Investigation of Genes Important to the Zymogenic Lineage in Mammalian Gastric Mucosa

Gastric cancer remains a prominent player in worldwide cancer mortality, due to its poor detection at early stages. Prevalence of the more common of the two canonical histologically defined gastric cancer subtypes, intestinal adenocarcinoma, has been steadily decreasing, likely due to worldwide decrease in H. pylori-induced gastritis; a condition that greatly increases cancer risk. On the other hand, diffuse adenocarcinomas, the rarer but more pernicious form, have been increasing, especially in more developed countries. Both forms of adenocarcinoma likely arise because of aberrant gastric epithelial renewal. Recent evidence implies that the zymogenic lineage specifically is the most at risk for producing aberrant patterns of differentiation that lead to cancer. Of the four primary lineages within the gastric mucosa, differentiation of the zymogenic lineage is most complex. Their precursors, mucous neck cells, descend out of the isthmal proliferative zone and migrate amongst the acid secreting parietal cells until they reach the base, whereupon they rapidly transition into secretory zymogenic cells. The differentiation of neck cells into zymogenic cells is meditated by the parietal cells, whose loss drives pre-neoplastic metaplasia. Parietal cell death induces spasmolytic polypeptide expressing metaplasia (SPEM). Spasmolytic polypeptide, aka trefoil factor two (TFF2), is normally a neck-cell specific secreted protein. During SPEM it and other neck cell genes are ectopically expressed in the base, in the region where zymogenic cells normally reside. SPEM has been directly associated with both intestinal metaplasia and adenocarcinoma. We identified a novel gene exclusively in neck cells. Brinc (GKN3) is a stomach-specific gene identified by expression profiling of laser-captured neck cells. GKN3 shares homology with two other gastric specific genes that have been identified as tumor suppressors: GKN1 and GKN2. Here, I will show that the three gastrokines have an interesting human evolutionary story, especially with regards to GKN3. Using human population genetic and evolutionary analyses, I (along with Carlo Lapid in the Templeton lab) have discovered that GKN1 is likely under more stringent selective control than the other two genes. Also, several selective sweeps have occurred in GKN3 that show population specific trends, the largest one likely leading to differential expression. I will show that GKN3 appears to be expressed during human gastric metaplasia and might, therefore, is a novel marker for SPEM. In parallel studies, I will show that parietal cells, though known to be necessary for normal zymogenic patterning, are not sufficient, as mice deficient in the secretory cell-organizing transcription factor MIST1 and the endocytic trafficking coordinating adaptor CD2AP, have parietal cells that are abundant in the base of their gastric units (near zymogenic cells) in addition to the neck; but this over- abundant distribution does not affect neck cells to zymogenic cell transition. The cause of parietal cell accumulation in the base appears to be loss of tensile strength of zymogenic cells, which normally exclude parietal cells via cell-cell homotypic interactions. Accordingly, Mist1−/−; Cd2ap−/− mice have aberrant E-cadherin localization. Conditional loss of E-cadherin in zymogenic cells in Mist1-/- mice recapitulates the Mist1-/-; Cd2ap-/- phenotype, though conditional loss of E-cadherin alone does not, indicating that a combination of MIST1 and E-cadherin endocytic trafficking is needed to establish a normal zymogenic base zone that excludes parietal cells. Interestingly, mutations in CDH1 (the gene encoding E-cadherin) have been directly linked to both hereditary and sporadic diffuse gastric cancers. I will show that, in mice, conditional loss of Cdh1 is sufficient to form signet rings cells, the hallmark histological form adopted by diffuse cancer cells, only under metaplastic conditions. These signet ring cells do not go on to form carcinomas, rather cells lacking Cdh1 likely die