MDM2 Controls the Timely Expression of Cyclin A to ...€¦ · 07/08/2009 · kinase (cdk) inhibitor p16 prevents MDM2-mediated inhibition of cyclin A expression, implicating its

MDM2 Controls the Timely Expression of Cyclin A toRegulate the Cell Cycle

Rebecca Frum, Mahesh Ramamoorthy, Lathika Mohanraj, Sumitra Deb, and Swati Palit Deb

Department of Biochemistry and Molecular Biology and the Massey Cancer Center,Virginia Commonwealth University, Richmond, Virginia

AbstractOverexpression of MDM2 has been related to oncogenesis.In this communication, we present evidence to show thatMDM2 controls the cell cycle–dependent expression ofcyclin A by using a pathway that ensures its timelyexpression. MDM2 does not inhibit cyclin D or E expression.Silencing of endogenous MDM2 expression elevates cyclinA expression. The p53-binding domain of MDM2 harbors aSWIB region homologous to a conserved domain of achromosome remodeling factor BRG1-associated protein.The SWIB domain of MDM2 inhibits cyclin A expression in ap53- and BRG1-dependent fashion, suggesting that MDM2interferes with p53 binding of the BRG1 complex freeing it torepress cyclin A expression. Silencing of cyclin-dependentkinase (cdk) inhibitor p16 prevents MDM2-mediatedinhibition of cyclin A expression, implicating its role in theprocess. MDM2-mediated repression of cyclin A expressioninduces G1-S arrest, which can be rescued by ectopicexpression of cyclin A. Cancer cells lacking p53, p16, orBRG1 escape MDM2-mediated repression of cyclin Aexpression and growth arrest. Our data propose a novelmechanism by which MDM2 controls the cell cycle in normalcells and how cancer cells may escape this important safetybarrier. (Mol Cancer Res 2009;7(8):1253–67)

IntroductionThe oncoprotein MDM2, coded by the human homologue of

the mouse double minute-2 (mdm2) gene, is frequently overex-pressed in various human cancers, particularly in breast tumorsand carcinomas and soft tissue sarcomas (1-3). Amplification ofthe mdm2 gene enhances the tumorigenic potential of murinecells (4, 5).

MDM2 interacts with several growth suppressors, includingthe tumor suppressor p53 and the retinoblastoma susceptibility

gene product Rb. These interactions are considered to be pos-sible mechanisms for the oncogenic function of MDM2 (1-3).MDM2 recognizes the transactivation domain of p53 and inac-tivates p53-mediated transcriptional activation (6-10). Workfrom our laboratory showed that this interaction is needed forthe inhibition of p53-mediated transactivation (7, 10). MDM2also degrades p53 by targeting p53 for ubiquitination (11-13).

Although MDM2 has been related to oncogenesis, overex-pression of MDM2 at a level similar to that found in cancercells induces growth arrest in cultured nontransformed cells(14-16). Two- to 50-fold overexpression of MDM2 efficientlyarrests normal human diploid cells at G1 (14, 16). This propertyof MDM2 is also evidenced in transgenic mice targeted to ex-press MDM2 in breast epithelial cells, which leads to inhibitedbreast development and morphogenesis (17). Generation oftransgenic mice expressing MDM2 splice variants showed alow success rate and selection for transgenic mice with muta-tions in the MDM2 splice variant (18). Tumor-derived cells areoften insensitive to the growth-inhibitory effect of MDM2 (14).Deletion mutants of MDM2 lacking the functional growth-in-hibitory domains enhance the tumorigenic potential of NIH3T3cells (14). Later studies revealed that splice variants of MDM2expressed in cancer cells lack the growth-inhibitory domainsand inactivate the full-length protein (19). These observationssuggest that the MDM2 splice variants found in cancer cells areincapable of inducing G1 arrest and may inactivate the growtharrest function of the full-length protein. The dichotomy in theoncogenic and G1 arrest functions of MDM2 led us to deter-mine the mechanisms for MDM2-mediated growth arrest.

To determine how MDM2 induces G1 arrest in normalcells and how transformed cells may escape this function,we examined the expression of endogenous G1-S cyclins incells after overexpressing or silencing MDM2. The resultsof this analysis show that in apparently normal diploid cells,MDM2 inhibits cyclin A expression but not the expression ofcyclin D or E. MDM2-mediated down-regulation of cyclin Aleads to G1 arrest, which can be rescued by ectopic expressionof cyclin A.

Cell cycle–dependent expression of cyclin A is ensured by ahighly orchestrated participation of cyclin-dependent kinases(cdk). The cyclin D–cdk4 inhibitor p16 prevents untimelyactivation of the cyclin E–dependent kinase and thereby theexpression of cyclin A by restricting cyclin E expression andalso through redistribution of cdk inhibitors p21 or p27 to thecyclin E–cdk2 complex (20, 21). Cyclin E–cdk2 releases thechromosome remodeling factor BRG1 from a multicomponentrepressor complex, which is known to inhibit cyclin A expres-sion (22). The tumor suppressor p53 interacts with BRG1 andother associated factors of the BRG1 repressor complex to

Received 7/15/08; revised 4/12/09; accepted 4/30/09; published OnlineFirst 8/11/09.Grant support: Jeffress Memorial Trust and National Cancer Institute grantsCA74172 (S.P. Deb), CA121144, and CA70712 (S. Deb).The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.Note: Supplementary data for this article are available at Molecular CancerResearch Online (http://mcr.aacrjournals.org/).Present address for R. Frum: Pathology and Laboratory Medicine, University ofNorth Carolina, Chapel Hill, NC 27514.Requests for reprints: Swati Palit Deb, Goodwin Research Laboratories,Virginia Commonwealth University, 401 College Street, Richmond, VA 23298.Phone: 804-828-9541; Fax: 804-827-1427. E-mail: [email protected] © 2009 American Association for Cancer Research.doi:10.1158/1541-7786.MCR-08-0334

Mol Cancer Res 2009;7(8). August 2009 1253

Published Online First on August 11, 2009

on April 22, 2021. © 2009 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst August 11, 2009; DOI: 10.1158/1541-7786.MCR-08-0334

http://mcr.aacrjournals.org/

up-regulate the cdk inhibitor p21, which inhibits cyclin Aexpression by blocking cyclin E–cdk2 activity. Our data showthat MDM2 intercepts this cell cycle–dependent pathwaythrough its p53-binding ability to control cyclin A expression.Transformed cells defective in BRG1, p53, or p16 functionare defective in MDM2-mediated cyclin A down-regulation,which may help the cells escape an important growth-regulatorypathway.

ResultsMDM2 Inhibits Cyclin A Expression in WI38 Cells

Because MDM2 induces G1 arrest (14, 16), we determinedif MDM2 inhibits expression of endogenous G1-S cyclins. Nor-mal diploid WI38 cells were nucleofected with an MDM2expression plasmid. The cell extracts were analyzed for expres-sion of cyclin A and cyclin E. Western blot and densitometricanalysis showed that MDM2 significantly down-regulates theexpression of cyclin A (Fig. 1A; Supplementary Fig. S1A) butdoes not down-regulate cyclin E (Supplementary Fig. S1B).Furthermore, quantitative PCR (QPCR) analysis of cDNA pre-pared from the nucleofected WI38 cells showed that MDM2down-regulates the cyclin A transcript levels in these cells(Supplementary Fig. S1C). This observation suggests thatMDM2 overexpression may lead to down-regulation of cyclinA and cell cycle arrest.

MDM2-Mediated Inhibition of Cyclin A Expression CanBe Detected by Confocal Microscopy

Because there should be heterogeneity in the expressionlevels of endogenous G1-S cyclins in asynchronously growingcultured cells, and Western analysis yields an average data, we

determined cyclin A expression in normal diploid WI38 cellsafter ectopic expression of MDM2 by immunostaining and con-focal imaging (Fig. 1B). Normal diploid WI38 cells were trans-fected with an MDM2 expression plasmid on sterile coverslips.MDM2-overexpressing cells were detected by immunostainingfixed cells with an anti-MDM2 antibody directly coupled withFITC, and endogenous cyclin A was detected with an anti–cyclin A antibody and a rhodamine-coupled secondary antibody.Transfected cells were also stained with anti–proliferating cellnuclear antigen (PCNA) antibody and rhodamine-coupled sec-ondary antibody as a control. Because cells that have crossedthe G1-S phase border, but not the G0-G1, should express cyclinA, both cyclin A–expressing and nonexpressing cells were de-tected in the untransfected population as expected. However,cells overexpressing MDM2 did not show cyclin A expression.Similar results were obtained in normal mouse embryo fibroblast(MEF) cells (data discussed in Fig. 6B, top). These results con-firm our finding that MDM2 can down-regulate cyclin A expres-sion in normal diploid cells. MDM2 did not significantly alterPCNA levels (Supplementary Fig. S2), suggesting that MDM2does not inhibit PCNA expression.

MDM2-Mediated Inhibition of Cyclin A Expression CanBe Detected by Flow Cytometry

In an asynchronous population, cells that crossed the G1-Sphase boundary should express cyclin A, whereas the cells inthe G0-G1 phase of the cell cycle should not. It is possible thatMDM2-expressing cells are asynchronous, some of which ex-press cyclin A and some do not. Because confocal microscopyyields results of few cells, our confocal analysis may have beenunable to find MDM2-overexpressing cells expressing cyclin A.

FIGURE 1. MDM2 overexpression down-regulates endogenous cyclin A expression in WI38 cells. A. Western blot analysis of cell extracts prepared fromWI38 cells nucleofected with vector (pCMV) or MDM2 expression plasmids (pCMV MDM2). Left, migration of MDM2, cyclin A, and actin. B. Confocal imaginganalysis of WI38 cells after overexpressing MDM2. Nuclei were stained with DAPI. Cyclin A expression was detected as red nuclear staining of cells by arhodamine-coupled anti–cyclin A antibody (small block arrows), and MDM2 expression was detected as green nuclear staining by a FITC-coupled anti-MDM2antibody. Long arrows, nuclear staining of the cells expressing MDM2 by DAPI, rhodamine, and FITC. Staining intensities in the rhodamine channel variedfrom 939 to 715 in untransfected cells, whereas the intensity of MDM2-overexpressing cells was 331 to 274 (close to background) in the fields shown. At least50 MDM2-expressing cells were analyzed in each experiment. Two representative fields are shown.

Frum et al.

Mol Cancer Res 2009;7(8). August 2009

1254




To test this possibility, we determined the percentage of MDM2-expressing cells that show inhibition of cyclin A expression byflow cytometry. After its ectopic expression, we sorted MDM2-overexpressing cells by flow cytometry and determined theexpression of G1 cyclins using methods reported previously(14-16). The flow cytometric method has been popularly usedfor detection and quantitation of protein expression (23).

WI38 cells were transfected with an MDM2 expression plas-mid. MDM2-overexpressing cells were identified by immunos-taining with an anti-MDM2 antibody directly coupled withphycoerythrin (PE). Endogenous levels of cyclin A were esti-mated by immunostaining the cells with an anti–cyclin A anti-body directly coupled with FITC. Because not all the cells weretransfected successfully, transfected cells overexpressingMDM2 were sorted from untransfected cells containing endog-enous levels of MDM2 using flow cytometry, and the endoge-nous cyclin A levels in the two populations were compared.

A representative result is shown in Fig. 2. Mock-transfectedWI38 cells immunostained with PE-coupled anti-MDM2 anti-body and FITC-coupled matched isotype of anti–cyclin A an-tibody (Fig. 2A) were used to determine background FITC andPE intensities. Vector-transfected cells or cells transfected with aplasmid expressing an irrelevant protein did not induce growtharrest (14). Figure 2B shows WI38 cells transfected withMDM2 expression plasmid and immunostained with PE-coupledanti-MDM2 and FITC-coupled anti–cyclin A antibodies. Thecells showing PE intensity similar to mock-transfected cells were

gated as untransfected cells expressing endogenous levels ofMDM2. Successfully transfected, MDM2-overexpressing cellsshowing higher PE intensity than the untransfected cells werethus identified (Fig. 2B, overexpressed MDM2).

To quantitate the increase in the level of MDM2 expressionover the endogenous level in transfected cells, the PE intensityof cells expressing endogenous levels of MDM2 or overexpres-sing MDM2 was plotted against cell number (Fig. 2C). Asevident from the figure, the MDM2-overexpressing cells showed2- to 50-fold increase in MDM2 expression over untransfectedcells. Single-parameter plot using mock-transfected cells showedsimilar results (Supplementary Fig. S3). This level of MDM2expression is well within the range of MDM2 overexpressionobserved in cancer cells (4, 24).

Comparative immunostaining with FITC-coupled anti–cyclin A antibody (Fig. 2D, curve 2) as opposed to a FITC-coupled IgG isotype of cyclin A antibody (Fig. 2D, curve 1)identified the cyclin A–expressing cells. The intensity of FITCstaining indicating endogenous cyclin A levels in cells expres-sing endogenous levels of MDM2 (Fig. 2E, curve 2) or over-expressing MDM2 (Fig. 2E, curve 3) was estimated andplotted against cell number.

In untransfected cells with endogenous MDM2, at least 41%of untransfected cells expressed approximately 8-fold highercyclin A levels over the background, whereas virtually all(97%) of the transfected (MDM2-overexpressing) cells showedFITC intensity similar to the background level (Fig. 2E, curve 3).

FIGURE 2. MDM2-mediated inhibition of cyclin A expression detected by flow cytometric analysis. A. Dot plot of mock-transfected WI38 cells immuno-stained with PE-coupled anti-MDM2 antibody and FITC-coupled IgG isotype of cyclin A antibody. The boundary of the mock-transfected cells was drawn toachieve 0.1% background staining. B. Dot plot of WI38 cells transfected with MDM2 expression plasmid and immunostained with PE-coupled anti-MDM2antibody and FITC-coupled cyclin A antibody. Transfected (overexpressed MDM2) and untransfected (Background) cells are indicated. C. Single-parameterhistogram showing MDM2 overexpression over background levels in transfected cells estimated by PE intensity. D. Single-parameter histogram showingendogenous cyclin A levels over background determined by comparative FITC intensity of cells stained with FITC-coupled IgG isotype of cyclin A antibody(curve 1) or FITC-coupled cyclin A antibody (curve 2). E. Single-parameter histogram comparing endogenous cyclin A expression in cells with endogenouslevels of MDM2 (curve 2) and MDM2-overexpressing cells (curve 3) obtained by plotting FITC intensity after immunostaining with FITC-coupled cyclin Aantibody.

MDM2 Regulates Cell Cycle–Dependent Cyclin A Expression


1255




The mean FITC intensity of the MDM2-overexpressing cellsis approximately 8-fold lower than that of the untransfectedWI38 cells that expressed cyclin A (Fig. 2E, comparecurves 3 and 2). This observation suggests that in WI38cells, MDM2 overexpression drastically inhibits cyclin Aexpression. An experiment using normal diploid MRC5 cellsshowed similar results (data not shown). Because cyclin Ais required for DNA replication (25, 26), this result alsosuggests a precise cell cycle inhibitory step modulated byMDM2. This observation is consistent with our previousreport that MDM2 induces G1 arrest in WI38 or MRC5cells (14).

MDM2 Does Not Prevent Cyclin E ExpressionBecause MDM2 inhibited cyclin A expression, we deter-

mined if MDM2 inhibits expression of cyclin E that precedescyclin A expression during normal cell cycling. As described inthe previous experiments, normal diploid WI38 cells weretransfected with an MDM2 expression plasmid. Transfectedcells overexpressing MDM2 were identified using a PE-coupledanti-MDM2 antibody, and the endogenous level of cyclin E wasdetermined by immunostaining with a FITC-coupled anti–cyclinE antibody.

As in the previous experiment, mock-transfected WI38 cellsimmunostained with PE-coupled anti-MDM2 antibody andFITC-coupled matched isotype of anti–cyclin E antibody

(Fig. 3A) were used to determine background FITC and PEintensities. The cells showing PE intensity similar to mock-transfected cells (Fig. 3A) were gated as untransfected cells(Fig. 3B, Background) and MDM2-overexpressing cells show-ing higher PE intensity than the untransfected cells were gatedas transfected (Fig. 3B, overexpressed MDM2). The levels ofoverexpressed MDM2 (3- to 50-fold over endogenous levels)in transfected cells were similar to the previous experiment(Fig. 3C).

Cells immunostained with FITC-coupled anti–cyclin E anti-body (Fig. 3D, curve 2) showed stronger FITC intensity thanthe FITC-coupled isotype (Fig. 3D, curve 1), identifying cellsexpressing cyclin E. Comparison of the FITC–anti-cyclin Estaining intensities of MDM2-overexpressing and untransfectedcells with endogenous MDM2 (Fig. 3E) showed that the meanFITC intensities of the transfected and untransfected cells didnot differ significantly (Fig. 3E, curves 2 and 3), suggestingthat MDM2 overexpression did not significantly alter the ex-pression of cyclin E in WI38 cells. These results suggest thatMDM2-induced G1 arrest is an event downstream of cyclin Eexpression.

Elimination of MDM2 Expression Elevates EndogenousCyclin A Levels

Because overexpression of MDM2 inhibits cyclin A expres-sion in normal cells, we determined if decrease in MDM2

FIGURE 3. Overexpression of MDM2 did not alter cyclin E expression in WI38 cells. A. Dot plot of mock-transfected WI38 cells stained with PE-coupledanti-MDM2 antibody and FITC-coupled IgG isotype of cyclin E antibody. The boundary of the mock-transfected cells was drawn to achieve 0.1% backgroundstaining. B. Dot plot of WI38 cells transfected with MDM2 expression plasmid and immunostained with PE-coupled anti-MDM2 antibody and FITC-coupledcyclin E antibody. Transfected (overexpressed MDM2) and untransfected cells with endogenous MDM2 (Background) are indicated. C. Single-parameterhistogram showing MDM2 overexpression in transfected cells over endogenous (Background) levels estimated by PE intensity. D. Single-parameter histo-gram comparing endogenous cyclin E levels over background obtained by plotting FITC intensity of cells immunostained with FITC-coupled IgG isotype ofcyclin E antibody (curve 1) and FITC-coupled cyclin E antibody (curve 2). E. Single-parameter histogram comparing endogenous cyclin E levels in cells withendogenous MDM2 (curve 2) and MDM2-overexpressing cells (curve 3) obtained by plotting FITC intensity after immunostaining with FITC-coupled cyclin Eantibody.

Frum et al.


1256




expression would elevate endogenous levels of cyclin A. How-ever, the absence of MDM2 stabilizes p53 (27-29). Silencing ofMDM2 expression in normal cells such as WI38 thereforeshould elevate p53 levels, up-regulating the cdk inhibitor p21and down-regulating cyclin A expression, masking the effect ofsilencing MDM2. Indeed, down-regulation of MDM2 expres-sion in WI38 cells using a lentivirus expressing MDM2 shRNAshowed drastic up-regulation of the p53 target protein p21 andinefficient silencing of MDM2 (data not shown). To circumventthis problem, we used a H460 cell line that harbors WT p53 andintact p53-inducible pathway (30, 31) but lower p21 transcriptlevels than WI38 cells (Supplementary Fig. S4D). H460 cellswere infected with lentiviral vectors expressing shRNA againstMDM2 or shRNA against a nonendogenous luciferase tran-script as control. After infection for different time periods,the cells were extracted to prepare RNA and cDNA. Levelsof MDM2, cyclin A, and cyclin E transcripts were determinedby QPCR. Transcript levels of glyceraldehyde 3-phosphate de-hydrogenase (GAPDH) were used as an internal control. Ourdata show that the MDM2 transcript level decreased consider-ably (4-fold) after 6 days of infection (Fig. 4A). The decrease inthe levels of MDM2 transcript was accompanied by a remark-able (44- to 50-fold) elevation in the level of cyclin A tran-scripts (Fig. 4B) although down-regulation of MDM2 and

elevation of cyclin A were detected at an earlier time point(Supplementary Fig. S4A and B). Consistent with our flowcytometric analysis (Fig. 3E), silencing of MDM2 did not showany increase in the cyclin D (Supplementary Fig. S4C) orcyclin E transcript levels (Fig. 4C). These data indicate thatendogenous MDM2 is involved in controlling cyclin A expres-sion but not cyclin D or E. Furthermore, down-regulation ofMDM2 in H460 cells showed increase in cell cycle progress(Supplementary Table S1).

To determine the effect of silencing MDM2 expression innormal cells, we prepared extracts of early-passage primaryMEF cells isolated from isogenic normal, p53 knockout, andp53 and MDM2 double-knockout mice (32). Cell extracts weresubjected to gel electrophoresis and Western blot analysis toanalyze and compare endogenous levels of cyclin A. The re-sults of this experiment show that elimination of endogenousp53 expression elevates cyclin A levels approximately 2-foldover normal MEF (Fig. 4D and E), whereas elimination of bothp53 and MDM2 expression increases cyclin A expression5-fold over normal MEF. This experiment corroborates withour data that endogenous MDM2 negatively regulates cyclinA expression in normal cells. Consistent with this observation,p53 and MDM2 double-knockout MEF cells showed fasterproliferation than normal or p53−/− MEF cells (Fig. 4F).

FIGURE 4. Down-regulation ofMDM2 expression elevates endog-enous cyclin A levels. Endogenouslevels of MDM2 (A), cyclin A (B),and cyclin E (C) transcripts inH460 cells determined by QPCRafter infection with lentiviruses ex-pressing shRNA against luciferase(shluc) or MDM2 (shMDM2-1,shMDM2-2) for the indicated period.The bar graphs show that the tran-script levels normalized to endoge-nous GAPDH levels. The assayswere done in triplicates. The errorbars are shown. D. Cyclin A proteinlevels in isogenic primary MEF, p53−/−, and p53−/− MDM2−/− MEFcells determined by Western blotanalysis. Migration of cyclin Aand β-actin are indicated. E.Quantitation of the cyclin A bandintensities normalized to β-actinband intensit ies is shown bythe bar graph. F. Cell prolifera-tion rates of normal, p53−/−,and p53−/− MDM2−/− MEF.



1257




The SWIB Domain of MDM2 Is Responsible for Down-Regulation of Cyclin A

MDM2 interacts with several cell cycle regulators such asRb or transcription factor E2F involved in the expression ofG1-S cyclins, and domains of MDM2 involved in these inter-actions have been determined previously (33-35). These inter-actions may account for negative regulation of cyclin A byMDM2. Therefore, we determined the domain of MDM2 in-volved in the down-regulation of cyclin A. Several deletionmutants of MDM2 were tested for their ability to inhibit cyclinA expression. As described above, plasmids expressing MDM2deletion mutants were transfected into WI38 cells. Endogenouscyclin A levels in untransfected (expressing endogenousMDM2) and transfected (overexpressing the deletion mutants)

cells were estimated by flow cytometry and compared asdescribed in the earlier experiments.

Comparative immunostaining with FITC-coupled anti–cyclin A antibody (Fig. 5A, C, and E, curve 2) as opposed toa FITC-coupled IgG isotype of cyclin A antibody (Fig. 5A, C,and E, curve 1) identified the cyclin A–expressing cells. The in-tensity of FITC staining indicating endogenous cyclin A levelsin cells expressing endogenous levels of MDM2 (Fig. 5B, D,and F, curve 2) or overexpressing MDM2 deletion mutants(Fig. 5B, D, and F, curve 3) was estimated and plotted againstcell number. The expression of MDM2 or its deletion mutantswas confirmed after sorting PE- or FITC-coupled anti-MDM2antibody-labeled cells and after performing Western blot analy-sis (Supplementary Fig. S5). The results of these experiments

FIGURE 5. The cyclin A in-hibitory domain overlaps withthe SWIB domain of MDM2.Cells expressing the MDM2deletion mutant Del 491-110were detected using a PE-coup led MDM2 ant ibody(N20). Cells expressing theNH2-terminal deletion mutantswere detected using a FITC-coupled antibody (SMP14).Endogenous cyclin A wasdetected by using either aFITC-coupled (A and B) orPE- coupled (C-F) anti–cyclinA antibody. The left panel (A,C, E) shows estimation ofendogenous cyclin A levelsover background by compara-tive FITC or PE intensity ofcells stained with FITC- orPE-coupled IgG isotype ofcyclin A antibody (curve 1)and FITC- or PE-coupled cy-clin A antibody (curve 2). Theright panel (B, D, F) showscomparison of endogenouscyclin A levels in untrans-fected (endogenous MDM2,curve 2) and transfected (over-expressed deletion mutants,curve 3) cells.

Frum et al.


1258




showed that overexpression of a deletion mutant of MDM2 (Del491-110) containing the NH2-terminal 109 amino acid residuesinhibited cyclin A expression drastically in WI38 cells (Fig. 5B,curves 2 and 3). Because this domain of MDM2 is required forits interaction with p53 (7, 10), these results suggest that adomain overlapping the p53 interaction domain of MDM2is responsible for cyclin A down-regulation in WI38 cells.Because the Rb-interaction domain of MDM2 maps outsidethe NH2-terminal 109 amino acid residues (35), these data alsoexclude the possibility that interaction with Rb is needed forMDM2-mediated cyclin A down-regulation.

The p53-interaction domain of MDM2 harbors a SWIB do-main (Fig. 5), homologous to a conserved domain of a chromo-some-remodeling factor BRG1-associated protein, structurallycapable of binding to p53 at the MDM2-binding site (36).BRG1 and its associated factors are known to form a repressorcomplex that restricts cyclin A expression until released bycyclin E in a cell cycle–dependent fashion (22, 37). The tumorsuppressor p53 binds with BRG1 and several components of thisrepressor complex to up-regulate p21 expression (38, 39). TheSWIB domain of MDM2 lies within 27 to 102 amino acid resi-dues (36), which are involved in its interaction with p53. Thus,MDM2may compete with the BRG1 complex and interfere withthe p53-BRG1 interaction, releasing BRG1 and its associatedcomplex for repression of cyclin A. If MDM2-mediated cyclinA repression is a result of releasing BRG1 from p53 interactionby competitive inhibition, the deletion mutant of MDM2 thatwould not stably interact with p53 but would still retain themajorpart of the SWIB domain should be able to compete with BRG1for the common interaction site on p53 to interfere with p53-BRG1 interaction. However, deletion mutants of MDM2 lackingthe SWIB domain should not inhibit cyclin A expression.

To test the contribution of the SWIB domain in MDM2-mediated cyclin A down-regulation, we chose two NH2-terminaldeletion mutants of MDM2, one of which (Del 1-58) harbors amajor part (amino acid residues 58 to 102) of the SWIB do-main, whereas the other (Del 1-120) lacks the SWIB domain.Both mutants are incapable of making a stable interaction withp53 (7, 10). As described above, plasmids expressing theMDM2 deletion mutants were transfected into WI38 cells. En-dogenous cyclin A levels in untransfected (expressing endoge-nous MDM2) and transfected (overexpressing the deletionmutants) cells were estimated and compared. The results ofthese experiments showed that overexpression of the deletionmutant Del 1-58 that harbors part of the SWIB domain retainedthe ability to down-regulate cyclin A expression (Fig. 5D,curves 2 and 3), whereas overexpression of the deletion mutantDel 1-120 did not interfere with cyclin A expression (Fig. 5F,curves 2 and 3). These results implicate the SWIB domain ofMDM2 in cyclin A down-regulation.

MDM2 Requires the Chromosome Remodeling FactorBRG1 and the Tumor Suppressor p53 to Down-Regulate Cyclin A

If the SWIB domain of MDM2 competes with the BRG1complex for the MDM2-binding site on p53 to repress cyclinA expression, MDM2 should not be able to down-regulatecyclin A expression in the absence of BRG1 or p53. To test thispossibility, a BRG1-specific siRNA (or a control siRNA) was

cotransfected with an MDM2 expression plasmid in WI38cells. MDM2-expressing cells were immunostained with aFITC-coupled anti-MDM2 antibody; cyclin A was detectedby a rhodamine-coupled anti–cyclin A antibody; and the abilityof the BRG1 siRNA to silence BRG1 expression was determinedby immunostaining the transfected cells with a rhodamine-coupled anti-BRG1 antibody. The immunostained cells wereanalyzed by confocal microscopy as described in Materialsand Methods. The results (Fig. 6A) show that whereas over-expression of MDM2 in the presence of control RNA (scram-bled RNA) inhibited cyclin A expression (top), silencing ofBrg1 indeed released inhibition of cyclin A expression byMDM2 (bottom). Successful silencing of BRG1 is shownon the right panel. These results suggest that MDM2 requiresBRG1 to inhibit cyclin A expression.

We next determined whether MDM2 requires p53 todown-regulate cyclin A. An MDM2 expression plasmid wastransfected into normal and p53−/−MEF seeded onto glass cov-erslips. The cells were fixed and subsequently immunostainedwith a FITC-coupled anti-MDM2 antibody and rhodamine-coupled anti–cyclin A antibody. Transfected MEF cells werealso stained with anti-PCNA antibody and rhodamine-coupledsecondary antibody as a control. The results in Fig. 6B showthat whereas cyclin A is down-regulated in MDM2-expressingnormal MEF cells (top), cyclin A is coexpressed with MDM2in p53−/− MEF (bottom). Flow cytometric analysis of cyclinA expression in p53−/− MEF-overexpressing MDM2 gene-rated similar results (data not shown). These observationssuggest that although a stable MDM2-p53 interaction is notessential for the ability of MDM2 to down-regulate cyclinA (Fig. 5D), MDM2 requires p53 for this function. As inthe case of WI38 cells, MDM2 did not significantly alterPCNA levels (Supplementary Fig. S6) in MEF cells, suggest-ing that MDM2 does not inhibit PCNA expression.

The cdk4 Inhibitor p16 Cooperates with MDM2 to InhibitCyclin A Expression

Cell cycle–dependent cyclin A expression has been shownto be restricted by the cyclin D–cdk4 inhibitor p16, whichrestricts expression and activation of cyclin E–cdk2 activity.Cyclin E–cdk2 in turn releases cyclin A expression fromBrg1-Rb–mediated repression (22). In cells lacking expressionof p16, cyclin E is expressed and the cyclin E–cdk2 inhibitorp21 relocates to the cyclin D–cdk4 complex, releasing cyclinE–cdk2 activity, and, in turn, cyclin A expression by the phos-phorylation of RB and BRG1 (22). Therefore, in the absence ofp16, MDM2-mediated release of the BRG1 complex from p53may not repress cyclin A expression effectively due to the pres-ence of active cyclin E–cdk2.

To test the above hypothesis, we determined whether silenc-ing of p16 expression interferes with MDM2-mediated cyclin Adown-regulation. A p16-specific siRNA (or a scrambled siRNAsequence) was cotransfected with MDM2 expression plasmid inWI38 cells. The efficacy of the siRNAwas checked by Westernblot analysis of p16 (Fig. 6C, right). MDM2-expressingcells were immunostained with a FITC-coupled anti-MDM2antibody; cyclin A was detected by a rhodamine-coupledanti–cyclin A antibody; and the cells were analyzed by confocalmicroscopy. The results (Fig. 6C) show that overexpression of



1259




MDM2 in the presence of nonspecific scrambled RNA inhibitedcyclin A expression (Fig. 6C, si control), but not in the presenceof p16-specific siRNA (Fig. 6C, si p16). Furthermore, in a sim-ilar experiment using p16−/− MEF cells, MDM2 overexpres-sion did not lead to down-regulation of cyclin A expression inp16−/− MEF cells (Supplementary Fig. S7). These results indi-cate that silencing of p16 expression interferes with MDM2-mediated cyclin A down-regulation, suggesting involvementof p16 in MDM2-mediated regulation of cyclin A expression.

The Cyclin A Down-Regulatory Domain of MDM2 InducesG1 Arrest in Normal Diploid Cells

Because MDM2 down-regulates cyclin A expression in nor-mal diploid cells such as WI38, MRC5, or primary MEF, and

cyclin A is required for DNA replication (25, 26), we testedwhether the cyclin A down-regulatory domain of MDM2 issufficient to induce G1 arrest in normal diploid WI38 orMRC5 cells.

To determine the growth-regulatory function of the cyclin Ainhibitory domain of MDM2, WI38 or MRC5 cells were trans-fected with a plasmid expressing MDM2 Del 491-110 harbor-ing the cyclin A inhibitory domain (Fig. 5). Cells expressingDel 491-110 were detected by immunostaining with an anti-MDM2 antibody against an NH2-terminal epitope of MDM2(N20) and a FITC-coupled secondary antibody. The mock-transfected cells were immunostained similarly to estimatebackground to gate the cell population expressing Del 491-110 and were then stained with propidium iodide (Fig. 7A).

FIGURE 6. Silencing of BRG1 or p53 or p16 prevents MDM2 to down-regulate cyclin A. The figure shows confocal imaging analysis of cyclin A expressionafter MDM2 overexpression in the presence or absence of BRG1 (A), p53 (B), or p16 (C). WI38 cells nucleofected with MDM2 expression plasmid (5 μg for3 × 106 cells) and (A) 80 pmol of either control RNA (scrambled sequence) or BRG1 siRNA, and (C) 80 pmol of either control RNA (si control) or p16 siRNA(si p16). Silencing of BRG1 expression (A) after transfection with 80 pmol of BRG1 or control siRNA is shown by confocal analysis using the same settings inthe right. BRG1 expression was detected as red nuclear staining of cells by a rhodamine-coupled anti-BRG1 antibody. Silencing of p16 expression (C) isshown by Western blot analysis in the right panel. B. Normal MEF (top, MEF p53+/+) or MEF p53−/− (bottom) were nucleofected with the MDM2 expressionplasmid (5 μg for 3 × 106 cells). In each experiment, nuclei were stained with DAPI (blue). Cyclin A expression was detected as red nuclear staining of cells bya rhodamine-coupled anti–cyclin A antibody; MDM2 expression was detected as green nuclear staining by a FITC-coupled anti-MDM2 antibody. Arrows,nuclear staining of the MDM2-overexpressing cells by DAPI, rhodamine, and FITC. Staining intensities in the rhodamine channel varied from 1,668 to 927 incontrol siRNA-transfected WI38 cells, whereas the intensity in MDM2-overexpressing cells was 331 to 268 (close to background) in the fields shown.Rhodamine channel intensities in BRG1 or p16 siRNA-transfected cells varied from 1,828 to 783 independent of MDM2 overexpression in the fields shown.Similarly, rhodamine intensity in MDM2-transfected normal MEF cells were close to background (357-152), whereas p53−/− cells showed 1,484 to 1,113independent of MDM2 expression. At least 50 MDM2-expressing cells were analyzed in each experiment. Two representative fields are shown in C.

Frum et al.


1260




Flow cytometric separation of Del 491-110–expressingWI38 cells is shown in Fig. 7B. As estimated by the FITCintensity, the transfected cells showed 2- to 40-fold increasein MDM2. The cells expressing Del 491-110 were gatedand their propidium iodide intensities were plotted against cellcount to generate a DNA histogram. The DNA histogram ofthe untransfected cells expressing endogenous MDM2 wasgenerated similarly and compared with the histogram of theDel 491-110–expressing cells (Fig. 7C). Our data suggest thatDel 491-110 expression resulted in a drastic decrease(24-6.6%) in the number of cells crossing the G1- to S-phaseboundary.

Thus, the cyclin A inhibitory domain of MDM2 is capableof inducing G1 arrest in WI38 cells. Similar expression of Del491-110 in normal human diploid MRC5 cells (Fig. 7D)showed comparable growth arrest. These results indicate thatthe cyclin A inhibitory domain of MDM2 situated within theNH2-terminal 109 amino acid residues can induce G1 arrestin normal human diploid cells. Overexpression of full-lengthMDM2 induced cell cycle arrest in WI38 cells as we reportedpreviously (Supplementary Fig. S8A; ref. 14). Consistent withour previous report, MDM2 deletion mutant Del 1-120 also in-duced cell cycle arrest because it harbors p53-independentgrowth arrest function (Supplementary Fig. S8B; ref. 14).

Overexpression of Cyclin A Rescues Cells from MDM2-Mediated G1 Arrest

If MDM2-mediated cyclin A down-regulation causes G1 ar-rest, overexpression of cyclin A should rescue WI38 cells fromthe G1 arrest mediated by the cyclin A down-regulatory domainof MDM2. This was tested in WI38 cells transfected with theDel 491-110 expression plasmid in the presence or absence ofa plasmid expressing human cyclin A. Cells expressing Del491-110 were detected by a PE-labeled anti-MDM2 antibodyand gated as described previously. Levels of cyclin A expres-sion were determined by immunostaining with a FITC-coupledanti–cyclin A antibody and background staining was detectedby a FITC-coupled IgG isotype of anti–cyclin A antibody.DNAwas stained with 7-aminoactinomycin D (7-AAD). Levelsof cyclin A expression in cells expressing Del 491-110 or en-dogenous MDM2 were quantitated by the intensity of FITC–anti-cyclin A staining and was compared by single-parametercurves plotting FITC intensity against cell number (Fig. 8Aand B). Intensities of 7-AAD were plotted against cell countto generate DNA histograms of the gated cells expressingDel 491-110 or only endogenous MDM2 (Fig. 8C and D).Our data show that consistent with our observation shown inFigs. 2 and 7, cells expressing Del 491-110 alone inhibit cyclinA expression (Fig. 8A, curves 1 and 2) and induce G1 arrest as

FIGURE 6 Continued.



1261




evidenced by less number of Del 491-110–expressing cellscrossing the G1-S boundary compared with untransfected cellsexpressing endogenous MDM2 (Fig. 8C, compare curves 1 and2). When cotransfected with the cyclin A expression plasmid,WI38 cells expressing Del 491-110 also expressed cyclin A ata level comparable with the endogenous levels of cyclin A(Fig. 8B compare curves 1 and 3) and failed to show G1 arrest(Fig. 8D, compare curves 1 and 3). These data show that cyclinA expression can rescue Del 491-110–mediated G1 arrest,implicating that MDM2-mediated down-regulation of cyclinA expression leads to cell cycle arrest.

Tumor-Derived or Immortal Cells Lacking p53 or p16Often Escape the Cyclin A Down-Regulatory Function ofMDM2

Our data suggest that at least one consequence of MDM2overexpression in normal diploid cells is down-regulation ofcyclin A and G1 arrest. Furthermore, the regulators of the cellcycle–dependent cyclin A expression such as BRG1, p53, andp16 are involved in this function. It is known that deletions ofBRG1, p53, and/or p16 mutations are frequent events in cancercells (40-42). Therefore, we determined if cancer cells defectivein expressing p53, p16, or BRG1 could escape MDM2-mediateddown-regulation of cyclin A and the consequent G1 arrest.

Flow cytometric experiments, as described previously, weredone to determine how MDM2 modulates cyclin A expressionin H1299, NIH3T3, 21PT, or MCF-7 cell lines that do not ex-

press functional p53, p16, or BRG1. NIH3T3 cells harbor de-letion of the p16/p19 gene, whereas H1299 cells harbor themethylated p16 gene (16, 43-45). H1299 and 21PT cells lackfunctional p53, whereas H1299 cells do not express the func-tional BRG1 (30, 42, 46).

A representative experiment done with H1299 cells isshown in Fig. 9. H1299 cells were transfected with anMDM2 expression plasmid and immunostained with PE-coupled anti-MDM2 and FITC-coupled anti–cyclin A antibo-dies to estimate the MDM2 expression and consequent changein the levels of endogenous cyclin A. As described previously,mock transfected H1299 cells were immunostained similarly todetect endogenous levels of MDM2 and cyclin A, and FITC-coupled isotype of cyclin A was used to determine backgroundimmunostaining.

Flow cytometric measurement of PE intensity showed thattransfected H1299 cells expressed MDM2 efficiently, with ma-jority of the cells expressing 4- to 50-fold MDM2 over back-ground levels (Fig. 9A). Comparative immunostaining withFITC-coupled anti–cyclin A antibody and FITC-coupled IgGisotype identified the endogenous levels of cyclin A in mock-transfected H1299 cells (Fig. 9B, curves 1 and 2). However,the levels of cyclin A expression in MDM2-overexpressingH1299 (Fig. 9C) cells did not differ significantly from the re-spective untransfected cells expressing endogenous levels ofMDM2, suggesting that MDM2 overexpression did not altercyclin A expression in these cells. Immortal NIH3T3 or breast

FIGURE 7. The cyclin A inhibi-tory domain of MDM2 is capable ofinducing G1 arrest in WI38 orMRC5 cells. Dot plot analysis ofmock-transfected (A) and MDM2expression plasmid-transfected(B) WI38 cells. The boundary ofthe mock-transfected cells wasdrawn to achieve 0.1% back-ground contaminat ion. FITC-labeled transfected cells (MDM2Del 491-110–expressing) and un-transfected cells with endogenousMDM2 expression (Background)are indicated. Comparison of theDNA histograms of transfected(Del 491-110 expressing) and un-transfected (endogenous MDM2)WI38 (C) and MRC5 (D) cells areshown. Because the number oftransfected cells was much lowerthan untransfected cells, enoughcells were analyzed to collect10,000 successfully transfectedcells, and the G0-G1 peak hasbeen normalized to the sameheight in the figure for comparison.Number of cells in G0-G1, S, andG2-M phases is shown in the bot-tom panel.

Frum et al.


1262




cancer–derived 21PT or MCF-7 cells showed similar results(ref. 14; data not shown). This observation suggests thatMDM2 fails to inhibit cyclin A expression in immortal or can-cer cells that do not express functional p53, BRG1, or p16.

To test whether the cyclin A inhibitory domain of MDM2can induce G1 arrest in H1299 cells, the deletion mutant ofMDM2, Del 491-110, was overexpressed in H1299 cells. Thetransfected (Del 491-110–expressing) and untransfected (en-dogenous MDM2) cells were sorted as described in the caseof WI38 cells, and their DNA histograms were generated. Un-like WI38 or MRC5 cells, overlay of the histograms of Del491-110 expressing and untransfected cells (expressing endog-enous MDM2) show that Del 491-110 did not induce G1 arrestin H1299 cells (Fig. 9D). Similar results were obtained inNIH3T3, 21PT, and MCF-7 cells (14; data not shown). Consis-tent with the observation that silencing of p53, BRG1, or p16attenuates MDM2-mediated cyclin A down-regulation in nor-mal diploid cells, attenuation of the cyclin A inhibitory functionof MDM2 in cancer cells defective in expressing p53, BRG1,and p16 argues that cancer cells can escape the MDM2-mediatedgrowth arrest that constitutes an important safety barrier innormal cells. Consistent with the data shown in Fig. 9 and inour previous report (14), H1299 cells exhibited an inefficientcell cycle arrest mediated by full-length MDM2 compared withWI38 cells (Supplementary Table S2).

DiscussionIn the present report, we show experimental evidences that

MDM2 inhibits expression of cyclin A in apparently normalhuman diploid cells such as WI38 (Figs. 1 and 2), MRC5 (datanot shown), or primary MEF cells (Fig. 6B). MDM2-mediatedcyclin A down-regulation prevents onset of the S phase in suchcells (Fig. 7). Because MDM2 does not prevent expression ofcyclin E (Figs. 3 and 4) or D (Supplementary Fig. S4D),

MDM2-mediated cyclin A down-regulation is a specific eventand not an artifact in which all G1 cyclin expression is inhibitednonspecifically. Ectopic expression of cyclin A rescues the G1

arrest mediated by the cyclin A inhibitory domain of MDM2(Fig. 8). We further show that down-regulation of MDM2 ex-pression remarkably elevates endogenous levels of cyclin A(Fig. 4). These observations indicate that MDM2 is involvedin regulating cyclin A expression.

Analysis of the functional domain of MDM2 indicated that aSWIB region, homologous to a conserved domain of a BRG1-associated factor (36), situated within the p53-binding domainof MDM2 was required to inhibit cyclin A expression (Fig. 5).This observation suggests that MDM2 may compete with theBRG1 complex for p53 binding, releasing BRG1 for its repres-sor function. Our data (Fig. 6A and B) show that MDM2 in-deed requires both p53 and BRG1 to down-regulate cyclin A.

In the absence of the cdk4 inhibitor p16, active cyclin D–cdk4 complex releases cyclin E–cdk2 activity to phosphorylateRb and Brg1 (22), attenuating their repressor function. Thus,the absence of p16 should also attenuate the repressor functionof BRG1 after its release from p53 caused by MDM2 overex-pression. Consistent with this model (22), our data also showthat silencing of p16 expression rescues MDM2-mediateddown-regulation of cyclin A expression (Fig. 6C). Conversely,silencing of MDM2 should encourage p53-BRG1 interactionby vacating the competitive binding site on p53. Therefore,up-regulation of cyclin A by silencing MDM2 did not requirethe presence of p16. We were able to detect transient up-regulation of cyclin A after silencing MDM2 in H460 cells thatdo not express p16 (Fig. 4A and B). Our data show thatalthough silencing of MDM2 up-regulates p21 expression, itcannot inhibit cyclin A expression efficiently when MDM2 isknocked down (Supplementary Fig. S4D). This suggests re-quirement of MDM2 expression for cyclin A down-regulation.Sustained silencing of MDM2, leading to stabilization of p53,

FIGURE 8. Ectopic expression of cy-clin A rescues WI38 cells from G1 arrestresulting from MDM2-mediated cyclin Adown-regulation. WI38 cells (5 × 105-6 × 105) were transfected with 10 μgDel 491-110 expression plasmid inthe presence or absence of 10 μgcyclin A expression plasmid. Del 491-110–expressing cells were detected bya PE-coupled anti-MDM2 antibody; cy-clin A expression was detected by aFITC-coupled anti–cyclin A antibody;and DNA was stained with 7-AAD asdescribed in Materials and Methods.FITC- and 7-AAD intensity of gatedtransfected (expressing Del 491-110 orDel 491-110 and cyclin A) and untrans-fected (endogenous MDM2 and cyclinA) cells were determined. Comparisonof cyclin A levels (A, B) and DNA histo-grams (C, D) of cells overexpressingDel 491-110 (A, C), or Del 491-110and cyclin A (B, D) with untransfectedcells expressing endogenous levels ofMDM2 and cyclin A, are shown.



1263




quenches MDM2 silencing and cyclin A up-regulation eventu-ally (Fig. 4B, days 8 and 10).

We also found that p53−/− MDM2−/− (a situation similarto silencing of MDM2 in cells with no p53 expression) early-passage MEF cells express higher levels of cyclin A comparedwith MEF or p53−/− MEF cells, suggesting increase in cyclinA expression in the absence of p53 and MDM2 (Fig. 4D andE). These data suggest that interaction of MDM2 with p53 fam-ily members may also have a role in down-regulation of cyclinA. It has been reported that MDM2 up-regulates p63-mediatedactivation of p21 promoter (47), which could lead to down-regulation of cyclin A expression in the absence of p53. Silenc-ing of MDM2 would reverse this effect.

Based on our results, we propose a novel pathway (Fig. 10)that overexpression of MDM2 discourages p53-BRG1 interac-tion through the SWIB domain of MDM2. As reported in theliterature, BRG1 released from p53 functions as a repressor ofcyclin A expression. This pathway may be important after on-cogenic challenges such as radiation-induced DNA damage thatstabilizes p53 and up-regulates MDM2. Inhibition of p21 ex-pression by MDM2 overexpression under such conditionswould be circumvented by release of BRG1 to prevent cyclinA expression and progress of the cell cycle. The repressorfunction of BRG1 and thus MDM2-mediated cyclin A down-

regulation would be discouraged in cell lines that lack p16expression.

Consistent with our earlier observations (14), we found thatthe cancer cells that lack expression of functional p53, BRG1,or p16 escape the cyclin A inhibitory function of MDM2. Thecell lines that are insensitive to the cyclin A inhibitory functionof MDM2 are also resistant to the G1 arrest mediated by its cy-clin A inhibitory domain (Fig. 9; ref. 14). Because p16 is fre-quently deleted or methylated in cancer cells (40), thisobservation explains the insensitivity of cancer cells toMDM2-mediated cyclin A down-regulation and G1 arrest. Con-sistent with this argument, overexpression of cyclin A is a fre-quent event in cancer cells (16, 48, 49).

It has been shown previously that MDM2−/− mouse cannotsurvive because of p53 accumulation (reviewed in ref. 3). Re-cently, Itahana et al. (50) have shown that a single amino acidmutation (C462A) at the ubiquitin ligase domain of MDM2produces a similar lethal effect, indicating that the amino acidresidues at 462 is required for controlling the lethal effect ofp53 accumulation during embryogenesis. This suggests that na-ture would avoid a MDM2−/− or MDM2 C462A mutation.

Although interaction of MDM2 with p53 and its ubiquitina-tion are required to prevent cell death, we report a novel p53-dependent cell cycle regulatory function of MDM2 in which

FIGURE 9. MDM2 doesnot inhibit cyclin A expressionin H1299 lung cancer cells.H1299 cells (5 × 105-6 × 105)were transfected with 10 μgMDM2 expression plasmid (ormock transfected) as indicatedin Materials and Methods. A.MDM2 overexpression intransfected cells over endoge-nous levels was estimatedby PE intensity plotted in asingle-parameter histogram.B. Estimation of endogenouscyclin A level over backgroundare shown by comparativeFITC intensity of cells stainedwith FITC-coupled IgG isotypeof cyclin A antibody (curve 1)and FITC-coupled cyclin A an-tibody (curve 2). C. Compari-son of endogenous cyclin Aexpression in cells with endog-enous levels of MDM2 (curve2) and MDM2-overexpressingcells (curve 3) estimated byFITC intensity. D. Comparisonof the DNA histograms oftransfected (Del 491-110 over-expressing) and untransfected(endogenous MDM2) cells.Sufficient cells were analyzedto collect more than 10,000successfully transfected cells,and the G0-G1 peak has beennormalized to the same heightin the figure for the sake ofcomparison. Number of cellsin G0-G1, S, and G2-M phasesis shown in the bottom panel.

Frum et al.


1264




MDM2 releases BRG1 by competing with p53, and a stableMDM2-p53 interaction is dispensable for restricting cyclin Aexpression. Our present data show that in normal cells,MDM2 is involved in regulating cell cycle by restricting cyclinA expression in cooperation with p53, p16, and BRG1. Duringan oncogenic challenge, induction of p53 would induce G1

arrest and apoptosis; however, it would also up-regulateMDM2. Our data signifies that MDM2-mediated cyclin Adown-regulation would provide an added level of protectionfrom the oncogenic consequences of MDM2 overexpression.Cancer cells that lack p53, p16, or BRG1 escape this cyclinA and cell cycle restricting function of MDM2.

Our identification of a p53-dependent G1 arrest functionof MDM2 is consistent with the finding of a recent report(51). In recent years, several oncoproteins have been shownto possess a growth arrest function (52). Activated Ras in-duces senescence (53). Raf-1 induces growth arrest that isovercome in immortal cells (54). These observations suggestthat normal cells are equipped to handle overexpression ofgenes that may harbor growth proliferative properties. Thecancer cells that overexpress these oncoproteins have ac-quired genetic damages to escape the growth arrest functionsof the oncoproteins.

Materials and MethodsPlasmids and MDM2 Deletion Mutants

The MDM2 cDNA and the cyclin A expression plasmidswere generous gifts from Bert Vogelstein (24) and RobertWeinberg (55). Construction of plasmids expressing MDM2has been described in detail (7, 10). The deletion mutantMDM2 Del 491-110 was generated by PCR by using primersflanking initiation and termination codons and inserting a NheIlinker with stop codons in all three reading frames. The mutantwas cloned in pGEM3zf (−) vector, sequenced, and subclonedin pCMV plasmid (7, 14) for in vivo expression.

Cells and TransfectionsWI38, MRC-5, H1299, NIH3T3, and MCF-7 cells were pur-

chased from the American Type Culture Collection and weremaintained in medium as suggested by the supplier. Normal,p53−/−, and p53−/− MDM2−/− MEF cells derived from iso-genic mice were kind gifts from Guillermina Lozano (32),and were cultured in DMEM in the presence of 10% fetal bo-vine serum. For transfection, cells were seeded 18 to 24 h beforetransfection at 5 × 105 to 6 × 105 cells per 10-cm dish, andplasmids or siRNAs were transfected either by the calciumphosphate method (10) or nucleofection using a nucleofectorkit from Amaxa Biosystem, using the supplier's protocol.Thirty to 48 h after transfection, the cells were collected andfixed. Control (5′-CATGTCATGTGTCACATCTC-3′), BRG1siRNA (5′-AACAUGCACCAGAUGCACAAGTT-3′), or p16siRNA (5′-TAACCATGCCCGCATAGAT-3′) were used.

Generation of Lentivirus Vectors and InfectionPlasmids (pLKO.1) expressing short hairpin (sh) RNA

against MDM2 from U6 promoter and harboring puromycin re-sistance gene were purchased from Open Biosystem and testedfor their ability to lower MDM2 expression. The control plas-mid expresses shRNA against nonendogenous luciferase gene.The lentivirus vectors were generated following a protocol pro-vided by the supplier. For silencing, the cells were infected withrespective lentiviruses for increasing time periods to determineoptimum silencing.

Western Blot AnalysisWestern blot analysis was carried out as described pre-

viously (10).

Immunofluorescent Staining and Confocal ImagingImmunofluorescent staining and confocal imaging were

done following a method reported by Lohrum et al. (56).WI38 cells (2 × 105) were seeded on sterile coverslips in asix-well culture dish and were transfected with 2 μg MDM2

FIGURE 10. Proposed mechanism for MDM2-mediated down-regulation of cyclin A presented bya diagram. The BRG1 repressor complex is shownby BRG1. Evidence for phosphorylation of BRG1by cdks in the presence or absence of p16 hasbeen published in the literature.



1265




expression plasmid. The cells were fixed 48 h after transfectionusing 4% paraformaldehyde, permeabilized in ice-cold PBScontaining 0.2% Triton X-100, and blocked in PBS containing0.5% bovine serum albumin (BSA). The fixed cells were con-secutively immunostained overnight at 4°C with a FITC-coupled anti-MDM2 antibody (N20, Santa Cruz Biotechnology)followed by overnight staining with a rhodamine-coupled anti–cyclin A antibody (SC-751 Tritc, Santa Cruz Biotechnology) inblocking solution. The slides were then washed, air dried, andmounted with Prolong Gold Antifade with 4′,6′-diamidino-2-phenylindole hydrochloride (DAPI, Molecular Probes) andanalyzed under a confocal microscope (Zeiss) at 40× magnifica-tion using a LSM image browser software (Zeiss).

Flow CytometryMethods for cell cycle analysis have been described in detail

(14-16). Transfected cells were harvested 30 to 48 h after trans-fection and fixed with 70% ethanol overnight at 4°C. The fixedcells were incubated with appropriate dye-coupled antibody inPBS, 0.5% BSA, and 0.5% Tween 20. Cells were then washedthrice in PBS and 0.5% BSA. To stain DNA with 7-AAD, thecells were incubated with 20 μg/mL 7-AAD, PBS, 0.5% BSA,and 0.5% Tween 20 buffer for 90 min at 0°C in the dark. Thesamples were analyzed in a fluorescence activated cell sorter(Elite, Coulter or FAC Starplus, Becton Dickinson) in a flow cy-tometry core facility. The dye-coupled antibodies were fromSanta Cruz. PE-coupled N20 was prepared using a PE conjuga-tion kit (Phycolink) from Prozyme.

After immunostaining with fluorescent dye-coupled anti-bodies, relative levels of MDM2 or cyclins were determinedby fluorescence intensity in the orange (575 nm) channel forPE, and green channel (525 nm) for FITC. Background fluores-cence was determined by immunostaining of mock-transfectedcells with FITC- or PE-coupled anti-MDM2 antibody andFITC- or PE-coupled IgG isotype of the respective cyclin anti-body. The intensity of 7-AAD staining was recorded in the red(675 nm) channel. A minimum of 8,000 MDM2-overexpressing(transfected) cells was analyzed in each experiment. Experi-ments were repeated at least three times.

To ensure that the flow cytometric data reasonably reflectcyclin A levels, we incubated serum-starved WI38 cells withFITC-coupled anti–cyclin A antibody. Normally growing cellswere also fixed and stained with FITC-coupled anti–cyclin Aantibody; the cyclin A–expressing cells and cells showing basalexpression were gated and analyzed by flow cytometry. Our da-ta show that the serum-starved cells aligned with the basal levelpeak and did not have the cyclin A–expressing peak (Supple-mentary Fig. S9).

RNA Extraction Generation of cDNA and QPCRTotal RNA was isolated from exponentially growing cul-

tured cell lines using the TRIzol reagent (Life Technologies,Invitrogen) following a protocol supplied by the manufacturer.The quality of RNAwas checked by 1.2% agarose Tris-borate-EDTA gel electrophoresis. cDNA was synthesized using theThermoscript reverse transcription-PCR system (Invitrogen).QPCR was done using a LightCycler system (Roche). Primerswere designed using OLIGO 5 software (Molecular Biology In-sights) and were synthesized by Sigma Genosys. Reactions

were done in triplicate using SYBR green dye, which exhibitsa higher fluorescence upon binding of double-strandedDNA. The methods have been described previously (57). TheQPCR primers used were as follows: (a) MDM2, 5′-CCCAA-GACAAAGAAGAGAGTGTGG-3 ′ and 5 ′-CTGGGCA-GGGCTTATTCCTTTTCT-3 ′ ; (b) GAPDH, 5 ′-GTCA-ACGGATTTGGTCGTATT-3′ and 5′-GATCTCGCTCCTG-GAAGATGG-3′; (c) human cyclin A, 5′-GACGGCGCTCCAA-GAGG-3′ and 5′-AATGGTGAACGCAGGCTGTT-3′; (d)human cyclin E, 5′-AGGTTTCAGGGTATCAGTGG-3′ and5′-TTTGCTCGGGCTTTGT-3′; (e) human p21, 5′-TTAG-CAGCGGAACAAGGAGT-3′ and 5′-AGCCGAGAGAAAA-CAGTCCA-3′; and (f) human cyclin D1, 5′-AATGACCCC-GCACGATTTCA-3′ and 5′-CTCCCCGCTGCCACCAT-3′.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

AcknowledgmentsWe thank Dr. Bert Vogelstein (Howard Hughes Medical Institute and the SidneyKimmel Comprehensive Cancer Center, Johns Hopkins Medical Institute, Balti-more, MD) for the MDM2 cDNA; Dr. Robert Weinberg (Whitehead Institute forBiomedical Research, MIT, Cambridge, MA) for cyclin A expression plasmid;Dr. Arnold Levine (The Institute for Advanced Studies, Princeton, NJ) for themonoclonal antibody 2A10; Dr. Guellermina Lozano (Department of MolecularGenetics, University of Texas M.D. Anderson Cancer Center, Houston, TX) fornormal, p53−/−, and p53−/− MDM2−/− MEF; Dr. Vimla Band (Department ofGenetics, Cell Biology and Anatomy, Tuft University, Boston, MA) for the 21PTcell line; and Drs. M.J. Tevethia (Department of Microbiology and Immunology,Pennsylvania State University College of Medicine, Hershey, PA) and BradWindle (Department of Medicinal Chemistry, Virginia Commonwealth Universi-ty, Richmond, VA) for reviewing the manuscript.

References1. Juven-Gershon T, Oren M. Mdm2: the ups and downs. Mol Med 1999;5:71–83.

2. Deb SP. Function and dysfunction of the human oncoprotein MDM2. FrontBiosci 2002;7:d235–43.

3. Iwakuma T, Lozano G. MDM2, an introduction. Mol Cancer Res 2003;1:993–1000.

4. Fakharzadeh SS, Trusko SP, George DL. Tumorigenic potential associatedwith enhanced expression of a gene that is amplified in a mouse tumor cell line.EMBO J 1991;10:1565–9.

5. Finlay CA. The mdm-2 oncogene can overcome wild-type p53 suppression oftransformed cell growth. Mol Cell Biol 1993;13:301–6.

6. Momand J, Zambetti GP, Olson DC, George D, Levine AJ. The mdm-2 onco-gene product forms a complex with the p53 protein and inhibits p53-mediatedtransactivation. Cell 1992;69:1237–45.

7. Brown DR, Deb S, Munoz RM, Subler MA, Deb SP. The tumor suppressorp53 and the oncoprotein simian virus 40 T antigen bind to overlapping domainson the MDM2 protein. Mol Cell Biol 1993;13:6849–57.

8. Chen J, Marechal V, Levine AJ. Mapping of the p53 and mdm-2 interactiondomains. Mol Cell Biol 1993;13:4107–14.

9. Oliner JD, Pietenpol JA, Thiagalingam S, Gyuris J, Kinzler KW, Vogelstein B.Oncoprotein MDM2 conceals the activation domain of tumour suppressor p53.Nature 1993;362:857–60.

10. Leng P, Brown DR, Shivakumar CV, Deb S, Deb SP. N-terminal 130 aminoacids of MDM2 are sufficient to inhibit p53-mediated transcriptional activation.Oncogene 1995;10:1275–82.

11. Haupt Y, Maya R, Kazaz A, Oren M. Mdm2 promotes the rapid degradationof p53. Nature 1997;387:296–9.

12. Honda R, Tanaka H, Yasuda H. Oncoprotein MDM2 is a ubiquitin ligase E3for tumor suppressor p53. FEBS Lett 1997;420:25–7.

13. Kubbutat MH, Jones SN, Vousden KH. Regulation of p53 stability byMdm2. Nature 1997;387:299–303.

14. Brown DR, Thomas CA, Deb SP. The human oncoprotein MDM2 arrests the

Frum et al.


1266




cell cycle: elimination of its cell-cycle-inhibitory function induces tumorigenesis.EMBO J 1998;17:2513–25.

15. Frum R, Deb SP. Flow cytometric analysis of MDM2-mediated growtharrest. Methods Mol Biol 2003;234:257–67.

16. Zhou R, Frum R, Deb S, Deb SP. The growth arrest function of the humanoncoprotein mouse double minute-2 is disabled by downstream mutation incancer cells. Cancer Res 2005;65:1839–48.

17. Lundgren K, Montes de Oca Luna R, McNeill YB, et al. Targeted expressionof MDM2 uncouples S phase from mitosis and inhibits mammary gland devel-opment independent of p53. Genes Dev 1997;11:714–25.

18. Schuster K, Harris LC. Selection for mutations in the cDNAs of transgenicmice upon expression of an embryonic lethal protein. Transgenic Res 2007;16:527–30.

19. Bartel F, Harris LC, Wurl P, Taubert H. MDM2 and its splice variant mes-senger RNAs: expression in tumors and down-regulation using antisense oligo-nucleotides. Mol Cancer Res 2004;2:29–35.

20. Obaya AJ, Sedivy JM. Regulation of cyclin-Cdk activity in mammalian cells.Cell Mol Life Sci 2002;59:126–42.

21. Pajalunga D, Mazzola A, Franchitto A, Puggioni E, Crescenzi M. The logicand regulation of cell cycle exit and reentry. Cell Mol Life Sci 2008;65:8–15.

22. Zhang HS, Gavin M, Dahiya A, et al. Exit from G1 and S phase of the cellcycle is regulated by repressor complexes containing HDAC-Rb-hSWI/SNF andRb-hSWI/SNF. Cell 2000;101:79–89.

23. Crawford YG, Gauthier ML, Joubel A, et al. Histologically normal humanmammary epithelia with silenced p16(INK4a) overexpress COX-2, promoting apremalignant program. Cancer Cell 2004;5:263–73.

24. Oliner JD, Kinzler KW, Meltzer PS, George DL, Vogelstein B. Amplificationof a gene encoding a p53-associated protein in human sarcomas. Nature 1992;358:80–3.

25. Girard F, Strausfeld U, Fernandez A, Lamb NJ. Cyclin A is required for theonset of DNA replication in mammalian fibroblasts. Cell 1991;67:1169–79.

26. Pagano M, Pepperkok R, Verde F, Ansorge W, Draetta G. Cyclin A is re-quired at two points in the human cell cycle. EMBO J 1992;11:961–71.

27. Midgley CA, Lane DP. p53 protein stability in tumour cells is not deter-mined by mutation but is dependent on Mdm2 binding. Oncogene 1997;15:1179–89.

28. Chen L, Lu W, Agrawal S, Zhou W, Zhang R, Chen J. Ubiquitous inductionof p53 in tumor cells by antisense inhibition of MDM2 expression. Mol Med1999;5:21–34.

29. Stommel JM, Wahl GM. Accelerated MDM2 auto-degradation induced byDNA-damage kinases is required for p53 activation. EMBO J 2004;23:1547–56.

30. Mitsudomi T, Steinberg SM, Nau MM, et al. p53 gene mutations in non-small-cell lung cancer cell lines and their correlation with the presence of rasmutations and clinical features. Oncogene 1992;7:171–80.

31. Corn PG, El-Deiry WS. Microarray analysis of p53-dependent geneexpression in response to hypoxia and DNA damage. Cancer Biol Ther 2007;6:1858–66.

32. Montes de Oca Luna R, Wagner DS, Lozano G. Rescue of early embryoniclethality in mdm2-deficient mice by deletion of p53. Nature 1995;378:203–6.

33. Martin K, Trouche D, Hagemeier C, Sorensen TS, La Thangue NB, KouzaridesT. Stimulation of E2F1/DP1 transcriptional activity by MDM2 oncoprotein. Nature1995;375:691–4.

34. Xiao ZX, Chen J, Levine AJ, et al. Interaction between the retinoblastomaprotein and the oncoprotein MDM2. Nature 1995;375:694–8.

35. Sdek P, Ying H, Zheng H, et al. The central acidic domain of MDM2 iscritical in inhibition of retinoblastoma-mediated suppression of E2F and cellgrowth. J Biol Chem 2004;279:53317–22.

36. Bennett-Lovsey R, Hart SE, Shirai H, Mizuguchi K. The SWIB and the

MDM2 domains are homologous and share a common fold. Bioinformatics2002;18:626–30.

37. Kang H, Cui K, Zhao K. BRG1 controls the activity of the retinoblastomaprotein via regulation of p21CIP1/WAF1/SDI. Mol Cell Biol 2004;24:1188–99.

38. Lee D, Kim JW, Seo T, Hwang SG, Choi EJ, Choe J. SWI/SNF complexinteracts with tumor suppressor p53 and is necessary for the activation of p53-mediated transcription. J Biol Chem 2002;277:22330–7.

39. Wang M, Gu C, Qi T, et al. BAF53 interacts with p53 and functions in p53-mediated p21-gene transcription. J Biochem 2007;142:613–20.

40. Ruas M, Peters G. The p16INK4a/CDKN2A tumor suppressor and its rela-tives. Biochim Biophys Acta 1998;1378:F115–77.

41. Reisman DN, Sciarrotta J, Wang W, Funkhouser WK, Weissman BE. Loss ofBRG1/BRM in human lung cancer cell lines and primary lung cancers: correla-tion with poor prognosis. Cancer Res 2003;63:560–6.

42. Medina PP, Carretero J, Ballestar E, et al. Transcriptional targets of thechromatin-remodelling factor SMARCA4/BRG1 in lung cancer cells. Hum MolGenet 2005;14:973–82.

43. Linardopoulos S, Street AJ, Quelle DE, et al. Deletion and altered regulationof p16INK4a and p15INK4b in undifferentiated mouse skin tumors. Cancer Res1995;55:5168–72.

44. Quelle DE, Ashmun RA, Hannon GJ, et al. Cloning and characterization ofmurine p16INK4a and p15INK4b genes. Oncogene 1995;11:635–45.

45. Kataoka M, Wiehle S, Spitz F, Schumacher G, Roth JA, Cristiano RJ. Down-regulation of bcl-2 is associated with p16INK4-mediated apoptosis in non-smallcell lung cancer cells. Oncogene 2000;19:1589–95.

46. Band V, Zajchowski D, Swisshelm K, et al. Tumor progression in four mam-mary epithelial cell lines derived from the same patient. Cancer Res 1990;50:7351–7.

47. Calabro V, Mansueto G, Parisi T, Vivo M, Calogero RA, La Mantia G. Thehuman MDM2 oncoprotein increases the transcriptional activity and the proteinlevel of the p53 homolog p63. J Biol Chem 2002;277:2674–81.

48. Keyomarsi K, Pardee AB. Redundant cyclin overexpression and gene ampli-fication in breast cancer cells. Proc Natl Acad Sci U S A 1993;90:1112–6.

49. Dutta A, Chandra R, Leiter LM, Lester S. Cyclins as markers of tumorproliferation: immunocytochemical studies in breast cancer. Proc Natl Acad SciU S A 1995;92:5386–90.

50. Itahana K, Mao H, Jin A, et al. Targeted inactivation of Mdm2 RING fingerE3 ubiquitin ligase activity in the mouse reveals mechanistic insights into p53regulation. Cancer Cell 2007;12:355–66.

51. Giono LE, Manfredi JJ. Mdm2 is required for inhibition of Cdk2 activity byp21, thereby contributing to p53-dependent cell cycle arrest. Mol Cell Biol 2007;27:4166–78.

52. Di Micco R, Fumagalli M, d'Adda di Fagagna F. Breaking news: high-speedrace ends in arrest-how oncogenes induce senescence. Trends Cell Biol 2007;17:529–36.

53. Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic rasprovokes premature cell senescence associated with accumulation of p53 andp16INK4a. Cell 1997;88:593–602.

54. Olsen CL, Gardie B, Yaswen P, Stampfer MR. Raf-1-induced growth arrestin human mammary epithelial cells is p16-independent and is overcome in im-mortal cells during conversion. Oncogene 2002;21:6328–39.

55. Hinds PW, Mittnacht S, Dulic V, Arnold A, Reed SI, Weinberg RA. Regu-lation of retinoblastoma protein functions by ectopic expression of human cyclins.Cell 1992;70:993–1006.

56. Lohrum MA, Ludwig RL, Kubbutat MH, Hanlon M, Vousden KH. Regula-tion of HDM2 activity by the ribosomal protein L11. Cancer Cell 2003;3:577–87.

57. Scian MJ, Stagliano KE, Anderson MA, et al. Tumor-derived p53 mutantsinduce NF-κB2 gene expression. Mol Cell Biol 2005;25:10097–110.



1267




Published OnlineFirst August 11, 2009.Mol Cancer Res Rebecca Frum, Mahesh Ramamoorthy, Lathika Mohanraj, et al. Regulate the Cell CycleMDM2 Controls the Timely Expression of Cyclin A to

Updated version

10.1158/1541-7786.MCR-08-0334doi:

Access the most recent version of this article at:

Material

Supplementary

http://mcr.aacrjournals.org/content/suppl/2009/08/11/1541-7786.MCR-08-0334.DC1

Access the most recent supplemental material at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://mcr.aacrjournals.org/content/early/2009/08/07/1541-7786.MCR-08-0334To request permission to re-use all or part of this article, use this link



http://mcr.aacrjournals.org/lookup/doi/10.1158/1541-7786.MCR-08-0334

http://mcr.aacrjournals.org/content/suppl/2009/08/11/1541-7786.MCR-08-0334.DC1

http://mcr.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mcr.aacrjournals.org/content/early/2009/08/07/1541-7786.MCR-08-0334


MDM2 Controls the Timely Expression of Cyclin A to ...€¦ · 07/08/2009 · kinase (cdk) inhibitor p16 prevents MDM2-mediated inhibition of cyclin A expression, implicating its

Documents