TGF-β Signaling Alterations and Colon Cancer
Colorectal cancer is the second most common cause of cancer-related death in the United States. Twin studies suggest that 35% of all colorectal cancer cases are inherited. High-penetrance tumor susceptibility genes account for at most 3–6% of all colorectal cancer cases and the remainder of the unexplained risk is likely due to a combination of low to moderate penetrance genes. Recent genome-wide association studies have identified several SNPs near genes belonging to the transforming growth factor beta (TGF-β) superfamily such as GREM1 and SMAD7. Together with the recent discovery that constitutively decreased TGFBR1 expression is a potent modifier of colorectal cancer risk, these findings strongly suggest that germline variants of the TGF-β superfamily may account for a sizeable proportion of colorectal cancer cases. The TGF-β superfamily signaling pathways mediate many different biological processes during embryonic development, and in adult organisms they play a role in tissue homeostasis. TGF-β has a central role in inhibiting cell proliferation and also modulates processes such as cell invasion, immune regulation, and microenvironment modification. Mutations in the TGF-β type II receptor (TGFBR2) are estimated to occur in approximately 30% of colorectal carcinomas. Mutations in SMAD4 and BMPR1A are found in patients with familial juvenile polyposis, an autosomal dominant condition associated with an increased risk of colorectal cancer. This chapter provides an overview of the genetic basis of colorectal cancer and discusses recent discoveries related to alterations in the TGF-β pathways and their role in the development of colorectal cancer.
Colorectal cancer is the fourth most common malignancy and the second most frequent cause of cancer-related death in the United States. In 2009, an estimated 146,970 cases of colorectal cancer were diagnosed and 49,960 people died from this disease . Institution of colonoscopy for at-risk individuals leads to earlier diagnosis of colon cancer which is amenable to curative surgery. Adjuvant therapy in patients with lymph node involvement has been demonstrated to have a benefit in overall survival . Patients with widespread disease at diagnosis or with recurrent disease are treated with chemotherapy agents, although surgery with curative intent also has a role in the treatment of some patients with metastatic disease. The use of antibodies against vascular endothelial growth factor and epidermal growth factor results in a small but significant increase in the survival of patients with metastatic or recurrent colon cancer . However, metastatic colorectal cancer is not a curable disease and therapy with current therapeutic agents is associated with significant morbidity.
The risk of developing colon cancer is approximately doubled in persons with a family history of colon cancer in a first-degree relative [4, 5, 6]. The risk increases with increasing number of first-degree relatives affected by colon cancer. Twin studies have suggested that genetic mutations contribute to the development of at least 35% of cases of CRC . Hence, it is reasonable to estimate that at least one third of colon cancer cases are attributable to genetic factors.
1 Genetics of Colorectal Cancer
The most well-known familial genetic syndromes predisposing to the development of colorectal cancer are familial adenomatous polyposis and Lynch syndrome, formally named hereditary non-polyposis colon cancer. Familial adenomatous polyposis (FAP) is an autosomal dominant disorder that affects 1 in 13,000 births . FAP is characterized by the formation of numerous polyps/adenomas throughout the large intestine in affected individuals, starting in their mid-twenties. The risk of these polyps/adenomas progressing to invasive carcinoma is 100%. Germline mutations in the adenomatous polyposis coli (APC) gene predispose individuals to develop numerous adenomatous polyps . In addition, they have an increased risk of developing desmoid tumors, thyroid cancer, gastric adenocarcinoma, duodenal adenocarcinoma, and/or ampullary carcinoma. A missense mutation in the APC gene known as I1307K (isoleucine changes to lysine) is associated with colon polyps and an increased risk (up to 1.5–2 times) of developing colon carcinoma. The mutation leads to a hypermutable region, thereby indirectly predisposing to cancer. Importantly, this allele is found in 6% of Ashkenazi Jews but at a very low level in the general population [10, 11].
Lynch syndrome, formerly named hereditary non-polyposis colon cancer (HNPCC), is an autosomal dominant condition characterized by early onset of colon cancer. The average age at diagnosis is 45 years, the tumors tend to develop in the proximal colon and show evidence of microsatellite instability (MSI) . Germline mutations in the DNA mismatch repair (MMR) enzymes predispose individuals to this syndrome. Deficiencies in these enzymes lead to numerous errors in DNA replication, especially in tandem repeat sequences, and cause lengthening of microsatellite sequences. Mutations in critical genes like BAX, TGFBR2, and E2F4 can then initiate or promote carcinogenesis [12, 13]. Patients have an 80% lifetime risk of developing colon cancer and an increased predilection to develop extra-intestinal tumors in the endometrium, ovary, stomach, small bowel, hepatobiliary tract, pancreas, upper uroepithelial tract, and brain .
Deficiency in the base excision repair gene MUTYH also predisposes to colon cancer . The MUTYH syndrome is inherited as a recessive trait and biallelic mutation carriers have almost a 100% risk of developing cancer. Variants in the MUTYH gene were identified in a family affected with multiple colorectal adenomas and carcinomas. Tumors from these individuals showed a predominance of somatic mutations in the APC gene. The majority of these APC mutations were G:C→A transversions, which suggests a defect in the base excision repair machinery . It has been proposed that monoallelic mutations in the MUTYH gene also confer an elevated risk of colorectal cancer, although this is controversial [16, 17, 18, 19, 20, 21, 22, 23]. A large population-based series of 9,628 patients with colorectal cancer and 5,064 controls were genotyped for MUTYH variants associated with colorectal cancer . Biallelic mutation status was associated with a 28-fold increase in colorectal cancer risk (95% CI, 17.66–44.06). Monoallelic mutation was not associated with an increased colorectal cancer risk. Cancers associated with MUTYH mutations are thought to progress through a MSI-independent pathway . It has not yet been fully determined as to why the MUTYH mutations predispose to the development of colorectal cancer .
The above-described mutations have a high penetrance with respect to colorectal cancer risk but, collectively, these various syndromes account for at most 3–6% of all colorectal cancers . The remaining fraction of familiar cancers and a majority of sporadic cancers are likely to be due to low-penetrance mutations, i.e., mutations that have low frequency of association with a specific phenotype . Genome-wide association studies have identified new genomic loci associated with colorectal cancer risk. A locus at 8q24 has been associated with a combined odds ratio of 1.17 (95% CI, 1.12–1.23; p = 3.16×10–11) [28, 29]. The association was confirmed in both sporadic and familial colorectal cancer. Single nucleotide polymorphisms (SNPs) near GREM1, SCG5 , and SMAD7  genes have also been found to be strongly associated with colorectal cancer risk. Other genomic loci associated with an increased risk of developing colorectal cancer have been identified at 18q21, 8q23.3, 10p14, 11q23, 14q22.2, 16q22.1, 19q13.1, and 20p12.3 [29, 32, 33].
Gene polymorphisms in specific signaling pathways have also been shown to modify the risk of colorectal cancer. Epidemiological studies have shown an association between colorectal cancer, obesity, and insulin resistance. Elevated circulating levels of C-peptide and insulin-like growth factor binding protein I (IGFBP1) are directly associated with colorectal cancer risk [34, 35, 36, 37, 38, 39, 40, 41, 42, 43]. Adiponectin, an endogenous insulin sensitizer, is a protein secreted by the adipose tissue. Adiponectin levels are decreased in patients with obesity and insulin resistance. A prospective clinical trial has demonstrated that men in the highest quintile of adiponectin levels have a decreased colorectal cancer risk when compared to men in the lowest quintile (relative risk, 0.42; 95% CI, 0.23–0.78) . The hypothesis that genetic polymorphisms in the adiponectin gene (ADIPOQ) and its type I receptor (ADIPOR1) may affect the risk of colorectal cancer was examined by testing for differences in single nucleotide polymorphisms of the respective genes. Genotyping of haplotype tagging SNPs of the ADIPOQ and ADIPOR genes in two case–control studies with a combined population of 640 patients and 857 controls showed that one ADIPOQ SNP (rs266729), tagging the 5′ end of the gene, is consistently associated with a decreased risk of colorectal cancer after adjustment for age, sex, race, and SNPs within the same gene (adjusted odds ratio, 0.73; 95% CI, 0.53–0.99) . An attempt at replicating these findings was recently conducted by Carvajal-Carmona et al. in two separate cohorts from the UK . The association of the ADIPOQ genomic region with colorectal cancer was studied using the Illumina Hap 300/370/550 arrays that genotype 82 markers covering 250 kb around the ADIPOQ gene, none of which includes the rs266729 SNP. This study did not find an association between any of these SNPs with colorectal cancer risk. However, the r2 value of the Illumina array SNP in strongest linkage disequilibrium with rs266729 was only 0.74. Furthermore, this SNP was located more than 7.7 kb upstream of rs266729. Thus, it is questionable that the Illumina array genotyping results truly excluded an association between rs266729 and colorectal cancer.
Adenomatous polyps have long been considered neoplastic lesions leading to the development of colorectal carcinoma. Another type of polyps, the hyperplastic (or serrated) polyps, have been regarded primarily as non-neoplastic polyps with no malignant potential of their own. However, several studies suggest that at least some serrated polyps may have malignant potential [46, 47, 48]. These serrated polyps named sessile serrated adenomas (SSA)  and dysplastic forms named serrated adenomas or SA  increase the likelihood of malignant transformation.
Genetic alterations observed in the sessile serrated adenomas and the serrated adenomas are different from those seen in the adenoma–carcinoma sequence [51, 52, 53]. For example, alterations in TP53 and APC and loss of heterozygosity are rare, whereas alterations in microsatellite sequences and hypermethylation of CpG islands are common. Sessile serrated adenomas are associated with mutations in BRAF and show high levels of CpG island methylation. These adenomas only rarely have KRAS mutations [54, 55, 56, 57]. Traditional serrated adenomas also show high levels of CpG island methylation but contain KRAS mutations more often than BRAF mutations [20, 57, 58]. Importantly, KRAS mutations and BRAF mutations have been found to be mutually exclusive [20, 59, 60].
Germline mutations in the TGF-β pathway are commonly found in patients diagnosed with familial juvenile polyposis (FJP), an autosomal dominant condition affecting 1 in 100,000 births . It is characterized by the presence of 10 or more juvenile polyps in the gastrointestinal tract. Patients have an increased risk of colon cancer, even though estimates of cancer incidence have varied in different studies . Mutations in the SMAD4 and BMPR1A genes have been identified in FJP patients and account for about half of FJP cases [62, 63, 64]. Additional mutations have been described in the endoglin gene (ENG), a co-receptor for TGF-β family receptors, but a causative role in FJP has not been conclusively proven . In addition to the germline alterations that confer an increased risk of colorectal cancer, alterations in the TGF-β pathway have been documented in a high percentage of sporadic colon carcinomas. These mutations have been documented in both carcinomas with microsatellite instability (MSI) and carcinomas with chromosomal instability . In this chapter, we will discuss the TGF-β signaling pathway alterations reported in colorectal cancer as well as our current understanding of the contribution of these alterations to colorectal carcinogenesis. A better understanding of this central pathway in colorectal carcinogenesis will be required to develop screening strategies and targeted therapies.
2 Overview of the TGF-β Pathway
TGF-β is a multifunctional cytokine with diverse effects on virtually all cell types and with key roles during embryonic development and tissue homeostasis . Members of the TGF-β superfamily ligands, such as TGF-β, activin, and BMP, transduce their signals through heterotetrameric complexes comprising two types of serine–threonine kinase receptors, the type 1 and type 2. Upon ligand binding, the type 2 receptor phosphorylates and activates the type 1 receptor, which in turn initiates downstream signaling by phosphorylating the receptor-regulated SMADs (R-SMADs). Specific ligands signal through a specific combination of type 2, type 1, and R-Smads . TGF-β binds to the TGF-β type 2 receptor (TGFBR2) and the TGF-β type 1 receptor (TGFBR1, formerly named TβRI or Alk5 for activin receptor-like kinase 5), although in endothelial cells it can also bind a complex comprising TGFBR2, ACVRL1, and TGFBR1 . The type I receptor dictates the specificity for the R-SMADs: TGFBR1, ACVR1B, and ACVR1C phosphorylate SMAD2 and SMAD3, whereas ACVRL1, ACVR1, BMPR1A, and BMPR1B phosphorylate SMAD1, SMAD5, and SMAD8. Once phosphorylated, these R-SMADs transduce the signal to the nucleus in cooperation with the common mediator SMAD, SMAD4, to transcriptionally activate or repress different targets genes . The SMAD4–R-SMAD complex has DNA binding capacity but association with additional DNA binding cofactors dictates which set of genes are transcriptionally regulated by this complex. The TGF-β superfamily pathways are also negatively regulated. The inhibitory SMADs, SMAD6 and SMAD7, bind the active receptor complexes and also recruit E3 ubiquitin ligase SMURF1/2 to the receptor complexes to degrade them [70, 71]. SMAD7 has also been shown to participate in a complex that dephosphorylates the active TGF-β receptor .
The TGF-β superfamily signaling pathways are involved in many different biological processes during embryonic development, and in adult organisms they play a role in tissue homeostasis . TGF-β has a role in inhibiting cell proliferation but also modulates processes such as cell invasion, immune regulation, and microenvironment modification. It is generally accepted that excessive production and/or activation of TGF-β by tumor cells can foster cancer progression by mechanisms that include an increase in tumor neoangiogenesis and extracellular matrix production, upregulation of proteases surrounding tumors, and inhibition of immune surveillance in the cancer host . They are also strongly implicated in cancer, since alterations of some specific and some common components of these different pathways have been identified in the majority of human tumors.
Two distinct types of genetic alterations have been identified: gain-of-functions in oncogenes that usually result in growth factor-independent cell proliferation and recessive loss-of-function mutations in tumor suppressors that allow evasion of growth inhibitory signals. The well-characterized growth inhibitory response of TGF-β , combined with the fact that up to 74% of colon cancer cell lines and 85% of lung cancer cell lines have become resistant to TGF-β antiproliferative effect [75, 76], led several groups to search for evidence of inactivation of components of the TGF-β pathway in human cancer.
Signaling alterations in the stromal compartment of tumors also have a pro-tumorigenic effect. It has been found that TGF-β secretion is abundant in many human cancers and the TGF-β-rich microenvironment is associated with poor prognosis, tumor vascularization, and metastasis . TGF-β plays an important role in the process of epithelial mesenchymal transition, myofibroblast generation, production of autocrine mitogens, and evasion of tumor immunity . The role of TGF-β signaling in the stromal compartment is of importance in processes important for carcinogenesis. Conditional knockout of Tgfbr2 (type 2 receptor) in mouse fibroblasts  led to hyperplasia in the adjacent epithelial tissue with subsequent progression to prostate intraepithelial neoplasia and gastric squamous cancer, respectively. These Tgfbr2-defective fibroblasts had increased levels of hepatocyte growth factor associated with increased activation of the hepatocyte growth factor receptor, Met in adjacent tissues. Disruption of the TGF-β pathway in fibroblasts leads to increased fibroblast proliferation and has been shown to promote mammary tumor metastasis in fibroblast-epithelial cell cotransplantation studies in mice .
Other crucial functions of TGF-β related to cancer development and progression are its ability to suppress immune and inflammatory responses. TGF-β acts as a central inhibitor of the multiple components of the native and the adaptive immune system. It also stimulates the generation of T-regulatory cells, which inhibit effector T-cell functions and IL-17 producing Th17 cells, which regulate NK cells and macrophages . These actions result in a context-dependent effect. Smad3 knockout mice develop colon cancers only after they are removed from a germ-free environment or infected with Helicobacter spp. . Conditional deletion of Smad4 in T cells has been associated with the development of colon carcinomas, and these lesions are heavily infiltrated with plasma cells . The loss of Smad4 expression results in skewed maturation toward a Th2 phenotype, with increased levels of cytokines including IL-4,-5,-6, and -13 in vivo and in vitro. Knockout mice produced through expression of Cre under control of the designed promoter went on to spontaneously develop carcinoma in the gastrointestinal tract. In addition, these mice also exhibit a high rate of oral squamous cell carcinoma . The chronic inflammation induced in these experimental systems by the loss of TGF-β favors tumorigenesis. On the other hand overexpression of TGF-β in certain tumors can lead to evasion from the immune system and have a pro-tumorigenic role. TGF-β also plays a role in epithelial mesenchymal transition (EMT) in human cancer . EMT is a well-coordinated process during embryonic development and a pathological feature in neoplasia and fibrosis . Cells undergoing EMT lose expression of E-cadherin and other components of epithelial junctions, produce a mesenchymal cell cytoskeleton, and acquire motility and invasive properties. It was first reported in mouse heart formation and palate fusion, in some mammary cell lines, and in mouse models of skin carcinogenesis that TGF-β is a potent inducer of EMT [83, 84]. TGF-β-induced EMT is observed in transformed epithelial progenitor cells with tumor propagating ability . EMT-like processes contribute to tumor invasion and dissemination owing to the cell junction free, motile phenotype they confer. Carcinoma cells with mesenchymal traits have been observed in the invasion front of carcinomas and may reflect a series of interconnected features: that carcinomas are propagated by transformed progenitor cells, that progenitor cells are competent to undergo EMT, that EMT is triggered at the invasion front, which ultimately augments the disseminative capacity of these cells [74, 85]. TGF-β promotes EMT by a combination of SMAD-dependent transcriptional events and SMAD-independent effects on cell junction complexes. SMAD-mediated expression of HMGA2 (high mobility group A2) induces expression of SNAIL, SLUG, and TWIST [86, 87]. Independent of SMAD activity, TGFBR2-mediated phosphorylation of PAR6 promotes the dissolution of cell junction complexes . In mouse tumors and cell lines, TGF-β-induced EMT is Smad-dependent and enhanced by Ras signaling . TGF-β also enhances cell motility by cooperating with ERBB2 signals, as observed in breast cancer cells overexpressing ERBB2 .
3 TGF-β Signaling Alterations in Colorectal Cancer
3.1 Alterations in TGFBR2
Mutations in TGFBR2 are the most common mechanism of loss of TGF-β signaling in colorectal cancer. It is estimated that approximately 30% of colorectal cancers harbor mutations in TGFBR2 [76, 90]. The TGFRB2 gene has a microsatellite sequence comprising an A(10) tract in exon 3 and GT(3) tracts in exons 5 and 7 called BAT-RII. These regions, especially the A(10) region, are prone to develop frameshift mutations in the presence of mutations in the DNA mismatch repair machinery. Almost 80–90% of colorectal tumors with microsatellite instability have mutations in TGFBR2 [91, 92]. Other poly(A) tracts of similar length are mutated in these tumors, but not as frequently as TGFBR2. It is commonly speculated that colorectal cancers acquire partial TGF-β resistance largely because of TGFBR2 genetic alterations. Interestingly, some colorectal cancer cell lines, which harbor homozygous mutations of TGFBR2, are growth-inhibited by TGF-β, which suggests that under certain circumstance, the cells can bypass TGFBR2 to retain TGF-β-mediated growth inhibition . Whether TGFBR2 mutations have a causative role in colorectal carcinogenesis or whether they arise because of the hypermutable phenotype observed in cells with defective mismatch repair machinery is still a topic of debate. Fifteen percents of colorectal cancer cell lines without any evidence of microsatellite instability also harbor mutations in TGFBR2 . The effect of Tgfbr2 loss in the intestinal epithelium in cancer formation was studied in a Tgfbr2 conditional knockout mouse model. Azoxymethane (AOM) was used to induce colon cancer. Adenoma and carcinoma formation were significantly increased and increased neoplastic proliferation was noted in the mice devoid of Tgfbr2 in the colonic epithelium ((4xat-132) Cre-Tgfbr2(flx/flx)) when compared with Tgfbr2(flx/flx) mice, which have intact Tgfbr2 in the colon epithelium. The increased proliferation suggested that loss of TGF-β-mediated growth inhibition contributes to carcinogenesis. The increased proliferation noted could be due to the failure to inactivate Cdk4 expression as Cdk4 expression is upregulated in MSI+ cancers . In addition, reconstitution of TGFBR2 expression in a colon cancer line with known microsatellite instability was associated with decreased proliferation and decreased Cdk4 expression and kinase activity .
Studies evaluating the effect of TGFBR2 mutations on the prognosis of patients with colorectal cancer have yielded conflicting results. The 5-year survival rate of patients with resected stage III colon cancer treated with adjuvant therapy was significantly higher in patients whose tumors exhibited microsatellite instability and TGFBR2 mutations (74%) when compared to patients whose tumors had microsatellite instability without evidence of TGFBR2 mutations (46%) . On the other hand, a population-based study evaluating the impact of TGFBR2 mutations on prognosis in MSI-positive tumors failed to reveal any significant difference in the age- and stage-adjusted risk of death associated with TGFBR2 mutations in unstable tumors (138 out of 174) when compared to unstable tumors without such mutations . However, another larger retrospective study suggested that TGFBR2 mutations are not associated with prognosis in patients with high-microsatellite instability (MSI-H) tumors .
4 TGFBR1Mutations and Polymorphisms in Colorectal Cancer
Mutations in TGFBR1 have been identified in human colorectal cancer cell lines but are uncommon . However, decreased TGFBR1 expression levels are frequently observed. In such cells, reconstitution of TGFBR1 expression has been shown to decrease tumorigenesis. TGFBR1*6A, a TGFBR1 polymorphism that consists of a deletion of three alanines within a nine-alanine repeat at the 3′ end of exon 1, results in an impairment of TGF-β-mediated anti-proliferative response and has been associated with increased cancer risk in several studies [99, 100, 101]. Liao et al.  recently published a meta-analysis of 32 studies including 13,662 cases and 14,147 controls. Overall, TGFBR1*6A was significantly associated with cancer risk in all genetic models (for allelic effect: OR = 1.11; 95% CI = 1.03–1.21; for 6A/6A vs. 9A/9A: OR = 1.30; 95% CI = 1.01–1.69; for 9A/6A vs. 9A/9A: OR = 1.08; 95% CI = 1.01–1.15; for dominant model: OR = 1.08; 95% CI = 1.02–1.15; for recessive model: OR = 1.29; 95% CI = 1.00–1.68). Genotyping of germline and tumor DNA has shown that TGFBR1*6A is somatically acquired in approximately 2% of primary colon and head and neck tumors . Exogenous TGF-β increases thymidine incorporation in breast cancer cells stably transfected with this variant and in colon cancer cells that endogenously harbor this allele , suggesting that TGFBR1*6A has oncogenic properties in established tumor cells. To determine the role of TGFBR1*6A in the tumor microenvironment, we microdissected tumors cells, stromal cell, and histologically “normal” epithelial cells adjacent to the tumor from individual with head and neck cancer and evidence of TGFBR1*6A somatic acquisition within the tumor tissue . In head and neck cancer we found that the TGFBR1*6A allele was present in the tumor, immediately juxtaposed “normal” squamous epithelium and stroma as well as in adjacent true vocal cord epithelium and stroma. In colon cancer we found that the TGFBR1*6A allele had been somatically acquired by stromal cells up to 2 cm away from the tumor’s edge. Importantly, we found higher TGFBR1*6A/TGFBR1 allelic ratios in tumor tissues compared with stromal and epithelial tissues . Hence, the amount of somatically acquired TGFBR1*6A allele in normal epithelial and stromal cells surrounding the tumor appears to be inversely proportional to the distance from the primary tumor, suggestive of tumor-centered centrifugal growth . This provides strong support for the concept that TGFBR1*6A somatic acquisition is a critical event in the early stages of cancer development that is associated with field cancerization . However, TGFBR1*6A is not a bona fide oncogene when transfected into NIH 3T3cells. Rather, its decreased TGF-β signaling capabilities result in reduced oncogenesis when compared with wild-type TGFBR1 . To test the hypothesis that constitutively decreased TGFBR1 signaling contributes to colorectal cancer development, we generated a novel mouse model of Tgfbr1 haploinsufficiency . We found that Tgfbr1 haploinsufficient mice crossed with mice carrying a mutation in the Apc tumor-suppressor gene develop two to three times more intestinal tumors than wild-type littermates. Importantly, invasive adenocarcinoma with features of human colon cancer is only identified among ApcMin/+;Tgfbr1+/– mice, not among ApcMin/+;Tgfbr1+/+ mice . These findings led us to study whether constitutively decreased TGFBR1 expression is associated with human cancer. We recently reported that constitutively decreased TGFRB1 expression is an inherited trait associated with significantly increased colorectal cancer risk . We also found that somatically acquired mutations of the TGFBR1 gene were significantly more common in the tumors of patients with constitutively decreased TGFRB1 expression (11.5%) than in the tumors of patients without constitutively decreased TGFRB1 expression (0%) . The mechanism for the constitutively decreased expression is currently under investigation.
5 SMADMutations in Colorectal Cancer
Alterations in the genes encoding proteins playing a role in the downstream pathways of TGF-β signaling have been associated with a variety of cancers. SMAD2 and SMAD4 both map to chromosome 18q, a region commonly deleted in colon adenocarcinomas . SMAD4 also known as DCC is mutated in 16–38% of colorectal tumors [108, 109, 110, 111]. SMAD2 also located on 18q21 is lost in 6% of sporadic colon cancers . SMAD2 and SMAD4 gene inactivation occurs by deletion of entire chromosomal segments, small deletions, frameshift, nonsense, and missense mutations . As mentioned earlier, germline mutations in SMAD4 have been noted in several juvenile polyposis families with an increased predisposition to colorectal cancer. Mice studies have shed more light into the role of Smad4 in carcinogenesis supporting its role as a tumor suppressor. Homozygous loss of Smad4 leads to death of mice in utero, but heterozygous mice are viable [113, 114, 115]. These mice develop gastric polyps which evolve into cancers at a late age. However, Smad4+/– mice do develop colorectal tumors but only in the context of a primed, Apc-defective genetic background [116, 117]. Smad4 deletion in the intestinal epithelium does not lead to tumor formation in the mice but a deletion in the T-cell compartment leads to the formation of numerous gastrointestinal tumors with infiltration by plasma cells . These data suggest that Smad4 plays an important and complex role in the interaction between the immune system, stroma and the epithelium, a disruption of which contributes to colorectal carcinogenesis. Clinically, the loss of SMAD4 is associated with late-stage colon cancer and metastatic disease [118, 119]. Low levels of SMAD4 protein or mRNA in the tumor are also predictive of a poor response to chemotherapy and significantly shorter survival when compared to patients with tumors expressing high levels of SMAD4 [120, 121].
SMAD3 mutations have been thought to be infrequent in cancers. Mutational analysis of 11 colorectal cancer cell lines revealed a novel missense mutation in SMAD3 (R273H) in the SNU-769A cell line. This mutation led to inhibition of the translocation of SMAD3 to the nucleus and decrease in the activity of SMAD3 during TGF-β-induced transcriptional activation . Genome-wide analysis of protein coding genes in breast and colorectal cancers revealed that SMAD3 is mutated at a significantly higher frequency than the background mutation rate in these tumors . Smad3 mutant mice are viable and fertile but develop colorectal adenocarcinomas between 4 and 6 months of age . These mice also enhance intestinal tumorigenesis with an increase in multiplicity and rapid onset of invasive adenocarcinomas when crossed with ApcMin/+ mice . However, two other Smad3 mutant mice generated independently did not reveal a higher incidence of colorectal malignancies but exhibited functional defects in the immune system [125, 126]. A potential explanation for this discrepancy may be related to the interaction of the immune system with the environment. When Smad3–/– mice are maintained in H. pylori-free environment, they do not develop colon cancer for up to 9 months of age. But infection of these mice with Helicobacter spp. leads to development of colon cancer in 55–60% of animals . When Smad3-deficient mice are crossed with mice deficient in both B and T lymphocytes (Rag2–/–), the progeny have a higher incidence of Helicobacter-induced diffuse inflammation, and adenocarcinoma of the colon when compared with Helicobacter-infected Smad3–/– or Rag2–/– mice. In addition, adoptive transfer of wild-type T-regulatory cells provided significant protection against colorectal cancer in the double knockout mice . This suggests that loss of Smad3 may contribute to colon cancer development by a combination of altered T-regulatory cell function, increased pro-inflammatory cytokines, and anti-apoptotic proteins leading to increased proliferation in colonic tissues.
6 Bone Morphogenetic Protein Pathway
Bone morphogenetic proteins are members of the TGF-β superfamily of proteins. The signaling cascade is similar to that described in the TGF-β pathway, involving the activation of the type 2 receptor by the ligand. The activated type 2 receptor phosphorylates the type 1 receptor and ultimately leads to the release of R-SMADs, which complex with SMAD4 and modulates target gene expression. The R-SMADs of the BMP pathway are SMAD1, SMAD5, and SMAD8. Activation of the BMP pathway can be assayed by using antibodies specific to phosphorylated forms of SMAD1, SMAD5, and SMAD8. BMP signaling inhibits intestinal stem cell renewal through suppression of the Wnt-β-catenin pathway . BMP signaling is also required for full maturation of secretory cell lineages, in the small intestine in vivo and may have a role in apoptosis of mature colonic epithelial cells . As mentioned earlier in the chapter, germline mutations in the BMPR1A gene have been described in patients with juvenile polyposis syndrome. A role for the alteration of the BMP pathway in sporadic colorectal cancer is emerging. At the ligand level, BMP2, BMP3, and BMP7 have been found to be growth suppressive . Downregulation of BMP3 was observed in 90% of colorectal cancer samples, in association with aberrant hypermethylation in the tumors and highly correlated with microsatellite instability. Approximately 76% of adenomas also exhibited downregulation of the promoter suggesting that silencing of BMP3 may be an early event in the progression of colorectal carcinogenesis via the serrated and the traditional pathways . However, BMP7 was noted to be overexpressed at the mRNA and protein level in colonic tumor tissue when compared to normal tissue. Overexpression of BMP7 was associated with liver metastasis and poor prognosis . Similarly, overexpression of BMP4, assessed by real-time RT-PCR and immunohistochemistry, was noted in late-stage adenocarcinomas and in tumors with liver metastasis when compared to normal tissue . Interestingly, genome-wide association studies have revealed that SNP rs4444235 which is 9.4 kb from the transcription start site of BMP4 predisposes to colorectal cancer (odds ratio 1.11, 95% CI 1.08–1.15, p = 8.1×10–10) . This association was significantly stronger in cases with microsatellite stable tumors compared with microsatellite unstable tumors.
It is possible that the BMP pathway may be inactivated during the transition from adenoma to carcinoma as almost 90% of adenomas have evidence of a functioning BMP pathway and loss of the pathway correlates with progression of adenoma to carcinoma . The BMP pathway, assessed by nuclear staining of pSMAD1/5/8 expression, is inactivated in up to 70% of sporadic colorectal cancers. The BMP receptor (BMPR2) expression is impaired in a majority of microsatellite unstable cancer cell lines. In addition, BMPR2 expression was significantly more frequently impaired in microsatellite unstable tumors than microsatellite stable tumors recapitulating the phenomenon seen with the type 2 TGF-β receptor (TGFBR2) .
7 SMAD Antagonists
SMAD6 and SMAD7 are inhibitory SMADs that negatively control TGF-β signaling in response to feedback loops and antagonistic signals . SMAD6 competes with SMAD4 for binding to receptor-activated SMAD1, and SMAD7 recruits SMURF to TGF-β and BMP receptors for inactivation. Overexpression of SMAD7 and suppression of TGF-β signaling has been reported in endometrial carcinomas and thyroid follicular tumors [136, 137]. Interestingly, a recent genome-wide association study has shown that common alleles of SMAD7 that lead to decreased SMAD7 mRNA expression are associated with colorectal cancer risk . SMAD function is also directly inhibited by transcriptional repressors such as SKI and SNON (SKI-like). Deletions as well as amplification of SKI and SKIL have been reported in colorectal and esophageal cancers, raising the possibility that these genes act as oncogenes or tumor-suppressor genes depending on the context .
8 Future Directions
The recent exciting discoveries from genome-wide association studies in colorectal cancer have unearthed alterations at multiple levels in the TGF-β pathway, including BMP4, SMAD4, and SMAD7. At first glance, it seems that the risk of colorectal cancer with the inherited genomic loci is only slightly increased with an odds ratio less than 1.5. But, germline allele-specific expression in TGFBR1  has been found to confer a substantially increased risk of colorectal cancer (odds ratio 8.7) even by conservative estimates. The findings need to be confirmed in larger studies and in different populations. Unlike most other human malignancies, we can screen for colorectal cancer effectively by fecal occult blood testing or colonoscopy. However, screening the entire population to identify early colon cancer is not a practical approach in terms of expense and availability of health-care workforce, and this practice may not benefit a large majority of the population. The immediate clinical application of the identification of high-risk genomic loci and allele-specific expression is that they can help us identify a group of individuals who are at a higher risk of developing colon cancer. Institution of thorough screening in such high-risk groups by screening individuals at an earlier age and/or more frequently than the general population may be a more effective approach.
The multiple pro-tumorigenic effects of the TGF-β pathway, enabling tumors to evade host immunity, facilitating invasion and metastasis make it a primary target for therapeutic interventions. However, the potential benefits of such a strategy have to be weighed against the potential complications associated with the inhibition of a pathway which has important roles in the maintenance of tissue homeostasis. A better understanding of this complex pathway with a focus on delineating the pro-tumorigenic effects and mechanisms in specific tumor types and at different phases of carcinogenesis and cancer progression is essential.
This work is supported by grants R01 CA108741, R01 CA112520, R01 137000, and P60 AR048098 from NIH.
- 54.O‘Brien MJ, Yang S, Mack C et al (2006) Comparison of microsatellite instability, CpG island methylation phenotype, BRAF and KRAS status in serrated polyps and traditional adenomas indicates separate pathways to distinct colorectal carcinoma end points. Am J Surg Pathol 30:1491–1501PubMedCrossRefGoogle Scholar
- 96.Samowitz WS, Curtin K, Leppert MF, Slattery ML (2002) The prognostic implications of BAX and TGF[BETA]RII mutations in colon cancers with microsatellite instability. Mod Pathol 15:143AGoogle Scholar
- 102.Liao RY, Mao C, Qiu LX, Ding H, Chen Q, Pan HF (2009) TGFBR1*6A/9A polymorphism and cancer risk: a meta-analysis of 13,662 cases and 14,147 controls. Mol Biol Rep (published online November 1, 2009)Google Scholar