Cancer chemotherapy is limited by toxicity sustained by normal tissues. To increase specificity of drug delivery antineoplastic agents have been linked to monoclonal antibodies (MABs) that bind to antigens selectively expressed on tumor cells. However, this strategy suffers from several drawbacks. Practical limits on drug loading or slow internalization of drug conjugates into cancer cells may limit tumor cell killing. In addition, cells expressing low levels of antigen in heterogeneous cell populations may escape destruction. In our preliminary work to overcome problems associated with chemoimmunoconjugates we prepared an enzyme-MAB conjugate (MABzyme) by chemically linking beta-glucuronidase to an anti- hepatoma MAB. This MABzyme localized to the target tumor cell surface where it specifically converted a nontoxic glucuronide conjugated prodrug to the parent compound, the potent alkylator, p-hydroxy aniline mustard. Genetically engineered MABzymes can overcome problems encountered in the preparation of MABzymes by standard chemical conjugation methods. In addition utilization of recombinant MAB fragments will increase tissue penetration, blood clearance and activity. Thus the overall goal of the present proposal is to prepare recombinant MABzymes linking a single chain version of an anti-hepatoma MAB to a cloned human beta-glucuronidase. During phase I we will clone and sequence heavy and light chain variable regions of hepatoma specific MAB hep1 and construct a single chain version of this antibody. We will evaluate binding activity of this scFV and construct recombinant MABzyme linking the scFv to cDNA of human beta- glucuronidase. Finally we will express and purify the rMABzyme to evaluate in vitro activity. Using this reagent to activate novel glucuronide prodrugs will provide a highly selective therapy for various forms of human cancer.