# High-affinity optimization potential of the virus neutralizing antibody with twin cysteine-stabilized complementarity-determining region 3

## TL;DR

This study shows that stabilizing the H3 CDR in antibodies with twin cysteines improves AI-driven optimization, leading to better virus neutralization.

## Contribution

Demonstrates that H3 CDR stabilization enhances AI-based antibody optimization for SARS-CoV-2 neutralization.

## Key findings

- Stabilized H3 CDR antibodies showed improved binding and neutralization against SARS-CoV-2.
- Optimized antibodies formed tighter interactions with the ACE2 receptor, correlating with efficacy.
- In vivo tests showed reduced viral loads in mice treated with optimized antibodies.

## Abstract

The optimization of neutralizing monoclonal antibodies (NMAbs) is crucial to counter viral evolution. The structural stability of the heavy-chain complementarity-determining region 3 (H3 CDR) significantly influences affinity maturation potential, yet its impact on computational optimization remains unclear.

This study employed an artificial intelligence (AI) model to optimize two categories of SARS-CoV-2 NMAbs: one featuring a conformationally stabilized H3 CDR via a twin cysteine motif, and another with flexible H3 CDR loops. Optimized antibody derivatives were evaluated for binding affinity to the SARS-CoV-2 spike protein, pseudovirus and live virus neutralization, and in vivo efficacy in a murine infection model. Structural analyses were conducted to elucidate interaction mechanisms with the angiotensin-converting enzyme 2 (ACE2) receptor.

H3 CDR stabilization via twin cysteines markedly enhanced AI-driven optimization efficacy. Optimized derivatives from the stabilized antibody category exhibited improved binding affinity and superior neutralization potency against both pseudotyped and authentic SARS-CoV-2 viruses. Structural analyses revealed optimized antibodies formed tighter interactions with the ACE2 receptor, including enhanced binding between key residues and ACE2, which correlated with biological efficacy. In contrast, antibodies lacking H3 CDR stabilization showed no affinity improvement after the same optimization process. In vivo, optimized antibodies effectively suppressed viral replication and reduced viral loads in infected mice. Mechanistically, the twin cysteine stabilization minimized structural perturbations caused by affinity-enhancing mutations, unlocking the optimization potential of the H3 CDR.

These findings establish that conformational stabilization of the H3 CDR in seed antibodies is a critical determinant for successful AI-driven affinity maturation. The study proposes a strategic framework for antibody development that prioritizes structurally stabilized H3 CDR regions, offering a robust approach to generating high-potency therapeutics against rapidly evolving viral pathogens.