**5. Choosing a model:** *in vitro* **and** *in vivo* **prostate cancer models and their relevance to human disease**

Unpredictable pre-clinical models have been credited for the absence of effective treatments in Phase II and III trials [180, 181]. In fact, up to 70% attrition has been reported in Phase II trials; lack of efficacy accounts for about 30% of failures [180]. In the past, drug efficiency was normally assessed during the preclinical phase by xenografts of human prostate cancer cells. The limited number of cells lines that have been utilized in the past mostly include Prostate Cancer (PC)3, Lymph Node Carcinoma of the Prostate (LNCaP) and DU145 (DU-145). *In vivo* (ectopic or orthotopic xenograft models) and *in vitro* studies employing such remain at a disadvantage as these cells lack features found in chemically-induced cancers. The relevance of such models to human prostate cancer is debatable. Nonetheless, these models are valuable in initial research and preclinical assessments. While disease heterogeneity of prostate cancer continues to challenge the development of clinically relevant models, prostate cancer treatment approaches are reliant upon the development and incorporation of relevant preclinical models to confirm appropriate therapeutic targets and biomarkers. Perhaps the most significant milestone and best solution to date in providing researchers with a clinically relevant model, is the evolution of mouse models designed to intentionally inhibit or express a specific gene function by introducing foreign DNA. Over the past 15 years, these Genetically Engineered Mice (GEM) have become valuable tools in studying any combination of oncogenes expressed in specific tissues or conditionally through the tissue-specific removal of tumor suppressors [182-186]. The goal of GEM modeling is to render the best clinical and molecular characteristics of human cancer.

#### **5.1. Prostate cancer cell lines**

An exhaustive list of prostate cancer cell lines is freely available via the British Columbia Cancer Research Centre's Prostate Cancer Cell Line Database and Health Canada (website: http://capcellines.ca). This site was created to provide prostate cancer researchers with valuable information regarding cell lines to be utilized in their research. Information is available for 114 cell lines in total with information inclusive of origin, tumor forming ability in mice, doubling time (and other relevant growth factors), karyotype, as well as the status of ribonucleic acid (RNA) and protein for androgen receptor (AR), Prostate Specific Antigen (PSA), prostate-specific human kallikrein (hK2) and creatine kinase (CK). Prostate cancer xenograft models are also included as well as other relevant markers, too numerous to mention. This site conveniently includes cell lines deemed controversial in nature as well as those that have been reported as contaminated. It would seem that currently, and based on a recent search of PubMed, that the most widely used prostate cancer cell lines include PC-3, LNCaP and DU145.

Cell culture experiments provide a practical and financially viable means of investigating PCa research questions by providing preliminary and mechanistic data that justifies further research. Cell culture approaches do have limitations. In general, consider that the conditions represented in cell culture are artificial. The ideal concentrations of factors replicated to characterize the ideal environment are by no means representative of physiologic conditions. Combine this with the fact that cell culture experiments often utilize a single, clonally identical cell line of similar origin in identically constant conditions (i.e: flask culture). This is especially problematic in prostate cancer research due to the complexity of the interconnected paracrine and endocrine communications system between epithelium, stroma and endocrine organs (i.e: the pituitary) [52]. This use of a single cell type removes the ability to detect effects that a distant organ may have on the agent being tested. A co-culture *in vitro* model may prove to be a more reasonable approach. Additionally, the epithelial layer of the prostate is surrounded by a stromal layer containing fibrobasts and blood vessels, among other cell types. In cell culture models, communication of surrounding and distant tissues is not represented. In this manner, the effect of a therapeutic to surrounding and distant normal tissue is speculative. Further‐ more, cell culture represents only one end of the carcinogenic spectrum as it is often conducted on transformed cells. Aside from the ability of such cells to develop various escape / survival mechanisms, this concept negates long-term exposure. Throughout a person's lifespan, exposure to agents may have varying effects prior to, during or after the transformation process to neoplastic tissue.

Choosing an appropriate cell line is largely dependent on the type of study being conducted. For example, based on the model of study represented in this chapter, the LNCaP line would be an appropriate choice to study n-3 PUFA and β-catenin as these cells are androgen sensitive, prostate-specific antigen (PSA) positive, and maintain malignant properties, even though they have low metastatic potential [187]. Furthermore, this cell line endogenously expresses βcatenin [188] and thus represents an ideal tool to elucidate the interactions of n-3PUFA/ βcatenin in PCa.

#### **5.2. Animal models of prostate cancer**

of such models to human prostate cancer is debatable. Nonetheless, these models are valuable in initial research and preclinical assessments. While disease heterogeneity of prostate cancer continues to challenge the development of clinically relevant models, prostate cancer treatment approaches are reliant upon the development and incorporation of relevant preclinical models to confirm appropriate therapeutic targets and biomarkers. Perhaps the most significant milestone and best solution to date in providing researchers with a clinically relevant model, is the evolution of mouse models designed to intentionally inhibit or express a specific gene function by introducing foreign DNA. Over the past 15 years, these Genetically Engineered Mice (GEM) have become valuable tools in studying any combination of oncogenes expressed in specific tissues or conditionally through the tissue-specific removal of tumor suppressors [182-186]. The goal of GEM modeling is to render the best clinical and molecular characteristics

228 Using Old Solutions to New Problems - Natural Drug Discovery in the 21st Century

An exhaustive list of prostate cancer cell lines is freely available via the British Columbia Cancer Research Centre's Prostate Cancer Cell Line Database and Health Canada (website: http://capcellines.ca). This site was created to provide prostate cancer researchers with valuable information regarding cell lines to be utilized in their research. Information is available for 114 cell lines in total with information inclusive of origin, tumor forming ability in mice, doubling time (and other relevant growth factors), karyotype, as well as the status of ribonucleic acid (RNA) and protein for androgen receptor (AR), Prostate Specific Antigen (PSA), prostate-specific human kallikrein (hK2) and creatine kinase (CK). Prostate cancer xenograft models are also included as well as other relevant markers, too numerous to mention. This site conveniently includes cell lines deemed controversial in nature as well as those that have been reported as contaminated. It would seem that currently, and based on a recent search of PubMed, that the most widely used prostate cancer cell lines include PC-3, LNCaP and

Cell culture experiments provide a practical and financially viable means of investigating PCa research questions by providing preliminary and mechanistic data that justifies further research. Cell culture approaches do have limitations. In general, consider that the conditions represented in cell culture are artificial. The ideal concentrations of factors replicated to characterize the ideal environment are by no means representative of physiologic conditions. Combine this with the fact that cell culture experiments often utilize a single, clonally identical cell line of similar origin in identically constant conditions (i.e: flask culture). This is especially problematic in prostate cancer research due to the complexity of the interconnected paracrine and endocrine communications system between epithelium, stroma and endocrine organs (i.e: the pituitary) [52]. This use of a single cell type removes the ability to detect effects that a distant organ may have on the agent being tested. A co-culture *in vitro* model may prove to be a more reasonable approach. Additionally, the epithelial layer of the prostate is surrounded by a stromal layer containing fibrobasts and blood vessels, among other cell types. In cell culture models, communication of surrounding and distant tissues is not represented. In this manner, the effect of a therapeutic to surrounding and distant normal tissue is speculative. Further‐

of human cancer.

DU145.

**5.1. Prostate cancer cell lines**

Although animal experiments are both costly and time-consuming, they are vital in the transition to Phase 1 clinical investigations. As PCa occurs naturally both in dogs and in certain strains of rats, these species represent options for animal modeling [189]. Although the dog most closely resembles human PCa characteristics [190], their use is limited and unrealistic for a variety of reasons including androgen independent tumor growth, long latency period, prohibitive costs, long gestation period, and difficulty of genetic manipulation [189]. Some strains of rats, although well characterized and capable of developing a wide range of PCa phenotypes [191], present different issues including scarcity of tumors, variability in pheno‐ types, long latency periods, and inability to represent metastatic disease [189]. Utilization of genetically engineered rat models for PCa will likely increase as the development of knockout rats advance [192]. In the meantime, despite anatomical differences between the murine and human prostate gland [189] and other weaknesses, mouse models have emerged as the choice for investigating the stepwise progression of PCa. Compared to earlier *in vivo* prostate cancer experiments, where rodent models were exposed to carcinogens or hormones, recent techno‐ logical advances (i.e.: tissue specific promoters and conditional gene deletions) have provided multiple models for researchers to investigate genetic defects [189]. Each model has its strengths and weaknesses. For example, the transgenic mouse models of PCa that express the Simian Virus 40 (SV40) early region have a highly prognostic gene signature for cancer, however mouse prostate expressing the large probasin promoter directed by SV40-large Tantigen (LPB–Tag) is not consistent in its ability to mimic the full spectrum of PCa as it only progresses to mPIN and rarely results in progression to adenocarcinoma formation [193]. Likely the most popular mouse model, TRAMP is based on such technology and develops cancer as a result of SV40 T antigen expression resulting in many genetic defects including p53 and Rb (tumor suppressor genes) loss [54]. Because cancer is not the result of a single genetic mutation, this multi-hit effect is desirable. However, consider that certain aspects of this model may not be relevant to human disease, including viral transformation. Additionally this model does not follow the normal human prostate cancer course as it is very aggressive and pro‐ gresses to advanced cancer very rapidly. Consider that it may take up to 20 years for PIN to develop, 10 or more years to advance from PIN to HGPIN and then to an early form of latent cancer, while a clinical diagnosis of PCa may occur anywhere from three to 15 years [11].

It would be reiterative to include a thorough review of mouse models utilized in prostate cancer research following the publication of such excellent work by both Valkenburg *et al.* [189] and Hensley *et al* [194]. In this section we will summarize commonly utilized transgenic mouse models; later we will discuss the reliability of mouse models to study disease effects, specifically n-3 PUFA. Table 2 summarizes the characteristics of several commonly used transgeneic mouse models.


**Table 2.** Common Transgenic Murine Models of Prostate Cancer

In choosing a mouse model, one must consider whether the study is aimed at treatment or prevention and the ability of a model to produce prostate phenotypes or other key features relevant to human disease. In dietary intervention studies, such as n-3 PUFA, it is important to consider whether or not models are resistant to the tumor suppressive effects of such treatments. For example, Smolinski (2011) indicates that the TRAMP model may not be ideal for studying the effects of bioactive lipids on prostate carcinogenesis. In fact, in the context of a high fat diet, including fish oil and corn oil as the primary source of lipids, the TRAMP model is reported to be resistant to the tumor suppressive effects of n-3 PUFA [200]. Thus, choosing an inappropriate mouse model may negate significant findings and lead to the conclusion that certain dietary interventions have no effect on a disease when in fact they may be responsible for disease risk or progression. When engaging in novel or controversial research such as discussed in this chapter, a rational tactic before confirming a mouse model may include the completion of a pilot study to ensure that mice are not resistant to the protective effects of n-3 PUFA and to confirm an appropriate cohort size. The goal should be to identify at least a statistically significant 15% difference in mean tumor size within an 80% confidence interval. Based on the proposed research model discussed in this chapter, the transgenic 12T-7s/Catnb lox(ex3)/ PB*Cre4* (designated as LPB–Tag/DA β-catenin) mouse would likely be ideal and was developed recently as a model to study the expression of stabilized β-catenin in PCa and represents the full spectrum of PCa progression to invasive adenocarcinoma [169]. Addition‐ ally, this model represents AR activity and offers a reasonable study period, with an end-point of 20–22 weeks [169].

develop, 10 or more years to advance from PIN to HGPIN and then to an early form of latent cancer, while a clinical diagnosis of PCa may occur anywhere from three to 15 years [11].

It would be reiterative to include a thorough review of mouse models utilized in prostate cancer research following the publication of such excellent work by both Valkenburg *et al.* [189] and Hensley *et al* [194]. In this section we will summarize commonly utilized transgenic mouse models; later we will discuss the reliability of mouse models to study disease effects, specifically n-3 PUFA. Table 2 summarizes the characteristics of several commonly used

**Model Genetic Manipulation Characteristics Reference**

Germline Knockout Invasive carcinoma age 30 weeks;

Conditional Knockout High grade PIN with advancement to

Myc transgene Expression PIN at age 2 weeks and invasive

SV40 Trangene Expression Neoplasis by age 10 weeks.

SV40 Transgene Expression Progressive PIN at age 6 weeks.

In choosing a mouse model, one must consider whether the study is aimed at treatment or prevention and the ability of a model to produce prostate phenotypes or other key features

PIN at age 4 weeks and micro-invasive

metastasis to lymph node and lung.

carcinoma at age 6-12 months with

Undifferentiated adenocarcinoma and invasive neuroendocrine tumors from age 5 - 9 months and metastasis to lymph node, liver and lung.

High-grade PIN at age 6 months; invasion by age 2 months and metastasis to lymph node.

Invasive adenocarcinoma and lymph node, liver, lung and skeletal metastasis by age 6 months. Large, poorly-differentiated neuroendocrine

tumors by age 16 weeks.

[195]

[196]

[194]

[197]

[194, 198]

[194, 199]

[182, 194]

carcinoma at age 8 months.

invasive metastasis.

lymphovascular invasion.

transgeneic mouse models.

PTEN (Phosphatase and tensin homologue deleted on chromosome 10)

Hi-Myc (high Myc transgene expression)

LADY (LPB promoter driving the large-T

TRAMP (TRansgenic Adenocarcinoma of the Mouse Prostate)

antigen)

Apt121/Rbf Fragment of SV40 Transgene

Nkx3.1/Pten Compound Deletion (Pten(+/-) Nkx3.1 (-/-))

**Table 2.** Common Transgenic Murine Models of Prostate Cancer

to dominantly inactivate Rb

230 Using Old Solutions to New Problems - Natural Drug Discovery in the 21st Century
