The estrogen receptor (ER) and androgen receptor (AR) are nuclear receptor (NR) transcription factors that can drive breast and prostate cancer growth, respectively, through cancer-specific DNA binding. In breast cancer, co-stimulating ER and progesterone receptor (PR) causes a PR-mediated relocation of ER to genetic regions associated with tumour suppression, including those that promote a more differentiated phenotype. ER reprogramming through activation of other NR family members is less well understood. Whether AR can be reprogrammed to promote differentiation in prostate cancer is not known. We screened over 3,000 NR small molecules, with the potential to reprogram ER or AR in breast and prostate cancer. Lead-like molecules were computationally selected based on 3D pharmacophore fingerprints with predicted activity against various NRs. Proliferation and morphology (as a readout of cancer cell differentiation) were assessed in 4 breast and 4 prostate cancer cell lines that encompass endocrine therapy-sensitive and -resistant disease. Cells were treated for 3 days and stained with DAPI (nuclear), CFMDA (whole-cell) and phalloidin (cytoskeleton) by immunofluorescence (IF). “Hits” were selected as (a) >30% growth inhibition (nuclear count) and (b) morphological changes measured by Mahalanobis distance (mp.value <0.01). Mahalanobis is a quantitative computation of phenotypic features incorporating various cellular components derived from nuclear, whole-cell and cytoskeleton IF staining. Supervised (partial least square; PLS) and un-supervised (hierarchical) clustering of features grouped library “hits” based on their deviation from vehicle treatment. PLS analysis distinguished library NR ligands (e.g. AR versus ER versus PR) against known NR ligands.
Our high-throughput screening pipeline represents an innovative tool to integrate growth inhibition with distinct morphology-based changes associated with differentiation to discover new therapeutic agents. Validation using 3D drug-screening, live-cell imaging and ChIP-seq analyses will determine whether “hits” also elicit ER or AR reprogramming. Lead “hits” will be tested using patient derived pre-clinical models with the aim of progressing an ER and AR reprogramming agent for treatment of breast and prostate cancer into the clinic.