Paradigms of Growth Control: Relation to Cdk Activation

See allHide authors and affiliations

Science's STKE  28 May 2002:
Vol. 2002, Issue 134, pp. re7
DOI: 10.1126/stke.2002.134.re7


The cyclin-dependent kinases (CDKs) play a key role in cell cycle control, and in this review, we focus on the events that regulate their activities. Emphasis is placed on the CDKs that function during the G1 phase of the cell cycle and on the CDK inhibitor p27Kip1. We discuss how CDK activation relates to two basic concepts of cell cycle regulation: (i) the need for multiple mitogens for the proliferation of nontransformed cells and (ii) the inhibitory effect of high culture density on proliferative capacity. We also describe how Cdk2 modulates the expression of the α subunit of the interleukin-2 receptor in T cells, and address the question of whether p27Kip1 functions as an activator or inhibitor of the CDKs associated with the D cyclins.

Paradigms of Cell Cycle Regulation

In the effort to elucidate the mechanisms that regulate cell proliferation, cultured cells have been invaluable tools. Early studies on cultured cells led to the identification of the four phases of the cell cycle: G1, S, G2, and M (1). Additional studies described a noncycling state (termed quiescence or G0) that was contiguous with G1 and defined the factors that influenced a cell's decision to enter or exit G0. For nontransformed cells, key factors include serum availability and the degree of cell-cell contact. When deprived of serum or grown to confluence, cycling cells cease proliferation, and when exposed to serum or replated at lower densities, quiescent cells reenter the cell cycle. The capacity of cultured cells to sense changes in the extracellular environment and respond accordingly mimics the in vivo situation in which cells remain quiescent for long periods of time and initiate proliferation only when conditions in the body signal the need for the expansion of a given cell population (for example, during development or tissue renewal). Transformed cells, on the other hand, grow in serum-depleted medium and with disregard of culture density; these properties form the basis of the transformed phenotype and account in part for tumor formation in vivo.

Many of the mitogens contained in serum have been purified and characterized. Such mitogens include epidermal growth factor (EGF), insulin-like growth factor I (IGF-1), platelet-derived growth factor (PDGF), and the interleukins (2-5). Growth factors bind to specific and saturable receptors on the cell surface (6). Ligand binding activates signal transduction mechanisms that ultimately impinge on gene expression. As a result, a variety of growth-regulatory proteins are expressed, and these proteins activate additional signaling cascades, induce the expression of secondary response genes, and perform numerous other tasks required for the execution of a proliferative response. Such tasks include the activation of the cyclin-dependent kinases (CDKs), which play a key role in cell cycle regulation. Reduced expression of growth-inhibitory proteins also contributes to CDK activation and cell cycle progression.

One of the basic tenets of cell cycle control is that mitogens act primarily during the G1 phase of the cell cycle. More specifically, mitogens are required for the passage of cells to a point in late G1 termed "the restriction point" (7, 8). Using cultures synchronized in G0, the growth factor requirements of numerous cell types were defined. On the basis of data obtained with Balb/c-3T3 mouse fibroblasts, we proposed a model of cell cycle regulation in which one set of mitogens initiates proliferation by rendering quiescent cells responsive to the growth-promoting actions of a second set of mitogens; such mitogens were designated competence and progression factors, respectively (Fig. 1) (9). In Balb/c-3T3 cells, PDGF induces competence whereas factors contained in platelet-poor plasma (PPP), such as IGF-1, mediate progression (9-12). Although not applicable to all cell lines, the competence-progression model has been used to describe the proliferative behavior of T cells and B cells (13, 14).

Fig. 1.

Regulation of G1 traverse by two sets of serum factors. PDGF, which is derived from the particulate portion of whole-blood serum, induces competence when added to density-arrested BALB/c-3T3 fibroblasts. Competent cells do not traverse G1 but acquire the capacity to respond proliferatively to factors contained in PPP, the fluid portion of whole-blood serum. PPP mediates the progression of competent cells to the restriction (R) point in late G1. After passing the R point, cells complete the remainder of the cell cycle in a growth factor-independent manner. S, S phase; M, M phase.

CDK activation is obligatory for cells to move through G1, and elucidation of the mechanisms that regulate this process has been a major focus of many laboratories, including our own. Distinct cyclin-CDK complexes become active at particular times during G1 as a result of an integrated interplay of events occurring at transcriptional, translational, and posttranslational levels. The purpose of this review is twofold: to present an overview of cyclins, CDKs, and CDK inhibitors and to discuss how two paradigms of growth control apply to cyclin-CDK activation. Specifically, we describe how different classes of mitogens regulate different aspects of CDK activation and how culture density alters the amount of CDK activity required for mitogenesis.

Cyclins, CDKs, and CKIs

Originally discovered in lower eukaryotes, CDKs are nuclear serine-threonine kinases that are present in all eukaryotic cell types (15, 16). CDKs are enzymatically active only when associated with cyclins, a group of proteins that are periodically synthesized and degraded during the cell cycle. CDKs also interact with a family of proteins collectively termed "CDK inhibitors" (CKIs). The abundance of the CKIs, like that of the cyclins, varies throughout the cell cycle and thus contributes to the timing of CDK activation. In mammalian cells, three sets of cyclin-CDK complexes are sequentially assembled and activated during G1: the D cyclins (D1, D2, D3) and Cdk4 or Cdk6; cyclin E and Cdk2; and cyclin A and Cdk2 (17). Beginning in G0, the following series of events occurs: (i) mitogens induce the expression of the D cyclins and decrease the amounts of the CKI, p27Kip1; (ii) D cyclin-associated Cdks and cyclin E-Cdk2 complexes become active and phosphorylate and inactivate the retinoblastoma protein (Rb), which functions as a transcriptional repressor; (iii) E2F target genes, including those encoding cyclin A and several DNA replication enzymes, are expressed; and (iv), cyclin E-Cdk2 and cyclin A-Cdk2 modulate additional processes required for the initiation and execution of DNA synthesis.

p27Kip1and Other Members of the CKI Family

Two classes of CKIs have been defined: the INK proteins (p16INK4a, p15INK4b, p18INK4c, and p19INK4d) and the Cip/Kip proteins (p21Cip1, p27Kip1, and p57Kip2) (18, 19). The INK proteins interact with monomeric Cdk4 or Cdk6 and thus preclude the association of these CDKs with the D cyclins. The Cip/Kip proteins bind complexes containing cyclins D, E, and A and their CDK partners. Although the Cip/Kip proteins are potent inhibitors of Cdk2 activity, their effects on Cdk4 and Cdk6 activity are controversial (see below). Of the CKIs, p27Kip1 is thought to be the primary modulator of proliferative status in most cell types, where it functions to induce and maintain the quiescent state. p27Kip1 accumulates in serum-starved and density-arrested cultures. Addition of mitogens to quiescent cultures reduces the abundance of p27Kip1, and this process is referred to as down-regulation (20-28). Ablation of p27Kip1 expression by antisense mRNA retards cell cycle exit, and consequently, mice lacking p27Kip1 exhibit multiple organ hyperplasia and are larger in size than their control littermates (24, 29-32). Conversely, conditions that block p27Kip1 down-regulation prevent quiescent cells from entering the cell cycle in response to mitogens; such conditions include treatment of cells with adenosine 3´, 5´-monophosphate (cAMP) analogs or rapamycin, hypoxia, and ectopic expression of p27Kip1 (21, 33-37). Collectively, these findings indicate that changes in p27Kip1 levels lead to rather than result from changes in cell cycle status.

In contrast to p27Kip1, p21Cip1 is present in larger amounts in cycling than in quiescent cells (20, 21 ,38-40). Thus, p21Cip1 may function as a "brake" that prevents excessive CDK activity in proliferating cells. The primary role of p21Cip1, however, is to inhibit the proliferation of cells exposed to DNA-damaging agents (41). p21Cip1 is abundant in irradiated cells and in cells treated with genotoxic drugs, and in its absence, cells with DNA damage do not efficiently arrest in G1 (42-44). p57Kip2 is expressed in a tissue-specific manner and, consequently, is not a universal regulator of cell proliferation (45, 46). The INK proteins mediate growth arrest in certain situations (for example, in cells treated with transforming growth factor-β or undergoing senescence), and increasing evidence links the development of a variety of human cancers to the loss of p16INK4a and (perhaps also p15INK4b) function (19, 47-49).

Three processes have been implicated in p27Kip1 down-regulation: accelerated degradation, translational inhibition, and transcriptional repression. Of these, degradation has been the most extensively studied, and the more rapid turnover of p27Kip1 in cycling as compared with quiescent cells is well established (25, 50-52). Like many cell cycle-regulatory proteins, p27Kip1 can be degraded by the ubiquitin-proteasome pathway (50). For p27Kip1, this pathway consists of three steps: (i) phosphorylation of p27Kip1 at threonine 187 by cyclin E-Cdk2; (ii) recognition of phosphorylated p27Kip1 by SKP2, a component of the SCFSKP2 ubiquitin-ligase complex; and (iii) SCFSKP2-directed destruction of p27Kip1 in the 26S proteasome (52-58). Degradation of p27Kip1 may also require its transport from the nucleus to the cytosol by a protein named Jab1 (59) and its interaction with the cytosolic adaptor protein Grb2 (60). The cell cycle dependence of p27Kip1 degradation reflects the cell cycle dependence of SKP2 expression and cyclin E-Cdk2 activation: minimal expression or activation in G0 cells and maximal expression or activation in S phase cells (61-66).

In cells exposed to mitogens, p27Kip1 levels fall (at least to some extent) before Cdk2 becomes active, and decreases in p27Kip1 abundance are required for Cdk2 activation. Thus, Cdk2-independent processes must initiate p27Kip1 down-regulation. Such processes may include the modulation of p27Kip1 stability by a kinase other than Cdk2 or by the extracellular signal-regulated kinase (ERK) pathway (67-70). Alternatively, or in addition, p27Kip1 expression may be translationally controlled in cells entering or exiting the cell cycle. Treatment of density-arrested Balb/c-3T3 cells with PDGF substantially reduces p27Kip1 synthesis but not p27Kip1 mRNA levels (25). p27Kip1 translation increases in fibroblasts grown to high density or expressing dominant-negative Ras and in HL60 cells induced to differentiate (26, 51 ,71). At present, the processes regulating p27Kip1 translation at the molecular level are incompletely understood. Efficient translation of the p27Kip1 transcript requires a uridine-rich element in its 5´-untranslated region, and proteins that bind this element are enriched in extracts prepared from cells arrested at the G1/S and G2/M borders (72). The 5´-untranslated region also contains an internal ribosome entry site that is thought to mediate cap-independent translation of the p27Kip1 mRNA in quiescent cells (73).

In some systems, changes in transcription rather than translation account for changes in p27Kip1 expression at times when Cdk2 is not active. Examples of systems in which p27Kip1 accumulation is accompanied by activation of the p27Kip1 promoter include hypoxic fibroblasts, myeloid leukemia cells undergoing differentiation, and B cells and hepatoma cells rendered quiescent by treatment with antibody to immunoglobulin M and dioxin, respectively (37, 74-76). Conversely, decreases in p27Kip1 amounts are paralleled by decreases in p27Kip1 transcription in quiescent smooth muscle cells treated with PDGF and in G0-arrested pre-B cells exposed to interleukin-3 (IL-3) (77, 78). Of the transcription factors implicated in the transactivation of the p27Kip1 gene, the most notable are the Forkhead factors, FKHR-L1 and AFX. These factors increase the activity of the p27Kip1 promoter and the abundance of p27Kip1 mRNA and protein when overexpressed in cells and function in the regulation of p27Kip1 expression in pre-B cells treated with or without IL-3 (77, 79).

The Ras proteins (collectively referred to as Ras) are thought to play a critical role in transducing extracellular signals to the intracellular machinery that controls p27Kip1 expression (Fig. 2). Activated forms of Ras reduce p27Kip1 levels when ectopically expressed in cells and, conversely, dominant-negative Ras mutants prevent the down-regulation of p27Kip1 by mitogens (67, 71 ,80-83). Ras activates phosphatidylinositol 3-kinase (PI3K) (84), and p27Kip1 is not down-regulated in cells cotreated with mitogens and either pharmacological PI3K inhibitors or supraphysiological amounts of the phosphoinositide phosphatase PTEN, a natural PI3K antagonist (77, 85-89). PI3K in turn activates the serine-threonine kinase Akt (90), and expression of activated forms of Akt or ablation of PTEN (which results in constitutive Akt activity) attenuate p27Kip1 expression (88, 91-93). Together, these findings indicate that the Ras-PI3K-Akt pathway elicits events that reduce the abundance of p27Kip1. Modulation of p27Kip1 abundance by the Ras-induced ERK pathway has also been reported (67, 68 ,70).

Fig. 2.

Ras pathways contributing to the expression of p27Kip1 and cyclin D1. See text for details. Processes that increase in response to Ras activation are indicated by "+"; processes that decrease in response to Ras activation are indicated by "–". RAF, a protein kinase; MEK, Mitogen-activated protein or extracellular signal-regulated kinase kinase.

The Forkhead transcription factors FKHR-L1 and AFX are phosphorylated by Akt and, consequently, are excluded from the nucleus (94, 95). Thus, the capacity of FKHR-L1 and AFX to activate the p27Kip1 promoter is impaired in situations in which Akt is active (for example., in PTEN-null cells) (92). Inhibition of PI3K activity reduces the abundance of SKP2 mRNA and protein (89). This finding suggests that PI3K (perhaps by activating Akt) reduces p27Kip1 accumulation by augmenting SKP2 expression and, consequently, the SKP2-dependent degradation of p27Kip1. The role of Ras in the translational regulation of p27Kip1 expression is unclear. Expression of a dominant-negative Ras mutant in NIH-3T3 cells increases p27Kip1 translation (71), whereas conditional expression of an activated form of Ras has no effect on p27Kip1 translation in a Balb/c-3T3-derived cell line (25).

The studies described in this section show that p27Kip1 is regulated in distinct ways at different times during the cell cycle. In G0, p27Kip1 accumulates as a result of increases in the transcription of its gene or the translation of its mRNA. Mitogenic stimulation of quiescent cells reduces the abundance of p27Kip1 by shutting off either its transcription or translation or perhaps by inducing its degradation by a non-Cdk2-dependent pathway. Beginning in late G1, cyclin E-Cdk2 complexes become active and phosphorylate p27Kip1, thus targeting p27Kip1 for destruction in the proteasome. These events set in motion a positive-feedback cycle in which p27Kip1 degradation facilitates cyclin E-Cdk2 activation, which in turn maintains p27Kip1 at low levels as cells traverse the remainder of the cell cycle. The possible involvement of the PI3K-Akt pathway in both the transcriptional activation of the p27Kip1 gene and the Cdk2-dependent degradation of the p27Kip1 protein implies that this pathway is operative at different times in the cell cycle. Indeed, PI3K is activated in both early and late G1 (96, 97).

D cyclin-Cdk4 and -Cdk6 Complexes

The D cyclins (D1, D2, and D3) form complexes with either Cdk4 or Cdk6, and different complexes predominate in different cell types. The major D cyclin complexes in T cells, for example, are cyclin D2-Cdk6 and cyclin D3-Cdk6, whereas in rodent fibroblasts, Cdk4 (with D1, D2, or D3) is the preferred D cyclin partner (98-100). All combinations phosphorylate Rb and the Rb-related proteins p107 and p130, and thus contribute to the inactivation of these transcriptional repressors (101). Although exceptions exist, the D cyclins are usually expressed in a mitogen-dependent manner. Small amounts of D cyclin mRNA and protein are present in G0 cells, and amounts of both increase concomitantly upon mitogenic stimulation (98, 102-106). The different D cyclins are induced with different kinetics and thus are not coordinately regulated (98, 102 ,104 ,106). Although the mechanisms controlling the expression of cyclin D2 and cyclin D3 are unresolved, the dependence of cyclin D1 expression on Ras activation is well documented (Fig. 2). Activated forms of Ras induce the expression of cyclin D1 when introduced into cells (80, 107-109), and expression of dominant-negative Ras mutants blocks the induction of cyclin D1 expression by mitogens (81, 83 ,110 ,111). Ras is thought to promote cyclin D1 expression by two pathways: (i) the ERK pathway, which increases the transcription of the cyclin D1 gene (112, 113), and (ii) a PI3K pathway, which, depending on the system, affects the transcription, translation, or degradation of cyclin D1 (Fig. 2) (114-117).

D cyclin-associated activity is first detected in mitogenically stimulated cells in mid G1 and peaks at or after the G1/S border (25, 99 ,100 ,118). Factors that promote D cyclin-Cdk4 and -Cdk6 assembly include p27Kip1, p21Cip1, the chaperonin complex, cdc37-Hsp90, the Cdk4-binding protein SEI-1, and ERK activity (28, 119 ,120). When bound to Cdk4 in S phase, cyclin D1 is phosphorylated on threonine 286 by glycogen synthase kinase-3β (GSK-3β) (Fig. 2) (114, 121). As a result, cyclin D1 is exported from the nucleus and ubiquitinated and degraded in the cytosol (114, 122). The cell cycle dependence of cyclin D1 degradation reflects the fact that GSK-3β is excluded from the nucleus and is inactive (as a result of phosphorylation by At) during G1 (114, 123 ,124). Uncomplexed cyclin D1 is a poor substrate for GSK-3β and is degraded by a ubiquitin-dependent process that does not require its phosphorylation on threonine 286 (125).

In addition to phosphorylating the Rb proteins, the D cyclins and their CDK partners also promote proliferation by sequestering p27Kip1 (23, 126 ,127). This latter function is important because Cdk2 activation requires both a mitogen-induced reduction in the abundance of p27Kip1 and the titration of residual p27Kip1 molecules. Given the capacity of p27Kip1 to inhibit Cdk4 and Cdk6 activity when overexpressed (33, 34), the question arises as to how D cyclin-Cdk4 and -Cdk6 complexes fulfill both their sequestration and enzymatic obligations. One explanation challenges the original designation of p27Kip1 as a Cdk4 and Cdk6 inhibitor. This viewpoint is based on data showing that recombinant p27Kip1 interacts with but does not inhibit the activity of cyclin D2-Cdk4 or -Cdk6 complexes at low stoichiometries and that antibody to p27Kip1 immunodepletes Cdk4 activity from lysates of mouse embryo fibroblasts (MEFs) (128, 129). Thus, individual D cyclin-Cdk4 or -Cdk6 complexes may simultaneously sequester p27Kip1 and phosphorylate Rb. The presence of p21Cip1 in active D cyclin-Cdk4 complexes has also been described (129-133).

Other studies suggest that p27Kip1 is indeed a bona fide Cdk4 (and perhaps also Cdk6) inhibitor. We found that antibody to p27Kip1 does not remove cyclin D1- or cyclin D3-Cdk4 activity from extracts of Balb/c-3T3 cells and that p27Kip1 inhibits Cdk4 activity when added to cell extracts at all concentrations that result in p27Kip1-Cdk4 association (134). Inactivation of recombinant Cdk4 by a wide range of p27Kip1 concentrations has also been reported (131). As a means of explaining how D cyclin-Cdk4 complexes accomplish their dual functions (CKI sequestration and Rb phosphorylation), we isolated cyclin D3 and associated proteins from exponentially growing wild-type MEFs. Most of the cyclin D3 was present in complexes containing Cdk4 and either p27Kip1 or p21Cip1 (Fig. 3) (134). A small amount of cyclin D3 was not bound to Cdk4, and an even smaller amount was complexed to Cdk4 in the absence of p27Kip1 and p21Cip1. Only the binary cyclin D3-Cdk4 complexes were active, thus indicating that a minor portion of the cyclin D3-Cdk4 pool supplies all of the activity required for cell cycle traversal. As a result, cells contain a large reservoir of cyclin D3-Cdk4 complexes that can function as a p27Kip1 or p21Cip1 sink. We suggest that Cip/Kip titration and catalytic activity are mutually exclusive functions of cyclin D3-Cdk4 complexes and that cycling cells contain sufficient amounts of cyclin D3-Cdk4 complexes for the execution of both processes.

Fig. 3.

Relative sizes of the cyclin D3 pools in wild-type and p27Kip1 and p21Cip1 double-null cells. In proliferating wild-type cells, only a small portion of the cyclin D3 pool cells consists of catalytically active cyclin D3-Cdk4 complexes. This leaves a large reservoir of cyclin D3-Cdk4 complexes that promote Cdk2 activation by sequestering p27Kip1 and p21Cip1. An intermediate-sized pool containing cyclin D3 not bound to Cdk4 also is present in wild-type cells. Owing to the lower stability of cyclin D3-Cdk4 complexes as compared with cyclin D3-Cdk4-CKI complexes, most of the cyclin D3 pool in cells lacking both p27Kip1 and p21Cip1 is not associated with Cdk4. Although limited in size in both wild-type and p27Kip1 and p21Cip1 double-null cells, the cyclin D3-Cdk4 pool is capable of supplying all of the enzymatic activity required for cell cycle traverse.

The D cyclins associate with Cdk4 less efficiently in MEFs lacking p27Kip1, p21Cip1, or both than in wild-type MEFs, and D cyclin-Cdk4 complexes dissociate at a lower rate in the presence than in the absence of p27Kip1 or p21Cip1 (129, 131 ,133 ,134). Thus, interaction with p27Kip1 or p21Cip1 increases the stability of D cyclin-Cdk4 (and perhaps also D cyclin-Cdk6) complexes. The inherent instability of binary D cyclin-Cdk4 may provide a means of keeping cyclin Cdk4 activity in check in conditions in which p27Kip1 and p21Cip1 are absent.

Cyclin E-Cdk2, Cyclin A-Cdk2 Complexes, and Pocket Proteins

Rb, the founding member of a family of "pocket proteins," is a constitutively expressed transcriptional repressor that interacts with various proteins through a highly conserved region called the "pocket" (135, 136). The most notable Rb targets are the E2Fs, a group of transcription factors that bind consensus sequences in gene promoters in combination with their dimerization partners DP1 and DP2 (137). By virtue of its capacity to associate with E2Fs (present in E2F-DP-DNA complexes), Rb inhibits transcription by two mechanisms: (i) it prevents the E2F transactivation domain from contacting the basal transcriptional machinery, and (ii) it recruits corepressors to promoters (138). In the latter mechanism, E2Fs do not function as transactivators but simply serve as DNA docking sites for Rb-associated corepressors, which silence transcription by effecting changes in chromatin structure that render promoters inaccessible to transacting factors. Such corepressors include the histone deacetylases (HDACs), which cause chromatin condensation (139-142), and BRG1 and BRM, both of which are adenine triphosphatases involved in the positioning of nucleosomes on promoters (143-145).

During G1, inactive forms of Rb (that is, forms that do not bind E2F) are generated by the carefully orchestrated phosphorylation of Rb by distinct CDK complexes (146, 147). Before the restriction point, Rb is phosphorylated by D cyclin-Cdk4 and -Cdk6 complexes to produce "hypophosphorylated" Rb. Hypophosphorylation does not inactivate Rb but is required for the subsequent "hyperphosphorylation" of Rb by cyclin E-Cdk2 complexes. Rb hyperphosphorylation occurs at the restriction point and results in the release of Rb from E2F and the expression of E2F target genes, including those encoding cyclin E, cyclin A, and several enzymes involved in DNA replication (148). Although E2F activation increases cyclin E expression, preexisting amounts of cyclin E (and thus of cyclin E-Cdk2 complexes) are presumably sufficient (at least initially) for Rb hyperphosphorylation (149). As an alternative or additional source of cyclin E, Rb hypophosphorylation may stimulate the expression of cyclin E by effecting the release of HDACs from Rb-E2F complexes (150, 151). In contrast to cyclin E-Cdk2 activity, which peaks at the G1/S boundary, cyclin A-Cdk2 activity appears in early S, a time that coincides with cyclin A accumulation, and continues to increase throughout S and G2 (61, 62). Once activated, cyclin A-Cdk2 complexes phosphorylate Rb and thus maintain Rb in an inactive state for the remainder of the cell cycle (132).

In addition to Rb, the pocket protein family also includes p107 and p130 (101, 152). Like Rb, p107 and p130 repress transcription in an E2F-dependent manner and are phosphorylated and inactivated by the G1 CDKs (118, 153-157). In several respects, p107 and p130 more closely resemble each other than they do Rb. For example, p107 and p130 interact with a different subset of E2Fs than does Rb and thus regulate the expression of a different subset of E2F target genes (158-163). Moreover, p107 and p130 (but not Rb) form stable complexes with cyclin E-Cdk2 and cyclin A-Cdk2 that either lack kinase activity (164, 165) or exhibit an altered substrate specificity (166). Most importantly, growth inhibition is dependent on the actions of both Rb and either p107 or p130, thus indicating that Rb and the p107 or p130 proteins are functionally distinct (167). Although similar in many respects, p107 and p130 are differentially expressed: p130 accumulates in G0-arrested cells (where it complexes primarily with E2F4), whereas p107 is most abundant in cycling cells (168, 169).

The requirement for Cdk2 activity for S phase entry is not limited to the inactivation of pocket proteins, and in this respect, Cdk2 differs from Cdk4 and Cdk6, whose only known physiological substrates are Rb, p107, and p130. The existence of additional Cdk2 targets is inferred from studies showing that ectopic expression of E2F-1 negates the capacity of p16INK4a but not that of p27Kip1, to arrest cells in G1 (170-172). Moreover, pocket protein inactivation allows the initiation of DNA synthesis only in cells in which Cdk2 is active (at least to some extent) or the need for Cdk2 activity is overridden (36, 172-174). Other than the Rb proteins, the Cdk2 substrate(s) that signals S phase entry is not known. However, a number of Cdk2 substrates that regulate other aspects of cell cycle traversal have been identified. E2F-1 and DP-1, for example, lose their affinity for DNA and their ability to invoke an S phase checkpoint when phosphorylated by cyclin A-Cdk2. p27Kip1 and cyclin E (when bound to Cdk2) are phosphorylated by cyclin E-Cdk2 and thus earmarked for destruction (175-180). Cyclin A is also phosphorylated by Cdk2 but is degraded in a Cdk2-independent manner (181).

Regulation of CDK Activity in Fibroblasts and T Cells by Multiple Mitogens

As detailed above, the activation of the G1 CDKs entails a series of events that is initiated by the addition of mitogens to quiescent cells. Mitogens induce the down-regulation of p27Kip1 and the expression of the D cyclins and the consequent sequestration of residual p27Kip1 molecules. These events facilitate the activation of the D cyclin-associated CDKs and cyclin E-Cdk2, which phosphorylate and inactivate the Rb proteins, thus leading to the expression of E2F target genes, including those encoding cyclins E and A. As a result, more cyclin E-Cdk2 complexes are formed, as are cyclin A-Cdk2 complexes, and these activities mediate the entry of cells into and through S phase. As examples of how multiple mitogens act in a combinatorial fashion to elicit this series of events, we present two sets of studies done in our laboratory. The first set focuses on cyclin-CDK activation in fibroblasts exposed to PDGF (or activated Ras) and PPP, whereas the second describes how T cell receptor (TCR) agonists, IL-2, and serum act in an integrated manner to promote cyclin-CDK activation in primary T cells.

PDGF and PPP in Fibroblasts

As exemplified by PDGF, competence factors allow quiescent Balb/c-3T3 cells to respond to progression factors, which are present in PPP and which mediate the passage of competent cells through G1 to the R point (9). PDGF-treated cells do not traverse G1 in the absence of PPP (that is, similar to quiescent cells, they are 12 hours from S phase) and quiescent cells do not respond proliferatively to PPP unless cotreated or pretreated with PDGF. To determine how the competence and progression model relates to CDK activation, we assessed the effects of PDGF and PPP, either alone or in combination, on cyclin expression, p27Kip1 abundance, and CDK activities in density-arrested Balb/c-3T3 cells. These parameters were also examined in a Balb/c-3T3-derived cell line (termed AC3) that inducibly expresses an activated form of Ras in response to dexamethasone. In AC3 cells, Ras functions as a competence factor and thus substitutes for PDGF (80). Similar results were obtained with both systems (Table 1 and Fig. 4).

Fig. 4.

CDK events associated with competence and progression in fibroblasts. Competence factors such as PDGF or activated Ras induce the expression of cyclin D1 and down-regulate p27Kip1 to a limited extent and in a transient manner. Competence factors also induce an unidentified event (designated "x") that subsequently allows progression factors contained in PPP to produce an additional and persistent decrease in p27Kip1 amounts. The secondary decline in p27Kip1 levels is accompanied by the activation of D cyclin-Cdk4 complexes, cyclin E-Cdk2 complexes, and after synthesis of cyclin A, cyclin A-Cdk2 complexes.

Competence factors contributed to cyclin-CDK activation in two respects: They induced the expression of cyclin D1, which associated with preexisting Cdk4, and they produced a limited and transient reduction in the total amount of p27Kip1 (Fig. 4 and Table 1) (23, 80 ,104). Accumulation of cyclin D1 and formation of cyclin D1-Cdk4 complexes resulted in sequestration of p27Kip1 and the near elimination of uncomplexed p27Kip1. However, neither cyclin D1-Cdk4 complexes nor constitutively expressed cyclin D3-Cdk4 complexes were active (27). Competence factors did not affect the expression of Cdk2, which was readily detectable in quiescent cells, and only weakly increased the expression of cyclins E and A. Although its cyclin partners were present in competent cells (albeit in lower amounts), Cdk2 was not active, thus indicating that p27Kip1 sequestration coupled with a partial decline in p27Kip1 abundance is not sufficient for Cdk2 activation.

In the absence of competence factors, progression factors did not down-regulate p27Kip1, activate CDKs, or induce cyclin expression (23). However, when added to cells in combination with competence factors, progression factors produced a secondary and more pronounced and persistent decline in p27Kip1 abundance that was accompanied by Cdk4 and Cdk2 activation (Fig. 4 and Table 1). Presumably as a result of consequent Rb phosphorylation and the derepression of E2F-dependent transcription, cyclin E and cyclin A were expressed in large amounts in cells receiving both competence and progression factors. Thus, in the context of cyclin-CDK activation, and although other actions cannot be excluded, an important function of progression factors is to maximally and productively down-regulate p27Kip1.

Although competence formation does not result in CDK activation in fibroblasts, it sets the stage for this event by promoting the sequestration of uncomplexed p27Kip1, by initiating the process of p27Kip1 down-regulation, and by eliciting an event that allows PPP to subsequently and more completely reduce p27Kip1 abundance. We suggest that translational inhibition and accelerated degradation account for the first and second phases of p27Kip1 down-regulation. PDGF markedly represses p27Kip1 translation within 2 to 3 hours when added to density-arrested Balb/c-3T3 cells in either the presence or absence of PPP (25). On the other hand, decreases in p27Kip1 stability occur at later times and only in cells receiving both PDGF and PPP. The mechanism responsible for this phase is not known; however, data showing that it does not occur in cells treated with trichostatin A, an inhibitor of histone deacetylases, suggest that expression of a specific gene product is required (182). Although Ras expression in AC3 cells induces the first phase of p27Kip1 down-regulation with kinetics similar to those observed in PDGF-treated Balb/c-3T3 cells, it does not affect the rate of p27Kip1 synthesis (25). This finding suggests that Ras regulates p27Kip1 expression posttranslationally and that PDGF inhibits p27Kip1 translation by a Ras-independent pathway.

Con A, Serum, and IL-2 in T Cells

In T cells, TCR agonists and IL-2 function as competence and progression factors, respectively (13). TCR agonists include cognate antigen, concanavalin A (Con A), and antibody to the CD3 component of the TCR (anti-CD3) (183). When added to resting T cells, TCR agonists induce the transcription of the genes encoding IL-2 and α subunit of the IL-2 receptor (IL-2Rα) (184). Thus, in this system, competence factors promote progression by providing cells with both IL-2 and an obligate component of the high-affinity IL-2 receptor. In conditions in which IL-2 signaling is precluded, TCR activation allows resting T cells to exit G0 and partially traverse G1 (185); in this latter respect, competent T cells differ from competent fibroblasts, which do not traverse G1 in the absence of progression factors. Continued progression through the cell cycle, however, is dependent on events elicited by both IL-2 and serum, which is a routine supplement of T cell culture media.

We examined the effects of serum on cyclin expression, p27Kip1 down-regulation, and CDK activation in primary splenocytes and purified T cells receiving Con A or anti-CD3 (185). Similar results were obtained in both populations and with either Con A or anti-CD3, and for the sake of simplicity, we limit our discussion to Con A-treated splenocytes (Fig. 5 and Table 2). In two major respects, the actions of Con A in splenocytes precisely mirrored those of PDGF in fibroblasts: in the absence of serum, Con A efficiently induced the expression of a D cyclin (in this case, cyclin D3) and caused a brief and relatively small decline in p27Kip1 abundance. However, in sharp contrast to cyclin D1-Cdk4 and cyclin D3-Cdk4 complexes in PDGF-treated fibroblasts, cyclin D3-Cdk6 complexes in Con A-treated splenocytes were fully active. The reason for this difference is not known; it is possible that D cyclin-CDK activation requires partial p27Kip1 depletion coupled with a second event that is induced by competence factors in splenocytes but by progression factors in fibroblasts. Alternatively, and perhaps due to differences between Cdk4 and Cdk6, D cyclin-CDK activation in fibroblasts may be more susceptible to p27Kip1-mediated inhibition than is D cyclin-CDK activity in splenocytes. Con A also mimicked PDGF in one minor respect (it increased cyclin A abundance to a small extent) and differed from PDGF in two minor respects (it increased the abundance of both cyclin E and a D cyclin-associated CDK).

Fig. 5.

CDK events associated with competence and progression in T cells. Competence factors such as Con A induce a partial and transient loss of p27Kip1. In the first part of progression (1), serum further reduces p27Kip1 amounts. As a result, cyclin E-Cdk2 complexes become active and increase the expression of IL-2Rα at a posttranscriptional level; IL-2Rα expression leads to the activation of IL-2 signaling pathways. In conjunction with cyclin D3-Cdk6 complexes, cyclin E-Cdk2 also increases the expression of cyclin A by phosphorylating Rb and derepressing E2F-mediated transcription. In the second part of progression (2), IL-2 signaling pathways optimize and sustain both cyclin E-Cdk2 and cyclin A-Cdk2 activities, which in turn further promote the accumulation of IL-2Rα.

Although ineffective in the absence of Con A, serum acted in conjunction with Con A to maximally and persistently down-regulate p27Kip1; as a result, catalytically active p27Kip1-free Cdk2 complexes were generated and cyclin A was highly expressed (Table 1 and Fig. 5). These actions of serum are reminiscent of those of PPP in fibroblasts. It is not known how Con A decreases p27Kip1 abundance and how serum potentiates this effect. Treatment of splenocytes with Con A and serum also elevated the expression of Cdk2, which may contribute to the formation of active cyclin E-Cdk2 and cyclin A-Cdk2 complexes in these cells. Despite some differences in cyclin-CDK expression and activation in fibroblasts versus T cells, the conservation of the major functions of competence factors (enhanced expression of a D cyclin and partial down-regulation of p27Kip1) and progression factors (Cdk2 activation and maximal down-regulation of p27Kip1 and Cdk2) in these cell types establishes a paradigm that may also be applicable to other cell types.

Fig. 6.

Model depicting the role of the pocket proteins in density-dependent growth arrest. In concert with Cdk4, Cdk2 phosphorylates the pocket proteins, thus resulting in their release from E2F-DP complexes and the consequent derepression of E2F-mediated gene expression (step 1). Gene products encoded by E2F target genes act in conjunction with other Cdk2-dependent events (step 2) to induce the entry of cells into S phase. One or more of these gene products also plays a key role in abrogating the antiproliferative barriers that are produced in cells at high densities by cell-cell contact and spatial and other constraints. To generate sufficient quantities of this gene product(s), high-density cells require large amounts of Cdk2 or Cdk4 activity, or both, to efficiently inactivate the pocket proteins. The requirement for high CDK activity in dense cultures can by bypassed by proteins encoded by DNA tumor viruses (e.g., the SV40 large T antigen); these viral proteins bind the pocket proteins, thus preventing their interaction with E2F-DP complexes. SV40-infected cells still require Cdk2 activity (although at minimal levels) for step 2.

Table 1.

Effects of PDGF, activated Ras, and PPP on cyclin-CDK activation in fibroblasts.

Table 2.

Effects of Con A and serum on cyclin-CDK activation in splenocytes.

Serum facilitates the expression of IL-2 and of surface-localized IL-2 receptors in antigen-treated human T cells, and IL-2 down-regulates p27Kip1 and activates Cdk2 in T lymphoblasts (21, 186). Consistent with these observations, we found that serum increases the expression of IL-2Rα by a posttranscriptional mechanism in Con A-treated splenocytes and that ablation of IL-2 signaling precludes Cdk2 activation in splenocytes receiving Con A and serum (187). These observations suggest that serum simply acts through IL-2, which in turn elicits the events required for p27Kip1 down-regulation and Cdk2 activation. However, Cdk2 activation not only results from but also is required for IL-2Rα accumulation and consequent IL-2 signaling in splenocytes (Fig. 5).

We suggest that serum promotes IL-2α accumulation by a process that is dependent on Cdk2 activity (185, 187). First, splenocytes exhibiting constitutive cyclin E-Cdk2 activity due to loss of p27Kip1 express large amounts of IL-2Rα in the absence of serum. Second, roscovitine, a potent and selective inhibitor of Cdk2 activity (188), reduces IL-2Rα expression when added to splenocytes or T lymphoblasts in combination with Con A and serum. Third, amounts of ectopically expressed IL-2Rα are lower in fibroblasts that conditionally express p27Kip1 than in control fibroblasts. In contrast, IL-2Rα mRNA amounts are similar in the presence and absence of exogenous p27Kip1, thus indicating that Cdk2 increases IL-2Rα accumulation at a posttranscriptional level.

On the basis of the data presented above, we suggest that the G1 phase of primary splenocytes consists of three parts (Fig. 5). In the competence phase, Con A partially down-regulates p27Kip1, activates cyclin D3-Cdk6, induces the transcription of IL-2Rα, and allows partial traverse through G1. In the first part of progression, serum maximally reduces p27Kip1 amounts, activates cyclin E-Cdk2 complexes, and consequently promotes IL-2Rα accumulation and the activation of IL-2 signaling pathways. Serum (through activation of Cdk2) up-regulates IL-2Rα expression at the level of translation (189). In the second part of progression, IL-2 signaling pathways optimize and sustain cyclin E-Cdk2 and cyclin A-Cdk2 activities, perhaps by keeping p27Kip1 amounts low. Thus, the first and second progression phases are interdependent: Cdk2 activation enhances IL-2Rα expression and IL-2 signaling increases Cdk2 activity. As a result of this regulatory loop, splenocytes complete their passage through G1 and enter S phase. At present, the mechanism by which Cdk2 promotes the accumulation of IL-2Rα remains to be determined.

Density-Dependent Effects of p27Kip1 on Proliferation

When exposed to mitogens, cells arrested at low densities exhibit a robust proliferative response, whereas those arrested at high densities undergo limited if any cell cycle progression. One explanation of this phenomenon is that signals that are sufficient for the proliferation of low-density cultures are insufficient for the proliferation of high-density cultures. Similarly, the continued proliferation of sparse, growing cells may necessitate increases in signal strength (or different types of signals) as culture density increases. We therefore examined the ability of high- and low-density cultures to grow in conditions in which CDK activity was minimized by enforced expression of p27Kip1. The results of these experiments show that culture density sets the threshold level of CDK activity required for the passage of cells through the cell cycle.

We prepared a Balb/c-3T3-derived cell line (termed p27-47) that ectopically expresses supraphysiological levels of p27Kip1 in response to isopropyl β-D-thiogalactopyranoside (IPTG) (36). Conditional expression of p27Kip1 in p27-47 cells markedly represses both Cdk4 and Cdk2 activities, as well as the CDK-dependent dissociation of p130-E2F4 complexes. To assess the effects of p27Kip1 overexpression and consequent CDK inactivation on culture growth, IPTG was added to p27-47 cells plated at low density in serum-containing medium, and cell number was determined at various times thereafter. Surprisingly, subconfluent cultures grew exponentially in the presence of IPTG, and the doubling times of control and treated populations were similar. Differences in growth rate became apparent as cells approached confluency; cells receiving IPTG grew more slowly and ceased proliferation at lower densities than did control cells. Although less dense, IPTG-treated populations consisted of cells that were larger in size than those in control populations and thus formed confluent monolayers. Regardless of density, p27Kip1 was highly expressed and functional in cells exposed to IPTG. These findings show that cells cultured at low density can proliferate in conditions in which CDK activity is severely curtailed (although not eliminated; see below).

We also grew p27-47 cells in medium containing different concentrations of serum; as a result, cells became confluent and quiescent at densities that increased as a function of serum concentration. Cells were then stimulated with PDGF and serum in the presence or absence of IPTG, and the percentage of cells that entered S phase was determined. Cells arrested at the lowest densities efficiently initiated DNA synthesis in medium containing mitogens and IPTG. As culture density increased, proliferative capacity decreased, and at the highest densities, cell cycle traversal was completely blocked by overexpression of p27Kip1. In the absence of IPTG, the ability of cells to enter S phase was also impaired at higher densities, although to a lesser extent than in IPTG-treated cultures. Thus, it appears that the degree of confluency (rather than confluency itself) dictates the responsiveness of IPTG-treated p27-47 cells to mitogenic stimulation.

Passage through the restriction point and into S phase is dependent on the phosphorylation and inactivation of the pocket proteins by Cdk4 and Cdk2. DNA tumor viruses such as SV40 and adenovirus encode proteins that bind the pocket proteins and thus prevent the interaction of pocket proteins with E2Fs in a CDK-independent manner (190-192). Viral oncoproteins, therefore, bypass the need for CDK activation for pocket protein inactivation. Consistent with this observation, we found that infection of p27-47 cells with SV40 counteracts the growth-inhibitory effect of high density and p27Kip1 overexpression (36). Because Cdk2 promotes proliferation by inducing events in addition to pocket protein inactivation, SV40 must either elicit or override these events or these events must still occur (although at reduced levels) in dense IPTG-treated cultures. In support of the latter possibility, we found that SV40 does not overcome p27Kip1-mediated growth arrest when presented to cells in combination with concentrations of roscovitine that would be expected to completely abolish Cdk2 activity (36). Similarly, SV40 large T antigen does not stimulate the proliferation of cells in which Cdk2 activation is ablated by expression of dominant-negative Cdk2 (173). Thus, SV40 acts in conjunction with (rather than substitutes for) Cdk2, and SV40-infected p27-47 cells apparently contain residual amounts of Cdk2 activity despite expression of large amounts of p27Kip1.

On the basis of the data presented above, we suggest that culture density specifies the amount of CDK activity required for cell cycle traversal. Cells at low density need only minimal amounts of Cdk4 and Cdk2 activity, and thus continue to proliferate in conditions that substantially reduce but do not abolish CDK activity. Cells at high density, on the other hand, require greater amounts of CDK activity, and in cells that overexpress p27Kip1 (or in control cells at very high densities), this amount of activity is not attained and proliferation does not occur. This premise is supported by data assessing the affects of density on the amounts of endogenous p27Kip1 in Swiss 3T3 cells (193). Serum reduced p27Kip1 abundance when added to cells arrested at high density, but not when presented to cells arrested at low density. Although cells at low density contained larger amounts of p27Kip1 than did cells at high density, both populations efficiently entered S phase in response to serum. Thus, in Swiss 3T3 cells, low density eliminates the need for p27Kip1 down-regulation for cell cycle traversal.

Cell-cell contact results in the generation of growth-suppressive signals that may increase in number and intensity as culture density increases (194). Moreover, it is likely that high density creates additional spatial (or other) constraints that further restrict proliferation. We suggest that these density-imposed cell cycle barriers can only be overcome by conditions (for example, large amounts of CDK activity or SV40 infection) that allow the efficient disruption of pocket protein complexes and the consequent expression of E2F target genes (Fig. 6). These genes presumably encode proteins that (when present at sufficient quantities) break down or negate the antiproliferative barriers present in cells at high densities. The capacity of dense SV40-infected p27-47 cells to enter S phase while overexpressing p27Kip1 indicates that high levels of Cdk2 activity are not required for Cdk2-dependent events other than pocket protein inactivation. The cell cycle barriers characteristic of cells at high densities would be absent from or diminished in cells at low densities; as a result, cells at low densities would not require high levels of the gene products required by cells at high densities and thus continue to proliferate even in conditions that severely limit the pocket protein inactivation.


Our studies dissected the actions of different classes of mitogens on the activation of the G1 CDKs in fibroblasts and T cells and identified the processes elicited by competence versus progression factors. The delegation of specific sets of mitogens to specific aspects of CDK activation explains (at least in part) why multiple mitogens are often required for the G1 traversal of various cell types. In other systems, all processes required for CDK activation are accomplished by a single mitogen. Our studies also suggest a mechanism by which high culture density inhibits growth. We found that the amount of CDK activity needed for cell cycle traversal increases as culture density increases, eventually reaching a point where the CDK activity threshold cannot be attained and, consequently, where cells are unable to proliferate.

On the basis of our (and other) experimental data, we argue in favor of an inhibitory role of p27Kip1 and p21Cip1 in the regulation of D cyclin-associated CDK activity and suggest that cells contain sufficient amounts of D cyclin-CDK complexes to both sequester these CKIs and phosphorylate Rb. Lastly, we describe a regulatory loop in T cells in which Cdk2 activation enhances IL-2Rα expression and IL-2 signaling increases Cdk2 activity. Whether Cdk2 modulates the posttranscriptional expression of other cell cycle regulators in other systems remains to be determined.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
  172. 172.
  173. 173.
  174. 174.
  175. 175.
  176. 176.
  177. 177.
  178. 178.
  179. 179.
  180. 180.
  181. 181.
  182. 182.
  183. 183.
  184. 184.
  185. 185.
  186. 186.
  187. 187.
  188. 188.
  189. 189.
  190. 190.
  191. 191.
  192. 192.
  193. 193.
  194. 194.
View Abstract

Stay Connected to Science Signaling

Navigate This Article