Emergence of Androgen Independence at Early Stages of Prostate Cancer Progression in Nkx 3 . 1 ; Pten Mice

Although androgen deprivation therapy is a widely used treatment for patients with advanced prostate cancer, it ultimately results in the emergence of a hormone-refractory disease that is invariably fatal. To provide insights into the genesis of this disease, we have employed an in vivo model to investigate how and when prostate epithelial cells can acquire the ability to survive and proliferate in the absence of androgens. In particular, we have been studying the evolution of androgen independence in Nkx3.1; Pten mutant mice, which develop prostatic intraepithelial neoplasia and adenocarcinoma as a consequence of aging, as well as androgenindependent phenotypes following castration. We now find that the prostate epithelial cells from these Nkx3.1; Pten mutant mice are capable of surviving and proliferating in the absence of androgens and that they develop androgenindependent phenotypes well before they display overt prostatic intraepithelial neoplasia or cancer phenotypes. Our findings in this mouse model show that acquisition of androgen independence can be uncoupled from overt cancer progression and raise the possibility that hormone-refractory disease can arise at early stages of prostate carcinogenesis. (Cancer Res 2006; 66(16): 7929-33)


Introduction
All aspects of prostate development and prostate carcinogenesis are critically dependent on the activity of the androgen receptor and its primary ligands, testosterone and dihydrotestosterone (1)(2)(3)(4)(5)(6)(7).In fact, androgen deprivation therapy remains the most widely used treatment for patients with advanced prostate cancer.However, although androgen deprivation initially results in regression of prostate tumors, the tumors eventually reemerge and the resulting hormone-refractory (androgen independent) disease is invariably fatal.Therefore, understanding how and when prostate cancer cells acquire the ability to survive and proliferate in the absence of androgens is critical for the treatment of patients with prostate cancer.
We have been investigating the evolution of androgenindependent prostate cancer in vivo using a mouse model based on the loss-of-function of the Nkx3.1 homeobox gene and the Pten tumor suppressor.NKX3.1 maps within a critical region of human chromosome 8p21 that undergoes deletion in prostate cancer (8,9).Furthermore, loss or reduction of NKX3.1 expression is a hallmark of prostate cancer progression in humans as well as in mouse models of this disease, and inactivation of Nkx3.1 in mutant mice leads to prostate intraepithelial neoplasia (PIN; refs.[10][11][12][13][14][15][16]. PTEN encodes a lipid phosphatase that is a key negative regulator of the PI-3 kinase signaling cascade, with one of its principal downstream targets being Akt/protein kinase B (17)(18)(19).PTEN maps to human chromosome region 10q23, which is deleted at high frequency in prostate cancer (8), and loss of PTEN protein expression is a frequent occurrence in advanced prostate cancer (20,21).In mice, loss-of-function of Pten leads to high-grade PIN and/or carcinoma and can synergize with inactivation of other relevant genes, such as Nkx3.1, p27 kip1 , or p53, in prostate cancer progression (15,(22)(23)(24)(25).
Previously, we have found that loss-of-function of Nkx3.1 and Pten leads to PIN and cancer with aging, as well as androgen independence following castration (22,25).In the present study, we have investigated the evolution of androgen independence in Nkx3.1;Pten compound mutant mice.We find that these mice can acquire androgen-independent phenotypes before the formation of overt PIN or cancer.Our findings shed new light on the relationship between androgen independence and cancer progression, which may have implications for the treatment of patients with prostate cancer.

Materials and Methods
Measurement of androgen levels.The Nkx3.1; Pten mutant mice have previously been described (11,22,25).Castration was done by surgical removal of the testes and epididymus, and mice were analyzed 1 to 3 days or 3 to 10 months following surgery.Because the adrenal glands provide an alternative source of androgens that may be used in the absence of testosterone (26), we also did bilateral adrenalectomy along with castration to fully deplete all circulating androgens in several mice.Serum testosterone levels were determined using a competitive enzyme immunoassay [Cayman Chemical (Ann Arbor, MI) Testosterone EIA Kit].Following castration, the levels of testosterone in the serum were reduced from the reference range of f4 to 10 ng/mL to <0.001 ng/mL (n = 10).
Histologic and immunohistochemical analyses.The prostate and other organs of the male urogenital system were collected from euthanized mice and processed for histopathologic analyses as described (11,22,25).Histologic criteria for designation of PIN and/or cancer phenotypes have been described (27); analyses shown are for anterior prostate, with similar results observed in other prostatic lobes. 5Procedures for immunohistochemical analyses were previously described (11,13,22,25).Antibodies and dilutions were as follows: Ki67 (Novocastra, Newcastle, United Kingdom; 1:1,000), androgen receptor (Sigma, St. Louis, MO; 1:2,000), and p-Akt (Ser473; 1:200).Terminal deoxyribonucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assays were done using an In Situ Cell Death Detection Kit (Roche Diagnostics, Mannheim, Germany).Cell counting for determination of the relative numbers of proliferative and apoptotic prostate epithelial cells was done as described (11,25).

Results
Survival and proliferation of Nkx3.1;Pten prostate epithelial cells following castration.In previous studies, we have shown that compound mutant mice lacking one or both alleles of Nkx3.1 as well as one allele of Pten (Nkx3.1 +/À or À/À ; Pten +/À , hereafter referred to as Nkx3.1;Pten mutant mice) develop low-grade PIN by 3 to 6 months of age, high-grade PIN by 6 to 9 months of age, invasive adenocarcinoma by 12 months, and androgen-independent phenotypes following androgen ablation (refs.22, 25; Fig. 1A).We have now used these Nkx3.1;Pten mutant mice to investigate the emergence of androgen independence in vivo.
Specifically, we have examined the consequences of surgical castration of these mice as a function of animal age and prostate cancer progression using four experimental regimens: group I, 0.75 months (3 weeks) of age, before complete maturation of the prostate; group II, 2 months, subsequent to prostate maturation but before the onset of PIN; group III, 6 months, subsequent to the onset of PIN, but before the onset of cancer; or group IV, 9 months or subsequent to the onset of cancer (Fig. 1A; Table 1).These mice were analyzed immediately following castration (1-2 days) for analyses of cell survival and proliferation, or were examined 3 to 6 months later for analyses of androgen-independent phenotypes.In this study, we define ''androgen independence'' as the ability of prostate epithelial cells to survive and display a robust proliferation index following removal of testicular androgens and for the mice to display PIN and/or cancer phenotypes several months after castration.
In striking contrast to wild-type control mice (n = 4), castration of Nkx3.1;Pten mutants (n = 6) did not induce massive apoptosis of the prostate at 2 days following castration, as shown by TUNEL analysis (Fig. 1B-G and Supplementary Fig. S1A-D).In particular, the high-grade PIN lesions of mutants castrated at 8 months and analyzed 2 days later (group III) displayed significantly less apoptosis (4%; n = 1,727 epithelial cells) than the prostate epithelium of wild-type littermates (20%; n = 976; Fig. 1B-G).Furthermore, the epithelial cells within these high-grade PIN lesions were highly proliferative, as determined by Ki67 immunoreactivity (16%; n = 1,188), compared with wild-type control epithelium (1%; n = 928; Fig. 1H-M).These findings suggest that at the time of castration, the high-grade PIN lesions of Nkx3.1;Pten mutant mice contain a significant percentage of cells that can survive and proliferate in the absence of androgens.
Interestingly, the Nkx3.1;Pten mutant mice castrated at 2 months of age and analyzed 2 days later (group II) also did not display massive apoptosis of the prostate, as shown by TUNEL analysis (Supplementary Fig. S1A-D) and instead showed a high proliferation index (Supplementary Fig. S1E-H).Together, these observations indicate that prostate epithelial cells from Nkx3.1; Pten mutant mice are able to survive and proliferate when testicular androgens are initially removed, even before when they display a PIN phenotype.
Acquisition of androgen-independent phenotypes before PIN or cancer formation in Nkx3.1;Pten mutant mice.When analyzed several months following androgen deprivation, a majority of the castrated Nkx3.1;Pten mutant mice displayed androgen-independent high-grade PIN and/or adenocarcinoma (Fig. 2D-G; Table 1).Such androgen-independent lesions were also observed in castrated Pten heterozygous single mutants (Fig. 3C  and F) but were never found in the wild-type mice or in the Nkx3.1 single mutant mice (Fig. 2A and B; Fig. 3A, B, D, and E; Table 1).These findings indicate that loss-of-function of Pten is sufficient for emergence of androgen independence in Nkx3.1;Pten mutant mice.
Because adrenal hormones can potentially activate androgen receptor in the absence of testosterone (28), we investigated whether high-grade PIN lesions also arise in Nkx3.1;Pten mutant mice that have undergone both castration and adrenalectomy, which would deplete all circulating androgens.We found that Nkx3.1;Pten mutant mice, but not the wild-type mice, displayed high-grade PIN lesions following castration and adrenalectomy  (22,25).Below are the experimental paradigms used in this study to investigate androgen independence in Nkx3.1;Pten mutants: I, castrated at 0.75 months and analyzed 1 to 2 days or 7 to 9 months later; II, castrated at 2 months and analyzed 1 to 2 days or 6 to 8 months later; III, castrated at 5 to 8 months and analyzed 1 to 2 days or 2 to 5 months later; IV, castrated at 9 months and analyzed 1 to 2 days or 3 to 6 months later.B to M, analyses of apoptosis and cellular proliferation in castrated Nkx3.1;Pten mutant mice.Wild-type (Nkx3.1 +/+ ; Pten +/+ ) or mutant (Nkx3.1 +/À ; Pten +/À ) mice at 8 months (group III) were analyzed immediately (0 hours) or at 24 or 48 hours after castration for apoptosis by TUNEL assay (B-G), or for proliferation by Ki67 immunostaining (H-M ).Note that 48 hours following castration, the Nkx3.1;Pten mutant mice do not display significant apoptosis (green staining, B-G) and are highly proliferative (brown staining, H-M).Similar data for group II mice are provided in Supplementary Fig. S1.
The histologic features of the androgen-independent high-grade PIN and/or cancer lesions that occur in the castrated Nkx3.1;Pten mice were comparable to lesions that occur in age-matched intact (noncastrated) Nkx3.1;Pten mutant mice (refs.22, 25, 27; Fig. 2, compare C with D-G).These androgen-independent lesions displayed cytoplasmic as well as nuclear localization of androgen receptor, consistent with depletion of endogenous androgens (Fig. 2K-N).Notably, they also displayed robust activation of membrane-localized Akt (Figs. 2Q and R-U), indicative of complete Pten loss-of-function.Activation of Akt was never observed in the wild-type mice either before or after castration (Fig. 2O and P).
To determine the earliest time point at which Nkx3.1; Pten mutant mice can acquire androgen-independent phenotypes, we compared the phenotype of mice castrated at 0.75 or 2 months (groups I and II) with those castrated at 6 months or older (groups III and IV).Surprisingly, we found that most of the Nkx3.1;Pten mutants castrated at 2 months (group II; n = 7 of 8) and half of those castrated at 0.75 months (group I; n = 7 of 12) displayed androgen-independent lesions when examined at 8 to 9 months of age (Table 1).Whereas the group I castrates small, focal PIN lesions, the group II castrates displayed high-grade PIN lesions that were histologically indistinguishable from those of the group III mice, which were castrated at 6 months or older (Fig. 2D and  E).Furthermore, regardless of their age when castrated, many Nkx3.1;Pten mutant mice analyzed at 8 months of age displayed Figure 2. Androgen-independent PIN and cancer lesions in Nkx3.1;Pten mutant mice.Wild-type (Nkx3.1 +/+ ; Pten +/+ ) or mutant (Nkx3.1 +/À ; Pten +/À ) mice were mock-castrated (Mock ) or were surgically castrated at 0.75 months (Group I ), 2 months (Group II ); or 6 months (Group III) and analyzed at 8 months of age, or were castrated at 10 months and analyzed at 15 months (Group IV ).Sections from the anterior prostate were analyzed by H&E staining or by immunostaining for androgen receptor or activated Akt (p-Akt).Representative data from group III for the wild-type mice (mock and castrated) and the noncastrated (mock) mutant mice.Note that in each experimental group, the castrated mutant mice, but not the wild-type mice, display PIN lesions that express androgen receptor and robust levels of p-Akt, although the PIN lesions in the group I mutants are small and focal.*The prostates of these mice were fully regressed without any evidence of PIN.cThe prostates of these mice displayed isolated regions of PIN.bThe prostates of these mice displayed isolated regions of high-grade PIN.
x The prostates of these mice displayed high-grade PIN throughout most of the tissue.
k The prostates of these mice displayed high-grade PIN throughout most of the tissue with areas of adenocarcinoma.
high-grade PIN, whereas those analyzed at 12 months or greater displayed high-grade PIN with invasion (Table 1), which resembles the time course of disease progression in their noncastrated counterparts (22,25).These results indicate that acquisition of androgen independence in Nkx3.1;Pten mice can occur before the onset of PIN or cancer, and therefore suggest that androgen independence can emerge in parallel with disease progression, rather than as an end-point.

Discussion
In this study, we have investigated the acquisition of androgen independence in vivo using a mouse model that is based on loss-offunction of Nkx3.1 and Pten.The major findings of our study are that Nkx3.1;Pten mutant mice can develop androgen-independent phenotypes before the occurrence of overt PIN or cancer and that Pten loss-of-function is sufficient for these androgen-independent phenotypes.Notably, our findings in these mutant mice complement a growing body of evidence supporting the idea that Pten loss-of-function is critical for the development of hormonerefractory cancer in humans.Indeed, recent studies have shown that Pten expression is reduced in hormone-refractory cancer (29) whereas forced expression of Pten in human prostate cancer cell lines promotes their androgen responsivity (30).
In contrast to Pten, we find that loss-of-function of Nkx3.1 is not sufficient to promote androgen independence.However, it is likely that Nkx3.1 loss promotes androgen independence in the Nkx3.1;Pten mutant mice by potentiating the effects of Pten inactivation (25).Indeed, our findings in the Nkx3.1;Pten mutant mice differ somewhat from a previous study showing that conditional deletion of Pten in the prostatic epithelium leads to androgen independence because, in this previous study, the prostate epithelium underwent apoptosis immediately following castration (15).Whereas this difference may be due to conditional deletion of Pten in the prostatic epithelium (previous study) versus germ-line deletion of Pten (our study), it may also reflect the collaborative activities of Nkx3.1 and Pten in our mouse model.The role of Nkx3.1 in this capacity will be interesting to explore in future studies.
Notably, an important distinction between the emergence of androgen independence in Nkx3.1;Pten mice versus the onset of hormone-refractory prostate cancer in humans is the sizeable proportion of prostate epithelial cells that can survive and proliferate immediately following androgen deprivation.This contrasts with the proportion of ''androgen-independent'' cells in human prostate cancer, which has been estimated to represent f1 in every 10 5 to 10 6 cells before androgen ablation (31).We believe that this difference reflects the preexisting uniform loss of one Pten allele in all prostate epithelial cells in the Nkx3.1;Pten mutant mice, greatly increasing the likelihood of complete Pten inactivation, whereas in humans, loss of both Pten alleles (or other possible molecular alterations that promote androgen independence) would presumably be stochastic and occur in a limited cell population.
Consequently, our findings favor a clonal selection mechanism for the emergence of hormone-refractory disease, which predicts that androgen-independent cells are present even in the absence of overt cancer progression, as opposed to an adaptation model in which androgen-dependent cells acquire hormone independence through genetic and/or epigenetic alterations (32).In particular, Craft et al. (31) proposed that androgen independence arises in two distinct stages, with an initial selection for preexisting cells that can survive in the absence of androgens and their subsequent clonal expansion.
In conclusion, our analysis of Nkx3.1;Pten mutant mice has enabled investigation of the relationship between cancer progression and the emergence of androgen-independent disease.Notably, our finding that androgen independence can emerge before overt cancer phenotypes in Nkx3.1;Pten mice has implications for interpreting the outcome of the recently concluded Prostate Cancer Prevention study, which investigated the long-term consequences finasteride, a 5(a)-reductase inhibitor (33).Somewhat paradoxically, although the finasteride-treated population displayed a significant reduction in prostate cancer incidence, those individuals who developed cancer had higher-grade disease (33).Whereas several possible explanations have been discussed (34,35), our current results support the interpretation that finasteride promotes an ''androgen-independent'' state in some prostate cancers, perhaps those with preexisting mutations of .Androgen-independent lesions following adrenalectomy and castration.H&E staining of sections from the anterior prostates of wild-type mice (Nkx3.1 +/+ ; Pten +/+ ) or Nkx3.1;Pten mutant mice (Nkx3.1 +/À ; Pten +/À ) that had undergone adrenalectomy or adrenalectomy as well as castration as indicated; surgeries were done on mice at 6 months and the mice were then analyzed at 8 months.Note that the mutant mice display high-grade PIN lesions following adrenalectomy and castration, which depletes all circulating androgens.Figure 3. Loss-of-function of Pten is sufficient for androgen independence.H&E staining of sections from the anterior prostates of wild-type mice (Nkx3.1 +/+ ; Pten +/+ ), Nkx3.1 single mutant mice (Nkx3.1 À/À ; Pten +/+ ), or Pten single mutant mice (Nkx3.1 +/+ ; Pten +/À ) that were castrated at 2 months (Group II ) or 6 months (Group III) and analyzed at 8 months of age, as indicated.Note that the prostate epithelia from the wild-type and Nkx3.1 single mutants are atrophic whereas the Pten single mutant displays PIN lesions.

Cancer Research
Cancer Res 2006; 66: (16).August 15, 2006 Pten or other genes that enable prostatic cells to survive under conditions of androgen deprivation.This interesting possibility can be tested in the context of the Nkx3.1;Pten mutant mice.

Figure 1 .
Figure 1.Analyses of androgen independence in Nkx3.1;Pten mutant mice.A, diagram of the experimental paradigm.On the top is shown the characteristic progression of prostate cancer in Nkx3.1;Pten mutant mice from low-grade PIN (LGPIN) to high-grade PIN (HGPIN ) to cancer as previously reported(22,25).Below are the experimental paradigms used in this study to investigate androgen independence in Nkx3.1;Pten mutants: I, castrated at 0.75 months and analyzed 1 to 2 days or 7 to 9 months later; II, castrated at 2 months and analyzed 1 to 2 days or 6 to 8 months later; III, castrated at 5 to 8 months and analyzed 1 to 2 days or 2 to 5 months later; IV, castrated at 9 months and analyzed 1 to 2 days or 3 to 6 months later.B to M, analyses of apoptosis and cellular proliferation in castrated Nkx3.1;Pten mutant mice.Wild-type (Nkx3.1 +/+ ; Pten +/+ ) or mutant (Nkx3.1 +/À ; Pten +/À ) mice at 8 months (group III) were analyzed immediately (0 hours) or at 24 or 48 hours after castration for apoptosis by TUNEL assay (B-G), or for proliferation by Ki67 immunostaining (H-M ).Note that 48 hours following castration, the Nkx3.1;Pten mutant mice do not display significant apoptosis (green staining, B-G) and are highly proliferative (brown staining, H-M).Similar data for group II mice are provided in Supplementary Fig.S1.

Figure 4
Figure 4. Androgen-independent lesions following adrenalectomy and castration.H&E staining of sections from the anterior prostates of wild-type mice (Nkx3.1 +/+ ; Pten +/+ ) or Nkx3.1;Pten mutant mice (Nkx3.1 +/À ; Pten +/À ) that had undergone adrenalectomy or adrenalectomy as well as castration as indicated; surgeries were done on mice at 6 months and the mice were then analyzed at 8 months.Note that the mutant mice display high-grade PIN lesions following adrenalectomy and castration, which depletes all circulating androgens.

Table 1 .
Summary of androgen-independent phenotypes NOTE: Summary of mice analyzed in each group and the phenotypes.