Cell death induced by ozone and various non-thermal plasmas: therapeutic perspectives and limitations

Non-thermal plasma has been recognized as a promising tool across a vast variety of biomedical applications, with the potential to create novel therapeutic methods. However, the understanding of the molecular mechanisms behind non-thermal plasma cellular effects remains a significant challenge. In this study, we show how two types of different non-thermal plasmas induce cell death in mammalian cell cultures via the formation of multiple intracellular reactive oxygen/nitrogen species. Our results showed a discrepancy in the superoxide accumulation and lysosomal activity in response to air and helium plasma, suggesting that triggered signalling cascades might be grossly different between different plasmas. In addition, the effects of ozone, a considerable component of non-thermal plasma, have been simultaneously evaluated and have revealed much faster and higher cytotoxic effects. Our findings offer novel insight into plasma-induced cellular responses, and provide a basis for better controlled biomedical applications.

span from increased proliferation 14,19 to cell death by necrosis 20 or apoptosis 15,16 through to the formation of intracellular ROS 14,15 .
Most biomedical plasma applications require well defined and controlled interactions between non-thermal plasma and living cells. However, our understanding of the interactions between plasma and living cells, as well as the underlying mechanisms remains rather limited. As of yet, no comparative study has been thoroughly performed that observes the effects of different chemical compositions of plasma on the physiological functions of the cells. Moreover, nonthermal plasma is capable of producing significant amounts of ozone 8,9 , which is known to have very aggressive effects on cells 21,22 . However, the information concerning comparison of the biological effects of non-thermal plasma and ozone are basically lacking. Indeed, ozone can directly induce cell death via the accumulation of intracellular oxidants. It has been shown that oxidative stress induced by ozone may initiate redox-sensitive MAPK signalling resulting in different biological processes, including inflammation and cell death 21,23 . Interestingly, ozone produced by non-thermal plasma may remain for a long time even after turning off the plasma generator 24 . Hence, it is of great importance to compare and discriminate cellular effects triggered by plasma from those induced by ozone.
Therefore, the aim of this study is to investigate and compare the effects of two different non-thermal plasmas and ozone on physiological and pathophysiological cellular functions, as well as gaining information on the potential molecular targets of non-thermal plasma.

Results
Characterization of the plasma jet. We utilized a single jet nonthermal plasma system with 200 kHz and 600 V of applied AC power to study the effects of plasma treatment on mammalian cells 25 . A characteristic picture of the plasma system with an image of a plasma torch is shown on Figure 1a, b. Non-thermal plasma was generated in a specifically designed nozzle bearing a plasma generation module (Fig. 1e). The designed system enables the chemical composition of the plasma to easily change with the application of distinct gases (air, helium, nitrogen, etc.). Generally, the plasma jet system operated under a specific atmospheric pressure is able to generate a variety of chemically active species, particularly oxygen and nitrogen atoms 8,9,26 . Different types of plasma devices (dielectric barrier discharge (DBD) or direct plasma generators) are also known to be used in biomedical applications. As a rule, in such devices the plasma treated sample is connected to one of the electrodes and as a result has a non-zero potential. Beside this, a high voltage (a few kV) is usually applied to electrodes in such devices. Both of these factors are unfavourable for the study of interactions between plasma ions and cells because the electric field itself has a direct impact on cell functionality, e.g. cell membrane electroporation effect. In spite of this, DBD and direct non-thermal plasma sources are also used for a variety of biomedical applications 7,9,27 . However, to separate the direct effects of the electric field and plasma -for cell interaction it is better to use generators in which plasma is produced relatively far from the treated cells. To study differentiated and comparable cellular effects, we exploited two types of different plasmas, i.e. air and helium plasma. First of all, we used optical emission spectrum analysis to validate the plasma composition in terms of particles and radicals generated by the air and helium plasma systems (Fig. 1c, d), which can mediate the cellular effects of plasma. Optical emission spectroscopy, which was performed over a wide range of wavelengths from 200 nm to 900 nm, revealed distinct chemical differences between the two types of plasma (Fig. 1c, d). In an effort to monitor the reactive species environment produced by the jet array in the vicinity of the cell cultures, optical emission spectra were recorded with the spectrometer OceanOptics USB 4000-UV-VIS. The spectra were recorded in the longitudinal geometry with the device probe in an axial position at a distance of 10 mm from the nozzle outlet. Acquisition time was 1 sec. Helium and air plasmas generated by microdischarges in porous ceramics were investigated. Molecular OH, atomic (He, O) radicals, and other active species, e.g.   Fig. 1c). Helium plasma of a similar spectra was successfully applied for wound healing as reported in 28 . The spectrum of air plasma is similar to that described in 27 . The emission spectrum of the air plasma is shown in Fig. 1d. The peaks correspond to nitrogen molecules. A small peak at 400 nm corresponds to atomic oxygen. The total Helium or air flow through the micronozzle was set to 4 L min 21 for each gas. The emitting plasmas were spatially localized well, showing the applicability of the proposed plasma reactor for controlled treatments of cells and tissue. The gas temperature of such plasma jets was measured by a K type thermocouple at a 10 mm distance from the nozzle and the temperature did not exceeded 36uC over the entire treatment time. Thus, a limited gas temperature implies the absence of thermal damage on living cells or tissue during plasma treatments. These results imply that both plasmas may represent an important source of reactive oxygen and nitrogen species production.
Antimicrobial effects of plasma. Non-thermal plasmas have been successfully shown to inactivate microorganisms 29 . The presence of a variety of microbicidal active agents in plasma makes it an appropriate tool for microbial decontamination 30 . In the light of a growing antibiotic resistance microbial problem, plasma treatment may be a promising technique to overcome this challenge. Therefore, we tested antimicrobial effects of air and helium plasmas in comparison with ozone on two types of bacteria, S. aureus and P. aeruginosa; these bacterial strains are often used as models of antibiotic resistant Gram-positive and Gram-negative bacteria. As one can see from Figure 2, for all treatment variations we found statistically significant antibacterial effects compared to the untreated control. Indeed, the bactericidal efficacy depended on treatment time and differed between air and helium plasma. Interestingly, bacterial inactivation of both strains P. aeruginosa and S. aureus by air plasma and ozone was the same (Fig. 2a, b), having nearly 99.99% antimicrobial efficacy. However, dramatically lower antimicrobial efficacy was found for the helium plasma (Fig. 2c, d). Both ozone and air plasma had comparable bacterial inactivation effects on both Gram-positive and Gram-negative bacteria (Fig. 2a, b). In contrast, helium plasma demonstrated higher efficacy in killing Gram-negative rather than Gram-positive bacteria (Fig. 2c, d).
Analysis of cytotoxic effects induced by different plasmas and ozone. There are several studies that demonstrate the acute cytotoxicity of two distinct types of plasmas on living cells 14,16,24,31,32 . However, a comparison of cytotoxic effects of different plasmas on the same cellular model system has not been performed. Furthermore, non-thermal plasma has been shown to produce considerable amounts of ozone 8,9,14 . It is well known that ozone is an extremely reactive gas and exposure to it results in acute cytotoxic effects 21,22,33 . For superficial cancer treatment and wound healing it is important to know the time and dose dependent effects of plasmas on pathogenic (cancerous) cells and healthy tissue. Here we analyse the time-dependent effects of the air and helium plasmas as well as ozone on glioma cells and 3T3 fibroblasts. It is worth noting that two lines of research currently intensively investigated are cancer treatment and wound healing by non-thermal plasma 3 . Indeed, cell lines of different origin and varying degrees of differentiation have frequently been used to model wound healing and cancer condition since primary tissue cells cannot be readily expanded ex vivo. Further, 3T3 fibroblast cell cultures are commonly used as a model of wound healing 34 . Due to their advanced differentiation state, the C6 glioma cell line is often used as a surrogate of in vitro glioblastoma tissue 35 . Therefore, we utilized these two commonly accepted cell cultures as model systems for our experiments using non-thermal plasma. The comparison of cellular effects of ozone and non-thermal plasma is of particular interest practically. This prompted us to perform a comparative study of the cellular effects triggered by the two types of plasma and ozone. As expected the effects of plasma on the viability of both glioma and 3T3 cells were dose-and time-dependent (Fig. 3). There was no significant difference between air and helium plasma in terms of triggering cytotoxicity (Fig. 3). However, ozone appeared to induce higher instances of cell death (Fig. 3a). Even short term exposure triggered cell death, whereas air and helium plasma were unable to  induce any signs of cytotoxicity (Fig. 3a). Both air and helium plasmas showed time-delayed cytotoxic effects only 24 h post exposure (Fig. 3a). Interestingly, the effects of both plasmas and ozone were cell type specific. Indeed, the glioma cell line (Fig. 3b) was much less affected by ozone or either plasmas in comparison with fibroblasts (Fig. 3a). However, for sufficient killing of glioma cells it may be possible that a higher dose or longer exposure is needed, as far as it has been shown that the peak of glioma cell death occurs after 72 h post plasma exposure 36 . We found that ozone possesses a high level of toxicity and fasted cellular responses within 24 h post plasma exposure, furthermore we focused our studies on this time scale to better reveal differences between plasmas and ozone treatment.
Exposure of fibroblasts to air, helium plasmas or ozone induced only very low signs of early apoptosis, namely translocation of phosphatidylserine to the outer cell membrane leaflet, as measured by binding of FITC-labeled annexin V (Fig. 4a, b). Instead, a concomitant increase in membrane permeability, as shown by propidium iodide exclusion (Fig. 4a, b), was the predominant effect induced by both plasmas and ozone. Indeed, the permeability of the cell membrane affects dramatically on the ability of propidium iodide to enter a cell 37 . Having an intact plasma membrane living or early apoptotic cells are not permeable for propidium iodide. In contrast, late apoptotic or necrotic cells have plasma and nuclear membranes with decreased integrity, allowing propidium iodide to pass through the membranes 37 . Thus, our observations of predominant propidium iodide nuclear staining suggest either late stage apoptotic or necrotic cell death induced by both plasmas and ozone. Consistent with cytotoxicity data, ozone was more reactive in triggering propidium iodide incorporation into the cell nucleus in comparison with air and helium plasma (Fig. 4). It is worth noting that helium plasma produced less propidium iodide positive cells in comparison with air plasma (Fig. 4).
Changes in cell membrane integrity upon plasma treatment. It is well known that propidium iodide does not penetrate through the plasma membrane of viable or early stage apoptotic cells 38 ; only cells that have lost plasma membrane integrity are permeable to this dye. Therefore, we investigated in detail membrane damage induced by both plasmas and ozone utilizing atomic force microscopy (AFM), for details see section ''Methods''. In fact, short exposure of cells to air or helium plasma had no significant effect on the cell surface morphology compared to the untreated control cells (Fig. 5a). However, 4 h post exposure we were able to detect changes in membrane morphology in response to both plasma treatments (Fig. 5a). Conversely, ozone treatment induced immediate changes in the cell surface morphology, which remained unchanged for 4 h post exposure (Fig. 5a). Interestingly, the roughness of the outer cellular membrane surface increased following treatment of cells with both plasma types 4 h post exposure (Fig. 5b, c), indicating a loss of membrane integrity. Moreover, 4 h after treatment helium plasma induced significantly more profound effects on membrane roughness than air plasma (Fig. 5c). Furthermore consistent with previous data, ozone induced a greater increase in membrane roughness (Fig. 5b, c). Flow of gas alone, either helium or air, through the plasma device did not induce changes in membrane integrity or roughness (see Fig. 5d and Fig. S1 in the Supplementary Information).

Generation of intracellular ROS and RNS by different plasmas and ozone.
Previous studies have reported that non-thermal plasma is able to catalyze the generation of reactive oxygen (ROS) and nitrogen species (RNS) in living cells 14,15 . Indeed, high levels of ROS may induce damage to cellular structures and may finally lead to cell death 39 . Moreover, it has been analyzed that non-thermal plasma is potent in the production of extracellular ROS and RNS and mediates their delivery into the liquid phase of the cell culture WST-1 assay of 3T3 fibroblasts treated with air, helium plasma or ozone for indicated time periods, measured 0 or 24 h after exposure. The data were normalized to control values (no particle exposure), which were set as 100% cell viability. (b) Cell viability as detected by the WST-1 assay of C6 glioma cell line treated with air, helium plasma or ozone for indicated time periods. Analysis was carried out as described in (a). *P ,0.05 **P ,0.01, mean 6 SD, n53.
www.nature.com/scientificreports SCIENTIFIC REPORTS | 4 : 7129 | DOI: 10.1038/srep07129 medium, and subsequently their diffusion into cells 16 . Therefore, we were interested in the generation of intracellular ROS following plasma treatment.
To address this question, we evaluated ROS/RNS generation by air plasma, helium plasma and ozone using the ROS-sensitive fluorescent assay (cellular ROS/superoxide detection assay kit). We used two distinct fluorescent probes. One probe used was indicative of cellular production of different ROS/RNS types, the other was superoxide (O 2 2 ) specific. This allowed us to monitor changes in the total ROS/RNS level as well as specifically verify the level of superoxide. As shown in Figure 6 both plasmas and ozone induced a time-dependent ROS production in fibroblasts. In line with the data on cell death assessment and cell surface morphology, ozone induced a significantly higher production of total ROS/RNS than air and helium plasmas even after short time exposure (Fig. 6a, c). Moreover, both plasmas and ozone triggered a time-dependent accumulation of superoxide (Fig. 6b). Again ozone was much more reactive than air or helium plasma following an exposure time of 15 s (Fig. 6b). However, 4 h post exposure helium plasma triggered a higher accumulation of superoxide than ozone and air plasma after 60 s of treatment (Fig. 6d), as revealed by fluorescent microscopy.
Effects of plasma and ozone on lysosomal integrity and mitochondrial membrane potential. Mitochondria have been identified as a major source of cellular ROS generation 40 . Moreover, the superoxide anion is generated as a by-product of mitochondrial oxidative phosphorylation 40 . The generation and accumulation of superoxide is primarily associated with cellular toxicity and mitochondrial dysfunction 40 . It has been shown that lysosomal rupture and the associated release of lysosomal enzymes promote the mitochondrial production of oxidants 41,42 . Therefore, we investigated the putative role of air, helium plasmas and ozone on lysosomal destabilization. To address the possible lysosomal leakage induced by plasmas and ozone, we used the lysosomotropic dye acridine orange (AO). The accumulation of AO in a lysosomal compartment leads to red fluorescence, which dissipates when the dye leaks from this compartment into the cytosol 43 . Indeed, treatment of fibroblasts with air plasma and ozone did not induce a significant decrease in the AO fluorescence (Fig. 7a, b) indicating the absence of lysosomal permeabilization. Additionally, helium plasma had no significant effect on lysosomal permeabilization (Fig. 7a, b). However, lysosomal acidification was revealed by the increasing detectability of AO fluorescence 4 h after helium plasma exposure (Fig. 7b).
The depolarization of the mitochondrial membrane and concomitant change of the mitochondrial potential (DY m ) has been implicated as an early event in the process of cell death 30,44 . The induction of mitochondrial membrane surface charge dissipation and corresponding change of DY m is sufficient to trigger apoptosis or necrosis 30,44 . Thus we used the fluorescent dye JC-1 to investigate whether air, helium plasma or ozone may perturb mitochondrial function. JC-1 represents a cationic dye that exhibits a potentialdependent accumulation in mitochondria, which can be monitored by a shift in fluorescence emission from green to red using fluorometer 45 . As expected, both plasmas and ozone induced depolarization of the mitochondrial membrane as indicated by a decrease of the redto-green fluorescence intensity ratio as revealed by fluorometric measurements (Fig. 7c) and confirmed using flow cytometry (see   Figure 7d indicate that the ion distributions exhibited a nearly Gaussian lateral profile with the mean ion ranges and widths at half amplitude: for helium (He) ions 11.6 nm and 5.8 nm, and for nitrogen (N) ions 4.8 nm and 2.1 nm, respectively. Moreover, during the irradiation a significant amount of He ions accumulated at the water/membrane interface, while it is still unreachable for N ions. Qualitatively, the same picture would be expected for any heavy ions of air plasma as they have a shorter penetration range than He ions (Fig. 7d). Thus, the simulated indepth profiles of the plasma ion distributions in a medium-cell system give an indication as to the differences in cell surface morphology (Fig. 5). Specifically, irradiation of cells by lighter He ions leads to a longer penetration profile than that of heavier N or O ion irradiation, resulting in observed differences in membrane roughness between helium and air plasma. Moreover, differences in ion penetration profiles could be the underlying physical reason behind biochemical outcomes of different plasma types.
Scavenging intracellular ROS abolishes the cytotoxic effects of air, helium plasma and ozone. To confirm the role of ROS in the induction of cell death and subsequent events by both air and helium plasmas as well as ozone, we used the ROS-scavenger N-acetyl-Lcysteine. N-acetyl-L-cysteine, a derivative of the dietary amino acid Lcysteine, is known to be a powerful free radical scavenger and has profound cytoprotective effects against ROS-induced cell death 46,47 . Furthermore, our previous study demonstrated the applicability of Nacetyl-L-cysteine as a free radical scavenger 48 . As expected, by scavenging intracellular ROS/RNS we were able to antagonize the cytotoxic effects

Discussion
Nowadays it is becoming clear that non-thermal plasmas have undoubtedly great potential. However, clinical applications of nonthermal plasmas are restrained due to the lack of knowledge about the molecular mechanisms of plasma-living cell interactions. Moreover, there is no information comparing the consequences of different plasma treatments on cells with pure ozone treatment (an abundant component of non-thermal plasmas). Thus, we selected two distinct plasmas (air and helium) for our study.
Our comparative study showed air plasma and ozone to have very high bactericidal potency (Fig. 2a,b), both producing nearly 99.99% efficacy. Moreover, both air plasma and ozone inactivated different bacterial strains P. aeruginosa and S. aureus to the same extent (Fig. 2a, b). In contrast, bacteria treatment using helium plasma had a much lower effectivity, reaching at most only 8.71% of bactericidal efficacy. Furthermore, helium plasma showed higher efficacy in killing Gram-negative P. aeruginosa rather than Gram-positive S. aureus bacteria (Fig. 2c,d). Interestingly, the cell wall of Gram-negative bacteria is composed of a thin layer of peptidoglycans, which itself is surrounded by an outer membrane containing lipopolysaccharide. In contrast, the Gram-positive bacteria envelop does not contain an outer membrane but is surrounded by layers of peptidoglycan many times thicker than in Gram-negatives 49,50 . On the other hand, from a physical point of view, under plasma ion bombardment the penetration depth of light ions (like He) is larger than that of heavier ions (ions of air). Thus, the differences in bacterial cell wall thickness do not, however, explain the similar effects of air plasma in P. aeruginosa and S. aureus deactivation.
We suppose that different cell wall chemical resistance and/or respiration mechanisms may play a key role in the sensitivities of these bacteria to the plasmas. Tentative explanations for the different efficacies of P. aeruginosa and S. aureus deactivation by He and air plasmas are as follows: 1) membrane damage by ion and electron bombardment; 2) membrane perforation by etching due to highly reactive species; 3) electrostatic disruption of cell parts due to mutual electrostatic repulsion of the retained ions.
In this study we show that both, air and helium plasma as well as ozone produce intracellular ROS and RNS, accumulation of which leads to cell death. Our study is in line with previous studies [14][15][16] identifying the potential role of ROS and RNS in the underlying biological effects of plasma. Interestingly, ozone showed higher cytotoxicity in comparison to air and helium plasma, having drastic toxicity even immediately after exposure (Fig. 3a). This indicates that ozone possesses unspecific cellular responses. In line with toxicity data, ozone rapidly triggered ROS/RNS formation more efficiently than both plasmas (Fig. 4a, b). However, later helium plasma was more potent than ozone and air plasma in the generation of superoxide (Fig. 4d). Indeed, ROS and RNS are key effectors in signal transduction 51 . When ROS is produced during normal physiological processes, they are rapidly scavenged by antioxidant enzymes. Hence, excess of ROS can induce apoptotic/necrotic cell death through prolonged activation of JNK signalling 39,51 . Furthermore, ROS/RNS may lead to an increase in the permeability of the outer mitochondrial membrane and damage of lysosomal membranes 52 .
Interestingly, the glioma cell line developed much lower toxic effects in response to ozone and both plasmas in comparison with fibroblasts (Fig. 3b). This observation can be explained by the fact that cancer cells have an upregulated antioxidant capacity developed in response to intrinsic oxidative stress 53 . This leads to the resistance of cancer cells to ROS-dependant drugs 53 . Suppression of ROS elimination mechanisms in combination with the application of ROSgenerating agents (such as non-thermal plasma) could result in a potent strategy to enhance cancer cell cytotoxicity.
We have demonstrated that exposure to plasmas led to delayed alteration in membrane morphology, while ozone treatment induced immediate dramatic changes in the cell surface (Fig. 5a). Following the treatment of cells with both plasmas, the outer cellular membrane increased in roughness (Fig. 5b, c), resulting in disruption of membrane integrity. Similarly, an increase in roughness of HeLa cellular membranes 3-6 h post air plasma jet treatment was recently revealed by AFM 54 .
Numerous studies have shown that lysosomal rupture and subsequent lysosomal enzymes leakage may lead to the production of oxidants in the mitochondria 41,42 . In fact, air and helium plasma as well as ozone treatment of fibroblasts resulted in the depolarization of the mitochondrial membrane (Fig. 6c). However, our results demonstrate that upon air or ozone application there is no significant effect on lysosomal membrane permeability (Fig. 6a,b). Interestingly, helium plasma exposure resulted in considerable lysosomal acidification (Fig. 6b). Recently it has been found that suppression of the mammalian target of rapamycin (mTOR) activity leads to enhanced acidification of lysosomes and subsequent autophagy induction 55 . Simulations of ion penetration through the cell membrane revealed that heavy ions of atmospheric plasma have a shorter penetration range than He ions. Such ion distribution and related ionization loss may enhance the temperature of the first and second layers, lead to the acceleration of diffusion processes and creation of ROS. Generally, our results imply that notwithstanding of having similar cytotoxicity by air and helium plasma (Fig. 3a,b) there is discrepancy in the superoxide accumulation, lysosomal activity and plasma membrane roughness. These findings suggest that triggered signalling cascades are different upon air and helium plasma treatment of living cells. Furthermore, ozone treatment showed that this compound is extremely reactive, showing significant toxicity with short term exposure of fibroblasts (Fig. 3a). In contrast, both plasmas had delayed toxic effects on fibroblasts. Nonetheless, despite inducing acute cellular damage the antimicrobial effects of ozone were of the same order as air plasma (Fig. 2a,b). This finding indicates that the potential application of air plasma in infectious wound healing might be advantageous over ozone treatment as it possesses significantly lower side effects on living tissue.
Furthermore, we showed that ROS scavenger N-acetyl-L-cysteine reduced cytotoxicity caused by ROS production induced by air, helium plasma and ozone indicating that dietary supplementation with antioxidants might be a suitable strategy to reduce oxidative damage. These data reveal the crucial role of ROS/RNS in triggering intracellular signalling by plasma.
In summary, our study shows that the antibacterial efficacy of air plasma was comparable to ozone, whereas helium plasma possessed significantly lower antibacterial effects. Furthermore, ozone exhibits dramatically higher toxic effects on living cells than air or helium plasma. Indeed, the extent of cytotoxicity may grossly differ between fibroblasts and phenotypically distinct tumour cell lines. Moreover, despite having the same cytotoxicity upon air and helium plasma treatment, our findings imply that the triggered signalling cascades may be grossly different. This indicates that these processes depend critically on the chemical composition of the plasma jet. In addition, proteins of the mTOR signalling pathway, which are overexpressed or deregulated in human cancers, represent a target for cancer therapy 56 , and a link between the deregulation of lysosomal activity and mTOR inhibition has been revealed 55,57 . In light of these facts our data on lysosomal acidification by helium plasma are of great value and bring new perspectives for the development of novel therapeutic strategies of cancer.
These results imply that the cytotoxic effects of non-thermal plasma require more intensive study in terms of the identification of molecular targets of two distinct types of plasmas and that they should be considered in biomedical applications.

Methods
Physicochemical characterization of the plasma. To produce uniform non-thermal plasma for biological applications, we utilized the plasma setup shown in Figure 1a.
The input voltage was about 600 V, electric current 167 mA and the power was 100 W; such a high voltage supply resulted in electron energy of about 0.5 keV. The schematic construction of the plasma jet nozzle is shown in Figure 1e, consisting of a plasma generation module and cases. The high voltage electrode, a porous ceramic membrane and ground electrode form a plasma generation module. The gas supply was administered through a gas inlet followed by gas ionization in the pores of the ceramic membrane utilizing an electric field between two electrodes. The gas temperature at the tip of the plasma jet was measured using a thermocouple embedded in an optical spectrograph USB 4000 (Ocean Optics Inc.). The temperature remained 37-40uC during the cell treatment. The optical emission spectrum of the non-thermal plasma was measured using an optical spectrograph USB 4000 (Ocean Optics Inc.).
Plasma treatment. Non-thermal plasma was produced using an experimental custom-designed system with a nozzle array to treat living cells as illustrated by Figure 1e. In order to alter the chemical composition of the plasma, we used either an air or helium gas supply. Gas was supplied through a gas inlet and then was ionized in the pores of the ceramic membrane. To compare effects of plasma with ozone, cells were exposed to ozone produced by an ozone generator with input voltage 230 V, the power 12 W and ozone gas production rate 400 mg h 21 . Cells grown to 70% confluence were exposed to plasma from the device located 10 mm away, for 15, 30 and 60 s. Before the treatment, the medium was removed from culture wells and then 50 ml of medium was added to prevent cells from drying up. After the plasma treatment, the remaining medium was replaced with the fresh one.
Cultivation of bacteria and inactivation by plasma. To study bactericidal effects of plasma we used the Gram-negative Pseudomonas aeruginosa (ATCC 27853) and Gram-positive Staphylococcus aureus (ATCC 6538) bacterial strains (Czech Collection of Microorganisms (Brno, Czech Republic). The gelatine pellets containing the bacterial strains were incubated in 9 ml of liquid media (Tryptic Soy Broth, MecrotubeH, Merc, NJ, USA) at 35uC for 18 hours, and then diluted in a phosphate buffer (PBS) to a concentration 6 3 10 6 colony forming units (CFU)/ml. A volume of 0.8 ml of diluted bacteria suspension was spread onto the agar plate (Caso-Agar, MercoplateH, Merc) and exposed to plasma or ozone from the device located 10 mm away, for 15, 30 60 s. The plates were incubated overnight and the number of CFU in the 16 cm 2 inhibition zone was counted using ImageJ software (National Institutes of Health, Bethesda, MD, USA). The tested layers were tested in triplicates. Bactericidal efficacy was calculated according to the following formula: Apoptosis assay. Apoptosis was assessed via annexin V/propidium iodide staining. Cells were treated with different plasmas and ozone for 15, 30 and 60 s as indicated and incubated further for 4 h. Phosphatidylserine expression as an early sign of apoptosis was determined via fluorescence microscopy analysis buy the binding of fluorescein isothiocyanate-labeled annexin V (Sigma-Aldrich); propidium iodide was used to differentiate necrotic cells. Fluorescence images were recorded with a Zeiss Axioscope 2 microscope (Carl Zeiss AG). ImageJ software was used for image processing and fluorescent micrograph quantification.
Quantification of mitochondrial membrane potential. Plasma-treated cells were further incubated for 4 h to measure mitochondrial membrane potential (DY m ). After 4 h of incubation, cells were loaded with 1 mM JC-1 (Invitrogen), a lipophilic cationic fluorescence dye with a dual emission wavelength for 30 min, in order to analyze the depolarization of the DY m . At low concentrations (due to low DY m ) JC-1 is predominantly a monomer that results in a green fluorescence with emission of 530 nm. At high concentrations (due to high DY m ) the dye aggregates, yielding a red emission of 590 nm. Thus a decrease in the aggregate fluorescent count displays a mitochondrial membrane depolarization whereas an increase exhibits a hyperpolarization. Following staining, cells were analyzed using a fluorescent microplate reader (Tecan InfiniteH 200 PRO). Readings were done in quadruplicates. Additionally, flow cytometry was performed using a Apogee Flow Cytometer (Auto 40).
Assessment of lysosomal integrity by acridine orange (AO) release. Cells were labeled with 5 mg/ml AO in DMEM culture medium for 15 min at 37uC. After the rinsing of cells in complete medium, they were exposed to different types of plasma and ozone. Following this, cells were cultured at 37uC for 4 h and then, orange fluorescence intensity was measured using a fluorescent microplate reader (Tecan InfiniteH 200 PRO). Readings were done in quadruplicates. Fluorescence images were recorded with a Zeiss microscope Axioscope 2 (Carl Zeiss AG) and ImageJ software (NIH) was used for image processing. where d is the length of the resistive plasma jet column, e is the charge of the electron, A is the cross section of the plasma jet, Z R is the real part of the single plasma jet column impedance and m e 51.13?10 21 m 2 V 21 s 21 is the electron mobility in helium at atmospheric pressure (10 5 Pa). An estimation was made using the described equation for the measured impedance Z R 53600 Om, d510 mm, and A510 mm 2 gives the ion concentration, n55?10 17 m 23 .
Atomic Force Microscopy (AFM). The effect of cell exposure to air and helium plasma and ozone on changes in cell membrane morphology was studied by AFM. A glass slide with fixed cells was removed from PBS, rinsed with MilliQ-water several times and dried in the laminar flow-box for 30 min. All images were acquired with Atomic Force Microscope Dimension Icon (Bruker) as topographical scans in Peak Force tapping mode under ambient conditions, using etched silicon tips MPP (TAP150A, Bruker). The images were taken with a scan rate within the range of 0.25-0.9 Hz. The images were analyzed using NanoScope Analysis software (Bruker). The 10 3 10 mm scans were fitted to the plane or flattened in order to remove the tilt of rough data and then the roughness of the membrane was determined. Five cells of each treatment type have been analyzed by randomly measuring several areas over each cell surface.
Statistical analysis. Quantitative results are expressed as mean 6 SEM. Results were analyzed by multi-group comparison Fisher's LSD and Newman-Keuls tests. Differences were considered statistically significant at *P , 0.05.