A comparison of three Peyer's patch "M-like" cell culture models: particle uptake, bacterial interaction, and epithelial histology.

Intestinal Peyer's patch (PP) microfold (M) cells transport microbes and particulates across the follicle-associated epithelium (FAE) as part of the mucosal immune surveillance system. In vitro human M-like cell co-culture models are used as screens to investigate uptake of antigens-in-nanoparticles, but the models are labour-intensive and there is inter-laboratory variability. We compared the three most established filter-grown Caco-2/Raji B cell co-culture systems. These were Model A (Kernéis et al., 1997), Model B (Gullberg et al., 2000), and Model C (Des Rieux et al. 2007). The criteria used were transepithelial resistance (TEER), the apparent permeability coefficient (Papp ) of [14 C]-mannitol, M cell-like histology, as well as latex particle and Salmonella typhimurium translocation. Each co-culture model displayed substantial increases in particle translocation. Truncated microvilli compared to mono-cultures was their most consistent feature. The inverted model developed by des Rieux et al. (2007) displayed reductions in TEER and an increased (Papp ), accompanied by the largest increase in particle translocation compared to the other two models. The normally-oriented model developed by Gullberg et al. (2000) was the only one to consistently display an increased translocation of Salmonella typhimurium. By applying a double Matrigel™ coating on filters, altering the medium feeding regime for Raji B cells, and restricting the passage number of B cells, improvements to the Gullberg model B were achieved, as reflected by increased particle translocation and improved histology. In conclusion, this is the first time all three designs have been compared in one study and each displays phenotypic features of M-like cells. While Model C was the most robust co-culture, the Model B protocol could be improved by optimizing several variables and is less complicated to establish than the two inverted models.