valproic acid induces growth arrest, apoptosis, and...

Valproic acid induces growth arrest, apoptosis, andsenescence in medulloblastomas by increasinghistone hyperacetylation and regulatingexpression of p21Cip1, CDK4, and CMYC

Xiao-Nan Li,1,2 Qin Shu,1,2 Jack Men-Feng Su,1,2

Laszlo Perlaky,2 Susan M. Blaney,2

and Ching C. Lau1,2

1Laboratory of Molecular Neurooncology, Cancer GenomicsProgram and 2Texas Children’s Cancer Center, Texas Children’sHospital, Baylor College of Medicine, Houston, Texas

AbstractValproic acid is a well-tolerated anticonvulsant that hasbeen identified recently as a histone deacetylase inhibitor.To evaluate the antitumor efficacy and mechanisms ofaction of valproic acid in medulloblastoma and supra-tentorial primitive neuroectodermal tumor (sPNET), whichare among the most common malignant brain tumors inchildren with poor prognosis, two medulloblastoma(DAOY and D283-MED) and one sPNET (PFSK) cell lineswere treated with valproic acid and evaluated with a panelof in vitro and in vivo assays. Our results showed thatvalproic acid, at clinically safe concentrations (0.6 and1 mmol/L), induced potent growth inhibition, cell cyclearrest, apoptosis, senescence, and differentiation andsuppressed colony-forming efficiency and tumorigenicityin a time- and dose-dependent manner. The medulloblas-toma cell lines were more responsive than the sPNET cellline and can be induced to irreversible suppression ofproliferation and significantly reduced tumorigenicity by0.6 and 1 mmol/L valproic acid. Daily i.p. injection ofvalproic acid (400 mg/kg) for 28 days significantlyinhibited the in vivo growth of DAOY and D283-MEDs.c. xenografts in severe combined immunodeficientmice. With Western hybridization and real-time reversetranscription-PCR, we further showed that the antitumoractivities of valproic acid correlated with induction of

histone (H3 and H4) hyperacetylation, activation of p21,and suppression of TP53, CDK4, and CMYC expression.In conclusion, valproic acid possesses potent in vitro andin vivo antimedulloblastoma activities that correlatedwith induction of histone hyperacetylation and regulationof pathways critical for maintaining growth inhibition andcell cycle arrest. Therefore, valproic acid may represent anovel therapeutic option in medulloblastoma treatment.[Mol Cancer Ther 2005;4(12):1912–22]

IntroductionEmbryonal tumors of childhood, including medullo-blastoma and supratentorial primitive neuroectodermaltumor (sPNET), are among the most common malignantbrain tumors of childhood. Although medulloblastomasand sPNETs are histologically similar, patients with medu-lloblastoma tend to have a better therapeutic response thansPNET when treated with same therapy. Furthermore,infants and young children with these tumors have a pooroverall survival rate. Even among survivors, many willhave long-term neurocognitive and neuroendocrine sequel-ae resulting from craniospinal radiation. Thus, effectivenew therapies and treatment paradigms are needed forthese diseases.Histone deacetylase (HDAC) inhibitors represent a novel

class of therapeutic agents that may provide an alternativeapproach for the treatment of these tumors. Recent studieshave shown that HDACs play an important role in theregulation of gene transcription and oncogenesis throughremodeling of chromatin structure and dynamic changes innucleosomal packaging of DNA (1–3). Inhibition of HDACincreases histone acetylation and maintains chromatinstructure in a more open conformation. This conforma-tional change may lead to restoration of transcriptionallysilenced pathways or suppression of aberrantly expressedgenes through recruitment of repressor proteins (2), result-ing in cell cycle arrest, apoptosis, and cellular differentia-tion in human cancers. Several structurally diverse HDACinhibitors have shown preclinical activities in a varietyof adult and pediatric tumor models (1), some of them,including suberoylanilide hydroxamic acid, depsipeptide,and MS-275, have recently entered clinical trials (4, 5). Formalignant pediatric brain tumors, such as medulloblas-toma and sPNET, however, there is still a lack of HDACinhibitors that are ready for clinical trials.Valproic acid, a well-tolerated anticonvulsant with an

extensively characterized toxicity profile, has been identi-fied recently as a HDAC inhibitor. It inhibits both class Iand II HDACs (excluding HDAC6 and HDAC10) with

Received 6/8/05; revised 9/9/05; accepted 9/30/05.

Grant support: Childhood Brain Tumor Foundation, John S. Dunn ResearchFoundation, Robert J. Kleberg, Jr., and Helen C. Kleberg Foundation,Gillson Longenbaugh Foundation, and Cancer Fighters of Houston.

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 18 U.S.C. Section 1734 solely toindicate this fact.

Requests for reprints: Xiao-Nan Li, Texas Children’s Cancer Center, TexasChildren’s Hospital, Baylor College of Medicine, 6621 Fannin Street,MC 3-3320, Houston, TX 77030. Phone: 832-824-4580;Fax: 832-825-4038. E-mail: [email protected]

Copyright C 2005 American Association for Cancer Research.

doi:10.1158/1535-7163.MCT-05-0184

1912

Mol Cancer Ther 2005;4(12). December 2005

on April 19, 2018. © 2005 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

http://mct.aacrjournals.org/

resultant hyperacetylation of histone H3 and H4 (3, 6–8).Altered expression of multiple genes, including the cyclin-dependent kinase inhibitor p21Cip1, glycogen synthasekinase-3h, and peroxisome proliferator-activated receptors,have been reported in cells exposed to valproic acidtreatment (8–10). Valproic acid has displayed potentin vitro and in vivo antitumor activities against neuroblas-toma (11, 12), glioma (13, 14), leukemia (15, 16), breastcancer (17) and prostate cancer (18), but effect of valproicacid in medulloblastoma and sPNET tumors remainsunknown.Valproic acid possesses several established and yet

special properties that make it an attractive drug fortreating brain tumors in children, especially in patientswith medulloblastomas and sPNETs. Valproic acid canpass the blood-brain barrier and has a long half-life of 9 to20 hours in human being; its concentration in cerebro-spinal fluid is nearly the same as the free valproic acidconcentration in plasma. Therefore, effective drug deliveryto brain tumors is feasible. In a pediatric patient withrelapsed sPNET, valproic acid was reported recently tohave induced histologically confirmed signs of tumor celldifferentiation (19). Furthermore, valproic acid is alreadya commercially available drug with very well definedpharmacokinetic properties; it has greater potential ofbeing quickly translated into clinical trials once its anti-tumor activities are established in preclinical models ofmedulloblastomas and sPNETs.The present study was therefore undertaken to assess the

antitumor activities of valproic acid in medulloblastomaand sPNET by using two medulloblastoma and one sPNETcell lines that are available from American Type CultureCollection (Manassas, VA). In this report, we describe thein vitro effects of valproic acid on cell proliferation, cellcycle regulation, apoptosis, differentiation, cellular senes-cence, colony-forming efficiency (CFE), and tumorigenicityin severe combined immunodeficient (SCID) mice as wellas the in vivo growth inhibition of medulloblastomaxenografts. In addition, we studied the changes in histone(H3 and H4) and TP53 acetylation and the alterations ofp21, TP53, p16, CDK4, and CMYC gene expression duringin vitro valproic acid treatment to investigate the molecularmechanisms of the antitumor effects of valproic acid. Ourfindings formed the basis of a recently approved phase Iclinical study of valproic acid in pediatric patients by theChildren’s Oncology Group.

Materials andMethodsValproic AcidValproic acid (2-propyl-pentanoic acid) was purchased

from Sigma (St. Louis, MO) and dissolved in serum-freemedium to make a stock solution of 360 mmol/L. Thestock solution was further diluted with cell culture mediumto yield final valproic acid concentrations that havebeen established in patients with epilepsy ranging from0.2 mmol/L (the typical cerebrospinal fluid concentration),0.6 mmol/L (the typical therapeutic serum concentration),

1 mmol/L (upper limit of antiepileptic range), 2.7 mmol/L(associated with limited toxicity, such as lethargy withbenign outcome), and 3.6 and 5.1 mmol/L (associated withlife-threatening toxicities, such as coma).

Cell LinesHuman medulloblastoma cell lines (D283-MED and

DAOY) and a sPNET cell line (PFSK) were obtained fromAmerican Type Culture Collection (20–22) and maintainedin DMEM supplemented with 10% fetal bovine serum(Mediatech, Herndon, VA).

Cell Proliferation AssayCells were seeded into 96-well plates at 2,000 to 3,000

live cells per well and treated with valproic acid (0.2–5.1 mmol/L) for up to 45 days. Culture medium wasreplaced every 3 to 4 days at which time the antiprolifer-ative effect of valproic acid was assessed using Cell CountKit-8 (Dojindo Molecular Technologies, Inc., Gaithersburg,MD). Washout experiments to assess the reversibility ofthe valproic acid–associated antiproliferative effect weredone at various time points by replacing drug-containingmedium with drug-free medium.

Cell CycleAnalysis with Flow CytometryCells treated with or without valproic acid (1 and

2.7 mmol/L) were harvested for flow cytometry analysison days 1, 2, 3, and 7 and weekly thereafter until day 42.Cells were fixed and stained with 0.1 mg/mL propidiumiodide for DNA analysis with Becton Dickinson FACScan(Franklin Lakes, NJ) as described previously (23).

Detection of ApoptosisApoptosis was evaluated with flow cytometry and on cell

smears using the terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assay (In situ CellDeath Detection kit, AP; Boehringer Mannheim GmbH,Mannheim, Germany). Samples were incubated with 50 ALof reaction mixture in a humidified chamber at 37jC for90 minutes as described previously (23). The percentageof apoptotic cells was determined by counting at least1,000 cells from 10 to 20 high-power fields (�400) underboth phase-contrast and fluorescent microscopy.

Cell SenescenceAssayHistochemical detection of senescence-associated expres-

sion of h-galactosidase activity (24) was done with aSenescence Detection kit (BioVision, Mountain View, CA)on fixed cells treated with or without valproic acid (0.6 and1 mmol/L). The development of cytoplasmic blue wasdetected and photographed using a Nikon (Nikon Instru-ments, Inc., Lewisville, TX) inverted microscope equippedwith a color CCD camera.

Immunofluorescent StainingProtein expression of glial marker glial fibrillary acidic

protein (GFAP) and neuronal marker synaptophysin wasevaluated during valproic acid treatment. Monoclonalantibodies against human GFAP (DAKO, Glostrup,Denmark) and synaptophysin (Boehringer Mannheim) wereused as primary antibody. FITC-conjugated secondary anti-bodies (Molecular Probes, Eugene, OR) were subsequentlyapplied. The staining intensity was scored as negative (�),marginal (F), low (+), medium (++), or high (+++).

Molecular Cancer Therapeutics 1913




CFE in Soft AgarTumor cells were resuspended in DMEM with 0.3% agar

and plated in 24-well plates at 2,000 per well on top of a0.5 mL precast semisolid 1% agar underlayer followingtreatment with valproic acid (1 or 2.7 mmol/L) for 1 or3 weeks as described previously (23). The CFE was definedas the percentage of plated cells that formed coloniesrelative to an untreated control.

Tumorigenicity and In vivo Treatment of DAOYandD283-MEDXenografts in SCIDMiceAll animal experiments were conducted according to

an Institutional Animal Care and Use Committee–approved protocol. RAG2 SCID mice, ages 8 to 12weeks, were bred and maintained in a specific patho-gen-free animal facility. Heterotransplantation was doneby s.c. injection of 107 live cells as described previously(23). Xenograft growth was measured weekly with a

sliding caliper. Tumor size (M) was calculated using theformula: M = a2b / 2, where a is the minimum widthand b is the maximum length. For tumorigenicityassay, cells were pretreated with valproic acid (0.6 and1 mmol/L) for 4 weeks before s.c. injection, and tumortake and xenograft growth were compared with untreat-ed cells. For efficacy of in vivo valproic acid treatment,tumors were allowed to reach f0.5 cm in diameterbefore the initiation of daily i.p. administration ofvalproic acid (400 mg/kg), which lasted up to 28 days.Tumor size was measured weekly, and at the end oftreatment, all mice were sacrificed and remnant tumorswere examined histologically.

Western HybridizationFor analysis of histone acetylation, histones were

prepared by acid extraction. For analysis of the remain-ing selected genes, protein pellets were collected withTrizol reagent (Invitrogen, Inc., Carlsbad, CA) anddissolved in 8 mol/L urea. Protein or histones (40 Ag)were separated with 4% to 20% SDS-polyacrylamide gels,which were either stained with Coomassie blue ortransferred to polyvinylidene difluoride membranes forblotting with primary antibodies against acetylatedhistone H3 (AcH3) and H4 (AcH4) and acetylated p53(Lys373 and Lys382; Upstate Biotechnology, Inc., Waltham,MA) and p21 (sc-397), TP53 (sc-6243), p16 (sc-9968),CDK4 (sc-260), and CMYC (N-262; Santa Cruz Biotech-nology, Santa Cruz, CA). Bound antibodies were visual-ized with horseradish peroxidase–conjugated secondaryantibody and Chemiluminescence Plus kit (Amersham,Piscataway, NJ). Because recent reports have shown thatlevels of housekeeping proteins can be affected byHDAC inhibitors (25), and glyceraldehyde-3-phosphatedehydrogenase may not be the optimal protein asinternal control for analyzing the effect of HDACinhibitors by Western blot (26, 27), we thereforemeasured the protein concentration in each sample andalso estimated the amount of proteins applied ontoeach lane by using a Coomassie blue–stained duplicateSDS-PAGE gel as protein loading control.

Table 1. Summary of GFAP and synaptophysin expression incells treated with valproic acid

Cells Valproicacid (mmol/L)

GFAP Synaptophysin

2 wk 4 wk 2 wk 4 wk

D283-MED 0 F F � �D283-MED 1 +++ +++ � +D283-MED 2.7 +++ ND � NDDAOY 0 � � � �DAOY 1 +++ +++ + ++DAOY 2.7 +++ ND ++ NDPFSK 0 � � � �PFSK 1 � + � �PFSK 2.7 F +++ � �

Abbreviation: ND, not done due to very few viable cells.

Figure 1. Time- and dose-dependent antiproliferative effects of valproicacid. D283-MED cells were the only ones that responded to 0.2 mmol/Lvalproic acid (A). Minimum required dose for inhibiting cell proliferation inDAOY and PFSK was 0.6 mmol/L, and the exposure time required in PFSKcells is much longer than in DAOY cells (B and C). Dotted lines, growthcurves of washout (W/O ) experiments. Experiments were repeated at leasttwice.

Antimedulloblastoma Effects of Valproic Acid1914




Quantitative Real-time PCRQuantitative real-time PCR was done with SYBR Green

Master Mix and ABI 7000 DNA Detection System (ABI,Columbia, MD) as described previously (23). Five normalbrain tissue cDNAs were employed as references, whichinclude two adult cerebellum and one fetal brain tissuespurchased from Clontech (Paulo Alto, CA) and ILSbio(Bethesda, MD) and two normal cerebellar tissuescollected from patients (ages 8 and 14 years) undergoingresection of benign tumors at Texas Children’s Hospitalin accordance to institutional review board–approvedprotocols. Gene-specific primers were designed to flankmore than one exon to ensure that all the expectedPCR products were generated from mRNA (Table 1,Supplementary Material).3 Gene expression levels weredetermined with standard DDCt method (23) andnormalized to the internal standard glyceraldehyde-3-phosphate dehydrogenase. All reactions were done induplicate on two occasions. Reaction specificity wasconfirmed with dissociation curves.

Statistical AnalysisThe effects of valproic acid on cell proliferation, CFE, cell

cycle arrest, apoptosis, and xenograft growth in SCID micewere analyzed with two-way ANOVA and presented asthe mean F SD.

ResultsValproic Acid Suppresses Cell ProliferationValproic acid suppressed tumor cell proliferation in a

time- and dose-dependent manner (P < 0.05) in all threecentral nervous systemembryonal tumor cell lines (Fig. 1). InD283 cells, treatment with 0.6 mmol/L valproic acid for10 days or with 0.2 mmol/L valproic acid for 14 daysresulted in >50% suppression of cell proliferation (Fig. 1A),whereas in DAOY cells it required 10 days of exposure to1 mmol/L valproic acid or 21 days to 0.6 mmol/L valproicacid (Fig. 1B). PFSK cells were the least responsive. Cellgrowth was initially suppressed on day 10 after exposure to1 mmol/L valproic acid; however, the cell number contin-ued to increase, albeit at a slower rate than untreated cells.PFSK cell growth was not completely suppressed following4 weeks of exposure to valproic acid concentrations as highas 3.6 mmol/L (Fig. 1C). It should be pointed out that thegrowth curves of the untreated cells in all three cell linesreached plateau in f21 days mainly due to the limited

3 Supplementary material for this article is available at Molecular CancerTherapeutics Online (http://mct.aacrjournals.org/).

Figure 2. Valproic acid – in-duced cell cycle arrest and apo-ptosis. Representative graphs ofcell cycle distribution analyzedwith flow cytometry in cells trea-ted with valproic acid (VPA ; 1 and2.7 mmol/L; A). Apoptosis wasexamined by analyzing subdiploidpopulation with flow cytometry(A) and comparing the positivelylabeled cells with terminal deoxy-nucleotidyl transferase–mediateddUTP nick end labeling assay, inwhich at least 1,000 cells fromrandomly selected high-powerfields were counted (B).





growth areas that were available in 96-well plates that wereused in the cell proliferation assay. This phenomenon,however, was not observed in any of the treated cells,suggesting that the differences of cell numbers betweentreated and control groups could have beenmore significanthad the untreated cells been given additional space.To evaluate whether cell lines regained proliferative

capacity following cessation of valproic acid exposure,washout experiments in which valproic acid–containingmedium was replaced with valproic acid–free mediumrevealed that following continuous exposure to 0.6 mmol/Lvalproic acid for at least 4 weeks suppression of cellproliferation became irreversible in the medulloblastomacell lines (Fig. 1A and B).

Valproic Acid Induces Cell CycleArrestTo investigate the effects of valproic acid on cell cycle

distribution, cells were treated with concentrations that areeither clinically safe (1 mmol/L) or with mild toxicities(2.7 mmol/L). In D283-MED and DAOY cells, shift of cellpopulation to G1-G0 phases started after 3 days of valproicacid (1 or 2.7 mmol/L) treatment. More significant cellcycle arrest, however, was detected on day 7, when cells inG1-G0 phases increased and cells in G2-M phases decreasedconcurrently. Higher concentration (2.7 mmol/L in D283-MED treated for 1 week) or longer exposure (up to 5 weeksin DAOY and D283-MED cells) also resulted in aremarkable increase of subdiploid apoptotic cells. In PFSKcells, however, valproic acid did not produce significantcell cycle arrest (Fig. 2A).

Valproic Acid Augments ApoptosisIn addition to flow cytometry analysis, terminal deoxy-

nucleotidyl transferase–mediated dUTP nick end labelingassay was also used to examine the induced apoptosis incells treated with valproic acid (1 and 2.7 mmol/L; Fig. 2B).The increased apoptosis was time and dose dependent.In D283-MED cells, following an initial induction ofapoptosis in the first 24 hours, apoptotic cells increasedf20-fold (10%) over untreated cells (0.5%) on day 7. Due toexcessive accumulation of dying D283-MED cells and debrisas a result of valproic acid treatment, exact quantification ofapoptotic assay beyond 7 days was unreliable. In DAOYcells, longer exposure time (up to 5 weeks) was required toelicit significant increase of apoptosis. In PFSK cells, onlyminimal increase (2- to 3-fold) in apoptosis was detectedafter 2 to 6 weeks of treatment.

Valproic Acid Induces Cellular SenescenceCellular senescence has been identified as one of the

mechanisms mediating the anticancer effects of chemo-therapies (24). One of the morphologic changes that wasfrequently observed in our valproic acid–treated cells isthe flattening of cells with increased granularity, whichis a typical morphologic change associated with cellularsenescence (24). By examining senescence-associated ex-pression of h-galactosidase activity (24), we confirmedthat cellular senescence was indeed induced in thoseflattened D283-MED and DAOY cells by valproic acid (0.6and 1 mmol/L) in a time- and dose-dependent manner(Fig. 3). More interestingly, in D283-MED cells treated

Figure 3. Cellular se-nescence induced by val-proic acid (0.6 and 1.0mmol/L). Arrowhead,changes found in multicellspheroids in D283-MEDcells; arrows, senescentcells with the typical flat-tened appearance. Magni-fication, �100.





with valproic acid (0.6 mmol/L), we observed bluestaining in the gradually dissociating cell spheroids aswell. The induced senescence started from day 3 andpeaked on day 7 when the whole spheroids were denselystained. A higher valproic acid concentration (1 mmol/L)led to more dramatic increase of cellular senescence asevidenced by further depletion of spheroids and increasedh-galactosidase staining of attached D283-MED cells.The PFSK cells showed only minimal increase of positiveh-galactosidase staining after valproic acid treatment.

Valproic Acid Induces Glial and Neuronal MarkerExpression inMedulloblastoma CellsTo evaluate the differentiation-inducing ability of

valproic acid, the upper limit of clinically achievableconcentration (1 mmol/L) and concentration with lowtoxicity (2.7 mmol/L) were included. For glial markerGFAP, treatment with 1 mmol/L valproic acid for 2weeks dramatically increased its expression from nega-tive (�) to high level (+++) in DAOY and D283-MEDcells (Table 1, Supplementary Fig. S1A-D).3 Such eleva-tion of GFAP expression in PFSK cells, however, was notobserved until after 4 weeks of treatment at 2.7 mmol/L(Supplementary Fig. S1E and F).3 For neuronal markersynaptophysin, elevated expression was observed inDAOY cells after 2 weeks of exposure to 1 mmol/Lvalproic acid (Supplementary Fig. S1G and H.),3 whereaslonger exposure for up to 4 weeks was required with

D283-MED cells (Table 1), suggesting that the predomi-nantly apoptotic and growth arrest effects of valproic acidon D283-MED cells delayed the few surviving cells frombeing induced to differentiate. Synaptophysin protein wasnot induced in PFSK cells even after 4 weeks of valproicacid (2.7 mmol/L) treatment (Table 1). Altogether, theseresults showed that the medulloblastoma cell lines (DAOYand D283-MED) are more responsive to valproic acid thanthe sPNET cell line PFSK and can be induced to moredifferentiated phenotype with concentration (1 mmol/L)that is clinically achievable.

Valproic Acid Suppresses CFETo examine the suppressive effects of valproic acid on

CFE, D283-MED and PFSK cells were treated with valproicacid (1 and 2.7 mmol/L) for 1 or 3 weeks. DAOY cells do notform colonies in soft agar and were not tested in this assay.Our results showed that valproic acid exerted time- anddose-dependant suppression of CFE in both D283-MED andPFSK cells (P < 0.01; Fig. 4). Treatment with 1 mmol/Lvalproic acid for 1 week resulted in 75% inhibition of colonyformation in D283-MED cells and >85% inhibition in PFSKcells. Higher concentration (2.7 mmol/L) and/or longertreatment time (3 weeks) produced more significantsuppressive effects in both cell lines (Fig. 4).

Valproic Acid Inhibits Tumorigenicity of Medullo-blastoma CellsTo examine the inhibitory effects of valproic acid on

tumorigenicity, cells were pretreated with valproic acid(0.6 and 1 mmol/L) for 4 weeks before heterotransplanteds.c. into SCID mice. As shown in Fig. 5A, valproic acidexerted dose-dependent suppression of tumorigenicityin the two medulloblastoma cell lines (D283-MED andDAOY). In D283-MED cells, pretreatment with 0.6 mmol/Lvalproic acid significantly reduced the tumor growth, andwith 1 mmol/L valproic acid, the tumor formation wascompletely suppressed (P < 0.01). In DAOY cells, bothtumor take and growth were inhibited (P < 0.01). In thesPNET cell line PFSK, however, no suppression oftumorigenicity was observed (data not shown).

Valproic Acid Suppresses the Growth of D283-MEDand DAOYS.c. Xenografts in SCIDMiceTo further assess the in vivo antimedulloblastoma effects

of valproic acid in s.c. xenografts, valproic acid was giventhrough daily i.p injection of 400 mg/kg, which is the mostcommonly used route and dose in mouse (28), forup to 28 days. Such treatment resulted in significantsuppression of xenograft growth from both cell lines(P < 0.01; Fig. 5B). Although no significant histologicchanges were observed, intense GFAP positivity wasdetected in themajority of DAOY cells in the treated tumors.The expression of synaptophysin, which was undetectablein the untreated tumors,was increased in a fraction of cells inxenografts treated with valproic acid (data not shown).

Valproic Acid Induces Hyperacetylation of HistoneH3 and H4 Both In vitro and in Animal ModelsTo evaluate the temporal changes of histone H3 and

H4 acetylation status, cells treated with valproic acid(1 mmol/L) for 0, 3, 7, 14, and 28 days in vitro were analyzed

Figure 4. Time- and dose-dependent suppression of CFE of D283-MED (A) and PFSK (B) following valproic acid treatment (P < 0.01).DAOY cells do not form colonies in soft agar and were not tested in thisassay.





by Western hybridization. As shown in Fig. 6A, increasedAcH3 and AcH4 levels were detected as early as day 3 in allthree cell lines. With extended treatment, the medulloblas-toma cell lines (D283-MED and DAOY) displayed progres-sive accumulation of AcH3 and AcH4 until day 28. In thesPNET cell line, however, no additional increases in AcH3or AcH4 levels were observed beyond day 7 of valproicacid treatment. To further examine the in vivo effects ofvalproic acid, remnant s.c. xenografts of D283-MED andDAOY cells were analyzed. Treatment with valproic acid(400 mg/kg/d i.p.) significantly increased AcH3 and AcH4levels in both D283-MED and DAOY xenografts (Fig. 6B).

Valproic Acid Activates p21Cip1 (p21) and InhibitsTP53 ExpressionTo perform detailed temporal analysis of gene expres-

sion alteration, both mRNA and protein levels before andafter valproic acid (1 mmol/L) treatment were examined.Using cDNAs from normal human brains, including twofrom age-matched cerebella, as references, we found thatthe intrinsic expression levels of p21 and TP53 weresignificantly different among the three cell lines andwere associated with some differences of their responsemodes toward valproic acid treatment. In agreement withpreviously published results (29), our data showed thatvalproic acid activated p21 gene by increasing its mRNAtranscription and/or protein translation (30– 33). InD283-MED cells, which showed near-normal level ofpretreatment p21 mRNA, the increased p21 protein levelwas observed on day 3 before the induction of a marginalincrease (<50%) in p21 mRNA transcripts on day 7. InDAOY and PFSK cells, both of which showed lower thannormal levels of pretreatment p21 mRNA, their increasesin p21 protein translation paralleled the increases inmRNA transcription (Fig. 7).

The tumor suppressor TP53 induces cell cycle arrest orapoptosis in response to a variety of stress signals (34).Because p21 is believed to be the main target for TP53-induced cell cycle arrest (35), we examined if valproic acidalso induces TP53 expression. As shown in Fig. 7, theexpression of TP53 mRNA and protein were induced onlyin DAOY cells, which is known to harbor a mutated TP53gene (36), whereas in D283-MED cells the expression ofTP53 was steadily down-regulated by valproic acid(1 mmol/L). In PFSK, the TP53 protein expression wasnot affected by 1 mmol/L valproic acid, although its near-normal expression of TP53 mRNA was slightly (<50%)decreased. Altogether, our data showed that effects ofvalproic acid on apoptosis, cell cycle arrest, and activationof p21 gene in medulloblastomas does not correlate withthe increased expression of tumor suppressor gene TP53.Post-translational modifications of TP53, such as ubiq-

uitination, phosphorylation, and acetylation, have profoundeffects on TP53 function (37). Acetylation of TP53, in

Figure 6. Induction of hyperacetylated histone H3 (AcH3 ) and H4(AcH4) during the treatment with valproic acid. For in vitro effects (A),temporal changes were investigated in the three cell lines during thetreatment with 1 mmol/L valproic acid for 28 d. For in vivo effects (B), theremnant tumors treated with (+) or without (�) valproic acid wereanalyzed with Western hybridization. Coomassie blue–stained histoneson polyacrylamide gel were used as loading control.

Figure 5. Effects of valproic acid ontumorigenicity and in vivo growth ofxenografts in SCID mice. To examinethe suppressive activities of valproicacid on tumorigenicity, tumor cellswere pretreated with valproic acid(0.6 and 1.0 mmol/L) for 4 wks beforeheterotransplanted s.c. into SCIDmice; both tumor take and growthrate were presented. For D283-MEDcells, two mice bearing tumors thatwere too big in the untreated groupwere euthanized on week 6 (arrow ;A). To examine the in vivo growthinhibitory effects of valproic acid,cells from the medulloblastoma celllines (D283-MED and DAOY) wereheterotransplanted s.c. into SCIDmice and allowed to grow for4 wks to reach approximate size of0.5 � 0.5 cm2 when daily i.p. injec-tions of valproic acid (400 mg/kg)were started and lasted 21 d forD283-MED and 28 d for DAOY xen-ografts (P < 0.01; B).





particular, can dramatically stimulate its sequence-specificDNA-binding activity in vitro , and treatment with trichos-tatin A, a HDAC inhibitor, has been shown to increase thelevels of acetylated TP53. To further investigate the effect ofvalproic acid on TP53 acetylation, Western hybridizationwas done using a monoclonal antibody against specificacetylated lysine residue (Lys373 and Lys382) in TP53 proteinin the three cell lines treated with 1 mmol/L valproic acidfor up to 28 days. In all the cell lines examined, valproic acidfailed to induce TP53 acetylation. Instead, it reduced thelevels of acetylated TP53 in both DAOY (with mutatedTP53) and PFSK (with wild-type TP53) cells starting fromday 7 of treatment while maintaining undetectable levels ofacetylated TP53 in D283-MED cells until the end of valproicacid treatment on day 28 (Fig. 7B), indicating that valproicacid did not activate p21 via increasing TP53 acetylation.

Effects of Valproic Acid on p16INK4a and CDK4ExpressionThe tumor suppressor p16INK4a arrests cells at G1

phase and mediates cellular senescence by inhibitingkinase activities of CDK4 and phosphorylation of RBtumor suppressor protein (24). Our results showed that,in the medulloblastoma cell lines, p16 mRNA levels weremuch lower compared with normal references and itsprotein expression was completely absent. Valproic acidtreatment had no effects on p16 expression in D283-MEDcells. Although p16 mRNA expression in DAOY wasinduced 6-fold by valproic acid on day 28, the mRNAtranscript level was still <30% of the levels in age-matched cerebellum, and no p16 protein was detected.

Contrary to medulloblastoma cells, PFSK expressednormal level of p16 mRNA, which was increased 2-foldwith valproic acid treatment, but p16 protein levelsremained unchanged. These results suggest that p16 didnot play a major role in valproic acid induced cellularsenescence.The expression of CDK4, however, was significantly

altered by valproic acid. Suppression of CDK4 mRNAexpression was most prominent in D283-MED cells(Fig. 7A), and a corresponding decline in protein levels,although less dramatic, was also observed (Fig. 7B). InDAOY cells, although the inhibition of mRNA transcriptlevels was not major, a dramatic depletion of CDK4protein was observed (Fig. 7B), suggesting that valproicacid treatment may also affect CDK4 protein levels in atranslational or post-translational manner. Taken together,our data suggest that altered CDK4 transcription and/ortranslation may mediate valproic acid–induced senes-cence in the two medulloblastoma cell lines. Because PFSKcells exhibited minimal growth inhibition and cellularsenescence despite significant decline in CDK4 mRNAand protein levels, we hypothesize that this sPNET cellline may possess redundant pathways and/or compensa-tory mechanisms to escape the antitumor effects ofvalproic acid.

Valproic Acid Down-Regulates the Expression ofOncogene CMYCA series of studies have documented that CMYC

regulates a wide range of genes involved in processes,such as proliferation, differentiation, and apoptosis,

Figure 7. Changes of gene expression induced by 1 mmol/L valproic acid in vitro . Quantitative real-time PCR was used to examine the temporal changesof mRNA transcripts during valproic acid treatment (up to 28 d) using normal cDNAs from two adult cerebellum (Adult ), two age-matched pediatriccerebella (Child ), and one fetal brain (Fetal ) as references (A). The concomitant changes of protein expression were analyzed with Western hybridization(B). Coomassie blue–stained proteins on polyacrylamide gel were used as loading control. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.





including p21 and CDK4 (38, 39). In our study, all three celllines expressed high levels (>6-fold of normal) of CMYCmRNA. Treatment with valproic acid (1 mmol/L) dramat-ically reduced CMYC mRNA transcription levels inD283-MED (>50% on day 28), DAOY (down to normallevels after day 14), and PFSK (45% on day 14) cells.Significant reductions in CMYC protein levels were alsoobserved, especially in D283-MED and DAOY cell lines,suggesting that suppression of CMYC correlated withvalproic acid responsiveness of medulloblastoma cells.

DiscussionIn the current study, we showed that the potentantimedulloblastoma activities of valproic acid in medul-loblastomas are correlated with induction of histone (H3and H4) hyperacetylation, activation of p21, restoration ofp16/CDK4 pathway, and suppression of CMYC oncogene.With absent cell cycle arrest, minimal apoptosis, andcellular senescence, the sPNET cell line PFSK seems lesssensitive to valproic acid than medulloblastoma cell lines.However, because of significant differences in geneexpression patterns among the three cell lines before theywere exposed to valproic acid treatment, some of thedifferential responses observed in these cell lines may berelated to their genetic backgrounds, including the genesand pathways that were not evaluated in the currentstudy.Our study showed that the in vitro antimedulloblastoma

effects of valproic acid were time and concentrationdependent, and irreversible inhibition of cell growth couldbe achieved with extended treatment. This finding is inagreement with our previous results with another HDACinhibitor phenylbutyrate (23). More importantly, we furthershowed that valproic acid possesses strong inhibitoryactivities on tumorigenicity of medulloblastoma cells.Pretreatment with 1 mmol/L valproic acid for 4 weeksbefore heterotransplantation into SCID mice resulted incomplete abrogation of tumorigenicity in D283-MED cellsand significantly reduced tumor take (down to 50%) andgrowth rate in DAOY cells. Even pretreatment with 0.6mmol/L valproic acid was able to significantly decrease thegrowth of xenografts from both medulloblastoma cell lines.These results provided strong evidence to support thenotion that irreversible epigenetic reprogramming hastaken place and are responsible for the reduced tumorige-nicity. Because long-term valproic acid administration inchildren is well tolerated (40, 41), these results suggest thatchronic treatment with valproic acid should be maintainedin children with medulloblastomas after radiation andchemotherapy, which may possibly decrease recurrenceand improve survival.The antimedulloblastoma effect of valproic acid was

confirmed in vivo using s.c. heterotransplanted D283-MED and DAOY xenografts in SCID mice. Because thehalf-life of valproic acid in mice is only 0.8 hourcompared with 9 to 18 hours in humans (42), weexpected that with once daily injection of 400 mg/kg

the xenografts were exposed to therapeutic valproic acidconcentrations for only 4 hours daily. Nonetheless, weobserved significant growth inhibition of the treatedxenografts. It is therefore reasonable to infer that moreprominent tumor suppression would have been observedhad steady-state therapeutic valproic acid concentrationsbeen maintained.In agreement with previous reports (7, 8), our data

showed valproic acid– induced histone (H3 and H4)hyperacetylation both in vitro and in vivo . We also foundthat the levels of accumulated AcH3 and AcH4 correlatedwith the degree of in vitro growth suppression in thevalproic acid–sensitive medulloblastoma cell lines, sug-gesting that the antimedulloblastoma effects of valproicacid were at least partly mediated through histone H3 andH4 hyperacetylation. Recent cDNA microarray profiles ofhuman medulloblastomas documented their overexpres-sions of HDAC1 and HDAC2 (43), lending further supportfor using HDAC inhibitors as novel agents for treatingthese tumors.The expression of cyclin-dependent kinase inhibitor p21

has been implicated in HDAC inhibitor–induced cell cyclearrest in numerous human cancers (29). In this study, weconfirmed that valproic acid is capable of activating p21gene in medulloblastoma cells. Our results also showedthat the p21 activation does not correlate with increasedp53 expression or with increased acetylation of TP53. Infact, the mRNA expression of TP53 gene was inhibited inD283 and PFSK cell lines, both of which have functionalwild-type TP53 gene. The p16/CDK4/RB pathway, nowbelieved to be the molecular link between cellular senes-cence and tumor suppression, also seemed to mediateHDAC inhibitor– induced senescence in human cells(44–46). Our results showed that, in medulloblastoma celllines lacking intrinsic p16INK4a expression, suppression ofCDK4 expression seemed to have compensated for losses ofp16 and restored significant cellular senescence in both celllines. In PFSK cells, although valproic acid inducedsignificant suppression of CDK4 mRNA and protein levels,induced cellular senescence was not observed, suggestingthat there may be redundant pathways or compensatorymechanisms allowing these cells to be resistant to the anti-tumor effects of valproic acid.Overexpression of CMYC had been frequently detected

in medulloblastomas and is associated with shortersurvival and tumor anaplasia (47–49). CMYC has alsobeen reported to promote cell cycle reentry and prolifera-tion (39) through repression of p21 expression and activa-tion of CDK4 mRNA transcription (50). Therefore, thesuppression of CMYC expression by valproic acid mayrender substantial therapeutic benefits in medulloblastomapatients by inhibiting the driving activities of CMYC in cellproliferation and cell cycle progression.In summary, we showed that valproic acid possesses

potent in vitro and in vivo antimedulloblastoma activitiesby suppressing cell proliferation, promoting apoptosis,inducing cell cycle arrest and cellular senescence, enhanc-ing cell differentiation, and inhibiting tumorigenicity at





concentrations within the established therapeutic ranges ofvalproic acid for epilepsy. These results may lay thegroundwork for further studies using specific geneticallyengineered models to establish the causal relationshipbetween valproic acid antitumor activity and specificgenetic pathways and to identify molecular markers thatwill predict drug responsiveness and guide the develop-ment of future clinical therapies.

References

1. Marks PA, Miller T, Richon VM. Histone deacetylases. Curr OpinPharmacol 2003;3:344–51.

2. Richon VM, O’Brien JP. Histone deacetylase inhibitors: a new class ofpotential therapeutic agents for cancer treatment. Clin Cancer Res 2002;8:662–4.

3. Gottlicher M, Minucci S, Zhu P, et al. Valproic acid defines a novel classof HDAC inhibitors inducing differentiation of transformed cells. EMBO J2001;20:6969–78.

4. Kelly WK, Richon VM, O’Connor O, et al. Phase I clinical trial of histonedeacetylase inhibitor: suberoylanilide hydroxamic acid administeredintravenously. Clin Cancer Res 2003;9:3578–88.

5. Piekarz R, Bates S. A review of depsipeptide and other histonedeacetylase inhibitors in clinical trials. Curr Pharm Des 2004;10:2289–98.

6. Kramer OH, Zhu P, Ostendorff HP, et al. The histone deacetylaseinhibitor valproic acid selectively induces proteasomal degradation ofHDAC2. EMBO J 2003;22:3411–20.

7. Blaheta RA, Nau H, Michaelis M, Cinatl J, Jr. Valproate and valproate-analogues: potent tools to fight against cancer. Curr Med Chem 2002;9:1417–33.

8. Phiel CJ, Zhang F, Huang EY, et al. Histone deacetylase is a directtarget of valproic acid, a potent anticonvulsant, mood stabilizer, andteratogen. J Biol Chem 2001;276:36734–41.

9. Chen G, Huang LD, Jiang YM, Manji HK. The mood-stabilizing agentvalproate inhibits the activity of glycogen synthase kinase-3. J Neurochem1999;72:1327–30.

10. Werling U, Siehler S, Litfin M, Nau H, Gottlicher M. Induction ofdifferentiation in F9 cells and activation of peroxisome proliferator-activated receptor y by valproic acid and its teratogenic derivatives. MolPharmacol 2001;59:1269–76.

11. Cinatl J, Jr., Kotchetkov R, Blaheta R, et al. Induction of differenti-ation and suppression of malignant phenotype of human neuroblastomaBE(2)-C cells by valproic acid: enhancement by combination withinterferon-a. Int J Oncol 2002;20:97–106.

12. Yuan PX, Huang LD, Jiang YM, et al. The mood stabilizer valproic acidactivates mitogen-activated protein kinases and promotes neurite growth.J Biol Chem 2001;276:31674–83.

13. Bacon CL, Gallagher HC, Haughey JC, Regan CM. Antiproliferativeaction of valproate is associated with aberrant expression and nucleartranslocation of cyclin D3 during the C6 glioma G1 phase. J Neurochem2002;83:12–9.

14. Kamitani H, Taniura S, Watanabe K, et al. Histone acetylation maysuppress human glioma cell proliferation when p21 WAF/Cip1 and gelsolinare induced. Neuro-oncol 2002;4:95–101.

15. Kawagoe R, Kawagoe H, Sano K. Valproic acid induces apoptosis inhuman leukemia cells by stimulating both caspase-dependent and-independent apoptotic signaling pathways. Leuk Res 2002;26:495–502.

16. Fischkoff SA, Walter E, Jr. Induction of neutrophilic differentiation ofhuman promyelocytic leukemic cells by branched-chain carboxylic acidanticonvulsant drugs. J Biol Response Modif 1984;3:132–7.

17. Olsen CM, Meussen-Elholm ET, Roste LS, Tauboll E. Antiepilepticdrugs inhibit cell growth in the human breast cancer cell line MCF7. MolCell Endocrinol 2004;213:173–9.

18. Thelen P, Schweyer S, Hemmerlein B, et al. Expressional changesafter histone deacetylase inhibition by valproic acid in LNCaP humanprostate cancer cells. Int J Oncol 2004;24:25–31.

19. Driever PH, Wagner S, Hofstadter F, Wolff JE. Valproic acid inducesdifferentiation of a supratentorial primitive neuroectodermal tumor. PediatrHematol Oncol 2004;21:743–51.

20. He XM, Ostrowski LE, von Wronski MA, et al. Expression of O6-methylguanine-DNA methyltransferase in six human medulloblastoma celllines. Cancer Res 1992;52:1144–8.

21. Friedman HS, Burger PC, Bigner SH, et al. Establishment andcharacterization of the human medulloblastoma cell line and transplantablexenograft D283 Med. J Neuropathol Exp Neurol 1985;44:592–605.

22. Fults D, Pedone CA, Morse HG, Rose JW, McKay RD. Establish-ment and characterization of a human primitive neuroectodermal tumorcell line from the cerebral hemisphere. J Neuropathol Exp Neurol 1992;51:272–80.

23. Li XN, Parikh S, Shu Q, et al. Phenylbutyrate and phenylacetateinduce differentiation and inhibit proliferation of human medulloblastomacells. Clin Cancer Res 2004;10:1150–9.

24. Schmitt CA. Senescence, apoptosis and therapy—cutting the lifelinesof cancer. Nat Rev Cancer 2003;3:286–95.

25. Finzer P, Ventz R, Kuntzen C, et al. Growth arrest of HPV-positivecells after histone deacetylase inhibition is independent of E6/E7 oncogeneexpression. Virology 2002;304:265–73.

26. Sakajiri S, Kumagai T, Kawamata N, et al. Histone deacetylaseinhibitors profoundly decrease proliferation of human lymphoid cancer celllines. Exp Hematol 2005;33:53–61.

27. Rombouts K, Knittel T, Machesky L, et al. Actin filament formation,reorganization and migration are impaired in hepatic stellate cells underinfluence of trichostatin A, a histone deacetylase inhibitor. J Hepatol2002;37:788–96.

28. Loscher W. Valproate: a reappraisal of its pharmacodynamicproperties and mechanisms of action. Prog Neurobiol 1999;58:31–59.

29. Archer SY, Meng S, Shei A, Hodin RA. p21(WAF1) is required forbutyrate-mediated growth inhibition of human colon cancer cells. Proc NatlAcad Sci U S A 1998;95:6791–6.

30. Blagosklonny MV, Robey R, Sackett DL, et al. Histone deacetylaseinhibitors all induce p21 but differentially cause tubulin acetylation, mitoticarrest, and cytotoxicity. Mol Cancer Ther 2002;1:937–41.

31. Gui CY, Ngo L, Xu WS, Richon VM, Marks PA. Histone deacetylase(HDAC) inhibitor activation of p21WAF1 involves changes in promoter-associated proteins, including HDAC1. Proc Natl Acad Sci U S A 2004;101:1241–6.

32. Lavelle D, Chen YH, Hankewych M, DeSimone J. Histone deacetylaseinhibitors increase p21(WAF1) and induce apoptosis of human myelomacell lines independent of decreased IL-6 receptor expression. Am JHematol 2001;68:170–8.

33. Richon VM, Sandhoff TW, Rifkind RA, Marks PA. Histone deacetylaseinhibitor selectively induces p21WAF1 expression and gene-associatedhistone acetylation. Proc Natl Acad Sci U S A 2000;97:10014–9.

34. Oren M. Decision making by p53: life, death and cancer. Cell DeathDiffer 2003;10:431–42.

35. Lagger G, Doetzlhofer A, Schuettengruber B, et al. The tumorsuppressor p53 and histone deacetylase 1 are antagonistic regulators ofthe cyclin-dependent kinase inhibitor p21/WAF1/CIP1 gene. Mol Cell Biol2003;23:2669–79.

36. Lee SH, Kang HS, Rhee CH, et al. Growth-inhibitory effect ofadenovirus-mediated p53 gene transfer on medulloblastoma cell line,Daoy, harboring mutant p53. Childs Nerv Syst 2001;17:134–8.

37. Brooks CL, GuW. Ubiquitination, phosphorylation and acetylation: themolecular basis for p53 regulation. Curr Opin Cell Biol 2003;15:164–71.

38. Patel JH, Loboda AP, Showe MK, Showe LC, McMahon SB. Analysisof genomic targets reveals complex functions of MYC. Nat Rev Cancer2004;4:562–8.

39. Hermeking H, Rago C, Schuhmacher M, et al. Identification of CDK4as a target of c-MYC. Proc Natl Acad Sci U S A 2000;97:2229–34.

40. Eiris JM, Lojo S, Del Rio MC, et al. Effects of long-term treatment withantiepileptic drugs on serum lipid levels in children with epilepsy.Neurology 1995;45:1155–7.

41. Guo CY, Ronen GM, Atkinson SA. Long-term valproate andlamotrigine treatment may be a marker for reduced growth and bonemass in children with epilepsy. Epilepsia 2001;42:1141–7.

42. Johannessen CU, Petersen D, Fonnum F, Hassel B. The acute effectof valproate on cerebral energy metabolism in mice. Epilepsy Res 2001;47:247–56.

43. Pomeroy SL, Tamayo P, Gaasenbeek M, et al. Prediction of central





nervous system embryonal tumour outcome based on gene expression.Nature 2002;415:436–42.

44. Ohtani N, Yamakoshi K, Takahashi A, Hara E. The p16INK4a-RBpathway: molecular link between cellular senescence and tumor suppres-sion. J Med Invest 2004;51:146–53.

45. Munro J, Barr NI, Ireland H, Morrison V, Parkinson EK. Histonedeacetylase inhibitors induce a senescence-like state in human cells by ap16-dependent mechanism that is independent of a mitotic clock. Exp CellRes 2004;295:525–38.

46. Oruetxebarria I, Venturini F, Kekarainen T, et al. P16INK4a is requiredfor hSNF5 chromatin remodeler-induced cellular senescence in malignantrhabdoid tumor cells. J Biol Chem 2004;279:3807–16.

47. Bigner SH, Friedman HS, Vogelstein B, Oakes WJ, Bigner DD.Amplification of the c-myc gene in human medulloblastoma cell linesand xenografts. Cancer Res 1990;50:2347–50.

48. Bruggers CS, Tai KF, Murdock T, et al. Expression of the c-Mycprotein in childhood medulloblastoma. J Pediatr Hematol Oncol1998;20:18–25.

49. Eberhart CG, Kratz J, Wang Y, et al. Histopathological and molecularprognostic markers in medulloblastoma: c-myc , N-myc , TrkC, andanaplasia. J Neuropathol Exp Neurol 2004;63:441–9.

50. Gartel AL, Ye X, Goufman E, et al. Myc represses the p21(WAF1/CIP1) promoter and interacts with Sp1/Sp3. Proc Natl Acad Sci U S A2001;98:4510–5.





2005;4:1912-1922. Mol Cancer Ther Xiao-Nan Li, Qin Shu, Jack Men-Feng Su, et al. CDK4, and CMYChyperacetylation and regulating expression of p21Cip1,senescence in medulloblastomas by increasing histone Valproic acid induces growth arrest, apoptosis, and

Updated version

http://mct.aacrjournals.org/content/4/12/1912

Access the most recent version of this article at:

Cited articles

http://mct.aacrjournals.org/content/4/12/1912.full#ref-list-1

This article cites 50 articles, 19 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/4/12/1912.full#related-urls

This article has been cited by 11 HighWire-hosted articles. Access the articles at:

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

Subscriptions

Reprints and

[email protected] at

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

Permissions

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

.http://mct.aacrjournals.org/content/4/12/1912To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/content/4/12/1912.full#ref-list-1

http://mct.aacrjournals.org/content/4/12/1912.full#related-urls

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

mailto:[email protected]



valproic acid induces growth arrest, apoptosis, and...

Documents