Machine learning-driven QSAR modeling combined with molecular dynamics suggests high-affinity CD33-targeting peptides...

Area of Science:

• Biochemistry
• Computational Biology
• Oncology

Background:

• Current antibody therapies for leukemia face challenges like immunogenicity, toxicity, and resistance.
• CD33 is a validated surface marker for leukemia targeting.

Purpose of the Study:

• To develop novel peptide-based therapeutics targeting CD33 in leukemia.
• To create a computational and experimental pipeline for designing targeted anticancer peptides.

Main Methods:

• Utilized machine learning quantitative structure-activity relationship (QSAR) modeling (R²=0.93).
• Employed molecular docking and 100-ns molecular dynamics simulations for peptide-CD33 interaction prediction.
• Experimentally validated peptide structure, self-assembly, selectivity, cytotoxicity, and safety.

Main Results:

• Identified two lead peptides, A3K2L2 and K4I3, with high binding affinities (-146.11 and -108.08 kcal/mol) and stable interactions.
• Peptides formed stable reverse β-sheet structures and self-assembling nanostructures.
• Demonstrated potent cytotoxicity against K-562 leukemia cells (IC₅₀ 60-90 μM) with minimal impact on normal PBMCs, low hemolytic activity (<5%), and induced cancer cell death.

Conclusions:

• Developed a successful integrated pipeline for designing targeted anticancer peptides.
• Identified promising peptide candidates that overcome limitations of antibody therapies for leukemia.
• The platform offers a generalizable framework for developing receptor-specific anticancer peptides, particularly for hematological malignancies.