Lovastatin Induction of Cyclin-Dependent Kinase Inhibitors in Human Breast Cells Occurs in a Cell Cycle-Independent Fashion1

CANCER RESEARCH 57. 604-609. February 15. I997J

Advances in Brief

Lovastatin Induction of Cyclin-dependent Kinase Inhibitors in Human Breast Cells Occurs in a Cell Cycle-independent Fashion1

Julie Gray-Bablin, Sharmila Rao, and Khandan Keyomarsi2

Division of Molecular Medicine, Laboratory of Diagnostic Oncology, Wadsworth (â€˜enter,Albany,New York 12201

Abstract dolichol production. As a result the pleiotropic cellular effects of lovastatin are not thoroughly understood or characterized. Lovastatin Cydin-dependent kinase inhibitors (CKIs) p2!, p27, p16, and p15 are has been used recently to arrest diverse cell types in both mitotic an essentialand integral part of cell cycleregulation.Studieson the (11â€”13)andmeioticcellcycles(14).Hengsteta!.(9)reportedonthe expression of these inhibitors in normal versus tumor human breast induction of p27 in cells synchronized via lovastatin treatment. More cancer cells revealed that although p27 and p16 are expressed at higher levelsintumorcells,p2! andp15expressionwerehigherin normalcells. recently, it was demonstrated that this induction, rather than being Analysisonthe expressionpatternof theseproteinsthroughoutthecell transcriptionally mediated, is at least partially attributable to transla cyclein synchronizedcellsdemonstratedasubstantialincreasein p21 tional control of p27 (10). The induction of p27 by lovastatin was during the S-phasein normal cellsand barely detectableexpressionof p21 assumed to be a cell cycle effect rather than a drug-specific effect. in anyphaseofthetumorcellcycle.Levelsofp15,p16,andp27remained Here, we report on the cell cycle expression of four CKIs of both the relatively constant throughout the cell cycle of normal and tumor cells. p21 and p16 families and demonstrate that their expression is modu Synchronizationoftumor cellsby lovastatin,whicharrestscellsin G1, lated by lovastatin via a cell cycle-independent mechanism. Synchro resulted in increased levels of p2! and p27 with a concomitant decrease in nization of human breast cells by means other than lovastatin treat cyclin-dependentkinase2-associatedkinaseactivity.Synchronizationof ment did not lead to the induction of any of the CKIs examined. cellsby double-thymidineblockdid not resultin the inductionofp21 or p27. TheseobservationssuggestthatlovastatinCaUSeSaprofoundcell Materials and Methods cycle-independentalteration of CM expressionwhich is distinct from growthfactordeprivationor thymidineblock. Materials, Cell Lines, and Culture Conditions. [methyl-3H]thymidine (81 Ci/mmol) and [a-32P]dCTP(3000Ci/mmol) were purchasedfromNew Introduction EnglandNuclear (Boston,MA). Lovastatinwas kindly providedby A. W. Alberts (Merck, SharpandDohmeResearchPharmaceuticals,Rahway,NJ). Cell proliferation, the ability of cells to traverse the cell cycle, is MevalonicacidlactoneandserumwerepurchasedfromSigmaChemicalCo. intricately regulated in normal cells by the coordinate activity of both (St.Louis,MO) andcell culturemediumfrom Life Technologies,Inc.(Grand positive and negative regulating proteins. However, tumor cells have Island, NY). All other chemicals used were of reagent grade. The scintillation defective cell cycle control mechanisms, resulting in the uncontrolled cocktailusedwasBudget-SolvefromResearchProductsInternational(Mount growth and proliferation characteristic of all cancers.The cell cycle is Prospect, IL). Before addition to cultures, lovastatin and mevalonic acid were driven forward by complexes of stable kinases termed cdks3 and converted from their inactive lactone prodrug form to their active dihydroxy unstable regulatory subunits called cyclins (1â€”4).Anadditional layer openacidasdescribedpreviously(12).The cultureconditionsfor 70N, 8lN, and76N normalcell strains,MCF-lOA cell line, andMCF-7,MDA-MB-l57, of cell cycle regulation has emerged with the discoveries of low MDA-MB-23l, MDA-MB-436, T47D, BT-20, HBL100, HsS78T, SKBR3, molecular weight CKIs which represent a novel mode of negative andZR7STtumorcell linesweredescribedpreviously(15,16).All cellswere regulation (5â€”7).Thereare two classesof CKIs: the CIPIKIP family culturedandtreatedat 37Â°Cina humidified incubatorcontaining6.5%CO2 @ including and @27Ki@Iandthe INK4 family including and maintained free of Mycoplasma as determined by the MycoTect kit (Life pl5@@4B and pl6@'@4â€•reviewedelsewhere (8). Both classes of Technologies,Inc.). inhibitors function to block the activity of cdks. p21 and p27 inhibit Synchronization and Flow Cytometry. Synchronization by lovastatin cdk2â€”5andcdk6, whereasp15 and p16 inhibit cdk4 and cdk6 (8). treatment or growth factor deprivation was performed as described previously CKI levels have recently been examined during the cell cycle by (12). Briefly, medium was removed 36â€”48 h after the initial plating of synchronization of cells with the drug lovastatin (9, 10), which is MDA-MB-l57 cellsandreplacedwithfreshmediumplus40 p@Mlovastatinfor routinely used to treat hypercholesteremia. Lovastatin inhibits the 36 h. At time 0 h, cells werestimulatedby replacingthe mediumwith fresh activity of the 3-hydroxy-3-methylglutaryl CoA (1-1MGCoA) reduc medium containing 4 mM mevalonic acid. Cells were harvested at the indicated times, and DNA synthesisand cell density were measuredas described tase enzyme, critical to the cholesterol biosynthetic pathway. Among previously(12). other effects, inhibition of this pathway disrupts protein prenylation The synchronizationof normal mammaryepithelial 76N cells by growth and, therefore, also disrupts the subcellular localization and function factor deprivationis asfollows: At 48 h following plating subconfluent76N of proteins such as ras and protein glycosylation via inhibition of cells,mediumwasremovedandcells werewashedthreetimesandincubated in DFCI-3 medium for 72 h. DFCI-3 medium is DFCI-l medium without Received10/24/96;accepted1/3/97. essential growth factors (17). At time 0 h, cells were stimulated by the addition The costs of publication of this article were defrayed in part by the payment of page of DFCI-l mediumandharvestedatthe indicatedtimesthereafter,andDNA charges. This article must therefore be hereby marked advertisement in accordance with synthesis rates were measured as described previously (12). 18 U.S.C. Section 1734 solely to indicate this fact. 76N normalmammaryepithelialcell strainandMDA-MB-l57 tumor cell I This research was supported in part by Grant DAMD- 17-94-J-408 1, AIBS I 579 from the U.S. Army Medical Research Acquisition Activity and by Grant No. R29CA66062 line were also synchronized at the -S boundary using a modification of the from the National Cancer Institute(both to K. K.). double-thymidineblockprocedure(18)asdescribedpreviously(16).For flow 2 To whom requests for reprints should be addressed, at Wadsworth Center, Empire cytometry studies, 106cells were centrifuged at 1000 X g for 5 mm, fixed with State Plaza,P.O. Box 509, Albany, NY 12201.Phone:(518) 486-5799;Fax: (518) 486-5798; E-mail: [email protected]. ice-cold70%ethanol(30 mm at 4Â°C),andwashedwith PBS(19).Cells were

3 The abbreviations used are: cdk, cyclin-dependent kinase; CKI, cdk inhibitor, HMG, suspendedin5 ml of PBScontaining10 @g/mlRNase,incubatedat37Â°Cfor 3-hydroxy-3-methylglutaryl. 30 mm, washedoncewith PBS,andresuspendedin1ml of 69 @Mpropidium 604

INDUCTION OF CKIs BY LOVASTATIN

iodidein 38mMsodiumcitrate.Cellswerethenincubatedatroomtemperature A: RNA in thedarkfor 30 rainandfilteredthrougha75-mmNitex mesh.DNA content Normal Tumor wasmeasuredona FACScanflow cytometersystem(BectonDickinson,San Jose,CA), and data were analyzedusing the CELLFIT software system @ (Becton Dickinson). jâ€”;-23 678910111213J RNA Isolation and Northern Blot Hybridization. Total cellular RNA wasextractedfromnormalandtumorcellsby guanidiniumisothiocyanateand p21 subjectedtocesiumchloridegradientpurification.For Northernblot analysis, 20 @goftotal RNA werefractionatedunderdenaturingconditionson a 1.2% agarose/0.66Mformaldehydegelandtransferredtoa Nytranfilter (Schleicher & Schuell,Keene,NH) for subsequenthybridization.The DNA probeswere preparedby randomprimed labeling (BoehringerMannheim,Indianapolis, p27 IN). Vectorcontainingp21wasprovidedby S.ElledgeandW. Harper,cDNA I. to p27wasprovidedby JoanMassague,cDNAsto p15andp16wereprovided by David Beach,andcDNAs to histoneH4 and36B4asdescribedpreviously (15).All cDNA insertswerelabeledwith [a-32PJdCTPtoa specificactivity of 1 X l0@dpm4@gDNA. p15 â€¢@ Western Blot and Immune Complex Kinase Analysis. Cell lysates and tissuehomogenateswerepreparedandsubjectedto Westernblot analysisas described previously (15, 20). Briefly, 50 p@gof protein from each tissue sample or cell line were electrophoresed in each lane of either a 10% SDS @ polyacrylamide gel (cyclin A), a 13% SDS-polyacrylamide gel (p21 and p27), p16 I $@ I a 15%SDS-polyacrylamidegel(p15 andp16),andtransferredtoImmobilon P. PrimaryantibodiesusedwerepRbmonoclonalantibody(PharMingen,San Diego,CA) atadilution of 1:100,monoclonalantibodyto p16(agift from Jim DeCaprio, Dana-Farber Cancer Institute, Boston, MA) at a dilution of I :20, 36B4 pl5, p27, and p21 polyclonal antibodies(SantaCruz Biotechnology,Santa Cruz, CA) at a dilution of 1:100, affinity-purified rabbit anti-human cyclin A antibody(agift from J.W. Harper,BaylorCollegeof Medicine,Houston,TX) at a dilution of 1:20,000,and monoclonalantibody to actin (Boehringer Mannheim)at 0.63 @g/mlBlono. ImmunoprecipitationsandHl kinaseassayswereperformedas described previously (16, 21). For immunoprecipitation followed by Western blot anal B: Protein ysis, 250 i.@gofprotein were used per immunoprecipitation with a polyclonal Normal Tumor antibodyto cdk2 in lysis buffer asdescribedabove.The lmmunoprecipitates @ were then electrophoresed on a 13% SDS-polyacrylamide gel, transferred to Ii23 6789lOlll2@@J Immobolin P, blocked, and incubated with either monoclonal antibody to p21 (Oncogene Science, Cambridge, MA) or p27 (Transduction Laboratories, p21 @ Lexington, KY) at a dilution of 1:100 and analyzed as described above. â€”@â€” - S

Results and Discussion

CKIs Are Differentially Expressed in Normal versus Tumor derived Exponentially Growing Cells. To determine the relative p27 ,. .@- @â€” levels of CKIs in normal versus tumor-derived breast epithelial cells, we initiated our studies by investigating the expression of four CKIs (p21, p27, plS, and p16) in three normal cell strains, one immortalized cell line, and nine mammary epithelial tumor cell lines (Fig. 1). The p15 three normal cell strains (70N, 81N, and 76N) were established from reduction mammoplasties obtained from three different individuals (17). The immortalized MCF-lOA cell line is a subline of a normal breast epitheial cell strain, MCF-lO, derived from human fibrocystic p16 mammary tissue which was immortalized after extended cultivation in medium containing low concentrations of calcium (22). The mam mary epithelial cell types, estrogen receptor, p53, pRb (retinoblasto Actin @-@--@u_@T'â€”--.--..* ma), and cycin E statusof these normal and the nine established tumor cell lines usedwere describedpreviously(15, 23). Total RNA

and protein were prepared from exponentially growing cells and Fig. I . Altered expression of CKls in exponentially growing normal versustumor cells. subjected to Northern or Western blot analyses,respectively. Loading Northem blot (A) andWestem blot (B) analysesofCKl expressioninnormalversustumor controls were performed for both Northern and Western blot analysis breastepithelial cells. RNA was analyzed on Northern blots (20 @gofRNA/lane). The list of nonnalcells(Lanes1â€”4)andtumorcelllines(Lanes5â€”13)ispresentedbelow.Blots using an invariant mRNA (i.e., 36B4 cDNA) and protein (i.e., actin were hybridizedwith the indicatedprobesor 36B4 (15) usedfor equalloading.B, Western antibody) as probes. As demonstrated, p21 mRNA is expressed at blot analysisof CKIs from cell extractsobtainedfrom the samecell lines usedin A. Fifty higher levels in normal versus tumor cells, whereas p27 and p16 1.Lgoftotal cell extractwererun on SDS-polyacrylamidegels.Proteinsweretransferred to Immobilon P and blots were incubatedwith the indicated anti-CKI antiserum,and mRNAs aremorehighly expressedinsometumorcells(Fig. 1A).The immunoreactiveproteins were detectedwith the enhancedchemiluminescencereagent pattern of p2l expression is interestingly different at the protein level (Amersham). Lane assignments: l-70N; 2-8lN; 3-76N; 4-MCF-1OA; 5-MCF-7; 6-MDA MB-157; 7-MDA-MB-231; 8-MDA-MB-436; 9-T47D; lO-BT-20; 1l-HBL-lO0; 12- with some tumor cell lines as demonstrated by Western blot analysis H5578T; and l3-2R75T. Blots were hybridized with the indicated probes or actin used for (Fig. 1B). For example,MCF-7, T47D, BT-20T, andZR7ST cells,all equalloading. 605

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 1997 American Association for Cancer Research. INDUCTION OF CKIs BY LOVASTATIN estrogen-receptor positive cell lines, express relatively high levels of A Time,hoursafterstimulation p21 protein,whereasthelevelsof p21 mRNA in thesecellswere very low. In addition, in all of the tumor cell lines that show increased p27 I0 3 691215182124273033 protein levels, the p21 levels are not as low as is typical in tumor cells, but instead are similar in levels to those of normal cells. Although the p21 levels of mRNA for each of the CKIs generally correspond to their protein levels, some particular discrepancies are noted with respect to p15 expression. In the normal cell strains, p15 protein is expressedat @ high levels, whereas mRNA levels are only increased in the 8lN cells p27 (Fig. I , Lane 2). Additionally, the tumor cell line Hs-578T (Fig. I, Lane 12) expresses very high levels of p15 mRNA and undetectable levels of p15 protein. These observations suggest that p15 levels may be regulated by translational or posttranslational mechanisms. p15 @â€” â€” â€”@ -.___ -@ â€” Although the differences in CKI expressionobservedbetweennormal breastcells and breastcancercells could be a result of tissuespecificity, @@@ it is more likely due to changes which occur during tumorigenesis. In p16 s-'-__.@..,â€”.@ â€”@- â€” most cases where a difference in expression was noted between normal versus tumor cells, all four normal cells (i.e., p21) or all three normal cell strains(i.e., pl5) consistentlyexhibited a clearly detectableexpressionof these two proteins compared to tumor cells, even though the normal cells cyclin A were derived from different individuals. â€”@â€” Expression of CKIs in Synchronized Normal and Tumor Cells. The CKIs have been proposed to establish a threshold of inhibition B that must be exceeded if cell cycle progressionis to occur (24). Therefore, a disruption in the levels of CKJs or cyclin-cdks could offset the threshold balance or result in the displacement of particular regulatory proteins. The net result of such an imbalance suggests an aberrant cell cycle progression. This hypothesis led us to analyze the pattern of expression and relative levels of the CKIs, P15, p16, p21, E and p27 throughout the cell cycle in synchronized populations of both 8' normal cell strains and tumor cell lines (Figs. 2â€”4). Initially, we assessedtheeffect ofgrowth factorson CM expressionby arresting normal 76N cells in G0 (Fig. 2; i.e., growth factor deprivation induced quiescence).This allowed us to determine whether a particular phaseof the cell cycle might enrich for the expression of these CKIs. Briefly, cells were cultured in growth factor-deficient medium for 72 h and then stimulated to reenterthe cell cycle with the addition of growth 0 6 12 18 24 30 factors. Reentry into the cell cycle and S-phase was monitored by Time, hrs after stimulation [3H]thymidine incorporation (Fig. 2B). At the indicated times after read Fig. 2. Expression ofCKIs in normal cells synchronized by growth factor deprivation. dition of growth factors,cells were harvestedandextractedproteinswere Normal 76N cells were arrestedin G0 via growth factor deprivationfor 72 h and then analyzed on Western blots with antibodies to p21, p27, p15, p16. and stimulatedtoreenterthecell cyclethroughtheadditionof growthfactors.A, at indicated cyclin A (Fig. 14). In normal 76N cells, the pattern of expression of times after growth factor stimulation,whole-cell lysateswere preparedand analyzedon Western blots. Blots were incubated with indicated anti-CKI antiserum and immunore cyclin A protein is consistent with that seen for other normal cell types activeproteinsvisualizedwith theenhancedchemiluminescencereagent.Detectiontime: with levels rising dramatically at the S-phaseanddisappearingby the end 1â€”30swith the exception of p16 for 24 h (see text). Blots were probed with actin used of G2-M. The patternofexpression of the four CKIs examinedfollowing for equalloading (datanot shown),and the patternwas identicalto that of p15 whose levelsdonotchangeduringthecell cycle.B, DNA synthesisratesasmeasuredby growth factor stimulation indicates that p2! levels increasesubstantially [3H)thymidineincorporation. (10-fold) during the S-phase,whereasp27, p15, and p16 levels remain relatively unchangedthroughout the cell cycle (Fig. 24). Theseobserva tions corroborate our hypothesis that these CKIs may each function pathway),which is routinelyusedto stimulatecells to reenterthe cell differently during the cell cycle, with p21 clearly having a pronounced cycle and advanceto the S-phase.Synchrony of tumor cells at various role during the S-phaseof cycling cells andp27 andp15 being important times after release from the lovastatin block was monitored by histone in quiescentcells as recently documentedin fibroblasts (25). H4 mRNA expressionand[3H]thymidineincorporation(Fig. 3, A and Induction of p2! and p27 by Lovastatin in Tumor Cells. Unlike E). Immediately following lovastatin treatment there is inhibition of normal cells, it is difficult to growth factor deprive tumor cells or DNA synthesis followed by a dramatic increase in incorporation of arrest them in G0 since most tumor cells have lost their growth factor [3H]thymidine as well as histone H4 expression.This increase is requirement. As such a pharmacological method has to be used to indicative of the cells being arrested in G5. synchronize tumor cells in early G1. For that purpose we used lovas Total RNA and protein were extracted from cells harvested at tatin, an inhibitor of the cholesterol biosynthetic pathway, to synchro various times after release from lovastatin and analyzed on Northern nize MDA-MB-l57 breast cancer cells in G1 (Fig. 3). We have andWesternblotsusingp2l@1'@'@, p27'@â€•,histoneH4,and36B4 previously reported on the use of lovastatin as an agent to synchronize cDNAs and p2! and p27 antisera as probes (Fig. 3, A and B). cells in early G1 (12, 13). Cells were cultured in lovastatin for 36 h, Unexpectedly, we found that lovastatin treatment dramatically in at which time lovastatin media were removed and replaced with media duced the expression of p21 and p27 in MDA-MB-157. These two containing mevalonate (the end product of the cholesterol biosynthetic CKJs are barely detectable in exponentially growing tumor cells (Fig. 606

Time,hrsafterLovastatln IB). The lovastatin induction is evident at both the transcriptional (p21) and translational (p21 and p27) levels as demonstrated by @ A i; 6912151821242730331 Northern (Fig. 3A) and Western blot (Fig. 3B) analyses. Although the CKI levels were substantially induced for 9 hours, the levels abruptly CIP1/WAF1 mRNA S.. diminished. The abrupt induction of p21 and p27 by lovastatin is not due to a responseto the medium change since earlier experiments demon stl-ated that p21 and p27 protein increase during lovastatin treatment (data KiplmRNA â€¢â€¢Ã¸...... sâ€¢Oâ€¢@ not shown) and continue until 9 h posttreatment(Fig. 3). To determine whether the transientinduction of the CKIs led to the inhibition of kinase activity, we measured the phosphorylation of histone Hl in anti-cdk2 immunoprecipitates prepared from extracts of synchronized cells (Fig. HistoneH4mANA 3C). These results indicate that the induced levels of p21 and p27 are accompaniedby a dramatic decreasein cdk2-associatedkinase activity which then increasesupon disappearanceof p21 and p27 proteins. To 36B4mRNA determine whether p21 and/or p27 can directly associatewith cdk2 and inhibit its activity, we performed a two-step immunoprecipitation/West Io3691215182124273033I em blot analysis.We immunoprecipitated lovastatin-treatedcell extracts B with anti-cdk2antiserafollowed by immunoblottingwith p21 or p27 antisera (Fig. 3D). Anti-p2l immunoblot analysis of cdk2-containing @ p21 complexesdemonstratesthatp21 is directly associatedwith cdk2 andthat the relative amount of p21 associated with cdk2 is inversely correlated with kinase activity of cdk2. @@ p27 p27 is also similarly associatedwith cdk2, although in relatively less abundancethan p21. Together theseanalysesindicate that not only are p21 andp27 inducedduring lovastatin synchronizationin MDA-MB-l57

Actin tumor cells but they also functionally complex with cdk2. Furthermore, the expression of p21 and p27 is inversely correlated with cdk2 kinase activity, suggestingthat theseCKIs directly inhibit cdk2. C Io3691215182124273033Induction of p21 and p27 in Tumor Cells by Lovastatin Is Cell Cycle Independent. We synchronizedboth normal and tumor cell cdk2/H1 Klnaee types in the G1-S-phase boundary by double-thymidine block [(16) and Fig. 4] to determine whether the lovastatin-mediated induction of p21 and p27 is cell cycle dependent or due to a direct (as of yet @@ D I 691215182124273@]unknown) effect of lovastatin. This synchronization trial was also used to compare the cell cycle pattern of expression of the four CKIs @ cdk2/pZl in normal and tumor cells. Synchrony of both cell types after release from block was monitored by flow cytometry [(16) and Fig. 4C]. Cells were harvested at various times after release from treatment, and cdk2/p27 extracted proteins were analyzed on Western blots with antisera to @ ,@ .- p21, p27, p16, pls (Fig. 4), and cyclin A (16). In both normal and E tumor cells, the expression pattern of cyclin A protein is consistent with that seenfor other cell types, the levels are tightly regulated such that peak expression occurs during the S-phase and early M as C MDA-MB-1 57 0 documented previously (16). Cs The pattern of expression of p21 in normal cells revealed cell cycle 0 0. regulation consistent with the growth factor deprivation results shown @.- 0@ in Fig. 2. In normal cells, with both methods of synchronization, p21 @1 .E E4 is cell cycle regulated with peak levels coinciding with the peak &- â€¢00 S-phase and early M (Fig. 4A, left panels). In tumor cells, however, Iâ€” p21 levels remain low and virtually undetectable during all phasesof the cell cycle. Thus, we were not able to enrich for p21 expression at C@) G1 of tumor cells synchronized by double-thymidine block (Fig. 4B, right panels). These observations suggest that the induction of p21 0 6 12182430 during lovastatin synchronization (Fig. 3) is not cell cycle dependent Time, hours but is rather a lovastatin- or mevalonate-specific effect. The level of p27 remained unchangedin both normal and tumor Fig. 3. Inductionof the CKIs p21 andp27 by lovastatinis inverselycorrelatedwith cdk2 kinase activity. MDA-MB-157 tumor cells were cultured in 20 @Mlovastatinfor 36 h, at which time lovastatinwas replacedby mevalonate.Sampleswerecollectedat indicated times after lovastatin removal, and RNA and proteins were extracted.A, Northern blot analysis:RNA extractedfrom cells at indicatedtimes after lovastatin with anti-cdk2 polyclonal antiserum and complexes were assayed for the ability to treatmentwasanalyzedby Northernanalysis(20 @gflane)andprobedwith 32P-labeled phosphorylate histone HI. The Hl-labeling reaction complexes were analyzed by SDS p21, p27, histone H4, or 36B4 cDNA. B, Western blot analysis: 50 @sgofprotein extracts PAGE and autoradiography. D, immunoprecipitation/Western blot analyses: 500 @gof from eachcondition were analyzedusing Westernanalysiswith anti-p21,anti-p27,or proteinextractswereimmunoprecipitatedwithanti-cdk2polyclonalantiseraandprecip anti-actin-specificantisera,andblots weredevelopedwith the enhancedchemilumines itatedcomplexeswereanalyzedusingWesternanalysiswith anti-p2l or anti-p27mono cencereagent.C,histoneHl kinaseassay:500 @sgofextractswereimmunoprecipitated clonal antisera.E, DNA synthesisrateswere monitoredby [3Hlthymidine incorporation. 607

INDUCTION OF CKIs BY LOVASTATIN

76N(Normal) MDA-MB-157(Cancer) TIn*, hours Timi, hours I036912151821242730331Io 4 8i2ie2832384ou1+v.c

@@ p21

@@@@ @1ww@@fw .@- .- d* p27

Fig. 4. Expression of CKIs in synchronized normal 76N and tumor MDA-MB-l57 breast cells. Both cell lineswere synchronizedusingthe p16 double-thymidine block procedure: 76N cells were incubated in 2 mM thymidine for 24 h, washed, and incubated in regular medium for 12 h and incubated in 2 m@ithymidine for an additional24 h. MDA-MB-157 cellswere treated p15 similarly, except incubation in thymidine was for 36 h and thymidine-freemediafor 24 h, as de scribedpreviously (16). At the indicatedtimes following releasefrom double-thymidineblock, cell lysateswere preparedandsubjectedtoWest em blot (A), histone Hl kinase (B), and flow cytometry (C) analyses. Fifty @sgofextracts of cells were subjected to Western blot analysis with antisera directed against the indicated CKI or cyclin A at indicatedtimes after double-thy midine block. Note: p21 and p15 +ve C are from 76N cell extracts and are used as a positive control. Blots were probedwith actin usedfor equal loading (data not shown). and the pattern -@e@ was identical to that of p15 in normal cells and p16 in tumor cells whose levels do not change during the cell cycle. For histoneHI kinaseac tivity, equal amount of proteins (500 @g)from cell lysatespreparedfrom eachcell line at the indicated times were immunoprecipitated with anti-CDK2 (polyclonal) coupled to protein A beadsusinghistoneHl as substrate.B, quantifi cation of the histoneHI-associated cdk2 kinase activities by scintillation counting, as described previously(16). C, relative percentageofcells in different phasesofthecell cycle for eachcell line was calculated from flow cytometric measure Time, hrs Time, hrs mentsof DNA content.

Time, Time, I...., (11 C 1!SISA hr@ @-@--@--.--.--0577130355100546514590972481001210033471592854018651004721628557246747692781386730100195733100186207666274710030206370122627

@10016502555206521422480194028100193632971938360025364075302544704025

cells synchronized by double-thymidine block. Similar to p21, p27 was constitutively overexpressed throughout the cell cycle in tumor @ levels were not induced further in of tumor cells synchronized by cells. p15 was expressedthroughout the normal cell cycle and was not double-thymidine block. This suggests that the induction of p27 detected in tumor cells even under prolonged exposure of the Western during lovastatin synchronization (Fig. 3B) is not a cell cycle-depen blots (i.e., 24 h). dent effect, but an effect of lovastatin. p16 levels were minimally and To comparethe kinaseactivity associatedwithcdk2 in normaland constitutively detected in normal cells, but only upon prolonged tumor cells, we measured the phosphorylation of histone Hi in anti exposure times (i.e., 10 mm versus 10â€”30sfor the other CKIs). p16 cdk2 immunoprecipitates prepared from synchronous cell extracts 608

[(16) and Fig. 4B]. There was a significant difference between Acknowledgments normal and tumor cells in the timing of cdk2 activity as described previously (16). In normal cells, the cdk2-associatedkinase activity is We thank Dr. E. Harlow for polyclonal antibody to cdk2; Dr. W. Harper for polyclonalantibodyto cyclin A; Dr. J.A. DeCapriofor monoclonalantibody cell cycle regulated,coinciding with the increasedlevels ofcyclins E and to pl6; Dr. W. HarperandS.Elledgefor cDNA to p2l; Dr. JoanMassaguefor A protein expressionas describedpreviously(16). Conversely,cdk2 cDNA to p27;Dr. D. Beachfor cDNAs to p15andp16;andDr. R. Sagerfor remains catalytically active throughout the cell cycle in tumor cells, providingnormalcell strains.WethankWendyToyofukuandM. PatFox for resulting in a near constitutive pattern of histone Hl phosphoiylation excellent technical assistance and other members of the KK laboratory (i.e., [(16) andFig. 4B]. Therefore,theobserveddecreaseincdk2-associated Don Porter,XiaomeiChen,Sheng-KaiYang,RichardHarwell,andThaddeus kinaseactivity in lovastatin-treatedMDA-MB-157 cells (Fig. 30 is most Herliczek) for the critical reading of the manuscript.We also gratefully likely a direct result of inhibition mediatedby the induction of p2! and acknowledge the use of Wadsworth Center's Immunology, Molecular Biology, p27 proteinsby lovastatin. Photography/Graphics,andTissueCulturecorefacilities. The studiesreportedhererepresentacomprehensiveanalysisof the References expression of four CKI mRNAs and proteins including both the p2l@â€•@' and Ink4 families in normaland tumorbreastepithelial 1. Draetta, G. cdc2 activation; the interplay of cyclin binding and Thrl6l phosphoryl ation. Trends Cell Biol., 3: 287â€”289,1993. cells. Our data indicate that striking differences exist among the 2. Pines,J. The cell cycle kinases.Semin.CancerBiol., 5: 305â€”313,1994. individual CKIs between normal and tumor cells. p21 mRNA and p15 3. Hunter, T., and Pines, J. Cyclins and cancer II: cyclin D and cdk inhibitors come of protein expressionare consistentlyhigherin normal cells while p27 age. Cell, 79: 573â€”582,1994. 4. Sheer,C. J. G, phase progression: cycling on cue. Cell, 79: 551â€”555,1994. and p16 mRNA and protein are overexpressed in some tumor cells. 5. Peter,M., andHerskowitz, I. Joiningthe complex: cyclin-dependentkinaseinhibitory Additionally, the cell cycle expression levels of the CKI proteins are proteins and the cell cycle. Cell, 79: 181â€”184,1994. 6. Sherr, C. J., and Roberts, J. M. Inhibitors of mammalian G, cyclin-dependent kinases. different, which suggests that these inhibitors may be functionally Genes Dcv.,9:1149â€”1163,1995. distinct throughout the cell cycle. We have found that the drug 7. Elledge, S. J., and Harper, J. W. Cdk inhibitors: on the threshold of checkpoints and lovastatin is capable of inducing p21 and p27 protein in a cell-specific development.Curr. Opin. Cell Biol., 6: 847â€”852,1994. 8. Elledge, S. J., Winston, J., and Harper, J. W. A question of balance: the role of manner via a cell cycle-independent mechanism. cyclin-kinase inhibitors in development and tumorigenesis. Trends Cell Biol., 6: Lovastatin, an inhibitor of HMG CoA reductase(the first enzyme in 388â€”392,1996. the isoprenyllipid biosyntheticpathway),is a widely useddrugfor the 9. Hengst, L., Dulic, V., Slingerland, J. M., Lees, E., and Reed, S. I. A cell cycle regulated inhibitor of cyclin-dependentkinases. Proc. Nail. Acad. Sci. USA, 91: treatment of hypercholesterolemia. Lovastatin prevents the first step 5291â€”5295,1994. of cholesterolsynthesis,which is the conversionof HMG into mev 10. Hengst, L., and Reed, S. I. Translational control of p2lKipl accumulation during the cell cycle. Science(WashingtonDC),271: 1861â€”1864,1996. alonic acid. The blockage of this pathway also prevents the isopre 11. Jakobisiak,M., Bruno, S., Skierski, J. S., and Darzynkiewicz, Z. Cell cycle-specific nylation of severalproteinssuchas Ras, Rap, and G by farnesyl, a effects of lovastatin. Proc. Natl. Acad. Sci. USA, 88: 3628â€”3632,1991. downstream product of the pathway. This inhibition of isoprenylation 12. Keyomarsi, K., Sandoval, L., Band, V., and Pardee, A. B. Synchronization of tumor and normal cells from G, to multiple cell cycles by lovastatin. Cancer Res.. 51: blocks the function of the aforementioned proteins (26, 27). Apart 3602â€”3609,1991. from its inhibitoryactionon HMG CoA reductase,lovastatinhasalso 13. Keyomarsi, K. Synchronization of mammalian cells by lovastatin. Methods Cell Sci., been used as an effective agent in cell synchronization for both tumor 18:1â€”6,1996. 14. Turner, J. E., Minkoff, C. G., Martin, K. H., Misra, R., and Swenson,K. I. Oocyte and normal cells (12, 13). activationand passagethroughthe metaphase/anaphasetransitionof the meiotic cell Recently, Hengst et a!. (9, 10) reported an elevation of p27 in cycle is blockedin clamsby inhibitorsof HMG-CoA reductaseactivity.J. Cell Biol., 128:1145â€”1162,1995. lovastatin-arrested HeLa cells. This increase was attributed to a cell 15. Keyomarsi, K., and Pardee,A. B. Redundantcyclin overexpressionandgeneampli cycle effect since a similar increase was observed in cells synchro fication in breastcancercells. Proc. Natl. Acad. Sci. USA, 90: 1112â€”1116,1993. nized by density-mediatedarrest and thymidine and nocodazole 16. Keyomarsi,K., Conte, D., Toyofuku, W., and Fox, M. P. Deregulationof cyclin E in breastcancer.Oncogene,I1: 941â€”950,1995. blocks (10). However, the increase in p27 levels, in density-mediated 17. Band, V., and Sager, R. Distinctive traits of normal and tumor-derived human arrested fibroblasts, was much lower than that in lovastatin-treated mammary epithelial cells expressed in a medium that supports long-term growth of HeLa cells. This was attributed to imperfect synchronization of the both cell types.Proc.NatI. Acad.Sci. USA, 86: 1249â€”1253,1989. 18. Rao, P. N., and Johnson, R. T. Mammalian cell fusion: studies on the regulation of fibroblastcells, indicatingthat the increasesin p27 levels seenmight DNA synthesis and mitosis. Nature (Lond.), 225: 159â€”164, 1970. be due to experimental design. Furthermore, HeLa cells are derived 19. Michalopoulos,G. K. Liver regeneration:molecularmechanismsof growth control. FASEB J., 4: 176â€”187, 1990. from a cervical carcinoma and carry the human adenopapilloma virus, 20. Keyomarsi, K., O'Leary, N., Molnar, G., Lees, E., Fingers, H. J., and Pardee, A. B. whereasmost cancercell lines in generaldo not carry this virus. Cyclin E, a potentialprognosticmarkerfor breastcancer.Cancer Res.,54: 380-385, We have shown here that although lovastatin is capable of 1994. 21. Bacus, S. S., Yarden, Y., Oren, M., Chin, D. M., Lyass, L., Zelnick, C. R., Kazarov, inducing both p21 and p27 in human breast tumor cells, this A., Toyofuku, W., Gray-Bablin, J., Beerli, R. R., Hynes, N. E., Nikiforov, M., induction is not due to cell cycle synchronization effects of by Haffner, R., Gudkov,A., and Keyomarsi,K. Neu differentiationfactor (heregulin) astatin. The induction of the inhibitors was not observed using activatesap53-dependentpathwayin cancercells.Oncogene,12:2535â€”2.547,1996. 22. Soule, H. D., Maloney, T. M., Wolman, S. R., Peterson, W. D., Jr., Brenz, R., other methods of cell synchronization such as double-thymidine McGrath, C. M., Russo,i., Pauley, R. J., Jones, R. F., and Brooks, S. C. Isolation and block in normal and tumor cells or growth factor deprivation in characterization of a spontaneously immortalized human breast epithelial cell line, MCF-lO. CancerRes.,50: 6075â€”6086,1990. normal cells. Furthermore, we used both normal and tumor cells 23. Gray-Bablin, J., Zalvide, J., Fox, M. P., Knickerbocker, C. J., DeCaprio, J. A., and that were derived from human mammary epithelial cells which are Keyomarsi,K. Cyclin E, a redundantcyclinin breastcancer.Proc.Nail. Acad.Sci. representativeof most types of cancer, since more than 90% of all USA, 93: 15215-15220,1996. 24. Massague,J., and Polyak, K. Mammalian anti-proliferativesignalsand their targets. human cancers are of epithelial origin. Although lovastatin is Curr. Opin. Genet. Dev., 5: 91â€”96,1995. capable of inducing these inhibitors in human epithelial cells, the 25. Rivard, N., L'Allemain, G., Bartek, J., and Pouyssegur, J. Abrigation of p2'lKipl by cDNA antisensesuppressesquiescence(G0state)in fibroblasts.J.Biol. Chem.,271: mechanism of induction is not through arrest of cells in a specific 18337â€”18341,1996. phase of the cell cycle, but through a lovastatin, drug-mediated 26. Maltese, W. A. Posttranslational modification of proteins by isoprenoids in mamma effect. Whether the CKI induction following lovastatin treatmentis han cells. FASEB J., 4: 3319â€”3328,1990. 27. Maltese, W. A., and Sheridan, K. M. Isoprenylated proteins in cultured cells: sub due to inhibition of any one reaction of the cholesterolbiosynthesis cellular distribution and changesrelatedto alteredmorphologyand growth arrest pathway remains to be elucidated. inducedby mevalonatedeprivation.J. Cell. Physiol., 133: 471â€”481, 1987.

609

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 1997 American Association for Cancer Research. Lovastatin Induction of Cyclin-dependent Kinase Inhibitors in Human Breast Cells Occurs in a Cell Cycle-independent Fashion

Julie Gray-Bablin, Sharmila Rao and Khandan Keyomarsi

Cancer Res 1997;57:604-609.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/57/4/604

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/57/4/604. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.