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. Cancer Res; 73(2); 978–89. ©2012 AACR.

Prostate cancer remains the most common nondermatologic cancer in the United States and the second leading cause of cancer-related death among men with approximately 190,000 new cases diagnosed and approximately 27,000 deaths annually. The adoption of screening with serum prostate-specific antigen (PSA) has led both to the earlier detection of prostate cancer and to the improved understanding of the biologic heterogeneity of the disease. Risk stratification following local therapy has used histopathologic criterion, Gleason scale, tumor grade, the degree of tumor differentiation, and serum PSA. The diverse clinical behavior of prostate cancer with some patients dying within 2 years of diagnosis and others living for 20 years with localized disease emphasizes the need for improved prognostic tests.

The molecular mechanisms contributing to prostate cancer recurrence and therapy resistance are poorly understood. Androgen ablation therapy results in a 60% to 80% initial response rate (1), however, approximately 30% of men who receive radical prostatectomy relapse, attributed to micrometastatic disease. Clinical classification nomograms are used that incorporate individual pre- and postoperative parameters. These clinically efficient models have limited value in predicting the diverse outcomes of patients with similar clinical, histologic, and biochemical parameters. A global pathway analysis view of the prostate cancer genome identified the most commonly altered pathways as the androgen receptor (AR), retinoblastoma (Rb), phosphoinositide 3-kinase (PI3K), and Ras/Raf signaling (2). The use of gene expression analysis has identified a subset of gene clusters that correlated in retrospective analysis with therapy outcomes (3, 4). Biomarkers derived from gene expression profiling studies that use the same dataset for signature discovery and validation (3, 5, 6) typically however, showed limited use when used to reinterrogate independent datasets (7).

Molecular analysis of human prostate cancer has shown increased c-Myc expression and/or copy number in up to 30% of prostate tumors. Murine prostate tumors induced by prostate-targeted c-Myc give rise to tumors with molecular features of human prostate cancer (8). c-Myc that is overexpressed in human prostate cancer is sufficient to induce prostatic intraepithelial neoplasia (PIN) in transgenic mice (8). In contrast, Ras mutations are relatively uncommon, however, induction of Ras effector pathways has been revealed by microarray and genomic analysis (2). Upregulation of Ras-mediated signaling cascades may also reflect the overexpression of autocrine and paracrine factors [EGF receptor (EGFR), TGF-α; ref. 9]. Although there are conflicting results over whether ErbB2 is overexpressed or amplified in prostate cancer, a subpopulation of patients with prostate cancer showed a gene expression pathway of activated ErbB2 signaling (10). Activation of the Src pathway is also found in primary prostate cancer associated with a truncated version of c-kit (11), and a subset of human prostate cancers expresses a Src signature (10).

The current studies were conducted to address several key needs. First, the development of new therapies for prostate cancer requires model systems that closely resemble human disease. Unfortunately, to date there have been several limitations with currently available cell lines. Although important transplantation experiments have been conducted using human prostate cancer cell lines in immune-deficient animals, the immune system plays an important role in prostate cancer onset and progression (12, 13) making it imperative to develop prostate cancer cell lines that can be studied in immune competent animals. Second, genome-wide expression studies of prostate cancer cell lines showed only a small number of genes with concordant expression in malignant prostate tissue (14). Third, prostate cancer in humans is associated with bone and lung metastasis, however, only a single cell line, a product of Myc and Ras transduction has been developed that develops bone metastasis in immune-competent mice (15). Fourth, although the transgenic mouse had been effectively deployed as a model to study the molecular basis of human cancers, the prostate cancer mouse models have long latency and often unpredictable metastasis (16).

The current studies aimed to generate oncogene-specific cell lines that could be grown in immune-competent mice, determine whether the lines recapitulate human prostate cancer, and validate the signature in distinct populations of patients as a prognostic test. Herein, oncogene transduction of murine prostate epithelium induced prostate cellular transformation with invasive and metastatic properties. Oncogene-specific molecular signatures were defined that distinguished benign from malignant human prostate and predicted outcome of patients with human prostate cancer. The generation of oncogene-specific prostate cancer cell lines that recapitulate molecular signatures of human prostate cancer will be of value in understanding molecular mechanisms and identification of new therapies for this disease.

Mice, cell culture, chemicals, and reagents

The experimental procedures with transgenic mice were approved by the ethics committee of Thomas Jefferson University (Philadelphia, PA). Mice were in the FVB strain. Mouse prostate epithelial cell cultures were isolated from prostate glands of 12-week-old male mice and maintained as previously described (17) and analyzed after 25 passages with at least 3 lines of each genotype. Transduction of cells by the retroviral expression vector encoding either c-Myc, Ha-Ras, v-Src, NeuT, in the vector pBABE-IRES-GFP, was previously described (18, 19).

Cellular growth assays

Cells were seeded in 24-well-plates at a concentration of 1 × 104 cells per well, with each sample in triplicate for 7 days. Transformed cells were grown in Dulbecco's Modified Eagle's Medium with 10% FBS, whereas control prostate epithlial cell (PEC) cells were cultured in prostate epithelial primary culture medium. Cells were harvested, suspended in 100 μL PBS, and an equal volume of 0.4% Trypan blue added after 5 minutes. Cells were counted using a Countess Automated Cell Counter (C10227, Invitrogen).

Colony formation in soft agar

Cells (3 × 103/mL) were seeded into 0.3% soft agar Sigma) in a suspension dish (Nalgene Nunc International). Colonies were stained by 0.04% crystal violet acetate and counted under a vertical microscope after 2 weeks of incubation.

Tumor formation assay and detection of lung metastasis

A total of 1 × 106 cells suspended in a 100 μL volume were injected subcutaneously into 7 to 8 week-old FVB male mice. The cell suspension was mixed with a 20% by volume BD Matrigel (BD Biosciences), resulting in a final cell concentration of 107 cells/mL. Tumor growth was measured by Vernier calipers twice a week. Tumor samples were harvested after 30 days (except NeuT-induced tumors, which were harvested after 16 days). Thirty mice were examined. At autopsy the presence of lung metastasis was examined by visualization and palpation. All lungs were subsequently analyzed for metastasis by histologic sections of hematoxylin and eosin–stained specimens. The prostate cancer lung metastases were confirmed by a trained pathologist to show similar histology to the parental line, with migration of cells through the basement membrane. Ten sections were examined per lung at 100-μm separation. No formal analyses of bone metastasis or lymph node infiltration were conducted in these studies.

Statistical analysis

Comparisons between groups were analyzed by 2-sided t test. A difference of P < 0.05 was considered to be statistically significant. All analyses were done with SPSS 11.5 software. Data are expressed as mean ± SEM.

Oncogene-transformed prostate cell lines convey contact-independent growth

Primary prostate epithelial cell cultures were established from the ventral prostates of FVB mice. Cells were transduced with retroviral expression vectors encoding a single distinct oncogene (c-Myc, Ha-Ras (V-12), v-Src, and NeuT, an activating mutant of ErbB2). The cellular morphology of the prostate epithelial cells was altered over the 4-week period (Fig. 1A). Individual colonies of oncogene-transduced cells were selected and characterized. Cellular growth assays were conducted by cell counting (Fig. 1B). A substantial growth advantage was observed in each oncogene-transduced cell line compared with primary prostate epithelial cells.

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, and 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 and 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 (C) and markers of basal (CK5) versus luminal (CK8) prostate cancer (D). GDI is used as a protein-loading control. E, soft agar assays of oncogene-transduced PEC. Nontransformed 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 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, and 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 and 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 (C) and markers of basal (CK5) versus luminal (CK8) prostate cancer (D). GDI is used as a protein-loading control. E, soft agar assays of oncogene-transduced PEC. Nontransformed 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.

Close modal

Western blot analysis was conducted to examine the relative expression of each of the oncogenes used to transduce the PEC. The presence of oncogenic c-Myc, Ha-Ras, ErbB2, and v-Src was identified by Western blot analysis (Fig. 1C). The increase in abundance of each oncogene was specific to each cell line (Fig. 1C). The prostate tumor cell lines expressed prostate epithelial cell markers including cytokeratin 5 (CK5) and CK8, and were either AR-positive (NeuT and v-Src) or AR-negative (c-Myc and Ha-Ras; Fig. 1D). Oncogene transformation of fibroblasts or murine epithelial cells conveys contact-independent growth in solid agar. The oncogene-transduced PEC lines were examined for growth in soft agar. Colony size and number were characterized for each oncogene (Fig. 1E). Nontransformed PECs failed to grow in soft agar as previously described. Oncogene transduction increased the size and number of colonies (Fig. 1F).

As changes in genome copy number and/or genome structure contribute to tumor progression (20, 21), we conducted genome copy number analysis using array comparative genomic hybridization (CGH; see Materials and Methods). The cell lines, such as human prostate cancer, exhibited genomic gains and losses (Fig. 2A and B). Genome abnormalities were similar within each oncogene-transformed line. The genome copy number abnormality profiles showed distinct levels of genomic complexity. Greater complexity and number of abnormalities was identified in the NeuT lines, followed by the c-Myc lines. Ha-Ras lines showed low-level abnormalities (Fig. 2A). Copy aberration associated with the human prostate cancer derived from 205 primary and metastatic prostate cancer samples (Materials and Methods), were then compared with the genomic aberration of the oncogene-transformed cell lines (Fig. 2C and D). The frequency and copy aberrations associated with the human prostate cancer samples are shown on the right of each schematic. Significant fidelity of conservation between genomic alterations in the prostate cell lines and alterations found in human prostate cancer was observed. Analysis of the individual genes within the regions deleted or amplified for each cell line was conducted and tabulated (Supplementary Table S1). Approximately 2,400 genes are included within these regions. A comparison of gene expression and copy number showed the relative frequency of coincident changes (Fig. 2D).

Figure 2.

Copy number aberrations in the 4 oncogene cell lines assessed by array CGH. A, the percentage of the 4 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 4 cell lines are shown as a function of genomic position. C, oncogenes are identified with mRNA overexpression (red), DNA amplification (yellow), or both (purple) among the 4 oncogene cell lines, with corresponding amplification in the Memorial Sloan-Kettering Cancer Center (MSKCC) 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 4 oncogene cell lines, with corresponding copy loss in the MKSCC prostate cancer database (listed on right-hand side). PCa, prostate cancer.

Figure 2.

Copy number aberrations in the 4 oncogene cell lines assessed by array CGH. A, the percentage of the 4 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 4 cell lines are shown as a function of genomic position. C, oncogenes are identified with mRNA overexpression (red), DNA amplification (yellow), or both (purple) among the 4 oncogene cell lines, with corresponding amplification in the Memorial Sloan-Kettering Cancer Center (MSKCC) 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 4 oncogene cell lines, with corresponding copy loss in the MKSCC prostate cancer database (listed on right-hand side). PCa, prostate cancer.

Close modal

Lung metastases of prostate cancer cell lines

Tumor formation studies were conducted in FVB mice. Each of the prostate tumor lines grew subcutaneously in immune-competent mice. Growth was sustained for c-Myc-, Ha-Ras-, and v-Src–transformed PECs (Fig. 3A). The extirpated tumors were hemorrhagic (Fig. 3B) with histologic features of prostate adenocarcinoma (Fig. 3C and Supplementary Fig. S1). Immunostaining of tumors for Von Willebrand factor (VWF) confirmed angiogenesis and showed significantly greater VWF staining in tumors induced by Ha-Ras (Supplementary Fig. S2A). Lung metastases were characterized at autopsy by histopathologic assessment as described in the Materials and Methods (Fig. 4A). The number of lung metastases derived from subcutaneous injection of the primary PECs was increased in the Ha-Ras, v-Src, and c-Myc sublines (Fig. 4B).

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 Supplementary Fig. S1).

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 Supplementary Fig. S1).

Close modal
Figure 4.

Oncogene-transformed prostate epithelial cell tumors metastasize to lung. A, hematoxylin and eosin stain of murine lung after tumor implantation showing 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 4.

Oncogene-transformed prostate epithelial cell tumors metastasize to lung. A, hematoxylin and eosin stain of murine lung after tumor implantation showing 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.

Close modal

Oncogene-specific molecular signatures in prostate cancer cell lines

To further characterize the molecular genetic signaling pathways regulated by specific oncogenes in prostate epithelial cells, mRNA was prepared from the oncogene-transformed PEC cell lines. Microarray analysis identified a total of 2,635 of 22,115 genes that were significantly altered in expression (at least 2-fold change) in oncogene overexpressing cell lines when compared with nontransformed prostate epithelial cell control samples (Fig. 5A). The rows of the heatmap represent unique genes and are displayed by their pattern of up- and downregulation for all 4 of the oncogene-induced cell lines. For example, group 1 contains genes that share concordant differential expression patterns across all 4 cell lines, whereas group 15 contains genes whose differential expression is specific to the v-Src cell lines. A sizable number of up- and downregulated genes were shared among all 4 cell lines (group 1; 251 genes). Genes with up and downregulation specific to Ha-Ras were the most prevalent (group 14; 584 genes), followed by c-Myc–specific genes (group 8; 332 genes), NeuT-specific genes (group 12; 277 genes), and v-Src–specific genes (group 15; 215 genes).

Figure 5.

Differential gene expression patterns in transformed PEC cell lines. A heatmap (A) provides an overview of the differential expression patterns among 4 distinct oncogene-transformed cell lines. Heatmaps of genes that are differentially expressed in the 4 oncogene-transformed PEC lines and differentially expressed genes in high grade versus low grade (B) and advanced stage versus early-stage prostate cancer (C; ref. 4). Heatmaps of the left-hand side represent the prostate cancer high grade and advanced-stage signatures, whereas heatmaps on the right represent genes that are differentially expressed in at least 1 of 4 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 5.

Differential gene expression patterns in transformed PEC cell lines. A heatmap (A) provides an overview of the differential expression patterns among 4 distinct oncogene-transformed cell lines. Heatmaps of genes that are differentially expressed in the 4 oncogene-transformed PEC lines and differentially expressed genes in high grade versus low grade (B) and advanced stage versus early-stage prostate cancer (C; ref. 4). Heatmaps of the left-hand side represent the prostate cancer high grade and advanced-stage signatures, whereas heatmaps on the right represent genes that are differentially expressed in at least 1 of 4 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.

Close modal

The murine prostate oncogene expression signature in high grade and advanced-stage human prostate cancer

The prostate “oncogene expression signature” (group 1) was defined as genes that were significantly altered in expression level and that were uniquely altered in expression by a specific oncogene compared with primary prostate epithelial cells (Fig. 5A). The oncogene expression signature was compared with gene signatures obtained from other published databases to identify similarities to other well-studied disease phenotypes and cell lines. Gene ontology enrichment analysis for each of the cell lines identified enrichment for pathways involved in tumor progression. The top 40 gene ontology biologic process terms sorted on the basis of their overall significance in the 4 oncogene cell lines are shown in Supplementary Fig. S3 and show enrichment of pathways involved in cellular proliferation, DNA replication, and reduced expression of pathways promoting adhesion and epithelial cell differentiation. Comparisons were conducted against gene signatures representative of differential expression in advanced state versus early-stage prostate cancer, high grade versus low-grade prostate cancer, recurrent versus nonrecurrent prostate cancer (4). The gene signature heatmaps representing advanced stage/early stage, high grade/low grade, and recurrent/nonrecurrent prostate cancer phenotypes (4; Fig. 5B and C) are shown on the left and the heatmaps on the right represent genes that are differentially expressed in the prostate oncogene expression signature. The heatmaps are labeled with the percentage of genes within the “oncogene expression signature” that are differentially expressed. P values for the statistical significance of the similarity between the genes expressed in the prostate cancer cell lines and the gene signature of the disease phenotype are shown. P values are based on the hypergeometric distribution and represent the probability of these genes being differentially expressed in the disease phenotype if they were selected at random.

A “high grade” prostate cancer gene signature was previously determined from 61 primary prostate tumors (4). Figure 5B shows that 34 genes from the prostate oncogene expression signature were common to the “high grade” gene signature (P = 2.97 × 10−5). For each oncogene-induced prostate cancer cell line, the proportion of significant genes contributing to that cell line is shown. For example, the overlap between the prostate oncogene expression signature and high-grade disease includes a combination of genes that are significant genes in c-Myc (47%), NeuT (53%), Ha-Ras (71%), and v-Src (62%). Figure 5C depicts the 72 genes that were common between the prostate oncogene signature and the “advanced stage” gene signature (P = 4.13 × 10−8). These results indicate a significant degree of similarity between the prostate oncogene expression signature and high-grade disease (P = 2.97 × 10−5) and between the oncogene expression signature and the advanced-stage disease phenotype (P = 4.13 × 10−8).

No significant overlap was identified between the prostate oncogene expression signature and the recurrent/nonrecurrent disease signature identified by Lapointe and colleagues (4). When the prostate oncogene expression signature was compared with the Lapointe data, the Ha-Ras cell line captured the highest level of similarity with high-grade disease (71%), whereas the v-Src cell line showed the highest similarity with advanced-stage disease (67%).

c-Myc–specific gene expression signature in prostate cancer epithelial cells resembles the c-Myc signature in fibroblasts and mammary tumors

A comparison was made between gene expression pathways regulated in the c-Myc cell line with previous studies of gene expression in transgenic mice expressing the c-Myc oncogene targeted to the prostate under control of the probasin promoter (8) double transgenic mice expressing c-Myc and NKX3.1 (22) and curated gene signatures (23; Supplementary Fig. S4A). These studies showed a significant concordant correlation between c-Myc–repressed and c-Myc–induced genes in the c-Myc PEC line, with the gene expression profiles of the transgenic mice and several curated gene signatures (Supplementary Fig. S4B).

In previous studies, we identified gene expression signatures that were specific to the oncogene used to transform fibroblasts (3T3 cells), which were recapitulated in mammary tumors induced by c-Myc or Ha-Ras (24). The previously defined Ha-Ras- and c-Myc–induced molecular signature was compared with the gene expression signature induced by these oncogenes in the prostate cancer epithelial cells (oncogene superset in Fig. 5A). A comparison was conducted of the Ha-Ras prostate oncogene expression signature against Ha-Ras–transduced fibroblasts (Fig. 6A). Significant overlap was identified between the prostate oncogene expression signature and genes differentially expressed upon Ras oncogene transduction in the mouse fibroblasts (33 genes, hypergeometric P = 3.7 × 10−3).

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 Ha-Ras (A) and c-Myc (B) 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. WT, wild type 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 ref. 2.

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 Ha-Ras (A) and c-Myc (B) 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. WT, wild type 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 ref. 2.

Close modal

The heatmaps in Fig. 6B depicts the genes shared between c-Myc–transduced fibroblasts (Fig. 6B), c-Myc–induced mammary tumors (Fig. 6C; left-hand heatmaps), and the c-Myc–induced prostate oncogene expression signature (right-hand heatmaps). A significant overlap was identified between the c-Myconcogene expression signature (Superset) and the genes differentially regulated by c–Myc transduction in mouse fibroblasts (108 genes, hypergeometric P = 5.84 × 10−12) or mammary tumors (363 genes; hypergeometric P = 7.5916e−012). Within the prostate oncogene expression signature, c-Myc cell lines showed the largest proportion of similarity with both the Myc-transduced fibroblasts (92%; Fig. 6B) and the c-Myc–induced mammary tumors (85%; Fig. 6C).

Kaplan–Meier analysis was used to evaluate the difference in recurrence-free survival associated with high expression versus low expression of these genes. Genes that correspond to the 14 most upregulated genes of the c-Myc signature were used to assign the samples as high (upper 25th percentile) or low (lower 75th percentile). Genes that were highly expressed in the c-Myc signature had a significant association with poor outcome when used to interrogate the clinical dataset of Taylor and colleagues (P < 0.005; Fig. 6D). Comparison was made between the c-Myc signature and a recently described gene prognostic signature (25) by using these signatures to interrogate a clinical dataset (2). The comparison of the c-Myc signature identified herein, compared favorably as a discriminator of outcome (Supplementary Fig. S5).

Prostate oncogene induced gene expression and recurrence-free survival in human prostate cancer

To examine the relationship between genes expressed in the oncogene-transformed cell lines and survival rates from human prostate cancer, a previously published microarray dataset of human prostate tumor samples with known clinical recurrence-free survival time was used (7). Figure 7A provides a heatmap showing the expression profiles of genes in the human prostate cancer samples from the Glinsky's dataset, which was upregulated in the c-Myc (Fig. 5A, group 8) oncogene-transformed prostate cancer cell lines.

Figure 7.

The c-Myc–specific expression profile distinguishes tumor from normal tissue. A, hierarchical clustering conducted 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 and colleagues dataset. ROC curves for the classifier performance are shown for the Lapointe and colleagues dataset (C) and the Taylor and colleagues MSKCC dataset (D), with AUC values of 0.990 and 0.977, respectively.

Figure 7.

The c-Myc–specific expression profile distinguishes tumor from normal tissue. A, hierarchical clustering conducted 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 and colleagues dataset. ROC curves for the classifier performance are shown for the Lapointe and colleagues dataset (C) and the Taylor and colleagues MSKCC dataset (D), with AUC values of 0.990 and 0.977, respectively.

Close modal

Canonical analysis was conducted using the c-Myc–specific genes that were also found to be differentially expressed between tumor and normal samples in 2 human prostate datasets. The first canonical variable, resulting from this analysis, was used to discriminate tumor from normal samples in each human prostate dataset. A plot of tumor and normal samples along the first canonical variable axis revealed a pronounced separation between tumor and normal samples (Fig. 7B). Receiver operating characteristic (ROC) curves have been used previously to evaluate the diagnostic ability of PSA (26, 27) as well as its ability to identify metastatic disease (28). A ROC analysis of tumor and normal samples, separated by the first canonical variable, produced area under the curve (AUC) values of 0.977 and 0.990 for the Taylor and colleagues (Fig. 7C) and Lapointe and colleagues (Fig. 7D) datasets, respectively (2, 4).

The relative paucity of murine prostate cell lines that can be grown in immune-competent mice to form metastases has limited preclinical studies of prostate cancer therapies in vivo. The immune system plays an important role in the onset and progression of prostate cancer (12). The current studies were conducted to generate murine prostate cancer cell lines that could be studied in immune-competent mice. The prostate cancer cell lines generated herein resembled human prostate cancer histologically, and grew in immune-competent mice in a reproducible manner and formed lung metastasis. The gene expression changes and genomic rearrangements resembled human prostate cancer. By more faithfully recapitulating the tumor host environment, these lines provide a valuable advance for preclinical testing new therapies in vivo.

These studies identified patterns of gene expression within newly engineered prostate cancer cell lines that are observed in human prostate cancer and are associated with poor outcome (Supplementary Fig. S5C). The algorithm distinguished recurrence-free survival (prognostic signature) and with superior characteristics to several published signatures (23, 25) and Gleason scale. Prior studies have shown a paucity of genes with concordant expression in established prostate cancer cell lines and human prostate cancer (29). Herein, the generation of oncogene-specific–transformed prostate cancer cell lines allowed a comparison with the molecular signatures of human prostate cancer. Comparison was made between whole tumor mRNA in our studies and whole tumor mRNA in the clinical studies (not laser capture microdissected samples). The prostate oncogene expression signature correlated with a previously defined high-grade prostate cancer and with advanced stage in a second patient population. The c-Myc signature correlated with the signature induced by c-Myc when targeted to the prostate gland in transgenic mice (Supplementary Fig. S4A and S4B) and gene enrichment analysis identified significant overlap with several recent published datasets and transgenic models of prostate cancer (8, 23). The c-Myc signature was used to derive algorithms that distinguished normal from cancerous prostate tissue with approximately 99% predictive value (diagnostic signature).

Herein, transduction of primary prostate epithelial cells of FVB mice by a single oncogene was sufficient for transformation. Oncogene genomic analysis identified loss of tumor suppressor regions and amplification of additional oncogenic regions that resemble genomic changes in human prostate cancer (Fig. 2). The cell lines reflect both AR-response and AR-unresponse prostate cancers (Supplementary Fig. S6) although each line expresses plasma Kallikrein 1B (PK1B; Supplementary Fig. S2B). Kallikrein gene expression is regulated by basal and androgen-independent factors, thus PK1B in the tumors may not necessarily be correlated directly with AR expression (Fig. 1D). The tumors expressed CK8, whereas the c-Myc and v-Src lines expressed CK5 by Western blot analysis (Fig. 1D) and by immunofluorescence in the primary cell lines (Supplementary Fig. S7A) as well as in the lung metastasis (Supplementary Fig. S7B). CK5 is considered a marker of basal cell type origin of prostate cancer and is more common in metastatic prostate cancer and after androgen deprivation (30). In previous studies, prostate epithelial cells have been transformed with DNA tumor viruses and murine prostate cancer cell lines have been derived from oncomice (31). Human telomerase reverse transcriptase (hTERT) together with SV40T/t immortalizes human PEC and together with either Ha-Ras or c-Myc results in cells that grow in tissue culture. However, unlike the lines developed herein, these PEC lines were unable to grow tumors after subcutaneous implantation or form metastasis in vivo (32).

The transition of normal somatic epithelial cells into an immortalized metastatic cell requires dysregulation of distinct cellular processes including cell survival motility, cell proliferation, altered metabolism, and angiogenesis. The transformation of human and murine cells has distinct requirements (33, 34). Human fibroblasts can be transformed by 2 oncogenes (35). Primary human cells can be immortalized by the introduction of the catalytic subunit of hTERT and SV40 large-T to inactivate the pRb and p53 pathways (36). Immortal human cells are converted into transformed cells with tumorigenic growth by the further induction of an oncogenic Ha-Ras allele and the SV40 early region oncoprotein small t antigen (35, 37, 38). The requirements for transformation are cell type specific (34). Human mammary epithelial cells immortalized by SV40 and hTERT undergo tumorigenic conversion with oncogenic Ras (36) or by the use of a mutant p53, a CDK4 mutant and cyclin D1 overexpression (39). SV40 Large T antigen contributes to transformation through inactivation of p53 and pRb, as attested by siRNA analysis. Cautionary caveats to the use of these cell lines include the potential importance of the local prostate microenvironment in progression of prostate cancer, the capacity of the FVB murine strain background to enhance tumor progression (40) and the tendency of human prostate cancer metastasis to target bone rather than lung (41). The mechanisms governing the spontaneous regression of the NeuT lines remains speculative at this time.

The current studies defined oncogene-specific signatures and then mapped these signatures to human prostate cancer. The oncogenes were chosen because of evidence that these oncogenes and/or tumor pathways are activated in human prostate cancer. The c-Myc signature was conserved in prostate, breast, and fibroblasts. The c-Myc gene copy number is increased in up to 30% of cases at the preneoplastic stage in patients with PIN (42, 43). The v-Src, ErbB2, Ha-Ras oncogenes have also been implicated in human prostate cancer, and a truncated version of c-kit identified in primary prostate tumors correlates with activation of the Src pathway (44). Although the role of ErbB2 in human prostate cancer is controversial, a subset of patients with prostate cancer expresses ErbB2 and its pathway, and activated ErbB2 is sufficient to induce prostate cancer in transgenic mice (45, 46). In vivo, the activating Ras mutation (Ha-Ras V12) induced low-grade PIN in mice when directed to prostate epithelial cells under the control of the minimal probasin promoter (47).

The current studies identified oncogene-specific signatures in murine prostate epithelium that were recapitulated within human prostate cancer samples. Oncogene-specific signatures have been proposed to exist within human prostate cancer (10). In human breast cancer, the identification of tumor subsets with patterns of transcriptional and biologic activity has allowed a deeper understanding of the disease. It has been anticipated that the identification of these patterns of gene function in otherwise heterogeneous biologic material may enable the identification of targets for therapies, and to thereby improve management of prostate cancer. The development herein of oncogene-specific prostate cancer cell lines may be of value in developing targeted therapies for prostate cancer.

R.G. Pestell has ownership interest (including patents) in ProstaGene, LLC and U.S. Provisional Patent Application Serial No. 10/532,009. PCT Patent Application No. PCT/US2003/033438: Caged Ligands and Uses Thereof. International Publication No. WO 2004/037983 A2. May 6, 2004 issued by the International Bureau of the World Intellectual Property Organization. No potential conflicts of interest were disclosed by the other authors. The Department specifically disclaims responsibility for any analysis, interpretations, or conclusions.

Conception and design: X. Ju, R.G. Pestell

Development of methodology: X. Ju, H. Meng, P. Fortina

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Ju, H. Meng, P.A. McCue, P. Fortina

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Ju, A. Ertel, M. Casimiro, P.A. McCue, P. Fortina, M.P. Lisanti

Writing, review, and/or revision of the manuscript: X. Ju, A. Ertel, P.A. McCue, P. Fortina, M.P. Lisanti, R.G. Pestell

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Ju, M. Casimiro, Z. Yu, H. Meng, R. Walters

Study supervision: X. Ju, M. Casimiro, M.P. Lisanti, R.G. Pestell

The authors thank David Strahan for the preparation of this article.

This work was supported in part by R01CA70896, R01CA75503, R01CA107382, R01CA86072 (to R.G. Pestell) and R01CA120876 (M.P. Lisanti). The Kimmel Cancer Center was supported by the NIH Cancer Center Core grant P30CA56036 (R.G. Pestell). This project is funded in part from the Dr. Ralph and Marian C. Falk Medical Research Trust and a grant from Pennsylvania Department of Health (R.G. Pestell).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Scher
HI
,
Sawyers
CL
. 
Biology of progressive, castration-resistant prostate cancer: directed therapies targeting the androgen-receptor signaling axis
.
J Clin Oncol
2005
;
23
:
8253
61
.
2.
Taylor
BS
,
Schultz
N
,
Hieronymus
H
,
Gopalan
A
,
Xiao
Y
,
Carver
BS
, et al
Integrative genomic profiling of human prostate cancer
.
Cancer Cell
2010
;
18
:
11
22
.
3.
Singh
D
,
Febbo
PG
,
Ross
K
,
Jackson
DG
,
Manola
J
,
Ladd
C
, et al
Gene expression correlates of clinical prostate cancer behavior
.
Cancer Cell
2002
;
1
:
203
9
.
4.
Lapointe
J
,
Li
C
,
Higgins
JP
,
van de Rijn
M
,
Bair
E
,
Montgomery
K
, et al
Gene expression profiling identifies clinically relevant subtypes of prostate cancer
.
Proc Natl Acad Sci U S A
2004
;
101
:
811
6
.
5.
Dhanasekaran
SM
,
Barrette
TR
,
Ghosh
D
,
Shah
R
,
Varambally
S
,
Kurachi
K
, et al
Delineation of prognostic biomarkers in prostate cancer
.
Nature
2001
;
412
:
822
6
.
6.
LaTulippe
E
,
Satagopan
J
,
Smith
A
,
Scher
H
,
Scardino
P
,
Reuter
V
, et al
Comprehensive gene expression analysis of prostate cancer reveals distinct transcriptional programs associated with metastatic disease
.
Cancer Res
2002
;
62
:
4499
506
.
7.
Glinsky
GV
,
Glinskii
AB
,
Stephenson
AJ
,
Hoffman
RM
,
Gerald
WL
. 
Gene expression profiling predicts clinical outcome of prostate cancer
.
J Clin Invest
2004
;
113
:
913
23
.
8.
Ellwood-Yen
K
,
Graeber
TG
,
Wongvipat
J
,
Iruela-Arispe
ML
,
Zhang
J
,
Matusik
R
, et al
Myc-driven murine prostate cancer shares molecular features with human prostate tumors
.
Cancer Cell
2003
;
4
:
223
38
.
9.
Weber
MJ
,
Gioeli
D
. 
Ras signaling in prostate cancer progression
.
J Cell Biochem
2004
;
91
:
13
25
.
10.
Creighton
CJ
. 
Multiple oncogenic pathway signatures show coordinate expression patterns in human prostate tumors
.
PLoS ONE
2008
;
3
:
e1816
.
11.
Paronetto
MP
,
Farini
D
,
Sammarco
I
,
Maturo
G
,
Vespasiani
G
,
Geremia
R
, et al
Expression of a truncated form of the c-Kit tyrosine kinase receptor and activation of Src kinase in human prostatic cancer
.
Am J Pathol
2004
;
164
:
1243
51
.
12.
De Marzo
AM
,
Platz
EA
,
Sutcliffe
S
,
Xu
J
,
Gronberg
H
,
Drake
CG
, et al
Inflammation in prostate carcinogenesis. Nature Rev
Cancer
2007
;
7
:
256
69
.
13.
Di Lorenzo
G
,
Buonerba
C
,
Kantoff
PW
. 
Immunotherapy for the treatment of prostate cancer
.
Nat Rev Clin Oncol
2011
;
8
:
551
61
.
14.
Bull
JH
,
Ellison
G
,
Patel
A
,
Muir
G
,
Walker
M
,
Underwood
M
, et al
Identification of potential diagnostic markers of prostate cancer and prostatic intraepithelial neoplasia using cDNA microarray
.
Br J Cancer
2001
;
84
:
1512
9
.
15.
Thompson
TC
,
Southgate
J
,
Kitchener
G
,
Land
H
. 
Multistage carcinogenesis induced by ras and myc oncogenes in a reconstituted organ
.
Cell
1989
;
56
:
917
30
.
16.
Hensley
PJ
,
Kyprianou
N
. 
Modeling prostate cancer in mice: limitations and opportunities
.
J Androl
2011
;
33
:
133
44
.
17.
Ilio
KY
,
Nemeth
JA
,
Lang
S
,
Lee
C
. 
The primary culture of rat prostate basal cells
.
J Androl
1998
;
19
:
718
24
.
18.
Li
Z
,
Wang
C
,
Jiao
X
,
Lu
Y
,
Fu
M
,
Quong
AA
, et al
Cyclin D1 regulates cellular migration through the inhibition of thrombospondin 1 and ROCK signaling
.
Mol Cell Biol
2006
;
26
:
4240
56
.
19.
Liu
M
,
Casimiro
MC
,
Wang
C
,
Shirley
LA
,
Jiao
X
,
Katiyar
S
, et al
p21CIP1 attenuates Ras- and c-Myc-dependent breast tumor epithelial mesenchymal transition and cancer stem cell-like gene expression in vivo
.
Proc Natl Acad Sci U S A
2009
;
106
:
19035
9
.
20.
Holcomb
IN
,
Young
JM
,
Coleman
IM
,
Salari
K
,
Grove
DI
,
Hsu
L
, et al
Comparative analyses of chromosome alterations in soft-tissue metastases within and across patients with castration-resistant prostate cancer
.
Cancer Res
2009
;
69
:
7793
802
.
21.
Robbins
CM
,
Tembe
WA
,
Baker
A
,
Sinari
S
,
Moses
TY
,
Beckstrom-Sternberg
S
, et al
Copy number and targeted mutational analysis reveals novel somatic events in metastatic prostate tumors
.
Genome Res
2011
;
21
:
47
55
.
22.
Anderson
PD
,
McKissic
SA
,
Logan
M
,
Roh
M
,
Franco
OE
,
Wang
J
, et al
Nkx3.1 and Myc crossregulate shared target genes in mouse and human prostate tumorigenesis
.
J Clin Invest
2012
;
122
:
1907
19
.
23.
Markert
EK
,
Mizuno
H
,
Vazquez
A
,
Levine
AJ
. 
Molecular classification of prostate cancer using curated expression signatures
.
Proc Natl Acad Sci U S A
2011
;
108
:
21276
81
.
24.
Huang
ES
,
Black
EP
,
Dressman
H
,
West
M
,
Nevins
JR
. 
Gene expression phenotypes of oncogenic signaling pathways
.
Cell Cycle
2003
;
2
:
415
17
.
25.
Ding
Z
,
Wu
CJ
,
Chu
GC
,
Xiao
Y
,
Ho
D
,
Zhang
J
, et al
SMAD4-dependent barrier constrains prostate cancer growth and metastatic progression
.
Nature
2011
;
470
:
269
73
.
26.
Catalona
WJ
,
Hudson
MA
,
Scardino
PT
,
Richie
JP
,
Ahmann
FR
,
Flanigan
RC
, et al
Selection of optimal prostate specific antigen cutoffs for early detection of prostate cancer: receiver operating characteristic curves
.
J Urol
1994
;
152
:
2037
42
.
27.
Thompson
IM
,
Chi
C
,
Ankerst
DP
,
Goodman
PJ
,
Tangen
CM
,
Lippman
SM
, et al
Effect of finasteride on the sensitivity of PSA for detecting prostate cancer
.
J Natl Cancer Inst
2006
;
98
:
1128
33
.
28.
Akimoto
S
,
Furuya
Y
,
Akakura
K
,
Shimazaki
J
,
Ito
H
. 
Relationship between prostate-specific antigen, clinical stage, and degree of bone metastasis in patients with prostate cancer: comparison with prostatic acid phosphatase and alkaline phosphatase
.
Int J Urol
1997
;
4
:
572
5
.
29.
Welsh
JB
,
Sapinoso
LM
,
Su
AI
,
Kern
SG
,
Wang-Rodriguez
J
,
Moskaluk
CA
, et al
Analysis of gene expression identifies candidate markers and pharmacological targets in prostate cancer
.
Cancer Res
2001
;
61
:
5974
8
.
30.
Schalken
JA
,
van Leenders
G
. 
Cellular and molecular biology of the prostate: stem cell biology
.
Urology
2003
;
62
:
11
20
.
31.
Roy-Burman
P
,
Wu
H
,
Powell
WC
,
Hagenkord
J
,
Cohen
MB
. 
Genetically defined mouse models that mimic natural aspects of human prostate cancer development
.
Endocrine-related Cancer
2004
;
11
:
225
54
.
32.
Berger
R
,
Febbo
PG
,
Majumder
PK
,
Zhao
JJ
,
Mukherjee
S
,
Signoretti
S
, et al
Androgen-induced differentiation and tumorigenicity of human prostate epithelial cells
.
Cancer Res
2004
;
64
:
8867
75
.
33.
Hahn
WC
,
Weinberg
RA
. 
Rules for making human tumor cells
.
N Engl J Med
2002
;
347
:
1593
603
.
34.
Rangarajan
A
,
Hong
SJ
,
Gifford
A
,
Weinberg
RA
. 
Species- and cell type-specific requirements for cellular transformation
.
Cancer Cell
2004
;
6
:
171
83
.
35.
Ruley
HE
. 
Adenovirus early region 1A enables viral and cellular transforming genes to transform primary cells in culture
.
Nature
1983
;
304
:
602
6
.
36.
Elenbaas
B
,
Spirio
L
,
Koerner
F
,
Fleming
MD
,
Zimonjic
DB
,
Donaher
JL
, et al
Human breast cancer cells generated by oncogenic transformation of primary mammary epithelial cells
.
Genes Dev
2001
;
15
:
50
65
.
37.
Hahn
WC
,
Counter
CM
,
Lundberg
AS
,
Beijersbergen
RL
,
Brooks
MW
,
Weinberg
RA
. 
Creation of human tumour cells with defined genetic elements
.
Nature
1999
;
400
:
464
8
.
38.
Land
H
,
Parada
LF
,
Weinberg
RA
. 
Tumorigenic conversion of primary embryo fibroblasts requires at least two cooperating oncogenes
.
Nature
1983
;
304
:
596
602
.
39.
Hahn
WC
,
Dessain
SK
,
Brooks
MW
,
King
JE
,
Elenbaas
B
,
Sabatini
DM
, et al
Enumeration of the simian virus 40 early region elements necessary for human cell transformation
.
Mol Cell Biol
2002
;
22
:
2111
23
.
40.
Bianchi-Frias
D
,
Pritchard
C
,
Mecham
BH
,
Coleman
IM
,
Nelson
PS
. 
Genetic background influences murine prostate gene expression: implications for cancer phenotypes
.
Genome Biol
2007
;
8
:
R117
.
41.
Fabozzi
SJ
,
Schellhammer
PF
,
el-Mahdi
AM
. 
Pulmonary metastases from prostate cancer
.
Cancer
1995
;
75
:
2706
9
.
42.
Jenkins
RB
,
Qian
J
,
Lieber
MM
,
Bostwick
DG
. 
Detection of c-Myc oncogene amplification and chromosomal anomloies in metastatic prostatic carcinoma by fluorescence in situ hybridization
.
Cancer Res
1997
;
57
:
524
31
.
43.
Sato
K
,
Qian
J
,
Slezak
JM
,
Lieber
MM
,
Bostwick
DG
,
Bergstralh
EJ
, et al
Clinical significance of alterations of chromosome 8 in high-grade, advanced, nonmetastatic prostate carcinoma
.
J Natl Cancer Inst
1999
;
91
:
1574
80
.
44.
Mori
S
,
Chang
JT
,
Andrechek
ER
,
Potti
A
,
Nevins
JR
. 
Utilization of genomic signatures to identify phenotype-specific drugs
.
PLoS ONE
2009
;
4
:
e6772
.
45.
Casimiro
M
,
Rodriguez
O
,
Pootrakul
L
,
Aventian
M
,
Lushina
N
,
Cromelin
C
, et al
ErbB-2 induces the cyclin D1 gene in prostate epithelial cells in vitro and in vivo
.
Cancer Res
2007
;
67
:
4364
72
.
46.
Gioeli
D
,
Kraus
S
,
Weber
MJ
. 
Signal transduction by the Ras-MAP kinase pathway in prostate cancer progression
. In:
Pestell
RG
,
Nevalainen
MT
,
editors
. 
Prostate cancer: signaling networks, genetics, and new treatment strategies
.
Totowa, NJ
:
Humana Press
; 
2008
.
p
.
223
56
.
47.
Scherl
A
,
Li
JF
,
Cardiff
RD
,
Schreiber-Agus
N
. 
Prostatic intraepithelial neoplasia and intestinal metaplasia in prostates of probasin-RAS transgenic mice
.
Prostate
2004
;
59
:
448
59
.