Background
Prostate cancer is a localized and indolent disease that becomes aggressive only in a small proportion of patients. Despite early diagnosis and the improved effectiveness of treatments, this cancer is the second leading cause of cancer death in males because of its high prevalence in elderly male population [
1].
In almost half of patients with prostate cancer, the tumor carries one of recurrent translocations that place one of the genes from the ETS family (
ERG, ETV1, ETV4, ETV5, FLI1) downstream to the promoter of a gene active in the prostate, with consequent aberrant overexpression of the respective
ETS gene [
2‐
5]. The role of the
ETS genes in prostate carcinogenesis has been investigated in transgenic mice models with a prostate-specific ETS overexpression [
6,
7]. The results have not been always concordant: some studies suggest that ERG or ETV1 overexpression promotes pre-malignant in situ lesions (equivalent to prostatic intraepithelial neoplasia, PIN) [
8‐
12], whereas other studies suggest that this overexpression is not sufficient to cause the onset of cancer [
13‐
18]. These variable results may be related to many factors such as transgene expression levels, transgene integration site, transgene structure, and what promoter drives transgene expression. The genetic background of mice and the timing of the analysis may also play a role, as in the case of human patients.
ETV4 is overexpressed in several cancers [
19‐
24] and in a relatively small fraction of prostate cancers [
25‐
29]. In vitro studies in human prostate cell lines suggested that ETV4 shares with other ETS proteins a major role in invasiveness [
30‐
32] and in cell migration [
33,
34]. We have previously found that, unlike other ETS proteins [
8‐
10], ETV4 increases the rate of proliferation of prostate cells and accelerates the progression through the cell cycle [
34].
Cyclin-dependent kinases inhibitors (CDKIs) are negative regulators of cell cycle progression. Specifically, p21/CIP1 (encoded by
CDKN1A gene) and p27/KIP1 (encoded by
CDKN1bB gene) [
35,
36] belong to the Cip/Kip family of CDKIs proteins, and they regulate the progression from quiescence to G1 and from G1 to S phase by inhibiting the activity of the cyclin/CDK complexes [
37,
38]. p21 and p27 have been regarded as tumor-suppressor genes and their loss has been associated with poor prognosis in several solid tumors [
39‐
43] including prostate cancer [
44‐
47]. However, the prognostic significance of these proteins in prostate cancer is still controversial [
48,
49], especially with respect to p21.
Overall, clinical evidence [
25,
50] and in vitro studies [
33,
34] strongly suggest that ETV4 plays a key role in prostate cancer in a non-negligible proportion of patients. However, the role of ETV4 overexpression in prostate cancer has never been investigated in vivo.
Here, we report a novel transgenic mouse model in which the overexpression of human ETV4 in the prostate results in late development of mouse prostatic intraepithelial neoplasia (mPIN). In these ETV4-overexpressing mice, we found an increased cell proliferation rate associated with the downregulation of p21 and p27. We further show that ETV4 downregulation of p21 (CDKN1A) is determined not only through direct binding of ETV4 to the CDKN1A promoter but also through the downregulation of the p53 protein.
Discussion
Aberrant overexpression of an ETS protein in the prostate is a common event in most of the patients with prostate cancer. This overexpression, caused by the translocation of an
ETS gene under the control of the promoter of a gene highly expressed in the prostate, plays a direct role in prostate cancer pathogenesis [
2,
4]. The most frequent translocation is that of
ERG gene downstream the promoter of
TMPRSS2, but rearrangements of
ETV1,
ETV5, and
ETV4 genes are also relatively common [
25]. The role of ERG and ETV1 in prostate cancer has been thoroughly studied, whereas the mechanisms whereby overexpression of ETV4 mediates oncogenesis in the prostate have not been investigated in depth.
Overexpression of ETV4 confers several neoplastic features onto prostate cell lines [
33,
34]; here, we have investigated whether this holds true also in vivo. To this end, we have generated two independent lines of transgenic mice in which the prostate specific expression of
ETV4 is driven by the probasin promoter (ETV4 mice). ETV4 mice showed an increased prostate expression of
MMP2,
MMP7, and
MMP9 suggesting that also in vivo ETV4 exerts a transcriptional upregulation of MMPs as observed in human prostate cell lines [
34] where it is associated with increased migration and invasion [
33,
34].
Two-thirds of 10-month-old ETV4 mice developed prostate focal lesions resembling the early modifications observed in human PIN: in some of these mice some lesions have more severe features. Thus, ETV4 overexpression promotes mPIN with long latency and partial penetrance. However, even in the older mice (15 months) and despite of MMPs overexpression, mPIN did not progress to prostate cancer, implying that additional genetic events are required. These findings are very similar to those reported in various model of ERG and ETV1 prostate transgenic mice [
8‐
10,
12].
The role of ETS proteins, such as ERG and ETV1, in regulating prostate cell proliferation in vivo is minimal: in fact, in ERG and ETV1 transgenic mice, there is no [
10,
13,
14] or only a slight increase [
8,
9,
12,
16‐
18] in cell proliferation. However, in a mouse model obtained by pronuclear injection of a bacterial artificial chromosome carrying the
TMPRSS2-ERG transgene, ERG drives proliferation and blocks the differentiation of prostate cells [
61]. In ETV4 mice, instead, we find that ETV4 plays a role in prostate cell proliferation in vivo. This is supported by the significant increase of proliferating cells observed in the prostate of ETV4 mice. This result is consistent with in vitro studies in which ETV4 increases proliferation of human prostate cell lines through progression of cell cycle [
34], although cell cycle progression has not yet been proven in ETV4 mice.
In prostate cell lines (RWPE and PC3), ETV4 expression is associated with an increased rate of proliferation and with downregulation of
CDKN1A [
34]. Accordingly, also in ETV4 mice, the increased proliferation rate is associated with reduced levels of both
CDKN1A mRNA and of its encoded protein p21 (Fig.
3). In addition, also in vivo ETV4 expression is associated with the reduction of p27 protein, but not of its mRNA (encoded by
CDKN1B gene) as previously found in RWPE cells transfected with ETV4 [
34]. ETV4 may promote cell proliferation also in other systems where it regulates a number of genes:
HER2 in ovarian (SKOV-3) and breast (MDA-MB-453) cancer cell lines [
62];
WT1 in CHO and COS7 cell lines [
63];
cyclin D3 in MDA-MB-231 breast cancer cell line [
42];
cyclin D1 in mammary tissues [
64];
NOTCH1 and
NOTCH4 in MCF-7, MDA-MB-231, and SKBr3 breast cancer cells [
65]. Taken together, these data suggest that ETV4 can control cell proliferation through a variety of cell type specific mechanisms.
In the prostate, ETV4 hinders the transcription of
CDKN1A and therefore the level of p21 through direct binding to a proximal ETV4-binding site within the
CDKN1A promoter (Fig.
5). This mechanism of downregulation of
CDKN1A in human prostate (RWPE, PNT1A, and PC3) cell lines is reminiscent of the downregulation induced by ETV4 through binding to other promoters, such as that of
ERBB2 promoter in breast and ovarian cancer cells [
62,
66] and that of collagenase-1 promoter in a breast cell line treated with all-trans retinoic acid [
67]. On the other hand, at variance with p21 (
CDKN1A), the ETV4-mediated reduction of the protein level of p27 (
CDKN1B), another cell cycle inhibitor controlling the progression at G1, seems indirect because it is not associated with significant variation of mRNA levels.
In the 1990s, ETV4 was regarded as a tumor suppressor gene because it was able to increase luciferase expression driven by the
CDKN1A promoter in SiHA cervical cancer cells whereas its deletion reduced
CDKN1A levels in Saos2 osteosarcoma cells [
68]. However, we find that ETV4 downregulates
CDKN1A in prostate cells and also in MCF7 breast cancer cell line (Fig. S
4). These observations suggest that the role of ETV4 in cancer is cell-type and tissue-dependent; it behaves as an oncogene in prostate and breast cells (and in many others tissues) [
69], whereas it acts as a tumor suppressor gene in osteosarcoma [
68] and in cervical cells [
62]. This apparent discrepancy could be explained by the fact that different cell types express different tissue-specific factors that, in turn, may influence the role of ETV4 as it happens for others ETS transcription factors [
5,
70].
Direct downregulation of the
CDKN1A promoter by ETV4 is only part of the story. In fact, ETV4 is able to reduce
CDKN1A expression even when its binding site to the
CDKN1A promoter is mutated (Fig.
6). Since this takes place only in p53 competent cells and even when the promoter contains only the p53-binding site [
55] led us to realize that regulation of
CDKN1A by ETV4 is mediated in part through p53. In keeping with this notion, the levels of p53 protein are reduced upon expression of ETV4 in normal human prostatic RWPE cells (Fig.
7b, c), and also in vivo in ETV4 mice (Fig.
7d, e). As already observed for p27, this ETV4-mediated reduction of p53 protein was not associated with any change in
TP53 mRNA, suggesting an indirect regulatory mechanism that remains to be identified.
Our main finding is that both in vitro and in vivo ETV4 can modulate cell cycle and, in turn, proliferation of prostate cells through multiple layers of regulation including both the direct regulation of transcription of some gene (CDKN1A-p21) and the indirect regulation of other genes (CDKN1B-p27, TP53). Finally, the finding that ETV4 reduces the level of p53 protein suggests its possible contribution to cellular processes, beyond cell cycle, in which p53 plays a role, such as apoptosis, genomic stability, and senescence.
ETV4 overexpression in the prostate is observed in only a relatively small subset of prostate cancer patients (about 1–5%) [
2‐
4,
29,
50]; however, since prostate cancer is common, these results are potentially relevant for a significant number of patients. In addition, Aytes and colleagues have reported the late increase of ETV4 expression in a prostate metastasis mouse model with several genetic alterations (pTen loss, NKX3.1 deletion, and a KRAS activating mutation) suggesting that ETV4 may have a role in the metastatic process even in prostate cancers that do not overexpress ETV4 initially [
71]. It is also noteworthy that in a recent study of biopsy cores from 120 patients ETV4 expression was mostly associated with high-grade cancer [
29]. Thus, the relevance of our results could be extended to an even larger number of patients.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.