High-throughput antibody humanization

摘要

The present invention relates to improved methods for antibody engineering, e.g., humanization. In particular, the disclosure provides a high-throughput antibody humanization process that can be automated by computer-implementation.

主权项

1. A method of producing a humanized variant of a non-human donor immunoglubin comprising the steps of:
(i) providing a collection of all possible human immunoglobulin light chain variable region (VL) sequences or heavy chain variable region (VH) sequences with the same CDR canonical structures and Kabat subgroup assignment as the VL or VH sequence of the non-human donor immunoglobulin; (ii) for each acceptor sequence in the collection, identifying the number of non-identical residues at all framework region (FR) positions between the donor sequence and each acceptor sequence; (iii) for each acceptor sequence in the collection, identifying the number of non-identical residues at key CDR positions (“strcdr”) between the donor sequence and each acceptor sequence; (iv) ranking the acceptor sequences in the collection based on a preference score (“diff”) which is a sum of the number of non-identical residues identified in step (ii) and (iii) for each acceptor sequence; (v) selecting the acceptor sequence in the collection with the lowest preference score and the lowest backmutation score (“fr_bm”); (vi) synthesizing a DNA segment encoding a humanized VL or VH sequence comprising CDRs from the donor immunoglobulin engrafted in the variable region framework from the selected acceptor sequence; and comprising key FR amino acids from the donor immunoglobulin that replace non-identical amino acids at corresponding amino acid positions in the acceptor variable region framework; (vii) introducing the DNA segment encoding the humanized VL or VH sequence and a DNA segment encoding a corresponding humanized VH or VL sequence into a cell; and (viii) expressing the DNA segments in the cell, wherein the lowest backmutation score is established by: (ix) providing a structural model of the donor immunoglobulin sequence; (x) identifying all key FR residues in the VL or VH sequence of the non-human donor immunoglobulin which are (a) within about 5 Angstroms of the CDRs of immunoglobulin and have a solvent exposure of less than 20% (“buried”); or (b) involved in the interaction with the CDR or another chain by specific hydrophobic, electrostatic or ion-ion chain interactions (“strltd”); and (xi) for each acceptor sequence in the collection, identifying the number of non-identical residues at the key FR positions of step (x) between the donor sequence and each acceptor sequence to establish a total backmutation score (“fr_bm”) for each acceptor sequence; (xii) ranking the acceptor sequences in the collection based on the backmutation score; (xiii) identifying the acceptor sequence in the collection with the lowest backmutation score, thereby producing a humanized variant of a non-human donor immunoglobulin.