In vitro and in vivo models for evaluation of GDEPT: quantifying bystander killing in cell cultures and tumors.

Abstract

The vectors currently available for gene therapy of cancer rarely achieve expression of therapeutic genes in more than a small fraction of the cells in solid tumors. This makes therapeutic success critically dependent on secondary events, known as bystander effects, by which transgene expression leads to the death of nontransduced tumor cells. An efficient bystander effect has the potential to compensate for spatially nonuniform expression of therapeutic genes, and its optimization is therefore an important goal in gene therapy of cancer. Here, we describe protocols for quantifying bystander effects using in vitro and in vivo experimental models. The generation of efficient bystander killing is the key rationale for gene-dependent enzyme–prodrug therapy (GDEPT), in which a suicide gene codes for a prodrug-activating enzyme (PAE) that generates a diffusible cytotoxin. In a well-studied example, the prodrug ganciclovir (GCV) is activated by herpes simplex virus-thymidine kinase (HSV-TK) to form a phosphorylated metabolite that diffuses between cells via gap junctions (1,2). In other cases, the bystander metabolite can diffuse across cell membranes, as for 5-fluorouracil (5-FU) generated from 5-fluorocytosine (5-FC) by bacterial or yeast cytosine deaminase (CD) (3,4). Several newer GDEPT systems generate freely diffusible (membrane-permeable) alkylating metabolites with the added attractive feature of killing noncycling as well as cycling tumor cells; these include oxidation of cyclophosphamide and other oxazaphosphorines by cytochrome P450 2B1/cytochrome P450 reductase (5), hydrolysis of glutamate derivatives of aromatic nitrogen mustards (prodrugs CMDA and ZD2767P) by Pseudomonas carboxypeptidase G2 (6,7), and reduction of the dinitrobenzamide CB 1954 (see Fig. 1) to its 4-hydroxylamine by the Escherichia coli aerobic nitroreductase NTR (8–10). The CB 1954/NTR