Dynamics of Singlet Oxygen-Triggered, RONS-Based Apoptosis Induction after Treatment of Tumor Cells with Cold Atmospheric Plasma or Plasma-Activated Medium

Treatment of tumor cells with cold atmospheric plasma (CAP) or with plasma-activated medium (PAM) leads to a biochemical imprint on these cells. This imprint is mediated by primary singlet oxygen, which is mainly generated through the interaction between CAP-derived H2O2 and NO2−. This imprint is induced with a low efficiency as local inactivation of a few membrane-associated catalase molecules. As sustained generation of secondary singlet oxygen by the tumor cells is activated at the site of the imprint, a rapid bystander effect-like spreading of secondary singlet oxygen generation and catalase inactivation within the cell population is thus induced. This highly dynamic process is essentially driven by NOX1 and NOS of the tumor cells, and finally leads to intercellular RONS-driven apoptosis induction. This dynamic process can be studied by kinetic analysis, combined with the use of specific inhibitors at defined time intervals. Alternatively, it can be demonstrated and quantified by transfer experiments, where pretreated cells are mixed with untreated cells and bystander signaling is determined. These studies allow to conclude that the specific response of tumor cells to generate secondary singlet oxygen is the essential motor for their self-destruction, after a singlet oxygen-mediated triggering process by CAP or PAM.


A.1 Kinetic analysis of the biological effects of primary singlet oxygen ( 1 O 2 )
derived directly from CAP. The experiment was performed as described in Supplementary Figure 1, with the modification that treatment with CAP was in the presence of the catalase mimetic EUK-134 instead of FeTPPS. Statistical analysis: A: Apoptosis induction at 5.5 h and later was highly significant (p<0.001). Inhibition by histidine during pretreatment and after the washing step was highly significant (p<0.001). B, C: Apoptosis induction at 5.5 h and later, as well as inhibition by histidine added at 0 min past washing and by taurine was highly significant (p<0.001).
When CAP treatment in the presence of either AEBSF plus FeTPPS or AEBSF plus EUK-134 was performed for 1 min or 5 min, and then the cells were washed and further incubated in fresh medium, cells that had been treated for 5 min started with apoptosis induction two hours earlier than the cells that had been treated for 1 min (Supplementary Figure 3). Control experiments in which histidine or FeTPPS were Supplementary Figure 3 Supplementary Figure 3. This figure analyses the kinetics of tumor cells that had been pretreated with CAP for 1 or 5 min under conditions that prevented primary 1 O 2 generation from long-lived species from CAP, and also inhibited the secondary 1 O 2 generation, i. e. in the presence of FeTPPS or alternatively EUK-134, and AEBSF. The data show the profound effect of treating time on the starting point of the kinetics of apoptosis induction.

Statistical analysis: Apoptosis induction defined by open crosses (3 h and later) and by open circles (5.5 h and later), as well as inhibition by all inhibitors was highly significant (p<0.001).
added to cells that had been treated with CAP for 5 min and then had been washed, showed that apoptosis induction after the washing step was dependent on secondary 1 O 2 with ONOO ─ as an essential intermediate for its generation. This finding demonstrates that the length of treatment with primary 1 O 2 directly derived from CAP determined the onset of apoptosis. This is consistent with the concept of bystander signaling shown in Figure 15, as more hits by primary 1 O 2 should increase the number of starting points for bystander signaling and thus have a direct impact on the onset of apoptosis induction.

A.2 Quantitation of the effect of secondary 1 O 2
The effect of secondary 1 O 2 was determined in experiments that utilized the potential of CAP-treated cells to trigger a bystander-like effect in untreated cells. The percentage of pretreated cells that are required to lead to apoptosis induction in an untreated population allow us to estimate the spreading of secondary 1 O 2 in the population.
Treatment of tumor cells with CAP for 1 min, immediately followed by a washing step was not sufficient to allow these cells to trigger apoptosis induction 5.5 h after mixing to untreated cells ( Supplementary Figure 4 A). However, when the CAP-treated cells were incubated in the same medium for 25 min before they were washed and added to untreated cells, a remarkable bystander signaling-inducing effect was observed.
As little as 0.4 % of treated cells within the population (i. e. 50 treated cells per 12,500 cells in total) allowed for nearly maximal apoptosis induction. This result indicates that during the 25 min incubation after CAP treatment, 1 O 2 derived from the long-lived species in plasma-treated medium and the resultant secondary 1 O 2 generated by the triggered tumor cells must have imprinted the potential to induce bystander signaling in a large number of cells. As selective inhibition of secondary 1 O 2 generation by the NOX1 inhibitor AEBSF caused drastically reduced bystander signaling (reflected in the need to transfer large numbers of treated cells to untreated cells for triggering apoptosis induction), secondary 1 O 2 generated by triggered tumor cells seemed to contribute the most to this effect, whereas the effect of primary 1 O 2 seemed to play a minor role only. In the presence of histidine, FeTPPS or EUK-134 during CAP treatment and initial 25 min incubation, no apoptosis induction was

B. 2 Visualization of the dynamic processes triggered by primary and secondary singlet oxygen ( 1 O 2 )
Supplementary Figure 17  Finally, catalase inactivation is sufficient to allow HOCl signaling that causes apoptosis through the mitochondrial pathway of apoptosis.
Inactivated membrane-associated catalase on the surface of tumor cells acts as an imprint that will allow triggering of secondary 1 O 2 in neighboring cells.
The transfer of few cells from a population with established autoamplification of catalase inactivation to untreated tumor cells is sufficient to trigger the onset of autoamplification of secondary 1 O 2 generation in this cell population. The number of cells that are necessary to transfer bystander signaling can be taken as a measure for the frequency of imprinted cells.
When the treatment of tumor cells with CAP or PAM is performed in the presence of the NOX1 inhibitor AEBSF (Suppl. Figure 18), the imprint by primary 1 O 2 is set, but the generation of secondary 1 O 2 and its autoamplification are hindered. As AEBSF is a reversible inhibitor, it can be washed away. Its removal will cause the start of secondary 1 O 2 generation and autoamplication. The kinetics of final apoptosis induction will be delayed compared to control assays without temporary AEBSF treatment. This scenario explains the kinetic differences seen in Figures 8-10 Fig.19 A). Most of the primary 1 O 2 from the gaseous phase of CAP will react with medium components on top of the assay, and only a few 1 O 2 from this source will reach tumor cells (in suspension) and inactivate catalase. H 2 O 2 and NO 2 ─ will generate 1 O 2 and a few 1 O 2 molecules from this source will inactivate catalase.
Our measurements showed that 1 O 2 derived from H 2 O 2 and NO 2 ─ is inactivating catalase more frequently than 1 O 2 derived directly from the gaseous phase of CAP.
At the site of inactivated catalase, secondary 1 O 2 is constantly generated through the interaction between NOX1 and NOS of the tumor cells (Suppl. Fig. 19 B; see Suppl. However, the generation of secondary 1 O 2 is prevented by FeTPPS as well. Therefore, resumption of the secondary 1 O 2 generation and its autoamplification can only start after FeTPPS has been removed from the system (D). As autoamplification and subsequent processes now start from very few initiation points, the resulting kinetics of apoptosis induction is strongly delayed, as has been shown in Figures 9 C, 10 C, 11 C and others of the main manuscript.
Supplementary Figure 19 Supplementary Figure 19

B.4 Autoamplification of secondary 1 O 2 generation on the initially triggered tumor cell and adjacent cells
The experimental proof for bystander effect-inducing potential of tumor cells with This process causes depletion of intracellular glutathione but is not sufficient by itself to cause apoptosis induction. Glutathione depletion renders the cells sensitive for the apoptosis inducing effect of HOCl and/or • NO/ONOO ─ signaling (#8), as lipid peroxidation through • OH generated by both pathways at the membrane of the tumor cells can no longer be repaired by glutathione peroxidase-4 in the absence of an optimal glutathione concentration. Lipid peroxidation then triggers the mitochondrial pathway of apoptosis, which is executed by caspase-9 and caspase-3 (#9) and leads to the cell death (#10).
CAP and PAM-mediated apoptosis induction shows some additional side reactions which may provide for further enhancement and synergistic effects, as recently outlined (Bauer, 2019 c). As the FAS receptor can also be activated by 1 O 2 (instead of its genuine ligand) (Zhuang et al.2001), primary and secondary 1 O 2 also have a chance to activate the FAS receptor (#11) instead of interacting with catalase. In most tumor cell systems, this signal will not be sufficient to cause FAS receptormediated apoptosis (Bauer, 2015), but will induce strong enhancement of NOX1 and NOS activities (#12) (Suzuki et al., 1998;Reinehr et al., 2005;Selleri et al, 1997;reviewed in Bauer, 2015). The resultant increase in free O 2 • ─ and • NO causes a profound enhancement of secondary 1 O 2 generation and thus essentially enhances the autoamplificatory process of 1 O 2 generation and potentially also the efficiency of subsequent apoptosis-inducing RONS signaling.
As SOD also carries an essential histidine residue in its active center (Escobar et al., 1996;Kim et al., 2001) (Bauer and Motz, 2016). Therefore, the potential of 1 O 2 to either inactivate catalase or SOD may be one of the clues for the efficiency of CAP and PAM action towards tumor cells, as it bears an inherent potential for the induction of a valuable synergistic effect.
Suppl. Figure 25: Suppl. Figure 25. Complete flow chart of apoptosis induction in tumor cells by CAP and PAM. Please find the explanation in the text.

B. 6 Comparison to other models for CAP and PAM-mediated apoptosis induction
The unique features of our model ( in chemical biology to achieve site-specific • OH generation, which would not be possible with randomly operating Fenton chemistry. The models suggested by Yan et al. (2015aYan et al. ( , 2017a and Van der Paal et al., (2017) follow a completely different line, as they are based on the concept that CAP or PAMderived H 2 O 2 is the essential factor that determines selective apoptosis induction in tumor cells. Yan et al. (2015aYan et al. ( , 2017a (Deichman and Vendrov, 1986;Deichman et al., 1989Deichman et al., . 1998Deichman 2000Deichman , 2002. This tumor progressionassociated resistance towards exogenous H 2 O 2 is based on the expression of membrane-associated catalase (Heinzelmann and Bauer, 2010;Bechtel and Bauer, 2009 a, b;Böhm et al., 2015),   Figure 19 of the main manuscript. Therefore, the model for CAP and PAM action, as proposed by us in this and the preceding manuscript also includes essential elements established by Keidar`s group.
Attri and Bogaerts (2019) 2018 a, b, c, d). Moreover, the enhancement of Fenton chemistry through the addition of ferrous ion to an assay with ongoing HOClmediated apoptosis induction in malignant cells caused a drastic inhibition of apoptosis induction, due to the shift from site-directed to random generation of • OH (Bauer, 2013).
Furthermore, as noted above, it is hard to see how models based on the concept that ROS/RNS in CAP and PAM are directly responsible for the induction of cell death in the target cells can explain autoamplification and the observed bystander effects. We note only a few 'activated' cells are sufficient to trigger a massive apoptotic response from a much larger group of non-activated cells. This is a central and strongly supported experimental effect. Any alternative model based on the aforementioned set of experiments would need to explain all of these observations.

B. 7 The potential connection between CAP-and PAM-mediated apoptosis induction and a subsequent specific immune response
One of the most exciting developments in oncology during the last years was the finding that classical tumor therapy, such as chemotherapy, radiotherapy, photodynamic therapy, and others, are not successful, unless they trigger a subsequent immunological attack on the treated tumor (Apetoh et al., 2008;Green et al., 2009;Golden and Apetoh, 2015;Krysko et al., 2012, Kroemer et al., 2013Garg et al., 22014;Candeias and Gaipl, 2016). These findings have led to the concept of "immunogenic cell death (ICD)", with its central theme that the activation of dendritic cells through defined signaling factors ("Damage-associated molecular patterns (DAMPs") released by the dead cells leads to the activation of a cytotoxic T cell response. This T cell response is essential for the complete eradication of the tumor and can also act at sites in the body that are distant of the tumor, e. g. metastases Hodge et al. (2013) have recognized that besides "classical ICD" (strictly defined by induction of cell death and by DAMPs), an " immunogenic modulation" of tumor cells can enhance killing by cytotoxic T lymphocytes. This mechanism is distinct from classical immunogenic cell death but leads to an analogous result. Certain chemotherapeutics, which do not cause immunogenic cell death, have been shown to have the potential for immunogenic modulation. In some cases, effective immunogenic modulation did not even require a preceding cell death, whereas in the case of immunogenic cell death, induction of the cell death is part of the definition.
Consistent with the concept of immunogenic modulation, HOCl has been shown to enhance the recognition of antigens by dendritic cells, and thus to enhance the control of tumors by cytotoxic T cells, both in vitro and in vivo (Chiang et al., 2006(Chiang et al., , 2008(Chiang et al., , 2015Prokopowicz et al., 2010;Biedron et al., 2015;Zhou et al., 2012).
Several groups have shown that cold atmospheric plasma induces immunogenic cell death in vitro and in vivo (Miller et al., 2016;a, b, 2018, Mizuno et al., 2017Bekeschus et al., 2018 a, b). ROS derived from plasma and/or generated within the treated cells seems to be involved in the triggering of ICD. In addition to the function of cytotoxic T cells, tumor cell killing by tumor necrosis factor type alpha from macrophages has been shown to be also involved in tumor cell killing after plasma treatment (Kaushik et al., 2018).
Therefore, it is an open and interesting question, whether RONS-dependent apoptosis induction, which is triggered by long-lived species derived from CAP or PAM and is established through autoamplificatory bystander signaling, is one of the triggers for ICD and/or immunogenic modulation. The answer to this question requires to clarify, whether defined species involved in intercellular RONS signaling are also inducers of DAMPs, or whether these processes are not directly connected.
As apoptosis induction after CAP and PAM treatment requires the action of the HOCl signaling pathway (as shown in our present study), it is not unlikely that it also activates immunogenic modulation according to the mechanisms described by Chiang et al., Prokopowicz, Biedron and Zhou. In this way, reactivation of intercellular HOCl signaling through the CAP-and PAM-mediated autoamplificatory catalase inactivation might be the trigger for another, equally specific superimposed