Molecular biology of prostate cancer progression

Prostate cancer is now the most common form of cancer and the second leading cause of cancer deaths in American men (Boring C.C. et al, CA 44:7-26, 1994). As with other forms of cancer, prostate cancer is a multistep disease process that involves the acquisition of multiple genetic alternations (Armitage P and Doll K, Br J Cancer 8:1-12, 1954). For prostate cancer, alternations in specific dominantly acting oncogenes including ras and myc and tumor suppressor genes including p53 and Rb have been reported. However, a simple phenotype-genotype correlation for prostate cancer progression may not be readily accessible because prostate cancer demonstrates remarkable genetic heterogeneity. Recent clinical data indicate that this heterogeneity exists both among the multiple cancer foci as well as within individual cancer foci. Furthermore, based on chromosomal analysis, it has been suggested that metastases do not necessarily seed from the largest index cancer focus at the primary site. Such observations imply that abrupt changes in gene expression may trigger metastatic behavior in relatively small cohorts of malignant cells present at the local site. This pattern of progression may result from compromised function of specific 'control' genes which could affect the activity of multiple downstream genes involved in specific pathways of malignant progression. Such a mechanistic framework involving networks of gene expression could explain the acquisition of the complex metastatic phenotype. Using the mouse prostate reconstitution (MPR) model system (Thompson et al, Cell 56:917-930, 1989) we demonstrated that progression of experimental prostate cancer to metastasis was invariably associated with functional inactivation of p53 (Thompson el al, Oncogene 10:869-879, 1995). Southern blotting analyses revealed that metastases do not necessarily originate from the most abundant clone in the primary carcinoma. Furthermore, the role of p53 as a potential metastasis suppressor 'control' gene in human prostate cancer was supported by studies using molecular biological and immunohistochemical techniques (Eastham et al, Clin Cancer Res 1:1111-1118, 1995 and Yang et al, Clin Cancer Res 2:399-401, 1996). Another possible ''control'' gene related to prostate cancer metastases may be the gene which encodes TGF-β1. We have previously shown that overexpression of TGF-β1 is associated with mouse and human prostate cancer and occurs predominantly in metastatic disease (Eastham et al, Lab Invest 73:628-635, 1995). To investigate a possible role of TGF-β1 in metastatic progression, we compared growth and extracellular matrix responses to TGF-β1 in six metastatic and six primary tumor cell lines derived from our metastatic mouse prostate cancer model system. The results indicated that tumor cell lines derived from focal pulmonary metastases secrete greater quantities of total TGF-β's and have lost most or all TGF-β1 growth inhibition, but respond to TGF-β1 through induction of type IV collagenase, matrix metalloproteinase-9. Cell lines derived from primary site tumors retain TGF-β1 growth inhibition, but lack TGF-β1-induced collagenase activity. Our results indicate that the elimination and/or subversion of TGF-β1 responsive pathways should be considered a mechanistic framework for metastatic events (Sehgal et al., Cancer Res 56:3359-3365, 1996). Both p53 and TGF-β1 can regulate the expression of downstream genetic targets, therefore, we are currently pursuing a strategy using differential display-polymerase chain reaction to elucidate additional changes in gene expression resulting from loss and/or subversion of function for these two putative ''control'' genes in prostate cancer metastasis. Hopefully, identification of these target genes will lead to greater understanding of the mechanisms of prostate cancer metastasis and possibly provide novel therapeutic targets