Machine learning-based discovery of GW3965 as a therapeutic compound against invasive emm92-type group A...

Area of Science:

• Microbiology
• Pharmacology
• Computational Biology

Background:

• Multi-drug resistant *emm92*-type Group A Streptococcus (GAS) strains are causing invasive infections, especially in people who inject drugs in the US.
• Emergence of antimicrobial resistance necessitates novel therapeutic strategies.

Purpose of the Study:

• To identify and repurpose FDA-investigated compounds as antimicrobials against multi-drug resistant *emm92*-type GAS.
• To discover novel therapeutics for invasive GAS infections.

Main Methods:

• A machine learning model was trained on the growth response of *emm92*-iGAS to 2,560 bioactive compounds.
• An *in silico* screen of 6,111 FDA-evaluated drugs was performed using the trained model.
• Experimental validation of the top-predicted compound (GW3965) in bacterial survival assays, a human skin equivalent model, and a mouse model of skin and soft tissue infection.

Main Results:

• The machine learning model successfully predicted GW3965 as a potent inhibitor of *emm92*-iGAS.
• GW3965 demonstrated a 99% reduction in iGAS survival at a Minimum Inhibitory Concentration (MIC) of 6.25 µM.
• GW3965 treatment resulted in complete wound closure in a human skin model and significantly reduced lesion size and bacterial burden in a mouse model.

Conclusions:

• Machine learning-driven drug repurposing expedited the discovery of GW3965 as a viable therapeutic candidate.
• GW3965 shows promise for treating invasive *emm92*-iGAS skin and soft tissue infections.