TM

LG

*GN

BgtisBeiettsRtipisiacSgsrmtSsiiBeS

Asvlsm

BA

SIC–

ALIM

ENTA

RY

TRA

CT

GASTROENTEROLOGY 2007;133:1272–1281

he Effect of Statins in Colorectal Cancer Is Mediated Through the Boneorphogenetic Protein Pathway

IUDMILA L. KODACH,* SYLVIA A. BLEUMING,* MAIKEL P. PEPPELENBOSCH,‡ DANIEL W. HOMMES,§

IJS R. VAN DEN BRINK,*,§ and JAMES C. H. HARDWICK*,§

Center for Experimental and Molecular Medicine, Academic Medical Center, Amsterdam, The Netherlands; ‡Department of Cell Biology, University Medical Centerroningen, University of Groningen, The Netherlands; and §Department of Gastroenterology & Hepatology, Leiden University Medical Center, Leiden, The

etherlands

ba

tdSCaaac(pdmRccmscsnp

achp

satwas

Ct

ackground & Aims: Epidemiological evidence sug-ests that statins prevent colorectal cancer (CRC), buthe biological mechanism remains obscure. Statinsnduce bone morphogenetic protein (BMP) expres-ion in bone cells. We have previously shown thatMPs act as tumor suppressors in CRC. We hypoth-sized that the action of statins in CRC involves thenduction of BMPs. Methods: We investigated theffects of statins on CRC cell lines using immunoblot-ing, measurements of apoptosis and cell prolifera-ion, and luciferase reporter assays. The effect oftatins was confirmed in a xenograft mouse model.esults: CRC cell lines show widely differing sensi-

ivities to statin treatment. Sensitive cell lines shownduction of BMP2 protein levels and a BMP2 re-orter construct, activation of the BMP pathway, and

nduction of the BMP target gene ID-2, whereas re-istant cell lines do not. The addition of the specificnhibitor of BMPs, noggin, completely prevents lov-statin-induced apoptosis in sensitive cells. Sensitiveell lines express the central BMP pathway elementMAD4, whereas the resistant cell lines do not. Tar-eted knockout of SMAD4 leads to the loss of statinensitivity and reconstitution with SMAD4, to theestoration of statin sensitivity. In a xenograft

ouse model, tumors from sensitive and insensi-ive cell lines were treated with oral simvastatin.ignificant inhibition of tumor growth using sen-itive cells but increased tumor growth when usingnsensitive cells was observed. Conclusions: Statinsnduce apoptosis in CRC cells through induction ofMP2. Statin therapy may only be effective in SMAD4-xpressing CRCs and may have adverse effects inMAD4-negative tumors.

recent trial specifically designed to investigate theincidence of colorectal cancer (CRC) in statin users

howed a highly significant 47% reduction.1 Several in-estigators are now suggesting that the time is ripe forarge-scale, prospective, clinical, or observational trials oftatin therapy as a chemopreventative or adjuvant treat-

ent in CRC,2,3 but the epidemiological evidence for a

eneficial effect is conflicting,4 and the biological mech-nism of action of statins in CRC remains obscure.

HMG-CoA reductase inhibitors are extensively usedo reduce serum cholesterol and to decrease the inci-ence of cardiovascular and cerebrovascular events.5,6

tatins prevent formation of mevalonate from HMG-oA by inhibiting the enzyme HMG-CoA reductasend thereby inhibiting cholesterol synthesis.7 As wells reducing cholesterol levels, statins inhibit the gener-tion of other products of the mevalonate pathway, in-luding mevalonate and the downstream isoprenoidsfarnesyl pyrophosphate and geranylgeranylpyrophos-hate). Posttranslational isoprenylation is important inetermining the membrane localization and function ofany cellular proteins including small GTPases such asas and Rho.8 In vitro studies show that statins inhibitellular proliferation and induce apoptosis in colon can-er cells.9,10 Because Ras mutations are frequent in tu-ors,11 and Rho proteins participate in growth-factor

ignaling,12 the study of the action of statins in tumorells has largely focused on their ability to inhibit thesemall GTPases.13 However, there is evidence that this mayot be the only mechanism by which statins inhibitroliferation and induce apoptosis.2,14

A screen of 30,000 compounds for their ability toctivate a BMP-2 promoter construct in mouse osteoblastell lines found that 2 statins (simvastatin and lovastatin)ad the highest activity and went on to show the anabolicroperties of statins on bone in vivo in mice.15

BMPs are part of the transforming growth factor-�uperfamily. They signal by cooperatively binding type Ind type II BMP receptors, which in turn phosphorylatehe BMP-specific SMADs1, 5, and 8, which then complexith SMAD4. The complex translocates to the nucleusnd modulates gene transcription of BMP target genesuch as ID-2. The BMP pathway has recently been impli-

Abbreviations used in this paper: BMP, bone morphogenetic protein;RC, colorectal cancer; HMG-CoA reductase, 3-hydroxy-3-methylglu-aryl coenzyme A reductase.

© 2007 by the AGA Institute0016-5085/07/$32.00

doi:10.1053/j.gastro.2007.08.021

ctjase

sttCdsatS

cDSscfcLmatS

wSe

M1tRhtMmetwe

(ibcmscTmwmbbrbbmmcSaIfwfpoDwLmL

ppccVwfwtaP

Bnl

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2007 STATINS ACT ON BMP PATHWAY IN COLON CANCER 1273

ated in CRC with the identification of germline muta-ions in BMPR1a and SMAD4 in families with familialuvenile polyposis syndrome.16 Affected individuals have

greatly increased risk of developing cancer.17,18 We havehown that BMP2 acts to promote apoptosis in maturepithelial cells in the colon.19

In this article, we have investigated the effects oftatins on CRC cell lines and xenografts in order to testhe hypothesis that statins act through their actions onhe BMP pathway. We show that statins induce BMP2 inRC cells and that noggin specifically blocks statin-in-uced apoptosis in colon cancer cell lines. Our resultsuggest that loss of SMAD4 confers resistance to statinsnd leads to growth promotional effects. This implieshat statin therapy may only be effective in a subgroup ofMAD4-expressing CRCs.

Materials and MethodsCell CultureDLD1, SW480, HT29, and HCT116 colon cancer

ell lines were obtained from the ATCC and cultured inulbecco’s Modified Eagle’s Medium (Gibco, Paisley,

cotland) with 4.5g/L glucose and L-glutamine. This wasupplemented with penicillin (50 U/mL) and streptomy-in (50 �g/mL) and, where serum was used, with 10%etal calf serum (Gibco). HCT116 SMAD4�/� cells wereultured in McCoy’s 5A Medium (Sigma Aldrich, St.ouis, MO) supplemented with 0.4 mg/mL G418 and 0.1g/mL hygromycin B. Cells were grown in monolayers inhumidified atmosphere containing 5% CO2. Simvasta-

in, lovastatin and pravastatin were all obtained fromigma Aldrich.

Cell CountCells were trypsinized, and 20 �L of cell suspension

as taken for counting using Z2 Coulter Particle Count andize Analyzer (Beckman Coulter B.V., Mijdrecht, The Neth-rlands) according to the manufacturer’s instructions.

MTT AssayCells were trypsinized and taken up in Dulbecco’s

odified Eagle’s Medium with 0.5% fetal calf serum. 5 �03 cells were seeded in flat-bottomed tissue-culture-reated 96-well plates (Greiner Bio-One B.V., Alphen a/dijn, The Netherlands) and allowed to adhere for 12ours. Cells were then stimulated with different concen-rations of statins at 24 and 48 hours. After treatment,

TT solution was added (final concentration 0.5 mg/L, stock solution 5 mg/mL MTT in phosphated buff-

red saline) for 3 hours. The medium was discarded, andhe cells were lysed in acidified 2-propanol. Absorbanceas measured at 550 to 560 nm. Ten wells were used for

ach treatment condition (n � 10). f

ImmunoblottingCells at 60% to 80% confluence from 6-well plates

Greiner Bio-One B.V., Alphen a/d Rijn) were washed ince-cold PBS and scraped into 200 �L of 2 � sampleuffer (125 mmol/L Tris/HCl, pH 6.8; 4% sodium dode-yl sulfate (SDS); 2% �-mercaptoethanol; 20% glycerol, 1g bromophenol blue). Protein concentration was mea-

ured using the RC DC protein assay kit (Biorad, Her-ules, CA) according to the manufacturer’s instructions.he lysates were sonicated and then heated at 95° for 5inutes. Fifty micrograms of protein from each sampleas loaded onto SDS-PAGE and blotted onto PVDFembrane (Millipore, Bedford, MA). The blots were

locked in block buffer (2% low-fat milk powder in Tris-uffered saline with 1% Triton [TBST]) for 1 hour atoom temperature and washed 3 � 10 minutes in TBSTefore overnight incubation at 4°C with primary anti-ody in primary antibody buffer (TBST with 0.2% low-fatilk powder). Primary antibodies to BMP2 (mouseonoclonal) were from R&D (Abingdon, UK). Goat poly-

lonal antibodies to pSMAD1, 5, and 8 were from Cellignaling Technology (Beverly, MA). Mouse monoclonalntibodies to SMAD4, rabbit polyclonal antibodies tod1, and rabbit polyclonal antibodies to �-actin wererom Santa Cruz Biotechnology (Santa Cruz, CA). Blotsere then washed 3 � 10 minutes in TBST and incubated

or 1 hour at room temperature in 1:2000 horseradisheroxidase-conjugated goat anti-rabbit, rabbit anti-goat,r goat anti-mouse secondary antibody (Dako, Glostrup,enmark) in block buffer. After a final 3- � 10-minuteash in TBST, blots were incubated for 5 minutes inumilite Plus (Boehringer-Mannheim, Mannheim, Ger-any), and chemiluminescence was then detected using a

umi-Imager (Boehringer-Mannheim).

Annexin V and 7-Amino-ActinomycinD AssaysControl and treated cells were collected and resus-

ended in 1 � binding buffer (0.01 mol/L Hepes/NaOH,H 7.4, 0.14 mmol/L NaCl and 2.5 mmol/L CaCl2) at aoncentration of 2 x 107 cells/mL. Subsequently, 100 �L ofell suspension was transferred to a 5-mL tube, and Annexin-APC (5 �L) and 7-amino-actinomycin D (7-AAD) (5 �L)as added. The cells were incubated at room temperature

or 15 minutes, after which 400 �L of 1 � binding bufferas added, apoptosis was analyzed by flow cytometry (Bec-

on Dickinson FACSCalibur, Rockville, MD), and data werenalyzed using the software Cell Quest Pro (BD Biosciencesharmingen, Breda, the Netherlands).

Luciferase Reporter AssayCells were transiently transfected either with

MP2-Luc vector or BRE-Luc vector reporter in combi-ation with a cytomegalovirus promoter-driven Renilla

uciferase vector (Promega, Madison, WI) using Lipo-

ectamine Plus (Invitrogen, Breda, The Netherlands) accord-

imba(nEfc

fcLatG

Camstsics

stetpvwabiwbv[whn1

Ss

t

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1274 KODACH ET AL GASTROENTEROLOGY Vol. 133, No. 4

ng to manufacturer’s instructions. After 24 hours of treat-ent with lovastatin, cells were lysed with passive lysis

uffer as provided by Promega, and luciferase activity wasssayed according to the Dual-Glo Luciferase Assay SystemPromega) protocol on a Lumat Berthold LB 9501 Lumi-ometer (Berthold Technologies, Bad Wildbad, Germany).ach firefly luciferase value was corrected for its cotrans-

ected cytomegalovirus-driven Renilla luciferase standard toorrect for transfection efficiency or dilution effects.

Plasmids and Transfection ProcedureHCT116 SMAD4�/� cells were transiently trans-

ected either with HAHA-SMAD4 plasmid or pmaxGFPontrol vector (Amaxa GmbH, Cologne, Germany) usingipofectamine Plus (Invitrogen, Breda, The Netherlands)ccording to manufacturer’s instructions. Efficiency ofransfection was determined by the measurement ofFP-positive cells and was at least 70%.

Tumor Growth in Nude MiceFemale NMRI nu/nu mice were obtained from

harles River (Maastricht, The Netherlands). The study waspproved by the Institutional Animal Care and Use Com-ittee of the Academic Medical Center, University of Am-

terdam, The Netherlands. Mice were sacrificed when theirumors reached 1,000 mm3 in size to avoid undue animaluffering. Groups of 8 female NMRI nu/nu mice werenjected subcutaneously in the flank with 1 � 106 HCT116ells or with 5 � 106 HT29 cells in Matrigel (BD, Bio-ciences). Mice were fed ad libitum with food containing

imvastatin (Arie Blok BV, Woerden, The Netherlands)hereby receiving 50 mg/kg per day for the duration of thexperiment, initiated when the tumor volume reached 100o 200 mm.3 We chose simvastatin as being the second mostotent statin in our in vitro experiments and because sim-astatin is licensed for use in humans in The Netherlands,hile lovastatin is not.20 The treatment was well toler-ted by all animals with no signs of toxicity (loss ofody weight, behavior changes). Controls received

dentical food without simvastatin. Tumor volumesere determined by external measurement performedy 2 different independent animal technicians. Tumorolume was determined according to the equation V �L � W2] � 0.5, where V is volume, L is length, and W isidth. After the mice were sacrificed, the tumors werearvested, frozen in liquid nitrogen, and later homoge-ized in cell lysis buffer (Cell Signaling Technology) withmmol/L Pefabloc (Sigma Aldrich).

Statistical AnalysisStatistical analysis was performed using 2-tailed

tudent’s t test, and P � .05 was considered statisticallyignificant. Data are shown as mean � SEM.

ResultsCRC Cell Lines Differ in Their Sensitivity toStatin TreatmentUsing a panel of CRC cell lines, a wide variation in

he ability of lovastatin to induce cell death was found using

Figure 1. The effects of statins onthe growth of CRC cell lines. (A) TheMTT cell viability assay was performedin 4 colon cancer cell lines treated for24 hours with various concentrationsof lovastatin. Values are expressed asa percentage of controls. Data arefrom 3 experiments. � SEM (n � 10).(B) Immunoblots of SW480, HCT116,HT29, and DLD1 colon cancer celllines for SMAD4 with �-actin as aloading control. (C) HCT116 andHT29 cells were treated for 24 hourswith various concentrations of lova-statin, simvastatin or pravastatin, andthe MTT assay was performed as in

panel A.

tHsa

Stw(

Fv�c3mcl�

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2007 STATINS ACT ON BMP PATHWAY IN COLON CANCER 1275

he same doses as previous in vitro studies9 (Figure 1A).CT116 and DLD-1 cell lines were significantly more sen-

itive than SW480 and HT29. At lower doses, there was evenslight growth stimulatory effect of statins in HT29 and

igure 2. Lovastatin activates the BMP pathway in HCT116 cells, buarious concentrations of lovastatin (�mol) for 24 hours. The expres-actin served as a loading control. (B) HCT116 cells were treated withombination with mevalonate (100 �mol) for 24 hours, and apoptosis w). (C) HCT116 cells were treated with lovastatin or lovastatin in combinicroscope. (D) HCT116 cells were treated with lovastatin or lovastat

ounted. Values are expressed as a percentage of controls. Data areovastatin or lovastatin in combination with noggin (0.5 �g/mL) for 24 hours

SEM (n � 3). In (F) cells were counted and values are expressed as a percen

W480 cells. The cell lines that were resistant to statinreatment are known to express no SMAD4 protein,hereas the sensitive cell lines express SMAD4 normally21

Figure 1B). We further investigated the effect of 3 different

in SW480 cells. (A) Immunoblots of colon cancer cells treated withof proteins was analyzed using the corresponding specific antibody.tatin, lovastatin in combination with noggin (0.5 �g/mL), or lovastatin inuantified by flow cytometry. Data are from 3 experiments � SEM (n �with noggin (0.5 �g/mL) for 24 hours and then photographed under aombination with noggin (0.5 �g/mL) for 24 hours and subsequently

3 experiments � SEM (n � 4). (E and F) SW480 cells were treated with) apoptosis was quantified by flow cytometry. Data are from 3 experiments

t notsionlovasas q

ationin in cfrom. In (E

tage of controls. Data are from 3 experiments � SEM.

saetrtciHlc

v

sBwn(auCsitan(twtnns

crscce

fH(ScTt(sm

rfawSL4le

st

F(2c�

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1276 KODACH ET AL GASTROENTEROLOGY Vol. 133, No. 4

tatins, lovastatin, simvastatin and pravastatin, in HCT116nd HT29 cells. Simvastatin and lovastatin show the sameffect with both leading to slight but significant increases inhe growth of HT29 cells at a concentration of 2 �mol andeduced cell viability at higher concentration, while leadingo significant reductions in HCT116 cell viability at alloncentration (Figure 1C). Pravastatin also leads to a similarncrease in the growth of HT29 cells but no effect on

CT116 cells. This is most likely due to the low lipophy-icity of pravastatin leading to low uptake into nonhepaticells.22

Lovastatin Induces Activation of theCanonical BMP Pathway in HCT116 CellsBut Not in SW480 CellsIn statin-sensitive HCT116 cells treated with lo-

astatin, levels of BMP2 protein increased with a corre-

igure 3. HCT116 cells were treated with 5-fluorouracil (A) or aspirinB) with or without the simultaneous addition of noggin (0.5 �g/mL) for4 hours and subsequently counted. Values obtained in the absence ofompound have been set at 100. Results represent the mean � SEM (n

u4) of 3 experiments.

ponding increase in levels of the activated form of theMP-specific SMADs (pSMAD1, 5, and 8) and ID2,hereas in SMAD4-negative SW480 cells, lovastatin didot lead to increases in BMP2, pSMAD1, 5, 8, or ID2

Figure 2A). Similarly, HCT116 SMAD4�/� cells gener-ted by targeted deletion of the SMAD4 gene23 show nop-regulation of BMP2, pSmad1, 5, 8, or ID2 (Figure 2A).onfirming the importance of BMP pathway activity in

tatin-induced apoptosis, the addition of a specific BMP-nhibitor, noggin24 in the form of an Fc chimera, was ableo prevent lovastatin-induced apoptosis in HCT116 cellss effectively as adding back mevalonate (Figure 2B), withear-complete inhibition of statin-induced apoptosis

Figure 2B and C). Similar data were obtained when cellsreated with lovastatin or with lovastatin and nogginere counted (Figure 2D). Noggin treatment was not able

o reverse the low levels of apoptosis induced in SMAD4-egative SW480 cells (Figure 2E), and cell counting didot show any difference between lovastatin and lova-tatin plus noggin-treated SW480 cells (Figure 2F).

When noggin was used in combination with otherytotoxic and proapoptotic agents such as 5-fluorou-acil, aspirin (Figure 3A, B), and FAS ligand (data nothown) it had no effect, suggesting that noggin spe-ifically blocks the cytotoxic effects of statins. The Fchimera, rituximab was used to control for nonspecificffects of the Fc chimera itself (data not shown).

The Sensitivity of Colon Cancer Cell Lines toStatins Is Influenced by SMAD4 ExpressionTo further investigate the importance of SMAD4

or the sensitivity of cell lines to statins, we studiedCT116 cells with a targeted deletion of SMAD4

SMAD4�/� cells) and compared them with theirMAD4-expressing counterparts. HCT116 SMAD4�/�ells express no SMAD4 protein as shown in Figure 2G.hese SMAD4�/� cells are far less sensitive to lovastatin

reatment than their SMAD4-expressing counterpartsFigures 4A and B). This is in contrast to their relativeensitivities to 5-fluorouracil where SMAD4�/� cells are

ore sensitive (data not shown).A reporter construct containing repeats of a BMP-

esponsive element (BRE) from the ID-1 promoter and aull-length BMP2 promoter reporter construct were bothctivated when transiently transfected HCT116 cellsere treated with lovastatin. In contrast, HCT116MAD4�/� cells show no increased BRE-Luc or BMP2-uc reporter activity upon lovastatin treatment (FigureC and D). SMAD4�/� cells also show higher basal

evels of BRE-Luc activation even in the absence of exog-nous ligand consistent with previous reports.25

Reconstitution of HCT116 SMAD4�/� cells by tran-ient transfection with SMAD4 restored lovastatin sensi-ivity and lovastatin-induced activation of BRE-Luc (Fig-

re 4E and F).

asbbctsic

t

HwmhhcohHdxt8p

FS�(HSwsce1HtacmaHsvatwaSwvwSftwStBc�btmD(

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2007 STATINS ACT ON BMP PATHWAY IN COLON CANCER 1277

The sensitive cell lines studied (HCT116 and DLD1) arelso both microsatellite unstable cell lines, whereas the re-istant cell lines (SW480 and HT29) are microsatellite sta-le. To investigate the possibility that microsatellite insta-ility could be responsible for statin sensitivity, HCT116ells where the mismatch repair defect has been corrected byhe addition of wild-type chromosome 3 were compared totandard HCT116 cells. The correction of microsatellitenstability did not alter the statin sensitivity of HCT116ells (data not shown).

Statin Treatment Inhibits the Growth ofHCT116 Xenografts in Mice and Promotesthe Growth of HT29 XenograftsNext, we evaluated the effect of orally adminis-

igure 4. The response of HCT116MAD4�/� and HCT116 SMAD4/� cell lines to lovastatin treatment.

A) The MTT assay was performed inCT116 SMAD4�/� or HCT116MAD4�/� cells treated for 24 hoursith various concentrations of lova-tatin. Values are expressed as a per-entage of controls. Data are from 3xperiments � SEM ***P � .001 (n �0). (B) HCT116 SMAD4�/� orCT116 SMAD4�/� cells were

reated with lovastatin for 24 hours,nd apoptosis was quantified by flowytometry. Data are from 3 experi-ents � SEM. *P � .05 (n � 3). (C

nd D) HCT116 SMAD4�/� orCT116 SMAD4�/� cells were tran-iently transfected with BMP2-Lucector (C) or the BRE-Luc vector (D)nd treated with various concentra-ions of lovastatin for 24 hours. Dataere normalized to Renilla luciferasectivity. Data are from 3 experiments �EM (n � 4). (D) as in C, except cellsere transfected with the BRE-Luc

ector. (E) HCT116 SMAD4�/� cellsere transiently transfected either withMAD4 or with GFP construct, treated

or 24 hours with various concentra-ions of lovastatin, and the MTT assayas performed as in A. (F) HCT116MAD4�/� cells were transiently co-

ransfected either with SMAD4 andRE-Luc or with GFP and BRE-Luconstructs, treated for 24 hours with 2mol of lovastatin or 100 ng/mL recom-inant human BMP2, and transcrip-ional activity was measured and nor-alized to Renilla-luciferase activity.ata are from 3 experiments � SEM

n � 4).

ered simvastatin on growth of the SMAD4 –wild-type H

CT116 cell line and SMAD4-negative HT 29 cell linehen these cells were implanted as xenografts in nudeice. Treatment of the nude mice with simvastatin in-

ibited tumor growth in HCT116 xenografts but en-anced the growth of HT 29 xenografts compared toontrols (Figure 5). These results are in agreement withur in vitro data where lower concentrations of lovastatinave growth-promotional effects in HT29 cells, whereasCT116 cells are highly sensitive to statin-induced celleath. To investigate whether the effects we see on theenografts in vivo are associated with the activation ofhe BMP pathway, we assessed levels of pSMAD1, 5, andin tumor lysates. As shown in Figure 5C and D, levels ofSMAD1, 5, and 8 are significantly increased in the

CT116 xenografts in mice receiving simvastatin com-

pax

Cccvcadla

al

omeltsCet

Firv�scpaw

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1278 KODACH ET AL GASTROENTEROLOGY Vol. 133, No. 4

ared to controls. Levels of pSMAD1, 5, and 8 are equalnd almost undetectable in both groups with HT29enografts (Figure 5C).

Low-Dose Statin Treatment Leads to CellDeath and Increased BMP2 Levels In VitroMuch of the evidence for the efficacy of statins in

RC comes from cohorts of patients taking standardholesterol-lowering doses of statins where serum statinoncentrations do not reach those commonly used in initro studies. To investigate the effect of much loweroncentrations of statins, we assessed both cell viabilitynd levels of BMP2 in HCT116 cells treated with low-ose lovastatin for longer periods. Figure 6 shows that

ovastatin leads to significant reductions in cell viability

igure 5. Growth curves of HCT116 and HT29 implants in nude mice. (n the flank with either 1 � 106 HCT116 cells (A) 1 � 106 HT29 cells (B) ineceiving 50 mg/kg per day, initiated when the tumor volume reached 1olumes were determined by external measurement. Mice were sacrific

SEM. *P � .05, **P � .01. (B), as in A, except 5 � 106 HT29 cells wacrificed due to tumor size. (C) Immunoblot analysis of lysates of HControls. Fifty micrograms of total protein was loaded per lane. Equal lohosphorylated SMAD1, 5, and 8 expression suggesting activation of Bnd 8 in lysates of HCT116 xenografts from mice treated with simvastatas set at 1. Error bars represent the SEM. ***P � .001.

nd to increased levels of BMP2 even at concentrations i

pproximating those achieved with standard cholesterol-owering doses of statins (� 0.1 �mol/L).26

DiscussionThe effects of statins on CRC has been the subject

f extensive previous investigation as reviewed by De-ierre et al.2 In humans, the evidence for the antitumor

fficacy of statins in CRC comes largely from epidemio-ogical studies in statin users. One such study showedhat using statins for at least 5 years was associated withignificant (47%) reduction in the risk for developingRC providing evidence for a CRC chemopreventative

ffect of statins in humans.1 However, results from fur-her observational studies in humans remain conflict-

B) Groups of 8 female NMRI nu/nu mice were injected subcutaneouslyigel. Mice were fed ad libitum with food containing simvastatin, thereby200 mm3. Controls received identical food without simvastatin. Tumoren the largest tumors reached 1000 mm3. Results represent the meaned, and simvastatin treatment was for 12 days before mice had to be

6 and HT29 xenografts from mice treated with simvastatin and fromwas confirmed by assessing �-actin. Simvastatin treatment increasesignaling. (D) Graph shows the relative mean expression of pSMAD1, 5,from controls analyzed by immunoblotting. Expression in control mice

A andMatr

00 toed where us

T11adingMP sin and

ng,4,27 whereas an observed reduction in stage IV cancers

ittdwbtctdt

5esecvto

nsct

ovewgdpWwcxodthpm

abaTnttHiticlmHtt

ascooatptspS

ccst

Fc1sVeto�

BA

SIC–

ALI

MEN

TARY

TRA

CT

October 2007 STATINS ACT ON BMP PATHWAY IN COLON CANCER 1279

n users may support a role for statins in the postinitia-ion phase of CRC in humans.27 Statins have been showno be effective antitumor agents in various different ro-ent models of CRC, both alone 28 –30 and in combinationith NSAIDs9,28 and COX-2 inhibitors.28,31 Efficacy haseen demonstrated in models of tumor initiation,31,32

umor growth,33 and metastasis33 compatible with bothhemotherapeutic effects at doses achievable in the shorterm in humans34 and chemopreventative effects usingoses at or below those currently used for hypercholes-erolemia in humans.29

Previous in vitro studies have used doses of betweenand 50 �mol/L.9,14 The mild growth stimulatory

ffects we observe in selected CRC cell lines are onlyeen at dose of 2 �mol. This may explain why thisffect has not been previously reported. That differentell lines have different sensitivities to statins has pre-iously been reported with HCT116 noted to be par-icularly sensitive and HT29 resistant, consistent withur findings.10

Our studies were designed to investigate the mecha-ism behind the effects of statins in CRC. The modelystems available to investigate this better mimic thehemotherapeutic use of statins than their chemopreven-

igure 6. Low concentrations of lovastatin induce BMP2 and reduceell viability. (A) HCT116 cells were plated in concentration 1000 and0,000 per well in the 6-well plates and treated for 3 weeks with lova-tatin (0.1 �mol). Cell viability was assessed using the MTT assay.alues are expressed as a percentage of controls. Data are from 3xperiments SEM (n � 10). **P � .01. (B) Immunoblots of HCT116 cellsreated with lovastatin (0.15 �mol) for 48 and 72 hours. The expressionf BMP2 was analyzed using the corresponding specific antibody.-actin served as a loading control.

ative use. However, previous studies using a wide range t

f doses and in models that better mimic the chemopre-entative setting have already been performed.29 –32 Ourxperiments in vivo investigate whether the differencese see in statin sensitivity between CRC cells of differentenetic backgrounds in vitro translate into the sameifferences in vivo. This is particularly relevant given therevious finding that statins block tumor angiogenesis.35

e have therefore chosen a xenograft model of CRChere tumors of different genetic backgrounds can be

ompared in vivo. Whereas previous studies of statins inenografts in mice have used intratumoral (100 mg/kg)36

r intra-peritoneal injections of statins (25 mg/kg)37 toemonstrate an antitumor effect, we chose oral adminis-ration of statins to more closely mimic their use inumans, and we administered doses in line with theserevious studies to ensure that we saw an effect in thisodel.Mechanistic studies have shown that statins lead to

poptosis in CRC cell lines and that this can be preventedy giving back products of the mevalonate pathway suchs mevalonate and geranylgeranylpyrophosphate.9,14,38

hese molecules are required for the prenylation andormal function of G proteins such as Ras and Rho, andhus the major current theory is that statins prevent CRChrough their inhibition of isoprenylation of G proteins.owever, the mutation status of these G proteins has no

nfluence on the sensitivity of cell lines to statins.39 Fur-hermore, adding back geranylgeranylpyrophosphate ornhibition of protein translation with cycloheximideompletely reverses the effects of statins (both morpho-ogical changes and apoptosis), while neither restore the

embrane localization of small GTPases such as Rho.14

ence, it has been suggested that statins may functionhrough an alternative mechanism involving the produc-ion of an unidentified protein.

Statins have anabolic effects on bone through specificctivation of the BMP2 promoter, increasing the expres-ion of BMP2 mRNA and protein by osteoblast-likeells.15 Our results show that statins have similar effectsn the BMP pathway in CRC cell lines. Presumably, thisccurs through direct activation of the BMP2 promoter,s in bone cells. This would be supported by the findinghat lovastatin activates a BMP2-promoter luciferase re-orter construct. We further show that CRC cell sensi-ivity to statins is influenced by the presence or ab-ence of SMAD4, a central and critical part of the BMPathway. Thirty percent of human CRCs express noMAD4.40

ConclusionIn conclusion, statins induce apoptosis in CRC

ells by the induction of BMP. CRC cells missing aentral element of the BMP pathway, SMAD4, are insen-itive to statin treatment and even show a tendencyoward increased growth. Our results might help to iden-

ify a subgroup of patients that could benefit from statin

taobTwit

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

BA

SIC–

ALIM

ENTA

RY

TRA

CT

1280 KODACH ET AL GASTROENTEROLOGY Vol. 133, No. 4

reatment and a subgroup of patients that could responddversely. Although several authors have proposed trialsf statins in CRC,2 these data suggest that caution woulde advised in trials of statin therapy in unselected CRC.he development of statins or statin-related moleculesith enhanced effects on the BMP pathway may offer

ncreased benefit in the treatment of CRC while main-aining the excellent safety profile of the statins.

References

1. Poynter JN, Gruber SB, Higgins PD, Almog R, Bonner JD, RennertHS, Low M, Greenson JK, Rennert G. Statins and the risk ofcolorectal cancer. N Engl J Med 2005;352:2184–2192.

2. Demierre MF, Higgins PD, Gruber SB, Hawk E, Lippman SM.Statins and cancer prevention. Nat Rev Cancer 2005;5:930–942.

3. Katz MS. Therapy insight: potential of statins for cancer chemo-prevention and therapy. Nat Clin Pract Oncol 2005;2:82–89.

4. Jacobs EJ, Rodriguez C, Brady KA, Connell CJ, Thun MJ, Calle EE.Cholesterol-lowering drugs and colorectal cancer incidence in alarge United States cohort. J Natl Cancer Inst 2006;98:69–72.

5. Downs JR. Primary prevention of acute coronary events withlovastatin in men and women with average cholesterol levels:results of AFCAPS/TexCAPS. Air Force/Texas Coronary Athero-sclerosis Prevention Study. 1998;279:1615–1622.

6. Furberg CD, Adams HP Jr, Applegate WB, Byington RP, EspelandMA, Hartwell T, Hunninghake DB, Lefkowitz DS, Probstfield J,Riley WA, et al. Effect of lovastatin on early carotid atherosclero-sis and cardiovascular events. Asymptomatic Carotid Artery Pro-gression Study (ACAPS) Research Group. Circulation 1994;90:1679–1687.

7. Alberts AW, Chen J, Kuron G, Hunt V, Huff J, Hoffman C, RothrockJ, Lopez M, Joshua H, Harris E, Patchett A, Monaghan R, Currie S,Stapley E, Albers-Schonberg G, Hensens O, Hirshfield J, Hoogs-teen K, Liesch J, Springer J. Mevinolin: a highly potent competi-tive inhibitor of hydroxymethylglutaryl-coenzyme A reductase anda cholesterol-lowering agent. Proc Natl Acad Sci U S A 1980;77:3957–3961.

8. Zhang FL, Casey PJ. Protein prenylation: molecular mechanismsand functional consequences. Annu Rev Biochem 1996;65:241–269.

9. Agarwal B, Rao CV, Bhendwal S, Ramey WR, Shirin H, Reddy BS,Holt PR. Lovastatin augments sulindac-induced apoptosis in co-lon cancer cells and potentiates chemopreventive effects of su-lindac. Gastroenterology 1999;117:838–847.

0. Agarwal B, Bhendwal S, Halmos B, Moss SF, Ramey WG, Holt PR.Lovastatin augments apoptosis induced by chemotherapeuticagents in colon cancer cells. Clin Cancer Res 1999;5:2223–2229.

1. Bos JL. ras oncogenes in human cancer: a review. Cancer Res1989;49:4682–4689.

2. Prendergast GC, Khosravi-Far R, Solski PA, Kurzawa H, LebowitzPF, Der CJ. Critical role of Rho in cell transformation by oncogenicRas. Oncogene 1995;10:2289–2296.

3. Wong WW, Dimitroulakos J, Minden MD, Penn LZ. HMG-CoAreductase inhibitors and the malignant cell: the statin family ofdrugs as triggers of tumor-specific apoptosis. Leukemia 2002;16:508–519.

4. Agarwal B, Halmos B, Feoktistov AS, Protiva P, Ramey WG, ChenM, Pothoulakis C, Lamont JT, Holt PR. Mechanism of lovastatin-induced apoptosis in intestinal epithelial cells. Carcinogenesis2002;23:521–528.

5. Mundy G, Garrett R, Harris S, Chan J, Chen D, Rossini G, BoyceB, Zhao M, Gutierrez G. Stimulation of bone formation in vitro and

in rodents by statins. Science 1999;286:1946–1949.

6. Howe JR, Bair JL, Sayed MG, Anderson ME, Mitros FA, PetersenGM, Velculescu VE, Traverso G, Vogelstein B. Germline muta-tions of the gene encoding bone morphogenetic protein receptor1A in juvenile polyposis. Nat Genet 2001;28:184–187.

7. Howe JR, Mitros FA, Summers RW. The risk of gastrointestinalcarcinoma in familial juvenile polyposis. Ann Surg Oncol 1998;5:751–756.

8. Brosens LA, van Hattem A, Hylind LM, Iacobuzio-Donahue C,Romans KE, Axilbund J, Cruz-Correa M, Tersmette AC, OfferhausGJ, Giardiello FM. Risk of colorectal cancer in juvenile polyposis.Gut 2007;56;965–967.

9. Hardwick JC, Van Den Brink GR, Bleuming SA, Ballester I, VanDen Brande JM, Keller JJ, Offerhaus GJ, Van Deventer SJ, Pep-pelenbosch MP. Bone morphogenetic protein 2 is expressed by,and acts upon, mature epithelial cells in the colon. Gastroenter-ology 2004;126:111–121.

0. Mantel-Teeuwisse AK, Klungel OH, Schalekamp T, VerschurenWM, Porsius AJ, de Boer A. Suboptimal choices and dosing ofstatins at start of therapy. Br J Clin Pharmacol 2005;60:83–89.

1. Woodford-Richens KL, Rowan AJ, Gorman P, Halford S, BicknellDC, Wasan HS, Roylance RR, Bodmer WF, Tomlinson IP. SMAD4mutations in colorectal cancer probably occur before chromo-somal instability, but after divergence of the microsatellite insta-bility pathway. Proc Natl Acad Sci U S A 2001;98:9719–9723.

2. McTaggart F, Buckett L, Davidson R, Holdgate G, McCormick A,Schneck D, Smith G, Warwick M. Preclinical and clinical pharma-cology of Rosuvastatin, a new 3-hydroxy-3-methylglutaryl coen-zyme A reductase inhibitor. Am J Cardiol 2001;87:28B–32B.

3. Zhou S, Buckhaults P, Zawel L, Bunz F, Riggins G, Le Dai J, KernSE, Kinzler KW, Vogelstein B. Targeted deletion of Smad4 showsit is required for transforming growth factor beta and activinsignaling in colorectal cancer cells 10.1073/pnas.95.5.2412.PNAS 1998;95:2412–2416.

4. Groppe J, Greenwald J, Wiater E, Rodriguez-Leon J, EconomidesAN, Kwiatkowski W, Affolter M, Vale WW, Belmonte JC, Choe S.Structural basis of BMP signalling inhibition by the cystine knotprotein Noggin. Nature 2002;420:636–642.

5. Beck SE, Jung BH, Fiorino A, Gomez J, Rosario ED, Cabrera BL,Huang SC, Chow JY, Carethers JM. Bone morphogenetic proteinsignaling and growth suppression in colon cancer. Am J PhysiolGastrointest Liver Physiol 2006;291:G135–145.

6. Pan HY, DeVault AR, Wang-Iverson D, Ivashkiv E, Swanson BN,Sugerman AA. Comparative pharmacokinetics and pharmacody-namics of Pravastatin and lovastatin. J Clin Pharmacol 1990;30:1128–1135.

7. Coogan PF, Smith J, Rosenberg L. Statin use and risk of colorec-tal cancer. J Natl Cancer Inst 2007;99:32–40.

8. Reddy BS, Wang CX, Kong AN, et al. Prevention of azoxymethane-induced colon cancer by combination of low doses of atorvasta-tin, aspirin, and celecoxib in F 344 rats. Cancer Res 2006;66:4542–4546.

9. Narisawa T, Fukaura Y, Tanida N, Hasebe M, Ito M, Aizawa R.Chemopreventive efficacy of low dose of Pravastatin, an HMG-CoA reductase inhibitor, on 1,2-dimethylhydrazine-induced coloncarcinogenesis in ICR mice. Tohoku J Exp Med 1996;180:131–138.

0. Narisawa T, Morotomi M, Fukaura Y, Hasebe M, Ito M, Aizawa R.Chemoprevention by Pravastatin, a 3-hydroxy-3-methylglutaryl-co-enzyme A reductase inhibitor, of N-methyl-N-nitrosourea-inducedcolon carcinogenesis in F344 rats. Jpn J Cancer Res 1996;87:798–804.

1. Swamy MV, Patlolla JMR, Steele VE, Kopelovich L, Reddy BS, RaoCV. Chemoprevention of familial adenomatous polyposis by lowdoses of atorvastatin and celecoxib given individually and incombination to APCM in mice. 10.1158/0008-5472. CAN-05-

4619. Cancer Res 2006;66:7370–7377.

3

3

3

3

3

3

3

3

4

DiJ

o

lgva –

TRA

CT

October 2007 STATINS ACT ON BMP PATHWAY IN COLON CANCER 1281

2. Kim KP, Whitehead C, Piazza G, Wargovich MJ. Combinatorialchemoprevention: efficacy of lovastatin and exisulind on theformation and progression of aberrant crypt foci. Anticancer Res2004;24:1805–1811.

3. Taras D, Blanc JF, Rullier A, Dugot-Senant N, Laurendeau I,Vidaud M, Rosenbaum J. Pravastatin reduces lung metastasis ofrat hepatocellular carcinoma via a coordinated decrease of MMPexpression and activity. J Hepatol 2007;46:69–76.

4. Holstein SA, Knapp HR, Clamon GH, Murry DJ, Hohl RJ. Pharma-codynamic effects of high dose lovastatin in subjects with ad-vanced malignancies. Cancer Chemother Pharmacol 2006;57:155–164.

5. Vincent L, Soria C, Mirshahi F, Opolon P, Mishal Z, Vannier JP,Soria J, Hong L. Cerivastatin, an inhibitor of 3-hydroxy-3-methyl-glutaryl coenzyme a reductase, inhibits endothelial cell prolifera-tion induced by angiogenic factors in vitro and angiogenesis in invivo models. Arterioscler Thromb Vasc Biol 2002;22:623–629.

6. Sebti SM, Tkalcevic GT, Jani JP. Lovastatin, a cholesterol biosyn-thesis inhibitor, inhibits the growth of human H-ras oncogenetransformed cells in nude mice. Cancer Commun 1991;3:141–147.

7. Khan SG, Saxena R, Bickers DR, Mukhtar H, Agarwal R. Inhibitionof ras p21 membrane localization and modulation of proteinkinase C isozyme expression during regression of chemical car-cinogen-induced murine skin tumors by lovastatin. Mol Carcinog1995;12:205–212.

8. Xia Z, Tan MM, Wong WW, Dimitroulakos J, Minden MD, Penn LZ.

Blocking protein geranylgeranylation is essential for lovastatin- (

induced apoptosis of human acute myeloid leukemia cells.Leukemia 2001;15:1398–1407.

9. Muller C, Bockhorn AG, Klusmeier S, Kiehl M, Roeder C, KalthoffH, Koch OM. Lovastatin inhibits proliferation of pancreatic cancercell lines with mutant as well as with wild-type K-ras oncogene buthas different effects on protein phosphorylation and induction ofapoptosis. Int J Oncol 1998;12:717–723.

0. Thiagalingam S, Lengauer C, Leach FS, Schutte M, Hahn SA,Overhauser J, Willson JK, Markowitz S, Hamilton SR, Kern SE,Kinzler KW, Vogelstein B. Evaluation of candidate tumour sup-pressor genes on chromosome 18 in colorectal cancers. NatGenet 1996;13:343–346.

Received February 23, 2007. Accepted July 12, 2007.Address requests for reprints to: James C. H. Hardwick, MD, PhD,

epartment of Gastroenterology & Hepatology, Leiden University Med-cal Center, P.O. Box 9600, 2300 RC, Leiden, The Netherlands. e-mail:.C.H.Hardwick@lumc.nl; fax: (31) 71-5248115.

JH is supported by the Dutch Cancer Society and the Netherlandsrganization for health research and development.HCT116 SMAD4�/� cells and HCT116 SMAD4�/� parental cell

ine as well as HAHA-SMAD4 plasmid were provided by Dr Bert Vo-elstein (Johns Hopkins University, Baltimore, MD). The BMP2-Lucector was provided by Lilly Research Laboratories (Indianapolis, IN),nd the BRE-Luc vector reporter was a gift from Professor ten Dijke

Leiden University Medical Center, Leiden, The Netherlands).

BA

SIC

ALI

MEN

TARY