Aroop Sircar - SnugDock: Paratope structural optimization during antibody-antigen docking compensates for errors in antibody homology models

Version 1

      Publication Details (including relevant citation   information):

      Aroop Sircar, Jeffrey J. Gray

      PLoS Comput. Biol. 6(1): e1000644, January 22, 2010.


      High resolution structures of antibody-antigen complexes are   useful for analyzing the binding interface and to make rational   choices for antibody engineering. When a crystallographic   structure of a complex is unavailable, the structure must be   predicted using computational tools. In this work, we illustrate   a novel approach, named SnugDock, to predict high-resolution   antibody-antigen complex structures by simultaneously   structurally optimizing the antibody-antigen rigid-body   positions, the relative orientation of the antibody light and   heavy chains, and the conformations of the six complementarity   determining region loops. This approach is especially useful when   the crystal structure of the antibody is not available, requiring   allowances for inaccuracies in an antibody homology model which   would otherwise frustrate rigid-backbone docking predictions.   Local docking using SnugDock with the lowest-energy   RosettaAntibody homology model produced more accurate predictions   than standard rigid-body docking. SnugDock can be combined with   ensemble docking to mimic conformer selection and induced fit   resulting in increased sampling of diverse antibody   conformations. The combined algorithm produced four medium   (Critical Assessment of PRediction of Interactions-CAPRI rating)   and seven acceptable lowest-interface-energy predictions in a   test set of fifteen complexes. Structural analysis shows that   diverse paratope conformations are sampled, but docked paratope   backbones are not necessarily closer to the crystal structure   conformations than the starting homology models. The accuracy of   SnugDock predictions suggests a new genre of general docking   algorithms with flexible binding interfaces targeted towards   making homology models useful for further high-resolution   predictions.

      Address (URL):