Comparison of Cellular Death Pathways after mTHPC-mediated Photodynamic Therapy (PDT) in Five Human Cancer Cell Lines

One of the most promising photosensitizers (PS) used in photodynamic therapy (PDT) is the porphyrin derivative 5,10,15,20-tetra(m-hydroxyphenyl)chlorin (mTHPC, temoporfin), marketed in Europe under the trade name Foscan®. A set of five human cancer cell lines from head and neck and other PDT-relevant tissues was used to investigate oxidative stress and underlying cell death mechanisms of mTHPC-mediated PDT in vitro. Cells were treated with mTHPC in equitoxic concentrations and illuminated with light doses of 1.8–7.0 J/cm2 and harvested immediately, 6, 24, or 48 h post illumination for analyses. Our results confirm the induction of oxidative stress after mTHPC-based PDT by detecting a total loss of mitochondrial membrane potential (Δψm) and increased formation of ROS. However, lipid peroxidation (LPO) and loss of cell membrane integrity play only a minor role in cell death in most cell lines. Based on our results, apoptosis is the predominant death mechanism following mTHPC-mediated PDT. Autophagy can occur in parallel to apoptosis or the former can be dominant first, yet ultimately leading to autophagy-associated apoptosis. The death of the cells is in some cases accompanied by DNA fragmentation and a G2/M phase arrest. In general, the overall phototoxic effects and the concentrations as well as the time to establish these effects varies between cell lines, suggesting that the cancer cells are not all dying by one defined mechanism, but rather succumb to an individual interplay of different cell death mechanisms. Besides the evaluation of the underlying cell death mechanisms, we focused on the comparison of results in a set of five identically treated cell lines in this study. Although cells were treated under equitoxic conditions and PDT acts via a rather unspecific ROS formation, very heterogeneous results were obtained with different cell lines. This study shows that general conclusions after PDT in vitro require testing on several cell lines to be reliable, which has too often been ignored in the past.


Introduction
Photodynamic therapy (PDT) is a method for the medical treatment of solid tumors based on the generation of reactive oxygen species (ROS) after photoactivation of a photosensitizer (PS). The PS is excited by light of a certain wavelength and then transfers its energy to triplet oxygen to form highly reactive singlet oxygen ( 1 O 2 ) or generates other ROS via a direct electron transfer to biological substrates, e.g., components of the cellular membrane. The ROS formed lead to oxidative stress within the cells of the treated tumor and eventually cause cellular death [1,2]. Thereby, a certain degree of selectivity can be achieved by restricting the illumination area to the site of the tumor.  were treated with mTHPC for 24 h in concentrations ranging from 0.001-5.0 µM and kept in the dark (dark toxicity) or illuminated with a light dose of 1.8 J/cm 2 (light-induced toxicity). MTT assay was carried out 24 h post illumination and the absorbance of the reduced formazan was measured at λ = 570 nm. The percentage of cell viability was calculated by dividing the absorbance for the treated group by the absorbance in the solvent dark control. IC50 and IC90 values were calculated by using Prism 6. Results are presented as means ± SD from at least three independent experiments.

Only Minor Induction of Necrosis after mTHPC-PDT
To investigate the role of necrosis as a part of the death mechanism after mTHPC-based PDT, release of lactate dehydrogenase (LDH) was measured after photodynamic treatment. LDH release is a classic assay for estimating damage to cell membranes, which is characteristic of necrosis [18]. Treatment with Triton X-100 resulted in cytotoxicity (see Table 2). Table 2. Established % cytotoxicity as assessed by lactate dehydrogenase (LDH) release assay after treatment of cells with 0.1% (v/v, 6 h) and 0.01% (v/v, 24 h) Triton X-100 positive control after  , and SISO (E) were treated with mTHPC for 24 h in concentrations ranging from 0.001-5.0 µM and kept in the dark (dark toxicity) or illuminated with a light dose of 1.8 J/cm 2 (light-induced toxicity). MTT assay was carried out 24 h post illumination and the absorbance of the reduced formazan was measured at λ = 570 nm. The percentage of cell viability was calculated by dividing the absorbance for the treated group by the absorbance in the solvent dark control. IC 50 and IC 90 values were calculated by using Prism 6. Results are presented as means ± SD from at least three independent experiments.

Only Minor Induction of Necrosis after mTHPC-PDT
To investigate the role of necrosis as a part of the death mechanism after mTHPC-based PDT, release of lactate dehydrogenase (LDH) was measured after photodynamic treatment. LDH release is a classic assay for estimating damage to cell membranes, which is characteristic of necrosis [18]. Treatment with Triton X-100 resulted in cytotoxicity (see Table 2). Table 2. Established % cytotoxicity as assessed by lactate dehydrogenase (LDH) release assay after treatment of cells with 0.1% (v/v, 6 h) and 0.01% (v/v, 24 h) Triton X-100 positive control after incubation in the dark and post illumination with a light dose of 1.8 J/cm 2 (24 h) or 3.5 J/cm 2 (6 h), respectively. Results as means ± SD from at least three independent experiments. The detected % cytotoxicity after treatment of five different cell lines with mTHPC did not exceed values >18.9% (at 0.1 µM mTHPC with 3.5 J/cm 2 6 h after PDT in SISO cells) under any experimental conditions (Figure 2A-E). In general, LDH release was higher 24 h after illumination with a light dose of 1.8 J/cm 2 compared to cells incubated in the dark or analyzed 6 h after illumination with a higher light dose of 3.5 J/cm 2 . LDH release was highest for A-427, RT-4, and SISO cells with peak values of 16.4%, 14.9%, and 17.1% cytotoxicity 24 h post illumination. Remarkably, these values were not observed at the highest, but at low or medium concentrations of mTHPC between 0.1-1.0 µM. The highest detected % cytotoxicity for BHY and KYSE-70 were 7.1% and 2.7% within the same concentration range.  , and SISO (E) were treated with mTHPC for 24 h in concentrations ranging from 0.01-10.0 µM and kept in the dark (dark toxicity) or illuminated with a light dose of 1.8 J/cm 2 or 3.5 J/cm 2 (light-induced toxicity). LDH release assay was carried out 24 or 6 h post illumination, respectively, and the absorbance of the reduced INT was measured at λ = 490 nm. The percentage of cytotoxicity was calculated by dividing the absorbance for the treated group by the absorbance at maximum LDH release. Results are presented as means ± SD from at least three independent experiments.

Increased ROS Generation after mTHPC-PDT
The formation of ROS during oxidative stress is known to cause cellular death after PDT [1,2]. ROS was detected by flow cytometric analysis of DCF fluorescence intensity after staining with 2 ,7 -dichlorodihydrofluorescein diacetate (H 2 DCF-DA) (see Figure A1 in Appendix A for representative analysis data). Fluorescent 2 ,7 -dichlorofluorescein (DCF) is formed after contact with ROS, except singlet oxygen ( 1 O 2 ) [19], and the fluorescence intensity is proportional to ROS levels [20,21]. The results of intracellular ROS analyses are shown in Figure 3A-E. Direct treatment with H 2 O 2 for 10 min led to increases of DCF fluorescence intensity. Solvent-treated cells that were kept in the dark served as the reference sample (fluorescence intensity for this sample was set to 1.0). No elevated ROS levels were observed after treatment with mTHPC in the absence of light or in solvent-treated and illuminated cells.
After mTHPC-PDT, elevated ROS levels were observed, but different mTHPC concentrations and light doses were required to increase ROS formation among the cell lines. After treatment with the IC 90 of mTHPC and PDT with the lowest light dose of 1.8 J/cm 2 , a 3.7-fold and a 3.1-fold increase in ROS levels were observed in KYSE-70 and SISO cells, respectively, but a higher light dose of 3.5 J/cm 2 was necessary for A-427 cells (1.3-fold increase). For BHY cells, a very high light dose of 7.0 J/cm 2 was required to produce a significantly generation of ROS (2.9-fold, IC 50 , and 3.5-fold, IC 90 ) of mTHPC. No significant increase in ROS formation was detected in RT-4 cells for any of the tested concentrations and light doses. the IC90 of mTHPC and PDT with the lowest light dose of 1.8 J/cm 2 , a 3.7-fold and a 3.1-fold increase in ROS levels were observed in KYSE-70 and SISO cells, respectively, but a higher light dose of 3.5 J/cm 2 was necessary for A-427 cells (1.3-fold increase). For BHY cells, a very high light dose of 7.0 J/cm 2 was required to produce a significantly generation of ROS (2.9-fold, IC50, and 3.5-fold, IC90) of mTHPC. No significant increase in ROS formation was detected in RT-4 cells for any of the tested concentrations and light doses.  Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 , 3.5 J/cm 2 , and 7.0 J/cm 2 or left in the dark, stained with H2DCF-DA and DCF fluorescence intensity was measured directly after illumination. Flow cytometric analysis of the single cell population was carried out using the FITC channel (λEx/Em = 488/525-550 nm). Fluorescence intensity was plotted with reference to nonilluminated, solvent-treated cells (fluorescence intensity was set to 1.0). Cells treated with 1.0-2.0 mM  Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 , 3.5 J/cm 2 , and 7.0 J/cm 2 or left in the dark, stained with H 2 DCF-DA and DCF fluorescence intensity was measured directly after illumination. Flow cytometric analysis of the single cell population was carried out using the FITC channel (λ Ex/Em = 488/525-550 nm). Fluorescence intensity was plotted with reference to non-illuminated, solvent-treated cells (fluorescence intensity was set to 1.0). Cells treated with 1.0-2.0 mM H 2 O 2 for 10 min were used as positive control. Data presented as means ± SD from at least three independent experiments. (** p < 0.01; *** p < 0.001; **** p < 0.0001).

Lipid Peroxidation (LPO) Only Plays a Minor Role after mTHPC-PDT
The detection of lipid peroxidation (LPO) was done with a flow cytometer after staining with the LPO sensor BODIPY 665/676 (see Figure A2 in Appendix A for representative analysis data). The dye, which localizes in the cellular membrane, is oxidized upon contact with hydroxyl (OH • ), alkoxyl (RO • ), and peroxyl radicals (ROO • ), leading to a change in the fluorescence spectrum [22,23]. The results of LPO analyses are shown in Figure 4A-E. Treatment with tert-butyl hydroperoxide (t-BHP) as a positive control led to more LPO in all cell lines. Solvent-treated, non-illuminated cells served as the reference sample (fluorescence intensity for this sample was set to 1.0). No enhanced formation of LPO was observed after treatment with mTHPC in the absence of light or in solvent-treated and illuminated cells in any cell line and at any time point.

Total Loss of Mitochondrial Membrane Potential (δψM) after mTHPC-PDT
To evaluate the effects of mTHPC-PDT on mitochondrial membrane potential (∆ψ m ), a widely used procedure was applied for staining the mitochondrion with the cationic dye JC-1. Green fluorescent JC-1 monomers and red fluorescent JC-1 aggregates were visualized by fluorescence microscopy. A decrease in JC-1 polymer aggregation within mitochondria indicates a decline in ∆ψ m , which is a characteristic sign of oxidative stress and apoptosis induction [24,25]. Cytosolic JC-1 monomers (green) were observed in all cell lines irrespective of the treatment 6 h after PDT or incubation in the dark ( Figure 5A-E). High amounts of JC-1 polymers (red) were observed in all cell lines after treatment with solvent or the IC 90 of mTHPC in the absence of light, indicating the presence of active mitochondria with normal ∆ψ m . As a positive control, cells treated carbonyl cyanide m-chloro phenyl hydrazone (CCCP) were used and showed a complete loss of red fluorescence.
Cancers 2019, 11, x 8 of 32 dark ( Figure 5A-E). High amounts of JC-1 polymers (red) were observed in all cell lines after treatment with solvent or the IC90 of mTHPC in the absence of light, indicating the presence of active mitochondria with normal Δψm. As a positive control, cells treated carbonyl cyanide m-chloro phenyl hydrazone (CCCP) were used and showed a complete loss of red fluorescence. After treatment with mTHPC and subsequent application of light, a decrease in red fluorescence was observed in all cell lines. These results indicate that mTHPC-PDT leads to a depolarization of the Δψm and a loss of mitochondrial activity, which is a sign of early apoptosis.  Table 1), illuminated with 1.8 J/cm 2 or left in the dark, and stained with the cationic dye JC-1 after an incubation period of  Table 1), illuminated with 1.8 J/cm 2 or left in the dark, and stained with the cationic dye JC-1 after an incubation period of 6 h post illumination or incubation in the dark. JC-1 monomers and aggregates were visualized with a fluorescence microscope equipped with a 63× oil/1.4 NA objective. JC-1 aggregates within active mitochondria are shown in red, whereas cytosolic JC-1 monomers display a green fluorescence. A decrease in red fluorescence indicates a decline in ∆ψ m , which is a sign of early apoptosis. Fluorescence images were captured with the FITC filter cube (green; λ Ex/Em = 460-500/512-542 nm) and the RHOD filter cube (red; λ Ex/Em = 541-551/565-605 nm). Solvent-treated and non-illuminated cells served as the negative control with active mitochondria. As a positive control, cells were treated with 50 µM CCCP, a mitochondrial oxidative phosphorylation uncoupling agent.
After treatment with mTHPC and subsequent application of light, a decrease in red fluorescence was observed in all cell lines. These results indicate that mTHPC-PDT leads to a depolarization of the ∆ψ m and a loss of mitochondrial activity, which is a sign of early apoptosis.

Induction of Phosphatidylserine Externalization after mTHPC-Mediated PDT
Apoptosis has been shown to be an important route of cellular death involved after photodynamic therapy [26,27]. For the detection of apoptotic cells after mTHPC-based PDT, a flow cytometric analysis was used to visualize cells double-stained with Annexin V-FITC and propidium iodide (PI) (see Figure A3 in Appendix A for representative analysis data). With the known anticancer drug doxorubicin (DOXO), apoptotic cells were detected at all time points ( Figure 6A-E). Compared to a solvent-treated, non-illuminated control, no increases of Annexin V-FITC single stained (apoptotic) or Annexin V-FITC/PI double-stained (late-apoptotic) cells were observed for solvent-treated and illuminated as well as mTHPC-treated, but non-illuminated controls at any time point.
For A-427, the IC 90 in combination with light led to significantly more apoptotic cells compared to the solvent-treated dark control independently of the incubation time. After 6 h, 28.3%, and after 24 h, 37.6% of the cells were Annexin V-FITC-positive, whereas this fraction dropped to 7.9% after 48 h. However, it is noteworthy that at this time point the fraction of late-apoptotic cells reached its peak at 55.1%. A similar pattern was observed after mTHPC-based PDT applied to BHY cells. The amount of apoptotic cells increased over time for the IC 90 from 13.8% (6 h) to 41.5% (48 h Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, stained with Annexin V-FITC and propidium iodide (PI) and analyzed 6 (left), 24 (middle), or 48 h (right) after photodynamic treatment. Flow cytometric analysis of the single cell population was carried out using the FITC channel (λEx/Em = 488/525-550 nm) for the detection of Annexin V-positive cells and the PI channel (λEx/Em = 488/655-730 nm) for PI-positive cells. The percentage of apoptotic cells is plotted on the left axis, while late-apoptotic cells can be seen on the right axis in opposite direction. Non-illuminated, solvent-treated cells served as the reference sample; cells treated with 0.5-5.0 µM DOXO were used as positive control. Data presented as means ± SD from at least three independent experiments. ( * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001).

PARP Cleavage Confirms Induction of Apoptosis after mTHPC-PDT
The induction of apoptosis was also investigated by western blot analysis of PARP and its cleaved form, which is involved in the process of apoptosis [28]. PARP cleavage was observed in all tested cell lines, but not under all conditions ( Figure 7A-E). In A-427, 90.6% of PARP were found in the cleaved form after treatment with the IC90 of mTHPC and illumination. Compared to the other cell lines, A-427 showed the highest amounts of cleaved PARP in the reference samples. In the other cell lines, the percentage of cleaved PARP relative to PARP was significantly increased to 38.1-86.4% (IC50) and 59.1-98.3% (IC90), respectively.  Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, stained with Annexin V-FITC and propidium iodide (PI) and analyzed 6 (left), 24 (middle), or 48 h (right) after photodynamic treatment. Flow cytometric analysis of the single cell population was carried out using the FITC channel (λ Ex/Em = 488/525-550 nm) for the detection of Annexin V-positive cells and the PI channel (λ Ex/Em = 488/655-730 nm) for PI-positive cells. The percentage of apoptotic cells is plotted on the left axis, while late-apoptotic cells can be seen on the right axis in opposite direction. Non-illuminated, solvent-treated cells served as the reference sample; cells treated with 0.5-5.0 µM DOXO were used as positive control. Data presented as means ± SD from at least three independent experiments. ( * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001).

PARP Cleavage Confirms Induction of Apoptosis after mTHPC-PDT
The induction of apoptosis was also investigated by western blot analysis of PARP and its cleaved form, which is involved in the process of apoptosis [28]. PARP cleavage was observed in all tested cell lines, but not under all conditions ( Figure 7A-E). In A-427, 90.6% of PARP were found in the cleaved form after treatment with the IC 90 of mTHPC and illumination. Compared to the other cell lines, A-427 showed the highest amounts of cleaved PARP in the reference samples. In the other cell lines, the percentage of cleaved PARP relative to PARP was significantly increased to 38.1-86.4% (IC 50 ) and 59.1-98.3% (IC 90 ), respectively.  Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, and total protein extracts were harvested 24 h after illumination and western blotting performed. Representative blots are shown and the % cleaved PARP relative to PARP was evaluated by densitometric analysis. GAPDH was used as a loading control. Data presented as dot plots from at least three independent experiments. Solvent-treated, non-illuminated cells served as the reference group. ( ** p < 0.01; *** p < 0.001; **** p < 0.0001).

PARP Cleavage at Least Partly Traced Back to Caspase 3-Activation
Among others, PARP is one of the main cleavage targets of the active effector caspase 3. Therefore, the activation of caspase 3, e.g., via initiator caspases 8 (extrinsic) or 9 (intrinsic), is another common hallmark of apoptotic cell death [29], and was investigated by western blotting. Solventtreated as well as mTHPC treated, but non-illuminated cells only displayed caspase 3 in its active form in low levels ( Figure 8A-E). In general, SISO cells showed a higher background activation of caspase 3 in controls.
Activation of (pro-)caspase 3 was detected in four out of five tested cell lines after mTHPC-PDT, but the fraction of active caspase 3 relative to inactive pro-caspase 3 was significantly increased only after treatment with the IC90. While caspase 3 activation could only be detected to a low extent in RT-4 cells (13.3%) with this concentration, 31.8-69.7% of active caspase 3 were found in the other cell lines. The IC50 of mTHPC initiated caspase 3-activation only in BHY, KYSE-70, and SISO cells (13.5-35.9%), but no significant differences to solvent-treated cells were observed.  Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, and total protein extracts were harvested 24 h after illumination and western blotting performed. Representative blots are shown and the % cleaved PARP relative to PARP was evaluated by densitometric analysis. GAPDH was used as a loading control. Data presented as dot plots from at least three independent experiments. Solvent-treated, non-illuminated cells served as the reference group. (** p < 0.01; *** p < 0.001; **** p < 0.0001).

PARP Cleavage at Least Partly Traced Back to Caspase 3-Activation
Among others, PARP is one of the main cleavage targets of the active effector caspase 3. Therefore, the activation of caspase 3, e.g., via initiator caspases 8 (extrinsic) or 9 (intrinsic), is another common hallmark of apoptotic cell death [29], and was investigated by western blotting. Solvent-treated as well as mTHPC treated, but non-illuminated cells only displayed caspase 3 in its active form in low levels ( Figure 8A-E). In general, SISO cells showed a higher background activation of caspase 3 in controls.
Activation of (pro-)caspase 3 was detected in four out of five tested cell lines after mTHPC-PDT, but the fraction of active caspase 3 relative to inactive pro-caspase 3 was significantly increased only after treatment with the IC 90 . While caspase 3 activation could only be detected to a low extent in RT-4 cells (13.3%) with this concentration, 31.8-69.7% of active caspase 3 were found in the other cell lines. The IC 50 of mTHPC initiated caspase 3-activation only in BHY, KYSE-70, and SISO cells (13.5-35.9%), but no significant differences to solvent-treated cells were observed.  Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, and total protein extracts were harvested 24 h after illumination and western blotting performed. Representative blots are shown and % cas 3 relative to pro-cas 3 was evaluated by densitometric analysis. GAPDH was used as a loading control. Data presented as dot plots from at least three independent experiments. Solvent-treated, non-illuminated cells served as the reference group. ( ** p < 0.01; *** p < 0.001; **** p < 0.0001).

mTHPC-PDT Induces G2/M Arrest and the Formation of Sub G1 Populations with Fragmented DNA, Emphasizing the Induction of Apoptosis
To further investigate whether the cell death mechanism is accompanied by growth inhibition or DNA fragmentation, which is another hallmark during apoptosis [30], cell cycle analysis was carried out after staining with PI. The analysis allowed for the assignment of cells in either sub G1 (fragmented DNA, apoptosis), G0/G1, S, or G2/m phase of the cell cycle (see Figure A4 in Appendix for representative analysis data). The effects on the cell cycle were examined 6, 24, and 48 h after mTHPC-based PDT. Again, no effects on cell cycle distribution were observed after treatment with mTHPC in the absence of light or after illumination of solvent-treated cells for any cell line at any time point compared to solvent-treated cells incubated in the dark ( Figure 9A-E).
For A-427 cells, an increase in sub G1 cells was detected 24 h after mTHPC-PDT leading to 9.6% (IC50) and 17.8% (IC90) of apoptotic cells, respectively. Similar results were obtained for RT-4 and SISO cells. For the RT-4 cell line, sub G1 population was significantly increased 6 h (8.9%) and 48 h (13.4%) after PDT with the IC90 of mTHPC. SISO cells displayed significantly more cells in the sub G1 phase for both concentrations after mTHPC-PDT at any tested time point   Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, and total protein extracts were harvested 24 h after illumination and western blotting performed. Representative blots are shown and % cas 3 relative to pro-cas 3 was evaluated by densitometric analysis. GAPDH was used as a loading control. Data presented as dot plots from at least three independent experiments. Solvent-treated, non-illuminated cells served as the reference group. ( ** p < 0.01; *** p < 0.001; **** p < 0.0001).

mTHPC-PDT Induces G 2 /M Arrest and the Formation of Sub G 1 Populations with Fragmented DNA, Emphasizing the Induction of Apoptosis
To further investigate whether the cell death mechanism is accompanied by growth inhibition or DNA fragmentation, which is another hallmark during apoptosis [30], cell cycle analysis was carried out after staining with PI. The analysis allowed for the assignment of cells in either sub G1 (fragmented DNA, apoptosis), G 0 /G 1 , S, or G 2 /m phase of the cell cycle (see Figure A4 in Appendix A for representative analysis data). The effects on the cell cycle were examined 6, 24, and 48 h after mTHPC-based PDT. Again, no effects on cell cycle distribution were observed after treatment with mTHPC in the absence of light or after illumination of solvent-treated cells for any cell line at any time point compared to solvent-treated cells incubated in the dark ( Figure 9A-E).
For A-427 cells, an increase in sub G 1 cells was detected 24 h after mTHPC-PDT leading to 9.6% (IC 50 ) and 17.8% (IC 90 ) of apoptotic cells, respectively. Similar results were obtained for RT-4 and SISO cells. For the RT-4 cell line, sub G 1 population was significantly increased 6 h (8.9%) and 48 h (13.4%) after PDT with the IC 90 of mTHPC. SISO cells displayed significantly more cells in the sub G 1 phase for both concentrations after mTHPC-PDT at any tested time point ( 6 h after PDT (17.2%) and dropped after 24 h (10.2%) and 48 h (7.6%). The opposite trend was observed for the IC90, where the sub G1 population increased over time (7.3%, 11.9%, and 21.8% after 6, 24, and 48 h) after mTHPC-PDT. Additionally, KYSE-70 cells showed a significant time-dependent increase in the G2/M population after treatment with the IC50 (16.3%, 35.9%, and 46.2% after 6, 24, and 48 h), whereas a decrease in this population with time was observed for the IC90 (18.1%, 17.2%, and 12.3%). For all five cell lines, changes in S phase distribution played only a minor role and changes in G0/G1 population were consequences of increased or decreased sub G1 and G2/M populations.  Table 1 of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, stained with  Table 1 of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, stained with PI and analyzed 6 (left), 24 (middle), or 48 h (right) after photodynamic treatment. Flow cytometric analysis of the single cell population was carried out using the PI channel (λ Ex/Em = 488/655-730 nm). Cells were assigned to either sub G 1 (black, fragmented DNA, apoptotic), G 0 /G 1 (green), S (red), or G 2 /M (blue) phase. Data presented as means ± SD from at least three independent experiments. Solvent-treated, non-illuminated cells served as the reference group. (* p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001).
Additionally, KYSE-70 cells showed a significant time-dependent increase in the G 2 /M population after treatment with the IC 50 (16.3%, 35.9%, and 46.2% after 6, 24, and 48 h), whereas a decrease in this population with time was observed for the IC 90 (18.1%, 17.2%, and 12.3%). For all five cell lines, changes in S phase distribution played only a minor role and changes in G 0 /G 1 population were consequences of increased or decreased sub G 1 and G 2 /M populations.

Autophagic Flux Analysis by LC3-II Levels in the Absence and Presence of Lysosomal Protease Inhibitors Revealed a Simultaneous Occurrence of Autophagy and Apoptosis after mTHPC-PDT
The involvement of autophagy in the fate of mTHPC-treated and illuminated cells was investigated by western blotting of LC3-II 6 and 24 h after PDT ( Figure 10A-E). LC3-II levels in the absence of lysosomal inhibitors only reflect the formation of autophagosomes without any information about the overall autophagic flux. Therefore, lysosomal degradation of LC3-II was inhibited by treatment with pepstatin A and E-64d and LC3-II levels compared to results without lysosomal protease inhibitors [31,32]. Furthermore, the PI3K inhibitor wortmannin was used to block autophagic sequestration and therefore inhibit autophagy ( Figure 11A-E) [31,33]. PI and analyzed 6 (left), 24 (middle), or 48 h (right) after photodynamic treatment. Flow cytometric analysis of the single cell population was carried out using the PI channel (λEx/Em = 488/655-730 nm). Cells were assigned to either sub G1 (black, fragmented DNA, apoptotic), G0/G1 (green), S (red), or G2/M (blue) phase. Data presented as means ± SD from at least three independent experiments. Solvent-treated, non-illuminated cells served as the reference group. ( * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001).

Autophagic Flux Analysis by LC3-II Levels in the Absence and Presence of Lysosomal Protease Inhibitors Revealed a Simultaneous Occurrence of Autophagy and Apoptosis after mTHPC-PDT
The involvement of autophagy in the fate of mTHPC-treated and illuminated cells was investigated by western blotting of LC3-II 6 and 24 h after PDT ( Figure 10A-E). LC3-II levels in the absence of lysosomal inhibitors only reflect the formation of autophagosomes without any information about the overall autophagic flux. Therefore, lysosomal degradation of LC3-II was inhibited by treatment with pepstatin A and E-64d and LC3-II levels compared to results without lysosomal protease inhibitors [31,32]. Furthermore, the PI3K inhibitor wortmannin was used to block autophagic sequestration and therefore inhibit autophagy ( Figure 11A-E) [31,33].
In the absence of lysosomal inhibitors, LC3-II levels were increased 6 h after PDT in BHY (4.6fold, IC50) and RT-4 (11.0-fold, IC90) cells compared to a solvent-treated, non-illuminated control. The LC3-II levels of these reference samples were normalized to 1.0. After 24 h, treatment with the IC50 of mTHPC in combination with light slightly elevated LC3-II levels in BHY (9.3-fold), KYSE-70 (14.0fold), and SISO (12.6-fold) cells, whereas RT-4 cells were less affected (4.5-fold) and A-427 cells remained unaffected. However, treatment with the IC90 led to a significant increase also in RT-4 (7.4fold) and A-427 cells (2.1-fold). This was also true for SISO cells (9.0-fold), indicating an involvement of autophagy in cellular death irrespective of the concentration, at least for this cell line. Interestingly, and unlike the IC50, the higher mTHPC concentration did not have a significant effect on LC3-II levels in BHY and KYSE-70 cells.   Table 1) of mTHPC between 0.02-0.3 µM, illuminated with 1.8 J/cm 2 or left in the dark, and total protein extracts were harvested 6 ( ) or 24 h ( ) after illumination and western blotting performed. Representative blots are shown and the amount of LC3-II relative to a solvent-treated, non-illuminated reference control (LC3-II level for this sample was set to 1.0) was evaluated by densitometric analysis. GAPDH was used for normalization. Data presented as dot plots from at least three independent experiments. (** p < 0.01; *** p < 0.001; **** p < 0.0001).
In the absence of lysosomal inhibitors, LC3-II levels were increased 6 h after PDT in BHY (4.6-fold, IC 50 ) and RT-4 (11.0-fold, IC 90 ) cells compared to a solvent-treated, non-illuminated control. The LC3-II levels of these reference samples were normalized to 1.0. After 24 h, treatment with the IC 50 of mTHPC in combination with light slightly elevated LC3-II levels in BHY (9.3-fold), KYSE-70 (14.0-fold), and SISO (12.6-fold) cells, whereas RT-4 cells were less affected (4.5-fold) and A-427 cells remained unaffected. However, treatment with the IC 90 led to a significant increase also in RT-4 (7.4-fold) and A-427 cells (2.1-fold). This was also true for SISO cells (9.0-fold), indicating an involvement of autophagy in cellular death irrespective of the concentration, at least for this cell line. Interestingly, and unlike the IC 50 , the higher mTHPC concentration did not have a significant effect on LC3-II levels in BHY and KYSE-70 cells.
Autophagic flux was investigated for each cell line only 24 h after PDT with mTHPC concentrations that led to increased levels of LC3-II in the absence of lysosomal protease inhibitors. After pretreatment with pepstatin A and E-64d, higher levels of LC3-II were observed in BHY (12.1-fold, IC 50  total protein extracts were harvested 6 (•) or 24 h (○) after illumination and western blotting performed. Representative blots are shown and the amount of LC3-II relative to a solvent-treated, non-illuminated reference control (LC3-II level for this sample was set to 1.0) was evaluated by densitometric analysis. GAPDH was used for normalization. Data presented as dot plots from at least three independent experiments. ( ** p < 0.01; *** p < 0.001; **** p < 0.0001).
Autophagic flux was investigated for each cell line only 24 h after PDT with mTHPC concentrations that led to increased levels of LC3-II in the absence of lysosomal protease inhibitors. After pretreatment with pepstatin A and E-64d, higher levels of LC3-II were observed in BHY (12.1fold, IC50), KYSE-70 (27.3-fold, IC50), and SISO cells (17.0-fold, IC50 and 22.0-fold, IC90) 24 h after mTHPC-PDT than in the absence of the inhibitors (Figure 11 A,D). However, no change in the levels of LC3-II were detected in A-427 and RT-4 cells after exposure to lysosomal protease inhibitors.
Likewise, blockage of autophagy induction by wortmannin was carried out for each cell line 24 h after PDT at mTHPC concentrations that led to increased levels of LC3-II in the absence of the PI3K inhibitor. Lower LC3-II levels were detected in all cell lines 24 h after mTHPC-based PDT and wortmannin pretreatment ( Figure 11A-E). More precisely, LC3-II formation was still increased in A-427 (1.9-fold, IC90), BHY (3.9-fold, IC50), KYSE-70 (1.3-fold, IC50), RT-4 (4.5-fold, IC90), and SISO cells (6.7-fold, IC50 and 2.0-fold, IC90) in comparison to the solvent-treated, non-illuminated controls, but was substantially lower than for the same treatment without wortmannin. and E-64d for autophagic flux detection or PI3K inhibitor wortmannin (Wort) for blocking of autophagic sequestration, respectively. Experiments were carried out with selected mTHPC concentrations that led to an increase of LC3-II levels ( Figure 10). Cells were treated with equitoxic concentrations (IC50 or IC90 in Table 1) of mTHPC that led to elevated LC3-II levels in the absence of the particular inhibitor. Only selected concentrations were tested for inhibitor pretreatment was carried out with 100 µM PSA and 10 µg/mL E-64d for 4 h or with 2 µM wortmannin for 1 h before PDT and cells were then illuminated with 1.8 J/cm 2 or left in the dark. Total protein extracts were harvested 24 h after illumination and western blotting performed. Representative blots are shown and the amount of LC3-II was evaluated by densitometric analysis. GAPDH was used for normalization. Data presented as dot plots from at least three independent experiments. and E-64d for autophagic flux detection or PI3K inhibitor wortmannin (Wort) for blocking of autophagic sequestration, respectively. Experiments were carried out with selected mTHPC concentrations that led to an increase of LC3-II levels ( Figure 10). Cells were treated with equitoxic concentrations (IC 50 or IC 90 in Table 1) of mTHPC that led to elevated LC3-II levels in the absence of the particular inhibitor. Only selected concentrations were tested for inhibitor pretreatment was carried out with 100 µM PSA and 10 µg/mL E-64d for 4 h or with 2 µM wortmannin for 1 h before PDT and cells were then illuminated with 1.8 J/cm 2 or left in the dark. Total protein extracts were harvested 24 h after illumination and western blotting performed. Representative blots are shown and the amount of LC3-II was evaluated by densitometric analysis. GAPDH was used for normalization. Data presented as dot plots from at least three independent experiments.
Likewise, blockage of autophagy induction by wortmannin was carried out for each cell line 24 h after PDT at mTHPC concentrations that led to increased levels of LC3-II in the absence of the PI3K inhibitor. Lower LC3-II levels were detected in all cell lines 24 h after mTHPC-based PDT and wortmannin pretreatment ( Figure 11A-E). More precisely, LC3-II formation was still increased in A-427 (1.9-fold, IC 90 ), BHY (3.9-fold, IC 50 ), KYSE-70 (1.3-fold, IC 50 ), RT-4 (4.5-fold, IC 90 ), and SISO cells (6.7-fold, IC 50 and 2.0-fold, IC 90 ) in comparison to the solvent-treated, non-illuminated controls, but was substantially lower than for the same treatment without wortmannin.

Discussion
The analysis of cellular viability after PDT with a light dose of 1.8 J/cm 2 by the MTT assay revealed that mTHPC-PDT was an effective photosensitizer against all five cell lines from different tissue origins ( Figure 1). Notably, large differences were observed in susceptibility towards the photosensitizer among the cell lines,  [17]. However, higher IC 50 values of 1.6 µM were determined by Abdulrehman et al. in colon cell lines SW480 and SW620 [35].
The determination of cellular membrane integrity after mTHPC-PDT was assessed by LDH release into the supernatant (Figure 2). In the dark, a dose-dependent increase in cytotoxicity was observed at concentrations starting with 1.0 µM or higher. After illumination, no substantially increased cytotoxicity was detected except for A-427, RT-4, and SISO cells. However, these differences were only seen at medium concentrations, while at higher concentrations LDH release surprisingly decreased to levels comparable to those observed without illumination. These results indicate that necrosis is only induced in some cell lines at rather low concentrations and does not contribute to cellular death after application of high mTHPC concentrations. These findings were confirmed by a subsequent MTT cell viability assay that has been carried out with the cells that remained in the 96-well plates after removal of the supernatant used for the LDH release assay (see Figure A5 in Appendix A). The MTT data revealed that cellular viability dropped to <5% in the absence of light at mTHPC concentrations of 5 µM (24 h) and 10 µM (6 h; <15% for BHY and <30% for RT-4 cells). More interestingly, after illumination with either 3.5 or 1.8 J/cm 2 , cellular viability was decreased to <5% at mTHPC concentrations of 0.1-0.3 µM (1.0 µM for KYSE-70), which implies that cellular membrane integrity was still intact, although cellular viability was nearly completely lost under these conditions. The presence of an intact cellular membrane accompanied by a nearly total loss of cellular viability clearly indicates that little or no necrosis occurred during cellular death. In the case of necrosis, a total loss of cellular viability would have led to high levels of LDH released into the supernatant due to a porous membrane. Treatment with Triton X-100 led to minimal residual cellular viability in the MTT assay (see Table A1 in Appendix A).
Altogether, these results indicate that the induction of necrosis plays only a minor role after mTHPC-PDT in the tested cell lines. to the values observed after incubation in the dark. These results were again consistent with those we found with all five tested cell lines. Additionally, Löw and colleagues used the WST-1 assay and detected a total loss of cellular viability at 7.5 and 1.5 µM after illumination with 5.0 J/cm 2 followed by an incubation period of 4 and 24 h, respectively. These results are again consistent with our results obtained with the MTT assay, where a total loss of cellular viability was observed 6 or 24 h after illumination with 3.5 J/cm 2 for mTHPC concentrations of 0.1-1.0 µM.
ROS-induced toxicity is one characteristic outcome of PDT [37]. Furthermore, membrane lipids are known targets for oxidation during oxidative stress [38]. In our study, ROS could be detected in KYSE-70 and SISO cells after illumination with 1.8 J/cm 2 , but higher light doses were required for A-427 and BHY cells (Figure 3). However, substantial loss of cellular viability was observed with doses of ≤1.8 J/cm 2 in all five cell lines. At time points 24 and 48 h after mTHPC-PDT, lipid peroxidation (LPO) was additionally detected in KYSE-70, RT-4 and SISO cells (Figure 4). Surprisingly, enhancement of LPO occurred in RT-4 cells, although no increased ROS levels were observed after illumination with 1.8 J/cm 2 . One explanation could be the limitation of H 2 DCF-DA to detect singlet oxygen ( 1 O 2 ) [19]. Indeed, 1 O 2 could even act as a quencher of DCF fluorescence [39]. The singlet oxygen quantum yield (Φ ∆ ) of mTHPC ranges from 0.3-0.4 in different solvents [40,41], meaning that 30-40% of the energy of each photon absorbed by mTHPC can be used for the generation of 1 O 2 from triplet oxygen ( 3 O 2 ). The remaining energy can be transferred for the generation of other ROS, e.g., superoxide anion (O 2 − ) or hydroxyl radicals (OH ), or will instead be lost via fluorescence emission or thermal radiation [1]. If the proportion of 1 O 2 on total ROS was in the same range of 30-40% in our ROS measurements, it may led to increases of ROS that were not detectable by the DCF assay, while detectable species like O 2 − or OH did not increase substantially. High but undetectable levels of 1 O 2 could led to the increase of LPO in KYSE-70, RT-4 and SISO cells after illumination with 1.8 J/cm 2 , although no increase in ROS levels were observed under these conditions by H 2 DCF-DA staining. A further explanation for not observing an increase in ROS levels could be a high efficacy of anti-oxidative pathways, e.g., due to high levels of catalase, glutathione peroxidase (GP X ), glutathione (GSH), or GSH-recovering enzymes like glutathione-disulfide reductase (GSR). For example, relatively high levels of GSH have been found for RT-4 cells by our group, while the other cell lines displayed lower GSH concentrations [42]. However, the generated ROS would still have to exist long enough to cause oxidative stress and induce a loss of cellular viability. No enhanced LPO as well as no increased ROS levels were observed in A-427 and BHY cells. Furthermore, no substantial loss of membrane integrity has been detected in the LDH release assay in any cell line after mTHPC-mediated PDT (Figure 2), suggesting that the cellular membrane is not the primary target of mTHPC-PDT. These results indicate that LPO plays only a minor role in photodamage after mTHPC-PDT (at least in A-427 and BHY cells) and is not correlated with ROS levels and cellular viability. The role of LPO in cellular death after PDT has been subject of discussion in the past. Ehrenberg and colleagues concluded that the toxicity after PDT with hematoporphyrin (HP) resulted from damage of proteins rather than from LPO-induced membrane damage [43]. Gaullier and co-workers also detected no correlation between LPO and cellular death after PDT with protoporphyrin IX (PpIX) [44]. With mTHPC as the applied PS, Klein et al. detected enhanced LPO and oxidation of proteins after treatment with mTHPC (5-50 µg/mg mitochondrial protein) and illumination with a high light dose of 5.3 J/cm 2 . However, experiments were carried out with isolated rat liver mitochondria and therefore do not reflect the situation in whole cells properly [45]. Melnikova and co-workers treated HT29 colon adenocarcinoma cells with 1.5 µM mTHPC and light doses of 2.3-6.8 J/cm 2 and concluded that LPO only plays a minor role in cell inactivation by mTHPC-PDT [46]. In a study of Kirveliene and colleagues with two rodent cell lines, cells were treated with 0.75 µM mTHPC and illuminated with a light dose of 1.8 J/cm 2 . Enhanced LPO was found as an early response to mTHPC-PDT, but the cells were able to restore LPO to initial levels already 2 h after light exposure. Additionally, the group found neither LDH nor ATP release after mTHPC-PDT.
These findings were consistent with our results from the LDH release assay ( Figure A5 in Appendix A), suggesting that a loss of membrane integrity by LPO and induction of necrosis do not occur [47].
The induction of oxidative stress triggers apoptosis accompanied or preceded by a loss of mitochondrial membrane potential (∆ψ m ) [24,25]. In our study, a collapse of ∆ψ m was observed after mTHPC-mediated PDT in all of the five tested cell lines ( Figure 5). These results indicate a direct targeting of the mitochondria by mTHPC-PDT and prove the induction of apoptosis at an early stage. These findings are consistent with results from Marchal and colleagues with the JC-1 dye in flow cytometric approaches. Depolarization was observed 4 and 24 h after PDT in HT29 human adenocarcinoma cells after treatment with 1.5-4.5 µM mTHPC and light doses of 0.06-1.9 J/cm 2 [48,49]. Treatment of human breast adenocarcinoma cell line MCF-7 with 1.5 µM mTHPC also led to a loss of ∆ψ m immediately and 24 h after illumination. In this study, the group also proved that mTHPC induced direct mitochondrial photodamage rather than an indirect damage via the translocation of the pro-apoptotic Bax protein [16].
The induction of apoptosis was investigated 6, 24 and 48 h after illumination with a light dose of 1.8 J/cm 2 by staining with Annexin V-FITC and PI ( Figure 6). It has to be mentioned that Annexin V-FITC/PI double-stained cells need to be considered as late-apoptotic and not necrotic one as the LDH release assay revealed no signs of necrosis involvement in cellular death. Misinterpretation of late-apoptotic as necrotic cells has been an issue in the past after mTHPC-PDT, where further confirmation of necrosis has not been carried out [48,49]. In all tested cell lines, a time-dependent increase in apoptotic (A-427, BHY) or late-apoptotic (A-427, KYSE-70, RT-4, SISO) populations can be observed for the high mTHPC concentration, indicating that even high-dose PDT leads to apoptosis rather than necrosis. Furthermore, a switch from apoptosis to late apoptosis was observed over time. Apoptosis induction after mTHPC-PDT was also observed via Annexin V-FITC/PI staining by Marchal and colleagues in HT29 human adenocarcinoma cells (1.5 µM, 1.92 J/cm 2 ) [49] and Abdulreham et al., who found apoptotic populations in colon carcinoma cell lines SW480 and to a lesser extent in SW620 with no evidence of late-apoptotic or necrotic cells (0.2-11.8 µM, 6.0 J/cm 2 ) [35]. Furthermore, Yow and colleagues visualized phosphatidylserine externalization in the human nasopharyngeal carcinoma cell line NPC/HK1 by confocal laser scanning microscopy (1.2 µM, 2.0 J/cm 2 ) [50].
Apoptosis induction was further studied and confirmed by the cleavage of PARP and the activation of caspase 3 (Figures 7 and 8). Our findings support the results observed by the Annexin V-FITC/PI method. However, the IC 50 led to PARP cleavage to a lesser extent and it should be emphasized that no or only little signs of apoptosis could be detected for this concentration, indicating that PARP might be additionally cleaved via a different pathway. Caspase 3-activation and PARP cleavage were observed after mTHPC-PDT under comparable conditions by other groups, e.g., in human breast carcinoma cell line MCF-7 (1.5 µM, 0.01-0.06 J/cm 2 ) [16] and HT29 adenocarcinoma cells (1.5-4.5 µM, 0.3-1.9 J/cm 2 ) [48,49]. The detection of a 89 kDa proteolytic fragment of cleaved PARP is characteristic after activation of the effector caspase 3, since PARP is a substrate of active caspase 3 [29]. In our study, caspase 3-activation was consistent with the results that were obtained by Annexin V-FITC/PI staining. Caspase 3 is directly connected to phosphatidylserine externalization via the inhibition of flippase and the activation of scramblase during apoptosis, two enzymes involved in the maintenance of membrane asymmetry in vital cells [51][52][53]. While caspase 3-activation was followed by PARP cleavage for the IC 90 concentration, it was not the case for the IC 50 concentration. These findings further support the hypothesis that PARP cleavage was also initiated by an additional route, e.g., by active caspase 7, which also produces a 89 kDa proteolytic PARP fragment [54][55][56]. Furthermore, the involvement of PARP as a suppressor in autophagy may also lead to an additional cleavage after oxidative stress [57,58]. Caspase-dependent PARP cleavage also circumvents necrotic cell death by preventing the depletion of NAD + and ATP by overactivated PARP [55,59,60], further indicating that necrosis plays only a minor role in mTHPC-PDT. It is a generally accepted statement that PDT leads to apoptosis with low and medium PS concentrations and rather necrosis with high PS concentrations in several cell types [61,62], which has also been shown with mTHPC as the selected PS [13,48,63]. However, our data do not point at necrosis as a cause of cell death for mTHPC-PDT.
Effects on the cell cycle were observed after mTHPC-PDT (Figure 9). Fragmented DNA can be detected in cells of the sub G 1 fraction and is characteristic for the final steps of apoptosis downstream of caspase 3/7-activation and PARP cleavage. Namely, caspase-activated DNase (CAD) is responsible for the degradation of DNA in the nucleus, but also caspase-independent pathways can be involved, e.g., after release of apoptosis-inducing factor (AIF) or the DNase EndoG from mitochondria [30,64,65]. In our studies, DNA fragmentation was observed and was in some cases accompanied by a G 2 /M arrest. However, BHY cells displayed a G 2 /M-arrested population, but no DNA fragmentation. The observed results for DNA fragmentation were consistent with the directly related caspase 3-activation and PARP cleavage for A-427, KYSE-70 and SISO cells, but divergent results were obtained with the BHY and RT-4 cell lines. In BHY, caspase 3-activation and PARP cleavage were clearly observed for mTHPC-PDT. However, neither the caspase 3-related activation of CAD nor the suppressed PARP-related DNA excision repair and repair of single-and double-strand breaks led to increased DNA fragmentation. Interestingly, significant G 2 /M arrest and even increased DNA replication were observed, indicating that BHY cells may try to compensate DNA-related damages, e.g., via an increased frequency of alternative repair mechanisms like hyper-homologous recombination repair [66,67]. Significant PARP cleavage was also detected in RT-4 cells, which was not preceded by substantial caspase 3-activation. Furthermore, no DNA fragmentation was observed, which might be explained by a missing caspase 3-related CAD activation. Therefore, it seems likely that PARP cleavage was initiated via an alternative route, but-like in BHY cells-the absent DNA repair activity did not lead to an accumulation of fragmented DNA in RT-4 cells. Nevertheless, a prominent sub G 1 population also indicated DNA fragmentation 48 h after PDT in RT-4 cells.
Autophagy was investigated via western blotting by detecting LC3-II, which can be correlated with autophagosome numbers ( Figure 10) and was confirmed by additional incubation with the PI3K inhibitor wortmannin, which blocks autophagic sequestration ( Figure 11). Elevated LC3-II levels have been detected in all cell lines following mTHPC-PDT. Incubation with wortmannin led to decreases in LC3-II levels, but not that clearly in A-427 cells. These results indicate that autophagy was induced after mTHPC in all tested cell lines except A-427 cells.
In the presence of PSA and E-64d, LC3-II levels were increased, and therefore autophagic flux detection confirmed the induction of autophagy in these cell lines ( Figure 11). However, autophagic flux was not enhanced in in A-427 and RT-4 cells. One explanation could be a late stage suppression of autophagy, e.g., by blocking autophagosome maturation or autophagosome-lysosome fusion. Similar results were obtained also by Kukcinaviciute et al. 24 h after mTHPC-PDT with HCT116 colorectal carcinoma cells after treatment with 0.15 µM mTHPC and light doses of 0.9-2.7 J/cm 2 . The amount of autophagosomes was raised, but no autophagic flux was observed [68], which are consistent with the results of François and colleagues [15]. The increase in LC3-II levels coincides with phosphatidylserine externalization (Figure 6), caspase 3-activation (Figure 8), and PARP cleavage (Figure 7) in A-427 and RT-4 cells, indicating that autophagy contributes to cellular death. However, no autophagic flux could be detected in these cell lines, leading to the conclusion that autophagy was inhibited at a late stage. In this case, no autophagy-associated cell death would occur, but the loss of pro-survival effects of autophagy may contribute to a further progression of apoptosis. For the lower concentration, neither apoptosis nor autophagy were detected, but PARP cleavage was substantial in RT-4 cells. In BHY and KYSE-70 cells, LC3-II levels were mainly increased at the lower mTHPC concentration, where no phosphatidylserine externalization and caspase 3-activation were observed. However, PARP cleavage occurred at this concentration, which could be due to a cleavage by active caspase 7. These findings indicate, that autophagy was rather a survival strategy, which counteracts apoptosis. This survival strategy ultimately fails for the lower concentration, probably leading to autophagy-associated cell death. At the high mTHPC concentration, autophagy was not involved in cellular death of BHY and KYSE-70 cells. An ambiguous picture has been produced by SISO cells. While phosphatidylserine externalization and caspase 3-activation were observed only in small amounts for the IC 50 , both were prominent after PDT with the IC 90 . PARP cleavage and increased LC3-II levels were, however, found with both mTHPC concentrations and autophagy induction was confirmed by autophagic flux analysis. These findings indicate that autophagy was cytoprotective after PDT with a low mTHPC concentration, which was confirmed by a missing increase in apoptotic cells 48 h after PDT. But at the high mTHPC concentration, autophagy tends to accompany cellular death and failed to prevent apoptotic cell death. Especially after PDT with PS that target the ER and the mitochondria, like mTHPC does [69,70], so called autophagy-associated cell death was observed after PDT. In contrast, also pro-survival effects of autophagy after PDT can occur, especially at low-level PDT [15,19,61,62,[71][72][73][74]. However, our results suggest that the role of autophagy even varies with the same PS in different cell lines.

Cell Culture
The five different human cancer cell lines A-427 (lung carcinoma; ACC 234), BHY (oral squamous cell carcinoma; ACC 404), KYSE-70 (esophageal squamous cell carcinoma; ACC 363), RT-4 (urinary bladder transitional cell carcinoma; ACC 412), and SISO (cervix adeno carcinoma; ACC 327) were obtained from Leibniz Institute DSMZ (Braunschweig, Germany) and routinely checked for mycoplasma. All lines were cultured in phenol red containing RPMI 1640 medium (PAN Biotech, Aidenbach, Germany) supplemented with 10% (v/v) fetal bovine serum (FBS; Sigma-Aldrich, Munich, Germany), 100 µg/mL streptomycin and 100 U/mL penicillin G (PAN Biotech, Aidenbach, Germany) at 37 • C and 5% CO 2 in a humidified atmosphere. During and post illumination the cells were cultured in the same medium, but without phenol red. Cells were subcultured by detachment with 0.5 g trypsin/0.2 g EDTA (Sigma-Aldrich, Munich, Germany) once a week. Individual experiments were done by using either transparent, flat-bottom 96-well plates with cells seeded out at a density of 2.0-5.0 × 10 3 cells per well in 100 µL medium for measurement of cellular viability and lactate dehydrogenase (LDH) release, T25 flasks with cells seeded out at a density of 5.0 × 10 5 cells in 5 mL medium for flow cytometric analyses (except ROS detection), 6-well plates with cells seeded out at a density of 2.5 × 10 5 cells per well in 2 mL medium for western blot analyses and detection of ROS generation, or 4-well cell culture chamber slides with cells seeded out at a density of 7.5 × 10 4 cells per chamber in 1 mL medium for fluorescence microscopy (all culture dishes from Sarstedt, Nümbrecht, Germany). After seeding, cells were allowed to attach and grow for 24 h before treatment.

Photosensitizer Treatment
mTHPC was kindly supplied by Biolitec AG (Jena, Germany). A 20 mM stock solution was prepared in propylene glycol/ethanol (60:40) and stored at 4 • C in the dark. The PS was added to the cells in RPMI 1640 medium containing 10% (v/v) FBS and left for 24 h. For detection of cellular viability and LDH release, concentrations between 0.001-5.0 µM were used. However, for the determination of phosphatidylserine externalization, cell cycle distribution, mitochondrial membrane potential and western blot analyses of caspase 3 (cas 3), poly(ADP-ribose) polymerase (PARP), and microtubule-associated protein light chain 3 (LC3-II), equitoxic concentrations corresponding to the IC 50 and IC 90 values (concentrations, where 50% and 90% of the measured effect, i.e., loss of cellular viability, was observed) established from the MTT cell viability assay data 24 h after illumination with 1.8 J/cm 2 were used for the treatment (see Section 2.1). Concentrations ranged between 0.02-0.3 µM depending on the cell line.

Photodynamic Treatment
The mTHPC-treated cells were washed with PBS and fresh phenol red free RPMI 1640 medium containing 10% (v/v) FBS was added before illumination. Cells were illuminated with an LED array based illumination device with 432 LEDs (light-emitting diodes) following the example of Pieslinger et al. [75]. The LEDs (Kingbright, Issum, Germany) produced a wavelength spectrum of λ = 640-660 nm and a light dose of 1.8 J/cm 2 , 3.5 J/cm 2 , and 7.0 J/cm 2 was applied at fluence rates of 3.0 mW/cm 2 and 5.8 mW/cm 2 , respectively. Cells were harvested by trypsinization 6, 24, or 48 h post illumination for analyses. For detection of ROS, cells were harvested immediately after illumination. For the determination of dark toxicity, cells were treated with mTHPC, but not illuminated. A sample treated with solvent in medium (equal to the highest solvent concentration used for dilution of mTHPC in the respective assay) and kept in the dark served as a reference control in the assays (solvent-treated control, SC).

MTT Cell Viability Assay
Cellular viability was measured 24 h after illumination of the cells by using the MTT (3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide; Alfa Aesar, Karlsruhe, Germany) assay. After photodynamic treatment, 20 µL of a 2.5 mg/mL MTT solution were added to each 100 µL medium per well and incubated at 37 • C and 5% CO 2 in a humidified atmosphere for 4 h. Supernatant was replaced afterwards with 50 µL DMSO per well and the absorbance of the reduced formazan was measured at λ = 570 nm with a microplate reader (SpectraMax Plus 384; Molecular Devices, Biberach, Germany). The percentage of cell viability was calculated by dividing the absorbance in the treated group by the absorbance in the solvent control. Calculation of the IC 50 values was done with the help of Prism 6 (GraphPad Software, La Jolla, CA, USA).

LDH Release Assay
Cells were illuminated with 1.8 J/cm 2 and 3.5 J/cm 2 in phenol red free medium containing 2.5% (v/v) FBS and incubated for 6 and 24 h post illumination, respectively. The loss of membrane integrity as a sign of necrosis was determined by measurement of released LDH into the medium supernatant as described elsewhere [18]. Briefly, after centrifugation of the 96-well plate at 2.000 rpm for 5 min, 50 µL/well of the supernatant were transferred to a new plate and 50 µL of LDH buffer was added (remaining cells were used for a subsequent MTT cell viability assay). LDH buffer consisted of 0.

Analysis of ROS Generation
The generation of ROS was analyzed after treatment with 2 ,7 -dichlorodihydrofluorescein diacetate (H 2 DCF-DA; Sigma-Aldrich, Munich, Germany) as described elsewhere [76]. Flow cytometric analysis of fluorescent DCF was carried out directly after illumination with 1.8 J/cm 2 , 3.5 J/cm 2 or 7.0 J/cm 2 with a MACS Quant flow cytometer (Miltenyi Biotech, Bergisch Gladbach, Germany). For each sample, 10,000 events were counted and gated for the single cell population. Fluorescent 2 ,7 -dichlorofluorescein (DCF) is formed within the cells after contact with ROS and the detected fluorescence intensity is increased with higher amounts of ROS [20,21]. The FITC channel (λ Ex/Em = 488 nm/525-550 nm) was used for the detection of DCF and data were analyzed with the MACS Quantify Software (Miltenyi Biotech, Bergisch Gladbach, Germany). As a positive control, cells were treated with 1.0-2.0 mM H 2 O 2 (Sigma-Aldrich, Munich, Germany) for 10 min.

Detection of Lipid Peroxidation (LPO)
The detection of lipid peroxidation (LPO) was carried out after staining with BODIPY 665/676 (Thermo Fisher Scientific, Waltham, MA, USA). Briefly, after treatment with equitoxic concentrations corresponding to the IC 50

Cell Cycle Analysis
Cell cycle was analyzed after staining with PI (AppliChem, Darmstadt, Germany). Cells were treated and 5.0 × 10 5 harvested 6, 24, and 48 h after photodynamic treatment as described for the detection of lipid peroxidation. Cells were washed twice with PBS and fixed by ice-cold 70% (v/v) ethanol at 4 • C for 30 min. Fixed cells were centrifuged at 4.000 rpm at 4 • C for 10 min and resuspended in PBS, containing 25 µg/mL PI and 100 µg/mL RNase A (Carl Roth, Karlsruhe, Germany). After staining at room temperature for 30 min in the dark, cells were analyzed with a MACS Quant flow cytometer. For each sample, 10,000 events were counted and gated for the single cell population. The PI channel (λ Ex/Em = 488/655-730 nm) was used for the detection of PI-positive cells. Data were analyzed with the MACS Quantify Software. After analysis, cells were assigned to either sub G 1 (fragmented DNA, apoptotic), G 0 /G 1 , S, or G 2 /M phase.

Fluorescence Microscopy
Evaluation of Mitochondrial Membrane Potential (∆ψ M ) Mitochondrial membrane potential (∆ψ m ) was detected with the BD MitoScreen JC-1 kit (BD Biosciences, San Diego, CA, USA) according to the kit instructions (modified for fluorescence microscopy). Briefly, living cells were washed with Assay Buffer within the wells of chamber slides 6 h after photodynamic treatment with the IC 90 of mTHPC and stained with 2 µM JC-1 solution at 37 • C for 20 min. Afterwards, cells were washed twice with Assay Buffer and JC-1 monomers and aggregates were visualized using a Leica DMi8 fluorescence microscope (Leica, Munich, Germany) equipped with a 63× oil/1.4 NA objective. The disruption of active mitochondria is an early sign of apoptosis after oxidative stress induction [24,25]. The cationic JC-1 (5,5 ,6,6 -tetrachloro-1,1 ,3,3 -tetraethylbenzimidazolylcarbocyanine iodide) dye accumulates in mitochondria with normal or hyperpolarized ∆ψ m , leading to the formation of aggregates with a red fluorescence at 590 nm, whereas cytosolic JC-1 monomers display a green fluorescence at 527 nm. A decline in ∆ψ m is thus indicated by a decrease in JC-1 aggregates, visualized by a decrease in red fluorescence [80]. Fluorescence images (1.392 × 1.040 px) of JC-1 monomers and aggregates were captured with the FITC filter cube (λ Ex/Em = 460-500/512-542 nm) and the RHOD filter cube (λ Ex/Em = 541-551/565-605 nm), respectively, and processed with the LAS X software (Leica, Munich, Germany). Solvent-treated and non-illuminated cells served as the negative control with active mitochondria. As a positive control, cells were treated with 50 µM carbonyl cyanide m-chloro phenyl hydrazone (CCCP; Alfa Aesar, Karlsruhe, Germany) at 37 • C for 5 min. CCCP is a well-known mitochondrial oxidative phosphorylation uncoupler [81].

Statistical Analysis
Data were presented as means ± standard deviation (SD) of at least three independent experiments. Significant differences were detected by one-way or two-way ANOVA followed by Dunnett's multiple comparisons test implemented by Prism 6 (GraphPad Software, La Jolla, CA, USA). A p-value < 0.05 was considered statistically significant.

Conclusions
In general, the overall phototoxic effects of mTHPC-PDT vary in dependency of concentration and time from cell line to cell line, suggesting that the cancer cells are not all dying by one defined mechanism, but rather succumb to an individual interplay of different cell death mechanisms. Triggering of oxidative stress by mTHPC-PDT was proven by a loss of mitochondrial membrane potential (∆ψ m ) and increased generation of ROS in all cell lines, although high light doses were required for enhanced ROS formation in some cases. However, lipid peroxidation (LPO) appeared to play only a minor role in most cell lines. Cellular death after high-dose mTHPC-PDT was mainly characterized by the induction of caspase-dependent apoptosis, which led to cleavage of PARP. The loss of ∆ψ m confirmed the induction of apoptosis. Cell cycle analysis revealed that sooner or later DNA fragmentation occurred (increased sub G 1 fraction), which was in some cell lines accompanied or preceded by an arrest in the G 2 /M cell cycle phase. At low-dose PDT, autophagy tends to have a pro-survival role, which counteracts apoptosis, while after high-dose PDT autophagy-associated apoptosis occurred. However, autophagy was just induced after either low-or high-dose PDT in most cell lines. Unregulated necrosis appeared to play only a minor role both after low-and high-dose PDT. Ideal dosage and incubation times in mTHPC-mediated PDT are rather individual for a specific tumor type, since the phototoxic effects vary widely in cells from different tissues. This study shows that general conclusions after PDT in vitro require testing on multiple cell lines to be reliable and that instead of applying one specific protocol for all cancer types, it would be advisable to individualize PS and light doses for the most effective outcome of a PDT treatment.   Figure A1. Representative flow cytometric analysis of ROS formation in SISO cells. Cells were treated with solvent (red and yellow), H2O2 (black), and the IC50 (green) or IC90 (blue) of mTHPC. Afterwards, cells were kept in the dark (left) or were illuminated with 1.8 J/cm 2 (right) and stained with H2DCF-DA directly after illumination. For each sample, 10,000 events were counted and gated for the single   solvent (red and yellow), t-BHP (black), and the IC50 (green) or IC90 (blue) of mTHPC. Afterwards, cells were kept in the dark (left) or illuminated with 1.8 J/cm 2 (right) and stained with BODIPY 665/676 24 h after illumination. For each sample, 10,000 events were counted and gated for the single cell population (not shown). BODIPY 665/676 fluorescence was detected with the APC channel (λEx/Em = 635/655-730 nm) of a MACS Quant flow cytometer. Data were analyzed with the MACS Quantify Software.     , and SISO (E) were treated with mTHPC for 24 h in concentrations ranging from 0.01-10.0 µM and kept in the dark (dark toxicity) or illuminated with a light dose of 1.8 J/cm 2 or 3.5 J/cm 2 (light-induced toxicity). MTT assay was carried out 6 h or 24 h post illumination and the absorbance of the reduced formazan was measured at λ = 570 nm. The percentage of cell viability was calculated by dividing the absorbance for the treated group by the absorbance in the solvent dark control. Results presented as means ± SD from at least three independent experiments. Table A1. Established % cellular viability as assessed by the MTT cell viability assay (with remaining cells from LDH release assay) after treatment of cells with 0.1% (v/v, 6 h) and 0.01% (v/v, 24 h) Triton X-100 positive control after incubation in the dark and post illumination with 1.8 J/cm 2 (24 h) or 3.5 J/cm 2 (6 h), respectively. Results as means ± SD from at least three independent experiments.