To reduce the risk that experimental results may be influenced by cell heterogeneity, we subcloned MCF-7 cells by limiting dilution. All clones analyzed (18 in total) ceased to proliferate in serum- and estrogen-free medium, and responded to mitogenic stimulation by E2 and insulin. Four clones were further analyzed and found to express the ER and PR (inducible by E2). One of these clones was used in all subsequent experiments.
1. The kinase function of Akt is required for the E2-dependent cell cycle progression.
In our previous work we showed that depletion of Akt1 and 2 prevented the mitogenic signaling by E2 in the MCF-7 cells. At the same time, E2 stimulation failed to induce the activating phosphorylation of Akt on Ser 473. This opened the possibility that Akt may have a function unrelated to its kinase activity, as has been suggested in a different context [16
]. We therefore produced Akt1 and Akt2 expression vectors carrying silent mutations in the sequence targeted by shRNA, as well as in the kinase domain. As reported by Nakatani et al. [20
] and Zinda et al. [21
], Akt3 is not expressed in the MCF-7 cells. We tested these constructs for their capacity to “rescue” the mitogenic action of E2 in cells exposed to shRNA targeting Akt1 and 2. The end-point was the activation of the promoter of the cyclin A gene cloned upstream of a luciferase coding sequence, as an indicator of late G1 phase.
When cells were transfected with the shRNA-expression vector Akt (1
2) directed against a sequence shared by Akt1 and 2 mRNAs, the activation of the cyclin A promoter by E2 was blocked and co-transfection of expression vectors coding for shRNA-resistant, wild type kinase variants of the Akt isoforms (Akt1R, Akt2R) restored the cyclin A promoter activation as revealed by the induction of luciferase. Akt2 appeared to be more efficient to restore the full mitogenic effect of E2 than Akt1 (Figure A).
Figure 1 Akt kinase activity is required for the E2 -induced reinitiation of the cell cycle progression.A: MCF-7 cells seeded in 35 mm dishes were transfected with shRNA Akt (1+2) (0.5 μg per dish) together with expression vectors of (more ...)
Next we compared the wild-type, shRNA-resistant Akt constructs with their kinase-dead counterparts Akt1R/KD and Akt2R/KD. In these experiments, the inclusion of the KD variants resulted in a reduced transfection efficiency documented by the diminished activity of the indicator β-galactosidase. Therefore, we treated groups of dishes with E2 and kept other groups of dishes as controls (in the serum-free medium supplemented with ICI 182780), to calculate the induction factor for the luciferase/β-galactosidase ratios. The results showed that with the kinase-dead mutants, there was only a partial restoration of luciferase induction (Figure B) as compared with the wild-type Akt2R used as a positive control. The results of these experiments demonstrate that the kinase function of exogenous Akt is required for efficient rescue of E2-inducible cell cycle progression when endogenous Akt is knocked down.
2. Cells deprived of serum in the absence of ICI 182780 continue to express cell cycle markers.
The arrest of proliferation by depriving the MCF-7 cells of exogenous mitogens was characterized by changes in the cell contents of certain markers of mitogenic signaling of the cell cycle (Figure ).
Figure 2 Changes in the levels of cell cycle-related and signaling proteins during serum- and E2 -starvation. MCF-7 cells were seeded in 60 mm dishes and allowed to attach overnight. They were then placed in serum- and phenol red-free medium containing or not (more ...)
Interruption of the mitogenic signaling is illustrated by the changes in the phosphorylation status of the Rb protein, a substrate of cyclin-dependent kinases and a modulator of late G1-phase gene expression. After incubation for 24 h or longer in serum and phenol red-free medium containing ICI 182780, Rb was dephosphorylated, whereas a significant fraction of Rb remained phosphorylated when ICI 182780 was omitted. This indicates that the suppression of ER by the antiestrogen is required for an efficient block of the induction of cyclin-dependent kinases. This conclusion is also supported by the presence of a residual cyclin A in cells deprived of serum in the absence of the antiestrogen whereas in the presence of the antiestrogen, the cyclin A signal is nearly eliminated (Figure ).
The cdk inhibitory proteins p21WAF1/CIP1 and p27 accumulated in cells deprived of serum. Whereas the addition of ICI 182780 in the starvation medium made no difference for p27, it led to a strongly reduced cell content of p21WAF1/CIP1 after a transient increase seen at 12 h (Figure ). The expression of IGF1R also showed a slightly higher level in cells deprived of serum in a medium without the antiestrogen. As the suppression of ER by ICI 182780 leads to a reduced expression of certain genes (see below, section 6), it is likely that the levels of their protein products result from the basal transcription-regulating activity of ligand-free ER.
As expected, in the cells serum-starved in medium with ICI 182780, ER was rapidly eliminated, the signal being near absent at 12 h. In spite of the continued presence of ICI 182780, ER became again detectable at later times. Starvation of serum and E2 in the absence of the antiestrogen led to a progressive accumulation of ER, as seen between 24 and 72 h.
It is to be noted that the cell contents of cyclin D1, a marker of early G1 phase, showed an early decrease at 12 h but then regained about the initial level and remained approximately constant throughout the 72 h incubation in serum-free medium. The presence of ICI 182780 did not reduce the level of cyclin D1 in mitogen-deprived cells (Figure ).
3. Serum and estrogen deprivation does not eliminate phospho-Akt.
Since the presence of the wild-type form of Akt is a prerequisite for the mitogenic signaling by E2 and since E2 does not induce the activating phosphorylation of Akt, we set out to verify by Western blotting the presence of phospho-Ser473-Akt (p-Akt) in the MCF-7 cells incubated in serum and estrogen-free medium. In these experiments the intensity of the p-Akt signal became weaker during serum deprivation but remained detectable, whether the cells had been incubated in a medium deprived of serum and exogenous estrogens, or in the same medium supplemented with ICI 182780. GSK3α/β a substrate of Akt kinase, showed a similar profile of phosphorylation (Figure ).
In order to verify that the signal detected with the anti-P-Ser473-Akt antibody represented the phosphorylated Akt rather than a non-specific antigen co-migrating incidentally with Akt, we treated the cell lysates with phosphatase. This treatment abolished the p-Akt signal both in cell lysates prepared from the quiescent MCF-7 cells and in cells treated for 1 h with insulin, a powerful inducer of the PI3K/Akt signaling (see Additional file 2
: Figure S2
The phosphorylation of Akt in the quiescent MCF-7 cells could be a consequence of signaling by an autocrine factor. To test this possibility, we harvested conditioned medium from cells after 48 h of incubation in the absence of serum and we compared the phosphorylation of Akt in quiescent cells placed in fresh DMEM with that detected in cells incubated with the conditioned medium. No difference was seen, suggesting that the Akt phosphorylation resulted from endogenous mechanisms and was not mediated by a secreted autocrine factor (see Additional file 3
: Figure S3
4. IGF1R signal transduction is not sufficient to drive the G1 phase progression.
Stimulation of the IGF1R signaling pathway induces a rapid and lasting phosphorylation of Akt. IGF-I and -II, as well as insulin at supra-physiological concentrations, are efficient mitogens in estrogen-deprived MCF-7 cells. Also, simultaneous stimulation of this pathway and of the ER acts in synergy to induce the MCF-7 cells’ proliferation. It has been reported by the laboratory of R. Sutherland that suppression of ER-dependent signaling by ICI 182780 prevents the mitogenic activity of insulin in these cells whereas antiestrogens of the type “SERM” do not show this effect [22
]. Varma and Conrad [12
] showed that the direct effects of IGF, phosphorylation of IGF1R and of Akt, are unaffected by ICI 182780, in contrast with the inhibition of the mitogenic action. We have addressed the mechanisms underlying the cooperation of the ER and IGF1R pathways. We analyzed the effects of E2 and insulin on the distribution of cells among the phases of the cell division cycle (Figure ; Additional file 4
: Figure S4
). Remarkably, even after 48 h incubation in serum-free medium, the MCF-7 cells did not become fully quiescent, with approximately 20% of the total population in S+G2/M-phase (Figure , time 0, lower panel). If the serum-free culture medium contained ICI 182780, after 48 h there remained practically no S+G2/M-phase cells. Stimulation with E2 or with insulin triggered the re-entry of G0/G1 arrested cells into the cell division cycle (Figure ). The most marked mitogenic effect was seen when the cells were fully synchronized by serum-starvation in the presence of ICI 182780 and subsequently stimulated by the addition of E2 (100-fold excess over ICI 182780). In these conditions, insulin produced only a weak and delayed effect. In contrast, insulin was an efficient mitogen when ICI 182780 was omitted from the culture medium (Figure , lower panel).
Figure 3 Reinitiation of the cell cycle progression in quiescent cells. Blocking the ER function inhibits the insulin-induced reinitiation of the cell cycle progression. MCF-7 cells were placed in serum- and phenol red-free medium containing or not 10 nM ICI 182780 (more ...)
These data confirm that pretreatment of the MCF-7 cells with ICI 182780 strongly reduces their sensitivity to the mitogenic action of insulin (Figure ) while the signal transduction by IGF1R is intact as documented by the strong induction of Akt phosphorylation by insulin in such cells, similar to that seen in cells deprived of serum in the absence of the antiestrogen (Figure , upper panel). We also observed an induction of cyclin D1 in cells starved of serum with and without ICI 182780, confirming that this process reflects direct IGFR1 signaling and is not sufficient for the cell cycle progression. There was though a correlation between the induction of cyclin D1 accumulation and the mitogenic action as shown by the FACS data: stronger induction by E2, weaker by insulin in antiestrogen-exposed cells.
Figure 4 Effect of the inhibition of PI3K by LY 294002 on the induction of cyclins D1 and A. MCF-7 cells were starved of serum and E2 during 48 h in the presence or absence of ICI 182780 (10 nM). Last 3 hours of serum deprivation (time −3 h to 0 h), one (more ...)
The fact that chemical inhibitors of PI3K block the mitogenic signaling in breast cancer cells has been reported earlier [9
]. This is also illustrated by the effect of LY 294002 on the expression of cyclin A (Figure , upper panel). In cells starved of mitogens in a medium without antiestrogen, cyclin A remained detectable, and its content did not diminish during a short (3 to 6 h) incubation with LY 294002 (Figure , upper panel). The expression of cyclin A in these conditions is probably the consequence of the incomplete quiescence (Figure ). When the cells were stimulated with E2 or with insulin for 19 h (late G1 phase), cyclin A was strongly induced and this induction was abolished by LY 294002 (Figure , lower panel).
As expected, the effect of IGF-I (10 nM) was the same as that of insulin (1 μM) (Additional file 5
: Figure S5
As ICI 182780 is a SERD-type antiestrogen (inducer of ER degradation), the lack of ER after pretreatment with this compound could be a reason for the diminished sensitivity of the cells to insulin. This is however unlikely to be the case as the reinitiation of the cell cycle progression by E2 in ICI 182780-pretreated cells is actually stronger than that of cells not pretreated with the antiestrogen, in spite of the strong reduction of the cell contents in ER (Figure ). The recent report of Wardell et al. [24
] demonstrates that the efficacy of ICI 182780 as an antiestrogen does not rely on its ability to induce ERα degradation.
5. Effect of the suppression of the PI3K pathway on the expression of cyclin D1 and c-myc protein and mRNA.
We were intrigued by the continuous presence of cyclin D1 in serum- and estrogen-deprived cells, non-suppressible by long-term treatment with ICI 182780. Signaling by the PI3K/Akt pathway favors the accumulation of the cyclin D1 protein by post-transcriptional mechanisms: accelerated translation [25
] as well as inhibition of degradation of the cyclin D1 protein due to the inhibition of GSK3 α/β through phosphorylation by Akt [27
In order to verify the role of the basal level of phosphorylated Akt in the expression of cyclin D1, we examined the effect of the PI3K inhibitor LY 294002. A 3 h incubation of serum-deprived cells with this drug strongly reduced the p-Akt signal, indicating that the basal phosphorylation of Akt seen in mitogen-deprived cells depended on PI3K activity. Further, our experiments showed a strong inhibition of the basal cyclin D1 expression by a 3 h exposure of the cells to LY 294002 (Figure , upper panel, t
0). The presence of LY294002 led to a reduction of the contents in cyclin D1 also when the cells were stimulated with either insulin or E2 (Figure , upper panel). Next we examined the transcriptional regulation of the CCND1
gene (Figure ). The presence of ICI 182780 during serum deprivation did not modify the level of cyclin D1 mRNA. After 48 h in serum-free medium, an incubation for 3 h with 20 μM LY294002 led to a 2 to 3-fold decrease of cyclin D1 mRNA contents, indicating that the basal activity of PI3K was required to maintain the expression of the CCND1
gene (Figure , bars 3, 4 vs. 1, 2). Stimulation of the quiescent cells with either E2 or insulin induced the accumulation of cyclin D1 mRNA (about 3-fold at 4 h). The amplitude of this induction paralleled the pattern of reinitiation of the cell cycle progression (Figure ): insulin was more efficient when serum deprivation had been carried out without ICI 182780 (Figure , bar 5 vs. 6), whereas the effect of E2 was more marked in cells rendered quiescent in the presence of ICI 182780 (bar 9 vs. 10). The induction of cyclin D1 mRNA by E2 was not prevented by LY 294002 (about 3-fold at 4 h compared with the level in control, LY 294002-exposed cells; Figure ); although the absolute level was lower than that reached without LY 294002, the induction of CCND1
transcription by estradiol apparently proceeded unhindered (compare bars 3 vs. 11 and 4 vs. 12; differences significant at respectively p ≤0.05 and p
0.01). On the other hand, the induction of the expression of the CCND1
gene by insulin was efficiently inhibited by LY294002.
Figure 5 Effect of LY 294002 on the induction of cyclin D1 and c-myc mRNA. MCF-7 cells were starved of serum and E2, in the presence or absence of ICI 182780, for 48 h. In one series of dishes, LY294002 was added for the last 3 h of serum deprivation (time −3 (more ...)
In contrast, in cells cultured in serum-free medium, a 3 h exposure to LY 294002 did not affect the level of the c-myc mRNA (Figure , bars 1 vs. 3 and 2 vs. 4). The same result was noted when the cells were stimulated with insulin (absence of effect of LY 294002; bars 5 vs. 7 and 6 vs. 8). The induction of c-myc mRNA accumulation by E2 was actually increased by LY294002 (difference significant for cells maintained without ICI 182780; bars 9
0.01). It is to be noted that ICI 182780 prevented the induction of c-myc mRNA accumulation by insulin (Figure , compare bars 5 vs. 6 and 7 vs. 8).
6. Transcriptional activity of unliganded ER in serum-deprived MCF-7 cells.
The essential consequence of the presence of ICI 182780 is the suppression of the basal level of ER-dependent gene expression. This was documented by monitoring the levels of two transcripts encoded by genes with estrogen response elements in their promoters, pS2 and PR (progesterone receptor). ICI 182780 caused a strong decrease in the expression of these genes (by approximately 90% after 48 h) whereas in the absence of the antiestrogen their mRNA levels decreased respectively by approximately 50% as compared to those observed in the exponential cells (Figure ).
Figure 6 ICI 182780 is required for the efficient suppression of ER-dependent endogenous transcripts during serum starvation. Exponentially growing MCF-7 cells (shaded bars) and cells incubated for 48 h in serum- and phenol red-free medium, with or without ICI (more ...)
In order to obtain a more direct information about the ER-dependent transcription in the absence of ligand, we evaluated the expression of luciferase in the MELN cell line derived from the MCF-7 cells by stable transfection with ERE-TK-LUC [28
]. When placed in serum- and phenol red-free medium, the cell content in luciferase varied little, whereas the addition of ICI 182780 led to a rapid extinction of the indicator enzyme, at a rate similar to that caused by the protein synthesis inhibitor cycloheximide, after a delay of about 3 h (Figure A). This delay is understandable: cycloheximide blocks all de novo
synthesis of luciferase protein whereas ICI 182780 prevents the synthesis of mRNA coding for luciferase and not the translation of pre-existing mRNA. To ascertain that the continued expression of luciferase was not due to a possible residual estrogen, we cultured the MELN cells for more than a month (a minimum of 4 passages) in estrogen-free medium supplemented with charcoal-stripped serum plus 100 nM Insulin. The cells were then placed in serum-free medium, without insulin, with or without ICI 182780. Similar results were obtained: ICI 182780 rapidly extinguished the expression of luciferase whereas in the absence of the antiestrogen the level of luciferase increased with time (Figure B)
Figure 7 ICI 182780 is required for the efficient suppression of ER-dependent luciferase during serum starvation of MELN cells.A. Exponentially growing MELN cells were placed in serum- and phenol red-free medium, with or without ICI 182780 (10 nM) or cycloheximide (more ...)
A possible explanation of these results is the existence of pathways that lead to the phosphorylation of the ER and of co-activators that participate at the regulation of its transcriptional activity. This possibility is sustained by the fact that phospho-Ser118-ER is detected in the serum-deprived MCF-7 cells (data not shown). The mechanism responsible for ER phosphorylation remains unknown at this moment. As in the case of the basal, constitutive phosphorylation of Akt, it is probably the result of an endogenous process, not requiring added or secreted factors.