Search

Cyclooxygenase-2 (COX-2), the inducible COX isozyme, plays a key role in intestinal tumorigenesis. We have demonstrated recently that COX-2 protein is induced in the polyp stroma near the intestinal luminal surface in the Apc(Delta716) mouse, a model for human familial adenomatous polyposis, and stimulate tumor angiogenesis. However, the precise cell types that express COX-2 are still to be determined. By immunohistochemical analysis, here we show that the majority of COX-2-expressing cells in the intestinal polyps of Apc(Delta716) mice are fibroblasts and endothelial cells. Furthermore, the COX-2-expressing cells in human familial adenomatous polyposis polyps are also fibroblasts and endothelial cells. In contrast, bone marrow-derived cells such as macrophages and leukocytes express little COX-2 protein in the intestinal polyps. These results clearly indicate that fibroblasts and endothelial cells play important roles in polyp expansion by expressing COX-2, resulting in tumor angiogenesis.

The loss of heterozygosity (LOH) in human chromosome 18q21 is found at high frequencies in advanced pancreatic and colorectal cancers. Several candidate tumor suppressor genes, such as SMAD2, SMAD4, and DCC, are located in this region. The homologues of these genes in the mouse are also clustered on chromosome 18. Mutations in the adenomatous polyposis coli (APC) gene are responsible for familial adenomatous polyposis, and we earlier constructed a mouse model for familial adenomatous polyposis, Apc(delta716). Although human APC is located on chromosome 5q, mouse Apc is on chromosome 18, 30 cM proximal to the Dcc-Smad4-Smad2 locus. Taking advantage of this fact, we constructed previously a cis-compound Apc(delta716) Smad4 mutant, the intestinal polyps of which progress to very invasive adenocarcinomas. To determine whether Smad2 mutations play similar roles in malignant progression, here we constructed compound mutant mice carrying Apc and Smad2 knockouts in the cis configuration. In contrast to the cis-compound Apc(delta716) Smad4 heterozygotes, the polyps in the cis-compound Apc(delta716) Smad2 heterozygotes showed no difference in the number, size, or histopathology from the polyps in the simple Apc(delta716) heterozygotes. These results suggest that, on human chromosome 18q21, the SMAD4 LOH plays a more significant role, and SMAD2 LOH is insufficient to cause malignant progression of colonic polyps.

To investigate angiogenesis during intestinal polyp development, we determined the microvessel density (MVD) in polyps of Apc knockout (Apc(Delta716)) mice, a model for human familial adenomatous polyposis. We scored MVD also in several compound mutants carrying Apc(Delta716), namely, mice with an additional mutation in Smad4, in which the polyps progress into invasive adenocarcinomas; mice with a cyclooxygenase (COX)-2 gene (Ptgs2) mutation, in which adenoma growth is suppressed; and mice with prostaglandin E(2) EP receptor gene mutations. In both simple Apc(Delta716) and compound Apc(Delta716) Smad4 mutants, MVD increased in a polyp size-dependent manner only in the polyps expanded beyond a threshold of about 1 mm in diameter. These results indicate that tumor angiogenesis is stimulated only after tumors grow to a certain size, and this angiogenic switch is common to both benign adenomas and malignant adenocarcinomas. In Apc(Delta716) polyposis attenuated by the COX-2 gene mutation, in contrast, MVD did not increase even in polyps larger than 1 mm. The same phenomenon was observed in the compound mutant mice with Apc(Delta716) and the EP(2) receptor gene mutations, but not in other EP compound mutants. We also immunohistochemically studied COX-2 and angiogenic factors, vascular endothelial growth factor and basic fibroblast growth factor. Interestingly, expression of these proteins was also increased in polyps larger than 1 mm. These results suggest that, in both benign and malignant mouse intestinal tumors, stromal expression of COX-2 results in elevated prostaglandin E(2) levels that stimulate cell surface receptor EP(2), followed by induction of vascular endothelial growth factor that causes tumor angiogenesis.

The SMAD4 (DPC4) gene was initially isolated as a candidate tumor suppressor from the convergent site of homozygous deletions on 18q in a panel of pancreatic carcinoma cell lines. It encodes a common cytoplasmic signaling molecule shared by the transforming growth factor-beta, activin, and bone morphogenic pathways. We recently inactivated its mouse homologue Smad4 and demonstrated its role in the malignant progression of benign adenomas to invasive adenocarcinomas by analyzing mice with Apc and Smad4 compound mutations. Although simple Smad4 homozygotes were embryonically lethal, the heterozygotes were fertile and appeared normal up to the age of 1 year. Upon further investigation, however, they have developed inflammatory polyps in the glandular stomach and duodenum. By PCR genotyping and immunohistochemical staining, the wild-type Smad4 allele has been lost in the polyp epithelial cells, ie., loss of heterozygosity. On the other hand, we have not found any mutations in such genes as K-Ras, H-Ras, N-Ras, p53, or PTEN. Histologically, the polyps are similar to human juvenile polyps showing moderate stromal cell proliferation and infiltrations by eosinophils and plasma cells. In addition, foci of adenocarcinoma with signet ring cells are also found. These results are consistent with a recent report that germ-line SMAD4 mutations are found in a subset of familial juvenile polyposis.

To determine the biological role of caudal-like homeobox gene CDX2, we constructed knockout mice in which its mouse homologue Cdx2 was inactivated by homologous recombination, placing a bacterial lacZ gene under the control of the Cdx2 promoter. Although the homozygous mutants died in utero around implantation, the heterozygotes were viable and fertile and expressed lacZ in the caudal region in early embryos and in the gut tissues in adults. The heterozygotes developed cecal and colonic villi by anteriorization and formed hamartomatous polyps in the proximal colon. The hamartoma started to develop at 11.5 days of gestation as an outpocket of the gut epithelium, which ceased to express the remaining Cdx2 allele. The outpocket then expanded as a partially duplicated gut but was contained as a hamartoma after birth. In adult mice, these hamartomas grew very slowly and took a benign course. None of them progressed into invasive adenocarcinomas, even at 1.5 years of age. Whereas the cecal and colonic villi expressed lacZ, the hamartoma epithelium did not, nor did it express Cdx2 mRNA from the wild-type allele. However, genomic DNA analysis of the polyp epithelium did not show a loss of heterozygosity of the Cdx2 gene, suggesting a mechanism of biallelic Cdx2 inactivation other than loss of heterozygosity. These results indicate that the Cdx2 haploin-sufficiency caused cecal and colonic villi, whereas the biallelic inactivation of Cdx2 triggered anomalous duplications of the embryonic gut epithelium, which were contained as hamartomas after birth.

Nonsteroidal anti-inflammatory drugs reduce the incidence and mortality of colorectal carcinoma. Their chemopreventive effects appear to be due to inhibition of cyclooxygenase (COX)-2. Here, we have studied the relationship between the COX-2 mRNA levels and pathological characteristics in 43 primary colorectal carcinomas. COX-2 levels were significantly higher in tumors with larger sizes and in those with deeper invasions but were not correlated with whether the patients had metastasis or not. These results suggest that larger carcinomas produce more COX-2 to support their own growth and that COX-2 inhibitors may be effective agents of carcinoma growth suppression.