G1 cyclin/cyclin-dependent kinase-coordinated phosphorylation of endogenous pocket proteins differentially regulates their interactions with E2F4 and E2F1 and gene expression.

Mitogenic stimulation leads to activation of G(1) cyclin-dependent kinases (CDKs), which phosphorylate pocket proteins and trigger progression through the G(0)/G(1) and G(1)/S transitions of the cell cycle. However, the individual role of G(1) cyclin-CDK complexes in the coordinated regulation of pocket proteins and their interaction with E2F family members is not fully understood. Here we report that individually or in concert cyclin D1-CDK and cyclin E-CDK complexes induce distinct and coordinated phosphorylation of endogenous pocket proteins, which also has distinct consequences in the regulation of pocket protein interactions with E2F4 and the expression of p107 and E2F1, both E2F-regulated genes. The up-regulation of these two proteins and the release of p130 and pRB from E2F4 complexes allows formation of E2F1 complexes not only with pRB but also with p130 and p107 as well as the formation of p107-E2F4 complexes. The formation of these complexes occurs in the presence of active cyclin D1-CDK and cyclin E-CDK complexes, indicating that whereas phosphorylation plays a role in the abrogation of certain pocket protein/E2F interactions, these same activities induce the formation of other complexes in the context of a cell expressing endogenous levels of pocket and E2F proteins. Of note, phosphorylated p130 "form 3," which does not interact with E2F4, readily interacts with E2F1. Our data also demonstrate that ectopic overexpression of either cyclin is sufficient to induce mitogen-independent growth in human T98G and Rat-1 cells, although the effects of cyclin D1 require downstream activation of cyclin E-CDK2 activity. Interestingly, in T98G cells, cyclin D1 induces cell cycle progression more potently than cyclin E. This suggests that cyclin D1 activates pathways independently of cyclin E that ensure timely progression through the cell cycle.

Ectopic expression of cyclin D1 and cyclin E in primary or immortal, nontransformed mammalian fibroblasts shortens the G 1 phase of the cell cycle (15)(16)(17). The relatively modest effects of ectopic expression of G 1 cyclins in primary or immortal, nontransformed mammalian fibroblasts are probably due to a requirement for additional events to ensure full activation of these complexes. Whereas cyclins are limiting subunits for activation of their corresponding CDKs, full activation of cyclin-CDK complexes requires other events also dependent upon mitogenic stimulation (reviewed in Refs. 18 -20). In agreement with this idea, microinjection of purified recombinant active cyclin D1-CDK4 or cyclin E-CDK2 complexes in human primary lung fibroblasts bypasses the requirement for mitogenic signaling (21). It has been suggested that cyclin D1-CDK effects trigger pRB inactivation and activation of E2F-dependent genes including cyclin E (21), whereas cyclin E-CDK2 effects might be independent of E2F (22,23).
Initial studies suggested that pRB was the only substrate of D-type cyclin-CDKs required for a p16-mediated G 1 arrest (24 -26). However, more recent studies demonstrated that, similar to the lack of pRB, lack of both p130 and p107 also prevents a p16-mediated G 1 arrest in primary mouse embryo fibroblasts, strongly suggesting that the three pocket proteins are important substrates for the cell cycle-regulatory functions of D-type cyclin-CDKs. Supporting this hypothesis further, it has been shown that both p130 and p107 are phosphorylated in vivo by D-type cyclin-CDK complexes at specific residues, which are not phosphorylated by cyclin E-CDK2 complexes (27,28). Phosphorylation of these sites modulates the growth suppressor activities of p130 and p107. It is likely that two parallel pathways, one controlled by pRB and the other by p130/p107, regulate distinct downstream events required for G 1 progression into S phase. In agreement with this possibility, different sets of E2F-dependent genes are deregulated in mouse embryo fibroblasts lacking pRB and mouse embryo fibroblasts lacking both p130 and p107 (29).
It is currently thought that pRB interacts with E2F1, E2F2, E2F3, and E2F4, whereas p130 and p107 interact with E2F4 but not with E2F1-3 (reviewed in Refs. 14 and 30). The unique ability of pRB to specifically interact with E2F1-3 supports the hypothesis that pRB controls a pathway, which is different from the pathway controlled by p130/p107.
Although it seems clear that cells will synthesize DNA when a certain threshold of cyclin D-CDK or cyclin E-CDK2 activity is induced in a cell, the concerted, as well as individual, effects of G 1 cyclin-CDK activities on the coordinated phosphorylation of the three endogenous pocket proteins and the subsequent effects on E2F-dependent gene expression are not well understood. To address these questions in more detail, we have studied the downstream events induced by both cyclin D1 and cyclin E and their dependence on each other in human T98G and Rat-1 cells. Both cyclins are sufficient to induce mitogenic independent growth. The effects of both cyclins in endogenous pocket protein phosphorylation are clearly distinct. Phosphorylation of certain pocket proteins is sufficient to disrupt, at least partially, a subset of pocket protein-E2F complexes. However, because of the cell cycle-coordinated expression of endogenous members of the pocket protein and E2F families, new pocket protein-E2F complexes are formed even in the presence of active G 1 CDKs. Surprisingly, we have found that both p130 and p107 specifically interact with E2F1 in a cell cycle-dependent manner. This work demonstrates that the interactions between pocket proteins and E2F family members are more complex than hereto anticipated and suggests cross-talk between these pathways.

EXPERIMENTAL PROCEDURES
Cell Culture and Cell Cycle Synchronization-T98G and Rat-1 cells were maintained in Dulbecco's modified Eagle's medium (Cellgro) supplemented with 10% fetal bovine serum (FBS) (Sigma) at 37°C in a humidified atmosphere with 5% CO 2 . Cells were synchronized in G 0 phase by contact inhibition followed by serum starvation. Briefly, cells were grown and kept overconfluent for 2 days. Cells were then trypsinized, counted, and seeded at 2 ϫ 10 6 cells/plate in 100-mm dishes in MCDB-105 medium without FBS (Sigma). After 12 h, medium was removed, and fresh Dulbecco's modified Eagle's medium without FBS was added. Cells were kept for another 60 h before starting experiments.
Roscovitine was used to specifically inhibit CDK2 activity in cells infected with cyclin D1 adenoviruses. Infections were performed in the presence of either the CDK2/CDC2 inhibitor roscovitine (31) at a concentration of 25 M, or vehicle (Me 2 SO).
For G 2 /M blockage and release experiments, serum-starved T98G cells restimulated with 10% FBS or infected with the indicated adenovirus for 22 h were incubated in the presence or absence of nocodazole (10 M) for an additional 22 h. When indicated, G 2 /M-synchronized cells were shaken off as previously described (32) and then reseeded in fresh medium. Cells were collected at the time points indicated under "Results" and processed for flow cytometric and/or Western blot analysis.
Adenovirus Production and Infection-Recombinant adenoviruses encoding cyclin D1 and cyclin E (Ad-Cyc D1 and Ad-Cyc E) were provided by J. Albrecht (33). Adenoviruses encoding p16 (Ad-p16) were provided by J. Fueyo. Adenoviruses encoding p21 (Ad-p21) were provided by W. El-Deiry. Adenoviruses encoding enhanced green fluorescent protein (Ad-EGFP) were a gift of P. Ruiz. Viral stocks were amplified using 293 cells and were purified by using CsCl density gradient centrifugation. Viral titers were determined by plaque assay (34). Titers obtained ranged from 5 ϫ 10 9 to 5 ϫ 10 10 plaque-forming units/ml. Infection conditions and optimal multiplicity of infection were previ-ously determined using adenovirus-carrying reporter genes (lacZ, EGFP). T98G and Rat-1 cells were infected at a multiplicity of infection of 50 -100 plaque-forming units/cell and 20 plaque-forming units/cell, respectively, for each adenovirus used. Infections were carried out by direct addition of the necessary volume of adenovirus stock to the medium.
Western Blot Analysis and Immunoprecipitations-Whole protein lysates were obtained essentially as described previously (12,35) by lysing cells in buffer containing 50 mM Tris-Cl (pH 7.4), 5 mM EDTA, 250 mM NaCl, 50 mM NaF, 0.1% Triton X-100, 0.1 mM Na 3 VO 4 , 2 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 4 g/ml aprotinin, and 4 g/ml pepstatin (lysis buffer). For cyclin D1 kinase assays and certain immunoprecipitations, we used DIP buffer (36) (see below). Immunoprecipitations and Western blots were performed as previously described (12,35). Briefly, protein extracts (250 -500 g) were incubated for 1 h at 4°C with specific antibodies, and immunocomplexes were precipitated with 25 l of Protein A-Sepharose beads (Pierce) for 2 h at 4°C and washed four times. Complexes were eluted from beads by adding 1.5 ϫ Laemmli sample buffer. Whole cell lysates or immunocomplexes were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Immobilon-P; Millipore Corp.) in 10 mM CAPS (pH 11) containing 10% methanol. 10 and 12% gels were run to determine the expression of cyclins, p16, p21, p27, and E2F proteins; and 6 and 8% gels to determine the phosphorylation status of pocket proteins. The transferred membranes were probed with specific primary antibodies and the corresponding horseradish peroxidase-conjugated secondary antibody (Amersham Biosciences). Bands were visualized by incubating the membranes with Enhanced Chemiluminescence reagent (PerkinElmer Life Sciences) and exposing the membranes to x-ray film.
In Vitro Kinase Assays-CDK2 kinase activity was determined from CDK2, cyclin E, and cyclin A immunopurified complexes as described previously (37,38). Complexes were immunoprecipitated with specific antibodies from whole protein extracts (40 g for CDK2 and 200 g for cyclin E). Kinase assays were performed at 30°C for 30 min in 20 mM HEPES-Na (pH 7.4), 10 mM magnesium acetate, 1 mM dithiothreitol, 20 M ATP (10 Ci/reaction), and 1 g of substrate (histone H1). CycD1 kinase activity was determined as described (36). Briefly, cells were lysed in DIP buffer (50 mM HEPES, pH 7.2, 150 mM NaCl, 1 mM EDTA, 2.5 mM EGTA, 10% glycerol, and 0.1% Tween 20) containing freshly added 1 mM dithiothreitol, 1 mM NaF, 0.5 mM phenylmethylsulfonyl fluoride, 0.5 mM Na 3 VO 4 , 10 mM ␤-glycerophosphate, 1 g/ml leupeptin, and 1 g/ml aprotinin. Active complexes were immunoprecipitated from 800 g of cell lysate by using CycD1 monoclonal antibodies, collected with protein A-Sepharose (30 l of 50% slurry containing beads preequilibrated in DIP with 4% bovine serum albumin), and washed four times with DIP buffer containing fresh inhibitors and two additional times with kinase buffer (50 mM HEPES, pH 7.2, 10 mM MgCl 2 , 5 mM MnCl 2 , and freshly added 1 mM dithiothreitol). Kinase assays were performed at 30°C for 30 min in kinase buffer containing 20 M ATP (10 Ci/reaction) and 1 g of substrate (GST-pRb C terminus). In all cases, reactions were stopped by adding 1 volume of 2ϫ Laemmli sample buffer and boiling samples for 10 min. Substrates were resolved in 12% SDS-PAGE, and gels were dried and visualized by autoradiography.
Flow Cytometry Analysis-All measurements were performed on a FACScan (BD Pharmingen). For DNA content quantification, cells were processed as described earlier (12). For the cell proliferation assay, cells were plated (see above) in serum-free medium, and 24 h later cells were washed twice with phosphate-buffered saline and stained with 20 M carboxymethyl fluorescein diacetate succinyl ester (Molecular Probes, Inc., Eugene, OR) in phosphate-buffered saline for 5 min at room temperature. CFSE incorporation was stopped by washing once with 10% FBS-containing medium and two additional times with serum-free medium. At time points indicated, cells were collected, washed with phosphate-buffered saline, and resuspended in 1% FBS-containing phosphate-buffered saline, and green fluorescence was quantified. Data from 10,000 cells were collected and analyzed using ModFitLT version 2.0 software.

RESULTS
We were interested in determining the effects that expression of G 1 cyclins have on the phosphorylation status of endogenous pocket proteins and the subsequent effects on pocket protein-E2F complexes, expression of E2F-dependent and independent gene products, and, ultimately, entry and progression throughout the cell cycle. To this end, we chose an immor-tal human cell line (T98G cells), which has been extensively used in cell cycle studies. We also used the rat embryo fibroblast cell line Rat-1. Both T98G and Rat-1 cells are effectively arrested in a quiescent state by serum starvation, and subsequent mitogenic stimulation leads to synchronous cell cycle entry. To ectopically express G 1 cyclins and cyclin kinase inhibitors, we utilized replication-defective recombinant adenoviruses. Under our experimental conditions, virtually all cells are transduced by the recombinant adenoviruses at multiplicity of infection values of 50 -100 as estimated by using recombinant adenoviruses encoding the green fluorescent protein and ␤-galactosidase (data not shown). Initial experiments demonstrated that transduction of serum-starved T98G cells with recombinant adenoviruses encoding cyclins D1 (Ad-Cyc D1) or E (Ad-Cyc E), but not ␤-galactosidase control adenoviruses (Ad-␤-gal), for 48 h was sufficient to induce cell cycle entry in the absence of any mitogenic stimuli as determined by fluorescence-activated cell sorting analysis following propidium iodide staining (data not shown; see below). This was also confirmed using Rat-1 cells.
To determine the effects of ectopic expression of cyclin D1 and E alone or combined on cell cycle entry, T98G cells were serum-starved for 2 days and then stimulated with FBS or infected with the indicated recombinant adenoviruses at a multiplicity of infection of 50 (Fig. 1). Cells were harvested at the indicated time points and processed for fluorescence-activated cell sorting and Western blot analysis. The expression of the ectopically expressed cyclins is shown in Fig. 1A. Cells maintained in the absence of serum did not show changes in cell cycle distribution (Fig. 1B). Similar results were obtained with cells transduced with the ␤-galactosidase control adenovirus in the absence of serum (data not shown; see below). As expected, stimulation with 10% FBS led to synchronous cell cycle entry. More than 40% of the cells were in S phase by 22 h, and by 30 h most cells were in the next G 1 phase following mitosis (Fig. 1B). Interestingly, ectopic expression of cyclin D1 and cyclin E individually or together led to synchronous cell cycle entry and progression throughout S phase. Cyclin D1 appeared more potent in inducing cell cycle progression than cyclin E (compare panels 3 and 4). Moreover, when both cyclins were coexpressed, the kinetics of cell cycle progression were very similar to that induced by cyclin D1 alone (compare panels 3 and 5). To determine whether the expression of cyclins D1 and E alone or combined was able to induce changes in the patterns of phosphorylation of pocket proteins similar to those changes induced by stimulation with FBS, we resolved protein extracts by 6% same experiment were collected and fixed with cold ethanol, and DNA content was determined by propidium iodide staining followed by flow cytometric analysis. Light gray bars represent the percentage of cells in G 0 -G 1 phase, dark gray bars represent the percentage of cells in S phase, and white bars represent cells in G 2 phase. C, the same lysates used in A were separated by 6% SDS-PAGE to resolve phosphorylated forms of pocket protein followed by Western blot analysis with the indicated antibodies. p130-phosphorylated forms are indicated as described earlier (12,32) and in subsequent studies. p107-P and pRB-P, hyperphosphorylated p107 and pRB, respectively. Note that the anti-pRB antibody mixture used for this experiment showed a preference for the hypophosphorylated form of pRB. SDS-PAGE followed by Western blot analysis with specific antibodies. Fig. 1C shows that expression of both cyclins appears sufficient to induce hyperphosphorylation of the three pocket proteins (see below). p130-phosphorylated forms found in quiescent T98G cells consist of forms 1 and 2 (12,32). Stimulation with FBS triggers phosphorylation of p130 to form 3 in mid-G 1 (8 -10 h in T98G cells) (Fig. 1C, panel 2) (12). p107 is expressed at low levels, and it is found in its hypophosphorylated form in quiescent cells. p107 becomes hyperphosphorylated and expressed at higher levels concomitantly with p130 hyperphosphorylation. pRB also becomes hyperphosphorylated in mid-G 1 . (Note that the anti-pRB mixture of monoclonal antibodies used in this experiment had a preference for the hypophosphorylated form; whereas the appearance of hyperphosphorylated forms is clear, the change in intensity of the different forms does not reflect a change in the expression of pRB.) Ectopic expression of cyclin D1 in the absence of FBS was apparently sufficient to induce the same changes in protein phosphorylation induced by serum stimulation (Fig. 1C, compare panels 2 and 3). Of note, cyclin E appeared less effective in inducing changes in pocket protein phosphorylation and cell cycle progression. Finally, coexpression of cyclins D1 and E exhibited the same effects of cyclin D1 alone. The results of this experiment demonstrate that ectopic expression of either cyclin D1 or cyclin E alone is sufficient to bypass the growth factor requirements necessary to trigger hyperphosphorylation of pocket proteins and cell cycle entry in serum-starved quiescent T98G cells. However, because activation of cyclin D1-CDK complexes is likely to lead to activation of endogenous cyclin E-CDK2 complexes and vice versa, we could not discern the individual effects of each cyclin on pocket protein phosphorylation and cell cycle progression.
Similar experiments were also performed using Rat-1 cells. The expression of either cyclin D1 or cyclin E was sufficient to trigger phosphorylation of pocket proteins and cell cycle progression in the absence of serum (data not shown).
Induction of Individual G 1 CDK Activities in Quiescent T98G Cells-A strategy to induce activation of cyclin D1-CDK complexes in the absence of CDK2 activity or activation of cyclin E-CDK2 complexes in the absence of D-type cyclin-CDK activity was designed. Our strategy consists of transducing serumstarved T98G cells with recombinant adenoviruses expressing either p16 (an inhibitor of D-type cyclin-CDK activity) (39), p21 (an assembly factor for D-type cyclin-CDK complexes and a potent inhibitor of cyclin E-CDK2 complexes) (reviewed in Ref. 20), or Ad-␤-gal (control). 5 h after transduction, (a) the cells infected with Ad-p16 were infected with Ad-Cyc E, (b) the cells infected with Ad-p21 were infected with Ad-Cyc D1, and (c) the cells infected with Ad-␤-gal were infected with Ad-Cyc D1, Ad-Cyc E, or Ad-␤-gal. These combinations were predicted to generate both individual and cooperative G 1 CDK activities in vivo. To determine whether this was the case, cells were harvested at the indicated time points after transduction with the second group of recombinant adenoviruses (Fig. 2). Whole cell protein lysates were obtained and used to perform kinase assays. Fig. 2A shows efficient expression of the ectopically expressed proteins, as determined by Western blot analysis. Under the shown exposure times, endogenous cyclin D1 is not detected (see Fig. 1). We determined the kinase activities associated with cyclin D1, cyclin E, and CDK2 by immunoprecipitating the kinase complexes with specific antibodies and performing kinase assays with the immunoprecipitates using the C-terminal domain of pRB and histone H1 as exogenous substrates as indicated in Fig. 2B. As expected, stimulation of T98G cells with FBS led to up-regulation of cyclin D1-, cyclin E-, and CDK2-associated kinase activities (lanes 1-3). Ectopic expression of cyclin D1 resulted in high levels of cyclin D1associated kinase activity, which was followed by induction of cyclin E-and CDK2-associated kinase activities (lanes 10 and 11). The differences between cyclin E-and CDK2-associated kinase activities are that the latter consists of cyclin E-CDK2 and cyclin A-CDK2 activities. Of note, coexpression of cyclin D1 and p21 also leads to induction of cyclin D1-associated kinase activity to levels comparable with those induced by serum stimulation (compare lanes 2 and 3 with lanes 12 and 13). However, cyclin D1 kinase activity was lower in cells coexpressing cyclin D1 and p21 than in cells expressing cyclin D1 alone (compare lanes 10 and 11 with lanes 12 and 13). This suggests that p21, when expressed at high levels, inhibits cyclin D1-CDK activity (20). Importantly, cyclin D1 activation in cells coexpressing p21 is not accompanied by activation of cyclin E or CDK2-associated kinase activities (compare lanes 10 and 11 with lanes 12 and 13). Thus, cotransfection of cyclin D1 and p21 results in individual activation of cyclin D1-CDK complexes. An independent strategy to activate D-type cyclin-CDK activity in the absence of cyclin E-CDK2 activity consisted of treating cells expressing exogenous cyclin D1 with roscovitine, an inhibitor of CDK2/CDC2 (see below).
On the other hand, expression of cyclin E leads to induction of cyclin E and CDK2-associated kinase activities and, to a lesser extent, cyclin D1-associated kinase activity (compare lanes 6 and 7 with lanes 4 and 5). Notably, coexpression of cyclin E and p16 prevented activation of cyclin D1 kinase activity but had little effect on the cyclin E-and CDK2-associated kinase activities (compare lanes 6 and 7 with lanes 8 and  9). Thus, we can selectively induce activation of cyclin E-CDK2 complexes without inducing D-type cyclin-CDK complexes.
Effects of Collective or Selective Activation of G 1 Cyclin-CDK Activities on the Phosphorylation Status of Endogenous Pocket Proteins in the Absence of Mitogens-Next, we determined the effects that collective or selective activation of G 1 cyclin-CDK activities had on the phosphorylation status of the three endogenous pocket proteins in comparison with the changes induced by stimulation with FBS at the indicated time points. Fig. 3A shows that whereas expression of cyclin D1 leads to hyperphosphorylation of the three pocket proteins, coexpression of cyclin D1 with p21, which blocks CDK2 activity, prevents hyperphosphorylation of both p130 and pRB (compare lanes 14 -16 with lanes 17-19). Interestingly, under these same conditions, p107 is efficiently hyperphosphorylated. However, the induction of p107 protein expression, which is believed to result from disruption of E2F-dependent repression of the p107 gene promoter, is mostly inhibited (see below). Similar results were obtained when T98G cells transduced with cyclin D1 were treated with roscovitine, which inhibits CDK2/CDC2 activity (Fig. 3B). Roscovitine blocked p107 up-regulation and inhibited S phase entry, but cyclin D1-induced hyperphosphorylation of p107 was undisturbed. Thus, the same result was obtained using two different strategies that block cyclin E-CDK2 activity in cyclin D1-overexpressing cells.
Similarly, as shown in Fig. 1C, cyclin E expression is sufficient to induce hyperphosphorylation of the three pocket proteins (Fig. 3A, lanes 8 -10), although it appears significantly less potent than cyclin D1 (Fig. 3A, compare lanes 8 -10 with   FIG. 3. Complete hyperphosphorylation of pRb and p130, but not p107, requires cooperation of cyclin D1-and cyclin E-associated kinase activities. However, cyclin D1-CDK activity is sufficient to hyperphosphorylate p107. Cell lysates were obtained as in Fig. 2 and resolved in 6% SDS-PAGE to clearly separate the different phosphorylated forms of each pocket protein. A, immunodetection of p130 (upper), p107 (middle) and pRb (lower panel). p130 phosphorylated forms are indicated as described earlier (12,32) and subsequent studies. p107-P and pRB-P, hyperphosphorylated p107 and pRB, respectively. *, phosphorylated form 2b of p130 (12,40), present in cells expressing exogenous p16 and cyclin E. B, T98G cells transduced with cyclin D1 adenoviruses for the indicated periods of time were incubated in the presence of the CDK2 inhibitor roscovitine or vehicle (Me 2 SO; DMSO). Protein lysates were analyzed by Western blot (WB) as in A by using anti-p107 antibodies. The upper and lower panels show two different exposures from the same Western blot to allow visualization of hyper-and hypophosphorylated forms of p107, which are expressed at very different levels (L denotes long exposure and S denotes short exposure). The percentage of cells in S phase was determined by flow cytometric analysis following propidium iodide staining. lanes 14 -16). Of note, coexpression of cyclin E and p16, which results in activation of cyclin E-CDK2 complexes in the absence of D-type cyclin-CDK activity, leads only to partial phosphorylation of p130 and pRB. Under the same conditions, p107 appears to remain in its hypophosphorylated form, but induction of its protein expression is unperturbed (Fig. 3A, lanes 11-13). p130 is partially hyperphosphorylated to form 2b (indicated by an asterisk), a partially phosphorylated form of p130 that we initially identified in 293 cells (12) and later found in other E1A-expressing cells (40). p130 form 2b migrates faster than form 3, but slightly more slowly than form 2. It is now established that this form of p130 is detected in cells in which D-type cyclin activity is inhibited by different mechanisms (41,42). These results also highlight that even when G 1 cyclins are overexpressed, the resulting CDK activities continue to exhibit selectivity in the phosphorylation of pocket proteins.
The differential phosphorylation of the three pocket proteins by the selective activation of G 1 CDK activities suggests that the disruption of pocket protein-E2F interactions also might be differentially affected by these kinase activities.
Regulation of Pocket Protein/E2F Interactions by G 1 CDKs-To explore this possibility, we first determined changes in p130/E2F4 interactions by performing immunoprecipitations followed by Western blot analysis using specific antibodies. Fig.  4A shows that 15 h following serum stimulation, p130-E2F4 complexes are mostly disrupted as a result of p130 hyperphosphorylation to form 3, as previously described (32). Expression of cyclin D1 in the absence of serum was sufficient to completely disrupt these complexes (Fig. 4A, lanes 10 and 11). However, coexpression of p21 effectively blocked complex disruption, strongly indicating that cyclin D1-and cyclin E-associated kinase activities cooperate to abrogate this interaction (Fig. 4A, compare lanes 10 and 11 with lanes 12 and 13). Expression of cyclin E alone led to partial disruption of p130-E2F4 complexes, although it was less efficient than cyclin D1 (Fig. 4A, compare lane 1 with lanes 6 and 7 and lanes 10 and  11). p16 had little effect on cyclin E-mediated disruption of these complexes (Fig. 4A, lanes 8 and 9). Altogether, these results demonstrate that cyclin D1-CDK and cyclin E-CDK2 complexes cooperate to disrupt p130-E2F4 complexes. It is also shown that cyclin E-CDK2 without cyclin D1-CDK activity, but not cyclin D1-CDK without cyclin E-CDK2 activity, can induce at least partial disruption of p130-E2F4 complexes.
We next determined the regulation of p107-E2F4 and pRB-E2F4 complexes in comparison with p130-E2F4 complexes (Fig. 4B, upper panels). Interestingly, conditions leading to abrogation of p130-E2F4 complexes led to similar disruption of pRB-E2F4 complexes. In contrast, p107-E2F4 complexes accumulated following disruption of p130-E2F4 and pRB-E2F4 complexes as a result of serum stimulation or activation of G 1 CDK activities. This indicates that although hyperphosphorylation of p130 and pRB is very effective at disrupting their interaction with E2F4, the increased expression of p107 appears to facilitate that a pool of hypophosphorylated p107 remains available for interaction with E2F4.
We then analyzed the formation/abrogation of pocket protein-E2F1 complexes (Fig. 4B, lower panels). Four major observations were made. First, disruption of pRB-E2F4 complexes coincides with formation of pRB-E2F1 complexes. Second, at least some partially hyperphosphorylated forms of pRB are detected in association with E2F1, although there is a selective preference for the hypophosphorylated form (compare pRB forms in E2F1 immunoprecipitates and E2F1-depleted lysates) (Fig. 4B, lower panels, and Fig. 4C; see also Fig. 5B). Third, and more surprisingly, disruption of p130-E2F4 and pRB-E2F4 complexes is followed by formation of p130-E2F1 and p107-E2F1 complexes (Fig. 4B). Fourth, the primary form of p130 that interacts with E2F1 is hyperphosphorylated and co-migrates with form 3 (longer exposures revealed trace levels of forms 1 and 2 migrating faster than the major form associated with E2F1) (data not shown). In addition, the migration of the p130 form associated with E2F1 matches precisely the migration of p130 form 3 detected in E2F1 immunodepleted lysates (Fig. 4, B and C). Of note, the kinetics of association of pocket proteins with E2F1 is distinct as can be seen by comparing the patterns of pocket protein/E2F1 association induced by serum and transduction with cyclin D1 adenoviruses (Fig. 4B). Notably, we did not observe formation of p130-E2F1 and p107-E2F1 complexes in cells transduced with cyclin E, which is consistent with the diminished ability of cyclin E in inducing abrogation of p130-E2F4 and pRB-E2F4 complexes and the lesser up-regulation of E2F1 (Figs. 4, A and B, and 6A).
To confirm the specificity of the E2F1 interaction with p130 and p107, we performed additional immunoprecipitations with various different antibodies. Fig. 5A shows that p107 is specifically coimmunoprecipitated with E2F1, since the antigenic peptide used to raise the C20 anti-E2F1 polyclonal antibody blocks immunoprecipitation of p107-E2F1 complexes. The C20 polyclonal antibody was raised against a 20-amino acid peptide corresponding to the C terminus of E2F1. We also performed immunoprecipitations using a monoclonal antibody specific for E2F1, KH95, which has been extensively used for the analysis of E2F1 complexes. This antibody confirmed the results obtained with the C20 anti-E2F1 polyclonal antibody. KH95 effectively immunoprecipitated p107 and pRB with E2F1. p130 was detected less efficiently (Fig. 5B). Moreover, anti-p107 antibodies also coimmunoprecipitated E2F1, which was detected by Western blot analysis with both C20 polyclonal and KH95 monoclonal anti-E2F1 antibodies (Fig. 5C). Similarly, anti-p130 antibodies also coimmunoprecipitated E2F1 (data not shown). Thus, an unexpected interaction between p130/ p107 and E2F1 was observed in vivo, which results from changes in G 1 CDK-induced cell cycle-dependent expression of p107 and E2F1 as well as disruption of p130-E2F4 complexes.
Previous studies have failed to detect interactions between endogenous p107/p130 and E2F1 (reviewed in Ref. 14). This is likely because these complexes are not very abundant and are only transiently formed at specific times during the cell cycle. It is also possible that E2F1 has a higher affinity for pRB than for p107/p130. Our strategy to detect these complexes was based on (a) using cells synchronized at specific points of the cell cycle to prepare the protein extracts; (b) preparing protein extracts in DIP buffer, which was previously described for the efficient immunoprecipitation of active D-type cyclin-CDK complexes (36); and (c) performing immunoprecipitations with anti-E2F1 antibodies by using up to 500 g of protein extracts.
Changes in the Expression of E2F-dependent and -independent Gene Products-Given that activation of G 1 cyclin-CDK activities exhibit specific effects on the regulation of pocket protein-E2F complexes, we analyzed changes in E2F-dependent gene expression. The expression of p107, E2F1, cyclin E, and cyclin A was determined by Western blot analysis, since it is known that the cell cycle-dependent up-regulation of these four proteins is regulated at the transcriptional level by E2F activities (Fig. 6A). p107 Western blots are shown again for clarity and to allow for direct comparison of changes in expression. (The same p107 Western blots were included in Fig. 3A to show changes in phosphorylation.) Serum stimulation clearly leads to the up-regulation of the four proteins, although with different kinetics (Fig. 6A, panel 1). Ectopic expression of cyclin D1 induces similar up-regulation of the four proteins, although there are noticeably slower kinetics in comparison with FBS stimulation, which is clearest in the case of cyclin A up-regulation (Fig. 6A, panel 5). This is consistent with slower progression through the cell cycle (see Fig. 1 and below). Interestingly, coexpression of p21 effectively blocked the up-regulation of all four proteins (Fig. 6A, panel 6), which is in agreement with the inability of cyclin D1 to disrupt E2F4-containing complexes in complexes. The lower panel shows E2F4 protein levels in total cell lysates. B, protein extracts from the indicated time points were immunoprecipitated with anti-E2F4 (upper panels) or anti-E2F1 (lower panels), and immunocomplexes were resolved in 8% SDS-PAGE, blotted, and probed with the indicated antibodies. pRB-pP, partially phosphorylated pRB. The migration of the IgGs is also indicated for the E2F4 Western blot. C, protein lysates that had been immunodepleted with anti-E2F1 antibodies were resolved in 8% SDS-PAGE in parallel with the immunocomplexes shown in B, blotted, and probed with anti-p130 and anti-pRB antibodies. Differently phosphorylated forms of pocket proteins are indicated as in Fig. 1C. the absence of cyclin E-CDK2 activity (compare Figs. 2, 3, 4, and 6A). Ectopic expression of cyclin E also led to the expression of p107, E2F1, and cyclin A (Fig. 6A, panel 3). Importantly, coexpression of cyclin E with p16, which effectively blocks D-type cyclin-associated kinase activity, did not effectively block the up-regulation of p107, E2F1, and cyclin A induced by cyclin E expression (Fig. 6A, panel 4). Thus, activation of cyclin E-CDK2 activity in the absence of D-type cyclin-CDK activity is sufficient to induce expression of E2F-dependent genes. Notably, accumulation of hypophosphorylated p107 does not prevent the expression of E2F-dependent genes (Fig. 6A, panel 4) or cell cycle progression, although it might delay these processes.
Whereas it is clear that G 1 cyclin-CDK complexes are involved in regulating the expression of E2F-dependent genes, these CDKs also regulate the expression of other cell cycleregulatory proteins. Fig. 6B shows the expression of p27. It is well known that serum stimulation leads to p27 down-regulation, which facilitates the activation of cyclin E-CDK2 complexes. Interestingly, ectopic expression of cyclin D1 was sufficient to induce partial down-regulation of p27 (Fig. 6B,  compare lane 1 with lanes 10 and 11). Interestingly, cyclin D1-mediated down-regulation of p27 did not require the activity of CDK2 complexes, since p21 expression did not effectively block p27 down-regulation (Fig. 6B, lanes 12 and 13). Cyclin E expression, however, failed to consistently induce down-regulation of p27 (Fig. 6B, lanes 6 -9).
Mitogenic Independent Growth-Finally, we determined the cell cycle and apoptotic effects associated with the cooperative FIG. 5. p130 and p107 form specific complexes with E2F1. A, protein extracts from synchronized T98G cells were immunoprecipitated with polyclonal anti-E2F1 antibodies raised against a 20-amino acid peptide corresponding to the C terminus of E2F1 in the presence or absence of antigenic peptide. Immunocomplexes were resolved by 6 and 8% SDS-PAGE, transferred to nylon membranes, and probed with anti-p107 and anti-E2F1 antibodies, respectively. B, protein extracts from synchronized T98G cells or cells transduced with Ad-Cyclin D1 were immunoprecipitated with monoclonal anti-E2F1 antibodies (KH95). Immunocomplexes and E2F1-depleted protein extracts were resolved by 8% SDS-PAGE, transferred to nylon membranes, and probed with the indicated antibodies. C, protein extracts from synchronized T98G cells were immunoprecipitated with anti-p107 antibodies. Immunoprecipitates were resolved by 8% SDS-PAGE followed by Western blot analysis using two different anti-E2F1 antibodies (C20 and KH95).

FIG. 6.
Regulation of E2Fdependent and independent gene expression by G 1 cyclin-CDK activities. Cell lysates used in Figs. 2A and 3 were immunoblotted to determine the expression of the E2F-dependent genes p107, E2F1, cyclin A, and cyclin E (A) and p27 Kip1 (B). Note that the high levels of cyclin E expression detected in panels 3 and 4 correspond to exogenously expressed cyclin E. or selective activation of G 1 cyclin-CDK complexes. Cell cycle progression was determined by flow cytometric analysis following propidium iodide staining (Fig. 7A). Ectopic expression of cyclin D1 induced growth factor-independent cell cycle progression through G 1 , S phase, and mitosis as well as entering the next G 1 phase of the cell cycle. Entry into the following G 1 was detectable by 30 h. This was entirely inhibited by coexpression of p21. Similarly, we showed that roscovitine inhibited cyclin D1-induced entry into S phase (Fig. 3B). Cyclin E expression led to progression through G 1 and S and entry into mitosis. Coexpression of p16 did not block cyclin E-mediated induction of cell cycle progression.
We also determined whether activation of G 1 cyclin-CDK activities was sufficient to confer growth factor-independent cell cycle division through multiple generations. To this end, we used two assays: (a) a nocodazole arrest, to assess accumulation of cells in mitosis, and (b) an assay that is readily used to count cell divisions of lymphocytes (CFSE content assay). In the first assay, T98G cells were serum-stimulated or transduced with the indicated adenoviruses as in Fig. 7A. 22 h later, nocodazole was added to the medium, and cells were incubated for 22 h. Cells were collected and prepared for flow cytometric analysis. It is clear that both cyclin D1 and cyclin E can induce accumulation of cells with a 4n DNA content (Fig. 7B). This indicates that these cells completed DNA replication successfully. Obviously, cells transduced with cyclin D1 and p21 remained in G 1 . This experiment also shows clearly that most cells stimulated with FBS or transduced with cyclin D1 were in mitosis. In contrast, a significant proportion of the cells transduced with cyclin E were still progressing through the G 1 and S phases. Nocodazole release resulted in entry of the cells into the next G 1 phase, although apoptosis was observed (Fig. 7B,  lower right panel, and see below). In the second assay, serumstarved T98G cells were incubated with CFSE. Next, we either stimulated cells with FBS or infected them as in the previous set of experiments. At the indicated time points, cells were collected, and the CFSE content per cell was determined by flow cytometric analysis. Fig. 7C shows that both cyclin E and D1 are sufficient to induce growth factor-independent cell division. However, cyclin D1 effects are blocked by p21. In contrast, inhibition of D-type cyclin activity in cells expressing cyclin E does not block growth factor-independent cell cycle progression. These results are in agreement with the data shown in Fig. 7, A and B. Whereas we observed that expression of G 1 cyclins was sufficient to induce growth factor-independent cell cycle progression through more than one generation, it was clear that these cultures were not growing as rapidly as those cultures stimulated with FBS. This is mostly because the cells transduced with the various combinations of adenoviruses were cycling more slowly than cells stimulated with FBS. Although it is not obvious from direct observation, there is also the possibility that a small portion of cells were dying by apoptosis due to the absence of survival factors present in serum. Thus, we determined induction of apoptosis by analyzing PARP cleavage. Fig. 7D shows that ectopic expression of G 1 cyclins in the absence of serum results in perceptible apoptosis as detected by PARP cleavage (appearance of an 87-kDa cleaved PARP fragment). PARP cleavage was more evident in cells ectopically expressing cyclin D1. In agreement with this result, we also observed apoptosis in cyclin D1-transduced T98G cells entering G 1 upon release from a nocodazole arrest (data not shown).

DISCUSSION
In this report, we show that expression of both cyclin D1 and cyclin E are sufficient to induce growth factor-independent cell cycle progression in human T98G and Rat-1 cells. Whereas cyclin E expression can induce cell cycle progression in the absence of D-type-CDK activity, activation of cyclin D1-CDK complexes in the absence of cyclin E-CDK2 activity is not sufficient to induce DNA synthesis. However, we have observed that expression of cyclin D1 alone was as potent as expression of both cyclin D1 and E in inducing phosphorylation of pocket proteins and changes in E2F-dependent gene expression and cell cycle progression (Table I).
Ectopic expression of cyclin D1 leads to downstream activation of endogenous cyclin E-CDK2 complexes and induces cell cycle progression more potently than cyclin E. In agreement with the more potent effects of ectopically expressing cyclin D1, the overall CDK2 activity, but not the cyclin E-CDK2 activity, is higher than in those cells ectopically expressing cyclin E, presumably because cyclin D1 induces activation of cyclin A-CDK2 complexes more effectively than cyclin E. This is likely to result from a combination of events induced by cyclin D1, which includes more effective up-regulation of the cyclin A protein, increased down-regulation of p27, and increased sequestration of p27. In contrast, whereas ectopic expression of cyclin E induces its associated kinase activity to levels similar to or higher than serum stimulation, the total CDK2 activity in cyclin E-transduced cells is lower than in FBS-treated cells. This is presumably because cyclin E induces downstream effects, such as activation of cyclin A-CDK2 complexes, less effectively than cyclin D1. Altogether, this suggests that although cyclin D1-mediated activation of endogenous cyclin E-CDK2 complexes is essential for cell cycle progression, cyclin D1 plays other roles that contribute to timely progression through G 1 and S phases.
Our data show that whereas transduction of either cyclin D1 or cyclin E is sufficient to induce coordinated hyperphosphorylation of the three endogenous pocket proteins, individual activation of cyclins D1 and E leads to distinct phosphorylation patterns. Both cyclins D1 and E are required for full hyperphosphorylation of pRB and p130. In contrast, cyclin D1 individually, but not cyclin E, can induce hyperphosphorylation of p107. Phosphorylation of p130 to form 2b by cyclin E-CDK2 complexes is sufficient to disrupt p130-E2F4 complexes. This is accompanied by the expression of a number of genes whose promoters are thought to be regulated by p130 and pRB including E2F1, cyclin A, and p107 (29,43). Of note, changes in the phosphorylation status of p107 can be uncoupled from changes in its protein expression levels. Cyclin D1 individually induces hyperphosphorylation of p107 without affecting its protein expression, whereas cyclin E induces p107 up-regulation without inducing p107 hyperphosphorylation. Importantly, cyclin E-induced dramatic accumulation of hypophosphorylated p107 does not prevent E2F-dependent gene expression or cell cycle progression, although we cannot rule out the possibility that the accumulation of hypophosphorylated p107 slows down cell cycle progression. Our data suggest that induction of cell cycle progression by cyclin E in the absence of D-type cyclin CDK activity is coupled to the expression of a number of E2F-dependent genes. Previous studies have shown that cyclin E can override a G 1 growth arrest induced by a phosphorylation-deficient pRB mutant that lacks certain CDK phosphorylation sites (22,23). This suggested that cyclin E was capable of inducing cell cycle progression independently of the pRB/E2F pathway. Our data suggest that cyclin E is sufficient to abrogate the ability of both p130 and pRB to inhibit E2F4. Recent reports indicate that the p130/p107 and pRB pathways are two separate pathways required for G 1 control (44 -46). Thus, it would be important to examine whether ectopic expression of cyclin E in cells expressing a phosphorylation-deficient pRB mutant is sufficient to induce the expression of at least certain E2F-dependent genes such as cyclin A and/or E2F1. A potential scenario is that cyclin E overrides a G 1 growth arrest induced by a phosphorylationdeficient pRB mutant because it bypasses the need for E2F activities that are specifically controlled by pRB and negatively regulate the expression of the cyclin E gene. Ectopic expression of cyclin E under these conditions is likely to be sufficient to disrupt p130-E2F4 complexes and allow for the expression of genes such as cyclin A and/or E2F1.
Unexpectedly, we found that endogenous E2F1 interacts with both p130 and p107. This interaction is modulated during the cell cycle in a manner that appears to depend on previous abrogation of p130-E2F4 and pRB-E2F4 complexes, which releases free p130 and up-regulates p107 and E2F1. In this regard, it is important to point out that E2F1 appears to interact with p130 form 3, a form of p130 that does not interact with E2F4. Thus, the availability of free p130 and the increased expression of E2F1 results in the formation of a transient complex, which probably disappears as result of p130 downregulation. Altogether, this suggests that the interactions between pocket proteins and E2F family members are more complex and interdependent than initially anticipated. Whereas the overall levels of p130-E2F1 and p107-E2F1 complexes in the cell might be low in comparison with pRB-E2F1 complexes, it is conceivable that the former complexes play a critical role in the regulation of certain E2F1 target genes. These results are not in disagreement with previous chromatin immunoprecipitation analysis of promoter occupancy in vivo in these and other cells (43,47). E2F4 and p130 were the major E2F and pocket proteins bound to the promoters of E2F1, p107, and cyclin A in G 0 and early G 1 . These promoters are bound by E2F1-3, but not p130 or E2F4, in late G 1 and S phases. Thus, it is conceivable that the E2F1-p130 and E2F1-p107 complexes that we have detected do not bind to these promoters. Instead, p130 and p107 may prevent binding of a portion of E2F1 to responsive promoters during late G 1 and perhaps S phase. Similarly, pRB has not been detected bound to any natural E2F promoters in vivo.
One question that remains is why cells of different origins exhibit distinct sensitivity to overexpression of G 1 cyclins. For instance, it has been reported that ectopic expression of G 1 cyclins in rodent and human fibroblasts accelerates the G 1 phase of the cell cycle but does not result in mitogen-independent growth (15)(16)(17)48). In contrast, expression of cyclin D1 in primary rat hepatocytes or human MCF-7 breast cancer cells is sufficient to induce mitogen-independent growth or growth in medium containing very limited serum, respectively (33,49). Thus, the most likely scenario is that the effects of overexpression of G 1 cyclins in different cells depend on the efficient activation of their associated CDKs. In many cells, activation of G 1 cyclin-CDK activities depends on steps that cannot be overridden simply by overexpression of their regulatory subunits. Of note, our results using Rat-1 cells are different from a previous study (16). In this previous report, overexpression of either cyclin D1 or E did not result in mitogen-independent growth. It is likely that the difference resides in the level of expression of the exogenous cyclins, which is presumably higher in our study.
Finally, our data also suggest that if constitutive activation of either cyclin D1-CDK or cyclin E-CDK2 complexes is achieved in quiescent cells, these are not only able to synthesize DNA as reported earlier (21) but are also able to grow independently of mitogens for multiple generations. This has in vivo implications for cells exhibiting certain genetic properties, because hyperactivation of G 1 cyclins as a result of mutations might result in autonomous growth in the absence of growth factors.