A major advantage of image-based phenotypic profiling of compounds is that numerous image features can be sampled and quantitatively evaluated in an unbiased way. However, since this assay is a discovery-oriented screening, it is difficult to determine the optimal experimental setup in advance. In this study, we examined whether variable cellular stimulation affects the efficacy of the image-based profiling of compounds. Seven different epidermal growth factor receptor (EGFR) ligands were used, and the expression of EGFR signaling molecules was monitored at various time points. Significant quantitative differences in image features were detected among the differentially treated samples. Next, 14 different compounds that affect EGFR signaling were profiled. Nearly half of the compounds were classified into distinct clusters, irrespective of differential ligand stimulation. The results suggest that image-based phenotypic profiling is quite robust in its ability to predict compound interaction with its target. Although this method will have to be validated in other experimental systems, the robustness of image-based compound profiling demonstrated in this work provides a valid basis for further study and its extended application.
Keywords: cancer and cancer drugs; cell-based assays; high-content screening; phenotypic drug discovery; signal transduction.