Artificial Intelligence-Powered Molecular Docking and Steered Molecular Dynamics for Accurate scFv Selection of
Nico Martarelli1, Michela Capurro1, Gizem Mansour2
1Institute of Hematology and Center for Hemato-Oncology Research, University of Perugia and Santa Maria della Misericordia Hospital, 06132 Perugia, Italy.
Abstract:
Chimeric antigen receptor (CAR) T cells represent a revolutionary immunotherapy that allows specific tumor recognition by a unique single-chain fragment variable (scFv) derived from monoclonal antibodies (mAbs). scFv selection is consequently a fundamental step for CAR construction, to ensure accurate and effective CAR signaling toward tumor antigen binding. However, conventional in vitro and in vivo biological approaches to compare different scFv-derived CARs are expensive and labor-intensive. With the aim to predict the finest scFv binding before CAR-T cell engineering, we performed artificial intelligence (AI)-guided molecular docking and steered molecular dynamics analysis of different anti-CD30 mAb clones. Virtual computational scFv screening showed comparable results to surface plasmon resonance (SPR) and functional CAR-T cell in vitro and in vivo assays, respectively, in terms of binding capacity and anti-tumor efficacy. The proposed fast and low-cost in silico analysis has the potential to advance the development of novel CAR constructs, with a substantial impact on reducing time, costs, and the need for laboratory animal use.
Insights
Artificial intelligence accurately predicts the best antibody fragments for CAR T-cell therapy, accelerating development. This computational approach reduces costs and lab work, improving chimeric antigen receptor (CAR) T-cell engineering.
Area of Science:
- Immunotherapy
- Computational Biology
- Biotechnology
Background:
- Chimeric antigen receptor (CAR) T cells are a revolutionary immunotherapy for cancer treatment.
- Effective CAR T-cell therapy relies on selecting optimal single-chain fragment variable (scFv) components derived from monoclonal antibodies (mAbs).
- Traditional methods for scFv selection are costly, time-consuming, and labor-intensive.
Purpose of the Study:
- To develop a fast, low-cost computational method for predicting scFv binding affinity before CAR-T cell engineering.
- To identify the most effective scFv for anti-CD30 CAR constructs using artificial intelligence (AI).
Main Methods:
- AI-guided molecular docking and steered molecular dynamics simulations were employed to analyze anti-CD30 mAb clones.
- Virtual computational scFv screening was performed to assess binding capacity.
- Results were compared with traditional in vitro (surface plasmon resonance) and in vivo assays.
Main Results:
- AI-guided virtual screening accurately predicted scFv binding capacity.
- Computational results correlated well with surface plasmon resonance (SPR) data.
- The in silico approach demonstrated comparable anti-tumor efficacy to in vitro and in vivo assays.
Conclusions:
- AI-driven in silico analysis offers a viable, cost-effective alternative for scFv selection in CAR-T cell development.
- This approach significantly reduces the time, cost, and need for animal models in engineering novel CAR constructs.
- The study highlights the potential of computational methods to accelerate immunotherapy advancements.
