Locally-invasive, castrate-resistant prostate cancer in a Pten/Trp53 double knockout mouse model of prostate cancer monitored with non-invasive bioluminescent imaging

Here we have improved an existing mouse model of prostate cancer based on prostate-specific deletion of Pten and Trp53 by incorporating a Cre-activatable luciferase reporter. By coupling the deletion of those genes to the activation of a luciferase reporter, we were able to monitor tumor burden non-invasively over time. We show that, consistent with previous reports, deletion of both Pten and Trp53 on a C57BL/6 background accelerates tumor growth and results in both the loss of androgen receptor expression and castrate resistant tumors as compared with loss of Pten alone. Loss of Trp53 results in the development of sarcomatoid histology and the expression of markers of epithelial-to-mesenchymal transition Zeb1 and vimentin, with kinetics and penetrance dependent on whether one or both alleles of Trp53 were deleted. Homozygous deletion of Trp53 and Pten resulted in uniformly lethal disease by 25 weeks. While we were able to detect locally invasive disease in the peritoneal cavity in aggressive tumors from the double knockout mice, we were unable to detect lymphatic or hematogenous metastatic disease in lymph nodes or at distant sites.


42
Prostate cancer is the most commonly diagnosed malignancy in men worldwide 43 (1). While the 5-year survival rate is high overall and has improved with prostate specific

57
The first genetically-engineered mouse model for prostate cancer was the 58 TRAMP model (5). TRAMP mice are engineered for prostate-specific expression of both 59 the large and small T-antigen, which results in the inhibition of Trp53 and Rb1 as well as 60 the activation of AKT. While this model exhibits metastasis, its pathology more closely 61 reflects rare high-grade, neuroendocrine-like disease, i.e. small cell carcinoma (6, 7). In 62 an attempt to build better models of prostate cancer, investigators have utilized 64 Homozygous deletion of Pten using Cre recombinase controlled by a prostate-65 specific probasin promoter initially demonstrated lethal metastatic prostate cancer in 66 mice (5). However, subsequent efforts found indolent disease in prostate-specific Pten 67 knockout mice extensively backcrossed onto the C57BL/6 background (9, 10). Chen,et 68 al. showed that combining deletion of Trp53 with Pten on a C57BL/6 background 69 unleashed a more aggressive, lethal phenotype with mice succumbing to bulky, locally prostate intra-epithelial neoplasia to adenocarcinoma and ultimately sarcomatoid 73 pathology, coupled with a progressive decrease in androgen receptor staining (11).

74
More recently, the Goodrich group has combined deletion of Rb1 with Pten and/or p53 75 resulting in aggressive, metastatic disease including bone metastasis (12). The addition 76 of a Trp53 deletion conferred castrate-resistance to the Rb1; Pten knockout tumors 77 (12). However, with the deletion of Rb1 in this model the resulting disease more closely 78 resembled high-grade neuroendocrine prostate cancer. 79 We have previously described a prostate-specific Pten knockout mouse with a 80 genetically engineered luciferase reporter that allowed quantitative, noninvasive 81 monitoring of disease progression (9). However, as mentioned above, this model did not 82 develop metastatic disease, and castration resulted tumor regression with no relapse.

83
Here we have combined loss of Trp53 with Pten with a luciferase reporter on an 84 extensively backcrossed albino C57BL/6 background to facilitate bioluminescence 85 imaging (BLI). We show that these mice rapidly develop fatal, bulky, locally invasive 86 tumors. We also note the consistent development of sarcomatoid pathology and 87 epithelial-to-mesenchymal transition (EMT)-like features in these tumors. Despite the 88 presence of these features, we did not detect disseminated metastatic disease via BLI 89 or histologic analysis, supporting previous findings with this model.

176
The longitudinal BLI imaging (Fig 1 A, B) demonstrated that tumors grew faster in the phase in all groups from 5-10 weeks, which reflects pubertal activation of Cre and 179 growth of the prostate, WT mice showed little growth out to 50 weeks consistent with 180 our previous studies (9). The DKO mice showed rapid tumor growth around 19 weeks of 181 age that resulted in formation of readily palpable primary tumors and a median survival 182 of 26.5 weeks (Fig 1B, C). In comparison to the DKO group, the HET group 183 demonstrated slower disease progression with tumor growth not increasing until mice 184 were around 29 weeks of age (Fig 1B). Moreover, the median survival for heterozygous 185 Trp53 null mice was closer to that of WT (41 and 50 weeks, respectively) (Fig 1C).  We performed histopathological analysis of prostates collected from DKO, HET,195 and WT mice euthanized at 25 weeks of age, as well as HET and WT mice at ~40 196 weeks (Fig 2A). At 25 weeks, 13/13 mice from the DKO group demonstrated  was metastatic disease in these mice. We evaluated residual tumor burden by BLI post-222 necropsy, after removal of the primary prostate tumor for 10 out of 16 mice (Fig 3A, D, 223 F). This was necessary because the intense signal from the primary tumor could 224 obscure weaker signals emanating from metastatic sites. We observed intraperitoneal 225 spread in 8 out of 10 mice. We then histologically evaluated BLI signal-positive tissue to 226 determine if there was metastatic disease (Fig 3B, C, E, G). Upon histological 227 evaluation of a mouse that appeared to have lumbar lymph node involvement (Fig 3A), 228 we found that instead of intralymphatic growth, the tumor invaded into the local 229 perilymphatic fat (Fig 3B, C). DKO mice (8/10) were found to have visceral BLI signal 230 upon post-necropsy imaging (Fig 3D), but this seems more likely to be peritoneal 231 seeding and local invasion than hematogenously-disseminated disease (Fig 3E). There 232 was also evidence of aggressive, invasive disease with BLI-positive tumor invading into 233 pelvic sidewall musculature (Fig 3F, G).

250
While the DKO tumors were not metastatic, we wished to investigate the androgen 251 sensitivity of this model. To assess this, we generated a cohort of five DKO mice which were 252 surgically castrated at 10 weeks of age. After castration, the BLI signal noticeably decreased 253 between 10 and 15 weeks of age (Fig 4A). Eight weeks after castration, the tumor burden 254 started to increase in all mice, demonstrating castrate-resistant prostate cancer. Histological 255 analysis from these tumors revealed that, like the intact DKO mice, they were all sarcomatoid 256 (Fig S2A, B). Due to the rapid and uniform onset of castrate-resistant growth, we determined if 257 the tumors from intact mice demonstrated alterations in AR expression (Fig 4B, C). From this 258 analysis, we found that ~75% of DKO mice lacked AR protein expression at 25 weeks, but all of 259 the HET and WT mice retained AR expression at this timepoint (Fig 4C). At 40 weeks, ~50%

260
HET mice showed loss of AR expression. Consistent with the observation that tumors from the 261 castrate DKO mice became androgen resistant, they also lacked AR expression (Fig S2C).    Acknowledgements: 356 We would like to thank Sophia Peterson for assistance with histopathology and Dr.