AbDesign: An algorithm for combinatorial backbone design guided by natural conformations and sequences.

Area of Science:

• Protein engineering and computational biology
• Immunology and antibody-based therapeutics

Background:

• Computational protein design has advanced significantly, enabling the creation of novel enzymes, binders, and nanomaterials.
• Designing new protein backbones with functional control remains a critical challenge, with most prior methods computing the mainchain de novo.

Purpose of the Study:

• To introduce AbDesign, a combinatorial algorithm for designing novel antibody backbones and sequences.
• To optimize antibody designs for high stability and binding affinity against specific target surfaces.

Main Methods:

• AbDesign leverages existing antibody sequences and structures to construct new antibody models.
• The algorithm docks models against target surfaces and optimizes sequence and backbone conformation.
• The approach was validated by designing antibodies targeting surfaces of nine natural high-affinity antibodies.

Main Results:

• Five designs achieved >30% interface sequence identity with natural antibodies.
• Four designs showed backbone conformation within 1 Å root-mean square deviation at the binding core.
• Designed antibodies recapitulated natural polar interaction networks and exhibited high amino acid sidechain rigidity.

Conclusions:

• AbDesign successfully generates novel antibody designs with high sequence and structural similarity to natural antibodies.
• The algorithm optimizes binding interfaces for enhanced stability and affinity, showing potential for therapeutic antibody development.
• Backbone conformation diversity in peripheral complementarity-determining regions (CDRs) may further enhance antibody affinity and specificity.