Modelling breast cancer requires identification and correction of a critical cell lineage-dependent transduction bias.

Clinically relevant human culture models are essential for developing effective therapies and exploring the biology and etiology of human cancers. Current breast tumour models, such as those from oncogenically transformed primary breast cells, produce predominantly basal-like properties, whereas the more common phenotype expressed by the vast majority of breast tumours are luminal. Reasons for this puzzling, yet important phenomenon, are not understood. We show here that luminal epithelial cells are significantly more resistant to viral transduction than their myoepithelial counterparts. We suggest that this is a significant barrier to generating luminal cell lines and experimental tumours in vivo and to accurate interpretation of results. We show that the resistance is due to lower affinity of luminal cells for virus attachment, which can be overcome by pretreating cells--or virus--with neuraminidase. We present an analytical method for quantifying transductional differences between cell types and an optimized protocol for transducing unsorted primary human breast cells in context.