The Nucleoside Analog Sangivamycin Induces Apoptotic Cell Death in Breast Carcinoma MCF7/Adriamycin-resistant Cells via Protein Kinase Cδ and JNK Activation*

Sangivamycin has shown a potent antiproliferative activity against a variety of human cancers. However, little is known about the mechanism of action underlying its antitumor activity. Here we demonstrate that sangivamycin has differential antitumor effects in drug-sensitive MCF7/wild type (WT) cells, causing growth arrest, and in multidrug-resistant MCF7/adriamycin-resistant (ADR) human breast carcinoma cells, causing massive apoptotic cell death. Comparisons between the effects of sangivamycin on these two cell lines allowed us to identify the mechanism underlying the apoptotic antitumor effect. Fluorescence-activated cell sorter analysis indicated that sangivamycin induced cell cycle arrest in the G2/M phase in MCF7/ADR cells. A marked induction of c-Jun expression as well as phosphorylation of c-Jun and JNK was observed after sangivamycin treatment of MCF7/ADR cells but not MCF7/WT cells. Sangivamycin also induced cleavage of lamin A and poly(ADP-ribose) polymerase (PARP) in MCF7/ADR cells, probably via activation of caspase-6, -7, and -9. Pretreatment with a caspase-9-specific inhibitor or pan-caspase inhibitor abolished sangivamycin-induced cleavage of lamin A and PARP but not sangivamycin induction of c-Jun expression and phosphorylation. Pretreatment of MCF7/ADR cells with SP600125, a specific inhibitor of JNK, or with rottlerin, a specific inhibitor of protein kinase Cδ (PKCδ), significantly reduced the sangivamycin-induced apoptosis and almost completely abolished sangivamycin-induced phosphorylation of c-Jun and cleavage of lamin A and PARP. Transfection of MCF7/ADR cells with PKCδ small interfering RNAs or PKCδ antibody or rottlerin pretreatment significantly suppressed the phosphorylation of JNK. Taken together, our data suggest that sangivamycin induces mitochondria-mediated apoptotic cell death of MCF7/ADR cells via activation of JNK in a protein kinase Cδ-dependent manner.

4-Amino-5-carboxamide-7-(D-ribofuranosyl)pyrrolo [2,3-d]pyrimidine (sangivamycin), 2 a naturally occurring deazapurine nucleoside analog antibiotic originally isolated from Streptomyces rimosus, was found to have antiviral activity (1)(2). It has potent cytotoxic activity in transformed and tumorigenic cells in vitro (3)(4)(5) and a significant antitumor activity without serious toxicity in vivo (6,7). Sangivamycin and its derivatives have been evaluated as potential anticancer agents against a variety of human cancers, including mammary carcinoma and leukemia (8 -13). Its ability to function as a protein kinase inhibitor has been demonstrated; it is a potent inhibitor of both protein kinase C (K i of 10 M) and rhodopsin kinase (K i of 180 nM). At higher concentrations sangivamycin also inhibits protein kinase A (IC 50 : 50 M) and ␤-adrenergic receptor kinase (K i of 67 M) (14 -16). Its inhibitory activity against partially purified nuclear protein kinases has also been reported; sangivamycin inhibited phosphorylation of histone H1 by partially purified nuclear protein kinase I (K i of 200 M) and phosphorylation of casein by nuclear protein kinase II (K i of 100 M) (17). Inhibition of cell cycle progression at the G 2 phase in tsFT210 cells, a temperature-sensitive mutant of the mouse mammary carcinoma cell line FM3A, by sangivamycin (10 M), has also been reported (18). Recently, a synthetic non-nucleoside analog of sangivamycin was shown to induce apoptosis in human leukemia cells via caspase-dependent Bid cleavage and release of cytochrome c from mitochondria (19). However, the mechanism of action underlying the antitumor activity of sangivamycin remains unknown.
Apoptosis, programmed cell death, plays an essential role in normal development and tissue homeostasis. Disruption of regulated apoptotic processes leads to various physiological abnormalities including cancer, autoimmune diseases, and neurodegenerative disorders (20 -24). Bcl-2 family member proteins and caspases have been associated with the apoptotic processes (25)(26)(27). The Bcl-2 family can be classified into two subsets, anti-apoptotic proteins such as Bcl-2, Bcl-x/L, and Mcl-1, and proapoptotic proteins such as Bax, Bak, Bid, and Bcl-x/S. Each subset can form homodimers as well as heterodimers with members of the other subset, resulting in the neutralization of opposing activities. Actually, the ratio between these two subsets helps determine in part the susceptibility of cells to death promoting signals (28,29). Activation of proapoptotic Bcl-2 family members leads to the release of cytochrome c and other apoptogenic factors, such as Smac, from the mitochondrial intermembrane space to the cytoplasm, resulting in activation of caspase cascades (caspase-9 and -3). Induction of caspase activities can also occur by interfering with caspase inhibitors (28 -30).
The apoptotic cell death triggered by many chemotherapeutic agents commonly involves stress-transducing kinases, culminating at the mitochondria. Involvement of protein kinase C in the process of apoptosis has been demonstrated (31,32). Protein kinase C␦ is a proapoptotic kinase that has been associated with cellular responses to DNA damage and apoptotic cell death (33)(34)(35). Increasing amounts of evidence suggest that proteolytic cleavage of PKC␦ by downstream effector caspases to a 40-kDa active kinase is required for the proapoptotic activity of PKC␦ (36 -38).
Activation of the stress-activated protein kinase pathway has been associated with the apoptotic response induced by some chemotherapeutic agents and other DNA damaging agents (39 -42). Activation of c-Jun NH 2 -terminal kinase (JNK) isoforms and the phosphorylation of transcription factors like c-Jun and ATF2 as well as pro-and anti-apoptotic Bcl-2 family members such as Bim, Bcl-2, and Bcl-x/L have been suggested to be important in the regulation of apoptosis (43)(44)(45).
In the present study we examined the molecular consequences of treating two human breast carcinoma cell lines, drug-sensitive MCF7/WT and multidrug-resistant MCF7/ ADR with sangivamycin. These studies focus on understanding why sangivamycin induces apoptosis only in the drug-resistant MCF7/ADR cell line.

EXPERIMENTAL PROCEDURES
Materials-Pan-caspase inhibitor (benzyloxycarbonyl-VADfluoromethyl ketone) and caspase 9-specific inhibitor (benzyloxycarbonyl-LEHD-fluoromethyl ketone) were purchased from Calbiochem. Rottlerin and SP600125 were from BioMol. Antibodies to 89-kDa cleaved poly(ADP-ribose) polymerase (PARP), lamin A, cleaved caspase 6, caspase-7 and -9, and c-Jun phosphorylated on Ser 63 or on Ser 73 and JNK phosphorylated on Thr 183 and Tyr 185 were from Cell Signaling Technology. Antibody to c-Jun was from Transduction Laboratories. Antibodies to Bcl-2, Bcl-x/L, Bax, and PKC␦ were purchased from Santa Cruz. Antibodies to p21 and Bcl-x/S were from Oncogene Sciences. PKC␦ ON-TARGETplus siRNA, ON-TARGETplus siCONTROL Non-targeting pool, c-Jun siRNA SMARTpool, Non-targeting siCONTROLpool, and DharmaFECT 1 transfection reagent were purchased from Dharmacon. Chariot protein transfection reagent was from Active Motif. Alexa Fluor 488-and Alexa Fluor 594-conjugated secondary antibodies and MitoTracker red were purchased from Molecular Probes. Sangivamycin was obtained from Natural Products Branch, Division of Cancer Treatment, NCI, National Institutes of Health.
Cell Culture-MCF7 human breast cancer cell lines (drugsensitive MCF7/WT cells and multidrug-resistant MCF7/ADR cells) were cultured in a complete medium; RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, penicillin (100 units/ml), and streptomycin (100 g/ml) at 37°C under a humidified atmosphere of 5% CO 2 in air. MCF7/ ADR cells were routinely monitored for their resistance to adriamycin. To maintain a highly drug-resistant cell population, the multidrug-resistant MCF7/ADR cells were periodically reselected by growing them in the presence of increasing concentrations of adriamycin (7-10 M) for ϳ2-month periods of time as described previously (46).
Cytotoxicity and Cell Cycle Analyses-For the sangivamycin dose-response experiment, MCF7 cells were seeded in 24-well plates (in triplicate) at a density of ϳ20,000 cells/well in a complete medium. After 24 h the medium was replaced with 1-ml aliquots of RPMI 1640 medium supplemented with 2 mM L-glutamine, penicillin (100 units/ml), streptomycin (100 g/ml), 10% heat-inactivated fetal bovine serum, and various concentrations of sangivamycin. After 72 h of incubation at 37°C, 100 l of MTT (5 mg/ml in Dulbecco's phosphate-buffered saline) was added to each well, and incubation was continued for another hour, and then the formazan precipitate was dissolved in 1 ml of Me 2 SO. The relative numbers of viable cells were estimated from the changes in absorbance at 590 nm (46). To determine the long-term cytotoxicity of sangivamycin, MCF7 cells were seeded in 6-well plates at a density of 7000 cells/well for MCF7/WT and 9000 cells/well for MCF7/ADR cells in a complete medium. After 24 h, cells were treated with or without 0.3 M sangivamycin in RPMI 1640 medium supplemented with 2 mM L-glutamine, penicillin (100 units/ml), streptomycin (100 g/ml), and 10% heat-inactivated fetal bovine serum for 6 days. Viable cells remaining in each well were trypsinized and counted using Coulter cell counter. For cell cycle analysis, the cells were harvested at the indicated time points and washed twice with cold phosphate-buffered saline. Cells (1-2 ϫ 10 6 ) were resuspended in 0.5 ml of ice-cold phosphate-buffered saline, and single cell suspension was verified microscopically. The cell suspension was gradually mixed with ice-cold absolute ethanol to the final volume of 2.0 ml and put on ice at least for 20 min before staining with propidium iodide. The cell cycle distributions were measured in a fluorescenceactivated cell sorter (FACSort) from BD Biosciences. DNA content was quantified using ModFit LT 3.0 software (Verity Software House, Inc.).
Transient Transfection of MCF7/ADR Cells with c-Jun siRNA or PKC␦ siRNA-MCF7/ADR cells were seeded in 35-mm dishes or in a 4-well chamber slide at a density of ϳ150,000 cells/dish or 25,000 cells/well, respectively, in complete medium. After 24 h cells were transfected with 100 nM siRNA or siCONTROL using DharmaFECT 1 transfection reagent as suggested by the manufacturer (Dharmacon). At 48 h post-transfection, cell culture medium was supplemented with a final concentration of 0.3 M sangivamycin or vehicle without removing the medium containing siRNA and the transfection reagent, and cells were continuously incubated for an additional 48 h until they were harvested or fixed for further analysis.
Delivery of PKC␦ Antibody into Live MCF7/ADR Cells Using Chariot-MCF7/ADR cells were seeded in 4-well chamber slides at a density of ϳ30,000 cells/well in the complete medium. After 24 h, cells were transfected with PKC␦ COOH terminus antibody using Chariot protein transfection reagent as instructed by the manufacturer (Active Motif). At 3 h post-transfection, cell culture medium was supplemented with a final concentration of 0.3 M sangivamycin or vehicle without removing the medium containing anti-PKC␦ and Chariot, and cells were continuously incubated for an additional 48 h until they were fixed for immunofluorescent microscopic analysis.
Immunoblot Analysis-Cells were washed three times with ice-cold phosphate-buffered saline, harvested by scraping from the culture dishes into ice-cold equilibration buffer containing 20 mM Tris-HCl (pH 7.5), 2 mM EDTA, 2 mM EGTA, and protease inhibitors (47) and collected by centrifugation at 800 ϫ g for 5 min at 4°C. Cell pellets were solubilized with 1% (w/v) CHAPS in equilibrium buffer on ice for 1 h. The total cell extract was then isolated by centrifugation at 50,000 ϫ g for 30 min at 4°C. Protein concentrations were determined with the Bradford assay (48) with bovine serum albumin as a standard. Equal amounts of proteins were denatured by boiling for 5 min in SDS sample buffer and subjected to SDS-polyacrylamide gel electrophoresis. The separated proteins were transferred to a polyvinylidene difluoride membrane, and nonspecific IgG binding sites were blocked by incubation with 5% nonfat dry milk for 1 h at room temperature. The membrane was then incubated overnight at 4°C with primary antibodies. Immune complexes were detected by enhanced chemiluminescence (Amersham Biosciences).
Confocal Immunofluorescent Microscopy-Cells were fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and blocked with 5% bovine serum albumin and 10% normal goat serum in phosphate-buffered saline. Primary antibody incubation was carried out overnight at 4°C, and Alexa Fluor-conjugated-secondary antibody incubation was carried out for 1 h at room temperature. Immunofluorescence was detected and analyzed using an Olympus Fluoview FV300 confocal laser scanning biological microscope system and software (Olympus America). All experiments were repeated more than three times, and the representative results were presented.

RESULTS
Differential Effects of Sangivamycin on MCF7/WT and MCF7/ADR Cell Growth-To identify biological activities of sangivamycin relevant to its antitumor activity toward human breast carcinoma cells, we first compared the chemosensitivity of drug-sensitive human breast carcinoma MCF7/WT cells and multidrug-resistant MCF7/ADR cells to sangivamycin by examining the cell growth rate and morphological changes. Sangivamycin exhibited differential cytotoxicity toward MCF7/WT cells and MCF7/ADR cells (Fig. 1A). When MCF7/ADR cells were treated with 0.1 M or higher concentrations of sangivamycin for 72 h, the number of viable cells was reduced to levels below that measured before the sangivamycin treatment, implying a cytocidal effect toward MCF7/ADR cells. In contrast, the numbers of viable MCF7/WT cells remaining after identical treatments with sangivamycin were slightly higher than the initial cell population, suggesting a cytostatic effect toward MCF7/WT cells. Similar results were observed when cells were treated with sangivamycin for 1 h followed by continued incubation in the absence of sangivamycin for 72 h, except that the dose-response curves were shifted to the right by one log scale, indicating irreversible cytocidal or cytostatic effects of sangivamycin in these cells (data not shown). A longer treatment (6 days) of these two cell lines with a lower concentration (30 nM) of sangivamycin resulted in similar responses (Fig. 1B). A similar pattern of cytocidal effect was also observed in another set of independently established drug-resistant MCF7 cell lines, MCF7/W0.3 and MCF7/W10 (46).
For the morphological study, a single pharmacological dose of sangivamycin was used, 0.3 M, a concentration that showed almost maximal cytocidal (for MCF7/ADR) or cytostatic (for MCF7/WT) effects ( Fig. 2A). Floating MCF7/ADR cells were observed within 24 h of sangivamycin addition, and most of the MCF7/ADR cells were floating by 72 h. The trypan blue dye exclusion test revealed that the majority of the floating MCF7/ ADR cells (95% at 72 h) had intact plasma membranes (data not shown). In contrast, no significant number of floating MCF7/WT cells was observed within 72 h, although growth inhibition was apparent by 48 h (Fig. 2A). When cells were further incubated for 7 days without replenishment of sangivamycin after the initial treatment, a slow recovery of cell proliferation was observed in MCF7/WT cells, suggesting a reversi-  ble cytostatic effect. However, a massive cell death was observed in MCF7/ADR cells as well as another multidrug-resistant metastatic breast cancer cell line, MDA-MB-231 (Fig.  2B), implicating an irreversible cytocidal effect of sangivamycin in these cells.

Sangivamycin-induced Inhibition of Cell Cycle Progression and Apoptotic Cell Death in MCF7/ADR Cells-To determine
if the cell growth arrest by sangivamycin is associated with the stages of cell cycle, the cell cycle distribution was measured by fluorescence-activated cell sorter analysis. In MCF7/ADR cells the relative population size of cells in the G 0 /G 1 phase continuously decreased over the 72-h period, whereas the relative S-phase population size was not affected significantly, and the G 2 /M phase cell population initially increased by 24 h and decreased thereafter. Concomitantly, a continuous increase in the sub-G 1 cell population was observed with a marked increase after 24 h and sustained over 72 h; more than 50% of the cell population contained hypodiploid DNA content (sub-G 1 peak), a hallmark of apoptosis (Fig. 2C). These results suggest that sangivamycin induces a cell cycle arrest at G 2 /M phase within 24 h, which is subsequently followed by a massive cell death in MCF7/ADR cells.
In contrast, MCF7/WT cells behaved differently in all subphases of cell cycle in response to sangivamycin; although there was an initial relative decrease in the G 0 /G 1 population by 24 h, it returned to the original level by 48 h, the G 2 /M-phase population increased by 24 h but stayed at that level, and the S-phase population was slightly increased by 24 h and then decreased significantly between 24 -48 h. Furthermore, no evidence for a significant increase in DNA fragmentation was observed in MCF7/WT cells upon exposure of cells to sangivamycin, with only 4% of the population containing a hypodiploid DNA content at 72 h. These results suggest that cell cycle progression of MCF7/WT cells is also arrested at G 2 /M phase within 24 h, but instead of massive cell death by 48 h, there is a second cell cycle arrest at the G 1 phase.
Apoptotic cell death observed in MCF7/ADR cells but not in MCF7/WT cells upon treatment with sangivamycin was further confirmed by determining annexin V binding to phosphatidylserine translocated from the inner to the outer membrane surface of the plasma membrane, a hallmark of early stage of apoptosis (Fig. 2D). In MCF7/ADR cells, annexin V binding was observed as early as 6 h after sangivamycin treatment, whereas binding of propidium iodide, a cell-impermeant dye, to the nucleic acids in the cell was observed after 12 h only in a minor fraction of annexin V-bound cells, implying cell death at the late stage of apoptosis. In MCF7/WT cells, in contrast, neither annexin V binding nor propidium iodide staining was observed even after 24 h of treatment with sangivamycin.
Because p53 plays a pivotal role in regulating a G 1 phase checkpoint and is required for G 2 /M arrest (49), the effect of sangivamycin on p53 accumulation was determined (Fig. 3). MCF7/WT cells express a wild type p53, whereas MCF7/ADR cells express an inactive mutant p53 (a 21-bp deletion at the splicing acceptor site of exon 5, resulting in a slightly smaller size of p53 protein) (50 -52). The basal level of p53 was significantly higher in MCF7/ADR cells, which gradually decreased to undetectable levels by 72 h after sangivamycin treatment. In contrast, sangivamycin treatment of MCF7/WT cells caused wild type full-length p53 to accumulate gradually, which was accompanied by a more rapid accumulation of 40 -50-kDa proteins, presumably proteolytic cleavage products of p53 as speculated from their cross-immunoreactivity to the p53 antibody. Concomitantly, a substantial amount of p21 CIP , a cyclin-dependent kinase inhibitor, was present in MCF7/WT cells, which further increased by 48 h upon exposure to sangivamycin and then remained at the elevated level in MCF7/WT cells but not in MCF7/ADR cells.
Effects of Sangivamycin on the Relative Level of Various Proteins Involved in Apoptosis-To identify possible molecular mechanisms underlying the apoptotic cell death of MCF7/ADR cells but not MCF7/WT cells upon exposure to sangivamycin, we first investigated the effect of sangivamycin treatment on the relative amounts of some proteins known to be involved in various apoptotic events (Fig. 3). Surprisingly, profound differences were seen even in the untreated cells. The amount of two anti-apoptotic proteins, Bcl-2 and Bcl-x/L, was substantially higher in untreated MCF7/WT cells than in untreated MCF7/ ADR cells. In contrast, the amount of proapoptotic protein Bclx/S was substantially higher in the untreated MCF7/ADR cells than in untreated MCF7/WT cells. Sangivamycin exposure increased these basal levels in both cell types except for Bcl-2 in MCF7/ADR cells, which remained undetectable. Thus, MCF7/ ADR cells contain substantially higher levels of a proapoptotic protein, Bcl-x/S, and substantially lower levels of two anti-apoptotic proteins, Bcl-2 and Bcl-x/L, compared with MCF7/WT cells both before and after sangivamycin treatment. However, a different pattern was observed with another proapoptotic protein, Bax, which was present at higher levels in untreated MCF7/WT cells than in untreated MCF7/ADR cells. Although the relative -fold induction of Bax in MCF/ADR cells after sangivamycin treatment appeared to be greater than in MCF7/WT cells, the total accumulation after 72 h in MCF7/ADR cells was merely more or less at the uninduced level in MCF7/WT cells.
Effects of Sangivamycin on Activation of Caspases and Cleavage of Their Substrate Proteins-Activation of caspases during apoptosis results in the cleavage of various cellular substrates including lamins and PARP (53)(54)(55). To determine if sangivamycin activates caspases in MCF7/WT and MCF7/ADR cells, we examined cleavage of lamin A and PARP after treatment with 0.3 M sangivamycin (Fig. 4A). Lamin A is a 70-kDa nuclear membrane protein and is cleaved to a 40 -45-kDa fragment and a 28-kDa fragment during apoptosis (53,56), resulting in nuclear disregulation and cell death (57,58). Cleavage of lamin A was detected using antibody specifically recognizing a 28-kDa fragment. MCF7/ADR cells have a higher basal level of lamin A than MCF7/WT cells. Upon exposure of MCF7/ADR cells to sangivamycin, a vast amount of cleavage of lamin A to a 28-kDa fragment was detected within 48 h. Lamin A cleavage was not detectable in MCF7/WT cells after exposure to sangivamycin.
Another caspase substrate, nuclear protein PARP, which is involved in DNA repair and is important for maintaining cell viability (59), is cleaved to a 89-kDa fragment and a 24-kDa fragment during apoptosis. Cleavage of PARP facilitates cellular disassembly and serves as a marker for identifying cells undergoing apoptosis (60). After sangivamycin exposure, cleavage of PARP was detected using an antibody recognizing the 89-kDa fragment but not full-length 116-kDa PARP. The 89-kDa PARP fragment was observed only in MCF7/ADR cells, where a profound amount was accumulated by 48 h after sangivamycin treatment and further accumulated by 72 h as a broad diffuse band tending toward lower mobility, implicating a further post-translational modification such as a poly(ADP)ribosylated or chromatin-associated form of the 89-kDa PARP fragment after 48 h. A similar result was also observed in two other multidrug-resistant metastatic breast cancer cell lines, MDA-MB-468 (supplemental Fig. 1) and MDA-MB-231 cells, but not in a non-malignant mammary epithelial breast cell line, MCF10A (data not shown). These data suggest that apoptotic cell death might be a prevalent mode of action of sangivamycin in a broad range of multidrug-resistant aggressive breast cancer cell lines.
Caspases exist as inactive pro-enzymes that undergo proteolytic processing to form activated enzymes, which play pivotal roles in the execution of apoptosis. Activation of caspase-6, -7, and -9 were measured by assessing the cleaved forms of these enzymes (Fig. 4B). As expected from the lamin A and PARP cleavage results (Fig. 4A), MCF7/ADR, but not MCF7/WT cells, contained the cleaved forms of caspase-6, -7, and -9 upon sangivamycin exposure. To further confirm that the sangivamycin-induced apoptosis of MCF7/ADR cells was caspase-dependent, MCF7/ADR cells were pretreated with a caspase-9-specific inhibitor or a pan-caspase inhibitor before sangivamycin treatment, and cleavage of lamin A and PARP was subsequently examined (Fig. 4C). Pretreatment with the pan-specific inhibitor completely abolished the cleavage of both lamin A and PARP, and the pharmacological inhibitor specific for caspase-9 also showed a significant inhibition of the cleavage of lamin A and PARP. These results strongly suggest that caspase-9 is cleaved and activated upon exposure of MCF7/ADR cells to sangivamycin and that the subsequent activation of the downstream caspase cascades accounts, at least in part, for the observed selective apoptotic cell death of MCF7/ADR cells.
Protein Kinase C␦-dependent Activation of JNK in Sangivamycin-treated MCF7/ADR Cells-It has been demonstrated that sustained activation of c-Jun is sufficient to trigger apoptosis, which involves the activation of caspase signaling (61). The possible involvement of c-Jun/JNK pathway in sangivamycininduced apoptosis of MCF7/ADR cells was investigated. JNK activation is mediated by phosphorylation of the threonine 183 and tyrosine 185 residues in the conserved TPY tripeptide motif in the kinase domain (42). Therefore, dually phosphorylated JNK at threonine 183 and tyrosine 185 was determined after 48-h treatment of cells with sangivamycin (Fig. 5A). A profound induction of p54 isoform as well as substantial induction of p46 isoform of phosphorylated JNK was observed in MCF7/ADR cells but not in MCF7/WT cells. Concomitantly, a time-dependent induction of c-Jun expression and sustained phosphorylation of c-Jun at serine 73 (Fig. 5B) and serine 63 (Fig. 5C) were observed in MCF7/ADR. Sangivamycin, however, did not induce c-Jun expression or phosphorylation in MCF7/WT cells where growth arrest occurs but not apoptosis.
Next, to determine if induction of c-Jun expression and phosphorylation by sangivamycin in MCF7/ADR cells was dependent on PKC␦ activity, we used three experimental approaches. First, pretreatment of MCF7/ADR cells with 20 M rottlerin, a specific inhibitor of PKC␦, for 1 h significantly reduced the induction of c-Jun expression and phosphorylation and suppressed the caspase-9 activation, cleavage of lamin A and PARP by sangivamycin (Fig. 5C). Secondly, transfection of MCF7/ ADR cells with PKC␦ antibody raised against the catalytic domain using protein transfection reagent, Chariot, significantly inhibited sangivamycin-induced JNK phosphorylation in MCF7/ADR cells (Fig. 5D). Statistical analysis of the immunofluorescence shown in Fig. 5D using Fluoview FV300 confocal laser microscope system software revealed that immunofluorescence of the activated form of phosphorylated JNK was increased by 18 -21-fold in sangivamycin-treated MCF7/ADR cells as compared with the untreated control cells (supplemental Fig. 2). Rottlerin pretreatment or transfection of MCF7/ ADR cells with PKC␦ antibody using Chariot reduced the sangivamycin-induced immunofluorescence by 12 or 10fold, respectively (data not shown). Lastly, the relationship between PKC␦ expression and phosphorylation of JNK was examined in MCF7/ADR cells treated with sangivamycin for 48 h after transfection with PKC␦ siRNAs (Fig. 5E). In a major-ity of cells in which the PKC␦ expression was knock down by PKC␦ siRNA, sangivamycin treatment did not induce phosphorylation of JNK; only a minor fraction of cells where PKC␦ expression was not abolished exhibited phosphorylation of JNK upon exposure to sangivamycin. Taken together, these data indicate that sangivamycin treatment of MCF7/ADR cells induces PKC␦-dependent activation of JNK, resulting in the induction of c-Jun expression and phosphorylation.

Activation of Caspases in Sangivamycin-treated MCF7/ADR
Cells Is Dependent on Activation of JNKs-Phosphorylation of c-Jun at serine 63 and serine 73 is mediated by JNK in response to various stimuli (62)(63)(64)(65)(66). Emerging evidence supports a role for JNK in stress-induced mitochondrial apoptotic pathways in a variety of cell systems (67)(68)(69)(70)(71). To confirm that the profound induction of c-Jun observed in sangivamycin-treated MCF7/ ADR cells was mediated through activation of JNK, c-Jun  induction was examined in MCF7 cells pretreated with SP600125, a specific inhibitor of JNK activity, before exposure to sangivamycin (Fig. 6A). Pretreatment of MCF7/ADR cells with 20 M SP600125 markedly reduced the induction of both phosphorylated form of c-Jun (upper band of c-Jun doublet) and nonphosphorylated form of c-Jun (lower band) after sangivamycin treatment. Reduction in the accumulation of total c-Jun is ascribed to the positive feedback regulation of c-Jun transcription. SP600125 alone did not affect c-Jun induction in either MCF7/ADR cells or MCF7/WT cells. Sangivamycin-induced cleavage of lamin A and PARP in MCF7/ADR cells was also significantly reduced when cells were pretreated with SP600125 (Fig. 6B), suggesting that the caspase cascade activation observed in MCF7/ADR cells treated with sangivamycin ( Fig. 4B) is downstream of JNK activation. Although pretreatment of MCF7/ADR cells with caspase-9 inhibitor or pan-specific caspase inhibitor abolished sangivamycin-induced cleavage of lamin A and PARP (Fig. 4C), such treatment did not reduce sangivamycin-induced c-Jun expression (Fig. 6C), which further confirms the notion that activation of caspases is downstream of JNK activation. Increased c-Jun activity has been linked to two apparently opposing physiological consequences, triggering either cell proliferation or apoptosis (61).
To investigate whether the marked induction of c-Jun in sangivamycin-treated MCF7/ADR cells is a prerequisite for the sangivamycin-induced apoptosis in these cells, a c-Jun siRNA expression vector was transiently transfected into MCF7/ADR cells, and sangivamycin-induced cleavage of lamin A and PARP was subsequently examined (Fig. 6D). Transfection of MCF7/ ADR cells with c-Jun siRNA almost abolished basal expression and greatly reduced induction of c-Jun by sangivamycin, whereas the control transfection with non-target siRNA did not affect either basal expression or induction of c-Jun. However, sangivamycin-induced cleavage of lamin A and PARP was not reduced in MCF7/ADR cells expressing c-Jun siRNA when compared with that in non-target siRNA-expressing cells. These data suggest that the marked induction of c-Jun observed in sangivamycin-treated MCF7/ADR cells is a cellular response to the sangivamycin-induced JNK activation but is dispensable for sangivamycin-induced apoptosis in these cells.
Protein Kinase C␦ and JNK-dependent Phosphorylation of Bcl-x/L in Sangivamycin-treated MCF7/ADR Cells-JNK-mediated phosphorylation of Bcl-x/L has been shown to render various cancer cells more susceptible to apoptotic cell death by inactivating the normal anti-apoptotic activity of Bcl-x/L (71)(72)(73)(74)(75). Phosphorylation of Bcl-x/L can be detected as more slowly migrating bands on Western blot (71)(72)(73). Treatment of MCF7/ADR cells with sangivamycin resulted in the induction of the slower migrating, presumably phosphorylated form of Bcl-x/L, and this induction was significantly inhibited when cells were pretreated with rottlerin, a specific inhibitor of PKC␦ (Fig. 7A), or SP600125, a specific inhibitor of JNK (Fig. 7B).
Under the same experimental conditions, pretreatment of MCF7/ADR cells with these inhibitors significantly block the production of apoptotic cells (Fig. 7C), strongly suggesting that activation of PKC␦ and JNK are involved in the process of sangivamycin-induced apoptosis.
To determine if the activated form (dually phosphorylated on threonine and tyrosine residues) of JNK was co-localized with mitochondrial protein Bcl-x/L, confocal immunofluorescence microscopic analysis was performed (Fig. 7D). As expected, no active form of JNK was detectable in both untreated MCF7/ ADR and MCF7/WT cells. After sangivamycin treatment, a significantly enhanced JNK activity was observed in MCF7/ADR cells, whereas a barely detectable level of JNK activity was observed in MCF7/WT cells. Consistent with the data shown in Chemosensitivity to sangivamycin with or without pretreatment with rottlerin or SP600125 was visually evaluated by using an Olympus Vanox microscopy. D, co-localization of Bcl-x/L and active form of dually phosphorylated JNK in sangivamycin-treated MCF7/ADR cells. Immunofluorescence of Bcl-x/L (red) and phosphorylated JNK (p-JNK) (green) was detected using an Olympus Fluoview FV300 confocal laser microscope system. E, co-localization of active form of dually p-JNK and Mitotracker Red in sangivamycin-treated MCF7/ADR cells. After treatment of cells with 0.3 M sangivamycin for 48 h, cells were stained with Mitotracker Red, a mitochondrion-selective cell permeant fluorescent stain, and subsequently subjected to immunocytochemical analysis. Immunofluorescence of phosphorylated JNK was detected by confocal microscopy. Co-localization of phosphorylated JNK (green) and Mitotracker Red (red) in sangivamycintreated MCF7/ADR cells is represented as yellow color immunofluorescence. Fig. 3, Bcl-x/L was abundant in MCF7/WT and was not changed significantly after sangivamycin treatment. In contrast, a lower basal level of Bcl-x/L was detected in MCF7/ADR cells, which was induced significantly after sangivamycin treatment. Moreover, the co-localization of phosphorylated JNK and Bcl-x/L was apparent only in sangivamycin-treated MCF7/ADR cells. Mitochondrial localization of the active form of dually phosphorylated JNK was further confirmed by using Mitotracker Red, a mitochondrion-selective cell-permeant fluorescent stain (Fig. 7E). The significant amount of Mitotracker Red was colocalized with the phosphorylated JNK present in sangivamycin-treated MCF7/ADR cells, whereas the co-localization in sangivamycin-treated MCF7/WT cells was hardly detectable. These data strongly suggest that sangivamycin-induced apoptosis in MCF7/ADR cells is mediated at least in part via phosphorylation of the mitochondrial protein Bcl-x/L in a manner that requires the sequential activation of PKC␦ and JNK.

DISCUSSION
Our data demonstrate that sangivamycin exerts differential cytotoxic effects in drug-sensitive MCF7/WT and multidrugresistant MCF7/ADR human breast carcinoma cells. Based on morphological observations, fluorescence-activated cell sorter analysis, and biochemical analysis of apoptotic marker proteins, submicromolar concentrations of sangivamycin induce significant apoptotic cell death in MCF7/ADR cells and significant growth arrest without detectable apoptosis in the parental MCF7/WT cells. Sangivamycin also induces early cell cycle arrest at the G 2 /M phase (within 24 h) in both MCF7/WT and MCF7/ADR cells and late arrest at G 1 phase (by 48 h) in MCF7/WT cells but not in MCF7/ADR cells. The apoptotic death of MCF7/ADR cells occurs in the G 2 /Marrested subpopulation.
Further characterization of various biochemical events accompanying growth arrest and apoptosis in these two cell lines demonstrate striking differences in the relative amounts of several proteins. First, treatment of MCF7/WT cells with sangivamycin results in increased p53 accumulation followed by increased accumulation of p21 Waf1/Cip1 after a 24-h lag period, consistent with a p53-dependent induction of p21 Waf1/Cip1 expression. Therefore, the sustained activation and induction of p53 and p21 as well as the persistent G 2 cell cycle arrest observed in MCF7/WT cells might account, at least in part, for the sangivamycin-induced prolonged growth arrest. In this view, both the irreversible G 2 cell cycle arrest and the subsequent apoptotic death of MCF7/ADR cells could be ascribed, at least in part, to an inactive p53 and non-detectable level of p21. Interestingly, our MCF7/WT data demonstrate a significant sangivamycin-induced accumulation of 40 -50-kDa proteins that have cross-immunoreactivity to the p53 antibody, consistent with recent reports showing proteolytic cleavage products of p53 (into 50-, 40-, and 35-kDa fragments) via its interaction with damaged DNA in vitro (77)(78)(79). Although sangivamycin promotes apparently similar p53 proteolytic cleavage, it does not induce a significant amount of DNA damage in MCF7/WT cells as assessed by monitoring the formation of foci of either phospho-p53 (serine 15) or a widely-used surrogate marker for DNA damage response, ␥-H2AX, although such foci formation was detected in sangivamycin-treated MCF7/ADR cells (supplemental Fig. 2A). DNA damage-induced foci were apparent in both cell lines after irradiation (supplemental Fig. 2B). Although further investigation is needed, we tentatively conclude that the presumed proteolytic p53 products produced in sangivamycin-treated MCF7/WT cells are not generated by DNA damage.
Second, relative amounts of pro-and anti-apoptotic proteins (at least four Bcl-2-related proteins) present in these two cell lines are very different. MCF7/WT cells contain relatively higher level of two anti-apoptotic proteins, Bcl-2 and Bcl-x/L, whereas MCF7/ADR cells contain the relatively higher level of a proapoptotic protein, Bcl-x/S. Furthermore, sangivamycin induces a persistent activation and translocation of JNK to mitochondria in MCF7/ADR cells, presumably phosphorylating Bcl-x/L in mitochondria, consistent with the recent findings that dimerization and post-translational modifications such as phosphorylation can alter the function of these proteins (72)(73)(74)(75)(76). Although a role for c-Jun in apoptosis has been demonstrated (61), our data with c-Jun siRNA indicate that the marked induction of c-Jun activity is dispensable for the sangivamycin-induced apoptosis of MCF7/ADR cells.
Stepwise analyses of our findings with PKC␦ siRNA and various pharmacological inhibitors (rottlerin, SP600125, caspase-9 inhibitor, and pan-specific caspase inhibitor) support that sangivamycin-induced apoptosis in MCF7/ADR cells is mediated through the sequential activation of PKC␦, JNK, the mitochondrial apoptotic pathway acting via phosphorylation of Bcl-x/L, and then activation of the caspase-9 cascade, leading to the activation of caspase-6 and -7 and subsequent cleavage of lamin A and PARP. Taken together, the sangivamycin-induced apoptotic pathway is supported by the observations that (i) sangivamycin induces persistent phosphorylation of c-Jun on serine 63 and 73 and a marked induction of c-Jun expression in MCF7/ ADR cells but not in MCF7/WT cells, (ii) sangivamycin induces phosphorylation of Bcl-x/L in MCF-7/ADR cells, (iii) sangivamycin induces cleavage of procaspase-9, -6, and -7 to active caspases in MCF7/ADR cells but not in MCF7/WT cells, (iv) sangivamycin induces cleavage of death-substrates lamin A and PARP in MCF7/ADR cells but not in MCF7/WT cells, (v) preincubation of cells with a specific JNK inhibitor, SP600125, abolishes sangivamycin-induced c-Jun phosphorylation and induction, Bcl-x/L phosphorylation, and cleavage of lamin A and PARP, (vi) preincubation with caspase-9 inhibitor significantly suppresses sangivamycin-induced cleavage of lamin A and PARP but not phosphorylation and expression of c-Jun, (vii) preincubation with the PKC␦ inhibitor, rottlerin, abolishes sangivamycin-induced c-Jun phosphorylation and expression, Bcl-x/L phosphorylation, caspase-9 activation, and cleavage of lamin A and PARP, and (viii) suppression of PKC␦ expression or activity by siRNA or anti-PKC␦, respectively, abolishes activation of JNK. These results suggest that the sangivamycininduced apoptosis of MCF7/ADR cells requires the induction of PKC␦ activity as an upstream signal transducer.
In conclusion, our data indicate that sangivamycin can exert chemotherapeutic activity by manipulating at least two key processes, cell cycle arrest and apoptosis. Depending upon the array or balance of cellular signal transducing proteins expressed in untreated cells, sangivamycin exposure triggers different signal transduction pathways, leading to different outcomes, apoptosis or growth arrest. Whether an apoptotic response will always occur in various multidrug-resistant breast cancer cells treated with sangivamycin whereas isogenic drug sensitive cells will respond with only growth arrest remains to be determined. Under our experimental conditions, sangivamycin-induced apoptosis in MCF7/ADR multidrug-resistant breast cancer cells is characterized by an irreversible G 2 /M cell cycle arrest, the sequential activation of PKC␦, JNK, and caspase-9, which is led to a massive apoptotic cell death. The molecular mechanism underlying G 2 /M cell cycle arrest remains unknown. The proposed working models (Fig. 8) for the molecular basis of two different sangivamycin-induced effects at the cellular level can be tested in multiple ways, which if confirmed would provide invaluable insights for approaches to the development of effective chemotherapy by targeting appropriate signal transducer that may, however, vary in different type of cancers.