Inhibition of Phosphorylation of BAD and Raf-1 by Akt Sensitizes Human Ovarian Cancer Cells to Paclitaxel*

We studied the roles of the phosphatidylinositol 3-kinase (PI-3K)-Akt-BAD cascade, ERK-BAD cascade, and Akt-Raf-1 cascade in the paclitaxel-resistant SW626 human ovarian cancer cell line, which lacks functional p53. Treatment of SW626 cells with paclitaxel activates Akt and ERK with different time frames. Interference with the Akt cascade either by treatment with PI-3K inhibitor (wortmannin or LY294002) or by exogenous expression of a dominant negative Akt in SW626 cells caused decreased cell viability following treatment with paclitaxel. Interference with the ERK cascade by treatment with an MEK inhibitor, PD98059, in SW626 cells also caused decreased cell viability following treatment with paclitaxel. Treatment of cells with paclitaxel also stimulated the phosphorylation of BAD at both the Ser-112 and Ser-136 sites. The phosphorylation of BAD at Ser-136 was blocked by treatment with wortmannin or cotransfection with the dominant negative Akt. On the other hand, the phosphorylation of BAD at Ser-112 was blocked by PD98059. We further examined the role of BAD in the viability following paclitaxel treatment using BAD mutants. Exogenous expression of doubly substituted BAD2SA in SW626 cells caused decreased viability following treatment with paclitaxel. Moreover, because paclitaxel-induced apoptosis is mediated by activated Raf-1 and the region surrounding Ser-259 in Raf-1 conforms to a consensus sequence for phosphorylation by Akt, the regulation of Raf-1 by Akt was examined. We demonstrated an association between Akt and Raf-1 and showed that the phosphorylation of Raf-1 on Ser-259 induced by paclitaxel was blocked by treatment with wortmannin or LY294002. Furthermore, interference with the Akt cascade induced by paclitaxel up-regulated Raf-1 activity, and expression of constitutively active Akt inhibited Raf-1 activity, suggesting that Akt negatively regulates Raf-1. Our findings suggest that paclitaxel induces the phosphorylation of BAD Ser-112 via the ERK cascade, and the phosphorylation of both BAD Ser-136 and Raf-1 Ser-259 via the PI-3K-Akt cascade, and that inhibition of either of these cascades sensitizes ovarian cancer cells to paclitaxel.

Paclitaxel, a natural product originally isolated from the bark of Taxus brevifolia, has significant anti-tumor activity in several human tumors, particularly in advanced ovarian and breast carcinomas (1,2). Unlike other antimicrotubule agents, paclitaxel increases tubulin polymerization, stabilizes microtubules, and prevents tubulin depolymerization, ultimately causing tubulin bundling (3)(4)(5). These effects of the drug are correlated with the arrest of cells in the G 2 /M phase of the cell cycle, as well as cellular toxicity (5)(6)(7). The inclusion of paclitaxel in the treatment of patients with newly diagnosed ovarian cancer has led to improved response rates and prolonged median survival compared with the results of prior regimens (8). Nevertheless, the majority of patients with advanced ovarian cancer are destined to relapse and subsequently to develop resistance to initially active drugs such as paclitaxel (9).
Insight into the mechanisms of drug resistance has been gained from a better understanding of the pathway of apoptosis. Apoptosis is the final common pathway of many if not all forms of chemotherapy-induced cell death (10,11). Molecules known to predispose cells to apoptosis have been shown to enhance sensitivity to a variety of chemotherapeutic agents that induce damage to DNA or to the mitotic spindle (10,12,13). Conversely, defects in the apoptotic pathway have been observed to confer insensitivity to the cytotoxic effects of chemotherapy and may therefore represent an important mechanism for the development of chemoresistance (14 -19). The apoptotic response of a cell damaged by chemotherapy partly depends upon the balance between proteins that predispose to apoptosis, such as BAX, and proteins that antagonize apoptosis, such as Bcl-X L or Bcl-2 (20,21). Recently, a signaling pathway by which extracellular stimuli suppress apoptosis has been characterized. One of the first reports on survival signaling demonstrated an association between activation of the mitogen-activated protein (MAP) 1 kinase cascade with survival in PC-12 cells (22). Recent studies (23)(24)(25) suggest that paclitaxel affects the activities of several intracellular tyrosine and serine/threonine protein kinases. For example, paclitaxel causes rapid activation of MAP kinase (26). Another signaling pathway requiring phosphatidylinositol 3-kinase (PI-3K) activity was shown to be associated with anti-apoptotic signaling in neurons, fibroblasts, and hematopoietic cells (27)(28)(29). Subsequently, the serine/threonine kinase termed Akt or protein kinase B was identified as a downstream component of survival signaling through PI-3K (30 -34). Akt plays a central role in promoting the survival of a wide range of cell types (30 -36).
Recently BAD, a pro-apoptotic member of the Bcl-2 family, was found to be a substrate of Akt, identifying an intersection point of pro-and anti-apoptotic regulatory cascades (37,38). Whereas BAD can be phosphorylated at either Ser-112 or Ser-136 (39), Akt phosphorylates BAD specifically at Ser-136 (37,38). The involvement of MAP kinase/ERK kinase (MEK) upstream of BAD phosphorylation (40) and the promotion of cell survival by the Ras-MAP kinase signaling pathway by phosphorylation of BAD at Ser-112 (41)(42)(43) were recently reported. BAD is capable of forming heterodimers with the anti-apoptotic proteins Bcl-X L or Bcl-2 and antagonizes their anti-apoptotic activity (44).
However, the effects of the chemotherapeutic agent paclitaxel, which induces damage to the mitotic spindle, on the PI-3K-Akt-BAD cascade, the ERK-BAD cascade, and the association of Akt and Raf- 1 have not yet been reported. Therefore, we sought to determine whether the PI-3K-Akt-BAD cascade, the ERK-BAD cascade, or the phosphorylation of Raf-1 by Akt plays a role in the cellular stress response to paclitaxel by using the paclitaxel-resistant SW626 human ovarian cancer cell line, which lacks functional p53. Here we provide evidence that paclitaxel induced the activation of Akt and ERK with different time frames, followed by phosphorylation of BAD. Paclitaxel also induced the phosphorylation and negative regulation of Raf-1 by Akt. Moreover, whereas inhibition of Raf-1 activity by geldanamycin markedly increased the cell viability following treatment with paclitaxel, inhibition of Akt and BAD markedly decreased the cell viability following treatment with paclitaxel.

EXPERIMENTAL PROCEDURES
Materials-Paclitaxel was a gift from Bristol-Myers Squibb Co. Wortmannin and geldanamycin were purchased from Sigma. LY294002 was purchased from Calbiochem. Geneticin was purchased from Invitrogen. ECL Western blotting detection reagents were obtained from Amersham Biosciences. PD98059, mouse monoclonal anti-phospho-ERK antibody, rabbit polyclonal anti-phospho-BAD Ser-112 and BAD Ser-136 antibody, rabbit polyclonal anti-BAD antibody, rabbit polyclonal anti-Akt antibody, rabbit polyclonal anti-phospho-Akt antibody, the Akt kinase assay kit including GSK-3 fusion protein and a phosphospecific GSK-3␣/␤ antibody, and rabbit polyclonal anti-phospho-Raf Ser-259 antibody were obtained from New England Biolabs (Beverly, MA). The Raf-1 immunoprecipitation kinase cascade assay kit was obtained from Upstate Biotechnology, Inc. (Lake Placid, NY). Rabbit polyclonal anti-ERK1 antibody, rabbit polyclonal anti-Raf-1 antibody, and rabbit polyclonal anti-HA antibody were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The Cell Titer 96-cell proliferation assay kit was obtained from Promega (Madison, WI).
Cell Cultures-Human ovarian cancer SW626 and Caov-3 (52, 53) cell lines were obtained from American Type Culture Collection (Manassas, VA). Human ovarian cancer A2780 cell line (52), derived from a patient prior to treatment, was kindly provided by Dr. T. Tsuruo (Institute of Molecular and Cellular Biosciences, Tokyo, Japan) and Drs. R. F. Ozols and T. C. Hamilton (NCI, National Institutes of Health, Bethesda). Cells were cultured at 37°C in Dulbecco's modified Eagle's medium with 10% fetal bovine serum in a water-saturated atmosphere of 95% air and 5% CO 2 .
Constructs-The plasmids encoding the HA-tagged form of kinasedead Akt (HA-AktK179M) and constitutively active Akt (HA-m⌬4 -129 Akt) and the BAD mutant (pCDNA3-BAD Ser to Ala at 112 and 136) used in this study have been described previously (37).
Clone Selection-A2780/PTX and Caov-3/PTX, both of which have acquired in vitro resistance to paclitaxel, were isolated as individual clones in a single-step selection by exposing the parental cells to 100 nM paclitaxel. SW626 cells were transfected for 12 h in 6-well tissue culture plates with 2 g of the empty vector (CMV-6) or CMV-6 containing the gene for HA-tagged Akt K179M and the neomycin resistance gene, or with the empty vector (pCDNA3), which contains a neomycin resistance gene, or pCDNA3-BAD2SA, encoding mutant BAD in which Ser is converted to Ala at 112 and 136, using LipofectAMINE plus (Invitrogen) (52). Clonal selection was performed by adding geneticin to the medium at 200 g/ml final concentration 2 days after the transfection. After 3 weeks, several clones were isolated using cloning rings. Selected clones were then maintained in medium supplemented with geneticin (100 g/ml), and only low passage number cells (p Ͻ 10) were used for the experiments described here. For analysis of the levels of expressed Akt and BAD protein products, empty vector (CMV-6)-or AktK179Mexpressing SW626 cells, or empty vector (pCDNA3)-and BAD2SAexpressing SW626 cells cultured in 100-mm dishes were lysed in icecold HNTG buffer (50 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl 2 , 1 mM EDTA, 10 mM sodium pyrophosphate, 100 M sodium orthovanadate, 100 mM NaF, 10 g/ml aprotinin, 10 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride) (53). The proteins immunoprecipitated with anti-HA antibody for analysis of the levels of expressed Akt protein or the whole-cell lysates for analysis of the levels of expressed BAD protein were analyzed by electrophoresis on 8% SDS-polyacrylamide gels. Transfer to nitrocellulose, Western blotting with anti-Akt or anti-BAD antibody and washing were performed as described elsewhere (53).
Cytotoxicity-A significant level of DNA repair was detected 5 days after, but not 24 -72 h after, cisplatin-induced DNA damage (54). Therefore, cell viability (52) was assessed by the addition of paclitaxel for 1 h 1 day after seeding 1.0 ϫ 10 2 test cells per well into 96-well plates followed by exchanging the medium with fresh medium. The number of surviving cells was determined 5 days later by determination of the A 590 nm of the dissolved formazan product after addition of 3-[4,5dimethylthiazol-2-yl]-5-[3-carboxymethoxy-phenyl]-2-[4-sulfophenyl]-2H-tetrazolium, inner salt, for 1 h as described by the manufacturer (Promega). By the 5th day, the cells without treatment were growing exponentially and had achieved a density of 6.0 -6.5 ϫ 10 3 cells/well. All experiments were carried out in quadruplicate, and the viability was expressed as the ratio of the number of viable cells with paclitaxel treatment to that without treatment.
Assay of Akt Kinase Activity and Akt Phosphorylation-Cells were incubated in the absence of serum for 16 h and then treated with various materials. They were then washed twice with phosphate-buffered saline (PBS) and lysed in ice-cold lysis buffer A (20 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM EGTA, 2.5 mM sodium pyrophosphate, 1 mM ␤-glycerophosphate, 1 mM sodium orthovanadate, 1 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride). The extracts were centrifuged to remove cellular debris, and the protein content of the supernatants was determined using the Bio-Rad protein assay reagent (Bio-Rad). Two hundred fifty g of protein from the lysate samples was incubated with gentle rocking at 4°C overnight with immobilized Akt antibody cross-linked to agarose hydrazide beads. After Akt was selectively immunoprecipitated from the cell lysates, the immunoprecipitated products were washed twice with lysis buffer and twice with kinase assay buffer (25 mM Tris, pH 7.5, 10 mM MgCl 2 , 5 mM ␤-glycerophosphate, 0.1 mM sodium orthovanadate, and 2 mM dithiothreitol), and the samples were resuspended in 40 l of kinase assay buffer containing 200 M ATP and 1 g of GSK-3␣ fusion protein. The kinase reaction was allowed to proceed at 30°C for 30 min and was stopped by the addition of Laemmli SDS sample buffer (55). Reaction products were resolved by 15% SDS-PAGE followed by Western blotting with a phospho-GSK-3␣ antibody.
For analysis of phosphorylated Akt or the total amount of Akt, 250 g of protein from the lysate samples was resolved by 8% SDS-PAGE, followed by Western blotting with anti-phospho-Akt antibody or anti-Akt antibody, respectively. For analysis of the effect of ectopically expressed Akt on Akt activity, empty vector (CMV-6)-or AktK179Mexpressing SW626 cells grown in 100-mm dishes were treated with 1 M paclitaxel for 3 h. The lysates samples were immunoprecipitated with anti-HA antibody. Immune complexes were precipitated with protein A-Sepharose, and the kinase reaction was carried out in the presence of cold ATP and GSK-3␣ fusion protein, as described above.
Assay of ERK Activity and ERK Phosphorylation-Cells were incubated in the absence of serum for 16 h and then treated with various agents. They were then washed twice with PBS and lysed in ice-cold HNTG buffer (53). The extracts were centrifuged to remove cellular debris, and the protein content of the supernatants was determined using the Bio-Rad protein assay reagent. Two hundred fifty g of protein from the lysate samples was used for immunoprecipitation by treatment with ERK1 rabbit polyclonal antibody at 4°C for 2 h. The immunoprecipitated products were washed once with HNTG buffer, twice with 0.5 M LiCl, 0.1 M Tris, pH 8.0, and once with kinase assay buffer (25 mM HEPES, pH 7.2-7.4, 10 mM MgCl 2 , 10 mM MnCl 2 , and 1 mM dithiothreitol), and then the samples were resuspended in 30 l of kinase assay buffer containing 10 g of myelin basic protein and 40 M [␥-32 P] ATP (1 Ci) as described previously (52,53). The kinase reaction was allowed to proceed at room temperature for 5 min and was stopped by the addition of Laemmli SDS sample buffer (55). Reaction products were resolved by 15% SDS-PAGE and visualized by autoradiography.
For analysis of phosphorylated ERK or the total amount of ERK, 250 g of protein from the lysate samples was resolved by 8% SDS-PAGE, followed by Western blotting with anti-phospho-ERK antibody or anti-ERK1 rabbit polyclonal antibody, respectively. Phosphorylation of BAD-Cells cultured in 100-mm dishes were transfected with 4 g of pCDNA3-BAD using LipofectAMINE plus. At 72 h after transfection, serum-deprived cells were treated with various materials. They were then washed twice with PBS and lysed in ice-cold HNTG buffer (53). The lysate samples were immunoprecipitated with phospho-BAD (Ser-112) or phospho-BAD (Ser-136) antibody. Immune complexes were precipitated with protein A-Sepharose, and the isolated proteins were analyzed by electrophoresis on 8% SDS-PAGE. Transfer to nitrocellulose, Western blotting with phospho-BAD (Ser-112) or phospho-BAD (Ser-136) antibody, and washing were performed as described elsewhere (53).
For analysis of the total amount of BAD, 250 g of protein from the lysate samples was resolved by 8% SDS-PAGE, followed by Western blotting with anti-BAD antibody. For analysis of the effect of expressed BAD on BAD phosphorylation, empty vector (pCDNA3)-and BAD2SAexpressing SW626 cells grown in 100-mm dishes were treated with 1 M paclitaxel for 3 h. The lysate samples was resolved by 8% SDS-PAGE, followed by Western blotting with phospho-BAD (Ser-112) or phospho-BAD (Ser-136) antibody, as described above.
Phosphorylation of Raf-Cells were incubated in the absence of serum for 16 h and then treated with various agents. They were then washed twice with PBS and lysed in ice-cold lysis buffer B (50 mM Tris, pH 7.5, 1% Triton X-100, 1 mM EDTA, 1 mM EGTA, 50 mM sodium fluoride, 5 mM sodium pyrophosphate, 10 mM sodium glycerophosphate, 0.5 mM sodium orthovanadate, 0.1% 2-mercaptoethanol, 1 g/ml leupeptin, 0.1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml pepstatin, and 1 M microcystin). The lysate samples were im-munoprecipitated with Raf-1 antibody. Immune complexes were precipitated with protein A-Sepharose, and the isolated proteins were analyzed by 8% SDS-PAGE. Transfer to nitrocellulose, Western blotting with either phospho-Raf-1 (Ser-259) antibody for analysis of the phosphorylation of Raf-1 or Raf-1 antibody for analysis of the total amount of Raf-1, and washing were performed as described elsewhere (53).
Assay of Raf Kinase Activity-Cells were incubated in the absence of serum for 16 h and then treated with various materials. They were then washed twice with PBS and lysed in ice-cold lysis buffer B. The extracts were centrifuged to remove cellular debris, and the protein content of the supernatants was determined using the Bio-Rad protein assay reagent. Five hundred g of protein from the lysate samples was incubated with gentle rocking at 4°C overnight with immobilized Raf-1 antibody cross-linked to protein G-agarose beads. After Raf-1 was selectively immunoprecipitated from the cell lysates, the immunoprecipitated products were washed twice in 500 l of lysis buffer B and once in 80 l of assay dilution buffer (20 mM MOPS, pH 7.2, 25 mM ␤-glycerophosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol). Immunoprecipitated Raf-1 was then incubated with 0.4 g of inactive GST-MEK 1 and 1 g of inactive GST-MAP kinase 2/Erk2 in 30 l of kinase assay buffer (20 mM MOPS, pH 7.2, 25 mM MgCl 2 , 25 mM ␤-glycerophosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol, 150 M ATP) for 30 min at 30°C. Four microliters of supernatant from the reaction were transferred to a new tube and incubated for 10 min at 30°C with 10 l of assay dilution buffer containing 20 g of myelin basic protein and 40 M [␥-32 P]ATP (1 Ci) as described previously (56). The kinase reaction was stopped by the addition of Laemmli SDS sample buffer (55). Reaction products were resolved by 15% SDS-PAGE and visualized by autoradiography.
Statistics-Statistical analysis was performed using Student's t test, and p Ͻ 0.05 was considered significant. Data are expressed as the mean Ϯ S.E.

RESULTS
Activation of Akt and ERK-We first evaluated whether Akt was activated or phosphorylated in response to paclitaxel treatment in SW626 human ovarian cancer cells. Cultured cells were exposed to 1 M paclitaxel for the indicated times (Fig.  1A). To examine Akt activation, cell lysates were immunoprecipitated with immobilized anti-Akt antibody, and the in vitro kinase reaction was carried out in the presence of cold ATP and GSK-3␣ fusion protein, followed by Western blotting with antiphospho-GSK-3␣/␤ antibody. To examine Akt phosphorylation, cell lysates were resolved by SDS-PAGE, followed by Western blotting with anti-phospho-Akt antibody. Induction of both activation and phosphorylation of Akt by paclitaxel in SW626 cells was detected at 30 min, reached a plateau at 3 h, and declined thereafter (Fig. 1A, upper and middle panels). We confirmed that the total amount of Akt in each lane was the same (Fig. 1A, lower panel). Because Akt is an effector of survival signaling downstream from PI-3K, we next determined whether stimulation of cells with paclitaxel increased the activity of Akt through a PI-3K-dependent mechanism. Cells were stimulated with paclitaxel in the presence of wortmannin, a PI-3K inhibitor, and the kinase activity of Akt was assayed. The induction of Akt activity and phosphorylation by paclitaxel was inhibited by wortmannin (Fig. 1A, upper and  middle panels, lane 7). These results indicate that paclitaxel induces Akt activity and phosphorylation through a PI-3K-dependent mechanism.
Recently it was reported that paclitaxel activates the MEK/ ERK cascade and that MEK inhibition enhances paclitaxelinduced tumor apoptosis (57). We therefore investigated whether paclitaxel induces the activation or the phosphorylation of ERK. Cells were treated with 1 M paclitaxel for the times indicated in Fig. 1B. To examine ERK activation, cell lysates were immunoprecipitated with anti-ERK antibody and examined for ERK activity by assaying the incorporation of 32 P into MBP (Fig. 1B, upper panel). To examine ERK phosphorylation, cell lysates were resolved by SDS-PAGE, followed by Western blotting with anti-phospho-ERK antibody (Fig. 1B, middle panel). We confirmed that the total amount of ERK in each lane was the same (Fig. 1B, lower panel). Both the activation and phosphorylation of ERK were fully induced by paclitaxel by 30 min and declined thereafter, a time frame different from that of the affects of paclitaxel on Akt.
Although pretreatment with a PI-3K inhibitor, wortmannin, completely abolished the induction of Akt activity and phosphorylation by paclitaxel (Fig. 1A, upper and middle panels,  lane 7), pretreatment with either wortmannin or LY294002 had no effect on the induction of ERK activity and phosphorylation by paclitaxel (Fig. 1B, upper and middle panels, lanes 8  and 9). In addition, although an MEK inhibitor, PD98059, completely abolished the induction of ERK activity and phosphorylation by paclitaxel (Fig. 1B, upper and middle panels,  lane 7), it had no effect on the induction of Akt activity or phosphorylation by paclitaxel (Fig. 1A, upper and middle panels, lane 8), indicating the absence of cross-talk between the Akt and ERK cascades in the activation induced by paclitaxel.
Kinase-deficient Akt, Wortmannin, LY294002, and PD98059 Sensitize SW626 Cells to Paclitaxel-To determine whether Akt activation is necessary for cell survival signaling after paclitaxel-induced cell damage, the effect of paclitaxel treatment on the viability of SW626 cells either pretreated with wortmannin or LY294002 ( Fig. 2A) or expressing a kinase-  1 and 2), wild-type Akt- (lanes 3 and 4), or AktK179M-expressing cells (lanes 5 and 6) grown in 100-mm dishes were treated with 1 M paclitaxel for 3 h (lanes 2, 4, and 6). The lysate samples were immunoprecipitated with anti-HA antibody. For analysis of the level of ectopically expressed Akt protein products (lower panel), immune complexes were precipitated with protein A-Sepharose, and the isolated proteins were analyzed by electrophoresis on 8% SDS-polyacrylamide gels, followed by Western blotting with anti-Akt antibody. For analysis of the effects of ectopically expressed Akt on Akt activity (upper panel), immune complexes were precipitated with protein A-Sepharose, and the kinase reaction was carried out in the presence of cold ATP and GSK-3␣ fusion protein, as described under "Experimental Procedures." C, cell viability was assessed in empty vector (CMV-6)-or AktK179M-expressing cells after treatment with the indicated concentrations of paclitaxel as described under "Experimental Procedures." deficient Akt (AktK179M), which is an Akt derivative rendered kinase-inactive by a point mutation within the catalytic domain (30, 37) (Fig. 2C), was compared with its effect on the viability of the parental SW626 cells or an empty vector (CMV-6)-expressing control line, respectively. We first confirmed the overexpression of ectopically expressed Akt protein products (Fig. 2B, lower panel) and the negative effects of the expression of HA-AktK179M on Akt activity (Fig. 2B, upper panel). The viability of SW626 cells was not detectably affected by increasing concentrations of paclitaxel of Ͼ200 nM. Further titrations revealed IC 50 values of 1900 and 1980 nM for parental (Table I) and empty vector-expressing (Table II) SW626 cells, respectively. In contrast, the cells pretreated with wortmannin or LY294002 exhibited an IC 50 as low as 125 or 130 nM, respectively, indicating over 15.2-or 14.6-fold greater sensitivity to paclitaxel than the untreated cells, respectively ( Fig. 2A and Table I). The AktK179M-expressing SW626 cells exhibited an IC 50 as low as 130 nM, indicating over 13.8-fold greater sensitivity to paclitaxel than the empty vector (CMV-6)-expressing SW626 cells (Fig. 2C and Table II). Expression of wild-type Akt did not affect the sensitivity to paclitaxel compared with the sensitivity of empty vector (CMV-6)-expressing control lines (data not shown). Thus, the sensitization to paclitaxel caused by pretreatment with either wortmannin or LY294002 or seen in the kinase-deficient Akt-expressing cells appeared to be because of interference with the activation of Akt.
Moreover, to determine whether ERK activation is necessary for cell survival signaling after paclitaxel-induced cell damage, the effect of paclitaxel treatment on the viability of SW626 cells pretreated with PD98059 was compared with the effect on the parental cells (Fig. 2A). The cells pretreated with PD98059 exhibited an IC 50 as low as 190 nM, indicating over 10.0-fold greater sensitivity to paclitaxel than the untreated cells ( Fig.  2A and Table I). Thus, the sensitization to paclitaxel caused by pretreatment with PD98059 appeared to be due to interference with the activation of ERK, as reported previously (57).
LY294002 and PD98059 Sensitize Both Paclitaxel-sensitive and -resistant Cells to Paclitaxel-We further examined whether blockade of the PI-3K-Akt and ERK cascades in ovarian cancer cells that are sensitive to paclitaxel has the same effect. A2780 and Caov-3 cells exhibited IC 50 values of 100 and 80 nM, respectively, indicating over 19.8-and 22.5-fold greater sensitivity to paclitaxel than SW626 cells, respectively ( Fig. 3C and Table III). Treatment of A2780 (Fig. 3, left panel) and Caov-3 (Fig. 3, right panel) cells with paclitaxel induced Akt activation (Fig. 3A, lane 2) and ERK activation (Fig. 3B, lane 2). Although pretreatment with LY294002 completely abolished the induction of Akt activity by paclitaxel (Fig. 3A, lane 4), pretreatment with LY294002 had no effect on the induction of ERK activity by paclitaxel (Fig. 3B, lane 3). In addition, although PD98059 completely abolished the induction of ERK activity by paclitaxel (Fig. 3B, lane 4), it had no effect on the induction of Akt activity by paclitaxel (Fig. 3A, lane 3), indicating the absence of cross-talk between the Akt and ERK cascades in the activation induced by paclitaxel in A2780 and Caov-3 cells, as in the case of SW626 cells (Fig. 1). A2780 and Caov-3 cells pretreated with LY294002 exhibited IC 50 values as low as 68 and 52 nM, respectively, indicating over 1.47-and 1.54-fold greater sensitivity to paclitaxel than the untreated cells, respectively ( Fig. 3C and Table III). A2780 and Caov-3 cells pretreated with PD98059 exhibited IC 50 values as low as 67 and 34 nM, respectively, indicating over 1.49-and 2.35-fold greater sensitivity to paclitaxel than the untreated cells, respectively ( Fig. 3C and Table III). Thus, blockade of each cascade in cells which are sensitive to paclitaxel also further sensitized the cells to paclitaxel.
Moreover, we examined whether blockade of the PI-3K-Akt and ERK cascades in A2780/PTX and Caov-3/PTX cells, both of which have acquired in vitro resistance to paclitaxel, has the same effect. A2780/PTX and Caov-3/PTX cells exhibited IC 50 values of 180 and 210 nM, respectively, indicating over 1.8-and 2.6-fold greater resistance to paclitaxel than the parental cells, respectively ( Fig. 3D and Table III). A2780/PTX and Caov-3/ PTX cells pretreated with LY294002 exhibited IC 50 values as low as 130 and 110 nM, respectively, indicating over 1.38-and 1.91-fold greater sensitivity to paclitaxel than the untreated cells, respectively ( Fig. 3D and Table III). A2780/PTX and Caov-3/PTX cells pretreated with PD98059 exhibited IC 50 values as low as 110 and 100 nM, respectively, indicating over 1.64-and 2.10-fold greater sensitivity to paclitaxel than the untreated cells, respectively ( Fig. 3D and Table III). Thus, blockade of either cascade in cells that have acquired in vitro resistance to paclitaxel also sensitized the cells to paclitaxel. These findings have broad implications for the potential clinical use of inhibitors of the PI-3K-Akt and ERK cascades for improving the response rate to paclitaxel in the treatment of  resistant tumors, an anticancer strategy similar to that proposed for MEK inhibitors (57). Phosphorylation of BAD-Recently, BAD was identified as an intersection point of pro-and anti-apoptotic regulatory cascades (37,38). BAD function is modulated by phosphorylation at two sites, Ser-112 and Ser-136 (39). The presence of two phosphorylation sites on BAD suggests that the simultaneous activation of different survival cascades may result in the con-  (lanes 2-4). Lysates were subsequently immunoprecipitated with immobilized anti-Akt antibody, and the kinase reaction was carried out in the presence of cold ATP and GSK-3␣ fusion protein, as described under "Experimental Procedures." After the reactions were stopped with Laemmli sample buffer, the samples were subjected to 12% SDS-PAGE followed by Western blotting with antiphospho-GSK-3␣/␤ antibody.  (37). In recent studies (37)(38)(39), the region surrounding Ser-136 in BAD was shown to conform to a consensus sequence for phosphorylation by Akt, and BAD was identified as a potential target of Akt, linking the PI-3K pathway directly to the apoptotic machinery. In addition, the promotion of cell survival by the Ras-MAPK signaling pathway by phosphorylation of BAD at Ser-112 (41)(42)(43)58) was recently reported. Therefore, we next examined the effect of paclitaxel on the phosphorylation of BAD at Ser-112 and Ser-136. Cells were transfected with pCDNA3-BAD and exposed to 1 M paclitaxel for 3 h. Cell lysates were immunoprecipitated with either anti-phospho-Ser-112 (Fig. 4A, upper panel) or anti-phospho-Ser-136 (Fig.  4B, upper panel) BAD antibody, followed by Western blotting with the same antibodies. Paclitaxel induced the phosphorylation of BAD at Ser-112 and Ser-136. Moreover, we confirmed that the total amount of BAD in each lysate was the same by Western blotting with anti-BAD antibody (Fig. 4, lower panel).
In accordance with reports showing that Akt phosphorylates BAD specifically at Ser-136 (37,38), paclitaxel-induced phosphorylation of BAD at Ser-136 was completely inhibited by wortmannin (Fig. 4B, lane 3) but was not completely inhibited by PD98059 (Fig. 4B, lane 4). In addition, paclitaxel-induced phosphorylation of BAD at Ser-136 was completely inhibited by the expression of a kinase-deficient Akt (AktK179M) (Fig. 4B,  lane 5). On the other hand, paclitaxel-induced phosphorylation of BAD at Ser-112 was not inhibited by wortmannin (Fig. 4A,  lane 3) but was completely inhibited by PD98059 (Fig. 4A, lane  4). Thus, paclitaxel induced the phosphorylation of BAD at Ser-112 via the ERK cascade and that at Ser-136 via the Akt cascade, as does cisplatin (58).
Interference with Phosphorylation of BAD at Ser-112 and Ser-136 Sensitizes SW626 Cells to Paclitaxel-To determine whether BAD phosphorylation is necessary for cell survival signaling after paclitaxel-induced cell damage, the effect of paclitaxel treatment on the viability of SW626 cells expressing mutant BAD constructs, in which both Ser-112 and Ser-136 were converted to alanine (BAD2SA) so that BAD could no longer be phosphorylated at these sites (37,58), was compared with the effect on the viability of an empty vector (pCDNA3)-expressing control line. We first confirmed the overexpression of the BAD protein products (Fig. 5A, lower panel) and the negative effects of the expression of BAD2SA on BAD phosphorylation at both Ser-112 (Fig. 5A, upper panel) and Ser-136 (Fig. 5A, middle panel). The BAD2SA-expressing SW626 cells exhibited an IC 50 as low as 90 nM, indicating over 20.0-fold greater sensitivity to paclitaxel than the empty vector-expressing SW626 cells ( Fig. 5B and Table II). Expression of wild-type BAD did not affect the sensitivity to paclitaxel compared with the sensitivity of cells expressing the empty vector (pCDNA3)-expressing control lines (data not shown). Thus, the sensitization to paclitaxel observed in BAD2SA-expressing cells appeared to be due to interference with the activation of BAD.
Negative Regulation of Raf-1 by Akt-Several studies have shown that paclitaxel-mediated apoptosis is mediated by activation of Raf-1 (25,45,46). The Hsp90 chaperone is required for maintenance of the Raf⅐Ras complex and for protecting Raf from degradation. An inhibitor of Hsp90, geldanamycin, disrupted the complex containing Raf, Ras, and Hsp90 and caused a marked decrease in the half-life of the Raf protein due to an increase of the rate of its degradation (59). To determine whether Raf-1 activation is necessary for paclitaxel-induced cell damage, the effect of paclitaxel treatment on the viability of SW626 cells pretreated with geldanamycin was compared with  panel of A and B), 250 g of protein from the lysate samples were resolved by 8% SDS-PAGE, followed by Western blotting with anti-BAD antibody. Experiments were repeated three times with essentially identical results. C, control. the effect on control cells not treated with geldanamycin. The cells pretreated with geldanamycin exhibited an IC 50 as low as 4200 nM, indicating greater resistance to paclitaxel than the untreated cells (Table I). This confirmed that the treatment with geldanamycin inhibited the paclitaxel-induced decrease of cell viability, as reported previously (25,45).
It was reported that the region surrounding Ser-259 in Raf-1 conforms to a consensus sequence for phosphorylation by Akt (49,50) and that the phosphorylation of Raf-1 by Akt inhibited the activation of Raf-1 (51). Therefore, we examined the effect of Akt on the regulation of Raf-1 in SW626 cells. To examine whether Akt and Raf-1 were physically associated, SW626 cells were stimulated with paclitaxel for 3 h, after which endogenous Raf-1 and Akt were immunoprecipitated (Fig. 6A). Endogenous Akt and endogenous Raf-1 were co-immunoprecipitated from SW626 cells that had been stimulated with paclitaxel. Next, we examined the effect of paclitaxel-induced Akt activation on the phosphorylation of Raf-1 at Ser-259 (Fig. 6B). Cell lysates were immunoprecipitated with anti-Raf-1 antibody, and the immunoprecipitates were subjected to Western blotting with anti-phospho-Raf-1 antibody. The treatment with paclitaxel for 3 h induced the phosphorylation of Raf-1 at Ser-259 (Fig. 6B, lane  2). Pretreatment with either of the PI-3K inhibitors, LY294002 (Fig. 6B, lane 3) or wortmannin (Fig. 6B, lane 4), inhibited paclitaxel-induced phosphorylation of Raf-1 at Ser-259, whereas pretreatment with PD98059 had no effect (Fig. 6B,  lane 5). Moreover, we confirmed that the total amount of Raf-1 in each lane was the same (Fig. 6B, lower panel). In addition, the expression of an activated Akt mutant that is constitutively targeted to the plasma membrane (HA-m⌬4 -129 Akt) (37,60) markedly induced the phosphorylation of Raf-1 at Ser-259 (Fig.  6D, ii). Moreover, we examined the effect of Akt on the activation of Raf-1. Cell lysates were immunoprecipitated with anti-Raf-1 antibody, and the immunoprecipitates were subjected to a coupled Raf-1 kinase assay. Whereas pretreatment with LY294002 inhibited the paclitaxel-induced Akt activation (Fig.  6C, iv) and the phosphorylation of Raf-1 at Ser-259 (Fig. 6C, ii), pretreatment with LY294002 increased the extent of activation of Raf-1 (Fig. 6C, i). Moreover, although expression of the activated Akt mutant induced Akt activation (Fig. 6D, iv) and the phosphorylation of Raf-1 at Ser-259 (Fig. 6D, ii), expression of the activated Akt mutant decreased the extent of activation of Raf-1 (Fig. 6D, i). We confirmed that the total amount of Raf-1 in each lane was the same (Fig. 6, C and D, iii). Thus, our results suggest that Akt antagonizes Raf-1 activity by direct interaction of Akt with Raf-1 and phosphorylation of Raf-1 at Ser-259. DISCUSSION The signaling pathway involving Raf, MEK (ERK kinase), and ERK functions downstream of the small guanine nucleotide-binding protein Ras and mediates several apparently contradictory cellular responses, such as proliferation, apoptosis, growth arrest, differentiation, and senescence, depending on the duration and strength of the external stimulus and on the cell type. For example, although prolonged activation of ERK by a GnRH agonist in ovarian cancer cells was reported to be involved in growth arrest (53), transient activation of ERK by cisplatin in ovarian cancer cells was reported to be involved in the resistance to cisplatin (i.e. proliferation) (52). It was reported recently (61) that prolonged activation of the ERK cascade induced by paclitaxel was not linked to activation of the cell death machinery. In the present report, we demonstrated that ERK is activated transiently by paclitaxel (Fig. 1B) and that interfering with the MEK/ERK cascade sensitizes SW626 cells to paclitaxel (Fig. 2). It was also reported more recently (57) that paclitaxel enhances the activation of the MEK/ERK pathway, which is expected to promote cell proliferation and survival. Another pathway that acts downstream of Ras involves PI-3 kinase and Akt and also regulates the cellular responses listed above, acting either synergistically with (62) or in opposition to (63) the Raf pathway. Overexpression of Akt was reported to confer resistance to paclitaxel (64). However, it was recently reported that paclitaxel induced apoptosis independently of Akt in human ovarian carcinoma cells that expressed wild-type p53 (65). In the present report, we showed that paclitaxel induced Akt activation transiently without cross-talk with the ERK cascade (Fig. 1A) and that interfering with the Akt cascade either by the addition of specific inhibitors wortmannin or LY294002 or by expression of dominant negative Akt sensitizes SW626 cells (which lacks functional p53) to paclitaxel (Fig. 2) as is also true in the case of cisplatin (58). Coordination of the two signaling pathways in a given cellular response may depend on the cell type or the stage of differentiation (66 -68). We recently reported (58) that the ERK and Akt signaling cascades converge at BAD to suppress the apoptotic effect of BAD in cisplatin-treated ovarian cancer  lanes 1 and 2), wild-type BAD (lanes 3 and 4), or mutant BAD construct (BAD2SA) in which both Ser-112 and Ser-136 were converted to alanine-expressing cells (lanes 5 and 6) grown in 100-mm dishes were treated with 1 M paclitaxel for 3 h (lanes 2, 4, and 6). For analysis of the level of BAD protein (bottom panel), the lysate samples were analyzed by electrophoresis on 8% SDS-polyacrylamide gels, followed by Western blotting with anti-BAD antibody. For analysis of the effects of the expressed BAD2SA on BAD phosphorylation, the lysate samples was resolved by 8% SDS-PAGE, followed by Western blotting with phospho-BAD (Ser-112) (upper panel) or phospho-BAD (Ser-136) (middle panel) antibody. B, cell viability was assessed in empty vector (pCDNA3)-or BAD2SA-expressing cells after treatment with the indicated concentrations of paclitaxel as described under "Experimental Procedures." C, control. cells. In the present study, we also demonstrated that paclitaxel induces the phosphorylation of BAD both at Ser-112 via the ERK cascade (Fig. 4A) and at Ser-136 via the Akt cascade (Fig. 4B). Moreover, interference with either of these cascades sensitizes SW626 cells to paclitaxel (Fig. 5B). It was recently shown (13) that paclitaxel-induced apoptosis of the SW626 ovarian cancer cell line is enhanced by stable BAX overexpression in a p53-independent manner. In addition, expression of HA-BAD in ovarian cancer cell lines was found to enhance significantly the cytotoxic effects of paclitaxel (69). Thus, the ERK and Akt signaling cascades converging at BAD are involved in the mechanisms of maintaining the cell viability following both cisplatin and paclitaxel treatment in ovarian cancer cells.
Although the expression of Raf-1 has no relationship with cisplatin (70), the level of paclitaxel-induced apoptosis seems to have some dependence on Raf-1 kinase activity. It was reported that phosphorylation by Akt inactivates the function of Akt substrates such as BAD (37), FKHR1 (71), and Raf-1 (51). The regions surrounding Ser-136 in BAD and Ser-259 in Raf-1 conform to a consensus sequence for phosphorylation by Akt (49,50). The present study also showed that paclitaxel induced the phosphorylation of BAD at Ser-136 (Fig. 4B) and Raf-1 at Ser-259 (Fig. 6B) in an Akt-dependent mechanism. Interference with the Akt-BAD cascade sensitized SW626 cells to paclitaxel (Figs. 2 and 5), suggesting that the phosphorylation of BAD at Ser-136 by paclitaxel inhibits the pro-apoptotic function of BAD. Akt reduced the level of activation of Raf-1, whose destabilization by geldanamycin increased the resistance of cells to paclitaxel (Table I) (Fig. 6, C and D), via direct interaction between Akt and Raf-1 (Fig. 6A) and the phosphorylation of Raf-1 at Ser-259 (Fig. 6B). Thus, Akt appears to play a central role in the mechanism of a protective response to paclitaxel.
Are there any differences among the ERK-BAD, Akt-BAD, and Akt-Raf-1 cascades with respect to the sensitivity to paclitaxel? We did not detect any differences in basal kinase activation or paclitaxel-induced kinase activation between SW626 cells, which are resistant to paclitaxel, and A2789 and Caov-3 cells, both of which are sensitive to paclitaxel (Figs. 1 and 3). In addition, blockade of the PI-3K-Akt and ERK cascades increased the sensitivity to paclitaxel, independent of the paclitaxel sensitivity of cells (Figs. 2 and 3). Extensive research has identified several potential mechanisms of paclitaxel-induced cell death; most prominent is the effect on Bcl-2 family members and p53. Several reports (45,(72)(73)(74)(75) have indicated that paclitaxel causes the phosphorylation and inactivation of Bcl-2 and its family members, whereas other studies (76 -78) have found that paclitaxel sensitivity varies with p53 status. SW626 cells show resistance to paclitaxel because they lack functional p53. The BAD protein is a newly described pro-apoptotic member of the Bcl-2 family which is capable of binding to both Bcl-XL and Bcl-2 and also of displacing BAX (79). Paclitaxelstimulated phosphorylations of c-Raf-1 and Bcl-2 are tightly coupled (45). Because interference with these cascades changed the sensitivity to paclitaxel (Fig. 5, Table I, and Table II), these cascades seem to be p53-independent mechanisms. In other words, these cascades might be involved in repair mechanisms of cells treated with paclitaxel, as in the case of cisplatin (52,58).
Raf-1 located at the plasma membrane phosphorylates ERK, which in turn phosphorylates Bcl-2 (45), thereby disrupting the association between Bcl-2 and the pro-apoptotic BAX protein (73), which then induces apoptosis. Prior exposure of human tumor cells to the drug geldanamycin has been reported to diminish concomitantly Raf-1 kinase activity and paclitaxel-induced apoptosis (45), as we showed in Table I. Thus, although Raf-1 is known to function upstream of ERK, the link between Raf-1 and ERK activation and paclitaxel-induced cell death might not be always straightforward. It is noteworthy that c-Raf-1 contains multiple potential serine and tyrosine phosphorylation sites (48,80). Phosphorylation of different residues has been shown to regulate differentially the enzymatic and biological activity of the protein (48,80). It was reported that at least some of the residues on c-Raf-1, which undergo phosphorylation in response to paclitaxel treatment, differ from those phosphorylated in response to growth factor stimulation, suggesting that the kinase(s) that are responsible for c-Raf-1 phosphorylation in response to paclitaxel treatment may be unique to this agent (46). In contrast, recent studies have shown that the Bcl-2 protein localizes Raf-1 at the mitochondrial membrane, where Raf-1 phosphorylates and inactivates the proapoptotic BAD protein (81). In addition, it was reported that Raf-1 kinase activity is a major determinant of paclitaxel resistance in human cervical tumor cells (82) and in ovarian cancer cells that harbor a mutated p53 protein (83). Thus, Raf-1 can act as both an agonist and antagonist of paclitaxelinduced apoptosis. It is possible that, depending on the subcellular localization of the phosphorylated Raf-1, the paclitaxelinduced elevation of Raf-1 kinase activity could lead to either high or low levels of apoptosis.
We were not able to clarify completely all of the mechanisms of resistance to paclitaxel in this work. One of these mechanisms might overlap the mechanism of resistance to cisplatin. There might also be specific mechanisms of resistance to paclitaxel. Recently, it was reported that Akt binds to Forkhead (71) and cAMP-response element-binding protein (41), which function in cascades known to be involved in transcription factordependent anti-apoptotic mechanisms. Further investigations will reveal whether these molecules are also involved in regulating cell viability following paclitaxel treatment. We are currently investigating these possibilities.