Ab initio structure prediction of the antibody hypervariable H3 loop.

Area of Science:

• Structural biology
• Computational chemistry
• Immunology

Background:

• Antibodies bind diverse antigens via complementarity determining regions (CDRs).
• The H3 loop is the most variable CDR loop, posing a challenge for structure prediction.
• Accurate H3 loop structure prediction is vital for computational antibody design and engineering.

Purpose of the Study:

• To develop and evaluate an ab initio method for predicting H3 loop structures.
• To assess prediction accuracy in crystallographic and non-crystallographic environments.

Main Methods:

• Utilized conformational sampling and energy calculations with the Prime program.
• Tested on a dataset of 53 H3 loops with lengths ranging from 4 to 22 residues.
• Evaluated predictions in both crystal and non-crystallographic (homologous antibody scaffold) environments.

Main Results:

• In a crystal environment, median backbone RMSD was 0.5 Å, with 91% of predictions < 2.0 Å.
• In a non-crystallographic environment, 70% of predictions achieved RMSD < 2.0 Å.
• Demonstrated high accuracy for ab initio H3 loop structure prediction.

Conclusions:

• Ab initio H3 loop structure prediction is feasible and accurate.
• The developed method shows promise for computational antibody modeling and engineering.
• Accurate H3 loop prediction can advance the design of novel antibodies with improved properties.