Phosphoinositide 3-Kinase Activation in Late G₁ Is Required for c-Myc Stabilization and S Phase Entry^▿

Plasmids and reagents.

The retroviral vectors encoding wild-type (WT) c-Myc-internal ribosome entry site-green fluorescent protein (GFP) or T58Ac-Myc-internal ribosome entry site-GFP (14) were kindly provided by S. Lowe (Cold Spring Harbor Laboratory, NY). pBabePuro encoding c-MycER (20) was generously donated by G. Evan (Cancer Research Institute, University of California at San Francisco, CA). Antibodies to c-Myc (C-19), cyclin E (M-20), and estrogen receptor alpha (ERα) (MC-20) were obtained from Santa Cruz Biotechnology; anti-pan-Ras and -α-tubulin were from Oncogene Research; anti-p-PKB(473) was from Cell Signaling Technologies; and anti-cyclin D3, -p27^Kip, and -MCM2 were obtained from BD Biosciences. Anti-Rb antibody was from Zymed Laboratories, anti-phospho(T58/S62)-c-Myc and anti-c-Myc (9B11) were from Cell Signaling Technologies, and anti-cyclin A and -phosphotyrosine were from Upstate Biotechnology. Monoclonal anti-β-actin antibody was from Sigma and anti-histone from Chemicon International. Horseradish peroxidase-conjugated secondary antibodies were purchased from Dako Cytomations. Enhanced chemiluminescence, l-[³⁵S]methionine, [α-³²P]dCTP, and [γ-³²P]ATP were from Amersham Biosciences. Lovastatin, herbimycin, and Ly294002 were from Calbiochem.

Cell lines and cell cycle analysis.

NIH 3T3 and Phoenix cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco-BRL) supplemented with 10% (vol/vol) fetal bovine serum (FBS), 2 mM glutamine, 10 mM HEPES, 100 U/ml penicillin, and 100 μg/ml streptomycin in a humidified atmosphere (5% CO₂, 37°C). To monitor the G₀-to-S transition accurately, we established a standard time course protocol for all experiments. Exponentially growing cells were seeded into dishes and rendered quiescent by incubation in DMEM-0.1% FBS (19 h). Under these conditions, cells exhibited a G₀ phenotype, examined as previously described (22). Quiescent cultures were rinsed with serum-free medium, and synchronous cell cycle entry was stimulated by readdition of serum (10% final concentration). Some samples were harvested immediately before serum addition (time zero). Cells were harvested at various times after serum stimulation. DNA synthesis was studied by DNA staining with propidium iodide and analyzed with a flow cytometer (Beckman-Coulter) as described previously (1) or by bromodeoxyuridine (BrdU) incorporation.

BrdU incorporation.

Cells were incubated with 10 μM BrdU for 90 min and harvested at indicated time points by using trypsin-EDTA (see Fig. 4). Cells were washed twice with phosphate-buffered saline (PBS)-1% FBS and fixed in ice-cold 80% methanol overnight. Cells were then washed twice and resuspended in PBS containing 1% FBS and 0.1 mg/ml RNase (30 min, room temperature). To extract histones and denature cellular DNA, we incubated cells with 1.5 N HCl and 0.5% Triton X-100 (30 min, room temperature). For direct immunofluorescence staining, cells were incubated (1 h) with fluorescein isothiocyanate-conjugated anti-BrdU antibody (Becton Dickinson). After being washed (in PBS-1% FBS), cells were resuspended in 500 μl PBS (containing 0.1 mg/ml RNase, 0.1% NP-40, and 5 μg/ml propidium iodide) and analyzed by flow cytometry.

Pulse-chase assay.

NIH 3T3 cells were incubated in DMEM-0.1% FBS for 19 h (as described above), and then the medium was replaced with Met-free RPMI medium (Gibco) containing 10% dialyzed FBS for 9 h, with 0.75 mCi [³⁵S]Met (each p100 dish) included for the last 6 h. At 8.5 h after serum addition, some of the samples were treated with Ly294002 (10 μM). After the 9-h incubation period described above (pulse), the [³⁵S]Met-containing medium was washed and replaced by DMEM-10% FBS containing 200 μM of cold Met and Cys alone or with Ly294002 (10 μM) and maintained until 12 h and 16 h after serum addition (chase).

Inhibitor treatments and retroviral transduction.

To activate c-MycER, we added 4-hydroxytamoxifen (4-OHT) (200 nM; Sigma) 6.5 h after serum stimulation. In some cases, cells were treated with 0.1% dimethyl sulfoxide (control), lovastatin (10 μM), herbimycin (2 μM), or Ly294002 (10 μM). When samples were collected at time zero, inhibitors were added 30 min before serum addition; otherwise, inhibitors were added after 4, 6, or 7 h after serum stimulation.

Phoenix cells were transfected using Jet Pei-NaCl (Poly plus transfection) according to the manufacturer's protocols; after 10 h, cells were washed and placed in DMEM-10% FBS. Retroviral gene transduction was performed as described previously (35). Infected NIH 3T3 (c-MycER) cells were selected for 2 days in medium containing 2 μg/ml puromycin (Sigma).

Cell lysis, immunoprecipitation, PI3K assay, WB, pull-down assay, subcellular fractionation, Northern blotting, and Cyclin/Cdk kinase assays.

Lysates were prepared in Triton X-100 lysis buffer (50 mM HEPES, pH 8.0, 150 mM NaCl, 1% Triton X-100) containing protease and phosphatase inhibitors (1 mM Na₃VO₄, 5 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 1 μg/ml aprotinin, and 10 μg/ml leupeptin). Immunoprecipitation, PI3K assays, and Western blotting (WB) were performed as described previously (15). For the PI3K assay, the cells were transfected with pSG5-myc-tagged-p110α (15), and the cells were synchronized 24 h later, as described above; PI3K was immunoprecipitated using anti-Myc-tag antibody. Ras-GTP was purified from cell extracts on Sepharose-Gex2T-RBD (the Ras-binding domain of Raf-1). Briefly, NIH 3T3 cells were harvested, lysed with glutathione S-transferase fluorescent in situ hybridization buffer (50 mM Tris-HCl, pH 7.5, 2 mM EDTA, 100 mM NaCl, 2 mM MgCl₂, 1% [vol/vol] NP-40, 5 mM NaF, 10% [vol/vol] glycerol, 1 mM phenylmethylsulfonyl fluoride, 1 μg/ml aprotinin, and 10 μg/ml leupeptin). Protein concentration, examined using the Micro bicinchoninic acid assay (Pierce), was normalized, and lysates were incubated (1 h, 4°C) with glutathione-Sepharose beads precoupled with glutathione S-transferase-RBD. Beads were washed three times in lysis buffer, and bound Ras-GTP was solubilized in 30 μl Laemmli buffer. Ras-GTP content was analyzed by WB. Total cell extracts, nuclear extracts, and chromatin fractions were isolated as described previously (26). Northern blot and cyclin/Cdk kinase assays were performed as described previously (22, 26, 35).

PI3K activity in late G₁ induces PKB activation that correlates with increased c-Myc protein levels.

PI3K is activated in late G₁; this activity peak is essential for S phase entry, since late-G₁ PI3K inhibition blocks S phase entry and PtdIns(3,4,5)P₃ addition in late G₁ induces cell cycle entry in the absence of serum (17, 18). To study the role of PI3K in late G₁, NIH 3T3 cells were driven into quiescence by serum deprivation and then released into G₁ by serum addition. Cells were committed to enter S phase at about 9 h after serum addition, with no further GF requirement (not shown). S phase began between 9 and 12 h after serum addition (Fig. 1A). In cells entering the cell cycle synchronously, we detected early-G₁ (1 h) and late-G₁ (∼9 to 15 h) PI3K activity peaks (Fig. 1B), as determined by examining the phosphorylation of the PI3K effector PKB (p-PKB) (1). c-Myc expression levels paralleled the PI3K activity peaks (Fig. 1C).

Ras and Tyr kinases activate PI3K in late G₁.

PI3K activation at G₀-to-G₁ transition is dependent on Tyr-K and Ras activities (6, 15); Ras is also activated in late G₁ (38). We examined Tyr-K activation by WB using an anti-pTyr antibody and Ras activity by pull-down assays. After GF addition, total Tyr-K activity increased transiently at 1 h and again between 6 and 16 h (Fig. 1D). Some of the Tyr-phosphorylated bands that appeared at 1 h differed from those visible at ∼9 h, suggesting that more than one Tyr-K is activated during G₁ (Fig. 1D). Ras-GTP increased at ∼1 h and again at 9 to 12 h after serum stimulation (Fig. 1E).

To determine whether Tyr-K or Ras stimulation is required for late-G₁ PI3K activation, we used small molecule inhibitors and examined the effects on p-PKB and c-Myc levels. Addition of the Ras inhibitor lovastatin (10) at 0, 4, or 6 h after serum stimulation reduced p-PKB levels at 9 and 12 h (Fig. 2A), suggesting that Ras activation is involved in late-G₁ PI3K activation. Addition of mevalonate, a lovastatin substrate, restored PKB phosphorylation (Fig. 2A). Herbimycin, a Src Tyr-K inhibitor (39), also reduced PKB activation at 9 and 12 h (Fig. 2A), whereas genistein, an inhibitor with high specificity for the epidermal GF receptor (40), did not affect the second G₁ PI3K activity peak (not shown). Combination of lovastatin and herbimycin treatments yielded a larger p-PKB reduction (Fig. 2A). The decrease in p-PKB levels correlated with a reduction in both c-Myc content and S phase entry at 12 h (Fig. 2A). The activities of the inhibitors in blocking Tyr-K and Ras activation were confirmed by WB; herbimycin reduced phospho-Tyr cellular content (Fig. 2B), and lovastatin reduced the active-Ras fraction, an effect that was reversed by mevalonate addition (Fig. 2C).

To confirm that late-G₁ PI3K activation requires Tyr-K and Ras, we examined PI3K activity in extracts from synchronous-cell cultures (as described above). Addition of lovastatin, herbimycin, or both at 4 h after serum addition resulted in reduced PI3K activity at 7 to 8 h after serum addition (Fig. 2D). As PI3K activity regulates cyclin levels (see below), we also examined whether lovastatin and herbimycin affected cyclin D3, E, and A levels. Tyr-K and Ras inhibition reduced the levels of these cyclins, most markedly those of cyclin A (Fig. 2E). Thus, Tyr-K and Ras cooperate in the induction of the second PI3K activity peak, which in turn regulates c-Myc and G₁ cyclin levels as well as S phase entry.

Late-G₁ PI3K inhibition reduces c-Myc and cyclin A levels as well as Cdk2 activity.

As late-G₁ PI3K activation correlates with increased c-Myc expression levels, we examined the consequences of inhibiting late-G₁ PI3K on c-myc mRNA levels by Northern blotting. Cells were synchronized as described above, and the PI3K inhibitor Ly294002 was added at 7 h after serum stimulation; cells were harvested at 9 h. This analysis showed that late-G₁ PI3K inhibition induces a reduction of c-myc mRNA levels of about 15% ± 5% (at 9 h, the mean value for three experiments) (Fig. 3A), whereas c-Myc protein reduction was systematically greater than 50%. In fact, late-G₁ PI3K inhibition markedly reduced c-Myc protein levels at 9 to 18 h poststimulation (Fig. 3A and B).

PI3K/PKB inhibits GSK3β (41), a kinase that phosphorylates c-Myc at Thr 58, thereby triggering Myc degradation (43). PI3K inhibition notably enhanced Thr 58-c-Myc phosphorylation, an event that correlated with c-Myc level reduction (Fig. 3B). The decrease in c-Myc protein levels correlated with the diminished expression of the Myc-transcriptional effectors cyclin D2 and Cdk4 (Fig. 3B). c-Myc stability was further examined in pulse-chase assays. Inhibition of PI3K activity in early G₁ (3 h after serum stimulation) blocked protein synthesis (not shown). Thus, for pulse-chase, we synchronized cells, labeled them with [³⁵S]Met between 3 and 9 h after serum addition, and harvested them at 9 h. At this time, the medium was replaced with nonradiolabeled Met/Cys-rich medium, alone or with Ly294002, and cells were collected at 12 and 16 h (Fig. 3C). For the sample treated with Ly294002 at the 9-h time point, the inhibitor was added 30 min before cell harvesting. PI3K inhibition greatly reduced c-Myc stability, an effect that was already evident 30 min after enzyme inhibition (Fig. 3C).

To further define the role of the second G₁ PI3K activity peak in S phase entry, cells were synchronized in G₀/G₁ and PI3K was inhibited at 7 h after serum stimulation; we examined the consequences for G₁-phase cyclin levels at different time points. Inhibition of late-G₁ PI3K activity greatly reduced cyclin A protein levels at 9 h, whereas cyclin D3 and E levels were more affected when Ly294002 had been present for prolonged periods (Fig. 3D). The Ras inhibitor lovastatin had a greater effect than Ly294002 in reducing cyclin D3 levels (Fig. 2E and 3D); this is likely due to the Ras/MAPK dependence for cyclin D synthesis (16).

We next tested the effect of inhibiting PI3K on Cdk2 activity. The Cdk2 substrate retinoblastoma protein (Rb) was hypophosphorylated following late-G₁ PI3K inhibition (Fig. 3D). Consistently, both cyclin E/Cdk2 and cyclin A/Cdk2 activities decreased after PI3K inhibition (Fig. 3E). The decrease in cyclin A expression paralleled the reduction of cyclin A/Cdk2 activity. Nonetheless, late-G₁ PI3K inhibition affected cyclin E/Cdk2 activity more markedly than cyclin E levels. We thus examined whether PI3K inhibition reduced cyclin E/Cdk2 activity by enhancing its association with the Cdk2 inhibitor p27^kip. PI3K inhibition increased the amount of p27^kip bound to Cdk2 (see below), explaining the reduction of cyclin E/Cdk2 activity by late-G₁ PI3K inhibition. Cyclin E/Cdk2 activity is required for loading of MCM2 onto chromatin (11, 13); PI3K inhibition in advanced G₁ resulted in a notable reduction in the amount of chromatin-bound MCM2 (Fig. 3F).

Conditional c-Myc-ER activation rescues S phase entry in PI3K inhibitor-treated cells.

PI3K inhibition reduced cyclin A levels and Cdk2 activity. c-Myc regulates G₁ cyclin expression, especially that of cyclin A, and the association of p27^kip with cyclin/Cdk2 (24, 29, 42). We thus hypothesized that the main function of late-G₁ PI3K activity may be to regulate c-Myc levels. To test this possibility, we used a c-Myc-estrogen receptor fusion protein (c-Myc-ER) that translocates to the nucleus after addition of an estrogen analogue such as 4-OHT (20). We examined whether the S phase entry defects induced by late-G₁ PI3K inhibition were counteracted by c-Myc-ER induction. Cells were infected with c-Myc-ER-expressing viruses (Fig. 4A), arrested in G₀, and released by serum addition. Some of the cells were treated with 4-OHT alone (at 6.5 h), with Ly294002 (at 7 h after serum addition), or with both. We collected cells at different times and examined S phase entry.

c-Myc-ER expression did not trigger S phase entry in the absence of serum (Fig. 4B and C). After serum addition, c-Myc-ER expression caused a slight increase in S phase entry compared to that in control cells, which was moderately enhanced upon 4-OHT addition (Fig. 4C). We found no notable differences when 4-OHT was added at 0 or 6.5 h (not shown). Ly294002 treatment reduced the proportion of cells in S phase by 50%, in both c-Myc-ER- and control ER vector-expressing cells (Fig. 4C). Nonetheless, c-Myc-ER induction at 6.5 h in cells treated with Ly294002 in advanced G₁ (7 h) showed almost normal S phase entry levels (∼85% recovery) compared to what was found for Ly294002-treated control cells (Fig. 4B and C). Induction of c-Myc-ER failed to compensate for the action of PI3K in S phase entry when PI3K was inhibited in early G₁ (0 to 4 h poststimulation) (Fig. 4C and data not shown). Examination of BrdU incorporation confirmed that c-Myc induction at 6.5 h counteracts S phase entry defects in cells treated with Ly294002 in advanced G₁ (Fig. 4D). Comparable results were obtained using the PI3K inhibitor wortmannin (not shown). These data suggest that the main function of late-G₁ PI3K activity is to regulate c-Myc protein levels.

Expression of a GSK3β-resistant c-Myc mutant rescues the cell cycle entry defects induced by inhibiting late-G₁ PI3K activity.

PI3K/PKB inactivate GSK3β, an enzyme that targets c-Myc for degradation (43). To confirm that c-Myc stabilization is the main role of PI3K activity in late G₁, we examined the effect of inhibiting PI3K in cells expressing the c-Myc_T58A substitution mutant, which is resistant to GSK3β action (14).

Cells were transfected with GFP control vector or with cDNAs encoding GFP fused to wild-type (WT) c-Myc or c-Myc_T58A (Fig. 5A). Transfected cells were sorted, and cultures were synchronized, released from arrest, and treated with Ly294002 at 7 h after serum addition. Cells were harvested at different time points. Overexpression of either WT c-Myc or c-Myc_T58A induced apoptosis and cell cycle entry in the absence of serum (Fig. 5B). Late-G₁ PI3K inhibition reduced cell cycle entry in control cells and to a lesser extent in cells overexpressing WT c-Myc; c-Myc_T58A expression, however, largely restored cell cycle entry (Fig. 5B). To reduce c-Myc expression levels, cells were infected with viruses encoding c-Myc_T58A. Under these conditions, c-Myc_T58A did not significantly induce cell cycle entry in the absence of serum (Fig. 5C). Synchronous-cell-infected cultures were treated with Ly294002 at 7 h after serum addition and harvested at different time points. PI3K inhibition blocked cell cycle entry in control cells, but cell cycle entry was nearly normal in cells expressing c-Myc_T58A (Fig. 5D). These results indicate that a stable form of c-Myc substitutes for PI3K action in late G₁.

c-Myc_T58A expression rescues cyclin A expression, Cdk2 activity, and MCM2 loading defects induced by late-G₁ PI3K inhibition.

To confirm that the primary effect of PI3K activity in advanced G₁ is to stabilize c-Myc, we examined whether c-Myc_T58A expression compensated for the cell cycle entry defects induced by late-G₁ PI3K inhibition. PI3K inhibition moderately affected cyclin D3 and E expression levels (see above). Similarly, c-Myc_T58A expression did not markedly alter cyclin D3 (not shown) or cyclin E levels (Fig. 6A). In contrast, cyclin A expression levels were greatly reduced upon late-G₁ PI3K inhibition (Fig. 6A). c-Myc_T58A expression increased cyclin A expression in Ly294002-treated cells and moderately increased basal cyclin A levels (Fig. 6A). Moreover, whereas hyperphosphorylated Rb levels, cyclin E/Cdk2, and cyclin A/Cdk2 kinase activities were reduced by late-G₁ PI3K inhibition in control cells, they were virtually unaffected in c-Myc_T58A-expressing cells (Fig. 6A to C). As c-Myc controls the levels of Cdk2 bound to p27^kip (29, 42), we tested whether p27^kip-Cdk2 association was affected by c-Myc_T58A expression. Ly294002 treatment at 7 h in synchronous-cell cultures increased the association of p27^kip with cyclin E/Cdk2 in controls, but association was lower and resistant to PI3K inhibition in c-Myc_T58A-expressing cells (Fig. 6D). Similar results were obtained using c-Myc-ER-expressing cells treated with 4-OHT (at 6.5 h), Ly294002 (at 7 h), or both simultaneously (Fig. 6E and data not shown). In fact, c-Myc-ER induction corrected the defects in S phase entry, cyclin A expression, and Rb phosphorylation induced by late-G₁ PI3K inhibition (Fig. 6E).

We also examined the consequences of expressing c-Myc_T58A on the loading of MCM2 onto chromatin. In control cells, MCM2 loading was still low at 9 h (similar to that observed at 0 h), increased at 12 to 16 h, and was blocked by PI3K inhibition. In contrast, in c-Myc_T58A cells, MCM2 loading increased by 9 h and remained insensitive to late-G₁ PI3K inhibition (Fig. 7).