Understanding the temporal sequence of genetic events that lead to prostate cancer progression and metastasis

Prostate cancer is the second most common cancer among men, with an estimated 903,000 newly diagnosed cases and 258,000 deaths per year worldwide (1). Although the prognosis is favorable in patients with localized disease, the 5-y survival rate drops to less than 30% for individuals with advanced, metastatic tumors (2). Although androgen deprivation therapy in the form of chemical or surgical castration is currently used to treat metastatic disease, response to treatment is usually temporary, and a significant percentage of patients invariably progress to castration-resistant prostate cancer (CRPC), which often arises due to micrometastases (3). The molecular mechanisms that contribute to tumor development, recurrence, and resistance to standard therapies are poorly understood, making it extremely difficult to predict which patients with treatable, early-stage prostate cancer will progress to lethal forms of the disease. Fortunately, global and computational analyses of the prostate cancer genome have provided us with part of the foundation needed to tackle this issue. One of the first genomewide studies was published in 2005 (4) and showed that ∼50% of all prostate tumors contain chromosomal rearrangements that involve the ETS family of transcription factors, primarily in the form of the TMPRSS2-ERG translocation, which fuse the ETS family member ERG with the androgen-regulated TMPRSS2 gene. This and other translocations involving ETS family genes are present even at early stages of the disease and are specific to prostate cancer (4). However, as the disease progresses and androgen receptor (AR) dependence is lost, these genetic lesions cooperate with signaling pathways, such those mediated by pRB, RAS/RAF, and phosphatidylinositol 3-kinase (PI3K)/PTEN/AKT, which are altered in a more “global” sense to promote metastasis (5). In PNAS, Aytes et al. (6) describe the temporal progression and the consequence of molecular activation of the PI3K and RAS signaling pathways during prostate cancer metastasis. Using a unique genetically engineered mouse model of prostate cancer that allows lineage tracing, the authors demonstrate that the spread of tumor cells is an early event that actually occurs well before the onset of metastasis. They further show that metastasis is temporally coincident with the expression of ETV4, a member of the PEA3 subfamily of ETS transcription factors that mediates late events associated with metastasis (Fig. 1).

Fig. 1.

Schematic representation of prostate cancer progression. Genes and pathways that are implicated at each stage are indicated.

Prior studies have demonstrated that the PI3K and RAS pathways are altered in ∼40% of primary prostate tumors, yet this number jumps to more than 90% in cases of metastatic disease (5). Both pathways often cooperate in the control of numerous cellular processes, including proliferation, survival, and metabolism of normal cells. In prostate cancer, PI3K activation is largely mediated by loss of function of the PTEN phosphatase, although there are anomalies seen in either the regulation or expression levels of other proteins that are part of this pathway (3). Loss of PTEN in mouse models results in the development of precancerous lesions in the prostate and, when combined with other mutations, leads to invasive carcinomas (7–9). Furthermore, PI3K activation correlates with decreased AR-mediated signaling, suggesting a key role for this pathway in castration-resistant tumors (8). Accordingly, in humans, prostate cancer in the absence of PTEN is associated with recurrence, shortened times to metastasis, resistance to therapy, and an overall poorer prognosis (3). This is likely due to the fact that androgen-resistant growth, as shown in mouse models, is inherent to PTEN-null prostate tumor cells, regardless of tumor stage (7).

Despite the fact that RAS/RAF signaling is also activated in prostate cancer, far less is known about its involvement in disease initiation and progression. Compared with PTEN levels that are highest in benign prostatic hyperplasia and untreated prostate cancers, phospho-MAPK (pMAPK), which reflects activation of the RAS pathway, increases significantly in neoadjuvant-treated, recurrent, and CRPC patients. Interestingly, pMAPK levels are also comparatively elevated in metastatic bone lesions, which is in stark contrast to that of PTEN (10). These data indicate that both pathways are activated and cooperate primarily during late stage disease development, as predicted by analysis of copy number alterations in prostate cancer genomes (5, 10). The requirement for multiple cooperating events is supported by the use of mouse models that show that activation of Kras (G12D) alone in the prostatic epithelium is insufficient to promote the development of prostate cancer. This observation in mice is similar to that previously shown for the TMPRSS2-ERG fusion, which was only weakly oncogenic when expressed as a transgene in the prostate (11). However, expression of activated Kras in the absence of PTEN (either as a homozygous or heterozygous deletion) resulted in early lethality and invasive carcinoma of the prostate. These mice also developed metastatic lesions in the lung and liver, with gene set enrichment analysis (GSEA) showing reductions in the levels of AR target genes. Furthermore, the gene expression signatures of PTEN-null/Kras^G12D overlap with those of human metastatic tumors. Interestingly, RAS activation in PTEN-null prostate epithelium also induces epithelial to mesenchymal transition (EMT) and promotes expansion of the tumor stem/progenitor cells, which have the ability to reconstitute metastatic disease in vivo (10). These data are in agreement with genomewide studies of patient samples and suggest that, although PTEN is required to initiate tumor development, RAS activation is required for the development of late stage, metastatic disease.

Aytes et al. (6) generate tamoxifen-inducible mouse models that both complement and extend the findings that highlight the roles played by both PTEN and RAS in prostate cancer development and metastasis. Similar to Mulholland et al. (10), tumors that develop in the absence of PTEN (WT RAS) are primarily high-grade prostatic intraepithelial neoplasias and in situ carcinomas, whereas Kras^G12D (WT PTEN) tumors are histopathologically normal. The authors also show that prostate tumors in mice with conditional loss of PTEN and expression of Kras^G12D are AR⁺, luminal epithelial in nature, highly proliferative, and metastasize primarily to the lungs and liver with 100% penetrance. These tumors also display features of EMT. GSEA shows that genes that are up-regulated in these tumors overlap with those expressed in human tumors, whereas the down-regulated genes correlate with those that are more frequently expressed in indolent prostate lesions. Although these data clearly confirm that the RAS and PI3K pathways cooperate during metastasis, the distinct roles played by each pathway and when each is required during this process are revealed using lineage tracing experiments with YFP that mark the dissemination of the prostatic epithelia of PTEN-deficient (designated NP by the authors) and PTEN-deficient:KrasG12D (designated NPK by the authors) cells. As expected, both NP and NPK mice show high levels of YPF fluorescence in the prostate; however, NPK mice also have high levels of fluorescence in the lungs and liver, reflecting metastatic lesions. The onset of prostate cancer in the NPK mice was extremely rapid and showed features of high grade prostatic intraepithelial neoplasia (PIN) and areas of invasion within 2 wk and adenocarcinoma within 1 mo following tamoxifen administration, respectively. However, metastasis to the lungs, which also corresponded with the appearance of EMT, was delayed and was not evident until 3 mo after the mice received tamoxifen. Interestingly, the lungs and lymph nodes contained micrometastases and individually disseminated tumor cells (in the bone marrow), which suggests that dissemination of prostate cancer cells occurs concurrently with the development of the primary tumor but before the appearance of metastases.

What genes, therefore, confer the latent metastatic property to NPK prostate tumor cells? The gene expression profiles of tumors that appeared at 3 mo vs. 1 mo revealed that Etv1, Etv4, and Etv5, members of the PEA3 ETS subfamily of transcription factors, were differentially activated in the metastatic tumors. ETV target genes were also differentially expressed in NPK vs. NP tumors. Among the PEA genes, only Etv4 was up-regulated in the 3-mo vs. 1-mo tumors and in NPK vs. NP tumors. Furthermore, ETV4 expression was also observed in the lung metastases of the NPK mice, suggesting that this gene is a key candidate. ETV4 is an oncogene that promotes EMT and is required for anchorage-independent growth of prostate cancer cell lines (12, 13). Studies have also demonstrated that ETV4 is involved in RAS signaling (14). In human tumors, ETV4 expression and its target genes correlate with Ras and PI3K signaling, a property that is not observed with other oncogenic ETS genes. Furthermore, knockdown of ETV4 in NPK tumor-derived and PC3 prostate cancer cell lines reduced the colony-forming potential and invasiveness of these cells in vitro. Although NPK cells expressing control and Etv4 shRNAs formed tumors in immunodeficient mice, those that expressed the Etv4 shRNA developed smaller and fewer metastases in the lung and liver. Together, the in vitro and in vivo data obtained in the mouse models, as well as that seen in the gene expression signatures of human tumors, strongly support a role for ETV4 in the metastasis of prostate cancer cells (6).

These findings by Aytes et al. are significant for two reasons. The first lies with the model itself and its ability to recapitulate prostate cancer as a metastatic disease in both spatial and temporal manner. The second is that these data establish a role for ETS family members in prostate cancer, in particular, how they promote metastasis in cooperation with activated PI3K and RAS signaling. ETS genes, including ETV4, ERG, ETV5, and ETV1, are known to regulate cellular invasion; however, of these, only ETV4 simultaneously controls proliferation and invasion (13). The authors did not observe an enrichment in ERG target genes in NPK tumors compared with those isolated from NP mice. Because ERG is primarily activated early during prostate cancer progression and is known to direct AR-dependent transcription, the authors speculate that ETS genes might function in a “cascade” and play distinct roles during phases of prostate cancer progression. There are several PI3K/AKT targeted therapies that are currently in various phases of clinical trials, either as single agents or in combination with antiandrogens (where appropriate) (3, 15). Given that ETV4 is a key player in the development of prostate cancer metastases, it is likely that cotargeting RAS and/or oncogenic ETS proteins, which can mirror activated RAS in prostate cancer cells (14), may turn out to be a more effective strategy to treat metastatic disease.