In antibody-targeted enzyme prodrug therapy, a monoclonal antibody (mAb) covalently linked to an enzyme is commonly exploited to concentrate the enzyme on the tumor cell surface prior to administration of a relatively nontoxic prodrug. The tumor-localized enzyme then converts the prodrug into a cytotoxic agent, which in turn diffuses into the tumor causing localized cell death. In this paper, we have substituted folic acid for the mAb as a mean of delivering an attached enzyme, penicillin-V amidase (PVA), to folate receptor (FR)-positive tumor cells. The enzyme PVA is capable of converting a doxorubicin-N-p-hydroxyphenoxyacetamide prodrug (DPO) into its potent parent drug, doxorubicin. For PVA targeting, each PVA molecule was covalently labeled with three molecules of folic acid via the formation of amide bonds. In vitro binding assays showed that folate-PVA-125I conjugates bind specifically to KB cells (FR-positive tumor cells) but not to A549 cells (FR-negative tumor cells). Moreover, in a series of in vitro cytotoxicity tests, folate-PVA conjugates were found to kill folate receptor positive but not receptor negative cells, and when bound to FR-positive cells, folate-PVA conjugates rendered the DPO prodrug as toxic as free doxorubicin (IC50, approximately 0.6 microM). Finally, preliminary in vivo plasma clearance studies in normal mice revealed that i.v. administered folate-PVA-125I and PVA-125I are both cleared from the blood within a 24 h time period, removing concern that nonspecifically trapped folate-PVA might activate prodrug in nontargeted tissues. In view of the fact that only a small number of folate-PVA molecules are required to mediate killing of target cells in vitro, these data argue that folate-targeted enzyme prodrug therapy should be considered for tumor eradication in vivo.