The mammalian target of rapamycin is a serine-threonine kinase that regulates cell cycle progression. Rapamycin and its analogues inhibit the mammalian target of rapamycin and are being actively investigated in clinical trials as novel targeted anticancer agents. Although cyclin D1 is down-regulated by rapamycin, the role of this down-regulation in rapamycin-mediated growth inhibition and the mechanism of cyclin D1 down-regulation are not well understood. Here, we show that overexpression of cyclin D1 partially overcomes rapamycin-induced cell cycle arrest and inhibition of anchorage-dependent growth in breast cancer cells. Rapamycin not only decreases endogenous cyclin D1 levels but also decreases the expression of transfected cyclin D1, suggesting that this is at least in part caused by accelerated proteolysis. Indeed, rapamycin decreases the half-life of cyclin D1 protein, and the rapamycin-induced decrease in cyclin D1 levels is partially abrogated by proteasome inhibitor N-acetyl-leucyl-leucyl-norleucinal. Rapamycin treatment leads to an increase in the kinase activity of glycogen synthase kinase 3β (GSK3β), a known regulator of cyclin D1 proteolysis. Rapamycin-induced down-regulation of cyclin D1 is inhibited by the GSK3β inhibitors lithium chloride, SB216763, and SB415286. Rapamycin-induced G1 arrest is abrogated by nonspecific GSK3β inhibitor lithium chloride but not by selective inhibitor SB216763, suggesting that GSK3β is not essential for rapamycin-mediated G1 arrest. However, rapamycin inhibits cell growth significantly more in GSK3β wild-type cells than in GSK3β-null cells, suggesting that GSK3β enhances rapamycin-mediated growth inhibition. In addition, rapamycin enhances paclitaxel-induced apoptosis through the mitochondrial death pathway; this is inhibited by selective GSK3β inhibitors SB216763 and SB415286. Furthermore, rapamycin significantly enhances paclitaxel-induced cytotoxicity in GSK3β wild-type but not in GSK3β-null cells, suggesting a critical role for GSK3β in rapamycin-mediated paclitaxel-sensitization. Taken together, these results show that GSK3β plays an important role in rapamycin-mediated cell cycle regulation and chemosensitivity and thus significantly potentiates the antitumor effects of rapamycin.
Rapamycin and its analogues are being actively investigated in clinical trials as novel targeted anticancer agents. The mammalian target of rapamycin (mTOR) is a serine-threonine kinase that regulates cell cycle progression. The two best-studied targets of mTOR, eukaryotic initiation factor 4E-binding protein 1 and ribosomal p70 S6 kinase-1, are thought to modulate translation; thus, rapamycin is thought to alter the translation of mRNA involved in control of the cell cycle. However, how rapamycin blocks cell growth and proliferation is not well understood.
Prior studies have suggested that cyclin D1 is a key target of mTOR (1–4). Cyclin D1 is overexpressed in a variety of tumor types and is proposed to contribute to cancer development. Cyclin D1 plays a critical role in G1 progression by activating cyclin-dependent kinases 4 and 6, leading to phosphorylation of tumor suppressor pRb, with depression of E2F-mediated transcription (5). Independent of cyclin-dependent kinase activity, cyclin D1 also modulates other transcription factors, such as estrogen and androgen receptors, signal transducers and activators of transcription 3, and PPARγ (6, 7). Recently, it has been reported that transcription factor CCAAT/enhancer-binding protein β is involved in regulating genes affected by cyclin D1 overexpression (8). Although the exact role of cyclin D1 overexpression in human tumors is controversial, there are several lines of evidence indicating that cyclin D1 plays a crucial role in mammary gland carcinogenesis. First, cyclin D1 is overexpressed in ductal carcinoma in situ and invasive ductal breast carcinoma and cyclin D1 overexpression is associated with a poorer prognosis (9, 10). Second, mammary gland–targeted cyclin D1 overexpression in mouse mammary tumor virus-cyclin D1 transgenic mice leads to mammary hyperplasia and development of carcinoma, suggesting that cyclin D1 at least is a weak oncogene (11). Third, mice lacking cyclin D1 are resistant to mammary carcinogenesis by the H-ras and HER-2/neu oncogenes, suggesting that cyclin D1 expression is critical to H-ras and HER-2/neu-mediated carcinogenesis (12). Finally, deregulation of cyclin D1 is implicated as the central pathway involved in chemical carcinogenesis models of breast cancer (13). Therefore, cyclin D1 may represent an important downstream target of signaling pathways that have a role in mammary carcinogenesis, making the effect of rapamycin on cyclin D1 of particular interest.
Cyclin D1 expression is regulated at multiple levels. Cyclin D1 transcription is up-regulated by mitogen stimulation through extracellular signal-regulated kinase 1/2 and 5, c-Jun NH2-terminal kinase, signal transducers and activators of transcription 5, nuclear factor-κB, and β-catenin and down-regulated by stress-activated kinase p38 (14–16). Cyclin D1 mRNA stability is regulated by a sequence in its 3′ untranslated region; its stability is increased by the phosphatidylinositol 3-kinase (PI3K) pathway and is decreased by prostaglandin A (14, 17). Cyclin D1 mRNA translation is thought to be controlled by the PI3K/Akt pathway and by mTOR signaling (18). Cyclin D1 protein stability is regulated by glycogen synthase kinase 3β (GSK3β) and p38, which phosphorylate the protein, triggering ubiquitination (19, 20). Cyclin D1 mRNA translation is often called the critical mode of cyclin D1 regulation by rapamycin. In recent work, Gera et al. have shown that rapamycin leads to a 5-fold decrease in cyclin D1 translation in cell lines that have high levels of Akt activity (3). However, in NIH3T3 cells, rapamycin was also found to decrease cyclin D1 mRNA and protein stability through unknown mechanisms (21). Therefore, the mechanism by which rapamycin down-regulates cyclin D1 expression needs further study.
In our previous work, we found that rapamycin leads to down-regulation of cyclin D1 levels in rapamycin-sensitive breast cancer cell lines MCF-7 and MDA-MB-468 but not in rapamycin-resistant cell lines MDA-MB-231, MDA-MB-435, and NCI/ADR-RES, suggesting that rapamycin-mediated cyclin D1 down-regulation may be critical to its growth-inhibitory effect (22). In this study, we found that cyclin D1 plays an important role in rapamycin-mediated cell cycle arrest and growth inhibition. We show that rapamycin mediates cyclin D1 down-regulation in part by accelerated proteolysis. Furthermore, we show that rapamycin activates GSK3β and that GSK3β activity is critical for rapamycin-mediated cyclin D1 down-regulation. We show that although GSK3β is not essential for rapamycin-mediated growth inhibition, it significantly enhances the growth-inhibitory effect of rapamycin. Further, rapamycin-mediated chemosensitization is also attenuated in GSK3β-null cells compared with GSK3β wild-type cells. Thus, we show that GSK3β plays a significant role in the antitumor effects of rapamycin.
Materials and Methods
Cell Cultures. Human breast cancer cell lines MCF-7 and MDA-MB-468 were obtained from the American Type Culture Collection (Manassas, VA). GSK3β+/+ and GSK3β−/− mouse embryo fibroblasts have been described previously (23, 24). Cells were cultured in DMEM/F-12 supplemented with 10% fetal bovine serum, 2 mmol/L glutamine, and 1% penicillin-streptomycin at 37°C and humidified 5% CO2.
Materials. Rapamycin and cycloheximide was purchased from A.G. Scientific, Inc. (San Diego, CA). N-acetyl-leucyl-leucyl-norleucinal (ALLN); GSK3β inhibitors lithium chloride (LiCl), SB216763, and SB415286; and antibodies against β-actin were purchased from Sigma Chemical Co. (St. Louis, MO). Antibodies against cyclin D1, p21, p27, Rb, phospho-Rb (Ser780), and phospho-GSK3β (Ser9), cytochrome c, caspase-9, caspase-3, caspase-7, poly(ADP-ribose)polymerase (PARP), Bad, Bad (Ser112), Bak, Bax, Bcl-xL, Bcl-2 (Ser70), survivin, and X-linked inhibitor of apoptosis were purchased from Cell Signaling Technology (Beverly, MA). Antibodies against cyclin A, cyclin E, GSK3β, and green fluorescent protein (GFP) were purchased from BD Biosciences PharMingen (San Jose, CA). Effectene transfection reagent was purchased from Qiagen, Inc. (Valencia, CA). Phospho-GSK3 Y216 antibody was purchased from Upstate Biotechnology (Lake Placid, NY). The GFP-cyclin D1 plasmid was a kind gift of Dr. Li-Huo Su (previously at University of Texas M.D. Anderson Cancer Center, currently at Cancer Cell). The pcDNA3.HA-tagged cyclin D1 and the pcDNA3.T286A cyclin D1 constructs were kind gifts of Dr. Doris Germain (University of Melbourne, Melbourne, Victoria, Australia).
Colony Formation Assay. Transfected cells were washed, trypsinized and, resuspended in DMEM. Cells were counted and plated in 100 mm dishes in triplicate, and cultured in medium containing 500 μg/mL G418 with or without 100 nmol/L rapamycin. The medium was changed twice a week for 3 weeks, and then colonies were stained with the crystal violet. Colonies of >120 um in diameter (20 cells) were counted using a microscope.
Cell Cycle Analysis. Cells were trypsinized and washed with PBS, and cells used in GFP experiments were fixed with 0.5% paraformaldehyde for 10 minutes. These cells were washed with PBS and then fixed with 75% ethanol overnight at 4°C. Following this, the cells were then washed twice with PBS and resuspended in hypotonic propidium iodide solution (10 μg propidium iodide, 10 μg RNase A, 0.5% Tween 20 in 1 mL PBS) for 0.5 hour at 37°C and kept in the dark at 4°C before analysis. Cell cycle distribution was determined with a FACScan flow cytometer and Cell Quest software (Becton Dickinson, San Jose, CA).
Western Blot Analysis. Cultured cells were washed with cold PBS and lysed in radioimmunoprecipitation buffer [20 mmol/L Tris (pH 7.5), 150 mmol/L NaCl, 5 mmol/L EDTA, 1% NP40, 1 mmol/L Na3VO4, 1 mmol/L phenylmethylsulfonyl fluoride, 50 mmol/L NaF] supplemented with complete protease inhibitors on ice. To extract cytosolic proteins for detection of cytochrome c release, 4 × 106 cells were harvested and washed twice with ice-cold PBS and resuspended in 300 μL ice-cold buffer [20 mmol/L HEPES-KOH (pH 7.0), 10 mmol/L KCl, 1.5 mmol/L MgCl2, 1 mmol/L EDTA, 1 mmol/L EGTA, 1 mmol/L DTT, 250 mmol/L sucrose, 1 μg/mL leupeptin and pepstatin, 2 μg/mL aprotinin]. After incubation on ice for 15 minutes, cells were homogenized with a Dounce homogenizer (B pestle per 25 strokes) and centrifuged at 1,000 × g for 10 minutes. The supernatants were centrifuged at 14,000 × g for 15 minutes in a microcentrifuge to pellet membranes, including mitochondria. The resulting supernatants were used as cytosolic extracts. Cell lysates or cytosolic extracts were separated by SDS-PAGE and transferred to a 0.2 μm polyvinylidene difluoride membrane (Bio-Rad Laboratories, Hercules, CA). Membranes were blocked with 5% nonfat dry milk in TBST (TBS with 0.1% Tween 20) and immunoblotted with antibodies. The immunoblots were visualized by an enhanced chemiluminescence detection system (Amersham Life Sciences, Arlington Heights, IL).
Kinase Assay. Cells treated with rapamycin (100 nmol/L) for 0, 3, or 6 hours or with the PI3K inhibitor LY294004 (10 or 20 μmol/L) for 1 hour were lysed in the NP40 immunoprecipitation lysis buffer (KPL, Gaithersburg, MD). Cell lysates (200 μg protein) were precleared for 1 hour at 4°C with 50 μL (50%) protein G-agarose beads (KPL). The precleared protein lysates were incubated with 3 μL rabbit anti-GSK3β antibody (Cell Signaling Technology) overnight at 4°C with gentle agitation. The immunoprecipitates were incubated with 50 μL (50%) protein G-agarose for 2 hours at 4°C with gentle agitation. The immobilized immunocomplexes were washed once with the NP40 lysis buffer (500 μL) and twice with kinase buffer (500 μL) [20 mmol/L Tris (pH 7.5), 5 mmol/L MgCl2, 1 mmol/L DTT]. The immunoprecipitates were equally aliquoted into two tubes for the Western blotting and the kinase assay, respectively.
In vitro kinase assays were carried out by mixing the beads with 30 μL kinase buffer [20 mmol/L Tris (pH 7.5), 5 mmol/L MgCl2, 1 mmol/L DTT, ATP (Sigma Chemical) 250 μmol/L, 1.4 μCi [γ-32P]ATP (MP Biomedicals, Inc., Irvine, CA), 0.1 μg/μL recombinant τ protein (Panvera, Madison, WI)] and subsequently incubated at 30°C for 20 minutes. The reaction was stopped by adding 6 μL of 6× Laemmli sample buffer and denaturing at 98°C for 5 minutes. Proteins were fractionated in 10% SDS-PAGE gels, vacuum dried, and exposed to a phosphoscreen. The phosphorimage was captured using PhosphorImager (Molecular Dynamics, Sunnyvale, CA) and followed by densitometry using Scion Image (Scion, Inc., Frederick, MD). Relative kinase activities were based on densities of the untreated samples. The efficiency of GSK3β immunoprecipitation was determined by immunodetection for GSK3β. A 30 μL of 1× Laemmli sample buffer was mixed with the beads and denatured at 98°C for 5 minutes. Proteins were fractionated in 10% SDS-PAGE, transferred to Immunoblot polyvinylidene difluoride membrane (Bio-Rad Laboratories), and underwent immunodetection using the anti-GSK3β antisera at 1:2,000.
Cell Growth Assays. For 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assays, cells were plated into 96-well, flat-bottomed plates at 2 × 103 to 4 × 103 cells/100 μL/well, and following the overnight incubation, triplicate wells were treated with varying concentration of rapamycin for 4 days. Relative percentage of metabolically active cells relative to untreated controls were then determined based on the mitochondrial conversion of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide to formazine. The results were assessed in a 96-well format plate reader by measuring the absorbance at a wavelength of 540 nm (A540 nm). The rates of DNA synthesis were determined by the percentage of cells showing [3H]thymidine incorporation into DNA. In brief, after the cells were treated with rapamycin in the same manner as in the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay, 0.5 μCi [3H]thymidine was added to each well and the cells were incubated for an additional 16 hours before being harvested. The incorporation of [3H]thymidine was measured by liquid scintillation counting. The rates of DNA synthesis in the treated cells were compared with the rates seen for control cells not treated with rapamycin in the same culture plate. Direct cell counts were done by seeding 10,000 cells/well and after 24 hours culturing the cells in the absence or presence of rapamycin. At the specified intervals, the cells were trypsinized and the viable cells were counted with a hemocytometer.
Annexin V Labeling. Apoptosis was determined using the ApoAlert Annexin V apoptosis kit (Clontech, Palo Alto, CA) according to the manufacturer's protocol. In brief, cells were trypsinized and rinsed in binding buffer. They were then resuspended in 200 μL binding buffer to which 5 μL Annexin V (20 μg/mL in Tris-NaCl buffer) and 10 μL propidium iodide (1 μg/mL) were added. Cells were incubated for 10 to 15 minutes at ambient temperature in the dark and then analyzed by flow cytometry.
Trypan Blue Exclusion Assay. A trypan blue exclusion study was done after treating cells with the signal transduction inhibitors. After 48 hours, cells were trypsinized and washed in PBS. Cells were mixed thoroughly and cell suspension (20 μL) was added to 20 μL of 0.08% trypan blue dye solution. At least 500 cells were microscopically counted in the hemocytometer. Cell death was expressed as the percentage of cells staining blue.
Statistical Analysis. Experiments were independently done three or more times. Two-tailed Student's t test or linear regression analysis was used to analyze growth inhibition assays. P < 0.05 was considered statistically significant.
Cyclin D1 Overexpression Partially Overcomes Rapamycin-Mediated Cell Cycle Arrest and Growth Inhibition. In previous work, we found that rapamycin-sensitive cell lines treated with 100 nmol/L rapamycin for 4 days showed a decline in cyclin D1 levels, whereas cell lines resistant to the growth-inhibitory effects of rapamycin did not (22). As rapamycin is known to mediate cell cycle arrest as one of its early effects, we evaluated the time course of rapamycin-mediated cyclin D1 down-regulation in two rapamycin-sensitive cell lines, MCF-7 and MDA-MB-468. We found that cyclin D1 levels started to decrease by 6 hours of rapamycin treatment (Fig. 1A). Rapamycin treatment also led to a decrease in Rb phosphorylation (Fig. 1B). We then evaluated the effects of rapamycin on a panel of cell cycle regulatory molecules. MCF-7 and MDA-MB-468 cells were maintained in 100 nmol/L rapamycin, and cell cycle analysis was done with flow cytometry after different durations of treatment (Fig. 1C). In both cell lines, the population of cells in G1 phase increased, whereas the population of cells in the S and G2 phases decreased after treatment with rapamycin (Fig. 1C). The expression of cyclins D1, A, and E and cell cycle inhibitors p21 and p27 were evaluated with Western blot analysis at the same time points. Cyclin D1 levels were found to decrease in both cell lines with rapamycin treatment, with variable changes in cyclin A levels and no change in cyclin E levels. Rapamycin did not increase levels of cell cycle inhibitors p21 and p27 in MCF-7 and MDA-MB-468 cells; rather, their levels were decreased within 3 days of treatment. Based on these data, we conclude that rapamycin induces G1 cell cycle arrest, and this is associated with a decline in cyclin D1 levels.
To determine the role of cyclin D1 down-regulation in the effects of rapamycin on breast cancer cells, we tested whether the forced overexpression of cyclin D1 is able to rescue cells from rapamycin-mediated cell cycle arrest and growth inhibition. We transfected MCF-7 cells with either a plasmid encoding for a GFP-tagged cyclin D1 protein (GFP-cyclin D1) or a plasmid encoding for GFP only. Cells were incubated in the absence or presence of 100 nmol/L rapamycin; then, cell cycle analysis was done on the GFP-positive cells (Fig. 1D,, GFP-positive cells, left) and on the entire cell population as a control (Fig. 1D,, total cells, right). Rapamycin induced G1 arrest in the cells transfected with GFP protein and in the MCF-7 cell population. In comparison, the effect of rapamycin on the cell cycle was less prominent in the cells transfected with GFP-cyclin D1 (Fig. 1D , GFP-positive cells, left).
To examine the effect of cyclin D1 on the inhibition of cell growth and proliferation mediated by rapamycin, we determined the effect of cyclin D1 overexpression on colony formation ability in the absence or presence of rapamycin. We transfected MCF-7 and MDA-MB-468 cells with either GFP control or GFP-cyclin D1. Cells were incubated with or without rapamycin for 3 weeks under G418 selection. Results in Fig. 1E show that rapamycin leads to a significant decrease in colony formation in both cell lines. Rapamycin treatment led to a 13.6-fold reduction in colony formation in control MCF-7 cells and a 4-fold reduction in cyclin D1–transfected MCF-7 cells. Rapamycin inhibited colony formation by the control cells 3.4-fold more than by cyclin D1–transfected cells (P < 0.05). Further, rapamycin led to a 3.7-fold reduction in colony formation in control MDA-MB-468 cells and a 1.7-fold reduction in cyclin D1–transfected MDA-MB-468 cells. Rapamycin inhibited colony formation by control cells 2.2-fold more than by cyclin D1–transfected cells (P < 0.05). Taken together, these results show that cyclin D1 overexpression partially overcomes rapamycin-induced cell cycle arrest and growth inhibition. This suggests that the rapamycin-mediated cyclin D1 down-regulation observed plays a critical role in the growth-inhibitory effects of rapamycin.
Rapamycin Accelerates the Proteolysis of Cyclin D1. To better delineate the mechanism of rapamycin-mediated cyclin D1 down-regulation, we tested the effect of rapamycin treatment on exogenously overexpressed cyclin D1 compared with endogenous cyclin D1. The transcription of the exogenous GFP-cyclin D1 construct is driven by a cytomegalovirus promoter; thus, it should not be under cyclin D1–specific transcriptional regulation. Further, the construct lacks the cyclin D1 5′ and 3′ untranslated regions that are proposed to be critical for cyclin D1 translational regulation and mRNA stability, respectively. MCF-7 cells were transfected with GFP-cyclin D1, GFP alone, or both plasmids together. Cells were cultured in the absence or presence of 100 nmol/L rapamycin, and cyclin D1 levels were evaluated with Western blotting (Fig. 2A). Rapamycin treatment did not affect exogenous GFP levels. In contrast, rapamycin reduced levels of both exogenous and endogenous cyclin D1 protein. Similarly, we observed that rapamycin decreased the expression of hemagglutinin-tagged cyclin D1 transfected in MDA-MB-468 cells (data not shown). Quantification of the effects in MCF-7 cells (Fig. 2A , right) shows that rapamycin reduced endogenous cyclin D1 levels more than exogenous cyclin D1 levels (70% versus 38%). The decrease in expression of exogenous cyclin D1 protein in this experiment suggests that rapamycin-mediated cyclin D1 down-regulation may be at least in part regulated at the level of protein stability. However, the fact that the expression of endogenous cyclin D1 was decreased more than that of the exogenous cyclin D1, which lacked sequence elements regulating transcriptional, translational, and mRNA stability, suggests that in addition to protein stability other mechanisms are likely at play in rapamycin-mediated down-regulation of the endogenous cyclin D1 gene.
To determine whether rapamycin affects cyclin D1 protein stability, we evaluated the effect of rapamycin on the half-life of cyclin D1 protein. MCF-7 and MDA-MB-468 cells were pretreated with rapamycin and then treated with an inhibitor of protein synthesis, 100 μg/mL cycloheximide, for different time intervals. The level of cyclin D1 protein in rapamycin-treated MCF-7 cells (Fig. 2B) and MDA-MB-468 cells (Fig. 2C) declined faster over time than it did in rapamycin-treated cells. In the average of three independent experiments, the half-life of cyclin D1 protein in untreated and rapamycin-treated MCF-7 cells was found to be 48 and 27 minutes, respectively, demonstrating a 44% decrease in cyclin D1 half-life on rapamycin treatment. The half-life of cyclin D1 protein in untreated and rapamycin-treated MDA-MB-468 cells was 44 and 36 minutes, respectively, demonstrating a more subtle, but reproducible, decrease in the cyclin D1 half-life on rapamycin treatment in three independent experiments. These results illustrate that rapamycin decreases the half-life of cyclin D1 protein and suggest that the degradation of cyclin D1 protein induced by rapamycin contributes to the decline of cyclin D1 levels after rapamycin treatment.
To further show the role of proteolysis in rapamycin-mediated cyclin D1 down-regulation, we carried out studies with the proteasome inhibitor ALLN. We incubated cells without inhibitors, with 100 nmol/L rapamycin alone, with ALLN alone, or with ALLN and rapamycin for 12 hours. As shown in Fig. 2D and E, ALLN inhibited the rapamycin-induced decrease of cyclin D1 levels. These results taken together show that rapamycin reduces the cyclin D1 protein levels at least in part by accelerating proteasome-dependent proteolysis.
Rapamycin Activates GSK3β. Because cyclin D1 is known to undergo proteasome-dependent proteolysis in a GSK3β-dependent fashion (19), we evaluated the effect of rapamycin treatment on GSK3β kinase activity. The kinase activity of GSK3β was evaluated in MCF-7 cells at baseline and after incubation with 100 nmol/L rapamycin (Fig. 3A). Rapamycin treatment was found to enhance GSK3β kinase activity by 110% and 65% at 3 and 6 hours after treatment, respectively. Next, GSK3β kinase activity was determined in MDA-MB-468 cells at baseline and after incubation with 100 nmol/L rapamycin. GSK3β kinase activity increased by 136% and 107% after 3 and 6 hours of rapamycin treatment, respectively. Alternatively, as GSK3β is inhibited by Akt (25), the cells were incubated with PI3K inhibitor LY294002 (Fig. 3B). Treatment with 10 or 20 μmol/L LY294002 increased GSK3β kinase activity by 73% and 163%, respectively. In three independent experiments, rapamycin induced a statistically significant increase in GSK3β kinase activity in MCF-7 and MDA-MB-468 cells (P < 0.001; Fig. 3C). Of note, treatment with the DMSO vehicle alone was not associated with GSK3β activation or cyclin D1 down-regulation (data not shown).
It is known that Ser9 phosphorylation of GSK3β decreases GSK3β activity, whereas Tyr216 phosphorylation is required for GSK3β activity (26, 27). Interestingly, although activation of GSK3β by LY294002 was associated with Ser9 dephosphorylation as expected, rapamycin-mediated activation was not (Fig. 3B). Further, no significant alteration in phosphorylation of GSK3β Tyr216 was observed in either cell line with rapamycin treatment (Fig. 3D). Our results suggest that rapamycin activates GSK3β through an as yet unknown mechanism.
GSK3β Inhibitors LiCl, SB216763, and SB415286 Inhibit Rapamycin-Mediated Down-Regulation of Cyclin D1. To investigate whether GSK3β is involved in the rapamycin-induced degradation of cyclin D1, we examined the effect of LiCl, a known inhibitor of GSK3β, on rapamycin-induced cyclin D1 down-regulation (28). We incubated cells with or without 100 nmol/L rapamycin in the presence or absence of 20 mmol/L LiCl. As shown in Fig. 4A, LiCl almost completely inhibited the decrease of cyclin D1 levels induced by rapamycin. We then tested the effect of two specific inhibitors of GSK3β, SB216763 and SB415286, at 30 and 10 μmol/L, respectively (29). As shown in Fig. 4B, SB216763 and SB415286 both inhibited the rapamycin-induced decline in cyclin D1 levels. These results suggest that rapamycin treatment activates GSK3β, which induces proteolysis of the cyclin D1 protein.
Cyclin D1 T286A Mutant Is Resistant to Rapamycin-Induced Down-Regulation. Because cyclin D1 proteolysis is thought to be mediated by phosphorylation of its Thr286 residue by GSK3β (21), we evaluated whether a T286A cyclin D1 mutant that is resistant to GSK3β-induced phosphorylation and degradation is also resistant to rapamycin. MDA-MB-468 cells were transiently transfected with pcDNA3 vector control or the same vector encoding for the T286A cyclin D1 mutant. Twenty-four hours later, the transfected cells were incubated in the absence or presence of 100 nmol/L rapamycin for another 48 hours. Western blot analysis showed a significant reduction in cyclin D1 levels in the control vector-transfected cells on rapamycin treatment, with a much less dramatic reduction in cyclin D1 levels in the T286A cyclin D1–transfected cells (Fig. 4C). These results suggest that GSK3β-induced T286 phosphorylation is critical in rapamycin-mediated cyclin D1 down-regulation.
GSK3β Wild-type Cells Are More Sensitive to Rapamycin-Mediated Growth Inhibition Than GSK3β-Null Cells. As GSK3β contributes to rapamycin-mediated cyclin D1 down-regulation, we hypothesized that GSK3β plays a significant role in rapamycin-mediated cell cycle arrest and growth inhibition. We first tested the effect of GSK3β inhibitors LiCl and SB216763 on rapamycin-mediated cell cycle arrest in MCF-7 cells. In three independent experiments, we found that coincubation with 20 mmol/L LiCl inhibited the G1 arrest induced after 24 hours of rapamycin treatment (Fig. 5A). SB216763 has been shown previously to inhibit GSK3β activity by 96% at 10 μmol/L (29); however, SB216763 was not able to inhibit rapamycin-mediated cell cycle arrest at 30 μmol/L (Fig. 5B) or at 10, 50, and 100 μmol/L (data not shown). These data suggest that although GSK3β activity plays a role in rapamycin-mediated cyclin D1 down-regulation, GSK3β is not essential for rapamycin-mediated cell cycle arrest. These data also suggest that LiCl may be inhibiting targets in addition to GSK3β to abrogate the effect of rapamycin on the cell cycle.
We then tested the role of GSK3β in rapamycin-mediated cell growth by comparing the growth-inhibitory effects of rapamycin in GSK3β wild-type (GSK3β+/+) and GSK3β-null (GSK3β−/−) fibroblasts (Fig. 5C; ref. 24). We first evaluated the effect of rapamycin on cell proliferation by direct cell counts (Fig. 5D). GSK3β wild-type and GSK3β-null cells were cultured in the absence or presence of 10 or 100 nmol/L rapamycin on triplicate plates. Three days after rapamycin treatment, compared with untreated GSK3β wild-type cells, a significant decline was seen in the number of GSK3β wild-type cells treated with 10 nmol/L rapamycin (10.7 × 104 versus 2.4 × 104; P = 0.0009) and 100 nmol/L rapamycin (10.7 × 104 versus 1.8 × 104; P = 0.0006). However, there was no statistically significant decline in the number GSK3β-null cells treated with rapamycin 3 days after treatment. By 6 days of rapamycin treatment, there was a significant decrease in the number GSK3β-null cells treated with rapamycin compared with untreated GSK3β-null cells, but there was a significantly greater inhibition in direct cell counts in GSK3β wild-type cells compared with GSK3β-null cells on treatment with both 10 nmol/L rapamycin (P = 0.0014) and 100 nmol/L rapamycin (P = 0.0010). These results were reproducible in three independent experiments.
Next, GSK3β wild-type and GSK3β-null cells were treated with increasing doses of rapamycin (1-1,000 nmol/L), and the metabolically active cells were determined based on the mitochondrial conversion of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide to formazine, assessed 4 days after treatment (Fig. 5E). As assessed by linear regression analysis, rapamycin led to a dose-dependent growth inhibition in both cell lines (P < 0.0001). However, at all four rapamycin concentrations tested, rapamycin caused a greater inhibition in GSK3β wild-type cells compared with GSK3β-null cells (P < 0.0001 for all). In addition, the data show that the effect of GSK3β on rapamycin-mediated growth inhibition increases as the rapamycin dose level increases (P < 0.0001).
We also determined the effect of rapamycin on GSK3β wild-type and GSK3β-null cell proliferation by culturing cells in the absence or presence of rapamycin for 4 days and then measuring DNA synthesis by [3H]thymidine incorporation (Fig. 5F). Rapamycin treatment decreased [3H]thymidine incorporation in both GSK3β wild-type and GSK3β-null cells (P < 0.05). However, treatment with 10 or 100 nmol/L rapamycin led to a significantly greater decline in DNA synthesis in GSK3β wild-type cells compared with GSK3β-null cells (P = 0.0011 and 0.0035, respectively). These results were reproducible in three independent experiments.
We further tested the effect of GSK3β on rapamycin on anchorage-dependent growth in the absence and presence of rapamycin long-term. GSK3β wild-type and GSK3β-null cells were trypsinized and plated in 100 mm Petri dishes (1 × 104 cells per dish), and after cell adherence (24 hours), cells were cultured in the absence or presence of 10 nmol/L rapamycin. Three weeks later, the cells were stained with crystal violet. On crystal violet assay, rapamycin completely inhibited anchorage-dependent growth in GSK3β wild-type cells but not in GSK3β-null cells (Fig. 5G). These results were reproducible in two independent experiments. These data taken together show that GSK3β activity is not essential for rapamycin-mediated cell cycle arrest; however, GSK3β-mediated mechanisms significantly enhance rapamycin-mediated growth inhibition.
Rapamycin Enhances Paclitaxel-Mediated Apoptosis in a GSK3β-Dependent Fashion. As GSK3β has been shown previously to be critical in regulation of cell survival (30), we next investigated whether GSK3β plays a role in rapamycin-mediated chemosensitivity. In our previous work, we observed additive interactions when rapamycin was given synchronously with doxorubicin or gemcitabine, and synergistic interactions when rapamycin was given synchronously with carboplatin, vinorelbine, or paclitaxel (31). As we observed the greatest synergy between rapamycin and paclitaxel, we explored the molecular mechanism behind this finding. MDA-MB-468 cells were treated with 0.01 μg/mL paclitaxel, 100 nmol/L rapamycin, or with the combination, and cells were harvested after 48 hours to determine apoptosis as determined by Annexin V labeling. The level of Annexin V labeling was much higher after treatment with the combination of rapamycin and paclitaxel than after treatment with each agent alone (Fig. 6A). After 24 hours, cells treated with the combination of rapamycin and paclitaxel, but not with either agent alone, showed cytoplasmic cytochrome c release and cleavage of caspase-9, caspase-3, caspase-7, and PARP (Fig. 6B). This suggests that the cytotoxicity of the combination of rapamycin and paclitaxel is mediated through the mitochondrial death pathway. As rapamycin has been reported to decrease the level of the antiapoptotic protein survivin (32, 33), we hypothesized that rapamycin-induced changes in expression of apoptotic and proapoptotic molecules may play a role in rapamycin-mediated chemosensitivity. We therefore determined the expression of a panel of proapoptotic and antiapoptotic molecules after 24 hours of treatment with paclitaxel, rapamycin, or the combination (Fig. 6C). We found no dramatic changes in the levels of Bad, phospho-Bad (Ser112), Bak, Bax, Bcl-xL, Bcl-2, Bcl-2 (Ser70), survivin, or X-linked inhibitor of apoptosis with the combination therapy. Thus, our data show that rapamycin enhances paclitaxel-induced cytotoxicity at least in part through the activation of the mitochondrial death pathway. However, this does not seem to be due to a rapamycin-mediated alteration in the levels of the proapoptotic and antiapoptotic molecules examined.
We next evaluated the role of GSK3β in rapamycin-enhanced cytotoxicity. We first evaluated the effect of GSK3β inhibitors on the cytotoxicity of rapamycin in combination with paclitaxel. After MDA-MB-468 cells were treated with 100 nmol/L rapamycin and 0.01 μg/mL paclitaxel for 24 hours, we evaluated apoptosis as shown by PARP cleavage. Coincubation of cells with rapamycin and paclitaxel along with GSK3β inhibitors SB216763 (20 μmol/L) or SB415286 (10 μmol/L) completely inhibited PARP cleavage (Fig. 6D). Coincubation with 10 mmol/L LiCl, however, did not inhibit PARP cleavage. When we evaluated apoptosis at 48 hours with Annexin V labeling, we showed that treatment with rapamycin in addition to paclitaxel significantly enhanced apoptosis compared with treatment with paclitaxel alone (7% versus 37%; P = 0.0003; Fig. 6E). Coincubation with GSK3β inhibitors SB216763 or SB415286 decreased the amount of apoptosis seen with rapamycin and paclitaxel (P = 0.0036 and 0.0012, respectively). Interestingly, in contrast, coincubation with LiCl did not decrease apoptosis but rather was associated with an increase in apoptosis. Following the same treatment schedule, when the percentage of nonviable cells was determined with a trypan blue dye exclusion assay, we found that coincubation with GSK3β inhibitors SB216763 or SB415286, but not with LiCl, decreased the percentage of nonviable cells observed with rapamycin and paclitaxel treatment (Fig. 6F). Of note, the treatments with SB216763 or SB415286 and especially LiCl alone led to an increase in cell death. Thus, inhibition of GSK3β itself may lead to cell death; this observation is consistent with the hypothesis that maintenance of baseline GSK3β activity is actually required for cell survival (30). In our experiments, inhibition of GSK3β decreases the rapamycin-enhanced paclitaxel cytotoxicity when GSK3β inhibitors SB216763 or SB415286 were used, suggesting that GSK3β activity may play a role in rapamycin-mediated chemosensitization. Interestingly, LiCl did not decrease the cell death induced by the combination of rapamycin and paclitaxel; this may be because treatment with LiCl alone induces more cell death or may be due to the differences in the specificity of LiCl compared with SB216763 or SB415286.
To further investigate the role of GSK3β in rapamycin-mediated chemosensitization, we evaluated the effect of rapamycin on paclitaxel-induced apoptosis in GSK3β wild-type and GSK3β-null cells. GSK3β wild-type and GSK3β-null cells were treated for 48 hours with 100 nmol/L rapamycin alone, 0.01 μg/mL paclitaxel alone, and combination. Annexin V–positive cells were determined by fluorescence-activated cell sorting (FACS) analysis (Fig. 6G). Rapamycin significantly enhanced paclitaxel-induced apoptosis in GSK3β wild-type cells (paclitaxel alone 6% versus paclitaxel and rapamycin 41%; P = 0.0128). In contrast, rapamycin did not significantly enhance paclitaxel-induced apoptosis in GSK3β-null cells. Furthermore, the combination of rapamycin and paclitaxel lead to cytoplasmic cytochrome c release, caspase-9, and PARP cleavage in GSK3β wild-type cells but not in GSK3β-null cells (Fig. 6H). These results show that the GSK3β activity is critical to rapamycin-mediated paclitaxel sensitization.
Rapamycin inhibits cell growth and proliferation and is currently in clinical trials as an anticancer agent. In the present study, we showed that rapamycin causes G1 cell cycle arrest in breast cancer cells, with a decrease in cyclin D1 levels. We showed that rapamycin mediates cyclin D1 down-regulation in part by accelerated proteolysis. We showed that rapamycin activates GSK3β and down-regulates cyclin D1 levels in a GSK3β-dependent fashion. Further, we showed that both rapamycin-dependent growth inhibition and rapamycin-mediated chemosensitization are attenuated in GSK3β-null cells compared with GSK3β wild-type cells. Taken together, these data show that GSK3β plays a significant role in rapamycin-mediated anticancer effects, including cell cycle regulation, growth inhibition, and chemosensitization.
We have reported recently that rapamycin leads to a reduction in cyclin D1 levels in rapamycin-sensitive MCF-7 and MDA-MB-468 cells but not in rapamycin-resistant MDA-MB-231, MDA-MB-435, and NCI/ADR-RES cell lines (22). Our findings are consistent with the results of Gera et al., who found that cyclin D1 is down-regulated in two rapamycin-sensitive cell lines with activated Akt but not in their isogenic counterparts that have less Akt activity and are rapamycin resistant (3). In this study, we show that cyclin D1 overexpression partially overcomes rapamycin-mediated cell cycle arrest and growth inhibition. Similar results obtained in glioblastoma cells were reported recently by Gera et al. (3). Rapamycin was also reported to inhibit the expression of cyclin D1 in hepatocytes, and transfection with cyclin D1 was found to overcome rapamycin-mediated cell cycle arrest in rat hepatocytes (2). These results taken together suggest that cyclin D1 is a key mediator of cell growth and proliferation downstream of mTOR and that down-regulation of cyclin D1 is critical to rapamycin-induced growth inhibition.
However, cyclin D1 is not down-regulated by rapamycin in all experimental systems. For example, cyclin D1 levels were not found to be altered by rapamycin analogue CCI-779 in multiple myeloma cells, which showed a growth-inhibitory response (34). Further, rapamycin treatment of the Eker rat model of tuberous sclerosis complex renal tumors led to a significant tumor response, without altering cyclin D1 levels (35). These results show that rapamycin does not uniformly lead to a decrease in cyclin D1 levels in all cell types. The effect of rapamycin on cyclin D1 may vary with the activity of different signaling pathways in each cell. For example, in the Eker rat kidney tumor model where Akt was found not to be phosphorylated (35), GSK3β may already be in an activated state due to lack of negative regulation by the PI3K/Akt pathway; thus, rapamycin may not lead to significant cyclin D1 proteolysis. Further, our results show that overexpression of cyclin D1 only partially overcomes the effect of rapamycin on cell cycle and growth. This may be not only because the exogenous cyclin D1 is down-regulated by rapamycin but also because cyclin D1 is not the only mediator of the downstream effects of rapamycin.
Rapamycin seems to down-regulate cyclin D1 expression at many levels. Rapamycin has been found to decrease cyclin D1 mRNA levels in U87 glioma cells and LAPC prostate cancer cells transfected with constitutively active Akt (3). Rapamycin treatment is associated with dephosphorylation of 4E-binding protein 1 (22, 31); the resulting decrease in eukaryotic initiation factor 4E availability may alter cyclin D1 mRNA export, as eukaryotic initiation factor 4E has been shown to facilitate the nuclear-cytoplasmic export of cyclin D1 mRNA (36, 37). Recently, Gera et al. showed that rapamycin decreases cyclin D1 translation 5-fold in U87MG glioblastoma cells and LAPC prostate cancer cells transfected with constitutively active Akt (3). In this study, we showed that rapamycin also down-regulates cyclin D1 by enhancing its proteolysis. Our results are consistent with those of Hashemolhosseini et al. who had reported previously that rapamycin decreases cyclin D1 protein stability in NIH3T3 cells (21). Our finding that the expression of endogenous cyclin D1 was decreased 70% by rapamycin treatment, whereas the expression of exogenous cyclin D1 that lacked transcriptional, translational, and mRNA stability elements decreased by 38% suggests that alterations in protein stability account for only a component of the effect of rapamycin on cyclin D1 levels in the cell lines we tested. Interestingly, inhibition of GSK3β dramatically inhibits rapamycin-mediated cyclin D1 down-regulation. Therefore, GSK3β may play a role not only in the proteolysis of cyclin D1 protein but also in other mechanisms of rapamycin-induced cyclin D1 down-regulation.
The role of mTOR signaling in regulation of glycogen synthase and GSK3 activity has been controversial to date. We show here that rapamycin treatment is associated with activation of GSK3β. In contrast, several reports have found no effect of rapamycin on GSK3 in other cell types, including Chinese hamster ovary cells, human myoblasts, rat hepatocytes, and rat epididymal fat cells (25, 38–42). In L6 muscle cells, rapamycin failed to block GSK3 inactivation in response to insulin but inhibited GSK3 inactivation in response to amino acids, suggesting that rapamycin may regulate GSK3 in a stimulus-dependent manner (43). The differences observed between the studies may also be related to differences between the cell lines and lineages studied, the activation status of the signaling pathways, and the baseline GSK3β activity of cells or to the time points studied. Further, most of these studies have evaluated the ability of rapamycin to block the inhibition of GSK3 by stimuli, such as insulin, rather than evaluating the effect of rapamycin treatment alone in the presence of serum and nutrients as was done in our study. Our results show that rapamycin activates GSK3β and suggest that GSK3β is a previously unrecognized critical downstream effector of the antitumor effects of rapamycin.
The mechanism of GSK3β activation by rapamycin observed in breast cancer cells is not known at this time. One possibility is that mTOR directly regulates one of the kinases that phosphorylates GSK3β. The target of mTOR, S6 kinase-1, has been shown previously to phosphorylate GSK3β in vitro (26); however, we did not observe a significant dephosphorylation of the in vitro GSK3β phosphorylation site Ser9 of S6 kinase-1 at the time points in which we observed kinase activation. Although rapamycin also did not affect the GSK3β Ser9 phosphorylation in multiple myeloma cells and baby hamster kidney cells (44, 45), it was reported to inhibit GSK3β Ser9 phosphorylation induced by opioids in a rat cardiac ischemia/reperfusion model (46). Thus, we cannot exclude the possibility that rapamycin affects GSK3β Ser9 phosphorylation in different models. Alternatively, GSK3β may be regulated by another mTOR-regulated kinase that has not been identified. Another mechanism through which rapamycin may regulate GSK3β is by activating a phosphatase that dephosphorylates GSK3β, such as protein phosphatase 2A (47). A third possible mechanism of GSK3β regulation by rapamycin may be indirect. In this scenario, a third molecule involved in GSK3β regulation may be regulated by inhibition of the mTOR signaling pathway at the translational or transcriptional level. Determination of the mechanism of GSK3β regulation by rapamycin may lead to identification of novel pathways or additional regulatory elements for GSK3β. Therefore, these possibilities deserve further study.
In our work, rapamycin enhances paclitaxel-induced cytotoxicity in GSK3β wild-type cells but not in GSK3β-null cells, suggesting that GSK3β is critical to rapamycin-mediated chemosensitization. It is unknown at this time which of the many putative substrates of GSK3β (48) mediates or potentiates paclitaxel-induced apoptosis. Enhanced GSK3β activity has been reported to mediate hypoxia-induced apoptosis via activation of the mitochondrial death pathway and has been attributed in part to alterations in glucose transport and metabolism (49). GSK3β has also been found to play a role in regulating survival downstream of PI3K/Akt signaling (50). In this context, GSK3β has been shown to control cell survival upstream of mitochondrial cytochrome c release as a result of phosphorylation of translation eukaryotic initiation factor 2B and global regulation of protein synthesis (50). Thus, eukaryotic initiation factor 2B may also be involved in rapamycin-mediated chemosensitization, and if so, this mechanism is likely to sensitize to a broad range of cytotoxic stimuli. Alternatively, the chemosensitizing effect of GSK3β may be mediated through its other phosphorylation targets, such as microtubule-associated proteins Tau and 1B (48); if so, the role of GSK3β may be more important for microtubule-targeted agents, such as paclitaxel, and thus more specific. Further studies to determine the mechanism of rapamycin-mediated chemosensitization are ongoing to help identify patients who will benefit the most from combination therapy with mTOR inhibitors.
Alterations in GSK3β has been associated with a variety of pathologic conditions, including diabetes, Alzheimer's disease, and bipolar disorder (28, 51, 52). Strikingly, in the phase I clinical trial of mTOR inhibitor CCI-779 (Wyeth, Collegeville, PA), higher doses induced euphoria followed by melancholy, mimicking bipolar disorder in three patients with no previous history of psychiatric disorders (53). Our finding that rapamycin activates GSK3β may be able to shed light on the etiology of this unexpected toxicity.
We also show that LiCl interferes with rapamycin-mediated cell cycle arrest, and SB216763 and SB415286 interfere with rapamycin-mediated paclitaxel sensitization. Although LiCl is an ATP noncompetitive inhibitor of GSK3β activity that has been widely used to evaluate the effects of GSK3β, it is known to inhibit other targets, such as casein kinase-2, p38 kinase, mitogen-activated protein kinase-2, polyphosphate 1-phosphatase, and inositol monophosphatase (29). SB216763 and SB415286 are ATP competitive compounds that are considered selective inhibitors of GSK3β (29); however, they may potentially have other targets. Thus, we cannot definitively determine whether our findings with these small molecule inhibitors are directly attributable to the inhibition of GSK3β. However, currently, LiCl is widely used for the treatment of bipolar disease, and GSK3β is also being considered as a therapeutic target for diabetes, stroke, and Alzheimer's disease (54). Our results emphasize that these GSK3β inhibitors may have previously unrecognized drug interactions that may decrease the antitumor activity of mTOR inhibitors. Further study is needed to determine the clinical relevance of these interactions.
Note: J. Dong and J. Peng contributed equally to this work.
Grant Support: Goodwin Foundation, NIH grant 1K08-CA-91895-01, 1 R01 CA112199-01, University of Texas M.D. Anderson Cancer Center Physician-Scientist Program (F. Meric-Bernstam), NIH grant 5T32CA09599 (Department of Surgical Oncology), and NIH Cancer Support grant P30-CA-16672 (University of Texas M.D. Anderson Cancer Center).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Ying Yang, Mark F. Munsel, and Kristine Broglia (Department of Biostatistics) for assistance with data analysis, Anne Packer for editorial assistance, and Marlen Banda for assistance with article preparation.