Adenovirus mediated over expression of the cyclin/cyclin dependent kinase inhibitor, p21 inhibits vascular smooth muscle cell proliferation and neointima formation in the rat carotid artery model of balloon angioplasty

Adenovirus-mediated over-expression of the

cyclin/cyclin-dependent kinase inhibitor, p21

inhibits vascular smooth muscle cell

proliferation and neointima formation in the rat

carotid artery model of balloon angioplasty.

M W Chang, … , K Barton, J M Leiden

J Clin Invest.

1995;

96(5)

:2260-2268.

https://doi.org/10.1172/JCI118281

.

Vascular smooth muscle cell (VSMC) proliferation after arterial injury is important in the

pathogenesis of a number of vascular proliferative disorders, including atherosclerosis and

restenosis after balloon angioplasty. Thus, a better understanding of the molecular

mechanisms underlying VSMC proliferation in response to arterial injury would have

important therapeutic implications for patients with atherosclerotic vascular disease. The

p21 protein is a negative regulator of mammalian cell cycle progression that functions both

by inhibiting cyclin dependent kinases (CDKs) required for the initiation of S phase, and by

binding to and inhibiting the DNA polymerase delta co-factor, proliferating cell nuclear

antigen (PCNA). In this report, we show that adenovirus-mediated over-expression of

human p21 inhibits growth factor-stimulated VSMC proliferation in vitro by efficiently

arresting VSMCs in the G1 phase of the cell cycle. This p21-associated cell cycle arrest is

associated both with significant inhibition of the phosphorylation of the retinoblastoma gene

product (Rb) and with the formation of complexes between p21 and PCNA in VSMCs. In

addition, we demonstrate that localized arterial infection with a p21-encoding adenovirus at

the time of balloon angioplasty significantly reduced neointimal hyperplasia in the rat carotid

artery model of restenosis. Taken together, these studies demonstrate the important role of

p21 in regulating Rb phosphorylation and cell cycle progression in VSMC, and suggest a

novel cytostatic gene therapy approach for […]

Research Article

Adenovirus-mediated Over-expression

of the Cyclin/Cyclin-dependent

Kinase

Inhibitor, p21 Inhibits Vascular Smooth Muscle Cell Proliferation

and Neointima

Formation in the Rat Carotid Artery Model of Balloon

Angioplasty

MarkW.Chang,* Eliav Barr,* Min Min Lu,* Kevin Barton,* and Jeffrey M. Leiden**

*Departments of Medicine and *Pathology, University of Chicago, Chicago, Illinois 60637

Abstract

Vascular smooth muscle cell (VSMC) proliferation after arterial

injury

isimportant inthepathogenesis ofanumber of vascular proliferative disorders, including atherosclerosis and restenosis after balloon angioplasty. Thus,abetter

un-derstanding of the molecular mechanisms underlying VSMC proliferation in response to arterial injury would have important therapeutic implications for patients with atherosclerotic vascular disease.Thep21 protein isa nega-tive regulator of mammalian cell cycle

progression

that functions both by inhibiting cyclin dependent kinases (CDKs) required fortheinitiation of Sphase, and by bind-ingtoand inhibiting the DNA polymerase 8 co-factor, prolif-erating cellnuclearantigen (PCNA).Inthisreport,weshow thatadenovirus-mediated over-expression ofhumanp21 in-hibitsgrowth factor-stimulated VSMC

proliferation

in vitro by efficiently arresting VSMCs in the G1phase ofthe cell cycle.Thisp21-associated cell cyclearrestisassociatedboth withsignificant inhibition of the

phosphorylation

ofthe reti-noblastoma gene product (Rb) andwith the formation of complexes between p21 and PCNAinVSMCs.Inaddition,

wedemonstratethatlocalized arterial infectionwitha

p21-encoding adenovirusatthe time of balloon

angioplasty

sig-nificantly reduced neointimal

hyperplasia

intheratcarotid artery model of restenosis. Taken

together,

these studies demonstrate the

important

role of

p21

in

regulating

Rb

phosphorylation

and cell cycle

progression

in

VSMC,

and suggestanovel

cytostatic

genetherapy

approach

for

reste-nosis and related vascular

proliferative

disorders. (J. Clin. Invest. 1995. 96:2260-2268.) Key words: restenosis * gene therapy *

adenovirus

* vascularsmooth muscle cells * cell cycle

Introduction

Vascular smooth muscle cells

(VSMCs)'

playacritical role in the normal and

pathological

responses of the arterial wall to

AddresscorrespondencetoJeffreyM. Leiden, MD, PhD, Department

ofMedicine, UniversityofChicago,5841 SMaryland Ave., Chicago,

IL60637.Phone: 312-702-1919;FAX: 312-702-1385.

Receivedfor publication2 June 1995 andacceptedinrevisedform

8August 1995.

1. Abbreviations used in thispaper: CDK, cyclin-dependent kinase; PCNA, proliferating cell nuclearantigen; PFU, plaqueforming units; PTCA, percutaneous transluminal coronaryangioplasty;Rb, retinoblas-toma;VSMC,vascular smooth muscle cells.

injury (

1).

VSMCs are normally present in the arterial tunica media in the resting (GO) state where they regulate vascular

tone andblood flow. Arterial injury leads to the proliferation of medial VSMCs and their subsequent migration from the mediato theintima,where they proliferate and elaborate extra-cellular matrix ( 1-3 ). While this VSMC proliferative response isimportant in arterial wound healing, it also plays an important role in thepathogenesisofanumberofcardiovasculardiseases, including atherosclerosis and restenosis which occurs in 30-50% ofpatients followingpercutaneous transluminal coronary angioplasty

(PITCA)

(2-5). Molecular approaches designed

to inhibit VSMC proliferation by targeting specific cell cycle regulatory pathways couldboth further elucidate the pathogene-sisof vascularproliferative disorders and lead to the rational designof novel gene-based therapiesfor these diseases.

Previous studies have demonstrated that many growth

fac-tors can stimulate theproliferation ofVSMCs in vitro and in vivo (4-8). However, the nuclear cell cycle regulatory mecha-nisms that govern the proliferative responses of VSMCs to growth factor stimulation and arterial injury have remained largelyunknown. We have recently demonstratedthat the reti-noblastomageneproduct (Rb)is an important negative

regula-tor of VSMC proliferation (9). In resting

(GO/Gi)

VSMCs, Rb is present in the unphosphorylatedform and, in that form, isapotent inhibitor of VSMC cell cycle progression. Phosphor-ylation (and inactivation) of Rb in response to growth factor stimulationis requiredfor VSMC totraversethe GI/S check-point of the cell cycle and proliferate. Consistent with these findings, localized infection of the arterial wall with an

adenovi-rus encoding aconstitutively activenonphosphorylatable form of Rb significantly reduced medial VSMC proliferation and restenosis intwo animal models of balloon angioplasty (9).

Althoughthesepreviousstudiesclearly demonstrate the im-portance of Rb incontrollingVSMCcell cycle progression, the molecularpathwaysthatregulate Rbphosphorylationin VSMC remain unknown. Previous work in other cell types has sug-gested that the formation ofcyclinD/cyclin-dependent kinase (CDK) complexes

with

serine/threonine kinaseactivity is

re-quiredforboth Rbphosphorylation, andfor.theability of cells

to traverse the

GI/S

checkpoint of the cell cycle (10-15). Moreover, cyclinD/CDK complexes can phosphorylate Rb in vitro leading to the suggestion that these complexes may di-rectly regulate the phosphorylation and activity of Rb in

re-sponsetogrowthfactor stimulation(14, 15).The kinase activi-ties of the cyclinD/CDKcomplexes can themselves be nega-tively regulated by several different molecules including

p16,

p21,

andp27 (16-23).

p21,

also known as Cip 1 (CDK-inter-acting protein) (17), Sdil (senescent cell-derived inhibitor) (19), and

WAFI

(wild-type

p53-activated

fragment) (20), is

an important inhibitor of cell cycle progression in fibroblasts and alsoplaysacritical role inprotecting cells againstcertain types ofinjury.Forexample, UV-inducedDNAdamage leads

J.Clin. Invest.

0 The AmericanSocietyforClinicalInvestigation,Inc.

0021-9738/95/11/2260/09 $2.00

to the induction of p53 which, in turn, activates transcription ofp21,resultingin cellcycle arrest and DNA repair prior to S phase-dependent chromosomal replication (24-26). Although p21 monomers can associate with active cyclinD/CDK

com-plexes inproliferating fibroblasts (27), over-expression of p21 in such fibroblasts has been shown to potently inhibit the in vitro kinase activity of these cyclin/CDK complexes and to arrestthese cells in the G1 phase of the cell cycle (17-20, 27) Thus, it has been suggested that p21 may inhibit cell cycle progression by inhibiting the cyclinD/CDK-dependent phos-phorylation of Rb (17-20, 27). However, to date this effect has not beendemonstrateddirectly in mammaliancells. In addi-tion to itsability to inhibit the kinase activities of cyclin/CDK complexes, p21 has also beenreported to bind to and inhibit the activity of the DNApolymerase 6 co-factor, PCNA (28-30). Recent studies have suggested that different regions of the p21 protein are required for its CDK and PCNA inhibitory activities(30). Thus, p21 appears to inhibit cell cycle progres-sion byatleasttwoindependent molecular mechanisms.

Thecytostatic and cytoprotective activitiesofp21 ledus to consider the hypothesis that forced over-expression of p21 might represent anovel means ofmodulating thepathologic proliferativeresponseofVSMCs to arterialinjury.Inthe studies described in this report, we have demonstrated that adenovirus-mediated over-expression ofp21 inhibits VSMC proliferation in response to serum stimulation in vitro. Moreover, we have shown thatthis effectis associated with acompleteinhibition of Rbphosphorylation and with the formation of p21/PCNA complexesin the cellsprogrammedtooverexpressp21. Finally, localized infection of the arterial wall at the time of balloon angioplasty with a replication-defective adenovirus encoding p21 markedly inhibited restenosis inthe ratcarotidarterymodel ofvascular injury. Taken together, these studies demonstrate theimportantrole of p21 inregulating VSMCproliferationin vitro and invivo,and suggest thatadenovirus-mediated overex-pression of p21 may represent anovelcytostatic genetherapy approachfor restenosis and related vascularproliferative disor-ders.

Methods

Constructionandpurification ofrecombinantadenoviruses.Adp2l and AdlacZareEl-andE3-deletedreplication-defective adenovirusvectors

derived from Ad5 sub360.Adp2l contains the humanp21cDNA(18) under the transcriptional control of the human elongation factor-la (EFla) gene promoter (31) and the human 4F2 heavy chain gene

transcriptionalenhancer(32).AdlacZ contains the bacterial lacZgene under thetranscriptionalcontrol ofthe chicken

fi-actin

gene promoter

and the cytomegalovirus transcriptional enhancer (33). Both viruses

wereprepared andgrown ashigh titer stocksin293 cellsasdescribed previously (9). Allvirus stockswerepurified bycentrifugation in dis-continuous CsCl gradients and dialyzed againstaHepes-buffered saline solution(9).

Cell culture andinfection. Isolation and infection ofprimary rat aortic VSMCs were performed as described previously (9, 34) For measurementsof cellproliferation,DNAsynthesis,and cellcycle

analy-sis passage 3 VSMCs wereincubatedin serum-free medium for 48 h before infection with eitherAdp2lorAdlacZ. Thisprotocolresults in

70-90% ofthe cultured cellsaccumulatinginG0/GIphasesof the cell

cycleasassessedby propidium iodide stainingand FACSanalysis (9).

24 hafter infection, Adp2l-orAdlacZ-infected cells were stimulated

toproliferate byincubation ingrowthmedium(45%DME,45% Hams

F-12,and 10%FBS) [GIBCOBRL,GrandIsland, NY) (9).To assay

[3H]thymidine incorporation, VSMCs were pulse-labeled for 4 h in

growth medium containing methyl-3H thymidine (1

_ttCi/ml,

2Ci/mmol; Amersham,Arlington Heights,IL)asdescribed (9, 35). Each experi-ment wasperformed intriplicate in 6-well tissue culture plates (Falcon, Franklin Lakes, NJ).

Propidium iodide staining and FACS analysis. VSMCswerefixed

overnightat 40C with 75% ethanol and stained for 30 min atroom

temperature with propidium iodide (50

_/ug/ml).

Cells were analyzed with a BectonDickinson FACScan and CellFit software. A totalof4. x 104cells werecounted for each sample.

Balloonangioplasty and adenovirus infection of rat carotid arteries. Adult maleSprague-Dawley rats werehoused and caredforaccording to NIH guidelines in the Carlson animal facility of theUniversity of Chicago. Localized infection ofballoon-injuredratcarotid arterieswas

performedasdescribed previously (9). Briefly, Sprague-Dawley rats

weresubjected to balloonangioplasty of the leftcommoncarotid artery usinga2 French Fogarty catheter(36). Immediatelyfollowing injury, 2 x 109 PFU ofAdp2l or AdlacZ in a total volume of 0.2 ml was

instilled into a1-cm segment of the distal common carotid artery for5

min using a24-gauge intravenous catheter.

Immunoprecipitation and immunoblot analyses. For immunoblot

analysisof p21 expressionin cultured cells, cell lysateswereprepared

asdescribed previously (9, 37) and 100

_kug

total protein from each sample was fractionatedby SDS-PAGE in 15% gels andsubjectedto

immunoblot analysisusing arabbitpolyclonala-p21 antibody (Phar-mingen, San Diego,CA) ( 1:1,000 dilution).Aperoxidase-labeled, goat a-rabbit antiserum(1:3,000 dilution) (GIBCO BRL)wasused in

con-junction with the ECL chemiluminescence system (Amersham) for de-tectionof boundprimary antibody. For detection of p21 /PCNA

com-plexes in VSMCs,quiescent(GO/GI)primaryrataortic VSMCswere

infected for 24 h with 40 PFU per cell of AdlacZorAdp2l, stimulated

toproliferate by incubation for 24 h in medium containing 10% FCS, and cell lysateswerepreparedin Tweenlysis buffer (50mMHepes, pH=7.5, 150mMNaCl, 2.5mMEGTA, 1 mMEDTA,0.1% Tween-20) (1.5 x 106 cells in 1 ml oflysis buffer). Cell lysateswere pre-clearedby incubationat4°Cfor 1 h with 100utlof protein A-sepharose beads(Pharmacia), mixed with 1

_/il

of rabbit polyclonal a-p21 antibody (Pharmingen), and incubated at4°C for 60 min. Theresulting beads werewashed three times in Tweenlysis buffer, added to pre-cleared infected cell lysates containing 700

jHg

of proteinasdetermined witha

commercially available kit(Pierce,Rockford,IL) and incubated for 60

minat 4°C withgentlerocking. The beads werewashed three times withTweenlysis buffer and theimmunoprecipitated proteinswere

re-leasedby boilinginSDS-PAGEloading buffer, fractionatedby electro-phoresisin 15%denaturing SDS polyacrylamide gels andsubjectedto

immunoblot analyses usingapolyclonal rabbita-p21 antibody ( 1:1,000 dilution) (Pharmingen) or a mouse a-PCNA mAb (1:100 dilution) (Santa Cruz, Santa Cruz, CA) asdescribed above. Forassessment of p21 expression in adenovirus-infectedratcarotidarteries, arterial

seg-ments wereharvested4dfollowing balloon angioplasty and adenovirus infectionasdescribed(9). Two arterial segments infected withAdp2l

and 2segments infected with AdlacZ(60 mg totalwetweight for each pair) wereimmediately placed in 200

_jul

of ice-coldhomogenization buffer (100mMNaCl, 1 mM EDTAand 20mMTris (pH8.0),1 mM

iodoacetamide, 0.23mMPMSF,77

_1LM

aprotinin, 1

_liM

leupeptin, 0.7 ,uMpepstatin A,and 77

_ptM

benzamidine). After briefhomogenization with a polytron homogenizer (5 s X 3)on ice, 200 _IL of 2X SDS-PAGE loading buffer (37)wasadded to eachsample. 200

,g

of total

protein from each sample was subjected to immunoblot analysis as

described aboveusingthea-p21 polyclonal antibody.

For assessment ofRb expression, passage 3 VSMCs cells were incubated in serum-freemediumfor96 h, infectedwith adenovirusas

described aboveand,24 hafterinfection,werestimulatedtoproliferate

byincubation ingrowthmedium. Celllysateswerepreparedasdescribed

previously (9, 37) and protein corresponding to 5 X 105 cells from eachsamplewasfractionatedby electrophoresisin7.5%SDS

polyacryl-amidegelsandsubjectedtoimmunoblotanalysis usingana-Rb

peroxidase-la-beled,goat a-mouse IgG antiserum(1:2000 dilution) (Gibco/BRL)in

conjunction withECL chemiluminescence system(Amersham).

Histologicalassessmentof restenosis. Ratcarotid arteries were

har-vested 20 dafter balloon injury and adenovirus infectionasdescribed

previously (9). 5-pm sections from paraffin-embedded arteries were

stained with hematoxylin and eosin and the neointimal and medial

boundariesweredeterminedoncodedslides byan investigator blinded

to the experimental conditions. Areas and ratios were determined by

digital planimetry of tissue sections using the Image Pro-Plus image

analysissystem(Fryer Co, Chicago, IL). The I/M ratiosweremeasured

from six sections ofeach artery subjected toballoon angioplasty and

adenovirus infection. The mean of these determinations was used to

calculatethe I/M cross-sectional ratios for eachanimal.

Statistical analysis. All values are expressed as mean±SEM. Statis-tical analyseswere performed using Sigmaplotsoftware(Jandel

Scien-tific, Corte Madere, CA) on a Macintosh Power PC 8100 computer. TheStudent'st testwasusedtoevaluatedifferences between the

experi-mentalgroups.

Results

Over-expression of p2] inhibits VSMC proliferation and cell cycle progression in response to growth factor stimulation in

vitro. To directly testthe effects of over-expression of p21 on

cell cycle progression inVSMCs,we constructeda

replication-defective adenovirus, Adp2l, that contains the human p21 cDNA (18) under the transcriptional control of the cellular elongation factor la gene promoter (31) and the 4F2 heavy chain gene enhancer (32) (Fig. I A). We have demonstrated previously that infection of primary cultured VSMC with 20-40 plaque-forming units (PFU) per cell of such adenovirus vectors results in the transduction of>90%of cultured VSMCs (9). To precisely define the levels and time course of p21

expression following infection with this virus, quiescent (GO/ GI)primary rat aortic VSMCs were infected with 20PFUper

cell of either Adp2l or a control virus, AdlacZ that encodes the bacterial,8-galactosidase enzyme (33). Cell lysates

pre-pared 24-48hours after infectionweresubjectedtoimmunoblot analysis using a polyclonal antibody reactive with human p21. Adp21-infected VSMCs expressed markedly elevatedlevels of p21 as compared toboth uninfected and AdlacZ-infected

con-trol cells (Fig. 1 B). High levels of p21 expression were

ob-served within 24 hoursofinfectionand the levels ofp21

expres-sion increased further between 24 and 48 h after infection (data not shown).Of note, insomeexperiments, both uninfectedand

AdlacZ-infectedcell lysates displayed low levels ofexpression of a 21kD protein that co-migrated with in vitro translated

hu-man p21 (Fig.1 _{B). This may reflect cross-reactivity of the}

polyclonal a-human p21 antibody with the endogenous rat p21 expressedin these cells. In addition, all ofthecell lysates

contained a faintnonspecificbandof slower mobility that

cross-reacted withthe polyclonal a-p2I antiserum. From these

experi-ment weconcluded that Adp21 can be used to program

high-level p21 expression following infection of quiescent VSMCs

in vitro.

To determine the effects of over-expression of p21 on

growth factor-stimulated VSMC proliferation in vitro, primary

rat VSMC were madequiescent byincubation for 48 h inserum

free medium. This treatment reproducibly results in 70-90%

of the culturedVSMCsbecoming arrested in theGO/GIphases

of the cell cycle as assayed by propidium iodide staining and

FACS

analysis

(9). The

quiescent

VSMC were _infected with 10, 20, or 40 PFU per cell of

Adp2

1, and then stimulated to

A

EFNa Pr 4F2Enh

_PolyA

Adp2l

MapUnits 01 p21 9

AE3

"7

100

B

Uninfected

I|**Ad/8cZ

_

_*AigZAdp21

C

S. ,_

Figure 1. Expression ofp21 in VSMCs and rat carotid arteries following infection with Adp2l. (A) Schematic representation ofAdp2l. Adp2l contains the human p21 cDNA (p21) under the transcriptional control of the human Elongation Factor la promoter (EFla Pr) andthehuman 4F2 heavy chain transcriptional enhancer (4F2 Enh). Adenovirus map

units are shown below each map. (B) Expression of p21 afterinfection of cultured rat aortic SMCs with Adp21.Cell lysates were prepared from cultured primary rat aortic SMCs 48 h after infection with20PFU per cell ofAdp21 or the AdlacZ control virus or from uninfected VSMCs and subjected to immunoblot analysis using an a-p21 antiserum. (C)

Expression ofp21 afterAdp2l infection ofballoon-injured ratcarotid arteries. Immediately after balloon angioplasty, rat carotid arteries were infected with 2x

109

PFU ofAdp2lor AdlacZ. Lysates of thearteries were prepared 4 d after infection and subjected to immunoblot analysis using ana-p21 antiserum. In vitro translated (IVT) p21 proteinwas

included on the immunoblots in B and C as a marker. The positions of

the

p21

bands are indicated with arrows.

proliferate by exposure to medium containing 10% fetal calf

serum (FCS). Control cells were infected with 40PFU per cell

of AdlacZ. As shown in Fig. 2 A, the AdlacZ-infected cells

proliferated rapidly during the first 48 h after serumstimulation,

undergoing approximately two doublings. In contrast,infection

with Adp2l resulted in dose-dependent reductions in VSMC

proliferation. At both the 24 and 48 h time points, eachof the

Adp21-infected

cell cultures (10, 20, and 40 PFU per cell)

demonstrated significantly less proliferation than the control

AdlacZ-infected cells (P < 0.02). Infection with40 PFU per

cell ofAdp2l resulted in a 60% reduction inVSMC

O

h

&-0 E

z

0

a

GOIGI

*AdkcZ4Opful

OAdp2ll0PEWe

V . 20OpU/

o * 4opuas

_ 40P*T

0 24 48

Time(hou)

24 h

Ad/acZ

GIG1

GOG1

I

G2!M S

Adp2l

GWG1

G2M G2M

DNA

Content

(PI)

Figure2.Inhibition of VSMCproliferationfollowinginfection with

Adp2l. (A) Replicatecultures ofquiescent, serum-starved, rataortic SMCswereinfected with differentamountsofAdp2l (open symbols) orAdlacZ (closed circles)andthenstimulatedtoproliferateby

incuba-tionin 10%FCS. Cellswereharvested0, 24,and 48 h after serum-stimulationandcell numbers determinedusingahemocytometer.Cell

viabilityasassessedbytrypanblueexclusionwas>97% in all cultures atthe end of theexperiment. The data is shownasmean±SEM. Error barsareincludedoneach datapointbuton somepointsaretoosmall

tobe visualized.(B) Replicatecultures ofquiescent,serum-starved

primaryrataortic SMCswereinfectedwith eitherAdp2l (open sym-bols)orAdlacZ(closed circles)andthen serum-stimulatedasdescribed in A.Cellswerepulse-labeledfor 4 h with[3H]thymidineand harvested

0, 24,and48 h after serum-stimulation. The dataareshownas mean [3H]thymidine incorporated±SEM.Error barsareincludedoneach data

pointbut insome cases aretoosmalltobe visualized. (C)Replicate

cultures ofquiescent,serum-starvedprimaryrataortic SMCswere

in-<0.0001)(Fig. 2 A). Importantly, > 97% of both the AdlacZ-andAdp2l-infected cells were viable at the end of these experi-mentsasdetermined by trypan blue exclusion (data not shown). Thus, the inhibition of cell proliferation observed afterAdp2l infection reflectedacytostaticasopposed to a cytotoxic effect.

If theobserved decrease in proliferation produced by

over-expression of p21 reflected its capacity to inhibit the kinase activities of cyclinD/CDK complexes, the Adp2l-infected VSMCs should be arrested in the GI phase of the cell cycle. Thishypothesis was tested directly using two different experi-mental approaches. First, we measured DNA synthetic activity ([3H]thymidine incorporation) after growth factor stimulation ofbothAdp2l- and AdlacZ-infected primary rat aortic SMCs (Fig. 2 B). Infection with Adp2l resulted in statistically sig-nificant (P < 0.05), dose-dependent reductions in [3H]-thymidineincorporation which paralleled the decreased prolifer-ative activity of these cells. Infection with 40 PFU/cell of Adp2l led to a >70% reduction in growth factor-stimulated [3H] thymidine incorporation as compared to the AdlacZ-in-fectedcontrol cells at 48 h (P<0.001) (Fig. 2 B).In asecond set ofexperiments, quiescent VSMCs were infected with 40 PFU/cell of Adp2l or AdlacZ, and stimulated to proliferate by incubation in 10% FCS. Cell cycle progression was assayed by propidium iodide staining and FACS analysis (Fig. 2 C). Before serum stimulation (0 h) 72% of the AdlacZ- and 71% of the Adp2l-infected VSMCs were in the GO/GI phases of the cell cycle. After 24 h of serum stimulation, 60% of the AdlacZ-infectedcells had progressed into the G2/M + Sphasesof the cell cycle (i.e., only 40% remained in GO/GI). In contrast, 69% of the Adp2l infected cells remained in GO/GL. Thus, Adp21 infection resulted in a 15-fold reduction in cell cycle progression during the first 24 h ofserumstimulation. Similar results were observed 48 hours after serum stimulation. How-ever, by this time point, the AdlacZ-infected cells were no

longer synchrously dividing and many cells had traversed an

entire cell cycle making precise quantitationofthese samples difficult. Takentogether, these results demonstrated that

over-expressionof p21 arrestsVSMCs in GI, before entry into the Sphase ofthe cell cycle.

Over-expression of p21 inhibits the growth factor-stimu-latedphosphorylationof Rbin VSMCs. Ithas previouslybeen suggestedthatoneof the mechanisms by which over-expression of p21 might inhibit cell cycle progression is by interfering directly with the cyclinD/CDK-mediated phosphorylation of Rb (17-20, 27). Although previous studies have shown that p21 caninhibit thecyclin/CDK dependent phosphorylation of Rb in vitro (17), a similar effect has notbeen demonstrated directlyin mammalian cells. Totesttheeffects of over-expres-sion of p21onRbphosphorylationin responsetogrowth factor stimulation, quiescent (GO/GI) primary ratVSMCs were in-fected with 40 PFU per cell of either Adp2l orAdlacZ, and then stimulated to synchronously enter and traverse the cell cycle byexposureto 10% FCS.Theexpressionand phosphory-lation of Rb after serum-stimulation was assessed by

immu-fected with40 PFU/cell of eitherAdp2l orAdlacZand then

serum-stimulated for 24 hasdescribed in A. Cellswerestainedwithpropidium

iodide andanalyzed byFACS withaBecton DickinsonFACScan and

CellFitsoftware.

A

B

I

E

I2

a

20 15

10

5

0

a

0 24 48

Timn (hour)

C

Time (hours):

0 24 48 0 24

48

Rbphos=

_

Rb

AdiacZ

Adp2l

Figure3.Over-expression of p21 inhibitsRbphosphorylationingrowth factor stimulated VSMCs. Passage3primary rataortic SMCswere

arrested inthe GO/GI stagesof the cellcyclebyincubation in serum-freemedium, infected with 40PFUpercell of AdlacZorAdp2l,and 24 hlaterstimulatedtoproliferate byincubation in mediumcontaining 10% FCS. Cellswereharvested 0, 24,and48 h after serum-stimulation

and cell lysates subjectedtoimmunoblotanalysiswithana-Rb mAb asdescribedpreviously (9). Unphosphorylated (Rb)and phosphory-lated formsofRb (Rbphs)areshownbyarrowstothe left of the

autoradiogram.

noblot analysis of whole cell extracts (Fig. 3). As described

previously, quiescent VSMCs contain only unphosphorylated

Rb (9). Serumstimulation, which results in therapid progres-sion of VSMCs into the S phase of the cell cycle (9), was

associatedwith theconcomitant, progressive phosphorylationof Rb both in uninfected VSMC(datanotshown)and inVSMCs infected with AdlacZ(Fig. 3).Thisfindingwasconsistent with

previousobservations that have demonstratedanimportantrole for Rb inregulatingVSMCproliferationinresponsetogrowth factor stimulation (9). In contrast, infection of the quiescent VSMCs with Adp2l completely inhibited the growth

factor-stimulated phosphorylationof Rb. Even 48hafterserum

stimu-lation, thepreponderance of Rb in these cells remained in the

unphosphorylated state (Fig. 3). These results demonstrated

directly thatover-expression ofp21 resultsindecreased phos-phorylationof Rb in vivo. When takentogetherwithour

previ-ousfindingthatunphosphorylatedRb inhibits VSMC cellcycle progression atthe GI/S checkpointof the cellcycle(9), they suggestedthattheabilityofp21 toinhibit VSMCproliferation

inresponsetogrowthfactor stimulationreflects,atleast inpart,

its ability toinhibit thecyclin/CDK-mediated phosphorylation

of Rb.

A physicalassociation between p21 and PCNA inAdp2J infected VSMCs. In additionto its abilitytoinhibit the kinase activities of cyclinD/CDK complexes, p21 can bind to and inhibit theactivityof the DNApolymerase6co-factor,PCNA

(28-30). Inhibition ofPCNA-dependentDNA synthesis may therefore reflect a second mechanism by which p21 inhibits

cell cycle progression. Recent studies havedemonstrated that differentregionsof the p21 proteinare involvedinbindingto

CDKs andPCNA, thereby suggestingthatp21 mayinhibit cell

cycle progression byatleasttwodistinct molecular mechanisms

(30).Totestdirectlywhetheroverexpressionofp21inVSMCs results in the formation ofp21/PCNAcomplexes, quiescentrat

aortic SMCswere infected with 40 PFUpercell ofAdp2l or AdlacZ and then stimulatedtoproliferatefor 24 hbyincubation in mediumcontaining 10% FCS. Lysates prepared fromthese

infectedcellswereimmunoprecipitated withana-p21 antibody and theimmunoprecipitates subjectedtoimmunoblot analyses

with either a-p21 (Fig. 4, left) or a-PCNA (Fig. 4, right) antibodies. Consistent withourprevious experiments, (Fig. 2A)

p21 wasimmunoprecipitated from theAdp2l-infected butnot

from theAdlacZ-infected VSMCs (Fig. 4, left panel). More

importantly,PCNAwasco-immunoprecipitated withp21 from

Adp21-infected but not from AdlacZ-infected VSMC lysates

7

CL

21kD- *4-p21

IP: a-p21 Immunoblot: a-p21

-o-PCNA

3YkD

-a-p21

a-PCNA

Figure4. Aphysicalassociation between p21 and PCNA in

Adp2l-infected VSMCs. Quiescent (GO/GI) primary rat aortic smooth muscle cells wereinfected with 40 PFU per cell ofAdp2l orAdlacZ. 24h afterinfection,cells werestimulated to proliferate by incubation in mediumcontaining10% FCS. Cell lysates prepared 24 h after stimula-tion were immunoprecipitated with an a-p21 antibody and the resultant immunoprecipitates were subjected to immunoblot analysis with a-p21 (left) or a-PCNA(right) antibodies. PCNA was co-immunoprecipitated withp21 in the Adp2l- but not in the AdlacZ-infected VSMCs. Size markersin kD areshowntothe leftof eachautoradiogram.

(Fig. 4, right panel). Thus, p21 forms complexes with PCNA followingAdp2l infection of VSMCs.

Over-expression of p21 inhibits neointimal hyperplasia in

theratcarotidarteryinjurymodel of restenosis. Theratcarotid artery balloon injury model represents a well-characterized, highly reproducible vascular proliferative disorder that is depen-dent on VSMC proliferation and migration (9, 36, 38-40). Previous studies using this modelhavedemonstrated that medial VSMC proliferation begins within 2 days of arterial injury, reaching a peak within 4 d. By 20 d after balloon angioplasty, nearly all of the injured arteries develop a stable neointimal lesion and demonstrate noevidence of VSMC proliferation in either the neointima or the media of the vessel wall (9, 36, 38-40).

Todeterminedirectlytheeffects of p21 over-expression on neointimaformation,ratcarotid arteries were subjected to bal-loon angioplasty andimmediately infected with 2 x 109PFU of eitherAdp2l orAdlacZ. We havepreviouslydemonstrated that this protocol results in the transduction of>70% of the medial VSMC in the injuredarterial segment (9). To demon-strateover-expression of recombinant p21 in these experiments, injured arterial segmentswereharvested 4 d afterinfectionand crudelysates prepared from these arteries were assayed for p21 expression byimmunoblotanalysis.AsseeninFig.1 C, arteries infected withAdp2l expressedmarkedly elevated levels of p21

ascomparedtotheAdlacZ-infectedarteries. Immunoblots from the AdlacZ- and Adp21-infected arteries containedequivalent levels ofasecond band of slowermobility.This band did not represent nativeratp21 because native rat p21 migrates with identical mobility tohuman p21 inourSDS-PAGE conditions (seeFig. 1 B).However,it may representadenatured form of

ratp21 or, alternatively non-specific binding of the antibody reagents used in theseimmunoblots.

In asecond series of experiments, rat carotid arteries were subjectedtoballoon angioplasty and immediately infected with

2 x 109PFU of either Adp2l orAdlacZ. Arteries were har-vested20 d after balloon injury and restenosis, as determined by the neointima/media (I/M) arearatio,was assessedby digital planimetry. Carotid arteries from the AdlacZ-infected control

arteries displayedanI/M ratio of 1.01±0.04 (n = 5). In

1.0

-0.8

-0.6

-0.4

0.2

-

O-I

*

Ad/acZ Adp2l

group (P < 0.001) (Fig. 5). Thus, localized ini replication-defective adenovirus encoding huma

time of arterialinjuryresulted inasignificantredui tima formation in theratcarotidarterymodel of V ation andrestenosis.

Severalexperiments wereperformedtoasses: localized arterial infection withAdp2l. First, n(

Figure5. Inhibition of neointima formation by Adp2l infection following balloon angi-oplasty in theratcarotidarteryinjury model. (A-D) Representative photomicrographs of

ratcarotid arteries 21 d following balloon injury and infection with 2x 109 PFU of AdlacZ (A and B)orAdp2l (C and D). Arrows denote the internal elastic lamina whichseparates

themedial and intimal layers of the arterial wall. The bars areequivalentto640

_MtM

in A and C andto50_OMin Band D. (E) Neointima formation in Adp2l- and AdlacZ-infected

ratcarotid arteries 20 d after balloon angioplasty. Immediately following balloon injury,

ratcarotid arterieswereinfected with 2 x 109 PFU of either the AdlacZ control virus(n

= 5)orAdp2l (n= 6).20 d after injury, the cross-sectionalareasof the neointimaand mediaweremeasured by digital planimetry in representative sections of the injured vessels. Thedataareshownasmean±SEM. Anasterisk signifies difference (P< 0.001)as com-paredtoAdlacZ-infected vessels.

fectionwith a formedby anindependent pathologist didnotreveal any

evi-in p21 at the denceofgross ormicroscopic tissuepathology in the

Adp2l-ctionin neoin- infectedrats.Moreover,serumelectrolytes,renal and liver

func-SMCprolifer- tion tests,completebloodcounts andclottingparameterswere allnormal 20daysafterballoonangioplastyandAdp2l infec-sthesafetyof tion (datanotshown). Finally, as shown inFigs. 5, C and D

ecropsies per- therewas noevidence of localized arterialinflammation,

necro-E

S

E

o

E

sis oraneurysm formation in the Adp21-infectedarteries 20 d

afterballoon angioplasty.

Discussion

Mammalian cells have evolved a set of specific and distinct

molecular responses to pathologicstimuli. Some cellsincluding VSMCs,fibroblasts, and lymphocytes proliferate in responseto

injury and foreign pathogen invasion. In other cases, such as

UV-inducedDNA damageorexposure to steroids, thesesame

cellscanundergo cell cyclearrestandevenapoptotic cell death. Although largely adaptive, these responsestoinjuryareinsome casespathological,leadingto avariety of human diseases. Thus, forexample, dysregulatedcellproliferationhasbeen associated

with both autoimmune disease (41) andmalignancy (42). Simi-larly,inappropriate apoptotic cell death may be associated with

avariety of neurodegenerative disorders (43). In the caseof VSMCs, unregulated cell proliferation in response to arterial injury has been implicated in theetiologyofavarietyof

vascu-larproliferative diseases including atherosclerosis and

resteno-sisfollowingballoon angioplasty (2-5).

Theplasticity of injury responses in asinglecell type sug-gested to us that it might be possible to manipulate these

re-sponsesbyover-expressing specificmolecules involved in regu-lating these pathways. In the studies described in this report, wehave shown thatover-expression ofonesuchmolecule, the cyclin/CDK inhibitor, p21 canfundamentallyalter the prolifer-ative responses of VSMCs to growth factors in vitro and to

arterialinjury in vivo. Adenovirus-mediatedover-expressionof p21 arrested VSMCs inGI,andsignificantly decreased VSMC proliferation. These growth inhibitory effects were associated both with an inhibition of Rb phosphorylation in these cells and with the formation ofp21/PCNA complexes. Inaddition, adenovirus-mediatedover-expressionofp21 inthe arterial wall

at thetimeof balloonangioplasty significantly inhibited neoin-tima formation in the rat carotid artery model of restenosis. Thesefindingshaveanumber of basic scientific andtherapeutic implications.

Previous studies have demonstratedimportantroles fortwo tumor suppressor genes, p53 and Rb in negatively regulating VSMC proliferation in vitro and in vivo (9, 44). It is also known thatp53canactivatetranscription ofp21 inseveral cell types (20, 24, 25) Moreover, over-expression ofp21 can di-rectly inhibit the kinase activities ofcyclinD/CDK complexes and the activity of PCNA in vitro (17-20, 27). When taken togetherwithourfindingthatover-expressionofp21 (a)leads

tothe formation ofp21/PCNAcomplexes,(b)inhibits Rb phos-phorylation,and (c) inhibits cellcycleprogressionin VSMCs, these findings define a distal regulatory pathway for VSMC proliferation in response to growth factors and arterial injury (Fig. 6). Future studies designed to examine how specific growth factor receptorsmodulate this cellcycle regulatory path-way in VSMCs should allow the construction of a complete map of the pathways leading from cell-surface receptors to nuclear cellcycle regulatorymolecules in these cells.

Inaddition to defining the molecularpathways underlying VSMCproliferation,ourfindings also have clinical significance. Coronary artery disease (CAD) due to atherosclerosis is the

leadingcause of morbidity and mortality in the United States. PTCA remains a mainstay of therapy for symptomatic CAD, with more than 400,000 procedures expected in 1995 in the United States alone. While theprocedureisinitiallysuccessful

n

Active

4

Active

Active

Inactive

Win

Inactive

...

...Iv

Inactive

G1I'-

S

Figure6.Thedistal cell cycle regulatory pathway inVSMC. Schematic illustrationofaportion oftheregulatory pathwaycontrolling the GI to

S transitionof the cellcycleinVSMC. p53 isknown toupregulatep21

transcription.p21 inhibits the kinase activity of the cyclinD/CDK

com-plexes.cyclinD/CDK complexesintheir activeformsphosphorylate (P)and inactivate theretinoblastoma protein(Rb). In its

unphosphory-latedform,Rb inhibits(11)VSMCs from traversingthe GI to S

check-point ofthe cellcycle. p21alsobindstotheDNApolymerase 6

co-factor, PCNA and inhibits its activity.

in relieving arterial stenoses in the vast majority of patients, clinically significantrestenosis continuestocomplicate the pro-cedure in upto40%ofcases(2-5). Numerouspharmacologic strategies including antiplatelet agents, anticoagulants, angio-tensin-converting enzyme inhibitors andcytotoxic agents have failed to significantly reduce the rate ofrestenosis following PTCA(45-49). Recently, a number of gene based therapies for restenosis have been described (9, 39, 40, 50-52). Adenovirus-mediated transferof the herpes simplex virus thymidine kinase gene(HSV-tk)followedbythesystemicadministration of gan-ciclovir has been showntoinhibit restenosis in both the rat and pigmodels of arterial injury (39, 40, 52). Although effective, there are several disadvantages of this cytotoxic approach as compared to cytostatic gene therapy using p21. First HSV-tk therapy involvestheinduction of VSMC death which is associ-ated with concomitantintravascular inflammationand the poten-tial for medial necrosis and aneurysm formation. In addition, thisapproach requires systemic ganciclovir therapy, which has been associated withneutropenia, thrombocytopenia, and

ven-triculararrhythmias in humans (53, 54). Biopolymer-mediated delivery of antisense oligonucleotides directed to c-myb or

PCNA andcdc2,havealso both been reportedtoinhibit neointi-malhyperplasiaintheratcarotid arteryinjurymodel (50, 51). However,recentstudieshavesuggestedthatantisense oligonu-cleotides maycausedegradationof multiple RNAspecies and have important non-specific effects on intracellular and cell

surface proteins (55-57). In addition, antisense oligonucleo-tides may be subject to significant batch-to-batch variability and to date it has been impossible to efficiently deliver antisense oligonucleotides to the vascular wall using nonviral, catheter mediated approaches (55-57). Finally, adenovirus mediated over-expression of a non-phosphorylatable constitutively active form of Rb (HAL\Rb) has also been shown to significantly reduce restenosis in both the rat carotid and pig iliofemoral artery models of balloon angioplasty (9).

Acytostatic gene therapy approach using catheter-mediated delivery ofAdp2l offers aclinically applicable, effective and non-toxic treatment for vascular proliferative disorders. The de-gree of inhibitionof restenosis achieved by Adp2l gene transfer in our studies (46%) was comparable to that achieved with over-expression ofHALARb (9), aswell aswith HSV-tk gene transfer withsystemic ganciclovir therapy (39, 40, 52), and the intraluminal delivery of antisense oligonucleotides (50, 51). Furthermore, administration of Adp2l was not associated with significantinflammatory responses norwith clinicaltoxicity in the presentstudy. From a clinical standpoint, the fact that Adp2l appearstoinhibittwoproliferative pathways (cyclin/CDKs and PCNA) may make it a more potent cytostatic agent than Ad-HAARb. On the other hand, the more pleiotropic activities of Adp2l may result in more undesirable side effects in vivo. It is also possible that co-administration of Adp2l andAdHAARb may prove to be more efficacious than the administration of eithervirus alone. The resolution of thesequestions will await the results of further pre-clinical trials which are currently in

progress. Finally, because cyclin/CDK kinases are important regulators ofproliferation inawidevariety of cell types, adeno-virus mediatedover-expression ofp21 mayprove useful for the

treatmentof other human diseases associated withdysregulated cellproliferation. Similarly, thefindingthatover-expression of both p21 and Rb can be used to provide effective cytostatic genetherapy suggests that other cellcycle regulatorymolecules suchasp15orp27, either alone, orincombination may

repre-sentuseful cytostatic genetherapyreagents.

Acknowledgments

This workwassupportedin partby agrant from the National Heart, Lung and Blood Institute (1ROlHL54592-01) toJ. M. Leiden and a

gift fromthe Falk Trust.

References

1.Ross,R. 1993. Thepathogenesisof atherosclerosis: aperspective for the

1990's. Nature(Lond.) 362:801-809.

2.Forrester, J. S., M. Fishbein,R.Helfant, and J.Fagin. 1991. Aparadigm

for restenosis basedoncellbiology:clues for thedevelopmentofnewpreventive therapies.J. Am. Coll. Cardiol. 17:758-769.

3.Ip,J.H.,V.Fuster, L.Badimon,J. Badimon,M. B.Taubman,and J. H. Chesebro. 1990.Syndromesof accelerated atherosclerosis: role of vascularinjury

andsmooth muscle cellproliferation.J. Am.Coll. Cardiol. 15:1667-1687. 4.Schwartz, R. S., D. R. Holmes, and E. J. Topol. 1992. The restenosis paradigm revisited:analternativeproposal for cellular mechanisms. J.Am.Coll. Cardiol. 20:1284-1293.

5.Liu,M.W., G. S. Roubin, and S. B. King. 1989. Restenosis after coronary

angioplasty.Potentialbiologicdeterminants and role of intimalhyperplasia. Circu-lation. 79:1374-1387.

6.Fingerle, J.,R.Johnson,A.W.Clowes,M. W.Majesky,and M.A.Reidy.

1989. Role ofplatelets in smooth muscle cellproliferationandmigrationafter vascularinjuryinratcarotid artery.Proc.Natl. Acad. Sci. USA. 86:8412-8416. 7. Libby, P., D. Schwartz, E. Brogi, H. Tanaka,and S. K. Clinton. 1992.

Acascade model forrestenosis: A special case of atherosclerosis progression.

Circulation.86(supplIII):IH-47-III-52.

8.Barr,E., and J. M. Leiden. 1994. Somatic gene therapy for cardiovascular disease: Recent advances. Trends. Cardiovasc. Med. 4:57-63.

9.Chang,M.W.,E. Barr,J. Seltzer, Y. Q. Jiang, G. J. Nabel, E. G. Nabel, M. S.Parmacek,and J. M.Leiden. 1995. Cytostatic gene therapy for vascular

proliferativedisorders withaconstitutivelyactive formof the retinoblastoma gene product. Science (Wash. DC). 267:518-522.

10.Nasmyth, K. 1993. Control of the yeast cell cycle by the Cdc28protein

kinase.Curr. Opin. Cell Biol. 5:166-179.

11.Hinds P. W., S. Mittnacht, V. Dulic, A. Arnold, S. I. Reid, and R. A. Weinberg. 1992. Regulation of retinoblastoma protein functions by ectopic expres-sion of humanscyclins. Cell. 70:993-1006.

12. Ewen, M. E., H. K. Sluss, C. J. Sherr, H. Matsushime, J. Kato, and D. M.Livingston. 1993. Functional interactions of the retinoblastomaproteinwith mammalian D-type cyclins. Cell. 73:487-497.

13.Dowdy, S. F., P. W. Hinds, K. Louie, S. 1. Reed, A.Arnold,and R. A. Weinberg. 1993. Physical interaction of the retinoblastoma protein with human Dcyclins.Cell. 73:499-51 1.

14. Matsushime, H., M. E. Ewen, D. K. Strom, J. Y. Kato, S. F. Hanks, M. F. Roussel, and C. J. Sherr. 1992. Identification and properties ofanatypical catalytic

subunit(p34PSK-J3/CDK4)for mammalian D type G1cyclins.Cell.71:323-334. 15. Kato, J., H. Matsushime, S. W.Hiebert, M. E.Ewen, and C. J. Sherr. 1993. Direct binding of cyclin Dtotheretinoblastoma geneproduct(pRb) and pRb phosphorylation by thecyclin D-dependent kinase CDK4. Genes & Dev. 7:331-342.

16.Serrano, M., G. H. Hannon, and D. Beach. 1993. Anewregulatory motif incell-cyclecontrolcausingspecificinhibition ofcyclinD/CDK4. Nature(Lond.)

366:704-707.

17.Harper, J. W., G. R. Adami,N. Wei, K. Keyomarsi, and S. J.Elledge.

1993. Thep21 CDK-interacting protein cipI is apotent inhibitor ofG1 cyclin-dependentkinases. Cell. 75:805-816.

18.Xiong, Y., G. J. Hannon, H. Zhang, D. Casso, R. Kobayashi, and D. Beach. 1993.p-21 is a universal inhibitor ofcyclin kinases. Nature (Lond.)

366:701-704.

19.Noda,A., Y.Ning,S. F.Venable, 0.M.Pereira-Smith,and J. R. Smith. 1994. Cloningofsenescentcell-derived inhibitors of DNA synthesis using an expressionscreen.Exp. Cell Res. 211:90-98.

20. El-Deiry, W.S.,T. Tokino, V. E.Velculescu,D. B.Levy, R.Parsons, J.M. Trent, D. Lin,W. E. Mercer, K. W. Kinzler, and B. Vogelstein. 1993. WAFI, apotential mediator of p-53tumorsuppression.Cell. 75:817-825.

21.Polyak,K., M. H. Lee, H. Erdjument-Bromage, A. Koff, J. M. Roberts, P.Tempst, and J. Massague. 1994.Cloningof p27KiP',acyclin-dependent kinase inhibitor and apotential mediator of extracellular antimitogenic signals. Cell. 78:59-66.

22.Toyoshima,H., and T. Hunter. 1994. p27,anovelinhibitor ofGI cyclin-CDKproteinkinaseactivity,is relatedtop21.Cell. 78(67-74).

23.Matthias,P., and 1. Herskowitz. 1994.Joiningthecomplex:

cyclin-depen-dent kinaseinhibitory proteinsand the cellcycle.Cell.79:181 -184.

24.Dulic, V.,W. K.Kaufmann, S. J. Wilson,T. D. Tlsty, E. Lees, J. W.

Harper,S. J.Elledge,and S. I. Reed. 1994.p-53-dependentinhibition of cyclin-dependentkinase activities in human fibroblasts duringradiation-induced G1

arrest.Cell. 76:1013-1023.

25.El-Deiry,W.S.,J. W.Harper, P. M.O'Connor,V. E.Velculescu,C. E.

Canman,J.Jackman,J. A. Pietenpol,M.Burrell, D.E.Hill,Y.Wang,and B.

Vogelstein. 1994. WAFl/CIPl is induced in p53-mediated G1 arrest and

apoptosis. Cancer Res. 54:1169-1174.

26.Li, R.,S.Waga,G. J.Hannon,D.Beach,andB.Stillman. 1994. Differen-tial effectsbythep21 CDK inhibitoronPCNA-dependentDNAreplicationand

repair.Nature(Lond.)371:534-537.

27. Zhang, H., G. J. Hannon,and D. Beach. 1994. p21 containing cyclin

kinases exist in bothactiveand inactivestates.Genes& Dev. 1750-1758. 28. Waga, S., G. J. Hannon, D. Beach,and B. Stillman. 1994. The p21

inhibitor ofcyclin-dependentkinases controls DNAreplication by interaction with PCNA. Nature(Lond.). 369:574-578.

29.Flores-Rozas, H.,Z.Kelman, F. B. Dean, Z.Q.Pan, J. W.Harper,S. J.

Elledge,and M.O'Donnell,J.Hurwitz. 1994.CDK-interacting protein 1directly

binds withproliferatingcell nuclearantigenandinhibits DNAreplication cata-lyzed bythe DNApolymerase.Proc.Natl. Acad. Sci. USA. 91:8655-8659.

30. Luo, Y., J. Hurwitz, and J. Massague. 1995. Cell-cycle inhibition by

independent CDK and PCNA binding domains in p2lCip. Nature (Lond.).

375:159-161.

31. Kim,D. W., T. Harada. I. Saito, and T.Miyamura. 1993. Anefficient

expressionvectorfor stableexpressioninhuman liver cells. Gene. 134:307-308. 32. Tripathy, S. K.,E. Goldwasser, M.-M. Lu,E. Barr, and J. M. Leiden. 1994.Stabledeliveryofphysiologiclevels of recombinanterythropoietintothe

systemiccirculation byintramuscularinjectionofreplication-defective

adenovi-rus.Proc.Natl. Acad. Sci. USA. 91:11557-11561.

33.Kozarsky,K., MM. Grossman, and J.M. Wison. 1993.

Adenovirus-medi-atedcorrectionof thegenetic defect inhepatocytes from patients with familial

hypercholesterolemia. Somat.Cell Mol. Genet. 19:449-458.

densitydependence of smoothmusclealphaactinexpressionin cultured rat aortic smoothmusclecells. J. Cell Biol. 107:299-306.

35. Owens,G.K., A.Loeb,D.Gordon,and M. M. Thompson. 1986.

Expres-sion of smooth muscle-specific a-isoactinin cultured vascular smooth muscle

cells:relationshipbetween growthandcytodifferentiation. J.Cell Biol. 102:343-352.

36. Clowes,A.W., M.A.Reidy, andM. M.Clowes.1983. Kinetics ofcellular proliferation after arterial injury. I.Smooth muscle growth in theabsence of endothelium. Lab Invest. 49:327-333.

37.Wang, C.-Y., B.Petryniak,C.Thompson, W.Kaelin,and J. M.Leiden. 1993. Regulation ofanEts-related transcription factor, Elf-I by bindingtothe retinoblastoma protein. Science (Wash. DC.) 260:1330-1335.

38. Lindner, V., N. E. Olson,A. W.Clowes,and M. A.Reidy. 1992. Inhibition ofsmoothmuscle cellproliferation in injuredratarteries.J.Clin.Invest. 90:2044-2049.

39.Guzman,R. J., E. A. Hirschowitz, S. L. Brody, R. G.Crystal, S. E.

Epstein, and T. Finkel. 1994. Invivo suppression of injury-induced vascular

smooth muscle cell accumulation using adenovirus-mediated transferof theherpes simplexvirus thymidine kinasegene. Proc. Natl. Acad. Sci. USA. 91:10732-10736.

40.Chang,M.W.,T, Ohno,D.Gordon,M.-M.Lu, G.J.Nabel,E.G.nabel,

and J. M. Leiden.1995. Adenovirus-mediatedtransfer of theherpes simplex virus thymidine kinase gene inhibits vascular smooth muscle cellproliferation and neointimaformation following balloon angioplasty ofthe ratcarotidartery.Mol.

Med. 1:172-181.

41.Steinman,L.1995. Escape from "horror autotoxicus:" pathogenesisand treatmentof autoimmunedisease. Cell. 80:7-10.

42. Bishop,J. M. 1991.Molecular themesinoncogenesis. Cell. 64:235-248. 43.Heintz,N. 1993. Cell death and the cellcycle:arelationshipbetween

transformationandneurodegeneration? TrendsBiochem. Sci. 18:157-159.

44.Speir, E.,R.Modali,E. S.Huang,M. B.Leon,F.Shawl, T. Finkel,and

S.E.Epstein. 1994. Potential role ofhumancytomegalovirusandp53interaction incoronaryrestenosis.Science(Wash.). 265:391-394.

45. Thornton, M. A., A. R. Gruntzig, J. Hollman, S. B. King, and J. S. Gouglas. 1984. Coumadin andaspirininpreventionofrecurrenceafter

translumi-nal coronaryangioplasty:arandomizedtrial.Circulation.69:721-727.

46.MERCATOR StudyGroup. 1992. "Doesthe newangiotensinconverting enzyme inhibitor cilazapril prevent restenosis afterpercutaneous transluminal coronaryangioplasty"Results of the MERCATORstudy:amulticenter,

random-ized double-blind placebocontrolledtrial. Am.Heart. 117:777-782.

47.Ellis, S. G., G. S. Roubin,J.Wilentz,J.S. Douglas,and S. B. King. 1989.

Effect ofthe18-24hourheparin administration forpreventionofrestenosisafter uncomplicatedcoronaryangioplasty.Am. Heart. 117:777-782.

48. Landau,C.,R. A.Lange,and L. D.Hillis. 1994.Percutaneoustransluminal

coronaryangioplasty.N.Engl.J.Med. 330:981-993.

49. Muller, D. W., S. G. Ellis, and E. J. Topol. 1991. Colchicine and

antineo-plastic therapy fortheprevention of restenosis afterpercutaneous coronary inter-ventions.J. Am.Coll.Cardiol. 17 (SupplB):126B-131B.

50.Simons, M.,E. R.Edelman,J. L.DeKeyser,R. Langer, and R. D.

Rosen-berg.1992.Antisensec-myboligonucleotides inhibitintimal arterial smooth

mus-cle cellaccumulation in vivo. Nature(Lond.). 359:67-70.

51. Morishita,R.,G.H.Gibbons,K. E.Ellison,M.Nakajima,L. Zhang, Y. Kaneda, T.Ogihara, and V.J. Dzau.1993. Singleintraluminal deliveryof anti-sensecdc2kinaseandproliferating-cellnuclearantigen oligonucleotidesresultsin chronic inhibitionof neointimalhyperplasia.Proc.Natl.Acad.Sci. USA. 90:8474-8478.

52.Ohno, T.,D.Gordon,H.San,V.J.Pompili,M. J.Imperiale,G. J.Nabel,

and E.G. Nabel. 1994.Genetherapy for vascular smooth muscle cell proliferation afterarterialinjury.Science(Wash. DC). 265:781-784.

53.Cohen,A.J.,B.Weiser, Q. Afzal, andJ.Fuhrer. 1990. Ventricular tachy-cardia intwopatients with AIDS receiving ganciclovir. Aids. 4:807-809.

54. Faulds, D., andR.C. Heel, 1990. Ganciclovir: Areview of its antiviral

activity,pharmacokinetic propertiesandtherapeutic efficacyincytomegalovirus infections. Drugs. 39:597.

55.Epstein, S.E., E. Speir, and T. Finkel. 1993. Do antisense approaches to theproblem ofrestenosis make sense?Circulation. 88:1351-1353.

56.Stein,C.A.,and Y.-C.Cheng.1993.Antisenseoligonucleotidesas

thera-peuticagents: Is thebulletreally magical?Science(Wash. DC). 261:1004-1012.

57.Guzman,L.A., C.L.Garrel,E.J.Poptic,P. E.DiCorleto,and E. J.Topol. 1994. Despitein-vivo cellular and nuclearuptake, antisense oligonucleotides do

nothaveasignificant anti-proliferative effect after vascular injury. Circulation. 90:I-147.