HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 276, Issue 48, 45433-45442, November 30, 2001

This Article Abstract Full Text (PDF) All Versions of this Article: 276/48/45433    most recent M104752200v1 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 Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Prall, O. W. J. Articles by Sutherland, R. L. Search for Related Content PubMed PubMed Citation Articles by Prall, O. W. J. Articles by Sutherland, R. L. Social Bookmarking                      What's this?

A Low Abundance Pool of Nascent p21^WAF1/Cip1 Is Targeted by Estrogen to Activate Cyclin E·Cdk2^*

Owen W. J. Prall, Jason S. Carroll^§, and Robert L. Sutherland^¶

From the Cancer Research Program, Garvan Institute of Medical Research, St. Vincent's Hospital, Sydney, New South Wales 2010, Australia

Received for publication, May 24, 2001, and in revised form, August 9, 2001

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Estrogens regulate cell proliferation in target tissues, including breast cancer by stimulating G₁-S phase transition. Activation of cyclin E·Cdk2 through abrogation of the ability of p21^WAF1/Cip1 to bind to and inhibit cyclin-CDKs is a pivotal event in this process in MCF-7 breast cancer cells. A proposed mechanism is p21 sequestration into cyclin D1·Cdk4/6 complexes driven by estrogen-induced transcriptional activation of cyclin D1 gene expression. However, we now show that some E₂-induced cyclin E·Cdk2 activation occurs in the absence of increased cyclin D1 levels and requires decreased p21 protein synthesis. Both mechanisms operate in the absence of major changes in total p21 protein levels and instead target a low abundance subset of newly synthesized p21. E₂-induced activation of cyclin E·Cdk2 is mimicked by targeted inhibition of nascent p21 expression by antisense p21 oligonucleotides. Cyclin E·Cdk2 activation is completely inhibited by a combination of antisense cyclin D1 oligonucleotide transfection and elimination of the decrease in nascent p21 by infection with adenoviral-p21. These findings strongly support a central role for p21 in the early phase of E₂-induced mitogenesis and highlight a major functional role for newly synthesized CDK inhibitory proteins.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Numerous lines of evidence indicate that activation of cyclin E·Cdk2 is crucial to G₁ to S phase progression. Increased cyclin E expression and cyclin E·Cdk2 activity have also been implicated in the development of breast cancer. Cyclin E overexpression induces mammary gland hyperplasia and carcinoma in mice (1), and deregulated cyclin E·Cdk2 activity causes chromosome instability in human breast epithelial cells (2). Furthermore, the levels of cyclin E and cyclin E-associated kinase activity in breast cancer cell lines are commonly elevated (3, 4) and correlate with poor outcome in breast cancer patients (5, 6). Estrogen-dependent activation of the estrogen receptor (ER)¹ is essential for normal breast and uterine cell proliferation (7) and approximately two-thirds of breast tumors retain estrogen sensitivity. However, the molecular mechanisms through which estrogen stimulates the proliferation of breast cancer cells have not been fully defined. Estrogens stimulate the expression of growth factor genes, which may exert indirect effects on proliferation through autocrine and paracrine mechanisms. In addition they induce the expression of several early- and intermediate-response genes with demonstrated roles in cell cycle progression, e.g. c-fos, jun-B, c-myc, and cyclin D1 (8). More recently, estrogens have been shown to stimulate the Src/Ras/MAP kinase pathway via mechanisms that require functional ER (9, 10). Although the interactions between these components of the estrogen response remain to be fully elucidated, there is now compelling evidence that activation of cyclin E·Cdk2 plays a pivotal role in linking ER signaling to S phase entry (11-14).

In fibroblast cell lines cyclin E·Cdk2 is activated prior to S phase entry (15, 16), and is both necessary (17-19) and sufficient (20-22) for the G₁-S phase transition. At least two mechanisms for cyclin E·Cdk2-dependent S phase entry have been proposed: phosphorylation of pRB, thus relieving pRB/E2F-mediated promoter repression of genes required for S phase entry (23, 24), and a less clearly defined pRB/E2F-independent pathway (25-27). Considerable biochemical and genetic evidence indicates multiple levels of regulation of cyclin E·Cdk2, including transcriptional activation of cyclin E gene expression, association with the Cip/KIP family of CDK inhibitor proteins (p21^Cip1/WAF1, p27^KIP1, p57^KIP2) and inhibition/activation through phosphorylation/dephosphorylation of specific amino acids on Cdk2 (28, 29).

Mitogens regulate cell proliferation by stimulating progression through G₁ phase of the cell cycle, however, descriptions of the molecular mechanisms of cyclin E·Cdk2 activation employed by different mitogens are relatively limited and largely confined to fibroblast cell lines. Cyclin E·Cdk2 activation in MCF-7 cells occurs 4-6 h after estradiol (E₂) treatment, increases to 3-fold above control levels at 8 h, and increases to ~7-fold at 16 h (12). E₂ treatment fails to induce detectable changes in the total cellular pools of cyclin E, p21, or p27 during the early phase of cyclin E·Cdk2 activation (12, 13). Rather, E₂ induces only a small number of highly active cyclin E·Cdk2 complexes, which are free of the CDK inhibitor p21. Interestingly, p21 is the major cyclin E·Cdk2 inhibitory protein during early cyclin E·Cdk2 activation in this model with little contribution from p27 (12, 13). Cyclin E·Cdk2 activation coincides with a decrease in the ability of endogenous p21 to associate with cyclin E·Cdk2 in vitro, but the mechanism remains undefined (12, 13). Furthermore, the Cdk2 component of these active cyclin E·Cdk2 complexes is phosphorylated on threonine 160, a target of CDK-activating kinase (CAK). Because CAK activity is unaltered by estrogen treatment, Cdk2 phosphorylation is most likely due to the absence of p21, which inhibits phosphorylation of this residue by CAK (30). Active cyclin E·Cdk2 complexes in MCF-7 cells are associated with the pRb-related pocket protein p130 (14). p130 can bind cyclin E through an RXL motif also present in p21, suggesting that these proteins may compete for association with cyclin E·Cdk2. These observations strongly suggest that the reduction in p21-associated cyclin E·Cdk2 is critical to cyclin E·Cdk2 activation by E₂. However, there are no major changes in the levels of p21, p130, cyclin E, or Cdk2 in cyclin E complexes following E₂ treatment (12).

Increased expression of cyclin D1 has been proposed to inactivate p21 by sequestration in cyclin D1·Cdk4/6·p21 complexes, thereby making p21 unavailable for association with cyclin E·Cdk2 (13). A p21/p27-sequestering role for cyclin D1 has also been implicated in mediating the effects of transforming growth factor  and INK4 protein inhibition on G₁ phase progression (31-34). However, although ectopic expression of cyclin D1 causes increased cyclin D1·Cdk4·p21 association, it is not sufficient to mimic E₂-dependent cyclin E·Cdk2 activation quantitatively (14). Further evidence for the existence of a cyclin E·Cdk2-activating mechanism independent of cyclin D1·Cdk4 sequestration comes from our data demonstrating that c-Myc can induce activation of cyclin E·Cdk2 and cell cycle progression in antiestrogen-arrested MCF-7 cells without concurrent increases in cyclin D1·Cdk4·p21 complexes (14). We now identify a low abundance pool of p21 that readily associates with cyclin E·Cdk2 in vitro and is targeted by E₂ in vivo by dual mechanisms: sequestration by cyclin D1; and decreased p21 protein synthesis secondary to decreased p21 gene transcription. The unexpected finding that substantial increases in cyclin E·Cdk2 activity were caused by decreases in this small nascent pool of p21 may have broader relevance to activation of cyclin E·Cdk2 by other mitogens.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell Culture-- Exponentially proliferating MCF-7 cells (from Michigan Cancer Foundation, Detroit, MI) were growth-arrested by pretreatment for 48 h with 10 nM of the steroidal antiestrogen ICI 182780 (7-[9-(4,4,5,5,5-pentafluropentylsulfinyl)nonyl]estra-1,3,5-(10)-triene-3,17-diol; a gift from Dr. Alan Wakeling, Astra-Zeneca Pharmaceuticals, Macclesfield, UK) and then treated with 100 nM E₂ or ethanol vehicle control as described previously (12). In some experiments a specific Cdk2 inhibitor, roscovitine (Calbiochem-Novabiochem, Alexandria, New South Wales, Australia), was added directly to cell culture medium. Working dilutions of roscovitine were prepared in Me₂SO at 1000-fold the required final concentration in cell culture medium. Analysis of cell cycle phase distribution by DNA flow cytometry was performed as described previously (14). The derivation of MCF-7 cell lines stably transfected with the pMT-cyclin D1 plasmid was described previously (MCF7.7cycD1 (14, 35)).

Gel Filtration Chromatography, Immunoprecipitation, and Immunoblotting-- Immunoprecipitation and protein separation by gel filtration chromatography on a HiLoad 16/60 Superdex 200 column (Amersham Pharmacia Biotech, Uppsala, Sweden) were performed as previously described (14). p21-containing complexes were re-immunoprecipitated from p21 immunoprecipitates with cyclin D1 antibodies by heating at 95 °C for 2 min in 100 µl of elution buffer (25 mM Tris, pH 7.5, 100 mM NaCl, 2 mM EDTA, 0.4% SDS, 2 mM 2--mercaptoethanol), neutralization with 10 mM iodoacetamide (Sigma Chemical Co, St. Louis, MO), and dilution to 1 ml with lysis buffer A containing 2% Triton X-100. Cyclin D1 was then re-immunoprecipitated from the supernatant with rabbit cyclin D1 antisera (12). Preparation of whole cell lysates as well as immunoblotting, GST pull-downs and cyclin E·Cdk2 activity assays were as previously described (14). Monoclonal antibodies directed against the following proteins were employed: cyclin D1 (DCS-6, Novacastra Laboratories Ltd., Newcastle-upon-Tyne, UK), pRB (G3-245, PharMingen, San Diego, CA), p21 (catalog number C24420, Transduction Laboratories, Lexington, KY), p27 (catalog number K25020, Transduction Laboratories), p53 (sc-126, Santa Cruz Biotechnology Inc., Santa Cruz, CA), and GST (B-14, Santa Cruz Biotechnology). Rabbit polyclonal antibodies against cyclin E (C-19), p21 (C-19), p27 (C-19), p107 (C-18), and p130 (C-20) and their corresponding immunogenic peptides were obtained from Santa Cruz Biotechnology.

Cyclin E·Cdk2 Kinase Assay-- Lysates prepared with lysis buffer A as described above, or gel filtration fractions were precleared with protein A-Sepharose (1 h, 4 °C) and then immunoprecipitated with protein A-Sepharose conjugated to anti-cyclin E polyclonal antibodies (Santa Cruz Biotechnology Inc., C-19) for 1 h at 4 °C. One round of immunoprecipitation was sufficient to recover >95% cyclin E. Preincubation with the immunizing peptide blocked immunoprecipitation of cyclin E, associated proteins, and kinase activity (data not shown). The immunoprecipitates were then washed once with ice-cold lysis buffer A, twice with ice-cold lysis buffer A containing 1 M NaCl, once again with ice-cold lysis buffer A and then three times with ice-cold 50 mM HEPES, pH 7.5, 1 mM dithiothreitol. The kinase reaction was initiated by resuspending the beads in 30 µl of kinase buffer (50 mM HEPES, pH 7.5, 1 mM dithiothreitol, 2.5 mM EGTA, 10 mM MgCl₂, 20 µM ATP, 10 µCi of [-³²P]ATP, 0.1 mM orthovanadate, 1 mM NaF, 10 mM -glycerophosphate) containing 10 µg of histone H1 (Sigma). This amount of substrate was found to be in excess and not limiting for kinase activity. After incubation with frequent agitation for 30 min at 30 °C, the reactions were terminated by the addition of 15 µl of 3× SDS sample buffer (187 mM Tris-HCl, pH 6.8, 30% (v/v) glycerol, 6% SDS, 15% (v/v) -mercaptoethanol). The samples were then incubated at 95 °C for 2 min and separated using 12% SDS-PAGE, and the dried gel was exposed to PhosphorImager screens (Molecular Dynamics, Sunnyvale, CA). Kinase activity detected by this method was consistent with Cdk2-mediated activity, i.e. it was substantially inhibited by incubation with GST-p21, roscovitine, or olomucine but not by GST or GST-p16.

Protein Half-life Analysis-- ³⁵S labeling and pulse-chase assays were performed by incubating cells in medium containing [³⁵S]methionine-cysteine. This medium was replaced with medium containing non-radioactive methionine-cysteine after which cells were harvested at intervals from 30 to 180 min. Incorporation of ³⁵S was analyzed by immunoprecipitation, SDS-PAGE, and autoradiography. Alternatively, protein synthesis was inhibited with 20 µg/ml cycloheximide, and remaining protein was analyzed by Western blot.

Northern Blotting and Nuclear Run-off Transcription Assays-- Northern blot analysis was performed as previously described (12). Sample loading was assessed by hybridization to glyceraldehyde-3-phosphate dehydrogenase. The origin of cDNAs used was as follows: p21, Dr. Bert Vogelstein, Johns Hopkins, Baltimore, MD; cyclin E, Dr. Steven Reed, Scripps Research Institute, La Jolla, CA. RNase protection analysis of p21, p27, and cyclin E was performed with RiboQuant kits (PharMingen, San Diego, CA) according to the manufacturer's instructions. Nuclear run-off analysis was performed according to standard techniques (36). Briefly, MCF-7 nuclei were incubated with ATP, CTP, GTP, and [-³²P]UTP to allow the elongation of partial transcripts. RNA was then isolated and precipitated with 50% (w/v) trichloroacetic acid/300 mM sodium pyrophosphate and washed with 5% (w/v) trichloroacetic acid/30 mM sodium pyrophosphate until the supernatant contained no free ³²P. RNA was reprecipitated and resuspended in TES buffer (10 mM TES, pH 7.4, 10 mM EDTA, 0.2% (w/v) SDS). Slot-blot membranes were hybridized with equivalent cpm levels of [-³²P]UTP-labeled RNA for 36 h at 65 °C. Membranes were washed extensively, treated with RNase A, and analyzed by exposure to PhosphorImager screens (Molecular Dynamics). Plasmids on the slot-blot membranes were: pUC, pUC-actin, pBS, and pBS-p21 (Dr. Steven Elledge, Baylor College of Medicine, Houston, TX).

Antisense Oligonucleotide Transfection-- Oligonucleotides (stocks of 200 µM) and CellFECTIN (1 mg/ml, Life Technologies, Inc.) were mixed together at a ratio of 0.2 nmol of oligo:1 µg of CellFECTIN in serum-free RPMI 1640 for 15 min at 37 °C then added to antiestrogen-arrested cells with 5% fetal calf serum. At least 95% of MCF-7 cells that were growth-arrested by ICI 182780 were transfected with fluorescein isothiocyanate-labeled oligonucleotides using this protocol as analyzed by flow cytometry (data not shown). The oligonucleotides (GeneWorks, Australia) used were: for p21, antisense (AS) to the region around the initiating codon 5'-TCC CCA GCC GGT TCT GAC AT-3' (37), sense (SN) 5'-ATG TCA GAA CCG GCT GGG GA-3', and scrambled (SC) 5'-ACG CGT TCA CTG CCT ACG TC-3'; for cyclin D1, AS 5'-GCT GGT GTT CCA TGG CTG GG-3' (human equivalent to a rat cyclin D1 AS oligonucleotide (38)), SN 5'-CCC AGC CAT GGA ACA CCA GC-3', SC 5'-CGG TCG GAT TCG GCG TGT TG-3'. A search of available data bases confirmed that the only sequences matched were p21 and cyclin D1 for p21 AS and cyclin D1 AS, respectively. Oligonucleotides contained 5'- and 3'-phosphorothioate modifications to minimize exonucleolytic degradation. Potentially toxic effects were minimized by only briefly exposing cells to an antisense oligonucleotide-CellFECTIN mixture (see figure legends for details) and then removed. Cells were then washed once and incubated in RPMI 1640/5% fetal calf serum with 10 nM ICI 182780 ± 100 nM E₂ as appropriate.

Adenoviral Infection-- The human p21 adenovirus (Ad-p21C) was obtained from Dr. Joseph Nevins (Duke University Medical Center) and amplified in HEK293 cells. Viruses were purified by CsCl density gradient centrifugation, and titers determined by a focus forming assay using standard techniques (39). MCF-7 cells were rescued from ICI 182780-induced growth arrest with E₂ as described above and infected with Ad-p21C from 3-8 h. Protein lysates were prepared 8 h after E₂ treatment.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A Minor Proportion of p21-containing Complexes Mediate Inhibition of Cyclin E·Cdk2-- Active cyclin E·Cdk2 complexes generated by E₂ treatment of MCF-7 breast cancer cells are deficient in the CDK inhibitor p21^Cip1/WAF1, but associate with the pRb-related protein p130 (12-14). This coincides with decreased p21-dependent "inhibitory activity" toward GST·cyclin E·Cdk2 and decreased association between endogenous p21 and GST-cyclin E·Cdk2 in vitro (12, 13). A series of experiments was undertaken to define the mechanism(s) by which E₂ treatment stimulated the formation of active cyclin E·Cdk2·p130 complexes in preference to inactive cyclin E·Cdk2-p21 complexes. The efficiency with which endogenous p21 associated with recombinant GST-cyclin E·Cdk2 in vitro was investigated following E₂ treatment. The expression of total p21 in cell lysates decreased 10 h after E₂ treatment and reached 16% of control levels at 16 h (Fig. 1A), thus lagging behind cyclin E·Cdk2 activation, which was first detected at 6 h in this experimental paradigm (12, 13). However, decreased association between p21 and GST-cyclin E·Cdk2 was evident 6 h after E₂ treatment (~40% of control levels, Fig. 1A), coinciding with the earliest detectable cyclin E·Cdk2 activation and decrease in in vitro cyclin E·Cdk2 inhibitory activity (12). The extent of the association between p21 and cyclin E·Cdk2 was further reduced at later times reaching ~10% of control levels at 16 h (Fig. 1A) coincident with maximal cyclin E·Cdk2 activity (12). At all time points the magnitude of the decrease in p21/cyclin E·Cdk2 association exceeded that of the decrease in total p21 protein. This in vitro assay was likely to provide an indication of the efficiency with which endogenous p21 associated with cyclin E·Cdk2 in vivo, suggesting that E₂ abrogated the cyclin E·Cdk2 inhibitory activity of p21 by a mechanism that was independent of its effect on total p21 levels. Although an E₂-regulated activity that modified the recombinant cyclin E·Cdk2 and altered its affinity for p21 could not be ruled out by these experiments, the most likely explanation for this effect was that E₂ modified the abundance or function of the subset of p21 that was readily available for association with cyclin E·Cdk2. Alternatively, the abundance or function of a molecule that competed with p21 for association with cyclin E·Cdk2 could have been modified.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   E2-induced activation of cyclin E·Cdk2 coincides with decreased levels of inhibitory complexes containing p21. MCF-7 cells were arrested with the antiestrogen ICI 182780 (10 nM) for 2 days prior to stimulation with 100 nM E2 (+) or vehicle () at 0 h, and lysates were prepared at various times thereafter. A, lysates were incubated with recombinant GST-cyclin E·Cdk2. GST complexes were retrieved on glutathione-agarose beads, and coprecipitating p21 was detected by immunoblot. Lysates were also analyzed by immunoblotting for total p21. B, lysates were prepared after 8 h of E2 treatment and immunodepleted with antibodies to p130 or p107. Depleted lysates were analyzed for p130 and p107, and p21 was retrieved by GST-cyclin E·Cdk2 pull-down as described in A. In C, lysates were prepared after 8 h and separated on a HiLoad 16/60 gel filtration column. Fractions (concentrated by acetone precipitation) were analyzed for total p21 and p21 retrieved by GST-cyclin E·Cdk2 pull-down. The approximate molecular mass of protein complexes can be estimated by comparison with the elution volume of the protein standards shown (ferritin 440 kDa, aldolase 158 kDa, ovalbumin 43 kDa). D, p21 levels (total and that recovered by GST-cyclin E·Cdk2 pull-down) were quantitated by densitometry, and the relative pull-down efficiency for each fraction was calculated and graphed (na denotes not applicable, because p21 was not detected in these fractions). E, lysates were prepared from MCF-7 cells arrested for 48 h with ICI 182780, boiled for 5 min, separated by gel filtration, and immunoblotted for p21.

Shared cyclin E-binding motifs, RXL, in p21 and p130 (40, 41) ensure that their association with cyclin E is mutually exclusive and potentially competitive. Therefore, we investigated the possibility that E₂ may target p130 and promote association of cyclin E·Cdk2 with p130 in preference to p21. Reduced association of p21 with GST-cyclin E·Cdk2 was still evident when lysates were immunodepleted of >95% p130 (Fig. 1B). In some experimental models the p107 and p130 pocket proteins are functionally redundant (42). However, E₂ treatment reduced p21/cyclin E·Cdk2 association when lysates were immunodepleted of p107, or both p107 and p130 (Fig. 1B), excluding the possibility that p107 may compensate for the loss of p130. Together these data indicated that E₂ decreased p21/cyclin E·Cdk2 association independently of cyclin E, Cdk2, p130, and p107 and were consistent with E₂ directly targeting p21 to reduce its ability to bind to cyclin E·Cdk2.

Modifications to the quaternary structure of p21 that could alter its cyclin E·Cdk2 binding affinity were next examined by analyzing the size distribution of p21-containing protein complexes by gel filtration chromatography. Fractionation of lysates by this method indicated that there was a substantial decrease in p21-containing complexes eluting below ~100 kDa (low molecular weight p21 (LMW p21)) in lysates from cells harvested 8 h after E₂ treatment (Fig. 1C), although at this time total p21 levels were similar in control- and E₂-treated lysates (Fig. 1A). There was a small increase in p21 in fraction 6 from E₂-treated lysates, suggesting that some of the decrease in LMW p21 may be due to sequestration and redistribution to larger complexes. However, the decrease in LMW p21 could not be wholly accounted for by this mechanism, indicating that E₂ treatment also decreased synthesis and/or increased degradation of LMW p21. LMW p21 accounted for only ~15% of total p21, and thus the loss of this pool in E₂-treated cells would not be expected to alter p21 levels measured by immunoblot analysis of the total lysate.

GST pull-down experiments were used to examine the efficiency with which endogenous p21 associated with GST-cyclin E·Cdk2 in the same gel filtration fractions. These experiments identified that p21 <100 kDa (i.e. LMW p21) was most efficiently recovered by GST·cyclin E·Cdk2 pull-down (Fig. 1, C and D); i.e. the relative efficiency of p21 recovery from fractions 8-12 was ~6- to 12-fold that from fractions 5 and 6, which contained the highest levels of total p21 (Fig. 1D). This was most evident when comparisons were made between fractions 4 and 8 (or between fractions 3 and 11) from vehicle-treated lysates, which contained similar levels of total p21 (Fig. 1C), but substantially more p21 was recovered by GST-cyclin E·Cdk2 pull-down in the LMW fractions (Fig. 1D, fractions 8 and 11). Importantly, the majority of LMW p21 was not monomeric, because free p21 in lysates that were boiled eluted predominantly in fraction 12 (~30 kDa, Fig. 1E). Therefore there were at least three distinct p21-containing complexes identified by these experiments: 1) high abundance, poorly recovered by cyclin E·Cdk2 pull-down (multimeric, eluting >100 kDa); 2) low abundance, efficiently recovered by cyclin E·Cdk2 pull-down (multimeric, eluting 30-100 kDa); and 3) very low abundance, efficiently recovered by cyclin E·Cdk2 pull-down (monomeric, eluting ~30 kDa). These properties accounted for the decrease in p21/GST-cyclin E·Cdk2 association in vitro in the absence of detectable changes in total p21 (Fig. 1A); i.e. LMW p21 (primarily complex types 2 and some type 3) comprised only a small fraction of total p21 but was efficiently recovered by GST-cyclin E·Cdk2 pull-down and was markedly decreased in abundance following estrogen treatment (Fig. 1, C and D). Therefore, the estrogen-induced decrease in cyclin E·Cdk2 inhibitory activity was likely caused by the decrease in LMW p21 abundance. The two possible mechanisms for this that were mentioned above (sequestration into complexes that eluted in fraction 6, and decreased expression through either decreased synthesis and/or increased degradation) were further investigated.

Increased Cyclin D1 Synthesis Is Sufficient but Not Necessary for Full Cyclin E·Cdk2 Activation-- LMW p21 may have been sequestered by cyclin D1 and thereby unavailable for association with cyclin E·Cdk2. E₂ treatment stimulates synthesis of cyclin D1 and the formation of cyclin D1·Cdk4·p21 complexes (12, 13, 43). We have previously derived clonal MCF-7 cell lines stably transfected with a heavy metal-responsive cyclin D1 gene (MCF7.7cycD1). In these cells zinc-induced expression of cyclin D1 is accompanied by the formation of cyclin D1·Cdk4·p21 complexes, reduced levels of LMW p21, and increased cyclin E·Cdk2 activity (14). However, cyclin E·Cdk2 activation is only quantitatively mimicked when cyclin D1 is expressed ectopically at ~2 × the level produced by E₂ treatment (14), consistent with a role for an additional mechanism(s) for cyclin E·Cdk2 activation. Zinc-induced ectopic cyclin D1 expression had no effect on total p21 levels, but caused redistribution of LMW p21 into gel filtration fraction 7 (Fig. 2A). Therefore, it was likely that the increase in p21 in fraction 7 following ectopic cyclin D1 expression (Fig. 2A) and in fractions 6 and 7 following E₂ treatment (Fig. 1C) represented cyclin D1·Cdk4·p21 complex formation.

View larger version (41K): [in this window] [in a new window]   Fig. 2.   Cyclin D1 antisense oligonucleotides partially inhibit E2-induced cyclin E·Cdk2 activation. A, a clonal MCF-7 cell line (MCF-7.7cycD1) stably transfected with a metallothionein gene promoter-cyclin D1 cDNA construct (14) was growth-arrested by treatment with 10 nM ICI 182780 for 48 h, then cyclin D1 expression was induced by treatment with 50 µM ZnSO4. Protein lysates were prepared after 8 h and analyzed for p21 levels in the total lysate or in HiLoad 16/60 gel filtration fractions as described in Fig. 1C. In B, MCF-7 cells were growth-arrested by treatment with 10 nM ICI 182780 for 48 h then rescued with 100 nM E2. After 1.5 h of E2 treatment, cells were transfected with cyclin D1 oligonucleotides (SN, sense; SC, scrambled; AS, antisense) using CellFECTIN liposomal reagent. The concentration of SN and SC control oligomers corresponded to the highest concentration of AS oligomers. The transfection mixture was replaced with medium (containing 10 nM ICI 182780 ± 100 nM E2 as appropriate) 1.5 h later. Lysates were prepared after 8 h of E2 treatment and analyzed for cyclin D1 expression by immunoblot, cyclin E·Cdk2 kinase activity (using histone H1 as a substrate), and p130 phosphorylation status. Immunoreactive bands corresponding to three differentially phosphorylated forms of p130 are indicated. The uppermost band is an unidentified cross-reacting protein that is a useful loading control.

Because ectopic expression of cyclin D1 is unable to quantitatively mimic the effect of E₂ on cyclin E·Cdk2 activation (14), cyclin D1-independent mechanisms may contribute to the activation of cyclin E·Cdk2 by E₂. To directly test this hypothesis, antisense cyclin D1 oligonucleotides were employed to inhibit the increased cyclin D1 expression following E₂ treatment. E₂-induced activation of cyclin E·Cdk2 was inhibited by ~25% following the transfection of sufficient antisense cyclin D1 oligomers to abrogate the majority of the increase in cyclin D1 expression following E₂ treatment (lane 5, Fig. 2B). In vivo phosphorylation of a cyclin E·Cdk2 target, p130, was also evident when the increase in cyclin D1 expression was abrogated. Indeed, E₂-induced activation of cyclin E·Cdk2 and p130 phosphorylation were still evident following suppression of cyclin D1 expression to below control levels by this method (lane 6, Fig. 2B). These results with ectopic cyclin D1 expression and antisense cyclin D1 oligomers both indicated that cyclin E·Cdk2 activation by E₂ occurred in part through a mechanism that was independent of an increase in cyclin D1-mediated sequestration of LMW p21. Cyclin E·Cdk2 activation following expression of ectopic c-Myc in MCF-7 cells also occurs via a cyclin D1-independent mechanism (14). It was likely that a basal level of cyclin D1 was necessary for complete cyclin E·Cdk2 activation, because the most marked inhibition of cyclin D1 synthesis did prevent some cyclin E·Cdk2 activation.

LMW p21 Is Recently Synthesized-- A second potential p21-targeting mechanism was that the E₂-induced decrease in LMW p21 was caused by changes in p21 synthesis/degradation. We first investigated whether LMW p21 was recently synthesized when compared with the total pool of p21. Gel filtration chromatography was performed on lysates from cells incubated with a 15-min pulse of ³⁵S-labeled methionine-cysteine 8 h after E₂ treatment. p21 immunoprecipitates from each fraction were analyzed for total p21 protein and ³⁵S-labeled p21 (Fig. 3). Because the half-life for p21 in these cells was ~130 min (data not shown), labeling for 15 min incorporated ³⁵S predominantly into newly synthesized p21. In vehicle-treated cells total p21 eluted as a single asymmetric peak at ~120 kDa with ~15% eluting <100 kDa (Figs. 1C and 3). ³⁵S-Labeled p21 in vehicle-treated cells also eluted as a single peak but had a molecular mass of ~70 kDa, with ~40% eluting at a molecular mass of <100 kDa (Fig. 3). This indicated that nascent p21 was preferentially distributed to LMW fractions. Following E₂ treatment there was a decrease in the level of total p21 eluting <100 kDa (Figs. 1C and 3) and a more substantial decrease in ³⁵S-labeled p21 eluting <100 kDa (Fig. 3). Therefore, although the proportion of LMW p21 that was recently synthesized could not be calculated from this experiment, a likely reason for the decrease in LMW p21 following E₂ treatment was a decrease in the level of nascent p21.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Low molecular weight p21 protein complexes contain nascent p21. The experimental design is described in Fig. 1. After 8 h of E2 treatment, cells were incubated with [35S]methionine-cysteine-containing medium for 15 min. Cell lysates were prepared from E2-treated and control cells, and proteins were separated by gel filtration chromatography as described in Fig. 1C. p21 immunoprecipitates were prepared from gel filtration fractions, separated by SDS-PAGE, and transferred to nitrocellulose membranes. Membranes were sequentially analyzed for 35S-labeled p21 (by using PhosphorImager screens and autoradiography) and total p21 by immunoblot, and their relative expression was plotted.

Estrogen Inhibits p21 Synthesis-- The effect of E₂ on the level of newly synthesized p21 was then examined by incubating cells with a 15-min pulse of ³⁵S-labeled methionine-cysteine at 2-10 h after E₂ treatment. Immunoprecipitates of p21 were analyzed for ³⁵S-labeled p21 by autoradiography and immunoblotted for total p21 (Fig. 4A). E₂ treatment had no effect on the uptake and incorporation of the Tran³⁵S-label into total cellular protein or a specific control cytoskeletal protein, cortactin (Fig. 4B). Decreased incorporation of ³⁵S into p21 was evident 4 h after E₂ treatment (~68% of control) and progressively declined (~52% of control at 10 h, Fig. 4, A and C; see also Figs. 5D and 7A). The decrease in nascent p21 occurred prior to a major detectable decrease in expression of total p21 (Figs. 1A and 4A) and coincided with decreased p21/cyclin E·Cdk2 association in the GST pull-down experiments (Fig. 1A). The prominently labeled 36-kDa co-immunoprecipitating protein, which was increased following E₂ treatment (Fig. 4A), was identified by re-immunoprecipitation experiments as cyclin D1 (Fig. 4A, top right panel), which is transcriptionally up-regulated between 2 and 4 h after E₂ treatment (12, 13, 43). The reduction in ³⁵S-labeled p21 may have been due to decreased stability of newly synthesized p21 following E₂ treatment, however, the half-lives of p21 incorporating ³⁵S during a 15-min pulse were identical in E₂- and vehicle-treated cells (Fig. 4D). Therefore, the decrease in nascent p21 identified in these experiments provided strong evidence that E₂ treatment decreased the rate of p21 protein synthesis.

View larger version (28K): [in this window] [in a new window]   Fig. 4.   E2 inhibits p21 protein synthesis. The experimental design is described in Fig. 1. Cells were incubated with [35S]methionine-cysteine-containing medium for 15 min at various times after E2 treatment. A, p21 immunoprecipitates were prepared and analyzed for 35S-labeled p21 and total p21 as described in Fig. 3 (top left panel). To confirm that the 36-kDa band was cyclin D1, p21 was immunoprecipitated and divided into two aliquots. One aliquot was re-immunoprecipitated for cyclin D1, then immunoprecipitated proteins from both samples were analyzed by SDS-PAGE/autoradiography (top right panel). B, total cellular lysate and immunoprecipitates of p21 and cortactin were analyzed for incorporation of 35S label into protein as described in Fig. 3. C, the expression of 35S-labeled p21 after E2 treatment (top left panel, Fig. 4A) and total p21 (bottom panel, Fig. 4A) was quantitated by densitometry and compared with time-matched controls. The points on the graph represent the mean value from three to six independent experiments ± S.E. , 35S-labeled p21; , total p21. D, the half-life of nascent p21 was determined by pulse-chase analysis 8 h after E2 treatment. Following 15 min of 35S labeling, the medium was replaced with non-radioactive medium and lysates were prepared up to 180 min later. 35S-Labeled p21 was analyzed as described in Fig. 3, and the relative expression was plotted on a logarithmic scale. , E2-treated t1/2 ~ 130 min; , vehicle-treated. t1/2 ~ 120 min.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   E2 inhibits p21 transcription. The experimental design is described in Fig. 1. A and B, RNA was prepared and p21 and 36B4 abundance were analyzed by Northern blot. Relative p21 mRNA expression (corrected for loading variation by 36B4 mRNA expression) is presented graphically. C, nuclei were harvested at various times after E2 treatment, and nascent mRNA transcripts were elongated in the presence of [32P]UTP. RNA was isolated and used to probe slot-blot membranes with pUC and pBS plasmids containing cDNAs for actin or p21 or no cDNA (there was no hybridization to empty pUC or pBS plasmid, data not shown). Slot-blot membranes were washed extensively and analyzed on PhosphorImager screens. The average level of expression of nascent p21 transcripts from two experiments was quantitated and was graphed relative to nascent actin transcripts (empty bars, +E2; filled bars, E2). Error bars indicate the range of ratios. D, cells were treated with either roscovitine (25 µM) or vehicle (Me2SO) at 0 or 6 h after E2 treatment. After a total of 8 h of E2 treatment, cells were incubated for 15 min with [35S]methionine-cysteine-containing medium. Immunoprecipitates of p21 were prepared from cell nuclei and analyzed for 35S-labeled p21.

Northern blot analysis showed that the level of p21 mRNA decreased by ~30% 4 h after E₂ treatment, and thereafter fell to ~30% of control levels at 24 h (Fig. 5, A and B), coinciding with the decrease in p21 protein synthesis (Fig. 4C). The same result was obtained by RNase protection analysis of mRNA from an independent experiment (data not shown). Because the half-life of p21 protein in E₂-treated and control cells was ~130 min (data not shown), it was not surprising that this level of inhibition of p21 synthesis was not detected as a decrease in total p21 protein until 8 h after estrogen treatment (Fig. 1A). Nuclear run-off transcription assays indicated that this effect was likely mediated by decreased p21 transcription rather than decreased p21 mRNA stability, because the level of p21 mRNA transcripts synthesized in the nuclear run-off transcription assay fell to ~60% of control by 6 h (mean of two independent experiments, Fig. 5C). It was possible that the decrease in p21 synthesis was a consequence rather than a cause of cyclin E·Cdk2 activation. To test this, cells were treated with roscovitine, a specific chemical inhibitor of cyclin E·Cdk2. Roscovitine treatment from 0 to 8 h or 6 to 8 h produced some decrease in p21 expression but did not prevent the decrease in p21 synthesis following E₂ treatment (Fig. 5D).

Collectively, these results suggested that E₂ decreased the rate of p21 protein synthesis by inhibition of p21 gene transcription. Reduced levels of nascent p21 protein occurred independently of cyclin E·Cdk2 activation and coincided with decreased LMW p21, decreased p21/cyclin-Cdk2 association in vitro, and activation of endogenous cyclin E·Cdk2. Therefore, inhibition of p21 synthesis was likely to constitute a cyclin D1-independent mechanism for cyclin E·Cdk2 activation by estrogen.

Decreased p27 Synthesis Is Secondary to Cyclin E·Cdk2 Activation-- Following E₂ treatment, p21 contributes the majority of cyclin E·Cdk2 inhibitory activity with little contribution from p27 (12, 13). However, the cyclin E·Cdk2-binding activity of p27 is also down-regulated by E₂ (11, 12), and this process may contribute to the maintenance of cyclin E·Cdk2 activation. Therefore, we investigated the mechanism for this decrease in p27 inhibitory activity (p57^Kip2 expression could not be detected in MCF-7 cells by RNase protection analysis and was therefore unlikely to play a major role in E₂ regulation of cell cycle progression). Similar to p21, the pool of p27 that was highly efficiently recovered by GST·cyclin E·Cdk2 pull-down was largely confined to relatively LMW p27-containing complexes, which were decreased in abundance following E₂ treatment (Fig. 6A). In contrast to p21, there was no evidence for sequestration of LMW p27 and elution at a different molecular weight. This may be because the increase in association between cyclin D1 and p27 following E₂ treatment is substantially less than that between cyclin D1 and p21 (12). Instead, decreased expression of LMW p27 was likely to be due to changes in p27 protein synthesis and/or stability. Surprisingly, p27 mRNA expression and protein synthesis rates were slightly increased (Fig. 6B and data not shown) following E₂ treatment, however, p27 protein stability was decreased (p27 protein half-life in antiestrogen-arrested control cells was >180 min and ~110 min after 10 h of E₂ treatment (Fig. 6C)). Although these assays analyzed total p27 rather than LMW p27 specifically, they suggested that the decrease in LMW p27 may have been secondary to decreased p27 stability. Proteasome-mediated degradation of p27 can be activated by cyclin E·Cdk2-dependent phosphorylation of p27 (44). Indeed, the cyclin E·Cdk2 inhibitor roscovitine was sufficient to completely prevent the E₂-induced decrease in p27 recovered by GST·cyclin E·Cdk2 pull-down (Fig. 6D). Most importantly, this result indicated that, in contrast to the decrease in availability of p21 for association with cyclin E·Cdk2, decreased p27 availability was likely to be a consequence, and not a cause, of E₂-induced cyclin E·Cdk2 activation. E₂ treatment stimulated a small increase in cyclin E synthesis, which was blocked by roscovitine, indicating that this was also secondary to cyclin E·Cdk2 activation (data not shown). Therefore, the order of events following E₂ treatment were: 1) decreased p21 synthesis and increased cyclin D1 synthesis, which both reduced the abundance of LMW p21; 2) cyclin E·Cdk2 activation; and 3) increased p27 degradation and cyclin E synthesis.

View larger version (14K): [in this window] [in a new window]   Fig. 6.   Decreased levels of inhibitory complexes of p27 are a consequence of cyclin E·Cdk2 activation. The experimental design is described in Fig. 1. A, lysates were prepared after 8 h of E2 treatment and separated on a HiLoad 16/60 gel filtration column. Fractions were analyzed for total p27 and p27 retrieved by GST·cyclin E·Cdk2 pull-down. B, cells were incubated with [35S]methionine-cysteine-containing medium for 15 min at various time points after E2 treatment. p27 immunoprecipitates were prepared and analyzed for 35S-labeled p27 and total p27. The expression of 35S-labeled p27, total p27, and p27 mRNA expression (analyzed by RNase protection and corrected for loading variation by glyceraldehyde-3-phosphate dehydrogenase mRNA expression) after E2 treatment (compared with time-matched controls) are presented. C, cells were treated with cycloheximide after 10 h of E2 treatment, and p27 levels were analyzed by Western blot. The percentage of p27 remaining is presented. , E2-treated; , vehicle-treated. D, cells were treated with 25 µM roscovitine or Me2SO vehicle 6 h after E2 treatment, and lysates were prepared at 8 h and analyzed for total p27 expression and p27 retrieved by GST-cyclin E·Cdk2 pull-down.

Inhibition of p21 Synthesis Is Sufficient and Necessary for Cyclin E·Cdk2 Activation-- Experiments were performed to determine whether a decrease in p21 synthesis, but without a major decrease in total p21 protein expression, was sufficient to activate cyclin E·Cdk2. Transfection of p21 antisense phosphorothioate oligonucleotides into antiestrogen-arrested cells produced concentration-dependent decreases in nascent p21 and activation of cyclin E·Cdk2 (Fig. 7A). Concentrations of p21 antisense oligomers that produced an approximately equal decrease in newly synthesized p21 to that following E₂ treatment activated cyclin E·Cdk2 but not to the same extent as E₂ (compare lanes 4 and 5 with 1). More pronounced inhibition of p21 synthesis (lane 6) matched cyclin E·Cdk2 activation by E₂. These concentrations of p21 antisense oligonucleotides did not produce observable decreases in total p21 after 8 h, although they did at later times (data not shown). Cyclin E·Cdk2 activation was accompanied by p130 phosphorylation in vivo. Although antisense-mediated inhibition of p21 synthesis is unlikely to completely mimic E₂ treatment, these experiments provided strong evidence that targeted inhibition of nascent p21 by E₂ can induce substantial activation of cyclin E·Cdk2.

View larger version (33K): [in this window] [in a new window]   Fig. 7.   Decreased p21 synthesis and increased cyclin D1 synthesis are required together for E2-induced cyclin E·Cdk2 activation. A, MCF-7 cells were growth-arrested by treatment with 10 nM ICI 182780 for 48 h then transfected with p21 oligonucleotides (SN, sense; SC, scrambled; AS, antisense) using CellFECTIN liposomal reagent. The concentration of SN and SC control oligomers corresponded to the highest concentration of AS oligomers. After 4 h the transfection mixture was replaced with medium containing 10 nM ICI 182780 ± 100 nM E2. Lysates were prepared 8 h later and analyzed by immunoblot for p21 expression, nascent p21 (as described in Fig. 4), cyclin E·Cdk2 kinase activity (using histone H1 as a substrate) and p130 phosphorylation status. B, MCF-7 cells were growth-arrested by treatment with 10 nM ICI 182780 for 48 h then treated with 100 nM E2. After 3 h, adenovirus carrying the p21C mutant cDNA or control virus was added to the medium. The m.o.i. of control virus corresponded to the highest m.o.i. of the p21C mutant virus. Cells were harvested 8 h after E2 treatment and analyzed for p21 expression and cyclin E·Cdk2 kinase activity (H1). Relative cyclin E·Cdk2 kinase activity is shown. C, cells were transfected with cyclin D1 oligonucleotides from 1.5 to 3 h after E2 treatment (SN, sense; AS, antisense), then infected with adenovirus carrying p21C mutant cDNA or control virus. Cells were harvested 8 h after E2 treatment and analyzed by immunoblot for expression of cyclin D1, p21 and p130, and for cyclin E·Cdk2 kinase activity (H1).

We next tested whether inhibition of p21 synthesis was required for cyclin E·Cdk2 activation by E₂. An adenovirus expressing human p21 (Ad-p21C) was used to quantitatively replace the E₂-induced decrease in endogenous p21. This mutant p21 lacks the C-terminal 21 amino acids but preserves the CDK-binding domain and N-terminal cyclin-binding domain sufficient for full inhibition of CDK activity (40, 45). Identification of endogenous- and adenoviral-synthesized p21 was facilitated by their differential electrophoretic mobility (p21C migrates at ~19 kDa). Cells were infected with various m.o.i. of Ad-p21C 3 h after E₂ treatment, and cyclin E·Cdk2 activity was analyzed at 8 h. Infection efficiency was >95% for all m.o.i. used when adenovirus carrying green fluorescence protein was employed in similar experiments (data not shown). As expected, increased synthesis of p21C produced concentration-dependent inhibition of cyclin E·Cdk2 activity (Fig. 7B). Most importantly, relatively small changes in p21 expression markedly inhibited cyclin E·Cdk2 activity, because increases of only 10 and 17% inhibited the increase in cyclin E·Cdk2 activity by 42 and 58%, respectively (lanes 7 and 8, Fig. 7B). These small increases in p21 quantitatively replaced the decrease in total p21 observed 8 h after E₂ treatment (as estimated by gel filtration, Fig. 1C), indicating that E₂ treatment induced some cyclin E·Cdk2 activation independently of decreased p21 expression.

Taken together, these data indicated that decreased p21 protein synthesis, prior to a detectable decrease in total p21 protein expression, was sufficient for cyclin E·Cdk2 activation following E₂ treatment and was necessary for complete cyclin E·Cdk2 activation. However, because manipulation of p21 expression did not quantitatively mimic cyclin E·Cdk2 activation following E₂ treatment, E₂ must also activate cyclin E·Cdk2 by a p21-independent pathway.

Full Activation of Cyclin E·Cdk2 Requires Inhibition of p21 Synthesis and Increased Cyclin D1 Synthesis-- Previous experiments in which we increased or decreased the expression of cyclin D1 (14) or p21 (Figs. 2B and 7B) indicated the presence of cyclin D1-independent and p21-independent pathways for E₂-induced cyclin E·Cdk2 activation. Therefore we tested whether inhibition of cyclin D1 synthesis and a concurrent increase in p21 synthesis would quantitatively inhibit the E₂-induced increase in cyclin E·Cdk2 activity. Cells were therefore treated with E₂ and then transfected with antisense cyclin D1 oligomers and infected with Ad-p21C. Either treatment alone inhibited the E₂-induced increase in cyclin E·Cdk2 activity by ~50%, and together they eliminated any increase in activity (Fig. 7C). Therefore, estrogen appears to utilize both molecular events to induce full activation of cyclin E·Cdk2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

E₂-dependent Transcriptional Regulation of p21 and Cyclin D1 Are Both Required for Full Activation of Cyclin E·Cdk2-- The proliferative effect of estrogens on breast epithelial cells is important for both breast development and oncogenesis. A critical target of E₂ is the cyclin E·Cdk2 complex, which stimulates G₁-S phase transition. Cyclin E·Cdk2 has been well studied in normal fibroblasts where it is also required for proliferation and is activated following exposure to serum or peptide growth factors by increased cyclin E expression (15, 16). However, in contrast to fibroblasts, estrogen-induced cell cycle progression in the MCF-7 breast cancer cell line is not associated with large increases in cyclin E expression (11-13), and increased cyclin E synthesis is secondary to cyclin E·Cdk2 activation.² Alternative mechanisms for cyclin E·Cdk2 activation include activation by the Cdc25A phosphatase and Cdk2 phosphorylation/dephosphorylation by CAK, and negative regulation of CDK inhibitor proteins. A recent paper has addressed the complex relationship between cyclin E·Cdk2 and Cdc25A during E₂-induced cell cycle reentry (46). E₂ induces the expression of Cdc25A mRNA and protein, and full Cdk2 activity is dependent upon Cdc25A expression and activation (46). However, Cdc25A activation is in turn dependent upon cyclin E·Cdk2 activity. Therefore, Cdc25A maintains and amplifies cyclin E·Cdk2 activity (and cyclin E·Cdk2 amplifies Cdc25A activity in a positive feedback loop) but is not likely to be an initiator of cyclin E·Cdk2 activation. Similarly, E₂ treatment stimulates CAK-dependent phosphorylation of Cdk2, but this is secondary to cyclin E·Cdk2 activation (12, 13), and therefore this mechanism is also likely to maintain, but not initiate, the cyclin E·Cdk2 activity.

In contrast, we now show that E₂-induced activation of cyclin E·Cdk2 is dependent upon the inhibition of transcription of the CDK inhibitor p21^WAF1/Cip1, but not vice versa, and therefore inhibition of p21 transcription is a strong candidate for the mechanism that initiates cyclin E·Cdk2 activation. We and others (47, 48) have recently proposed that p21 is essential for antiestrogen-mediated inhibition of cyclin E·Cdk2 activity. p21 is also required to maintain a quiescent, G₀ state in normal foreskin fibroblasts (49). The major cyclin E·Cdk2 inhibitory protein during early cyclin E·Cdk2 activation in the MCF-7 breast cancer cell model is p21 (and not p27 or p57), association between p21 and cyclin E·Cdk2 is inhibited by estrogen treatment, and active cyclin E·Cdk2 complexes are deficient in p21 (12, 13). Our data demonstrate that p21 exists in three distinct protein complexes in MCF-7 cells, which were differentiated by complex size, i.e. <30, 30-100, and >100 kDa. In MCF-7 cell lysates p21 was confined predominantly to the two larger complexes, of which the 30-100 kDa forms were the least abundant but made the greatest contribution to the p21 available for association with GST-cyclin E·Cdk2. E₂ treatment targeted this LMW p21 pool and resulted in substantial changes in cyclin E·Cdk2 activity without major changes in the abundance of total cellular p21. Because LMW p21 was primarily a nascent protein, it could also be preferentially targeted by experimental manipulation. In such experiments, the expression of p21 was manipulated quantitatively to mimic the effects of E₂ treatment. Treatment with p21 antisense oligonucleotides was sufficient to activate cyclin E·Cdk2, and reversing the effect of E₂ on p21 by infection with a p21-expressing adenovirus prevented E₂-induced cyclin E·Cdk2 activation. We and others (47, 48) have performed what is conceptually the reverse experiment and shown that antisense to p21 can inhibit the growth-arrest induced by potent specific antiestrogens. LMW p21 was targeted in vivo by two E₂-dependent mechanisms: sequestration of p21 through increased cyclin D1 expression and decreased p21 protein synthesis. The early activation of cyclin E·Cdk2 could only be partially prevented by quantitatively reversing either of these events but was completely prevented by reversing both events simultaneously. Therefore, it is necessary to extend the current model of E₂-induced cyclin E·Cdk2 activation, which proposes that p21 is sequestered away from cyclin E·Cdk2 by increased cyclin D1 expression (13, 14), to include decreased nascent p21 via inhibition of p21 transcription and protein synthesis. Our results indicate that this mechanism is of equivalent importance to that of increased cyclin D1 expression and represents a novel cyclin E·Cdk2-activating mechanism. This mechanism may have been overlooked previously, because, in contrast to cyclin D1, changes in p21 mRNA expression are not rapidly reflected in p21 protein levels due to the relatively long half-life of p21 protein.

We focused on initial phases of cyclin E·Cdk2 activation, but our results also indicate that cyclin E transcription and p27 degradation are likely to contribute to subsequent increases in activity. These molecules were both downstream of initial cyclin E·Cdk2 activation in our model but may then serve to further activate cyclin E·Cdk2 in positive feedback loops. Cyclin E transcription can be E2F-dependent and initiated by cyclin E·Cdk2 phosphorylation of pRB (50, 51), whereas cyclin E·Cdk2 can phosphorylate p27 and stimulate its degradation (44, 52). Coupling of p21 synthesis to p27 expression has been recently reported in astrocytes in which antisense p21 oligonucleotides prevented interleukin-4-mediated inhibition of Cdk2 activity and elevation of p27 expression (53), thus providing precedence for the proposed model of cyclin E·Cdk2 activation induced by E₂.

The Functional Subset of p21 Is Newly Synthesized-- The two smallest p21-containing protein complexes (30-100 and <30 kDa) contained p21 that was efficiently recovered by GST-cyclin E·Cdk2 pull-down assays. Monomeric p21 was confined to <30 kDa complexes that were in very low abundance in vivo, and the 30-100 kDa p21 complexes contained newly synthesized p21. This suggests that p21 molecules pass through at least three distinct phases following synthesis: nascent p21 molecules rapidly associate with unidentified proteins (forming the 30-100 kDa complexes) and are available for binding to and inhibiting cyclin·CDK complexes; as cyclin·CDK complexes become available, their association with p21 results in the accumulation of the >100 kDa complexes. Cyclin·CDK·p21 are relatively stable, because they can be disrupted by boiling (12) but not by antibodies raised against the CDK-binding site on p21 (54). This stable interaction would account for the relatively low availability of p21 from the >100 kDa complexes for binding to cyclin E·Cdk2 in vitro.

We have previously reported that only a small proportion of the total cyclin E is associated with active Cdk2 in breast cancer cell lines and is also associated with p130 (4, 12). This pool of "active" cyclin E, like the inhibitory complexes of p21, is also recently synthesized.³ Association between active cyclin E and p130 in our model (14) is likely to occur as a consequence of the decrease in available p21. p130 and p21 associate with cyclin E through their RXL motifs. Prior to E₂ treatment the relatively high rate of p21 protein synthesis ensures that cyclin E associates with, and is inhibited by, nascent p21. Following E₂ treatment the decreased abundance of LMW p21 permits the activation of recently synthesized cyclin E and allows association with another RXL protein. The high levels of p130 in G₀/G₁ phase-arrested MCF-7 cells (47) is likely to ensure association of cyclin E·Cdk2 with p130 rather than other RXL-containing proteins, e.g. p27, p57, p107, E2F1-3, and Cdc25A (40, 55, 56). We consistently observe association between p130 and active cyclin E·Cdk2 following estrogen treatment, or ectopic expression of c-Myc or cyclin D1 expression in MCF-7 cells (14). Because p130 can inhibit cyclin E·Cdk2 activity (57, 58), the mechanism that permits activation of p130-associated cyclin E·Cdk2 in MCF-7 cells remains unknown.

In summary, we present data identifying inhibition of nascent p21 as a novel mechanism for cyclin E·Cdk2 activation following estrogen stimulation. Recent evidence demonstrating that p21 gene expression is repressed by c-Myc (59, 60) provides a potential mechanism through which estrogen and other mitogens may repress p21 transcription and activate cyclin E·Cdk2. Thus this study provides a previously undefined mechanism of action of estrogen, which may have broad relevance to cell cycle control by diverse mitogenic hormones and growth factors.

	ACKNOWLEDGEMENTS

We thank Ruth Lyons, Ly Q. Nguyen, and Drs. Eileen M. Rogan and Lee Carpenter for assistance with some experiments and Dr. Elizabeth A. Musgrove for critically reviewing this manuscript. Dr. Joseph R. Nevins generously provided the adenoviral p21, and Dr. Boris Sarcevic provided the recombinant GST·cyclin E·Cdk2 complexes.

	FOOTNOTES

^* This work was supported in part by grants from the National Health and Medical Research Council of Australia (NHMRC) and the New South Wales Cancer Council.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Supported by an NHMRC Medical Postgraduate Research Scholarship and a Leo & Jenny Cancer Foundation Cancer Research Grant.

^§ Supported by an Australian Postgraduate Award.

^¶ To whom correspondence should be addressed: Cancer Research Program, Garvan Institute of Medical Research, 384 Victoria St, Darlinghurst, Sydney, New South Wales 2010, Australia. Tel.: 61-2-9295-8322; Fax: 61-2-9295-8321; E-mail: r.sutherland@garvan.org.au.

Published, JBC Papers in Press, October 1, 2001, DOI 10.1074/jbc.M104752200

² O. W. J. Prall, unpublished data.

³ O. W. J. Prall, unpublished observations.

	ABBREVIATIONS

The abbreviations used are: ER, estrogen receptor; E₂, estradiol; CDK, cyclin-dependent kinase; CAK, CDK-activating kinase; ICI 182780, 7-[9-(4,4,5,5,5-pentafluropentylsulfinyl)nonyl]estra-1,3,5-(10)-triene-3,17-diol; GST, glutathione S-transferase; PAGE, polyacrylamide gel electrophoresis; TES, 2-{[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]amino}ethanesulfonic acid; AS, antisense; SN, sense; SC, scrambled; Ad, adenovirus; LMW, low molecular weight; m.o.i., multiplicity of infection.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES