Enzyme prodrug therapy has the potential to remedy the lack of selectivity associated with the systemic administration of chemotherapy. However, most current systems are immunogenic and constrained to a monotherapeutic approach. We developed a new class of fusion proteins centered about the human enzyme β-glucuronidase (βG), capable of converting several innocuous prodrugs into chemotherapeutics. We targeted βG to phosphatidylserine on tumor cells, tumor vasculature and metastases via annexin A1/A5. Phosphatidylserine shows promise as a universal marker for solid tumors and allows for tumor type-independent targeting. To create fusion proteins, human annexin A1/A5 was genetically fused to the activity-enhancing 16a3 mutant of human βG, expressed in chemically defined, fed-batch suspension culture, and chromatographically purified. All fusion constructs achieved >95% purity with yields up to 740 μg/l. Fusion proteins displayed cancer selective cell-surface binding with cell line-dependent binding stability. One fusion protein in combination with the prodrug SN-38 glucuronide was as effective as the drug SN-38 on Panc-1 pancreatic cancer cells and HAAE-1 endothelial cells, and demonstrated efficacy against MCF-7 breast cancer cells. βG fusion proteins effectively enable localized combination therapy that can be tailored to each patient via prodrug selection, with promising clinical potential based on their near fully human design.