p53 and p21 Form an Inducible Barrier that Protects Cells against Cyclin E-cdk2 Deregulation

Current Biology, Vol. 12, 1817–1827, October 29, 2002, 2002 Elsevier Science Ltd. All rights reserved.

PII S0960-9822(02)01225-3

p53 and p21 Form an Inducible Barrier that Protects Cells against Cyclin E-cdk2 Deregulation Alex C. Minella,1,2 Jherek Swanger,1 Eileen Bryant,1 Markus Welcker,1 Harry Hwang,1 and Bruce E. Clurman1,2,3 1 Divisions of Clinical Research and Human Biology Fred Hutchinson Cancer Research Center 2 Department of Medicine University of Washington School of Medicine Seattle, Washington 98109

Summary Background: Cyclin E, in conjunction with its catalytic partner cdk2, is rate limiting for entry into the S phase of the cell cycle. Cancer cells frequently contain mutations within the cyclin D-Retinoblastoma protein pathway that lead to inappropriate cyclin E-cdk2 activation. Although deregulated cyclin E-cdk2 activity is believed to directly contribute to the neoplastic progression of these cancers, the mechanism of cyclin E-induced neoplasia is unknown. Results: We studied the consequences of deregulated cyclin E expression in primary cells and found that cyclin E initiated a p53-dependent response that prevented excess cdk2 activity by inducing expression of the p21Cip1 cdk inhibitor. The increased p53 activity was not associated with increased expression of the p14ARF tumor suppressor. Instead, cyclin E led to increased p53 serine15 phosphorylation that was sensitive to inhibitors of the ATM/ATR family. When either p53 or p21cip1 was rendered nonfunctional, then the excess cyclin E became catalytically active and caused defects in S phase progression, increased ploidy, and genetic instability. Conclusions: We conclude that p53 and p21 form an inducible barrier that protects cells against the deleterious consequences of cyclin E-cdk2 deregulation. A response that restrains cyclin E deregulation is likely to be a general protective mechanism against neoplastic transformation. Loss of this response may thus be required before deregulated cyclin E can become fully oncogenic in cancer cells. Furthermore, the combination of excess cyclin E and p53 loss may be particularly genotoxic, because cells cannot appropriately respond to the cell cycle anomalies caused by excess cyclin E-cdk2 activity. Introduction Cell cycle transitions are catalyzed by the cyclin-dependent kinases (cdks). Cdks are composed of a catalytic subunit (the cdk) and a regulatory subunit (a cyclin), and cyclin binding activates the cdk [1]. Progression through the G1 phase of the mammalian cell cycle is driven by the D-type cyclins, which activate the cdk4 and cdk6 kinases, and cyclin E, which activates cdk2. Cyclin 3

Correspondence: [email protected]

E-cdk2 is rate limiting and is required for G1 progression, and cyclin E abundance oscillates during the cell cycle [2–5]. A pulse of cyclin E transcription in late G1 is regulated by E2F activity, and this in part determines cyclin E periodicity [6]. Cyclin E is an unstable protein, and ubiquitin-mediated proteolysis also determines its abundance. Cyclin E turnover is regulated by site-specific cyclin E phosphorylations, by binding to cdk2, and by the Fbw7/SCF ubiquitin ligase [7–10]. The Retinoblastoma protein (Rb) is phosphorylated by cyclin D- and cyclin E-associated cdks during G1 progression. Progressive phosphorylation inactivates Rb, and this promotes S phase entry. Most cancer cells contain mutations in the Rb-cyclin D pathway, although the precise mechanisms through which these mutations promote tumorigenesis are largely unknown [11]. These mutations include gain-of-function mutations of cyclin D and cyclin E and the loss of function of two classes of cdk inhibitor proteins that inhibit Rb phosphorylation (the Cip/Kip and INK4 families), as well as Rb itself. Virtually all of these mutations may lead to inappropriate cyclin E-cdk2 activity, which may then contribute directly to cell transformation. In established cell lines, ectopic cyclin E expression causes faster progression through G1 phase, prolongs the duration of S phase, and leads to decreased mitogen requirements, altered cell size, and increased cell saturation density [4, 5, 12, 13]. Ectopic cyclin E expression in established cell lines has also been associated with genetic instability and abnormal centrosome duplication [13, 14]. These activities have been suggested to underlie the association between cyclin E deregulation and the development of cancer. Deregulated cyclin E expression has also been linked to tumorigenesis in vivo. Cyclin E abnormalities have been described in many human cancers, and cyclin E status has been correlated with poor clinical and/or histologic indices [15–18]. Mice with transgenes directing cyclin E expression in the mammary epithelium develop breast carcinomas [19]. However, these tumors developed in only 12% of animals with cyclin E transgenes and after a long latent period. Similarly, transgenic mice with cyclin E expression directed to the thymus did not spontaneously develop lymphomas, but they exhibited increased lymphomagenesis induced by carcinogens [20]. Thus, cyclin E deregulation was not sufficient for the development of these cancers, but it may have directly contributed to the multistep process of tumorigenesis. In order to further characterize the consequences of cyclin E deregulation in normal cells, we enforced cyclin E expression in primary human and murine fibroblasts. Unexpectedly, we found that prolonged expression of cyclin E in primary fibroblasts led to decreased cdk2 activity and inhibited, rather than promoted, cell cycle progression. The decreased cdk2 activity was associated with increased p21cip1 expression, and this required intact p53 function. Furthermore, both p53 and p21 were required to suppress ectopic cyclin E in mouse embryo fibroblasts (MEFs). The activation of p53 by

Current Biology 1818

Figure 1. Deregulated Cyclin E Expression in Primary Human Fibroblasts Leads to cdk2 Inhibition and Increased p21cip Expression WI-38 cells were transduced with retroviral vectors expressing cyclin E, a hyperactive cyclin E mutant, cyclin E (T380A), or control vector. Cells were harvested 2 weeks after transduction and selection. Western analysis of cyclin E and p21 abundance, and total cdk2 activity measured in an anti-cdk2 immunoprecipitate kinase assay, are shown.

cyclin E was associated with increased p53 serine 15 phosphorylation that was inhibited by caffeine or wortmannin, suggesting that it resulted from the ATM/ATR family of kinases. Moreover, when p53 function was disabled, excess cyclin E-cdk2 activity caused aberrant cell cycle regulation and genetic instability. These data indicate that excess cyclin E in primary cells initiates a homeostatic p53-dependent program that limits cdk2 activity via induction of p21. This response protects normal cells from inappropriate cyclin E-cdk2 activity. Results Deregulated Cyclin E Expression in Primary Human Fibroblasts Causes the Induction of p53 and p21 We constructed retroviral vectors encoding either cyclin E or cyclin E (T380A), a mutant that is hyperactive because it is resistant to ubiquitin-mediated proteolysis [7, 10]. Transduction of WI-38 primary human diploid fibroblasts with either cyclin E or cyclin E (T380A) vectors led to significant cyclin E overexpression that exceeded the amount of endogenous cdk2 (Figures 1 and S1A, see the Supplementary Material available with this article online). Although cyclin E normally activates cdk2, we paradoxically found that WI-38 cells with prolonged cyclin E expression contained substantially reduced cdk2 kinase activity (Figure 1). Concomitant with the inactivation of cdk2, we observed a large increase in the abundance of the endogenous p21cip1 cdk inhibitor

(Figure 1). The increased p21 abundance was due to increased p21 mRNA, and not altered p21 turnover (Figures S1B and S1C). In contrast, there was no increased abundance of the related p27Kip1 cdk inhibitor (not shown). Thus, deregulated cyclin E expression in primary human fibroblasts led to increased p21cip1 abundance and cdk2 inactivation. Because p21 is a key transcriptional target of p53, we next examined if the increased p21 abundance resulted from increased p53 activity. Figure 2A shows that cyclin E overexpression in primary human foreskin fibroblasts (HFFs) caused increased p53 protein abundance that paralleled p21 abundance. To determine if increased p53 activity was required for the p21 induction, we first transduced cells with a vector encoding the human papilloma virus E6 protein (HPV-E6), which abrogates p53 function by catalyzing p53 degradation [21]. Thus, HPVE6-expressing cells are effectively p53 null, although E6 may have activities in addition to p53 turnover. We found that expression of HPV-E6 completely inhibited the induction of p21 by cyclin E in WS1 human diploid fibroblasts, HFFs, and WI-38 cells (Figures 2B and 2C and not shown). Furthermore, HPV-E6 expression reversed the inhibition of total cdk2 and cyclin A-cdk2 activity caused by cyclin E and allowed the cyclin E to form active kinase complexes in all three primary fibroblast isolates (Figure 2C and not shown). The loss of p21 expression did not significantly alter the amount of cdk2 that coprecipitated with cyclin E, but it instead resulted in a larger proportion of active cyclin E-cdk2 complexes (Figure 2D). These data are consistent with the idea that the induction of p21 was due to increased p53 activity. The suppression of cdk2 activity resulting from cyclin E expression was also p53 dependent, because, in the absence of p53, this response no longer occurred. Both p21 and p53 Are Required for the Suppression of Ectopic Cyclin E in Primary Mouse Fibroblasts Because HPV-E6 may have pleiotropic effects in addition to p53 degradation, we also used MEFs with targeted deletions in either p21 or p53 to directly test the role of these proteins in suppressing deregulated cyclin E activity [22, 23]. We found that, whereas ectopic cyclin E had minimal kinase activity in normal MEFs, the exogenous cyclin E was fully active in cells lacking either p21 or p53 (Figures 3A and 3B). Similarly, cyclin E overexpression caused profound cell cycle anomalies in p21 null and p53 null MEFs compared with normal cells (Figures 3A and 3B). Enforced cyclin E expression only minimally altered the cell cycle profile of asynchronous normal MEFs (Figure 3A). Like other workers, we consistently detected a small fraction of 8N cells in normal MEFs [24]. In contrast, cyclin E expression in either p21 or p53 null MEFs had two major effects on the cell cycle: accumulation of cells in S phase and an increased proportion of cells with higher ploidy, with cells reaching 16N DNA content (Figures 3A and 3B). Thus, in primary murine fibroblasts, the suppression of ectopic cyclin E activity required p21 and p53 function. In contrast, MEFs with deletions of either p19ARF or the p27Kip1 cdk inhibitor resembled wild-type cells and were quite resistant to excess cyclin E (Figure 3C and not shown).

p53-Dependent Cyclin E Suppression 1819

Figure 2. The Induction of p21 Expression in Primary Human Fibroblasts Is Dependent upon p53 Function (A) Human foreskin fibroblasts (HFF) were lysed 2 weeks after transduction and selection with the indicated vectors. Western analysis of p53 and p21 expression is shown. (B) WS1 cells were transduced with either control virus (LXSH) or a vector expressing HPV-E6 prior to transduction with the indicated vectors. Expression of cyclin E and p21 is shown. (C) HFFs transduced with either control or E6 vectors were then transduced with the indicated vectors. Western analysis of cyclin E and kinase assays for cyclin E-cdk2, cyclin A-cdk2, or total cdk 2 activity is shown. (D) HFFs were transduced with the indicated vectors. Lysates were immunoprecipitated with anti-cyclin E antibody after normalizing total protein for total cyclin E protein. The amount of cdk2 (p-cdk2 indicates active threonine 160 phosphorylated cdk2) and p21 bound to cyclin E is shown.

The Induction of p53 by Cyclin E Is Associated with Increased p53 Serine 15 Phosphorylation, but Not Increased p14ARF Expression Enforced expression of dominant oncogenes such as myc and ras in MEFs also initiates a response involving p53 and p21 induction, followed by premature senescence [11]. This response is mediated by the 19ARF protein (p14ARF in human cells), which increases in abundance and prevents p53 degradation, in addition to other activities. p14ARF is an E2F target gene (in at least some cell types), and the increased ARF expression under these conditions may be due to increased E2F activity [25]. Because cyclin E overexpression may also lead to increased E2F activity through phosphorylation of the Retinoblastoma protein, we examined p14ARF expression in HFFs and WI-38 cells transduced by cyclin E vectors. In agreement with a recent report, we could not detect p14ARF protein in primary human fibroblasts, either in control or cyclin E-transduced cells (not shown) [26]. We then used a sensitive real-time PCR assay to examine p14ARF mRNA abundance. WI-38 and HFF cells contained approximately 20% of the amount of p14ARF mRNA detected in HeLa cells, which are known to overexpress p14ARF (Figure 4A). No increased p14ARF mRNA was observed in cells expressing cyclin E, although the assay was sensitive enough to detect 2-fold changes in p14ARF expression at this level of expression (Figure 3A and not shown). We also assayed p14ARF mRNA expression very early after transduction with cyclin E retroviruses, in case there was a transient pulse of p14ARF induction, but we failed to detect in-

creased p14ARF expression at any point after transduction (not shown). Thus, the increased p53 abundance resulting from cyclin E expression was not associated with increased p14ARF expression. Another common mechanism of signaling cellular stress to p53 is phosphorylation, and many distinct phosphorylation sites affect p53 function in different physiologic contexts [27, 28]. We thus examined sitespecific p53 phosphorylation in cyclin E-expressing cells and found that phosphorylation of p53 on serine 15 was significantly elevated compared to control cells (Figure 4B). In contrast, no increased p53 phosphorylation was noted on several other sites, including serine 20 (not shown). Because the total abundance of p53 increased, it was difficult to determine if serine 15 phosphorylation specifically increased in response to cyclin E, or if this simply reflected increased p53 abundance. We thus treated control cells with the proteasome inhibitor MG-132 so that the amount of p53 in the control cells approximated the amount of p53 in the cyclin E-expressing cells. The cyclin E-expressing cells contained significantly more p53 serine 15 phosphorylation compared with the MG-132-treated controls, and this indicated that cyclin E expression led to a specific increase in p53 serine 15 phosphorylation (compare lanes 2 and 3, Figure 4C). Several members of the PI3 kinase family are physiologically important p53 serine 15 kinases. The ataxiatelangiectasia protein, ATM, and the related ATR protein both phosphorylate p53 on serine 15 after DNA damage, as well as in response to other cellular stresses such as

Current Biology 1820

Figure 3. Both p21 and p53 Are Required to Suppress Deregulated Cyclin E Expression in Primary Murine Embryo Fibroblasts (A and B) MEFs derived from p21 ⫹/⫹, p21 ⫺/⫺, and p53 ⫺/⫺ littermates were transduced with the indicated viral vectors. Asynchronous cell cycle profiles, cyclin E abundance, and cyclin E-associated kinase activity are shown. (C) MEFs derived from p19ARF null embryos were transduced with the indicated viruses and were harvested for cell cycle analysis. The ploidy of the various cell populations is indicated (2N, 4N, 8N, 16N).

replication blocks [29, 30]. DNA-PK also phosphorylates serine 15 in vitro, but its physiologic importance is less clear [31]. We treated cells with two distinct pharmacologic inhibitors of these enzymes, caffeine and wortmannin. We found that the amount of p53 serine 15 phosphorylation in cyclin E-overexpressing cells was substantially decreased after a short treatment with ei-

ther drug (Figure 4C). Furthermore, when we performed a time course analysis after wortmannin treatment of HFFs expressing cyclin E, we found that the abundance of p21 decreased roughly in parallel with serine 15 phosphorylation (Figure 4D). These data are consistent with the idea that p53 serine 15 phosphorylation, at least in part, drives p21 expression. To further define the rele-

p53-Dependent Cyclin E Suppression 1821

Figure 4. The Induction of p53 by Cyclin Is Associated with Increased p53 Serine 15 Phosphorylation, but Not Increased p14ARF Expression (A) HFF or WI-38 cells were transduced with either control (Babe) or cyclin E vectors, and RNA was harvested after 2 weeks. A quantitative real-time PCR assay was used to determine the amount of p14ARF mRNA compared with a HeLa cell standard. (B) HFFs were transduced as indicated, and Western analyses of p53 serine 15 phosphorylation (as detected by a phospho serine 15-specific antibody), total p53, and p21 are shown. (C) HFFs transduced with cyclin E (T380A) were treated with either caffeine or wortmannin. Control cells were treated with the proteasome inhibitor Mg-132 to increase total p53 abundance to levels comparable to cyclin E (T380A)-expressing cells. Western analysis of serine 15 phosphorylation and total p53 is shown. (D) HFFs were transduced with cyclin E vector and were treated for the indicated times with wortmannin prior to harvesting. Western analysis of serine 15 phosphorylation, total p53, p21, and cdk2 expression is shown. (E) ATM null primary fibroblasts (GM1627) were transduced with the indicated vectors, and total p53 abundance and p53 serine 15 phosphorylation were examined. (F) Human foreskin fibroblasts and ATM null fibroblasts transduced and passaged in parallel. p21 expression in passage 4 cells is shown.

Current Biology 1822

Table 1. Cytogenetic Analysis of WI-38 Cells 2 Weeks after Transduction with the Indicated Vectors Vector

Karyotype (20 Metaphases Total)

Babe ⫹ E6

14/20 normal 1/20 add (4p) 1/20 dic (4;8) 1/20 del(11q),dic(21;22) 1/20 dic(2;11) 1/20 del(Xq) 1/20 add(Xp) 14/20 normal 1/20 add(20p) 1/20 del(5q) 1/20 del(7q),t(X;9) 1/20 del(5q),del(6),⫹r 1/20 add(20q) 1/20 add(10p) 13/20 normal 1/20 dic(3;11),⫺21,⫹mar 1/20 ⫹mar 1/20 del(1),t(6;16) *1/20 46,XX,del(1)q11,del(3)(q21),del(18)(p11) *1/20 49,XX,⫺7,add(9)(q34),⫺16,⫹21,⫹r,⫹marx3 *1/20 80 ⬍3n⬎,XXX,der(1)t(1:11)(p36;q13), ⫹del(1),⫺2,⫺3,⫺4,⫺5, del(5)(q23), ⫹del(6)(p21),⫹7,del(8)(q13q22),⫹9x2,⫺10, ⫺12,⫹13,der(17)t(17;18),⫹18,der(18)t(17;18),⫹21x2, ⫹22,⫹marx5 *1/20 45,XX,add(X)(q28),del(1)(p22), del(7)(q22),add(13)(p11),⫺13,⫹mar

Cyclin E ⫹ E6

Cyclin E(T380A) ⫹ E6

In each sample, 20 metaphases were analyzed. Asterisks denote metaphases containing complex cytogenetic abnormalities.

vant p53 serine 15 kinase, we transduced primary AT fibroblasts (which are derived from a patient with ataxia telangiectasia and lack functional ATM protein) with the cyclin E retroviral vectors. We found that the induction of p21, p53, and p53 serine 15 phosphorylation occurred in the AT cells indistinguishably from normal human fibroblasts (Figures 4E and 4F). Thus, ATM is not essential for this response. Deregulated Cyclin E-cdk2 Activity Causes Genetic Instability and Delayed S Phase Progression, but Does Not Impair the G1 and G2 Checkpoints Reed and colleagues found that prolonged expression of cyclin E in established cells caused genetic instability that was manifested primarily by chromosome loss [13]. In order to determine if deregulated cyclin E activity in primary cells also caused genetic instability, we analyzed WI-38 cells 2 weeks after transduction with cyclin E retroviruses. In these experiments, cells were first transduced with the HPV-E6 virus to insure that the cyclin E was fully active. We found that the elevated cyclin E-cdk2 kinase activity associated with the expression of cyclin E (T380A) rapidly caused massive chromosomal abnormalities in a substantial fraction of the cells analyzed (Table 1). In contrast, when parallel experiments were performed in control WI-38 cells that did not express E6, sufficient metaphases could not be obtained for cytogenetic analysis due to the severe inhibition of WI-38 proliferation caused by prolonged cyclin E and cyclin E (T380A) expression. We next examined the integrity of the G1 and G2 checkpoints in primary cells expressing cyclin E. Expo-

nentially growing HFFs expressing cyclin E, HPV-E6, both, or neither were irradiated or mock irradiated and were harvested 24 hr later (Figure 5A). Wild-type cells arrested in G1 and G2 following irradiation. Cyclin E expression increased the proportion of S and G2 cells in the cycling cells but had no effect on either the G1 or G2 arrest after irradiation. In contrast, HPV-E6 expression almost completely prevented the G1 arrest, and the irradiated cells arrested instead in G2. Furthermore, when cells were transduced with both E6 and cyclin E vectors, they arrested in G2 following ␥-irradiation, indicating that ectopic cyclin E, even in combination with E6, does not override the G2 checkpoint. To further investigate the effects of cyclin E on S phase entry, HFFs expressing cyclin E, E6, or neither were density arrested, then released and either irradiated or mock irradiated. As expected, a large proportion of the cells expressing HPV-E6 expression entered S phase after ␥-irradiation (Figure 5B). In contrast, cyclin E expression in HFF did not lead to abnormal G1 checkpoint control, and there was no entry into S phase. Similarly, we found that, when cells were continuously labeled with BrDU after irradiation, nearly 50% of the E6expressing cells synthesized DNA, compared with only 5% of the cells expressing cyclin E (Figures 5C and S2). Thus, deregulated cyclin E activity does not grossly alter the G1 and G2 checkpoints that sense DNA damage after ␥-irradiation in HFFs, and this is unlikely to be the mechanism through which cyclin E promotes genetic instability. Although there was a slight increase in BrDU incorporation in irradiated cyclin E expressors versus controls, this finding may simply reflect the slightly higher proportion of cells in S phase in cells expressing

p53-Dependent Cyclin E Suppression 1823

Figure 5. Enforced Cyclin E Expression Does Not Impair G1 and G2 Checkpoint Controls (A) HFFs transduced with the indicated vectors were irradiated (10 Gy) or mock irradiated and were analyzed by flow cytometry 24 hr after irradiation. (B) HFFs transduced with the indicated vectors were density arrested, then trypsinized and irradiated (10 Gy), or mock irradiated, and replated. Flow analysis was performed 24 hr after replating. (C) HFF transduced with the indicated vectors were density arrested, then trypsinized and irradiated (20 Gy) or mock irradiated, and replated. BrDU was added 1 hr after replating, and incorporation was measured by flow cytometry 24 hr later.

cyclin E compared with controls after the density arrest and is of unclear significance (data not shown and Figure 5B). We also examined cyclin E overexpressors for DNA breaks and did not observe any increased TUNEL positivity in cells expressing cyclin E compared with controls (data not shown). Finally, we found that deregulated cyclin E activity caused defective S phase progression. We preformed

these experiments in HPV E6-expressing cells to segregate the consequences of excess cyclin E-cdk2 activity from cell cycle affects resulting from the p53-dependent response to excess cyclin E. When these cultures were synchronized in early S phase with aphidicolin and then released, we found that cyclin E did not affect the timing of S phase entry after release (Figure 6). Instead, we found that the rate at which cyclin E-expressing cells

Current Biology 1824

Figure 6. Enforced Cyclin E Expression Delays S Phase Progression in Synchronized Cells HFFs were first transduced with LXSH-E6, then with the indicated vectors. Cells were treated with aphidicolin overnight, then washed and placed in normal growth media for the indicated time periods.

progressed through S phase to G2 was decreased compared with control cells. Whereas almost half of the cyclin E-expressing cells remained in S phase 12 hr after aphidicolin release, almost all of the control cells progressed to G2 by this time. Thus, deregulated cyclin E-cdk2 activity, and not the p53-dependent response to cyclin E, impaired S phase progression in primary human fibroblasts. Discussion Although cyclin E normally activates cdk2, we found that prolonged and deregulated cyclin E expression in primary fibroblasts paradoxically inhibited cdk2. This was because deregulated cyclin E expression initiated a homeostatic cellular response that “buffered” the excess cyclin E by inducing p21 expression and thereby limiting inappropriate cdk2 activation. Disruption of either p53 or p21 rendered this response nonfunctional, and the excess cyclin E became fully active. Initially we thought it likely that the p14ARF protein mediated this response. However, we could not detect increased p14ARF mRNA or protein in cells with cyclin E-induced p53 activation. Furthermore, cells with cyclin E overexpression hyperaccumulated in S and G2 phases, but not in G1, as is usually observed in cells with p14ARF induction. Finally, p19ARF null mouse embryo fibroblasts were not hypersensitive to cyclin E overexpression. Thus, the accumulated data do not support a role for p14ARF in the signaling pathway that links cyclin E deregulation and p53 induction in human fibroblasts. In fact, the signals that induce p14ARF in human fibroblasts are quite different than those in MEFs, and two recent studies found that oncogenic ras does not lead to human p14ARF induction [26, 32]. Although the essential features of the p53-p21 system in restraining excess cyclin E activity are conserved in human and murine fibroblasts, there is at least one difference that deserves comment. In human fibroblasts,

cyclin E-dependent p21 expression is an induced response. In MEFs, however, although both p21 and p53 are required to suppress excess cyclin E, the amount of endogenous p21 synthesis is sufficient to inhibit the activity of the ectopic cyclin E without additional p21 induction. This likely reflects the high rate of p53-dependent p21 synthesis that occurs in MEFs shortly after being placed in culture, which limits their continued proliferation in vitro [33]. We have observed increased p53 serine 15 phosphorylation in cells with excess cyclin E that is sensitive to both wortmannin and caffeine. We directly tested the role of the ATM protein and found that ATM null fibroblasts accumulated serine 15 phosphorylated p53 and induced p21 in response to cyclin E. However, although it is not essential, it is important to realize that ATM may function redundantly with other kinases and contribute to this response in normal cells. One such kinase may be the ATR protein. ATR is involved in signaling replication blocks to p53, and the ATR knockout phenotype in mice has revealed that ATR probably plays an essential role in normal cell cycle progression [30, 34, 35]. Mechanisms other than p53 serine 15 phosphorylation may also contribute to the activation of p53 by cyclin E. For example, Schreiber and colleagues have found that ATR inhibition decreased p53 serine 15 phosphorylation after DNA damage but did not prevent p53 activation [36]. Instead, p53 and ATR acted independently and synergistically in the establishment of an S phase checkpoint response. Analogously, the caffeine/wortmannin-sensitive serine15 phosphorylation might comprise only a portion of the response that couples cyclin E deregulation to p53 activation. How might deregulated cyclin E cause S phase abnormalities that activate an S phase checkpoint? In yeast, S phase-promoting cyclins inhibit the transition of replication origins to the prereplicative state [37]. Furthermore, when early-firing origins are inhibited by hydroxyurea, then the stalled early origins inhibit late origins through a checkpoint that requires the Mec1 protein

p53-Dependent Cyclin E Suppression 1825

(the budding yeast ATM/ATR homolog) [38]. Similarly, inhibition of ATR function in a human cell line by a kinase-inactive ATR mutant renders these cells hypersensitive to treatments that prolong DNA synthesis, and cyclin E overexpression was synthetically lethal with ATR inhibition [39]. Thus, perhaps cyclin E deregulation leads to aberrant licensing of replication origins, and the resultant S phase progression defect may be sensed by a protein such as ATR, which then enforces an S phase checkpoint. Furthermore, the stalled replication origins associated with this prolonged S phase may be fragile and constitute the precursors to genetic instability. Another mechanism through which enforced cyclin E expression might impair normal cell cycle progression is by cyclin A-cdk2 inhibition, since we have found that cyclin A-cdk2 activity (and cyclin A expression) drops substantially in cells with ectopic cyclin E expression (Figures 2C and S3). However, cyclin E-induced cell cycle anomalies persist in E6-expressing cells with high levels of cyclin A-cdk2 kinase activity, so cyclin A-cdk2 activity cannot be the principle cause of the cyclin E-associated S phase phenotype. One unanswered question regarding the cyclin E-p53 homeostatic pathway concerns the signal that maintains p21 expression once cyclin E-cdk2 activity has been initially suppressed. That is, if the signal from cyclin E to p53 is simply excess cyclin E-cdk2 activity, once p21 is induced, this signal should be quenched. However, our data indicate that p21 expression is constitutively maintained once the response is activated. Perhaps mechanisms other then elevated cyclin E-cdk2 kinase activity per se might maintain p53 activation by enforced cyclin E expression. For example, increased abundance of either monomeric cyclin E or cyclin E-cdk complexes, or uncoupling of cyclin E expression from its normal cell cycle regulation, might provide this signal. Our data indicate that inactivation of the p21-p53 pathway in tumors may be necessary for deregulated cyclin E expression to become fully oncogenic. Interestingly, mice expressing a T cell-specific cyclin E transgene do not contain increased amounts of cdk2 activity in thymic extracts, whereas lymphomas induced in these mice do exhibit elevated cdk2 activity [20]. Perhaps loss of p53 function is one of the required events in the neoplastic progression of these lymphomas and allows the exogenous cyclin E to form active complexes with cdk2. The combination of cyclin E-cdk2 deregulation and p53 loss may be particularly potent at inducing genetic instability. This is because unchecked cyclin E-cdk2 activity may promote genetic instability through the mechanisms discussed above, and the loss of p53 function greatly impairs the cell’s ability to respond to this damage appropriately. This model also suggests that tumors with deregulated cyclin E-cdk2 activity may be very susceptible to therapies that modulate or restore p53 or ATR function.

Conclusions Deregulated cyclin E expression in primary fibroblasts initiates a cellular response leading to p53 and p21 activation and thereby preventing excessive cdk2 activity. p53 and p21 thus form an inducible barrier that protects

primary cells from excess cyclin E activity. When this barrier is breached through the loss of either p53 or p21, then deregulated cyclin E expression results in elevated cyclin E-cdk2 activity and causes severe cell cycle anomalies and genetic instability. Experimental Procedures Cell Lines, Tissue Culture, Drug Treatments, Cell Cycle Analyses, TUNEL Assay, and Cytogenetics Cells and cell lines were obtained as follows: WI-38 and WS-1, ATCC; HFFs, W. Carter (FHCRC); ATM mutant fibroblasts (GM1627), J. Simon (FHCRC); p21⫺/⫺ and p21 ⫹/⫹ MEFs, J. Roberts (FHCRC); p53⫺/⫺ and p53 ⫹/⫹ MEFs, C. Kemp (FHCRC); p19ARF ⫺/⫺ and p19ARF ⫹/⫹ MEFs, M Roussell (St. Judes); P27 ⫺/⫺ MEFs, M. Fero (FHCRC). All cells were grown in DMEM with 10% FCS. Drug treatments: caffeine (5 mM) and wortmannin (50 ␮M) were placed in growth media for 3 hr prior to cell harvest. For the induction of p53, cells were treated with MG-132 (Calbiochem, 5 ␮g/ml) for 1 hr prior to harvest. To measure S phase transit, cells were synchronized with aphidicolin (Calbiochem, 5 ␮g/ml), then released into media containing nocodazole (Sigma, 40 ng/ml). Cell cycle analyses: cells were trypsinized and fixed in cold 70% ethanol, then washed and resuspended in PBS containing propidium iodide and RNase A. Analyses were performed on a Becton Dickinson FACSCAN and utilized Multicycle and Cellquest software. G1 and G2 checkpoint assays: cells were trypsinized after being density arrested for 4 days, or after asynchronous growth, and were irradiated from a Cesium source. Cells were harvested 24 hr after irradiation. BrDU labeling: cells were incubated continuously in 20 ␮M BrDU (Sigma) after replating following mock treatment or irradiation at the indicated dosage. After 24 hr, cell nuclei were isolated, and BrDU was detected with a fluorescein-conjugated monoclonal anti-BrDU antibody (Amersham). TUNEL staining was performed on HFF by using the In Situ Cell Death Detection Kit (Roche Diagnostics). The cytogenetic studies were performed by the University of Washington Clinical Cytogenetics Laboratory. Retroviral Vector Cloning and Transduction DNAs encoding cyclin E or cyclin E (T380A) were cloned into pBabePuro [40]. PLXSH-E6 was provided by D. Galloway (FHCRC). Viral stocks were prepared by transfecting appropriate vectors into the Phoenix Ampho or Eco producer cells for transducing human or mouse cell lines, respectively, and stocks were filtered and frozen (producer cells provided by G. Nolan, Stanford). Cells were transduced by viral supernatants in DME and 5 ␮g/ml polybrene and were selected with puromycin (3 ␮g/ml for human cells, 5 ␮g/ml for MEFs) or hygromycin (400 ␮g/ml) 48 hr after transduction. Antibodies The following antibodies were used for Western blotting: monoclonal anti-p21 (Transduction Labs), monoclonal anti-human cyclin E HE12 (Pharmingen), anti-p53 (D01) (Santa Cruz Biotechnology), phospho-p53 Ser15 antibody (9284) (Cell Signaling), and anti-phospho p53 Sampler Kit (Ser6, Ser9, Ser15, Ser20, Ser37, Ser46, Ser392) (Cell Signaling). The following antibodies were used for kinase assays: rabbit polyclonal anti-cyclin E, monoclonal anti-human cyclin E (HE111), rabbit polyclonal anti-cyclin A (gift of J Roberts [FHCRC]), and rabbit polyclonal anti-cdk 2 (M2) (Santa Cruz Biotechnology). Immunoblotting and Kinase Assays Cells were lysed in NP-40 buffer (0.5% Np-40, 10 mM Tris [pH 7.4], 0.15 M NaCl, 10 mg/ml each of aprotonin, leupeptin, and pepstatin, 50 mM NaF, 1 mM Na vanadate). Lysates were quantitated, electrophoresed, and transferred to PVDF membranes as described [7]. After incubation with primary antibodies, proteins were visualized by incubation with HRP-conjugated anti-rabbit or anti-mouse secondary antibodies as appropriate (1:10,000, Amersham), followed by ECL per the manufacturer’s instructions (Pierce). IP kinase assays: cell lysates normalized for protein concentration were incubated at 4⬚C for 1 with appropriate dilutions of primary antibodies

Current Biology 1826

and 30 ␮l of a 1:1 slurry of PBS-Gammabind (Pharmacia). Precipitates were washed, and phosphorylation of histone H1 was performed as previously described [7]. Quantitative Real-Time PCR RNA was isolated from cultured cells (10-cm dishes) with Trizol reagent (Invitrogen), and cDNA was synthesized by oligo dT priming with TaqMan Reverse Transcription Reagents (Applied Biosystems) as per manufacturer’s instructions. Quantitative real-time PCR was performed with a Prism 7700 sequence detector (Applied Biosystems) in triplicate 50 ␮l reactions containing 5 ␮l cDNA, 25 ␮l 2X Taqman Universal Master mix (Applied Biosystems), and probe/ primer mix for p14arf (Applied Biosystems) or PGK (Applied Biosystems). Quantitation for each probe/primer combination was determined relative to serial dilutions of HeLa cell cDNA with Sequence Detection System Software (Applied Biosystems). Supplementary Material Supplementary Material including data referenced in the text but not included in the published manuscript is available at http:// images.cellpress.com/supmat/supmatin.htm. Acknowledgments We thank our colleagues for their many kind gifts of reagents. We also thank John Diffley, Mike Kastan, Scott Lowe, Jim Roberts, and Charles Sherr for helpful discussions. B.E.C. is a W.M. Keck Distinguished Young Scholar in Medical Research. This work was supported by National Institutes of Health (NIH) R01 CA84069 (B.E.C.) as well as NIH T32 CA80416-05 (H.C.H.) and T32 CA0951517 (A.C.M). H.H is an E.D. Thomas Scholar of the Jose Carreras Foundation. M.W. is a fellow of the Lymphoma and Leukemia Society. Received: June 26, 2002 Revised: August 15, 2002 Accepted: September 6, 2002 Published: October 29, 2002

10.

11. 12. 13. 14.

15.

16.

17.

18. 19.

20.

21.

References 1. Morgan, D.O. (1995). Principles of CDK regulation. Nature 374, 131–134. 2. Dulic, V., Lees, E., and Reed, S.I. (1992). Association of human cyclin E with a periodic G1-S phase protein kinase. Science 257, 1958–1961. 3. Koff, A., Giordano, A., Desai, D., Yamashita, K., Harper, J.W., Elledge, S., Nishimoto, T., Morgan, D.O., Franza, B.R., and Roberts, J.M. (1992). Formation and activation of a cyclin E-cdk2 complex during the G1 phase of the human cell cycle. Science 257, 1689–1694. 4. Ohtsubo, M., Theodoras, A.M., Schumacher, J., Roberts, J.M., and Pagano, M. (1995). Human cyclin E, a nuclear protein essential for the G1-to-S phase transition. Mol. Cell. Biol. 15, 2612– 2624. 5. Resnitzky, D., Gossen, M., Bujard, H., and Reed, S.I. (1994). Acceleration of the G1/S phase transition by expression of cyclins D1 and E with an inducible system. Mol. Cell. Biol. 14, 1669–1679. 6. Geng, Y., Eaton, E.N., Picon, M., Roberts, J.M., Lundberg, A.S., Gifford, A., Sardet, C., and Weinberg, R.A. (1996). Regulation of cyclin E transcription by E2Fs and retinoblastoma protein. Oncogene 12, 1173–1180. 7. Clurman, B.E., Sheaff, R.J., Thress, K., Groudine, M., and Roberts, J.M. (1996). Turnover of cyclin E by the ubiquitin-proteasome pathway is regulated by cdk2 binding and cyclin phosphorylation. Genes Dev. 10, 1979–1990. 8. Koepp, D.M., Schaefer, L.K., Ye, X., Keyomarsi, K., Chu, C., Harper, J.W., and Elledge, S.J. (2001). Phosphorylation-dependent ubiquitination of cyclin E by the SCFFbw7 ubiquitin ligase. Science 294, 173–177. 9. Strohmaier, H., Spruck, C.H., Kaiser, P., Won, K.A., Sangfelt, O., and Reed, S.I. (2001). Human F-box protein hCdc4 targets

22.

23.

24.

25.

26.

27.

28. 29. 30. 31.

32.

cyclin E for proteolysis and is mutated in a breast cancer cell line. Nature 413, 316–322. Won, K.A., and Reed, S.I. (1996). Activation of cyclin E/CDK2 is coupled to site-specific autophosphorylation and ubiquitindependent degradation of cyclin E. EMBO J. 15, 4182–4193. Sherr, C.J., and McCormick, F. (2002). The RB and p53 pathways in cancer. Cancer Cell 2, 103–112. Ohtsubo, M., and Roberts, J.M. (1993). Cyclin-dependent regulation of G1 in mammalian fibroblasts. Science 259, 1908–1912. Spruck, C.H., Won, K.A., and Reed, S.I. (1999). Deregulated cyclin E induces chromosome instability. Nature 401, 297–300. Mussman, J.G., Horn, H.F., Carroll, P.E., Okuda, M., Tarapore, P., Donehower, L.A., and Fukasawa, K. (2000). Synergistic induction of centrosome hyperamplification by loss of p53 and cyclin E overexpression. Oncogene 19, 1635–1646. Dou, Q.P., Pardee, A.B., and Keyomarsi, K. (1996). Cyclin E–a better prognostic marker for breast cancer than cyclin D? Nat. Med. 2, 254. Erlanson, M., Portin, C., Linderholm, B., Lindh, J., Roos, G., and Landberg, G. (1998). Expression of cyclin E and the cyclindependent kinase inhibitor p27 in malignant lymphomas-prognostic implications. Blood 92, 770–777. Porter, P., Malone, K., Heagerty, P., Alexander, G., Gatti, L., Firpo, E., Daling, J. R., and Roberts, J.M. (1997). Expression of cell-cycle regulators p27Kip1 and cyclin E, alone and in combination, correlate with survival in young breast cancer patients. Nat. Med. 3, 222–225. Steeg, P.S., and Zhou, Q. (1998). Cyclins and breast cancer. Breast Cancer Res. Treat. 52, 17–28. Bortner, D.M., and Rosenberg, M.P. (1997). Induction of mammary gland hyperplasia and carcinomas in transgenic mice expressing human cyclin E. Mol. Cell. Biol. 17, 453–459. Karsunky, H., Geisen, C., Schmidt, T., Haas, K., Zevnik, B., Gau, E., and Moroy, T. (1999). Oncogenic potential of cyclin E in T-cell lymphomagenesis in transgenic mice: evidence for cooperation between cyclin E and Ras but not Myc. Oncogene 18, 7816– 7824. Scheffner, M., Huibregtse, J.M., Vierstra, R.D., and Howley, P.M. (1993). The HPV-16 E6 and E6-AP complex functions as a ubiquitin-protein ligase in the ubiquitination of p53. Cell 75, 495–505. Deng, C., Zhang, P., Harper, J.W., Elledge, S.J., and Leder, P. (1995). Mice lacking p21CIP1/WAF1 undergo normal development, but are defective in G1 checkpoint control. Cell 82, 675–684. Donehower, L.A., Harvey, M., Slagle, B.L., McArthur, M.J., Montgomery, C.A., Jr., Butel, J.S., and Bradley, A. (1992). Mice deficient for p53 are developmentally normal but susceptible to spontaneous tumours. Nature 356, 215–221. Lanni, J., and Jacks, T. (1998). Characterization of the p53dependent postmitotic checkpoint following spindle disruption. Mol. Cell. Biol. 18, 1055–1064. Bates, S., Phillips, A.C., Clark, P.A., Stott, F., Peters, G., Ludwig, R.L., and Vousden, K.H. (1998). p14ARF links the tumour suppressors RB and p53. Nature 395, 124–125. Wei, W., Hemmer, R.M., and Sedivy, J.M. (2001). Role of p14(ARF) in replicative and induced senescence of human fibroblasts. Mol. Cell. Biol. 21, 6748–6757. Giaccia, A.J., and Kastan, M.B. (1998). The complexity of p53 modulation: emerging patterns from divergent signals. Genes Dev. 12, 2973–2983. Prives, C., and Hall, P.A. (1999). The p53 pathway. J. Pathol. 187, 112–126. Abraham, R.T. (2001). Cell cycle checkpoint signaling through the ATM and ATR kinases. Genes Dev. 15, 2177–2196. Zhou, B.B., and Elledge, S.J. (2000). The DNA damage response: putting checkpoints in perspective. Nature 408, 433–439. Jimenez, G.S., Bryntesson, F., Torres-Arzayus, M.I., Priestley, A., Beeche, M., Saito, S., Sakaguchi, K., Appella, E., Jeggo, P.A., Taccioli, G.E., et al. (1999). DNA-dependent protein kinase is not required for the p53-dependent response to DNA damage. Nature 400, 81–83. Ferbeyre, G., de Stanchina, E., Querido, E., Baptiste, N., Prives,

p53-Dependent Cyclin E Suppression 1827

33. 34.

35.

36.

37.

38.

39.

40.

C., and Lowe, S.W. (2000). PML is induced by oncogenic ras and promotes premature senescence. Genes Dev. 14, 2015–2027. Sherr, C.J., and DePinho, R.A. (2000). Cellular senescence: mitotic clock or culture shock? Cell 102, 407–410. Brown, E.J., and Baltimore, D. (2000). ATR disruption leads to chromosomal fragmentation and early embryonic lethality. Genes Dev. 14, 397–402. de Klein, A., Muijtjens, M., van Os, R., Verhoeven, Y., Smit, B., Carr, A.M., Lehmann, A.R., and Hoeijmakers, J.H. (2000). Targeted disruption of the cell-cycle checkpoint gene ATR leads to early embryonic lethality in mice. Curr. Biol. 10, 479–482. Nghiem, P., Park, P.K., Kim Ys, Y.S., Desai, B.N., and Schreiber, S. (2001). ATR is not required for p53 activation but synergizes with p53 in the replication checkpoint J. Biol. Chem. 277, 4428– 4434. Dahmann, C., Diffley, J.F., and Nasmyth, K.A. (1995). S-phasepromoting cyclin-dependent kinases prevent re-replication by inhibiting the transition of replication origins to a pre-replicative state. Curr. Biol. 5, 1257–1269. Santocanale, C., and Diffley, J.F. (1998). A Mec1- and Rad53dependent checkpoint controls late-firing origins of DNA replication. Nature 395, 615–618. Nghiem, P., Park, P.K., Kim, Y., Vaziri, C., and Schreiber, S.L. (2001). ATR inhibition selectively sensitizes G1 checkpoint-deficient cells to lethal premature chromatin condensation. Proc. Natl. Acad. Sci. USA 98, 9092–9097. Morgenstern, J.P., and Land, H. (1990). Advanced mammalian gene transfer: high titre retroviral vectors with multiple drug selection markers and a complementary helper-free packaging cell line. Nucleic Acids Res. 18, 3587–3596.