Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2013 Jan 15;73(2):978-89.
doi: 10.1158/0008-5472.CAN-12-2133. Epub 2012 Nov 30.

Novel Oncogene-Induced Metastatic Prostate Cancer Cell Lines Define Human Prostate Cancer Progression Signatures

Free PMC article

Novel Oncogene-Induced Metastatic Prostate Cancer Cell Lines Define Human Prostate Cancer Progression Signatures

Xiaoming Ju et al. Cancer Res. .
Free PMC article


Herein, murine prostate cancer cell lines, generated via selective transduction with a single oncogene (c-Myc, Ha-Ras, and v-Src), showed oncogene-specific prostate cancer molecular signatures that were recapitulated in human prostate cancer and developed lung metastasis in immune-competent mice. Interrogation of two independent retrospective cohorts of patient samples using the oncogene signature showed an ability to distinguish tumor from normal prostate with a predictive value for prostate cancer of 98% to 99%. In a blinded study, the signature algorithm showed independent substratification of reduced recurrence-free survival by Kaplan-Meier analysis. The generation of new oncogene-specific prostate cancer cell lines that recapitulate human prostate cancer gene expression, which metastasize in immune-competent mice, are a valuable new resource for testing targeted therapy, whereas the molecular signatures identified herein provides further value over current gene signature markers of prediction and outcome.

Conflict of interest statement

Conflicts of Interest: R.G.P. holds minor (< $10,000) ownership interests in, and serves as CSO/Founder of the biopharmaceutical companies ProstaGene, LLC and AAA Phoenix, Inc. R.G.P. additionally holds ownership interests (value unknown) for several submitted patent applications.


Figure 1
Figure 1. Oncogene transduced PEC lines form colonies in soft agar
(A)Phase contrast microscopy of oncogene induced cell lines were transduced by distinct oncogenes (c-Myc, NeuT, Ha-Ras, v-Src). Photo of individual colonies derived from oncogene-transduced PEC that were selected and characterized. (B) Growth curves of PEC lines determined by cell counting. Data are mean ± SEM of N>3 separate experiments. (C, D) Western Blot analysis of 3 separate clones of each oncogene induced PEC with antibodies as shown for detection of c-Myc, NeuT, Ha-Ras and v-Src and (D) markers of basal (CK5) vs luminal (CK8) prostate cancer. GDI is used as a protein loading control. (E) Soft agar assays of oncogene transduced PEC. Non-transformed PEC failed to grow in soft agar. The size and number (F) of colonies from oncogene transduced PEC lines are shown as mean ± SEM of N>5 separate experiments.
Figure 2
Figure 2. Copy number aberrations in the four oncogene cell lines assessed by array CGH
(A) The percentage of the four cell lines sharing copy gain or loss regions is shown as a function of genomic position. (B) Regions of copy gain (red) or loss (blue) for each of the four cell lines are shown as a function of genomic position. (C) Oncogenes are identified with mRNA over-expression (red), DNA amplification (yellow), or both (purple) among the four oncogene cell lines, with corresponding amplification in the MKSCC prostate cancer database (listed on right-hand side). (D) Tumor suppressor genes are identified with mRNA under expression only (blue) or both mRNA under expression and DNA copy loss (orange) among the four oncogene cell lines, with corresponding copy loss in the MKSCC prostate cancer database (listed on right-hand side).
Figure 3
Figure 3. Prostate epithelial cell lines grow in immune competent mice
(A) PEC tumor diameter determined by vernier caliper measurement is shown as days after inoculation in FVB mice. The diameter mean ± SEM for N>5 separate experiments. (B) Photograph of representative tumor derived from oncogene-induced lines. NeuT induced tumors were harvested at 15 days after cell injection. (C) Hematoxylin and eosin staining at low and high magnification (see also Supplemental Fig. 1).
Figure 4
Figure 4. Oncogene transformed prostate epithelial cell tumors metastasize to lung
(A) Hematoxylin and eosin stain of murine lung post tumor implantation demonstrating representative example of lung metastasis. (B) Frequency of lung metastases were detected in mice for c-Myc, NeuT and v-Src PEC groups 5 weeks after subcutaneous injection. The rates were 100% frequency in Ha-Ras and v-Src groups.
Figure 5
Figure 5. Differential gene expression patterns in transformed PEC cell lines
A heatmap (A) provides an overview of the differential expression patterns among four distinct oncogene transformed cell lines. Heatmaps of genes that are differentially expressed in the four oncogene transformed PEC lines and differentially expressed genes in (B) high grade vs. low grade and (C) advanced stage vs early stage prostate cancer (4). Heatmaps of the left-hand side represent the prostate cancer high grade and advanced stage signatures, while heatmaps on the right represent genes that are differentially expressed in at least one of the four prostate cancer cell lines. The percentage of the 34 high-grade and 72 advanced-state genes that are differentially expressed within each individual prostate cancer cell line is shown in the respective columns along with the p value for the degree of similarity with the high grade and advanced stage phenotypes.
Figure 6
Figure 6. c-Myc- and Ha-Ras-specific oncogene signatures in prostate tumors are conserved in other tissues
Heatmaps show genes that are differentially expressed in the oncogene-induced prostate cancer cell lines and in (A) Ha-Ras and (B) c-Myc fibroblasts (3T3 cell line). (C) A heatmap shows the intersection of genes that are differentially expressed in the c-Myc prostate cancer cell line and mouse mammary tumor samples. The p values shown under each prostate cell line heatmap represent the significance of the overlap between the prostate and fibroblast/mammary tumor signatures. (D) Kaplan Meier curves are shown for high (upper 25th percentile) and low (lower 75th percentile) expression populations for the c-Myc overexpression signature used to interrogate the clinical data of (2).
Figure 7
Figure 7. The c-Myc-specific expression profile distinguishes tumor from normal tissue
(A) Hierarchical clustering performed in the subset genes exclusively deregulated in the c-Myc cell lines separates normal (green), localized tumor (light blue), and tumor metastasis (pink) samples. (B) A classifier based on canonical analysis of c-Myc signature distinguishes human tumor (red) from normal tissue (light blue), along the x-axis, in the Lapointe 2004 dataset. ROC curves for the classifier performance are shown for (C) the Lapointe 2004 dataset and (D) the Taylor 2010 MSKCC dataset, with AUC values of 0.990 and 0.977, respectively.

Similar articles

See all similar articles

Cited by 13 articles

See all "Cited by" articles

Publication types