AI-Driven Acceleration of Fluorescence Probe Discovery.

Area of Science:

• Biochemistry
• Molecular Biology
• Medical Imaging

Background:

• Fluorescence imaging probes are crucial for clinical and preclinical research but their discovery is hindered by limited scaffolds and slow, costly trial-and-error methods.
• Developing target-specific probes requires overcoming challenges in identifying suitable fluorophore scaffolds and predicting photophysical properties.

Purpose of the Study:

• To develop a hybrid strategy integrating Artificial Intelligence (AI) with bioassays to accelerate the discovery of target-specific fluorescence imaging probes.
• To create an AI model (PROBY) capable of identifying fluorescent molecules and predicting key photophysical properties.

Main Methods:

• Developed PROBY, an AI model trained on over one million molecular entries from nine datasets to identify fluorescent molecules and predict seven photophysical properties.
• Applied PROBY to a library of 26,416 target-validated molecules to identify candidates with target affinity and favorable optical characteristics.
• Validated AI-identified candidates for clinically relevant targets (tau, BCL-2, TDP-43) and optimized a lead compound (PE859) through chemical modification.

Main Results:

• PROBY identified thousands of candidate fluorescent probes with target affinity and desirable optical properties.
• Discovered PE859, obatoclax, and B3, which were validated for applications including spectral analysis, drug screening, pathological labeling, and in vivo imaging.
• Optimized derivative 859-2 enabled in vivo two-photon imaging of tau pathology in transgenic mice.

Conclusions:

• The hybrid AI-bioassay strategy significantly expands the accessible scaffold landscape for designing target-specific fluorescence probes.
• This approach provides a scalable, efficient, and cost-effective framework for the next generation of fluorescence probe discovery.
• The developed AI model and validated probes offer powerful tools for advancing clinical navigation and preclinical research.