Identification of a Selective G1-Phase Benzimidazolone Inhibitor by a Senescence-Targeted Virtual Screen Using...

Area of Science:

• Oncology
• Computational Biology
• Drug Discovery

Background:

• Cellular senescence acts as a tumor suppressor in normal cells.
• Tumor cells can enter senescence in response to genotoxic stress, presenting a therapeutic target.
• Current senescence inducers often cause apoptosis, necessitating more selective agents.

Purpose of the Study:

• To identify novel, selective inducers of cellular senescence using a machine learning-based in silico screen.
• To discover compounds that promote senescence as a cancer therapy strategy.

Main Methods:

• Differential biological process network profiling from gene expression data under induced telomere dysfunction in colorectal cancer cells.
• Training a neural network classifier on 3517 compounds active against 17 protein targets.
• Screening a virtual library of ~2 million lead-like compounds using the trained model.
• Validating virtual hits through growth inhibition and senescence-associated β-galactosidase assays.

Main Results:

• A benzimidazolone compound, CB-20903630, was identified as a potent senescence inducer.
• CB-20903630 demonstrated low micromolar IC50 for HCT116 cell growth inhibition.
• The compound selectively induced senescence-associated β-galactosidase activity without cytotoxicity or apoptosis.
• Growth suppression was mediated by G1 cell cycle blockade, increased p21, and suppressed cyclin B1, CDK1, and CDC25C.
• CB-20903630 inhibited multicellular spheroid growth and long-term population kinetics.
• Expression analysis suggested a PI3K/AKT-inhibitor-like profile in normal cells.

Conclusions:

• Machine learning can effectively identify selective senescence inducers.
• CB-20903630 represents a promising lead compound for cancer therapy due to its selective induction of senescence.
• The compound's mechanism involves cell cycle arrest and distinct pathway modulation in cancer cells.