MCF7 cells were grown in Dulbecco’s modified eagle medium (DMEM) with phenol-red, supplemented with 10% fetal calf serum (FCS), penicillin (100 units/mL), streptomycin (100 μg/mL) and 2 mmol/L glutamine. LCC1 and LCC9 cells (kindly provided by Prof. Robert Clarke, Vincent T. Lombardi Cancer Research Center, Georgetown University Medical School, Washington, D. C., USA) [9, 10] were maintained in phenol-red free containing DMEM supplemented with 5% dextran activated double charcoal stripped FCS (DCSS) + additives. MDA-MB-231 cells were obtained from American Type Culture Collection (ATCC) and were grown in the same medium as MCF7 cells. All cell lines were maintained in a humidified atmosphere at 37°C and 5% CO₂ and confirmed mycoplasma free.

Cells were counted using a Beckman Z2 Coulter counter. Log-phase cells were seeded into 24-well tissue culture plates (at cell densities optimized between 2–5 × 10⁴ cells mL⁻¹) and treated with 5% DCSS phenol red-free DMEM containing 17β-estradiol (E₂, 1 nmol/L), transforming growth factor alpha (TGFα, 1 nmol/L) or heregulin-beta (HRGβ, 1 nmol/L). Groups of wells were trypsinized on days 0, 3, 5, and 7, and cells were counted.

Cells were seeded into 96-well plates in a humidified atmosphere of 5% CO₂/95% O₂ at 37°C and cell number was determined using the sulphorhodamine B (SRB) assay [44]. MCF7 cells were seeded in 5% DCSS phenol red-free DMEM 48 h prior to treatment while resistant cell lines were seeded 24 h prior to treatment. Cells were then treated as indicated and the plates were incubated. The experiment was stopped at the times specified by the addition of 50 μL/well 25% trichloroacetic acid, 1 h, 4°C. Plates were then washed in water and when dry, 50 μL/well 4% SRB was added for 30 min at room temperature. Plates were washed in 1% acetic acid, left to dry and 150 μL/well 1.5 mol/L tris was added. After 1 h, the optical density of each plate at 540 nm was obtained using a Perkin Elmer plate reader. All ligands and reagents were obtained from Sigma-Aldrich unless otherwise stated. Seliciclib was obtained from Cyclacel Pharmaceuticals, Inc, Dundee and pertuzumab was kindly provided by Roche Diagnostics, Penzberg.

Cells grown in DMEM/FCS were treated with either 1 nmol/L E₂, TGFα or HRGβ for 72 h or in studies with seliciclib were treated with 20 μmol/L seliciclib for 24 h (with 0.1 nmol/L E₂ ). For cell cycle analysis, cells were resuspended in 100 μL of citrate buffer and stored at –20°C. Cells were subjected to flow cytometric DNA analysis using a FACSCalibre™ flow cytometer (Beckton Dickinson) as described previously [45]. Apoptosis was measured using the TACS annexin V-FITC kit (R&D Systems).

Cells were treated as indicated with E₂, TGFα, HRGβ or seliciclib in DMEM media containing 5% DCSS. Cell lysates were prepared as previously described [46]. Lysates were electrophoretically resolved on 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, with the exception of anti-human Rb where a 6% gel was used) and transferred to immobilon-P membranes. After transfer, membranes were probed overnight at 4°C with the appropriate primary antibody. All antibodies used were from Cell Signaling Technology, except for anti-human Rb (BD Pharmingen); anti-total ERα (Santa Cruz Biotechnology), anti-actin (Oncogene Research Products), anti-p53 (Oncogene), anti-CDK2 (Upstate Biotechnology, Inc); anti-tubulin (Abcam) and were used at 1:1,000 with the exception of actin (1:120,000). Immunoreactive bands were detected using enhanced chemiluminescent reagents (Roche) and hyperfilm enhanced chemiluminescence (ECL) film (Amersham^TM). Integrated optical density (IOD) absorbance values were obtained by densitometric analysis using a gel scanner and analyzed by “Labworks^TM” gel analysis software (UVP Life Sciences).

For immunoprecipitation experiments, cells were lysed as described above and a volume of lysate containing 100 μg of protein was agitated overnight at 4°C with 1–10 μL of relevant antibody. Following overnight incubation, protein-G-agarose beads were washed in lysis buffer and 50 μL of the bead slurry was added to the lysate/antibody mix. The samples were then incubated for 3 h as before. Beads were washed three times in ice-cold lysis buffer. The lysate/antibody/bead slurry mixture was centrifuged for 2 min at 2,000 rpm (4°C) at the end of the three-hour incubation and the supernatant was collected using a syringe. Lysis buffer (minus protease inhibitors, 500 μL) was then added to the solution followed by another spin at 2,000 for 2 min. The supernatant was collected again, and the wash was repeated twice more as described. Loading buffer (20 μL) was added to the bead solution and the sample was heated at 95°C for 5 min. Samples were then centrifuged at 13,000 rpm for 1 min and the supernatant was removed for loading onto a polyacrylamide gel for western blot analysis.

AKT small interfering RNAs (siRNAs; 2.5 μL) were diluted in 250 μL of Opti-MEM in 6-well plates. Lipofectamine RNAiMAX (Invitrogen, 7.5 μL) was then added to each well, mixed gently and incubated for 10–20 min at room temperature. Cells were harvested as previously described and all cell lines were diluted to a concentration of 1.7 × 10⁵ cells/mL in 5% DCSS DMEM (-phenol red). The cell dilution (1.5 mL) was added to the RNAi-lipofectamine complexes and incubated for 48 h before RNA collection. For protein extraction plates, they were incubated for 72 h. siRNAs were used at a concentration of 50 nmol/L from a stock solution of 20 nmol/L. siRNA sequences were as follows: AKT1: 5’ACCTGACCAAGATGACAG; AKT2: 5’AAGTGGGTCCGCTGGT; AKT3: 5’AGGAGGTACAAGCTTTTTA (Applied Biosystems, UK); negative control siRNA (Upstate Biotechnology, Inc, M-003401).

Extraction of total RNA from whole cells was performed using tri-reagent (Sigma, Poole, Dorset) as per manufacturer instructions. RNA concentration was measured using a spectrophotometer. QuantiTect SYBR Green reverse transcription polymerase chain reaction (RT-PCR) kit (Quiagen Cat. #204243) was used according to the manufacturer’s instructions for one-step RT-PCR in a total of 15 μL reaction volumes including 10 μmol/L each primer and 40 ng RNA. The RotorGene PCR cycle conditions were: reverse transcription step 50°C for 30 min, Taq activation step 95°C for 15 min, PCR 40 cycles of 94°C for 15 s, 57°C for 30 s, and 72°C for 30 s. After a final extension of 72°C for 60 s, a PCR product melt curve was performed 60°–99°C. The following primer pairs were used:

β-actin: GATGGAGCCGCCGATCCACACGG, CTACGTCGCCCTGGACTTCGAGC

The vector pSUPER.gfp+neo (VEC-pBS-0006) into which was inserted a sequence encoding an effective siRNA targeting p53 isolated from the vector pSUPER.p53 (VEC-p53.0001) was used. Sequence was checked by direct sequencing through the siRNA expression cassette and validated by lipofectin mediated transfection of the relinearized pSUPER.p53.gfp+neo vector into the MCF7 cell line which resulted in effective p53 knockdown in these cells. The p53 knockdown status in these cells in culture was maintained by growth in 400 ug/mL geneticin. These cell lines are referred to as MCF7-p53-KD.

Groups were compared by analysis of variance (ANOVA) followed by the Tukey-Kramer multiple comparison test.

The LCC1 and LCC9 cell lines represent models of increasing degrees of estrogen insensitivity relative to wild-type estrogen-sensitive MCF7 cells. We first assessed whether the reduced response to estrogen was also associated with changed sensitivity to the HER activating ligands TGFα and HRGβ. These growth factors activate signaling via HER dimerization, predominantly HER1 homodimers or HER1/HER2 heterodimers for TGFα and HER3/erbB2 for HRGβ. The effects of E₂ on growth and cell cycle distribution were compared with those of TGFα and HRGβ. MCF7 cells grew poorly in DCSS FCS but were growth stimulated by treatment (1 nmol/L) with E₂, TGFα, and HRGβ (Figure 1A). In contrast, the LCC1 and LCC9 cell lines proliferated in the absence of ligand stimulation. LCC1 cells demonstrated a minimal growth response to E₂, and were insensitive to TGFα and HRGβ compared to wild type MCF7 cells, while LCC9 cells were insensitive to all treatments (Figure 1A).

Consistent with the growth effects in MCF7 cells, cell cycle analysis demonstrated that 1 nmol/L E₂, TGFα, and HRGβ increased the percentage of cells in S- and G2/M phases (Figure 1B). In contrast, LCC1 and LCC9 cells had a much higher percentage of cells initially in S-phase in the absence of hormone or growth factors and this percentage was not increased by their addition (Figure 1B). These data indicate collateral resistance between E₂ and the growth factors in these progressively resistant cell lines. Protein expression levels of HER2 were lower in LCC1 and LCC9 cells relative to MCF7 cells while HER4 expression was increased. HER3 expression was similar across the 3 cell lines. EGFR protein was below the limit of detection for the cell lines but shown to be present in MDA-MB-231 cells (as a positive control) (Figure 2).

We next investigated whether the AKT pathway was differentially activated in the sensitive and resistant cell lines (Figure 3). Expression of P-AKT(Ser473) was significantly increased in both LCC1 and LCC9 cells relative to MCF7 cells in the absence of ligands, with levels of total AKT (T-AKT) unchanged (Figure 3A, 3B). Addition of either TGFα or HRGβ increased P-AKT expression in all three lines while E₂ had no effect (Figure 3A, 3B).

The anti-HER2 antibody pertuzumab could reverse the growth stimulations produced by either TGFα or HRGβ in MCF7 cells as previously reported [47], however it was ineffective in the LCC1 and LCC9 cell lines (Figure 3C), again indicating a changed growth dependence on these pathways in the resistant cell lines.

Since AKT pathway activation was increased in the resistant cell lines, we next investigated whether there was differential expression of individual isoforms of AKT in these cell lines. Expressions of AKT1 and AKT2 mRNA were significantly lower in LCC1 and LCC9 cells in comparison to parental MCF7 cells (Figure 4A). In contrast, AKT3 mRNA expression was elevated in LCC1 and LCC9 cells. Assessment of protein expression by western analysis indicated similar expression of AKT1 and AKT2 across the three cell lines, but increased expression of AKT3 in LCC1 and LCC9 cells relative to MCF7 cells (Figure 4B). Despite these differences in individual isoforms, the total level of the three isoforms combined was similar in the three cell lines (Figure 4B).

To assess phosphorylation levels of specific AKT isoforms, immunoprecipitation was performed using AKT isoform specific antibodies. The immunoprecipitates were then probed with anti-P-AKT(Ser473) antibody to reveal which isoforms were activated in each of the cell lines. In the three cell lines, AKT3 is highly phosphorylated (Figure 4C). The level of AKT3 phosphorylation progressively increases in LCC1 and LCC9 in comparison to MCF7, and levels of AKT2 phosphorylation are also increased (Figure 4C). Activation of signaling downstream of AKT was then investigated. In LCC1 and LCC9 cells, both phospho-mTOR and phospho-S6 expression were increased, consistent with enhanced AKT activation, although total mTOR is already increased in LCC1 and LCC9 cells (Figure 4D). Furthermore, ERα was phosphorylated at its Ser167 residue consistent with the pathway proposed by Campbell et al. [48] (Figure 4E). We have previously reported ERα expression to be increased in LCC1 and LCC9 cells [49].

To test the effects of inhibiting individual AKT isoform expression, an siRNA approach was adopted. Specificity of mRNA knockdown was first demonstrated using LCC9 cells (Figure 5A) and this cell line was used as these cells express all 3 isoforms (Figure 4B). Cell growth of parental MCF7 cells were sensitive to AKT1 and AKT2 specific siRNAs as cell number is reduced while AKT3 siRNA had little effect (Figure 5B). LCC1 cells were sensitive to all three AKT siRNAs as they all reduced growth while LCC9 cells were also inhibited by all three siRNAs but less so with AKT2 siRNA (Figure 5C, 5D).

Estrogen binding to ERα results in phosphorylation of Ser118 of the receptor and this can also be achieved by growth factor signaling resulting in ligand-independent activation [50]. The effects of E₂ and TGFα on phospho-ERα(Ser118) expression in the cell lines were compared over a time-course (Figure 6). In MCF7 cells, both E₂ and TGFα markedly increased P-ERα(Ser118) expression over several hours (Figure 6A, 6B). In LCC1 cells, E₂ increased P-ERα(Ser118) expression after 5 min for a short period of time but this rapidly reversed while TGFα had no effect on expression in this cell line (Figure 6C). Neither E₂ nor TGFα increased P-ERα(Ser118) expression in LCC9 cells (Figure 6D).

We next investigated whether stepwise acquisition of increasing endocrine resistance is accompanied by a progressive loss of control of the estrogen-mediated cell cycle drive. Treatment with E₂ leads to increased MYC mRNA expression in MCF7 cells after 6 h which is maintained at 24 h (Figure 7A). By comparison, levels of MYC in LCC1 and LCC9 cells are already elevated relative to MCF7 in the absence of E₂ stimulation and are not induced further by E₂ (Figure 7A). These differences in MYC expression are also reflected at the protein level (Figure 7B, 7C).

Altered basal expression of MYC in LCC1 and LCC9 and E₂ induction of MYC in MCF7 are reflected in progressive changes in expression and regulation of a number of cell cycle regulators (Figure 8A). Levels of cyclin A1 mRNA are elevated in LCC1 and LCC9 relative to MCF7 cells and are strongly increased in both MCF7 and LCC1 by E₂ but not in LCC9 cells consistent with growth response. Expression of cyclin D1 (but not cyclin E) is elevated in LCC9 but not LCC1 cells and E₂ increases expression of cyclin D1 in both E₂-growth responsive MCF7 and LCC1 cells. Expression of the CDK1 and CDK2 (but not CDK4) are increased on exposure to E₂ in MCF7 and LCC1 cells but decreased in LCC9 which has elevated basal expression. Expression of the cell cycle inhibitors p21, p27, and p15 are reduced in the MCF7 and LCC1 cell lines on exposure to E₂. Expression of p21 and p27 are highest in untreated MCF7 cells and the most marked repression of p21 by E₂ is seen in this cell line.

Elevated expression of CDK2 protein in LCC1 and LCC9 cells was confirmed by western blotting (Figure 8B). Although E₂ does not markedly affect its expression in MCF7 cells, a profound difference was seen in its activation (phosphorylation of Thr160) when cells were exposed to E₂. LCC1 cells have elevated basal activation of CDK2 (compared to MCF7 cells) but this was further increased by E₂, while the elevated level of phosphorylated CDK2 seen in LCC9 was not further increased when cells were exposed to E₂. Seliciclib reduced both the levels of total CDK2 and phosphorylated CDK2 in MCF7 cells. While seliciclib increased CDK2 phosphorylation in LCC1 cells in the absence of E₂, it had little effect on total CDK2 in these cells and no effects in LCC9 cells.

The target for active cyclin E/CDK2 is Rb and therefore its hyper-phosphorylation status was evaluated after treatment with E₂ and/or seliciclib (Figure 8C). Both LCC1 and LCC9 cells showed increased basal levels of ppRb compared to MCF7 and exposure to E₂ resulted in decreased hyperphosphorylation in these cells (Figure 8C). Conversely, in MCF7 cells, exposure to estrogen leads to a very significant increase in ppRb. Seliciclib reduced hyperphosphorylation in both the presence and absence of E₂. Collectively, these data indicate that in MCF7 cells, estrogen modulates cell cycle regulators. In LCC1 cells some elements of estrogen-control are lost, e.g., MYC induction and an increase in ppRb whilst some estrogen responsiveness is retained (phosphorylated CDK2, cyclin A1), consistent with a residual degree of estrogen cell growth drive. The LCC9 cell line shows no estrogen-mediated growth response and, consistent with this, estrogen has no impact on these signaling events.

These cell lines were next treated with seliciclib; all three cell lines were inhibited to a similar degree (EC₅₀ of 11 μmol/L for MCF7; 14 μmol/L for LCC1 and 12.5 μmol/L for LCC9) (Figure 9A). Cell cycle studies indicated that all three cell lines showed a similar percentage of cells accumulating in G2/M on exposure to seliciclib (Figure 9B).

MCF7 cells have the highest basal level of apoptosis but seliciclib increased apoptosis to a comparable extent in all three cell lines as determined by measurement of annexin V (Figure 9C) and PARP cleavage (Figure 9D).

The expression and activation profiles of p53 are altered in both LCC1 and LCC9 cells compared with MCF7 cells (Figure 10A, 10B). Low basal p53 expression in MCF7 is increased by estrogen and seliciclib to a level comparable with that found in unstimulated LCC1 and LCC9 cells. No increase in p53 expression was seen on exposure of these latter cell lines to either seliciclib or estrogen. Seliciclib increased phosphorylation at the Ser46 residue of p53 in MCF7 cells and this is augmented by estrogen (Figure 10A). Phosphorylation of this residue is important in regulating the ability of p53 to induce apoptosis [37]. In LCC1 cells, seliciclib alone does not increase phosphorylation of Ser46 but is able to do this in the presence of estrogen. In LCC9 cells, seliciclib does not increase the level of Ser46 phosphorylation. Expression of p21, PUMA, and p53AIP1 mRNAs are strongly increased in MCF7, LCC1 and LCC9 cells by seliciclib (Figure 10B), but induction of p53AIP1 mRNA after 6 h is much greater in MCF7 than in the other cell lines (Figure 10B). This is consistent with earlier observations that p53AIP1 expression is specifically induced when p53 is phosphorylated on Ser46 in response to seliciclib [51].

To further evaluate the role of p53 in mediating the cellular effects of seliciclib, we stably expressed p53 targeting siRNA in MCF7 cells resulting in knockdown of p53. We compared clones with p53 knockdown with vector controls and wild type cells (Figure 11). We observed that induction of p21, PUMA and p53AIP1 by seliciclib is p53-dependent, and clones lacking p53 expression show corresponding reduction in induction of these known p53 transcriptional targets. Loss of p53 also results in a reduced level of MDM2 mRNA (Figure 11C) but does not affect the levels of protein expression (Figure 11A). In MCF7 parent and vector control cells, exposure to seliciclib results in increased phosphorylation of MDM2 Ser166, an event normally mediated by AKT which allows MDM2 to gain nuclear entry and thereby regulate levels of p53 [52]. When p53 expression is reduced by stable siRNA expression in MCF7 cells, we see loss of this activation. Together, these observations imply a reduction in turnover of MDM2 following siRNA mediated p53 knockdown. Despite these effects, the degree of seliciclib-induced apoptosis as demonstrated by PARP cleavage in these clones does appear independent of the degree of siRNA mediated p53 knockdown (Figure 11A). Cell growth experiments using a range of seliciclib concentrations shows that p53 status does not alter the overall growth inhibition of MCF7 derived clones by seliciclib (Figure 11D).