The observation that familial JP was an autosomal dominant disorder in 19667,8 has been confirmed over the following decades. Twenty to 50 percent of cases have a family history, 74-76 while the remaining cases presumably arise de novo from new mutations, rather than recessive inheritance as hypothesized by Sachatello.57 The mean age at diagnosis has varied between studies, with Veale estimating it at 4.5 years in sporadic cases and 9.5 years in familial cases.8 Coburn's review of the literature, including both sporadic and familial cases, found a mean age of diagnosis of 18.5 years, 75 while Howe et al. reported the mean age at presentation to be 30.5 years in the Iowa kindred.23 The age of diagnosis is clearly influenced by whether patients are detected by screening endoscopy, or whether they present with symptoms, and thus far there have been no reliable prospective or retrospective studies to carefully determine the age of onset. For these reasons, the age-related penetrance in JP is also unknown, but there is evidence to suggest that it is incomplete by middle age. Howe et al. reported that one member of the Iowa kindred inheriting a predisposing germ line mutation was asymptomatic and has had both negative upper and lower endoscopy at the age of 44.5 years.94 There is variable expression in terms of upper and lower GI polyposis within and between families, and the association of multiple juvenile polyps with a variety of genetic conditions suggest that there is significant genetic heterogeneity, with multiple chromosomal loci influencing the development of these polyps. A slight preponderance of cases in males have been described in some reports (17 of 23 by Desai, 80 15 of 27 by Restrepo, 32 80 of 149 by Coburn75), but the sex distribution in the Iowa kindred was essentially equal (14 males of 29).23
Cytogenetic and Loss of Heterozygosity (LOH) Studies
In 1997, Jacoby and colleagues reported a 4-year-old patient with multiple colonic juvenile polyps, hypoplastic ears, tricuspid insufficiency, widely spaced canthi, and who was <5th percentile for head circumference, height, and weight. On cytogenetic analysis, this patient was noted to have an interstitial deletion of 10q22.3-q24.1, leading the authors to hypothesize that this region contained a tumor-suppressor gene associated with JP.24 Based on the findings in this patient, they evaluated this region for LOH in juvenile polyps from 13 unrelated JP patients (5 of whom had less than 5 polyps) and 3 with solitary juvenile polyps. The marker D10S219 was deleted in 39 of 47 (83 percent) polyps from these 16 patients and the minimal region of overlap of all deletions was an approximately 3-cM region between markers D10S219 and D10S1696. They next performed fluorescent in situ hybridization on tissue sections from juvenile polyps using a cosmid clone mapping near this region, and compared the staining in epithelial versus lamina propria cells. They found that 30 of 39 (77 percent) polyps examined had somatic deletions from 10q, predominantly within lymphocytes and macrophages in the lamina propria. These findings suggested the presence of a tumor-suppressor gene predisposing to juvenile polyps on 10q22 (named JP1), 25 in the same vicinity as the locus recently mapped for Cowden syndrome.38
Prior to these reports, linkage studies in familial JP had been limited to one study that had excluded APC and MCC as the gene for JP in an Australian family.95 After Jacoby's report of 10q deletions in juvenile polyps, Marsh and associates genotyped 4 microsatellite markers from this region on 47 members of 8 JP families, and were able to exclude linkage of the JP gene by multipoint lod scores less than −2.0 over the entire putative JP1 interval. Howe et al. genotyped 43 members (13 affected, 24 at-risk, 6 spouses) of the Iowa kindred using 5 microsatellite markers from this region and also found no evidence for linkage.26
In the latter study, markers were also examined from the regions of several genes known to play an important role in either the development of sporadic colorectal cancer or hamartomatous polyps. These included MSH2 (2p16), MLH1 (3p21), MCC, APC (5q21-22), HMPS (6q21), JP1, PTEN (10q22-24), KRAS2 (12p12), TP53 (17p13), DCC (18q21), and LKB1 (19p13), as well as CDKN2 (9q21). No evidence for linkage was found with markers from most of these regions, but linkage was detected with markers from 18q21. Genotyping and linkage analysis with 27 microsatellite markers from 18q21 resulted in lod scores >3.0 with 7 different markers, with a maximum lod score of 5.00 (at θ = 0.001) with the marker D18S1099. Haplotype analysis revealed five affected individuals with recombination events, allowing for localization of the JP gene to an 11.9-cM interval between markers D18S1118 and D18S487 on 18q21.1.28 This interval was known to contain two tumor-suppressor genes, DCC (deleted in colorectal carcinoma96), and DPC4 (deleted in pancreatic cancer 4, also known as Smad4 97), and was also known to be commonly deleted in both sporadic colorectal96 and pancreatic carcinomas.98 This is the only locus for JP identified by linkage thus far, but efforts continue to identify additional loci for this apparently genetically heterogeneous syndrome.
The PTEN gene on 10q23 is a tumor-suppressor gene that is frequently mutated in glioblastomas, prostate cancers, and breast cancers, with homology to tyrosine phosphatases and the protein tensin.99,100 It has also been shown to be the predisposing locus for both CS39 and BRR40 (see Chap. 45 for more detail on the PTEN gene). There have been two reports describing PTEN mutations in JP patients.27,101 Lynch and colleagues reported a R334X germ line mutation in two members of a family believed to have both CS and JP. The affected father had a history of small intestinal cancer, skin lesions, and macrocephaly, while his affected son also had skin lesions, macrocephaly, and small intestinal and colonic polyps. Eng and Ji have suggested that these patients might be more accurately classified as having CS.102 Olschwang and associates reported PTEN mutations in 3 of 14 patients presenting with GI bleeding and >10 juvenile polyps. One patient had a deletion at codon 232 leading to a premature stop at codon 255, was 74 years of age with upper and lower GI juvenile polyps, had a laryngeal cancer treated by radiotherapy 2 years earlier, and a heterogeneous thyroid nodule. Another patient, 10 years of age with generalized JP (with no extraintestinal manifestations of CS), was found to have a M35R substitution, while a third patient was 14 years old with colonic JP (and no family history of CS or BRR) with a mutation in a splice donor site of exon 6.27 Eng and Ji believed the first patient was suggestive of having CS, and made the point that CS could not be ruled out in the other two patients (aged 10 and 14 years) because the penetrance is <10 percent below the age of 15 years. These authors cautioned that the diagnosis of JP should be made after exclusion of other syndromes such as CS and BRR, and if PTEN mutations were found in these patients, then a high index of suspicion for the diagnosis of CS or BRR should be maintained.102
This recommendation was also influenced by two studies in which PTEN mutations were not found in JP patients. Riggins and associates sequenced the PTEN gene in 11 patients with familial JP and found no mutations.103 Marsh and coworkers found no PTEN mutations in members of 14 JP families and 11 sporadic cases, and concluded that PTEN was either not a predisposing gene for JP or that it was only involved in a small group of cases. Furthermore, they raised the possibility that if there were a susceptibility locus on 10q22-24, as suggested by Jacoby, 25 then the 3 cM region as proposed for JP1 is 7 cM centromeric to the PTEN gene.104 In summary, mutations of PTEN or another nearby gene on 10q22-24 may be responsible for a subset of patients with JP, but it is important to rule out CS or BRR in these patients because approximately 80 percent of CS patients and 60 percent of BRR patients have PTEN mutations105 and may also have hamartomatous polyps. Patients with CS need to be followed closely for the development of breast and thyroid neoplasms, while JP patients do not.
Smad4 Mutations on 18q21.1
The finding of linkage to chromosome 18q21.1 in an interval containing two tumor-suppressor genes thought to play a role in the development of GI cancers led us to screen these genes for mutations in affected members of the Iowa JP kindred. After sequencing all 11 exons of Smad4 and 14 of 29 DCC exons, a 4 base-pair (bp) deletion was found in exon 9 of the Smad4 gene. All 13 affected members of this kindred had this deletion, as did 4 of 26 individuals at-risk, while 7 spouses and 242 control patients did not (Fig. 35-4). The maximum 2-point lod score of this deletion with the JP phenotype in this family was 5.79 at θ = 0.00. This deletion occurred between nucleotides 1372 and 1375 (codons 414 to 416) and resulted in a frameshift, creating a new stop codon at the end of exon 9 (codon 434).28
A Denaturing and B nondenaturing gels of Iowa JP kindred members showing the exon 9 PCR product. Affected individuals 4, 5, 6, and 11, as well as one at-risk (*), 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; reproduced from Howe et al.28 with permission).
To further study the importance of the Smad4 gene in JP, all exons were sequenced in eight other unrelated patients. The same 4 bp deletion in exon 9 was also found in affected members of JP families from Mississippi (originally described by Subramony et al.86) and Finland.28 Since this report, an additional Caucasian kindred from Texas has also been found with the same deletion. The sharing of this common deletion from exon 9 raises the possibility that these kindreds may either have a common ancestor or that this region is a mutational hotspot. Genotyping of members of all four families with several markers from 18q21 revealed that there was no shared haplotype between these families for markers closest to the Smad4 gene, suggesting that this area in exon 9 is indeed more susceptible to mutation.106
In the six other unrelated patients studied, four had the wild-type sequence for all exons of the Smad4 gene. One patient with sporadic generalized JP was found to have a 2 bp deletion at codon 348 of exon 8, which caused a frameshift and stop at codon 350. Another patient with sporadic JP diagnosed at age 6 with 30 to 40 colonic juvenile polyps had a 1 bp insertion between codons 229 and 231 of exon 5. This mutation, which added a guanine to six sequential guanines, caused a frameshift and stop at codon 235.28 A summary of the results of Smad4 sequencing in these patients is shown in Table 35-1.
Table 35-1: Smad4 Sequencing Results in 10 Unrelated JP Patients |Favorite Table|Download (.pdf) Table 35-1: Smad4 Sequencing Results in 10 Unrelated JP Patients
|Patient ||Type ||Codon (exon) ||Nucleotide Change ||Predicted Effect ||Control Pts. |
|1-13* ||Familial ||414–416 (9) ||4 bp deletion ||Frameshift, stop at codon 434 ||0/242 |
|M-1* ||Familial ||414–416 (9) ||4 bp deletion ||Frameshift, stop at codon 434 ||0/242 |
|T-1* ||Familial ||414–416 (9) ||4 bp deletion ||Frameshift, stop at codon 434 ||0/242 |
|JP 5/1* ||Familial ||414–416 (9) ||4 bp deletion ||Frameshift, stop at codon 434 ||0/242 |
|JP 11/1 ||Sporadic ||348 (8) ||2 bp deletion ||Frameshift, stop at codon 350 ||0/101 |
|JP 10/1 ||Sporadic ||229–231 (5) ||1 bp insertion ||Frameshift, stop at codon 235 ||0/101 |
|JP 6/1 ||Sporadic ||— ||— ||— || |
|JP 4/1 ||Familial ||— ||— ||— || |
|JP 1/1 ||Sporadic‡ ||— ||— || |
|JP 2/13† ||Familial ||— ||— ||— || |
Fraction of Cases due to Smad4 and PTEN
There is little doubt that there is genetic heterogeneity in JP patients, but the degree of this heterogeneity and the number of genes involved remain to be defined. As discussed above, we have found Smad4 mutations in 60 percent of the familial and sporadic cases thus far examined. However, no evidence for linkage to 18q21 markers has been seen in other JP families, and gel shifts were observed in only 1 of 20 unrelated individuals studied by conformation-sensitive gel electrophoresis of the Smad4 gene.107 It would, therefore, appear that the frequency of Smad4 mutations in JP patients might range from as low as 23 percent to as high as 60 percent. The role of PTEN mutations in JP remains to be clarified. Riggins et al.103 and Marsh et al.104 found no mutations in 36 sporadic and familial cases, while Olschwang and coworkers found 3 of 14 (21 percent) of their patients had germ line PTEN mutations.
The Smad4 Gene and Human Tumors
Hahn and associates identified a new tumor-suppressor locus by virtue of the fact that approximately 90 percent of pancreatic cancers had deletions of chromosome 18q and 30 percent had homozygous losses within a common interval on 18q21.1, which did not include the DCC gene.98 They found three expressed sequences from this region, one of which showed significant homology to the Drosophila Mad (mothers against decapentaplegic) gene. This gene was named DPC4 based on its being the fourth deletion locus described in pancreatic cancers, and it encodes for a 552 amino acid protein. Somatic mutations of this gene were found in 6 of 27 pancreatic cancers without homozygous deletions on 18q21, implicating it as a tumor suppressor predisposing to pancreatic cancer.97 DPC4 is now referred to as Smad4, a nomenclature that combines the terms for the homologous Mad genes in Drosophila melanogaster and sma genes in Caenorhabditis elegans.108
Subsequent studies have shown that the rates of Smad4 mutations are modest in other GI tumors, and are distinctly uncommon in extraintestinal tumors. Thiagalingam and coworkers found loss or mutation of Smad4 in 5 of 18 colorectal carcinoma cell lines exhibiting 18q loss. These tumors were taken from a panel of 55 tumors with 18q loss (out of a total of 100 colorectal tumors), suggesting an overall mutation rate of approximately 15 percent.109 Tagaki and associates found 5 Smad4 mutations in 31 primary colorectal cancers, for a total mutation rate of 16 percent.110 MacGrogan and colleagues found Smad4 mutations in 4 of 21 (19 percent) primary colorectal carcinomas and cell lines.111 Hoque and coworkers found LOH on 18q21 and a Smad4 mutation in one of six (17 percent) colitis-associated colorectal cancers.112 Lei and collaborators found no mutations of Smad4 in 10 gastric, 10 esophageal, and 10 colitis-associated colorectal cancers.113 Powell and coworkers found one case with inactivation of both copies of Smad4 (one deleted and the other with nonsense mutation at codon 334) in a panel of 35 gastric adenocarcinomas.114 Hahn and colleagues reported 5 mutations in 32 (16 percent) carcinomas of the biliary tract, even though they only examined exons 8 through 11.115 Moskaluk and colleagues sequenced the Smad4 gene from members of 11 families with familial pancreatic cancer (defined as 2 affected first-degree relatives) and found no mutations in these patients. They speculated that this situation could be similar to that seen in Li-Fraumeni syndrome, in which affected family members have germ line mutations of the p53 gene but do not develop colorectal cancers, while sporadic colorectal cancers have a high rate of p53 mutation.116
Schutte and associates analyzed 64 non-GI tumors (11 prostate, 8 breast, 8 ovary, 7 bladder, 6 hepatocellular carcinoma, 6 lung cancers, 5 head and neck carcinomas, 4 melanomas, 3 osteosarcomas, 3 renal cell carcinomas, 2 glioblastomas, and 1 medulloblastoma) displaying 18q loss for sequence changes in the Smad4 gene, and found only 2 alterations, 1 in a breast and the other in an ovarian cancer. They concluded that another tumor-suppressor gene from 18q might be involved in the development of these tumors.117 Kim and coworkers found 2 mutations in head and neck tumors derived from cell lines of 11 patients and 20 primary tumors (6 percent).118 MacGrogan and collaborators found no mutations in 45 primary and metastatic prostate cancers.111
Function of the Smad4 Gene
The Smad4 protein is the common mediator involved in the transforming growth factor-β (TGF-β), activin, and bone morphogenetic protein (BMP) signal-transduction pathways. Members of the TGF-β superfamily initiate a wide spectrum of effects on a variety of cell types, including cell differentiation, proliferation, and apoptosis.119,120 Currently, there are eight known Smad genes in vertebrates. The Smad2 and Smad3 proteins function as cytoplasmic effectors in the TGF-β and activin pathways. Their counterparts in the BMP pathway are Smad1, Smad5, and possibly Smad9. Smad6 and Smad7 function as inhibitors of all three pathways by binding to type I receptors and interfering with phosphorylation.120
An overview of the sequence of proteins involved in the TGF-β signaling pathway is shown in Fig. 35-5. TGF-β binds to plasma membrane serine/threonine kinases, and, specifically, to the type II TGF-β receptor (TβR-II). This then complexes with the type I receptor (TβR-I), causing phosphorylation in a serine- and threonine-rich domain of TβR-I.121 These activated type I receptors phosphorylate cytoplasmic monomers of Smad2 or Smad3, allowing these to form oligomers and to associate with Smad4 monomers or oligomers.122,123 Hetero-oligomers of Smad2 or Smad3 and Smad4 then migrate to the nucleus and regulate transcription in conjunction with DNA-binding proteins.124
Overview of the TGF-β signaling pathway. TGF-β binds to TβR-II, which then phosphorylates TβR-I, thereby activating it. TβR-I then phosphorylates Smad2 or Smad3, allowing them to form homo-oligomers or hetero-oligomers with Smad4. Hetero-oligomeric complexes of Smad4 with Smad2 or Smad3 associate with DNA-binding proteins. These complexes then bind to sequences in the promoter regions of genes under TGF-β control, regulating their transcription.
The mechanism by which this transcriptional regulation occurs is just beginning to be understood. Within the nucleus, the Smad4 protein appears to bind directly to DNA, 125 and it has been shown that both Smad3 and Smad4 efficiently bind the 8-bp nucleotide sequence GTCTAGAC.126 Another sequence that binds Smad3 and Smad4 (AG(C/A)CAGACA, dubbed the “CAGA box”) has been described at positions −730, −580, and −280 within the plasminogen activator inhibitor-1 promoter, which is strongly inducible by TGF-β. Experiments cloning CAGA boxes upstream of various promoters resulted in markedly increased responses to TGF-β, and mutations within these sites decreased these responses.127 The recently identified human DNA-binding protein hFAST-1 also appears to bind to specific DNA sequences in response to TGF-β in the presence of Smad2 and Smad4.128 In Xenopus, it has been proposed that FAST-1 binds to the C-terminus region of Smad2, and Smad4 stabilizes these proteins by binding to Smad2.129 This complex appears to be required for transcription of TGF-β and activin target genes, which is achieved by hFAST-1 and Smad4 associating with their sequence-specific binding elements.128
The majority of Smad4 mutations thus far described in human cancers have been between codons 330 and 526, 97,110,117,118 within several highly conserved domains. This C-terminus of Smad4 is important in the formation of Smad4 homo-oligomers (initially believed to be homo-trimers), which then form hetero-oligomers with other Smad proteins.122 Recently, it has been suggested that the majority of intracellular Smad4 is found in the form of monomers rather than homo-trimers, and that Smad4 competes with Smad2 and Smad3 in the formation of hetero-trimer complexes.123 It has been demonstrated that mutations which disrupt Smad4 homo- and hetero-oligomerization lead to loss of TGF-β superfamily induced signaling pathways.130 Lagna and associates showed that loss of the terminal 38 amino acids of Smad4 leads to a dominant negative effect on the induction of mesoderm in Xenopus embryos by Smad2, and that mutant and wild-type proteins form oligomers which may be responsible for their loss of activity.131
Zhou and colleagues constructed heterozygous and homozygous Smad4 deletions by homologous recombination in the colorectal cancer cell line HCT116, which has a truncating mutation of TβR-II and requires restoration of this gene in order to mediate TGF-β signaling. When TβR-II was reintroduced into these cells, cell lines with heterozygous Smad4 deletions generated an equal response to TGF-β as cell lines without deletion, while those having homozygous deletions had no signaling. Similar results were observed when an activated TβR-I receptor (TβR-IT204D) was introduced into these cells, or when they were stimulated by the addition of activin. Furthermore, the proliferation of cell lines in which a functional TβR-II gene had been introduced was substantially decreased in the parental cell line, and in those with heterozygous Smad4 deletions, when placed in medium containing TGF-β. Similar cell lines with homozygous Smad4 deletions had a significantly lower level of growth inhibition. These results indicated that Smad4 mutation was a potential mechanism by which tumor cells could escape the antiproliferative effects of TGF-β.132
Generation of transgenic mice with heterozygous and homozygous inactivation of Smad4 has added insight into the role of this gene in embryogenesis and tumor formation. Sirard and colleagues created Smad4 mutant embryonic stem cells by homologous recombination, using a construct replacing exons 8 and 9 with a neomycin resistance gene. They followed 25 Smad4 heterozygous mutant (+/−) mice for 11 months, and found no increase in tumors relative to wild-type mice, suggesting that loss of the other allele was necessary for tumor development. Smad4 homozygous mutant (−/−) mice died in utero, predominantly at 7.5 days of embryogenesis. These embryos manifested impaired growth, with poor separation of embryonic and extraembryonic boundaries. Histologic comparison of wild-type and −/− embryos at 6.5 days revealed that −/− embryos had no mesoderm formation, as well as abnormal development of the visceral endoderm. This defect in gastrula formation could be rescued by aggregation with tetraploid embryonic cells, but the resulting embryos had impaired development of anterior structures.133 Yang and coworkers confirmed these findings using a similar model, and concluded that Smad4 is necessary for epiblast proliferation, egg-cylinder formation, and induction of the mesoderm.134
Takaku and associates created transgenic mice by homologous recombination using a construct disrupting the Smad4 gene within exon 1, and generated compound Smad4/Apc mutant mice, taking advantage of the fact that these genes are approximately 30 cM apart on mouse chromosome 18. Smad4 −/− mutant mice died in utero, while Smad4 +/− mutants were viable, fertile, and had no histologic abnormalities within their intestines or pancreas relative to wild-type litter mates. Compound ApcΔ716 and Smad4 heterozygous mutant mice had only about 12 percent of the number of intestinal polyps seen in ApcΔ716 +/− mice, but their average size was larger (1 to 2 mm vs. 0.5 mm). Fifty-three percent also had epidermoid cysts, and 20 percent developed adenocarcinoma of the ampulla of Vater, which was not seen in Apc or Smad4 +/− mice. Intestinal polyps from the compound mutant mice also showed stromal proliferation similar to that observed in juvenile polyps, and an increased incidence of malignant tumors relative to ApcΔ716 +/− mice. Examination of tumor DNA from compound heterozygous mutant mice revealed that adenocarcinomas had lost the wild-type chromosome 18 and reduplicated the compound mutant chromosome.135 Together, these studies demonstrate that Smad4 most likely functions as a tumor-suppressor gene, for heterozygous mutants did not display increased numbers of tumors and were phenotypically normal. Furthermore, formation of tumors may require somatic mutations in addition to loss of Smad4, as seen in polyps of cis-compound Smad4/APC mutants, which had histologic changes reminiscent of those seen in JP patients.
Mutations of Other Genes in the TGF-Β Pathway
Mutations within the TβR-II gene have been demonstrated in 90 percent of tumors with microsatellite instability, occurring within a 10-bp polyadenine tract of this gene.136-138 Restoration of functional TβR-II expression in deficient cell lines reduces their tumorigenicity in cell cultures and nude mice.139 There has been one report of a germ line mutation at codon 315 (T315M) of TβR-II in an hereditary colorectal cancer family without microsatellite instability, which suggests that mutations outside the polyadenine tract in this gene may be another mechanism for the development of colorectal cancer, independent of defects in DNA mismatch-repair.140
Smad2 has been mapped to chromosome 18q21.1 approximately 3 Mb centromeric to the Smad4 gene, appears to be ubiquitously expressed, 141 and consists of 11 exons and 467 amino acids.142 It may play a role in the development of sporadic colorectal cancers, but mutations within Smad2 appear to be infrequent in these tumors. Analysis of Smad2 in 18 colorectal cancer cell lines revealed homozygous loss in one tumor, and a truncated protein in another.143 Of 66 sporadic colorectal carcinomas, 4 (6 percent) were found with missense mutations in Smad2. Adjacent normal mucosa from these patients did not have these mutations, indicating that these events were somatic within the tumors. No germ line mutations of Smad2 were identified in a panel of 15 patients with a strong family history of colorectal cancer or early age of onset.142
Smad3 has been mapped to chromosome 15q21-22, 143 and consists of 9 exons and 424 amino acids.144 Examination of 167 cancer cell lines (70 colorectal, 22 breast, 22 brain, 15 lung, 12 pancreas, 8 head and neck, 6 ovary, 4 esophagus, 4 stomach, and 4 prostate) by in vitro synthesized protein assays for Smad3, found no truncated proteins in any of these tumor lines.145 Study of the Smad3 gene in 35 sporadic and 15 HNPCC colorectal cancers revealed no mutations, but 2 of 17 informative tumors showed LOH from this region.144 These studies suggest that both Smad2 and Smad3 may not significantly contribute to colorectal carcinogenesis, but do not address the question as to whether alterations in these genes contribute to the development of hamartomatous polyps.
Currently, little is known about Smad6 and Smad7 except that they are inhibitory through their binding to type I receptors, thus interfering with phosphorylation of downstream Smad proteins. They both have high homology to other Smad genes in their C-terminus, but appear to lack conserved regions in their amino-terminal domains common to other Smad genes.120,145 Smad6 encodes a 235 amino acid protein, and has been localized to chromosome 15q21-22, the same region as the Smad3 gene.143,145 Smad7 has been shown to inhibit TGF-β signaling by binding to the type I receptor and preventing the phosphorylation of Smad2. A truncated Smad7 protein demonstrated loss of this inhibition.146 Although it has not yet been demonstrated, it is conceivable that other mutations could lead to stable binding with type I receptors and sustained inhibition of the antiproliferative effects of TGF-β.