FoxO3a Transcriptional Regulation of Bim Controls Apoptosis in Paclitaxel-treated Breast Cancer Cell Lines*

Paclitaxel is used to treat breast cancers, but the mechanisms by which it induces apoptosis are poorly understood. Consequently, we have studied the role of the FoxO transcription factors in determining cellular response to paclitaxel. Western blotting revealed that in a panel of nine breast cancer cell lines expression of FoxO1a and FoxO3a correlated with the expression of the pro-apoptotic FoxO target Bim, which was associated with paclitaxel-induced apoptosis. In MCF-7 cells, which were paclitaxel-sensitive, the already high basal levels of FoxO3a and Bim protein increased dramatically after drug treatment, as did Bim mRNA, which correlated with apoptosis induction. This was not observed in MDA-231 cells, which expressed low levels of FoxOs and Bim. Gene reporter experiments demonstrated that in MCF-7 cells maximal induction of Bim promoter was dependent on a FoxO binding site, suggesting that FoxO3a is responsible for the transcriptional up-regulation of Bim. Gene silencing experiments showed that small interference RNA (siRNA) specific for FoxO3a reduced the levels of FoxO3a and Bim protein as well as inhibited apoptosis in paclitaxel-treated MCF-7 cells. Furthermore, siRNA specific for Bim reduced the levels of Bim protein and inhibited apoptosis in paclitaxel-treated MCF-7 cells. This is the first demonstration that up-regulation of FoxO3a by paclitaxel can result in increased levels of Bim mRNA and protein, which can be a direct cause of apoptosis in breast cancer cells.

Breast cancer is one of the most common malignancies affecting women in the western world and arises following the accumulation of a series of somatic changes that serve to increase the rate of cellular proliferation and/or reduce the levels of apoptosis. These changes are often found to involve deregulation of key signal transduction pathways. Signal transduction pathways within the cell transmit the extracellular signals to transcription factors, resulting in changes in gene expression. These changes in gene expression are often cell type-specific and lead to cell growth, differentiation, or apoptosis, thereby regulating normal tissue structure and function. One key signal transduction pathway highly conserved in eukaryotes is the phosphatidylinositol 3-kinase (PI3K) 1 pathway (1), which is stimulated by a number of growth factors, including insulin and insulin-like growth factor 1 (1). Once a receptor tyrosine kinase has become activated by binding a specific ligand, it becomes autophosphorylated and binds PI3K heterodimers either directly, or indirectly via insulin receptor substrate (IRS) adaptor proteins (1)(2)(3). The interaction between the p85 regulatory subunit of PI3K and the receptor brings the catalytic domain (p110 subunit) of PI3K into contact with the plasma membrane, where it is able to phosphorylate its targets. Alternatively, PI3K can be juxtaposed to the plasma membrane via interactions with activated RAS family members. PI3K phosphorylates phosphatidylinositol (4,5)diphosphate at the 3 position, resulting in the formation of phosphatidylinositol (3,4,5)-triphosphate. Phosphatidylinositol (3,4,5)-triphosphate binds to protein kinase B (PKB) also called AKT, a serine/threonine kinase, via its pleckstrin homology domain and recruits it to the inner surface of the cell membrane where it becomes activated by another pleckstrin homology domain-containing protein, PDK1 (1)(2)(3). PKB phosphorylates a number of key substrates, including glycogen synthase kinase 3␤ and mammalian target of rapamycin, which is involved in cell growth and translation, and the transcription factor NFB (1).
Genetic studies on the nematode Caenorhabditis elegans identified the DAF16 protein as a key downstream substrate of the insulin/PI3K/PKB signaling pathway, and showed that DAF16 is inactivated by PKB phosphorylation (4,5). The mammalian orthologues of the DAF16 protein are the FoxO class of the forkhead family of winged helix transcription factors, FoxO1a, FoxO3a, and FoxO4 (formally known as FKHR, FKHR-L1, and AFX, respectively) (4,5). In the absence of growth factor stimulation, FoxOs are localized in the nucleus, where they function as transcription factors. Upon stimulation of the PI3K signaling cascade, FoxOs become phosphorylated by PKB on highly conserved serine and threonine residues (6 -8). These phosphorylations result in impairment of DNA binding ability and increased binding affinity for the 14-3-3 protein (6 -8). Newly formed 14-3-3-FoxO complexes are then exported from the nucleus (9), thereby inhibiting FoxO-dependent transcription. One key FoxO target gene is the cyclin-dependent kinase inhibitor (CKI) p27 Kip1 (10,11), which serves to inhibit cell cycle progression by inhibiting the kinase activity of cyclin-CDK holoenzymes, thereby reducing cell proliferation. Another important target gene is Bim, a BH3 domain protein that is capable of inducing apoptosis (11)(12)(13). Consequently, activation of the PI3K pathway serves to repress FoxO-mediated growth arrest and apoptosis. However, regulation of FoxO target genes is multifactorial, and therefore other transcription factors and post-translation regulatory events will also influence the final level of protein expression. It has been demonstrated that some promoters lacking the FoxO consensus site are still capable of being activated by FoxOs. Furthermore, there is evidence that FoxOs can interact strongly with other transcription factors such as the estrogen receptor-␣ (14), suggesting that the FoxOs can also modulate transcription via interaction with other transcription factors (14). These results suggest that changes in gene expression resulting from the FoxO arm of the PI3K/PKB signaling pathway are likely to be cell type-dependent. Interestingly, overexpression of PKB (15) and inactivation of the PI3K/PKB pathway inhibitor, PTEN (phosphatase and tensin homologue deleted on chromosome TEN), are frequently observed in breast cancer (16,17), indicating a potential role for FoxOs in modulating both cell cycle and apoptosis during tumorigenesis and treatment.
Typically, treatment of breast cancer involves surgery and chemotherapy using drugs that target the estrogen receptor, such as tamoxifen (18). Although anti-estrogen therapy can be effective in treating estrogen-dependent tumors, some tumors exhibit intrinsic or acquired resistance to these modalities, and alternative treatments are needed (18). Of these, the taxanes are one of the most frequently used classes of drugs, which include paclitaxel and its derivatives (19,20). These drugs act by interacting with the cellular microtubules, and in particular, with the microtubules associated with the spindle apparatus during mitosis (19,20). Although the primary cellular targets of the taxanes are microtubules, our understanding of the signal transduction pathways involved in the induction of apoptosis by taxanes in breast cancer cells is not complete. Consequently, we have studied the role of the PI3K/PKB pathway, and chiefly the FoxO family of transcription factors, in mediating the apoptosis and survival of breast cancer cell lines treated with paclitaxel. In this report, we demonstrate that in a drug-sensitive cell line accumulation of hypophosphorylated FoxO3a is responsible for the up-regulation of the pro-apoptotic protein Bim, whereas in a less sensitive cell line there is a reduction in hypophosphorylated FoxO3a, and significantly less Bim expression and apoptosis. We also propose that the regulation of FoxO family members by the PI3K/PKB pathway may be of prognostic value in the prediction of paclitaxel sensitivity in breast cancer.
Transfections and Gene Reporter Assays-Based on the published mouse Bim promoter sequence (21), a 790-bp fragment of the human Bim promoter, containing a single inverse Forkhead box TAAACAC (nucleotide Ϫ266/Ϫ259 relative to the 3Јterminus of the promoter) was amplified from genomic DNA by PCR using the following primers: 5Ј-AAGCTTCCCGCCCTCACCCGGGA-3Ј and 5Ј-GAGCTCCAACAAACT-GCAGACC-3Ј, cloned in pGEM-T easy vector (Promega, Southampton, UK) and validated by sequencing. The human Bim promoter was then mutated to introduce a single G to C substitution in the core of the Forkhead box (GTAAACAC) by site-directed point mutagenesis using the following primers: 5Ј-TTACTCCGGTAAAGACGCCAGGGA-3Ј and 5Ј-TCCCTGGCGTCTTTACCGGAGTAA-3Ј. Subsequently, a double PCR was performed using the human Bim promoter primers described before. The pGL2-hBim and pGL2-mutant hBim plasmids were then created by cloning the respective Klenow-blunted EcoRI promoter fragment into the SmaI-linearized pGL2Basic vector (Promega, UK). MCF-7 and MDA-MB-231 cells were transfected using the calcium phosphate co-precipitation method as described previously (22). Briefly, calcium phosphate precipitates containing 1 g of wild type Bim promoter firefly-luciferase reporter plasmid (pGL2-hBim) or mutant plasmid (pGL2-mutant-hBim) together with 0.2 g of a Renilla luciferase transfection control (pRL-TK; Promega) were incubated overnight with subconfluent cell cultures in each well of a 24-well plate. The cells were then washed twice in PBS, treated with paclitaxel, and harvested for firefly/Renilla luciferase assays using the Dual-Luciferase Reporter Assay System (Promega).
Apoptosis Determination Using Annexin V Staining-To determine the extent of apoptosis, Annexin V staining was used according to the manufacturer's instructions (R&D Systems). Briefly, paclitaxel-treated cells were trypsinized and washed in PBS containing 2% bovine serum albumin, before being stained in fluorescein isothiocyanate-labeled annexin V and propidium iodide (R&D Systems). Cells were analyzed by flow cytometry using a FACScan (BD Biosciences), and data were processed using CellQuest software (BD Biosciences).
Cell Cycle Analysis-Cell cycle analysis was performed using propidium iodide staining as described previously (23). Briefly, cells were trypsinized, washed in PBS, then fixed in 90% ethanol. Fixed cells were then washed twice in PBS and stained in 50 M propidium iodide containing 5 g/ml DNase-free RNase for 1 h, then analyzed by flow cytometry using a FACScan (BD Biosciences) and analyzed using Cell Quest software (BD Biosciences).
Northern Blotting Analysis-Total RNA was isolated from treated cells using the RNeasy kit (Qiagen) and quantitated using UV spectrophotometry, and gene expression was analyzed by Northern blotting as detailed previously (22). Briefly, 40 g of RNA was size-fractionated using 1.5% (w/v) formaldehyde-agarose gel electrophoresis. Following electrophoresis, RNA was transferred onto a Hybond N ϩ nitrocellulose membrane (Amersham Biosciences) using capillary transfer Northern blotting as described previously (22). Bim, p27 Kip1 , and GAPDH mRNAs were detected using 32 P-labeled human cDNA probes described previously (24,25).
Real-time Quantitative RT-PCR-Total RNA was isolated as before and DNase I-treated. Equal amounts of total RNA (2 g) was reversetranscribed using the Superscript First-Strand Synthesis System for RT-PCR (Invitrogen), and the resulting first strand cDNA was diluted and used as template in the real-time quantitative-PCR analysis. All measurements were performed in triplicate. The mRNAs analyzed were Bim and GAPDH, which served as internal control and was used to normalize for variances in input cDNA. The following gene-specific primer pairs were designed using the ABI Primer Express software: GAPDH-sense (5Ј-ATTTGGTCGTATTGGGCGCCTGGTCACC-3Ј) and GAPDH-antisense (5Ј-GAAGATGGTGATGGGATTTC-3Ј); Bim-sense (5Ј-CACAAAACCCCAAGTCCTCCTT-3Ј) and Bim-antisense (5Ј-TTCA-GCCTGCCTCATGGAA-3Ј). Specificity of each primer was determined using NCBI BLAST module. Detection of Bim expression was per-formed with SYBR Green (Applied Biosystems) and an ABI PRISM 7700 Sequence Detection System (Applied Biosystems), using the relative standard curve method.
Gene Silencing with Small Interfering RNAs-The siRNA oligonucleotides were purchased from Dharmacon Research Inc. (Lafayette, CO). MCF-7 and MDA-MB-231 cells were cultured in six-well plates until 60% confluent. Cells in 2 ml of culture medium were transfected with 4 g of annealed oligonucleotides using LipofectAMINE (Invitrogen, UK) according to manufacturer's instructions. 24 h after transfection the cells were treated with 10 nM paclitaxel or Me 2 SO vehicle. After 48 h of treatment, both the suspension and the adherent cells were collected for either Western blot analysis, or Annexin V/propidium iodide staining. The FoxO3a siRNA sequence corresponds to the coding region 46 -64 relative to the first nucleotide of the start codon (Gen-Bank TM accession number BC020227) and does not match any other genomic sequence, except for that of the pseudogene FoxO3b in a Blast search from the NCBI website (www.ncbi.nlm.nih.gov). FoxO3a sense 5Ј-ACUCCGGGUCCAGCUCCAC(dTdT)-3Ј; FoxO3a antisense 5Ј-GUG-GAGCUGGACCCGGAGU(dTdT)-3Ј. The Bim siRNA sequence corresponds to the coding region 39 -57 relative to the first nucleotide of the start codon of the published the Bim EL cDNA (GenBank TM accession number NM_138621). This sequence is present in all known Bim mRNA splice forms, but does not match any other sequence, in the GenBank TM . Bim sense 5Ј-CAAUUGUCUACCUUCUCGG(dTdT)-3Ј; Bim antisense 5Ј-CCGAGAAGGUAGACAAUUG(dTdT)-3Ј; control sense 5Ј-UUCUCCGAACGUGUCACGU(dTdT)-3Ј; and control antisense 5Ј-ACGUGACACGUUCGGAGAA(dTdT)-3Ј.

Apoptosis Induced by Paclitaxel on a Panel of Breast Carcinoma Cell
Lines-To delineate the mechanisms involved in paclitaxel-induced apoptosis in breast cancer, a panel of breast cancer cell lines were treated with 10 nM paclitaxel for 48 h, and the level of apoptosis was measured using annexin V staining. The data shown in Fig. 1A clearly demonstrate that there is extensive heterogeneity in the extent of apoptosis induction in the panel of cell lines. The least paclitaxel-sensitive cell lines were MDA-MB-231 and T47D, which displayed an increase in apoptosis of less than 5%. Intermediate sensitivity (an increase of between 5 and 10%) was observed in the ZR-75-1 and CAL-51 cell lines. The most drug-sensitive cell lines were HMT3552, MCF-7, 734B, CAMA-1, and SKBR-7, which all exhibited increases in apoptosis of greater than 10% following paclitaxel treatment. Consequently, Western blotting experiments were performed to identify differences in expression of potential apoptotic regulators that could explain the heterogeneous apoptotic response.

The Protein Expression of FoxO Family Members and FoxO Targets Is Heterogeneous in a Panel of Breast Carcinoma Cell
Lines-It has been shown that the FoxO family of transcription factors can induce the pro-apoptotic BH3-only protein Bim in hematopoietic cells, thereby causing cell death (11,25). Furthermore, because the PI3K/PKB pathway is often deregulated in breast cancer (26,27), and the FoxO family of transcription factors is subject to inhibitory phosphorylation by PKB, we sought to examine the expression levels of FoxO3a and FoxO1a to determine whether its expression correlated with the levels of apoptosis induced by paclitaxel in a panel of breast cancer cell lines.
The expression pattern of total FoxO3a shows that expression is highest in MCF-7 and 734B and T47D, with intermediate levels of expression in HMT3522 and ZR-75-1 (Fig. 1B). The expression of FoxO3a is lowest in CAL-51, CAMA-1, MDA-MB-231, and SKBR-7. In the case of FoxO1a, the levels of total protein and the phosphorylated form are highest in MCF-7 and 734B. The expression levels in the remaining cell lines are much lower.
Western blotting experiments demonstrated that the level of the FoxO target p27 Kip1 was high in MCF-7, 734B, ZR-75-1, T47D, and CAL-51, with expression in HMT3552, CAMA-1, MDA-MB-231, and SKBR-7 being much lower. The expression of Bim also shows heterogeneity, its expression being high in HMT3552, MCF-7, 734B, and ZR-75-1, and very low in comparison with the remaining cell lines. The expression of ␤-tubulin was determined as a control for equal loading of the gels, and showed little variation. Taken together, these data suggest that in the MCF-7, 734B, T47D, and ZR-75-1, and to a lesser extent, the HMT3552 cell lines, there is a correlation between the expression of FoxO1a and -3a, and the expression of one or other of the known FoxO targets p27 Kip1 and Bim. What is clear from the data is that the high level of expression of Bim in HMT3552, MCF-7, and 734B correlates with the induction of apoptosis by paclitaxel. Conversely, the low levels of Bim expression in T47D and MDA-MB-231 also correlated with the lack of apoptosis induction. The exceptions to this are ZR-75-1, which expresses high levels of Bim, but does not undergo apoptosis readily, and SKBR-7 and CAMA-1, which display high levels of apoptosis despite low levels of Bim, suggesting other mechanisms are responsible for regulating apoptosis in these cells.
Paclitaxel Induces Cell Death in a Drug-sensitive Cell Line, but G 2 /M Arrest in a Less sensitive Cell Line-To investigate in more detail the role of FoxO family members and Bim in paclitaxel-induced apoptosis, we chose to perform further experiments in two cell lines: MCF-7, which expressed high levels of FoxOs and Bim and underwent apoptosis readily, and MDA-MB-231 cells, which expressed low levels of FoxOs and Bim and was refractory to paclitaxel-induced apoptosis. The cell-cycle phase distribution in these two cell lines was measured using flow cytometry of ethanol-fixed propidium iodide-stained cells after treatment with 10 nM paclitaxel. The data shown in These results would suggest that the MCF-7 cells arrest in G 2 /M phase before death. In contrast, the MDA-MB-231 cells accumulate in the G 2 /M phase for the first 24 h and then become polyploid but do not undergo extensive cell death.
Paclitaxel Induces Increased Protein Expression of FoxO3a, Bim, and p27 Kip1 in a Drug-sensitive Breast Cancer Cell Line-To study the role of FoxOs in determining the apoptotic response to paclitaxel in MCF-7 and MDA-MB-231 cells, we compared the changes in protein expression following treatment with paclitaxel at either 10 or 75 nM. Whole cell protein extracts were prepared 24, 48, and 72 h after drug treatment, which from the previous experiment were the times when apoptosis was initiated (24 h) and then was observed to be maximal (48 and 72 h). The protein expression was then analyzed by Western blotting. The data clearly demonstrated that in both cell lines that there was an increase in the levels of both phospho-and total FoxO1a 48 h after 10 nM paclitaxel treatment (Fig. 3A). Following treatment with 75 nM paclitaxel, the induction is still observable in the MDA-MB-231 cells, and is less pronounced in the MCF-7 cells, which underwent extensive cell death (data not shown). What is most striking is the observation that expression of both phosphorylated and total FoxO3a dramatically increased 48 h after 10 nM paclitaxel treatment in the MCF-7 cells but not in the MDA-MB-231 cells. Interestingly, this expression pattern is also observed in the 75 nM MCF-7 treatment group, but to a lesser extent. These data clearly demonstrate that FoxO3a is induced in the paclitaxelsensitive MCF-7, but to a much lesser extent in less sensitive MDA-MB-231. It is notable that, although the phosphorylated form of FoxO3a was detected in both MCF-7 and MDA-MB-231 cells, large increases in the protein levels of total FoxO3a were only observed in the MCF-7 cell line. This may reflect the fact that the phospho-FoxO3a antibody was more sensitive than the total FoxO3a antibody and suggests that high levels of hypophosphorylated, and consequently transcriptionally active, FoxO3a were accumulating in MCF-7 cells following paclitaxel treatment. Therefore, we analyzed the expression of the FoxO targets, p27 Kip1 and Bim. What is evident is that there was a large increase in both p27 Kip1 and Bim protein expression in the MCF-7 cell line 48 and 72 h after paclitaxel treatment, which paralleled that of FoxO3a, but not in MDA-MB-231.
These results strongly suggest that the increase in FoxO3a in the MCF-7 cells caused by paclitaxel treatment induces expression of both Bim and p27 Kip1 . To confirm this, parallel cell cultures were treated with paclitaxel, total RNA was isolated, and quantitative real-time PCR was performed to determine whether the increase in Bim expression was due to an increase in steady-state Bim mRNA levels. The data from cells revealed that following treatment with paclitaxel the steady-state levels of the three Bim transcripts of sizes 5.7, 3.8, and 1.4 kb were higher in the MCF-7 cell line than in the MDA-MB-231, which correlated with our observations regarding the protein expression in untreated cells (Fig. 4A). Furthermore, it is clear that the expression of Bim increased following treatment with paclitaxel in the MCF-7 cell line after 4 h and remained high until 72 h. In the MDA-MB-231 cells, however, the transcripts only became elevated after 72 h. Data from real-time quantitative PCR, which measured all the isoforms of Bim (Fig. 4B), also demonstrated that there was an increase in Bim mRNA and that the induction was greater in the MCF-7 than in the MDA-MB-231 cells, confirming the Northern blot result. These data demonstrate that the increase in Bim protein expression probably results from increased transcription.
Induction of Bim Transcription by Paclitaxel Cell Is Increased by FoxOs-From the previous data it is clear that paclitaxel-induced increase in Bim protein and mRNA was more pronounced in the MCF-7 than in MDA-MB-231 cells. Although the expression of FoxOs correlated with that of Bim, there was no direct evidence that FOXO family members regulated Bim expression in these cell lines. To delineate the role of FOXOs in transcriptional regulation of Bim transcription we used reporter genes containing firefly luciferase under the control of either the wild type Bim promoter or a mutant plasmid in which the FOXO consensus sequence had been altered to make it refractive to FOXO transactivation in transient transfections (data not show). From the data in Fig. 5, it is clear that in untreated cells the basal levels of luciferase activity in both the mutant and wild type promoters are higher in MCF-7 cells than in MDA-MB-231, which agrees with our previous observations (Fig. 5). It is noticeable that in untreated control cells the relative luciferase activity of the mutant reporter is higher than that of the wild type (Fig. 5). This may in part be due to repression of transcription caused by other transcription factors interacting at this site directly or indirectly. Treatment of cells with 10 nM paclitaxel resulted in an increase in the relative luciferase activity from both the mutant and the wild type promoter in the MCF-7 cell line, whereas the changes in levels of luciferase activity in the MDA-MB-231 cell lines were much smaller. This result in the MCF-7 cell line suggests that the activation of the Bim promoter is multifactorial, and other transcription factors may play a role in Bim activation. When a direct comparison is made regarding the relative induction of the Bim promoter in the MCF-7 cell line, the mutant reporter is only activated 1.5-fold by paclitaxel, whereas the wild type is activated nearly 4.2-fold, indicating that FoxOs play a key role in the regulation of Bim transcription (Fig. 5). In the MDA-MB-231 cell line there is a difference in the relative activation between the mutant and wild type Bim reporters, the levels of luciferase induction are so low as to suggest that, although FoxO family members may play a role in Bim transcriptional activation in these cells, this role is minor.
Silencing are strongly induced by paclitaxel in MCF-7 and not MDA-MB-231 cells, and that this effect is most likely due to the increased expression of FOXO3a, it is unclear as to whether the increase of FOXO3a and Bim are directly responsible for the induction of apoptosis in MCF-7 cells. To address this question, we performed gene silencing experiments. Recently, it has been demonstrated that the use of specific small inhibitory RNA (siRNA) duplexes are capable of reducing the expression of target genes to study their function. In these experiments, cells were transfected with siRNAs specific for FoxO3a, and then treated with paclitaxel. Because the siRNA, and thus the transfection reagent, must be present throughout the experiment to achieve a reduction in the expression of FoxO3a, control groups consisting of mock-transfected cells were also included. Fig. 6A shows the result of Western blotting experiments that clearly show that 10 nM paclitaxel treatment for 48 h results in an increase in FoxO3a as well as the protein products of the target genes Bim and p27 Kip1 in MCF-7 but not MDA-MB-231 cells. It is also clear that in untreated MCF-7 cells transfection with a FoxO3a-specific siRNA had reduced the protein levels of FoxO3a and, consequently, the levels of the targets Bim and p27 Kip1 . In the case of MCF-7 cells that had been transfected with FoxO3a-specific siRNA, although there was a clear increase in the levels of the FoxO targets Bim and p27 Kip1 , the increase failed to rise above the levels of those observed in the mock transfected, untreated MCF-7 cells. In the case of the MDA-MB-231 cells we failed to observe these effects. To confirm the specificity of the FoxO3a siRNA oligonucleotides, we also analyzed the expression of FoxO1a, which increased in response to paclitaxel in the MCF-7 cell line. However, the levels were not greatly affected by the siRNA, suggesting that the gene knockdown was specific for FoxO3a.
Although the Western blotting experiments demonstrated that the use of siRNA did result in the efficient knock down of FoxO3a, we performed further experiments to determine whether the reduction in Bim and p27 Kip1 levels resulting from FoxO3a siRNA treatment could inhibit apoptosis. When the cell cycle phase distribution of MCF-7 and MDA-MB-231 cells was analyzed in both untransfected and mock transfected cells, it was clear that paclitaxel treatment resulted in 20 and 23% of cells being in the sub-G 1 peak, which represents dead or dying cells (Fig. 6B). However, in the cells that received FoxO3a siRNA treatment, the levels were reduced to 11%. Interestingly, as seen before, in the MDA-MB-231 cells, the cells appeared to pass through mitosis and accumulate in a cohort with a DNA content, which was higher than tetraploid (i.e. 4N or mitotic). The presence of the transfection reagent appeared to inhibit this hyper-diploid accumulation, whether siRNA was Although it is clear that the sub-G 1 peak represents cells that have died, we also observed a number of MCF-7 cells accumulating in S-phase it maybe that cells accumulating in S-phase may have undergone cell death in G 2 /M as well. Additionally, the presence of cells in a sub-G 1 peak gives little clue as to the mechanism of cell death, i.e. apoptosis or necrosis. Consequently, we repeated the siRNA experiments and measured paclitaxel-induced apoptosis regardless of cell cycle phase distribution using annexin V staining. The data shown in Fig.  6C clearly show that there is an inhibition of paclitaxel-induced apoptosis in the MCF-7 cells (reducing from 23% in the control to 7% in the presence of FoxO3a-siRNA). In the MDA-MB-231 cell line, there was a negligible increase in apoptosis following paclitaxel treatment, and no inhibition following application of FoxO3a-siRNA.
Silencing of Bim by siRNA Results in a Reduction in the Paclitaxel-induced Apoptosis in MCF-7 Cells-Although the FoxO3a gene silencing experiments clearly showed that there was a significant reduction of FoxO3a, the FoxO3a target Bim, and apoptosis following paclitaxel treatment, there is no evidence that Bim is directly inducing apoptosis in the MCF-7 cell line following paclitaxel treatment. Thus, we performed further gene silencing experiments using an siRNA specific for Bim that targets all known isoforms. The Western blot shown in Fig. 7A clearly demonstrates that in non-treated MCF-7 cells, the transfection of an siRNA specific for Bim caused a dramatic decrease in the expression of Bim. As before, paclitaxel treatment of MCF-7 cells increased the level of Bim expression in the mock transfected cells, but in cells transfected with a Bim siRNA, there was no Bim expression. Again, to confirm the specificity of the siRNA oligonucleotides, we analyzed the expression of the other BH3 domain-containing proteins Bcl-2 and Bax. Although the levels of both Bax and Bcl2 decreased slightly in the MCF-7 cells following drug treatment, protein expression was not altered in either cell line by the gene silencing, suggesting the siRNA is specific for Bim. Further experiments demonstrated that siRNA specific for Bim caused a reduction in paclitaxel induced apoptosis in MCF-7 cells as determined by propidium iodide staining (Fig. 7B) and annexin V staining (Fig. 7C). This effect was negligible in MDA-MB-231 cells. These results clearly show a functional correlation between the levels of Bim expression and the extent of apoptosis induced by paclitaxel. DISCUSSION While the PI3K signaling pathway has been studied extensively in breast cancer, little is known about its role in the regulation of apoptosis in response to paclitaxel treatment. When we surveyed a panel of breast cancer cell lines we found that the protein expression of the downstream mediators of PI3K signaling, FoxO3a and FoxO1a, as well as their transcriptional targets Bim and p27 Kip1 was heterogeneous, but that cell lines that generally had high levels of FoxO family members and Bim underwent apoptosis in response to paclitaxel. The data presented here clearly demonstrate that in the paclitaxel- Also shown is the expression of GAPDH as a control for RNA loading and blotting. B, the expression of Bim RNA was also analyzed in parallel by real-time PCR, and normalized to the level of GAPDH. The data shown is a representative of three independent experiments that gave similar results sensitive MCF-7 cell line, drug treatment resulted in an increase in the levels of FoxO3a, and its targets Bim and p27 Kip1 . When this accumulation of FoxO3a was blocked using an siRNA approach, the paclitaxel-induced increases in FoxO3a, p27 Kip1 , and Bim were greatly reduced, as were the levels of drug-induced apoptosis. Additionally, when Bim itself was knocked down using a specific siRNA, apoptosis was again inhibited. We failed to observe these effects in a less sensitive breast cancer cell line, MDA-MB-231. This is the first report that shows that FoxO3a can induce the expression of Bim in breast cancer cells, or indeed that Bim itself plays a highly significant role in drug-induced apoptosis regulation. The data shown here clearly demonstrate that the expression of FoxO3a is induced in MCF-7 cells by paclitaxel treatment and that the expression of both Bim protein and mRNA also increased. The results also suggest that the increase in FoxO3a activity is a result of an increase in the levels of FoxO3a protein, rather than an increase the ratio of hypophosphorylated to hyper-phosphorylated FoxO3a. Further experiments demonstrated that a Bim promoter construct was less inducible by paclitaxel if the FoxO consensus site was mutated, and that the induction of Bim by paclitaxel was blocked by the introduction of a FoxO3a-specific siRNA. These data suggested that FoxO3a is functioning as a transcriptional regulator of Bim expression in the MCF-7 breast cancer cell line. It has been demonstrated previously in hematopoietic cells that following interleukin-3 withdrawal, FoxO family members can induce Bim, and thus cell death (12). FoxO3a and FoxO1a have also been shown to modulate survival in response to c-Kit ligand in hematopoietic progenitor cells (6). In these systems it is the presence of survival factors that activates PI3K/PKB signaling resulting in inhibitory phosphorylation of FoxO family members, and thus inhibition of Bim transcription. These examples demonstrate that FoxOs function in Bim induction and apoptosis. There is very little evidence that FoxO3a plays a role in breast cancer, however, EGF treatment has been shown to result in PKB phosphoryl-ation of FoxO3a and thus exclusion from the nucleus in MDA-MB-231 cells (28). Also, transforming growth factor ␤2 was shown to be a transcriptional target of FoxO3a in breast and pancreatic tumor cell lines (29). FoxO3a has also been shown to modulate apoptosis in response to nitric oxide suppression; however, this is mediated by Rho-associated kinase (ROCK) kinase independently of PKB (30). Consequently, the data shown here represent the first direct demonstration that FoxO3a can induce the expression of Bim and mediate apoptosis in breast cancer cells.
The results shown here demonstrate that there is a correlation between the expression of FoxO family members and the pro-apoptotic protein Bim, and the sensitivity to paclitaxel. With respect to the apoptosis data, it is clear that three of the most sensitive cell lines were HMT3552, MCF-7, and 734b, which all expressed high levels of Bim protein. Notably, MCF-7 and 734b have the highest expression levels of FoxO3a and FoxO1a. The cell lines CAMA-1 and SKBR-7 also displayed high paclitaxel sensitivity but expressed low levels of Bim. Of the least sensitive cell lines MDA-MB-231, T47D, and CAL-51 all expressed low levels of Bim, FoxO3a, and FoxO1a. Therefore, in some cell lines, Bim expression can correlate with paclitaxel sensitivity. However, in the panel of cell lines studied here ZR-75-1 only underwent moderate apoptosis despite expressing high levels of Bim, suggesting that there are other mechanisms of apoptosis regulation that are Bim-independent in breast cancer cell lines.
While the data shown here and elsewhere demonstrate that FoxO family members have been shown to induce Bim expression, there are other mechanisms of Bim regulation. One such pathway is via the AP-1 transcription factor pathway, because c-Jun has been shown to regulate Bim transcription (31). In healthy cells Bim binds to dynin in the cytoskeleton (31,32). The c-Jun N-terminal kinases components of the stress response pathway, have been shown to phosphorylate Bim resulting in the disassociation of Bim from dynin (33,34). Therefore, activation of the c-Jun N-terminal kinase arm of the stress response pathway is capable of not only increasing the expression of Bim, but also of inducing its dissociation from dynin. Upon the induction of apoptotic stimuli, Bim dissociates from dynin and binds to Bcl-2 and Bcl-XL via its BH3 domain, thereby sequestering Bcl-2 from the pro-apoptotic Bax and Bad, which are then free to initiate an apoptotic cascade (32). Consequently, the pro-apoptotic function of Bim requires Bax or Bad (35). The protein stability of Bim can also be regulated by phosphorylation (36). Thus low levels of Bax and Bad or high levels of Bcl-2 or Bcl-XL could result in inefficient induction of apoptosis by Bim, and this could in part explain the low sensitivity of ZR-75-1 cells to paclitaxel despite the high level of Bim expression. Taken together, these data suggest that Bim expression can influence the apoptotic response to paclitaxel in some cells (MCF-7, 734B, and HMT3552), and in some cases this is FoxO-dependent. Although the results shown here clearly demonstrate that increases in the transcriptional activity of FoxO3a up-regulate Bim expression, and that Bim in turn regulates apoptosis, the regulation of apoptosis by Bim is multifactorial and dependent on the levels of expression, sequestration of Bim by the cytoskeleton, the levels of Bcl-2 family members, and Bim protein stability. However, in the case of MCF-7 cells, it is clear that, following paclitaxel treatment, FoxO3a regulation is very important in mediating Bim-directed apoptosis. Paclitaxel binds to microtubules and cause the stabilization of spindle fibers at mitosis, thereby inducing mitotic arrest (20). It has been shown that the use of extremely high concentrations of paclitaxel used in some experiments to highlight biological effects in cell lines in vitro cause other effects that are not associated with mitotic interference (20). Indeed, Blagosklonny and Fojo (20) suggest that, although plasma levels of paclitaxel of 3-100 M are achievable in patients, these levels do not represent the levels seen in tumors and suggest that a physiological concentration lies in the range of 2-200 nM (20). The concentrations of paclitaxel used in these experiments lie well within this range and would support the assertion that our observations regarding the role of FoxOs and Bim in determining the sensitivity to paclitaxel are of clinical relevance.
It is notable that unlike MCF-7, the MDA-MB-231 cells accumulate in the G 2 /M phase and then become polyploid without undergoing apoptosis following paclitaxel treatment. This FIG. 6. The effect of FoxO3a-specific small interfering RNA on apoptosis and the expression of FoxO3a, Bim, and p27 Kip1 proteins following paclitaxel treatment. MCF-7 and MDA-MB-231 cells were either mock transfected or transfected with a FoxO3a-specific siRNA and then treated with 10 nM paclitaxel for 48 h. The expression of FoxO3a, Bim, and p27 Kip1 protein was analyzed by Western blotting. B, parallel cell cultures were also fixed in ethanol, and the DNA content was analyzed by flow cytometry following propidium iodide. The percentage of cells in each phase of the cell cycle (sub 2N, G 0 /G 1 , S, and G 2 /M) are indicated. C, additional cultures were also analyzed by flow cytometry following staining with annexin V to determine the extent of apoptosis as before. The data shown is a representative of three independent experiments that gave similar results.
suggests that normal S-M phase coupling during cell cycle progression failed in these cells following drug treatment. The reasons for this remain unclear; however, FoxO3a has been demonstrated to modulate the expression of several genes, including Gadd45, cyclin B1, and Polo-like kinase, which have a role in mediating the G 2 /M checkpoint in response to stress (37,38). It is therefore possible that the failure of S-M coupling in the paclitaxel-resistant cells, like MDA-MB-231, is due to their low levels of FoxO3a expression. The potential role of FoxO3a in S-M coupling is currently being investigated.
Our knowledge of the mechanisms by which apoptosis is induced by paclitaxel is limited, however, it has been shown that Bcl-2 and Bax are key effectors of paclitaxel-induced apoptosis (20,39). Several studies have shown that a reduction in Bcl-2 levels increased the sensitivity to paclitaxel in breast cancer (40,41) and multiple myeloma cells (42). The dependence of Bim-induced apoptosis on Bax and Bad, coupled with the role of Bcl-2 in paclitaxel-induced apoptosis would support our assertion that Bim does have a role in modulating the apoptotic response to taxanes. Consistent with our finding, a previous report showed that loss of Bim renders thymocytes resistant to taxol-induced apoptosis (52). Interestingly, the PI3K/PKB pathway also influences the sensitivity in breast cancer, because expression of constitutively active PKB results in decreased sensitivity to paclitaxel (43), an observation that reinforces the evidence for a role for FoxO family members in mediating paclitaxel sensitivity.
The role of FoxOs in the cellular response to paclitaxel shown here, and elsewhere, in the induction of Bim and p27 Kip1 would suggest that they may play a role in the stress response. Indeed, it has been shown that one gene target is GADD45, which functions in DNA repair (38). Furthermore, FoxO3a has been shown to function to induce G 2 /M checkpoint arrest in response to oxidative stress (44,45). The fact that FoxO3a can induce cell cycle arrest at G 1 and G 2 /M, as well as activate apoptosis, further suggests that it has a role in the stress response. The results obtained from MCF-7 cells support the hypothesis that FoxO3a may function as a modulator of the stress response.
In summary, the data presented here represent the first demonstration that paclitaxel treatment in breast cancer can induce the expression of FoxO proteins, which are then able to induce the expression of the pro-apoptotic protein Bim. Fur- FIG. 7. The effect of Bim-specific siRNA on the expression of Bim, following paclitaxel treatment and apoptosis. A, MCF-7 and MDA-MB-231 cells were either mock transfected or transfected with a Bim-specific siRNA, and then treated with 10 nM paclitaxel for 48 h. The expression of Bim, Bcl-2 (28 kDa), and Bax (23 kDa) protein was analyzed by Western blotting. B, parallel cell cultures were also fixed in ethanol, and the DNA content was analyzed by flow cytometry following propidium iodide. The percentage of cells in each phase of the cell cycle (sub 2N, G 0 /G 1 , S, and G 2 /M) are indicated. C, additional cultures were also analyzed by flow cytometry following staining with annexin V to determine the extent of apoptosis as before. The data shown is a representative of three independent experiments which gave similar results. thermore, in a survey of a panel of breast cancer cell lines we determined that sensitivity to paclitaxel in some cell lines correlated with Bim expression and that the most sensitive of all the cell lines tested had high expression of FoxO family members. This finding may indicate a prognostic role for both Bim and FoxO3a in predicting paclitaxel responsiveness in breast cancer. Interestingly, the expression of two FoxO targets, namely p27 Kip1 (11) and cyclin D1 (46), are already regarded as prognostic indicators in breast cancer. It has been shown that cyclin D1 is often overexpressed in early stage breast cancer and is often amplified (47). It has also known that p27 Kip1 loss can correlate with a poorer prognosis (48 -50). Whether FoxO family members actually regulate cyclin D1 and p27 Kip1 in breast cancer is not known. However, given this evidence and the data shown here, it is attractive to speculate that FoxO family members, and their transcriptional targets, play a significant role in breast cancer. A role for Bim in breast biology was demonstrated when it was shown that Bim expression in the breast rose dramatically after conception and remained high thereafter (51). Consequently, to delineate a role for FoxO3a and Bim in breast cancer, especially in determining the response to paclitaxel, we are currently investigating the expression of Bim and FoxO3a in patient samples before, during, and after paclitaxel treatment.