Inhibition of TPO-induced MEK or mTOR activity induces opposite effects on the ploidy of human differentiating megakaryocytes

Megakaryocytes (MKs) are characterized by high levels of ploidy during their normal development. During the MK life cycle, an initial proliferative stage (mitosis) is followed by an endomitotic process (successive rounds of DNA synthesis in the absence of cytokinesis), which results in a polyploid DNA content, before final maturation and platelet fragmentation (Zimmet and Ravid, 2000). The major regulator of megakaryocytic development, thrombopoietin (TPO), fully supports MK differentiation in vitro, including pro-platelet and platelet formation (Guerriero et al., 1995; Kaushansky et al., 1995; Cramer et al., 1997). As observed for other cytokines, its receptor binding might regulate specific cell functions activating several signaling pathways. An extensive cross-talk between these pathways might be crucial for controlling a wide array of biological processes (Rojnuckarin et al., 2001). However, in spite of the numerous studies carried out, the cellular and molecular mechanisms underlying MK polyploidization remain largely unknown.

Several groups have reported that components of Janus kinase/signal transducer and activator of transcription (JAK/STAT) (Drachman et al., 1995; Wendling and Vainchenker, 1998) and Shc-Ras-MAPK pathways (Nagata and Todokoro, 1995; Yamada et al., 1995; Rouyez et al., 1997) are strongly activated in response to TPO. In particular, the importance of the mitogen-activated protein kinase (MAPK) pathway in megakaryocytopoiesis was corroborated by introduction of constitutively activated mutants of MAPK kinase (MEK) in several cell lines (Whalen et al., 1997; Racke et al., 1997; Melemed et al., 1997) and by treatment of cell lines and primary cells with pharmacological inhibitors of MEK such as PD98059 and UO126 (Rojnuckarin et al., 1999; Fichelson et al., 1999). Nevertheless, the conclusions from these studies on the role of TPO-induced MAPK activity were often controversial. These discrepancies may depend on the experimental conditions and/or on the source of cells used: cells lines, murine mature MKs, murine CD41-selected cells and human cord blood CD34⁺ cells.

In addition to JAK/STAT and Ras-Raf-MAPK, the phosphoinositide 3-kinase (PI3K) pathway is important for mpl (TPO receptor) signaling (Sattler et al., 1997; Miyakawa et al., 2001). PI3K has been shown to be activated by many growth factors involved in hematopoiesis, and plays an important role in promoting cell survival (Songyang et al., 1997; Datta et al., 1999), as well as in driving some cell types to transit from G1 to S phase, in both mitotically and endomitotically cycling cells (Geddis et al., 2001).

The best investigated PI3K target is AKT, which mediates and activates several pathways implicated in the suppression of apoptosis and growth-factor-mediated survival (Datta et al., 1999). A downstream target of AKT is mTOR (the mammalian target of rapamycin), which regulates p70 ribosomal S6 kinase (p70S6K). Several studies have suggested that p70S6K mediates PI3K-AKT signaling during G1 cell-cycle progression in a variety of cell types (Feng et al., 2000; Gao et al., 2003; Alvarez et al., 2003; Shao et al., 2004). The regulation of the cell cycle that leads to megakaryocytic polyploidization is not clear, but it is thought that an orchestrated change in expression of several genes might be involved. It has been suggested that MK endomitosis is characterized by a significantly short G1 phase (Wang et al., 1995) and an incomplete mitosis with development of large polyploid cells (Roy et al., 2001).

The D-type cyclins, in conjunction with their catalytic partners cdk4 and cdk6, appear to regulate G1 phase progression and play a role in G1-S transit: they are partially cell-type specific, the D3 being the only D-cyclin that is upregulated during MK differentiation (Wang et al., 1995; Furukawa et al., 2000). Moreover, the cyclin D3 is expressed and upregulated by ploidy-promoting factors, such as TPO and phorbol ester, in both megakaryocytic cell lines and primary MKs (Zhang et al., 1996; Zimmet et al., 1997), whereas its over-expression results in an increased ploidy (Zimmet et al., 1997). The molecular mechanisms and signaling pathways inducing megakaryocytic cyclin D3 expression and its nuclear translocation have remained largely unknown.

Studies on human megakaryocytopoiesis have been hampered by the lack of pure and abundant MK progeny. In our previous study (Guerriero et al., 1995), we optimized a serum-free liquid culture system for gradual hematopoietic progenitor cell (HPC) differentiation along the MK lineage, giving rise to a virtually pure MK population (97-99% of cells were CD61⁺/CD41⁺), thus providing an in vitro experimental tool to dissect the cellular and molecular basis of megakaryocytopoiesis. So far, `clear-cut' signaling studies on the polyploidization of human differentiating MKs are still lacking. In this study, we have investigated the role of MAPK and PI3K-AKT pathways during MK differentiation and maturation. CD34⁺ progenitor cells, purified from human adult peripheral blood, were induced to unilineage megakaryocytic differentiation in the presence of TPO alone, or in combination with PD98059 or rapamycin, which inhibit MEK1/2 and mTOR activation, respectively. The results clearly showed an opposite effect of these inhibitors on MK polyploidization.

HPCs purified from human peripheral blood, grown in fetal calf serum (FCS)-free liquid suspension culture supplemented with plateau levels of TPO (100 ng/ml), undergo a gradual wave of differentiation and maturation along the MK lineage, giving rise to a virtually pure MK population (97-99% of the cells were CD61⁺/CD41⁺) as previously reported (Guerriero et al., 1995). In an attempt to analyze the effects of MAPK inhibition during MK differentiation, PD98059 (a specific inhibitor of the activation of MEK1/2 by Raf kinase) or UO126 (an inhibitor that acts directly upon MEK1/2) were added at a final concentration of 10 μM from day 0 and during the first three days of unilineage MK cultures.

The results clearly demonstrate that cell growth was affected by the addition of both these inhibitors: Fig. 1 shows the cell growth curves of TPO alone (control) or in combination with these two inhibitors. The reduced PD98059 growth curve (about 50% of the control) seems to be the result of a decrease in cell proliferation rather than the result of cell death, as shown in Table 1. Interestingly, this cell growth decrease is associated with a significant increase in polyploidization (from 45 to 70%) as observed by morphological and DNA content analysis. PD98059-treated cells were larger (Fig. 2A) and had a significant increase in the number of nuclear lobes (Fig. 2B) and DNA content (Fig. 3), compared with the control cell cultures.

Moreover, the effect of PD98059 on megakaryocytic cultures was clearly evident when PD98059 addition to HPC culture medium started at day 0 and was repeated for the first 3 days of culture; on the contrary, it was abolished when the inhibitor was added only at day 0 or at later days of culture (i.e. after day 3 of culture), when the cells were already committed to MK precursors (data not shown). These observations suggest prolonged PD98059 treatment is necessary in the early differentiative stages for the induction of polyploidization.

Analysis of the cell membrane phenotype in differentiating MKs grown with or without PD98059 showed a similar percentage of positive cells for specific megakaryocytic markers (i.e. CD61, CD41a, CD62 and CD42b) (Fig. 4). In both culture conditions, the expression of cell-surface antigens confirmed the same gradual CD34 decrease, together with an increase in megakaryocytic markers, as previously reported in our control culture system (Guerriero et al., 1995). By contrast, the addition of inhibitor U0126 to HPC cultures induced a rapid and strong decrease in the total number of cells (Fig. 1), which were smaller than the control cells and died rapidly during the first days of culture (Table 1).

It has been just suggested that the PI3K-AKT pathway plays an important role in the G1-S cell-cycle transition (Geddis et al., 2001); in this report, we focused our interest on exploring its involvement in MK endomitosis. HPCs, induced to MK differentiation by TPO, were treated at day 0 and for the first three days of culture with LY294002 (a PI3K inhibitor) at 10 μM final concentration. As shown in Fig. 1, the total number of LY294002-treated MKs decreased in comparison with the control culture, and the cells observed at day 6 were small and died rapidly. However, the elevated percentage of non-viable cells reported in Table 1 indicated that the inhibition of PI3K probably gave rise to the activation of downstream targets involved in the apoptotic process.

In addition to regulating anti-apoptotic pathways, PI3K-AKT also activated mTOR (Feng et al., 2000; Gao et al., 2003; Alvarez et al., 2003; Shao et al., 2004). Therefore, in an attempt to investigate whether the inhibition of mTOR and p70S6K could affect G1 cell-cycle progression, MK differentiating cell cultures were treated with rapamycin (a specific inhibitor of mTOR activity). The growth curve of MKs cultured in the presence of 50 nM rapamycin was partially inhibited (about 50%) (Fig. 1), and the MK cells were smaller than the control cells (Fig. 2A), with a significant decrease in their number of nuclear lobes (about 99% of the cells exhibited a single nuclear lobe) (Fig. 2B) and DNA content (Fig. 3). Interestingly, as observed above for PD98059, the phenotypic analysis of control and rapamycin-treated cells did not show significant differences (Fig. 4), and the cell growth decrease did not depend on an increased cell death as shown in Table 1.

The culture experiments described above indicated a role of inhibitor PD98059 in potentiating MK polyploidization when it was added at day 0 of the culture; this observation suggests that early TPO-induced signaling events play a key role in MK polyploidization. Thus, in order to evaluate phospho-ERK 1 and phospho-ERK 2 (P-ERK 1/2) and phospho-AKT (P-AKT) expression in human progenitors, purified CD34⁺ cells were starved overnight in serum-free medium, and were then pre-treated for a brief period (20 minutes) in the presence or absence of specific MAPK or PI3K pathway inhibitors, followed by TPO incubation (40 minutes) and then analyzed by western blotting (Fig. 5A).

In agreement with other studies, we observed that a TPO stimulus induced both P-ERK and P-AKT expression. Our results showed that PD98059 pre-treatment partially inhibited TPO-induced P-ERK expression, from a 2.4-fold (TPO alone) to a 1.4-fold (PD98059+TPO) increase compared with unstimulated cells but, surprisingly, increased P-AKT expression (from 1.6- to 2.7-fold) (Fig. 5A,B). The use of inhibitor UO126 fully blocked P-ERK expression, but had no effect on AKT phosphorylation (Fig. 5A).

In an attempt to confirm P-AKT involvement in TPO-induced MK differentiation and/or polyploidization, we analyzed P-AKT expression by treatment with LY294002 (upstream) and rapamycin (downstream) inhibitors. LY294002 led to a block in AKT phosphorylation, whereas rapamycin induced an accumulation of P-AKT (from a 1.6- to 2.6-fold increase); the activation of ERK 1/2 in both conditions was decreased (Fig. 5A,B).

We then tested whether p70S6K, a downstream regulator of the PI3K-AKT-mTOR pathway, was activated by TPO. Western blotting analysis was performed using phospho-p70S6K antibodies, including anti-Thr389 antibody and anti-Thr421/Ser424 antibody. TPO induced a strong phosphorylation at both sites of p70S6K; this phosphorylation was minimally affected by PD98059 treatment but, remarkably, was totally inhibited by rapamycin (Fig. 5A,B). Furthermore, the p70S6K phosphorylation was diminished at Thr389 by LY294002 and at Thr421/Ser424 by U0126 (Fig. 5A).

We have evaluated the subcellular localization of cyclin D1 and D3 in differentiating MK progenitor cells grown in the presence of TPO alone (control) or in combination with PD98059 (inhibitor of ERK 1/2) or rapamycin (inhibitor of mTOR). In the control cells, cyclin D1 was expressed during all stages of MK differentiation and maturation, remaining predominantly localized at the level of cytoplasmic compartments; no difference in its expression and localization was observed in MK cells grown in the presence of both inhibitors (Fig. 6A). It is noteworthy that cyclin D1 showed cytoplasmic localization also in polyploid MKs.

By contrast, cyclin D3 showed a predominant cytoplasmic localization in immature MK cells, and shifted to the nuclear compartment as soon as they became polyploid (Fig. 6B). In fact, at day 3 of culture, a faint anti-cyclin D3 reactivity was restricted to the cytoplasm of early MK precursors; at day 8 of culture, the majority of MK precursors displayed cyclin D3 both in the nucleus and in the cytoplasm; and, at day 10 of culture, a bright cyclin D3 nuclear reactivity was observed in polyploid MKs. This phenomenon was accelerated and potentiated in polyploid PD98059-treated MKs, whereas it was completely blocked in cells grown with rapamycin (Fig. 6B), all showing one nuclear lobe and displaying a faint cytoplasmic cyclin D3 reactivity.

In an attempt to investigate if rapamycin induces cell-cycle arrest, the cell-cycle distribution of CD34⁺ cells, grown in the presence of TPO alone (control) or in combination with rapamycin, was examined after 24, 48 and 72 hours. At 24 hours, only a small fraction of CD34⁺ cells were cycling (control: 0.95±0.12% of cells in the S-G2-M phases; rapamycin treated: 1.02±0.19% of cells in the S-G2-M phases). At 48 hours and 72 hours, the proportion of the cells in S-G2-M phases greatly increased in control cultures (48 hours: 11.74±1.8%; 72 hours: 19.89±3.7%), whereas it was markedly lower in rapamycin-treated cultures (48 hours: 3.36±0.66%; 72 hours: 10.98±1.45%). The difference in cell cycle between control and rapamycin-treated cells was not significant at 24 hours, but it was significantly lower at 48 hours (P=0.0016) and 72 hours (P=0.018) in cells grown in the presence of rapamycin, compared with untreated cells.

The molecular mechanisms regulating MK endomitosis and polyploidization remain largely unknown; TPO, the major regulator of megakaryocytic development activates both ERK and PI3K-AKT pathways (Rojnuckarin et al., 1999; Miyakawa et al., 2001; Majka et al., 2002). In this study, we investigated the involvement of the ERK and PI3K-AKT pathways in promoting polyploidization of human MK-induced HPCs by means of specific inhibitors used in combination with TPO.

The ERK 1/2 signaling cascade is a tightly controlled pathway, in which the magnitude and duration of kinase activity determines the physiological response. Several studies have previously reported controversial results on the MK polyploidization effects of MAPK inhibition (Rojnuckarin et al., 1999; Fichelson et al., 1999; Minamiguchi et al., 2001). These discrepancies might depend on the source and maturative stage of the cells used (cells lines, murine mature MKs, murine CD41-selected cells, human cord blood CD34⁺ cells) and/or, most likely, on the different experimental conditions used. In our previous study, according to Rojnuckarin et al. (Rojnuckarin et al., 1999), we reported that a high concentration of the MAPK inhibitor PD98059 interfered with normal MK polyploidization (Guerriero et al., 2001). However, in the present study, we observed that early and prolonged treatment with a low PD98059 concentration (10 μM) is crucial for the investigation of megakaryocytopoiesis. Therefore, using this experimental approach, we obtained a marked decrease in TPO-induced progenitor proliferation (about 50%), associated with enhanced cell polyploidization (about 70% of MKs showed more than 2N) and unchanged phenotypic features. Interestingly, our results demonstrated that the inhibitor PD98059 drastically reduced ERK activation (about 40% inhibition), without blocking it completely.

By contrast, the inhibitor U0126 acts directly upon MEK, inducing a rapid decrease in the total number of cells followed by their progressive death. In this case, the activation of ERK was completely inhibited. The different effects, shown by these two MAPK inhibitors on TPO-induced megakaryocytopoiesis, could be ascribed to the partial or total block of ERK activation, indicating that at least a moderate MAPK activation was essential for MK proliferation and differentiation. The observation that activation of ERK is necessary to induce proliferation and differentiation, whereas its downmodulation is crucial to promote polyploidization, prompted us to investigate whether the PI3K-AKT pathway is hyperactivated and involved in endomitosis during ERK downregulation. In support of this hypothesis, a 1.6- to 2.7-fold increase of TPO-induced P-AKT levels in PD98059-treated cells was observed in our experiments.

As expected, given the importance of PI3K as a mediator of growth-factor-regulated survival signals, the addition of its inhibitor, LY294002, to HPC unilineage MK cultures induced a rapid cell death, probably through the activation of specific apoptotic targets. Geddis et al. have previously shown that PI3K transduces a signal promoting cell cycling in response to TPO, both in a factor-dependent leukemic cell line and in primary murine MKs, and that this signal can be distinguished from its effect on cell survival (Geddis et al., 2001). In an attempt to explore whether the involvement of PI3K in cell cycling, and perhaps in endomitosis, could be ascribed to a downstream PI3K-AKT target, HPCs were treated with rapamycin (an inhibitor of mTOR, a downstream target of AKT and upstream regulator of p70S6K). Surprisingly, the culture addition of rapamycin gave rise to a fully differentiated population of small MKs displaying a specific membrane phenotype with only one nuclear lobe. Our results, described for the first time in human megakaryocytic differentiating cells, indicate a role of the PI3K-AKT-mTOR pathway in cell-cycle progression and cell size, in line with previous studies reported in other systems (Feng et al., 2000; Bodine et al., 2001; Rommel et al., 2001; Fingar et al., 2002; Gao et al., 2003). The opposite effect obtained on MK polyploidization by the addition of either PD98059 or rapamycin to TPO-induced HPC cultures encouraged us to investigate the possible role of PI3K-AKT-mTOR pathway in MK endomitosis.

When the TPO-induced HPCs were treated with rapamycin, giving rise to a small and mononuclear differentiated MK progeny, the phosphorylation of p70S6K (mTOR target) was completely inhibited, suggesting its possible role in cell size and endomitosis; by contrast, in PD98059-treated HPCs, which give rise to large polyploid MK cells, the p70S6K was activated, enforcing its possible role in cell size and endomitosis. Furthermore, both inhibitors (PD98059 or rapamycin) surprisingly showed a similar effect in reducing ERK activation and enhancing AKT phosphorylation. The rapamycin-induced mechanism responsible for the P-ERK decrease and P-AKT increase is not clear; however, as also reported by other investigators (Rommel et al., 1999), AKT activation might inhibit the ERK pathway. In any case, our results indicate that the PI3K-AKT-mTOR pathway could mediate TPO signaling by activating p70S6K during MK endomitosis.

Furthermore, the subcellular localization of cyclin D3, the primary D-type cyclin expressed in MK cells, is predominantly cytoplasmic in control cells and became nuclear in polylobated MKs grown in the presence of PD98059; by contrast, it was completely cytoplasmic in MKs grown in the presence of rapamycin, suggesting that mTOR-p70S6K was involved in promoting cyclin D3 nuclear relocation. These results, described for the first time in human MK differentiation, are in line with previous studies on mouse spermatogonia that showed that stem cell factor (SCF) promotes cell-cycle progression by a rapamycin-sensitive PI3K-p70S6K-cyclin D3 pathway (Feng et al., 2000), and induces a rapid G1-S transition by ERK 1/2 and PI3K activation followed by cyclin D3 nuclear redistribution (Dolci et al., 2001). The mechanism through which rapamycin inhibits cyclin D3 activation remains to be determined, but it is seemingly related to the capacity of this compound to inhibit cell-cycle progression (G1-S transition). In fact, in line with previous reports on different cellular systems (Breslin et al., 2005; Decker et al., 2003), we observed that rapamycin induces the inhibition of cell-cycle progression in CD34⁺ cells induced to MK differentiation.

Moreover, our results on cyclin D1, unlike those obtained for cyclin D3, show that it is constantly localized in the cytoplasm during all stages of MK maturation, in the control, and in cells treated with the inhibitors. Previous studies have failed to show a role for cyclin D1 in megakaryocytic differentiation: in fact, cyclin D3, but not cyclin D1, was upregulated during MK differentiation (Furukawa et al., 2000), and in vivo, cyclin D1 over-expression induced only a weak effect on MK ploidy (Sun et al., 2001).

Therefore, although TPO simultaneously activated both the ERK and PI3K-AKT-mTOR pathways, the effect of the latter on MK polyploidization appeared to depend on the duration and intensity of MAPK activation. However, the mechanisms responsible for the activity modulation of either pathway remain unclear. Rommel et al. reported that the inhibition of the ERK pathway by the PI3K-AKT pathway was dependant upon the cell differentiation stage during muscle cell hypertrophy (Rommel et al., 1999); it is possible that a cross-talk regulation between the ERK pathway and the PI3K-AKT pathway might be similarly important in MK polyploidization. More recently, Kanaji et al. developed a murine model in which sequestration of the signal proteins by the glycoprotein Ibα modulates MK ploidy and proliferation (Kanaji et al., 2004), whereas Tong and Lodish reported that the duration and intensity of MAPK activation in murine CD41⁺ MKs is affected by a negative regulator of TPO-mediated signaling (Tong and Lodish, 2004).

Finally, our results could have future practical and clinical implications. In fact several pathological conditions are associated with a decreased or increased platelet production or with the accumulation of leukemic megakaryocytic progenitors/precursors. Therapy for these diseases could take advantage of treatment with MAPK inhibitors (Sebolt-Leopold and Herrera, 2004) and mTOR inhibitors (Bjornsti and Houghton, 2004), similar to those used in the present study and recently introduced in clinical trials. These agents, used in combination with standard drugs, could improve the prognosis of these pathological conditions, which are often fatal.

In conclusion, although the molecular mechanisms regulating cell growth and cell-cycle progression in human megakaryocytopoiesis are currently unclear, our findings demonstrate that rapamycin-treated HPCs proliferate and differentiate into small MKs without undergoing endomitosis, suggesting mTOR may be a key element of this process. In particular, the rapamycin-induced cytoplasmic cyclin D3 localization suggests that the endomitotic process could be mediated by a cascade mechanism in which TPO activates a rapamycin-sensitive PI3K-AKT-mTOR-p70S6K-cyclin D3 pathway (Fig. 7).

Adult peripheral blood was obtained from 20-40-year-old healthy male donors after informed consent. Low-density mononuclear cells (MNCs) were isolated by Ficoll-Hypaque density-gradient centrifugation and CD34⁺ HPCs were then purified by MACS columns (Milteny) according to the manufacturer's instructions. Purified cells were more than 90% CD34⁺, as evaluated by fluorescence-activated cell sorting (FACS) analysis.

Purified HPCs were grown in FCS-free unilineage MK liquid culture (Guerriero et al., 1995) [1×10⁵ cells/ml, in the presence of a saturating dose of TPO (100 ng/ml)] alone or in combination with 10 μM PD98059, a specific inhibitor for the activation of MEK1/2 by Raf Kinase, or 10 μM UO126, an inhibitor that acts directly upon MEK1/2, or 10 μM LY294002, a specific inhibitor of PI3K, or 50 nM rapamycin, a specific inhibitor of mTOR (all from Calbiochem). All inhibitors were dissolved in dimethylsulfoxide (DMSO). In the mock culture, an equivalent amount of DMSO (<0.1%) was added. Cells were incubated in a fully humidified atmosphere of 5% CO₂, 5% O₂, 90% N₂.

Cells collected at different days of culture were cytocentrifuged onto glass slides, stained with May-Grünwald Giemsa (Sigma) and then identified by morphological analysis.

The following monoclonal antibodies (mAbs) directly conjugated with fluorescein isothiocyanate (FITC) or phycoerythrin (PE) were used to characterize the membrane phenotype of cell samples: anti-CD34, anti-CD61, anti-CD62, anti-CD42b (Becton Dickinson) and anti-CD41a (Serotec). Cells were analyzed by FACScan (Becton Dickinson) by means of the Lysis II program for fluorescence intensity analysis.

MK ploidy was analyzed by flow cytometry after DNA staining with propidium iodide (PI) (Sigma) according to the procedure described by Dolzhanskiy et al. (Dolzhanskiy et al., 1996). Cells were washed and resuspended in medium containing 0.5% Tween-20 for 30 minutes to permeabilize the cell membranes. Then, an equal volume of medium containing 0.5% Tween-20 and 2% paraformaldehyde was added. After 5 minutes at 4°C, the cells were pelletted, and freshly prepared PI was added. The suspension was stored overnight in the dark at 4°C. After an overnight incubation, 50 μg/ml RNAse was added for 30 minutes at room temperature in the dark, and the cells were analyzed by flow cytometry.

Cell viability was evaluated by flow cytometry after staining intact cells with PI (non-viable cells fluoresce, whereas viable cells are non-fluorescent).

The activation of extracellular signal-regulated kinase 1/2 (ERK 1/2), AKT and p70S6K was analyzed by western blotting. CD34⁺ progenitor cells (2×10⁵cells/condition) were incubated overnight in serum-free medium in the absence of growth factors, to allow maximal dephosphorylation of cellular proteins, then the cells were stimulated with protein kinase inhibitors (10 μM PD98059, 10 μM U0126, 10 μM LY294002 and 50 nM rapamycin) for 20 minutes, followed by 40 minutes with 100 ng/ml TPO. Cells were washed with ice-cold phosphate-buffered saline (PBS), lyzed with sample buffer, loaded on SDS-PAGE and transferred to a nitrocellulose filter. Blots were blocked using 5% non-fat milk in Tris-Buffered Saline Tween-20 (TBST) for 1 hour at room temperature, or alternatively, as suggested by manufacturer's instruction (Cell Signaling Technology). ERK 1/2, AKT and p70S6K activity was measured by an immunoblot of whole-cell extracts with activated specific polyclonal antibodies against: the dually phosphorylated forms of p42ERK2 and p44ERK1 (P-ERK 1/2), phospho-AKT (Ser473), phospho-p70S6K (Thr389), phospho-p70S6K (Thr421/Ser424) (all from Cell Signaling Technology). After hybridization with secondary antibodies conjugated to horseradish peroxidase, the immunocomplex was detected with the ECL detection reagent (Pierce). Total ERK, AKT and p70S6K amounts were determined by reprobing the same membranes with ERK 1/2 (Promega), AKT (Cell Signaling Technology) and p70S6K (Cell Signaling Technology) antibodies. The densitometric analysis of phosphorylated protein blots was performed with the chemidoc program Biorad, and expressed as fold increase of unstimulated cells.

The expression and localization of cyclin D1 and D3 in developing MK cells were explored by confocal microscopy after immunofluorescence labeling. Preparations of MK cells cytocentrifugated at low speed (3 minutes at 100 g) on polylysinated slides (Sigma) were fixed for 15 minutes at room temperature with 2% paraformaldehyde and then permeabilized for 5 minutes at room temperature with 0.2% Triton X-100 in PBS. After washing, the cells were incubated for 30 minutes at room temperature with 5 μg/ml anti-cyclin D1 antibody or anti-cyclin D3 antibody directly conjugated with fluorescein (Pharmingen). After extensive washing in PBS, the cells were analyzed by confocal microscopy (Olympus, Flow View FV500).

Cell-cycle analysis was carried out by PI staining using the CycleTEST™ PLUS DNA Reagent Kit from Becton-Dickinson. Briefly, the method involves dissolving the cell membrane lipids with a non-ionic detergent, eliminating the cell cytoskeleton and nuclear proteins with trypsin, digesting the cellular RNA with an RNase and stabilizing nuclear DNA with spermine (Vindelov et al., 1983). PI is then added and stoichiometrically bound to the isolated nuclei, which are run on a flow cytometer equipped with electronic doublet-discrimination capability (Martens et al., 1981).

We thank M. Fontana, M. Blasi and S. Hourshid for editorial assistance, and A. Zito for graphics.