Haploid loss of the tumor suppressor Smad4/Dpc4 initiates gastricpolyposis and cancer in mice

Xiaoling Xu1,3, Steven G Brodie1,3, Xiao Yang1,3, Young-Hyuck Im2, W Tony Parks2, Lin Chen1,Yong-Xing Zhou1, Michael Weinstein1, Seong-Jin Kim2 and Chu-Xia Deng*,1

1Genetics of Development and Disease Branch, 10/9N105, National Institute of Diabetes, Digestive and Kidney Diseases, NationalInstitutes of Health, Bethesda, Maryland, MD 20892, USA; 2Laboratory of Cell Regulation and Carcinogenesis, National CancerInstitute, National Institutes of Health, Bethesda, Maryland, MD 20892, USA

The tumor suppressor SMAD4, also known as DPC4,deleted in pancreatic cancer, is a central mediator ofTGF-b signaling. It was previously shown that micehomozygous for a null mutation of Smad4 (Smad47/7)died prior to gastrulation displaying impaired extraem-bryonic membrane formation and endoderm di�erentia-tion. Here we show that Smad4+/7 mice began todevelop polyposis in the fundus and antrum when theywere over 6 ± 12 months old, and in the duodenum andcecum in older animals at a lower frequency. Withincreasing age, polyps in the antrum show sequentialchanges from hyperplasia, to dysplasia, in-situ carcino-ma, and ®nally invasion. These alterations are initiatedby a dramatic expansion of the gastric epithelium whereSmad4 is expressed. However, loss of the remainingSmad4 wild-type allele was detected only in later stagesof tumor progression, suggesting that haploinsu�ciencyof Smad4 is su�cient for tumor initiation. Our data alsoshowed that overexpression of TGF-b1 and Cyclin D1was associated with increased proliferation of gastricpolyps and tumors. These studies demonstrate thatSmad4 functions as a tumor suppressor in the gastro-intestinal tract and also provide a valuable model forscreening factors that promote or prevent gastrictumorigenesis. Oncogene (2000) 19, 1868 ± 1874.

Keywords: Smad4; Dpc4; Juvenile polyposis; gastriccancer; TGFb1; Cyclin D1

Introduction

Gastric carcinomas are a leading cause of cancerincidence and mortality worldwide (Kinzler andVogelstein, 1996; Pisani et al., 1993). As in manytumor types, gastric carcinogenesis re¯ects a multistepprocess of morphologic abnormalities associated withthe accumulation of genetic alterations, includingactivation of oncogenes and/or inactivation of tumorsuppressor genes (reviewed in Correa, 1992; Kinzlerand Vogelstein, 1996; Moss, 1998). It was recentlydemonstrated that one mechanism of gastric cancerprogression is the inactivation of the TGF-b signalingpathway (Markowitz and Roberts, 1996; Yang et al.,1999a), where SMAD4/DPC4 is the central mediator(Heldin et al., 1997; Massague, 1998).

Human SMAD4/DPC4 is located in chromosome18q21.1, in which loss of heterozygosity (LOH) isfrequently demonstrated in a wide variety of tumors,including pancreatic, colon and gastric adenocarcinomas(Hahn et al., 1996; Howe et al., 1998a; Tamura et al.,1996). SMAD4/DPC4 was ®rst cloned as a tumorsuppressor that is deleted or mutated in about 40 ±50% of pancreatic cancers (Hahn et al., 1996). Mutationsin Smad4 have also been identi®ed in about 30% of coloncarcinomas and less than 10% of other cancers(Nagatake et al., 1996; Schutte et al., 1996). Recently,it was reported that germline mutation of humanSMAD4 contributes to familial juvenile polyposis, anautosomal dominant disorder characterized by predis-position to hamartomatous polyps and gastrointestinalcancer (Friedl et al., 1999; Howe et al., 1998b).

In addition to SMAD4, eight other SMAD proteinshave been described in the TGF-b signaling pathways(reviewed in Heldin et al., 1997; Massague, 1998). Thefunctions of several SMAD family members have beenstudied using gene targeting. By introducing mutationsinto di�erent functional domains in these genes, weand others have demonstrated that Smad2 playsmultiple roles in mesoderm induction, left-rightsymmetry, and anterior-posterior axis formation (No-mura and Li, 1998; Waldrip et al., 1998; Weinstein etal., 1998). Smad3 mediates TGF-b signals for mucosalimmunity and tumor suppressor activity in coloncancer (Yang et al., 1999c; Zhu et al., 1998). Smad5is indispensable for normal angiogenesis during earlyembryogenesis (Chang et al., 1999; Yang et al., 1999b).However, the most severe phenotype was observed inSmad4-null embryos which died at embryonic day 6.5without signs of gastrulation (Sirard et al., 1998; Yanget al., 1998). They also exhibited impaired extraem-bryonic membrane formation and decreased epiblastproliferation (Yang et al., 1998). Interestingly, whenSmad4 heterozygous mice were bred with Apc hetero-zygous mice, the cis-compound Smad4+/7; Apc+/7 micedeveloped intestinal carcinomas with all the tumorsshowing LOH for both wild-type alleles (Takaku et al.,1998). This observation indicates that reduced TGF-b/Smad4 signals in combination with other factorspredispose mice to cancer.

To study whether loss of one Smad4 allele caninitiate tumorigenesis, we closely monitored our Smad4heterozygous mice (referred to as Smad48+/7 because ofthe targeted disruption in exon 8 of Smad4) (Yang etal., 1998) for up to 20 months. We found thatSmad48+/7 mice began to develop polyps in the fundus,antrum, duodenum and cecum of the gastrointestinaltract at di�erent ages. The antral polyps in older mice

Oncogene (2000) 19, 1868 ± 1874ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00

www.nature.com/onc

*Correspondence: C-X Deng,3The ®rst three authors contributed equally to this studyReceived 9 November 1999; revised 21 January 2000; accepted 31January 2000

progressed to adenocarcinoma through a multistepprocess of morphological changes, including hyperpla-sia, dysplasia, and in-situ and invasive carcinoma.These morphological changes were accompanied by aseries of alterations including increased epithelialproliferation and upregulation of TGF-b1 and CyclinD1. Of note, LOH of the remaining Smad4 wild-typeallele occurred at later stages of tumorigenesis. Thisstudy provides compelling evidence that SMAD4 is atumor suppressor gene in the gastrointestinal (GI) tractand also provides an animal model to study themechanisms of gastric cancer formation.

Results

Smad48+/7mice developed polyposis in the GI tract

We have previously showed that the Smad487/7 micedied prior to gastrulation, while the Smad48+/7 micewere phenotypically normal up to 8 months in terms ofgrowth rate, morphology, health and fertility (Yang etal., 1998). Because human SMAD4/DPC4 has beenimplicated in the formation of pancreatic and GI tractcancers or polyps at varying frequencies (Friedl et al.,1999; Hahn et al., 1996; Howe et al., 1998b; Nagatakeet al., 1996; Schutte et al., 1996), we followedSmad48+/7 mice to determine the e�ect of gene dosageon tumor formation. After examination of the GItracts of Smad48+/7 mice over 1 year of age, weidenti®ed masses that developed in the lower region ofthe stomach (Figure 1a). Upon dissection, it was clearthat the masses were arising above the pyloricsphincter, and that the enlargement was caused bypolyps formed in the antrum (Figure 1b ± d). Next, wecarefully examined Smad48+/7 mice of varying ages inan attempt to describe and follow tumor progression.Histologically, polyps were ®rst detected in the fundusregion of some six month old Smad48+/7 mice. Thea�ected areas increased in size as the mice aged. Thefundal polyps were very small in size and clustered inthe a�ected area (Figure 1f), and rarely developed intoadenomas, remaining the same size in most olderanimals. Antral polyps with varying sizes were found inall Smad48+/7 mice that were over one year of age(n=25). Small polyps were sometimes found in young-er animals (Figure 1b). Compared with fundal polyps,the antral polyps were much larger, but fewer innumber. Polyps and tumors were also found in thececum (2/25, Figure 1g), and duodenum (5/25, Figure1h) of older animals at lower frequencies. This was anintriguing ®nding since heterozygous SMAD4 muta-tions were recently identi®ed in human colorectalcancers and juvenile polyposis (Friedl et al., 1999;Howe et al., 1998b; Miyaki et al., 1999).

To further analyse the antral tumors, we performedhistologic analysis on Smad48+/7 mice and littermatecontrols. We analysed antral tissues for changes inglandular architecture and cellular morphology. AsSmad48+/7 mice increased in age, we found the antralglands became more severely tortuous, dilated, andelongated compared to controls (Figure 2a ± g). Thesetumors progressed with age from simple hyperplasia(Figure 2b) to atypical hyperplasia (Figure 2c), dysplasia(Figure 2d), carcinoma in situ (Figure 2e), and ultimatelyto invasive carcinoma (Figure 2f, g). Increases in stromal

tissue and leukocyte in®ltration were also observed(Figure 2e,f). Most animals over 12 months exhibitedpolyps with severe dysplasia (Figure 2d). The in situcarcinoma had characteristic changes including promi-nent nucleoli, increased mitoses, decreased gland spa-cing, and anaplastic cells with nuclear enlargement andhyperchromasia (Figure 2e). The ®ndings were identi®edin three animals over 18 months of age. Two of theseanimals also had regions suspicious for invasion, and oneanimal had a de®nitive invasive carcinoma (Figure 2f,g).Testing for H. felis infection, which presents animportant carcinogenic risk factor in gastric cancer(Fox et al., 1996, 1997), was negative, indicating thatthe gastric tumors are indeed caused by genetic changesand not by environmental insult. In the fundal region wefound variously increased proliferation of foveolar cells,pariental cells and neck cells (data not shown).Hyperplasias were identi®ed in both the duodenal cryptsand colonic glands (Figure 2h). One duodenal lesionproved to be an invasive adenocarcinoma (Figure 2i).

LOH of Smad4 wild-type allele occurs late in gastrictumorigenesis

Loss of heterozygosity is a common mechanism forinactivation of tumor suppressor genes. Therefore, wewere interested in determining the status of the wild-type Smad4 allele during tumor progression. We

Figure 1 GI polyps and tumor formation in Smad48+/7 mice.(a ± d) Representative samples of antral polyps/tumors (arrows)found in 8 month (a,b), 15 month (c), and 18 month (d) oldSmad48+/7 mice. (An) Antrum, (De) duodenum, (Fu) fundus. (e,f) Hyperplastic polyps found in the fundus of a 7 month oldmouse (f), but not in its littermate control (e). (g) Polyposis foundin the cecum (Ce) of a 8 month old Smad48+/7 mouse (arrow).(h) A tumor (arrow) found in duodenum of a 15 month oldSmad48+/7 mouse. Magni®cations: 3.26for a ± c, g and h; 26ford; 106for e and f

Oncogene

Smad4/Dpc4 and gastric cancerX Xu et al

1869

performed Southern blot analysis using antral tumorsfrom Smad48+/7 animals ranging in age from 3 ± 18months. Five of eight tumors having diameters of over0.4 cm demonstrated loss of heterozygosity (LOH) atthe Smad4 locus (Figure 3a). In contrast, only one of®ve tumors (or polyps) of smaller sizes showed LOH.In addition, we have also performed micro-dissectionfollowed by PCR and con®rmed that the polyps/tumors that did not show LOH contained the wild-typeallele of Smad4 (data not shown). Histologic analysisindicated that tumors showing LOH were from animalswith severe antral dysplasia and/or carcinoma (seeFigure 2e ± g). Consistent with the LOH data, thesetumors showed no Smad4 protein as revealed byimmunohistochemical staining (see Figure 4a,b). Alto-gether, these results indicate LOH of Smad4 occurs as

a late event in antral tumor progression, but is not anobligate step in tumor initiation.

Many oncogenes and tumor suppressor genes havebeen implicated in the pathogenesis of gastric andcolon cancer, and are also altered in other forms ofcancer. These genetic changes include the inactivationof p53, mutations in K-ras, decreased expression ofp21, ampli®cation of c-met and c-ErbB-2, andmicrosatellite instability (reviewed in Kinzler andVogelstein, 1996; Moss, 1998; Tahara, 1995). In anattempt to further characterize tumor progression, weperformed PCR and SSCP analysis on genes com-monly mutated in gastrointestinal cancers. We analysedSmad4 (exon 9), p53 (exons 5, 6, 7), K-ras (exon 1 and2), H-ras (exon 1) and Apc, and found no detectablechanges (Figure 3c). In addition, we performedimmunohistochemistry using antibodies speci®c for c-myc and ErbB-2 but found no signi®cant di�erences instaining of antral tumors compared to control tissue.These data suggest that many of the tumorigenicmechanisms involving these oncogenes and tumorsuppressor genes are not contributing factors for antraltumor progression in Smad48+/7 mice.

Gastric polyps and tumors exhibited increasedproliferation

Tumor progression involves an increase in cellularproliferation and/or decrease in apoptosis. Because

Figure 2 Histologic analysis of polyps and tumors found in theantrum, fundus, and cecum of Smad48+/7 mice. (a ± f) Progres-sion of antral tumor formation demonstrating hyperplasia (b),atypical hyperplasia (c), dysplasia (d). (e ± g) Representative in-situcarcinoma (e,f) and invasive carcinoma (g) with prominentnucleoli, increased mitoses decreased cytoplasmic to nuclear ratio[arrows point to glands, which are losing their normal structure,and arrowheads point to in®ltrating lymphocytes (e), and invasivecancer cells (f,g)]. These abnormalities were not found in wild-type mice (a). (h ± i) Dysplasia (h) and invasive carcinoma of theduodenum (i). Arrow in (i) points to muscle. Magni®cations:296for a ± d; 976for e and f; 1456for g; and 486for h and i

Figure 3 Southern blot and mutation analyses. (a) LOH in theSmad4 locus by Southern blot using EcoRV digestion whichproduces a fragment of 9.5 kb and 3.6 kb from wild-type andmutant allele, respectively. Genotype as indicated. Lanes 1 and 2are DNA's from spleen of a wild-type and a Smad48+/7 mouse,respectively. Others are either from antral tumors (T) or polyps(P). (b) SSCP mutation analysis of GI tumor hot spots.Representative SSCP samples from three advanced tumors (T)and control (N) animals. Exon 5 of p53, exon 1 of K-ras, exon 9of Smad4, and Apc around Min mutation site were examined inthis study. No mutations were detected

Smad4/Dpc4 and gastric cancerX Xu et al

1870

Oncogene

TGF-bs are known to inhibit cellular proliferation andinduce apoptosis in many cell types through Smadsignaling cascade (Ko et al., 1998; Yamamoto et al.,1996), we were interested in determining whether one of

these two mechanisms or a combination was responsiblefor antral tumor progression. To assess proliferation, weperformed 3H-thymidine labeling in control andSmad48+/7 mice. In control antrum and non-hyperplasticregions of tumors, we identi®ed [3H]thymidine positivecells (S-phase cells) primarily at the base of the gastricglands (Figure 4c,e). In contrast, [3H]thymidine positivecells were detected throughout the expanded hyperplasticor dysplastic regions of antral tumors (Figure 4c,d, andnot shown). To determine whether decreased apoptosisalso occurred, we performed TUNEL assays and foundno signi®cant change in apoptotic ®gures betweentumors and controls (not shown). These data suggestthat increased cellular proliferation is a major factorcontributing to antral polyps and tumor progression inSmad48+/7 mice.

Gastric polyps and tumors exhibited increased expressionof TGF-b1 and cyclin D1

TGF-b signals are mediated through the interactions ofSmad2 and Smad3 with the common mediator Smad4.This signaling cascade ultimately leads to activation ofgenes responsible for regulation of growth, di�erentia-tion, embryonic development, and tissue morphogen-esis (Massague, 1990). In order to understandtumorigenesis associated with Smad4 loss, we usedimmunohistochemistry to screen for proteins expressedin gastric tissues to monitor changes associated withtumorigenesis. TGF-b1 is normally expressed in thegastric epithelial cells (Figure 4f). Its expression wasincreased in antral tumors in a non-uniform fashion(Figure 4g). Similar non-uniform increased expressionof TGF-b RII in tumors was also observed (notshown). These are interesting ®ndings, since TGF-b1overexpression is found in some human gastric cancers(Naef et al., 1997; Yoshida et al., 1989).

TGF-b1 inhibits proliferation of most epithelial celltypes by blocking progression of the G1 phase of thecell cycle and inhibiting Cyclin D1 expression (Ko etal., 1998). Overexpression of Cyclin D1 was shown toattenuate the responsiveness of cells to TGF-b throughdown-regulation of TGF-b RII expression (Okamotoet al., 1994). Our data showed that Cyclin D1 isnormally expressed at a low level in the gastricepithelial cells (Figure 5h). In Smad48+/7 mice, slightlyincreased expression of Cyclin D1 was found in areasof hyperplasia (not shown). However, stronger stainingwas found in advanced tumors (Figure 4i). We showedearlier that the advanced tumors exhibited LOH ofSmad4 (Figure 3a), and failed to express Smad4

Figure 4 Proliferation and immunohistochemical analyses ofpolyps/tumors in Smad48+/7 mice. (a,b) Immunohistochemistryusing an antibody to Smad4 to show its expression in controlantrum epithelium (a, arrow) and a tumor (b), which is negativefor the staining. (c) 3H-thymidine labeling of a highly proliferativehyperplasic antral polyp (arrow) and adjacent non-hyperplasticregion (arrowhead) from a 12 month old Smad48+/7 mouse. (d)An enlarged image of hyperplastic antrum shown in (c). (e) 3H-thymidine labeling of the antrum from an age matched controlmouse, showing proliferative cells near the base of gastric glands.(f,g) TGF-b1 expression near the apical region in control antrumepithelium (f), and in a tumor from Smad48+/7 mouse at anincreased level (g, arrow). (h,i) Cyclin D1 expression in isthmus ofcontrol antrum (h, arrow), and in tumor showing higher intensity(i). Magni®cations: 506for a, b, d, e, f and g; 156for c; and37.56for h and i

Figure 5 Summary of morphological and molecular changes associated with antral tumor progression in Smad48+/7 mice. Smad4mediated signals are required to maintain normal growth of gastric epithelium. However, haploinsu�ciency caused by the loss ofone Smad4 allele results in increased proliferation of the antral epithelium, which initiates polyp formation. Other events, includingincreases in Cyclin D1 lead to further increased rates of proliferation and the loss of TGF-b1 cell cycle control, leading to geneticalterations such as LOH for Smad4 and subsequent dysplasia. Additional genetic changes would eventually result in carcinomaformation

Oncogene

Smad4/Dpc4 and gastric cancerX Xu et al

1871

protein (Figure 4b), thus, the higher expression ofCyclin D1 in these tumors is consistent with a viewthat TGF-b/Smad4 signals inhibit cell proliferationthrough repressing the activity of this gene.

Discussion

The human SMAD4/DPC4 gene was initially cloned asa tumor suppressor in patients with cancers of thepancreas and other tissues (Hahn et al., 1996).However, homozygous disruption of Smad4 in micecaused early embryonic lethality, preventing studies ofSmad4's functions during later stages of developmentand tumorigenesis. In this study, we assessed the tumorsuppressor function of Smad4 in the GI tract usingSmad48+/7 mice. We showed that all Smad48+/7 micedeveloped hyperplasia of the fundus and antrum. Thefundal hyperplasia rarely developed into tumors whilethe polyps found in the antrum increased in size andnumber, eventually developing into adenocarcinoma asmice aged. Polyps were also found in the duodenumand cecum of Smad48+/7 mice but they occurred atlower frequencies. These observations indicated thatSmad4 is a major tumor suppressor gene in the GItract, especially in the stomach.

The development of gastric cancer in Smad48+/7 miceshowed a multistep process of morphological changesranging from hyperplasia to dysplasia, in-situ andlikely invasive carcinoma. This prolonged complexprocess shares striking similarity with the adenocarci-noma sequence (Kinzler and Vogelstein, 1996; Vogel-stein, 1990). However, SSCP analysis andimmunohistochemical studies for genes involved incolorectal tumorigenesis showed no changes in struc-ture (p53, K-ras, H-ras, Apc and Smad4) and/orexpression (c-Myc, ErbB-2, E-cad and b-catenin),suggesting di�erences in molecular mechanisms be-tween gastric and colorectal tumor progression. Inaddition, Smad4 LOH was not detected until the laterstages of antral carcinogenesis suggesting that thehaploid loss of Smad4 initiated polyps and tumors.However, further tumor progression required the lossof Smad4, which serves as a rate-limiting step foradvanced tumor formation.

Germline mutations in SMAD4 have been found in asubset of families with familial juvenile polyposis (FJP)(Howe et al., 1998b). FJP is an autosomal dominantdisease, characterized by predisposition to hamartoma-tous polyps and gastrointestinal cancer (Jarvinen andFranssila, 1984). In a large Iowa FJP kindred, all 13a�ected individuals carried a 4 bp deletion in SMAD4(exon 9), which created a premature stop codon leadingto protein truncation. In addition, SMAD4 mutationanalysis of nine unrelated FJP patients demonstratedthat three patients carried the same 4 bp deletion, onecarried a 2 bp deletion in exon 8 and one carried a1 bp insertion in exon 5. The most interesting ®ndingwas that only one of 11 tumors examined demon-strated LOH of the wild-type SMAD4 allele, suggestingthat FJP is caused by reduced SMAD4 signals.Therefore, both the murine and human forms of GIpolyposis share similar mechanisms in that they areboth initiated by SMAD4 haploinsu�ciency.

It is important to gain an understanding of howhaploinsu�ciency of SMAD4 results in tumor initia-

tion. SMAD4 is a common mediator for all TGF-bfamily members, which constitute a large family withover 40 members. Therefore, loss of one SMAD4 allelecould signi®cantly reduce signals from all familymembers involved in gastric epithelial proliferation. Inaddition, the TGF-b signaling pathway represents adirect pathway, through which membrane bound TGF-b RI phosphorylates the SMADs, which translocate tothe nucleus and trigger transcription of target genes.Because there are no ampli®cation steps betweenmembrane signals and downstream transcription tar-gets, the relative quantity of the intracellular mediator(SMAD4) becomes critical. Indeed, developmentalhaploinsu�ciency is also observed in other Smads.Nomura and Li (1998) recently reported that approxi-mately 10% of Smad2+/7 mice displayed abnormalitiesin mesoderm induction. In another study, mice doublyheterozygous for Smad2 and Smad3 mutations diedduring gestation caused by reduced signaling levels ofthese two genes (Weinstein and Deng, unpublishedobservation). Indeed, tumor initiation without the lossof the wild-type allele has been reported for a number oftumor suppressor genes, including p53 (Venkatachalamet al., 1998), p27 (Fero et al., 1998), and TGF-b1 (Tanget al., 1998), suggesting that haploinsu�ciency of tumorsuppressor genes will become a common phenomenonassociated with various tumor types as a consequence ofimprovement of the detection approaches.

Of all TGF-b family members, TGF-b1 and itsreceptors have been most strongly implicated in gastriccancer formation. TGF-b1 is produced by gastricepithelial cells which, in turn, acts through anautocrine loop to inhibit cellular proliferation bydecreasing Cyclin D1 expression and cyclin-cdk4association (Ko et al., 1998; Naef et al., 1997; Sandhuet al., 1997; Yoshida et al., 1989). The TGF-b1 signalswere also found to induce apoptosis in cultured gastriccancer cells after they were arrested in the G1/S phaseof the cell cycle (Yamamoto et al., 1996). However,many cultured gastric cells lines lose their ability torespond to TGF-b1 inhibition by reducing expressionof TGF-b RI (Ito et al., 1992) or generating mutationsin TGF-b RII (Park et al., 1994; Yang et al., 1999a).Microsatellite instability in TGF-b RII, resulting in thefunctional loss of the receptor, is also found in manygastric tumor cells (Grady et al., 1999; Myero� et al.,1995). The Smad4 heterozygous and homozygoustumor cells (LOH of Smad4) from our study aresimilar to human gastric cancer cells carrying reducedor ine�ective TGF-b receptors, in that they both su�erdownstream disruptions in the TGF-b signaling path-way. We have detected increased TGF-b1 and CyclinD1 expression in both antral polyps and tumors. TGF-b RII staining was generally enhanced in hyperplasticand dysplastic regions of pyloric tumors. The increasedexpression of TGF-b1 and its receptor may be aconsequence of a feedback response caused byblockage in downstream pathways. This overexpressionis also observed in human gastric cancers (Naef et al.,1997; Yoshida et al., 1989). On the other hand, theincreases in Cyclin D1 signaling may be important forfurther tumor progression. Cyclin D1 is an importantcomponent of G1/S cell cycle check point. Its over-expression is thought to inactivate Rb and trigger E2Frelease resulting in increased cellular proliferation(Oswald et al., 1994; Schulze et al., 1994; Sherr,

Smad4/Dpc4 and gastric cancerX Xu et al

1872

Oncogene

1996). Thus, it is conceivable that upregulation ofCyclin D1, whose expression and activity are normallyinhibited by TGF-b1 (Ko et al., 1998; Sandhu et al.,1997), is responsible for the over proliferation ofepithelial cells in antral hyperplasia and tumors.

In summary, we have shown that Smad4 is expressedin gastric epithelium and serves as a negative regulator ofepithelial cell proliferation. The haploinsu�ciency due tothe loss of one wild-type allele in Smad48+/7 mice, overtime, causes an upregulation in epithelial proliferation,which is in part mediated by increased levels of CyclinD1 (Figure 5). This lack of cell cycle control leads tohyperplasia and dysplasia. The increases in TGF-b1 andits receptors may be triggered by epithelial overgrowthleading to compensation or increased expression as ameans to prevent cellular proliferation. However, thisoverexpression may not be e�ective due to the reducedquantity of the down stream mediator (Smad4). TheLOH of Smad4 ultimately leads to further geneticchanges and invasive tumor formation. Because metas-tasis was not observed in Smad48+/7 mice, we suggestthat further genetic changes, in addition to loss ofSmad4, would be required.

The demonstration that Smad4 is a tumor suppres-sor in gastric antral cancer does not exclude thepossibility that Smad4 also plays a role in repressingtumor formation along the entire GI tract. However,the requirement of Smad4 functions in other segmentsof the GI tract appears not as critical as that of gastrictissues. For examples, colon cancer does not initiate inSmad4+/7 mice until an Apc heterozygous mutation isintroduced (Takaku et al., 1998). Similarly, otherunknown factors must exist to account for the lowerfrequencies of duodenal and cecal polyps and tumorsin Smad48+/7 mice. Even in stomach, polyps did notdevelop until several months after birth, suggestingadditional hits, besides the reduced Smad4 signals,must occur to initiate this multistep gastric cancerseries. The identi®cation of these factors deserves a fullinvestigation in the near future.

Materials and methods

Genotype and LOH analysis

Genotyping of Smad48+/7 and control mice, which were in129/Black Swiss or 129/C57 B6 background, was performedusing PCR, and Smad4 LOH studies were performed usingSouthern blot analysis as described (Yang et al., 1998).

Preparation for histologic evaluation and antibody staining

Control and transgenic mice were dissected through midlineincision of the abdomen. The stomach was isolated in wholeby cutting both ends of the stomach after ligation of thelower esophageal end and proximal part of duodenum. Oneml of 10% formalin was injected into gastric cavity using a25 gauge needle, in¯ating the stomach. The in¯ated stomachwas then submerged in ®xatives for 10 min. Next, stomachtissue was cut alongside the greater curvature, washed with

PBS, and opened to lie ¯at on the Whatman paper. Thestomach was wrapped with Whatman paper securelyenclosing the sample, then submerged in ®xatives overnightfor histologic preparation. Sections (5 mm) were stained withhematoxylin and eosin (H&E), or subject to immunohisto-chemical analysis with antibodies speci®c for TGF-b1, TGF-b RII, Smad4, E-cadherin, ErbB-2, b-catenin, Cyclin D1(Santa Cruz, Santa Cruz, CA, USA). Antibody detection wasperformed using histomouse SP kit (Zymed Laboratories,South San Francisco, CA, USA).

In situ hybridization, thymidine labeling and TUNEL assay

In situ hybridization using Smad4 probes was performed asdescribed (Yang et al., 1998). [3H]Thymidine (Amersham)was IP injected at a dosage at 20 mCi/gram. Animals weresacri®ced 2 h post injection. Slides were dipped in emulsion(Kodak NTB-2) and exposed for 4 ± 15 days beforedeveloping. TUNEL staining was performed using ApopTagin situ apoptosis detection kit (Oncor).

PCR±SSCP

DNA was extracted from the gastric tumor tissue and age-matched control tissue using standard proteinase K digestionand phenol/chloroform extraction for examining mutations inthe p53, Ras, Apc, and Smad4 genes. The PCR primer pairsfor the ampli®cation of these genes are showing below. p53exon 5: 5'-TCTCTTCCAGTACTCTCCTC-3'/5'-AGGCGG-TGTTGAGGGCTTAC-3'. Exon 6: 5'-GGCTTCTGACTTA-TTCTTGC-3'/5'-CAACTGTCTCTAAGACGCAC-3'. Exon7: 5'-TCACCTGGATCCTGTGTCTT-3'/5'-CAGGCTAAC-CTAACCTACCA-3'. K-ras exon 1: 5'-ATGACTGAGTA-TAAACTTGT-3'/5'-GCAGCGTTACCTCTATCGTA-5'.Exon 2: 5'-TTCTCAGGACTCCTACAGGA-3'/5'-GATT-TAGTATTATTTATGGC-3'. H-ras exon 1: 5'-GATTGG-CAGCCGCTGTAGAA-3'/5'-GGCAGAGCTCACCTCTA-TAG-3'. Apc (150 pb fragment containing the Min mutation):5'-GTTCTCGTTCTGAGAAAGAC-3'/5'-CTTCCATAACT-TTGGCTATC-3'. Smad4 exon 9: 5'-GAGGTTGCACATA-GGCAAAG-3'/5'-ACCTTTATATACGCGCTTGG-3'. Onehundred ng of genomic DNA was suspended in a 20 ml totalvolume reaction mixture containing 10 mM Tris HCl (pH8.3), 50 mM KCl, 200 ng of each primer, 200 mM of eachdNTP, 1.5 mM MgCl2, 0.25 ml of [a-32P] dCTP (3000 Ci/mmol, 10 Ci/ml), and 0.1 unit of Taq DNA polymerase. PCRwas performed using the following conditions: 35 cycles ofdenaturation at 958C for 1 min, annealing at 54 ± 588C for1 min and extension at 728C for 1 min. Following the PCRreaction, 40 ml of SSCP loading bu�er (95% formamide,10 mM NaOH, 0.025% bromophenol blue, and 0.025%xylene cyanol) were added to the 4 ml aliquots of theradioactive PCR products. The reaction mixture was heatdenatured for 5 min at 958C, and then chilled on ice. Next,electrophoretic analysis was performed using 2 ml; of thereaction mixture on a 6% nondenaturing gel containing 10%glycerol for 12 ± 16 h at 10 W in a cold room. The gel wasdried and exposed to X-ray ®lm overnight at 7808C.

AcknowledgmentsWe thank W Qiao for technical assistance and A Robertsand L Ferrin for critically reading the manuscript andhelpful discussion.

References

Chang H, Huylebroeck D, Verschueren K, Guo Q, MatzukMM and Zwijsen A. (1999). Development, 126, 1631 ±1642.

Correa P. (1992). Cancer Res., 52, 6735 ± 6740.

Fero ML, Randel E, Gurley KE, Roberts JM and Kemp CJ.(1998). Nature, 396, 177 ± 180.

Fox JG, Dangler CA, Whary MT, Edelman W, KucherlapatiR and Wang TC. (1997). Cancer Res., 57, 3972 ± 3978.

Oncogene

Smad4/Dpc4 and gastric cancerX Xu et al

1873

Fox JG, Li X, Cahill RJ, Andrutis K, Rustgi AK, Odze Rand Wang TC. (1996). Gastroenterology, 110, 155 ± 166.

Friedl W, Kruse R, Uhlhaas S, Stolte M, Schartmann B,Keller KM, Jungck M, Stern M, Lo� S, Back W, ProppingP and Jenne DE. (1999). Genes Chromosomes Cancer, 25,403 ± 406.

Grady WM, Myero� LL, Swinler SE, Rajput A, Thiaga-lingam S, Lutterbaugh JD, Neumann A, Brattain MG,Chang J, Kim SJ, Kinzler KW, Vogelstein B, Willson JKand Markowitz S. (1999). Cancer Res., 59, 320 ± 324.

Hahn SA, Schutte M, Hoque AT, Moskaluk CA, da CostaLT, Rozenblum E, Weinstein CL, Fischer A, Yeo CJ,Hruban RH and Kern SE. (1996). Science, 271, 350 ± 353.

Heldin CH, Miyazono K and ten Dijke P. (1997). Nature,390, 465 ± 471.

Howe JR, Ringold JC, Summers RW, Mitros FA, NishimuraDY and Stone EM. (1998a). Am. J. Hum. Genet., 62,1129 ± 1136.

Howe JR, Roth S, Ringold JC, Summers RW, Jarvinen HJ,Sistonen P, Tomlinson IP, Houlston RS, Bevan S, MitrosFA, Stone EM and Aaltonen LA. (1998b). Science, 280,1086 ± 1088.

Ito M, Yasui W, Kyo E, Yokozaki H, Nakayama H, Ito Hand Tahara E. (1992). Cancer Res., 52, 295 ± 300.

Jarvinen H and Franssila KO. (1984). Gut, 25, 792 ± 800.Kinzler KW and Vogelstein B. (1996). Cell, 87, 159 ± 170.Ko TC, Yu W, Sakai T, Sheng H, Shao J, Beauchamp RD

and Thompson EA. (1998). Oncogene, 16, 3445 ± 3454.Markowitz SD and Roberts AB. (1996). Cytokine Growth

Factor Rev., 7, 93 ± 102.Massague J. (1990). Annu. Rev. Cell. Biol., 6, 597 ± 641.Massague J. (1998). Ann. Rev. Biochem., 67, 753 ± 791.Miyaki M, Iijima T, Konishi M, Sakai K, Ishii A, Yasuno M,

Hishima T, Koike M, Shitara N, Iwama T, Utsunomiya J,Kuroki T and Mori T. (1999). Oncogene, 18, 3098 ± 3103.

Moss SF. (1998). Aliment Pharmacol. Ther., 12 Suppl 1, 91 ±109.

Myero� LL, Parsons R, Kim SJ, Hedrick L, Cho KR, OrthK, Mathis M, Kinzler KW, Lutterbaugh J, Park K, et al.(1995). Cancer Res., 55, 5545 ± 5547.

Naef M, Ishiwata T, Friess H, Buchler MW, Gold LI andKorc M. (1997). Int. J. Cancer, 71, 131 ± 137.

Nagatake M, Takagi Y, Osada H, Uchida K, Mitsudomi T,Saji S, Shimokata K, Takahashi T and Takahashi T.(1996). Cancer Res., 56, 2718 ± 2720.

Nomura M and Li E. (1998). Nature, 393, 786 ± 790.Okamoto A, Jiang W, Kim SJ, Spillare EA, Stoner GD,

Weinstein IB and Harris CC. (1994). Proc. Natl. Acad. Sci.USA, 91, 11576 ± 11580.

Oswald F, Lovec H, Moroy T and LippM. (1994).Oncogene,9, 2029 ± 2036.

Park K, Kim SJ, Bang YJ, Park JG, Kim NK, Roberts ABand Sporn MB. (1994). Proc. Natl. Acad. Sci. USA, 91,8772 ± 8776.

Pisani P, Parkin DM and Ferlay J. (1993). Int. J. Cancer, 55,891 ± 903.

Sandhu C, Garbe J, Bhattacharya N, Daksis J, Pan CH,Yaswen P, Koh J, Slingerland JM and Stampfer MR.(1997). Mol. Cell. Biol., 17, 2458 ± 2467.

Schulze A, Zerfass K, Spitkovsky D, Henglein B and Jansen-Durr P. (1994). Oncogene, 9, 3475 ± 3482.

Schutte M, Hruban RH, Hedrick L, Cho KR, Nadasdy GM,Weinstein CL, Bova GS, Isaacs WB, Cairns P, Nawroz H,Sidransky D, Casero Jr RA, Meltzer PS, Hahn SA andKern SE. (1996). Cancer Res., 56, 2527 ± 2530.

Sherr CJ. (1996). Science, 274, 1672 ± 1677.Sirard C, de la Pompa JL, Elia A, Itie A, Mirtsos C, Cheung

A, Hahn S, Wakeham A, Schwartz L, Kern SE, Rossant Jand Mak TW. (1998). Genes Dev., 12, 107 ± 119.

Tahara E. (1995). World J. Surg., 19, 484 ± 488; discussion489 ± 490.

Takaku K, OshimaM,Miyoshi H, Matsui M, Seldin MF andTaketo MM. (1998). Cell, 92, 645 ± 656.

Tamura G, Sakata K, Nishizuka S, Maesawa C, Suzuki Y,Terashima M, Eda Y and Satodate R. (1996). J. Pathol.,180, 371 ± 377.

Tang B, Bottinger EP, Jakowlew SB, Bagnall KM, MarianoJ, Anver MR, Letterio JJ and Wake®eld LM. (1998). Nat.Med., 4, 802 ± 807.

Venkatachalam S, Shi YP, Jones SN, Vogel H, Bradley A,Pinkel D and Donehower LA. (1998). EMBO J., 17,4657 ± 4667.

Vogelstein B. (1990). Nature, 348, 681 ± 682.Waldrip WR, Biko� EK, Hoodless PA, Wrana JL and

Robertson EJ. (1998). Cell, 92, 797 ± 808.Weinstein M, Yang X, Li C, Xu X, Gotay J and Deng CX.

(1998). Proc. Natl. Acad. Sci. USA, 95, 9378 ± 9383.Yamamoto M, Maehara Y, Sakaguchi Y, Kusumoto T,

Ichiyoshi Y and Sugimachi K. (1996). Cancer, 77, 1628 ±1633.

Yang HK, Kang SH, Kim YS, Won K, Bang YJ and Kim SJ.(1999a). Oncogene, 18, 2213 ± 2219.

Yang X, Castilla LH, Xu X, Li C, Gotay J, Weinstein M, LiuPP and Deng CX. (1999b). Development, 126, 1571 ± 1580.

Yang X, Letterio JJ, Lechleider RJ, Chen L, Hayman R, GuH, Roberts AB and Deng C. (1999c). EMBO J., 18, 1280 ±1291.

Yang X, Li C, Xu X and Deng C. (1998). Proc. Natl. Acad.Sci. USA, 95, 3667 ± 3672.

Yoshida K, Yokozaki H, Niimoto M, Ito H, Ito M andTahara E. (1989). Int. J. Cancer, 44, 394 ± 398.

Zhu Y, Richardson JA, Parada LF and Gra� JM. (1998).Cell, 94, 703 ± 714.

Smad4/Dpc4 and gastric cancerX Xu et al

1874

Oncogene