Correlation between cell line chemosensitivity and protein expression pattern as new approach for the design of targeted anticancer small molecules

BACKGROUND AND RATIONALE: Over the past few decades, several databases with a significant amount of biological data related to cancer cells and anticancer agents (e.g.: National Cancer Institute database, NCI; Cancer Cell Line Encyclopedia, CCLE; Genomic and Drug Sensitivity in Cancer portal, GDSC) have been developed. The huge amount of heterogeneous biological data extractable from these databanks (among all, drug response and protein expression) provides a real foundation for predictive cancer chemogenomics, which aims to investigate the relationships between genomic traits and the response of cancer cells to drug treatment with the aim to identify novel therapeutic molecules and targets. In very recent times many computational and statistical approaches have been proposed to integrate and correlate these heterogeneous biological data sequences (protein expression – drug response), with the aim to assign the putative mechanism of action of anticancer small molecules with unknown biological target/s. The main limitation of all these computational methods is the need for experimental drug response data (after screening data). From this point of view, the possibility to predict in silico the antiproliferative activity of new/untested small molecules against specific cell lines, could enable correlations to be found between the predicted drug response and protein expression of the desired target from the very earliest stages of research. Such an innovative approach could allow to select the compounds with molecular mechanisms that are more likely to be connected with the target of interest preliminary to the in vitro assays, which would be a critical aid in the design of new targeted anticancer agents. RESULTS: In the present study, we aimed to develop a new innovative computational protocol based on the correlation of drug activity and protein expression data to support the discovery of new targeted anticancer agents. Compared with the approaches reported in the literature, the main novelty of the proposed protocol was represented by the use of predicted antiproliferative activity data, instead of experimental ones. To this aim, in the first phase of the research the new in silico Antiproliferative Activity Predictor (AAP) tool able to predict the anticancer activity (expressed as GI50) of new/untested small molecules against the NCI-60 panel was developed. The ligand-based tool, which took the advantages of the consolidated expertise of the research group in the manipulation of molecular descriptors, was adequately validated and the reliability of the prediction was further confirmed by the analysis of an in-house database and subsequent evaluation of a set of molecules selected by the NCI for the one-dose/five-doses antiproliferative assays. In the second part of the study, a new computational method to correlate drug activity data and protein expression pattern data was proposed and evaluated by analysing several case studies of targeted drugs tested by NCI, confirming the reliability of the proposed method for the biological data analysis. In the last part of the project the proposed correlation approach was applied to design new small molecules as selective inhibitors of Cdc25 phosphatase, a well-known protein involved in carcinogenic processes. By means of this innovative approach, integrated with other classical ligand/structures-based techniques, it was possible to screen a large database of molecular structures, and to select the ones with optimal relationship with the focused target. In vitro antiproliferative and enzymatic inhibition assays of the selected compounds led to the identification of new structurally heterogeneous inhibitors of Cdc25 proteins and confirmed the results of the in silico analysis. CONCLUSIONS: Collectively, the obtained results showed that the correlation between protein expression pattern and chemosensitivity is an innovative, alternative, and effective method to identify new modulators for the selected targets. In contrast to traditional in silico methods, the proposed protocol allows for the selection of molecular structures with heterogeneous scaffolds, which are not strictly related to the binding sites and with chemical-physical features that may be more suitable for all the pathways involved in the overall mechanism. The biological assays further corroborate the robustness and the reliability of this new approach and encourage its application in the anticancer targeted drug discovery field.