Vol. 10, 1597–1604, March 1, 2004 Clinical Cancer Research 1597 Featured Article Missense Mutations of MADH4: Characterization of the Mutational Hot Spot and Functional Consequences in Human Tumors Christine A. Iacobuzio-Donahue,1 Jason Song,5 Introduction Giovanni Parmiagiani,4 Charles J. Yeo,2,3 Human pancreatic ductal adenocarcinomas inactivate the Ralph H. Hruban,1,2 and Scott E. Kern2 tumor suppressor gene MADH4 (DPC4, SMAD4) with a high frequency (1). This inactivation occurs most commonly by Departments of 1Pathology, 2Oncology, 3Surgery, and 4Public Health, The Johns Hopkins Medical Institutions, Baltimore, Maryland, and homozygous deletion (HD), but some tumors may also inacti- 5Temple University School of Medicine, Philadelphia, Pennsylvania vate the gene by loss of heterozygosity (LOH) coupled with a mutation in the remaining allele. Inactivation by nonsense mu- tation may cause the loss of protein expression by enhanced Abstract proteosomal degradation (2, 3). Even when expressed as protein, Purpose and Experimental Design: The mutational spec- missense mutations may result in loss of a specific function of trum of MADH4 (DPC4/SMAD4) opens valuable insights the Madh4 protein such as DNA binding or Smad protein into the functions of this protein that confer its tumor- interactions (2–9). Thus, the location of these mutations can suppressive nature in human tumors. We present the provide clues to key structural features that mediate the tumor- MADH4 genetic status determined on a new set of pancre- suppressive function of MADH4. atic, biliary, and duodenal cancers with comparison to the Members of the Smad protein family, including Madh4, mutational data reported for various tumor types. have two evolutionarily conserved regions termed the MH1 Results: Homozygous deletion, followed by inactivating and MH2 domains (Mad homology 1 and 2). The NH2-termi- nonsense or frameshift mutations, is the predominant form nal MH1 domain (codons 1 through 142) is responsible for of MADH4 inactivation in pancreatic cancers. Among the sequence-specific DNA binding (5–7, 10–12), whereas hetero- naturally occurring MADH4 missense mutations, the MH2 merization and transactivation functions have been largely at- domain is the most frequent target (77%) of missense mu- tributed to the MH2 domain [codons 319–552 (6, 8, 13)]. In tations in human tumors. A mutational hot spot resides addition, the MH2 region has been shown to partially interfere within the MH2 domain corresponding to codons 330 to 370, with the DNA-binding function of the MH1 region (5, 7, 8, 12, termed the mutation cluster region (MCR). A relationship 14). These domains are separated by a linker region that con- was found between the locations of the missense mutations tains a 48-amino acid segment called the Smad4 activation (the MH1 domain, the MH2-MCR, and the MH2 outside of domain required for the activation of Smad4-dependent signal- the MCR) and the tumor types, suggesting environmental or ing responses (15). selective influences in the development of MADH4 muta- In the seminal study by Hahn (1), six pancreatic carcino- tions. Immunohistochemical studies for Madh4 protein in mas were found to have intragenic mutations of the MADH4 nine archival cancers (six pancreatic cancers, two duodenal gene, only one of which was a missense mutation. Schutte et al. cancers, and one biliary cancer) with known missense mu- (16) reported additional missense mutations in a pancreatic tations indicated that all mutations within the MH1 or MH2 cancer cell line and an ovarian carcinoma. Missense mutations domain COOH-terminal to the MCR (seven of nine cases) have since been reported in biliary cancers (17), neuroendocrine had negative or weak labeling, whereas two cancers with tumors (18), colorectal cancer (19–23), juvenile polyposis syn- mutations within the MCR had strong positive nuclear la- drome (24–28), ovarian cancer (29), and lung cancers (30). beling for Madh4 protein. Thus, the mapping of these mutations to the known domains of Conclusions: These findings have important implica- the MADH4 gene and their presumed effect on the function of tions for in vitro functional studies, suggesting that the ma- the Dpc4 protein are in need of update. We present new data jority of missense mutations inactivate Madh4 by protein characterizing the mutations of the MADH4 gene in a large degradation in contrast to those that occur within the MCR. series of additional pancreatic cancer xenografts and cell lines. These combined data help to better define the spectrum of mutations in the MADH4 gene with respect to clustering and potential effects on tumor-suppressive function in pancreatic Received 7/8/03; revised 12/8/03; accepted 12/8/03. cancers. We compare our findings with those reported for other Grant support: Supported by the NIH Specialized Programs of Re- human tumor types in an effort to consider the possible associ- search Excellence in Gastrointestinal Cancer Grant CA 62924 and Grant ation between the mutation location and tumor types. CA 68228. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked Materials and Methods advertisement in accordance with 18 U.S.C. Section 1734 solely to Generation of Xenografts and Cell Lines. The genera- indicate this fact. tion of xenografted tumors derived from pancreatic, biliary, and Requests for reprints: Scott E. Kern, Department of Oncology, Room 461, Cancer Research Building, 1650 Orleans Street, The Johns Hop- other tumor types has been described in detail previously (31). kins University School of Medicine, Baltimore, Maryland 21231. Human pancreatic cancer cell lines Panc 9.06, Panc 8.13, Panc Phone: (410) 614-3316; Fax: (410) 614-9705; E-mail: [email protected]. 3.27, Panc 2.8, and PL45 are low-passage pancreatic carcinoma Downloaded from clincancerres.aacrjournals.org on September 30, 2021. © 2004 American Association for Cancer Research. 1598 MADH4 Missense Mutations in Human Tumors cell lines kindly provided by Dr. Elizabeth Jaffee (32) or de- the MH2] in relation to the primary tumor type was determined scribed previously (33). All were recently made available by ␹2 analysis. Values of Յ0.05 were considered significant. through the American Type Culture Collection (Manassas, VA). All cell lines were cultured in DMEM supplemented with 10% fetal bovine serum and antibiotics (100 units/ml penicillin and Results 100 ␮g/ml streptomycin). Cells were incubated at 37°Cina MADH4 Gene Inactivation in Pancreatic Cancer. We humidified atmosphere of 5% CO2 in air. analyzed the MADH4 gene in 63 new cases representing 54 Determination of LOH. Previous work demonstrated a pancreatic cancer xenografts, 2 duodenal cancer xenografts, 2 high frequency of LOH at 18q21.1 in these pancreatic cancer biliary cancer xenografts, and 5 pancreatic cancer cell lines. xenografts by use of dinucleotide markers, described previously LOH at 18q21 was determined to be present for all but one of in detail (34). the duodenal cancers. Homozygous deletion was found in 13 Amplification and Sequencing of Exons 0–11 of cases (10 xenografts and 3 cell lines of pancreatic cancer). Ten MADH4. PCR amplification of exons 0–11 from genomic of these HDs involved the entire coding region of MADH4. In DNA was performed as described previously (16, 35). PCR- three additional cases, only a portion of the coding region was amplified products were purified using QIAquick (Qiagen) and homozygously deleted, corresponding to exons 7–11 in cell line studied by automated sequencing using nested primers and an Panc 8.13, exons 5–11 for xenograft PX191, and exons 1–4 for ABI Prism model 3700 (Applied Biosystems, Foster City, CA). xenograft PX194. In all 13 cases, the presence of a HD was Sequence analysis used Sequencher version 4.0.5 software confirmed by the failure of PCR to amplify contiguous DNA (Gene Codes, Ann Arbor, MI). Verification of the mutation was segments in the presence of appropriate positive controls. accomplished by sequencing of a second PCR product derived For the 50 cases in which no HD was found, exons 0–11 of independently from the original template. the MADH4 gene were PCR amplified and sequenced. Eighteen Immunohistochemistry. Unstained 5-␮m sections were different intragenic mutations in 17 cases were identified, cor- cut from the archival paraffin blocks of eight pancreatic, biliary, responding to 13 pancreatic cancer xenografts, two duodenal or duodenal cancers with tumorigenic missense mutations of the cancer xenografts, one biliary cancer xenograft, and one pan- MADH4 gene. Paraffin blocks corresponding to one of the creatic cancer cell line (Table 1). In four pancreatic cancer duodenal carcinomas having a missense mutation were unavail- able. For this case, samples of the normal duodenal mucosa and xenografts and the cancer cell line, an insertion/deletion muta- xenografted tumor were fixed overnight in 10% buffered for- tion was found, and each was predicted to cause a frameshift in malin and embedded in paraffin, and sections were cut. Slides the coding sequence. In three additional pancreatic cancer xe- were deparaffinized by routine techniques followed by incuba- nografts, a nonsense mutation was found, and in nine cases (six tion in 1ϫ sodium citrate buffer (diluted from 10ϫ heat-induced pancreatic cancer xenografts, two duodenal cancer xenografts, epitope retrieval buffer; Ventana-Bio Tek Solutions, Tucson, and one biliary cancer xenograft), a missense mutation was AZ) before steaming for 20 min at 80°C. Slides were cooled for identified in the retained MADH4 allele. One of the missense 5 min and incubated with a 1:100 dilution of monoclonal anti- mutations in a pancreatic cancer xenograft occurred in a location body to Madh4 protein (clone B8; Santa Cruz Biotechnology, predicted to cause a splice site alteration at exon 8. One of the Santa Cruz, CA) using the Bio Tek-Mate 1000 automated duodenal cancer xenografts (PX255) contained two different stainer (Ventana-Bio Tek Solutions).