Principles of using Cold Atmospheric Plasma Stimulated Media for Cancer Treatment

To date, the significant anti-cancer capacity of cold atmospheric plasma (CAP) on dozens of cancer cell lines has been demonstrated in vitro and in mice models. Conventionally, CAP was directly applied to irradiate cancer cells or tumor tissue. Over past three years, the CAP irradiated media was also found to kill cancer cells as effectively as the direct CAP treatment. As a novel strategy, using the CAP stimulated (CAPs) media has become a promising anti-cancer tool. In this study, we demonstrated several principles to optimize the anti-cancer capacity of the CAPs media on glioblastoma cells and breast cancer cells. Specifically, using larger wells on a multi-well plate, smaller gaps between the plasma source and the media, and smaller media volume enabled us to obtain a stronger anti-cancer CAPs media composition without increasing the treatment time. Furthermore, cysteine was the main target of effective reactive species in the CAPs media. Glioblastoma cells were more resistant to the CAPs media than breast cancer cells. Glioblastoma cells consumed the effective reactive species faster than breast cancer cells did. In contrast to nitric oxide, hydrogen peroxide was more likely to be the effective reactive species.

The reactive species accumulate in the CAPs media with a dose-dependent and liquid surface-dependent pattern. Among dozens of species, NO in RNS and H 2 O 2 in ROS are thought to play key roles in killing cancer cells 5,35 . In addition, hydroxyl free radicals (.OH) in CAP are also proposed to kill cancer cells 36 . To date, most of these conclusions are based on the research for the direct CAP treatment on cancer cells. The understanding on the RNS and ROS accumulation in the CAPs media is far from clear. In this study, we first studied the generation of NO and H 2 O 2 in the CAPs complete media (90% DMEM + 10% FBS) via Griess Reagent System (Promega) and Fluorimetric Hydrogen Peroxide Assay Kit (Sigma-Aldrich), respectively.
Next, we harnessed methylene blue (Fisher) to qualify the generation of .OH. In contrast to terephthalic acid, which has also been used to qualify the .OH generation after plasma treatment 36 , methylene blue (MB) is a cheaper and more convenient probe. MB strongly reacts with .OH in the aqueous solution and changes its color from blue into colorless, which can be detected by spectrophotometer at 664 nm 37,38 . The atmospheric plasma jet is a proven MB decomposition tool 39 . We proved that MB was just sensitive to the species with a short half-life (Fig. S1). Because MB is strongly absorbed by the proteins in the complete media, we investigated the generation of .OH in the CAPs deionized water in this study.
Reactive species accumulate in the CAPs media with different patterns. To better illustrate these patterns, all data in Fig. 2 have been normalized to be the relative values via dividing the data from experimental group by the data from corresponding control group. RNS (Fig. 2a) and H 2 O 2 (Fig. 2b) in the CAPs media both increase as the treatment time increases. Because MB will be consumed after the reaction with .OH, corresponding relative absorbance of the CAPs MB solution should be less than 1 (Fig. S1). However, as shown in Fig. 2c, noticeable .OH generation was not observed even the treatment time was extended to 2 min. We performed the same experiment in the MB solution with a larger volume (2 mL) and found that the generation of .OH was proportional to the plasma treatment time again (Fig. 2d). This volume-dependent pattern may be due to the interaction between plasma jet and solution. During the CAP treatment, the liquid below the plasma jet would be extruded, pushing the solution to the perimeter of well (Fig. S2a). When the volume of liquid was just 1 mL, the plasma jet could not touch liquid due to the exposure of the bottom of well. When the volume of liquid was up to 2 mL, the extrusion of liquid would be weaker and enabled the plasma jet to touch the liquid layer. However, even when the plasma jet does not touch the liquid, the reactive species are still able to dissolve into the liquid. Yonemori et al. observed that when an atmospheric-pressure plasma jet touched a glass surface, it flowed radically over the glass surface and formed a large area containing reactive species on the glass surface 40 . Thus, the reactive species in the plasma jet should affect an area of liquid that is significantly larger than the diameter of the jet. The half-life of .OH is only a few microseconds 41 , however, which eliminates the possibility that .OH diffuses over the liquid surface (Fig. S2a). In contrast, H 2 O 2 and NO with much longer half-life may enter the media by the diffusion over the whole surface of liquid. We denote that H 2 O 2 /NO area and .OH area to represent the area mainly affected by H 2 O 2 /NO and .OH on the liquid surface covered by plasma flow, respectively (Fig. S2a). Together, when the volume of media is relative small, .OH will not be a main factor to directly affect the anti-tumor capacity of the CAPs media. .OH may react with .OH to form H 2 O 2 and affects the anti-tumor capacity indirectly.
If H 2 O 2 /NO area and .OH area do exist as we depicted in Fig. S2, it is reasonable to deduce that the surface of media irradiated by the plasma may affect the accumulation of H 2 O 2 /NO but not .OH in the CAPs media. We proved this deduction by measuring the production of H 2 O 2 /NO and .OH in the CAPs media and CAPs MB solution in distinct multi-well plates, respectively. The well diameters on 48-well, 24-well, 12-well, and 6-well plate were 10.2 mm, 15.4 mm, 21.4 mm, and 35.0 mm, respectively. We found that the H 2 O 2 /NO concentration in the CAPs media varied significantly with the well size on the plate. The larger diameter of well, the more H 2 O 2 /NO accumulated in the CAPs media (Fig. 3a,b). By contrast, the .OH generation in the CAPs MB solution doesn't significantly vary with the well size (Fig. 3c), even when the volume of MB solution is up to 2 mL (Fig. 3d). The .OH generation in the CAPs MB solution from a 6-well plate is noticeably lower than that from other plates (Fig. 3c). Results are presented as the mean ± s.d. of three repeated experiments performed in triplicate. Student's t-test was performed, and the significance compared with the first bar is indicated as *p < 0.05, **p < 0.01, ***p < 0.005.
Scientific RepoRts | 5:18339 | DOI: 10.1038/srep18339 A schematic illustration in Fig. S2b depicts the underlying mechanism of the well size-effect on the reactive species accumulation. A smaller well creates a smaller H 2 O 2 /NO area on the media surface. More H 2 O 2 /NO will diffuse into the CAPs media when the well size becomes larger. Additionally, the thickness of media in the well increases as the well size decreases. The calculated liquid thickness of 1 mL media in 48-well, 24-well, 12-well, 6-well plate are 12.2 mm, 5.4 mm, 2.8 mm, and 1.0 mm, respectively (Fig. S2b). Thus, the extrusion of media due to the plasma jet pressure will be weakened as the diameter of well decreases. When the volume of MB solution is just 1 mL, compared with the MB solution in 6-well plate, the MB solution in other multi-well plates is more likely to contact the .OH in the CAP. Thus, more MB is consumed in the CAPs MB solution from 48-well, 24-well, and 12-well plate than that from 6-well plate (Fig. 3c). Even when the volume of MB solution is 2 mL, .OH is only able to affect a small area on the media directly touched by the CAP jet, so the .OH generation in the CAPs MB solution changes little when the well size is noticeably altered. In short, the distinct half-life among H 2 O 2 /NO and .OH underlies the different well size-effect of reactive species observed in this study.
The anti-tumor capacity of the CAPs media is dose-dependent and well size-dependent. We further investigated the anti-tumor capacity of CAPs media on glioblastoma cells (U87), breast cancer cells (MDA-MB-231 and MCF-7) with distinct cell confluences. It was found that the anti-tumor capacity of the CAPs media increases as the treatment time (dose) increases and decreases as the cell seeding confluence decreases (Fig. 4a-c). Thus, it is the dose of CAP treatment exerting on a unit cell, rather than the whole CAP treatment dose, that determines the fate of cancer cells. In addition, MDA-MB-231 cells and MCF-7 cells are more vulnerable to the CAPs media than U87 cells. The response of MDA-MB-231 cells and MCF-7 cells to the CAPs media is similar, though MCF-7 cells are a little easier to be killed.
We further investigated effect of well size on the anti-tumor capacity of the CAPs media on the same three cancer cell lines. We found that the anti-tumor capacity of the CAPs media decreased as the size of the wells decreased ( Fig. 4d-g). For U87 cells, the residual viability of cells cultured in the CAPs media from a 6-well plate is about 1/3 of the viability of cells cultured in the CAPs media from a 48-well plate (Fig. 4d). In other words, at least 2/3 of the anti-cancer ability of the CAPs media is wasted in the 48-well plate. A similar trend is also observed on MDA-MB-231 cells (Fig. 4e). However, the well size-effect on MCF-7 cells only appears when the seeding confluence is as high as 4 × 10 4 cells/ml (Fig. 4g), which may be due to the fact that even the CAPs media from a 48-well plate is adequate to kill almost all MCF-7 cells with a seeding confluence as low as 2 × 10 4 cells/ml (Fig. 4f). Accordingly, the 96-well plate should waste more reactive species. By surveying all publications about the application of plasma jet on cancer treatment, it was determined that only 25% publications used 6-well plate or dish with same size in their CAP treatment (Fig. S3). Thus, most investigations wasted significant amount of plasma, gas and media over the past decade. The dose-dependent and the well size-dependent anti-cancer features of the CAPs media is consisted with the dose-dependent and liquid surface-dependent reactive species accumulation patterns in the CAPs media.  The anti-tumor capacity of the CAPs media varies with the gap between plasma source and media. For the media in the well with identical volume, when the well size decreases, the height of the media in the well increases. Thus, the well size-dependent effect may be due to the altered size of the gap between the surface of the media and the plasma source. We investigated the generation of NO and H 2 O 2 in CAPs media by altering the gap between the bottom of plate and the nozzle of quartz tube. We found that the NO and H 2 O 2 in the CAPs media shown a distinct response to a change in gap. As the gap increases from 2 cm to 4 cm, the concentration of NO in the CAPs media shows a parabolic response. It reaches a peak at the gap of 3 cm and then decreases as the gap is increased to 4 cm (Fig. 5a). The distribution of NO along the axial direction of the atmospheric pressure plasma jet has been measured by laser induced fluorescence and showed similar trend as we observed 42 . In contrast, the concentration of H 2 O 2 shows a stepwise change upon the gap increase. The concentration of H 2 O 2 in the CAPs media remains fairly constant over the gap of 2 cm to 3 cm. When the gap increased from 3 cm to 4 cm, the concentration of H 2 O 2 in the CAPs media decreased about 26% (Fig. 5b). Next, we investigated the viability change of three cancer cell lines upon the gap change. As shown in Fig. 5c-e, the anti-tumor capacity of the CAPs media dose not noticeably change until the gap increases to 3.5 cm or 4 cm. The gap-effect on the anti-cancer capacity is consistent with the gap-effect on the generation of H 2 O 2 but significantly differs from the gap-effect on the generation of NO. It indicates that H 2 O 2 rather than NO may dominate the death of cancer cells. Ultimately, Fig. 5 proves that the well size-effect on the CAPs media is not due the gap-effect. Because the smaller gap only tends to generate more reactive species, the actual well size-effect should be stronger than that shown in Fig. 4. Results are presented as the mean ± s.d. of three repeated experiments performed in triplicate (a,b) or in sextuplicate (c,d,e). Student's t-test was performed, and the significance compared with the first bar is indicated as *p < 0.05, **p < 0.01, ***p < 0.005.
Scientific RepoRts | 5:18339 | DOI: 10.1038/srep18339 The anti-cancer capacity is volume-dependent. As early as 2007, it was found that melanoma cells could be protected in FE-DBD treatment by varying depth of cell growth media 3 . However, the mechanism underlying the protecting role of media remains obscure. As we have demonstrated, the strong anti-cancer capacity of the CAPs media always accompanies a high concentration of species in the media. Because the increase of media depth means the increase of media volume in the same container, the protecting role of media may just due to the dilution of reactive species in the media. We investigated the volume-effect on the anti-cancer capacity of the CAPs media. It was found that both the concentrations of H 2 O 2 (Fig. 6a) and NO (Fig. 6b) decreased as the volume of media increased. Furthermore, the killing capacities of the CAPs media on three cancer cell lines significantly decreased as the volume of media increased (Fig. 6c-e). In short, the protecting role of media is due to the dilution-effect of the reactive species. According to this principle, the optimized anti-cancer capacity of CAP treatment, direct or indirect, can be achieved when cells are surrounded by a few media.
Cysteine and tryptophan are the main targets of effective species in the CAPs media. Here, we denote effective species to represent the effective reactive species which kill cancer cells. Understanding the chemical essence of the reaction between the effective species and the intracellular molecules is a prerequisite to understanding the mechanism underlying the anti-cancer capacity of CAP treatment. However, to study the intracellular reaction between RNS/ROS with the thousands of intracellular molecules directly is too challenging to be performed. In this study, we focused on revealing which amino acids significantly reacted with the effective species. The interaction between CAP and the amino acid solution have been studied via mass spectra. It was found that the sulfur-containing and aromatic amino acids in the aqueous solution were preferentially consumed in the CAP treatment 43 . Methionine, cysteine, tryptophan, phenylalanine, and tyrosine were the five most consumed . Student's t-test was performed, and the significance compared with the first bar is indicated as *p < 0.05, **p < 0.01, ***p < 0.005.
Scientific RepoRts | 5:18339 | DOI: 10.1038/srep18339 reactive amino acids upon the CAP treatment 43 . However, some amino acids may be consumed by the species which do not kill cancer cells. The revealed reaction types between amino acids and plasma have not answered the question of which amino acids tend to react with the effective species.
We harnessed a novel strategy to reveal the reaction strength among effective species and specific amino acids. We used cancer cells as a cell probe to investigate whether the effective species in the CAPs media would be consumed by particular amino acids via comparing the anti-cancer capacity of a specific 2.4 mM CAPs amino acids rich DMEM with the corresponding untreated amino acids rich DMEM. Each amino acids rich DMEM was prepared by dissolving specific amino acids powers in DMEM.
Among 20 amino acids, cysteine and tryptophan showed the strongest reactivity towards the effective species in CAPs media. In contrast to DMEM, the cysteine rich DMEM consumes most effective species and almost completely eliminates the anti-tumor capacity of the CAPs DMEM on U87 cells (Fig. 7a), MDA-MB-231 cells (Fig. 7b), and MCF-7 cells (Fig. 7c). Tryptophan has a similar but weaker capacity to consume effective species. A control group was also analyzed for each amino acid tested, because if a specific amino acid rich DMEM is toxic to cancer cells, the corresponding control group has a very low cell viability, which makes the relative cell viability in Fig. 7 appears very large. As a result, we may get a delusion that a toxic amino acid consumes most effective species. In this study, for U87 cells and MCF-7 cells, tryptophan shows a strong resistance to growth of cancer cells (Fig.  S4a,c). The mechanism underlying this toxicity is unclear. We also found that the tryptophan-rich DMEM did not resist the growth of MDA-MB-231 cells (Fig. S4b). Thus, the weak anti-tumor capacity of the CAPs tryptophan rich DMEM on MDA-MB-231 cells conclusively demonstrates that tryptophan is the second most sensitive amino acid to the effective species. In addition, the CAP treated DMEM tends to cause the most cell death among most CAPs amino acid rich DMEM. Thus, the effective species in CAP reacts with wide range of amino acids. Other than cysteine and tryptophan, arginine, lysine, asparagine, and glutamine also react significantly with the effective species in the CAPs media. Thus, they should also be the hot targets of effective species.
Furthermore, we studied the generation of H 2 O 2 and NO in the CAPs amino acid rich DMEM, which will reveal the species reacting strongly with specific amino acids. In contrast to NO, H 2 O 2 is more likely to be the effective species, because no significant difference of NO generation exists between different CAPs amino acids rich DMEM (Fig. 8a), while H 2 O 2 in the CAP simulated amino acid DMEM varied specifically with the amino acids in DMEM. Most of the hot targeted amino acids of effective species are also capable of consuming H 2 O 2 significantly in the CAPs DMEM (Fig. 8b). There are 15 amino acids that consume the H 2 O 2 generated in the CAPs DMEM, which is consistent with the conclusion made above that many amino acids react with the effective species in the CAP treated DMEM. Among them, cysteine rich DMEM consumes almost all H 2 O 2 . Despite tryptophan is the second most sensitive amino acids to H 2 O 2 , it just consumes about 23% of H 2 O 2 as cysteine does.

Cancer cells consume the reactive species with a cell line-dependent pattern.
Due to the obscure essence of the effective species in the CAP, the consumption speed of the effective species by cancer cells has not been reported. Briefly, in this study, we harnessed one cancer cell line (MDA-MB-231) as the cell probe to investigate how much the effective species were left in the CAPs media which had been used to culture three cancer cell lines for a period of time (Fig. S5). We denoted such CAPs media and such time as residual media and consumption time, respectively. The viability of cell probes (MDA-MB-231) was inversely proportional to the concentration of residual effective species in the residual media. The detailed protocols were illustrated in Fig. S5. It has been found that the effective species in the residual media are gradually consumed by all three cell lines, which causes the viability of cell probe (MDA-MB-231 cells) cultured in residual media to increase as the consumption time increases (Fig. 9a). In addition, the cell probe (MDA-MB-231 cells) cultured in the residual media which has been used to culture U87 cells obtains higher cell viability than the cell probe cultured in the residual media which has been used to culture MDA-MB-231 cells and MCF-7 cells (Fig. 9a). In other words, U87 cells consume the effective species significantly faster than MDA-MB-231 cells and MCF-7 cells. For U87 cells, the residual CAPs media almost completely lose its anti-cancer capacity 3 hours after treatment. In contrast, for MDA-MB-231 cells and MCF-7 cells, the CAPs media still maintain significant anti-cancer capacity 3 hours after treatment. MDA-MB-231 cells and MCF-7 cells consumes the effective species similarly.
Because H 2 O 2 was regarded as the main effective species in the CAPs media, we further investigated the decay speed of H 2 O 2 in the residual media by measuring the evolution of H 2 O 2 in the CAPs media which had been used to culture three cell lines. We found that the H 2 O 2 in the residual media which had been used to culture U87 cells decayed noticeably faster than the cells in the residual media which had been used to culture MDA-MB-231 cells and MCF-7 cells (Fig. 9b)

Discussion
So far, the anti-tumor capacity of CAP treatment has been regulated by controlling the treatment time 3 , the gas sources composition 34,45 , the gas flow rate 34 , and the supply voltages 46 . Gold nanoparticles 6,9 and some small molecules such as osmolytes 47 and 2-deoxy-d-glucose 48 were also used to obtain a synergistic anti-cancer effect. In this study, we demonstrated several methods to obtain stronger anti-cancer capacity of the CAPs media. Specifically, the CAP treatment should be performed in a container with a large diameter. In addition, the gap between the plasma source and the media should be adequately small. To obtain high reactive species concentration, the volume of media should be relative small. Because these principles are fit for all cell lines involved in this study, they may be universal for other cancer cells.
Because 20 amino acids are the building blocks of all proteins, cysteine, the most reactive amino acid determined by cell probes, should also be the main targeted amino acid residue on the intracellular proteins or other small molecules. Intracellular ROS levels and redox balance are tightly regulated by multiple antioxidant defense systems, including small antioxidant molecules such as glutathione, NAD(P)H and ROS-scavenging enzymes such as catalase, superoxide dismutase, glutathione peroxidase, glutathione reductase, thioredoxin, thioredoxin reductase, and peroxiredoxin 49 . Redox status of thiol group on the cysteine residue directly determines the function of glutathione peroxidase 50 , thioredoxin 51 , thioredoxin reductase 52,53 , and peroxiredoxin 54 . When thiold group on the cysteine residue was oxidized, corresponding normal function of some anti-oxidant enzymes or small molecules will be completely lost. Here, we take glutathione, the most abundant small molecular thiol inside mammalian cells 55 , as an example. Reduced form of glutathione (GH) is a tripeptide with a cysteine residue in the middle. When GH is oxidized, the cysteine residue will form a disulfide bond with the cysteine residue on another GH. Such product is denoted as GSSG. GSSG loses the ability to be a anti-oxidant molecule 55 . Actually, the weakend antioxidant system has been observed in the plasma treated cancer cells 15,56 . The weakened intracellular anti-oxidant system facilitates the attack of extracellular ROS on cells, and ultimately results in serious ROS rise 4,24 and oxidative damage including DSB 1,17 and carbonyl content formation 15 .  In contrast to H 2 O 2 , NO was not likely to be the main effective species. However, the real role of NO in killing cancer cells has not been deeply explored in this study. Theoritically, NO may result in the general observed ROS increase and the death of cancer cells after CAP treatment. NO is able to increase intracellular ROS levels by blocking the electron transport chain in mitochondria 58 and inactivating the glutathione peroxidase 59 . NO reacts with superoxide to generate peroxynitrite, which will not only attack DNA but also weaken the antioxidant system via the inactivation of manganese-superoxide dismutase 60,61 , glutathione peroxidase 62 , glutathione reductase 63,64 , catalase 65 , and peroxiredoxin 66 . In addition, other effective species which have not been studied may also contribute to the death of cancer cells. Thus, the number of species that codetermine the anti-cancer capacity of CAPs media is still unknown.
Recently, two trends regarding the response of cancer cells to the CAP treatment have been found. First, the anti-tumor capacity of CAP on cancer cells is proportional to the growth speed of cancer cell lines 67 . Second, the cancer cells carrying mutated p53 genes are more vulnerable to the CAP treatment than the cancer cells carrying wild p53 genes 68 . The second trend might explain the first trend to some extent. Cancer cell lines carrying the mutated p53 gene tend to obtain malignant 69 or metastasis phenotypes 70,71 and overcome growth arrest and senescence 72 . The cancer cells in metastasis stage own prosperous metabolism and high ROS level 73,74 . Thus, the fast growing cancer cells without a normal, functional p53 gene are more vulnerable to the CAP treatment. In this study, however, we found a new trend that the cancer cells that could absorb or eliminate the effective species in the surrounding environment faster would be more resistant to the CAPs media. Despite the fact that the intrinsic relationship between these trends was obscure, we demonstrated that the absorption capacity on the effective species and H 2 O 2 by cancer cells significantly varied with the cell lines.
The absorption of reactive species, mainly H 2 O 2 in this study, directly relates to the diffusion speed of reactive species across the cellular membrane. H 2 O 2 has been regarded as a molecule that is able to freely cross the phospholipid membrane 75 . Recent investigations revealed that diffusion of H 2 O 2 across phospholipid membrane was limited by the membrane composition 75 . Due to the highly similarity among H 2 O 2 and water, aquaporins (AQPs), a membrane protein family facilitating the transport of water across the cellular membrane, also plays an important role in facilitating the passive diffusion of H 2 O 2 76 . Not all AQPs are able to transport H 2 O 2 . So far, only the transport of H 2 O 2 by AQP1 and AQP8 has been investigated. AQP1 with a relative smaller pore diameter in the selectivity filter region (2.7 Å) cannot transport H 2 O 2 75,77 . By contrast, due to the larger pore diameter (3.2 Å) 56 , AQP8 is able to transport H 2 O 2 across the cellular membrane 78 . AQPs are expressed to varying degrees in different types of human tumors. For example, AQP1, 4, and 5 highly express in breast cancer cell lines 79,80 . AQP1, 4, 8, and 9 highly express in glioblastoma cell lines [81][82][83] . The distinct expression pattern of AQP8 in glioblastoma cells and breast cancer cells can explain why H 2 O 2 is consumed faster by U87 cells than MCF-7 and MDA-MB-231. Nonetheless, if H 2 O 2 is the main effective species to kill cancer cells, the intracellular H 2 O 2 level should be at least codetermined by the diffusion speed of H 2 O 2 across the cell membrane and the intracellular H 2 O 2 scavenging system 75 . Thus, we may not be able to predict the vulnerability of cancer cells to the CAPs media only based on the distinct consumption speeds of cancer cells. More research focusing the intracellular H 2 O 2 scavenging capacity in distinct cancer cells should be carried out in the future.

Conclusions
In summary, we demonstrated several principles to optimize the anti-tumor capacity of the CAPs media on glioblastoma cells and breast cancer cells. Specifically, a larger well, a closer gap between plasma source and media, and To obtain cC, we first measured the H 2 O 2 concentration in the media which has been treated by CAP for 0.5, 1, 1.5, and 2 min. Then, the linear fitting between the measured H 2 O 2 concentration and the treatment time helped us to get cC. Results are presented as the mean ± s.d. of three repeated experiments performed in sextuplicate. Student's t-test was performed, and the significance compared with the first bar (nominal concentration of 1 μ M) is indicated as *p < 0.05, **p < 0.01, ***p < 0.005. a smaller volume of media produce a stronger anti-cancer CAPs media. Breast cancer cells are more vulnerable to the CAPs media than glioblastoma cells. In addition, compared with NO, H 2 O 2 in the CAPs media is likely to be the main effective species to kill cancer cells. The effective species in the CAPs media mainly react with cysteine, which explains the rise of intracellular ROS in the CAP treated cancer cells. H 2 O 2 and the CAPs media cause distinct selective killing patterns in cancer cells, indicating that other reactive species may also affect the death of cancer cells. Glioblastoma cells are able to consume effective species and H 2 O 2 in the CAPs media significantly faster than breast cancer cells, which may relate to the distinct expressions of cell membrane proteins on cancer cells. The dose-dependent NO, H 2 O 2 accumulation in the CAPs media. First, 1 mL of complete media in a well on 6-well plate (Falcon) was treated by CAP for 0.5 min, 1 min, 1.5 min, and 2 min, respectively. The gap between the bottom of the quartz tube and the bottom of plate was 3 cm. 50 μ L of CAPs complete media was immediately transferred to a well on a black 96-well clear bottom plate (Falcon) in triplicate. As the control, 50 μ L of untreated complete media was also transferred to a well on the same plate in triplicate. Next, according to the standard protocols provided by Promega and Sigma-Aldrich, the NO and H 2 O 2 concentration in the CAPs media were measured, respectively. The absorbance at 540 nm and the fluorescence at 540/590 nm were read using a H1 microplate reader (Hybrid Technology).

Methods
The liquid surface-dependent NO, H 2 O 2 accumulation in the CAPs media. First, 1 mL of complete media in a well on 48-well, 24-well, 12-well, and 6-well plate (Falcon) were respectively treated by CAP for 1 min. The gap between the outlet of the quartz tube and the bottom of plate was 3 cm. Then, 50 μ L of CAPs media from different multi-well plates were immediately transferred to a well on the black 96-well clear bottom plate in triplicate. For the control, 50 μ L of untreated CAPs media was also transferred to a well on the same plate in triplicate. Ultimately, we measured the NO/H 2 O 2 concentration in the CAPs media.
The dose-dependent .OH accumulation in the MB solution. The MB solution was prepared by dissolving MB powder into deionized water. Then, 1 mL of 0.01 g/L MB solution in a well on 6-well plate was treated by CAP for 0.5 min, 1 min, 1.5 min, and 2 min. The gap between the outlet of the quartz tube and the bottom of plate was 3 cm. 100 μ L of CAPs MB solution was immediately transferred to a well on the black 96-well clear bottom plate in triplicate. As the control, 100 μ L of untreated MB solution was also transferred to a well on the same plate in triplicate. Ultimately, we measured the absorbance at 664 nm using a H1 microplate reader (Hybrid Technology).
The liquid surface-dependent .OH accumulation in the MB solution. First, 1 mL of 0.01 g/L MB solution in a well on 48-well, 24-well, 12-well, and 6-well plate were respectively treated by CAP for 1 min. The gap between the outlet of quartz tube and the bottom of plate was 3 cm. Next, 100 μ L of CAPs MB solution was transferred to a well on the black 96-well clear bottom plate in triplicate. 100 μ L of untreated MB solution was also transferred to a well on the same plate in triplicate as the control. Ultimately, we measured the absorbance at 664 nm using a H1 microplate reader (Hybrid Technology).
The dose-dependent anti-cancer capacity of CAPs media. For each cell line, the protocol was identical. Here, we used U87 cells as an example. First, U87 cells were seeded in 96-well plate with three confluencies (2 × 10 4 cells/ml, 4 × 10 4 cells/ml, and 8 × 10 4 cells/ml) and cultured in an incubator for 24 hours under standard conditions. Next, 1 mL of complete media in a well on 6-well plate was respectively treated by CAP for 0.5 min, 1 min, 1.5 min, and 2 min. The gap between the outlet of the quartz tube and the bottom of plate was 3 cm. 100 μ L of CAPs media were immediately transferred to culture U87 cells in a well on the 96-well plate in sextuplicate. 100 μ L of untreated complete media was also transferred to culture U87 cells in a well on the same plate in sextuplicate as the control. Before this step, the media that was used to culture U87 cells overnight was discarded. After that, U87 cells were cultured in the CAPs media for 72 hours. Ultimately, according to the standard method, the viability of U87 cells were qualified by MTT test and were read by a H1 microplate reader (Hybrid Technology) at the absorbance of 570 nm.
The well size-dependent anti-cancer capacity of CAPs media. The protocols for three cell lines were identical. Here, we used U87 cells as an example. First, U87 cells were seeded in a 96-well plate with a confluence of 2 × 10 4 cells/ml and were cultured in an incubator for 24 hours under standard conditions. Next, 1 mL of complete media in a well on 48-well, 24-well, 12-well, and 6-well plate were treated with CAP for 1 min. The gap between the outlet of the quartz tube and the bottom of plate was 3 cm. Then, 100 μ L of CAPs media were immediately transferred to culture U87 cells in a well on the 96-well plate in sextuplicate. As the control, 100 μ L of untreated media was also transferred to culture U87 cells on the same plate in sextuplicate. Before this step, the media that had been used to culture U87 cells overnight was discarded. After that, U87 cells were cultured in the CAPs media for 72 hour. Ultimately, the cell viability was measured.
The gap-dependent NO/H 2 O 2 accumulation in the CAPs media. First, when the gap between the outlet of the quartz tube and the bottom of the plate varied from 2 cm to 4 cm, 1 mL of complete media in a well on 6-well plate was treated by CAP for 1 min. Then, 50 μ L of CAPs media was immediately transferred to a well on the black 96-well clear bottom plate in triplicate. 50 μ L of untreated complete media was also transferred to a well on the same plate in triplicate as the control. Ultimately, we measured the NO/H 2 O 2 concentration in the CAPs media.
The gap-dependent .OH accumulation in the MB solution. First, when the gap between the outlet of the quartz tube and the bottom of the plate varied from 2 cm to 4 cm, 1 mL of 0.01 g/L MB solution in a well on 6-well plate was treated by CAP for 1 min. Next, 100 μ L of CAPs MB solution was transferred to a well on the black 96-well clear bottom plate in triplicate. 100 μ L of untreated 0.01 g/L MB solution was also transferred to the same plate in triplicate for the control. We measured the absorbance at 664 nm using a H1 microplate reader (Hybrid Technology).
The gap-dependent anti-cancer capacity of the CAPs media. The protocol for the three cell lines was identical. Here, we used U87 cells as an example. First, U87 cells were seeded in 96-well plate with a confluence of 2 × 10 4 cells/ml and were cultured in incubator for 24 hours under standard conditions. Next, when the gap between the outlet of the quartz tube and the bottom of plate varied from 2 cm to 4 cm, 1 mL of complete media in a well on a 6-well plate was treated by CAP for 1 min. Then, 100 μ L of CAPs media were immediately transferred to culture U87 cells in a well on 96-well plate in sextuplicate. 100 μ L of the untreated complete media was also transferred to culture U87 cells in a well on 96-well plate in sextuplicate as the control. Before this step, the media that was used to culture U87 cells overnight was discarded. U87 cells were then cultured in the CAPs media for 72 hours. Ultimately, the cell viability was measured.
The volume-dependent NO/H 2 O 2 accumulation in the CAPs media. First, 1 mL, 2 mL, 3 mL, and 4 mL of complete media in a well on 6-well plate were treated by CAP for 1 min. The gap between the outlet of the quartz tube and the bottom of the plate was 3 cm. 50 μ L of CAPs complete media was immediately transferred to a well on the black 96-well clear bottom plate in triplicate. 50 μ L of the untreated complete media was also transferred to a well on the same plate in triplicate for the control. Ultimately, we measured the NO/H 2 O 2 concentration in the CAPs media.
The volume-dependent .OH accumulation in the MB solution. 1 mL, 2 mL, 3 mL, and 4 mL of 0.01 g/L MB solution in a well on 6-well plate were treated by CAP for 1 min. The gap between the outlet of the quartz tube and the bottom of the plate was 3 cm. 100 μ L of CAPs MB solution was transferred to a well on a black 96-well clear bottom plate in triplicate. 100 μ L of untreated 0.01 g/L MB solution was also transferred to a well on the same plate in triplicate for the control. Ultimately, we measured the absorbance at 664 nm using a H1 microplate reader.
The volume-dependent anti-cancer capacity of the CAPs media. The protocols for three cell lines were identical. Here, U87 cells were used as an example. First, U87 cells were seeded in 96-well plate with a confluence of 2 × 10 4 cells/ml and were cultured in incubator for about 24 hours under standard conditions. Second, 1 mL, 2 mL, 3 mL, and 4 mL of complete media in a well on 6-well plate were treated by CAP for 1 min. The gap between the outlet of the quartz tube and the bottom of the plate was 3 cm. Then, 100 μ L of CAPs complete media was immediately transferred to culture U87 cells in a well on a 96-well plate in sextuplicate. 100 μ L of the untreated complete media was also used to culture U87 cells in a well on same plate in sextuplicate for the control. Before this step, the media which has been used to culture U87 cells overnight was discarded. After that, U87 cells were cultured in the CAPs media for 72 hours. Ultimately, the cell viability was measured.
The anti-cancer capacity of CAPs amino acids rich DMEM. The protocol for three cell lines was identical. Here, U87 cells were used as an example. First, U87 cells were seeded in 96-well plate with a confluence of 2 × 10 4 cells/ml and were cultured in an incubator for 24 hours under standard conditions. Second, we respectively prepared 2.4 mM specific amino acids rich DMEM by dissolving specific quantities of amino acids powers (Sigma-Aldrich) in DMEM (1% ABS). Because some amino acid rich DMEM would be gradually oxidized by the air during long storage even in the refrigerator, all prepared amino acid rich DMEM were discarded and renewed every two weeks. Next, all 20 specific amino acid rich DMEM and normal DMEM were treated by CAP for 1 min in a well on 6-well plate. The volume of solution in each well was 1 mL. The gap between the outlet of the quartz tube and the bottom of the plate was 3 cm. 100 μ L of the CAPs amino acids rich DMEM and the CAPs normal DMEM were transferred to culture U87 cells in a well on 96-well plate in sextuplicate. As the control, 100 μ L of the untreated amino acid rich DMEM and normal DMEM were also used to culture U87 cells in a well on 96-well plate in sextuplicate. Before this step, the media which has been used to culture U87 cells overnight was discarded. After that, U87 cells were then cultured in the CAPs media for 72 hours. Ultimately, the cell viability was measured.

Measurement of NO/H 2 O 2 in the CAPs amino acids rich DMEM.
First, all 20 specific amino acids rich DMEM and normal DMEM were treated by CAP for 1 min in a well on 6-well plate. The volume of solution in each well was 1 mL. The gap between the outlet of the quartz tube and the bottom of the plate was 3 cm. 50 μ L of the CAPs amino acid rich DMEM and the CAPs normal DMEM were immediately transferred to a well on the black 96-well clear bottom plate in triplicate. As the control, 50 μ L of the untreated amino acid rich DMEM and normal DMEM were also transferred to a well on the same plate in triplicate. Ultimately, we measured the NO/ H 2 O 2 concentration in the CAPs media.

Measurement of the consumption speed of H 2 O 2 by cancer cells.
The protocol for the three cell lines is identical. Here, we used U87 cells as an example. First, U87 cells were seeded in 96-well plate with a confluence of 1 × 10 4 cells/ml and cultured in the incubator for 6 hours under the standard conditions. Then, 1 mL of complete media was treated by CAP in 6-well plate for 1 min. After that, 120 μ L of the CAPs media was transferred to culture Scientific RepoRts | 5:18339 | DOI: 10.1038/srep18339 U87 cells on a 96-well plate. Since then, until the third hour, 50 μ L of the residual media that was used to culture U87 cells was transferred to a well on the black 96-well clear bottom plate in triplicate. As the control, 100 μ L of new complete media was also transferred to a well on the same plate in triplicate. Ultimately, we measured the H 2 O 2 concentration in the CAPs media.
The anti-cancer effect of H 2 O 2 rich media. The protocol for the three cell lines was identical. Here, U87 cells were used as an example. First, U87 cells were seeded in 96-well plate with a confluence of 2 × 10 4 cells/ml and were cultured in an incubator for 24 hours under standard conditions. Then, 1 μ M, 5 μ M, 10 μ M, 20 μ M, 25 μ M, 50 μ M, and 100 μ M H 2 O 2 rich media was prepared by mixing 30 wt % H 2 O 2 solution (Sigma-Aldrich) with the complete media. 100 μ L of these H 2 O 2 rich media was transferred to culture U87 cells in a well on 96-well plate in sextuplicate. 100 μ L of normal complete media was also used to culture U87 cells in a well of the same plate in sextuplicate for the control. Before this step, the media which was used to culture U87 cells overnight was discarded. After that, U87 cells were then cultured in the incubator for 72 hours. Ultimately, the cell viability was measured.