An oncogene-tumor suppressor cascade drives metastatic prostate cancer by coordinately activating Ras and nuclear factor-kappaB

Abstract

Metastasis is responsible for the majority of prostate cancer-related deaths; however, little is known about the molecular mechanisms that underlie this process. Here we identify an oncogene-tumor suppressor cascade that promotes prostate cancer growth and metastasis by coordinately activating the small GTPase Ras and nuclear factor-kappaB (NF-kappaB). Specifically, we show that loss of the Ras GTPase-activating protein (RasGAP) gene DAB2IP induces metastatic prostate cancer in an orthotopic mouse tumor model. Notably, DAB2IP functions as a signaling scaffold that coordinately regulates Ras and NF-kappaB through distinct domains to promote tumor growth and metastasis, respectively. DAB2IP is suppressed in human prostate cancer, where its expression inversely correlates with tumor grade and predicts prognosis. Moreover, we report that epigenetic silencing of DAB2IP is a key mechanism by which the polycomb-group protein histone-lysine N-methyltransferase EZH2 activates Ras and NF-kappaB and triggers metastasis. These studies define the mechanism by which two major pathways can be simultaneously activated in metastatic prostate cancer and establish EZH2 as a driver of metastasis.

Figure 1. DAB2IP-suppression induces prostate tumor development

(a) Immunoblots showing expression of NF1, DAB2IP or p120RasGAP in E1A and DNp53-expressing MEFs, following lentiviral shRNA expression (left). Colony formation in soft agar (right). Significance determined by a Students t-test. (P = 3.7 × 10⁻⁸, P = 8.6 × 10⁻⁹). (b) Relative size of colonies resulting from loss of DAB2IP versus NF1. Significance determined by a Students t-test. (P = .046) (c) Colony formation of immortalized PrEC cells infected with lentiviruses expressing DAB2IP shRNA sequences. (d) Immunoblots of cells from c to assess DAB2IP knock-down and activation of Ras, ERK and AKT as described in Methods. (e) Colony growth induced by DAB2IP-suppression using an shRNA directed to the 3′UTR, or rescued with wild-type DAB2IP. (f) Immunoblots of cells used in e to assess DAB2IP expression and activation of Ras, ERK and AKT. (g) Histological micrographs of tumors resulting from orthotopic injection of PrECs. Top row: hematoxylin and eosin (H&E), middle row: LgT immunohistochemistry, bottom row: Ki67 staining. Note: the first column of images represent prostate tissue from an animal in which a control lesion had regressed. The second column of images is from a small control lesion. The last column is from a prostate in which a “rescued” lesion had regressed. (h) Immunohistochemical staining of H-Ras^V12 and DAB2IP-deficient tumors using antibodies to cytokeratin 8, p63, and PSA. (i) Volume of orthotopic tumors over time, calculated by Xenogen imaging (Students t-test (p=.13)). (j) Final weight of orthotopic tumors at the time of death (Students t-test (p=.34). All scale bars = 200μM.

Figure 2. DAB2IP-loss, but not H-Ras^V12, promotes invasion and widespread metastasis

(a) Histological micrographs of primary tumors from mice injected with H-Ras^V12-expressing PrEC cells. Scale bar = 5mm. (b) Histological micrographs of primary tumors from mice injected with DAB2IP-deficient PrEC cells. Immunohistochemistry using a LgT antibody is shown in the second panel. Scale bar = 5mm, except lumbar muscle and bladder, scale bar = 1mm. (c) Histological micrographs of metastases from mice injected with DAB2IP-deficient PrEC cells. Immunohistochemistry using a LgT antibody on an adjacent section is shown at higher power below each metastasis, with the exception of the lymph node, where on H&E staining is shown. Scale bars (Liver and SV) = 5mM; (LN, testis, VD) = 1nM. (d) H&E stained sections of tissues containing tumor cells (T) within blood vessels (BV and arrows) and lymphatic vessels (LV). When specified by lines higher magnification images are shown. Two panels show LgT staining. Scale bars (panels 1, 2, 4) =1 mM; (panels 3, 5, 6) = 200μM. (e) Xenogen images of mice orthotopically injected with luciferase-expressing control, DAB2IP-deficient, or H-Ras^V12-expressing cells. (f) Survival curves of animals with DAB2IP-deficient versus H-Ras^V12-expressing tumors.

Figure 3. The RasGAP activity of DAB2IP underlies some but not all of its tumor/metastasis suppressor function

(a) Immunoblots comparing DAB2IP expression in PrEC, PC-3, and PC-3 cells expressing wild-type DAB2IP or DAB2IP-R289L. (b) Immunoblots assessing effects of DAB2IP and DAB2IP-R289L on Ras, ERK and AKT activation in PC-3s, as described in Methods (c) Xenogen images of mice injected with PC-3 cells from a and b. (d) Volume of PC-3 derived tumors calculated by Xenogen images. (e) Dot plot depicting number of metastases per mouse in animals injected with PC-3 cells expressing wildtype DAB2IP or DAB2IP-R289L. (f) DAB2IP immunoblot corresponding to e. (g) (Left) Bar graph showing final weight of tumors from PrECs. (Right) A dot plot depicting number of metastases per mouse in animals injected with PrEC cells expressing DAB2IP shRNA1 or 2 alone or with the wild-type or DAB2IP-R289L. (Mann-Whitney U; P = 0.0030) (h) (Top row, left panel) Phase images of immortalized PrEC cells expressing a lentiviral vector control, a DAB2IP-specific shRNA, or an activated Ras allele. (Middle row, left panel) Immunofluorescent staining of the mesenchymal marker vimentin or (Bottom row, left panel) the epithelial marker E-cadherin. (Middle panel) Expression of DAB2IP, ectopic Ras, fibronectin, vimentin, E-cadherin as detected by immunoblot. (Right panel) Real time PCR was performed to quantify vimentin, E-cadherin and twist expression in response to DAB2IP-loss or an activated Ras allele. (i) Confocal images of immunofluorescent staining of vimentin (green) and E-cadherin (red) in DAB2IP-deficient or H-Ras^V12-expressing tumors. All scale bars = 100μM.

Figure 4. DAB2IP-loss promotes tumorigenesis and metastasis via concomitant effects on Ras and NF-κB

(a) (Left) NF-κB activity reported as relative luciferase units (RLU) in response to DAB2IP suppression in PrEC cells. (Middle) NF-κB activity in response to DAB2IP reconstitution in PC-3 cells. (Right) DAB2IP immunoblot. (b) Expression of NF-κB targets as determined by Q-PCR in PrEC cells with and without the DAB2IP shRNA in the absence of TNF-α, (c) Expression of NF-κB targets as determined by Q-PCR in PC-3 cells with and without the DAB2IP cDNA, in the presence of TNF-α. (d) DAB2IP immunoblots of reconstituted PC-3 cells. IκB super-repressor expression shown in Supplementary Fig. 7b. (e) Tumor volumes derived from PC-3 cells calculated from xenogen images. (f) A dot plot depicting the number of metastases detected in each mouse injected with PC-3 cells. (g) Xenogen images of mice injected with PC-3 cells from d-f. The Ras and NF-κB status noted above summarizes the biochemical effects of expressed mutants. The numbers below the images represent the number of animals that develop metastases/the number of animals injected. (h) Final weight of tumors derived from control PrEC cells, PrEC cells with the 3′UTR shRNA (3′UTR), 3′UTR cells reconstituted with DAB2IP (cDNA) or the S604A mutant. Expression was determined to be equivalent (not shown). (i) A dot plot depicting the number of metastases detected in each mouse injected with the PrEC cells in h. (j) Immunohistochemistry using antibodies that recognize NF-κB, pERK, and pAKT in DAB2IP-deficient and H-Ras^V12-expressing tumors. Scale bars = 200μM.

Figure 5. EZH2 promotes tumorigenesis and metastasis via suppression of DAB2IP

(a) Histological images of tumors and metastases from orthotopic injection of EZH2-expressing PrEC cells. Tumor cells observed within lymphatic vessels (LV) and blood vessels (BV), circled. Metastasis data for all animals is in Supplementary Table 1. (b) (Left) DAB2IP immunoblots of immortalized PrEC cells expressing a vector control or EZH2. The top band is EZH2. (Right) Immunoblots of cells in b reconstituted with ectopic DAB2IP-V5. (c) Suppression of endogenous DAB2IP mRNA expression in response to EZH2 as determined by real time PCR. (d) Chromatin immunoprecipitation of EZH2 bound to the DAB2IP promoter. (e) Tumor volume calculated from xenogen images of animals injected with cells expressing EZH2 or EZH2 and DAB2IP. (f) A dot plot depicting the number of metastases detected in each mouse injected with EZH2-expressing or EZH2/DAB2IP-expressing PrEC cells. (g) Survival curves of mice injected with orthotopically injected EZH2-expressing or EZH2/DAB2IP-expressing PrEC cells. (h) Immunohistochemistry of EZH2-expressing or EZH2/DAB2IP-expressing PrEC derived tumors using p-ERK, p-AKT and NF-κb (p50) antibodies. (Right panels). Confocal immunofluorescence images of of EZH2-expressing or EZH2/DAB2IP-expressing PrEC derived tumors using human vimentin (green) e-cadherin (red) antibodies. (i) Immunoblots assessing effects of EZH2 and DAB2IP reconstituion on Ras, ERK and AKT activation in PrEC cells, as described in Methods (j) Final weight of tumors and number of metastases detected in mice injected with cells shown in i. All scale bars = 200μM, except confocal images where scale bars= 100μM.

Figure 6. DAB2IP is suppressed in human prostate cancer

(a) Model of how EZH2 regulates DAB2IP, Ras and NF-κB to promote tumor development and metastasis. (b) DAB2IP immunohistochemical staining of normal mouse prostate, DAB2IP-deficient PrEC-derived tumors or Ras-driven tumors. (c) (Top) Immunohistochemical staining with a DAB2IP antibody in normal human prostate epithelium. (Bottom) Immunohistochemical DAB2IP staining in normal human tissue and adjacent tumor tissue (T) within the same biopsy. (d) Immunohistochemical staining with a DAB2IP antibody in human prostate cancer tissues. (Right) High power magnification of images. Arrows depict stromal cells. (e) Histogram of DAB2IP protein expression in prostate intraepithelial neoplasia (PIN) and in prostate cancer (PCa) compared to the benign prostate tissue (P = 1.8 × 10⁻²⁷). (f) Histogram comparing DAB2IP expression and Gleason grade (not score), evaluated as the most prevalent pattern on each individual core (P = 0.001). Because the tissue sections on the arrays are relatively small, they were assigned a Gleason grade (1–5), rather than a Gleason score (1–10), the latter of which represents the combined total of the two most prevalent grades in an entire tumor. As such, Grade 4 and 5 tumors are high-grade tumors. (g) Box plots of DAB2IP mRNA expression data in prostate cancer compared to normal tissue (P = 2.0 × 10⁻⁷) and in high grade tumors as opposed to low grade and benign tissue (P = 0.003). (h, i) Expression of DAB2IP levels versus EZH2 levels in individual tumors and metastatic lesions. All scale bars = 400 μM.