Machine learning approaches to optimize small-molecule inhibitors for RNA targeting

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Molecular Biology

Background:

• Data-driven algorithms aid in discovering small molecules with specific biological targets.
• The ribosomal peptidyl transferase center (PTC) is a validated target for antimicrobial drug development.

Purpose of the Study:

• To develop data-driven algorithms for identifying phenylthiazole-containing molecules that bind to the Mycobacterium tuberculosis ribosomal PTC RNA hairpin.
• To guide the chemical space search for potent inhibitors of bacterial translation.

Main Methods:

• Utilized complementary data-driven algorithms to analyze molecular features for binding affinity.
• Synthesized 10 computationally identified small molecules for functional validation.
• Assessed inhibition of the ribosomal PTC in Mycobacterium tuberculosis.

Main Results:

• Identified key visual, geometrical, and chemical features enhancing binding to the target RNA hairpin.
• Four out of 10 synthesized molecules demonstrated potent inhibition of hairpin 91 in the ribosomal PTC.
• The identified inhibitors effectively halt bacterial translation.

Conclusions:

• Data-driven approaches are effective in discovering novel antimicrobial agents.
• Phenylthiazole derivatives targeting the M. tuberculosis ribosomal PTC represent a promising new class of antibiotics.
• Halting translation via inhibition of ribosomal function is a viable strategy against tuberculosis.