Comprehension characterization of prostate cancer cell lines identified JAK-STAT3 signaling in lineage-switched non-neuroendocrine cells.

Abstract

e17005 Background: Prostate cancer (PCa) is the most diagnosed cancer type among American men. Androgen receptor (AR) targeted therapy as the gold standard of PCa has been challenged by treatment resistance. Lineage plasticity enables PCa cells to transit from prostate luminal epithelia to alternative lineages to bypass their dependency on AR signaling. Previously we established prostatic HOX code to study the prostatic lineage. We found that cells with neuroendocrine (NE) phenotype have a lineage switch. Besides, a subgroup of AR negative, NE negative samples also display lineage switch, among which is DU145, a human PCa cell line. Methods: RNA-seq data of PCa cell lines were obtained from GEO then aligned and quantified using human GRCh38 genome with STAR-Salmon pipeline. Downstream differential expression and enrichment analysis was performed with R. Immunohistochemistry (IHC) was used to assess protein levels on cell line derived xenografts. Results: 45 RNA-seq datasets of 6 prostate adenocarcinoma cell lines (VCaP, LNCaP, C4-2, 22RV1, PC3 and DU145) and 8 RNA-seq datasets of neuroendocrine prostate cancer cell line (H660) passed quality control. Principle component analysis indicates that samples from the same cell line had similar transcriptomes. Differential expression analysis followed by gene function enrichments analysis identified the marker genes and key signaling pathways of each cell line. HOX code analysis confirmed previous conclusion that all PCa cell lines but DU145 and H660 samples maintained prostatic HOX code. IHC staining on cell line-derived xenografts confirmed the protein expression of selected marker genes (TUBB2B for H660, HGF for VCaP and 22RV1, CXCL1 and KRT7 for DU145 and PC3). The three AR-negative PCa cell lines were further characterized in terms of AR/ NE activity scores, transcriptome identities, HOX codes. These analyses demonstrated that PC3 is a prostate progenitor like cell line, H660 is a lineage-switched NE cell line, and DU145 is a lineage-switched non-NE cell line. Further, signaling enrichment analysis revealed that DU145 cells had higher JAK-STAT3 signaling activity compared to other cell lines. IHC staining detected strong p-STAT3 signals on DU145-derived tumors but not on any other cell line-derived tumors. Conclusions: Human PCa cell lines were fully characterized based on their transcriptomes with high sample numbers. PC3 represents cells dedifferentiated into progenitor stage. DU145 and H660 represent cells that further trans-differentiate into alternative lineages. Neuron master regulators drive H660 cells into NE lineage and STAT3 drives DU145 cells into a non-NE lineage. DU145, representing a non-NE PCa class that have lost prostatic lineage, would be a suitable model system to study JAK-STAT targeting therapy.