptomic data and qPCR/ELISA readings for genes of interest. A combination of plasma-generated ROS and RNS was responsible for the observed alterations: the strongest response on gene or protein level was observed for 75% O₂ and 25% N₂, the condition with the lowest total amount of ROS/RNS. In contrast, the changes were found for cases with the highest amount of reactive species (100% N₂ and 75% N₂: 25% O₂). With increasing oxygen amount in the shielding gas and rising ROS production in the gas phase (and likely the free oxygen radicals in the liquid phase as indicated in Fig. 5 a) (71), a stronger activation of cellular defense mechanisms was observed. By varying the shielding gas composition, we found an important change of the antioxidant response element signaling pathway (e.g., nuclear factor (erythroid-derived 2)-like 2, also known as Nrf2) (72) controlling ROS levels and cell death progression together with strong upregulation of antioxidant enzymes HMOX-1 and OSGIN1, especially when the shielding gas mixtures contained high fractions of O₂. Furthermore, the observed upregulation of CHAC1 indicates the presence of protein stress (e.g., unfolded proteins due to thiol group oxidation) and an increased demand of glutathione to maintain intracellular ROS levels (73). OSGIN1 is a key regulator of both inflammatory and antiinflammatory molecules, underpinning that shielding gas composition can direct cellular responses, e.g., cytokine secretion (Fig. 5 b) (74, 75, 76). HaCaT cells are able to generate antioxidant enzymes like catalase, thioredoxin reductase, and glutathione peroxidase upon plasma treatment, e.g., via activation of the Nrf2 signal transduction pathway modulating cell proliferation, and migration ex vivo by upregulation of growth factor expression like VEGF, CSF2, etc. (12, 33, 40). In contrast, in monocytic immune cells no cellular changes in response to diverse compositions of the shielding gas were observed (77) underlying the fact that monocytes have a significantly increased antioxidant defense base level, which may be of high importance in their resistance to cold plasma-induced cell death (78).

Cytokines and chemokines are major regulators of inflammation in wound healing (79), being effectors and signal transducers of upstream pathways. Plasma treatment showed to have a strong impact on gene transcription and translation of these factors, and this was markedly modulated by the shielding gas composition. In general, hydrogen peroxide largely contributes to this observation (80). Yet, the increase in expression was mostly emphasized in oxygen-rich shielding gases, emphasizing repeatedly the role of non-H₂O₂ ROS as discussed above. Especially in the case of IL-6, shielding gas composition had a significant impact on both mRNA and full protein expression—inversely correlating with H₂O₂ deposition. Corroborating changes in IL-6 receptor complex consisting of IL-6 protein and interleukin 6-signal transducer (IL6ST/GP130/IL6-β) were detected under plasma treatment with high oxygen shielding. IL-6 is a proinflammatory cytokine, involved in UV-light and ROS-related signaling (81, 82). It also acts as an early wound response pleiotropic cytokine that regulates cell growth and differentiation and plays an important role in the immune response (83). Both singlet oxygen and ozone have been reported to trigger its cellular release, suggesting the potential prevalence of these species in the plasma-treated medium in the case of high oxygen shielding gas. This is further emphasized by the changes in HBEGF expression. Identically to the treatments without a shielding device, a release of HBEGF was not detectable (63), despite a strong upregulation of HBEGF mRNA similar to the IL-6 mRNA. The reasons are not known, yet HBEGF is a strongly folded protein containing seven cysteine residues between amino acids 108 and 142. Accordingly, misfolding due to thiol group oxidation may contribute to either a stop of HBEGF release or a loss of antibody specificity and consequent failure of the ELISA assay. Changes of isolated protein structure and/or functionality have been reported for cold plasmas (84). As discussed for the transcriptomic data, CHAC1 expression indicated protein stress due to oxygen-rich shielding gas plasma treatment, corroborating this notion. Finally, PTGS, a cyclooxygenase, is regulated by specific stimulatory events such as inflammation and mitogenesis underlining the importance of plasma-induced immunomodulation for wound healing processes.

Conclusion

It is one of the major aims of plasma source engineering to tune the plasma specifically for desired biological responses. Yet to do so, plasma medicine research must provide input on which species this may be. We here altered reactive species generation of the kINPen plasma jet by modulating the effluent’s ambient environment composition in a wide range. We were able to identify distinct reactions in human skin cells on a molecular biology level. Oxygen-dominated shielding showed the highest impact on not only the number of genes being regulated but also the level of their upregulation. Among those were targets associated with both antioxidant defense and cytokine signaling. Our results suggest that oxygen-shielded plasma induces a cell response more efficiently despite a decrease of hydrogen peroxide deposition, which was previously shown to be a major player in plasma-cell regulation, emphasizing the role of non-H₂O₂ primarily or secondarily generated ROS or RNS like singlet oxygen, ozone, or ONOO⁻. This provides ambient plasma shielding a potential role in manipulating biological outcomes and delivers input for plasma engineering.

Author Contributions

A.S., H.J., and A.B. designed the study. A.S. and A.B. performed gene expression experiments and analyzed data. H.J. measured concentration of ROS and RNS in liquids. K.W. and A.B. conducted ELISA experiments. A.S., S.B., H.J., K.-D.W., and K.W. wrote and edited the manuscript.

Editor: Catherine Galbraith.