A colonic tissue architecture assay applied to human colon carcinoma cells

Summary A two-component tissue architecture assay system has been devised that tests the ability of human colon carcinoma cells to conform to the specific three-dimensional cell-cell and cell-substratum interactions characteristic of normal colonic tissues. Dissociated fetal rat colonic cells (FRCC) were allowed to reaggregate in suspension with or without the addition of different proportions (0.1%, 1%, and 10% of the total cells) of the human colon carcinoma cell lines, SW-1222 and LS-174T. Cellular aggregates obtained after 36 hours’ incubation exhibited cell sorting by the formation of recognizable epithelial colonic crypt-like structures with glandular lumens in a mesenchyme-like background. Carcinoembryonic antigen (CEA)-positive SW-1222 cells in 10% mixed aggregates were organized into numerous well-formed glandular structures with a polarized apical distribution of CEA. LS-174T cells, on the other hand, were self-sorted but structurally disorganized with a continuous cell surface CEA distribution. Pure FRCC and mixed aggregates were implanted under the kidney capsules of Swiss nu/nu (nude) or CD-1 nu/nu mice and allowed to grow for a period of 7–10 days. Whereas the normal FRCC readily formed colonic tissue, the SW-1222 cells exhibited a capacity for differentiation into colonic crypts which became progressively less normal and more tumor-like as the proportion of carcinoma cells in the aggregates was increased. The LS-174T cells demonstrated poor differentiation at all concentrations. Cell surface levels of CEA and the CEA family member nonspecific crossreacting antigen (NCA), both overexpressed in colon cancer, were higher in LS-174T than in SW-1222 cells, whereas family member biliary glycoprotein (BGP), downregulated in colon carcinoma was higher in the SW-1222 cells. These results thus support the suggestion that deregulated expression of CEA family members can be involved in the ability of colonocytes to differentiate and conform to normal tissue architecture as assessed by the assay. The assay is therefore amenable to genetic analysis of normal and perturbed architectural phenotypes. This work was supported by grants from the National Cancer Institute of Canada and the Medical Research Council of Canada. C. I. is a recipient of a studentship from “le Fonds pour la Formation de Chercheurs et l’Aide à la Recherche.”