Development of an organotypic 3D in vitro model of normal human breast tissue: a tool for cancer initiation studies

The mechanisms involved in breast cancer initiation are not well understood. This may in part be due to a lack of an in vitro model that faithfully recapitulates the morphology, phenotype and in vivo architecture of the normal human mammary gland. Most in vitro models of normal breast have relied on the use of reconstituted basement membrane gels to induce luminal epithelial cell polarity and have neglected the role of myoepithelial cells and fibroblasts in this process. The aim of this thesis was to develop a three dimensional in vitro culture system of normal breast which included three of the major functional cell types of breast embedded in a more physiologically relevant collagen I matrix. It was then sought to use the system to investigate the mechanisms behind breast cancer initiation via genetic manipulation of well-known oncogenes and tumour suppressors involved in breast cancer progression.
To achieve this, myoepithelial cells (Myo1089, originally isolated from breast reduction mammoplasty sample, gift of Dr Mike O’Hare) and fibroblasts (isolated and immortalised from breast reduction mammoplasty samples collected with ethical approval in house) were characterised by immunofluorescence to assess their suitability. Following characterisation, these were virally transfected with Turbo Green Fluorescent Protein (tGFP) and dsRed protein respectively to enable tracking. Three-dimensional tri-cultures were established in collagen I and included the non-tumorigenic luminal epithelial cell line HB2 with GFP Myo1089 cells and dsRed fibroblasts. Cells were cultured for three weeks in Transwell™ cell culture inserts. Following fixation these were analysed by haematoxylin and eosin staining, confocal microscopy and immunohistochemistry. Morphology and immunostaining profiles were compared to sections of a normal human in vivo breast tissue specimen.
Immunohistochemical characterisation using the following antibodies: E-cadherin, epithelial membrane antigen, vimentin, laminin 5, collagen IV plus luminal and basal cytokeratins, demonstrated polarised epithelial structures with lumen formation and basement membrane
production with a similar immunostaining profile to normal breast tissue. The importance of including myoepithelial cells and fibroblasts in maintaining these structures was demonstrated. We established this model was amenable to genetic engineering by overexpressing HER2 and HER3 in HB2 cells, and knocking out ERβ1 in Myo1089 cells and DOCK4 in fibroblast cell lines using siRNA/shRNA techniques respectively. These were included in separate models with morphological and phenotypic effects determined by haematoxylin and eosin staining, immunohistochemistry and quantification of HB2 structures formed. We further investigated the intracellular signalling cascades stimulated by heregulin in order to validate our findings upon overexpression of HER2 and HER3 and to investigate the cancer initiation potential of heregulin in the breast.
In summary, an in vitro model of normal breast tissue that includes three of the major functional breast cell types cultured in a physiologically relevant three dimensional matrix has been developed. The morphology and protein expression profile of the model was validated against a human breast tissue specimen and confirmed that it is a suitable model of normal breast. The model proved to be reproducible, suitable for experimentation using genetic engineering and cell behaviour could be easily visualised using standard laboratory techniques. To conclude, this is a robust in vitro model of normal breast tissue offering an alternative cost-effective method of studying genes and processes involved in breast cancer initiation.