Fig. 5

Cellular and EV proteomics results. a + d. Venn diagrams of differentially expressed cellular (a) and EV (d) proteins identified by label-free proteomics in the different vandetanib doses compared to DMSO. The gene names of the differentially expressed proteins in all vandetanib treatments compared to DMSO were added to the diagrams. b + e. Two-dimensional reduction of the 500 proteins with the highest variance in the quantile-normalized proteomics dataset by principal component analysis (PCA) in the cellular (b) and EV (e) proteome. Black squares, yellow dots, orange triangles and red diamonds indicate the DMSO control, EC10, EC50, and EC80 dose, respectively. Ellipses represent 95% CIs. c + f. Three-dimensional reduction of the quantile-normalized cellular (c) and EV (f) proteomics data by sparse partial least-squares discriminant analysis (sPLS-DA). Tuning of the model resulted in choosing three components with 47, 41 and 45 factors each (10, 3 and 24 for EVs). The plots show the samples in the first two dimensions. Black squares, yellow dots, orange triangles and red diamonds indicate the DMSO control, EC10, EC50 and EC80 dose, respectively. Ellipses represent 95% CIs. g + h. Heatmap representing the quantile-normalized (QN) expression data of the 43 (g) / 11 (h) proteins that were picked by the sPLS-DA model and significantly differentially expressed in at least one vandetanib dose compared to the DMSO control on the cellular (g) / EV (h) level.