The physiological role of cyclin D2

To elucidate the physiological role of cyclin D2 in the mouse prostate a conditional knockout mouse model with a prostate-specific cyclin D2 deletion was established during the present project. To investigate especially the effect of cyclin D2 deletion on PCa development and progression the cyclin D2 knockout mice were crossed with TRAMP mice.

In the Apc(Min/+) mice, a popular animal model for studies of human colon cancer (Leclerc et al. 2004), it was shown that a knockout of cyclin D2 dramatically reduced tumor growth and development (Cole et al. 2010). Cyclin D2 is known to be overexpressed in 53% of colon cancers (Mermelshtein et al. 2005), whereas it is inactivated in PCa patients due to promoter hypermethylation (Padar et al. 2003). It was shown that restoration of the cyclin D2 expression in LNCaP PCa cells resulted in reduced proliferation. Witt et al. (2013) could also show that re-expression of cyclin D2 in human and mouse PCa cells, induced by VPA treatment was associated with decreased proliferation rates, whereas fibroblast cells, in which cyclin D2 expression was not further increased by VPA treatment, showed no such proliferation inhibition, except for the L-cells. Moreover, in non-small cell lung cancer, reduced cyclin D2 expression is correlated with a poor recurrence-free survival (Ko et al. 2012). These data indicate that at least in some cancers, including PCa, increased or restored cyclin D2 expression is associated with anti-tumorigenic effects. To examine whether this holds true in vivo, in the present study a prostate-specific cyclin D2 knockout mouse model was generated, because to date such a mouse model has not been established. These conditional knockout mice were further crossed with TRAMP mice to induce PCa formation and to study the consequences of a cyclin D2 knockout on tumor development and tumor progression. If cyclin D2 indeed exerts a tumor suppressive function in PCa then TRAMP mice lacking cyclin D2 expression in the prostate epithelium should develop PCa at an earlier age or exhibit a more aggressive PCa.

Multiple conventional knockout mouse models already exist with either a single D-type cyclin knockout, double D-type cyclin knockout or triple D-type cyclin knockout (cyclin D-null mouse).

Studies on single D-type cyclin knockout mice showed that they were all viable but exhibited different phenotypes. In cyclin D1 knockout mice the proliferation of breast epithelium was impaired during pregnancy whereas in female cyclin D2 knockout mice the proliferation of ovarian granulosa cells in response to follicle-stimulating hormone (FSH) was inhibited, resulting

in infertility. Knockout of cyclin D2 resulted in cerebellar abnormalities and impaired neurogenesis (Huard et al. 1999, Kowalczyk et al. 2004). Cyclin D3 knockout mice exhibited disturbed development of immature T-lymphocytes. The predisposition for different cancer types was also divert in the single D-type cyclin knockout mice. A knockout of cyclin D1 expression led to resistance of developing breast cancer promoted by Ras and Neu oncogenes. Cyclin D2 knockout mice had a decreased susceptibility for cancer of the gonads, whereas cyclin D3 knockout mice had a decreased susceptibility for skin papilloma and Myc-promoted oral mucosa tumors (summarized from Kozar and Sicinski 2005).

Studies on mice that express only a single D-type cyclin showed that cyclin D1-only mice developed severe megaloblastic anemia, cyclin D2-only mice presented neurological abnormalities, and cyclin D3-only mice lacked normal cerebella (Ciemerych et al. 2002). These studies indicate that the D-type cyclins possess some individual functions, but also general functions that can be compensated by the other D-type cyclins. Even the lack of all three D-type cyclins in cyclin D-null mice (cyclin D1(^-/-)D2(^-/-)D3(^-/-) showed that cyclins are dispensable for cell cycle progression. These mice developed phenotypically and histopathologically comparable to control mice until embryonic day 13.5 (E13.5, Kozar et al. 2004) when most of the organs are already completely developed. At E17.5 cyclin D1(^-/-)D2(^-/-)D3(^-/-) mice died due to heart abnormalities combined with a severe anemia. Furthermore, this study showed that cyclins are critically required for the expansion of hematopoietic stem cells, whereas cyclin D-deficient fibroblasts proliferate nearly normally. Additionally, mouse embryonic fibroblast (MEF) cells lacking D-cyclins displayed reduced susceptibility to oncogenic transformation (Kozar et al.

2004).

For the present study, Ccnd2^fl/fl mice and PB-Cre4⁺ mice with a C57/Bl6 background were purchased and their colonies established by mating with C57/Bl6 wild type mice. PB-Cre4⁺ mice contain a second-generation composite probasin promoter, the ARR2PB promoter, which was modified to contain two androgen response elements (ARR) in the PB promoter (Zhang et al.

2000). This 0.5-kb PB promoter fragment maintains reliable prostate-specificity and simultaneously provides very high transgene expression in transgenic mice (Zhang et al. 2000).

Mating of Ccnd2^fl/fl mice with PB-Cre4⁺ mice resulted in homozygous Ccnd2^fl/fl/Cre⁺ mice as verified by genotyping PCR. The prostate-specific knockout of cyclin D2 could be partially confirmed by PCR analysis of different mouse tissues of homozygous Ccnd2^fl/fl/Cre⁺. Only in the prostate tissue the band for the deleted Ccnd2 allele could be detected, however, also a specific band for the floxed Ccnd2 allele was observed. Quantitative real-time PCR and western blot analyses of different tissues revealed that cyclin D2 expression in the prostate was not reduced,

which is contrary to the expectation from a homozygous, prostate-specific cyclin D2 knockout mouse.

One possible explanation for this observation could be that not a pure population of epithelial cells of the prostate was used for genotyping PCR, quantitative real-time PCR and western blot analyses. The probasin promoter in the knockout construct ensures specific expression of the Cre recombinase exclusively in prostate epithelial cells, therefore only these cells should exhibit a cyclin D2 knockout. For genotyping PCR, quantitative real-time PCR and western blot analyses the whole prostate gland was used, which consists of two generic cell types, i.e. epithelial cells and stromal cells. Epithelial cells form glands that are composed of the luminal secretory and basal cell types and rare neuroendocrine cells. The stroma surrounding the prostatic glands contains smooth muscle cells and fibroblasts (Cunha et al. 1996). Blood vessels, peripheral nerves and ganglia, and tissue-infiltrating white blood cells are additional constituent cell elements of the normal adult human prostate. The fact that the prostate is a solid organ makes the isolation of a specific cell type for a knockout verification problematic (Liu and True 2002).

Ongoing experiments to verify the prostate-specific knockout of cyclin D2 include immunohistochemistry, but to date no suitable antibody for mouse cyclin D2 has been identified, and RNA in situ hybridization (RNA-ISH). The cyclin D2-specific RNA probes were generated during the present study, but still need to undergo digoxigenin (DIG) labeling for the actual ISH procedure. If neither one of these two approaches should be successful in demonstrating the prostate-specific cyclin D2 downregulation, then microdissection should be conducted as a last alternative approach. Using microdissection, a single cell type, in this case prostate epithelial cells, can be specifically isolated from tissues consisting of multiple cell types. The isolated pure prostate epithelial cells could then be used for repetition of PCR, quantitative real-time PCR and western blot analyses.

The putative conditional cyclin D2 knockout mice were further mated with TRAMP mice (T⁺) to study the effects of prostate-specific loss of cyclin D2 expression in a PCa mouse model.

Thereby, especially the effect on tumor development and progression ought to be studied. If cyclin D2 proves to be a tumor suppressor in PCa, then it is expected that mice lacking cyclin D2 expression in the prostate epithelium would develop PCa earlier as compared to normal TRAMP mice or exhibit a more aggressive PCa. The likelihood to receive homozygous Ccnd2^fl/fl/Cre⁺/T⁺ mice when mating Ccnd2^fl/+/Cre⁺/T⁺ mice with Ccnd2^fl/fl mice is only 12.5%. To date, there are seven putative homozygous Ccnd2^fl/fl/Cre⁺/T⁺ mice of which none has developed a palpable PCa. The oldest of these mice is 18 weeks old, therefore, this mouse is too young to have developed an adenocarcinoma yet, as compared to single-transgenic TRAMP mice which develop an adenocarcinoma at app. 28 weeks of age. Furthermore, one 16-week-old

heterozygous Ccnd2^fl/+/Cre⁺/T⁺ mice had already developed a palpable PCa and had to be sacrificed. Based on the age of this heterozygous mouse it should not have developed a PCa but rather a PIN according to the tumorigenic time course in the TRAMP mouse model.

Development of PCa in heterozygous double transgenic Ccnd2^fl/+/Cre⁺/T⁺ mice at such a young age could suggest that even the loss of one cyclin D2 allele in the prostate of TRAMP mice could be sufficient to induce a more severe PCa as compared to single transgenic TRAMP mice.

Nonetheless, at the present state of the experiment no definite conclusion can be drawn on the physiological role of cyclin D2 in PCa development and progression in vivo (Table 4.5, section 4.4.2).

The fact that not all TRAMP mice in our institute developed PCa and that these mice often develop seminal vesicle carcinoma, which very rarely is accompanied with development of PCa, as observed in our institute and also known from the literature (Tani et al. 2005, Yeh et al. 2009), led to the search for a more suitable PCa mouse model. In the literature there are other disadvantages of the TRAMP mouse model stated, for example that the carcinoma developed from TRAMP mice are of neuroendocrine origin (Chiaverotti et al. 2008) and not of epithelial origin as compared to the majority of human PCa cases (Abrahamsson 1999). Also, the PCa in TRAMP mice rarely metastasizes to the bone, as it is the case in humans, but rather to lymph nodes and lung (Gingrich et al. 1997). Another limiting factor for the use of TRAMP mice in PCa studies is the inherent use of the probasin promoter which is regulated by androgens (Matuo et al. 1989, Kasper et al. 1998). When it comes to interpreting consequences of castration this will be confounded by the possibility that observed androgen sensitivity is due to downregulation of transgene expression. Furthermore, as already mentioned, use of the non-physiological SV40 T antigen leads to prostate epithelium-specific inactivation of pRb and p53, the major targets of SV40 T antigen, resulting in formation of neuroendocrine tumors with metastasis potential to distant organs. The same is observed by prostate-specific inactivation of pRB and p53 (Zhou et al. 2006). Consequently, the phenotype of TRAMP mice may reflect the consequences of RB and p53 pathway inactivation. Besides, the relatively short kinetics of PCa development differ from the characteristically slow development of PCa in humans (Gingrich et al. 1999).

The PCa mouse model in which the tumor suppressor gene phosphatase and tensin homolog deleted from chromosome 10 (Pten, Song et al. 2012) is excised seems to be a suitable alternative to the TRAMP mouse model. It is the most frequently used genetically-engineered mouse model, although very few human PCa patients have loss of both PTEN alleles.

Approximately 23% of human high grade PIN, 69% of localized PCa (Yoshimoto et al. 2006) and 86% of metastatic castration resistant PCa (Holcomb et al. 2009) exhibit PTEN deletions. PTEN is a phosphatase which removes a phosphate group from phosphatidylinositol

3,4,5-triphosphate (PIP3,4,5,) resulting in downregulation of the Akt/m-Tor signaling pathway, leading to decreased cell proliferation and survival (Cantley and Neel 1999). PTEN is of considerable importance for PCa because of its relevance for regulating androgen receptor signaling (Abate-Shen and (Abate-Shen 2000, (Abate-Shen and Abate-(Abate-Shen 2007, 2010) and its loss has been linked to many cancers, including PCa (Cairns et al. 1997). Due to the fact that Pten conditional knockout mice resemble the principal driving event of human PCa in PCa of mice and also represents the human course of the disease from PIN to metastatic, castration resistant PCa of epithelial origin (Wang et al. 2003, Trotman et al. 2003) it was chosen for mating with Ccnd2^fl/fl/Cre⁺ mice instead of TRAMP mice. One other major benefit in using Pten conditional knockout mice instead of TRAMP mice for the generation of conditional cyclin D2 knockout mice is that there will be no further interference with other cell cycle- associated factors, solely the cyclin D2 knockout. PTEN is known to induce cell cycle arrest by negatively regulating especially cyclin D1, but also the other D-type cyclins (Radu et al. 2003, Diao and Chen 2007). PTEN was also shown to downregulate cyclin D2 expression (Huang et al. 2007). Therefore, loss of PTEN in conditional cyclin D2 knockout mice should have no further negative effect on cell cycle regulators. Using TRAMP mice, the SV40 T antigen would inactivate pRb and p53, two major cell cycle components, which would distort the outcome of a conditional cyclin D2 knockout. Thus, the phenotype of TRAMP mice with a conditional cyclin D2 knockout may also reflect the consequences of RB and p53 pathway inactivation. The idea is to solely interfere with cyclin D2 and no other cell cycle components since the physiological role of cyclin D2 ought to be investigated. The Pten conditional knockout mice fulfills this criteria and was therefore purchased and the line just recently established. Due to the advanced stage of this thesis and the accompanied time limitation it was not possible to set up the mating of Pten conditional knockout mice with prostate-specific cyclin D2 knockout mice during this project. Once the physiological function of cyclin D2 in PCa is fully understood, this knowledge can be transferred to reveal its pathologic function and might help in the development of new therapeutic strategies for the treatment of PCa.