Prostate tumor development and androgen receptor function alterations in a new mouse model with ERG overexpression and PTEN inactivation

Abstract

Gene fusions involving ETS transcription factors (predominantly ERG and ETV1) and PTEN deletions are prevalent in the prostate cancer genome. This report describes a novel mouse model that overexpresses ERG and lacks PTEN with the majority of mice developing prostate tumors by 6 mo. Biological mechanisms suggest increased/altered binding of the male hormone receptor in the genome. This model will be useful in pre-clinical evaluation of new drugs targeting these common prostate cancer genomic alterations.

Keywords: :

• prostate
• androgen receptor
• ERG
• PTEN
• mouse model

This article refers to:

The highly prevalent TMPRSS2-ERG fusion is present in approximately half of all prostate cancers in western countries. This gene fusion results in androgen-dependent and prostate tumor-specific expression of the near full-length ERG protein with a 32 amino acid deletion at the N-terminus.^1-3 Studies using ERG knock-down or its overexpression in prostate cancer cell lines showed that ERG induces cell invasion or blocks differentiation.^4,5 Previous studies using transgenic mouse models designed to express the human ERG gene, where ERG expression was selectively induced into a transgenic mouse prostate through the Probasin promoter-driven ERG expression vector, resulted in subtle and variable phenotypes with some showing prostatic intraepithelial neoplasia (PIN) lesions while others did not exhibit any phenotype.^5-8 In cooperation with Pten^+/− loss, the ERG transgenic mice develop PIN or cancerous lesions with variable results.^7,8 A recent publication by Chen et al. published in Nature Medicine described the development and characterization of an engineered mouse model that combined the overexpression of human ERG gene with the complete deletion of the mouse Pten gene in the mouse prostate.⁹ The ERG^+/+/Pten^−/− mice consistently develop prostate cancer within 6–8 mo. Further evaluation of the ERG^+/+/Pten^−/− mice for biological mechanisms highlighted the activation of the androgen receptor (AR) functions through reprogramming of the AR cistrome (genome wide binding sites of the AR transcription factor). These observations were further extended in the context of human prostate cancer with ERG and PTEN gene defects.

Chen et al. generated an ERG knock-in mouse in which the TMPRSS2-ERG gene was inserted at the Rosa 26 locus in a silent configuration. The Rosa26^ERG/ERG mice then expressed ERG in the mouse prostate only after crossing with the Pb-Cre4 mice. The Pb-Cre4 × Rosa26^ERG/ERG mice showed uniform expression of the ERG protein in the dorso-lateral lobes of the prostate by 8 wk and in the anterior lobes by 3 mo. When compared with the wild type (WT) mice, heterozygous (Pb-Cre4 × Rosa26^{ERG/+ )} or homozygous (Pb-Cre4 × Rosa26^ERG/ERG) ERG mice prostates did not show a significant difference in AR protein expression, tissue histology or cell proliferation. Despite the robust ERG protein expression in prostate, these mice did not exhibit precancerous or cancerous lesions, and these observations are similar to several earlier reports. Pten^flox/flox mice that had the inactivation of both copies of the mouse Pten genes developed PIN but did not progress to adenocarcinoma. When the Pb-Cre4 × Rosa26^ERG/ERG mice were crossed with Pten^flox/flox mice, the resulting mice (Pb-Cre4 × Rosa26^ERG/ERG × Pten^flox/flox:ERG^+/+;Pten^−/−) developed highly penetrant PIN lesions by 8 wk, which progressed to invasive adenocarcinoma. Eighty percent of the ERG^+/+;Pten^−/− mice developed prostate adenocarcinoma by 6 mo and these tumors were highly proliferative by Ki-67 (cell proliferation marker protein) staining as compared with PIN lesions. These mouse prostate tumors were ERG-positive, CK 5-positive, p63-negative, and AR-positive. The 12 mo or older mice with the ERG^+/+;Pten^−/− genotype developed poorly differentiated tumors with focal neuroendocrine differentiation similar to Gleason grade 5 human prostate cancers. The ERG^+/+;Pten^−/− mice showed significantly shorter survival in comparison to Pten^−/− mice. These data are different from the previous transgenic mice models showing variable rate of PIN or adenocarcinoma in ERG transgenic mice with a Pten heterozygous deletion.

Having developed this promising mouse model, the authors evaluated the biological mechanisms of prostate cancer development in ERG^+/+;Pten^−/− mice. Qualitative and quantitative features of ERG and AR transcription factor binding were analyzed in the genomes of mouse prostates with ERG^+/+, Pten^−/−, or ERG^+/+;Pten^−/− genotypes. An unexpected and intriguing finding of this study showed four times more AR binding peaks in the prostate genome of the Rosa26 ^ERG/ERG mice (14 889) in comparison to the WT mice (3476). This observation was not due to increased levels of the AR protein or circulating testosterone levels in ERG^+/+ mice. Evaluations of the AR binding sites revealed the presence of pre-existing sites in WT mice and additional sites in the ERG^+/+ mice. Since a large percent of new AR sites (77%) mapped to genes containing ERG sites, these observations suggested that ERG functions as a pioneering factor whose field effects reprograms AR binding.

The authors made another intriguing finding by showing that ERG induces only a small number of transcriptional changes in ERG^+/+ mice prostates (expression of only 20 genes changed by 1.5-fold) in comparison to WT mice. In contrast ERG^+/+;Pten^−/− mice prostates exhibited a large number of ERG-dependent expression changes (expression of 800 genes changed by 1.5-fold) when compared with Pten^−/−. Closer evaluations of the data showed that overall magnitude of the ERG mediated gene expression changes was amplified in ERG^+/+;Pten^−/− in comparison to Pten^−/− mice. To assess the similarity of the ERG-dependent gene expression changes in ERG^+/+;Pten^−/− mice and ERG-positive human cancers, the authors used gene set enrichment analyses (GSEA) to evaluate existing gene expression data sets. These sets consisted of upregulated genes from a comparison ERG-positive and ERG-negative prostate cancers and downregulated genes in response to ERG knockdown in VCaP cells. This analysis showed a similarity of ERG regulated genes that also had a similar expression relationship of two selected genes, AMPD3 and TFF3, with ERG expression between mouse model and human prostate cancers. Functional assessment of genes altered by ERG in ERG^+/+;Pten^−/− mice included genes involved in inflammation, cell migration, and angiogenesis. Using the GSEA analyses, Chen et al. showed that AR regulated genes were enriched in ERG^+/+;Pten^−/− mice by comparing downregulated genes in response to castration in mouse prostates or the androgen induced genes in VCaP cells. Three androgen regulated genes (Pbsn, Nkx3.1, and Msmb), were shown to be upregulated even after castration in ERG^+/+;Pten^−/− mouse prostate in contrast to Pten^−/− mouse prostate. Analysis of human prostate tumors in the context of PTEN loss exhibited decreases in AR target gene signature, which was partially restored in tumors harboring ETS fusion and PTEN loss.

In summary this study reports a new and robust mouse model of prostate cancer with ERG overexpression and Pten deletions, two genes that are frequently altered in human prostate cancer. The reprogramming of AR binding sites by ERG or ETV1 in the presence of PTEN loss is an intriguing novel observation. The results of this study showing enhancement of the AR function by ERG contrasts some previous reports showing attenuation of AR function by ERG in a VCaP cell line model.^6,10 These differences may be due to the differences of biological and experimental contexts in which ERG functions have been studied. This study underscores a possible need for biological stratification based treatment strategies for prostate cancer. For example, ERG fusion-positive and PTEN-negative human prostate cancers with enhanced AR functions may need to include additional therapeutic strategies (chemotherapy or vaccine) earlier in treatment. The ERG^+/+;Pten^−/− mice described here provide a useful animal model for pre-clinical evaluations of the next generation of AR inhibitors. This model also offers new opportunities for testing novel agents targeting ERG activation or PTEN inactivation in prostate cancer.

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.