Principles for computational design of binding antibodies.

Area of Science:

• Protein engineering
• Computational biology
• Structural biology

Background:

• Natural proteins require stable conformations for molecular function.
• Computational protein design has excelled at creating stable proteins with "ideal" folds.
• Designing proteins with non-ideal features like loops and polar networks for function remains challenging.

Purpose of the Study:

• To develop principles for designing stable and functional antibody variable fragments (Fvs).
• To overcome limitations in computational design for non-ideal protein structures.
• To generate precise and functional antibodies and enzymes.

Main Methods:

• Utilized five design/experiment cycles for iterative learning.
• Applied sequence-design constraints from antibody multiple-sequence alignments.
• Maintained stabilizing framework-loop interactions in complementarity-determining regions (CDRs) during backbone design.

Main Results:

• Designed Fvs exhibited midnanomolar ligand-binding affinities.
• Engineered Fvs demonstrated stability comparable to natural antibodies.
• Crystallographic analysis confirmed atomic accuracy in frameworks and CDRs of designed Fvs.

Conclusions:

• Established generalizable principles for designing stable, functional proteins with non-ideal folds.
• Demonstrated successful design of antibody variable fragments with high affinity and stability.
• Showcased the potential for designing precise antibodies and enzymes for various applications.