The androgen receptor transcriptional program in castration-resistant prostate cancer

Abstract

The androgen receptor (AR) is central to the initiation and progression of prostate cancer, even after castration. Its transcriptional activity has previously been studied in cell lines. A group at the University of Cambridge recently outlined the AR transcriptional program in tissue samples, with an emphasis on castration-resistant tumors. AR binding sites, gene-expression changes (in xenografts), and potential transcription factor interactions were notably different from those observed in cultured cells. These discrepancies suggest a distinct signaling network for the AR in vivo and serve as a reminder that results from in vitro models should be checked against clinical realities.

Keywords: :

• androgen receptor
• binding site
• castration-resistant prostate cancer
• transcription factor
• transcriptional program

This article is related to:

Disruption of androgen receptor (AR) signaling is a major goal of prostate cancer (PCa) treatments. The first therapeutic step in metastatic PCa is androgen deprivation therapy (ADT), which works by limiting androgen/AR interaction through surgical or chemical castration. While ADT succeeds in shrinking or slowing tumors, progression to castration-resistant prostate cancer (CRPC) inevitably follows. AR signaling persists despite castrate levels of androgens. The reasons for this continued AR activity are not fully understood; contributing factors are thought to include extratesticular androgen production and constitutive, androgen-independent AR activation.¹ As the disease shifts to CRPC, the AR activates both pre-castration target genes and new ones.

Elucidating the transcriptional program of the AR is important in understanding the receptor’s role in PCa, including CRPC. Multiple studies have described AR binding sites (ARBS) in PCa cells, leading to identification of activated pathways.^2-4 However, these studies used cultured PCa cell lines derived from metastases.⁵ Not only do such cells experience in vitro environments, but they likely have different AR-regulated expression than do primary tumors.^6,7 In a recent paper in Cancer Cell, Sharma et al. studied ARBS from tissue samples (benign, treatment-responsive, and castration-resistant).⁸ Their overall findings suggested in vivo AR signaling is quite different from that observed in cultured cell lines.

Chromatin immunoprecipitation sequencing (ChIP-seq) was used to identify thousands of genome-wide ARBS in prostate tissue samples from 12 patients. Untreated and CRPC tissue samples had much more overlap of ARBS (30%) than untreated tissue and cell lines (3%). More than half of the ARBS found in CRPC tissue were not found in cell lines. The authors also aimed for a more direct look at AR target gene activation in CRPC tissue samples by performing ChIP-seq for H3K4me1/me3, two histone marks enriched at active enhancers and promoters.⁹ Nearly all genes bound by AR in CRPC tissue were also marked by H3K4me1/me3 (95%/89%). In xenograft tumors of LNCaP cells (a commonly used line of androgen-sensitive human prostate cancer cells of metastatic origin), a majority of these genes were differentially regulated following castration, demonstrating in vivo AR regulation. However, 44% of the tissue ARBS genes were not AR-regulated in LNCaPs in vitro.

The authors focused on which specific genes were contributing to the apparent discrepancy between cultured cell and tissue sample transcriptional activity. They monitored gene-expression changes in response to changes in androgen levels. Genes at both cell line ARBS and CRPC tissue sample ARBS were downregulated in xenografts following castration, but the genes located at untreated tissue ARBS showed much less upregulation than the cell line genes in response to in vitro androgen stimulation. Gene expression data from a previous study in primary tumors¹⁰ revealed that cell line genes were not greatly affected by chemical castration. Tissue ARBS, especially those that were not activated in cell lines in vitro, experienced greater gene expression changes. Through these analyses, a subset of in vivo-restricted AR-regulated genes was identified.

Sharma et al. suspected that cellular context could be partly responsible. In vivo, cells might be receiving signals that lead to activation of other pathways and binding of other transcription factors, which could interact with the AR. Motif analysis of the sequences at ARBS was used to identify which other transcription factors could bind at those sites. Here again, the in vitro and in vivo AR-regulated genes diverged, with certain transcription factor sequences found only at in vivo ARBS. Furthermore, AR binding in cultured LNCaP cells was redirected to sites occupied exclusively in PCa tissue by simulating in vivo conditions with a cocktail of cytokines. These changes confirmed the importance of cellular context in determining AR transcriptional patterns.

Certain categories of genes were differentially regulated in cell lines and tissue samples. The Genomic Regions of Enrichment Annotations Tool (GREAT)¹¹ was used to identify which genes were controlled by ARBS found at regulatory domains. ARBS from tissue were more enriched for genes associated with metabolic stress (increased cholesterol and body fat and decreased circulating insulin) while ARBS from cell lines were more often associated with genes that respond to in vitro stimulation by androgens. GREAT analysis also reinforced the concept of a divergent transcriptional program for the AR in vivo—those genes upregulated in tissue samples (and certain metabolic genes) were enriched only near ARBS from CRPC tissue.

As a potential application of these newly-identified CRPC AR targeted genes, Sharma et al. created a 16 gene signature that could be used to make prognoses or monitor the progression of CRPC. These core genes captured the hallmarks of the in vivo-restricted AR-regulated group: overall increased expression in CRPC tissue, downregulation after castration, and reemergence after the transition to CRPC. When pitted against a published 250 gene signature,¹² this more succinct set was better able to identify CRPC. In addition, 4 of the 16 genes were able to predict survival in two clinical data sets.^13,14

This study suggests a couple different directions for future research on the role of the AR in CRPC. The discovery that transcription factors unique to in vivo PCa samples leads to different AR transcriptional activity warrants further studies on transcription factor interplay, both in CRPC and other cancers. Similarly, the redirected AR binding induced by addition of cytokines to cultured LNCaP cells will draw attention to the importance of cytokines in PCa. Researchers might benefit from adding such a cytokine cocktail to in vitro cells in order to better model AR activity. The identification of the specific signaling pathways activated by in vivo-restricted ARBS should lead to research on the role of these pathways in PCa progression. Finally, further investigation of newly identified ARBS might highlight novel therapeutic targets.

PCa cell lines such as LNCaP, DU-145, and PC3 are widely used in prostate cancer research. They work well as in vitro cancer cell models and are relatively easy to manipulate in a lab setting. However, one must remember that these cultured cell lines sometimes behave differently from in vivo CRPC tumor cells. Sharma et al. have demonstrated that something as integral to PCa biology as AR transcriptional activity can be significantly different in cell lines vs. clinical samples. This study reveals new information about the signaling pathways connected to the AR, highlights the importance of cellular context, and cautions that results from established cell lines paint an incomplete picture.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Disclaimer

The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organization imply endorsement by the US Government. The views in this manuscript are those of the authors and may not necessarily reflect NIH policy. No official endorsement is intended nor should be inferred.

10.4161/cbt.27149