In silico design of multi-target inhibitors against breast cancer-related proteins

Breast cancer is one of the most common types of cancer among women, and is the second leading cause of cancer death in women. In Portugal, for example, about 4,500 new cases of breast cancer are detected annually, and 1,500 women die of this disease. Although chemotherapy has been shown to be effective against this malignant neoplasm, depending on several factors (such as histology, degree and stage of disease, among others), there have been increasing cases of resistance to drugs used in current chemotherapy treatments. Thus, it is urgent to discover new drugs that are so powerful and more versatile against breast cancer. In the present work, we present a computational model for the in silico design of new chemical compounds with specific inhibitory activity in relation to nineteen proteins involved in breast cancer. The model was developed based on 14868 chemical compounds, which were previously tested on these proteins, and taking into account simultaneously other used experimental conditions. Such a model demonstrated to have a predictive capacity of around 93% and also allowed us to confirm which parts of the same compounds (molecular fragments) further contributed to their inhibitory activity. It was thus possible to design based on the physicochemical interpretation of the same model and on these molecular fragments eight new compounds with a strong anticancer activity. In addition, these lead compounds have been shown to have the desired pharmacological properties to be used as more effective drugs in the fight against breast cancer. Finally, the present work shows how computational models are inexpensive and effective alternative tools, and can therefore accelerate the design of new drugs.

Authors and Affiliations:

Alejandro Speck-Planche¹ e· M. Natália D. S. Cordeiro1

¹LAQV@REQUIMTE/Department of Chemistry and Biochemistry, Faculty of Sciences, University of Porto, 4169-007 Porto, Portugal

Abstract:

Breast cancer is the most frequent cancer reported in women, being responsible for hundreds of thousands of deaths. Chemotherapy has proven to be effective against this malignant neoplasm depending on different biological factors such as the histopathology, grade, and stage, among others. However, breast cancer cells have become resistant to current chemotherapeutic regimens, urging the discovery of new anti-breast cancer drugs. Computational approaches have the potential to offer promising alternatives to accelerate the search for potent and versatile anti-breast cancer agents. In the present work, we introduce the first multitasking (mtk) computational model devoted to the in silico fragment-based design of new molecules with high inhibitory activity against 19 different proteins involved in breast cancer. The mtk-computational model was created from a dataset formed by 24,285 cases, and it exhibited accuracy around 93% in both training and prediction (test) sets. Several molecular fragments were extracted from the molecules present in the dataset, and their quantitative contributions to the inhibitory activities against all the proteins under study were calculated. The combined use of the fragment contributions and the physicochemical interpretations of the different molecular descriptors in the mtk-computational model allowed the design of eight new molecular entities not reported in our dataset. These molecules were predicted as potent multi-target inhibitors against all the proteins, and they exhibited a desirable druglikeness according to the Lipinski’s rule of five and its variants.

Journal: Molecular Diversity

Link: http://link.springer.com/article/10.1007%2Fs11030-017-9731-1