Interplay between Cellular Uptake, Intracellular Localization and the Cell Death Mechanism in Triphenylamine-Mediated Photoinduced Cell Death

Triphenylamines (TPAs) were previously shown to trigger cell death under prolonged one- or two-photon illumination. Their initial subcellular localization, before prolonged illumination, is exclusively cytoplasmic and they translocate to the nucleus upon photoactivation. However, depending on their structure, they display significant differences in terms of precise initial localization and subsequent photoinduced cell death mechanism. Here, we investigated the structural features of TPAs that influence cell death by studying a series of molecules differing by the number and chemical nature of vinyl branches. All compounds triggered cell death upon one-photon excitation, however to different extents, the nature of the electron acceptor group being determinant for the overall cell death efficiency. Photobleaching susceptibility was also an important parameter for discriminating efficient/inefficient compounds in two-photon experiments. Furthermore, the number of branches, but not their chemical nature, was crucial for determining the cellular uptake mechanism of TPAs and their intracellular fate. The uptake of all TPAs is an active endocytic process but two- and three-branch compounds are taken up via distinct endocytosis pathways, clathrin-dependent or -independent (predominantly caveolae-dependent), respectively. Two-branch TPAs preferentially target mitochondria and photoinduce both apoptosis and a proper necrotic process, whereas three-branch TPAs preferentially target late endosomes and photoinduce apoptosis only.

Over the past decade, there has been a growing interest in developing and studying two-photon absorption probes for cell imaging and biomedical applications. Indeed, near infrared (NIR) and infrared (IR) wavelengths used for two-photon excitation minimize autofluorescence and maximize tissue penetration due to smaller light scattering/absorption by the tissue and endogenous biomolecules 1 . Such probes were used for several types of applications including fluorescent markers such as nucleus and organelle markers or chemical probes [2][3][4][5] . Alternatively, two-photon excitation may allow to trigger photochemical/photobiological processes, for instance in the case of (i) optogenetics for the control of protein activity 6,7 , or (ii) photodynamic therapy (PDT) for light-dependence production of reactive oxygen species (ROS) [8][9][10][11][12] .
Regarding the two-photon absorption/fluorescence properties, all TPAs display high σ 2 values comprised between 325 and 764 GM and maximum NIR excitation wavelengths (λ exc,2-hν ) in the 750-880 nm range, depending on the compound. By contrast to what observed for the Φ F value, the σ 2 value in each pair is higher for the 3-branch (octupolar arrangement) than for the 2-branch (quadrupolar arrangement) compound: 764 and 495 GM for TP3Bzim and TP2Bzim, respectively; 388 and 365 GM for TP3Pyo and TP2Pyo, respectively; 500 and 325 GM for TP3Py and TP2Py, respectively. However, taking into account that the σ 2 values are less influenced by the nature of electron acceptor group, the hierarchy of in vitro two-photon brightness (σ 2 xΦ F ) principally follows the hierarchy based on the Φ F value: 267 and 258 GM for TP2Bzim and TP3Bzim, respectively; 62 and 31 GM for TP2Pyo and TP3Pyo, respectively; 26 and 5 GM for TP2Py and TP3Py, respectively.
Photoinduced effects of TPAs in living cells. One-photon photoactivation. Living HeLa cells, treated with TP3Py, TP2Bzim, TP2Pyo or TP3Pyo and irradiated by visible light (458 nm), displayed similar features than those previously reported for TP3Bzim and TP2Py 16,17 . All cells displayed an initial localization of the TPA fluorescence signal which is cytoplasmic (Fig. 2). Under prolonged illumination, we consistently observed a nuclear translocation of the fluorescence signal with the beginning of this translocation occurring on a time scale of 20-230 min (after the start of illumination), depending on the compound. These values for TP2Bzim (20 min) and TP3Py (50 min) remain comparable with those characterizing TP3Bzim and TP2Py (20 and 40 min, respectively) 17 , although TP2(3)Pyo led to slower kinetics (>90 min). The stronger fluorescence emission in the nucleus compared to the initial cytoplasmic emission is in agreement with in vitro studies showing the large fluorescence enhancement of TPAs upon DNA-binding (Supplementary Table S1). As previously found with TP3Bzim/TP2Py, the nuclear translocation of the four new analogues coincides with both cell shrinkage and the appearance of plasma membrane blebs (see images in the transmission mode on Fig. 2), two morphological hallmarks of cell apoptosis 18 . Of note, both phenomena, i.e. the TPA relocalization to the nucleus and membrane blebbing, were strictly dependent on the photoactivation process since only cells in the illuminated area displayed (i) a strong nuclear fluorescence signal and (ii) a dramatic morphological change (Fig. 2, panels B, D, F, H). The cells outside this area displayed a constant cytoplasmic location of fluorescence and a normal morphology. These results obtained with the new analogues indicate that all TPAs share -at least qualitatively -similar pro-apoptotic properties upon prolonged visible light irradiation.
Two-photon photoactivation. Two-photon experiments in living HeLa were performed in a similar manner than one-photon experiments except that each TPA was excited by its optimal two-photon excitation wavelength ( Fig. 3A-D). Conversely to one-photon experiments, the new analogues clearly display differential responses regarding their ability to induce cell death upon two-photon excitation. Only cells treated with TP2Bzim underwent significant apoptosis with a concomitant nuclear translocation of fluorescence (Fig. 3A), in a similar manner to that previously observed with TP3Bzim/TP2Py 16 . By contrast, no nuclear translocation or morphological change of treated cells were observed with TP3Py, TP3Pyo or TP2Pyo ( Fig. 3B-D). Their fluorescence signals were low or continuously decreased without nuclear relocalization, suggesting that photobleaching preferentially occurs for these compounds under continuous two-photon illumination. The susceptibility of TPAs to photobleaching under two-photon illumination was then measured in vitro using a FCS (fluorescence correlation spectroscopy) set-up as described in Methods. In vitro results show a continuous decrease of fluorescence intensities for TP3Py, TP2Pyo and TP3Pyo, while fluorescence intensities remained stable for TP3Bzim, TP2Bzim and TP2Py (Fig. 3E), in agreement with observations in the cell context. Altogether, these results indicate that, in contrast to TP3Bzim, TP2Bzim and TP2Py, the three other compounds are unable to induce a two-photon-dependent cell death process, most likely for reasons related to faster photodamage.
Structural determinant of the initial subcellular tpA localization. All cationic TPAs presented in Fig. 1 show an exclusive cytoplasmic localization before light irradiation and subsequent nuclear translocation. The global positive charge (2 + to 3 + for 2-and 3-branch compounds, respectively) is important for the cytoplasmic retention. Accordingly, we have previously shown that living cells treated with a neutral TPA displayed a non-specific and homogeneous staining pattern, with both cytoplasmic and nuclear staining, even before prolonged illumination 17 . Nevertheless, the fluorescence signal pattern of cationic TPAs in the cytoplasm differs between compounds. Indeed, we have previously shown that TP2Py mainly colocalized with mitochondria (together with a slight localization in late endosomes that was consistently observed) while TP3Bzim only partially colocalized with mitochondria 16 . For TP3Bzim, most of the colocalization signal originated from late endosomes. Thus, we extended here the colocalization study to the other compounds to identify the structural features of TPAs that govern their intracellular fate. The colocalization of TP2Bzim, TP3Py, TP2Pyo and TP3Pyo with either mitochondria or late endosomes was then investigated in HeLa cells. Although all TPAs (except TP3Py) displayed a dual location, the strongest colocalization signals were obtained with mitochondria for 2-branch derivatives (TP2Bzim and TP2Pyo) and with late endosomes for 3-branch derivatives (TP3Py and TP3Pyo)  Supplementary Fig. S1): the initial observation (time 0), the beginning of the nuclear translocation of TPA fluorescence, the appearance of plasma membrane blebs (indicated by blue arrows in DIC -differential interference contrast -transmission images) and the time of maximum fluorescence intensity in nuclei (t max ). Enlarged field of observation is shown at t max for each TPA (panels B, D, F, H), showing that only cells illuminated at t = 0 are characterized by both TPA nuclear translocation and membrane blebbing. Cells outside the irradiated area (delineated by dashed lines) display normal morphologies and cytoplasm-located fluorescence. Cells indicated by a red arrow serve as a landmark. (2020) 10:6881 | https://doi.org/10.1038/s41598-020-63991-9 www.nature.com/scientificreports www.nature.com/scientificreports/ ( Fig. 4). Altogether, our results show that the number of branches, rather than the chemical nature of the electron acceptor group, is critical to determine the subcellular TPA localization with 2-and 3-branch derivatives preferentially targeting mitochondria and late endosomes, respectively.
Relationship between the subcellular tpA localization and the subsequent photoinduced cell death mechanism. Phosphatidylserine (PS) recognition by Annexin V is commonly used to probe dead cells: (i) apoptotic cells which are characterized by PS externalization occurring at the plasma membrane but also (ii) necrotic cells with damaged plasma membranes allowing Annexin V staining even if PS remains on the inner face of the membrane 16,19 . The use of 4' ,6'-diamino-2-phenylindole (DAPI) in combination with Annexin V allows to differentiate between the two subpopulations of dead cells since DAPI does not permeate cells with intact plasma membrane 16 . Thus, the double Annexin V/DAPI treatment followed by flow cytometer analysis allows discrimination of living (Annexin V − /DAPI − ), apoptotic (Annexin V + /DAPI − ) and necrotic cells (Annexin V + / DAPI + ). To address the precise TPA-induced cell death mechanism, non-adherent Jurkat cells were pre-treated with TPAs, subjected to light illumination and analysed by flow cytometry just after 30 min-illumination (= time 0) and at various times after illumination, up to 8 h ( Fig. 5 and Supplementary Fig. S2).
First, the overall kinetics of cell death (apoptotic + necrotic cells) show that TP2(3)Pyo are much less efficient than other compounds for triggering cell death upon photoactivation (Fig. 5A). For other compounds, TPAs are efficient in the following order: TP3Bzim > TP2Bzim = TP3Py > TP2Py. Their ability to induce cell death is related to the kinetics of TPA nuclear translocation after photoactivation (Fig. 5B), confirming that the nuclear relocalization process, as observed in fluorescence images of irradiated TPA-treated cells, accounts for the beginning of the cell death process. Note that no simple relationship exists between absorption properties of TPAs (relative to the band-pass filter used in flow cytometry experiments: TP3Pyo > TP3Bzim > TP2Bzim > TP3Py > TP2Pyo > TP2Py) (Fig. 5C) and their ability to trigger cell death, although the nature of the electron acceptor group appears to be critical for the overall cell death efficiency with the following hierarchy: Bzim> Py >> Pyo, involving other properties (not mutually exclusive) such as ROS generation or TPA cellular uptake.
We next quantified at various times after illumination the two subpopulations of dead cells. For each TPA, the difference between apoptotic and necrotic cells is reported as a function of time in Fig. 5D-I. At time 0, the necrotic cell subpopulation was consistently more represented than the apoptotic cell subpopulation, accounting for a small population of apoptotic cells already present before illumination (as observed in TPA-treated www.nature.com/scientificreports www.nature.com/scientificreports/ cells without illumination; Supplementary Fig. S3) for which the apoptosis->late apoptosis/necrosis transition is accelerated upon illumination. All kinetic profiles show first a significant increase in the apoptotic subpopulation, regardless of the TPA, which is also strictly dependent on illumination (compare with control experiments: without illumination +/−TPA; Supplementary Fig. S3). However, if TP3Py, TP3Bzim and TP3Pyo displayed a www.nature.com/scientificreports www.nature.com/scientificreports/ continuous increase in the apoptotic population, 2-branch TPAs displayed a distinct profile with the presence of a second phase where the necrotic subpopulation becomes significantly more represented than the apoptotic subpopulation. Importantly, such a differential qualitative behaviour in terms of cell death mechanism is independent of the overall cell death kinetic, suggesting that the Annexin V + /DAPI + subpopulation of TP2-treated cells is more related to a necrotic process than to a progression of apoptotic cells toward late apoptosis. Such a differential behaviour was already suggested for the TP3Bzim/TP2Py pair 16 . Beyond the confirmation that light irradiation may trigger different cell death mechanisms among the TPA compound family, our results show that, as observed for the subcellular TPA localization, the number of branches is the main structural feature accounting for the apoptosis-necrosis balance upon light irradiation. All TPAs primarily induce apoptosis but 2-branch compounds also induce a proper necrotic process. Altogether, our results suggest that the subcellular localization of TPAs influences the precise mechanism of cell death involved (apoptosis vs necrosis) upon photoactivation. intracellular RoS generation upon tpA photoactivation. The photoinduced TPA-mediated production of ROS was measured first in the cell context by flow cytometric analysis, using two fluorescent ROS To note, the ability of TPAs to efficiently photoinduce intracellular ROS production is not predictable by their photophysical properties as measured in vitro. Indeed, all TPAs have singlet oxygen ( 1 O 2 ) quantum yield close to 0 (Φ Δ ≤ 0.01; Supplementary Table S1 and Fig. S4A). However, they can photoinduce weak production of O 2 •− and • OH, as measured with the Diphenylisobenzofuran (DPBF) [26][27][28] and Aminophenyl fluorescein (APF) 29 Fig. S4B,C). Such a weak production via photochemical reaction cannot explain by itself the large amount of ROS detected in the cell context, but could be sufficient to initiate mitochondrial permeability transition that, in turn, amplifies ROS production, in accordance with the observed protection effect of cyclosporine A, an inhibitor of the mitochondrial permeability transition, on the TPA-mediated ROS production 17 . Furthermore, from the present flow cytometry data as shown www.nature.com/scientificreports www.nature.com/scientificreports/ in Fig. 6, for each TPA pair (TP-Bzim, TP-Py and TP-Pyo), the 2-branch compound was systematically more efficient for ROS production than its 3-branch counterpart, which is consistent with the preferential mitochondrial localization of 2-branch TPAs and probably also explains their specific effect on necrosis. Nevertheless, the number of branches is not determinant for the overall cell death efficiency which primarily relies on the chemical nature of their electron acceptor group as mentioned above (Bzim > Py >> Pyo), the exact reason is not clear but may be due to more subtle considerations about the spatio-temporal intracellular distribution of TPAs that influences both the timing and subcellular localization of ROS generation.
Relationship between the cellular uptake mechanisms and the intracellular fates of tpAs. In order to determine first whether TPA uptake is an active (energy-dependent endocytosis) or passive (energy-independent) process 31,32 , HeLa cells were incubated with TPAs at 37 °C (normal conditions) or 4 °C (active processes are blocked). For quantification reasons, the most fluorescent TPAs (TP2Py, TP3Bzim and TP2Bzim) were compared (Fig. 7A). Lowering the temperature led to a near complete inhibition of TPA uptake, regardless of the compound, demonstrating that TPA uptake occurs exclusively via endocytosis.
In order to determine the endocytic pathway 33 involved in TPA uptake, cells were pre-treated (30 min-incubation) with various endocytosis inhibitors before addition of TPA (additional 90 min-incubation) and analysed by fluorescence microscopy (Fig. 7B,C) or flow cytometry (Fig. 7D). Chlorpromazine that specifically inhibits clathrin-dependent endocytosis 31,34 had slight or no effect on TP3Bzim uptake whereas TP2Py/2Bzim uptakes were significantly affected. Consistent with the inhibition of TP2Py/2Bzim uptake by chlorpromazine, we also observed a significant inhibition of the uptake of TP2 compounds by dynasore, an inhibitor of dynamin which is involved in clathrin-dependent endocytosis 33,35 . TP3Bzim uptake which primarily corresponds to a clathrin-independent mechanism was even more sensitive to dynasore, suggesting the involvement of caveola-dependent endocytosis which is a clathrin-independent and dynamin-dependent process 33,35-39 . Accordingly, a slight but significant inhibition effect of genistein, a tyrosine kinase inhibitor that disrupts caveola-related endocytosis 40,41 , was observed on TP3Bzim uptake (after 90 min-incubation of TPAs) but not on TP2 compounds. This inhibition was even more pronounced after shorter times of incubation, 30 or 60 min ( Supplementary Fig. S5). This suggests that TP3Bzim uptake occurs predominantly via a caveola-dependent endocytosis mechanism. However, the decrease in the genistein-mediated inhibition of TP3Bzim uptake as a function of time indicates a compensatory effect through the upregulation of a clathrin-and caveola-independent mechanism. It is worth noting that the mechanism by which genistein enhances cell uptake of TP2Py/2Bzim could correspond to a related phenomenon although the precise mechanism is still unclear and requires further investigations. Treatment with cytochalasin-D, an inhibitor of F-actin elongation involved in macropinocytosis 31 , had no effect on TP2 compounds uptake. Although cell-imaging data did not show significant effect of cytochalasin-D on TP3Bzim uptake, a moderate but reproducible inhibition effect was observed in flow cytometry experiments, characterized by a better statistical significance (30,000 and 50 cells analyzed in flow cytometry and microscopy experiments per condition, respectively). Alternatively, and not mutually exclusive with the caveola-dependent endocytosis, the clathrin-and dynamin-independent macropinocytosis process 33,35,38,42,43 could be also involved in TP3Bzim uptake. In this context, the dynasore-mediated inhibition of TP3Bzim uptake could be also related to a previously described dynamin-independent inhibition mechanism of dynasore on macropinocytosis 35 . As expected, all efficient inhibitors (chlorpromazine for TP2Py (Fig. 7E) and TP2Bzim (Supplementary Fig. S6A); genistein for TP3Bzim ( Supplementary Fig. S6B)) delayed the appearance of membrane blebs and the concomitant TPA nuclear translocation, corresponding in each case to a direct effect on the intracellular TPA concentration since no qualitative influence of endocytosis inhibitors was observed on the subcellular localization of TPAs ( Supplementary Fig. S7). •-, • OH) as measured in vitro, the apparent PDT effect of photoactivated TPAs in the cell context mainly relies on high amplification of endogenous ROS production, due at least in part to their mitochondrial targeting properties. Regarding two-photon photoactivation, only TP2Py and TP2(3)Bzim efficiently triggered cell death, making them good candidates for two-photon PDT. Other compounds, TP3Py and TP2(3)Pyo, were found to be inefficient to trigger cell death upon two-photon excitation. Besides the nature of the electron acceptor group which already quantitatively influences the overall photoinduced cell death efficiency in one-photon experiments (TP-Bzim > TP-Py >> TP-Pyo), such inefficiency seems to be also related to susceptibility to photobleaching in two-photon experiments, more than to proper two-photon absorption properties (there is no correlation between the ability of TPAs to induce cell death upon two-photon excitation and their σ 2 values). Furthermore, we found that some differences exist between TPAs with respect to the precise cell death mechanism involved upon photoactivation. Clearly, the number of vinyl branches, but not the chemical nature of the acceptor group, plays a key role in determining the cell death mechanism with 2-branch TPAs inducing a proper necrotic process whereas all TPAs induce apoptosis. Such a qualitative and mechanistic difference is related to the initial subcellular TPA localization (before photoactivation) which itself is related to the cellular uptake mechanism. 2-branch compounds are taken up via clathrin-dependent endocytosis and preferentially target mitochondria whereas 3-branch compounds are taken up via clathrin-independent endocytosis (predominantly caveolae-dependent, alternatively macropinocytosis) and preferentially target late endosomes. Although this structure-activity relationship of TPAs is clear, further studies are required to precisely explain the interconnection between the endocytosis pathway, the subcellular localization and the cell death mechanism involved.  In (A,B), fluorescence images are shown for each TPA (top: TP2Py; middle: TP3Bzim; bottom: TP2Bzim) with corresponding DIC images (insets). In (C,D), quantitative analysis of TPA fluorescence intensity is shown for inhibitor-treated cells (normalized by the intensity level obtained without inhibitor). Results shown in panels B-D correspond to a TPA-incubation time of 90 min. Except the "TP3Bzim + genistein" condition, similar results were obtained with TPA-incubation times of 30 or 60 min (not shown). Results corresponding to the "TP3Bzim + genistein" condition with incubation times of 30 or 60 min are shown in Supplementary Fig. S5. The bar graphs in panels C,D show mean ± SD values from 3 independent experiments. (E) Kinetics of nuclear accumulation of TP2Py in HeLa cells treated (red) or not (blue) with chlorpromazine and subjected to continuous light illumination (light illumination conditions are similar to those described in the Fig. 2 legend). Results show that chlorpromazine delays both TPA nuclear translocation (∆ translocation ≈ +45 min) and membrane blebbing (∆ blebbing ≈ +40 min; vertical dashed line indicates the appearance of first membrane blebs). Fluorescence/DIC images are also shown at the end of the process and confirm that only cells in the irradiated area (delineated by dashed lines) display both a nuclear translocation of the TPA emission signal (fluorescence image) and membrane blebbing (blue arrows in DIC image).

Synthesis of TP2(3)Pyo is described in Supplementary
Methods. 1 H and 13 C NMR spectra are shown in Supplementary Fig. S8.
More details about experimental conditions for the determination of spectroscopic/fluorescence parameters are reported in the legend of the Supplementary Table S1. photophysical properties : In vitro measurements of 1 o 2 , o 2 ·and • oH generation upon light excitation of tpAs. The absorption spectra were recorded on a UV-3600 UV-Vis double beam spectrophotometer (Shimadzu). The fluorescence spectra were recorded on a Fluorolog FL3-222 spectrofluorimeter (HORIBA Jobin Yvon) equipped with a 450 W Xenon lamp, a thermo-stated cell compartment (25 °C), a UV-Vis photomultiplier R928 (Hamamatsu), and an InGaAs infrared detector (DSS-16A020L Electro-Optical System Inc). The excitation beam was diffracted by a double ruled grating SPEX monochromator (1200 grooves/ mm blazed at 330 nm). 1 O 2 emission was detected through a double ruled grating SPEX monochromator (600 grooves/mm blazed at 1 mm) and a long-wave pass (780 nm). All the emission spectra (fluorescence and singlet oxygen luminescence) were monitored using the same absorbance (less than 0.2) in ethanol (Sigma-Aldrich) or dimethylformamide (DMF; Fisher chemical) and corrected for the lamp and photomultiplier. Rose Bengal (Acros Organics) was chosen as a reference solution thanks to its high 1 O 2 quantum yield in ethanol (Φ ∆0 = 0.68) 45  In vitro measurements of tpA photobleaching under continuous two-photon excitation. Photobleaching measurements were performed on a on a home-built FCS system using a 80-MHz mode-locked Mai-Tai ® Ti:Sapphire tunable laser and a Nikon-TE2000 inverted microscope 47 . Two-photon excitation wavelengths are reported in the Fig. 3E legend. Fluorescence emission was separated from the excitation IR light by a Chroma 700DCSPXR dichroic mirror and an additional Chroma E700SP-2p filter. The output signal was focused onto a single-photon counting avalanche photodiode (SPCM-AQR-14, PerkinElmer). Measurements of emission intensity as a function of time were carried out in a 50-μl TPA solution (5-µM in Tris-buffer) dropped on a coverslip pre-treated with dimethyldichlorosilane. fluorescence imaging. HeLa cells were seeded in glass-bottom WillCo-Dish ® (WillCo Wells) at 6 × 10 4 cells/ml in DMEM (GibcoBRL) 48 h before treatments. Time-lapse (one-and two-photon) and colocalization imaging were obtained using a SP2 confocal microscope (Leica MicroSystem) equipped with an oiled immersion x63 objective (NA, 1.32) and an incubation chamber (37 °C, 5% CO 2 ). A continuous laser line (458 nm) was used for TPA excitation in the confocal mode (emission slit settings: 530-690 nm for TP3Bzim and TP2Bzim; 560-720 nm for TP2Py, TP3Py, TP2Pyo and TP3Pyo). The excitation source for two-photon images was a 80-MHz mode-locked Mai-Tai ® Ti:Sapphire tunable laser (720-920 nm, 100-fs laser pulse; Spectra Physics). The irradiance was calculated by measuring the power at the exit of the objective using a Vega power meter (Ophir). The power measurement was performed in the x-y scan mode (image size, 512 × 512 pixels; field of view, 140 × 140 µm; scanning frequency, 800-Hz). z-stack in two-photon microscopy corresponded to 8-10 slices (512×512 pixels; z-step, 1 µm). For comparison with one-photon confocal imaging, the time scale accounting for the effective illumination time in two-photon imaging experiments was adjusted according to the z-stack as previously described 16 . Images of HeLa cells treated with both TPA and endocytosis inhibitors were obtained with a SP8 confocal microscope (Leica MicroSystem) equipped with an oiled immersion x100 objective (NA, 1.4). The continuous laser line (458 nm) was used for TPA excitation (same emission slit settings as indicated above).
Colocalization experiments (+/− endocytosis inhibitors) are described in Supplementary Methods. www.nature.com/scientificreports www.nature.com/scientificreports/ immediately (t = 0) treated for Annexin V/DAPI staining whereas the rest of cells were further incubated for various times (up to 8 h) in the dark at 37 °C before Annexin V/DAPI treatment. Annexin V-FITC or Annexin V-Cy5 (Invitrogen) was used for experiments in the presence of TP2Py, TP3Py, TP2Pyo, TP3Pyo (red emission) or in the presence of TP2Bzim, TP3Bzim (green emission), respectively. For Annexin V staining, cells were washed in ice-cold PBS 1×(Gibco ® ) and stained according to manufacturer's instructions (BD Biosciences): the cell pellet was resuspended in 100-µl Annexin V-HEPES solution (Annexin V diluted in the binding buffer: 10-mM HEPES-NaOH, pH 7.4, 140-mM NaCl, 2.5-mM CaCl 2 ) and incubated for 15 min in the dark at room temperature. Cells were then further incubated with 400-µl of binding buffer containing 200-nM DAPI (Invitrogen) for 10 min. Cells were counted by flow cytometry (FACSCantoII flow cytometer; Becton-Dickinson) (≈100,000 cells analysed per condition). The laser/detection channels for fluorescence were 405-B for DAPI (excitation: 405 nm/emission: 450 ± 25 nm), 488-E for TP2Bzim, TP3Bzim and Annexin V-FITC (excitation: 488 nm/emission: 530 ± 15 nm), 488-D for TP2Py, TP3Py, TP2Pyo, TP3Pyo (excitation: 488 nm/emission: 585 ± 20 nm) and 633-C for Annexin V-Cy5 (excitation: 633 nm/emission: 660 ± 10 nm). before experiments. Cells were pre-treated with endocytosis inhibitors (see previous section) or with the solvent used to dissolve the inhibitor for 30 min before addition of 2-µM TPA and further incubation for 90 min at 37 °C (TPA-incubation time of 30 min was also performed in the case of genistein). Cells were washed with PBS buffer, trypsinized and transferred in FACS tubes. After centrifugation (1,500 rpm; 5 min), cell pellets were resuspended in 500-µl of fresh medium before measurement. The measurement was done by a FACSCalibur flow cytometer (Becton-Dickinson) using the 488 nm laser line for TPA excitation. Emission was measured at 585 ± 21 nm (FL2 channel) for TP2Py and 530 ± 15 nm (FL1 channel) for TP2(3)Bzim. The analysis of results (3 independent experiments for each condition; ≈30,000 cells analyzed per experiment) is explained in Supplementary Information  (Supplementary Fig. S9).