Colon cancer, the second leading cause of cancer death among adult Americans, accounts for 150,000 new cases and 60,000 deaths yearly. Our laboratory has identified several key pathways that are genetically inactivated in inherited and sporadic forms of human colorectal cancer.
Epigenetic Inactivation of DNA Repair Genes Is a Cause of Colon Cancer
Germline mutations in genes encoding components of the DNA mismatch repair system give rise to kindreds having hereditary nonpolyposis colon cancer (HNPCC). In collaboration with Bert Vogelstein's laboratory (HHMI, Johns Hopkins University), we observed that inactivation of DNA mismatch repair is also present in 15 percent of nonfamilial sporadic colon cancers. Unexpectedly, mutations of mismatch repair genes were not detected in these sporadic cancers that are functionally repair deficient. Rather, our laboratory found that in these cancers the MLH1 repair gene is silenced due to methylation of the MLH1 gene promoter at both the maternal and paternal MLH1 alleles. Pharmacological reversal of MLH1 methylation induces reexpression of the MLH1 gene. We hypothesize that cancers target MLH1 for methylation by a specific aberrant mechanism, which we hope to elucidate.
Repair-Deficient Cancers Mutationally Inactivate the TGFβ Receptor Tumor-Suppressor Gene
Perhaps as a primitive defense against cancer, the epithelial cells of the colon that are exposed to environmental mutagens are killed off on a weekly basis by apoptotic cell death and shedding. We found that transforming growth factor β (TGFβ), a hormone present in the gut, potently induces apoptosis in normal colon cells, but that colon cancer cells are resistant. In colon cancers deficient in mismatch repair, we demonstrated that TGFβ resistance is due to ubiquitous mutational inactivation of the type II receptor for recognizing TGFβ (RII).
The RII gene is particularly susceptible to mutation because it contains a unique coding region 10–base pair adenine repeat (encoding a lysine triplet). In the absence of effective mismatch repair, this sequence is destabilized and becomes a hot spot for frameshift mutations, generally on both alleles. These mutations encode truncated receptors that are inactive for TGFβ signaling. We find frequent RII mutations in repair-deficient colon and gastric cancers. Moreover, the timing of colonic RII mutations coincides with the transformation of benign, cancer precursor, colon adenomas to frankly invasive colon cancers. This suggests that TGFβ suppresses the progression of neoplastic cells to invasive cancers. Consistent with this hypothesis, transferring a normal RII gene back into colon cancer cells abrogates their ability to form tumors when injected into immune-deficient mice. Thus, structurally and functionally, RII acts as a tumor-suppressor gene.
Mutation of TGFβ Receptors in Repair-Proficient Colon Cancers
We hypothesized that if TGFβ mediates a colon cancer–suppressor pathway, this pathway should also be inactivated in colon cancers that are mismatch repair proficient. Indeed, by functional assay we find that 80 percent of these cancers have inactivated TGFβ signaling. In 15 percent of these cases this is again due to RII mutations that have targeted the RII kinase domain rather than the RII polyadenine repeat. In collaboration with the Vogelstein laboratory, we have demonstrated mutational inactivation in other repair-proficient cancers of TGFβ signaling via mutations in either SMAD2 or SMAD4 genes, which encode postreceptor elements of the TGFβ signaling pathway.
TGFβ Antagonizes the COX-2 Oncogene by "Aspirin-like" Activity
We have recently identified 15-prostaglandin dehydrogenase (15-PGDH) as a novel effector of TGFβ-mediated colon cancer suppression. 15-PGDH catalyzes the rate-limiting step in prostaglandin degradation, and thus, like aspirin, antagonizes the enzymatic activity of the COX-2 (cyclooxygenase 2) colon cancer oncogene. We find that 15-PGDH is a direct transcriptional target of the TGFβ pathway, is expressed by normal colonic epithelial cells, and is ubiquitously absent in colon cancers (which universally inactivate TGFβ signaling). Restoring wild-type 15-PGDH expression strongly suppresses the tumorigenic capacity of colon cancer cell lines. In contrast, knockout of 15-PGDH alleles from the mouse genome confers marked susceptibility to colon tumor development in the 15-PGDH–null murine model. We are examining a library of small molecules to determine if it is possible to use drugs that stimulate non-TGFβ-dependent pathways to reactivate 15-PGDH expression in colon cancers.
Novel Genes Targeted for Epigenetic Silencing in Colon Cancer
Our recent studies support the hypothesis that epigenetic silencing is a major mechanism of gene inactivation in human cancer. For example, individuals who inherit a germline mutant E-cadherin gene develop gastric cancers upon inactivation of their remaining normal E-cadherin copy. In half of these gastric cancers, epigenetic methylation provides the mechanism for inactivating the second E-cadherin allele. More recently, we have found that helicase-like transcription factor (HLTF) is a tumor-suppressor gene that is inactivated exclusively by epigenetic methylation found in nearly half of human colon cancers. This demonstrates a new connection of the DNA-helicase gene family to human cancer development.
Most recently, we have demonstrated that a novel gene, SLC5A8, is inactivated by epigenetic mechanisms in 60 percent of colon cancers, and found that SLC5A8 encodes a new transporter that mediates colonocyte uptake of butyrate from the colonic lumen. Butyrate is a potent differentiating agent and a major breakdown product of dietary fiber. SLC5A8 thus provides a novel molecular connection between diet and a colon cancer–suppressor pathway.
The Metastatic Colon Cancer Genome
To initiate a genetic dissection of the pathway that generates colon cancer metastasis, we have developed a resource composed of cancer metastasis samples from more than 30 colon cancer patients, cell lines established from these metastases, and tissue from the primary colon tumors matching these metastases. In collaboration with the Vogelstein laboratory, we have sequenced the complete "cancer genome" from these samples, identifying 140 "cancer genes" that each showed somatic mutations in two or more patient's tumors. All of the mutations we detected in these metastatic cancers were present in the primary colon cancer tumors that preceded the metastasis, suggesting that the genetic instructions enabling metastases preexisted prior to the actual metastatic spread of these tumors.
Novel Colon Cancer Susceptibility Locus
Our recent genetic studies have focused on a group of 300 individuals from families in which multiple persons have developed colon cancers during middle age. A whole-genome scan of these individuals has determined that in one-third of these families, colon cancer development is linked to an autosomal-dominant disease locus residing on a 15-Mb interval on chromosome 9q22.2–31.2. We are searching for the identity of this gene.
Molecular Strategies for Early Detection of Colon Cancer
To develop molecularly based assays for early detection of colon cancer, we have searched for marker sequences highly targeted for aberrant methylation in cancer. We defined one such sequence within exon-1 of the human vimentin gene. A methylation-specific PCR (polymerase chain reaction) assay was developed against this sequence and was shown to be sensitive enough to detect this aberrant methylation event in DNA extracted from the feces of colon cancer patients. In a clinical trial of 200 patients, this assay detected nearly half of all colon cancers, including half of curable early-stage cancers. Additionally, the test has proved able to detect half of all advanced colon adenomas, premalignant colon lesions that are the direct precursors of colon cancers. This noninvasive approach, which detects colon cancer by testing stool DNA, has been endorsed by the American Cancer Society and is now commercially available for patient use. We are working to define additional complementary markers that will further increase the sensitivity of this assay.