Abstract
The identification and validation of biomarkers for clinical applications remains an important issue for improving diagnostics and therapy in many diseases, including prostate cancer. Gene expression profiles are routinely applied to identify diagnostic and predictive biomarkers or novel targets for cancer. However, only few predictive markers identified in silico have also been validated for clinical, functional or mechanistic relevance in disease progression. In this study, we have used a broad, bioinformatics-based approach to identify such biomarkers across a spectrum of progression stages, including normal and tumor-adjacent, premalignant, primary and late stage lesions. Bioinformatics data mining combined with clinical validation of biomarkers by sensitive, quantitative reverse-transcription PCR (qRT-PCR), followed by functional evaluation of candidate genes in disease-relevant processes, such as cancer cell proliferation, motility and invasion. From 300 initial candidates, eight genes were selected for validation by several layers of data mining and filtering. For clinical validation, differential mRNA expression of selected genes was measured by qRT-PCR in 197 clinical prostate tissue samples including normal prostate, compared against histologically benign and cancerous tissues. Based on the qRT-PCR results, significantly different mRNA expression was confirmed in normal prostate versus malignant PCa samples (for all eight genes), but also in cancer-adjacent tissues, even in the absence of detectable cancer cells, thus pointing to the possibility of pronounced field effects in prostate lesions. For the validation of the functional properties of these genes, and to demonstrate their putative relevance for disease-relevant processes, siRNA knock-down studies were performed in both 2D and 3D organotypic cell culture models. Silencing of three genes (DLX1, PLA2G7 and RHOU) in the prostate cancer cell lines PC3 and VCaP by siRNA resulted in marked growth arrest and cytotoxicity, particularly in 3D organotypic cell culture conditions. In addition, silencing of PLA2G7, RHOU, ACSM1, LAMB1 and CACNA1D also resulted in reduced tumor cell invasion in PC3 organoid cultures. For PLA2G7 and RHOU, the effects of siRNA silencing on proliferation and cell-motility could also be confirmed in 2D monolayer cultures. In conclusion, DLX1 and RHOU showed the strongest potential as useful clinical biomarkers for PCa diagnosis, further validated by their functional roles in PCa progression. These candidates may be useful for more reliable identification of relapses or therapy failures prior to the recurrence local or distant metastases.