BONE EXTRACELLULAR-MATRIX INDUCES HOMEOBOX PROTEINS INDEPENDENT OF ANDROGENS - POSSIBLE MECHANISM FOR ANDROGEN-INDEPENDENT GROWTH IN HUMAN PROSTATE-CANCER CELLS

BACKGROUND. Differences in gene expression in prostate cells are belie ved to be secondary tb epithelial-stromal interactions. We theorized t hat bone matrix may provide a fertile ''soil'' for prostate cancer by inducing androgen-dependent genes and allowing for androgen-independen t growth. METHODS. Human prostate cancer cells (LNCaP) were grown unde r different conditions and analyzed for differential expression of mRN A. LNCaP cells were grown in the presence of 10 nM dihydrotestosterone (DHT), on extracellular matrix (ECM) derived from bone cells (without exogenous DHT), and on plastic culture dishes without exogenous DHT. A differential display of mRNA produced by LNCaP cells grown in the ab ove conditions was then analyzed. RESULTS. Multiple unique transcripts were present in cells that were grown in the presence of DHT and on b one ECM (without exogenous DHT), but not on plastic culture dishes wit hout exogenous DHT. Nine of these transcripts were then cloned and ana lyzed. Many (5/9) of these transcripts were found to contain multiple ATTA motifs in their corresponding 3'-untranslated regions. ATTA motif s have been shown to be homeobox protein-binding sites. Homeobox prote ins and their target genes are thought to regulate cellular differenti ation. Consistent with this, we demonstrated by reverse transcription polymerase chain reaction (PCR) that homeobox genes were differentiall y expressed in LNCaP cells when the cells were grown in the presence o f DHT and on bone ECM (without exogenous DHT), but not on plastic cult ure dishes without exogenous DHT. Furthermore, we assayed LNCaP/fetal fibroblast chimeric tumors (n = 8) that were grown in male nude mice. Some of those tumors continued to grow in these mice despite treatment with surgical castration. In blinded studies, we were able to determi ne which tumor samples were androgen independent by their expression o f homeobox genes. All samples that were androgen independent (n = 4) e xpressed the homeobox genes. Finally, gel retardation assay demonstrat ed that the homeobox proteins were able to bind to our cloned DNA sequ ences. Furthermore, footprinting analysis showed that the homeobox pro teins bound to the ATTA motif in the 3'-region of our target DNA. CONC LUSIONS. Bone ECM, in the absence of DHT, has the ability to regulate androgen-responsive genes. Furthermore, many of these genes contain ho meobox binding sites and the expression of homeobox genes may itself b e regulated by bone ECM. if so, this may partially explain the clinica l observation that bone provides a fertile ''soil'' for prostate cance r growth and metastasis. (C) 1996 Wiley-Liss, Inc.