Relationship between subcellular localisation of Foscan® and caspase activation in photosensitised MCF-7 cells

The present study investigates the relationship between the subcellular localisation of Foscan® and intrinsic apoptotic pathway post Foscan®-based photodynamic therapy (PDT). With this purpose, mammary carcinoma MCF-7 cells were incubated with Foscan® for 3 or 24 h and then subjected to equitoxic light doses. Fluorescence microscopy revealed very good Foscan® co-localization to endoplasmic reticulum (ER) and Golgi apparatus after 3 h incubation with MCF-7 cells. Progressive increase in incubation time shows leakage of Foscan® from Golgi apparatus. Twenty-four hours incubation yielded a fluence-dependent enhanced induction of the ER-resident glucose-regulated protein 78 (Bip/GRP78), along with a weak mitochondrial damage, thus underscoring the ER as the main site of photodamage after prolonged incubation. Analysis of events implicated in apoptotic pathway after 24 h incubation demonstrated photodamage to Bcl-2 protein in total cellular extract, but not in the mitochondrial fraction. We further determined an increase in caspases-7 and -6 activation, which was strongly related to the expression of GRP78. The above findings demonstrate that Foscan® localisation in ER improves the photoactivation of the caspase-7 apoptotic pathway, which is poorly related to mitochondrial damage.

Photodynamic therapy (PDT) has been approved as a treatment modality in various cancerous and non-cancerous lesions (Dougherty, 2002). Photodynamic therapy, through the combination of three elements, a photosensitising agent, light and oxygen, triggers the liberation of highly reactive oxygen species, leading to damage of cellular components and eventually to apoptotic and/or necrotic cell death.
Several distinct apoptotic pathways induced by photooxidative stress have recently been established (Oleinick et al, 2002). These include the death-receptor activation at the surface of tumour cells (extrinsic pathway) or mitochondrial outer membrane permeabilisation (intrinsic pathway). Intrinsic apoptosis can be initiated from signals that originate from or converge to intracellular organelles. An important role in the initiation of intrinsic apoptosis following PDT has been attributed to oxidative stress of the endoplasmic reticulum (ER), which proceeds through disruption of calcium homeostasis and accumulation of misfolded proteins (Mak et al, 2004;Wong et al, 2004;Kessel et al, 2005;Buytaert et al, 2006). How and to what extent photodamaged-ER contributes to apoptosis induced by PDT is still under investigation.
The short migration distance of singlet oxygen (less than 0.02 mm) suggests a narrow relationship between the sites of subcellular localisation of the photosensitiser and photodamage to nearby organelles involved in apoptotic and/or necrotic cell death (Oleinick et al, 2002). It has been assumed that photosensitisers primarily localised in mitochondria are able to induce early apoptosis by rapid loss of mitochondrial transmembrane potential and/or release of apoptosis-inducing factors such as cytochrome c (cyt c), itself leading to post-mitochondrial caspase activation (Kessel and Luo, 1999;Chiu and Oleinick, 2001). Several reports investigated the possibility to modulate the mechanism of PDTinduced cell death using protocols that selectively target proapoptotic organelles (Fabris et al, 2001;Hsieh et al, 2003). Prolonged incubation of photosensitisers (24 h) with cells compared with shorter incubation times (2 -3 h) was accompanied by a more efficient apoptosis after photoirradiation. This was attributed to photosensitiser relocalisation to mitochondria or Golgi apparatus during incubation (Fabris et al, 2001;Hsieh et al, 2003).
Foscan s (meta-tetrahydroxyphenylchlorin), a second-generation photosensitiser has been granted European approval for palliative treatment of advanced head and neck cancers. The efficacy of Foscan s -PDT in the treatment of early squamous cell carcinoma Lou et al, 2004) and other malignancies has also been recently reported (Campbell et al, 2004;Lovat et al, 2005;Shikowitz et al, 2005). Endoplasmic reticulum and Golgi apparatus have been demonstrated as preferential sites of Foscan s accumulation in cultured tumour cells after 3-h incubation (Teiten et al, 2003a) leading to primary photodamage of these organelles upon irradiation (Teiten et al, 2003b). Despite that mitochondria are not specifically targeted by Foscan s , a fluence-dependent mitochondrial depolarisation has been observed, concomitant with cyt c release immediately after Foscan s -based PDT (Chen et al, 2000;Teiten et al, 2003b;Marchal et al, 2005) and consecutive post-mitochondrial apoptotic induction (Chen et al, 2000;Marchal et al, 2005). These results have suggested that initiating apoptotic events originate from ER/ Golgi and/or mitochondria. The present study investigates the relationship between the subcellular localisation of Foscan s and intrinsic apoptotic pathway post Foscan s -based PDT. With this purpose, we specified subcellular distribution of Foscan s with time in mammary carcinoma MCF-7 cells and evaluated photoinduced caspases activation in relation to the dye subcellular localisation.

Photosensitiser
Foscan s stock solution (6 Â 10 À3 M) was performed in a solvent mixture of ethanol/polyethylene glycol/water (20 : 30 : 50, by vol) and was kept at 41C in the dark. Further dilution was performed in RPMI 1640 medium supplemented with 2% (v v À1 ) FCS to reach a final Foscan s concentration of 1.5 Â 10 À6 M.
Confocal laser scanning microscopy MCF-7 cells (1 Â 10 4 cells ml À1 ) were plated into eight-well chambers Slideflask (Nunc, Roskilde, Denmark), incubated in the dark at 371C with 1.5 Â 10 À6 M Foscan s for different times (from 3 to 24 h), then rinsed in the medium and incubated with organellespecific fluorescent probes. The ER was labelled with DiOC 6 dye, applied for 1 min at a final concentration of 2.5 mg ml À1 . To identify Golgi apparatus, cells were labelled with 5 Â 10 À6 M BPC for 30 min at 41C, then rinsed with Hank's buffered saline solution containing 10 Â 10 À3 M N-2-hydroxyethylpiperazine-N 0 -2-ethanesulphonic acid (HEPES), pH 7.4 (Hanks' balanced salt solution (HBSS)/HEPES) and re-incubated at 371C for the next 30 min. The staining of mitochondria was performed by cells incubation with 500 Â 10 À9 M MitoTracker Green for 30 min at 371C. Before observation, dyes were removed by gentle rinsing with RPMI 1640 or HBSS/HEPES buffer for BPC. Double-stained cells were observed with a confocal laser-scanning microscope (SP2 AOBS LCSM, Leica microsystem, Wetzlar, Germany). An optimal pinhole size of 60.64 mm was used to exclude fluorescence light emitted from out-of-focus plane above and below the focusing plane. An oil immersion objective ( Â 63) was used to capture each image of 512 Â 512 pixels size. Confocal microscopy was coupled with microspectrofluorimetry to define spectral profiles of Foscan s and organelle probes in the same focal plane. Organelle-specific fluorescent probes were excited with a 488 nm argon/argon krypton laser and Foscan s was excited with a helium/neon laser at 633 nm. Band-pass emission filters of 505 -550 and 640 -660 nm were respectively used to discriminate organelle probes (channel 1, green) from Foscan s (channel 2, red) fluorescence.

Measurements of Foscan s intracellular concentrations
MCF-7 cells were incubated with 1.5 Â 10 À6 M Foscan s for 3 or 24 h. After incubation, cells were washed twice with cold phosphate-buffered saline (PBS), trypsinised, washed in PBS, resuspended in methanol and sonicated for 10 min. After centrifugation (3500 r.p.m., 5 min), supernatant was introduced into a 10 Â 10-mm quartz cuvette. Fluorescence intensity from cells lysates was measured with respect to a calibration curve ranging from 0 to 0.45 Â 10 À6 M Foscan s in methanol. Fluorescence intensity of each sample was normalised to the protein content determined by the DC protein assay (Bio-Rad laboratories, Marnes-la-Coquette, France). Emission spectra were carried out using SAFAS luminescence spectrometer. Spectra were collected between 600 and 700 nm (excitation wavelength 422 nm; photomultiplier voltage 650 V).
Foscan s mitochondria concentration was assessed according to Laville et al (2003). Briefly, mitochondria were isolated by differential centrifugation in 0.25 M ice-cold sucrose solution. Cell membranes were disrupted, centrifuged and the supernatant was further centrifuged for an additional 15 min at 6800 g. Sucrose was added to the pellet and the suspension was centrifuged twice for 15 min at 10 000 g. The last pellet consisted of mitochondrial fraction. Foscan s concentration was measured by spectrofluorimetry according the procedure described above.
Mitochondrial preparations were analysed for contamination with ER by Western blotting for ER-marker GRP 78. By such analysis, mitochondrial fraction was judged to be 475% free of ER contamination.

Photodynamic treatment
Four days before treatment, 4 Â 10 4 cells ml À1 were seeded in Petri dishes and then logarithmically growing MCF-7 cells were washed twice and incubated with fresh medium containing 2% (v v À1 ) FCS with 1.5 Â 10 À6 M Foscan s for 3 or 24 h before light exposure. Before photosensitisation, cells were washed three times, incubated with RPMI 9% (v v À1 ) FCS, then irradiated at room temperature with a 650-nm laser diode (F-System, Coherent) at a fixed fluence rate of 2.12 mW cm À2 . Following irradiation, cells were maintained in a humidified 5% (v v À1 ) CO 2 incubator at 371C for various times periods until experiments. seeded in triplicate into six-well plates at a density of 500 cells per well. Nine days after treatment, medium was removed, colonies were fixed with 70% (v v À1 ) ethanol and stained with 1% (w v À1 ) crystal violet (Pointet Girard, Clichy, France) for 5 min. Dye excess was carefully washed off and colonies composed of more than 50 cells were counted with a robotised image analysis system (Clemex, Longueil, Canada). Each experiment was done at least three times. Cell death percentage was obtained by referring treated samples to non-irradiated culture (drug, no light).

Immunoblotting analysis
For immunoblotting analysis, unless otherwise indicated, MCF-7 cells were collected by scraping immediately, 4 and 24 h after Foscan s -PDT. When experiments were performed in the presence of caspase-3/7 inhibitor, a 20 Â 10 À6 M Ac-DEVD-CHO was added to cells 30 min before irradiation and maintained with cells until analysis.
The procedure used for the Western blot detection has been described in detail (Marchal et al, 2005). Briefly, after protein extraction, the samples were subjected to electrophoresis in SDSpolyacrylamide gels (SDS-PAGE gels). Before immunoblotting, nonspecific binding was blocked with 0.1% (v v À1 ) Tween-20 in Tris-buffered saline containing 5% (wv À1 ) non-fat dry milk for 1 h at room temperature. Afterwards, membranes were probed overnight at 41C with an adapted concentration of each antibody followed by the appropriate HRP-conjugated antibody for 1 h at room temperature (1 : 2000). The immune complexes were detected by enhanced chemiluminescence system (GE Healthcare, Orsay, France) and visualised by autoradiography.
Probing with a mouse anti-actin antibody was used as loading control for the blots.

Flow cytometry analysis of cyt c release and mitochondrial membrane depolarization
The measurements of cyt c release and mitochondrial membrane depolarization (Dj m ) after PDT was assessed by flow cytometry technique (FACS Calibur, BD BioSciences) as described previously (Teiten et al, 2003b). Briefly, for measurements of cyt c release, permeabilised cells were labelled with PE-APO2.7 for 15 min at room temperature and subjected to flow cytometry analysis (l ex ¼ 488; l em ¼ 585742 nm; FL2).
The measurement of Dj m was performed by the use of JC-1 probe. Cells were centrifuged, the cell pellet was suspended in 1 ml medium containing 1 ml of JC-1 (final concentration 5 mg ml À1 ) and the resulting suspension was measured by flow cytometry after a 15-min incubation at 371C. The fluorescence of JC-1 mitochondriasequestered aggregates (l ex/em ¼ 488/590 nm) was detected in channel FL2 with a 585742-nm band -pass, whereas cytoplasmic monomer fluorescence (l ex/em ¼ 488/527 nm) was detected in channel FL1 with a 530730-nm band-pass filter.

Statistical analysis
Mann -Whitney's U-test was employed to determine the statistical significance with a limit set to Po0.05 using Staview 5.0 software.

Foscan s subcellular localisation in respect to time of incubation
In our previous study, we demonstrated that after 3 h incubation, Foscan localised primarily in the Golgi apparatus and ER of MCF-7 cells (Teiten et al, 2003a). In the present study, Foscan localisation in MCF-7 cells after 24 h incubation, assessed by confocal images of double-stained cells with Foscan s and specific-organelle fluorescent probes together with topographic profiles are depicted in Figure 1. Good superposition between Foscan s and specific probes is revealed in yellow.
Co-staining images and topographic profiles of Foscan s and MitoTracker green revealed a weak correlation, thus pointing out a scarce localisation of Foscan s in mitochondria ( Figure 1A and B). At the same time, Foscan s was particularly well localised in ER, as demonstrated by the perfect overlap of Foscan s and DiOC6 in topographic profiles ( Figure 1D). In contrast to 3 h incubation, characterised by the good dye localisation in Golgi apparatus (Teiten et al, 2003a), the staining patterns and topographic profiles of Golgi probe BPC and Foscan after 24 h incubation did not show any overlap ( Figure 1E and F). Foscan s accumulation in the Golgi apparatus was further mapped in the time span of 3 -24 h (Figure 2). After 3 -6 h incubation, a good localisation of Foscan s in the Golgi apparatus was evidenced by a yellow pattern in the dual-staining images and a very good overlap in fluorescence topographic profiles (Figure 2). However, from 12 h incubation a progressive extrusion of Foscan s from the Golgi becomes noticeable with increasing mismatch between Foscan s and BPC images and profiles.

Fluence-dependent cell photoinactivation with respect to incubation time
Cells incubated with Foscan s for 3 -24 h were subjected to the range of light fluences and their photocytotoxicity was further assessed by clonogenic assay. Table 1 displays the light fluences at which the same levels of photocytotoxicity (equitoxic light doses) were achieved under both incubation conditions. For all selected lethal doses (from LD 63 to LD 97 ), an incubation time of 24 h necessitated a six-to 12-fold lower light fluence than 3 h incubation.
GRP78 protein photoinduction in MCF-7 cells with respect to incubation time GRP78 induction, a stress associated ER resident protein, was assessed 4 h and 24 h post-PDT after both incubation times using Western blotting (Figure 3). Cells subjected to short incubation periods with Foscan s did not reveal any changes in GRP78 expression 4 h post-PDT ( Figure 3A), whereas an upregulation of GRP78 was observed at all light fluences 24 h after illumination ( Figure 3A). When cells were incubated with the dye for 24 h, protein induction occurred as soon as 4 h post-PDT starting from LD 85 and was significantly enhanced at all light doses 24 h post-PDT ( Figure 3B). Table 2 presents Foscan s content measured by spectrofluorimetry following extraction from whole cells and mitochondria. After 3 and 24 h incubation, whole cellular content was 9.470.4 Â 10 À11 mol Foscan s mg À1 protein and 76.075.7 Â 10 À11 mol Foscan s mg À1 protein, respectively. Foscan s concentrations in mitochondria were 3.372.2 Â 10 À11 mol Foscan s mg À1 protein and 7.772.7 Â 10 À11 mol Foscan s mg À1 protein, respectively, for 3 and 24 h incubation. The mitochondrial fraction of Foscan s represented 33.673.7 and 10.271.2% of the total cellular content at 3 and 24 h incubation, respectively.

Photoinduced mitochondrial damage with respect to incubation time
Flow cytometry evaluation of cyt c release and collapse of Dj m from photodynamically treated cells were performed immediately 24 h post-PDT.

Mitochondrial damage in cells incubated 3 h with Foscan s
and assessed immediately after PDT was dose-dependent as shown in Figure 4A and C. For low fluences (LD 63 and LD 85 ) both cyt c release and Dj m was not significantly different from control cells (P40.05), whereas at LD 93 and LD 97 we observed a significant increase in both parameters ( Figure 4A and C). Mitochondrial damage 24 h post-PDT demonstrated a significant dose-dependent increase in cyt c release and loss of Dj m ( Figure  4A and C).
Cells subjected to 24 h incubation with Foscan s demonstrated much less pronounced mitochondrial photodamage compared with 3 h incubation, irrespective of the time post-PDT ( Figure 4B and D). For instance, measurements carried out 24 h post-PDT showed from 5 (LD 63 ) to 45% (LD 97 ) of damaged cell ( Figure 4B and D), whereas equitoxic doses applied after 3 h incubation resulted in mitochondrial damage of 20 and 85% of cells ( Figure  4A and C). Therefore, the mitochondrial response was considerably decreased when cells were submitted to prolonged Foscan s incubation.

Activation of post-mitochondrial caspases cascade pathway
We further investigated the activation of post-mitochondrial apoptotic events by Western blot analysis of the proteolytic cleavage of caspases -9, -7, -6 and PARP in Foscan s photosensitised cells at 24 h post-PDT.
Immunoblotting revealed a dose-dependent cleavage of caspases-9, -7, -6 and PARP after both incubation times ( Figure 5) with an increased expression at first three light fluences, followed by reduced cleavage at the highest applied fluence. Compared with cleaved caspase-7 and PARP, caspase-9-and caspase-6-cleaved fragments displayed discrete expression. For long incubation times, expression of cleaved caspase-7, -6 and PARP was obviously higher and enzyme cleavage occurred at lower light doses (LD 63 for 24 h vs LD 93 for 3 h).

Effect of Foscan s -PDT on Bcl-2 protein expression
Expression of Bcl-2 protein was assessed immediately after irradiation of cells preloaded with Foscan s for 24 h as well as from whole cell extract from mitochondrial fraction ( Figure 6). Bcl-2 measured in the whole extract was unaffected at LD 85 but decreased with increasing light doses. In opposition, this protein remained unchanged at all applied fluences in mitochondrial fraction.

DISCUSSION
Mechanism of cell death induced by photooxidative stress is tightly related to the sites of intracellular photosensitiser accumulation. As such, a better comprehension of the role of specific organelles Confocal microscopy investigation of Foscan s intracellular distribution together with the evaluation of enzymatic post-irradiation activity of selected organelles after 3 h incubation provided an unambiguous evidence of Foscan s localisation in the ER and Golgi apparatus (Teiten et al, 2003a). In the present report, confocal microscopy investigation of Foscan s intracellular distribution showed that prolonging incubation time to 24 h induced variations in its subcellular distribution. Very good Foscan s localisation in ER along with a weak distribution in mitochondria was maintained with time ( Figure 1). The main difference observed was the poor localisation in the Golgi apparatus after 24 h, suggesting extrusion of Foscan s from Golgi with time. Cogent evidence was given by kinetic of Foscan s relocalisation over a 24h period. Good localisation of Foscan s in Golgi apparatus observed at 3 h incubation vanished progressively with time and starting from 12 h, the BPC and Foscan s fluorescence profiles were clearly different, indicating a poor affinity of Foscan s for this organelle (Figure 2). Based on this localisation pattern, and also that only 10% of total intracellular Foscan s content was found in mitochondria (Table 2), we supposed a predominant accumulation of Foscan s in the ER after 24 h incubation. Further confirmation of the enhanced presence of Foscan s in the ER was obtained in the study of the ER-resident GRP78 protein expression after PDT. This chaperone protein is known to be induced by the unfolded protein response to alleviate ER stress and maintain cell survival (Rao et al, 2002). By using mitochondria-and/or ER-targeting photosensitisers, dose-dependent elevation of GRP78 has been observed in few hours post-irradiation, thus assuming a classic ER oxidative stress response following PDT (Mak et al, 2004;Wong et al, 2004).

Compared with cells incubated for 3 h, cells incubated with
Foscan s for 24 h and subjected to equitoxic light doses exhibited higher upregulation of GRP78 (Figure 3). These combined results indicate that prolonged incubation favours ER localisation. It is pertinent to notice that clinical therapy with Foscan s uses 96 h time interval between administration and irradiation suggesting that the ultimate localisation pattern involves the ER. Short incubation time favoured massive mitochondrial membrane injury irrespective of time post-PDT ( Figure 4). Moreover, all light fluences, except the lowest one (LD 63 ), yielded significant cyt c release and collapse of Dj m right after PDT ( Figure 4A and C), thus presuming direct mitochondrial photodamage. Alternative explanation for the cyt c release and loss of Dj m could be a translocation of the proapoptotic protein Bax during irradiation (Kessel and Castelli, 2001). The process of the Bax insertion into the mitochondrial membrane is sensitive to temperature (Pryde et al, 2000). Therefore, in the next step we conducted both cell irradiation (LD 93 ) and measurements of Dj m at 151C, the temperature at which Bax penetration into mitochondria is suppressed. Upon these conditions, we observed an immediate loss of Dj m comparable with that at 371C (27.374.2 vs 33.376.1%), thus ruling out indirect mitochondria damage.
We studied further the apoptotic events induced by PDT in relation to specific Foscan s localisation.
It has been shown that PDT-induced activation of caspase-9 requires the cyt c release from mitochondria (Oleinick et al, 2002).
The massive photoinduced cyt c release after 3 h, but not after 24 h incubation ( Figure 4A), may suggest an elevated expression of cleaved caspase-9 under 3 h incubation. However, Figure 5 indicates a weak expression of cleaved caspase-9, without obvious difference in its activation at both incubation times. A plausible explanation could be an observation of the close relationship between the presence of active caspase-3 and procaspase-9 processing. Indeed, cyt c-mediated processing of procaspase-9 was reported to be strongly impaired in caspase-3-deficient cells (Blanc et al, 2000;Xue et al, 2001b).
Activation of caspases 6, -7 and PARP cleavage were studied 24 h post-PDT ( Figure 5). For both incubation times we observed the cleavage of caspases 6, -7 and PARP at all light doses, except the highest one ( Figure 5). This could be related to the inhibition of apoptosis in favour of necrosis on excess of oxidative damage (Marchal et al, 2005). In the presence of Ac-DEVD-CHO, a specific inhibitor of caspase-7, a complete inhibition of the proteolytic process of caspase-7, -6 and PARP was registered (data not shown), thus indicating that apoptotic pathway was primarily governed by caspase-7. The major role of caspase-7 in apoptotic mechanism in MCF-7 cells has been already reported (Mooney et al, 2002).
Compared to 3 h incubation, Western blotting demonstrated much stronger expression of cleaved caspases-6, -7 and PARP in cells subjected to equitoxic doses after 24 h incubation ( Figure 5A and B). These results are consistent with the improved GRP78 induction ( Figure 3) along with the weak mitochondrial damage ( Figure 4B) in cells after prolonged incubation. The mechanism underlying ER stress-associated apoptosis is poorly understood. It has been proposed recently that caspase-7 could be involved in ER -stress-induced apoptosis through its association with a subpopulation of GRP78 existing as an ER transmembrane protein (Reddy et al, 2003;Rao et al, 2004;Wu et al, 2005). Basically, caspase-7 and GRP78 bind to each other (Wu et al, 2005) preventing the activation of caspase-7 (Reddy et al, 2003). However, under severe ER stress the complex could disrupt resulting in caspase-7 activation (Rao et al, 2002;Wu et al, 2005). This explanation can also stand for the results obtained in the present study and account for an apoptotic pathway that is different from the classical post-mitochondrial process. Another scenario for the observed apoptotic effects could be offered by the photodamage to Bcl-2 protein. Photochemical destruction of both mitochondrial-and ER-localised Bcl-2 followed by apoptosis has been demonstrated in several studies (Kessel and Castelli, 2001;Xue et al, 2001a). Therefore, rther examined Bcl-2 photodamage after prolonged incubation of cells with Foscan s . We observed dose-dependent loss of Bcl-2 in whole cell extract, whereas no obvious signs of its photodestruction were noticed in mitochondrial fraction ( Figure 6). These results point out that ER-localised Bcl-2 could be a possible target of Foscan s -induced photodamage. It has recently been demonstrated that ER-resident Bcl-2 controls apoptosis through sequestration/inactivation of proapototic mitochondria BH3-only proteins that activate Bax (Thomenius et al, 2003). However, we failed to establish a clear relationship between Bcl-2 photodestruction and caspases activation as caspase-6, -7 and PARP-cleavage was maximum at LD 85 ( Figure 5B), whereas expression of Bcl-2 was unaffected at this fluence ( Figure 6A).
The duration of photosensitiser contact with the tissue is a probable determining parameter in adjusting cell-death mechanism. Applying protocols that elicit ER as the main target of Foscan s -mediated PDT could be an efficient proapoptotic strategy.