In this report, we have investigated the relationship between androgen levels and prostate tumorigenesis in Nkx3.1; Pten mutant mice, a genetically engineered mouse model of human prostate cancer. By experimentally manipulating serum levels of testosterone in these mice for an extended period (i.e., 7 months), we have found that prolonged exposure of Nkx3.1; Pten mutant mice to androgen levels that are 10-fold lower than normal (the “Low-T” group) resulted in a marked acceleration of prostate tumorigenesis compared with those exposed to androgen levels within the reference range (the “Normal-T” group). We found that prostate tumors from the Low-T mutant mice share a similar gene expression profile as androgen-independent prostate tumors from these mutant mice, which includes the deregulated expression of several genes that are up-regulated in human hormone-refractory prostate cancer, such as Vav3 and Runx1. We propose that exposure to reduced androgens may promote prostate tumorigenesis by selecting for molecular events that promote more aggressive, hormone-refractory tumors. [Cancer Res 2007;67(19):9089–96]
- androgen deprivation
- mouse models
Androgen receptor (AR) signaling is essential for prostate development as well as all stages of prostate tumorigenesis ( 1– 3), which was the foundation for the landmark work of Huggins and Hodges ( 4) who introduced androgen deprivation therapy (ADT) for treatment of prostate cancer. Although ADT remains a widely used treatment for patients with advanced disease ( 3, 5– 8), the relationship between androgen levels and the evolution of prostate cancer remains unresolved. Indeed, serum levels of testosterone are known to decline with aging and aging represents one of the most significant risk factors for prostate cancer ( 9– 11); yet, whether this relationship is causal or coincidental remains unclear. Furthermore, several studies have shown a correlation between low levels of serum androgens and the occurrence and/or outcome of prostate cancer ( 12– 16). In contrast, elevated levels of androgen are known to promote carcinogenesis in rodent models, although this may be due in part to its conversion to estrogens ( 10, 11). Understanding these relationships is further complicated by the fact that serum testosterone levels may not reflect their actual levels in prostate tumors, which may be capable of synthesizing androgens ( 17). Indeed, it seems plausible that variations in androgen levels, either as a result of experimental manipulation or by virtue of aging, may have different consequences for individuals who are cancer prone or not.
To directly investigate the relationship between androgen levels and prostate tumorigenesis, we have used a mutant mouse model based on the combined loss of function of the Nkx3.1 homeobox gene and the Pten tumor suppressor, both of which are relevant for human prostate cancer ( 18– 22). NKX3.1 is located in a region of human chromosome 8p21, which is frequently lost at early stages of prostate cancer, whereas loss of function of Nkx3.1 in mutant mice leads to impaired prostate differentiation and predisposes to prostate cancer (reviewed in ref. 21). PTEN is a broad-spectrum tumor suppressor gene that is deregulated in many types of cancer, and particularly prostate cancer, where its function is critically associated with AR signaling (reviewed in refs. 18– 20). Mutant mice having germ-line loss of function of one allele of Nkx3.1 and Pten (herein referred to as Nkx3.1; Pten mutant mice) develop prostate intraepithelial neoplasia (PIN), which progresses to adenocarcinoma with metastatic potential as a consequence of aging, as well as hormone-refractory tumors following androgen deprivation ( 23– 26).
We have now used these Nkx3.1; Pten mutant mice to investigate the relationship between androgen levels and prostate tumorigenesis. We find that sustained delivery of low levels of androgens accelerates prostate cancer in these mutant mice and that the resulting tumors share molecular features in common with androgen-independent tumors from the Nkx3.1; Pten mutant mice. These findings provide experimental evidence to support the idea that limiting androgens promotes prostate tumorigenesis, which is due in part to the selection for a more aggressive, hormone-refractory phenotype.
Materials and Methods
The Nkx3.1; Pten mutant mice have been described previously ( 23, 25). Analyses were done on a hybrid 129/SvImJ and C57BL/65 background. Cohorts of age-matched (6 weeks old) wild-type (Nkx3.1+/+; Pten+/+) or mutant (Nkx3.1+/−OR−/−; Pten+/−) mice were castrated to eliminate endogenous testicular androgens. Testosterone propionate (hereafter, testosterone; 4-androsten-17β-ol-3-one 17-propionate; Sigma) or vehicle (polyethylene glycol 400, Fluka) was provided by osmotic pumps (Alzet minipumps model 2004, 0.25 μL/h), which were implanted into the midscapular and replaced every 28 days. Pilot studies were done to define the actual concentration of testosterone (0–25 mmol/L; 0–7.5 mg/mL) that would result in the desired levels in the serum (<20–1,500 pg/mL; Fig. 1A ). Aside from the delivery of testosterone, all other variables, such as diet and environmental factors, were maintained under controlled conditions for the duration of the experiment (8.5 months).
At the time of sacrifice, serum was collected by cardiac puncture and testosterone levels were measured using a Cayman Chemical Assay Service. The prostate lobes (anterior, dorsolateral, and ventral) were dissected individually and bilaterally with one side fixed in formalin and embedded in paraffin and the other snap frozen in OCT (Sakura Finetek). Histologic and immunohistochemical analyses were done on paraffin-embedded tissues as described ( 23, 25). Criteria for grading PIN phenotypes in Nkx3.1; Pten mutant mice and for distinguishing low-grade and high-grade PIN have been described previously ( 27). Grading of the phenotypes was done blindly and independently by three individuals (W.B-P., H.G., and C.A-S.), each inspecting multiple ( 3– 5) sections from several distinct regions of the anterior and dorsolateral prostatic lobes for each experimental animal. Data are shown for the anterior lobe; similar results were obtained for the dorsolateral prostate (data not shown).
For gene expression profiling, laser capture microdissection (Arcturus Pixcell IIe) was done to isolate prostate epithelial cells from cryosections of the OCT-embedded anterior prostate of mice from the following groups: normal, dysplasia, low-grade PIN, high-grade PIN, cancer, androgen-independent high-grade PIN, androgen-independent cancer, and low testosterone (“Low-T”) mutants. RNA was prepared using the PicoPure RNA isolation kit (Arcturus) followed by RNA linear amplification and labeling using Small Sample Labeling Protocol VII (Affymetrix). Biotin-labeled RNA samples were hybridized to Affymetrix GeneChips (MOE430A). GeneChips were scanned for data acquisition using a GeneChip Scanner 3000 (Affymetrix). Affymetrix Microarray Suite 5.0 was used to generate “CEL” files that were then processed using Robust Multichip Analysis in Bioconductor/R ( 28). Gene expression level values for each transcript in each sample were set to its ratio relative to the median expression of the measurements of that transcript across the normal (wild-type) samples.
The ANOVA method (P ≤ 0.1) was used to identify differential gene expression between the following sample groups: normal, dysplasia/low-grade PIN, high-grade PIN/cancer, androgen-independent high-grade PIN, and androgen-independent cancer. Two-way hierarchical tree clustering was done across the previously mentioned samples and, additionally, the Low-T mutants. K-means clustering was used to identify the principal patterns of gene expression. Statistical comparisons were done using GeneSpring GX v7.3.1 (Agilent Technologies). Results from the primary analysis were corrected for multiple testing effects by applying the Benjamini et al. ( 29) false discovery rate (FDR) correction (FDR ≤ 0.1). Additional details will be published elsewhere. 4
Results and Discussion
Investigating the relationship between androgen levels and prostate tumorigenesis. To investigate the relationship between androgen levels and prostate cancer in Nkx3.1; Pten mutant mice, we experimentally manipulated androgen levels for an extended period (7 months) and then assessed the consequences for prostate tumorigenesis ( Fig. 1A). Specifically, cohorts of littermate wild-type or Nkx3.1; Pten mutant mice were enrolled at puberty (6 weeks of age), by which time the murine prostate is fully mature ( 1), but before the occurrence of PIN or cancer phenotypes in the mutant mice ( 23, 25). The mice were castrated to remove the endogenous source of testicular androgens and implanted with osmotic pumps for delivery of controlled levels of testosterone (or vehicle) for an additional 7 months.
Before initiating this study, we determined that the serum levels of testosterone in these mice are in the range of 1,000 to 2,500 pg/mL, similar to that reported for other rodents ( 9, 11). Next, we empirically determined the concentration of testosterone in the osmotic pumps that was needed to achieve serum levels that were negligible (i.e., <20 pg/mL for the “No-T” group), 10-fold lower than normal (i.e., 200 pg/mL for the Low-T group), or within the reference range (i.e., 1,500 pg/mL for the “Normal-T” group), which were 0, 3, and 25 mmol/L, respectively. The osmotic pumps, which provide continuous delivery for 28 days, were implanted when the mice were 6 weeks of age and replaced monthly for 6 consecutive months such that all mice were exactly 8.5 months at the conclusion of the study ( Fig. 1A). Because the continuous delivery of testosterone via osmotic pumps does not replicate the normal diurnal variation of testosterone ( 9, 11), we maintained a cohort of mice that had not been castrated or otherwise manipulated, which provided a control for endogenous levels of testosterone (i.e., the “Mock” group). At the conclusion of the study, the mice were sacrificed and the consequences for prostate tumorigenesis were assessed by histologic and molecular analyses.
Therefore, this study compared the following groups of Nkx3.1; Pten mutant and wild-type mice: (a) those without exogenous testosterone (the No-T group), (b) those with testosterone levels that were 10-fold lower than normal (the Low-T group), (c) those with testosterone levels within the reference range (the Normal-T group), or (d) those with endogenous (unmanipulated) levels of testosterone (the Mock group).
The Low-T group display accelerated prostate cancer progression. The prostate cancer phenotype in the Nkx3.1; Pten mutant mice is highly penetrated and displays a well-characterized time course of progression ( 23, 25). In particular, these mice develop low-grade PIN by 6 months of age, which progresses to high-grade PIN by 9 months and adenocarcinoma by 12 months, whereas androgen ablation at 9 to 12 months leads to androgen independence ( Fig. 1A; refs. 23, 25). Notably, at 8.5 months, which was the age of the mice at the conclusion of this experiment, the Nkx3.1; Pten mutant mice characteristically display low- and high-grade PIN but not adenocarcinoma ( 23, 25).
To investigate the effect of varying circulating testosterone levels for prostate tumorigenesis, we examined the histologic phenotype of the experimentally manipulated wild-type and Nkx3.1; Pten mutant mice, focusing on the anterior and dorsolateral prostate ( Fig. 1B; data not shown), which are the lobes that display a tumor phenotype in these mutant mice ( 25). We found that the level of circulating testosterone had a dramatic effect on the occurrence of PIN and cancer in the mutant mice ( Fig. 1B and C; Table 1 ). Specifically, whereas all of the mutant mice in the Mock (n = 6) and Normal-T groups (n = 15) displayed low- and high-grade PIN, but not cancer, as is characteristic of these mutant mice at 8.5 months ( 25), a majority (17 of 24) of the mutant mice in the Low-T group displayed high-grade PIN with invasive carcinoma ( Fig. 1C; Table 1). This represents a significant acceleration of the phenotype of the Low-T group relative to the Mock and Normal group (P < 0.0001 for Low-T versus Mock and Low-T versus Normal-T).
We interpret the consequences of the reduced androgens as an acceleration of the cancer phenotype because the histologic appearance is similar to that seen in aged mutant mice (i.e., ≥12 months; refs. 23, 25). Importantly, the wild-type mice did not display PIN or cancer phenotypes regardless of their serum testosterone levels ( Fig. 1B; Table 1), indicating that the consequences of reduced androgens for acceleration of the cancer phenotype were dependent on loss of function of Nkx3.1 and Pten in these mutant mice. Furthermore, although some of the mutant mice in the No-T group developed low-grade PIN, none displayed high-grade PIN or cancer ( Fig. 1B and C; Table 1). Therefore, the reduction, but not depletion, of serum testosterone promotes a more severe prostate cancer phenotype in the Nkx3.1; Pten mutant mice.
The accelerated cancer phenotype of the Low-T mutant group was further evident by immunohistochemical analyses using well-characterized markers of cancer progression in the Nkx3.1; Pten mutant mice ( Fig. 2 ; refs. 23, 25). We found that the Low-T mutant group displayed comparable and sometimes elevated levels of AR protein expression compared with the Mock and Normal-T groups ( Fig. 2A–D). In addition, the stroma of the Low-T mutant group was significantly attenuated, as evident by staining with smooth muscle actin, a marker of invasion in these mice ( 23, 25). Finally, all of the mutant mice displayed robust levels of activated Akt as expected due to their mutation of Pten ( 25), whereas activation of extracellular signal-regulated kinase (ERK) mitogen-activated protein kinase (MAPK) was primarily elevated in the Low-T group ( Fig. 2I–P). This is consistent with our previous findings in which we observed ERK activation in advanced and androgen-independent tumors in these mutant mice ( 24). Taken together, these findings show that reduced levels of serum accelerate cancer progression in Nkx3.1; Pten mutant mice.
Prostate tumors of the Low-T group share a similar gene expression profile as androgen-independent prostate tumors. To investigate the molecular basis for the accelerated cancer phenotype of the Low-T mutant mice, we did Affymetrix gene expression profiling ( Fig. 3A ). These analyses were facilitated by our extensive characterization of gene expression changes that occur during cancer progression in the Nkx3.1; Pten mutant mice, where we have examined differentially expressed genes at all stages of tumorigenesis. 4 Thus, we did two-way hierarchical tree clustering to compare probe sets that were differentially regulated between normal, dysplastic, low-grade PIN, high-grade PIN, cancer, and androgen-independent high-grade PIN and cancer 4 to the Low-T mutants.
We found that the Low-T mutants displayed a gene expression profile that was strikingly similar to that found in androgen-independent tumors from Nkx3.1; Pten mutant mice. In particular, using K-means clustering to identify the principal patterns of gene expression, we found that one cluster displayed remarkable degree of overlap between the androgen-independent and the Low-T groups ( Fig. 3A; Table 2 ; Supplementary Table S1). Notably, this cluster includes several genes that have been associated with prostate cancer progression and hormone-refractory disease in humans, including AR, MMP9, Ets1, Runx1, and Vav3 ( Fig. 3B; Table 2; e.g., refs. 30– 32). Therefore, reduced levels of androgens promote a molecular phenotype that is similar to androgen independence in the Nkx3.1; Pten mutant mice.
The relationship of serum androgen levels for prostate cancer progression has important implications for treatment of prostate cancer patients as well as evaluating the risk of individuals predisposed to prostate cancer. Considering the impracticality of addressing this relationship experimentally in humans, we have used a relevant mutant mouse model based on loss of function of Nkx3.1 and Pten to manipulate serum levels of testosterone and assess the consequences for cancer progression. Our findings show that sustained exposure of the mutant mice to reduced levels of testosterone leads to an accelerated cancer phenotype, which displays a similar molecular profile as androgen-independent prostate tumors from these mice. Therefore, at least in certain circumstances, prolonged exposure to low androgen can promote, rather than prevent, prostate carcinogenesis, potentially by providing a selective advantage for the outgrowth of androgen-independent prostate cancer cells.
We have shown previously that Nkx3.1; Pten mutant mice display androgen-independent phenotypes before the onset of cancer phenotypes and that androgen independence can develop in parallel with, rather than as a consequence of, cancer progression ( 26). The present findings, in combination with our previous ones, have important implications for experimental manipulation of androgen levels in the human population for chemoprevention, as opposed to the use of ADT for the treatment of patients with advanced prostate cancer. In particular, our findings suggest that some individuals may be harmed rather than helped by treatments that reduce androgen levels. Indeed, the findings of the Prostate Cancer Prevention Trial reported that, although finasteride (an agent that blocks production of dihydrotestosterone) reduced the incidence of prostate cancer ∼25%, some individuals who received finasteride and developed prostate cancer displayed a higher grade of the disease ( 33). In light of our current findings, one interpretation is that these individuals were already predisposed to develop prostate cancer and the finasteride facilitated this by promoting for the selection of more aggressive (i.e., androgen independent) prostate cancer cells.
It is widely recognized that aging is coincident with a progressive decline in androgen levels, while at the same time aging represents the single most important risk factor for prostate cancer. Our findings support the idea that these phenomena are causally linked. Further studies in the human population will be required to evaluate this possible relationship and develop strategies to reverse this phenomenon, perhaps via androgen supplementation.
Grant support: CA076501, CA115717, and UO1-CA84294 (C. Abate-Shen) and T32 HL07382-30 (W.J. Jessen).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
↵4 W.J. Jessen et al., in preparation.
- Received July 30, 2007.
- Accepted August 3, 2007.
- ©2007 American Association for Cancer Research.