Engineering stable and non-immunogenic immunoenzymes for cancer therapy via in situ generated prodrugs.

Engineering human enzymes for therapeutic applications is attractive but introducing new amino acids may adversely affect enzyme stability and immunogenicity. Here we used a mammalian membrane-tethered screening system (ECSTASY) to evolve human lysosomal beta-glucuronidase (hBG) to hydrolyze a glucuronide metabolite (SN-38G) of the anticancer drug irinotecan (CPT-11). Three human beta-glucuronidase variants (hBG3, hBG10 and hBG19) with 3, 10 and 19 amino acid substitutions were identified that display up to 40-fold enhanced enzymatic activity, higher stability than E. coli beta-glucuronidase in human serum, and similar pharmacokinetics in mice as wild-type hBG. The hBG variants were two to three orders of magnitude less immunogenic than E. coli beta-glucuronidase in hBG transgenic mice. Intravenous administration of an immunoenzyme (hcc49-hBG10) targeting a sialyl-Tn tumor-associated antigen to mice bearing human colon xenografts significantly enhanced the anticancer activity of CPT-11 as measured by tumor suppression and mouse survival. Our results suggest that genetically-modified human enzymes represent a good alternative to microbially-derived enzymes for therapeutic applications.