Modelling lymphangioleiomyomatosis (LAM) using two-dimensional and three-dimensional in vitro co-culture systems

Background: Lymphangioleiomyomatosis (LAM) is a rare progressive neoplastic cystic lung disease that primarily affects women of child bearing age leading to lung destruction, respiratory failure and death. Thought to be a consequence of dysregulated protease expression, cells of unknown origin accumulate in the lung, often forming clusters or nodules of cells with both melanocytic and smooth muscle properties. Some of these cells, known as LAM cells, have bi-allelic mutations in TSC2 resulting in constitutive mTOR activation. However LAM nodules are heterogeneous structures and genotyping analyses suggest that cells without LOH for TSC2 including wild-type fibroblasts are also common within LAM nodules. Hypotheses and aims: We hypothesise that LAM cells recruit wild-type fibroblasts and modify their properties to generate a permissive microenvironment, akin to a tumour stroma including the production and activation of matrix-degrading proteases which contribute to the destruction of the lung parenchyma. This study has therefore deigned in vitro co-culture models with an aim to study the expression patterns and activation of proteases in a LAM lung leading to matrix destruction. Another aim was also to characterise transcriptional differences normal human lung fibroblasts (NHLFs) and LAM-associated fibroblasts (LAFs) and to investigate changes in their gene expression when cultured together with a model LAM cell line, 621-101 angiomyolipoma cells which were derived from a LAM patient and have bi-allelic loss of TSC2. Methods: In vitro 2-dimensional (2D) and 3-dimensionalD (3D) co-culture models were designed and validated using fibroblasts characterised and isolated from 4 LAM lung donors, now termed LAFs, and 621-101 cells. The 3D extracellular matrix (ECM) incorporated the two cell types in a 10:1 ratio embedded in a basement membrane extract (BME) mimicking the lung matrix. An organotypic spheroid model was also developed incorporating both cell types thereby mimicking a LAM nodule. 6 LAM lung and 3 normal lung tissue donors were screened for candidate proteases in LAM pathology using qRT-PCR and identified upregulated proteases which may contribute to a role in LAM pathology. These findings were verified in the 2D and 3D in vitro models as well as ex vivo tissue using a variety of immunostaining techniques, activity assays and ELISA. Lastly, commercially bought NHLF (n=3) and LAF (n=3) were cultured in the presence or absence of 621-101 cells in the 2D Boyden chamber co-culture model. LAF and NHLF RNA was analysed using Affymetrix Human Genome U133 Plus 2.0 Arrays and Genomics Suite and Pathway (Partek). Findings were validated by multiplex assay and immunohistochemistry in 2D and 3D in vitro models and tissue respectively. Inflammatory cell migration and function was examined in co-culture model and LAM tissue. Results: The 3D BME model showed that TSC2-/- 621-101 cells and fibroblasts spontaneously form aggregates and clump together akin to a LAM nodule. The two cell types exhibited strong heterotypic cell-cell adhesive forces and resulted in strictly spherical spheroids. The 3D models designed all showed expression of markers of LAM nodules thereby representing LAM nodules in a dish. Of 30 proteases screened, cathepsin K gene expression was increased almost 15-fold in LAM lung compared to normal tissue and was also found to be elevated in 3D BME model. Cathepsin K in LAM tissue was expressed in the LAM nodules associated with cysts and was expressed exclusively by fibroblasts in the 3D spheroid model. As cathepsin K requires low pH for activity it was determined if LAFs and TSC2-/- cells can acidify the extracellular space. TSC2-/- cells but not LAFs decreased extracellular pH, over 24 hours and pH values