Objective:

To identify a GBM-associated membrane target using machine learning techniques and develop a corresponding cyclic peptide ligand for targeted radionuclide therapy.

Approach:

• Data Integration: Utilized transcriptomic profiling, differential expression analysis, and machine learning techniques to identify potential targets, integrating structural modeling and molecular docking.
• Target Identification: Prioritized candidate genes using LASSO regression, gradient boosting, and SHAP interpretation.
• Peptide Development: Designed and synthesized a cyclic peptide ligand, IMP-3, and validated its binding properties.

Key Findings:

• DLL3 was identified as a top candidate target with high discriminative power and prognostic relevance in GBM.
• The cyclic peptide candidate IMP-3 showed favorable predicted binding to DLL3 and stable interaction during molecular dynamics simulation.
• The probe-labeled derivative MPA-IMP-3 selectively accumulated in GBM cells, demonstrating targeting specificity with significantly stronger fluorescence signals compared to normal astrocytes.

Interpretation:

DLL3 is identified as a target for GBM therapy, and the developed cyclic peptide ligand has potential for selective binding.

Limitations:

• The study primarily relies on computational methods, which may require further experimental validation in clinical settings.
• DLL3-directed delivery systems for GBM-targeted radionuclide therapy are still underdeveloped.

Conclusion:

The integrated AI-driven workflow provides a rational framework for developing targeted theranostic strategies in GBM.