Cdk2-dependent Phosphorylation and Functional Inactivation of the pRB-related p130 Protein in pRB(−), p16INK4A(+) Tumor Cells*

The retinoblastoma family proteins pRB, p107, and p130 are phosphorylated and released from E2Fs in the late G1 phase of the cell cycle. This phosphorylation is thought to contribute to the derepression of E2F-responsive genes and to be mediated, in part, by Cdk4 and Cdk6. Evidence that Cdk4/6 activity is inhibited by p16INK4A in most pRB(−) cells suggests that p107 and p130 may be underphosphorylated and remain associated with E2Fs during G1-S progression in cells that lack pRB. To examine this, we evaluated the cell cycle-dependent phosphorylation and E2F binding abilities of p107 and p130 in pRB(−), p16(+) Saos-2 osteosarcoma cells. p130, but not p107, was phosphorylated and released from E2F-4 in late G1 and S phase cells, although p130 phosphorylation differed qualitatively in these and other pRB(−), p16(+) cells as compared with pRB(+), p16(−) cell types. p130 phosphorylation occurred in the absence of cyclin D-Cdk4/6 complexes, coincided with cyclin E- and Cdk2-associated kinase activity, and was prevented by expression of dominant negative Cdk2. Moreover, dominant negative Cdk2 prevented the dissociation of endogenous p130-E2F-4 complexes and inhibited E2F-4-dependent transcription. These findings show that p130 can be phosphorylated and functionally inactivated in a Cdk2-dependent process, and they highlight the involvement of distinct Cdks in the regulation of different pRB family proteins.

In mammalian cells, proliferative decisions are frequently made in the late G 1 phase of the cell cycle, when under appropriate conditions cells progress through the restriction point and become committed to proceed through the S, G 2 , and M cell cycle phases (1,2). G 1 -S progression is influenced by diverse growth signaling pathways that converge on the control of cyclin-dependent kinases (Cdks), 1 including Cdk4 or Cdk6 in conjunction with D type cyclins and Cdk2 in conjunction with cyclin E (3)(4)(5)(6)(7)(8). When active, G 1 Cdks are thought to promote the G 1 -S transition by coordinately phosphorylating specific substrates (for review see Ref. 9).
Whereas pRB phosphorylation is thought to be crucial for G 1 -S progression as well as for S phase completion in most cell types (69 -72), the importance of p107 and p130 phosphorylation has not been established. Because the Cdk4/6-dependent phosphorylation of p107 and p130 coincides with the release of p107 and p130 from E2Fs and with the induction of p107/p130responsive genes in late G 1 , it has seemed likely that p107 and p130 phosphorylation by Cdk4/6 is required for release from E2F, for induction of p107/p130-responsive genes, and for normal G 1 -S progression. However, the requirement for such phosphorylation is brought into question in cells that lack functional pRB, because pRB(Ϫ) tumor cells generally express high levels of the Cdk4/6 inhibitor p16 INK4A (hereafter referred to as p16) (73)(74)(75)(76)(77) and are insensitive to ectopically expressed p16 as well as to other agents that antagonize Cdk4/6 activity (78 -84). Accordingly, the growth inhibitory functions of p107 and p130 appear to be circumvented in pRB(Ϫ) cells without p107 and p130 being phosphorylated by the Cdk4/6 kinases. Moreover, the mechanism that is used to circumvent p107-and p130-mediated growth inhibition would appear to be highly effective, because pRB(Ϫ), p16(ϩ) tumors only very rarely have inactivating mutation of the p130 gene (see Ref. 85 for the only example to date), and none have been reported to have mutation of p107.
At present, mechanisms by which the growth inhibitory effects of p107 and p130 are overcome in pRB(Ϫ), p16(ϩ) cells are unknown. As one possibility, p107 and p130 may remain constitutively underphosphorylated, bound to E2Fs, and able to repress p107/p130-responsive genes, with this having little effect on cell proliferation due to the more profound consequences of the loss of pRB. Arguing against this idea, however, is the notion that p107/p130-responsive genes may be important for progression through various stages of the cell cycle and are indeed expressed in pRB(Ϫ) cells. 2 One means by which transcriptional repression by p107 and p130 might be bypassed is through the exclusion of p107-E2F-4 and p130-E2F-4 complexes from the nucleus in late G 1 and S phase cells (86 -89), although it is currently unclear whether this can occur in the absence of p107 and p130 phosphorylation. An alternative means by which the growth inhibitory effects of p107 and p130 might be bypassed in pRB(Ϫ), p16(ϩ) cells, and which is addressed in the current study is through the release of p107 and p130 from E2Fs, due to their phosphorylation by kinases other than Cdk4 or Cdk6. Here, we show that p130 but not p107 is indeed phosphorylated and released from E2F-4 in a Cdk4/6independent but Cdk2-dependent process. These findings provide a mechanism by which p130 is inactivated in pRB(Ϫ), p16(ϩ) cells and highlight the involvement of distinct Cdks in the regulation of different pRB family proteins.
Cell Cycle Synchronization and Restimulation-Subconfluent Saos-2 and HaCaT cells were trypsinized and plated at 3 ϫ 10 6 /100-mm dish and 1.5 ϫ 10 6 /100-mm dish, respectively. On the following day, cells were rinsed once with phosphate-buffered saline (PBS) and maintained in serum-free medium for 3 days. Cells were restimulated by addition of complete medium, in some cases in the presence of 1 mM hydroxyurea.
[ 3 H]Thymidine Incorporation and Flow Cytometry Analyses-Cells were seeded to 24-well plates at 2.5 ϫ 10 4 cells/well for SAOS-2 or 6 ϫ 10 4 cells in 0.5 ml of medium/well for HaCaT, serum starved as above, and restimulated in the presence of 10 Ci/ml, [ 3 H]thymidine for 1 h. Labeling was terminated by adding ascorbic acid to 250 mM, cells were washed three times with PBS and once with ice-cold 5% trichloroacetic acid, incubated in 5% trichloroacetic acid on ice for 30 min, washed with H 2 O, and lysed overnight in 1 M NaOH at 37°C, and [ 3 H]thymidine incorporation was measured by scintillation counting. Samples were assayed in triplicate, and the means and standard errors of [ 3 H]thymidine incorporation levels were calculated. Flow cytometry was performed as described previously (15).
Transfection and Selection of Transfected Cells-For Hook selection, 2.2 ϫ 10 6 Saos-2 cells in 10-cm dishes were co-transfected with 2 g of pHook-2-CC plus 20 g of expression vector using the calcium phosphate method (95), and serum starved and stimulated as above. Cells were then rinsed with PBS, incubated in 0.06% trypsin (1:4 dilution of 0.25% trypsin in Hanks' balanced salt solution) at 37°C for 7-10 min, collected by centrifugation, and resuspended in 1 ml of medium containing 1.5 ϫ 10 6 hapten-coated magnetic beads (InVitrogen). Mixtures were rotated for 1 h at 6 rpm at 37°C, and cells were collected by end-over-end rotation in a magnetic stand for 1 min and washed with Hanks' balanced salt solution twice while in the stand.

Impaired p130 and p107
Phosphorylation and Production of a Novel p130 Phosphoisomer, "Form IIb," in Late G 1 Saos-2 Cells-In this study, we sought to determine whether p107 and p130 are phosphorylated and functionally inactivated during G 1 -S progression in pRB(Ϫ), p16-expressing cells. The Saos-2 osteosarcoma cell subclone 2.4 (14) was used for these analyses. These pRB(Ϫ), p16(ϩ) cells (77,96) can be growth inhibited by serum starvation and induced to enter S phase in a synchronous manner upon serum stimulation. As shown in Fig. 1, [ 3 H]thymidine incorporation increased over basal levels between 15 and 24 h after stimulation and subsequently decreased, indicating that a discrete cohort of Saos-2 cells had been induced to enter and proceed through S. For comparison purposes, p107 and p130 phosphorylation was also examined in pRB(ϩ) HaCaT keratinocytes, which are thought to have active Cdk4 and Cdk6 (66). Fortuitously, HaCaT cells entered S phase with kinetics similar to that of Saos-2 (Fig. 1).
The phosphorylation of p107 and p130 in serum starved and restimulated cells was assessed by immunoblotting for electrophoretically distinct p107 and p130 isoforms. Two p130 species, here termed Forms I and IIa, were detected in serum starved Saos-2 cells and persisted for at least 9 h after stimulation ( Fig.  2A). Between 9 and 12 h after stimulation, p130 Forms I and IIa declined, and a slightly more slowly migrating p130 species termed Form IIb appeared. This species comprised the majority of p130 from 15 to 24 h after stimulation. Similar to the situation in Saos-2 cells, two p130 isoforms were also present in serum starved HaCaT cells, and a more slowly migrating species appeared after serum stimulation (Fig. 2B). However, whereas serum starved Saos-2 and HaCaT cells contained p130 species with identical mobilities, corresponding to Forms I and IIa, the p130 species produced in stimulated HaCaT cells (termed Form III) migrated substantially more slowly than the p130 Form IIb in Saos-2 (Fig. 2C). The distinct mobilities of p130 Forms IIa, IIb, and III are most clear when compared with that of an invariant protein band that cross-reacts with the p130 antibody (labeled with an asterisk in Fig. 2C). For both Saos-2 and HaCaT cells, the different p130 forms represented distinct phosphoisomers, because each was converted by phosphatase treatment to a single, more rapidly migrating species (data not shown), consistent with earlier phosphatase analyses (57,97).
In contrast to the situation with p130, no mobility shift of p107 was evident in serum-stimulated Saos-2 cells (Fig. 2, A and C). This behavior differed from that seen in HaCaT cells, where p107 mobility decreased in late G 1 , coincident with the appearance of hyperphosphorylated p130 and pRB (Fig. 2, B and C). Thus, phosphorylation of both p107 and p130 differed in pRB(Ϫ) Saos-2 cells, compared with that of pRB(ϩ) HaCaT keratinocytes, as indicated by the absence of the most slowly migrating phosphorylated forms of these proteins.
Impaired p130 Phosphorylation Correlates with p16 Status-A possible explanation for the difference in p107 and p130 phosphorylation in Saos-2 and HaCaT cells may be that the p16 that is expressed in Saos-2 cells may impair Cdk4/6-dependent phosphorylation. If this is the case, then p107 and p130 phosphorylation may be similarly affected in other pRB(Ϫ), p16(ϩ) cell types. To evaluate this, the expression of the different p130 phosphoisomers was compared in a series of pRB(Ϫ) and pRB(ϩ) breast cancer cell lines. As shown in Fig. 3, the slowly migrating p130 Form III was detected in each of the six pRB(ϩ) cell types examined (lanes 4 -9), but only the more rapidly migrating p130 species were detected in three pRB(Ϫ) lines (lanes 1-3). As found in many other tumor cell types, p16 was readily detected in each of the pRB(Ϫ), but not in the pRB(ϩ)  1 and 3) and HaCaT (H, lanes 2 and 4) cells. Positions of p130 Forms I, IIa, IIb, and III are indicated by arrows. In C, the different mobilities of p130 Forms IIa and IIb is most apparent when compared with that of an invariant protein band (asterisk) that cross-reacts with the anti-p130 antibody.
Impaired Binding of E2F-4 to p130 Form IIb, but Not to p107, in Late G 1 and S Phase Saos-2 Cells-Earlier studies showed that p107 and p130 were hyperphosphorylated and lost their ability to bind E2F-4 in late G 1 in various pRB(ϩ) cell types (58 -60, 68, 98). These findings suggested that the release of p107 and p130 from E2F-4 in late G 1 might be important for relieving p107-and p130-mediated cell cycle blocks. However, because p107 and p130 were not fully phosphorylated in Saos-2 cells, it was unclear whether they would dissociate from E2F-4 in the late G 1 and S cell cycle phases. To address this issue, we evaluated the E2F-4 binding abilities of p107 and p130 in Saos-2 cells that had been serum starved and restimulated for 21 h. In these experiments, hydroxyurea was added to the culture at the time of serum stimulation to prevent asynchronous G 0 /G 1 cells within the culture from obscuring the pattern of E2F association occurring in late G 1 and S.
To monitor the association of E2F-4 with p107 and p130, lysates prepared from serum starved and restimulated Saos-2 cells were immunoprecipitated, using either an antibody directed against E2F-4 or an antibody directed against p107 and p130, and the presence of p107, p130, or E2F-4 in the immunoprecipitates was detected by immunoblotting. After serum starvation, p130 Forms I and IIa, as well as p107, co-immunoprecipitated with E2F-4 ( Fig. 4, lanes 1 and 2). After 21 h of stimulation, p130 Form IIb was the predominant p130 species and failed to co-immunoprecipitate with E2F-4 (Fig. 4, lanes 3  and 4). In this regard, p130 Form IIb behaved similar to the p130 Form III that is present in HaCaT and other pRB(ϩ) cells, which also lacks E2F-4 binding ability (60). 3 In contrast, the co-immunoprecipitation of p107 with E2F-4 was unchanged after serum stimulation, indicating that p107 had retained the ability to bind E2F-4 in late G 1 and S phase Saos-2 cells.
Absence of Cyclin D-Cdk4/6 Complexes in Late G 1 Saos-2 Cells-Because Saos-2 cells failed to produce p130 Form III, p130 phosphorylation and release from E2F-4 must have come about through a mechanism different from that which operates in pRB(ϩ) cell types. Accordingly, we sought to define the mechanism of p130 Form IIb phosphorylation by evaluating the involvement of known G 1 Cdks.
We first investigated the role of Cdk4 and Cdk6 in p130 Form IIb phosphorylation. Although Saos-2 cells express p16 (77), this does not preclude their having active Cdk4 or Cdk6, because such kinases may be mutated in a way that prevents binding to p16 and that permits sustained binding to D type cyclins (99). Accordingly, we assessed whether active Cdk4 or Cdk6 were present in late G 1 Saos-2 cells by determining whether these kinases were in complexes that contained Dtype cyclins but that were free of p16. The analyses focused on Cdk4/6 association with cyclin D3, because cyclins D1 and D2 were not detectably expressed in Saos-2 cells (data not shown), consistent with previous observations (100), and because cyclin D3 may be particularly important for p130 phosphorylation and functional inactivation (98). The assays were performed 15 h after serum stimulation, when most stimulated cells were in late G 1 , when cyclin D3 levels peaked (data not shown), and when p130 Form IIb was being produced.
To determine whether cyclin D3-Cdk4 or cyclin D3-Cdk6 complexes that were free of p16 had formed, cell lysates were first subjected to three rounds of immunoprecipitation with anti-p16 antibody. This treatment cleared all p16 as well as p16-associated Cdk4 and Cdk6 from the lysate (Fig. 5, lanes  1-3). In addition, a small amount of cyclin D3 was present in the anti-p16 immunoprecipitate (lane 1), suggesting the existence of inactive p16-cyclin D3-Cdk4 or p16-cyclin D3-Cdk6 complexes, as previously observed in other cell types (101). We then determined whether cyclin D3 that remained in the lysate was associated with Cdk4 and Cdk6, by immunoprecipitating the cleared lysates with anti-cyclin D3 and immunoblotting with anti-Cdk4 and anti-Cdk6 antibodies. The results showed that cyclin D3 but no Cdk4 or Cdk6 subsequently precipitated with anti-cyclin D3 antibody (Fig. 5, lane 4). The cyclin D3 immunoprecipitation was nearly quantitative, because only small amounts of cyclin D3 precipitated in subsequent immunoprecipitation rounds (Fig. 4, lanes 5 and 6). In parallel experiments using HaCaT cells, substantial amounts of Cdk4 co-precipitated with anti-cyclin D3, confirming that such complexes would have been readily detected if they had been present in the Saos-2 cell lysates (data not shown).
These findings indicate that late G 1 Saos-2 cells lacked cyclin D3-Cdk4 or cyclin D3-Cdk6 complexes that were free of p16 and thus most likely lacked Cdk4 and Cdk6 kinase activities at the time when p130 Form IIb was being produced. This view was strengthened by our inability to detect Cdk4-associated pRB kinase activity during G 1 progression (data not shown) and by the failure of ectopically expressed p16 to inhibit the production of p130 Form IIb (see below). Thus, we conclude 3 L. Cheng, unpublished data. .  1 and 2) or for 21 h in the presence of hydroxyurea (HU; lanes 3 and 4), and lysates were immunoprecipitated (IP) with antibody recognizing both p107 and p130 (using 3 mg of lysate) or with antibody directed against E2F-4 (using 6 mg of lysate), as indicated. The immunoprecipitates were then subjected to Western blotting, and the upper portion of the blot was sequentially probed with anti-p130 and anti-p107 antibodies (top and middle panels), and the lower portion was probed with anti-E2F-4 (bottom panel). The position of the p130 IIb isoform that fails to immunoprecipitate with E2F-4 is indicated.
that p130 was phosphorylated in late G 1 Saos-2 cells by kinases other than Cdk4 or Cdk6.

Appearance of p130 Form IIb Coincides with Cyclin E-Cdk2 Kinase Activity and Is Impaired by Dominant Negative Cdk2-
Previous studies indicated that Cdk2 is needed for S phase entry in Saos-2 cells (7). To assess whether Cdk2 together with cyclin E contributes to the late G 1 phosphorylation of p130 in these cells, cyclin E and Cdk2 expression and their associated kinase activities were measured during G 1 -S progression.
As shown in Fig. 6A, cyclin E levels increased moderately after serum stimulation but were thereafter relatively unchanged. This constitutive expression of cyclin E in Saos-2 cells is similar to that seen in other pRB(Ϫ) cell types (31). Cdk2 was also expressed at nearly constant levels in serum starved and stimulated Saos-2 cells, although a fast migrating Cdk2 species that is thought to represent the active Cdk2 isoform (102) appeared between 9 and 12 h after stimulation (Fig. 6B). Similarly, cyclin E-and Cdk2-associated histone H1 kinase activity appeared between 9 and 12 h and increased progressively thereafter (Fig. 6, C and D), corresponding to times when p130 Form IIb was being produced (Fig. 2A). These results are consistent with cyclin E-Cdk2 contributing to and being ratelimiting for the production of p130 Form IIb.
To determine whether Cdk2 is required for the late G 1 phosphorylation of p130 in Saos-2 cells, we evaluated the effects of dominant negative (DN) Cdk2 on the production of p130 Form IIb. Saos-2 cells were co-transfected with expression vectors encoding DNCdk2 (7) together with a chimeric "Hook" protein composed of the platelet-derived growth factor receptor transmembrane domain and a single chain antibody (103). Transfected cells were then serum starved, restimulated for 21 h, and selected using hapten-coated magnetic beads, and the phosphorylation state of the endogenous p130 protein in the selected cells was assessed by Western analysis.
Selected cells that had been co-transfected with Hook and control expression vectors expressed predominantly p130 Form IIb (Fig. 7A, lane 5). However, selected cells that had been co-transfected with the Hook and DNCdk2 vectors produced p130 Form IIa but no detectable Form IIb (Fig. 7A, lane 2). The mobility difference between p130 Form IIa in DNCdk2-transfected cells and p130 Form IIb in control cells is best appreciated by referring to a consistently migrating protein band that cross reacts with the p130 antibody (labeled with an asterisk in Fig. 7A) and was apparent in three independent experiments. Unselected cells obtained from Hook ϩ DNCdk2 transfected cultures expressed p130 Form IIb that migrated identically to that present in unselected vector-or p16-transfected cultures (Fig. 7A, lanes 6 -8). Thus, the absence of p130 Form IIb in Hook-selected cells depended on their transfection with both DNCdk2 and Hook DNAs.
We next determined whether inhibitors of G 1 Cdks other than Cdk2 impair production of p130 Form IIb. Ectopic expression of p16 and DNCdk3 had no effect on the production of p130 Form IIb in selected cells after 21 h of serum stimulation (Fig.  7A, lanes 1 and 3-5), despite their high expression levels compared with endogenous p16 and Cdk3 (Fig. 7B). Furthermore, fluorescence-activated cell sorter analysis of transfected cells showed that both DNCdk2 and DNCdk3 prevented Saos-2 cell S phase entry (Fig. 7C), as shown previously (7,104). This implies that Cdk2 had a specific role in p130 phosphorylation and that the ability of DNCdk2 to inhibit p130 phosphorylation was not solely a consequence of the DNCdk2-mediated cell cycle block. In summary, production of p130 Form IIb depended on Cdk2 but not on Cdk3, Cdk4, or Cdk6 activity.
DNCdk2 Prevents Dissociation of p130 from E2F-4 and Inhibits E2F-4-dependent Transactivation-The above studies showed that p130 was phosphorylated in a Cdk2-dependent process that coincided with its dissociation from E2F-4. To determine whether the dissociation of p130 from E2F-4 depended on Cdk2, Saos-2 cells were co-transfected with pBABE-Puro DNA (encoding puromycin resistance), together with either a DNCdk2 expression vector or an empty vector control. The cells were then serum starved in the presence of puromycin to simultaneously kill untransfected cells and arrest transfected cells in G 1 , and they were restimulated with complete medium for 21 h in the presence of puromycin (to complete selection) and hydroxyurea (to prevent re-entry into G 1 ). The association of E2F-4 with p130 and p107 in the surviving cells was then measured by immunoprecipitation with anti-E2F-4 antibody followed by Western analysis. As shown in Fig. 8A, E2F-4 immunoprecipitated in both DNCdk2-transfected and vector-transfected cells. However, p130 co-precipitated with E2F-4 in lysates from DNCdk2-transfected cells but was not detected in immunoprecipitates from the vector-transfected controls. In contrast, p107 was present at similar levels in both immunoprecipitates (Fig. 8A), consistent with its remaining associated with E2F-4 during G 1 -S progression as in earlier experiments (Fig. 4). Because DNCdk2 did not affect the overall levels of p130 (Fig. 7A), the experiments indicated that DNCdk2 prevented the dissociation of p130-E2F-4 complexes.
The ability of DNCdk2 to prevent dissociation of p130 from E2F-4 suggested that DNCdk2 may also prevent E2F-4-dependent transcription in these pRB(Ϫ) cells. To test this, FIG. 6. Production of p130 Form IIb coincides with activation of cyclin E-Cdk2. Saos-2 cells were serum starved and restimulated, and at the indicated times protein lysates were prepared and used for immunoblotting with anti-cyclin E (A) or anti-Cdk2 (B) antibodies or for immunoprecipitation with anti-cyclin E (C) or anti-Cdk2 (D) antibodies followed by assay for histone H1 kinase activity.
Saos-2 cells were transfected with an E2-CAT reporter construct whose expression depended on the E2F binding sequence (94), together with expression vectors for E2F-4 and DP-1, and CAT activity was determined after 48 h. As shown in Fig. 8B, E2F-4 and DP-1 stimulated E2-CAT expression by ϳ10-fold, whereas transfection of DNCdk2 abolished E2F-4-mediated transactivation. The ability of DNCdk2 to inhibit E2F-4-dependent transactivation is consistent with Cdk2-dependent phosphorylation of p130 being required for E2F-4-mediated transcription in pRB(Ϫ), p16(ϩ) cells. DISCUSSION The phosphorylation of p107 and p130 by cyclin D-Cdk4/6 has been thought to contribute to the release of p107 and p130 from E2Fs, to the derepression of E2F-responsive genes, and to G 1 -S progression (for reviews see Refs. 11,21,22,and 105). However, evidence that most pRB(Ϫ) cells express high levels of p16 raised the possibility that p107 and p130 might not be phosphorylated or released from E2Fs in such cells. Defining whether p107 and p130 are released from E2Fs in late G 1 pRB(Ϫ) cells has important implications. If p107 and p130 are released from E2Fs, then such release may contribute to the growth of pRB(Ϫ) tumors. Alternatively, if p107 or p130 are not released from E2Fs, then other mechanisms may function to overcome the growth inhibitory effects of p107-E2F and p130-E2F complexes and to promote the growth of pRB(Ϫ) malignancies.
To address these issues, the current study examined the cell cycle-dependent phosphorylation and E2F-4 binding abilities of p107 and p130 in pRB(Ϫ), p16(ϩ) Saos-2 cells. We found that p130 was phosphorylated and released from E2F-4 in late G 1 , yet this phosphorylation differed from that occurring in numerous pRB(ϩ) cell types. In pRB(ϩ) cells, p130 phosphorylation was accompanied by a substantial electrophoretic mobility shift corresponding to p130 Form III, whereas in pRB(Ϫ) Saos-2 cells, phosphorylation was accompanied by a minor mobility shift corresponding to p130 Form IIb. This p130 Form IIb phosphorylation coincided with activation of the cyclin Eand Cdk2-associated kinases and dissociation of p130 from  1 and 6), DNCdk2 (lanes 2 and 7), or DNCdk3 (lanes 3 and 4), or with empty expression vector (lanes 5 and 8) and were serum starved, restimulated for 21 h, and selected using magnetic beads. Protein lysates prepared from the selected (lanes [1][2][3][4][5] or unselected (lanes 6 -8) cells were then subjected to Western blotting with anti-p130 antibody. The position of the p130 Form IIb is indicated. The impaired production of Form IIb in DNCdk2-expressing cells (lane 2) is most readily appreciated by comparing p130 mobility to that of a consistently migrating species that cross reacts with the p130 antibody (asterisk). B, the high level expression of p16 (lane 1), DNCdk2 (lane 2), and DNCdk3 (lane 4) relative to their endogenous counterparts in transfected and selected cells was confirmed by Western blot analysis using a mixture of the corresponding antibodies. C, the ability of DNCdk2 and DNCdk3 to induce a G 1 block was confirmed by cotransfecting Saos-2 cells with an expression vector for CD20 together with vectors encoding either DNCdk2, DNCdk3, DNCdk4, or p16, as indicated, or with the "empty" cDNA3 vector control. At 48 h after transfection, cells were fixed, stained with fluorescein isothiocyanatelabeled anti-CD20 antibody and with propidium iodide, and DNA content of CD20(ϩ) cells was determined by fluorescence-activated cell sorter analysis. The percent increase in G 1 cells as compared with cDNA3 vector transfected controls was calculated. Error bars indicate standard deviation for duplicate transfections.

FIG. 8. DNCdk2 impairs dissociation of p130 from E2F-4 and inhibits E2F-4-dependent transactivation.
A, Saos-2 cells co-transfected with pBABE-Puro and expression vectors for either DNCdk2 or an empty vector control were serum starved and selected in the presence of puromycin and were then restimulated for 21 h in the presence of puromycin and hydroxyurea. Lysates (2 mg) were immunoprecipitated with anti-E2F-4 antibody and subjected to Western blotting, and the upper portion of the blot was probed sequentially with anti-p130 and anti-p107 antibodies, and the lower portion was probed with anti-E2F-4. B, Saos-2 cells were co-transfected with E2-CAT, CMV-lacZ, and expression vectors for E2F-4 ϩ DP-1, DNCdk2, or the empty expression vector control, as indicated. CAT and ␤-galactosidase activity were measured 48 h after transfection, and CAT activity normalized to ␤-galactosidase was determined. Error bars indicate standard deviation for duplicate transfections.
E2F-4, and it was insensitive to ectopically expressed p16. However, p130 Form IIb phosphorylation and dissociation from E2F-4 were impaired by expression of dominant negative Cdk2, suggesting that Cdk2-dependent phosphorylation of p130 contributed to the dissociation of p130-E2F-4 complexes in Saos-2 cells. It seems likely that p130 is phosphorylated and inactivated in a similar fashion in other pRB(Ϫ), p16(ϩ) cell types, because pRB(Ϫ) cells generally require Cdk2 kinase activity for proliferation (7,8,82) but have high levels of p16 and presumably lack Cdk4/6 kinase activity (76). In contrast, in pRB(ϩ) cells, cyclin D-Cdk4/6 is implicated in the production of p130 Form III and in the release of p130 from E2F-4 ( Fig. 9).
Whereas p130 was phosphorylated and released from E2F-4 in pRB(Ϫ) Saos-2 cells, p107 did not display a phosphorylationdependent mobility shift and remained associated with E2F-4 in the late G 1 and S phases of the cell cycle. This observation is consistent with earlier evidence that p107 is phosphorylated and inactivated in response to Cdk4/6 but not in response to Cdk2 (58,59). Thus, the different pRB family proteins exhibit distinct sensitivities to different G 1 Cdks (Fig. 9). Under circumstances where cyclin-Cdk complexes are present at physiologic levels, pRB dissociation from E2F requires Cdk4/6 as well as Cdk2 (67). In contrast, p107 dissociation from E2F requires Cdk4/6 but is unaffected by Cdk2 (58,59). Finally, as shown here, p130 phosphorylation and dissociation from E2F requires Cdk2 in pRB(Ϫ) cells but likely involves Cdk4/6, perhaps in conjunction with Cdk2, in pRB(ϩ) cell types.
The ability of Cdk2 to promote phosphorylation of p130 but not p107 may reflect sequence differences in the vicinity of the phosphorylation sites of these proteins that mediate recognition by the various Cdks, as well as differences in the numbers and positions of such sites and structural differences that determine whether phosphorylation results in dissociation from E2F. In this regard, it may be significant that p130, but not p107, has a potential Cdk phosphorylation site within its pocket A box at residue Ser 603 , corresponding precisely to a site in pRB at Ser 567 (25). In pRB, phosphorylation of this residue depends upon cyclin E-Cdk2, as well as upon the prior phosphorylation of pRB C terminus by cyclin D-Cdk4, and is crucial for cyclin E-Cdk2-mediated disruption of the pRB pocket and dissociation of pRB-E2F complexes (64). An ability of cyclin E-Cdk2 to phosphorylate the homologous site in p130 might underlie Cdk2-dependent p130 phosphorylation and release from E2F-4, whereas the lack of such a site may underlie the failure of p107 to be inactivated by Cdk2 in pRB(Ϫ), p16(ϩ) cells. If cyclin E-Cdk2 does indeed phosphorylate p130 Ser 603 in Saos-2 cells, as seems likely, then it would do so in the absence of prior phosphorylation of p130 by cyclin D-Cdk4/6. This suggests an additional distinction in the regulation of pRB family proteins by Cdks; whereas the Cdk2-dependent phosphorylation of pRB Ser 567 and disruption of the pRB pocket requires prior phosphorylation of the pRB C terminus by Cdk4/6 (64), Cdk2-dependent phosphorylation of the homologous position in p130 and disruption of the p130 pocket may be independent of Cdk4/6 activity.
The association of p107 with E2F-4 in late G 1 and S phase Saos-2 cells demonstrates that the persistence of p107-E2F-4 complexes did not prevent S phase entry. One implication of this finding is that p107-mediated repression of E2F-responsive genes may be circumvented without p107 being released from E2Fs. This might occur if p107-E2F-4 complexes are unable to bind to promoter E2F binding sites in late G 1 or S, as suggested by in vivo footprint analyses of the cdc-2 and B-myb genes (106,107), and is consistent with evidence that p107-E2F-4 complexes localize to the cytoplasm in S phase-enriched cultures (89).
Interestingly, if p107-E2F-4 complexes localize to the cytoplasm in late G 1 Saos-2 cells, then the structurally similar p130-E2F-4 complexes might also localize to the cytoplasm were they not disrupted through p130 phosphorylation. Although Cdk2 activity was required for E2F-4-dependent transcription and S phase entry in Saos-2 cells, it is unclear whether Cdk2-mediated phosphorylation of p130 was required for these processes. Accordingly, it remains possible that Cdk2mediated p130 phosphorylation and release from E2F-4 is dispensable for overcoming the growth inhibitory effects of p130-E2F-4 complexes. The importance of Cdk2-dependent p130 phosphorylation for the proliferation of pRB(Ϫ), p16(ϩ) cells may be resolved by analyzing the effects of p130 phosphorylation site mutants.
If p107 phosphorylation and release from E2F is dispensable for S phase entry in pRB(Ϫ) tumor cells, as this study suggests, then this raises the question as to the general importance of p107 and p130 phosphorylation and dissociation from E2Fs in the diverse cell types where it has been observed. As one possibility, phosphorylation of p107 and p130 may be required for S phase entry in pRB(ϩ) cells, whereas pRB loss may permit the effects of p107-E2F or p130-E2F complexes to be circumvented in pRB(Ϫ) cells. Alternatively, p107 and p130 phosphorylation may modulate the expression of E2F-responsive genes without being required for S phase entry in either pRB(ϩ) or pRB(Ϫ) cell types. Finally, given that p107 and p130 may be needed to control proliferation in relatively few embryonic or adult tissues (72, 108 -110), their phosphorylation may contribute to G 1 -S progression only in certain cell types or at selected times during development, particularly in contexts where other mechanisms for overcoming the effects of p107-E2F and p130-E2F complexes do not exist. To address these issues, it will be of interest to define the requirement for p107 and p130 phosphorylation in diverse biological contexts.
Acknowledgments-We thank R. Medema, E. Harlow, C. Sardet, and S. Chellappan for plasmid DNAs and S. Chellappan for critical reading of the manuscript. Cdk2 also contributes to the phosphorylation and inactivation of pRB and possibly p130 but does not contribute to phosphorylation and inactivation of p107. Most pRB(Ϫ) tumor cells express p16 and lack Cdk4 and Cdk6 activity as well as the ability to fully phosphorylate and dissociate p107 from E2F-4. However, p130 is phosphorylated and dissociated from E2F-4 in pRB(Ϫ), p16(ϩ) cells in a Cdk2-dependent process. In the figure, a indicates the active, E2F binding form of pRB family proteins, whereas i indicates phosphorylated pRB family proteins that are unable to bind E2Fs. For p130, the i species corresponds to p130 Form III in pRB(ϩ) cells and to p130 Form IIb in pRB(Ϫ), p16(ϩ) cells.