This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Pestell, R. G. Articles by Arnold, A. Search for Related Content PubMed PubMed Citation Articles by Pestell, R. G. Articles by Arnold, A.

The Cyclins and Cyclin-Dependent Kinase Inhibitors in Hormonal Regulation of Proliferation and Differentiation¹

The Albert Einstein Cancer Center (R.G.P., C.A., A.T.R., R.J.L.), Department of Developmental and Molecular Biology and Department of Medicine, Department of Anatomy and Structural Biology (J.E.S.), Albert Einstein College of Medicine, Bronx, New York 10461; and Center for Molecular Medicine (A.A.), and Division of Endocrinology and Metabolism, University of Connecticut Health Center, Farmington, Connecticut 06030

	Abstract

Top Abstract I. Introduction II. Regulation of the... III. Endocrine Regulation of... IV. Cyclins and CKIs... V. Transcription Factors... VI. Conclusion Note Added in Proof References

 

I. Introduction

II. Regulation of the Cell Cycle G₁ Phase

A. The cyclins

B. The Cdk-inhibitory proteins

C. Regulation of transcription by cell cycle-control proteins

III. Endocrine Regulation of Cyclin and Cdk-Regulatory Proteins

A. Single-transmembrane segment receptors

B. Seven-transmembrane receptors

C. Steroid hormones

D. Intracellular second messengers

IV. Cyclins and CKIs in Endocrine Tumors

A. Parathyroid adenomas

B. Thyroid cancer

C. Pituitary tumors

D. Adrenal tumors

E. Ovarian and testicular cancer

F. Mechanisms of oncogenic transformation by cyclin D1

G. CKIs in treatment of endocrine disease

V. Transcription Factors Regulating the Cell Cycle

VI. Conclusion

	I. Introduction

Top Abstract I. Introduction II. Regulation of the... III. Endocrine Regulation of... IV. Cyclins and CKIs... V. Transcription Factors... VI. Conclusion Note Added in Proof References

 
THE LAST decade has seen an enormous expansion in our understanding of the molecular mechanisms governing cell cycle regulation. Several recent reviews have described the importance of the coordinated regulatory interactions between components of the cell cycle-regulatory apparatus and their aberrations in tumorigenesis (1, 2, 3, 4, 5, 6, 7). More recently the mechanisms by which hormone signals affect the cell cycle-regulatory apparatus have been investigated. These studies have provided important new insights into the mechanisms by which hormones regulate diverse processes including differentiation, proliferation, and apoptosis. This review describes the role of the cell cycle-regulatory apparatus in hormone-regulated alterations in cellular phenotype.

The normal mammalian cell cycle consists of several temporally distinct phases (Fig. 1). One current model of the cell cycle envisages transitions between different cell cycle states by passage through checkpoints (1, 2, 3, 5, 8) (Fig. 1). Examples of these states are the initiation and completion of DNA replication (S) phase and of cell division or mitosis (M). Between these phases are gaps (G). One important checkpoint in mammalian cells is the restriction point in late G₁, also known as START in yeast. This is the point at which the cell commits itself irrevocably to another round of DNA replication. Passage through the restriction point is promoted by a group of G₁ cyclins, which include in mid-G₁, the D type cyclins, and in late G_1, cyclin E. These cyclins can heterodimerize with specific catalytic subunits, the cyclin-dependent kinases (Cdks), to form holoenzymes. Some substrates of these holoenzymes, which are inactivated upon phosphorylation, are the retinoblastoma tumor suppressor protein, pRB (retinoblastoma protein) (Fig. 2B) and the related proteins, p130 and p107. It is thought that phosphorylation and inactivation of pRB leads to progression through the restriction point. The ability of the cyclin/Cdk holoenzymes to phosphorylate pRB is inhibited by a family of small molecular weight proteins, known as cyclin-dependent kinase inhibitors (CKIs) (Fig. 2A).

View larger version (124K): [in this window] [in a new window]   Figure 1. Schematic representation of the mammalian cell cycle. Competence factors such as PDGF and FGF promote entry into the early G1 phase. Sequential treatment with progression factors, IGF or EGF, promote progression through the G1 phase restriction point. Competence of the cyclin D1/Cdk4 complex is induced by mitogens. The cyclin D1/Cdk4 complex phosphorylates the pRB protein leading to sequential phosphorylation by cyclin E/Cdk2 and release of free E2F. The phosphorylation of pRB and relief of transcriptional repression by pRB induces genes involved in the induction of S-phase entry.

View larger version (119K): [in this window] [in a new window]   Figure 2. Cdk Inhibitors. A, Schematic representation of the CKI proteins. The p21 protein contains an amino-terminal Cdk-inhibitory domain, a nuclear localization signal (NLS) that lies carboxy terminal to a putative caspase cleavage site (DVHD112L) indicated by an asterisk (474 ), and a C-terminal domain that binds to PCNA. Although cyclin A or cyclin E bind only the amino terminus, two distinct cyclin-binding motifs [Cy1 and Cy2 (87 )], which bind cyclin E/Cdk2 or cyclin A/Cdk2 complexes are indicated. Cyclin D1 can bind to the amino terminus, either alone or in the presence of Cdk4 and binds independently of the K region (87 ). B, CKI function. The cyclin D1 gene product binds its catalytic subunit partner (Cdk4) in the presence of an assembly factor. The cyclin D1/Cdk4 holoenzyme is phosphorylated by a Cdk-activating kinase (475 ) (CAK), which consists of several subunits. The Ink4 proteins, shown as p16Ink4a, can compete with D cyclins to form independent binary complexes with Cdk4 (476 ), thereby inhibiting the activity of the cyclin D1/Cdk4 holoenzyme. The role of the p21 CKI family, shown as p27Kip1, in regulating activity of the cyclin D1/Cdk complex is controversial. In some circumstances p27Kip1 is thought to inhibit activity of the complex (75 ). In other circumstances, p27Kip1 does not inhibit activity of the complex (92 ) and acts as an assembly factor (80 ). These findings suggest that the stoichiometry or cell type may be important in the action of p27Kip1.

	II. Regulation of the Cell Cycle G1 Phase

Top Abstract I. Introduction II. Regulation of the... III. Endocrine Regulation of... IV. Cyclins and CKIs... V. Transcription Factors... VI. Conclusion Note Added in Proof References

 
A. The cyclins
The cyclin D1 protein is a regulatory subunit of a holoenzyme that phosphorylates and inactivates the tumor suppressor pRB (5, 9, 10). The catalytic subunits, Cdk4 and Cdk6, are the primary heterodimeric partners of cyclin D1 (1). Overexpression of cyclin D1 acts in G₁ to promote progression through the cell cycle in mammalian cells. Immunoneutralization and antisense experiments have demonstrated that the abundance of cyclin D1 is rate-limiting in serum-induced G₁ phase progression of Rat1 fibroblasts and MCF7 human breast cancer cell lines (11, 12, 13, 14, 15, 16, 17, 18). Cyclin D1 is not required for G₁-phase progression in cultured mammalian cells that lack functional pRB (19). In Drosophila melanogaster, cyclin D is also necessary, but not sufficient, for the initiation of events leading to S phase entry (20, 21).

The human cyclin D1 gene was cloned as an endocrine tumor oncogene in human parathyroid adenomas. During structural analysis of the PTH gene, a pericentromeric inversion was observed in chromosome 11 (22) (Fig. 3A). Cloning and sequencing of the chromosomal breakpoint in these tumors revealed the presence of an overexpressed gene, downstream of the PTH gene promoter (22). The PRAD1 cDNA (parathyroid adenoma 1) revealed structural homology to a class of proteins previously identified in yeast known as cyclins (23). Cyclins were known to regulate the first Gap phase (G₁) in nonmammalian cells. The PRAD1 gene product was thus the first putative mammalian G₁ cyclin cloned. Furthermore, the PRAD1 cDNA was capable of rescuing a yeast mutant in its G₁ CLN cyclins, which suggested the PRAD1 gene may encode a functional cyclin now called a D-type cyclin (24). Independently, the murine homolog of PRAD1 was cloned as a colony-stimulating factor-1 (CSF-1)-responsive gene product (25).

View larger version (47K): [in this window] [in a new window]   Figure 3. A, Structural organization of the PTH gene in a parathyroid adenoma demonstrates the pericentric inversion of chromosome 11. The cyclin D1 gene is positioned 3' to the parathyroid (PTH) gene-regulatory region. B, The cyclin D1 protein is encoded by five exons. Conserved motifs include a cyclin box, which makes contact with the Cdk. This site is conserved between the cyclins. In cyclin A the crystal structure revealed two sequential 90-amino acid repeats, referred to as cyclin folds, separated by a short linker peptide. The first fold corresponds to the cyclin box. An LXCXE motif conserved with several viral oncogenes is shown, and a site capable of interacting with coactivator proteins (LLESSL) is indicated. The carboxy terminus of cyclin D1 interacts with HLH proteins involved in myogenesis.

The 35-kDa cyclin D1 protein is encoded by 5 exons in the structurally similar human and mouse genes (26). The cyclin D1 protein shares structural homology to the other cyclins. The amino terminus of cyclin D1 contains a motif Leu-X-Cys-X-Glu (where X represents any amino acid). This motif, which is shared by the viral oncoproteins E1A, simian virus (SV) 40 large T antigen, and papillomavirus E7, is involved in binding the pRB pocket domain (Fig. 3B). A region located carboxy terminal to the pRB-binding domain is known as the "cyclin box" because it is conserved between the known cyclins. The acidic rich carboxy terminus of cyclin D1 inhibits myogenic helix loop helix (HLH) protein function (27). An alternate splice form of cyclin D1 encodes a protein with an altered carboxy-terminal domain (CTD) (28). The cyclin D1 protein is quite unstable, with a half-life of less than 20 min, with degradation occurring through ubiquitin proteosome-mediated degradation (29). In quiescent cells, cyclin D1 protein levels are low. However, nuclear abundance increases as cells progress through G₁ phase (11). As cells pass into S phase, cyclin D1 moves from the nucleus to the cytoplasm. Exclusion of cyclin D1 from the nucleus is required for progression into S phase in human fibroblasts (11). Proliferating cell nuclear antigen (PCNA), which is required for DNA polymerase activity, binds cyclin D1 in the nucleus. As cells enter S phase, cyclin D1 protein no longer binds PCNA and is extruded from the nucleus (30).

Two other structurally related cyclins, D2 and D3, are also capable of heterodimerizing with Cdk4/6 and phosphorylating pRB in vitro. The amino acid identity between the human D-type cyclins is 53–63%. Cyclin D2 was mapped to chromosome 12 at 12p13 and cyclin D3 was mapped to 6p21. It is currently thought that each of the D-type cyclins subserve multiple functions, some of which are shared, such as phosphorylation of the pRB protein, and some of which are distinct. The mRNA distribution and expression profiles differ between the D-type cyclins. Steady state cyclin D2 mRNA levels, for example, peaked in late G₁ (31) and for cyclin D3 peaked in S phase rather than early in G₁ as seen for cyclin D1. Unlike the relatively ubiquitous distribution of cyclin D1 expression, the expression of cyclin D2 is somewhat more restricted with mRNA expressed abundantly in T lymphocytes and gonadal cells, although several different transformed cell lines also express cyclin D2 (25, 31, 32, 33).

Cyclin E was first identified by screening human cDNA libraries for genes that would complement G₁ cyclin mutations in Saccharomyces cerevisiae and has subsequently been found to have specific biochemical and physiological properties that are consistent with a G₁ function in mammalian cells. mRNA levels for cyclin E peak later in G₁ phase than cyclin D1. Mammalian cells express several isoforms of cyclin E protein. These proteins are encoded by alternatively spliced mRNAs and are localized to the nucleus during late G₁ and early S phase (34). The cyclin E-Cdk2 complex is maximally active at G₁/S, and overexpression of cyclin E decreases the time it takes the cell to complete G₁ and enter S phase. The destruction of cyclin E is linked to ubiquitin-mediated degradation. Cyclin E phosphorylation is coupled to cyclin E turnover via site-specific phosphorylation, which acts as a signal for ubiquitination and proteasome processing (35, 36).

Cyclin D1 and cyclin E appear to subserve at least partially overlapping functions. pRB is phosphorylated at an overlapping subset of sites by these two distinct kinases (37). Several different scenarios may explain why two different kinases subserve this function. Phosphorylation of pRB at distinct sites may alter the ability of pRB to interact with a distinct subset of transcription factors or substrates (38). Alternatively, phosphorylation at distinct sites by both kinases may alter the epitopes of pRB required for full phosphorylation. It is clear, however, that the effect of cyclin D1 to promote cell cycle progression in cultured cells requires pRB, while the effect of cyclin E occurs independently of pRB (19, 39, 40). Thus, fibroblasts engineered to constitutively overexpress cyclin E showed elevated cyclin E-dependent kinase activity and a shortened G₁ phase of the cell cycle (40, 41). Under certain circumstances, in cultured Rat-1 cells, cyclin D1, but not cyclin E, induced pRB phosphorylation, suggesting cyclin D1 and cyclin E promote G₁ phase progression through different mechanisms (42). However, recent studies in NIH3T3 cells demonstrated that cyclin E can phosphorylate pRB and located a pRB-binding motif VxCxE (43). Mutation of the pRB-binding motif abolished the ability of cyclin E to promote S phase entry in NIH3T3 cells (43). It is clear that, in addition to pRB, alternate substrates for cyclin D/Cdk exist (below). Histone H1 is also phosphorylated well by cyclin E/Cdk2. Because cyclin E enhancement of G₁ phase progression occurs independently of pRB (14, 40), a search has begun for alternate cyclin E/Cdk substrates. Recent studies have identified one new substrate for cyclin E/Cdk2, known as the NPAT protein (nuclear protein mapped to the ATM locus) (44). It is thought that phosphorylation of NPAT by cyclin E/Cdk2 may promote S phase entry.

Cyclin A plays an essential role in the progression through the cell cycle as a regulatory component of the Cdk2 and cdc2 kinases (1, 45). The mRNA and protein of cyclin A accumulate at the end of the G₁ phase. Microinjection of immunoneutralizing antibodies to cyclin A or antisense expression vectors indicate a requirement for cyclin A in DNA replication (46). Complexed to Cdk2 in cooperation with cyclin E, cyclin A has been shown to trigger S phase entry of G₁ nuclei from HeLa cells (47). Cyclin A has been shown to promote S phase entry by associating with transcription factors and by regulating target genes involved in cell growth regulation.

In normal human fibroblasts, cyclin A/Cdk2 forms a quaternary complex with p21^Cip1 and PCNA and cyclin A forms associations with p107 and E2F transcription factors. Cyclin A/Cdk2 binds to the amino terminus of the E2F-1–3 transcription factors. The E2F multigene family contains six members, and preferential binding of cyclin A/Cdk2 to E2F-1–3 likely alters the repertoire of E2F transcriptional target genes. The E2F proteins bind to DNA with an heterodimeric binding partner from the DP protein family and activate gene transcription through the E2F enhancer sequence. This DNA-binding site was initially identified as an adenovirus AdE2a enhancer sequence and has since been identified in the promoter of a number of genes induced during G₁/S phase transition. E2F-1 is phosphorylated by cyclin A/Cdk2, down-regulating its ability to bind DNA and activate transcription (48, 49, 50). The region of E2F-1 phosphorylated by cyclin A/Cdk2 is conserved among E2F proteins (E2F-1, E2F-2, -E2F-3), implying each of the protein’s binding activity may be regulated in this manner. E2F-1 mutants defective in cyclin A binding cause apoptosis and deregulated cell cycle progression, indicating the importance of E2F-1 phosphorylation-dependent inactivation (49, 50).

A number of other cyclins and Cdks have been identified and are touched on only briefly herein. The abundance of cyclin F fluctuates during cell cycle progression like cyclin A, peaking in G₂ (51). Overexpression of cyclin F increases the proportion of cells in G₂ phase (51). The expression of cyclin G (52) is induced upon nerve injury in the motor neurons during the early phase of the nerve regeneration process (53). A neuron-specific cyclin p35 has been described that is expressed in postmitotic neurons of the central nervous system (CNS) (54, 55). The heterodimeric partner for p35 is Cdk5 (55, 56). Mice lacking p35 develop abnormal lamination likely due to abnormal neuronal migration (57). The related cyclin p39 is also required for neurite outgrowth in a cultured hippocampal cell line model (58). Cyclin H and Cdk7 form a Cdk-activating kinase (CAK), which regulates activities of the cyclin/Cdk complexes and activity of the RNA Pol II CTD. Unlike many of the other cyclins the abundance of CAK does not change during cell cycle progression, although its activity and substrates may (discussed below). Mitotic entry is signaled by the accumulation of cyclin B-cdc2 (59). Although accumulating in S and G₂ phases, the activity of cyclin B/cdc2 is inhibited by phosphorylation on Tyr-15 and Thr-14 by Wee1/Mik1/Myt1-related protein kinases. The Cdc25c phosphatase is stimulated to dephosphorylate T14/Y15 and activate Cdc2. Destruction of the B-type cyclins through a ubiquitin-dependent proteolysis is required for progression past anaphase (60, 61).

Thus, specific cyclins convey cell-type dependent developmental functions and regulate specific steps in progression through the cell cycle of dividing cells. Recent attention has been drawn to aberrations in function of these proteins in tumorigenesis. The normal function of these cyclins in the mature animal in nondividing cells, however, remains to be fully understood.

B. The Cdk-inhibitory proteins
Activity of the holoenzyme containing cyclin D1 (cyclin D1/Cdk4 and cyclin D1/Cdk6) referred to here as cyclin D1 kinase (CD₁K), is modulated in vitro by the Cdk-regulatory proteins (62), which are divided into two broad categories (Fig. 2). The first family, the Ink4s (p16^Ink4a, p15^Ink4b, p18^Ink4c, and p19^Ink4d), inhibit specifically Cdk4 or Cdk6 and contain a set of highly conserved ankyrin ring motifs (Fig. 2A). The second group include the Cip/KIP family (p21^Cip1, p27^Kip1, and p57^Kip2), which share partial structural homology and possess the ability to inhibit cyclin/Cdk complexes. All CKIs can cause G₁ arrest when overexpressed in transfected cells (reviewed in Ref. 2). The expression and subcellular distribution of the CKIs is regulated by hormones in a complex and cell-type specific manner (Fig. 2).

Overexpression of Ink4 proteins in vitro dissociates the CD₁K complexes and overexpression of p16^Ink4a is associated with reduced levels of cyclin D1-holoenzyme complexes (Fig. 2B). The p16 (INK4a/ARF) locus encodes two alternate transcripts, the p16^Ink4a and the p19^ARF (alternate reading frame). Overexpression of either of these transcripts can induce cell cycle arrest and block transformation (63, 64, 65). Transgenic mice homozygously deleted of the p16 locus spontaneously developed a variety of malignancies including lymphomas and fibrosarcomas (66), and deletion of the p19^ARF gene resulted in animals with a similar phenotype (67). Unlike p16^Ink4a, however, the cell cycle arrest induced by p19^ARF is p53 dependent. p19^ARF directly binds MDM2 and stabilizes p53 (65, 68). The human equivalent to p19^ARF, p14^ARF, also binds MDM2, resulting in stabilization of both p53 and MDM2 (69) (Fig. 2B). Consistent with a negative feedback loop, p53 in turn inhibits p14^ARF (69) and p19^ARF expression (70). Unlike p53, however, p19^ARF is not involved in the DNA damage response (69). Together these studies suggest that the alternate reading frames of the CdkN2A locus function to inhibit cell cycle progression through two different mechanisms (71). The evolutionary pressure to sustain the presence of these two tumor suppressor genes, possibly as the result of gene duplication, in such close proximity and given their vulnerability to co-deletion, remains an area of considerable speculation. The possibility that each transcript plays a role in hormonal signal transduction specificity has yet to be explored.

The second class of inhibitors, the p21 family, includes p21^Cip1 (72, 73), p27^Kip1 (74, 75), and p57^Kip2 (76, 77, 78). Although intially referred to as CKIs, with biochemical properties supporting this notion (79), strong evidence is accumulating that p21^Cip1 and p27^Kip1 function as assembly factors in vitro (80) and in vivo (81) to enhance cyclin D-dependent activity. The p21 family members have a conserved region near the amino terminus that is necessary and sufficient for binding to and inhibiting Cdk2 (82, 83, 84). At certain concentrations, however, p21^Cip1 does not inhibit cdc2 or Cdk2 kinase activity (85). Thus although p21^Cip1 inhibits Cdk4 and Cdk6 kinase activity, with an inhibition constant (K_i) of 0.5–15 nM, p21^Cip1 is a poor inhibitor of cdc2/cyclin B in vitro with a K_i of 400 nM (86). In subsequent studies, functional subdomains of p21^Cip1 were defined with fine deletional analysis. For example, the binding of p21^Cip1 to cyclin E/Cdk2 and cyclin A/Cdk2 was shown to involve both a Cdk2-binding domain and either an amino-terminal or carboxy-terminal cyclin-binding domain whereas binding by cyclin D1 involved only the amino-terminal cyclin-binding domain (87) (Fig. 2A). The carboxy-terminal region of p21^Cip1 allows it to associate with PCNA, a processivity subunit of the DNA polymerase holoenzyme (82, 83, 84, 88, 89). Because the binding of p21^Cip1 to PCNA inhibits the processivity of polymerization, but does not affect excision repair, it was suggested that p21^Cip1 may serve to coordinate DNA replication with cell cycle progression (90).

The p21^-/- embryonic fibroblasts derived from mice homozygously deleted of the p21^Cip1 gene (91) exhibit a significant growth alteration in vitro, achieving a saturation density as high as that observed in p53^-/- cells. In addition, p21^-/- cells are significantly deficient in their ability to arrest in G₁ in response to DNA damage and nucleotide pool perturbation. Although the p21 family of proteins inhibit the activity of Cdk2-containing complexes, recent studies demonstrated that the p21 family may also function under some circumstances as assembly factors to promote the association of Cdk4 with D-type cyclins (80, 92). The p21 family proteins enhance activity of the cyclin D1-kinase (CD₁K) complex at low concentrations, while inhibiting CD₁K complex at high concentrations.

The p27^Kip1 protein was initially characterized as a protein homologous to the tumor suppressor p21^Cip1. When overexpressed in fibroblasts, cell cycle progression was delayed and antisense p27^Kip1 experiments resulted in mitogen-independent G₁ phase progression, indicating a critical role for p27^Kip1 in the establishment or maintenance of cellular quiescence (93, 94). The abundance of p27^Kip1 is regulated primarily at a posttranslational level, although translational control also contributes. Thus, p27^Kip1 mRNA levels remain relatively unchanged during the cell cycle transition; however, the addition of mitogens reduces p27^Kip1 protein levels. For example, in quiescent 3T3 cells p27^Kip1 protein levels decrease after mitogenic stimulation (95, 96, 97). The growth factor-mediated reduction in p27^Kip1 protein levels is mediated primarily through enhanced ubiquitin-mediated degradation (98). It will be of interest to determine whether progression through the G₁/S phase of the cell cycle is associated with induced expression of proteolytic enzymes, such as members of the caspase family. p27^Kip1 abundance was also implicated as an important mediator of the cytostatic effects of rapamycin and cAMP (75, 99, 100), although these implications were not borne out in all studies using transgenic mice models (below).

The cyclin/Cdk complex to which p27^Kip1 is bound determines its functional activity. p27^Kip1 is found associated with cyclin E in a variety of cell types during quiescence (95, 99). When bound to cyclin D1/Cdk4, p27^Kip1 may not be inhibitory (74, 99, 101, 102), whereas cyclin E/Cdk2 activity is inhibited by p27^Kip1. It is thought that the removal of p27^Kip1 from the cyclin E/Cdk complex is an essential step for S-phase entry. Through binding cyclin D1/Cdk4, p27^Kip1 is sequestered from cyclin E/Cdk2, reducing its inhibition by p27^Kip1 (74, 99, 101, 102).

The degradation of p27^Kip1 upon mitogen stimulation is dependent upon prior phosphorylation. Expression of cyclin E/Cdk2 in murine fibroblasts was found to induce phosphorylation of p27^Kip1 on T187 (103). A mutant of p27^Kip1 at amino acid T187, to alanine, created a p27^Kip1 protein that caused a G₁ block resistant to cyclin E overexpression and whose level of expression was not modulated by cyclin E. Phosphorylation of p27^Kip1 by cyclin E/Cdk2, enhanced degradation of p27^Kip1, thereby promoting G₁-S phase transition. Thus, the cyclin/Cdk complexes promote cell cycle progression in mammalian cells by also enhancing degradation of the CKI.

Together these studies suggest that p27^Kip1 interacts with cyclin E/Cdk2 in two distinct ways. First, through tightly binding to cyclin E/Cdk2, p27^Kip1 inhibits the ability of the holoenzyme to phosphorylate target substrates, such as pRB, or the NPAT protein (44). Second, p27^Kip1 is phosphorylated by cyclin E/Cdk2, leading to the release and subsequent degradation of p27^Kip1. The ability of p27^Kip1 to either block Cdk activity or serve as a substrate for the Cdk appears to be determined by the ambient concentration of ATP within the reaction. At low ATP concentrations (<50 mM) p27^Kip1 is primarily a CKI, but at ATP concentrations approaching physiological levels (>1 mM), p27^Kip1 is more likely to be a substrate.

The homozygous deletion of the p27^Kip1 gene resulted in animals with organomegaly, intermediate lobe pituitary tumors, and testicular and ovarian cell hyperplasia (104, 105). Because p27^Kip1 expression was induced upon oligodendrocyte differentiation, the role of p27^Kip1 in this process was closely examined in mice homozygously deleted of the p27^Kip1 gene [p27 knockout (KO)]. The oligodendrocytes derived from p27 KO mice underwent prolonged proliferation and delayed differentiation, suggesting a role for p27^Kip1 in promoting oligodendrocyte proliferation (106). In view of previous studies demonstrating the induction of p27^Kip1 by transforming growth factor-ß (TGFß) and rapamycin, signaling by these agents was assessed in the p27 KO mice. Surprisingly, initial analysis of p27^Kip1 KO T cells suggested their proliferation rates in response to TGFß and rapamycin were similar to those of wild type, suggesting that p27^Kip1 was not required for the cytostatic effect of these two agents (104, 105). In subsequent experiments, however, exponentially growing p27^-/- fibroblasts were found to have an impaired antiproliferative response to rapamycin. The inhibition of DNA synthesis by rapamycin was approximately half of control p27^+/+ fibroblasts (107). In addition, the p27^-/- T lymphocytes were 15- to 30-fold more sensitive to the growth-inhibitory effect of rapamycin (107). The different results obtained by these investigators may be due to methodological differences, but together provide support for the role of p27 in rapamycin-dependent inhibition of cellular proliferation. The p27^Kip1 gene product is not a significant source of specific selected inactivating mutations, however, raising suspicions that this locus may not encode a human tumor suppressor gene. The results of studies exposing p27^Kip1 KO mice to carcinogenic agents or matings to transgenic tumor-prone animals may provide important insights into the potential tumor suppressor function of the p27^Kip1 gene product.

p57^Kip2 was also cloned as a protein related to p21^Cip1 and p27^Kip1. Overexpression of p57^Kip2 was found to arrest cells in G₁ (77). Both the amino-terminal cyclin/Cdk binding domain and the carboxy-terminal PCNA binding domains of p57^Kip2 were required for full antimitogenic activity and inhibition of cellular transformation much like p21^Cip1 (78). Unlike p21^Cip1, however, p57^Kip2 was not regulated by p53. Homozygous deletion of the murine p57^Kip2 allele by one group of investigators resulted in animals with abnormal endochondral ossification (108), while other investigators observed increased prenatal growth, adrenal cortical hyperplasia, and cytomegaly (109), features found in patients with Beckwith-Wiedemann syndrome. The pituitary tumors, together with the testicular and ovarian cell hyperplasia in the p27^Kip1 KO mice and the adrenal cortical hyperplasia and cytomegaly in the p57^Kip2 KO mice, launched an excited search for a role of these proteins in the regulation of normal function in these endocrine tissues (below).

Since the original cloning of the CKI proteins, it has become clear that the abundance and activity of these proteins are regulated in a complex manner by hormonal stimuli. Frequently the overexpression of the protein may be sufficient to induce cell cycle arrest, but equally frequently the overexpression of the protein, as observed during differentiation, has been shown to be insufficient to recapitulate the differentiated phenotype induced by a particular hormone. Furthermore, in most circumstances, homozygous deletion of the CKIs has not affected the signaling pathway that induced the abundance of the CKI. Together, these types of findings have led to an understanding that the precise timing and orchestration of the alteration in expression of the CKI, together with changes in their subcellular distribution and the formation of multimeric complexes in the cell, may be critical for the induction of the differentiated phenotype. Thus, although the instruments have been identified, the mechanisms of orchestration are poorly understood and are critical for the hormonal induction of the cellular phenotype.

C. Regulation of transcription by cell cycle-control proteins
As noted above, hormonal signals regulate the abundance of cell cycle-control proteins. The altered levels of the cyclins/Cdks in turn regulate downstream target genes. These genetic effects are conveyed by altering the activity of transcription factors and/or by regulating the basal transcription apparatus. Transcription factors that coordinate hormone-mediated signal transduction can be regulated by the cyclin/Cdk complexes through several different mechanisms (Fig. 4). These effects can be thought of as either regulating transactivation function (Fig. 4, A and B), by altering DNA binding (Fig. 4, C and D), or by altering protein/protein interaction (Fig. 4, E and F).

View larger version (104K): [in this window] [in a new window]   Figure 4. Regulation of transcription factors by the Cdks. A, Enhanced transcription factor binding activity. The phosphorylation of p53 by cyclin A/Cdk2 (110 ) enhances the specificity of its binding to target DNA sequences in the p21Cip1 promoter. B, Reduced transcription factor binding activity. Phosphorylation of E2F-1 at the amino terminus by cyclin A/Cdk2 inhibits E2F/DP-1 binding activity (477 ). The phosphatase PP2A is also capable of enhancing phosphorylation and thereby reducing DP-1 binding activity (367 ). The loss of E2F/DP binding reduces the transcriptional enhancement of target genes containing E2F-binding sites. C, Induction of transcription factor activity. Cyclin A/Cdk2 holoenzymes phosphorylate B-myb to enhance its transcriptional activity through the myb-binding site (MBS). (D). Inhibition of transcription repression. The transcriptional repressor Id2 heterodimerizes with basic HLH proteins that normally activate gene transcription through E box sequences. As Id2 is incapable of binding DNA, it functions as trans-dominant negative regulator of E box-mediated transcription. Cyclin E/Cdk2 or cyclin A/Cdk2 phosphorylate and inhibit Id2 function, promoting E box-dependent transcription. E, Protein-protein interactions. Cell cycle-regulatory proteins form protein-protein interactions with known transcription factors and alter their activity independently of the kinase activity. For example, cyclin D1 binds to the ER and enhances ERE enhancer activity (114 115 ).

Cell cycle-regulatory proteins can also affect the transactivation function of specific transcription factors. The transcription factor B-Myb, for example, which regulates activity of gene expression through the Myb-binding site (PyAACG/TG), is phosphorylated and activated by cyclin A/Cdk2 (111) (Fig. 4C). Several nuclear receptors are also phosphorylated and activated by the cyclin/Cdks, including the glucocorticoid and estrogen receptors (Section III.C below). The glucocorticoid receptor, for example, is phosphorylated by cyclin E/Cdk2 (112). Cell cycle-control proteins can also regulate gene expression by promoting transcription factor binding through inhibiting repressors. The HLH protein Id2 resembles the basic HLH proteins that govern gene expression through E box sequences (Fig. 4D). The Id family of proteins lack basic DNA-binding domains and therefore function as transdominant negative regulators through E box sequences. Phosphorylation of Id2 by cyclin E/Cdk2 or cyclin A/Cdk2 leads to a reduction in the transcriptional repressor function of Id2 and enhances activity through the E box sequence (113). As noted above, the transcriptional activity of the E2F-1 factor is inhibited by cyclin A/Cdk2 binding and phosphorylation (48, 49, 50).

Cell cycle-regulatory proteins can also affect signaling by forming direct protein-protein interactions with known transcription factors. For example, cyclin D1 can bind the estrogen receptor (ER), enhancing the activity of the ER through a synthetic estrogen response element (114, 115). Cyclin D1 also binds a Myb-related protein, DMP1, and modulates its activity (116). Although the nature of the association remains to be fully defined, cyclin D1 also inhibits v-Myb transcriptional activity independently of Cdk activity (117) and cyclin D1 interacts with HLH proteins to inhibit their activity, at least in part, independently of the Cdk domain of cyclin D1 (27). Finally, a growing list of transcription factors are regulated in their activity by either the pRB or pRB-related proteins. Activity of the Sp1 transcription factor is induced by the pRB protein (118). Although the mechanisms by which pRB regulates the activity of these other transcription factors remain to be fully determined, other transcription factors capable of binding to and/or regulated by pRB include Elf-1, E2Fs, PU.1, c-Myc, Sp-1, MyoD, ATF-2, NF-IL6, and UBF (reviewed in Refs. 119, 120). For the time being, these interactions are loosely classified as pRB dependent (Fig. 4E).

Components of the basal transcription apparatus are also regulated by the cyclin/Cdks (121). The RNA polymerase II large subunit contains an essential CTD. The CTD is phosphorylated on a fraction of the RNA polymerase II molecules in vivo and is phosphorylated by the general transcription factor TFIIH in vitro. TFIIH is composed of at least nine subunits, which include Cdk7, cyclin H, and MAT1 (p36). These three subunits (Cdk7, cyclin H, MAT1) form a ternary complex, Cdk-activating kinase (CAK). The kinase activity of the TFIIH complex is directed toward the RNA Pol II CTD (122, 123). CAK preferentially phosphorylates Cdk2, whereas TFIIH preferentially phosphorylates the CTD (124). CAK also phosphorylates the threonine residue of the Cdk when cyclin D1-Cdk has formed a dimeric complex, and CAK is required for full cyclin D1/Cdk4 activity (1, 125). Recent studies have demonstrated that the cyclin H/Cdk7 complex preferentially phosphorylated Cdk2, whereas the multimeric complex that included cyclin H/Cdk7 and MAT1(p36) preferentially phosphorylated the RNA Pol II CTD (126). Thus, the substrate specificity of cyclin H/Cdk7, which is abundant within the cell, is dictated by the presence of MAT1 (124, 126). The CTD is also phosphorylated by a complex known as P-TEFb which includes Cdk9 and cyclin T (which includes T1, T2a, and T2b) (127).

RNA Pol II and Pol III-dependent transcription is reduced during mitosis in part through cell cycle-regulatory proteins (128, 129). RNA Pol III activity is increased in pRB-deficient mice and pRB targets TFIIB. As pRB contains regions of homology to TATA-binding protein and BRF, components of TFIIB, it is thought that pRB disrupts TFIIB by mimicking these two components (130). Mitosis-specific exclusion of transcription factors from chromatin may also be important in cell cycle-specific gene regulation. Phosphorylation and inhibition of poly (A) polymerase by cyclin B/Cdc2 may contribute to the overall decrease in RNA and protein synthesis that occurs during mitosis (129, 131, 132).

Cell cycle-regulatory proteins also affect the high molecular weight coactivator proteins of the p300/CBP family. The p300 protein was noted to undergo phosphorylation during cell cycle transition and during cell cycle exit in myocyte differentiation (133). These high molecular weight coactivators, (p300/CBP), interact with a broad array of nuclear receptors and general transcription factors, including members of the AP-1 family, which are involved in transducing hormone signals (134, 135). The general effects of coactivators on gene transcription are mediated in part through modulating histone structure either through direct histone acetylation activity of the transcription coregulator (136) or through the ability of the coregulator to recruit other proteins with histone-regulatory function (137, 138). The p300/CBP family also regulate cell cycle-regulatory transcriptional targets including p21^Cip1 and cyclin D1 (139, 140, 141). To date, relatively little is known about the hormonal regulation of these associations, although these interactions provide important evidence for cross-talk between the cyclins and the basal transcription apparatus.

	III. Endocrine Regulation of Cyclin and Cdk-Regulatory Proteins

Top Abstract I. Introduction II. Regulation of the... III. Endocrine Regulation of... IV. Cyclins and CKIs... V. Transcription Factors... VI. Conclusion Note Added in Proof References

 
A. Single-transmembrane segment receptors
1. Protein tyrosine kinase receptors.
a. CSF-1.
CSF-1 was the first of a group of factors purified (142) that stimulate the formation of colonies of mature myeloid cells from immature hemopoietic precursors (143). Binding of CSF-1, a homodimeric glycoprotein, to its high-affinity receptor, encoded by the c-fms protooncogene, initiates dimerization, autophosphorylation, and signaling (143, 144). SH2 domain-containing substrates binding to autophosphorylated c-Fms have been implicated in signaling by CSF-1 and include STAT (signal transducer and activator of transcription)-1, Grb2, Src, and phosphatidylinositol 3-kinase (PI3-kinase).

As noted above, the murine homolog of PRAD1 was initially cloned as a CSF-1-responsive gene product (25). Many studies of CSF-1 signaling were performed in the murine macrophage cell line, BAC1.2F5. In these cells CSF-1 induces dimerization and autophosphorylation of the CSF-1 receptor, as well as tyrosine phosphorylation of the p85 subunit of PI3-kinase, c-Cbl, SHP-1, and Tyk2. CSF-1 induces the activity of STAT proteins, Raf-1, MEK, ERKs (extracellular signal-regulated kinases), and S6 kinase. Although it is clear that the expression of c-fos, c-myc, and cyclin D1 are enhanced by CSF-1, there is controversy as to whether the induction of c-fos and c-myc both lie downstream of a Ras-signaling pathway (145). Overexpression of oncogenically activated Raf-1 protein in BAC1.2F5 macrophages induced cellular proliferation and immediate early gene expression without induction of ERK activity, demonstrating that ERK signaling was not required for cellular proliferation (146).

The introduction of ligand-activated human c-Fms receptor can also induce proliferation when ectopically expressed in other cell types, including NIH-3T3 cells. The use of dominant negative expression vectors in NIH3T3 cells helped map an important role for a proliferative pathway involving Ras, Ets proteins, and c-Myc (147). This pathway, although necessary, was not sufficient for CSF-1 signaling to promote proliferation (147). In NIH3T3 cells overexpressing a mutant c-Fms receptor, Y809F, the defective mitogenic signaling and c-myc expression could be rescued by cyclin D1 overexpression (148). Studies by Barone and Courtneidge (149) suggested that the induction of DNA synthesis by CSF-1 in NIH3T3 cells required both a Ras-dependent pathway that induced Fos and a Ras-independent pathway that induced c-myc transcription, suggesting that Fos and c-myc may function in parallel pathways.

b. Platelet-derived growth factor (PDGF).
The presence of mitogens in platelets was first suggested by Balk (150) who observed that serum-supplemented culture medium supported the proliferation of chicken fibroblasts better than plasma-supplemented medium. Each of the three possible isoforms of this dimeric peptide, PDGF, made of A and B chains, have been isolated from natural sources including human platelets. The most prominent activity of PDGF is to stimulate proliferation. Growth factors have been categorized as either competence- or progression-inducing agents (151). Transient exposure of BALB/c3T3 cells to PDGF or fibroblast growth factor (FGF) induces a state of competence to undergo DNA synthesis, while insulin-like growth factors (IGFs), epidermal growth factor (EGF), and other substances that are present in plasma permit progression through the G₁ phase of the cell cycle (152). In BALB/c-3T3 cells, two arbitrary control points were described in the G₁ phase: the V-point, typically occurring 6 h into G₁ phase — a point at which the cell fails to progress further without the addition of essential nutrients including amino acids, and the W point, which occurs later at the G₁/S border (153).

In keeping with the known effect of PDGF early in G₁ phase, PDGF treatment of a variety of cell types was associated with an induction of cyclin D1 abundance (17, 154, 155) and CD₁K activity (17, 155). In primary tracheal myocytes, PDGF induced cyclin D1 expression with a sequential induction of pRB phosphorylation (17). The induction of S phase traversal was also blocked by immunoneutralizing antibodies to cyclin D1 in these cells (17). The induction of cyclin D1 by PDGF in CHO cells was dependent upon Ras (155).

PDGF directly affects the abundance and subcellular localization of the CKI, p27^Kip1. PDGF has been shown to reduce p27^Kip1 levels in fibroblasts (96); however, the relative abundance of p27^Kip1 in a p27^Kip1/cyclin D1/Cdk4 multiprotein complex increased with PDGF treatment of fibroblasts and was associated with increased pRB kinase activity, consistent with a model in which p27^Kip1 binding to specific partners is important in determining pRB kinase activity (156). PDGF treatment of BALB/c-3T3 cells resulted in both a reduction in p27^Kip1 protein synthesis (156) and an increase in p27^Kip1 protein degradation. In PDGF-treated BALB/c-3T3 cells, p27^Kip1 protein associated with cyclin D1 increased as the abundance of p27^Kip1 bound to cyclin E complexes decreased. The induction of p27^Kip1 degradation by PDGF in Chinese hamster embryo fibroblasts was blocked by dominant negative mutants of Ras and RhoA (155). Because the induction of cyclin D1 by PDGF was dependent only upon Ras, these findings suggest distinct Ras family members regulate the effect of PDGF on distinct components of the cell cycle-regulatory apparatus.

c. Basic FGF (bFGF).
bFGF was the first member of a family of structurally related proteins, isolated as a biochemically distinct mitogen for 3T3 fibroblasts from pituitary gland crude extracts (157). Together with the 11 other members of the FGF family (aFGF, int-2, hst, FGF-5, FGF-6, KGF, AIGF, GAF, FGF-10, FGF-11, and FGF-12), bFGF has been shown to exert mitogenic and differentiation inducing potential in a wide variety of cell types. The high-affinity FGF receptors are single-chain transmembrane proteins of the Ig-like receptor superfamily. The binding of ligand induces dimerization, tyrosine transphosphorylation, and phosphorylation and activation of a broad array of substrates including c-Src and the F-actin-binding protein cortactin, which correlates with the induction of DNA synthesis (158). The induction of SHC/Grb2/SOS complex formation with Ras and subsequently ERK activation has been linked to the induction of DNA synthesis primarily in cell types that are involved in tissue repair and revascularization such as fibroblasts, chondrocytes, myoblasts, smooth muscle, and endothelial cells (159).

FGF can induce either proliferation or differentiation depending upon the cell type. It was anticipated, therefore, that distinct changes in the abundance or composition of cell cycle-control proteins may mediate these distinct responses. To examine this possibility, a number of different groups analyzed cell cycle-control protein abundance in FGF-treated cells. In fibroblasts, FGF treatment is associated with the induction of DNA synthesis and the induction of cyclin D1 levels. In MCF7 cells, FGF induces ERK activity with an initial mitogenic and subsequently a cytostatic signal. Cyclin D1 protein levels are initially induced in bFGF-treated MCF7 cells, and then increasing p21^Cip1 levels contribute to reduced pRB phosphorylation and the inhibition of cellular proliferation (160). Differentiation of skeletal myoblasts in culture is negatively regulated by certain growth factors, including bFGF. Cyclin D1 was induced in myoblasts by bFGF, and stable expression of cyclin D1 inhibited C2C12 myoblast differentiation. Cyclin D1 inhibited myoblast differentiation by certain growth factors (161) and directly inhibited myogenic HLH proteins through an acidic region in cyclin D1’s carboxy terminus (27). Forced expression of cyclin D1, but not cyclins A, B or E, inhibited the ability of MyoD to transactivate muscle-specific genes and correlated with phosphorylation of MyoD. There are also pRB-independent mechanisms by which cyclin D1 inhibits growth factor-mediated differentiation (162). Transfection of myoblasts with CKIs p21^Cip1 and p16^Ink4a augmented muscle-specific gene expression in cells maintained in high concentrations of serum, suggesting that an active cyclin-Cdk complex suppresses MyoD function in proliferating cells (163). The direct binding and inhibition of MyoD by Cdk4, together with the ability of cyclin D1 to promote nuclear translocation of Cdk4, may contribute to the inhibition of myocyte differentiation by cyclin D1 (164). The possibility that the carboxyl termiunus of cyclin D1, which was required for inhibition of MyoD (27), may contribute to inhibition of differentiation through distinct mechanisms, including recruitment of histone acetylase-regulatory proteins (below), remains to be examined.

d. IGFs.
As noted above, IGF-I has been classified as a progression factor in the cell cycle. In density-arrested BALB/c-3T3 cells made competent with PDGF treatment, IGF-I with EGF induced G₁ phase progression and initiation of DNA synthesis. Subnanogram amounts of IGF-I with EGF allow progression to the V-point but not S phase. Once cells have progressed to the V-point within G₁, IGF-I alone is sufficient to promote progression into S-phase (165). Transcription of new mRNAs do not appear to be required for IGF-I to function as a progression factor, and IGF-I is known to stimulate posttranslational modifications of several proteins (166, 167).

IGF-I and IGF-II were isolated from human plasma as peptides with insulin- and somatomedin-like properties (168). Signaling by IGF-I requires a tetrameric type II IGF receptor. The requirement for IGF and the IGF receptor in normal growth was well illustrated by studies using targeted disruption of the IGF-I receptor. These mice were growth retarded, their size being 30% of wild-type control (169, 170). The mouse embryo fibroblasts (MEFs) derived from these animals also showed defective mitogenic capabilities, failing to grow in medium supplemented with growth factors that sustained the growth of wild-type cells (171). In the knockout MEFs, all phases of the cell cycle were delayed (171, 172), and this abnormality was reverted with reintroduction of the wild-type receptor (171). The IGF-I receptor is also required for EGF receptor function to exert its mitogenic effect (173).

IGF-I induces DNA synthesis in a broad array of cell types, and intracellular signaling involves activation of Ras via a Grb2-mSOS pathway. There is also a requirement for a docking protein IRS-1 in IGF signaling, which in turn regulates the activity of SH2 domain-containing proteins such as PI3 kinase, Grb-2, Nck, and Syp (174, 175). The induction of cell cycle progression by IGF-I treatment correlated with the induction of cyclin D1 mRNA levels in T47D breast cancer cells (176), in quiescent WI-38 cells (177), and MG63 osteosarcoma cells (178). The induction of cyclin D1 mRNA was, at least in part, regulated at the level of transcription, and the human cyclin D1 promoter linked to a luciferase reporter gene was induced by IGF-I in cultured cells (179). Recent studies suggested that IGF-I may also regulate cyclin D1 levels in vivo. During development, cyclin D1 is expressed in proliferating cells of the developing CNS (179, 180) and retina (181, 182). Protein-calorie malnutrition inhibits neonatal cerebellar growth and development, in part, through a reduction in IGF-I. Developmental delay induced by malnutrition was associated with reduced cyclin D1 protein and CD₁K activity assessed using glutathione-S-transferase-pRB as a synthetic substrate (179, 180). Nutritional replacement induced cyclin D1 protein levels and kinase activity in conjunction with a partial restoration of normal cerebellar development (179, 180).

As with PDGF, IGF also regulates the abundance of p27^Kip1. The induction of p27^Kip1 protein levels that occurs with growth factor withdrawal is likely mediated through reduced ubiquitin-mediated degradation (98). Although the reduction in p27 by interleukin-2 was blocked by rapamycin (183), rapamycin did not block insulin-mediated inhibition of p27^Kip1 levels in Swiss 3T3 cells (184), suggesting a rapamycin-independent pathway is involved in insulin-mediated DNA synthesis (184). Using a surrogate in vivo model of IGF-I deficiency, nutritional deprivation-induced cerebellar developmental arrest was associated with a reduction in p27^Kip1 levels from the same tissues in which CD₁K activity was decreased (180). These findings are consistent with the growing evidence from in vitro experiments that p21-related proteins may function to enhance pRB kinase activity (80).

e. EGF.
EGF, like IGF, functions as a progression factor. EGF is a potent mitogen for cells of ectodermal and mesodermal origin and is the prototype of a large superfamily of ligands that signal through a family of related receptors. EGF binding leads to receptor aggregation, autophosphorylation and phosphorylation of intracellular substrates including phospholipase C-, MAP kinase, and the Ras GTPase-activating protein (GAP). Binding of EGF to the EGF receptor (EGFR) can also induce heterodimerization with other members of the EGFR family, including c-ErbB-2 (185).

EGF binding to the EGFR induces expression of immediate early genes such as c-jun, c-fos, and c-myc (186). In fibroblast cells induced to proliferate by EGF, cyclin D1 levels were increased (154). The induction of cyclin D1 protein was preceded by the induction of cyclin D1 promoter activity in JEG-3 and CHO cells (187). Dominant negative mutants of Ras, mitogen-activated protein kinase (MAPK), and Ets inhibited EGF induction of the cyclin D1 promoter, mapping a likely signal transduction pathway from mitogenic signals directly to the cell cycle-regulatory apparatus (187). The EGF-related ligand TGF in esophageal and colon cancer cells (188) and activating mutants of c-ErbB-2 in MCF7 breast cancer cells (R. J. Lee, C. Albanese, G. Watanabe, G. K. I. Haines, P. M. Siegel, W. J. Muller, M. C. Hung, and R. G. Pestell, submitted) were also shown to regulate cyclin D1 abundance and promoter activity. Point mutation in the context of the native cyclin D1 promoter demonstrated that the induction of cyclin D1 by c-ErbB-2 involved the E2F- and Sp1-binding sites (R. J. Lee, C. Albanese, G. Watanabe, G. K. I. Haines, P. M. Siegel, W. J. Muller, M. C. Hung, and R. G. Pestell, submitted), and deletion of both the E2F and Sp1 sequences reduced TGF induction (188).

In PC12 cells, EGF stimulated cellular proliferation and increased the levels of several cell cycle progression factors including Cdk2, Cdk4, and cyclin B1 (190). p27^Kip1 levels were reduced by EGF in NIH3T3 cells, and dominant negative mutants of Ras blocked the EGF-induced down-regulation of p27^Kip1, consistent with dual roles of Ras in early G₁ and late G₁/S (191). As with PDGF, EGF-mediated induction of cyclin D1 and inhibition of p27^Kip1 abundance appears to be a common finding in several different cell types.

f. Nerve growth factor (NGF).
The differentiation and survival of CNS and peripheral nervous system (PNS) neurons is influenced by neurotrophic factors including NGF, brain-derived neurotrophic factor (BDNF), ciliary neurotrophic factors, and the fibroblast growth factors. One of the better characterized neurotrophins, NGF, is required for the differentiation and survival of sympathetic and sensory neurons of the PNS. NGF binds to two transmembrane proteins, p140trk and p75, on the cell surface (192). PC12 cells, which are derived from a neural crest-derived pheochromocytoma cell line, respond to NGF by withdrawing from the cell cycle, extending neurites and acquiring features of sympathetic neurons. NGF induces a sustained activation of the MAPK pathway and induces a pp90^Rsk kinase (193). Activation of the MAPK pathway by an activating Raf mutant in NIH3T3 cells overexpressing the TrkA receptor is sufficient to induce p21^Cip1 and growth arrest (194).

The response of cell cycle-regulatory proteins to NGF suggests a concurrent induction of G₁ cyclins, and CKIs may participate in coordinating the differentiation signal. NGF induces expression of cyclin D1 and the CKI p21^Cip1 (195, 196, 197). In association with these increases in cyclin D1 and p21^Cip1, the protein levels and associated activities of Cdk4, Cdk6, and cdc2 decreased. The induction of the cyclin D1 and p21^Cip1 genes by NGF requires DNA sequences in their respective promoters that include Sp1-binding sites (198).

The induction of cyclin D1 is not sufficient for the induction of neurite outgrowth in PC12 cells, however, raising the possibility that the induction of cyclin D1 by NGF subserves a distinct function (196). Because the addition of Cdk chemical inhibitors to undifferentiated PC12 cells promoted cell death, it was suggested that cyclin D1 may convey a cell survival function (199). During neuronal apoptosis, however, CD₁K activity was stimulated and cyclin D1 was required for the induction of apoptosis in R970B cells (200). The induction of apoptosis by cyclin D1 did not require p53 and was inhibited by Bcl-2 and the 21-kDa E1B protein, suggesting cyclin D1 inhibited cell survival (200). Expression of cyclin D1 in the HC11 breast cell line (201) or NIH3T3 cells (200) also induced apoptosis, and the induction of apoptosis in rat fibroblasts was observed with cyclin D1 but not with cyclin E (202). It remains possible, therefore, that cyclin D1 may either promote apoptosis or function as a sensor of impending cell death. It has been suggested that the induction of apoptosis by cyclin D1, particularly after ionizing radiation, represents a G₁/S checkpoint function of cyclin D1 (30, 203) (for a review see Ref. 4). Under these circumstances, cyclin D1 is actually required for the cell cycle arrest, and a similar function was observed for cyclin D1 in p53-mediated cell cycle arrest (204). In part, this cell cycle-arrest function of cyclin D1 may be through its ability to bind PCNA and sequester its activity in the nucleus.

Recent studies have suggested a role for Cdk5 in neurite outgrowth, and overexpression of a Cdk5 dominant negative mutant blocked neurite extension in cultured cortical neuron cultures (205). The heterodimeric partners for Cdk5, p35, and the related p39 are required for normal neurite extension in a rat hippocampal cell line (58). The role of the cyclin/Cdk complexes in NGF signaling remains to be fully explored.

The abundance of other cell cycle-control proteins are altered during NGF treatment of PC12 cells, although their role in mediating the differentiated phenotype is not understood. One of the earlier changes induced in PC12 cells treated with NGF is a decrease in cyclin F, raising the possibility that cyclin F is also involved in NGF-mediated cell cycle events during the differentiation of PC12EY cells (206). In addition the DNA binding activity of the E2F/DP proteins, which bind DNA enhancer sequences in a number of different genes involved in promoting cell cycle progression, was reduced with NGF treatment (195).

2. Serine/threonine kinase receptors.
a. TGFß.
The TGFß family of cytokines regulates diverse functions including differentiation and inhibition of cell growth (207). In addition, TGFß can induce cellular proliferation in certain cell types, such as subconjunctival and adult lung fibroblasts (208, 209). The proliferative effects are likely in part related to release of paracrine growth factors such as connective tissue growth factor (210). TGFß inhibits cellular growth in most epithelial cells and can act at both early and late points during the prereplicative G₁period (211, 212, 213). The diverse effects of TGFß are, in part, regulated by the cell type, the state of cellular differentiation, local paracrine context, and the relative abundance of the TGFß receptors. The signaling pathway induced by TGFß is dependent upon binding of ligand to the TGFß type II receptor, which forms a heteromultimeric transphosphorylated complex with the TGFß1 receptor.

Studies of pathways activated by TGFß indicate important roles for members of a multigene family related to Drosophila Mad (mothers against decapentaplegia) in intracellular signaling. In addition the CKIs appear to play an important role in the cytostatic phenotype induced by TGFß. In the developing Drosophila eye (214), signaling mediated by the TGFß-related gene decapentaplegia (dpp) was required for the synchronization of the cell cycle. dpp May affect cell cycle synchronization by promoting cell cycle progression. This synchronization is critical for the precise assembly of the Drosophila eye. More than six mammalian proteins related to Drosophila Mad, Smads, have been cloned and are implicated in transducing TGFß signaling to the nucleus, with Smad3 and Smad4 functioning as DNA sequence-specific binding proteins (215, 216, 217, 218, 219).

The effect of TGFß on components of the cell cycle-regulatory apparatus are quite complex (211). TGFß functions in part through both transcriptional effects and posttranscriptional and posttranslational mechanisms. For example, the inhibitory effect of TGFß on CCL64 cells late in G₁ occurred in the presence of RNA synthesis inhibitors (220). Unifying themes are appearing, however, indicating a critical role for the CKIs (p21^Cip1, p27^Cip1 and p15^Ink4b) in which the transcriptional induction and subnuclear localization appear to mediate their cell cycle inhibitor effects. In addition, TGFß has been shown to inhibit expression of Cdk2, cyclin D2, and cyclin A and to reduce pRB phosphorylation (221). TGFß inhibits growth in fibroblasts and Mv1Lu cells in association with a delayed reduction in Cdk4 abundance (222) while in several cell types it appears that the earliest and critical steps in the cytostatic function of TGFß is the induction of the Cdk4/6 inhibitor p15^(Ink4b/MTS2). TGFß stabilizes p15^Ink4b protein, increases p15^Ink4b-Cdk4 complexes, and inhibits cyclin D1/Cdk4 association in human mammary epithelial cells (223). The recent studies from Massague’s laboratory (92) have been particularly insightful in understanding TGFß signaling to the CKIs with their recent finding that the subcellular locations of p15^Ink4b and p27^Kip1 coordinate their inhibitory interactions with Cdk4 and Cdk2. In lung epithelial cells, treatment with TGFß led to an induction of p15^Ink4b bound to cyclin D1/Cdk4 and reduced binding of cyclin D1/Cdk4 by p27^Kip1. The displacement of p27^Kip1 led to its binding to cyclin E/Cdk2, which forms an inhibitory complex. These studies suggested the subcellular distribution, and the proteins to which p27^Kip1 was bound may be critical in mediating the cytostatic effect of TGFß (92) (Fig. 5). In cells lacking p15^Ink4b, however, TGFß is still capable of arresting cells in G₁ through a reduction in cyclin D1 and an induction of p21^Cip1 and p27^Kip1 (224).

View larger version (102K): [in this window] [in a new window]   Figure 5. A, TGFß has been shown to induce p21Cip1, p27Kip1, and p15 Ink4b in different cell types. The induction of p21Cip1 is p53 independent, as with the induction of p21Cip1 by BRCA1. The role of BRCA1 in TGFß induction of p21Cip1 is not known. It is known that cyclin D1/Cdk activity is reduced in TGFß-treated cells and that p21Cip1 can inhibit cyclin D1. B, The addition of TGFß alters the associations of p27Kip1 from cyclin D1/Cdk4 to a cyclin E/Cdk2 complex, the activity of which it inhibits. p15Ink4b replaces p27Kip1 bound to cyclin D1/Cdk4, inhibiting its activity. In this model the amount of p27Kip1 may not change in the TGFß-treated cell.

b. Activins.
Activins are members of the TGFß family encoded by two closely related genes, activin-ßA and activin-ßB. Activins inhibit cellular proliferation and induce differentiation of mesoderm, erythroid precursors, and other cell types (229, 230, 231). Activin also inhibited basal and stimulated proliferation in prostate cancer cell lines (232). In plasmacytic cells, the induction of cell cycle arrest by activin A was associated with an induction of p21^Cip1, but not the other CKIs (p27^Kip1, p16^Ink4a, or p15^Ink4b), and a suppression of cyclin D2 (233). Because pRB phosphorylation was reduced in association with the induction of p21^Cip1 in activin A-treated cells, the cell cycle arrest induced by activin was attributed to the induction of 21^Cip1 expression.

B. Seven-transmembrane receptors
1. Rhodopsin family.
a. Angiotensin II (AII).
The octapeptide AII binds to specific high-affinity receptors present in the adrenal cortex, liver epithelial cells, and in vascular smooth muscle cells where it elicits a vast array of biological effects. AII increases DNA synthesis, cell proliferation, and steroidogenesis in cultured adrenal cortical cells, both in cells derived from the adrenal fasciculata and glomerulosa cell layer (234, 235). AII functions as a growth factor in cardiac fibroblasts, myocytes, vascular smooth muscle cells, and adrenal cells (236, 237, 238). Many of the known biological actions of AII, including enhanced DNA synthesis, are mediated by stimulation of the AT₁ receptor (234, 235).

The AT₁ receptor is a member of the G protein-coupled seven-transmembrane spanning receptor family (239, 240). Binding of AII to the AT₁ receptor activates phospholipase C, which initiates a rapid release of inositol triphosphate and diacylglycerol from phosphatidylinositol 4,5-bisphosphate, causing intracellular calcium release (241). The intracellular transmission of signaling by Ca⁺² and activation of cytosolic phospholipases involves, in part, sequential activation of Ras (242, 243, 244) and thereby protein kinases, including MAPK (240, 245). Previous studies showed that AII can stimulate phosphorylation of several intracellular signaling protein kinases at tyrosine residues including the ERKs in vascular cells (246, 247) and the related stress-activated protein kinases [SAPKs or Jun N-terminal kinases (JNKs)] in hepatic cells (248). In addition, AII stimulates tyrosine phosphorylation of p44/p56^SHC (246), the Jak family proteins Jak2 and Tyk2 (249), and focal adhesion kinase (p125^FAK) in vascular smooth muscle cells (246).

In the human adrenal cell line H295R, AII (10^-6 M) stimulated G₁ phase progression within 12 h, with a maximal effect after 72 h (250). This action was preceded by the induction of cyclin D1 mRNA, the presence of nuclear cyclin D1 protein, and CD₁K activity. Acting through the AT₁ receptor, AII induced cyclin D1 promoter activity 4-fold via an enhancer sequence at -954 bp. c-Fos and c-Jun bound the cyclin D1 -954 enhancer sequence, and the abundance of c-Fos within this complex was increased by AII treatment (250). AII induced ERK activity 7-fold, and dominant negative mutants of either Ras or ERK reduced AII-stimulated cyclin D1 promoter activity. These findings suggest AII may stimulate mitogenesis by increasing CD₁K activity through a Ras/ERK/AP-1 pathway.

b. Parathyroid hormone (PTH).
Osteoblasts are the major direct cellular target of PTH action in bone. The high-affinity receptors for PTH generate a variety of intracellular signaling pathways, including changes in intracellular cAMP, inositol triphosphate, diacyl glycerol, Ca⁺², membrane potential, and pH. The effect of PTH on osteoblast proliferation and differentiation is complex, and PTH may either inhibit or induce cellular proliferation depending upon the cell system examined and upon whether the exposure to PTH is tonic or intermittent. Acute exposure to high concentrations of PTH leads to an inhibition of many osteoblastic functions. Continuous exposure to high concentration of PTH in vivo results in a progressive bone loss and osteopenia, although osteoblasts may increase in number. PTH is known to inhibit proliferation of the UMR-106 osteoblast cell lines and reduce the proportion of cells entering S-phase. Associated with these changes, PTH increased p27^Kip1, but not p21^Cip1 levels. This effect was mimicked by 8-bromo-cAMP, but not by phorbol 12-myristate 13-acetate. The protein kinase A inhibitor KT5720 abolished the PTH induction of p27^Kip1. PTH increased Cdk2-associated p27^Kip1 without affecting the levels of Cdk2 (251). Extracellular calcium inhibits parathyroid cell proliferation. In a rat epithelial parathyroid cell line, PT-r, cyclin D1 mRNA levels oscillate during the cell cycle, and increasing amounts of calcium in the incubation medium reduced the expression of rat cyclins D1 and D2 (252).

c. TSH.
TSH is a member of a dimeric family of glycoprotein hormones that contain a common - and a unique ß-subunit. The binding of TSH to its cognate G protein-coupled receptor (Gs) initiates a cascade of intracellular signaling, including the induction of cAMP, and in the human the activation of phospholipase C (253). The cAMP pathway induces DNA synthesis with kinetics comparable to the effect of TSH and is associated with the induction of immediate early genes and PCNA (254). In the rat thyroid cancer cell line FRTL5, TSH induces cellular proliferation and cell cycle progression. A FRTL5 cell line containing a dominant negative mutant of CREB (cAMP reponse element binding protein) had reduced induction of DNA synthesis in response to TSH, providing supportive evidence that the cAMP/CREB pathway was linked to DNA synthesis in thyroid cells (255). The proliferative effects of TSH are associated with an enhanced rate of G₁ phase progression and induction of cyclin D1 and cyclin E (256). The phosphatase inhibitor okadaic acid, which inhibits protein phosphatase 2A (PP2A) and protein phosphatase 1, stimulates the TSH-induced G₁-S phase transition in thyroid cells (257). In dog thyrocytes, microinjection of human GST-p16 inhibited BrdU incorporation induced by TSH (18). Cyclin D3 abundance increased in the nucleus of TSH-treated dog thyrocytes, and microinjection of immunoneutralizing cyclin D3 antibodies blocked TSH-induced BrdU uptake, suggesting a specific role for cyclin D3 in G₁ phase entry induced by TSH (258).

d. FSH.
The binding of FSH to its seven membrane-spanning receptor leads to induction of cAMP within the cell through heterodimeric G protein coupling to adenyl cyclase (259). FSH induces ovarian folliculogenesis and male spermatogenesis, and the mitogenic effect of FSH is, in part, induced through cAMP (259, 260). Many actions of FSH can be mimicked by the addition of cAMP to cultured ovarian cells. Cyclin D2 was found in FSH-responsive Sertoli cells and spermatagonia; furthermore, testicular germ cell tumors express very high levels of cyclin D2 (261). cAMP treatment of granulosa cells induced cyclin D2 mRNA with similar kinetics to the induction by FSH (261, 262). Serum addition or induction of the protein kinase C (PKC) pathway by mitogens did not induce cyclin D2 mRNA levels, suggesting activation of the cAMP pathway by FSH was critical for the induction of cyclin D2 mRNA. Homozygous deletion of the cyclin D2 gene resulted in female mice with hypoplastic granulosa cells, unresponsive to FSH, suggesting a critical role for cyclin D2 in FSH-induced cellular proliferation in vivo (261). The male cyclin D2 KO mice displayed hypoplastic testis, with normal differentiation and normal serum testosterone levels, suggesting an important role for cyclin D2 in cellular proliferation required for the formation of a critical mass of tissue during development.

C. Steroid hormones
1. Estrogens. Steroid hormones mediate diverse effects on cellular proliferation in association with modulating activity of the cyclins and the CKIs. The proliferative effects of estrogen on responsive tissues, including breast and uterus, have been well documented (for reviews see Refs. 263, 264, 265). The use of animals homozygously deleted of the ER- gene (ER KO mice) confirmed the requirement of the ER for several normal proliferative processes including normal mammary gland