*Article* **Cold Atmospheric Pressure Plasma Treatment Modulates Human Monocytes/Macrophages Responsiveness**

**Letizia Crestale 1,†, Romolo Laurita 2,† , Anna Liguori 2,3,†, Augusto Stancampiano 2,†,‡, Maria Talmon 1,†, Alina Bisag 2, Matteo Gherardi 2,3, Angela Amoruso 4, Vittorio Colombo 2,5,\* and Luigia G. Fresu 1,\***


Received: 30 September 2018; Accepted: 19 October 2018; Published: 29 October 2018

**Abstract:** Monocytes are involved in innate immune surveillance, establishment and resolution on inflammation, and can polarize versus M1 (pro-inflammatory) or M2 (anti-inflammatory) macrophages. The possibility to control and drive immune cells activity through plasma stimulation is therefore attractive. We focused on the effects induced by cold-atmospheric plasma on human primary monocytes and monocyte-derived macrophages. Monocytes resulted more susceptible than monocyte-derived macrophages to the plasma treatment as demonstrated by the increase in reactive oxygen (ROS) production and reduction of viability. Macrophages instead were not induced to produce ROS and presented a stable viability. Analysis of macrophage markers demonstrated a time-dependent decrease of the M1 population and a correspondent increase of M2 monocyte-derived macrophages (MDM). These findings suggest that plasma treatment may drive macrophage polarization towards an anti-inflammatory phenotype.

**Keywords:** cold atmospheric pressure plasma; dielectric barrier discharges; monocytes; monocytes-derived macrophages

#### **1. Introduction**

Monocytes are versatile mononuclear phagocytes involved in innate immune surveillance, establishment and resolution of inflammation. When activated, monocytes are recruited from the bloodstream to the site of inflammation where they differentiate into macrophages [1]. The local inflamed microenvironment drives polarization towards activated M1 macrophages, which display a pro-inflammatory phenotype and are involved in severe inflammation leading to tissue damage [2], or towards activated M2 macrophages [3], that display anti-inflammatory properties and are involved in tissue remodeling, wound healing, and efficient phagocytic activity [4]. M1 macrophages express high levels of major histocompatibility complex class II (MHC II) proteins, including the CD68 marker, and costimulatory molecules CD80 and CD86 [5]. They release reactive oxygen intermediates and several pro-inflammatory cytokines [6]. M2 macrophages are instead characterized by the expression of specific phenotypic markers, such as the mannose receptor-1 (CD206), the scavenger receptors CD163 and CD36 [7,8]. The molecular mechanisms underlying macrophage polarization have not been completely understood, because of the broad spectrum of stimuli affecting the process. However, it has been widely reported that this physiological process is altered in pathological conditions, for example in cancer or autoimmune diseases [9].

In this context, great interest has arisen towards physical stimuli that are able to modulate M1/M2 macrophage polarization [10]. In particular, the possibility to stimulate immune cells through cold atmospheric pressure plasma (CAP) treatment in order to control and drive their activity may pave the way to a vast field of medical applications. CAPs are characterized by a non-equilibrium in temperature between electrons and the heavy species of plasma (i.e., ions, excited species and neutrals) generated by the application of an electric field to a neutral gas. Besides presenting a negligible heat component, CAPs are characterized by several biological active components, such as ions, electrons, reactive oxygen (ROS) and nitrogen (RNS) species, UV radiation, playing a synergic action in the interaction of CAP with biological substrates [11]. Therefore, CAPs may be used for the treatment of cells and biological tissues by properly selecting the most suitable source and operating conditions for plasma generation to avoid any thermal damage [12,13]. Indeed, the Authors have previously shown an effect of CAP in a number of different eukaryotic cell types [14–17]. Immune cells have not been investigated yet thoroughly, although there are indications that their function may be affected. As an example, Kaushik et al. [18], reported the cytotoxic effects of CAP on monocytic lymphoma U937 cells and Bekeschus et al. [19] demonstrated the differential sensitivity of blood mononuclear cell subpopulations to plasma treatment. Focusing on the effects of plasma on monocytes, the plasma treatment of primary human monocytes can activate the pro-proliferative or pro-apoptotic intracellular signaling cascades, depending on plasma treatment time [20]. Several papers have focused on the plasma treatment of macrophagic populations [21–25]. In particular, the direct or indirect plasma treatment of macrophages can increase migration, an important immune cell function against diseases, and anti-tumor function [26,27].

In the present work, we focused our attention on the effects of a Dielectric Barrier Discharge (DBD) CAP source operated in open air on primary human monocytes and in monocyte-derived macrophages (MDM), evaluating their viability, ROS production and membrane markers expression.

#### **2. Materials and Methods**

#### *2.1. Monocytes Isolation and Differentiation*

The study was conducted in accordance with the Declaration of Helsinki. Ten healthy volunteers were enrolled, after approval of the Research Protocol by the Ethic Committee of Azienda Ospedaliera Maggiore della Carità, Novara, Italy (241CE), and informed written consent. Human monocytes were isolated from heparinised venous blood samples by standard technique of dextran sedimentation and histopaque (density = 1.077 g·cm<sup>−</sup>3, Sigma-Aldrich, St. Louis, MO, USA) gradient centrifugation (400× *g*, 30 min, room temperature) and recovered by thin suction at the interface, as previously described [28]. Purified monocytes populations were obtained by adhesion (90 min, 37 ◦C, 5% CO2) in serum free RPMI 1640 medium (Sigma-Aldrich) supplemented with 2 mM glutamine and antibiotics (Invitrogen, Carlsbad, CA, USA). After 90 min, medium was changed with RPMI added by 10% foetal bovine serum (FBS, Euroclone, Pero, Italy). Cell viability (trypan blue dye exclusion) was usually >98%. For differentiation into monocyte-derived macrophages (MDM), freshly isolated monocytes were cultured in 20% FBS-enriched medium for six days [29,30]. For plasma treatment cells were plated into 12-well plates in 1 mL of fresh medium.

#### *2.2. Plasma Treatment*

The CAP adopted in this study and reported in Figure 1A is a DBD source, already tested for biological applications [14,16,31,32] and consisting of a cylindrical brass electrode, 10 mm in diameter, having a hemispherical tip, with 5 mm of curvature radius. The electrode is coated with a 1 mm thick borosilicate glass (relative permittivity ε<sup>r</sup> = 4.7), as dielectric layer. The DBD plasma source was operated in open air and powered by a micropulsed generator producing high-voltage quasi-sinusoidal pulses with peak voltage (PV) of 25 kV, pulse repetition frequency (*f*) of 20 kHz, and a duty cycle of 7.5%. In order to enable the plasma generation between the high voltage electrode and the liquid surface, the multiwell plate was positioned onto a grounded counter-electrode (aluminum foil with thickness of 0.13 mm). Plasma treatments were performed by setting the gap between the tip of the plasma source and the surface of the liquid medium at 2 mm. The experimental setup employed for the treatment is schematically reported in Figure 1B. In all the experiments, the temperature of the medium after CAP treatment resulted below the threshold of cytotoxicity. After the treatment cells were incubated 2 h at 37 ◦C, and 5% CO2 before further analysis. The setup reported in Figure 1C was used for the time-resolved record of the plasma discharge electrical parameters. A high voltage probe (Tektronix P6015A, Tektronix, Beaverton, OR, USA) was used to measure the voltage waveform. The discharge current was measured by means of a current probe (Pearson 6585, Pearson Electronics, Palo Alto, CA, USA) mounted on the ground cable. Both signals were recorded with an oscilloscope (Tektronix DPO 40034, Tektronix) and subsequently elaborated to estimate the average power.

**Figure 1.** Dielectric Barrier Discharge (DBD) plasma source and setup. (**A**) Picture of plasma generated by a Dielectric Barrier Discharge source on cells; experimental setups employed for (**B**) the plasma treatment of monocytes and monocyte-derived macrophages and (**C**) the electrical characterization.

#### *2.3. Detection of Reactive Oxygen and Nitrogen Species in Plasma-Treated Medium*

The Amplex® Red Hydrogen Peroxide Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) and nitrate/nitrite colorimetric assay (ROCHE, Basel, Switzerland) were used, according to the manufacturer's protocol, to measure the concentrations of hydrogen peroxide and nitrites induced by plasma treatment in 1 mL of cell culture medium. Plasma treated medium was diluted 100 fold in phosphate buffered saline (PBS, a water-based salt solution containing 10 mM PO4 <sup>3</sup>−, 137 mM NaCl, and 2.7 mM KCl, at pH 7.4) immediately after treatment. The absorbances were measured photometrically with a microplate reader (Rayto, Shenzhen, China).

#### *2.4. Viability Test*

To assess potential plasma toxicity in monocytes and MDM, cell viability was evaluated using the methylthiazolyldiphenyl-tetrazolium bromide (MTT) assay. Cells (1 × 105 cells) were treated with plasma in the above described conditions. Two hours after the treatment, the medium was replaced by the MTT assay solution (1 mg·mL−1; 2 h, 37 ◦C 5% CO2; Sigma-Aldrich). Supernatant was removed and DMSO (Sigma-Aldrich) was added in order to dissolve the purple formazan; the absorbance was measured at 580 and 675 nm. Treatment times were 5 s, 10 s and 20 s for monocytes, and 10 s, 20 s and 30 s for macrophages.

#### *2.5. ROS/RNS and Superoxide Anion (O2* −*) Production*

2 h after the treatment, the medium was changed and reactive species production was evaluated. O2 − production was evaluated by the superoxide dismutase-sensitive cytochrome C reduction assay [33]. Briefly, 10<sup>6</sup> monocytes/macrophages treated by CAP were incubated with the cytochrome C (1 mg·mL−1; 2 h, 37 ◦C 5% CO2; Sigma-Aldrich) and then the supernatant was read at the spectrophotometer (Perkin Elmer Victor LightPerkin Elmer, MA, USA) at 550 nm. Indeed, cytochrome C reacting with the O2 − is reduced in ferrocytochrome C whose absorbance is detectable at 550 nm. The results were expressed in nmol cytochrome C reduced/10<sup>6</sup> cells/30 min, using an extinction coefficient of 21.1 mm [34], and correlated with the amount of superoxide anion produced by analyzed cells. Moreover, ROS/RNS and O2 − productions were also evaluated by flow cytometry analysis (FACS Calibur, BD, San Jose, CA, USA) using the Cellular ROS/Superoxide Detection Assay Kit (AbCam, Cambridge, UK) according to the manufacturer's instructions. The kit provides two fluorescent dye reagents: One (ROS/RNS, green) recognizing reactive species of both oxygen and nitrogen except for the superoxide anion that is detected by the second probe (O2 −, Orange). Results were analyzed by Flowing Software (version 2.5; PerttuTerho, Turku Centre for Biotechnology, Turku, Finland) and expressed as percentage of cells expressing ROS/RNS or O2 −. The cut-off for the analysis was based on non-stained sample. Moreover, we analyzed with the same software the mean fluorescence intensity (MFI) as indicator of the mean amount of O2 − produced by each single cell.

#### *2.6. Flow Cytometry Analysis*

Evaluation of surface markers expression was performed by multi-parametric analysis by flow cytometry (FACS Calibur, BD) and analyzed by Flowing Software (version 2.5; PerttuTerho, Turku Centre for Biotechnology). Monocytes and MDM were treated by CAP and after 2 h cells were mechanically detached and stained for FACS analysis. The following antibody panels were used: Anti-CD14 (APC, eBioscience, MA, USA), anti-CD16 (FITC, eBioscience), anti-CD36 (FITC, eBioscience), anti-CD86 (PE, eBioscience), anti-CD163 (PE, eBioscience), and anti-CD206 (PerCp, eBioscience), with 10.000 events acquired. The monocytes and MDM population were defined as CD14+ cells. Data were therefore expressed as the number of CD16+, CD86+, CD36+, CD163+ or CD206<sup>+</sup> cells over the number of CD14<sup>+</sup> cells (the gating strategy was shown in Figure 2). CD16 and CD86 are representative of M1 phenotype, while CD36, CD163 and CD206 of M2 phenotype. Comparison between treated and untreated cells was performed and data were expressed as percentage of positive events.

**Figure 2.** Gating strategy for flow cytometry (FACS) analysis of surface markers in monocytes and monocyte-derived macrophages (MDM). Population was first defined using forward scatter (FSC) and side scatter (SSC) to find viable cells and exclude debris. Then we gated CD14+ cells and on this population we analyzed the expression of the other markers. All the gates were firstly set using unstained control.

#### *2.7. Statistical Analysis*

Statistical analyses were performed using GraphPad Prism 5. Data are presented as mean ± SEM (standard error of the mean) of "*n*" independent experiments performed in triplicate on monocytes/macrophages. Data were analyzed by one-way ANOVA non-parametric (Kruskal-Wallis and Dunn's test). A value of *p* < 0.05 was considered significant.

#### **3. Results**

#### *3.1. Plasma Treatment and Viability Assay*

In Figure 3 is reported the temporal evolution of voltage and current waveforms. Concerning the current waveform is possible to observe multiple peaks both in the positive and negative half-periods of the voltage pulse that may be associated with multiple discharge events. The subsequently processing of waveforms of three independent experiments allowed to estimate an average power density of 1.7 ± 0.1 W.

In Figure 4, the concentrations of hydrogen peroxides and nitrites in 1 mL of culture media after plasma treatments are reported. CAP treatment induced the production of similar concentrations of nitrites (up to about 340 μM after 60 s) in both monocytes and MDM culture medium. On the other hand, hydrogen peroxide concentrations in monocyte culture medium were significantly higher compared to those produced in MDM culture medium.

**Figure 3.** Typical voltage (blue) and current (red) waveforms recorded during the plasma treatment.

**Figure 4.** Nitrite and hydrogen peroxide concentrations in plasma treated media. Data are expressed as mean ± SEM of three independent experiments (*n* = 9).

Plasma exposure times were selected by evaluating the cytotoxicity induced by CAP in both cell models. Monocytes resulted to be very sensitive to CAP exposure for periods longer than 20 s (data not shown). Therefore, as reported in Figure 5A, the range of exposure times were limited to 5–20 s, despite a significant reduction in cell viability was observed after only 10 s. MDM were at first treated for 20 s, 40 s and 60 s, highlighting a strong reduction of viability after 40 s and 60 s of plasma exposure (data not shown). Accordingly, treatment times of 10 s, 20 s, and 30 s were then performed. In these operating conditions, CAP did not affect MDM viability (Figure 5B), with only a non-significant reduction (20%) of cell viability being registered after 30 s of plasma treatment.

**Figure 5.** Cell viability after cold atmospheric pressure plasma (CAP) treatment. (**A**) Monocytes treated with CAP for 5 s, 10 s, and 20 s showed a significant reduction of viability already after 10 s of treatment. Data are expressed as mean ± SEM analyzed by Kruskal-Wallis test of 10 independent experiments (*n* = 30). \*\* *p* < 0.005, \*\*\* *p* < 0.001 vs. untreated control (Ctrl). (**B**) MDM treated by CAP for 10 s, 20 s, and 30 s.

#### *3.2. Plasma Treatment Effects on Monocytes*

#### 3.2.1. ROS Production by Treated Monocytes

Since human monocytes are phagocytes and release oxy-radicals upon challenge with appropriate stimuli, we first investigated CAP ability to affect superoxide anion (O2 −) production, using both an indirect and a direct method (superoxide dismutase-sensitive cytochrome C reduction assay and flow cytometry analysis, respectively), as reported in Materials and Methods. Basal O2 − production of monocytes amounted to 0.46 ± 0.018 nmol cytochrome C reduced/106 cells, while after 20 s of CAP treatment, this was significantly increased by about 0.15 nmol (Figure 6A). This small, but significant, increase might be a direct consequence of the reduced viability observed. The cytofluorimetric assay confirmed this result demonstrating that the percentage of cells producing radical oxygen species moved from about 19 ± 4% of untreated cells to 47 ± 1% after 20 s of CAP treatment (Figure 6B). Moreover, the analysis of MFI, that in this instance represents the amount of superoxide anion produced by cells, revealed a progressive increase (Figure 6C) that become significant at 10 s, confirming that CAP treatment induced not only the oxidative burst in a greater number of cells, but also each cell to produce a higher amount of O2 <sup>−</sup>. PMA (positive control of oxidative burst induction) 10−<sup>6</sup> M was used as positive control of oxidative burst induction [35].

ROS/RNS evaluation in monocytes after CAP treatment could not be performed, since the percentage of expressing cells in the untreated group in our test conditions was high and therefore poorly responsive to treatments (untreated monocytes 79%±0.7 vs. monocytes treated with PMA 88%±2).

**Figure 6.** Effect of plasma treatment on monocytes superoxide anion production. (**A**) Monocytes were exposed to CAP for 5 s, 10 s and 20 s and results are expressed as levels of nmol reduced cytocromeC/10<sup>6</sup> cells. Data are expressed as mean ± SEM of five independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparisons. \* *p* < 0.05 vs. untreated cells (Ctrl); (**B**) Cytofluorimetric assay to assess positive monocytes to the superoxide anion staining. NS, unstained control used to set acquisition parameters; Ctrl, untreated control; PMA, positive control of oxidative burst induction. Results are expressed as percentage mean ± SEM of three independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparisons. (**C**) Mean fluorescence intensity (MFI) analysis expressed as mean ± SEM of three independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparison. \* *p* < 0.05; \*\* *p* < 0.005 vs. corresponding control.

#### 3.2.2. Surface Marker Expression

The expression of specific monocyte surface markers was then analyzed (Figure 6). Freshly isolated monocytes treated with CAP showed a significant reduction (about 40%) of CD86, CD36, CD163 and CD206 already after 5 s of treatment, (Figure 7) while the reduction of CD16 expression became significant after 10 s of plasma exposure.

**Figure 7.** Effect of plasma treatment on membrane markers expression of monocytes. CD14<sup>+</sup> cells, treated and untreated (Ctrl), were stained with the indicated antibodies and analyzed by flow cytometry. Results are expressed as the percentage of positive events for each marker on the total of CD14+. Data are expressed as mean ± SEM of 10 independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparison. \* *p* < 0.05; \*\* *p* < 0.005; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001 vs. corresponding control. We have previously shown that untreated monocytes do not change marker expression in a two-hour incubation and therefore the untreated control also depicts the baseline expression [36].

#### *3.3. Plasma Treatment Effects on Monocytes-Derived Macrophages (MDM)*

#### 3.3.1. ROS Production by MDM

The production of both O2 − and other reactive oxygen and nitrogen species (ROS/RNS) in MDM exposed to CAP is shown in Figure 6. MDM had a basal production of O2 <sup>−</sup> of about a 0.39 ± 0.04 nmol of cytochrome C reduced/10<sup>6</sup> cells (Figure 8A), and CAP treatments of 10, 20 and 30 s were ineffective in inducing an oxidative burst, as demonstrated by the unchanged percentage of positive cells to the ROS/RNS staining (Figure 8B). In contrast, the number of positive MDM to the O2 − staining was significantly increased after the longest time treatment (Figure 8C). It is interesting to note that after plasma treatment we can observe the onset of a peak of low-expressing O2 − cells (represented by the circled peak in the histogram), that gradually increases in parallel with CAP-treatment time. This can be correlated with the emergence of a new population of MDM (in treated cells vs. Ctrl) able to produce basal levels of O2 −. Moreover, it is important to note that the MFI did not increase (Figure 8D).

#### 3.3.2. Surface Marker Expression

As shown in Figure 9A, in MDM exposed to CAP the expression of the single surface markers was not significantly influenced by CAP treatment, except for CD36, whose expression significantly increased, and CD16, that decreased already after 10 s. This result is in line with the previously presented data: In fact, CAP treatment did not induce the MDM respiratory burst nor a decrease in cell survival even after the longest treatment time. The possibility that CAP exposure could drive polarization of MDM versus M1 or M2 phenotypes was investigated. In fact, freshly isolated monocytes, cultured in 20% FBS-enriched medium, spontaneously differentiate to MDM, defined as M0, an intermediate phenotype (not M1 nor M2), that under different stimuli can shift to one phenotype or to the other. We performed a co-expression analysis for the specific markers of M1 (CD16/CD86) and M2 (CD163/206) on the total of CD14<sup>+</sup> population. As demonstrated in Figure 9B, there was a time dependent reduction of M1 population, defined as CD14+CD16+CD86+, becoming significant at 30 s, with a correspondent increase to the M2 population, defined as CD14+CD163+CD206+.

**Figure 8.** Effect of plasma treatment on reactive species production in MDM. (**A**) Superoxide anion production. Cells were irradiated by CAP for 10, 20 and 30 s, and results are expressed as levels of nmol reduced cytocromeC/10<sup>6</sup> MDM. Data are expressed as mean of five independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparison; (**B**) Cytofluorimetric assay to assess MDM positive to the reactive oxygen (ROS)/RNS detection probe staining after 10, 20 and 30 s of treatment with CAP. Results are expressed as percentage mean of MDM producing ROS/RNS ± SEM of three independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparison; (**C**) Cytofluorimetric assay to assess MDM positive to the superoxide anion detection probe staining after 10, 20 and 30 s of treatment with CAP. Results are expressed as percentage mean of MDM producing O2 <sup>−</sup> ± SEM of three independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparison. \* *p* < 0.05 vs. untreated cell (Ctrl); (**D**) MFI analysis expressed as mean ± SEM of three independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparisons. \* *p* < 0.05 vs. corresponding control. NS, unstained control used to set acquisition parameters, Ctrl, untreated control; PMA, positive control of oxidative burst induction.

**Figure 9.** Effect of plasma treatment on MDM phenotype. (**A**) Cyofluorimetric analysis of MDM stained with anti-CD14, anti-CD86, anti-CD36, anti-CD163 and anti-CD206. Results are expressed as the percentage mean of positive events for each marker on the total of CD14<sup>+</sup> <sup>±</sup> SEM of 10 independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparison. \* *p* < 0.05, \*\* *p* < 0.005 vs. untreated control (Ctrl); (**B**) Cytofluorimetric analysis of M1 and M2 populations represented by the co-expression of CD14/CD16/CD86 and CD14/CD163/CD206 respectively. \* *p* < 0.05 vs. untreated cell (Ctrl). Data are expressed as mean ± SEM of 10 independent experiments analyzed by Kruskal-Wallis test and Dunn's test for multiple comparisons. We have previously shown that untreated MDM do not change marker expression in a two-hour incubation and therefore the untreated control also depicts the baseline expression.

#### **4. Discussion**

In this study, the effect of a microsecond pulsed DBD on human monocytes and monocyte-derived macrophages was investigated.

The CAP treatment of monocytes and MDM culture medium induced the production of nitrites and hydrogen peroxides. The lower the concentration of FBS, the higher the concentration of hydrogen peroxide, while the concentration of nitrites resulted similar in both treated media. This difference can be ascribed to the differences between MDM and monocytes culture medium in terms of FBS concentrations, scavenger of hydrogen peroxide, but not of nitrites [37]. Moreover, the longer the treatment time, the higher was the concentration of both species in the treated media and the reduction of cell viability for both monocytes and MDM. Similar findings have been reported for several cell lines [16,20,38].

Focusing on monocytes, CAP treatment up to 20 s induced cells to produce of O2 − and a decrease of surface markers. This effect is unlikely to be due to a change in cell polarization, but is most likely a physiochemical modification, since compensatory increases in cell surface markers were not observed to counteract these decreases. While we did not investigate this further, the decrease of marker expression might be explained by a direct membrane damage (for example peroxidation and subsequent loss of membrane fluidity and elasticity [39]), to a direct protein oxidation [40] or by other mechanisms, which we have nonetheless not evaluated further. Anyway, it is noteworthy to highlight a slight recovery of the surface marker expression after the impressive5sreduction. This could be explained with a rapid anti-oxidative response of cell to the insult [41], but this defense mechanism was not sufficient to restore membrane integrity and keep cell survival. The fact that monocytes are significantly more sensitive to CAP compared to MDM would be in accordance with Bundscherer et al. [42] who demonstrated significant differences between different immune cell types regarding survival after plasma treatment.

On the other hand, CAP treatment for up to 30 s of MDM did not cause an increase in O2 − production, but only induced a greater number of cells to produce the superoxide anion. As a possible consequence, MDM, terminally differentiated cells, resulted more resistant to CAP treatment. Indeed, CAP treatment did not affect significantly MDM survival. We suggest that these results are strictly connected to the absence of an increase of ROS/RNS and the augmented percentage of O2 −-producing MDM. In fact, as demonstrated by Zhang et al. [43] the polarization versus M2 phenotype is sustained by a proper amount of O2 −. Indeed, when monocytes are induced to differentiate there is an increased superoxide anion production that triggers the biphasic ERK activation that has a pivotal role for M2 differentiation. In our experiments, CAP stimulated more M0 to secrete basal amounts of superoxide necessary to polarize versus M2, that were represented by the new population previously described in Figure 8C. Moreover, marker expression modifications in MDM were reputed as a true sign of polarization change, since CD16 decrease was counter-balanced by an increase in CD36, while CD163 and CD206 were not affected. In the present study, M2 subsets most represented in CAP treated MDM have not been evaluated. In the future, it will be interesting to explore this aspect as the different M2 sub-populations are involved in different physiological and pathological processes [4,44,45]. This information will therefore provide clues on the clinical applications of CAP.

#### **5. Conclusions**

In conclusion, our results suggest that CAP treatment may be able to selectively modulate the effect of some cells over others. Furthermore, CAP could drive macrophage polarization supporting the idea that at the proper operating conditions of plasma treatment it could be possible to direct macrophages versus an anti-inflammatory phenotype. At the same time, our results also suggest the differences in terms of viability of macrophages compared to monocytes after CAP exposure may be due to the different concentration of FBS between the cell culture media. In fact, as reported by Yan et al. [46], it is possible to harness the medium to kill glioblastoma cells by altering the concentration of the serum demonstrating therefore the significant role of media in CAP treatment. In prospect, our results may suggest that CAP technology may find clinical applications in those pathological settings in which macrophages play a prominent role, and in which it can be delivered locally, for example in wound healing, treatment of non-metastatic melanoma difficult to surgically remove or in topical applications for autoimmune disorders, such as psoriasis.

We acknowledge that our results should be read in light of the following limits: (a) We have evaluated changes in monocytes and MDM 2 h after CAP treatment, but events occurring immediately after treatment might be relevant; (b) we have not fully characterized the M2 sub-populations represented in CAP treated cells, neither by FACS nor by real time PCR. This last point, in the future, will yield more precise indications on the possible therapeutic applications of CAP.

**Author Contributions:** L.C., A.A., M.T., R.L., A.L., A.S., A.B.: methodology and investigation A.A., V.C., M.G.: conceptualization, L.G.F., R.L., A.L., A.S.: writing—original draft preparation, L.G.F., M.G., V.C.: supervision, formal analysis, M.G., V.G, L.G.F. Writing—review & editing.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Possible Mechanism of Glucose Uptake Enhanced by Cold Atmospheric Plasma: Atomic Scale Simulations**

#### **Jamoliddin Razzokov \* , Maksudbek Yusupov and Annemie Bogaerts**

Research Group PLASMANT, Department of Chemistry, University of Antwerp, Universiteitsplein 1, B-2610 Antwerp, Belgium; maksudbek.yusupov@uantwerpen.be (M.Y.); annemie.bogaerts@uantwerpen.be (A.B.)

**\*** Correspondence: jamoliddin.razzokov@uantwerpen.be; Tel.: +32-03-265-2382

Received: 9 May 2018; Accepted: 6 June 2018; Published: 8 June 2018

**Abstract:** Cold atmospheric plasma (CAP) has shown its potential in biomedical applications, such as wound healing, cancer treatment and bacterial disinfection. Recent experiments have provided evidence that CAP can also enhance the intracellular uptake of glucose molecules which is important in diabetes therapy. In this respect, it is essential to understand the underlying mechanisms of intracellular glucose uptake induced by CAP, which is still unclear. Hence, in this study we try to elucidate the possible mechanism of glucose uptake by cells by performing computer simulations. Specifically, we study the transport of glucose molecules through native and oxidized membranes. Our simulation results show that the free energy barrier for the permeation of glucose molecules across the membrane decreases upon increasing the degree of oxidized lipids in the membrane. This indicates that the glucose permeation rate into cells increases when the CAP oxidation level in the cell membrane is increased.

**Keywords:** cold atmospheric plasma; reactive oxygen and nitrogen species; glucose uptake; molecular dynamics; permeation free energy

#### **1. Introduction**

Diabetes is a chronic disease related to an abnormal increase of the glucose level in the blood. Approximately 75% of glucose in the body is consumed by skeletal muscle cells that are stimulated by insulin [1,2]. Failure of glucose uptake leads to the presence of a high level of sugar in the blood. This is due to one of two mechanisms or a combination of both, i.e., (a) disproportional production of insulin or (b) inadequate sensitivity of cells to insulin [3–5]. Most of the currently available pharmaceuticals are inefficient in the treatment of diabetes, thus alternative insulin independent healing methods need to be found in order to increase glucose uptake by muscle cells.

Recently, Kumar et al. investigated the impact of cold atmospheric plasma (CAP) on the regulation of glucose homeostasis [6]. The experimental results showed that the glucose uptake is significantly enhanced in skeletal muscle cells after plasma treatment, exhibiting the beneficial effects of CAP. Furthermore, higher levels of intracellular Ca2+ and reactive oxygen species (ROS) in CAP treated cells were observed. An increase in intracellular ROS and Ca2+ ions helps to increase the glucose uptake by skeletal muscle cells [7,8]. Increases in the Ca2+ ion concentration as well as the uptake of middle-sized, membrane-impermeable molecules after direct CAP exposure were also observed in [9]. The authors attributed these effects to an increase in cell permeability caused by CAP-generated electric stimulation and the delivery of OH radicals into cells. Moreover, Vijayarangan et al. investigated CAP parameters and conditions for drug delivery across HeLa cells [10]. They determined the efficient treatment time (i.e., low toxicity) and the range of frequencies with an optimal number of pulses as key parameters for the cell membrane permeability [10]. In addition, Leduc et al. determined the maximum radius for

macromolecules that are capable of ingressing into HeLa cells, applying a specifically designed CAP source [11]. Hence, an increase in the cell membrane permeability might play an important role in the delivery of the abovementioned species into the cell. The underlying mechanisms, however, still remain unclear, and need more thorough investigation. Computer simulations might be a useful tool to gain insight into the atomic level processes.

In the context of plasma medicine, we have already performed several computational studies, by means of molecular dynamics (MD) simulations, using a phospholipid bilayer (PLB) as a model system for the plasma membrane. Specifically, we studied the effect of lipid oxidation on phosphatidylserine translocation across the plasma membrane, which plays a vital role in apoptosis signaling [12]. Furthermore, we investigated the ROS oxidation of the head groups and lipid tails in the membrane [13], the permeation of ROS across oxidized and non-oxidized membranes, including the synergistic effect of plasma oxidation and electric field [14], and the hampering effect of cholesterol [15,16]. In general, our investigations showed that oxidation of the membrane leads to an increase in its fluidity and permeability to ROS, thereby affecting the abovementioned processes.

In this study, we investigate glucose translocation across native and oxidized membranes in order to provide a possible explanation for the mechanism of glucose uptake observed in previous experiments. In particular, we perform MD simulations to calculate the free energy barriers for glucose transport through oxidized and non-oxidized membranes. Comparison of the latter shows the effect of CAP oxidation on the glucose transport across the membrane, which might explain the increased level of glucose uptake after the plasma treatment of cells, as observed experimentally.

#### **2. Computational Details**

#### *2.1. Simulation Setup*

We performed MD simulations in order to study the glucose translocation across both intact and oxidized PLBs. The PLB is considered to be a simple model system that represents the eukaryotic cell membrane, since it determines the thickness of the bilayer. A schematic representation of the intact PLB is given in Figure 1a. It consists of 128 phospholipids (PLs), covered by 6000 water molecules, organized in two lamellae (i.e., 64 PLs, with a corresponding water layer at the top, and 64 at the bottom). As the PL molecule, we used 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), as depicted in Figure 1b. Glucose molecules were randomly placed in the xy-plane of the upper side of the PLB, i.e., in the water phase, as well as in the lipid tail region, see Figure 1a (and see below for more details).

Simulations were carried out using the GROMACS package (version 5.1) [17], by applying the GROMOS (43A1-S3) force field [18]. To study the effect of oxidized PLs, we used aldehyde oxidation products (see Figure 1b, DOPC–ALD), which are one of the major oxidation products [19]. Note that CAP yields a cocktail of reactive species and thus, could possibly form a range of products upon oxidation of the PLB, but the formation of aldehyde groups (i.e., DOPC–ALD) was prominently observed in CAP-treated vesicles [13]. These oxidation products were also used in our recent simulation studies on various properties of the PLB [13–15]. The force field parameters for the aldehyde groups in the oxidized PLs were obtained from [20] and the parameters of glucose were based on [21]. The Packmol package was employed to create initial configurations of the intact and oxidized PLB systems [22]. Two aldehyde oxidation products were created from the non-oxidized (i.e., native) PLBs containing 128 PLs, by replacing 32 and 64 DOPC molecules with DOPC–ALD, corresponding to concentrations of 25% and 50%, respectively. The presence of 25% and 50% DOPC–ALD in the oxidized PLB does not necessarily correspond to experimental oxidation levels, but performing the calculations for a lower degree of oxidation would require excessive calculation times to give the same qualitative conclusions. Hence, this oxidation degree was high enough to observe the effect of oxidation within an acceptable calculation time, but low enough so that pore formation did not occur within the simulated time scale [15]. Thus, in total, three model systems were studied in our simulations, i.e., native (0%) as well as aldehyde oxidized (25% and 50%) PLBs. For each system, we created four different structures

(e.g., four native PLBs) extracted from the last 40 ns trajectory of the 200 ns equilibration run, with a time interval of 10 ns, in order to obtain an average free energy profile (FEP) of glucose transition across each system (see next section). All structures (in total 12, including the native PLB) are initially optimized using the steepest descent algorithm and then equilibrated for 200 ns by so-called NPT simulations (i.e., at a constant number of particles (*N*), pressure (*P*) and temperature (*T*)), at 310 K and 1 bar, employing the Nose–Hoover thermostat [23] with a coupling constant of 0.2 ps as well as the semi-isotropic Parrinello–Rahman barostat [24] with a compressibility and coupling constant of 4.5 × <sup>10</sup>−<sup>5</sup> bar−<sup>1</sup> and 0.1 ps, respectively. For the non-bonded interactions, a 1.1 nm cut-off was applied. The long range electrostatic interactions were described by the particle mesh Ewald (PME) method [25], using a 1.1 nm cut-off for the real-space interactions in combination with a 0.15 nm spaced grid for the reciprocal–space interactions. Subsequently, a series of umbrella sampling (US) simulations [26] were run for 20 ns, applying, again, the NPT ensemble (see next section), of which the last 10 ns was used for further analysis. In all simulations, a time step of 2 fs was used. Periodic boundary conditions were applied in all three directions.

**Figure 1.** (**a**) Intact or native 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) phospholipid bilayer (PLB), together with two glucose molecules in the water and lipid tail regions. For the sake of clarity, the N and P atoms of DOPC are shown with larger beads. The bilayer center is indicated by the red dashed line; (**b**) Schematic illustration of native (DOPC) and oxidized (DOPC–ALD) PLs. The head group consists of choline, phosphate and glycerol, whereas the lipid tails are two fatty acid chains.

#### *2.2. Umbrella Sampling (US) Simulations*

We performed US simulations in order to determine the free energy profiles (FEPs) of glucose translocation across the native, as well as the oxidized, PLBs. In total, we obtained 12 FEPs for all model systems (i.e., four for the native, four for the 25% oxidized, and four for the 50% oxidized PLBs, see previous section) and averaging was performed over four FEPs for each system. For the calculation of each energy profile, we extracted 32 to 36 windows (36 for the native, 34 for the 25% and 32 for the 50% oxidized PLB, due to a decrease of the bilayer thickness) along the *z*-axis, which were separated by 0.1 nm. These windows were obtained by pulling glucose molecules against the *z*-axis direction (see Figure 1a) for 500 ps, using a harmonic bias between these glucose molecules and the center of mass of the PLB, with a force constant of 2000 kJ·mol−1·nm−<sup>2</sup> and a pulling rate of 0.01 nm·ps<sup>−</sup>1. Note that, in principle, slow pulling rates can be used in the pulling simulations. However, dragging the glucose from water phase into the center of the PLB requires a long computation time. Higher pulling rates can solve this issue, but they can lead to significant disturbances in the PLB. Thus, by using an

appropriate (standard) pulling rate, we performed short pulling simulations to save computation time with minimum perturbations on the PLB. One of the glucose molecules (i.e., the glucose in the upper water phase, see Figure 1a) was pulled until it reached the center of mass of the PLB, while the second one moved towards the lower water phase from the center of the bilayer. Thus, each US simulation involved two glucose molecules. In this way, we saved computational resources, thereby increasing the number of sampling points. Note that these two glucose molecules were separated from each other at least by 3 nm in the z direction; hence, there was no interaction between these two molecules. In principle, we could have used three glucose molecules in each US simulation. However, in order to cause minimal disturbance to the PLB system, we chose two glucose molecules instead of three. It should be mentioned that, in reality, adsorption or chemical reaction of glucose might take place in the PLB. However, these processes cannot be studied with conventional non-reactive MD simulations, due to the limitations in time and reactivity. Nevertheless, the US simulations can predict how often the glucose transport occurs across the PLB before and after oxidation, through calculation of the FEPs.

As mentioned above, we extracted 32 to 36 US windows from our 500 ps pulling simulation. Hence, 32 to 36 US simulations were performed to construct a single FEP. Each US simulation lasted 20 ns, and the last 10 ns were used for analysis, i.e., to collect the US histograms and calculate the FEPs. Note that the pulling simulation trajectory was used only to extract windows/frames for the further US simulation to obtain the FEPs. During the pulling simulations, the glucose dragged water molecules with it into the hydrophobic core of the bilayer, but these molecules escaped from the hydrophobic core within the initial 10 ns of US simulation. Thus, the last 10 ns of the US simulation was an adequate time for calculating the FEPs, as there were not any water defects or hydration layers in the hydrophobic part of the membrane. In each US simulation, the glucose molecules were able to freely travel in the xy-plane, but their movement in the z-direction was restricted by applying a harmonic bias with a spring constant of 2000 kJ·mol−1·nm<sup>−</sup>2.

The FEPs were constructed using a periodic version of the weighted histogram analysis method (WHAM) [27], as implemented in GROMACS. We analyzed our simulation systems to identify underestimated possible "hidden barriers" based on previous literature [28]. We did not define any hidden barriers because in our US simulations, the data was sufficiently sampled. Indeed, in the four model systems used, the glucose was randomly positioned in the xy-plane to escape trapped metastable states and to allow estimation of the error bars that are associated with choosing the initial model systems. As noted above, the final energy profiles were obtained by averaging four independently-built FEPs for each system which differed from one another based on their starting structure to allow for some statistical variation. The uncertainties associated with the FEPs were obtained by calculating the standard deviations between these four FEPs for each system. Thus, in total, 410 US simulations were performed for the calculation of the FEPs.

#### **3. Results and Discussion**

The US MD simulations allowed us to elucidate the glucose translocation across the native and oxidized PLBs which gave insight into the possible mechanism of glucose ingress triggered by plasma oxidation of the cell membrane. Figure 2 illustrates the symmetric FEPs of glucose transfer across native (0%) as well as oxidized (25% and 50%) PLBs.

It is clear that, in all three cases, the ΔG started to drop when glucose entered the hydrophilic head group region from the water phase (see Figure 1b). This means that the head group of the PLB is energetically the most favourable region, showing the minimum energy for glucose transfer. Moreover, this energy minimum slightly shifted towards the centre of the bilayer in the case of oxidized PLBs. We know from our previous studies that when lipid tails are oxidized, this eventually leads to a drop of the bilayer thickness [13,14]. In other words, the shortened tail and aldehyde products move towards the water phase, thereby increasing the area and fluidity of the PLB. Consequently, this results in a decrease in PLB thickness, e.g., the native DOPC bilayer has a thickness of around 3.89 nm, whereas, in the case of a 50% DOPC–ALD bilayer, the thickness decreases to 3.33 nm [14]. Therefore, the free

energy minima in Figure 2 are found at around |z| = 1.9 nm and |z| = 1.76 nm for native and 50% oxidized PLBs, respectively.

Continuing the motion of glucose towards the hydrophobic tail region leads to an increase of the free energy for translocation, representing the role of the membrane as a permeation barrier. In case of the native PLB, the free energy barrier for the permeation of glucose was 51 ± 3 kJ/mol, which is within the range of experimental results [29,30]. On the other hand, for the 25% and 50% oxidized PLBs, this barrier for the translocation of glucose decreased to values of 37 ± 4 kJ/mol and 28 ± 4 kJ/mol, respectively (see Figure 2). Hence, the obtained results show that the free energy barrier for the transport of glucose molecules across the PLB decreases when the oxidation degree is increased. This, in turn, leads to an increase in the probability of glucose permeation to the cell interior. The obtained simulation results can be correlated with the experimental observations [6,9] as the plasma treatment of cells most probably gives rise to oxidation of the cell membrane, thereby increasing the glucose (or other middle-sized molecules, as well as Ca2+) translocation rate.

**Figure 2.** (**a**) Symmetric free energy profiles for the translocation of glucose across the native and oxidized PLBs. A PLB is schematically illustrated in the background, to indicate the position of the water layer, the head groups and the lipid tails. For clarity, the zoomed extrema of the profiles are shown in (**b**,**c**). Errors associated with the umbrella sampling (US) calculations are depicted in pale color.

Note that our simulations only provide one possible explanation for the increased level of glucose [6] in cells after plasma treatment, while other mechanisms might play a role as well [9–11]. Therefore, further investigations should be performed to obtain a more complete picture of the plasma effect on the membrane permeability. This might be achieved, for instance, by silencing the glucose transporter proteins (or GLUTs) and measuring the concentration of intracellular glucose molecules after CAP treatment. The latter would clearly reveal the role of cell membrane permeability (induced by CAP oxidation of lipids) in transporting glucose molecules.

#### **4. Conclusions**

We performed MD simulations in order to understand the possible mechanisms of cellular glucose uptake induced by CAP treatment. The obtained free energy profiles of glucose across native and oxidized membranes revealed that the plasma induced oxidation of the membrane lipids decreases the barrier for translocation of glucose across the membrane. This, in turn, might possibly explain the increased concentration of glucose observed by experiments using CAP. Hence, this computational study provides an atomic level insight into the possible process of glucose permeation through the membrane.

**Author Contributions:** Formal analysis, J.R.; Methodology, J.R. and M.Y.; Visualization, J.R. and M.Y.; Writing—Original Draft Preparation, J.R.; Writing—Review & Editing, M.Y. and A.B.; Supervision, A.B.

**Funding:** This research was funded by the Research Foundation—Flanders (FWO), grant No. 1200216N.

**Acknowledgments:** The computational work was carried out using the Turing HPC infrastructure at the CalcUA core facility of the Universiteit Antwerpen, a division of the Flemish Supercomputer Center VSC, funded by the Hercules Foundation, the Flemish Government (department EWI), and the Universiteit Antwerpen.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **New Hopes for Plasma-Based Cancer Treatment**

#### **Hiromasa Tanaka 1,\* , Masaaki Mizuno 2, Kenji Ishikawa <sup>3</sup> , Shinya Toyokuni 4, Hiroaki Kajiyama 5, Fumitaka Kikkawa <sup>5</sup> and Masaru Hori <sup>1</sup>**


Received: 30 July 2018; Accepted: 16 August 2018; Published: 18 August 2018

**Abstract:** Non-thermal plasma represents a novel approach in cancer treatment. Both direct and indirect plasma treatments are available, with clinical trials of direct plasma treatment in progress. Indirect treatments involve chemotherapy (i.e., plasma-activated medium) and immunotherapy. Recent studies suggest that integrated plasma treatments could be an extremely effective approach to cancer therapy.

**Keywords:** plasma cancer treatment; plasma-activated medium (PAM); plasma-assisted immunotherapy

#### **1. Introduction**

Plasma-based cancer treatments represent a critical area in the field of plasma medicine [1–3]. Some pioneering in vitro [4] and in vivo [5] works have shown that non-thermal plasma exerts anti-tumor effects. Currently, two options for plasma cancer treatment are available: direct and indirect (Figure 1). Clinical trials of cancer treatments using non-thermal plasma (direct plasma treatments) are ongoing in Germany [6] and the USA [7]. Two different types of indirect plasma treatment have been proposed: plasma-assisted immunotherapy [8] and plasma-activated medium (PAM) [9,10]. Plasma is also considered as an adjuvant therapy, and the three main options are plasma in combination with chemotherapy [11,12], plasma to modulate tumor microenvironment [13,14], and plasma in association with electrotherapy [15]. These approaches have dramatically broadened the ways of using non-thermal plasma for treating cancers and other diseases.

Currently, two options for plasma cancer treatment are available: direct and indirect. Clinical trials of direct plasma treatments are ongoing. Indirect treatments include plasma-assisted cancer immunotherapy and plasma-activated medium (PAM) therapy.

**Figure 1.** Two types of plasma-based cancer treatments.

#### **2. Direct Treatments**

Direct plasma treatments are the most straightforward. A variety of plasma sources have been developed for medical applications such as cancer treatment [16–20]. Metelmann et al. used a plasma source known as kINPen MED to treat advanced head and neck carcinoma ulcerations and patients in the final stages of disease [6,21]. Plasma treatment reduced both pain and odor. Only a few myeloid cells were present in tumor tissue of patients that received frequent plasma treatment, whereas numerous myeloid cells were found in tissue sections of patients that did not receive plasma treatment. Canady et al. treated liver cancer using a Canady Helios Cold Plasma Scalpel to remove cancerous tissue without damaging the blood supply to the remaining liver.

Since plasma needle was used for treatment of culture cells [22], various plasma sources have been developed for cancer treatment. The plasma jet and dielectric barrier discharge (DBD) have been developed and widely used in plasma cancer treatment. A pulsed DBD with microsecond pulses was used for treatment of xenograft model mouse of human glioblastoma cells [23]. Recently, nanosecond-pulsed plasma have been developed as potential tools in cancer treatment [24,25].

#### **3. Indirect Treatment: Plasma-Assisted Cancer Immunotherapy**

Several researchers have proposed the use of plasmas as immune modulators for treating cancer. The number of cells in the human body is estimated at about 40 trillion, and a small portion of these cells acquire mutations and become cancerous every day. However, the immune system typically removes mutated cells. It is only when cancerous cells avoid the immune system that cancerous disease develops. Recently, a variety of anti-cancer therapies designed to modulate the immune system have been developed [26–28]. These approaches include the use of cytokines, cell-based therapies, and immune checkpoint blockade. For example, the US Food and Drug Administration (FDA) approved the first cellular immunotherapy (sipuleucel-T) for prostate cancer in 2010 [29]. The FDA also approved the anti-PD1 monoclonal antibody, nivolumab, for adjuvant treatment of patients with melanoma involving the lymph nodes and patients with metastatic disease who have undergone complete resection [30].

A better understanding of the interactions between cancer cells and the immune system has increased interest in immunotherapies over the last decade [31,32]. Radiation and some chemotherapeutic drugs increase immunogenicity by triggering immunogenic cell death (ICD). Damaged or stressed cancer cells present "danger signals" known as damage-associated molecular pattern (DAMP) molecules. High-mobility group box 1, ATP, and calreticulin (CRT) are well-known DAMP molecules that are retained inside the cell in the healthy state and released only in response to stress or cell damage. Cancer cells usually induce immunosuppression; however, ICD-associated

DAMP molecules can reactivate anti-cancer immunity by triggering dendritic cell maturation and antigen presentation.

Several recent studies have suggested that non-thermal plasma treatment induces ICD and stimulates macrophages [25,33–35]. Non-thermal plasma treatment was shown to stimulate extracellular ATP secretion and enhance cell death via ICD-mediated macrophage stimulation. Plasma-generated reactive oxygen species (ROS) are major effectors of ICD. Non-thermal plasma elicits surface exposure of CRT, and N-acetyl cysteine, which is an ROS scavenger, reduces the externalization of CRT. These results suggest that intracellular ROS are responsible for plasma-induced CRT production. Tumor necrosis factor–alpha released from plasma-activated macrophages induces tumor cell death [36].

In the future, non-thermal plasma will be used to induce ICD in tumors to help dendritic cells find, eat, and present cancer cell antigen to elicit robust T cell immune responses [37]. It was shown that naïve T helper cells were less sensitive toward non-thermal plasma treatment, suggesting that plasma could be used as a tool to redox-control T cell phenotypes in cancer immunology [38]. Flow cytometric technique for microparticle characterization was established, and the number and size of microparticles released were shown to be modulated in THP-1 monocytes, polymorphonuclear leukocytes (PMN), and peripheral blood mononuclear cells (PBMC) after plasma exposure [39]. Interestingly, abscopal effects of non-thermal plasma treatment on tumor growth were observed, suggesting that plasma activated innate immune response [24].

#### **4. Indirect Treatment: PAM**

PAM has been proposed as a type of cancer chemotherapy. Various in vitro experiments have demonstrated that PAM exerts anti-tumor effects on many kinds of cancer cells [9,10,40,41]. In most cases, PAM induces intracellular ROS production and subsequent apoptosis of cancer cells. The mechanism through which PAM induces the apoptosis of cancer cells depends on the cell type [42,43]. In glioblastoma cells, down-regulation of survival and proliferation signaling networks plays a critical role in PAM-induced apoptosis [9,42,44]. Aquaporins, which transport hydrogen peroxide into the plasma membrane, are also key factors in apoptosis induction [45]. Many in vivo experiments have also demonstrated the anti-tumor effects of PAM against a variety of cancers. In a xenograft mouse model, PAM inhibited the growth of ovarian and pancreatic cancer tumor cells [10,41]. Intraperitoneal injection of PAM/plasma-activated Ringer's lactate (PAL) inhibited the metastasis of ovarian, gastric, and pancreatic cancer tumors in disease model mouse experiments examining peritoneal metastasis [46–48]. However, in apoptosis induced by PAL, less ROS are produced in comparison with PAM [49], suggesting that components generated in PAM control the redox balance.

#### **5. Conclusions**

Two major options are available for plasma-based cancer therapies: direct and indirect treatments. Direct plasma treatment methods have already been introduced clinically, whereas indirect plasma treatment methods such as plasma-assisted cancer immunotherapy and PAM therapy are new approaches currently under study. In the future, the overall survival of cancer patients could be significantly improved by combining direct and indirect plasma treatments.

**Funding:** This work was funded in part by Grants-in-Aid for Scientific Research on Innovative Areas "Plasma Medical Innovation" (No. 24108002 and No. 24108008), a Grant-in-Aid for Young Scientists (A) (No. 15H05430), a Grant-in-Aid for Challenging Exploratory Research Grant (No. 15K13390), and Grant-in-Aid for Scientific Research (C) (No. 18K03599) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

MDPI Books Editorial Office E-mail: books@mdpi.com www.mdpi.com/books

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com

ISBN 978-3-03936-604-0