ABSTRACT: Prostate cancer (PrCa) is one of the most common malignancies in Western countries. However, the pathways that promote invasion, especially in late-stage castration-resistant PrCa (CRPC), are poorly understood. Heat shock factors (HSFs) are transcriptional regulators essential for cell survival upon proteotoxic stress, and HSF1 is also identified as a driver of oncogenesis. Here, our aim was to elucidate the molecular mechanisms that contribute to CRPC and invasion. In silico analyses showed that high HSF1 expression levels correlate with poor survival and high Gleason score in PrCa. In the same data set, HSF2, which has not previously been linked to cancer, displayed decreased expression. Using 3-dimensional (3D) organotypic PrCa cultures that spontaneously undergo an invasive conversion, opposite effects of HSF1 and HSF2 were observed; tumor spheroids lacking HSF1 grew slowly, were polarized and devoid of invasive structures, and depletion of HSF2 potentiated invasiveness. In the in vivo xenograft chorioallantoic membrane model, HSF1 silencing caused tumor regression and fibrosis, and knockdown of HSF2 generated large invasive tumors, verifying the results from the 3D-cultures. Gene expression profiling revealed enrichment of genes connected to translational control when either HSF1 or HSF2 was silenced. HSF1-specific targets were associated with progression of tumorigenesis, whereas HSF2 targets were involved in focal adhesion, critical for invasion. This study provides the first evidence for HSF2 functioning in cancer, i.e. as a tumor suppressor. Moreover, human PrCa tissue microarrays demonstrated increased nuclear HSF1 expression, which significantly correlated with high-grade Gleason score. Cytoplasmic HSF1 was detected in a subset of tumors and intriguingly, correlated with decreased disease-specific survival. This was most pronounced in intermediate Gleason score 7 tumors and in metastases. We propose the use of HSF1 as a biomarker to predict PrCa outcome, thus facilitating clinical decision-making and supporting individualized treatment choices. HSF1 and HSF2 siRNA transfection was performed in triplicates (biological replicates), for 5 and 8 days prior to harvesting total RNA; and compared to PC-3 cells transfected with scrambled control siRNAs for 5 and 8 days, respectively.
