Cyclin D1 overexpression induces progestin resistance in T-47D breast cancer cells despite p27(Kip1) association with cyclin E-Cdk2.

Long-term growth inhibition, arrest in G(1) phase and reduced activity of both cyclin D1-Cdk4 and cyclin E-Cdk2 are elicited by progestin treatment of breast cancer cells in culture. Decreased cyclin expression, induction of p18(INK4c) and increased association of the CDK inhibitors p21(WAF1/Cip1) and p27(Kip1) with cyclin E-Cdk2 have been implicated in these responses. To determine the role of decreased cyclin expression, T-47D human breast cancer cells constitutively expressing cyclin D1 or cyclin E were treated with the progestin ORG 2058. Overexpression of cyclin E had only a modest effect on growth inhibition. Although cyclin E expression was maintained during progestin treatment, cyclin E-Cdk2 activity decreased by approximately 60%. This was accompanied by p27(Kip1) association with cyclin E-Cdk2, indicating that both cyclin E down-regulation and p27(Kip1) recruitment contribute to the decrease in activity. In contrast, overexpression of cyclin D1 induced progestin resistance and cell proliferation continued despite decreased cyclin E-Cdk2 activity. Progestin treatment of cyclin D1-overexpressing cells was associated with increased p27(Kip1) association with cyclin E-Cdk2. Thus the ability of cyclin D1 to confer progestin resistance does not depend on sequestration of p27(Kip1) away from cyclin E-Cdk2, providing evidence for a critical function of cyclin D1 other than as a high-capacity "sink" for p27(Kip1). These data indicate that regulation of cyclin D1 is a critical element of progestin inhibition in breast cancer cells and suggest that breast cancers overexpressing cyclin D1 may respond poorly to progestin therapy.

Long-term growth inhibition, arrest in G 1 phase and reduced activity of both cyclin D1-Cdk4 and cyclin E-Cdk2 are elicited by progestin treatment of breast cancer cells in culture. Decreased cyclin expression, induction of p18 INK4c and increased association of the CDK inhibitors p21 WAF1/Cip1 and p27 Kip1 with cyclin E-Cdk2 have been implicated in these responses. To determine the role of decreased cyclin expression, T-47D human breast cancer cells constitutively expressing cyclin D1 or cyclin E were treated with the progestin ORG 2058. Overexpression of cyclin E had only a modest effect on growth inhibition. Although cyclin E expression was maintained during progestin treatment, cyclin E-Cdk2 activity decreased by ϳ60%. This was accompanied by p27 Kip1 association with cyclin E-Cdk2, indicating that both cyclin E down-regulation and p27 Kip1 recruitment contribute to the decrease in activity. In contrast, overexpression of cyclin D1 induced progestin resistance and cell proliferation continued despite decreased cyclin E-Cdk2 activity. Progestin treatment of cyclin D1overexpressing cells was associated with increased p27 Kip1 association with cyclin E-Cdk2. Thus the ability of cyclin D1 to confer progestin resistance does not depend on sequestration of p27 Kip1 away from cyclin E-Cdk2, providing evidence for a critical function of cyclin D1 other than as a high-capacity "sink" for p27 Kip1 . These data indicate that regulation of cyclin D1 is a critical element of progestin inhibition in breast cancer cells and suggest that breast cancers overexpressing cyclin D1 may respond poorly to progestin therapy.
The female sex steroid progesterone and its synthetic analogues, progestins, have complex effects on cell proliferation and can either stimulate or inhibit cell proliferation, depending on the cell type, tissue, or treatment regimen (1). For example, in the uterus progesterone acts synergistically with estrogen to stimulate stromal proliferation but inhibits estrogen-induced epithelial proliferation. The latter effect has led to the addition of progestin to hormone replacement therapies to counteract the increased risk of endometrial cancer arising from treat-ment with estrogen alone (2). Synthetic progestins have an established role in the therapy of breast and endometrial cancers (2,3), demonstrating a growth-inhibitory effect on breast and endometrial cancer cells, although whether progesterone is stimulatory or inhibitory for normal breast epithelium remains controversial. Progestins have a biphasic effect on the proliferation of breast cancer cells in culture (4), initially stimulating G 1 cells to enter S phase, but the predominant effect is long term growth inhibition. Several recent studies have focused on the molecular mechanisms for this growth inhibition (5)(6)(7).
Progestin-mediated growth inhibition is preceded by decreased expression of the major G 1 cyclins in breast cancer cells, cyclin D1 and cyclin E (5,6), and preferential formation of cyclin E-Cdk2 complexes that contain the CDK 1 inhibitor p27 Kip1 and are therefore inactive (6,7). The related CDK inhibitor p21 Cip1 appears to play a minor role since immunodepletion experiments indicate that few of the cyclin D1-or cyclin E-containing complexes contain p21 Cip1 following progestin treatment (7). The increased association of p27 Kip1 with cyclin E-Cdk2 occurs prior to any increase in p27 Kip1 abundance. Since in breast cancer cells cyclin D-Cdk4 complexes bind a significant fraction of the total cellular p27 Kip1 , a decrease in their abundance will make p27 Kip1 increasingly available to associate with other molecules and this likely contributes to increased p27 Kip1 -cyclin E-Cdk2 association. The increased formation of p27 Kip1 -cyclin E-Cdk2 complexes also reflects increased expression of another CDK inhibitor, p18 INK4c (7). In contrast with p27 Kip1 and p21 Cip1 , which associate with both cyclin D-Cdk4 and cyclin E-Cdk2, p18 INK4c specifically inhibits the activity of cyclin D-associated kinases, restricting cyclin D binding to Cdk4/6 and thereby making p27 Kip1 and p21 Cip1 available to bind other proteins (8). Thus, it is apparent that progestins target multiple elements of the cell cycle control machinery which may contribute to inhibition of CDK activity and consequent inhibition of proliferation. However, the relative importance of regulating cyclin abundance remains unclear, and it was the objective of this study to address this issue by using constitutive overexpression of cyclin D1 or cyclin E as a means of maintaining cyclin expression during progestin treatment.
Additional impetus for these experiments comes from the frequent overexpression of cyclin D1 or cyclin E in breast cancer. Cyclin D1 is overexpressed in ϳ50% of breast cancers (9 -11). The consequences of this for patient prognosis are not clear, with conflicting data from early studies (12)(13)(14)(15)(16). More recent data indicate that cyclin D1 overexpression is an indicator of poor prognosis specifically in estrogen receptor (ER)-positive breast cancers (17). Cyclin E is present as low molecular weight isoforms in breast cancer but not in normal breast epithelium and is overexpressed in ϳ30% of breast cancer specimens (18 -21). This is associated with significant increases in the risk of death or relapse (18,21). One possible explanation for poor outcome associated with cyclin overexpression is impaired response to therapies that modulate cyclin D1 expression, but there are to date few studies addressing the question of whether cyclin overexpression might alter response.
This manuscript demonstrates that overexpression of cyclin D1 induces resistance to progestin-mediated growth inhibition, but overexpression of cyclin E has little effect. Maintenance of cyclin D1 expression after progestin treatment did not prevent increased p27 Kip1 association with cyclin E-Cdk2. This observation is inconsistent with the suggestion, supported by recent genetic evidence (22,23), that sequestration of p27 Kip1 is the physiologically relevant function of cyclin D1. Rather, it indicates that in the context of progestin inhibition, other functions of cyclin D1 are critical. Overall, the data presented here indicate that regulation of cyclin D1 is a central element of progestin inhibition of breast cancer cell proliferation and suggest that breast cancers overexpressing cyclin D1 may respond poorly to progestin therapy.

EXPERIMENTAL PROCEDURES
Cell Lines and Cell Culture-The generation of T-47D human breast cancer cells constitutively expressing either cyclin D1 or cyclin E is described fully elsewhere. 2 In brief, a T-47D clone stably transfected with the tet-responsive transcriptional activator tTA (T-47D tTA-17) was transfected with a pTRE vector containing full-length cyclin D1 or full-length cyclin E cDNA. Stable clones were selected by hygromycin treatment (200 g/ml) following co-transfection with pTK-Hyg to produce cell lines overexpressing cyclin D1 (D1 17-1) or cyclin E (E 17-2, E 17-3). Western blot analysis of cell lysates demonstrated overexpression of the cyclin proteins in the absence of tetracycline. T-47D tTA-17 and clonal derivatives stably transfected with the empty pTRE vector were used as control cell lines. T-47D tTA-17 derivatives were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum, insulin (10 g/ml), and tetracycline (2 g/ml).
T-47D-EcoR-p cells for retroviral infection were generated by stable transfection of a vector encoding the murine (ecotropic) retroviral receptor (pBabePuro-EcoR; provided by Dr. Gordon Peters, Imperial Cancer Research Fund, London, UK) into T-47D breast cancer cells. A clone was selected based on high retroviral infectability and normal steroid responsiveness and expanded for subsequent experiments.
RPMI 1640 medium supplemented with 5% fetal calf serum, insulin (10 g/ml), and gentamicin (20 g/ml) was used to culture the cell lines for progestin treatment experiments. The synthetic progestin ORG 2058 (16␣-ethoxy-21-hydroxy-19-norpregn-4-en-3,20-dione; Amersham Biosciences) was dissolved in ethanol at 1,000-fold final concentration and added to cells in exponential growth. Control cultures received ethanol vehicle to the same final concentration.
Colony-forming Assay-The sensitivity of the cell lines to ORG 2058 treatment was assessed in colony-forming assays. Cells (8000 cells/ plate) were plated into duplicate 6-cm diameter dishes in RPMI 1640,5% fetal calf serum. After 24 h, the cells were treated with a range of ORG 2058 concentrations (0.01-100 nM) or ethanol vehicle (control) and incubated for up to 35 days (typically 18 -21 days) until the control dishes for each cell line reached similar colony size. The medium was changed, and ethanol or ORG 2058 treatment repeated every 7 days during this time period. The cells were fixed and stained using the Diff-Quik Stain Set (64851, Lab Aids Pty. Ltd., Narrabeen, Australia). The number of colonies in each dish was quantitated using Bio-Rad Quantity One 4.2.1 GelDoc software (Bio-Rad Laboratories, Hercules, CA).
Retroviral Infection-Retroviral vectors were constructed as follows: pLib-D1 was made by digesting pLib (CLONTECH Laboratories, Palo Alto, CA) with EcoRI/NotI and pHsCYCD1-H124 (David Beach, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) with EcoRI/ HindIII, followed by agarose gel electrophoresis purification. The fragments were end-blunted with DNA polymerase and ligated. The pLib-cycE vector was constructed similarly, using a cyclin E long-form insert isolated from pBSSK-cycE (Steven Reed, Scripps Research Institute, La Jolla, CA) by EcoRI/HindIII digestion.
Ecotropic retroviruses expressing cyclins D1 and E were packaged in Phoenix-Eco cells (a gift of Philip Achacoso and Garry Nolan, Stanford University Medical Center, Stanford, CA) by transient transfection. Phoenix-Eco cells were seeded into 10-cm diameter dishes and 24 h later were transfected with 20 g of vector DNA using 60 l of Fu-GENE-6 reagent (Roche Diagnostics Australia, Castle Hill, New South Wales, Australia) per dish. The medium was changed 24 h later, and the cells incubated at 32°C for a further 24 h. Viral supernatant was harvested by filtration (Millex-HV Durapore, Millipore Bedford, MA), and polybrene (Sigma) added to a final concentration of 16 g/ml. Viral supernatant was added immediately to target cells by 1:4 dilution in culture medium or stored at Ϫ80°C for later use. As a means of estimating infection efficiency a GFP-expressing virus (pLib-EGFP, CLONTECH) was packaged in parallel with the cyclin constructs.
T-47D-EcoR-p cells (5 ϫ 10 5 cells/plate) were plated into 10-cm diameter dishes in RPMI 1640,5% fetal calf serum. After 24 h, the cells were infected with either the GFP, cyclin D1, or cyclin E viral supernatant and incubated with occasional swirling. The medium was removed 24 h later, and the cells were reinfected with fresh viral supernatant to maximize infection efficiency. Following a further 24 h of incubation, the retroviral-infected cells were plated (8000 cells/plate) into replicate 6-cm diameter dishes in RPMI 1640,5% fetal calf serum, subsequently treated with ORG 2058 or ethanol vehicle, stained after 18 days incubation, and quantitated as described above for the colonyforming assay. Parallel dishes were harvested for analysis of cyclin expression levels by Western blotting. GFP-expressing cells were also harvested for flow cytometric analysis 48 h after infection to estimate the proportion of infected cells, which was typically 30 -50%.
Western Blot Analysis and Immunoprecipitation-Cell lysates (600 g) were immunoprecipitated by incubation (40 min, 4°C) with p27 Kip1 (C-19-G) antibody from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) followed by incubation (1 h, 4°C) with protein G-Sepharose beads (Zymed Laboratories Inc., San Francisco, CA). Alternatively, lysates were additionally incubated with p21 Cip1 (C-19) antibody from Santa Cruz Biotechnology that had been previously chemically cross-linked to protein A-Sepharose beads (Zymed Laboratories Inc.) by incubation in dimethyl pimelimidate (5 mg/ml),sodium tetraborate (0.2 M, pH 9.0), for 30 min at room temperature, essentially as described previously (25). A control with no antibody (mock immunoprecipitation) was included for each sample. Following three rounds of immunoprecipitation, the combined immunoprecipitated proteins were washed with lysis buffer and eluted by the addition of glycine (0.1 M, pH 2.7) for 30 min at room temperature. The beads were cleared by centrifugation and the eluted proteins neutralized with the addition of Tris-HCl (1 M, pH 9.0) before addition of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (63 mM Tris-HCl, pH 6.8, 10% (v/v) glycerol, 2% SDS, 5% ␤-mercaptoethanol). Following the addition of ferritin carrier protein, the immunodepleted supernatant was acetone-precipitated overnight at Ϫ80°C and then resuspended in SDS-PAGE sample buffer.
Samples of immunoprecipitated or total protein (50 -80 g) in SDS-PAGE sample buffer were heated to 95°C for 3 min, then separated by SDS-PAGE, and transferred to nitrocellulose. The membranes were incubated (2 h at room temperature or overnight at 4°C) with the following primary antibodies: cyclin E (HE12) antibody from Santa Cruz Biotechnology; cyclin D1 (DCS6) antibody from Novocastra Laboratories, Newcastle-upon-Tyne, United Kingdom; p27 Kip1 (K25020) antibodies from Transduction Laboratories, Lexington, KY.; pRb (14001A) antibody from PharMingen, San Diego, CA; and phospho-Rb (Ser-780) antibody from New England Biolabs Inc., Beverley, MA. Following incubation (1 h at room temperature) with horseradish peroxidaseconjugated anti-mouse or anti-rabbit secondary antibody (Santa Cruz Biotechnology), specific proteins were visualized by chemiluminescence (PerkinElmer Life Sciences). Where the proteins of interest were of sufficiently different mobilities, membranes were incubated either sequentially or simultaneously with several primary antibodies.
Kinase Assays-The histone H1 kinase activity of cyclin E immunoprecipitates from 100 to 250 g of cellular protein was measured as previously described (24), using 10 g of histone H1 as substrate. The degree of background histone H1 phosphorylation, estimated from parallel control samples immunoprecipitated using beads without antibody, was typically near the limit of detection. Kinase activity of Cdk4 immunoprecipitates from 500 g of cellular protein was measured using 10 g of GST-pRb 773-928 fusion protein substrate as previously described (24). The degree of background phosphorylation in pRb phosphorylation assays was estimated from parallel control samples immunoprecipitated following blocking of specific antibody binding with the appropriate antigenic peptide. Following termination of kinase reactions, samples were incubated at 95°C for 3 min in SDS-PAGE sample buffer and separated by SDS-12% PAGE.
Image and Data Analysis-Images captured by PhosphorImager (Molecular Dynamics 445 SI; Molecular Dynamics, Sunnyvale, CA) or, for chemiluminescence, by densitometer scanning (Molecular Dynamics PDSI) of x-ray film, were quantitated using IP Lab Gel H analysis software (Signal Analytics, Vienna, VA). Quantitation of protein levels by this method was linear over the range of intensities measured. All figures were compiled using Deneba Canvas 5.0 software.
Flow Cytometry-Flow cytometric analysis was performed on a FACSCalibur (Becton Dickinson Immunocytometry Systems, San Jose, CA) using CELLQuest 2.0 (Becton Dickinson Immunocytometry Systems) software. The proportion of cells in the G 1 , S, and G 2 /M phases of the cell cycle were calculated from the resulting DNA histograms using ModFit LT analysis software (Verity Software House, Inc., Topsham, ME).

Overexpression of Cyclin D1 but Not Cyclin E Induces Progestin Resistance-
To determine the effect of cyclin overexpression on sensitivity to growth inhibition by progestins we used clonal derivatives of T-47D breast cancer cells that had been transfected with either cyclin D1 or cyclin E. The levels of cyclin expression achieved were ϳ5-fold greater than in parental or vector-transfected control cells for the cyclin D1-overexpressing line and Ͼ3-fold for the cyclin E overexpressing lines (Fig. 1A). This level of cyclin overexpression had no effect on the expression of other G 1 cyclins, nor on the levels of the CDK inhibitors p21 Cip1 and p27 Kip1 (Fig. 1A and data not shown).
Colony formation over 3 weeks of monolayer culture was used to test the sensitivity of the cyclin-overexpressing clonal cell lines to long term inhibition of proliferation following progestin treatment. Vector-transfected cells were profoundly inhibited by the presence of the progestin ORG 2058. Very few colonies were apparent after treatment with Ն1 nM ORG 2058, representing Ͼ99% inhibition of colony formation (Fig. 1, B and C). The concentration-dependence of the inhibition of proliferation was similar to that previously observed in the parental T-47D cells using different methodology (26). In marked contrast, the cyclin D1-overexpressing cell line (cyclin D1 17-1) was poorly inhibited by ORG 2058 treatment even at the highest concentration used, 100 nM (Fig. 1, B and C). In two cyclin E-overexpressing cell lines (cyclin E 17-2 and 17-3) ORG 2058 clearly inhibited colony formation, to ϳ25% of control at 1-100 nM, although significant numbers of colonies were still apparent (Fig. 1, B and C).
These data suggested that cyclin D1 overexpression attenuated the antiproliferative effects of progestins, but cyclin E overexpression had a more modest effect, still allowing substantial inhibition of proliferation. Measurement of estrogen and progesterone receptor levels by Western blot confirmed similar receptor expression in all the cell lines and demonstrated progestin-mediated down-regulation of both receptors in all cell lines (not shown), indicating that other progestin responses remained intact and thus that the apparent resistance was not simply due to defects in progesterone receptor expression or signaling.
To further confirm that the alteration in progestin sensitivity documented in Fig. 1 was due to cyclin overexpression rather than clonal variation, we infected T-47D cells with retroviruses expressing either GFP (control), cyclin D1, or cyclin E. Up to 50% of the original population was infected with virus. The resulting level of cyclin expression was similar to that in the cyclin-overexpressing clonal T-47D cell lines (Fig. 2) and was maintained over the 3 weeks of the experiment (not shown). Colony formation of cells infected with GFP-expressing virus was inhibited by ORG 2058 in a manner similar to other control T-47D derivatives, but retroviral expression of cyclin D1 resulted in marked resistance to ORG 2058-mediated growth inhibition (Fig. 2). Retroviral expression of cyclin E still allowed significant inhibition with the number of colonies reduced by ϳ60%, but again more colonies were apparent than in progestin-treated control cells. These experiments confirmed the conclusion obtained with the clonal cell lines overexpressing the cyclins i.e. that cyclin D1 overexpression markedly reduced sensitivity to progestin-mediated inhibition of proliferation, but cyclin E overexpression was much less effective.

Acute Effects of Progestin Treatment on Cyclin-overexpress-
ing T-47D Cells-As a basis for a more detailed examination of the mechanisms responsible for the progestin resistance documented in Figs. 1 and 2, responses in the first 4 days of treatment were characterized. Measurement of cell number over this timeframe revealed a modest decrease in relative cell number after 2 days of treatment of vector-transfected cells (Fig. 3A). The initial continuation of cell division in the presence of ORG 2058 is consistent with the early G 1 site of progestin action and consequent ability of cells past this point in the cell cycle to complete a round of DNA replication and mitosis (26). Between days 2 and 4, no further increase in cell number was seen following treatment with either 0.2 or 10 nM ORG 2058 (Fig. 3A). Growth curves of cyclin E-overexpressing cells were similar to control cell lines, with an increase in cell number over the first 2 days of treatment but no subsequent increase in the presence of 0.2 or 10 nM ORG 2058 (not shown). In contrast, while the response of cyclin D1-overexpressing cells after 2 days was very similar to that of vector-transfected cells (Fig. 3B), thereafter these cells continued to proliferate rather than becoming growth-arrested (Fig. 3B).
The relative numbers of vector-transfected cyclin D1 17-1 and cyclin E 17-3 cells after 96 h of treatment with 0.1-10 nM ORG 2058 indicated decreased proliferation following treatment with 0.2 or 10 nM ORG 2058 (Fig. 3C). The cyclin D1overexpressing cell line was relatively insensitive, and the cyclin E-overexpressing cell line was of intermediate sensitivity, consistent with the longer term data presented in Fig. 1. Similarly, measurement of S phase fraction after 48 h of ORG 2058 treatment yielded data (Fig. 3D) that paralleled those from the colony-forming assay: the S phase fraction of vector-transfected cells decreased to Ͻ20% of control, while the response was substantially attenuated in cyclin D1-overexpressing cells. The two cyclin E-overexpressing cell lines again were inhibited, although to a lesser degree than control cell lines.
After treatment with maximally effective concentrations of ORG 2058 (0.2 or 10 nM) the S phase fraction of vector-transfected control cells decreased from 15-18% to Ͻ5% after treatment for Ն30 h (Fig. 4A), consistent with previous data (6,26). The cyclin D1-overexpressing cell line, cyclin D1 17-1, had a higher initial S phase fraction (ϳ22%), although this did not result in a significant increase in proliferation rate (Fig. 3B), consistent with other studies of cyclin D1 overexpression (27)(28)(29). The S phase fraction was transiently reduced to 11-12% at 18 -30 h but maintained at ϳ18% thereafter (Fig. 4B), i.e. a value similar to that of untreated vector-transfected cells. This S phase value is consistent with the maintenance of proliferation despite the presence of ORG 2058 apparent in Fig. 3B. Both cyclin E-overexpressing cell lines also had an increased S phase fraction of 20 -22% during exponential growth, again consistent with previous data (28). This was reduced by Ͼ50% by treatment with 10 nM ORG 2058, with 0.2 nM ORG 2058 being slightly less effective (Fig. 4, C and D).
Maintenance of Cyclin Expression Following Progestin Treatment of Cyclin-overexpressing Cells-Western blots of lysates from progestin-treated cells indicated that cyclin D1 expression declined only slightly in cyclin D1 17-1 cells following treatment with either 0.2 or 10 nM ORG 2058 (Fig. 5B), in contrast with the ϳ60% decrease in cyclin D1 expression in either vector-transfected or cyclin E overexpressing cell lines (Fig. 5,  A, C and D). Cyclin E levels decreased by 50% after 24 -30 h in the cyclin D1-overexpressing cells but recovered to ϳ80% of control by 48 h (Fig. 5B). In neither cyclin E-overexpressing cell line was cyclin E expression significantly reduced following progestin treatment, but in both cyclin D1 expression decreased to an extent similar to that in control cells (Fig. 5, A, C  and D). Thus the level of the exogenously expressed cyclin D1 and E was maintained after progestin treatment.

Both Decreased Cyclin E Expression and p27 Kip1 Association with Cyclin E-Cdk2 Contribute to the Progestin-mediated Decrease in Cyclin E-Cdk2
Activity-To determine the effects of cyclin E overexpression on cyclin E-Cdk2 activity following progestin treatment, cyclin E-associated kinase activity was measured using an in vitro kinase assay (Fig. 6, A and B). The cyclin E-overexpressing cells displayed a basal level of cyclin E-Cdk2 activity ϳ2-fold higher than control cells. This was decreased by only ϳ60% following 10 nM ORG 2058 treatment of cyclin E-overexpressing cells (Fig. 6, B and C) compared with the ϳ90% decrease in control cells (Fig. 6, A and C). Because of the higher initial level of activity, despite this relative decrease the residual cyclin E-Cdk2 activity after progestin treatment was similar to that in untreated control cell lines (compare Figs  6, A with B). In contrast the S phase fraction of progestintreated cyclin E 17-3 cells was ϳ7%, much lower than the 16 -17% S phase fraction of untreated control cells (Fig. 4).
We also examined the phosphorylation of pRb, an endogenous substrate for cyclin E-Cdk2, in lysates of cyclin E-overexpressing cells after ORG 2058 treatment. Hyperphosphorylated, low mobility pRb remained readily detectable following progestin treatment of the cyclin E-overexpressing cell lines but was in low abundance after 32-48 h of ORG 2058 treatment of T-47D tTA-17 cells (Fig. 6, A and B). Since pRb is phosphorylated by both cyclin E-Cdk2 and cyclin D-Cdk4, progestin effects on Cdk4 activity in cyclin E-overexpressing cells were examined. Cdk4 activity, measured in an in vitro assay using pRb substrate, was decreased in both vector-transfected and cyclin E-overexpressing cells to near-baseline levels by 48 h (Fig. 6D). Similarly, specific pRb phosphorylation at Ser-780, a site targeted by cyclin D-Cdk4 but not cyclin E-Cdk2 (30), decreased to essentially undetectable levels after 48 h of treatment of either vector-transfected or cyclin E-overexpressing cells (Fig. 6, A and B). Thus Cdk4 activity was very low following ORG 2058 treatment of either vector-transfected or cyclin E-overexpressing cells, suggesting that the kinase activity responsible for the maintenance of pRb phosphorylation in the cyclin E-overexpressing cells was likely cyclin E-Cdk2. This appears to be sufficient for partial resistance to progestin treatment but not proliferation at control levels.
We previously postulated that the decrease in cyclin E-Cdk2 activity following progestin treatment resulted from both decreased cyclin E expression and increased association of p27 Kip1 with the remaining cyclin E-Cdk2 complexes (6,7). To determine the role of p27 Kip1 association with cyclin E in the decrease in cyclin E-Cdk2 activity in cyclin-overexpressing cells, p27 Kip1 immunoprecipitates were Western blotted in parallel with the depleted supernatant. In vector-transfected cells few cyclin E-p27 Kip1 complexes were present in vehicle-treated cells and cyclin E was not significantly depleted by p27 Kip1 immunoprecipitation (Fig. 7A). Cyclin E overexpression increased the number of cyclin E-p27 Kip1 complexes present in vehicle-treated cells. However, in both cell lines increased cyclin E-p27 Kip1 association accompanied the decrease in cyclin E-Cdk2 activity following progestin treatment (Fig. 7, A and B). After ORG 2058 treatment of vector-transfected cells, almost all the cyclin E was associated with p27 Kip1 , so that little remained in the supernatant after p27 Kip1 immunoprecipitation (Fig. 7A). In contrast, the overexpressed cyclin E was not significantly depleted (Fig. 7B). Although cyclin E-p21 Cip1 complexes were much more abundant in the cyclin E-overexpress- ing cell lines than in control cell lines, they remained in lower abundance than the cyclin E-p27 Kip1 complexes. Immunoprecipitation of both p27 Kip1 and p21 Cip1 did not deplete significant amounts of cyclin E from the cyclin E-overexpressing cells, nor was any change in p21 Cip1 association with cyclin E observed following progestin treatment (not shown). Thus increased p27 Kip1 association likely accounts for the 60% decrease in activity following ORG 2058 treatment of cyclin E-overexpressing cells.
Cyclin D1-overexpressing Cells Proliferate in the Presence of ORG 2058 Despite Low Cyclin E-Cdk2 Activity-Measurement of cyclin E-associated kinase activity following progestin treatment of cyclin D1-overexpressing cells revealed an initial decrease of similar magnitude to that observed in control cells i.e. to below 25% of that observed in vehicle-treated cells (Fig. 8A). However, by 48 h kinase activity had recovered to ϳ50% of control (Fig. 8A), paralleling increased cyclin E abundance at that time point (Fig. 5B). Low mobility (hyperphosphorylated) pRb was clearly present following 48 h of progestin treatment of D1 17-1 cells but not vector-transfected cells (Fig. 8B), consistent with higher CDK activity in the former cells. Despite the maintenance of cyclin D1 levels following ORG 2058 treatment of cyclin D1-overexpressing cells, specific phosphorylation of pRb at Ser-780 decreased, although it remained at a level significantly above vector-transfected cells treated in parallel (Fig. 8B). Measurement of Cdk4 activity using in vitro kinase assays confirmed that decreased cyclin D-Cdk4 activity resulted from ORG 2058 treatment of cyclin D1-overexpressing cells but that the decrease in Cdk4 activity was attenuated compared with the Ͼ90% decrease in vector-transfected control cells at 48 h (Fig. 8, B and C). The basal level of kinase activity in these cells was higher than in vector-transfected cells (Fig.  8B), such that the residual activity represents ϳ70% of the level in exponentially proliferating control cell lines.
Recent data showing that lack of p27 Kip1 can rescue the defects in proliferation resulting from the lack of cyclin D1 (22,23) argue that sequestration of p27 Kip1 is important for the physiological role of cyclin D1. However, the similar initial decrease in cyclin E-Cdk2 activity in control and cyclin D1overexpressing cell lines raised the possibility that p27 Kip1 redistribution was still occurring in the latter cell line. In vehicle-treated D1 17-1 cells, as in vector-transfected cells, p27 Kip1 immunoprecipitation did not significantly deplete cyclin E, indicating that little cyclin E was associated with p27 Kip1 before progestin treatment. ORG 2058 treatment of cyclin D1-overexpressing cells increased p27 Kip1 -cyclin E association at both 33 h, when cyclin E-Cdk2 activity was at its minimum, and 48 h, when it had partially recovered (Fig. 7C). At 33 h, the majority of cyclin E was depleted by p27 Kip1 immunoprecipitation but at 48 h a minority was bound to p27 Kip1 . Immunoprecipitation of both p27 Kip1 and p21 Cip1 confirmed that little cyclin E was bound to p21 Cip1 after ORG 2058 treatment. Thus much of the cyclin E was not bound to either inhibitor, consistent with the significant cyclin E-Cdk2 activity present at that time point. The increased availability of p27 Kip1 for cyclin E binding after ORG 2058 treatment was likely due to decreased cyclin D3-Cdk4 complex abundance since there was no detectable decrease in p27 Kip1 -cyclin D1 association (not shown). Overall these data indicated that cyclin D1 overexpression resulted in progestin resistance independent of effects on p27 Kip1 availability to bind cyclin E-Cdk2. DISCUSSION This study has used T-47D breast cancer cells overexpressing cyclin D1 or cyclin E as a means of investigating the role of these cyclins in progestin inhibition of proliferation. Overex-pression of cyclin D1 induced substantial progestin resistance, but cyclin E overexpression did so weakly. This was apparent both in cell lines stably transfected with each cyclin and following retroviral expression of the cyclins. The greater effectiveness of cyclin D1 in inducing progestin resistance does not appear to simply result from higher relative expression, since cell lines overexpressing either cyclin displayed a similar increase in S phase fraction compared with control cell lines, with the implication that the overexpressed cyclins were at functionally equivalent levels. These data complement our previous demonstration that cyclin D1 induction in progestin-treated cells results in re-initiation of cell cycle progression (6) and indicate that regulation of cyclin D1 is a critical element of progestin inhibition of proliferation in breast cancer cells.
The antiproliferative effects of progestins are often viewed as antiestrogenic, and there are a number of similarities in the molecular mechanisms by which antiestrogens and progestins inhibit breast cancer cell proliferation. Both types of compound decrease cyclin D1 expression, consequently decreasing cyclin D1-Cdk4/6 activity and triggering redistribution of CDK inhibitors which in turn contributes to decreased cyclin E-Cdk2 activity (5,6,31,32). Antiestrogen-mediated arrest of breast cancer cells can be overcome by cyclin D1 induction (33,34), but constitutive cyclin D1 expression does not lead to long term antiestrogen resistance (35). 2 The differences in response to cyclin D1 overexpression indicate important differences between the mechanisms of action of progestins and antiestrogens, i.e. that antiestrogens activate growth-inhibitory pathways that are capable of counteracting the effects of cyclin D1 overexpression, whereas progestins do not.
Our previous examination of possible mechanisms for progestin inhibition of cyclin E-Cdk2 activity showed that after progestin treatment essentially all the cyclin E present was in complexes also containing p27 Kip1 and therefore inactive (7). These data demonstrate a key role for p27 Kip1 association in the decrease in activity (6). Since much of the cellular complement of cyclin E in T-47D cells is in inactive complexes (7,36) and there is a poor relationship between cyclin E levels and cyclin E-Cdk2 activity in a series of breast cancer cell lines (36), the contribution of the concomitant decrease in cyclin E abundance was unclear. The present study demonstrates that decreased cyclin E abundance and p27 Kip1 association with cyclin E-Cdk2 make similar contributions to decreased cyclin E-Cdk2 activity since maintenance of cyclin E levels attenuated the response by approximately half.
Progestin-treated cyclin E-overexpressing cells maintained cyclin E-Cdk2 activity at levels similar to those in exponentially proliferating control cells. This was, however, not sufficient for S phase entry at the same rate as in untreated control cells, which displayed a much higher S phase fraction. A likely explanation is that the accompanying decrease in cyclin D-Cdk4 activity impaired cell cycle progression, although this is perhaps unexpected given evidence from a number of model systems that cyclin E-Cdk2 activity can compensate for lack of cyclin D1-Cdk4 activity. For example, expression of cyclin E can overcome the G 1 block resulting from expression of the cyclin D-Cdk4/6-specific inhibitor p16 INK4a or unphosphorylated pRb (37)(38)(39)(40). Furthermore, mice in which the coding sequence of the cyclin D1 gene has been replaced by that of cyclin E do not display the proliferative defects in the retina and mammary gland resulting from lack of cyclin D1, indicating that cyclin E can functionally substitute for cyclin D1 in these tissues (41). However, inhibition of cyclin D1 expression or function by antisense or antibody microinjection, by INK4a family inhibitor expression or by specific chemical inhibitors of Cdk4 (42-45) is sufficient to inhibit cell cycle progression. One FIG. 8. CDK activity, pRb phosphorylation, and cyclin E-p27 Kip1 association following ORG 2058 treatment of cells overexpressing cyclin D1. T-47D cells constitutively expressing cyclin D1 (D1 17-1) or vector-transfected control cells were treated with progestin (ORG 2058, 10 nM). A, the kinase activity of cyclin E immunoprecipitates from whole cell lysates was measured using histone H1 substrate. Quantitation of data from two separate experiments using D1 17-1 cells is presented as mean Ϯ range. Control data presented in Fig. 6 are reproduced for comparison. B, whole cell lysates were Western blotted for total pRb protein and phosphorylation at Ser-780 (Rb-P-Ser 780), a residue phosphorylated by cyclin D1-Cdk4 and not cyclin E-Cdk2. The kinase activity of Cdk4 immunoprecipitates from whole cell lysates was measured using GST-pRb substrate, with the background pRb phosphorylation estimated by the addition of antigenic peptide (ϩ pep) to parallel control immunoprecipitates. C, quantitation of data from two separate experiments with each cell line, presented as mean Ϯ range. interpretation of these data consistent with our observations is that supra-physiological levels of cyclin E-Cdk2 are necessary to overcome a lack of cyclin D1-Cdk4 activity, i.e. that the residual level of cyclin E-Cdk2 activity following progestin treatment is not sufficient to compensate for the reduction in cyclin D-Cdk4 activity.
Cdk4 activity was reduced by progestin treatment of cyclin D1-overexpressing cells in the absence of a significant decrease in cyclin D1 abundance. This response may be a consequence of p18 INK4c induction, since we have previously shown a transient induction of INK4c mRNA 6 -18 h after progestin treatment (7), or may result from decreased cyclin D3 levels (5-7) and consequent decrease in cyclin D3-Cdk4 activity. Despite the decrease in Cdk4 activity, these cells maintained an S phase fraction and proliferation rate only slightly lower than that of untreated, exponentially proliferating control cells. Thus, the residual cyclin D-Cdk4 activity after progestin treatment of cyclin D1-overexpressing cells, representing ϳ70% of that in exponentially proliferating control cells, was sufficient to allow proliferation to continue in the presence of an ϳ50% decrease in cyclin E-Cdk2 activity. This contrasts with the failure of cyclin E-Cdk2 activity at a level similar to that in untreated control cells to compensate for a lack of cyclin D-Cdk4 activity in progestin-treated cyclin E-overexpressing cells, pointing to functional differences between the two cyclins despite the ability of cyclin E to substitute for cyclin D1 in some model systems (41).
While the ability of cyclin D1 to activate Cdk4/6 has been well studied as a mechanism for its roles in cell cycle progression and oncogenesis, it has additional functions that may be at least as important. These include CDK-independent interactions with transcription factors including a Myb-like protein (DMP1) (46,47), STAT3 (48) and the estrogen and androgen receptors (49 -51). Recent genetic evidence implicates the ability of cyclin D1 to sequester p27 Kip1 as an essential physiological function for cyclin D1 and has focused attention on this role. Inactivation of the gene encoding p27 Kip1 corrects the defects in retinal and mammary development resulting from lack of cyclin D1 (22,23), and substantially overcomes the delay in cell cycle re-entry after serum starvation of fibroblasts lacking Cdk4 (52). These data suggest that increased p27 Kip1 availability rather than lack of cyclin D1-Cdk4 activity leads to proliferative defects in the retina and mammary gland of mice lacking cyclin D1. In contrast with the prediction of this conclusion, cyclin D1 overexpression in progestin-treated cells prevented inhibition of proliferation but not the decrease in cyclin E-Cdk2 activity resulting from p27 Kip1 association with these complexes. Thus, this effect of cyclin D1 is not dependent on the ability of cyclin D1 to sequester p27 Kip1 , providing evidence for a critical function for cyclin D1 other than as a high-capacity sink for CDK inhibitors. The observation that cyclin D1 is necessary for cell cycle progression in vitro only in cells with functional pRb (53) points to activation of Cdk4/6 and consequent phosphorylation of pRb as the most likely alternative.
The mechanism by which progestins regulate cyclin D1 and E expression remains unknown, although the correspondence between changes in mRNA and protein levels (6) argues that it is via regulation of mRNA abundance. However, the decreases in expression implicated in inhibition of proliferation do not occur within the first 6 -12 h of progestin treatment and thus are not likely to be direct effects of progestin. Identifying the pathway(s) responsible for regulation of these cyclins, particularly cyclin D1, will offer further insight into the mechanisms by which progestins inhibit cell proliferation.