To understand the cellular events leading to a leaky gut in pathologies such as inflammatory bowel diseases, an in vitro approach using cell models is required. The intestinal cell lineage Caco-2 is the cell model most commonly used to reproduce the features of the human small intestine epithelium [17, 18]. Nevertheless, the Caco-2 cell model is different in some respects to the human epithelium. For example, it has a low paracellular permeability and allows access of highly diffusible small molecules to the microvilli too easily [14, 17, 19, 30, 31], owing to the lack of an adequate mucus layer. As an alternative, therefore, we analyzed the properties of a co-culture of Caco-2 cells and HT29-5M21 cells differentiating spontaneously into mucus-secreting cells. This co-culture system exhibits the two main important cellular types encountered in the human intestinal epithelium. In order to make this procedure reliable and reproducible, we replaced the usual serum-containing culture medium with a serum-free medium supplemented with hormones and amino acids. Although the proportions of both cellular types are determined at culture seeding, they can rapidly modify during culture owing to the ability of each cell line to adhere to and to proliferate on the culture substrate. More, the heterogeneity of Caco-2 cells between 0 and 30 days of confluence  can deeply influence the co-culture composition.
A brief characterization of the Caco-2/HT29-5M21 co-culture, initially seeded with a 3:1 ratio, was performed. Results showed a monolayer containing few cells with condensed DNA (ethidium bromide accumulation). The ethidium bromide accumulation may reflect an early proliferative/apoptotic activity and modified nucleic acid expression. Indeed, an increase in DNA content is a feature of actively-proliferating cells that would therefore exhibit higher fluorescence peaks when labeled by ethidium bromide . Moreover, previous studies reported the appearance of an increase in DNA staining by ethidium bromide during the early phases of apoptosis, known as the hyperdiploid peak, . On the other hand, intracellular ethidium bromide fluorescence could also be enhanced by an increase in DNA base pair accessibility associated with the differentiation state . Then, a recent study indeed suggests that cell proliferation and apoptosis can co-exist . In our case we can not speculate between these possibilities. Whatever the solution may be, our results show that our co-culture has a high proportion of living cells.
For this study, we used alkaline phosphatase as an indicator of the cell proportion in co-culture models. The alkaline phosphatase is a membrane-bound enzyme localized in the brush border of enterocytes in the human intestinal epithelium and was used previously by Matsumoto et al.  to monitor the differentiation of enterocyte-like cells. In order to analyze the correlation between the proportion of the enterocyte-like cells and the alkaline phosphatase in the Caco-2/HT29-5M21 co-cultures, we estimated the enzyme SA. This enzyme activity is increased with the Caco-2 proportion. However, the alkaline phosphatase SA in co-cultures seems to be higher than in cell mixtures with the same proportions (1:3, 2:2 and 3:1). This result suggests a positive interaction between Caco-2 and HT29-5M21 cells. This interaction could lead to an increase of the gene expression, an increase of the translation rate or a decrease of enzyme degradation. This last possibility was already proposed by Matsumoto et al.  during cell differentiation.
The alkaline phosphatase activity was detected in co-cultures by the cytochemistry method. Our results show that the alkaline phosphatase was restricted to the enterocyte-like cell and could then be used as a potential marker for the monitoring of these cells under several co-culture conditions. The discrepancy between the alkaline phosphatase-positive area and the Caco-2 seeding proportion in culture could be explained by the heterogeneity of the cell lineage . Indeed, this cell line differentiates spontaneously in enterocyte-like cells, but also in other cell types which do not express alkaline phosphatase. Our results underline this heterogeneity, as we observed, by cytochemistry at cellular level, several red-scales (meaning alkaline phosphatase activity) in the same culture. This differentiation seems to be influenced by the Caco-2/HT29-5M21 co-culture and thus by a cell-cell interaction.
In order to collect more data on the proportionate composition of the monolayer, we investigated the expression level of the alkaline phosphatase coding gene and of the MUC5AC gene. Though the two cell lineages were both of an adenocarcinoma origin [16, 21, 24], we showed that the IAP and MUC5AC gene expressions are restricted, respectively, to Caco-2 cells and HT29-5M21 cells. We showed by real-time PCR that the expression level of the IAP and MUC5AC genes in the monolayer is independent of the cell proportion and thus of the Caco-2 or HT29-5M21 abundancy. However, the specific activity of the alkaline phosphatase was dependant on the Caco-2 proportion. Consequently, there is a complex control of the gene expression, enzyme translation and degradation regarding the cell-cell interaction.
Monolayer permeability is influenced by the presence of HT29-5M21 cells in the system. This influence is in accordance with an increase of the permeability. Indeed, we showed a decrease of the TEER and an increase of the Lucifer yellow and 20 kDa dextran permeability with the increase of HT29-5M21 proportion in the monolayer at seeding. The cell ratio is established at seeding and is probably not indicative of the situation when permeability is measured. However, HT29-5M21 cells seem to remain in sufficient proportion in the culture to increase cell permeability. The range of molecular masses (Lucifer yellow 452 Da & dextran 20 kDa) used in this study to explore monolayer permeability allow to characterize para-cellular permeability [37, 38]. Lucifer yellow is not absorbed by epithelial cell but is able to pass through cell junctions as mannitol does. Dextrans are largely used as macromolecule permeability tracer.