Cold Atmospheric Plasma induces silver nanoparticle uptake, oxidative dissolution and enhanced cytotoxicity in Glioblastoma multiforme cells

Silver nanoparticles (AgNP) emerged as a promising reagent for cancer therapy with oxidative stress implicated in the toxicity. Meanwhile, studies reported cold atmospheric plasma (CAP) generation of reactive oxygen and nitrogen species has selectivity towards cancer cells. Gold nanoparticles display synergistic cytotoxicity when combined with CAP against cancer cells but there is a paucity of information using AgNP, prompting to investigate the combined effects of CAP using dielectric barrier discharge system (voltage of 75 kV, current is 62.5mA, duty cycle of 7.5kVA and input frequency of 50-60Hz) and 10nm PVA-coated AgNP using U373MG Glioblastoma Multiforme cells. Cytotoxicity in U373MG cells was >100-fold greater when treated with both CAP and PVA-AgNP compared with either therapy alone (IC50 of 4.30 μg/mL with PVA-AgNP alone compared with 0.07 μg/mL after 25s CAP and 0.01 μg/mL 40s CAP). Combined cytotoxicity was ROS-dependent and was prevented using N-Acetyl Cysteine. A novel darkfield spectral imaging method investigated and quantified AgNP uptake in cells determining significantly enhanced uptake, aggregation and subcellular accumulation following CAP treatment, which was confirmed and quantified using atomic absorption spectroscopy. The results indicate that CAP decreases nanoparticle size, decreases surface charge distribution of AgNP and induces uptake, aggregation and enhanced cytotoxicity in vitro.


Introduction
Glioblastoma Multiforme (GBM) is an aggressive grade IV astrocytoma. It is the most dominant form of central nervous system malignancy, accounting for 47.1% of all tumours diagnosed in the CNS [1]. The current predominant treatment is surgical resection followed by radiotherapy and chemotherapy with Temozolomide [2]. However, conventional treatments are rarely successful and are plagued with poor target delivery, poor efficacy and systemic toxicity. Despite intensive therapeutic strategies and medical care, approximately 5% of GBM patients survive five years after diagnosis [1]. In addition, more than 90% of GBM patients reveal recurrence at or near the primary site after treatment, underlining the need for a new effective therapeutic approach [3].
Nanoparticles (NP) have been used for therapy in diverse fields as a radiosensitiser [4], fluorescent labels [5], transfection vectors [6] and as drug carriers [7]. The size, shape and material composition confer advantageous properties on NP for various applications [8]. In particular, silver nanoparticles (AgNP) are the most widely used nanomaterial in consumer products such as household, cosmetics and healthcare-related products due to its antimicrobial properties through the release of silver ions [9]. The generation of reactive oxygen species has been associated with AgNP toxicity, making it useful for anticancer therapy [10]. Studies have shown that AgNP induce alterations in metabolic activity, cell morphology and decreased cell viability. Treatment with AgNP showed higher selectivity towards aggressive brain cancer human glioblastoma cells (U251) compared with a normal human lung fibroblast cells (IMR-90), leading to mitochondrial damage and increased in reactive oxygen production, resulting with DNA damage [11]. The combined NPs advantages of small size and large surface area has led to their use as drug delivery systems. In particular, AgNPs have attracted attention due to their intrinsic anticancer activity and effective drug delivery agents in previous studies [12][13][14]. Studies have shown AgNPs capability of crossing the blood brain barrier (BBB), providing opportunities and tackling challenges associated with NP drug delivery to the central nervous system (CNS) [15][16][17].
Recently, metal nanoparticles have been utilised to enhance cytotoxicity in cancer cells using oxidising treatments such as radiation therapy [18,19]. Liu [4]. These studies resulted to multitude applications of employing NPs to cancer treatments through a vast number of strategies.
Plasma treatment has shown potential as a future cancer therapy. Plasma is the fourth state of matter next to solid, liquid and gas. It can be artificially produced for its versatile applications [20]. Plasmas are classified as either thermal or non-thermal, also known as cold atmospheric plasma (CAP). The non-thermal nature of CAP coupled with a wide range of biological effects has led to the emergence of CAP across a range of biomedical applications including wound healing, dentistry and sterilisation [21]. CAP has low power requirements and is achieved at low or atmospheric pressure. CAP provides a rich environment of reactive oxygen species (ROS) such as singlet oxygen (O 2 ), superoxide (O 2 -), ozone (O 3 ), hydroxyl radicals (OH -), hydrogen peroxide (H 2 O 2 ) and generates reactive nitrogen species (RNS) such as nitric oxide (NO) nitrite and nitrate anions (NO 2 and NO 3 ) [22]. Recent studies have shown CAP's potential application in cancer therapy with biochemical features of cancer cells including high levels of ROS due to oncogenic transformation and with the application of CAP triggered self-perpetuating process of RONS induction, which effectively showed to induce apoptosis selectively against cancer cells overcoming the problem with conventional treatments [23]. In contrast to NPs systemic application, the effects of CAP are mostly associated with the location of treatment with few systemic effects observed and hence the recent trend in research of CAP is the interaction at cellular level [24]. The localised interaction of CAP with mouse fibroblast cells, BEL-7402 liver cancer cells and PAM212 cancer cells demonstrated detachment from extracellular matrix when treated [25]. CAP's ability to change biochemical signalling intracellularly without thermal and electrical damage creates a suitable biomedical application [26]. The operating system of plasma discharged used in this study is the dielectric barrier discharge (DBD), DIT 120, which generates high voltage output of non-thermal plasma between two aluminium electrodes [27]. CAP induced by the DIT 120 system has previously been reported to induce cell death at higher exposures and enhance uptake of gold nanoparticles at lower exposures in U373MG glioblastoma multiforme cancer cells, the cell line used in this study [28,29].
A synergistic cytotoxic effect has been reported when CAP and various nanoparticles are combined, as first reported in 2009 by Kim, et al., who found a 5-fold increase in cytotoxicity on G361 melanoma cancer skin cells when treated with ambient air CAP combined with antibody-conjugated AuNP [30]. Since then, studies using nanomaterials with various sizes and compositions have been used. For example, electrosprayed core-shell nanoparticle fabricated with 5-Fluorouracil synergistically inhibited cell growth of epithelial breast cancer cells MBA-MD-231 when used with CAP [31]. The plasma jet device using helium and oxygen gas in combination with Iron NPs significantly decreased viability of human breast adenocarcinoma cancer cells, MCF-7 [32]. Our own group unveiled 25-fold enhanced cytotoxicity on U373MG cells when 20 nm citrate-capped AuNP were combined with non-toxic doses of CAP, demonstrating enhanced AuNP endocytosis and subcellular trafficking [29]. Meanwhile, interest in AgNP has shifted beyond antimicrobial use to potential additional anticancer applications [33][34][35]. Evidence is emerging that oxidative stress induced by low dose AgNP is implicated in their cytotoxicity [36][37][38][39]. Despite AgNP being the main commercial nanomaterial used worldwide, there are limited reports on combining AgNP with other current therapies to investigate its possible enhanced effect in comparison to various type of nanoparticles studied.
In consideration of the advantages of both AgNP and CAP, the importance of oxidative stress in both modes of cytotoxicity, the enhancement of AgNP toxicity when combined with other oxidising treatments and the finding that CAP can induce cellular uptake of nanomaterials, we chose to investigate whether synergistic cytotoxicity exists between AgNP combined with CAP and to explore the interaction using the U373MG GBM cell line model.

Chemicals
All chemicals used were obtained from Sigma-Aldrich (Vale Road, Arklow, County Wicklow, Ireland) unless specified otherwise.

AgNP preparation
The top-down synthesis used in the study was previously reported by Mavani et al [40]. Cold synthesis was employed with chemical reduction of silver nitrate (AgNO 3 ) of 0.001M with icecold reducing agent sodium borohydride (NABH 4 ) of 0.002M in the presence of a stabiliser formed AgNP. 2ml of 1% polyvinyl alcohol (PVA) in millipore water (Simplicity 185, 18.2 MΩ.cm at 25°C resistivity) was added with 2ml of silver nitrate and mixed well [41]. The icecold reducing agent was stirred for 20 minutes and 2 ml of mixed AgNO 3 /PVA were added 1 drop per second approximately and reaction was stopped. PVA stabilised silver nanoparticles (PVA-AgNP) mixture was concentrated by using an ultrafiltration tube of 3kDa (Sartorius, UK) with centrifuge, Heraeus Megafuge 16R (Thermo Scientific) at 5000 rpm.
PVA-AgNP were stored away from direct exposure to light at 4ºC. The chemical reduction of AgNO 3 can be written as:

Cold atmospheric plasma device
The CAP system used was a dielectric barrier discharge (DBD) device. It is a novel prototype atmospheric low temperature plasma generator [27]. to CAP with a total of 100 µl in 96 well plates and 3 ml in 60mm petri dishes. The samples were treated at 75 kV at different exposure times from 0-80s.

Cell viability assay
Alamar Blue was used as a parameter for measuring cytotoxicity [28,42]. hours, the cells were then exposed to CAP from 0-80s at 75kV as outlined above. The plates were incubated for a total of 48h post CAP treatment at 37°C. Cells were then washed with PBS and 10% Alamar blue was added into the wells for 3h at 37°C. The fluorescence was measured using excitation wavelength of 530nm and emission wavelength of 590nm on plate reader (SpectraMax M3, Molecular Devices (UK) Ltd). The protective effect of N-Acetyl Cysteine (NAC) was evaluated by pre-treating 4mM NAC on U373MG cells for 1h followed by treatment of PVA-AgNP at stated concentrations for 24h and exposed to CAP at 0s, 25s and 40s for another 24h.

Flow cytometry for live and dead cell staining
Propidium iodide (PI) was used to demonstrate live and dead cell staining with flow cytometer BD Accuri C6 (BD, Oxford, UK). U373MG cells were seeded in 6-well plate at 2.5 x 10 5 cells per well and was incubated at 37ºC overnight to allow adherence. Cells were treated with PVA-AgNP at low-nontoxic dose of 0.07μg/mL for 24h and were exposed to CAP at 75kV for another 24h. Cells were harvested including pre-existing media and centrifuge to form a pellet. The pelleted cells were resuspended in 1ml PBS and was stained with PI for 1 minute with concentration of 10μg/mL. PI fluorescence was detected using FL2 vs FSC demonstrating binding of PI to nuclear degradation from dead cells.

Measurement of CAP effect on AgNP size
The effect of CAP on PVA-AgNP size was measured as follows: concentrated PVA-AgNP obtained from ultrafiltration to remove excess unwanted reactants including NaBH 4 was resuspended in millipore water, in synthesis solution containing unwanted reactants and resuspending concentrated PVA-AgNP in 4mM NaBH 4 . The samples were exposed to CAP at different exposure times from 0-80s. The hydrodynamic size of the NPs was determined using DLS, Zetasizer Nano ZS. The morphology was determined using STEM (Hitachi SU 6600), as described above.

Darkfield spectral imaging of AgNP uptake: image acquisition and uptake analysis
Optical observation of AgNP uptake was performed using darkfield spectral imaging (SI). PCA was applied to the false colour images to create a mask between the cell and background (thresholding the first principal component score image such that all pixels with a values below 2 were set to one provided sufficient separation between cells and background). In some cases, multiple cells were attached to each other in the mask images.
In order to be able to analyse each cell individually, these cells were separated manually by drawing a line across the narrowest connecting point in the mask image. In addition, some cells were only partially represented in a given field of view -such cells were removed from the analysis. Further, some pixels that were not identified as cells were highlighted in the mask -in order to remove them from the analysis, any regions with a pixel size smaller than 300 pixels were automatically removed from the analysis.
Standard normal variate pre-processing (i.e. subtracting the mean from each spectrum and dividing by the standard deviation) was applied to spectra and a partial least squares discriminant analysis model was developed to discriminate between NPs and cells. This model was subsequently applied to all images, facilitating identification of NPs from their spectra. Using the 'regionprops' function of the MATLAB image processing toolbox, the size and circularity of each cell and number and size of NPs per cell was calculated. The median number of NPs identified per cell was then calculated for each treatment.

AAS measurement of AgNP uptake
Atomic Absorption Spectroscopy (AAS) was further employed to measure uptake of AgNP.
U37MG cells were grown in petri dishes with cell density of 2.5 x10 6 cells per dish. The samples were negative control with U373MG cells alone, cells treated with AgNP, cells exposed to CAP, the combined therapy on U373MG cells and positive control treated with 1mM H 2 O 2 . The cells were treated with AgNP for 24h and after with 25s CAP for another 24h. The cell samples were trypsinised and resuspended in the existing culture media. The samples were then analysed by AAS for AgNP detection. Triplicate readings were analysed for each sample and the results were expressed as the mean amount of Ag in pg/cell.

Statistical analysis
The data of the experiments are expressed as mean ± standard deviation of replicates from at least three independent experiments, unless specified otherwise. Statistical analysis and curve fitting presented in results were completed using Prism 7 (GraphPad Software).
Analysis of data distribution was performed using two-way ANOVA and three-way ANOVA where indicated to analyse the differences of significance between the control group and 1 0 treated groups. The following P values were deemed statistically significant, **P<0.01, ***P<0.001, ****P<0.0001. Therapeutic synergism between PVA-AgNP and CAP was evaluated using isobologram analysis. CompuSyn software determined the combinational index (CI) where, CI>1 is antagonism, CI=1 is additive and CI<1 is synergism. Descriptive statistics (median, standard deviation) were calculated on cell size, circularity, NP size and number as derived from spectral imaging measurements.

Data Availability
All datasets can be viewed in tabular form in Supplementary Information. PVA-AgNP were prepared and characterised as indicated in the methods section. STEM analysis of PVA-AgNP confirmed production of nanoparticles that are spherical in shape, approximately 10nm in size and well dispersed (see Figure 1a). The presence of a plasmon absorption band (400 nm), which is a main characteristic of AgNP, was evident and remained relatively unchanged over 6 months indicating the production of highly stable

Silver Nanoparticles Characterisation
AgNP (see Figure 1b)[43]. The particle size range by DLS analysis was determined to be 8-

Cytotoxic Effect of AgNP in Combination with CAP on Cellular Viability
Cytotoxicity was examined using Alamar Blue and propidium iodide. μg/mL). From previous work with the same DBD prototype used in this study, it was determined that the IC 50 with CAP treatment alone is 74.26s (95% confidence range of 47.24-116.8s) on U373MG cells [28]. In the current study, we combined treatment of AgNP with a range of low CAP exposures (i.e. 5s, 10s, 25s and 40s at 75kV). We found that cells

AgNP-induced Cytotoxicity Protected by NAC
The main component of CAP is the generation of RONS and many have linked oxidative stress with AgNPs toxicity in previous studies [36][37][38][39]. NAC is a scavenger of oxygen-free radicals and directly interacts with reactive ionised species. The protective effect of NAC on AgNP and in combination with CAP can be seen in figure 3a. The IC 50 value for AgNP control was 4.9 μg/ml with 95% confidence range of 3.759 to 6.081μg/ml. The IC 50 value for AgNP with pre-treated NAC was 6.57 μg/ml with 95% confidence range of 5.989 to 7.130 μg/ml. The IC 50 value for the combinational AgNP-CAP without NAC at 25s was 0.06 μg/mL

The effect of CAP on AgNP size
Our data indicate that ROS-dependent cytotoxicity is induced in GBM cells treated with a combination of CAP and AgNP. The toxicity is likely due to one or more of the following: cellular rate of uptake, increase in cell membrane permeability, changes to nanoparticle size and morphology, altered dispersion or agglomeration and rate of dissolution. It has been reported that NPs can be prepared using an electrical discharge [44][45][46]. We therefore investigated the effect of CAP on AgNP. 10nm freshly synthesised AgNP including unwanted reactants in the existing solution containing excess NaBH 4 (4mM) were exposed to the DBD-CAP device (75kV, 0-80s). The size of AgNP significantly decreases in a dose-dependent manner when exposed to CAP (Figure 4a). In contrast to this, purified AgNP resuspended in fresh millipore water without unwanted reactants with CAP did not change significantly in size. We confirmed this using STEM. AgNP in a solution of the reductive agent, NaBH 4 (4mM) showed decrease in size (5 nm) when exposed to 25s CAP compared with controls without CAP treatment (10 nm) (see Figure 4b). Nanoparticle morphology and dispersion did not change during CAP exposure, and AgNP remained spherical in shape, uniformly dispersed with no visible aggregation in the samples was observed.

6
Zeta potential of AgNP before and after CAP treatment in millipore water (mH 2 O) and culture media (DMEM). Representation of zeta potential of AgNP before CAP treatment with -55.6 ±8.06mV to -26.8 ±9.08mV after 25s CAP exposure in mH 2 O and -18.0 ± 8.70mV without CAP exposure to 11.40 ± 5.06mV after CAP exposure at 75kV in culture media.
The recognised value for zeta potential that is anything higher than positive or negative 30mV is a stable suspension [47]. Figure 4c shows an electrostatically stabled nanoparticle suspension of -55.6 ± 8.08mV with conductivity of 0.360mS/cm and after CAP exposure of 25s the nanoparticle suspension has zeta potential of -26.80 ± 9.08mV with conductivity of 0.25mS/cm for PVA-AgNPs suspended in millipore water. In contrast to PVA-AgNP in millipore suspension, PVA-AgNPs in culture media (DMEM-F12) has changed the zeta potential value of -18.0 ± 8.70mV with conductivity of 0.011mS/cm and after CAP exposure of PVA-AgNP in DMEM-F12 the zeta potential decreased further to -11.40 ± 5.06mV with conductivity of 0.30mS/cm. The change of zeta potential of NPs was previously reported in other studies that affects the internalisation process during uptake of NPs by cells. Uptake of PVA-AgNP was further explored and can be seen in figure 5.

Uptake of AgNP in U373MG cells
The study further investigated whether the significant difference on cytotoxicity of AgNP treated alone compared to combination of AgNP-CAP could be explained by differences in cellular uptake and localisation intracellularly. Visualisation of cell morphology, nanoparticle distribution and particle localisation after treatment with low dose of 0.07 µg/ml AgNP alone and in combination with 25s CAP at 75kV was investigated using Spectral imaging (SI) (see signatures that is highly repeatable in approx. 50 images per sample (see Figure 5a). The spectral profile is graphed as wavelength (nm) versus intensity of scattered light (a.u.) presenting spectral signatures and interaction of NPs to cells with spectral response showing a broad scattering spectrum of U373MG cells alone are at 520nm with low intensity, PVA-AgNPs spectral response in cells is at 520nm with higher intensity and an enhanced intensity of PVA-AgNP showing a shift of the resonance peak to 570nm presents NPs aggregating into larger sizes.  Figure 2d). This correlates with our previous study (He et al) when we found that combination of gold nanoparticles with CAP increased nanoparticle uptake [29].

Discussion
The  [23,[27][28][29]. Despite the reports of CAP sensitivity to cancer cells, we and others have demonstrated that the Glioblastoma multiforme cell line U373MG is relatively resistant to CAP treatment [28] and approaches to overcome this inherent resistance will be necessary in a clinical setting. In addition, with the ever-growing interest of combined cancer therapy with nanotechnology and plasma medicine, the combination of AgNP with CAP has not yet been reported. In this study, the combined synergistic effect of PVA stabilised AgNP with 1 9 CAP on U373MG was evaluated and the interaction between CAP and AgNPs physical properties and its effect on cell morphology were demonstrated.  [36][37][38][39]. Interestingly as viewed in figure 2a, the combination of AgNP with CAP increased more than 100-fold of cytotoxicity in comparison to AgNP treatment alone with IC 50 of AgNP combined with 25s CAP at 75kV is 0.077 µg/ml and when combined with 40s CAP at 75kV, IC 50 was 0.0087 µg/ml. CAP similarly generates ROS and can be localised by selective application process at a milli or micro scale to cancer cells [50,51]. Consistent with the findings regarding to oxidative stress correlation to cell death with AgNP and CAP continuously mounts the evidence to show that generation of ROS is highly related to the mechanism of AgNP and CAP with the results in current study showed that the cytotoxicity induced by AgNP alone, as well as the combinational treatment AgNP-CAP were efficiently prevented prior NAC treatment up to 10-fold (see Figure 3). The results determine that oxidative stress is responsible for the cytotoxicity of AgNPs and CAP, which is compliant with previous studies portraying the protective effect of NAC when treated with either AgNP or CAP resulting to recovery of proliferative cells [52][53][54][55].
Tseng et al, has previously reported metal nanoparticle fabrication using CAP in the form of electric discharge machine, where the generation of arc discharge between two electrodes disintegrate silver rod in liquid producing silver nanoparticles [45,56,57]. Due to this phenomenon, the current study next investigated the effects of CAP on AgNP physical properties. CAP evidently reduced the diameter size of AgNP with longer exposure time in 2 0 presence of reducing agent (see Figure 4a). A study has investigated hydrolysis of NaBH 4 in aqueous solution, it was reported that the stability of NaBH 4 decreases with elevated temperature [58][59][60]. It can be hypothesised with the findings that CAP's effect on AgNP leads to generation of discharge between two electrodes producing a thermal effect on AgNP solution with borohydride present, this increases temperature and thus accelerates hydrolysis reaction. Furthermore, CAP's alteration on PVA-AgNP size in our study resulted in altered electronic properties on the surface of NP (see figure 4C). Several studies have shown size-dependent effects on cytotoxicity using silver nanoparticles [61][62][63]. In many cases, the levels of cell death are increased when smaller nanoparticles are used, and this is believed to be due to a larger surface area and enhanced rates of endocytosis. However, the effect of size on cytotoxicity in these studies is relatively small (approx. 2-5 fold) and unlikely to be solely responsible for the synergistic cytotoxicity between CAP and AgNP observed in our study. Many reports stated the standard stable nanoparticle suspension is anything higher than positive or negative 30mV. The higher value of positive or negative zeta potential has been studied to show in nature to repel each other and not come together. The particles tend to aggregate and flocculate with lower zeta potential values due to the absence of repulsive forces that hinders agglomeration [64][65][66]. Studies have reported that the greater the negative charge the less toxic the nanoparticles [67,68]. In our study CAP have decreased the zeta potential of PVA-AgNP, which may be associated with absence of repulsive forces at the double layer leading to the likelihood of agglomeration that can be seen in uptake of PVA-AgNP in figure 5. Another reason for the agglomeration of NPs evident in the uptake of PVA-AgNP with CAP in figure 5a is due to serum proteins in culture media absorbed on NPs surface. Studies have reported the nanoparticle protein corona adds complexity to biological system interactions that cannot be limited to electrostatic binding alone. The new biological identity of the nanoparticle influences cell behaviour interaction [69][70][71].
Nanoparticle detection in cells and tissues are often achieved by employing electron microscopy techniques and confocal microscopy to investigate translocation of NPs in cells [72][73][74]. While these techniques have extensively accomplished identification of NPs, they lack the potential to validate NPs presence in cells or tissue via spectral mapping. The spectral imaging technique provides each pixel of SI image a spectral response for each spatial area of a pixel [75]. SI of NPs in cells provides the feasibility of detecting NPs, partial size, surface modification, spatial location, presence of NP agglomeration and wavelength differentiation [76,77]. In this study, we used SI to asses uptake of PVA-AgNP when exposed to CAP resulting to an enhanced PVA-AgNP uptake when cells are exposed to CAP than of NP treatment alone. This was quantified using atomic absorption spectroscopy where we 1 confirmed a 50-fold increase in Ag/cell following CAP treatment. Our group demonstrated the direct and indirect chemical effects generated by CAP DIT-120 is a mediator of the uptake increase of AuNPs [29]. Our data here provide further evidence that CAP DIT-120 can stimulate uptake of nanomaterials of different sizes and compositions in addition to significantly enhancing cytotoxicity. Interestingly, Au/cell is increased by 50% following exposure to CAP, whereas Ag/cell is increased 50-fold under similar conditions. This may be due to nanomaterial size, the direct effect of CAP on the AgNP or due to a cellular process.
Further investigation of the biological processes that regulate CAP-stimulated uptake of nanomaterials and cytotoxicity in GBM are ongoing and will offer future insights into adapting these combinational approaches for development of therapeutics for treatment of GBM and other solid tumours.
In conclusion, the current study reports the enhanced synergistic cytotoxic effect of the combined PVA-AgNP and CAP on U373MG cells in vitro. The study showed the ROSdependent toxicity of the combined therapy, which was prevented by NAC. Enhanced uptake of PVA-AgNP followed by CAP treatment was confirmed using spectral imaging and AAS.
Overall, the results indicate the effect of CAP on the physical properties of PVA-AgNP leading to decrease in nanoparticle size, decrease in surface charge distribution and inducing enhanced uptake, aggregation and synergistic cytotoxicity. The findings in the study demonstrated the combination therapy of PVA-AgNP with CAP can be further evaluated for its potential use in cancer therapy.