Vol. 7, 1609-1615, December 1996 Cell Growth & Differentiation 1�

Up-Regulation and Functional Role of p2lWafl/CIPI duringGrowth Arrest of Human Breast Carcinoma MCF-7Cells by Phenylacetate

Myriam Gorospe, Sonsoles Shack,Kathryn Z. Guyton, Dvorit Samid, andNikki J. Holbrook1Laboratory of Cellular and Molecular Biology, Gerontology ResearchCenter, National Institute on Aging, NIH, Baltimore, Maryland 21224[M. G., S. S., K. Z. G., N. J. H.], and Experimental TherapeuticsProgram, University of Virginia Cancer Center, Charlottesville, Virginia22908 [D. S.]

AbstractPhenylacetate (PA) and related aromatic fatty acidsconstitute a novel class of relatively nontoxicantineoplastic agents. These compounds induce tumorcytostasis and growth inhibition and differentiation ofcancer cells, but little is known regarding themolecular events mediating these biological effects.Using human breast carcinoma MCF-7 cells as amodel, we show here that PA-induced growth arrest isassociated with enhanced expression of the cyclin-dependent kinase inhibitor p21��/C1� anddephosphorylation of the retinoblastoma protein (pRB).The induction of p21WAFI/CIPI mRNA by PA wasindependent of the cellular p53 status. To directlyassess the contribution of p2iW&1/CIPI to PA-mediatedcytostasis, we compared the effects of PA in parentalMCF-7 cells and cells expressing reduced levels ofp21W�� protein (clones AS.3 and AS.4),accomplished through constitutive expression ofantisense p2IW81l/CIPI transcripts. In contrast toparental cells, AS.3 and AS.4 cells did not showreduced pRB phosphorylation following PA treatment,indicating that p2IW&l/CIPI induction by PA is requiredfor dephosphorylation (inactivation) of pRB, a knownmediator of cell cycle control. A prominent role for� in mediating PA-induced growth arrest wasfurther supported by the demonstration that embryonalfibroblasts derived from a p21WAFI/CIPI knockout

mouse (p21 mouse embryonal fibroblasts) did notgrowth arrest following PA treatment, whereas PAeffectively induced p21WAFI/CIPI mRNA and growthinhibition of the wild-type mouse embryonalfibroblasts. Taken together, our findings strongly

Received 6/14/96; revised 9/16/96; accepted 9/30/96.The costs of publication of this article were defrayed In part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 1 8 U.S.C. Section 1 734 solely to mdi-cate this fact.1 To whom requests for reprints should be addressed, at Laboratory ofCellular and Molecular Biology, Box 31 , Gerontology Research Center,National Institute on Aging, National Institutes of Health, 4940 EastemAvenue, Baltimore, MD 21224. Phone: (410) 558-8197; Fax: (410) 558-8335.

support a role for p21W&l/CIPI in the PA-mediatedinhibition of cell growth.

Introduction

Compounds that induce growth arrest and differentiationmay provide an attractive alternative to conventional cyto-toxic cancer treatment. This approach is based on the un-derlying hypothesis that driving the relatively dedifferentiatedcancer cell population through a maturation program willlessen its proliferative and invasive capacity. A number ofdifferentiation inducers have been used clinically for treat-

ment of various cancers, including butyrate, DMSO, interter-ons, 5-azacytidine, hexamethylene bisacetamide, dimethyl-formamide, and retinoids (1-6). However, their success has

been hampered by both practical concerns regarding theinability to attain effective plasma concentrations as well as

by potential toxic and carcinogenic side effects. PA2 andrelated aromatic fatty acids constitute a new class of differ-

entiation inducers that hold considerable promise because of

their relatively low toxicity at clinically effective doses. These

compounds have been shown to cause cytostasis and mat-uration of various human cancer lines, such as glioblastomas(2), leukemias (7), and prostate carcinomas (1), as well asmultiple drug-resistant cell lines (8). Moreover, PA inhibitstumor growth in vivo (2). A number of mechanisms have been

proposed to contribute to the antitumor actions of PA: (a)depletion of circulating glutamine, on which cancer cells areexquisitely dependent for nucleic acid and protein synthesis(9, 1 0); (b) inhibition of protein isoprenylation, notably that of

Ras and lamins (1 1); (c) hypomethylation of DNA (1 2); and (d)

activation of the peroxisomal proliferator-activated nuclearreceptor, a transcription factor shown to regulate the expres-

sion of genes controlling lipid metabolism (13). However, theway in which these proposed mechanisms interplay ingrowth arrest and differentiation by PA and its potential

molecular mediators is not well understood.Cellular proliferation, growth arrest, differentiation, and

death are highly regulated processes. The identification andcharacterization of a growing list of molecular effectors gov-

erning these processes has suggested numerous potentialtargets for growth-arresting agents such as PA. The ERK/

MAPK signaling cascade, which involves the sequential ac-tivation of growth factor receptors, Ras, Raf, MAPK kinase,and ERK (for review, see Ref. 14), constitutes one importantlevel of cell growth control. A different level of regulation of

2 The abbreviations used are: PA, phenylacetate; MAPK, mitogen-acti-

vated protein kinase; ERK, extracellular-regulated kinase; MEF, mouseembryonal fibroblast; RB, retinoblastoma; cdk, cyclin-dependent kinase;PGA2, prostaglandin A�; WT, wild-type; FBS, fetal bovine serum.

-.-- MCF7neo

time (days)

untr.

C FBS

PA

C FBS

1610 Up-Regulation of p21 Wafi/Cipi by Phenylacetate

A

100

� 20

0 � 4 6 8 10

PA (mM)

cell proliferation is implemented by modulators of cell cycle

progression, which is primarily driven by the orchestrated

activation of cdks. Cdks phosphorylate key regulatory fac-

tors, such as pRB, and are themselves regulated through

interaction with other proteins, which exert positive (cyclins)

and negative (cdk inhibitors) influences on kinase activities

(15). p2lWafl/ciPl (also known as Sdil , Cap2O, and Mda6) is

an inhibitor of various cdks (1 6), and its expression is en-

hanced by a variety of treatments that result in growth arrest.

Induction of p21waf1/ci�1 by DNA damage relies on the pres-

ence of the tumor suppressor protein p53 (1 7), presumablythrough interaction with p53-binding sites present in the

promoter region of the p21wAF1/CIP1 gene. However, most

inducers of p21wAFl/cIPl, including cytokines, growth fac-

tors, PGA2, and even genotoxic agents (18-20; for review,

see Ref. 21), operate to a large degree throughp53-independent mechanisms. In addition, enhanced

p21 wafi/cipi expression has been found to correlate tightly

with the process of cellular differentiation (22, 23), an obser-

vation that has led to the postulation that p21 Wafi/Cipi plays

a key role in the onset and/or maintenance of the differenti-

ated phenotype.

Here, we report that exposure of breast carcinoma MCF-7

cells to PA results in the inhibition of cell proliferation asso-

ciated with p21wafl/ciPl expression and loss of pRB phos-

phorylation, regardless of cellular p53 status. PA-mediated

inhibition of proliferation was less effective in MCF-7 cells

expressing reduced levels of p21 Wafi/cipi through constitu-

tive expression of �21wAF1/cIP1 antisense transcripts and in

embryonal fibroblasts derived from p21 WAF1/CIP1 knockout

mice. Taken together, these findings support a direct role for

� in the implementation of PA-induced growth

arrest.

ResultsPA Treatment Is cytostatic for MCF-7 Cells. The growth-

inhibitory effect of PA was examined in both MCF-7 cells with

W.r p53 status and in MCF-7 cells rendered p53 deficient

through stable transfection with the viral oncoprotein E6 (24).

As shown in Fig. 1A, PA treatment resulted in a dose-depen-

dent inhibition of growth in MCF-7 cells that was independ-

ent of cellular p53 status. Maximal inhibition without signifi-

cant toxicity was seen with a concentration of 10 m�i PA. The

Fig. 1. Inhibitory effect of PA on growth of MCF-7cultures. A, parental MCF-7 cells (neo) and E6-over-expressing MCF-7 cells (E6) were treated for 6 dayswith varying doses of PA, and cell numbers weredetermined at the end of the treatment period. B,exponentially growing MCF-7 cells (neo and E6)were treated with 1 0 mM PA, and cell numbers weredetermined every 24 h thereafter, using a hemocy-tometer.

Ti: MBP

Fig. 2. ERK activation by serum in untreated and PA-treated MCF-7cells. Following serum depletion in the absence (untr.) or presence of 10mM PA (PA) for 72 h, MCF-7 cultures were subjected to serum stimulationfor 20 mm, whereupon ERK activity was measured. ERK was immuno-precipitated, and the kinase activity present in the immunoprecipitateswas assayed as described in “Materials and Methods.” c, control.

growth curves for MCF-7 neo and MCF-7 E6 cells cultured in

10 mM PA for 6 days are shown in Fig. lB. Differences in the

cellular proliferative status become apparent after day 2 of

PA treatment.

Loss of Proliferation Is Not Due to Alterations in MAPKSignaling. In certain cell types, PA has been reported to

inhibit isoprenylation of important cellular proteins, including

the Ras oncoprotein (25). Ras plays a key role in activation of

ERK in response to serum, as well as other extracellular

signals (1 4). Therefore, as a reflection of altered Ras activity,

we tested whether PA-mediated growth inhibition was asso-

ciated with an attenuation of ERK activation by serum.

MCF-7 cells were maintained in 10 m� PA for 72 h prior to

stimulation of ERK by serum. As shown in Fig. 2, greater than

20-fold activation in ERK kinase was observed following

serum stimulation, which was not affected by pretreatment

with PA. This finding indicates that alterations in the MAPK

signaling cascades may not constitute the primary mecha-

nism of PA-induced cytostasis and led us to examine directly

the effects of PA on the regulation of components of the cell

cycle machinery.

Induction of �21wAF1/cuP1 mRNA Expression in PA-treated Cells. Since induction of p21Wafl/CiP1 correlates

with the inhibition of cell proliferation in many growth arrest

paradigms, we examined the expression of this cdk inhibitor

in PA-treated MCF-7 cells. PA caused a time-dependent

increase in p21WAF1/CIP1 mRNA expression; a maximal in-

duction of 8-fold was achieved by day 3 (Fig. 3), coincident

with cessation of growth (Fig. 1A). Given that p21wafl/ciPl

10.

-�0- E6

8

6

4

A time(days) 0 2 3

cciu[I�1

B

I

2

cl 234

p21�[�1 .�.a,O.I

zE

I-

0

cJ

0

Fig. 3. Expression of p21WAF1/CIP1 mRNA in PA-treated MCF-7 cells.MCF-7 cells were treated with 10 m� PA for the indicated times, andp21WAFh/�1 mRNA levels were studied by Northern blot analysis. Ribo-somal 18S RNA was used to normalize for RNA loading and transferringamong samples. The graph represents the quantitation of p21 WAF1/CIP1

mRNA normalized to 1 8S signals. Bars, SD; c, control. Inset, represent-ative Northern blot, MCF-7 neo cells.

issIOSOSOl0 1 2 3 4 5

time (days)

2343

J

C time (days) 0

Western p21 - LI

Immunoprecipitation: anti-cdk2 pre--immune

Fig. 4. Cdk2 expression and activity following PA treatment. A, after

treatment of MCF-7 cells with 10 m� PA for the times indicated, cdk2protein levels were analyzed by Western blot analysis. B, Cdk2 wasimmunoprecipitated in PA-treated MCF-7 cells, and cdk2 kinase activitywas monitored using histone Hi as a substrate. Incorporation of 32P onHistone Hi was quantitated using a Phosphorlmager. C, Cdk2 was im-munoprecipitated from lysates of PA-treated MCF-7 cells, and the amountof p21Wafl/CIP1 present in the Cdk2 immunocomplex was detected byWestern blot analysis of the immunoprecipitated proteins.

Cell Growth & Differentiation 1611

induction is dependent, in some instances, on the presence

of functional p53, we also examined the expression of

p21WAF1�P1 mRNA in MCF-7 cells that were rendered p53deficient through constitutive expression of the oncoprotein

E6 (24). As shown in Fig. 3, parental and p53-deficient cells

displayed similar kinetics and magnitude of p2lwafl/cIPl in-

duction. These findings are in keeping with the ability of PA

to arrest MCF-7 cells expressing E6 (Fig. 1) and suggest that

PA-dependent induction of p2lWafl/CuPl occurs independ-

ently of p53 status.

Effect of PA on cdk2 Expression and Activity.p21wa�ch1� � an established regulator of cdk2 activity. The

physical association of cdk2 with p2lwafl/ciPl has been ex-

tensively characterized in intact cell types subjected to a

variety of growth-inhibitory treatments, and it has been es-

tablished that this interaction results in the inactivation of

cdk2 (16, 1 8, 26). Therefore, we next examined whether

increased p21Wafl/CIP1 expression by PA was associated

with altered cdk2 activity. Although the expression of the

cdk2 protein does not change with PA treatment (Fig. 4A),

there was a progressive loss of cdk2 activity (Fig. 4B), which

correlated with the observed increase in p2lWafl/clPl expres-

sion (Fig. 3). By 2 days of PA treatment, cdk2 kinase activity

(measured using histone Hi as a substrate) had decreased

to 40% of that seen in untreated cultures; by day 4, PA-

treated cells showed less than 20% of the cdk2 activity seen

in untreated cultures. Since cdk2 protein levels were un-

changed by PA, we postulated that the loss in cdk2 kinase

activity was likely due to the inhibitory interaction with

p2iwa��� . To test this hypothesis, lysates from PA-treated

roo�

� 80

�60

rj 40

0)

� 20

0�time(days) 0

cells were subjected to immunoprecipitation of cdk2-con-

taming immunocomplexes followed by Western blot analysis

of cdk2-associated proteins. As shown in Fig. 4C, the pres-

ence of p21wafl/cIPl in the immunocomplex is enhanced in

PA-treated cells, presumably aiding the inhibition of cdk2

kinase activity.

Influence of Altered p21wafl/ciPl Expression on PA-mediated Effects. pRB is an endogenous substrate for

cdk2, and the pRB phosphorylation pattern is an established

indicator of the proliferative status of the cell (27). In partic-

ular, hypophosphorylated (active) pRB is found in G1 cells,

whereas hyperphosphorylated (inactive) pRB is characteris-

tic of proliferating cells in S-phase, G2, and M-phase. As

shown in Fig. 5, exposure of MCF-7 parental (neo hygro) cells

to PA resulted in the accumulation of hypophosphorylated

1612 Up-Regulation of p21 Wafi/Cipi by Phen�acetate

3 Unpublished results.

neo.hygro AS.3 AS.4

time(days)02 4 0 2 4 0 24

Fig. 5. Changes in p21 wafl/c,pl expression and RB phosphorylation fol-lowing PA treatment of MCF-7 parental, AS.3, and AS.4 cells. ParentalMCF-7 cells (neo.hygro) as well as MCF-7 cells constitutively expressinga transcript complementary to p21wAFl/cIPl mANA (AS.3 and AS.4) weretreated with i 0 mM PA for the times indicated and monitored for expres-sion of p2lwafl/ctPl and RB by Western blot analysis. Faster-migratingbands in the RB doublet represent hypophosphorylated forms; slower-migrating bands represent hyperphosphorylated forms of RB. The highermolecular weight cross-reactive band is nonspecific.

pRB (evidenced by faster migration on Western blots). Itappears, therefore, that PA-induced growth arrest is associ-ated with an increase in p21 Wafi/CipI expression, leading to

reduced cdk2 activity and inhibition of pRB phosphorylation.

A moderate loss in the total amount of pRB was noted in thePA-treated populations. To further examine the potential role

that p21 Wafi/Cipi plays in mediating growth arrest by PA, weexamined the response to PA in MCF-7 clonal populations

that produced even smaller amounts of p2lwafl/CIP1 due totheir constitutive expression of p21WAF1/CIP1 antisensetranscripts. These clonal isolates (AS.3 and AS.4) have

been described previously (28). Basal and PA-induciblep2iWafh/Cir� levels were dramatically lower in these cell lines(Fig. 5). Accordingly, PA-mediated pRB dephosphorylationwas substantially reduced in AS.3 and AS.4 cells (Fig. 5). This

finding supports the hypothesis that p2lwafi/ciPl induction

by PA constitutes an important player in the molecular eventsthat are associated with PA-mediated growth arrest. Ifp21 Wafi/Cipi is required for the execution of the growth arrestprogram, we postulated that in the presence of PA, AS.3 and

AS.4 clones would continue to proliferate, since they are onlyable to express very low levels of p21 Wafi/Cipi � Unexpect-edly, however, when compared with that seen in the parentalcells, PA treatment of AS.3 and AS.4 cells was markedlycytotoxic (not shown). This response was unusual, since PAis typically cytostatic, but it was not without precedent. We

have recently shown that PGA� treatment of MCF-7 cells,

which is also cytostatic and leads to enhanced p21 wafi/CiP1

expression, is highly toxic to AS.3 and AS.4 cells (28). Al-though, taken together with the effects of PGA�, these ef-

fects of PA raise important considerations regarding the roleof p2lwafl/CiP1 for cell survival, they pose a significant limi-tation for addressing the role of p2iwafl/CiP1 in PA-mediatedgrowth cessation. Thus, we used an alternative model, em-bryonal fibroblasts derived from a p2lWafl/CiP1 knockout

mouse (p21 “ MEFs), to address the role of p21Wafl/CIP1 in

PA-mediated growth arrest.p21��CI��deflcient Fibroblasts Show Attenuated

Growth Inhibition by PA Treatment. As shown in Fig. 6A,PA treatment of WT MEFs resulted in a time-dependentelevation of �21Waf1/CiP1 mRNA levels. Maximal p21 WAF1/CIP1

mRNA expression was reached within 2 days of treatmentand was accompanied by growth arrest (Fig. 6B). By con-

trast, PA was significantly less effective in inhibiting thegrowth of the p2iWAFl�’��deficient cells (Fig. 6B). This dif-

ference was evidenced by a clear separation of the dose-

response curves for PA-induced growth arrest in WT versus

p21 “ MEFs, with an increase in the IC50 of PA from 3.5 m�in WT MEFs to greater than 1 0 m� in p21 � MEFs.

Discussion

PA exhibits selective growth-inhibitory activity in both exper-imental models and in humans (1 , 2, 8, 25, 29, 30). In thisarticle, we provide three lines of evidence indicating thatp21Wafl��� is involved in mediating cytostasis induced byPA. First, exposure of MCF-7 cells to PA was found to resultin a time- and dose-dependent inhibition of growth associ-ated with induction of p21wafl/ciPl expression, loss of cdk2

kinase activity, and dephosphorylation of pRB. Second, areduction in p2iwafl/CiP1 expression (by expressing anti-sense p21WAF1/CIP1 transcripts) attenuated the accumulation

of the hypophosphorylated form of pRB (which otherwiseresulted from PA treatment), indicating that PA-mediatedinhibition of pRB phosphorylation can be blocked

by preventing p21Wafl/CiP1 expression. And third, PA-medi-

ated growth arrest of embryonal fibroblasts derived from

p2lWA�h/’�l knockout mice was markedly attenuated, com-pared with an effective growth inhibition by PA on the wild-type counterparts.

As mentioned above, PA treatment of AS.3 and AS.4MCF-7 cultures did not result in an increase in cell numberrelative to parental cells but, rather, resulted in increasedcytotoxicity.3 This intriguing effect is similar to that we havereported comparing the response of MCF-7 cells with

that of AS.3 and AS.4 cells treated with another growth-

inhibitory agent, PGA2, in which failure to elicit normal induc-

tion of p21 Wafi/Cipi resulted in greater toxicity (28). That

p21WAFl�’�I’l�deficient MEFs continue to grow, whereas� MCF-7 cells encounter enhanced tox-

icity, could reflect fundamental differences between PA of-

fects on transformed and untransformed cells. Alternatively,these differences could reveal adaptive changes whereincells completely devoid of p2iwafl/ciPl (p21 “ MEF5)may have developed compensatory mechanisms that AS.3and AS.4 cells have not had the opportunity to acquire.

Moreover, inhibition of p21 wafi/cipi expression throughtransfection of p21WAF1/CIP1 antisense oligonucleotides was

recently demonstrated to block growth factor-induced differ-

entiation of SH-SY5Y neuroblastoma cells and instead resultedin apoptosis (31). We hypothesize that induction of p21wafl/ciPl

during stress exerts a protective function, and cytotoxicity mayresult from the lack of this protective influence. Further analysisis required to extend this hypothesis to PA.

The precise mechanism through which PA treatment in-duces p2iwafi/ciPl expression is not understood. In vivo, PA

has been postulated to deplete circulating glutamine (9, 10).If such a mechanism were operating in this in vitro model, thelack of availability of an important nutrient (glutamine con-

0 1 2 3 4 5 0 2 4 6 8 1012

time(days) PA (mM)

Cell Growth & Differentiation 1613

Fig. 6. Effect of PA treatmenton wt and p21 � ‘ MEFs. A, The

graph represents p21WAF1/CIP1

mRNA expression in wt MEFstreated with PA. Analysis wascamed out as described in “Ma-terials and Methods,” normal-ized by hybridization to i8S, andquantitated using a Phospho-rlmager. (As expected, there

was no p2iwAF1/c�1 mRNApresent in p21 �‘ MEFs). B, Fol-lowing PA treatment of p21 �and wt MEFs PA for 6 days withPA at the concentrations mdi-cated, cells were fixed andstained with crystal violet. Cellnumbers were quantitated asdescribed in “Materials andMethods.” Bars, SD.

A B

stitutes a major source of nitrogen) would likely represent a

metabolic stress, to which cells might respond by slowing

down their rate of growth. Enhanced expression of

p2iW/c1r� following nutrient depletion has been docu-

mented (32), and the direct contribution of p21Wafl/CIP1 to

the implementation of growth arrest in the absence of an

essential nutrient has been established. For example, inresponse to treatment with N-(phosphoacetyl)-L-aspartic

acid, a drug that depletes nitrogen for nucleic acid and

protein synthesis (32), p21 “ MEFs failed to arrest,

whereas the WT MEFs showed significant inhibition of

proliferation.

The pRB protein is a key regulator of cell proliferation

(reviewed in Ref. 27). According to a well-established model,

active (underphosphorylated) pRB sequesters transcription

factor members of the E2F family, thus preventing the tran-

scription of key genes that are critical for proliferation, (i.e.,

c-myc, cyclin Dl , DHFR, and thymidine kinase; Refs. 33-35).Here, we have shown that PA treatment led to a substantial

loss of pRB phosphorylation due, at least in part, to theinhibition of pRB kinases (cdks) by p21 Wafi/Cipi � An important

question that arises from these findings centers around the

implication of pRB in PA-mediated effects: is pRB a primary

target for the effects of PA? That is, does PA treatmentrequire that pRB be activated to initiate the growth arrest?

Alternatively, is pRB dephosphorylation a secondary effect

that arises as a consequence of cellular arrest, implemented

by other means? If PA treatment has such dramatic effects

on pRB, it will be of interest to study how PA affects theproliferative capacity of cell lines lacking pRB or expressing

mutated forms of pRB. We are currently addressing this

question in several RB-deficient cell types. The result from

such studies will shed light on important aspects of the

growth-inhibitory potential of these agents.

Our findings also indicate that PA action on MCF-7 cells

occurs independent of p53, because cells lacking functional

p53 exhibit a similar response to PA treatment as do cells

with functional p53. This notion is further supported by the

fact that PA treatment also results in growth inhibition and

p21 Wafi/Cipi induction in a variety of other cell types null for

p53 function.3 Considering that mutations affecting p53

function occur very frequently (50% of all cancers), the find-ing that PA effects occur irrespective of p53 function makes

its use particularly attractive. Indeed, many conventionalchemotherapeutic agents, which act through the induction of

apoptotic cellular death, rely on the presence of p53 foreffectiveness (36). Our results reported here suggest that

PA-mediated action does not require functional p53, thusexpanding the spectrum of malignancies that might poten-tially respond to PA. Consistent with this hypothesis, ongo-

ing laboratory and clinical studies show that PA is active

against high-grade gliomas (37), a disease characterized byfrequent loss of WI p53 activity (38-41) and resistance to

conventional cytotoxic chemotherapy.

Materials and MethodsCell Culture and Treatment Condftions. Human breast carcinomaMCF-7 parental cells and derivatives expressing the viral oncoprotein E6protein (MCF-7 E6) were maintained in a humidified atmosphere contain-ing 5% CO2 in air and cultured in RPMI 1640 (Ufe Technologies, Inc.,

Bethesda, MD) supplemented with 10% FBS (Hyclone Laboratories, Lo-gan, UT) and 50 �g/ml gentamicin (Ufe Technologies). MCF-7 cells ex-

pressing antisense p21wAFl/cIPl transcripts (AS.3 and AS.4) were rou-tinely cultured in the presence of 1 50 pg/mI hygromycin B and used within4 weeks of selection (28). Hygromycin B (Sigma Chemical Co.) wasremoved from the medium prior to experimental treatment with PA. MEFs

derived from � (p21 � MEFs) or p21 knockout mice (p21 ‘ MEFs;Ref. 32) were maintained in DMEM supplemented with FBS and penicillinand streptomycin. Sodium PA (Elan Pharmaceutical Research Corp.,Gainesville, GA), dissolved in distilled water (neutralized to pH 7.4), wasadded directly into the medium to the indicated final concentrations. Cell

counts were performed using a hemocytometer.Northern and Western Blot Analysis. Total RNA was isolated with

STAT-60 B (Tel-Test, Friendswood, TX). Twenty-�tg RNA samples weredenatured, size-fractionated by electrophoresis in i .2% agarose-formal-dehyde gels, and transferred onto GeneScrean Plus nylon membranes(DuPont NEN Research Products, Boston, MA). For the detection ofp2i�l/cIF’l mRNA in MCF-7 cells, p21wAFl/cuPl cDNA was excised from

pCEP-WAF1 (1 7) and labeled with [a-32P]dCTP with a random primerlabeling kit (Boehringer Mannheim, Indianapolis, IN). For the detection of

p2i WAF1/C�P1 mRNA in MEFs and for normalization of differences inloading and transfer of RNA, oligomers complementary to the mouse

1614 Up-Regulation of p2lwafl/cuPl by Phenylacetate

i . Samid, 0., Shack, S., and Myers, C. Selective growth arrest and

phenotypic reversion of prostate cancer cells in vitro by nontoxic phar-

p21��’1�”1 mRNA (5’-CTCCGTGACGAAGTCAA.AGUCCACCGT-TCTCGGGCCTCCTGGAGACAGCC-3’) and to the 185 rRNA (5’-ACGGTATCTGATCGTCTTCGAACC-3’), respectively, were 3’ end labeled with[a-32P]dATP by terminal deoxynucleotidyl transferase (Life Technologies).

Hybridizations and washes were performed according to the method ofChurch and Gilbert (42), and the hybridization signals were quantified

using a Phosphorimager (Molecular Dynamics, Sunnyvale, CA). Signalsobtained with the p21waf1/cI�l probes were normalized to i8S signalsobtained on the same blot to control for variations in loading and transfer

among samples.For Western blot analysis of cdk2, p21wafl/cIPl , and pRB, 50-p.g sam-

pIes of total cell lysates or proteins present in cdk2-immunoprecipitateswere size-fractionated by SOS-PAGE and transferred onto polyvmnylidene

difluoride membranes using standard techniques (43). Following incuba-tion with monoclonal mouse antihuman p2iwa�ciPl (Calbiochem, LaJolla, CA) monoclonal mouse antihuman pAB, and monoclonal mouse

antihuman cdk2 (PharMingen, San Diego, CA), respectively, proteins weredetected by the enhanced chemiluminescence system following the man-

ufacturer’s instructions (Amersham, Arlington Heights, IL).

Cdk2 Immunoprecipltation and Kinase Assays. For immunoprecipi-tation of cdk2, cells were washed twice with ice-cold PBS and harvestedin odk2 lysis buffer containing 50 m� Tris-HCI (pH 7.4), 250 m�.i NaCI, and

0.1 % Triton X-100, supplemented with phosphatase and protease inhib-

itors (5 mM NaF, 0.1 m� sodium orthovanadate, 5 pg/mI leupeptin, 10pg/mI aprotinin, 50 pg/mI phenylmethylsulfonyl fluoride, and 5 pg/mIpepstatmn A). Soluble extracts were prepared by centrifugation at 16,000 xg for 1 0 mm at 4”C. Following normalization of protein content, lysateswere precleared by incubation with protein A-Sepharose (Sigma) andpreimmune rabbit serum for 30 mm at 4”C. Endogenous cdk2 was im-munoprecipitated for 3 h at 4”C using a rabbit polyclonal antihuman cdk2

antiserum (PharMingen). Immunoprecipitates were washed twice withcdk2 lysis buffer and four times with Hi kinase buffer [50 m�i Tris-HCI (pH7.4), 10 mM MgCI2, and 1 m�i OTT]. The kinase activity associated with

cdk2 immunocomplexes was assayed in 50 �.tl of Hi kinase buffer sup-plernented with 2 mM EGTA, 10 �Ci of [y-32P]ATP, and 10 �ig of histoneHi (Ambion, Austin, TX) for 30 mm at 30”C. Nonradioactive ATh was then

added to the reaction mixture to a final concentration of 30 �M to reduce

the background signal. Reactions were stopped by addition of SOS sam-pie buffer, and the reaction products were electrophoresed in i 2% poly-

acrylamide gels, visualized by autoradiography, and quantitated with aPhosphorlmager.

ERK Immunopreclpitation and Kinase Assays. Cells (60-80% con-fluent) were either left untreated or treated with PA for 3 days in complete

growth medium. During the last 18 h of the 3-day period, treated anduntreated cells were cultured in serum-free medium prior to addition ofserum (to final 10%) for 20 mm. Cells were then washed twice with

ice-cold PBS and lysed in buffer containing 20 m� HEPES (pH 7.4), 50 m�f3-glycerophosphate, i % Triton X-iOO, 10% glycerol, 2 m� EGTA, 1 m�

OTT, iO mM sodium fluoride, i m� sodium orthovanadate, 2 �M leupeptin,

2 �i aprotinin, 2 �M pepstatmn A, 1 m�a phenylmethylsulfonyl fluoride, and

0.5 �sM okadaic acid. Soluble extracts were prepared by centrifugation at1 0,000 x g for 1 0 mm at 4#{176}C.Following normalization of protein content,

endogenous ERK2 was immunoprecipitated from cell extracts using rab-bit polyclonal antibodies against p42ERK2 (Santa Cruz Biotechnology,Santa Cruz, CA). Kinase activity was assayed for 20 mm at 37#{176}Cin thepresence of 6 �Lg of substrate (myelin basic protein), 30 �i ATP, and 20

�.tCi of [y-32P]ATP in 55 �l of assay buffer [20 m� 3-(N-morpholino)pro-panesulfonic acid (pH 7.2), 2 mM EGTA, and 20 m�i MgCI2]. After corn-

pletion of kinase assays, the proteins were resolved by SDS-PAGE, andthe gels were dried and subjected to autoradiography. The incorporation

of 32P was visualized using a Phosphorlmager.

AcknowledgmentsWe thank Or. P. Leder for providing the MEFs, Dr. B. Vogelstein for

providing the pCEP4-WAFi constructs, and Dr. A. J. Fornace, Jr., forproviding the MCF-7 nec and MCF-7 E6 call lines.

References

macological concentrations of phenylacetate. J. Clin. Invest., 91: 2288-

2295, i993.

2. Sarnid, 0., Ram, Z., Hudgins, W. A., Shack, S., Liu, L, Walbridge, S.,Oldfield, E. H., and Myers, C. E. Selective activity of phenylacetate against

malignant gliomas: resemblance to fetal brain damage in phenylketonuria.

Cancer Res., 54: 891-895, 1994.

3. Sacks, L Control of normal cell differentiation and the phenotypicreversion of malignancy in myeloid leukemia. Nature (Lond.), 274: 535-

539, i978.

4. Golomg, H. M., Ratain, M. J., Mick, A., and Daly, K. The treatment ofhairy cell leukemia: an update. Leukemia (Baltimore), 6: 24-27, 1992.

5. Momparler, A. L, Rivard, G. E., and Gyger, M. Clinical trial on 5-aza-

2-deoxycytidmne in patients with acute leukemia. Pharmacol. & Ther., 30:277-286, i995.

6. Muindi, J. A., Frankel, S. A., Huselton, C., DeGrazia, F., Garland, W. A.,Voung, C. W., and Warrell, A. P. J. Clinical pharmacology of oral aII-trans-retinoic acid in patients with acute promyelocytic leukemia. Cancer Res.,

52: 2i38-2i42, 1992.7. Samid, 0., Veh, A., and Prasanna, P. Induction of erythroid differenti-

ation and fetal hemoglobin production in human leukemic cells treatedwith phenylacetate. Blood, 80: 1576-1581 , 1992.

8. Shack, S., Miller, A., Uu, L, Prasanna, P., Thibault, A., and Samid, 0.Vulnerability of multidrug-resistant tumor cells to the aromatic fatty acidsphenylacetate and phenylbutyrate. Clin. Cancer Res., 2: 865-872, 1996.9. Weber, G. Biochemical strategy of cancer cells and the design of chem-otherapy: G. H. A. memoriei lecture. Cancer Res., 43: 3446-3492, 1983.

1 0. Medina, M. A., Sanchez-Jimenez, F., Markez, F. J., Perez-Rodriguez,J., Quesada, A. A., and Nunez de Castro, I. Glutamine and glucose asenergy substrates for ehrhch ascites tumor cells. Biochem. Int., 16: 339-347, 1988.

1 i . Goldstein, J. L., and Brown, M. S. Regulation of the mevalonate

pathway. Nature (Lond.), 343: 425-430, 1990.

12. Bird, A. The essentials of DNA methyIation. Cell, 70: 5-8, 1992.

i3. Tontonoz, P., Hu, E., and Spiegelman, B. M. Stimulation of adipo-

genesis in fibroblasts by PPAR�y2, a lipid-activated transcription factor.

Cell, 79: ii47-ii56, 1994.

i4. Seger, A., and Krebs, E. G. The MAPK signalling cascade. FASEB J.,9: 726-735, 1995.

is. Morgan, 0. 0. Principles of CDK regulation. Nature (Lond.), 374:

i3i-i34, i995.

16. Xiong, V., Hannon, G. J., Zhang, H., Casso, 0., Kobayashi, R., andBeach, 0. p2i is a universal inhibitor of cyclmn kinases. Nature (Lond.), 366:

70i-704, i993.

i 7. El-Dairy, W. S., Tokino, T., Velculescu, T., Levy, 0. B., Parsons, R., Trent,

J. M., Un, 0., Mercer, W. E., Kinzler, K. W., and Vogelstein, B. WAF1, apotential mediator of p53 tumor suppression. Cell, 75: 8i 7-825, i 993.

18. Gorospe, M., Liu, V., Xu, 0., Chrest, F. J., and Holbrook, N. J. Inhi-bition of G1 cyclin-dependent kinase activity during prostaglandin A�-mediated growth arrest. Mol. Cell. Biol., 16: 762-770, i996.

19. Liu, V., Martindale, J. L, Gorospe, M., and Holbrook, N. J. Regulation

of p21wAFl/cIPl expression through mitogen-activated protein kinase sig-

nailing pathway. Cancer Res., 56: 1-5, 1996.

20. Michieli, P., Chedid, M., Lin, 0., Pierce, J. H., Mercer, W. E., and Givol,

0. Induction of WAF1/CIP1 by a p53-independent pathway. Cancer Res.,54: 339i-3395, 1994.21 . Gorospe, M., Martindale, J. L, Sheikh, M. S., Fornace, A. J., Jr., andHolbrook, N. J. Regulation of p21 Cipi/WAF1 expression by cellular stress:

p53-dependent and p53-independent mechanisms. Mol. Cell. Differ., 4:47-65, 1996.

22. Macleod, K. F., Sherry, N., Hannon, G., Beach, 0., Tokino, T., Kinzler,K., Vogelstein, B., and Jacks, T. p53-dependent and independent expres-sion of p2i during cell growth, differentiation and DNA damage. GenesDcv., 9: 935-944, 1995.

23. Nakanishi, M., Adami, G. R., Robetorye, R. S., Noda, A., Venable, S.

F., Dimitrov, 0., Pereira-Smith, 0. M., and Smith, J. A. Exit from G0 andentry into the cell cycle of cells expressing p21 Sdii antisense RNA. Proc.Nati. Acad. Sci. USA, 92: 4352-4356, 1995.

Cell Growth & Differentiation 1615

43. Harlow, E., and Lane, 0. Antibodies: A Laboratory Manual. Cold

Spring Harbor, NV: Cold Spring Harbor Laboratory, 1988.

24. Fan, S., Smith, M. L., Rivet, 0. J., Duba, 0., Zhan, Q., Kohn, K. W.,Fornace, A. J., Jr., and O’Connor, P. M. Disruption of p53 function san-

sitizes breast cancer MCF-7 cells to cisplatin and pentoxifyllmne. CancerRes., 55: 1649-1654, 1995.

25. Shack, S., Chen, L-C., Miller, A. C., Danesi, A., and Samid, 0.Increased susceptibility of ras-transformed cells to phenylacetate is as-sociated with inhibition of p2i� isoprenylation and phenotypic reversion.

Int. J. Cancer, 63: 124-129, 1995.

26. Zhang, H., Xiong, V., and Beach, 0. Proliferating cell nuclear antigenand p21 are components of multiple cell cycle kinase complexes. Mol.

Biol. Cell, 4: 897-906, 1993.

27. Weinberg, A. A. The retinoblastoma protein and cell cycle control.Cell, 81: 323-330, 1995.

28. Gorospe, M., and Holbrook, N. J. Role of p21 in prostaglandin A�-mediated cellular arrest and death. Cancer Aes., 56: 457-479, 1996.

29. Samid, 0., Shack, S., and Sherman, L T. Phenylacetate. A novelnontoxic inducer of tumor cell differentiation. Cancer Res., 52: 1988-

i992, i992.

30. Thibault, A., Samid, 0., Cooper, M. A., Figg, W. 0., Tompkins, A. C.,

Patronas, N., Headlea, 0. J., Kohler, 0. A., Venzon, 0. J., and Myers, C.E. Phase I study of phenyl-acetate administered twice daily to patientswith cancer. Cancer (Phila.), 75: 2932-2938, i995.

31. Poluha, W., Poluha, 0. K., Chang, B., Crosbie, N. E., Schonhoff, C. M.,Kilpatrick, 0. L, and Ross, A. H. The cyclmn-dependent kinase inhibitor

p2i�’1 is required for survival of differentiating neuroblastoma cells.Mol. Cell. Biol., 16: 1335-1341, 1996.

32. Deng, C., Zhang, P., Harper, J. W., Elledge, S. J., and Leder, P. Micelacking p2iCIPl/WAF1 undergo normal development, but are defective in

G1 checkpoint control. Cell, 82: 675-684, i995.

33. Oswald, F., Lovec, H., Moroy, T., and Upp, M. E2F-dependent reg-ulation of human MYC: trans-activation by cyclins 01 and A overrides

tumour suppressor protein functions. Oncogene, 9: 2029-2036, i994.

34. Nevins, J. A. E2F: alink betweanthe Rbtumor suppressor protein andviral oncoproteins. Science (Washington DC), 258: 424-429, 1992.

35. Lamangue, N. B. DRTF1IE2F: an expanding family of heterodimerictranscription factors implicated in cell cycle control. Trends. Biochem.

Sci., 19: i08-ii4, i994.

36. Kastan, M. B., Canman, C. E., and Leonard, C. J. p53, cell cyclecontrol and apoptosis: implications for cancer. Cancer Metastasis Rev.,14: 3-i5, 1995.

37. Prados, M. 0., Spence, A., Schold, C., Mehta, M., Kuhn, J., Rector, 0.,Chang, S., and Gilbert, M. A Phase II trial of phenylacetic acid for therecurrent malignant glioma, preliminary report of the North American braintumor consortium. Proc. Am. Soc. Clin. Oncol. Annu. Meet., 288: 156,

1996.

38. Louis, 0. N. The p53 gene and protein in human brain tumors. J. Neu-ropathol. & Exp. Neurol., 53: ii-2i, i994.

39. van Meyel, 0. J., Ramsey, 0. A., Casson, A. G., Keaney, M.,

Chambers, A. F., and Caimcross, J. G. p53 mutation, expression, andDNA ploidy in evolving gliomas: evidence for two pathways of progres-sion. J. NatI. Cancer Inst., 86: 1011-1017, 1994.

40. Van Meir, E. G., Kikuchi, T., Tada, M., Li, H., Diserens, A. C., Wojcik,B. E., Huang, H. J., Friedmann, T., de Tribolet, N., and Cavenea, W. K.Analysis of the p53 gene and its expression in human glioblastoma cells.

Cancer Res., 54: 649-652, i994.

41 . Anker, L, Ohgaki, H., Ludeke, B. I., Herrmann, H. 0., Kleihues, P., andWestphal, M. p53 protein accumulation and gene mutations in human

glioma cell lines. nt. J. Cancer, 55: 982-987, i993.

42. Church, G. M., and Gilbert, W. Genomic sequencing. Proc. NatI.Acad. Sci. USA, 81: 1991-1995, 1984.