Studies of antibody conformational ensembles towards better understanding of structure and function

Thesis

In this thesis, we examine relationships between structural, biophysical and dynamic properties of antibodies. First, we investigate determinants of the binding rates of antibodies, showing that the flexibility of the third complementarity-determining region of the heavy chain (CDR-H3) and the epitope are predictors of this, while analysis of static structures does not yield information on binding rates. We also show that consideration of the conformational ensemble of an antibody gives an enriched view of the surface properties that contribute heavily to developability of an antibody therapeutic. Next, we investigate the sources of CDR-H3 flexibility and structural diversity. We provide evidence in support of the long-standing hypothesis that CDR-H3 structure and flexibility are fine-tuned by the rest of the antibody and propose a model for a steric cage that other CDR loops form around CDR-H3, thereby modulating its conformational space. We also demonstrate the effects of different genetically encoded structural attributes on the flexibility and structural variability of CDR-H3, showing that certain genetic lineages of antibodies have preferences for CDR-H3 loops of differing levels of flexibility. Lastly, we train deep learning models to improve predicted models of antibody structures, demonstrating that errors made by state-of-the-art structure prediction algorithms can effectively be learned.

Authors: Guloglu, B

• DOI: 10.5287/ora-2aa0m1qnr
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: structure refinement; antibody structure; therapeutic antibodies; binding kinetics; antibody binding; structure modelling; molecular dynamics; antibody on rates
• Subjects: Computational Biology; Biophysics; Structural Biology; Immunoinformatics