RapidCaP, a novel GEM model for metastatic prostate cancer analysis and therapy, reveals myc as a driver of Pten-mutant metastasis

Abstract

Genetically engineered mouse (GEM) models are a pillar of functional cancer research. Here, we developed RapidCaP, a GEM modeling system that uses surgical injection for viral gene delivery to the prostate. We show that in Pten deficiency, loss of p53 suffices to trigger metastasis to distant sites at greater than 50% penetrance by four months, consistent with results from human prostate cancer genome analysis. Live bioluminescence tracking showed that endogenous primary and metastatic disease responds to castration before developing lethal castration resistance. To our surprise, the resulting lesions showed no activation of Akt but activation of the Myc oncogene. Using RapidCaP, we find that Myc drives local prostate metastasis and is critical for maintenance of metastasis, as shown by using the Brd4 inhibitor JQ1. Taken together, our data suggest that a "MYC-switch" away from AKT forms a critical and druggable event in PTEN-mutant prostate cancer metastasis and castration resistance.

Figure 1. Stable transgene delivery to epithelial prostate cells by virus injection

A. The RapidCaP system. 1- Design and production of lentivirus harboring candidate cancer genes and marker genes. 2- Surgical injection of virus directly into anterior prostate gland. 3- Non-invasive bioluminescence imaging to monitor disease progression and therapeutic effects. Note that Adenovirus is depicted for purely aesthetic reasons. B. Live imaging of lentiviral luciferase (LV-Luci) injected Bl6 mice and non-injected control at 60 days post injection suggests persistent transgene expression. C. Post-mortem autopsy imaging of LV-Luci and non-injected control mice at 98 days post injection. The genito-urinary (GU) tract comprises bladder (B), seminal vesicle (SV) and anterior prostate (AP). Luminescence signal is found in the (injected) right anterior prostate lobe and also in the seminal vesicle. Note that FACS sorting experiments fail to identify infected (tomato-FP) seminal vesicle cells, potentially suggesting SV signal to be extracellular. PCR of prostate genomic DNA reveals the luciferase transgene (300 bp) only in the LV-Luci injected prostate lobe. D. Anti-luciferase immunofluorescence on LV-Luci- and non-injected control prostate reveals luciferase expression in prostate luminal epithelium. Scale bar, 47μm. E. H&E staining of LV-Luci injected and non-injected control prostate reveals retention of normal glandular architecture. Scale bar, 100μm.

Figure 2. Prostate specific LV-Cre/ Luci delivery results in focal disease

A. Live and post-mortem images of LV-Cre/Luci injected Pten^loxP/loxP; Trp53^loxP/loxP mouse at 69 days post injection. Dissection reveals similar distribution to that shown in Fig. 1C. Luminescence signal was seen in LV-Cre/Luci injected prostate and seminal vesicle. B. PCR analysis confirms recombination of Pten and Trp53 in LV-Cre/luci injected Pten^loxP/loxP; Trp53^loxP/loxP specifically in the injected right (Rt) anterior prostate. C. Immunohistochemistry analysis of LV-Cre/luci injected Pten^loxP/loxP; Trp53^loxP/loxP prostate reveals a focal lesion in the luminal epithelium of one gland (H&E), which also displays increased proliferation, specific to epithelial cells (Ki-67). Loss of Pten expression, activation of Akt (pAkt), and increased proliferation are also found in this region (middle panels), which is a unique focus of disease in the cross section of the entire AP (right panels). Note that surrounding stroma remains unaffected in all panels. Scale bars, 100μm (top right), 50μm (all other panels).

Figure 3. Pten/ Trp53 deletion triggers disease dissemination

A. Time course of a typical RapidCaP experiment (showing one LV-Cre/Luci injected Pten^{loxP/ loxP}; Trp53^loxP/loxP mouse) reveals disease spreading to distant sites starting at 72 days post injection. B. Quantification of signal intensity from mouse shown in Fig. 3A. Total body, secondary and primary signal measurements are shown. C. Kaplan-Meier analysis of disease dissemination reveals early onset and high penetrance, specifically in Pten^loxP/loxP; Trp53^loxP/loxP mice (p<0.0001). D. Live (219 days post injection) and post-mortem autopsy of an LV-Cre/Luci injected Pten^loxP/loxP; Trp53^loxP/loxP mouse reveals luciferase positive secondary organ sites (lymph-nodes, LN, spleen and liver) and a minor difference in size of prostate lobes (injected, inj. vs. non-injected, ni). Scale bar, 1cm. E. PCR analysis reveals Pten/ Trp53 recombination in the secondary tissues (weakest in kidney). Scale bar, 1cm. F. H&E staining of an LV-Cre/Luci injected Pten^loxP/loxP; Trp53^loxP/loxP prostate lesion (192 days post injection) reveals a typical budding gland with features of focal invasion. Scale bar, 50μm.

Figure 4. Histogenic analysis confirms prostate cancer metastasis to lung

A. Autopsy imaging of the mock-castrated RapidCaP mouse (#3) from Fig. 4 reveals disease dissemination to lung, lymph nodes, spleen and liver at the trial endpoint (10 months post injection). B. Identification of histogenic markers that define prostate cancer cells by immunohistochemistry (IHC) analysis of prostate in animal from A: Pten & Ck8 (black arrows, see black arrowheads for region with normal protein levels) and Pten & Ar (red arrows, see red arrowheads for region with normal protein levels). Note that ‘loss’ of AR-staining is a hallmark of Pten/p53-null prostate (see text). Scale bar, 100μm. C, D. IHC analysis of lung from above mouse reveals metastatic nodules that are Pten-, Ck8- and AR-negative and positive for the prostate epithelial marker Nkx3.1. Boxes indicate areas of magnification. Scale bars, 100μm (top & bottom left panels), 50μm (all other panels). E. Double-immunofluorescence (IF) labeling confirms double-negativity for Pten & Ck8 in a metastatic lung nodule (nod) shown in (C-D) (yellow circles & arrowheads). Note that in contrast, co-labeling of Pten and Ck8 in adjacent normal lung epithelia (nl, white dashed circles and arrowheads) shows double positive staining. Scale bar, 50μm. F. Anti-luciferase antibody staining and IF analysis of lung from (C-F) confirms that metastatic nodules are luciferase positive (white arrows) and Pten / Ck8 deficient (red arrowheads). Scale bar, 100μm.

Figure 5. Metastatic signals respond to castration

A. Live imaging time course of two castrated RapidCaP mice (LV-Cre/Luci injected Pten^loxP/loxP; Trp53^loxP/loxP mouse). Surgical castration was performed at 5 months post injection on mice harboring distant secondary disease (see Castr.). Imaging analysis reveals differential response and recurrence of disease after castration. B. 3D-plot of signal intensity time course from (A), including control mice: 1, wt untreated mouse- 2, RapidCaP mouse- 3, mock-castrated RapidCaP mouse- 4&5, castrated RapidCaP mice. C. Quantification of luminescence signals in castrated animals reveals a sharp increase in the rate of disease progression after relapse. “C” denotes castration. D. Close-up analysis of graphs from (C) shows that both primary and secondary disease respond to castration, confirming their hormone dependence.

Figure 6. Analysis of castration resistant prostate cancer

A. Post-mortem analysis of Cast-1 animal after relapse (see Fig. 4A) shows metastasis to distant organs and massive prostate enlargement. B. Castration results in prostate epithelium atrophy and diffuse cytoplasmic Ar staining (upper panels). Note that Pten/ Trp53-loss in RapidCaP does not cause a significant increase in prostate size (lower panels). Scale bar, 100μm. C. IHC analysis of Cast-1 prostate shows loss of Akt-activation, lost/ low cytokeratin epithelial and basal cell markers Ck8 and CK5, respectively. In contrast, there is strong activation of Myc in tumor cells in the therapy resistant prostate tumor. Scale bars, 1mm (Myc overview), 50μm (all other panels). D. Western blotting analysis of PI 3-Kinase reveals no feedback (p-S6) or parallel pathway (p-Pdk1) activation of castration resistant tumor shown in (C). E. A strong increase in Myc mRNA expression is detected in the castration resistant tumor from. Error bars are s.d., ***p = 0.0001 (C). F. FISH analysis reveals Myc gene amplification in disease progression from prostate to lung metastasis of the same animal. Myc gene amplification is also observed in a castration resistant tumor (CRPC). Percentage of nuclei with > 2 copies is indicated, scale bar 5μm.

Figure 7. Myc succeeds Akt activation and is required for growth of metastatic prostate cancer

A. IHC analysis of phospho-Akt (Ser473) in metastatic lung nodules of LV-Cre/Luci injected Pten^loxP/loxP; Trp53^loxP/loxP mice reveals absence of pAkt activation. In contrast, Myc levels are strongly increased, especially in the cytoplasm (compared to prostate, Fig. S3F). Note the apparent further Myc increase in castration resistant disease (Fig. 6C) compared to this mock-castrated metastasis. Scale bar, 50μm. B. Top, typical live imaging of RV-Myc-luci injected Pten^hy/+ mice shows lower abdominal signals and absence of distant metastasis (in contrast to LV-Cre/Luci injected Pten^loxP/loxP; Trp53^loxP/loxP mice, Fig. 3A). Note that Pten^hy/+ mice express only 75% Pten throughout their body. Bottom, Kaplan-Meier analysis of disease spread in indicated RapidCaP models reveals that Myc expression can drive local dissemination of prostate cancer and cooperate with Pten-suppression. Log-rank (Mantel-Cox) testing shows that wt, Pten^hy/+, as well as negative control study arms are significantly different (p = 0.004). C. Autopsy analysis of RV-Myc-luci injected mice reveals local, mostly lower abdominal disease dissemination from prostate (yellow arrows) to secondary sites (green arrows), including local lymph nodes. See also Supplementary Fig. S7C. D. Treatment of metastasis in LV-Cre/Luci injected Pten^loxP/loxP; Trp53^loxP/loxP mice using the Myc antagonizing Brd4 inhibitor JQ1 (at 50 mg/kg/day) reveals regression of metastatic disease compared to the disease progression in DMSO treated mice. E. Quantification of disease regression from (D) shows significant effect on metastasis (*p = 0.026) but not primary disease. Error bars are s.d.