Genetic interaction between Tmprss2-ERG gene fusion and Nkx3.1-loss does not enhance prostate tumorigenesis in mouse models

Abstract

Gene fusions involving ETS family transcription factors (mainly TMPRSS2-ERG and TMPRSS2-ETV1 fusions) have been found in ~50% of human prostate cancer cases. Although expression of TMPRSS2-ERG or TMPRSS2-ETV1 fusion alone is insufficient to initiate prostate tumorigenesis, they appear to sensitize prostate epithelial cells for cooperation with additional oncogenic mutations to drive frank prostate adenocarcinoma. To search for such ETS-cooperating oncogenic events, we focused on a well-studied prostate tumor suppressor NKX3.1, as loss of NKX3.1 is another common genetic alteration in human prostate cancer. Previous studies have shown that deletions at 8p21 (harboring NKX3.1) and 21q22 (resulting in TMPRSS2-ERG fusion) were both present in a subtype of prostate cancer cases, and that ERG can lead to epigenetic silencing of NKX3.1 in prostate cancer cells, whereas NKX3.1 can in turn negatively regulate TMPRSS2-ERG fusion expression via suppression of the TMPRSS2 promoter activity. We recently generated knockin mouse models for TMPRSS2-ERG and TMPRSS2-ETV1 fusions, utilizing the endogenous Tmprss2 promoter. We crossed these knockin models to an Nkx3.1 knockout mouse model. In Tmprss2-ERG;Nkx3.1+/- (or -/-) male mice, although we observed a slight but significant upregulation of Tmprss2-ERG fusion expression upon Nkx3.1 loss, we did not detect any significant cooperation between these two genetic events to enhance prostate tumorigenesis in vivo. Furthermore, retrospective analysis of a previously published human prostate cancer dataset revealed that within ERG-overexpressing prostate cancer cases, NKX3.1 loss or deletion did not predict biochemical relapse after radical prostatectomy. Collectively, these data suggest that although TMPRSS2-ERG fusion and loss of NKX3.1 are among the most common mutational events found in prostate cancer, and although each of them can sensitize prostate epithelial cells for cooperating with other oncogenic events, these two events themselves do not appear to cooperate at a significant level in vivo to enhance prostate tumorigenesis.

Fig 1. Nkx3.1-loss modestly increases the Tmprss2 promoter activity in vivo.

A. Progressive Nkx3.1 transcript loss was confirmed in wild type (black) and heterozygous (dark gray) and homozygous (light gray) Nkx3.1 knockout mice by real-time RT-PCR (left). Immunohistochemical (IHC) staining of anterior prostates (APs) using a mouse-specific Nkx3.1 antibody also validated Nkx3.1 protein loss. B. Real-time RT-PCR showing slight but statistically significant increase in the Tmprss2-ERG expression in T-ERG;Nkx3.1 ^+/- double heterozygous males. C. IHC staining of APs showing increase in ectopic ERG expression at the protein level from the T-ERG knockin allele under the Nkx3.1-null background (T-ERG;Nkx3.1 ^-/-). H-scores were calculated as 81 and 127 for T-ERG and T-ERG;Nkx3.1 ^-/- sections, respectively. D. FACS analysis showing progressive increase in the percentage of GFP⁺ cells in the prostates of T-ERG;Nkx3.1 ^+/- and T-ERG;Nkx3.1 ^-/- males, compared to those of males with T-ERG alone. Statistics: p values from Student t-test are indicated; ns = not significant. Scale bars represent 50 μm.

Fig 2. Heterozygous Nkx3.1-loss does not strongly cooperate with Pten-loss and Tmprss2-ERG expression.

A. Representative anterior prostate (AP) histology of male mice with the indicated combinations of Nkx3.1 ^+/-, Pten ^+/-, and T-ERG knockin. Note HG-PIN lesions developed in all prostate lobes of T-ERG;Pten ^+/- and T-ERG;Pten ^+/- ;Nkx3.1 ^+/- males due to cooperation between Pten ^+/- and T-ERG. Representative HG-PIN lesions developed in the APs of T-ERG;Pten ^+/- and T-ERG;Pten ^+/- ;Nkx3.1 ^+/- males are shown (red arrows). Scale bars represent 100 μm. B. Histology summary of aged Pten ^+/- (left) and Pten ^+/- ;Nkx3.1 ^+/- (right) male mice with or without the T-ERG knockin allele. Notable cooperation was detected with T-ERG (p = 0.05 under the Pten ^+/- background and p = 0.04 under the Pten ^+/- ;Nkx3.1 ^+/- background). HG-PIN in any prostate lobe was diagnosed by a trained rodent pathologist.

Fig 3. Total Nkx3.1-loss does not cooperate with Tmprss2-ERG gene fusion to promote prostate tumorigenesis.

A. Representative anterior lobe (AP) histology of Nkx3.1 ^-/- (left) and T-ERG;Nkx3.1 ^-/- (right) mouse prostates stained with H&E. Scarce pleomorphic nuclei are evident (red arrows). Scale bars represent 100 μm. B. Graphical summary of histological findings of Nkx3.1 ^-/- and T-ERG;Nkx3.1 ^-/- male mice. There was no significant difference in AP hyperplasia frequency (p = 0.63). Histology was diagnosed by a trained rodent pathologist. C. IF staining for respective basal keratin 5 (K5, red) and luminal keratin 8 (K8, green) to visualize AP architecture in Nkx3.1 ^-/- and T-ERG;Nkx3.1 ^-/- mice. Nuclei counterstained with DAPI (blue). Scale bars represent 50 μm.

Fig 4. NKX3.1-loss in patients harboring ERG rearrangements is not predictive of biochemical relapse.

Patient data from Taylor et al. [29] was used to compare via Kaplan-Meier analysis the disease-free survival of patients overexpressing ERG, which is highly predictive of harboring TMPRSS2-ERG fusion. Within this 'ERGup' cohort, patients who exhibited NKX3.1 downregulation (red line, n = 4) compared to those who expressed normal levels of NKX3.1 (blue line, n = 65) were not more likely to display biochemical relapse. Logrank test p value was 0.35. Analysis performed using the cbioportal software [31].