Functional microRNA-Targeting Drug Discovery by Graph-Based Deep Learning.

Area of Science:

• Oncology
• Bioinformatics
• Molecular Biology

Background:

• MicroRNAs are crucial in cancer development, yet challenging to target with small molecules.
• Specific microRNAs like microRNA-21 (miR-21) drive breast cancer progression.
• Developing effective small molecule inhibitors for microRNAs is a significant unmet need in cancer therapy.

Approach:

• Introduced RiboStrike, a deep learning framework for identifying microRNA-targeting small molecules.
• Applied RiboStrike to screen nine million molecules for miR-21 inhibitors, incorporating counter-screening against DICER to ensure specificity.
• Utilized auxiliary models for toxicity assessment and candidate selection.

Key Points:

• RiboStrike identified eight candidate molecules, with three demonstrating significant anti-miR-21 activity in reporter assays and RNA sequencing.
• One validated compound effectively reduced lung metastases in preclinical mouse models of breast cancer.
• The framework successfully screened a large chemical library for specific microRNA-targeting agents.

Conclusions:

• RiboStrike demonstrates high efficacy in discovering potent and specific microRNA-targeting small molecules for cancer therapy.
• The validated anti-miR-21 compound shows therapeutic potential for breast cancer treatment.
• This deep learning approach offers a powerful new strategy for developing novel microRNA-based cancer drugs.