Insulin-like growth factor I triggers nuclear accumulation of cyclin D1 in MCF-7S breast cancer cells.

Stimulation of the breast cancer-derived MCF-7S cell line with insulin-like growth factor I (IGF-I; 20 ng/ml) leads to enhanced expression of cyclin D1, hyperphosphorylation of pRb, DNA synthesis, and cell division. 17beta-Estradiol (E(2); 10(-9) m) is not able to stimulate proliferation of MCF-7S cells, although addition of E(2) to serum-starved cells does result in induction of cyclin D1. However, in combination with submitogenic amounts of IGF-I (2 ng/ml), E(2) induces cell proliferation. We have previously shown that the synergistic action of E(2) and IGF-I emanates from the ability of both hormones to induce cyclin D1 expression and that IGF-I action is required to induce activity of the cyclin D1-CDK4 complex, which triggers cell cycle progression. Here, we show that IGF-I (but not E(2)) is able to induce nuclear accumulation of cyclin D1 by a phosphatidylinositol 3-kinase-dependent mechanism. Nuclear accumulation of cyclin D1 and cell cycle progression were also observed when LiCl, a known inhibitor of GSK3beta, was added to E(2)-stimulated cells. Thus, inhibition of GSK3beta activity appears to trigger nuclear accumulation of cyclin D1 and cell cycle progression. This notion was confirmed by overexpression of constitutively active GSK3beta, which blocks IGF-I-induced nuclear accumulation of cyclin D1 as well as S phase transition.

The importance of estrogen action in the regulation of breast cancer cell proliferation has been demonstrated both in vivo and in vitro (1). Epidemiological studies have shown that estrogen exposure is a causal factor in the onset of breast cancer (2). Clinical studies in which the effects of pharmacological agents inhibiting the synthesis or action of estrogen have been studied have demonstrated the fundamental importance of estrogen-induced mitogenesis in breast cancer (3). From recent research, it has become clear that estrogens act in concert with other mitogens in the regulation of breast epithelial cell proliferation. In particular, synergistic effects on proliferation have been described with members of the insulin-like growth factor (IGF) 1 family. The IGFs play im-portant roles in the normal regulation of cell proliferation, differentiation, and apoptosis. Breast tumors have been shown to have defects in the regulation and function of the IGF ligands, binding proteins, and receptors (4,5), and clinical trials targeting IGF signaling have been very successful in treating breast carcinoma (5,6).
Originally, it was considered that IGF-I and 17␤-estradiol (E 2 ) manifest their mitogenic actions through separate pathways, but a growing body of evidence suggests that the IGF-Iand E 2 -mediated signaling pathways are intertwined. We have reported previously that quiescent MCF-7S breast tumor epithelial cells resume the cell division cycle when IGF-I (20 ng/ml) is added to their phenol red-and serum-free medium (7). We have demonstrated that E 2 by itself at concentrations ranging from 0.1 to 100 nM does not induce cell cycle progression in quiescent MCF-7S cells. However, a combination of submitogenic amounts of IGF-I and E 2 synergistically induces cell cycle progression and proliferation (8).
It should be noted that, in contrast to other laboratory MCF-7 cell lines, MCF-7S cells are almost completely growtharrested in G 0 /G 1 by serum deprivation in estrogen-free medium, without the need for estrogen antagonists to reach quiescence. However, when the cells are cultured for Ͼ30 passages, they start to grow more rapidly and no longer show complete growth arrest upon serum withdrawal. The serumstarved MCF-7S cells become progressively less dependent on IGF-I with passage number. Eventually, proliferation of serum-starved MCF-7 cells can be induced by E 2 without addition of IGF-I, as is observed in most laboratory strains of MCF-7. In vitro cultures of cancer cells are known to change their growth characteristics with time. The cultures become increasingly more independent of exogenously added growth factors because of the selective growth advantage of cells producing growth factors themselves. For this reason, no MCF-7S cells with a passage number over 25 were used in our experiments.
Using the MCF-7S cells as a model system, we investigated the pathways involved in the synergistic activation of cell cycle progression by IGF-I and E 2 . In contrast with several published reports, we could not detect any activation of cytoplasmic signaling cascades by E 2 and any synergy in the activation of these cascades by IGF-I and E 2 . In terms of cell cycle-related molecules, we found that IGF-I dose-dependently raised cyclin D1 levels in serum-starved cells. Subsequent activation of the cyclin E-CDK2 complex, hyperphosphorylation of pRb, and DNA synthesis were detected only in cells treated with mitogenic concentrations of IGF-I. Treatment of cells with E 2 also led to induction of cyclin D1, but, in the absence of IGF-I, did not lead to cell cycle progression. The absence of CDK4-specific pRb Ser 780 phosphorylation after treatment of the cells with E 2 suggested that the cyclin D1-CDK4 complex was not activated in these cells. We have shown that co-exposure of the cells to E 2 and a submitogenic amount of IGF-I did activate the cyclin D1-CDK4 complex. From this, we concluded that IGF-I signaling is required for G 1 -to-S phase transition. If E 2 is present to induce cyclin D1 levels, a non-mitogenic amount of IGF-I suffices to activate the cyclin D1-CDK4 complex and to enter S phase. The nature of the role of IGF-I in the activation of the complex is not clear. The activity of the cyclin D1-CDK4 complex is dependent on multiple factors, including composition and subcellular localization of the complex and phosphorylation status of CDK4.
Here, we investigated the mechanism by which IGF-I induces the activation of the cyclin D1-CDK4 complex. We show that cyclin D1, although strongly induced, was not translocated to the nucleus in late G 1 phase in E 2 -treated MCF-7S cells. Nuclear accumulation of cyclin D1 and activation of the cyclin D1-CDK4 complex were observed only in cells with elevated cyclin D1 levels subjected to the action of per se non-mitogenic amounts of IGF-I.
Cell Culture-MCF-7S human breast cancer cells were cultured in Dulbecco's minimal essential medium/Ham's F-12 medium (1:1) supplemented with 5% fetal calf serum, 300 g/ml glutamine, 100 IU/ml penicillin, and 100 g/ml streptomycin. Cell cultures of 40 -60% confluence were used for experiments. Prior to experiments, MCF-7S cells were made quiescent by culturing the cells for 24 h in phenol red-free medium containing 5% dextran-coated charcoal-treated serum, followed by a 24-h incubation in phenol red-and serum-free medium supplemented with 0.2% BSA, 10 g/ml transferrin, and 30 nM sodium selenite.
[ 3 H]Thymidine Incorporation-Assays were performed in 24-well plates. Serum-starved cells were stimulated with IGF-I, E 2 , or a combination of the two hormones. After 24 h of incubation, [ 3 H]thymidine (2 Ci/ml) was added. 30 h after addition of the stimuli, cells were fixed with 10% trichloroacetic acid; washed with PBS; and lysed in 0.1 M NaOH, followed by liquid scintillation counting.
For detection of proteins in total lysates, cells were seeded in six-well plates and 25-or 75-cm 2 cell culture flasks. After stimulation with hormones, cells were washed with ice-cold PBS and lysed in sample buffer. Lysates were boiled for 5 min, and protein concentrations were determined using the BCA protein assay (Pierce). Equal amounts of protein (5-50 g) were size-separated on an SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride membrane.
Specific proteins were detected by Western blotting. In brief, the polyvinylidene difluoride membrane was blocked with 2-3% nonfat milk or 0.5-2% BSA in 150 mM NaCl, 10 mM Tris-HCl, pH 8.0, and 0.2% Tween 20 (Tris-buffered saline/Tween) for 1 h. Incubation with the primary antibody was performed overnight at 4°C in 0.1ϫ blocking solution. The membrane was washed with Tris-buffered saline/Tween and exposed to horseradish peroxidase-coupled secondary antibody for 1 h in Tris-buffered saline/Tween. For detection, the Western blot chemiluminescence reagent plus system (PerkinElmer Life Sciences) was used.
CDK2 Kinase Assay-The CDK2 assay was performed as described previously (7). Briefly, cells were treated with the stimuli and then washed with PBS and lysed in immunoprecipitation buffer. After preclearing, the lysates were incubated overnight at 4°C with 50 l of 10% protein A beads coupled to anti-CDK2 antibody. Beads were washed several times and resuspended in kinase mixture. The mixture was incubated for 15 min at 30°C, after which the reaction was stopped with 4ϫ sample buffer containing 400 mM dithiothreitol. The suspension was boiled and centrifuged, and the supernatant was run on SDS-polyacrylamide gel. The gel was dried and exposed to x-ray film. Levels of 32 P incorporation in the substrate were quantified using a PhosphorImager (Amersham Biosciences).
Transfections-Cells were seeded in 6-or 12-well plates (10 6 cells/ plate) 1 day before transfection. Transfections were done using the DAC30 transfection reagent (Eurogentech, Seraing, Belgium). 2-3 g/ well DAC30 and 1-2 g/well plasmid DNA were used, each diluted in 500 l of HBS (150 mM NaCl and 20 mM HEPES, pH 7.4). The DNA and DAC30 dilutions were then combined, gently mixed, and left at room temperature for 30 min. Immediately before transfection, the cell medium was refreshed using 1 ml/well. The DNA/DAC30 mixture was then added dropwise to the cells under continuous swirling, after which they were incubated for 3-4 h at 37°C. Subsequently, the transfection mixture was replaced by phenol red-and serum-free medium to serumstarve the cells.
Immunofluorescence Staining and Microscopy-For detection of specific immunofluorescence, cells were seeded in 12-well plates (10 6 cells/ plate) on 12-mm-diameter glass coverslips. After stimulation with hormones, cells were washed twice with ice-cold PBS, and the coverslips were transferred into humid incubation chambers, where they were kept until the immunofluorescence staining procedure of the cells was completed. To fix the cells, they were incubated with paraformaldehyde for 15 min at room temperature. Next, the coverslips were washed three times for 5 min with HBS (135 mM NaCl, 10 mM KCl, 0.4 mM MgCl 2 , 1 mM CaCl 2 , 10 mM HEPES, pH 7.5), and the remaining fixative was quenched with 100 mM Tris-HCl, pH 8.0. Subsequently, the cells were treated with methanol at Ϫ20°C for 6 min, after which they were washed twice with HBS. Nonspecific sites were blocked with 1% BSA/ HBS for 1 h at room temperature before specific staining with the appropriate antibodies. For cyclin D1 staining, the cells were incubated 1 h at room temperature with anti-cyclin D1/2 antibody (1:50 dilution in 1% BSA/HBS), followed by a 1-h incubation with biotin-labeled goat anti-mouse antibody (1:40 dilution in 1% BSA/HBS) and finally with Cy2-labeled streptavidin (1:1000 dilution in 1% BSA/HBS). For detection of transfected HA-GSK3␤(S9A), cells were stained with anti-HA probe antibody (1:250 dilution in 1% BSA/HBS) for 1 h, followed by TRITC-conjugated goat anti-rabbit antibody (1:500 dilution in 1% BSA/ HBS), also for 1 h. After incubation with an antibody or Cy2-labeled streptavidin, the coverslips were washed five times for 5 min with HBS. The coverslips were mounted in Immumount (Shandon, Pittsburgh, PA). Stained cells were visualized with a Nikon Optiphot-2 microscope fitted with appropriate fluorescence filters.

IGF-I and E 2 Do Not Affect the Composition of the Cyclin D1-CDK4 Complex-
We have previously shown that submitogenic amounts of IGF-I in combination with E 2 synergistically induce proliferation of the MCF-7S human breast tumor epithelial cell line, whereas E 2 on its own does not induce proliferation in this particular subline of MCF-7 (8). Although both hormones induce cyclin D1 protein expression, we demonstrated that in these cells, IGF-I action is required to induce cyclin D1-CDK4 complex-specific phosphorylation of serine 780 in pRb (8). To establish this particular role of IGF-I, the composition of the cyclin D1-CDK4 complex was studied after addition of IGF-I and/or E 2 to the cells. Serum-starved MCF-7S cells were given no mitogens, mitogenic amounts of IGF-I (20 ng/ml), submitogenic amounts of IGF-I (2 ng/ml), a combination of submitogenic amounts of IGF-I (2 ng/ml) and E 2 (1 nM), or E 2 (1 nM). After 16 h of stimulation, the cells were lysed. In the lysate, elevated cyclin D1 levels were detected in cells treated with IGF-I (20 ng/ml), with E 2 , and with a combination of IGF-I (2 ng/ml) and E 2 . The hormones did not significantly alter CDK4 and p27 levels, but p21 was up-regulated when the cells were treated with mitogenic amounts of IGF-I ( Fig. 1, Total lysate). For detection of proteins associated with cyclin D1, cyclin D1 was immunoprecipitated. Co-immunoprecipitation of CDK4, p21, and p27 was detected by Western blotting. Although the absolute amount of immunoprecipitated cyclin D1 and of co-immunoprecipitated CDK4, p27, and p21 varied in the five immunoprecipitated samples, we found no obvious differences when comparing the relative content of these proteins in the cyclin D1 complexes from cells treated with a mitogenic stimulus (20 ng/ml IGF-I or 2 ng/ml IGF-I and 1 nM E 2 ) or with a non-mitogenic stimulus (2 ng/ml IGF-I or 1 nM E 2 ) (Fig. 1, IP). Conceivably, other proteins associated with the complex may regulate the activity of CDK4. A number of reports have shown CDC37, HSP90, and calmodulin to be involved in the regulation of CDK4 activity (9 -11). However, we could not detect any of these proteins in our co-immunoprecipitation studies, although they were readily detected in the total lysate samples. Moreover, treatment of the cells with the different hormones had no effect on the levels of any of the three proteins in the total lysates (data not shown). E 2 in Combination with IGF-I Induces Nuclear Accumulation of Cyclin D1 during Late G 1 Phase-To establish whether IGF-I affects the subcellular localization of cyclin D1 during the G 1 phase of the cell cycle, we stained serum-starved and hormone-treated cells for cyclin D1 expression in a time course experiment for up to 20 h after stimulation. We found that mitogenic stimuli were able to change the subcellular localization of cyclin D1. In serum-starved cells, a low level of cyclin D1 was present, as shown in Fig. 1. The localization of cyclin D1 in these cells was predominantly perinuclear ( Fig. 2A). In cells treated with non-mitogenic concentrations of either IGF-I or E 2 , the localization of the cyclin D1 protein was not changed. Only in a small percentage of cells was nuclear fluorescence detected, 4 and 2% for IGF-I (2 ng/ml)-and E 2 -treated cells, respectively (Fig. 2, A and B). However, treatment of the cells with mitogenic concentrations of IGF-I or with non-mitogenic concentrations in combination with E 2 did result in a change in localization. We first observed this change around 16 h after addition of the stimulus (data not shown). Therefore, the cells  (20) ), 2 ng/ml IGF-I (I (2) ), 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 (I (2) E2) was added for 16 h. Cell lysates were subjected to immunoprecipitation (IP) with anti-cyclin D1/2 monoclonal antibody. The precipitated proteins were size-separated on an SDSpolyacrylamide gel and analyzed by Western blotting. Co-immunoprecipitation of CDK4, p27, and p21 was visualized with the appropriate antibodies. Presence of the proteins in the total lysate is shown on the same blot. In the immunoprecipitated samples of the p27 blot, an additional band was detected, migrating slightly faster than p27. This is the IgG light chain for the immunoprecipitation antibody that also reacted with the anti-mouse secondary antibody.

FIG. 2. Immunofluorescence detection of cyclin D1 subcellular localization in IGF-I-and E 2 -treated MCF-7S cells.
A, MCF-7S cells were grown on glass coverslips in 12-well plates and serumstarved for 26 h. 20 ng/ml IGF-I (I (20) ), 2 ng/ml IGF-I (I (2) ), 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 (I (2) E2) was added for 17 h. After fixation, the cells were immunostained for cyclin D1 expression. Magnification is 400-fold. B, the bar graph shows the percentage of cells with nuclear cyclin D1 staining from a total of 1000 cells in three independent experiments.
in the experiments shown were all fixed and stained after 17 h of stimulation. In 94 and 87% of cells treated with 20 ng/ml IGF-I or with 2 ng/ml IGF-I and 1 nM E2, respectively, the cyclin D1 staining was nuclear with a uniform distribution (Fig. 2, A and B), except for the nucleoli that appeared unstained. This accumulation in the nucleus seems essential for cell cycle progression, which may be explained by the fact that the pRb protein, one of the major CDK4 targets, is also local-ized in the nucleus during G 0 /G 1 phase (12,13).
Our data show that mitogenic concentrations of IGF-I (20 ng/ml) and submitogenic amounts of IGF-I (2 ng/ml) in combination with E 2 induced nuclear accumulation of cyclin D1, whereas IGF-I (2 ng/ml) and E 2 (1 nM) separately did not. This suggests that an IGF-I-activated signal is essential for nuclear accumulation. However, the elevation of cyclin D1 levels seems to be essential as well because IGF-I (2 ng/ml) did not induce nuclear accumulation. This suggests that a threshold in cyclin D1 levels must be reached before an IGF-I-activated signal facilitates the accumulation of the protein in the nucleus.
Nuclear Accumulation of Cyclin D1 Is Regulated via the PI3K Pathway-Previously, we demonstrated that neither the ERK2 kinase pathway nor the PI3K pathway is significantly activated after treatment of MCF-7S cells with E 2 , whereas stimulation with IGF-I leads to the activation of both pathways in a dose-dependent manner (8). Using specific inhibitors, we have shown activation of the PI3K pathway by IGF-I to be essential for cell cycle progression. In particular, this pathway was shown to be involved in the regulation of cyclin D1 mRNA and protein levels (7,14). Here, we investigated whether the PI3K or ERK pathway is involved in the transduction of the IGF-I signal leading to nuclear accumulation of cyclin D1. To this end, inhibitors of these pathways were added to serumstarved MCF-7S cells 1 h prior to stimulation with hormones. The PI3K inhibitor LY294002 (25 M) completely abolished nuclear accumulation of the cyclin D1 protein in late G 1 phase. In cells treated with the LY294002 inhibitor and mitogenic amounts of IGF-I or with a combination of submitogenic amounts of IGF-I and E 2 , the percentage of cells showing nuclear accumulation of cyclin D1 was reduced to background levels (Fig. 3). The ERK pathway is not critically involved in induction of nuclear accumulation because the MEK inhibitor PD098059 (40 M) did not significantly block translocation of cyclin D1 to the nucleus. Some residual cytoplasmic staining was visible in cells treated with PD098059 in combination with either 20 or 2 ng/ml IGF-I together with 1 nM E 2 (Fig. 3A). However, this effect may well be due to an aspecific stress response to the presence of the inhibitor in the medium for 18 h.
Submitogenic Concentrations of IGF-I Inactivate GSK3␤ Kinase Activity by Phosphorylation of Serine 9 -GSK3␤ is a well documented downstream target of PI3K. Activation of PI3K leads to activation of protein kinase B (AKT-1), which has been  (20) ), 2 ng/ml IGF-I (I (2) ), 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 (I (2) E2) was added for 17 h. After fixation, the cells were immunostained for cyclin D1. The bar represents 10 m. B, the bar graph shows the percentage of cells with nuclear cyclin D1 staining from a total of 1000 cells in three independent experiments.

FIG. 4. Phosphorylation of GSK3␤ in IGF-I-and E 2 -treated MCF-7S cells.
Serum-starved cells were stimulated with 20 ng/ml IGF-I (I (20) ), 2 ng/ml IGF-I (I (2) ), 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 (I (2) E2) and harvested after 30 min, 1, 2, 4, 8, or 10 h. 10 g of protein was used for Western blotting. The blot was probed for phospho-GSK3␣/␤ Ser 21/9 (pGSK 3␤) and, after stripping off the antibodies, probed for total GSK3␤. shown to inactivate GSK3␤ by specific phosphorylation. Recently, GSK3␤ has been implicated in the regulation of the relocalization of cyclin D1 from the nucleus to the cytoplasm during S phase (15). To investigate whether nuclear accumulation of cyclin D1 in G 1 phase is regulated by GSK3␤, we first determined whether the GSK3␤ kinase was inhibited upon  (20) ), 2 ng/ml IGF-I (I (2) ), 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 (I (2) E2) was added in the presence or absence of 10 mM LiCl for 17 h. After fixation, the cells were immunostained for cyclin D1. The bar represents 10 m. In B, the bar graph shows the percentage of cells with nuclear cyclin D1 staining from a total of 800 cells in three independent experiments. C, effect of LiCl on cyclin D1 protein levels. Serum-starved cells were stimulated with 20 ng/ml IGF-I, 2 ng/ml IGF-I, 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 in the presence or absence of 10 mM LiCl for 16 h. 5 g of protein was used for cyclin D1 detection by Western blotting. D, effect of LiCl on CDK2 activity. Serum-starved MCF-7S cells were stimulated with 20 ng/ml IGF-I, 2 ng/ml IGF-I, 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 in the presence or absence of 10 mM LiCl for 24 h. Cell lysates were subjected to immunoprecipitation with anti-cyclin E antibody, and associated CDK2 activity was measured in a kinase assay using histone H1 as substrate. E, effect of LiCl on pRb phosphorylation. Serum-starved MCF-7S cells were incubated with 20 ng/ml IGF-I, 2 ng/ml IGF-I, 1 nM E 2 , or a combination of 2 ng/ml IGF-I and 1 nM E 2 in the presence or absence of 10 mM LiCl for 18 h. Cell lysates were used for immunoprecipitation of total pRb and subsequent Western blot analysis for cyclin D1-specific phosphorylation of Ser 780 . The blot was then stripped and reprobed for total pRb protein.
The hyperphosphorylated form of pRb (ppRb) was detected as a slower migrating band compared with the hypophosphorylated form, pRb. F, effect of LiCl on induction of DNA synthesis. Serum-starved cells grown in 24-wells plates were given no mitogens (Ϫ), 20 ng/ml IGF-I, 2 ng/ml IGF-I, 2 ng/ml IGF-I in combination with 1 nM E 2 , or 1 nM E 2 in the presence or absence of 10, 25, or 50 mM LiCl and were harvested after 30 h. addition of non-mitogenic concentrations of IGF-I to serumstarved MCF-7S cells. In vivo, GSK3␤ is rendered inactive by phosphorylation of Ser 9 , which may be detected by Western blotting using an antibody specifically recognizing phospho-GSK3␤ Ser 9 . As shown in Fig. 4, Ser 9 of GSK3␤ remained phosphorylated for at least 4 h after addition of IGF-I (2 ng/ml) to the cells, whereas addition of E 2 did not lead to increased phosphorylation of GSK3␤ at any time. Addition of mitogenic concentrations of IGF-I led to a phosphorylated state of GSK3␤ for at least 16 h.
LiCl, an Inhibitor of GSK3␤, Induces Nuclear Accumulation of Cyclin D1 and, in Combination with E 2 , Leads to G 1 -to-S Phase Transition-Subsequently, we used LiCl, a chemical inhibitor of GSK3␤, to establish the role of the kinase in the regulation of the subcellular localization of cyclin D1 during G 1 phase. LiCl inhibits GSK3␤ by competition for Mg 2ϩ binding, but not for ATP or substrate binding, and has been shown to be highly specific for GSK3␤ (16). We determined whether inhibition of GSK3␤ by LiCl leads to nuclear accumulation of cyclin D1 and to cell cycle progression. Immunofluorescence staining of cyclin D1 showed that treatment of quiescent cells with 10 mM LiCl for 17 h led to an increase in nuclear cyclin D1. The percentage of cells showing nuclear cyclin D1 in samples treated with mitogenic amounts of IGF-I or with submitogenic amounts of IGF-I in combination with E 2 was not significantly changed by LiCl addition (Fig. 5, A and B). In untreated and IGF-I (2 ng/ml)-treated cells, however, nuclear staining was observed in 25 and 61% of the cells, respectively, compared with 1 and 4% without LiCl. The largest difference was observed with E 2 , showing 83% of the cells with nuclear localization of cyclin D1 in the presence of LiCl versus only 2% in the absence of LiCl. Because of considerable variations in LiCl concentrations used in published reports, we initially tested concentrations ranging from 2.5 to 50 mM. In cells treated with E 2 in the presence of higher concentrations of LiCl, no significant rise in the percentage of cells with nuclear cyclin D1 was observed compared with the percentage of cells treated with E 2 in the presence of 10 mM LiCl. In cells treated with E 2 in the presence of lower concentrations of LiCl, the percentage of cells with nuclear localization of cyclin D1 was not or only slightly raised compared with background levels (data not shown). In subsequent experiments, 10 mM LiCl was used to inhibit endogenous GSK3␤ activity.
A number of studies have shown that the cyclin D1 protein is stabilized by inhibition of GSK3␤ (15,17). Therefore, we investigated whether LiCl by itself has an effect on cyclin D1 levels in MCF-7S cells. Fig. 5C shows that when LiCl was present in the medium of serum-starved or hormone-treated cells for 16 h, a rise in cyclin D1 levels was observed in all of the samples compared with their counterparts without LiCl. The higher level of cyclin D1 in IGF-I (2 ng/ml)-treated cells in the presence of LiCl may explain the observed increase in numbers of cells showing nuclear localization of cyclin D1.
To investigate whether the nuclear accumulation of cyclin D1 observed in cells stimulated with IGF-I (2 ng/ml) and E 2 in the presence of LiCl leads to cell cycle progression, we studied the levels and activities of proteins involved in G 1 -to-S phase transition. The activity of the cyclin E-CDK2 complex was monitored in cells stimulated for 22 h with hormones in the presence or absence of 10 mM LiCl. The cells were lysed, and CDK2 was immunoprecipitated. Kinase activity was measured using histone H1 as substrate. Fig. 5D shows that addition of IGF-I (20 ng/ml) as well as of a combination of IGF-I (2 ng/ml) and E 2 (1 nM) in the absence of LiCl led to a clear increase in active cyclin E-CDK2 complex activity in the cells, whereas treatment of the cells with either IGF-I (2 ng/ml) or E 2 increased the activity of the kinase only slightly. In serumstarved and IGF-I (2 ng/ml)-treated cells, addition of 10 mM LiCl raised background CDK2 activity slightly, whereas it had little additional effect on the already active CDK2 in IGF-I (20 ng/ml)-or IGF-I (2 ng/ml)/E 2 -treated cells. However, a large effect was seen in cells treated with E 2 . In the presence of LiCl, the CDK2 kinase activity became almost equal to the activity in samples of cells treated with mitogenic concentrations of IGF-I or with a combination of IGF-I and E 2 . This again suggests that LiCl can mimic the action of IGF-I when used in combination with E 2 . Surprisingly, in cells treated with only low amounts of IGF-I in the presence of LiCl, nuclear accumulation of the weakly induced cyclin D1 was observed, but no activation of CDK2 ensued.
Phosphorylation of the pocket protein pRb is a subsequent important event in G 1 -to-S phase progression. We studied the phosphorylation status of pRb in cells treated with E 2 , IGF-I, or a combination of the two hormones in the presence or absence of 10 mM LiCl. Both the phosphorylation status of Ser 780 of pRb, which is phosphorylated specifically by the cyclin D1-CDK4 complex (18,19), and the total phosphorylation status of pRb after 18 h of stimulation were examined. As we have shown previously (8), addition of IGF-I (20 ng/ml) as well as of a combination of IGF-I (2 ng/ml) and E 2 (1 nM) to MCF-7S cells clearly led to the phosphorylation of Ser 780 as well as of other residues in pRb (Fig. 5E). Treatment of the cells with IGF-I (2 ng/ml) or E 2 (1 nM) had hardly any effect on the phosphorylation state of pRb. Addition of LiCl to serum-starved cells caused some Ser 780 phosphorylation and a small shift in the total phosphorylation assay. Also in cells treated with IGF-I in nonmitogenic amounts, some hyperphosphorylated pRb was detected when LiCl was present. Much stronger hyperphosphorylation of pRb was seen with IGF-I (20 ng/ml) or IGF-I (2 ng/ml)/E 2 , independent of the presence LiCl. Again, the strongest effect of LiCl was observed when E 2 was added to the cells. In these cells, the level of hyperphosphorylation was comparable with the level in cells treated with mitogenic concentrations of IGF-I or with a combination of IGF-I and E 2 .
Progression through S phase was further monitored by a DNA synthesis assay. Serum-starved MCF-7S cells were treated with IGF-I and E 2 in the presence or absence of LiCl. After 24 h, [ 3 H]thymidine was added; the cells were harvested after further incubation for 6 h; and [ 3 H]thymidine incorporation was measured by liquid scintillation counting. Fig. 5F (left panel) shows that E 2 by itself induced only a 13-fold increase in [ 3 H]thymidine incorporation compared with background incorporation in unstimulated cells during the same time period. Mitogenic amounts of IGF-I induced a 142-fold increase in [ 3 H]thymidine incorporation, whereas IGF-I at a non-mitogenic concentration induced only a small increase, comparable with the amount of incorporation observed with E 2 . A combination of IGF-I (2 ng/ml) and E 2 induced a 134-fold increase. Addition of LiCl to serum-starved cells induced an 8-fold increase in background levels of [ 3 H]thymidine incorporation, but did not significantly affect incorporation in IGF-I (2 or 20 ng/ml)-or IGF-I/E 2 -treated cells. In cells treated with E 2 in the presence of LiCl, DNA synthesis was induced to 106-fold over background level, comparable with levels with IGF-I or E 2 combined with submitogenic concentrations of IGF-I. This correlates well with the observed activation of CDK2 and hyperphosphorylation of pRb in E 2 -treated cells in the presence of LiCl. DNA synthesis in E 2 -stimulated cells in the presence of higher concentrations of LiCl up to 25 mM was not significantly different from that observed in E 2 -treated cells in the presence of 10 mM LiCl (Fig. 5F, right panel). Higher concentrations of LiCl (50 mM) resulted in low levels of [ 3 H]thymidine incorporation, although nuclear accumulation of cyclin D1 was observed. This may be due to cytotoxicity or aspecific inhibition of enzymes essential for progression through the cell cycle.
We conclude that inhibition of GSK3␤ by LiCl leads to nuclear accumulation of cyclin D1 and raises steady-state cyclin D1 levels. Under these conditions, addition of E 2 to the cells leads to CDK2 activation, pRb phosphorylation, and DNA synthesis. LiCl treatment thus substitutes for IGF-I, which is able to suppress GSK3␤ activity via PI3K. Inhibition of GSK3␤ thus seems a key event in the regulation of cell cycle progression in MCF-7S cells.
The GSK3␤(S9A) Mutant Inhibits Nuclear Translocation of Cyclin D1-To further investigate the involvement of GSK3␤ in induction of cyclin D1 accumulation in the nucleus during G 1 phase, MCF-7S cells were transfected with the HA-tagged GSK3␤(S9A) mutant. This mutant lacks the Ser 9 phosphorylation site, making it inert to inactivation by protein kinase B. However, GSK3␤(S9A) can still be inactivated by LiCl (20). After transfection, the cells were steroid hormone-deprived and serum-starved for 26 h. Subsequently, no mitogens, IGF-I (20 ng/ml), or IGF-I (2 ng/ml) in combination with E 2 was added in the presence or absence of 25 mM LiCl. 25 mM LiCl was used to ensure optimal inhibition of the overexpressed GSK3␤ without aspecific inhibitory effects on cell cycle progression in MCF-7S cells (Fig. 5F, right panel). 17 h after addition of the hormones, cells were fixed and stained for HA and cyclin D1 expression. We found that transfection of the cells with GSK3␤(S9A) had no effect on the localization of cyclin D1 in serum-starved cells. Stimulation of GSK3␤(S9A)-transfected cells with mitogenic amounts of IGF-I or with a combination of non-mitogenic amounts of IGF-I and E 2 did not result in nuclear accumulation of cyclin D1, whereas cyclin D1 was nuclear in the non-transfected cells in the same experiment. In the presence of LiCl, the nuclear accumulation in the stimulated transfected cells was restored (Fig. 6). GSK3␤ inactivation is thus a necessary step in the accumulation of cyclin D1 in the nucleus in late G 1 phase. DISCUSSION Estrogen and IGF-I are hormones involved in a wide variety of processes regulating proliferation, apoptosis, and differentiation in mammalian cells (21,22). In MCF-7S cells, a combi-nation of both hormones synergistically induces proliferation (23). We have previously shown that the synergistic action of E 2 and IGF-I derives from the ability of both hormones to induce cyclin D1 expression and that IGF-I action is required to induce activity of the cyclin D1-CDK4 complex, which triggers cell cycle progression (8). The objective of this study was to gain more insight into the nature of the IGF-I signal required to activate the cyclin D1-CDK4 complex.
In this report, we have shown that no obvious differences were detected in the composition of the cyclin D1-CDK4 complex in cells treated with mitogenic stimuli (20 ng/ml IGF-I or a combination of 2 ng/ml IGF-I and E 2 ) or non-mitogenic stimuli (2 ng/ml IGF-I or E 2 ). However, IGF-I did affect the subcellular localization of the cyclin-CDK4 complex. We found that IGF-I (but not E 2 ) induced the accumulation of cyclin D1 in the nucleus in late G 1 phase. This IGF-I signal was PI3K-dependent and completely inhibited by LY294002. GSK3␤, a well established downstream target of PI3K, has previously been shown to be involved in the regulation of the subcellular localization of cyclin D1 (15,17,24,25). Diehl et al. (15,17) elegantly showed that active GSK3␤ phosphorylates the cyclin D1 protein during S phase, which increases the rate of nuclear export relative to nuclear import by facilitating the association of cyclin D1 with the nuclear exportin CRM1. We investigated whether inactivation of GSK3␤ is involved in the regulation of nuclear accumulation of cyclin D1 in late G 1 phase. We demonstrated that addition of LiCl, a well documented inhibitor of GSK3␤, to E 2 -stimulated cells induced nuclear cyclin D1 accumulation, CDK2 activation, pRb hyperphosphorylation, and DNA synthesis. In addition, we found that the GSK3␤(S9A) mutant, which cannot be inactivated by IGF-I, blocked nuclear accumulation. This shows that GSK3␤ inactivation is an essential step in the regulation of the subcellular localization of cyclin D1 during G 1 phase. Two observations make it unlikely that the observed nuclear accumulation of cyclin D1 in late G 1 phase is regulated by the same mechanism as described by Diehl et al. First, if (lack of) phosphorylation of cyclin D1 is responsible for the translocation in late G 1 , nuclear accumulation of the complex in E 2 -treated cells should be inhibited by GSK3␤-induced cyclin D1 phosphorylation. Addition of IGF-I prevented this phosphorylation by inactivating GSK3␤ and thus directly induced nuclear accumulation of cyclin D1 once GSK3␤ was inactivated. However, we found that non-mitogenic concentrations of IGF-I (in the presence or absence of E 2 ) inhibited the GSK3␤ kinase for only 4 -6 h and that nuclear accumulation of cyclin D1 was first observed after 16 h. If inactivation of GSK3␤ induces nuclear accumulation of cyclin D1 directly by regulating cyclin D1 phosphorylation, we should be able to observe the accumulation much before 16 h after addition of IGF-I to the cells. Second, if nuclear accumulation is exclusively caused by decreased nuclear export by impeding the association of CRM1 with cyclin D1, this would suggest that there is constitutive import of the complex into the nucleus. Addition of leptomycin B, a known inhibitor of CRM1, to MCF-7S cells should thus lead to nuclear accumulation of cyclin D1. However, we found that addition of leptomycin B to the cells for 3 h in mid-late G 1 phase did not lead to nuclear accumulation of cyclin D1 (data not shown). Taken together, these data suggest that nuclear accumulation of cyclin D1 is indirectly regulated by GSK3␤ via a process of activated import rather than inhibited export via CRM1.
Treatment of MCF-7S cells with E 2 resulted in induction of cyclin D1, but not in nuclear accumulation of cyclin D1, as long as GSK3␤ was not inactivated. Treatment of serum-starved MCF-7 cells with submitogenic amounts of IGF-I (2 ng/ml) inactivated GSK3␤, but still did not induce nuclear accumulation of cyclin D1. Only a combination of the two hormones induced nuclear accumulation of cyclin D1, suggesting that inactivation of GSK3␤, as well as a certain threshold level of cyclin D1, is required for the change in cyclin D1 localization to occur. In fibroblasts, a protein named SSeCKS has been shown to bind cyclin D1 in a phosphorylation-dependent manner and to act as a cytoplasmic anchor (26 -28). Stimulation of the fibroblasts with mitogens was shown to cause down-regulation of the protein and, in addition, to induce phosphorylation of the protein by protein kinase C. Both effects together result in nuclear accumulation of cyclin D1 (26 -28). Conceivably, this mechanism also operates in MCF-7S cells. This would suggest that cyclin D1 is bound to SSeCKS, retaining it in the cytoplasm, as long as GSK3␤ is active. When GSK3␤ is inactivated by IGF-I stimulation of the cells, the scaffolding protein is down-regulated and phosphorylated by protein kinase C, making it unable to retain the cyclin D1 protein in the cytoplasm as soon as the threshold level is reached. However, in MCF-7S cells, inhibition of protein kinase C did not result in retention of cyclin D1 in the cytoplasm (data not shown). This finding suggests that SSeCKS is not involved in the regulation of the subcellular localization of cyclin D1 in MCF-7S cells. Nevertheless, other scaffolding proteins may bind cyclin D1 and retain it in the cytoplasm until both GSK3␤ is inactivated and the cyclin D1 threshold level is reached.
We found that low concentrations of IGF-I in the presence of LiCl to some extent induced cyclin D1 protein levels as well as nuclear accumulation of cyclin D1, but that no cell cycle pro-gression ensued. This would imply that not only a threshold level of cyclin D1 must be reached before the protein can translocate to the nucleus, but that a threshold level of cyclin D1 or of cyclin D1-CDK4 complex activity must be reached as well before the cells progress through the cell cycle.
Based on our findings, we propose a model (Fig. 7) in which the synergy of IGF-I and E 2 signaling is explained by regulation of cyclin D1 levels and by GSK3␤ inactivation. High levels of IGF-I are capable of inducing cyclin D1 and of inactivating GSK3␤. E 2 may substitute for IGF-I in its capacity to enhance cyclin D1 levels. Because a much lower concentration of IGF-I is needed to suppress GSK3␤ activity, per se non-mitogenic amounts of IGF-I in synergy with E 2 may now trigger proliferation of MCF-7S cells.