Gabriele Gamerith, Johannes Rainer, Julia M. Huber, Hubert Hackl, Zlatko Trajanoski, Stefan Koeck, Edith Lorenz, Johann Kern, Reinhard Kofler, Jens M Kelm, Heinz Zwierzina, Arno Amann

This work evaluated gene expression differences between a hanging-drop 3D NSCLC model and 2D cell cultures and their in-vivo relevance by comparison to patient-derived data from The Cancer Genome Atlas. Gene expression of 2D and 3D cultures for Colo699 and A549 were assessed using Affymetrix HuGene 1.0 ST gene chips. Biostatistical analyses tested for reproducibility, comparability and significant differences in gene expression profiles between cell lines, experiments and culture methods. The analyses revealed a high interassay correlation within specific culture systems proving a high validity. 979 genes were altered in A549 and 1106 in Colo699 cells due to 3D cultivation. The overlap of changed genes between the cell lines was small (149), but the involved pathways in the reactome and GO- analyses showed a high overlap with DNA methylation, cell cycle, SIRT1, PKN1 pathway, DNA repair and oxidative stress as well known cancer-associated representatives. Additional specific GSEA-analyses revealed changes in immunologic and endothelial cell proliferation pathways, whereas hypoxic, EMT and angiogenic pathways were downregulated. Gene enrichment analyses showed 3D-induced gene up-regulations in the cell lines 38 to be represented in in-vivo samples of NSCLC patients using data of The Cancer Genome Atlas. Thus, our 3D NSCLC model might provide a tool for early drug development and investigation of microenvironment-associated mechanisms. However, this work also highlights the need for further individualization and model adaption to address remaining challenges.