Potential use of an estrogen-glucocorticoid receptor chimera as a drug screen for tissue selective estrogenic activity.

SERMs act as ER agonists in bone despite their antagonistic properties in other tissues. As well as inhibiting bone remodelling, this effect may involve stimulation of osteoblast activity, in light of evidence from recent in vivo studies. However, progress in exploring this action has been hampered by a lack of accurate in vitro models. For example, ER antagonists are reported to stimulate reporter assays based on estrogen target genes in osteoblasts, contrary to their inhibitory effects in vivo. We examined whether evaluating global aspects of ER function provides a more accurate reflection of ER activation in osteoblasts, based on the use of morphological and/or transcriptional read-outs with green fluorescent protein (GFP)-receptor chimeras. Osteoblast-like (ROS and U2OS) and breast cancer (MCF7) cells were transfected with a human ERalpha-GFP fusion protein, and treated with ER agonists (17beta-estradiol, and dienestrol), antagonists (ICI 182,780 and ZK 164015) and SERMs (tamoxifen, raloxifene, 4-hydroxytamoxifen (4-HT) and hexestrol). We investigated cellular compartmentalisation of these constructs by fluorescence microscopy, nuclear mobility by fluorescence recovery after photobleaching (FRAP), and global activation of estrogenic transcription using a ERE-luc reporter. SERMs caused a modest increase in ERE-luc activity in osteoblast-like cells (but not in breast cells), and a reduction in nuclear mobility in breast (but not osteoblast-like) cells. These studies were then repeated using a GFP chimera where the human GR ligand binding domain (LBD) was replaced by the human ERalpha LBD (ERGR-GFP), combined with a GRE-luc reporter. Interestingly, SERMs increased both cytoplasmic to nuclear translocation of ERGR-GFP, and GRE-luc reporter activity, in osteoblast-like (but not breast) cells. Indeed, transcriptional responses to SERMs in osteoblast-like cells were considerably greater with the ERGR/GRE-luc than the ERalpha/ERE-luc system, 4-HT inducing 300 and 25% increases in reporter activity respectively. ER antagonists were entirely without effect. We conclude that evaluation of global estrogenic activity, as opposed to activation of a specific target gene, provides a more accurate read-out for osteoblast stimulation. In particular, ERGR-mediated GRE-luc activity provides a high signal response to estrogen agonists and SERMs, in a cell context dependent manner closely resembling that observed in vivo. Further studies utilising this system are justified to explore the mechanistic basis for estrogenic stimulation of osteoblast activity, and to identify newer SERMs capable of targeting this activity.