Mechanics of CD20 recognition by anti-CD20 antibodies, and its relationship with direct cytotoxicity. (A) RTX and OBZ bind roughly the same epitope on the larger loop of CD20 but, depending on the recognition of Asn76 (OBZ and type II mAbs) or not (RTX and type I mAbs),⁷  the paratope shifts by a few notches, modifying the orientation of the Fab arm (blue arrow) relative to CD20 and the target cell membrane. (B) During the process of humanization of the parental murine mAb of GA101 (OBZ), the white-circled purple residue at position 13 (IMGT unique numbering) was demonstrated to be critical direct cytotoxicity, and must be a valine.⁶  This residue is very close to Phe150 (Eu numbering) in the CH1 domain, and is thought to be responsible for the wider elbow angle of OBZ relative to RTX.⁷  Tightening or loosening the screw responsible for this elbow hinge is therefore another way to influence direct cytotoxicity. (C) Replacing the serine residues 131 (CH1) or 229 (hinge) (Eu numbering) of RTX by cysteines, as in human IgG2, enhances RTX-mediated apoptosis and tends to create conditions of half maximal binding to CD20, as for type II mAbs.²¹  These residues probably create different inter–heavy chain disulfide bridges, rendering the hinge more rigid and also widening the Fab-Fab angle. Here again, tightening or loosening the screw of this hinge is another way to influence direct cytotoxicity.