Mutations in the SMAD4/DPC4 Gene in Juvenile Polyposis

See allHide authors and affiliations

Science  15 May 1998:
Vol. 280, Issue 5366, pp. 1086-1088
DOI: 10.1126/science.280.5366.1086


Familial juvenile polyposis is an autosomal dominant disease characterized by a predisposition to hamartomatous polyps and gastrointestinal cancer. Here it is shown that a subset of juvenile polyposis families carry germ line mutations in the geneSMAD4 (also known as DPC4), located on chromosome 18q21.1, that encodes a critical cytoplasmic mediator in the transforming growth factor–β signaling pathway. The mutant SMAD4 proteins are predicted to be truncated at the carboxyl-terminus and lack sequences required for normal function. These results confirm an important role for SMAD4 in the development of gastrointestinal tumors.

Familial juvenile polyposis (JP) is an autosomal dominant disease in which individuals are predisposed to hamartomatous polyps and gastrointestinal cancer. Gastrointestinal malignancy develops in 9 to 68% of JP patients (1). Two groups have reported that a subset of JP patients harbor mutations in the protein phosphatase gene PTEN, located on chromosome 10q23 (2). PTEN is somatically mutated in many human tumor types and is the gene responsible for Cowden disease and Bannayan-Ruvalcaba-Riley syndrome (3). Other groups have found no evidence of linkage to markers on 10q or PTEN mutations in JP families (4). These results suggest that there is genetic heterogeneity in JP families, or that JP patients previously described with 10q abnormalities may have actually been Cowden disease or Bannayan-Ruvalcaba-Riley syndrome patients (5).

We recently mapped a gene predisposing to JP to chromosome 18q21.1, between markers D18S1118 and D18S487 (6), an interval that contains the two putative tumor suppressor genes DCC andSMAD4 (7). The high incidence of colorectal cancer (as well as one case of pancreatic cancer) in affected members of the JP kindred displaying 18q21 linkage (the Iowa JP kindred) (8) led us to hypothesize that one of these tumor suppressor genes could be the gene predisposing to JP. Because of the complexity of DCC [29 exons spanning 1.4 Mb (9)], we initially searched for germ line mutations by single-strand conformation polymorphism (SSCP) analysis of five family members (three affected, two unaffected) (10). Shifts were detected in exons 1, 8, and 16, but these did not cosegregate with the disease. We then changed our mutation screening strategy and began sequencing genomic polymerase chain reaction (PCR) products generated from one affected individual for each exon of DCC andSMAD4 (11). After sequencing 14 DCCexons and all 11 SMAD4 exons, we detected a 4–base pair (bp) deletion in exon 9 of SMAD4. The patient's affected brother had the same heterozygous deletion and his unaffected mother had the wild-type sequence for exon 9. To confirm this mutation, we subcloned the exon 9 PCR product from this patient into a plasmid vector and sequenced the individual alleles (12). One allele was the wild type and the other had a 4-bp deletion (Fig.1) between nucleotides 1372 and 1375 (codons 414 to 416) of the cDNA sequence [GenBank accession numberU44378 (13)]. This deletion causes a frameshift that creates a new stop codon at the end of exon 9 (nucleotides 1432 to 1434 of the wild-type sequence, codon 434).

Figure 1

Sequences of the wild-type (upper) and mutant (lower) alleles of SMAD4 exon 9 (nucleotides 1365 to 1382) from an affected member of the Iowa JP kindred. The rectangle indicates the 4 bp deleted in the mutant allele (arrow).

We next analyzed exon 9 of SMAD4 from all 46 members of the Iowa JP kindred by PCR amplification and denaturing polyacrylamide gel electrophoresis. The altered allele was present in all 13 affected individuals, none of 7 spouses, and 4 of 26 individuals at risk [two-point lod score of 5.79, θ = 0 (the lod score is the logarithm of the odds favoring linkage and θ is the recombination fraction)]. This altered allele was also readily observed on SSCP gels (Fig.2). To exclude the possibility that this alteration represented a polymorphism, we amplified exon 9 from 242 unrelated individuals (484 chromosomes). The altered allele was not observed in this population. DNA extracted from gastrointestinal polyps was also used to amplify SMAD4 exon 9. This analysis revealed loss of the wild-type allele in 1 of 11 tumors derived from five affected individuals (Fig. 3).

Figure 2

(A) Denaturing and (B) nondenaturing gels of Iowa JP kindred family members showing theSMAD4 exon 9 PCR product. Affected individuals 4, 5, 6, and 11, as well as one at risk (8), all have an extra band [arrow in (A)] on denaturing gels that is produced by the 4-bp deletion. The mutant allele is also seen as a shift by SSCP analysis [arrows in (B)].

Figure 3

PCR amplification of SMAD4 exon 9 from microdissected polyps. Pedigree numbers correspond to affected individuals as described (6). Loss of the wild-type allele (arrow) is seen in a juvenile polyp from patient IV-17 (fourth lane from the left). DNA was extracted from paraffin-embedded polyps after microdissection (28). Amplification of exon 9 was performed with the primers 5′-TAGGCAAAGGTGTGCAGTTG-3′ and 5′-TGCACTTGGGTAGATCTTATGAA-3′, which generate a 152-bp product from within the exon. C, colon; S, stomach; VA, villous adenoma; AP, adenomatous polyp; JP, juvenile polyp.

Eight additional unrelated JP patients were subsequently analyzed for mutations of all exons of SMAD4 by SSCP and genomic sequencing (Table 1). Two JP kindreds were found that segregated a similar 4-bp deletion in exon 9. Because of the nature of the sequence in this region, these deletions can begin at any of four consecutive nucleotides and result in the same mutant sequence and new stop codon. The three kindreds segregating these deletions were all Caucasian and originated from Iowa, Mississippi, and Finland. There was no common ancestral haplotype, as assessed by analysis of microsatellite markers close to SMAD4. Sequencing did not reveal any intragenic polymorphisms that would be useful in evaluating common ancestry, and it is unclear whether this defect is an ancestral founder mutation or a mutational hotspot. A patient with colonic and gastric JP (whose father has a history of gastrointestinal symptoms but has not been evaluated clinically) was found to have a 2-bp deletion in exon 8 of SMAD4, at nucleotides 1170 to 1171 (codon 348). This deletion causes a frameshift that creates a stop codon at nucleotides 1178 to 1180 (codon 350). Another patient diagnosed with 30 to 40 colonic juvenile polyps at age 6 but with no family history of JP (four siblings and both parents unaffected) was found to have a 1-bp insertion between nucleotides 815 and 820 of exon 5; this change added a guanine to a stretch of six sequential guanines in the wild-type sequence and created a frameshift and a new stop codon at nucleotides 830 to 832 (codon 235). NoSMAD4 mutations were found in four other unrelated JP patients.

Table 1

Analysis of SMAD4 mutations in nine unrelated JP patients. Under “Type,” F refers to familial and S to sporadic JP. wt, wild type.

View this table:

Somatic mutations in SMAD4 have been reported in up to 50% of human pancreatic tumors (13, 14) and 15% of colorectal tumors (15). The occasional development of pancreatic cancer and the high incidence of colorectal cancer (40% in the Iowa JP kindred) in JP families is consistent with these findings in sporadic tumors. It remains to be determined whether the locus on 18q21 involved in the development of sporadic colorectal cancers is SMAD4, DCC, another closely linked gene, or a combination of these genes.

SMAD4 is a member of the SMAD family of genes, which code for cytoplasmic mediators in the transforming growth factor–β (TGF-β) signaling pathway (16). This pathway mediates growth inhibitory signals from the cell surface to the nucleus. Upon activation by TGF-β or related ligands, serine-threonine kinase receptors phosphorylate various SMAD proteins, which then form heteromeric complexes with SMAD4 in the cytoplasm (17). These complexes then migrate to the nucleus, where they are thought to regulate transcription through association with various DNA binding proteins (18). The growth inhibitory effect of TGF-β on pancreatic cancer cell lines requires functional SMAD4 (19).

SMAD4 is a 552–amino acid protein (13). Its COOH-terminus appears to be important for the formation of SMAD4 homotrimers, which then complex with other SMAD proteins. Mutations that disrupt homotrimer formation lead to loss of TBF-β signaling (20). A SMAD4 mutant lacking 38 COOH-terminal amino acids has a dominant negative effect on SMAD2-mediated mesoderm induction inXenopus embryos and forms oligomers with wild-type SMAD4 that may be responsible for this loss of activity (17). The majority of somatic mutations described in SMAD4 map to the COOH-terminus between codons 330 and 526 (13, 14, 21, 22) within several highly conserved domains. The 4-bp deletion detected in three JP families is predicted to produce a COOH-terminally truncated protein of 433 amino acids, with loss of regions critical for normal function. The 1-bp insertion and 2-bp deletion seen in two other patients are predicted to result in truncated proteins of 234 and 349 amino acids, respectively. Although deletion of the wild-type allele was seen in only one of 11 polyps, some of these may have been contaminated with normal cells during microdissection. Alternatively, other somatic SMAD4 mutations may have been present in these samples, or germ line mutation of SMAD4 may induce tumors through a dominant negative effect.

One of the features of the gastrointestinal polyps seen in compoundApc-Smad4 mutant heterozygote mice is the increased proliferation of stromal cells (23), which is one of the characteristic features of juvenile polyps seen in humans. It has also been shown in Xenopus embryos that wild-typeSMAD4 induces mesodermal markers and that mixtures of mutant and wild-type SMAD4 inhibit this response (17). In JP patients, it would appear that germ line SMAD4 mutations predispose to focal abnormalities of mesenchymal development (hamartomas) and cancer through disruption of the TGF-β signaling pathway. JP may be a genetically heterogeneous condition, as evidenced by the fact that not all families are linked to 18q markers (24) and not all families studied had germ lineSMAD4 mutations. It is possible that germ line mutations in genes encoding different components of the TGF-β signaling pathway may be present in these other JP kindreds. The roles of the Cowden disease gene (PTEN) and Peutz-Jeghers syndrome gene [LKB1 (25)] in cell growth control remain unclear, although PTEN may be down-regulated by TGF-β (26). Further studies on the components of the TGF-β pathway may add to our understanding of these hamartomatous polyposis syndromes.

  • * To whom correspondences should be addressed. E-mail: james-howe{at}


View Abstract

Navigate This Article