Our laboratory has developed two cellular models of human prostate cancer progression. The LNCaP prostate cancer progression model is based upon the well-known cellular interaction between human prostate or bone stromal cells and LNCaP cells in vivo. The marginally tumorigenic LNCaP cells acquired tumorigenic and metastatic potential upon cellular interaction with either prostate or bone fibroblasts. A subline termed C4-2 was observed to grow readily in castrated animals and acquired metastatic potential spreading from the primary tumor site to the lymph node, the seminal vesicles, and the axial skeleton, resulting in an intense osteoblastic reaction. The second model is ARCaP, where prostate cancer cells derived from the ascites fluid of a man with metastatic disease exhibited an Androgen- and estrogen-Repressed Prostate Cancer cell growth and tumor formation in either a hormone-deficient or a castrated environment. However, the growth of either the tumor cells in vitro or the tumors in vivo was suppressed by both estrogen and androgen. While the tumor cells expressed low levels of androgen receptor and prostate-specific antigen (PSA), they were highly metastatic when inoculated orthotopically. Distant metastases to a number of organs were detected, including the liver, lung, kidney, and bone. We have employed a human prostate cancer progression model as a system to study the efficacy of gene therapy. Results of the study show that whereas universal promoters, such as Cytomegalovirus (CMV) and Rous Sarcoma Virus (RSV) promoter-driven tumor suppressors (e.g. p53, p21, and p16), were effective in inhibiting prostate tumor growth, the advantages of driving the expression of therapeutic toxic genes using a tissue-specific promoter prostate-specific antigen (PSA) and a tumor--but not tissue-specific promoter, osteocalcin (OC), are preferred. In the case of the PSA promoter, we can achieve cell-kill in PSA-producing human prostate cancer cells. To circumvent the supporting role of bone stroma for prostate cancer epithelial growth, we have recently developed a novel concept where the expression of therapeutic toxic genes is driven by a tumor--but not a tissue-specific OC promoter. Osteocalcin-thymidine kinase (OC-TK) was found to efficiently eradicate the growth of osteosarcoma, prostate, and brain tumors both in vitro and in vivo. We observed that androgen-independent human prostate cancer cells lines expressed OC-TK at higher levels than androgen-dependent human prostate cancer cell lines. We have obtained data to suggest that Ad-OC-TK plus a pro-drug acyclovir (ACV) may be used as an effective therapy to treat prostate cancer bone metastasis in models where the growth of androgen-independent PC-3 and C4-2 tumors in the bone has occurred.