Genome-wide analysis of AR binding and comparison with transcript expression in primary human fetal prostate fibroblasts and cancer associated fibroblasts

C Nash, N Boufaied, IG Mills, OE Franco… - Molecular and cellular …, 2018 - Elsevier
C Nash, N Boufaied, IG Mills, OE Franco, SW Hayward, AA Thomson
Molecular and cellular endocrinology, 2018Elsevier
The androgen receptor (AR) is a transcription factor, and key regulator of prostate
development and cancer, which has discrete functions in stromal versus epithelial cells. AR
expressed in mesenchyme is necessary and sufficient for prostate development while loss of
stromal AR is predictive of prostate cancer progression. Many studies have characterized
genome-wide binding of AR in prostate tumour cells but none have used primary
mesenchyme or stroma. We applied ChIPseq to identify genomic AR binding sites in primary …
Abstract
The androgen receptor (AR) is a transcription factor, and key regulator of prostate development and cancer, which has discrete functions in stromal versus epithelial cells. AR expressed in mesenchyme is necessary and sufficient for prostate development while loss of stromal AR is predictive of prostate cancer progression. Many studies have characterized genome-wide binding of AR in prostate tumour cells but none have used primary mesenchyme or stroma.
We applied ChIPseq to identify genomic AR binding sites in primary human fetal prostate fibroblasts and patient derived cancer associated fibroblasts, as well as the WPMY1 cell line overexpressing AR. We identified AR binding sites that were specific to fetal prostate fibroblasts (7534), cancer fibroblasts (629), WPMY1-AR (2561) as well as those common among all (783). Primary fibroblasts had a distinct AR binding profile versus prostate cancer cell lines and tissue, and showed a localisation to gene promoter binding sites 1 kb upstream of the transcriptional start site, as well as non-classical AR binding sequence motifs.
We used RNAseq to define transcribed genes associated with AR binding sites and derived cistromes for embryonic and cancer fibroblasts as well as a cistrome common to both. These were compared to several in vivo ChIPseq and transcript expression datasets; which identified subsets of AR targets that were expressed in vivo and regulated by androgens. This analysis enabled us to deconvolute stromal AR targets active in stroma within tumour samples.
Taken together, our data suggest that the AR shows significantly different genomic binding site locations in primary prostate fibroblasts compared to that observed in tumour cells. Validation of our AR binding site data with transcript expression in vitro and in vivo suggests that the AR target genes we have identified in primary fibroblasts may contribute to clinically significant and biologically important AR-regulated changes in prostate tissue.
Elsevier