Phosphorylation of Bcl-2 in G2/M Phase-arrested Cells following Photodynamic Therapy with Hypericin Involves a CDK1-mediated Signal and Delays the Onset of Apoptosis*

The role of Bcl-2 in photodynamic therapy (PDT) is controversial, and some photosensitizers have been shown to induce Bcl-2 degradation with loss of its protective function. Hypericin is a naturally occurring photosensitizer with promising properties for the PDT of cancer. Here we show that, in HeLa cells, photoactivated hypericin does not cause Bcl-2 degradation but induces Bcl-2 phosphorylation in a dose- and time-dependent manner. Bcl-2 phosphorylation is induced by sublethal PDT doses; increasing the photodynamic stress promptly leads to apoptosis, during which Bcl-2 is neither phosphorylated nor degraded. Bcl-2 phosphorylation involves mitochondrial Bcl-2 and correlates with the kinetics of a G2/M cell cycle arrest, preceding apoptosis. The co-localization of hypericin with α-tubulin and the aberrant mitotic spindles observed following sublethal PDT doses suggest that photodamage to the microtubule network provokes the G2/M phase arrest. PDT-induced Bcl-2 phosphorylation is not altered by either the overexpression or inhibition of p38 mitogen-activated protein kinase (p38 MAPK) and c-Jun NH2-terminal protein kinase 1 (JNK1) nor by inhibiting the extracellular signal-regulated kinases (ERKs) or protein kinase C. By contrast, Bcl-2 phosphorylation is selectively suppressed by the cyclin-dependent protein kinase (CDK)-inhibitor roscovitine, completely blocked by the protein synthesis inhibitor cycloheximide and enhanced by the overexpression of CDK1, suggesting a role for this pathway. However, in an in vitro kinase assay, active CDK1/cyclin B1 complex failed to phosphorylate immunoprecipitated Bcl-2, suggesting that this protein kinase may not directly modify Bcl-2. Mutation of serine-70 to alanine in Bcl-2 abolishes PDT-induced phosphorylation and restores the caspase-3 activation to the same levels of the vector-transfected cells, indicating that Bcl-2 phosphorylation may be a signal to delay apoptosis in G2/M phase-arrested cells.

Photodynamic therapy (PDT) 1 is an emerging and very attractive therapeutic procedure suitable for the management of a variety of tumors and nonmalignant disorders. PDT involves the administration of a photosensitizing compound, which accumulates in the target cells, followed by selective irradiation of the lesion with visible light. This procedure results in a sequence of photochemical events that generate reactive oxygen species (ROS), which induce oxidative damage ultimately causing the killing of cancerous cells or other targets of therapeutic interest (1)(2)(3)(4)(5).
Hypericin is a natural photosensitizer present in Hypericum perforatum (St. John's wort) with powerful in vitro and in vivo photocytotoxic properties (6), which have pointed to its potential as a valuable tool for clinical PDT. The molecular mechanisms underlying PDT in general, and of hypericin-PDT in particular, are not completely understood, although it has become clear that cell photosensitization initiates multiple signaling pathways that ultimately lead to cell death (6 -8). We have recently shown that PDT with hypericin induces either apoptosis or necrosis, depending on the intracellular hypericin concentration and/or the light-activating dose (9 -11). In HeLa cells, apoptosis induced by hypericin-PDT is associated with a rapid and sustained activation of c-Jun NH 2 -terminal protein kinase 1 (JNK1) and p38 mitogen-activated protein kinase (p38 MAPK), which is independent of caspase activation and functionally antagonizes cell death (12). Although hypericin does not accumulate in mitochondria but localizes in membranes of the endoplasmic reticulum and Golgi apparatus (6), its photoactivation induces a rapid loss of the mitochondrial membrane potential and release of cytochrome c, which is followed by a sharp increase in the activity of caspase-3 and apoptotic cell death (9,10,13). Furthermore, recent studies have shown that hypericin-PDT-induced cytochrome c release, caspase-3 activation, and apoptosis can be substantially retarded in cells overexpressing the anti-apoptotic Bcl-2 protein (10,13).
Members of the family of Bcl-2 proteins have emerged as key regulators of apoptosis by acting either as promoters or as suppressors of the cell death process (14). In addition to its anti-apoptotic function, Bcl-2 has also been found to affect cell * This work was supported in part by the Interuniversitaire Attractiepolen of the Federal Belgian Government and by the Geconcerteerde Onderzoeksacties (Catholic University of Leuven, Leuven, Belgium). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Research Director from the Fonds National de la Recherche Scientifique (Brussels, Belgium).
Because Bcl-2 family members are important guardians of the cell death process, post-translational modifications of Bcl-2 proteins may have a great impact on cell fate. Modifications such as degradation and phosphorylation have emerged as important devices for modulating the anti-apoptotic properties of Bcl-2 (18). Bcl-2 phosphorylation occurs in a large unstructured loop of the protein located between the BH4 and BH3 regions (amino acids 32-80), a proline-rich region called the "loop region," which contains several serine and threonine residues and is therefore amenable to phosphorylation (19). Several protein kinases including Raf-1 (20), protein kinase C␣ (PKC␣) (21,22), cAMP-dependent protein kinase (23), the extracellular signal-regulated kinase 1/2 (ERK1/2), p38 MAPK, and JNK1 (24 -32), as well as the cell cycle-regulated kinase cyclin-dependent protein kinase 1 (CDK1) (33)(34)(35), have been suggested to be implicated in these phosphorylations.
However, the functional significance of the Bcl-2 phosphorylation remains highly controversial (19). Some Bcl-2 phosphorylations have been reported to enhance the cytoprotective effects of Bcl-2, whereas others rendered the cells more susceptible to apoptosis. The Bcl-2 phosphorylation induced by growth factor withdrawal, interleukin-3, or PKC agonists enhanced the anti-apoptotic Bcl-2 function (21, 22, 24, 36 -38). By contrast, paclitaxel (Taxol) and other chemotherapeutic agents that target microtubules induce phosphorylation of Bcl-2 with the abrogation of its anti-apoptotic function (23,29,31,39,40). Some other studies suggested that the functional consequences of Bcl-2 phosphorylation might be associated with mitotic arrest rather than with the regulation of apoptosis (33,34,41,42). Hence, as recently suggested (19,38), it is likely that the role of Bcl-2 phosphorylation in promoting or inhibiting apoptosis may depend on the phosphorylation site(s) involved and the cellular context where this event takes place, as well as on its duration.
Given the established relevance of the mitochondrial pathway of caspase activation in photodynamic cell killing (6,7), several recent studies have addressed the role of Bcl-2 in PDT. In some systems overexpression of Bcl-2 has been shown to prevent or delay apoptosis-related events (10,43,44). This suggests that increased levels of Bcl-2, as found in many tumors, may hamper the efficacy of PDT. By contrast, a rapid photodestruction of Bcl-2 or its down-regulation with antisense oligonucleotides resulting in a higher Bax:Bcl-2 ratio and an increased sensitivity to cell photokilling, has been recently reported in studies utilizing porphyrin-related photosensitizers (45)(46)(47).
These conflicting data on the role of Bcl-2 in PDT prompted us to undertake a study of the potential regulation of the Bcl-2 function by post-translational modifications, i.e. degradation and/or phosphorylation, and its relevance for the hypericininduced apoptosis.
In this study we report that photosensitized hypericin does not induce the loss of Bcl-2 protein but results in a dosedependent Bcl-2 phosphorylation in the mitochondria. This event is not observed in apoptotic cells but is associated with a hypericin-PDT-induced G 2 /M growth arrest, which precedes the onset of apoptosis. The co-localization of hypericin with ␣-tubulin suggests that photodamage to the microtubule network may lead to the observed formation of dysfunctional mitotic spindles, blocking the progression of mitosis in PDTtreated cells. Interfering with the p38 MAPK, JNK1, ERK, or PKC signaling pathways does not alter the level of Bcl-2 phos-phorylation in photosensitized cells. The Bcl-2 phosphorylation is enhanced in CDK1-overexpressing cells; it is inhibited by the CDK inhibitor roscovitine, and it is dependent on de novo protein synthesis. Mutation of residue Ser 70 in Bcl-2 drastically blocks PDT-induced Bcl-2 phosphorylation in mutant cells and increases caspase-3 activation to levels observed in vectortransfected cells. These results suggest that Bcl-2 phosphorylation may be a signal to restrict premature apoptosis of photodamaged cells.

EXPERIMENTAL PROCEDURES
Materials-Hypericin was prepared, purified, and dissolved in Me 2 SO as described in Ref. 48. All cell culture products were obtained from Invitrogen (Paisley, Scotland, UK). Myelin basic protein (MBP), Hoechst 33342, paclitaxel, and cycloheximide were obtained from Sigma; histone H1 was from Roche Diagnostics. The p38 MAPK inhibitor PD169316, the MEK inhibitor PD98059, and the CDK inhibitor roscovitine were purchased from Calbiochem (Bierges, Belgium), whereas staurosporine was from ICN (Costa Mesa, CA). Sytox Green was bought from Molecular Probes (Eugene, OR). protein phosphatase (400,000 units/ml) was obtained from New England Biolabs (Beverly, MA), and c-Jun-(79) was from Santa Cruz Biotechnology (Santa Cruz, CA). The p13 suc1 -Sepharose beads and cAMP-dependent protein kinase peptide inhibitor were kindly provided by Dr. J. Goris (Division of Biochemistry, KU Leuven, Belgium). Anti-poly(ADP-ribose) polymerase (PARP) antibody was from Biomol Research Laboratories (Plymouth, PA), and anti-Bcl-2 antibodies were obtained from BD PharMingen (San Diego, CA). Anti-phospho-p38 MAPK (Thr 180 /Tyr 182 ) monoclonal antibody, which specifically recognizes the phosphorylated form of the kinase, and anti-p38 MAPK antibody, as well as the anti-phospho-p44/42 MAPK (Thr 202 /Tyr 204 ) monoclonal antibody, were purchased from New England Biolabs, Inc. Monoclonal anti-␣-tubulin antibody was from Sigma. Horseradish peroxidase-conjugated secondary antibodies were from DAKO (Denmark), whereas the Alexa Fluor ® 488 goat anti-mouse antibody was from Molecular Probes, Inc. (Eugene, OR). JNK1 and ERK2 antibodies were prepared in the laboratory as described in Ref. 49.
Cell Photosensitization-HeLa cells were preincubated with hypericin for 16 h in subdued light conditions (Ͻ1 microwatt/cm 2 ). The cells were subsequently irradiated in hypericin-free medium by placing the samples on a plastic diffuser sheet 5 cm above a set of seven L18W/30 fluorescent lamps (Osram, Berlin, Germany) as described previously (9). At the surface of the diffuser, the uniform fluence rate was 4.5 milliwatts/cm 2 , as measured with an IL 1400 radiometer (International Light, Newburyport, MA). The fluence or light dose (J/cm 2 ) was calculated by multiplying the fluence rate with the time. All inhibitors used were added to the cell culture medium 1 or 2 h prior to photosensitization at concentrations indicated in the figure legend.
Cell Cycle Analysis and Assessment of Cell Viability and Apoptosis-The cell cycle profile was analyzed by flow cytometry after staining of the DNA with Sytox Green. At the indicated time points after irradiation, the cells were trypsinized, rinsed in PBS, and fixed in 70% (v/v) ethanol overnight at Ϫ20°C. After fixing, the cells were washed in ice-cold PBS and resuspended in DNA extraction buffer containing 0.2 M Na 2 HPO 4 , pH 7.8, and 0.1 M citric acid, and kept at room temperature for 5 min. The suspension was centrifuged again and resuspended for 30 min in DNA staining buffer containing 1 M Sytox Green in PBS containing 0.2 mg/ml DNase-free RNase A. Finally the cells were analyzed by flow cytometry using a 488-nm argon laser for excitation. The green fluorescence was measured at 525 nm (FACSCalibur, BD Phar-Mingen, Mountain View, CA). For each sample, 10 4 cells were measured and the data were analyzed by using the Cell Quest software.
Apoptotic cells, characterized by cellular shrinkage and membrane blebbing, were scored by phase contrast microscopic analysis following PDT in at least 10 fields with a minimum of 50 cells/field and expressed as percentage of apoptotic cells over the total number of cells. Caspase-3 activation was measured by DEVD-amc cleavage exactly as described in Ref. 10.
Preparation of Cell Extracts and Western Blotting-Cell extracts were prepared at the indicated time points following photosensitization.
Samples (50 -100 g of protein) were separated by SDS-PAGE and electrophoretically transferred to nitrocellulose (Protran ® , Schleider & Schuell GmbH, Dassel, Germany) or polyvinylidene difluoride membranes (Bio-Rad). The membranes were blocked in 5% nonfat dry milk in 50 mM Tris, pH 7.4, 150 mM NaCl, and 0.1% Tween 20 for 1 h at room temperature and incubated with primary antibody overnight at 4°C. The membranes were then washed in 50 mM Tris, pH 7.4, 150 mM NaCl, and 0.1% Tween 20 and incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibodies. Detection was carried out with the enhanced chemiluminescence detection system (PerkinElmer Life Science Products).
Phosphatase Treatment-Hypericin-photosensitized cells were lysed 16 h after treatment in a buffer consisting of 10 mM Tris, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, and 2 g/ml aprotinin. The cell extracts (150 g of proteins) were then treated with protein phosphatase (400 units) as described previously (50). Control samples containing the phosphatase in the presence of the phosphatase inhibitors EDTA (20 mM) and sodium orthovanadate (4 mM) or samples without phosphatase addition were incubated either at 30°C or at 4°C for the same length of time prior to analysis by gel electrophoresis.
Preparation of Subcellular Fractions-HeLa cells (6 ϫ 10 7 ) were washed in ice-cold PBS and resuspended in 3 ml of a hypotonic buffer (HB) containing 10 mM Tris, pH 7.4, 10 mM NaCl, 5 mM EGTA, 1.5 mM MgCl 2 , 200 nM microcystin, and protease inhibitors and incubated on ice for 25 min. The cells were then disrupted by repeated passage through a 25-gauge needle and centrifuged at 2000 ϫ g for 5 min. The supernatant (S1) was saved, and the pellet was resuspended in HB. This cell disruption procedure was repeated three times. The pellet fraction (P1) was further purified by centrifugation (Beckman SW27 rotor at 17,000 ϫ g) over a 1.6 M sucrose cushion. The nuclei were recovered at the bottom of the tubes. The supernatant S1 was centrifuged for 30 min at 13,000 ϫ g. This supernatant (S2) was kept and centrifuged at 100,000 ϫ g for 45 min to give the cytosolic cell fraction. The pellet fraction (P2), corresponding to the mitochondrial fraction, was suspended in HB and further purified by centrifugation after layering it on top of a discontinuous sucrose gradient consisting of 1 M sucrose in HB on top of 1.5 M sucrose in HB. Mitochondria were recovered at the interface of the 1 and 1.5 M sucrose layers. All manipulations were carried out at 4°C. The purity of the fractions was confirmed by assessing the localization of fraction-specific proteins like cytochrome c oxidase for the mitochondria, caspase-8 for the cytosol, and the nuclear splicing factor CDC5L (51) for the nuclei.
Immunoprecipitation and Protein Kinase Assays-JNK1 and ERK2 activities were measured by immunocomplex assays essentially as described in Ref. 49. Briefly, cell lysates normalized to contain 250 g of proteins were incubated for 2-3 h at 4°C in the presence of polyclonal anti-JNK1 or anti-ERK2 antibodies. Immune complexes were recovered with the aid of protein A-TSK beads. The beads were washed three times with PBS containing 1% Nonidet P-40 and 2 mM Na 3 VO 4 , and twice with kinase reaction buffer (20 mM Hepes, pH 7.4, 15 mM MgCl 2 , 2 mM EGTA, 1 mM DTT, 0.1% Triton X-100, 1 mM Na 3 VO 4 ). Kinase reactions were performed by resuspending the beads for 20 min at 30°C in 30 l of kinase reaction buffer containing 1 g of c-Jun-(79) protein (JNK assay) or 50 g of MBP (ERK assay), in the presence of 20 M ATP and 0.5 Ci of [␥-32 P]ATP. After boiling the samples in Laemmli sample buffer, the proteins were resolved by SDS-PAGE and the incorporation of [ 32 P]phosphate was visualized by autoradiography.
For measurement of the histone H1 kinase activity, cell extracts containing 100 g of protein were incubated for 4 h with p13 suc1 -Sepharose beads. The beads were washed once with a buffer containing 0.2% Nonidet P-40, 0.5 M NaCl, 80 mM ␤-glycerophosphate, pH 7.3, 20 mM EGTA, 15 mM MgCl 2 , 1 mM DTT, and 1 mM benzamidine, and twice with kinase buffer containing 50 mM Tris, pH 7.4, 10 mM MgCl 2 , 5 mM EGTA, and 1 mM DTT. The coupled p13 suc1 -Sepharose beads were then assayed for histone H1 kinase activity by incubating the samples for 20 min at 30°C in the presence of 0.15 mg/ml histone H1, 200 M ATP, and 2.4 Ci of [␥-32 P]ATP, 20 mM MgCl 2 , 1 M okadaic acid, and 10 M cAMP-dependent protein kinase peptide inhibitor (TTYADFIASGRT-GRRNAIHD). The reaction was stopped by spotting the mixture onto P-81 phosphocellulose (Whatman) papers. Phosphorylated histone H1 was separated from free [␥-32 P]ATP by five washes of 5 min with 0.5% phosphoric acid. The papers were dried, and radioactivity was measured by liquid scintillation counting.
Transient Transfections-HeLa cells were seeded at 0.7 ϫ 10 6 cells/ 10-cm Petri dish. After an overnight culture, the constructs were transiently co-transfected using the FuGENE 6 transfection protocol (Roche Diagnostics Corp., Indianapolis, IN) with pEGFP-C 3 , a CMV promoterdriven enhanced green fluorescent reporter construct (CLONTECH, Palo Alto, CA) (4:1) to evaluate the transfection efficiency. After 24 h of transient transfection, cells were incubated in fresh complete medium with or without hypericin for 16 h in the dark. The cells were then irradiated as described above. In each experiment the transfection efficiency yields ϳ80% as evaluated by calculating the percentage of fluorescent cells over the total number of cells in 10 independent microscopic fields. Total lysates were prepared and subjected to Western blotting.
Immunofluorescence Microscopy-Subconfluent HeLa cells, plated on two-well chambers (LabTek), were fixed with 10% paraformaldehyde in PBS for 30 min at room temperature 24 h after hypericin-PDT. Cells were then permeabilized with cold methanol and washed three times with PBS. After blocking for 3 h in PBS containing 1% bovine serum albumin, the cells were incubated overnight at 4°C with a monoclonal anti-␣-tubulin antibody (1:1000 in 1% bovine serum albumin in PBS) followed, after washing, by incubation with the Alexa Fluor ® 488 goat anti-mouse antibody for 1 h at room temperature. Preparations were then counterstained for 15 min with 3 M Hoechst 33342 to stain the nuclei, and, after washing, the glass slides were mounted in Vectashield antifade reagent (Vector Laboratories) and examined by fluorescence microscopy (Diaplan, Leitz, Germany) equipped with a digital camera (DC 200, Leica Microsystems, Wetzlar, Germany). ␣-Tubulin was visualized using a filter set consisting of a 450 -490-nm bandpass and 515-nm long pass filter for excitation and emission, respectively, whereas Hoechst 33342 was visualized using the filter set containing a 340 -380-nm bandpass filter and a 425-nm long pass filter. The hypericin fluorescence was recorded at 610 nm (bandpass filter 610/75 nm) after excitation at 535 nm (bandpass filter 535/50 nm). All filters were from Leica Microsystems.

Bcl-2 Is Phosphorylated and Not Degraded in Hypericin-
photosensitized Cells-To test the hypothesis that post-translational modifications of endogenous Bcl-2, e.g. cleavage or phosphorylation, were involved in hypericin-PDT, HeLa cells were subjected to two different protocols varying in the intensity of the photodynamic stress induced. PDT treatment with 125 nM hypericin promptly induced apoptotic cell death as already reported in previous studies (9,10,12). This is demonstrated here by the rapid kinetics of PARP cleavage and the time-dependent increase in cells with apoptotic morphology and reduced DNA content (Figs. 1A and 3). Lowering the concentration of photoactivated hypericin to 60 nM retarded the onset of PARP cleavage and apoptosis, which only became significant 48 h after irradiation (Fig. 1A). Consistently, 24 h after irradiation 80% of the cells survived this low dose PDT treatment (Fig. 1A).
Putative Bcl-2 modifications under apoptotic (125 nM hypericin) or low dose PDT (60 nM hypericin) conditions were examined over a 48-h time course. Fig. 1A shows that during apoptotic PDT the Bcl-2 protein levels remained unaltered, ruling out the proteolytic degradation of Bcl-2 as a signal that could exacerbate hypericin-induced photokilling. However, Fig. 1A clearly shows that the electrophoretic mobility pattern of the 26-kDa Bcl-2 band was modified in a time-dependent manner when cells were exposed to low dose PDT. Under these conditions, the Bcl-2 protein was discernible as two or three bands of different electrophoretic mobility; the 26-kDa fastest migrating band corresponds to the unmodified Bcl-2 and the slower migrating band(s) represent phosphorylated form(s) of the Bcl-2 protein. As a control, HeLa cells were also treated with paclitaxel, a microtubule-targeting drug that has been shown to induce Bcl-2 phosphorylation in a variety of cellular systems (33,52). As expected, paclitaxel caused a Bcl-2 mobility shift that was maximally induced 24 h after treatment, similar to what is observed after hypericin-PDT (Fig. 1B). To prove that the Bcl-2 reduced electrophoretic mobility was the result of a phosphorylation event, lysates from PDT-treated cells were subjected to treatment with protein phosphatase (Fig. 1C), a vanadate/EDTA-inhibitable phosphatase that dephosphorylates phosphoserine, phosphothreonine, and phosphotyrosine residues in proteins (53). The phosphatase treatment reverted the electrophoretically retarded Bcl-2 bands to their 26-kDa form in a vanadate/EDTA-inhibitable fashion (Fig. 1C), confirming that these bands are phosphorylated Bcl-2 products.
Bcl-2 phosphorylation was transiently induced by low dose PDT; it became apparent 7 h after irradiation and peaked at 16 -24 h, whereas after 48 h Bcl-2 appeared to be fully dephosphorylated (Fig. 1A). This transient Bcl-2 phosphorylation occurred in preapoptotic cells because it preceded the low dose PDT-induced PARP cleavage and the onset of apoptosis (Fig. 1A).
To define the threshold for the PDT-induced Bcl-2 phosphorylation, HeLa cells were treated with a range of subapoptotic PDT doses (18 -60 nM hypericin, 4 J/cm 2 ). Photo-activation of 60 nM hypericin resulted in the most prominent Bcl-2 mobility shift compared with lower PDT doses (data not shown), and this photodynamic treatment was therefore chosen to further investigate the relevance of this process.

PDT-induced Bcl-2 Phosphorylation Is Associated with Mitotic Arrest and Involves the Mitochondrial Fraction of Bcl-2-
Light microscopic analysis of asynchronously growing HeLa cells treated either with low dose PDT or with paclitaxel revealed analogous features. Treatments for 24 h induced an increase in a cell population that was phenotypically characterized by cell volume reduction, lack of adhesion, and rounded morphology (Fig. 2); these morphological changes are strikingly similar to those of cells undergoing early apoptosis, mitosis, or mitotic arrest. Cell viability started to decrease rapidly 48 h after a low dose PDT or paclitaxel, and both morphological (Fig.  2) and biochemical (Fig. 1A) analysis confirmed that this was primarily caused by the induction of apoptosis. As expected, apoptotic PDT conditions rapidly induced the typical features of cells undergoing apoptosis such as cell shrinkage, cytoplasmic blebbing (Fig. 2), and nuclear condensation (data not shown).
To better characterize these cellular responses, we analyzed the effects of low or apoptotic PDT doses on the cell cycle profile by FACS analysis after staining the cells with Sytox Green. The majority of the asynchronously growing control cells contained 2N DNA consistent with the large fraction of cells in the G 1 phase. Low dose PDT caused a slow accumulation of cells harboring G 2 /M DNA (48.6% compared with 25.4% in control cells, Fig. 3), which was maximal 24 h after irradiation, whereas at 48 h the percentage of cells with a sub-G 1 DNA content (apoptotic cells) was markedly increased (39.3% versus 12.4% in control cells, Fig. 3). Similarly, paclitaxel has been shown to induce mitotic arrest followed by apoptosis in a number of cell lines (33,42,52). The cells arrested at the G 2 /M boundary by low dose PDT and isolated 24 h after irradiation by "mitotic shaking" (54) underwent apoptosis when further cultured in complete medium in the absence of hypericin and light (data not shown), indicating that the growth arrest imposed by PDT was irreversible.
When increasing the PDT dose, the majority of the cells did not accumulate in the G 2 /M phase but rapidly underwent apoptosis, as also demonstrated by the significant accumulation of the cells with subdiploid DNA (58.2% compared with 7.9% in control cells), as early as 7 h after light exposure (Fig. 3).
The kinetics of Bcl-2 phosphorylation induced by low dose PDT closely overlapped with the time-dependent accumulation of cells with a G 2 /M DNA content 24 h after irradiation (Fig. 3). An inverse correlation between the level of Bcl-2 phosphorylation and the induction of apoptosis is evidenced by the obser- vation that 48 h after irradiation, when the sub-G 1 cell fraction was significantly increased, the Bcl-2 protein was fully dephosphorylated, as also observed in cells exposed to apoptotic PDT conditions (Figs. 1A and 3).
When compared with paclitaxel treatment, low dose PDT induced mitotic arrest in a smaller fraction of the cells, and this might explain the lower amount of phosphorylated Bcl-2 observed in the treated cells (Fig. 1). However, both PDT-and paclitaxel-induced Bcl-2 phosphorylation and mitotic arrest preceded the induction of apoptotic cell death in HeLa cells.
Signaling Pathways Involved in the PDT-induced Bcl-2 Phosphorylation-Because other studies have involved different members of the MAPK family in the Bcl-2 phosphorylation (24 -32), and in particular JNK1 has been suggested as a major Bcl-2 kinase in paclitaxel-treated cells (29,31), the question was raised whether these signaling pathways contributed to the PDT-induced Bcl-2 phosphorylation in our system as well.
To address this possibility, the kinetic profile of ERKs, JNK1, and p38 MAPK activity changes following low dose PDT was monitored in parallel with the Bcl-2 phosphorylation. The notable activation state of ERKs in untreated cells (Fig. 4A) is in agreement with the asynchronous growth of the HeLa cell population, which throughout the period examined is prevalently found in the G 1 phase, a cell cycle phase associated with elevated ERK activity (55). Similar to the effect of cell exposure to apoptotic concentrations of photo-activated hypericin (Ref. 12; Fig. 4A), low dose PDT caused a rapid inhibition of both ERK1 and ERK2 as measured by immunoprecipitation kinase assay and Western blot analysis with anti-phospho-ERKs an-tibody (Fig. 4A). Significantly, a transient increase in ERK activity, albeit to a lower extent when compared with the untreated cells, was observed between 16 and 24 h after irradiation in concomitance with maximal Bcl-2 phosphorylation (Fig. 4A) and cell cycle arrest (Fig. 3), whereas 48 h after irradiation, when cells started to die, the ERK activity leveled off. In contrast to the ERKs, untreated HeLa cells exhibited low levels of JNK1 and p38 MAPK activity, whereas following PDT these enzymes became rapidly activated (Fig. 4A). Low dose PDT-induced JNK1 and p38 MAPK activation showed a remarkable biphasic pattern; both kinase activities increased within the first hours after PDT, when Bcl-2 was fully unphosphorylated, and declined thereafter, to substantially increase again 16 -24 h after irradiation.
Although no detectable change in the different MAPK protein levels was observed during the entire time course of PDTtreated cells (Fig. 4C), the finding that cell pretreatment with the protein synthesis inhibitor cycloheximide considerably reduced the late MAPK activation phase without substantially interfering with the initial responses suggests that the second MAPK-activation signal correlates with the PDT-induced reentry in the cell cycle and that it is dependent on de novo protein synthesis (data not shown).
To substantiate the possible role of these signaling pathways in the PDT-induced Bcl-2 phosphorylation, inhibitor and transfection experiments were set up. Fig. 4B shows that blocking either the ERK-or the p38 MAPK-signals with the specific MEK-1 inhibitor PD98059 or the specific p38 MAPK inhibitor PD169316, respectively, did not change the degree of PDT- induced Bcl-2 phosphorylation, although the activation of the respective kinases was clearly suppressed by the inhibitors (data not shown). Similar results were obtained in cells pretreated with the PKC-specific inhibitor bisindolylmaleimide. In addition, overexpression of p38 MAPK or JNK1 did not affect this event either, although the transfected constructs (HAtagged) were highly expressed (Fig. 4C) and activated by PDT (data not shown). The inhibition of the JNK1 pathway by overexpressing a dominant negative mutant of SEK1 (SEK-AL), the upstream activator of JNK1 (56), did not alter the level of Bcl-2 phosphorylation in agreement with the results shown in Fig. 4C. These observations suggest that MAPK-or PKCmediated signaling pathways are unlikely to be involved in the phosphorylation of Bcl-2 in PDT-treated HeLa cells.
Another candidate for a protein kinase activated at the G 2 /M boundary is CDK1, which has recently been shown to act as a Bcl-2 kinase in okadaic acid-or in nocodazole-treated cells (34 -35). To examine the involvement of this cell cycle-regulated kinase in our system, we used the specific interaction between CDK1 and p13 suc1 -Sepharose beads to affinity-isolate this protein kinase activity from lysates of PDT-treated cells (57). The precipitates were then analyzed for CDK1 kinase  2 ), and harvested at the indicated time points after irradiation. JNK1 and ERK2 activities were determined by immunocomplex kinase assays using c-Jun-(79) and MBP as substrate, respectively. Bcl-2 phosphorylation and ERK and p38 MAPK activation were measured by Western blot using anti-Bcl-2, anti-phospho-ERKs, and anti-phospho-p38 MAPK antibodies. The blots shown are representative of at least three independent experiments. B, HeLa cells were preincubated for 1 h with or without PD98059 (20 M) or PD169316 (10 M) prior to the photo-activation of hypericin (60 nM, 4 J/cm 2 ). Lysates prepared at the indicated time points after irradiation were then subjected to Western blot analysis using anti-Bcl-2 antibodies. C, HeLa cells were transiently co-transfected with the pEGFP-C 3 reporter plasmid and the HA-JNK1 or HA-p38 MAPK expression vectors as described under "Experimental Procedures," followed by incubation with 60 nM hypericin for 16 h. The pcDNA3 vector was used to equalize the total transfected DNA. The transfection efficiency was over 80%, as revealed by fluorescence and phase-contrast microscopy. Bcl-2 phosphorylation and JNK1 and p38 MAPK expression were evaluated at the indicated time points after irradiation (4 J/cm 2 ) by Western blot with the specific antibodies as indicated. activity with histone H1 as substrate. Low dose PDT caused a time-dependent increase in the phosphorylation of histone H1, which closely correlated with the kinetics of Bcl-2 phosphorylation in the same lysates (Fig. 5A). The CDK1-like kinase activity associated with p13 suc1 increased at 7 h, peaked between 16 and 24 h, and declined to basal level 48 h after light exposure. Western blot analysis with a specific antibody against CDK1 showed that the phosphorylation/dephosphorylation pattern of this protein kinase (58), as evidenced by the electrophoretic mobility shift on SDS-PAGE, mirrored the kinetics of activation of the histone H1 kinase (Fig. 5A). Accordingly, a dramatic increase in cyclin B1 levels overlapped the time-dependent CDK1 activity changes (Fig. 5A). Cell pretreatment with cycloheximide prior to irradiation blocked the synthesis of cyclin B1 and the subsequent CDK1 activation (data not shown).
To investigate the functional correlation between CDK1 activation and Bcl-2 phosphorylation in PDT-treated cells, we evaluated the effects of cell pretreatment with roscovitine, a known specific CDK inhibitor (IC 50 ϭ 0.45 M for CDK1) (59, 60) and staurosporine, a nonspecific Ser/Thr kinase inhibitor to which CDKs are particularly sensitive (IC 50 ϭ 3.2 nM for CDK1) (61). Pretreatment of HeLa cells with roscovitine or staurosporine prior to irradiation caused a significant suppression or a complete inhibition of both PDT-induced Bcl-2 phos-phorylation (Fig. 6A) and CDK1 kinase activation as measured by histone H1 phosphorylation (Fig. 6B), respectively. In addition, Bcl-2 phosphorylation did not occur in cycloheximidepretreated cells (Fig. 6A), strengthening the functional relationship between this event and the changes in CDK1 activity.
To assess the potential effect of the inhibition of the PDTinduced CDK1 activation and Bcl-2 phosphorylation on the cell cycle arrest, we analyzed the DNA pattern of cells after low dose PDT, pretreated or not with either staurosporine or cycloheximide. Fig. 6C shows that the combination of low dose PDT with either cycloheximide or staurosporine clearly blocked the accumulation of the cells in the G 2 /M phase of the cell cycle at 24 h after irradiation, the time point with the most prominent G 2 /M growth arrest and the peak of Bcl-2 phosphorylation after low dose PDT. The increased amount of cells with an apoptotic hypoploid (sub-G 1 ) DNA content after PDT in the presence of the protein kinase inhibitor staurosporine (Fig. 6C) suggests that inhibition of a staurosporine-sensitive kinase blocks the progression into mitosis and accelerates the onset of apoptosis in photodamaged cells. In addition, cells photosensitized in the presence of the protein synthesis inhibitor cycloheximide (Fig.  6C) are blocked in the G 1 phase of the cell cycle. This is expected because de novo synthesis of, e.g., cyclins dictates the progression into the different phases of the cell cycle. Together with the results of Fig. 6 (A and B), these observations indicate that Bcl-2 phosphorylation is not mediated by a protein kinase active in the G 1 phase of the cell cycle but requires a staurosporine-sensitive protein kinase, which is active during the early phases of mitosis and which is dependent on de novo synthesis of proteins.
The involvement of a CDK1-regulated signal in the Bcl-2 phosphorylation was further corroborated by the findings that overexpression of wt CDK1 significantly increased the level of Bcl-2 phosphorylation in photosensitized cells (Fig. 6D). It is worth noting that, although under apoptotic PDT conditions, neither Bcl-2 phosphorylation (Fig. 1) nor CDK1 activation (Fig. 5B) was induced, the JNK/p38 MAPK pathways were permanently activated. In addition, the observation that both the PDT-induced stress-activated kinase activations were not affected by the CDK inhibitors roscovitine and staurosporine (data not shown), which also blocked Bcl-2 phosphorylation, further rules out the involvement of p38 MAPK and JNK1 in our system. CDK1 Co-localizes with Phosphorylated Bcl-2 in the Mitochondria-Previous studies have shown that, in addition to its predominant mitochondrial localization, a substantial fraction of the Bcl-2 protein is found in the nuclear envelope and in the contiguous endoplasmic reticulum membranes (62,63). The observation that, in PDT-treated cells, Bcl-2 phosphorylation correlates with CDK1 activation prompted us to examine whether both phosphorylated Bcl-2 and CDK1 were found in the same subcellular compartment. Subcellular fractionation of low dose PDT-treated cells revealed that the phosphorylated Bcl-2 was found exclusively in the mitochondrial fraction, whereas a substantial amount of unmodified Bcl-2 was also located in the nuclear fraction (Fig.  7). Bcl-2 was absent in the cytosol (Fig. 7), whereas only a very tiny amount was found to be associated with the light membrane fraction and it did not appear to be phosphorylated (data not shown). Interestingly, PDT-activated CDK1 was found to be distributed mainly in the cytosol and also in the nuclear and mitochondrial fraction of the cells (Fig. 7). This observation was confirmed by immunocytochemistry (data not shown) and is in agreement with previous studies showing that CDK1 and cyclin B1 prevalently localize in the cytosol of HeLa cells and that active CDK1 becomes associated to the nucleus during mitosis (64 -66). The purity of the isolated subcellular fractions was assessed by immunodetection of specific proteins such as cytochrome c oxidase for mitochondria, CDC5L for the nuclei, and After normalization for protein content, the fractions were subjected to Western blotting using anti-Bcl-2 and CDK1 antibodies. Cytochrome c oxidase (Cyt. Ox.), CDC5L, and caspase-8 antibodies were used to verify the purity of the mitochondrial, nuclear, and cytosolic fractions, respectively.
FIG. 6. Bcl-2 phosphorylation is blocked by pretreatment of the cells with cycloheximide, roscovitine, and staurosporine and is enhanced by the overexpression of CDK1. A, HeLa cells were pretreated for 2 h either with the protein synthesis inhibitor cycloheximide (1 g/ml) or with the CDK1 inhibitors roscovitine (25 M) or staurosporine (80 nM) prior to photo-activation of hypericin (60 nM, 4 J/cm 2 ). Cell lysates were prepared 7 or 24 h after irradiation and examined for Bcl-2 phosphorylation. B, CDK1 activation as measured by histone H1 phosphorylation as described under "Experimental Procedures" is inhibited in photosensitized cells pretreated with the CDK inhibitors roscovitine and staurosporine. C, control cells or low dose PDTtreated cells pretreated with staurosporine or cycloheximide or not were analyzed by FACS at 24 h after irradiation for DNA content after staining of the cells with Sytox Green as described under "Experimental Procedures." Analysis of the data was done by means of the Cell Quest software. Data are representative for two independent experiments. D, Bcl-2 phosphorylation induced by low dose PDT is enhanced by the transient overexpression of wt CDK1 in HeLa cells at the different time points after irradiation. The level of transfection was over 80% determined by the co-transfection with the pEGFP-C 3 reporter plasmid as described under "Experimental Procedures." The expression of CDK1 in vector and CDK1-transfected cells is shown at 16 h after irradiation. Results shown are representative of three independent experiments. caspase-8 for the cytosol. These proteins were found only in the expected fractions, thus ruling out the possibility of cross-contaminations.
Although CDK1 was found to co-localize with Bcl-2 in the mitochondrial fraction and immunoprecipitation experiments revealed a Bcl-2-associated kinase capable of phosphorylating histone H1 in vitro, we failed to detect CDK1 in Bcl-2 immunoprecipitates and could not show phosphorylation of Bcl-2 immunoprecipitates by exogenously added purified cyclin B1-CDK1.
Phosphorylation of Ser 70 in Photosensitized Cells Reduces the Level of Caspase-3 Activation-Ser 70 has been shown to be one of the major Bcl-2 residues phosphorylated in response to different cellular stresses, including microtubule-active drugs such as paclitaxel and nocodazole (27,29,31,35,67,68). This Bcl-2 site features a consensus sequence for phosphorylation by proline-directed protein kinases such as CDK1 and JNK1, and both protein kinases have been proposed to mediate Ser 70 phosphorylation in several cellular systems (27,29,31,35).
To evaluate the involvement of the Ser 70 as a possible residue modified in Bcl-2 during hypericin-PDT, we transfected HeLa cells with either wt Bcl-2 or a Bcl-2 mutant in which Ser 70 residue was replaced with the nonphosphorylatable alanine. Immunoblot analysis of lysates of wt Bcl-2-transfected cells shows that Bcl-2 was heavily phosphorylated following low dose PDT or paclitaxel treatment (Fig. 8). Mutation of the Ser 70 to alanine blocked PDT-mediated Bcl-2 phosphorylation and severely reduced the paclitaxel-induced effects (Fig. 8A); similar results were also obtained in nocodazole-arrested cells (data not shown). Intriguingly, analysis of caspase-3 activation in vector-transfected cells or in cells overexpressing wt Bcl-2 or the S70A mutant protein treated with either low dose PDT or paclitaxel revealed important differences. The overexpression of wt Bcl-2 in photosensitized cells significantly reduced caspase-3 activation 40 h after irradiation by 40% compared with the vector-transfected cells, whereas, in the case of paclitaxel, Bcl-2 overexpression had a relative minor effect on the level of caspase-3 activation (Fig. 8B). However, it is clear that in both cases Bcl-2 was hyperphosphorylated (Fig. 8A). In addition, in photosensitized cells, the increased expression of the nonphosphorylatable S70A Bcl-2 mutant counteracted the cytoprotective effect of the Bcl-2 overexpression and restored the caspase-3 activation to levels comparable with that in control cells. In contrast, the S70A mutant appeared to still lower the caspase-3 activation in paclitaxel-treated cells, which is in agreement with other reports (29,31).
Hypericin-induced Photodamage to Microtubule Network Triggers Alterations in Mitotic Spindle Organization and Chromosome Segregation-Intrigued by the strong correlation in the effects induced by hypericin-PDT and microtubule-targeting drugs, we explored the possibility that the hypericin-induced photodamage to microtubules may be the cause of the mitotic block and apoptosis. Because ␣-tubulin is the major structural component of the microtubule network, we examined by immunofluorescence labeling the changes in microtubule organization in cells arrested in mitosis following low dose PDT and other microtubule-targeting treatments. For comparison the Vinca alkaloid vinblastine, which interferes with the cell's ability to properly form the mitotic spindle by preventing the normal microtubule polymerization; paclitaxel, which stabilizes the microtubules; and the microtubule depolymerizer nocodazole (69) were also analyzed.
␣-Tubulin immunofluorescent staining of control interphase HeLa cells revealed a well developed microtubular network (Fig. 9A, a), whereas in dividing cells an ordered bipolar spindle apparatus with congressed chromosomes (Fig. 9A, a and f) was observed. By contrast, mitotic arrest induced by low dose PDT produced a marked increase in cells showing dramatically altered mitotic spindles with clear alterations in chromosomal segregation (Fig. 9A, b and g). Comparing the effects induced by low dose PDT with those caused by the classical antimitotic drugs vinblastine, paclitaxel, and nocodazole, we found that cells arrested at mitosis after PDT or vinblastine treatments looked very much alike, as they showed strikingly similar abnormal mitotic spindles (Fig. 9A).
Given that low dose PDT appears to have a severe effect on the microtubule functions, it is reasonable to assume that hypericin localizes in close proximity to microtubules, because the primary sites of photodamage coincide with the subcellular distribution of the photosensitizer and result from the ROS (e.g. singlet oxygen) locally produced during its light activation (2,4,5). Indeed, Fig. 9B shows that the hypericin fluorescence is absent in the nucleus and concentrates in the perinuclear area, defined as the membranes of the endoplasmic reticulum and Golgi complex (data not shown and Refs. 6 and 49), and overlapped closely with ␣-tubulin immunofluorescent labeling (Fig. 9B). Hence, it is highly possible that ROS photodynamically generated by hypericin impart damage to the microtubule network, leading to aberrant chromosomal segregation and mitotic block. DISCUSSION This study was initiated to investigate the potential involvement of Bcl-2 post-translational modifications in the hypericin photodynamic cell killing. It shows that Bcl-2 degradation is not occurring in hypericin-photosensitized cells, which therefore indicates that the loss of Bcl-2 protein is not a general phenomenon of the PDT cytotoxic actions, as recently proposed for porphyrin-related photosensitizers (45)(46)(47). Importantly, Bcl-2 undergoes a time-and dose-dependent phosphorylation in HeLa cells photosensitized with hypericin. This Bcl-2 modification is only seen when the cells are exposed to sublethal doses of PDT, and it occurs with kinetics that precede the induction of apoptosis in the photosensitized cells. Increasing the photodynamic stress by incubating the cells with higher hypericin concentrations results in the rapid induction of the apoptotic program, which is not associated with either phosphorylation or degradation of the Bcl-2 protein (Fig. 1).
We further provide evidence for a temporal correlation between Bcl-2 phosphorylation and the accumulation of cells in the G 2 /M phase of the cell cycle (Fig. 3). These mitotically arrested cells are not able to progress through the cell cycle and eventually die by apoptosis, likely because of the inability of the cells to cope with the hypericin-induced photodamage. It has indeed been proposed that prolonged mitotic arrest stimulates the apoptotic program, which probably represents an in-built safety mechanism to eliminate cells with deregulated cell cycle components (42).
Interestingly, our comparative analysis of the effects induced by PDT and the microtubule-damaging agent paclitaxel, a known inducer of Bcl-2 phosphorylation, strongly suggests that a common event underlies the mitotic arrest and the concomitant Bcl-2 phosphorylation observed in response to these anticancer treatments. In keeping with this hypothesis, we show that hypericin co-localizes with ␣-tubulin, a major component of the microtubule network, which is required for the movement of endoplasmic reticulum and Golgi membranes where this photosensitizing agent is known to accumulate ( Fig. 9B and Refs. 6 and 49). Because the proximity of the photosensitizer to its cellular target is of primary relevance in PDT (2,4,5), it may be assumed that, at the time of irradiation, the membrane-associated hypericin leads to damage of close microtubule structures resulting in an irreversible impairment of their function, an effect also observed with other photosensitizers (70 -72). The microtubule network and its dynamic behavior are major targets of a number of anticancer drugs, as their integrity is crucial to many important cellular functions, such as signaling, motility, cell shape, and cell division. In particular, during mitosis the assembly of a functional bipolar spindle is required for the passage through the metaphase/ anaphase checkpoint during which chromosomes are aligned and segregate. The inability of the photosensitized cells to form functional mitotic spindles and to segregate the chromosomes ( Fig. 9) is also observed in cells treated with paclitaxel or Vinca alkaloids (73,74) and likely explains the block in cell cycle progression in PDT-treated cells. Therefore, the capacity of PDT with hypericin (this study) and a number of other structurally unrelated drugs, including paclitaxel, vinblastine, and nocodazole, to all induce Bcl-2 phosphorylation (33, 35, 40 -42) suggests that this event is associated with the arrest at the G 2 /M phase of the cell cycle, and probably represents a checkpoint for the fidelity of chromosome segregation before cell division may take place. Increasing the photodynamic stress by using higher doses of hypericin pushes the photodamage over a threshold level, which results in the prompt activation of the cell death machinery.
Studies of the signaling pathways involved in Bcl-2 phosphorylation in response to microtubule-directed drugs, such as paclitaxel and vinblastine, have pointed to JNK1 as a major Bcl-2 kinase (29 -32). In paclitaxel-treated cells, JNK1 mediated Bcl-2 phosphorylation at multiple sites, including Ser 70 , Ser 87 , and Thr 69 , resulting in an inactivation of the Bcl-2 antiapoptotic function (29). Because of the similarity in the microtubule effects and subsequent mitotic arrest, as well as in the kinetics of Bcl-2 phosphorylation, observed in paclitaxel-or low dose PDT-treated cells, we examined the possible involvement of different MAPK signaling pathways.
Low dose PDT induces, albeit to different extents, the activation of different members of the MAPKs. In particular, the kinetics of JNK1 activation (second phase) are in good correlation with the accumulation of the cell population in the G 2 /M phase of the cell cycle and the timing of Bcl-2 phosphorylation. However, we show that specific blockage of the ERK, p38 MAPK, or JNK1 pathways or increased expression of p38 MAPK or JNK1 do not affect the extent of Bcl-2 phosphorylation in photosensitized cells (Fig. 4), suggesting that MAPKs are not involved in the PDT-mediated Bcl-2 phosphorylation.
In eukaryotic cells the progression from G 2 to the mitotic phases requires the activation of CDK1, activity of which is tightly regulated by a series of phosphorylation/dephosphorylation events and protein-protein interactions (58). In accordance with the PDT-induced increase in the cellular fraction with G 2 /M DNA content, we found a remarkable temporal correlation between Bcl-2 phosphorylation, cyclin B1 up-regulation, and CDK1 activation, suggesting that this protein kinase may participate in the Bcl-2 phosphorylation.
The results of this study favor the involvement of CDK-1 in the Bcl-2 phosphorylation in response to PDT. First, the Bcl-2 phosphorylation in photosensitized cells is strongly inhibited by pretreatment of the cells with roscovitine or staurosporine (Fig. 6A), which specifically block CDK1 without affecting the PDT-induced JNK1 activation. Second, only the overexpression of CDK1, and not of p38 MAPK or JNK1, causes a significant increase in the level of phosphorylated Bcl-2 in photosensitized cells (Fig. 6C). Third, in apoptotic PDT conditions, which cause a sustained activation of p38 MAPK and JNK1 (12), Bcl-2 is not modified and CDK-1 is not activated (Figs. 1 and 5B). In addition, subcellular fractionation and immunofluorescence analysis show that a pool of CDK1 is found in the same subcellular compartment where Bcl-2 phosphorylation takes place: the mitochondria (Fig. 7). Our observations support the view that phosphorylation of Bcl-2 is regulated by a CDK1-dependent signal, although we failed to prove that CDK1 directly phosphorylates Bcl-2 in an in vitro kinase assay. This could be explained by the removal of a necessary co-factor during the immunoprecipitation, or else the recognition of the Bcl-2 protein by CDK1 may require a specific mitochondria-bound conformation. Alternatively, CDK1 may not be the direct Bcl-2 kinase, as also proposed by Scatena et al. (42) in Taxol-induced growth arrest.
An interesting observation from this study is that in HeLa cells Ser 70 is a major Bcl-2 site phosphorylated in response to both PDT and paclitaxel treatment. This conclusion is based on the findings that substituting Ser 70 of Bcl-2 with a nonphosphorylatable alanine drastically inhibits Bcl-2 phosphorylation in both hypericin-PDT and paclitaxel-treated cells (Fig. 8). In a recent study, Bcl-2 phosphorylation induced by the microtubule-depolarizing agent nocodazole was shown to be prevented by the CDK inhibitor flavopiridol and, in agreement with the results reported here, mutation of Ser 70 residue efficiently blocked Bcl-2 phosphorylation in nocodazole M-arrested cells (35).
The functional role of the Bcl-2 phosphorylation is still a matter of debate. As suggested in two recent reviews (19,75), it is very likely that the physiological role of the Bcl-2 phosphorylation may be influenced by the phosphorylation status of additional sites, besides Ser 70 , which are located in the "loopregion" of the Bcl-2 protein. This would explain why cellular stresses, such as interleukin-3 and growth factor withdrawal, which induce phosphorylation of Bcl-2 uniquely at Ser 70 (monosite phosphorylation) (24,27,36), result in an enhancement of the cytoprotective function of the protein, whereas others, such as Taxol, inducing multisite phosphorylation of Bcl-2 (Ser 70 , Ser 87 , Thr 69 ) inactivate its anti-apoptotic function (29). In addition, the functional consequence of the Bcl-2 phosphorylation could also depend on post-phosphorylation events, inasmuch as phosphorylated Bcl-2 in nocodazole-treated cells was found to associate with Pin1, a mitotic WW domain-containing peptidylprolyl isomerase that binds to phosphorylated Ser/Thr-Pro peptide bonds, thereby changing their conformational and activity state (76).
Although the function of the Bcl-2 phosphorylation in hypericin-photosensitized cells requires further clarification, our results show that counteracting Bcl-2 Ser 70 phosphorylation by overexpressing a nonphosphorylatable mutant protein increases the activation of the executioner caspase-3 in photosensitized cells (Fig. 8), thus pointing to an activating role for this event.
In our experimental system, single mutation of Ser 70 only marginally decreased paclitaxel-induced caspase-3 activation, a result that is probably explained by the fact that multiple mutations, abrogating all phosphorylation sites involved in paclitaxel-mediated Bcl-2 phosphorylation, would be required to obtain a full effect. Hence, it seems that, although cell photosensitization or treatment with the classical microtubuleactive agent paclitaxel leads to dysregulation of microtubule functions that in turn causes mitotic arrest, the Bcl-2 phosphorylation associated with these responses is mediated by different signaling pathways, which appear to converge into Bcl-2 phosphorylation on Ser 70 . It is tempting to speculate that the functional outcome of such Bcl-2 phosphorylations would then be influenced by the phosphorylation status of other sites leading to an inactivation (paclitaxel) if multiple sites are involved, or to an increased anti-apoptotic function for a more selective phosphorylation of Ser 70 as observed in photosensitized cells.
Collectively taken, our data suggest that, in response to the hypericin-induced photodamage of microtubules, cells are unable to proceed into mitosis and arrest at the metaphase/anaphase checkpoint; this process is associated with the phosphorylation of the mitochondria-bound Bcl-2 protein on Ser 70 , through a pathway involving CDK1. This Bcl-2 phosphorylation in turn would temporary increase the cytoprotective function of Bcl-2 and delay apoptotic cell death following moderate photodynamic stress, likely until a futile attempt by the cell to repair the damage has taken place.
Although the effects of hypericin-PDT on microtubule dynamics and their functional link with Bcl-2 phosphorylation and mitotic arrest deserve to be further investigated, the present study shows that depending on the level of photodynamic stress induced and on the distinct intracellular pathways activated, photosensitized hypericin can rapidly induce cell death (apoptosis/necrosis; Refs. 9 and 10) or cell cycle arrest. Understanding how these responses are regulated and cross-talk will contribute to elucidate the antitumoral action of this drug.