Conversion of murine antibodies to human antibodies and their optimization for ovarian cancer therapy targeted to the folate receptor.

We previously developed murine and chimeric antibodies against a specific epithelial ovarian carcinoma (EOC) marker, named folate receptor (FR), and promising results were obtained in phase II trials. More recently, we successfully generated a completely human Fab fragment, C4, by conversion of one of the murine anti-FR antibodies to human antibody using phage display and guided selection. However, subsequent efforts to obtain C4 in a dimer format, which seems especially desirable for EOC locoregional treatment, resulted in a highly heterogeneous product upon natural dimerization and in a very poor production yield upon chemical dimerization by a non-hydrolyzable linker to a di-Fab-maleimide (DFM). We therefore designed, constructed and characterized a large Fab dual combinatorial human antibody phage display library obtained from EOC patients and potentially biased toward an anti-tumor response in an effort to obtain new anti-FR human antibodies suitable for therapy. Using this library and guiding the selection on FR-expressing cells with murine/human antibody chains, we generated four new human anti-FR antibody (AFRA) Fab fragments, one of which was genetically and chemically manipulated to obtain a chemical dimer, designated AFRA-DFM5.3, with high yield production and the capability for purification scaled-up to clinical grade. Overall affinity of AFRA-DFM5.3 was in the 2-digit nanomolar range, and immunohistochemistry indicated that the reagent recognized the FR expressed on EOC samples. (131)I-AFRA-DFM5.3 showed high immunoreactivity, in vitro stability and integrity, and specifically accumulated only in FR-expressing tumors in subcutaneous preclinical in vivo models. Overall, our studies demonstrate the successful conversion of murine to completely human anti-FR antibodies through the combined use of antibody phage display libraries biased toward an anti-tumor response, guided selection and chain shuffling, and point to the suitability of AFRA5.3 for future clinical application in ovarian cancer.