[CANCER RESEARCH57, 1569-1574. April 15, 1997J

ABSTRACT

Cyclin Dl plays an important role in regulating the progression of cellsthrough the Gi phase of the cell cycle. This gene is frequently overexpressed in human colon cancer. To address the role of cyclin Dl in growthcontrol and tumorigenesis in this disease, we have overexpressed anantisense cyclin Dl cDNA construct in the human colon cancer cell lineSW48OES,which expresseshigh levelsof cycin Dl. The integration andexpression of the antisense construct was verified by Southern and Northens blot analyses, respectively, and resulted In decreased expression of thecyclin Dl protein. This was associated with decreased levels of the Rb andp27―@―proteins. In addition, the hypophosphorylated form of Rb was

increased in these cells. The SW4SOES antisense cydlin Dl cells displayedan increased doubling time, a decrease in saturation density, decreasedplating efficiency and anchorage-independent growth, and a loss of tumorigenicity in nude mice. These findings provide direct evidence that

increased expression of cydin Dl in colon tumor cells contributes to theirabnormal growth and tumorigenicity. The ability to revert the transformed phenotype of these cells with antisense cydlin Dl suggests thatcydlin Dl or its associated cydin-dependent kinase 4 may be useful targets

in the therapy of colon cancer.

INTRODUCTION

In mammalian cells, critical transitions in the cell cycle are regulated by the sequential activation of a series of cyclins and CDKs.3These cyclins and CDKS now include cyclins A, B!, B2, Dl—3,andE andCDKs1—7(forreview,seeRefs.1—7).CyclinDl hasbeenimplicated in controlling the G1 phase of the cell cycle. The expression of cyclin Dl mRNA and protein peaks during mid-G1 whengrowth factor-deprived cells are restimulated to enter the cell cycle (3,8, 9). Inhibition of cyclin Dl function mediated by microinjection ofantibodies or antisense plasmids into normal fibroblasts or tumor cellsin mid-G1 can prevent entry ofthese cells into the S phase (3, 10—12).On the other hand, overexpression of cyclin Dl in rodent fibroblastsleads to a shortened duration of the G1 phase, decreased cell size, andreduced serum dependence (13—15).

Because the major regulatory events leading to mammalian cellproliferation and differentiation occur in the G1 phase of the cell cycle(16), deregulated expression of the G1 cyclins and CDKs might causeloss of cell cycle control and thus enhance oncogenesis. Indeed,rearrangement, amplification, and/or increased expression of the cyc!in Dl gene, which is located on the human chromosome I 1q13

Received 9/25/96; accepted 2/17/97.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

IThisstudywassupportedbyfundingfromtheDorotFoundation(toN.A.),theNational Dairy Council and OSG Foundation (to P. R. H.), the Italian Association forCancerResearch(toA.S.),anda NationalCancerInstitutegrantROl-63467andawardsfrom the National Foundation for Cancer Research and the Lucille Markey CharitableTrust (to I. B. W.).

2Towhomrequestsforreprintsshouldbeaddressed,atHerbertIrvingComprehensiveCancer Center, College of Physicians and Surgeons, Columbia University, HHSC-l509,701 West 168th Street, New York, NY 10032. Phone: (212) 305-6921; Fax: (212)305-6889.

3 The abbreviations used are: CDK, cyclin-dependent kinase; Rb, retinoblastoma.

region, have been found in several types of human cancer, includinghuman parathyroid adenomas, B cell lymphomas, breast, colon, lung,bladder and liver cancers, and squamous carcinomas of the esophagus,head and neck (for review, see Refs. 3—7,17, and 18). Overexpressionof cyclin Dl in rat fibroblasts enhances their growth and tumorigenicity (14), and cyclin Dl collaborates with an activated ras oncogene(19) or a defective adenovirus EJA oncogene (20) to increase thetransformation of primary rodent fibroblasts. Furthermore, overexpression of cyclin Dl in mammary cells in transgenic mice results inabnormal cell proliferation and the development of mammary tumors(21). Cyclin Dl also cooperates with the myc oncogene in lymphomagenesis in transgenic mice (22—24).Taken together, these results indicate that cyclin D 1 can function as an oncogene.

Cyclin Dl is not amplified in human colon cancer. However, thecyclin D 1 protein is overexpressed in about 30% of adenocarcinomasof the colon (25, 26). This overexpression appears to be a relativelyearly event in colon carcinogenesis, because it is also seen in about30% of adenomatous polyps of the colon (26). Previous studiesindicate that primary human tumors and tumor derived cell lines oftendisplay a correlation between the levels of expression of cyclin Dland the Rb tumor suppressor proteins (3, 27—29).A correlationbetween the expression of these two proteins and the CDK inhibitorprotein @27kiPlhas also been seen in esophageal and breast cancer celllines (29—31). Furthermore, epithelial cell lines engineered to overexpress cyclin Dl display increased expression of the @27k*@@Iprotein(29, 3 1). These findings seem paradoxical, given that cyclin Dlenhances the G1-to-S progression of the cell cycle, whereas Rb and

@27k@ have the opposite effect.

The present study was undertaken to obtain more direct evidencethat cyclin Dl can play a critical role in establishing and maintainingthe transformed phenotype of colon cancer cells and to explore furtherthe above-described relationship between the expression of cyclin D 1,Rb, and @27ki@1@ To this end, we constitutively expressed an antisensecyclin Dl cDNA in SW480E8 cells, a well-characterized human coloncancer cell line that expresses high levels of cyclin Dl . This articledemonstrates that stable expression of antisense cyclin Dl cDNA ledto a decreased level of the endogenous cyclin D I protein, reduction ofin vitro CDK4 kinase activity, a relative decrease in the phosphorylated form of Rb, a decrease in the level of the p27k1Pt protein,inhibition of cell proliferation, and loss of tumorigenicity in nudemice. These findings provide direct evidence that the cyc!in DI gene

can play a critical role in maintaining the malignant phenotype ofcolon cancer cells. It may, therefore, be a valuable biomarker and alsoa potential target for cancer prevention and therapy.

MATERIALS AND METHODS

Cell Culture and Transfection Procedure. The human colon carcinomacell line 5W480E8 (32) and its derivatives were grown in DMEM (Life

Technologies, Inc., Grand Island, NY) containing 10% FCS. The medium forcell lines containing the neomycin resistance gene was supplemented with 500p@g/mlor 1 mg/mI of G4l8 (Life Technologies, Inc.).

1569

Antisense to Cyclin Dl Inhibits the Growth and Tumorigenicity of Human Colon

Cancer Cells1

Nadir Arber, Yuichiro Doki, Edward K-H. Han, Alessandro Sgambato, Ping Zhou, Nae-Hwa Kim, Tom Delohery,Michael G. Klein, Peter R. Holt, and I. Bernard Weinstein2

Herbert Irving Comprehensive Cancer Center, College of Physicians and Surgeons, Columbia University, New York, New York 1fXI32 (N. A.. 1'.D.. E. K-H. H., A. S., P. 1, N-H. K., M. G. K., P. R. H., L B. W.J; Flow Cytometry Core Facility. Memorial Sloan-Kettering Cancer Center, New York, New York 10021 IT DI: Division of Gastroenterology.St. Luke ‘s-RooseveltHospital Center, New York@New York llAiI9 [N. A., P. R. H.]; and Department of Medicine. College of Physicians and Surgeons. Columbia Unis'ersitv, NewYork New York 10032 [1. B. W.J

Research. on January 25, 2021. © 1997 American Association for Cancercancerres.aacrjournals.org Downloaded from

ANTISENSE CYCL1N Dl REVERSES TUMORIGENICITY

The SW480E8 cells are a well-characterized epithelial clone originallyisolated in this laboratory from a mixed population of SW480 cells (32). Inmonolayer culture, these cells have an epithelial and cuboidal morphology andare very adhesive. They also display anchorage-independent growth in softagar and are tumorigenic in nude mice (32). They have a modal chromosomenumber of 58, have amplification of the c-myc gene, and have mutations in thec-K-ras and p53 genes (32).

The plasmid pMV7D1AS, containing a complete human cyclin Dl cDNAsequence in its antisense orientation, was constructed as described below. Thispla.smidor the vector control pMV7 plasmid was transfected into GPAM12cells by the calcium-phosphate method to generate defective recombinantretrovirus particles (33). The viral supernatants of pMV7D1AS- or pMV7-transfected cells were harvested and then used to infect the SW480E8 cells.

After infection, transduced cells were selected in complete DMEM containing

I mg/ml 0418 for 2—3weeks (33). Following 0418 selection, 18drug-resistant(neo+) clones were picked randomly from four different plates (designatedAS1—ASI8).As controls, 10 neo+ clones (designated VCI—VCIO)werepicked from SW480E8 cultures infected with the pMV7 vector lacking theantisense cyclin Dl cDNA sequence and studied following clonal expansion.

Construction of the Antisense Cyclin Dl Expression Plasmid. The

1.1-kbhuman cyclin Dl cDNA containing the entire coding sequence (14) wassubcloned in its antisense orientation into the HindIlI site of the retroviral

expression vector pMV7, as described previously (33). The resulting plasmidwas designated pMV7DI AS and contained a 5â€M̃oloney murine leukemiavirus long terminal repeat, followed by the I.I-kb antisense cyclin Dl sequence, a herpes simplex thymidine kinase promoter linked to the neomycin

phosphotransferase gene (041 8 resistance gene, neo+), and a 3' Moloneymurine leukemia virus long terminal repeat (33).

Protein Extraction and Western Blotting. Proteins were extracted fromexponentially growing cells and were subjected to Western blot analysis asdescribed previously (3, 14). The membranes were incubated with a 1:1000dilution ofmonoclonal antihuman cyclin Dl (clone DCS-6) antibody, a p27@'(clone DCS-72.F6) antibody (NeoMarkers, Fremont, CA), or a monoclonal

antibody to Rb (Pharmingen, San Diego, CA). The intensities ofcyclin Dl, Rb,and p27kIPIproteins were quantified by densitometric scanning. All experiments were repeated at least three times and gave similar results.

DNA and RNA Analysis. As described previously (14), total genomicDNA was isolated from the cell lines, and 10 j.@gwere digested to completionwith the restriction enzymes HindIll or EcoRI (Boehringer Mannheim, Indianapolis, IN) and used for Southern blot analysis. Total RNA was preparedfrom the cell lines, and 10 @gwere used for Northern blot analysis, asdescribed previously (14).

The probes used for these hybridization studies were a gel-purified 1.8-kbBgl-2 cDNA fragment of the Rb-containing plasmid and a 1.1-kb HindlIlcDNA fragment of human cyclin Dl (30). It was essential to use probes thatwere devoid of plasmid sequences to avoid cross-hybridization to plasmidsequences present in the DNA of the cyclin Dl antisense cells. Because thetranscripts of the exogenous antisense cyclin Dl and the endogenous sense

cyclin Dl were similar in size (4.8 versus 4.7 kb), the blots were hybridizedwith a specific [a-32PJCTP-labeled ribonucleoside cyclin Dl sense probe.

Hybridization was carried out as described previously (33). Equivalent loading

of lanes on the Northern and Southern blots was confirmed by ethidiumbromide staining. All experiments were repeated at least three times.

Doubling Times and Saturation Densities. Cells were plated in triplicateat a density of 1 times l0'@per well in six-well (35 mm diameter) plates, in 3ml of DMEM plus 10% or 1% FCS. The number of cells per well was countedusing a Coulter counter every 2 days for 21 days. Doubling times andsaturation densities of each cell line were calculated by standard methods (14).All experiments were repeated at least three times and gave similar results.

Cloning Efficiency. Cells were seeded in triplicate, with 1000 cells per10-cm dish in complete medium and were refed with fresh medium onalternate days. After 10 days of growth the cells were fixed and stained with5% Giemsa stain, and the number of grossly visible colonies was counted (33).All experiments were repeated at least three times and gave similar results.

Soft Agar Assay. Cells (8 X 10'@)were suspended in DMEM plus 10%FCS containing 0.3% Noble agar (DIFCO Laboratories) and plated in triplicatein six-well plates on top of a 1% agar base (33). After 3 weeks of growth, thecells were stained and the colonies were counted by microscopy (see Table 1).These experiments were repeated twice and gave similar results.

Tumorigenicity Assays. To assess tumorigenic capacity, each cell line was

injected s.c. into athymic (nude) mice, at four separate sites using 10@cells persite and two sites per mouse. The mice were monitored for tumor formation

every week and sacrificed 2 months later. Tumors were excised and theirvolumes estimated by multiplying length X width X depth and expressed ascm3. Tumorigenicity was verified by histology of the excised lesions.

Flow Cytometric Analysis. Exponentially dividing cells were collected,stained with propidium iodide, and analyzed for cell cycle distribution by flow

cytometry using a standard protocol (33). These experiments were repeatedthree times.

In Vitro Assays for Cydlin Dl- and Cyclin E-associated Kinase Activity.Assays for cyclin Dl- and cyclin E-associated in vitro kinase activity were

performed as described previously, with minor modifications (33). In brief,200 @gof total cell lysates obtained from exponentially growing cells wereincubated with 2 j.@gof mouse monoclonal antihuman cyclin Dl antibody(NeoMarkers, Fremont, CA) or 1 ,@gof rabbit polyclonal antihuman cyclin Eantibodies (Upstate Biotechnology, Inc., Lake Placid, NY), at 4°C for 2—3h

and then incubated with 30 pJ of protein A-Sepharose beads (50% v/v) for 2 h.The beads were then collected by centrifugation and washed. Kinase assayswere performed by using either 0.4 @gof glutathione S-transferase-Rb fusionprotein or I i.@gof histone HI as substrates for cyclin Dl- and cyclin E-associated kinase assays, respectively.

RESULTS

Integration and Expression of the Antisense Cydin Dl Construct in Derivatives of SW4SOES Cells. Northern blot analyses onRNA extracted from the parental SW480E8 cells, a vector controlclone, and three SW480E8 antisense cyclin Dl clones (ASIO, AS 14,and AS18) are shown in Fig. 1A. When hybridized with a specific[a-32P]CTP sense cyclin Dl ribonucleotide probe, high levels of the4.8-kb exogenous antisense cyclin Dl mRNA were detected in thethree antisense clones. As expected, no hybridization signals wereseen with either the 5W480E8 parental or the vector control cell lineswith this probe (Fig. IA).

Southern blot analysis after Hindu! digestion of genomic DNA,using a 32P-labeled cyclin Dl cDNA probe, revealed 23- and 4-kbDNA bands in all of these cell lines, reflecting the endogenous cyclinDl gene (Fig. 1B). An additional 1.1-kb band was seen in the AS1Oand ASI4 derivatives (Fig. IB). This was anticipated, because theantisense construct was cloned within HindHI sites. In the AS18 cells,the extra band had a higher molecular size (16 kb) most probably dueto rearrangement of the integrated antisense cyclin Dl cDNA (Fig.lB). When similar studies were done after digestion of genomic DNAwith EcoRl, all five cell lines displayed 5- and 2.6-kb bands, representing the endogenous cyclin Dl gene. The three antisense clonescontained an additional 6.2-kb band in the AS1O and AS18 cells anda 5.2-kb band in the ASI4 cells, representing the integrated antisensecyclin Dl cDNA (Fig. 1C).

We next examined whether integration and expression of the cxogenous antisense cyclin Dl sequence was accompanied by decreasedexpression of the cellular cyclin Dl protein. Western blot analysis wasperformed to determine the levels of cyclin Dl protein expression inthe above series of cell lines using a monoclonal antihuman cyclin Dlantibody. Representative data are shown in Fig. 2. High levels of asingle Mr 36,000 cyclin Dl protein band were detected in the parentalSW480E8 and vector control cells. The levels of the Mr 36,000 cydlinDl protein band were three to five times lower in the three antisenseclones than in the control cells (Fig. 2).

Taken together, these results indicated that the antisense cyclin Dlsequence was integrated into the SW480E8 cellular genome, and thecorresponding mRNA was expressed at a relatively high level. This

caused a reduction in the expression of the endogenous cyclin Dlprotein in these derivatives.

1570

Research. on January 25, 2021. © 1997 American Association for Cancercancerres.aacrjournals.org Downloaded from

ANTISENSE CYCLIN DI REVERSES TUMORIGENICITY

@ the control cells and they had relatively low cloning efficiencies (6,I-,@ 1 5, and 9%, respectively) when plated at low densities (Table 1).

@ @- 0@@ When the cells were grown in medium with 1% FCS, the doubling

;;; ;;; &;@@ timesoftheparentalandvectorcontrolcellswere30and31h,andthe< < < doublingtimeof thethreeantisensecloneswere46, 38,and40 h,

respectively. Taken together, these results demonstrate that expressionof antisense cyclin Dl cDNA in SW480E8 cells inhibited cell proliferation, especially when the cells were plated at low cell density or

A @. - 5 Kb maintained in 1% rather than 10% FCS.. Cell cycle parameters were examined in exponentially growing

cultures of the SW480E8 control and antisense cyclin Dl cells usingflow cytometric analysis (fluorescence-activated cell sorting). Curi

.@ ously, the three antisense clones displayed a somewhat lower percent

.@.)Ø4@@@ - 23 Kb age of cells in the G1 phase than the control cells (Table I ). However,

@@ in culturesmaintainedatconfluencefor 2 days,thethreeantisenseB @,@@@ :@ , clones displayed a marked increase in the proportion of cells in the

—G1phase when compared to the SW480E8 Vector control and. parental cells (Table 1). The latter results probably indicate that the

antisense clones stop dividing and enter the —G1phase when they- 4 Kb reach confluence, whereas the parental cells continue to divide at

confluence, which is consistent with their higher saturation densityand their piling up on one another.

Down-Regulation of Cydin Dl Is Associated with DecreasedExpression of the Rb and p27kt@)l Proteins. We found that the

@. @a - 1.1 Kb parentalSW480E8cellsandthevectorcontrolderivativeexpressedrelatively high levels of the Rb protein (Fig. 2). Indeed, separatestudies (not shown here) indicated that the Rb gene is amplified about3-fold in this cell line. This result is consistent with previous studies

C 6Kb indicatingthattheRbgeneissometimesamplifiedinprimaryhumancolon carcinomas, and that its level of expression is often relatively

- 5 Kb high in these tumors (36—38). In the control SW480E8 cells, most of

the Rb protein was hyperphosphorylated (note the more slowly migrating Rb-P band in Fig. 2). All three of the cyclin Dl antisenseclones displayed about a 4-fold decrease in the total amount of the Rb

- 2.6 Kb protein, and there was a relative decrease in the ratio of the phospho

Fig. 1. Integration and expression ofthe antisense cyclin Dl cDNA. The following celllines were analyzed: the parental SW480E8 cells, the vector control derivative, and three o @-antisense clones (AS1O,AS14, and AS18). A, Northern blot analyses for the expression of @Q 0 o @-@the antisense cyclin Dl RNA. Ten ,sg of total cellular RNA from the indicated cell lines@@ - -were electrophoresed through a 1.0% denaturing agarose gel, blotted onto nylon mem-@ • Cfl ti) (1)branes, and hybridized with a 32P-labeled cyclin Dl sense RNA probe. The position ofthe @/) > < <5.0-kb (28S) RNA marker is indicated on the right. B and C, Southern blot analysis afterHindffl (B) or EcoRl (C) digestion ofgenomic DNA and hybridization with a 32P-labeledcycin Dl eDNA probe. The sizes of the major DNA bands are indicated on the right. The@@@@@ .extra bands in the AS1O,ASI4, and AS18 derivatives represent the integrated antisense 1@1@ @_—-cyclin Dl cDNA. Rb

Characterization of the Morphology, Growth Properties, andCell Cycle Kinetics of the Antisense Cydlin Di-expressing Cells.The SW480E8 antisense clones AS1O, AS14, AS18 displayed an .@@ @,_____epithelial morphology and an aneuploid karotype similar to that of the Cycun Dl@ ,,@- -parental and vector control cells (data not shown). When grown in .complete medium with 10% FCS, the SW480E8 parental cells displayed an exponential doubling time of 29 h, a saturation density of2.9 times 106 cells (on 35-mm plates), and a cloning efficiency of 44%(Table 1). They also displayed a loss of contact inhibition, because .they formed dense foci when maintained as confluent cultures (data p27@@@not shown). In parallel studies, the exponential doubling times of theASlO, AS 14, and AS18 clones were 36, 33, and 35 h, respectively, Fig.2.WesternblotanalysisfortheexpressionofcyclinDl, Rb,and @27k@@ Fiftygsgand they showed a decrease in saturation density, because these values of total cellularproteinfromeach of the indicatedcell lines were subjectedto 10%

‘ 6 . . SDS-PAGE, transferred to nitrocellulose, and immunoblotted with a 1 : 1000 dilution of

were 1.8, 2.6, and 2.0 times 10 cells (per 35-mm dish), respectively. monoclonalantihumancyclinDl, Rb,orp27@―antibody.AllofthesedatawereobtainedFurthermore, in all three of the latter clones, the cells were larger than fromthe samegel.

1571

Research. on January 25, 2021. © 1997 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table 1Growth properties and cell cycle parameter ofparental and antisense cvdin DI celllinesVariableParental

SW480E8 cellsVector control cloneAntisense

clones1OASI4AS18ASDoubling

time (h)10% FCSl%FCS

Saturation density (X l06)@@Cloning efficiency (%)Growth in soft agar (%)Tumorigenicity in nude mice (cm3)'@28

303.0

416.5

5.2/1.90.7/0.629

312.9

397.1

1.7/1.51.2/0.836

461.861.8

0/00/033

382.4

152.6

0/00/035

402.092.0

0/00/0

433942 3834 5939 51347038

4940 4846 3644 40463418

1218 1320 517 1020 5

ANTISENSE CYCLIN Dl REVERSES TUMORIGENICITY

G1 phase (%)Exponential growth'Confluent culture

S phase (%)Exponential growthcConfluent culture

G2-M phase (%)Exponential growth'Confluent culture

a Represents the number of cells at the saturation density per 35 mm-diameter plate (mean of three experiments).b Volumes of tumors removed from four individual injection sites; with some of the AS cells, small (2-mm) nodules were noted at the injection sites, the histology of which revealed

fibrotic tissue.C Cells were grown in 10% FCS and harvested during exponential growth at about 50% confluence, or 48 h after they had reached confluence. The results are the mean values of

three experiments.

rylated to the unphosphorylated bands (Fig. 2). The three antisenseclones also displayed a marked (about 5-fold) decrease in the level ofthe p27kIPI protein when compared to the control cells (Fig. 2). Therewas no change or only a slight decrease (< 1.5-fold) in the levels ofthe following proteins: cyclins E and A, CDC25A, and p2lwafl (datanot shown).

Effects of Expression of Antisense Cyclin Dl cDNA on Cyclin

Dl- and Cyclin E-associated Kinase Activity. Cyclin Dl associateswith the cyclin D1-dependent kinases CDK4 and CDK6 and activatestheir function. Therefore, we examined the effects of expression ofantisense cyclin Dl in SW480E8 cells on cyclin Dl-associated kinaseactivity. In vitro kinase assays were performed on cyclin Dl immu

noprecipitates prepared from extracts of the parental, vector control,

and antisense SW480E8 cells, using a glutathione S-transferase-Rbfusion protein as the substrate (see “Materialsand Methods―).Representative data are shown in Fig. 3. The immunoprecipitates from the

parental and vector control SW480E8 cells displayed high levels ofcyclin Di-associated kinase activity, but only very low levels were

detected with the three SW480E8 antisense clones. Densitometricanalysis indicated that the in vitro phosphorylation of the recombinantRb protein by the extracts of the latter three cell lines was only10—20%of that obtained with the control cells. The decreased in vitrokinase activity is consistent with our findings of a relative decrease inRb protein phosphorylation in vivo (Fig. 2). In contrast, in vitro assaysfor cyclin E-associated kinase activity indicated only a slight decreasein this activity in the three antisense clones, when compared to thecontrol cells (Fig. 3), possibly reflecting the somewhat slower growthof these three clones.

Expression of Antisense Cyclin Dl Reduces Anchorage-independent Growth and Tumorigenicity. The SW480E8 parental cellsexhibit anchorage-independent growth in cell culture and inducetumors in athymic (nude) mice (32). We examined whether theseproperties were altered in the derivatives of SW480E8 cells thatexpress antisense cyclin Dl RNA. As shown in Table 1, when grownin suspension in 0.3% agar in complete medium with 10% FCS, theSW480E8 control cells formed large colonies (about 3 mm in diameter) with a colony-forming efficiency of 44%. In contrast, the threeantisense SW480E8 clones showed a colony-forming efficiency ofonly 6—15%,and the average colony size was about four times smallerthan that seen with the SW480E8 cells (data not shown). Therefore,the expression of antisense cyclin Dl markedly decreased the anchorage-independent growth of SW480E8 cells.

Tumorigenicity of the SW480E8 control and antisense cyclin Dlexpressing cells was assayed in nude mice by s.c. injection of 1 timesl0@cells/injection site. The SW480E8 control cells produced tumorsat all four of the injection sites (Table 1). Thus, by 8 weeks followinginjection of the cells, s.c. tumors that ranged in volume from 0.6 to 5.2cm3 were detected. In contrast, cells from the three SW480E8 antisense clones failed to form detectable tumors at any of the injection

sites even after 16 weeks (Table 1). These results provide directevidence that expression of antisense cyclin Dl in SW480E8 cellsmarkedly inhibits tumorigenicity.

The present studies demonstrate that the stable expression of anantisense cyclin Dl cDNA in the human colon cancer cell line

LU0

C., - I-@: @, ci@ Cl) cnC,) > < < <

Fig. 3. Studies on cyclin-associated in vitro kinase activity. Top and bottom. resultsobtained when extracts of the indicated cell lines were assayed for cyclin Dl- or cyclinE-associated kinase activity. respectively.

1572

CyclinDlassociatedkinase activity

:: :::@@

Cyclin Eassociatedkinase activity DISCUSSION

Research. on January 25, 2021. © 1997 American Association for Cancercancerres.aacrjournals.org Downloaded from

ANTESENSECYCLIN DI REVERSES TUMORIGENICITY

SW480E8 leads to a decrease in the otherwise high level of expressionof the endogenous cyclin Dl protein in this cell line, and that thisdecrease is maintained through several cell passages. These derivatives also show an increase in the fraction of the cellular Rb proteinthat is hypophosphorylated and a decrease in in vitro cyclin Dlassociated Rb kinase activity. Presumably as a consequence of thesechanges, the cells of the antisense derivatives are larger in size. Theyare also impaired markedly in their growth, because they have alonger doubling time, a decrease in saturation density, and a decreasein plating efficiency. They also display a reduction in anchorageindependent growth and, what is perhaps most remarkable, a loss oftumorigenicity when injected s.c. into athymic (nude) mice. Weshould emphasize that the antisense derivatives still express appreciable levels of the cyclin Dl protein (Fig. 2). Therefore, our results arenot simply due to the fact that we have completely deleted theexpression of an essential gene. Instead, our findings provide evidencethat the relatively high level of expression of cyclin Dl seen in theparental SW480E8 cells and in about 30% of primary human colontumors (see “Introduction―and Refs. 25 and 26) plays a critical rolein maintaining their tumorigenic phenotype. The reversion of thephenotype of the antisense SW480E8 cells toward normalcy is remarkable, because the original parental cells are highly aneuploid andcarry an activating mutation in the c-K-ras oncogene and an mactivating mutation in the p53 tumor suppressor gene (32). In this sense,our results resemble those of other investigators (see Ref. 34 forreview), in which restoration of a single tumor suppressor gene to acancer cell line can block its tumongenicity despite the presence ofseveral other genetic abnormalities in the recipient cells. Nor are theimplications of the present findings on cyclin Dl restricted to theSW480E8 colon cancer cell line, given that our laboratory demonstrated that introduction of the antisense cyclin Dl cDNA into ahuman esophageal cancer cell line (HCE7), in which the cyclin DIgene is overexpressed due to amplification, led to a decreased level ofthe cyclin Dl protein, marked growth inhibition, and loss of tumorigenicity, even though in these cells the residual level of the cyclin Dlprotein also remained relatively high and these cells also retainedother genetic abnormalities (33). These findings have led us to suggestthat the intracellular circuitry of cancer cells that overexpress cyclinDl requires a higher level of this protein than that required by cellsthat developed through pathways in which this gene was not overexpressed. In this sense, the former cells are “addicted―to cyclin Dl,which may explain the profound growth-inhibitory effects we haveseen with antisense cyclin Dl in both the 5W480E8 and HCE7 cells.Additional studies are in progress to examine this hypothesis.

Previous studies indicated that when cyclin Dl was overexpressedin rodent fibroblasts, this led to a shortening of the fraction of cells inthe G1 phase of the cell cycle, which is consistent with its role inenhancing the G1-to-S progression in the cell cycle (4, 13—15).On theother hand, the increased exponential doubling time of the threeantisense SW480E8 clones obtained in the present study appeared tobe due to a lengthening of all phases of the cell cycles, not just the G1phase. The slower growth rate of the cyclin Dl antisense derivative ofHCE7 cells was also not due simply to a lengthening of the G1 phase(33). These findings suggest that the functions of cyclin Dl in malignant epithelial cells are more complex than in nontransformedrodent fibroblasts. The mechanisms responsible for these pleiotropiceffects are not, however, apparent at the present time.

Cyclin Dl complexes with and activates CDK4 and CDK6 (4). Themajor substrate for the cyclin Dl-CDK4 complex is the Rb protein,but it is not known whether Rb is the only in vivo substrate. In thepresent study, we found that cyclin Dl immunoprecipitates fromSW480E8 cells expressing antisense cyclin Dl had a marked decreasein their ability to phosphorylate the Rb protein (Fig. 2), which is

consistent with the fact that these cells had a marked reduction inexpression of the cyclin Dl protein (Fig. 2). This could explain theinhibition of growth and reversion of tumorigenicity seen in theantisense derivatives. However, our data do not exclude the possibilitythat there are other functions of cyclin Dl, in addition to phosphorylation of Rb, that play a role in growth control and oncogenesis.

The present studies may also shed light on feedback pathwaysbetween components of the cell cycle circuitry. We noted that decreased expression of the cyclin Dl protein in the three antisenseclones of SW480E8 cells was associated with decreased levels of boththe Rb and p27kiPl proteins (Fig. 2). We have also observed decreasedlevels of the Rb and @27k1Plproteins in the HCE7 antisense cyclin Dlcells (33)•4These findings complement previous evidence that increased expression of cyclin Dl in esophageal cancer and mammaryepithelial cells is associated with increased expression of the Rb and

@27k@ proteins (28—3 1 ). Therefore, these feedback pathways appar

ently exist in several cell types, perhaps to maintain homeostaticcontrol between positive and negative acting factors involved in cellcycle progression. It is curious, however, that despite the reduction inthe total level of the Rb and p27'―@proteins in the antisense cells,their growth was impaired, presumably because of the above-descnbed marked decline in cyclin DI-associated kinase activity.

Regardless of the underlying mechanisms, the present studies provide the first evidence that inhibition of the expression of cyclin Dl incolonic tumor cells that overexpress this gene tends to revert theirtransformed phenotype and inhibits their tumorigenicity. These findings may be of clinical significance because of the fact that about 30%

of human primary colon tumors display increased expression of thisgene (25, 26), and this gene is frequently expressed at high levels inseveral other types of human cancer (see “Introduction―).Takentogether, these observations suggest that therapeutic agents directedagainst cyclin Dl, or cyclin D1/CDK kinase activity, might be usefulin the therapy of the subset of human colon tumors that overexpresscyclin Dl.

REFERENCES

2.Pines, J. Cyclins: wheels within wheels. Cell Growth & Differ., 2: 305—310.1991.Reed, S. I. The role ofp34 kinases in G1 to S phase transition. Annu. Rev. Cell Biol.,8: 529—561,1992.

3. Jiang, W., Zhang, Y-J., Kahn, S., Hollstein, M., Santella, R., Lu, S-H., Harris, C.,Montesano, R., and Weinstein, I. B. Altered expression of the cyclin DI andretinoblastoma genes in human esophageal cancer. Proc. NatI. Acad. Sci. USA. 90:9026—9030,1993.

4. Sherr, C. J. Mammalian GI cyclins. 1993. Cell, 73: 1059—1065,1993.5. Hunter, T., and Pines, J. Cyclins and cancer II: cyclin D and CDK inhibitors come of

age. Cell, 79: 573—582,1994.6. Sherr, C. J. Gl phase progression: cycling on cue. Cell, 79: 551—555,1994.7. Weinstein, I. B., and Thou, P. Cell cycle controlgenedefectsand humancancer.in:

J. R. Bertino (ed), Encyclopedia of Cancer, Vol. 1, pp. 256—267.New York:Academic Press, 1996.

8. Won, K. A., Xiong, Y., Beach, D., and Gilman, M. Z. Growth regulated expressionof D-type cyclin genes in human diploid fibroblasts. Proc. Natl. Acad. Sci. USA, 89:9910—9914,1992.

9. Sewing, A., Burger, C., Brusselbach, S., Schalk, C., Lucibello, F. C., and Muller, R.Human cyclin Dl encodes a labile nuclear protein whose synthesis is directly inducedby growth factor and suppressed by cyclic AMP. J. Cell Sci., 104: 545—554,1993.

10. Baldin, V., Lukas, J., Marcote, M. i., Pagano, M., and Draetta, G. Cyclin Dl is anuclear protein required for cell cycle progression in G I. Genes Dcv., 7: 812—821,I993.

11. Lukas, J., Pagano, M., Staskova, Z., Draeua, G., and Bartek. J. Cyclin Dl proteinoscillates and is essential for cell cycle progression in human tumor cell lines.Oncogene,9: 707—718, 1994.

12. Tam, S. w., Theodoras, A. M., Shay, J. W., Draetta, G. F., and Pagano, M.Differential expression and regulation of cyclin Dl protein in normal and tumorhuman cells: association with CDK4 is required for cyclin D 1 function in G 1progression. Oncogene, 9: 2663—2674, 1994.

13. Quelle, D. E., Ashmun, R. A., Shurtleff, S. A., Kato, J-Y., Bar-Sagi, D., Roussel,M. F., and Sherr, C. J. Overexpression of mouse D type cyclins accelerates G I phasein rodent fibroblasts. Genes Dcv., 7: 1559—1571.1993.

4P.Zhouand‘N.Jiang,unpublishedstudies.1573

Research. on January 25, 2021. © 1997 American Association for Cancercancerres.aacrjournals.org Downloaded from

ANTISENSE CYCLIN Dl REVERSES TUMORIGENICITY

14. Jiang, ‘W.,Kahn, S. M., Thou, P.. Zhang, Y. J., Cacace, A. M., Infante, S. D., Santella,R. M., and Weinstein, I. B. Overexpression of cyclin Dl in rat fibroblasts causesabnormalities in growth control, cell cycle progression and gene expression. Oncogene. 8: 3447—3457,1993.

15. Resnitzky, D., Gossen, M., Bujard, H., and Reed, S. Acceleration of the Gl/Stransition by expression of cyclins Dl and E with an inducible system. Mol. CellBiol., 14: 1669—1679, 1994.

16. Pardee, A. B. GI events and regulation of cell proliferation. Science (WashingtonDC), 246: 603—608,1989.

17. Hunter, T., and Pines, J. Cyclins and cancer. Cell, 66: 1071—1074,1991.18. Motokura, T., and Arnold, A. Cyclin D and oncogenesis. Curr. Opin. Genet. Dcv., 3:

5—10,1993.19. Lovec, H., Sewing, A., Lucibello, F. C., Muller, R., and Moroy, T. Oncogenic activity

of cyclin Dl revealed through cooperation with Ha-ras: link between cell cyclecontrol and malignant transformation. Oncogene, 9: 323—326,1994.

20. Hinds, P. W., Dowdy, S. F., Eaton, E. N., Arnold, A., and Weinberg, R. A. Functionof a human cyclin gene as an oncogene. Proc. NatI. Acad. Sci. USA, 91: 709—713,1994.

21. Wang, T. C., Cardiff, R. D., Zukerberg, L., Lees, E., Arnold, A., and Schmidt, E. V.Mammary hyperplasia and carcinoma in MMTV-cyclin Dl transgenic mice. Nature(Lond.), 369: 669—671, 1994.

22. Bodrug. S. E., Warner, B. J., Bath, M. L., Lindeman, G. J., Hanis, A. W., and Adams,A. M. Cyclin DI transgene impedes lymphocyte maturation and collaborates inlymphogenesis in the myc gene. EMBO J., 13: 2124—2130,1994.

23. Lovec, H., Grzeschiczek, A., Kowalski, M., and Moroy, T. Cyclin Dl/bcl-l cooperates with myc genes in the generation of B-cell lymphoma in transgenic mice. EMBOJ., 13: 3487—3495,1994.

24. Lukas, J., Jadayel, D., Bartkoba, J., Nacheva, E., Dyer, M. i. S., Strauss, M., andBartek, 3. BCI-l/cyclin Dl oncoprotein oscillates and subverts the Gl phase controlin B-cell neoplasms carrying the t(l 1:14) translocation. Oncogene, 9: 2159—2167,1994.

25. Bartkova, J., Lukas, 1., Strauss, M., and Bartek, J. The PRAD-l/cyclin Dl oncogene

prod uct accumulates aberrantly in a subset of colorectal carcinomas. mt. J. Cancer,58: 568—573,1994.

26. Arber, N., Hibshoosh, H., Moss, S. F., Suner, T., Zhang. Y., Begg, M., Wang, S.,Weinstein, I. B., and Holt, P. R. Increased expression of cyclin DI is an early eventin multistage colorectal carcinogenesis. Gastroenterology, I 10: 669—674, 1996.

27. Muller, H., Lukas, J., Schneider, A., Warthoc, P., Bartek, J., Eilers, M., and Strauss,M. Cyclin Dl expression is regulated by the retinoblastoma protein. Proc. NatI. Aced.Sci. USA, 91: 2945—2949, 1994.

28. Arber, N., Sutter, T., Miyake, M., Kahn, S., Venkatraj, V. S., Sobrino, A., Warburton,D., Holt, P. R., and Weinstein, I. B. Increased expression of cyclin Dl and the Rbtumor suppressor gene in c-K-ras transformed rat enterocytes. Oncogene, 12: 1903—1908. 1996.

29. Doki, Y., Imoto, M., Han, E. K-H., Sgambato, A., and Weinstein, I. B. Correlationsbetween the levels of expression of cyclin Dl and other cell cycle-related genes inhuman esophageal cancer cell lines. Carcinogenesis, in press.

30. Kitahara, K., Yasui, W., Yokozaki, H., Semba. S., Hamamoto, T., Hisatsugu, T., andTahara, E. Expression of cyclin Dl, CDK4, and p27@―@1is associated with the

pI6―@@gene status in human esophageal carcinoma cell lines. J. Exp. Ther. Oncol.,1: 7—12,1996.

31. Han, E. K-H., Begemann, M., Sgambato, A., Soh, J-W., Xing, W-Q., Lui, W., andWeinstein, I. B. Increased expression of cyclin Dl in a murine mammary epithelialcell line induces p27KPl, inhibits growth, and enhances apoptosis. Cell Growth &Differ., 7: 699—710, 1996.

32. Tomita, N., Jiang, W., Hibshoosh, H., Warburton, D., Kahn, S. M., and Weinstein,I. B. Isolation and characterization of a highly malignant variant of the SW480 humancolon cancer cell line. Cancer Res., 52: 6840—6847,1992.

33. Thou, P., Jiang, W., Shang, Y-J., Kahn, S. M., Schieren, I., Santella, R. M., andWeinstein, I. B. Antisense of cyclin Dl inhibits growth and reverses the transformedphenotype of human esophageal cancer cells. Oncogene, Ii: 571—580,1995.

34. Dowdy, S., Fasching, C., Scanlong, D., Araujo, D., Livanos, E., Lai, K., Weissman,B., and Stanbridge, E. Suppression oftumorigenicity in Wilms' tumor by the pl4:pl5region of chromosome I 1. Science (Washington DC), 254: 293—295,1991.

1574

Research. on January 25, 2021. © 1997 American Association for Cancercancerres.aacrjournals.org Downloaded from

1997;57:1569-1574. Cancer Res   Nadir Arber, Yuichiro Doki, Edward K-H. Han, et al.   of Human Colon Cancer CellsAntisense to Cyclin D1 Inhibits the Growth and Tumorigenicity

  Updated version

  http://cancerres.aacrjournals.org/content/57/8/1569

Access the most recent version of this article at:

   

   

   

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

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

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

  Permissions

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

.http://cancerres.aacrjournals.org/content/57/8/1569To request permission to re-use all or part of this article, use this link

Research. on January 25, 2021. © 1997 American Association for Cancercancerres.aacrjournals.org Downloaded from