v-Src Oncogene Induces Trop2 Proteolytic Activation via Cyclin D1

The cyclin D1 gene encodes the regulatory subunit of the holoenzyme that phosphorylates and inactivates the retinoblastoma (pRb) protein, promoting G₁–S phase cell-cycle entry. Cyclin D1 enhances prostate cellular proliferation in vivo and endogenous cyclin D1 maintains prostate cancer cellular proliferation in vitro (1). In human prostate cancer, cyclin D1 abundance is increased in many patients and a cyclin D1–regulated gene expression signature predicts poor outcome (1). The abundance of cyclin D1 is induced by activating mutations of Src via transcriptional mechanisms (2). The abundance of cyclin D1 is rate limiting in growth of a variety of tumors in vivo, including ErbB2-induced breast cancer (3, 4) and a 50% reduction in cyclin D1 abundance in cyclin D1 heterozygote mice is sufficient to reduce the onset and progression of gastrointestinal tumorigenesis induced by the Apc/Min mutation (5).

The Src family of kinases (SFK) includes nonreceptor tyrosine kinases with nine homologous members that encode an SH4 domain governing cytoplasmic membrane association. Src kinase conveys a functional role in both initiation and progression of murine prostate cancer (6, 7). Coexpression of wild-type Src and the androgen receptor (AR) enhances the formation of murine prostate adenocarcinoma (7). Genomic analysis of human prostate carcinoma demonstrated that mutations of activating tyrosine kinases are rare. However, phosphotyrosine peptide analysis with quantitative mass spectrometry demonstrated increased Src kinase activity in most patients with prostate cancer, and >80% of patients are candidates for Src-inhibitor treatment (8). In contrast, very few patients were positive for activated states of receptors for MET, ErbB2, or EGFR, despite their detection in prostate cancer cell lines (8). Given the frequency of Src kinase activation in human prostate cancer, an enhanced understanding of Src-mediated transformation of prostate epithelium is fundamental to improving prostate cancer patient treatments.

A variety of independent analyses have provided supporting evidence for a role of stem cells in the onset and progression of tumorigenesis (9, 10). The tumor-associated calcium signal transducer 2 (Trop2) identifies a subpopulation of prostate cells with stem cell characteristics in both murine and human prostate (11). Trop2 is also highly expressed in the proximal region of the prostate. Trop2 has oncogenic activity (12) demonstrated by chimeric cyclin D1–Trop2 fusions in many cancer types, and silencing the fusion protein inhibits tumor growth (13). The related Trop2 family member, EpCAM, is considered a therapeutic target through regulation of cell–cell adhesion (14–16). Trop2 associates with the α5 integrin subunit, and thereby displaces focal adhesion kinase from focal contacts to promote an invasive phenotype. Consistent with this finding, Trop2 is upregulated in human prostate cancer with extracapsular extension (stages pT3/pT4) as compared with organ-confined (stage pT2) prostate cancer (16). Intramembrane proteolysis of Trop2 occurs by the TNFα converting enzyme (TACE), followed by γ-secretase. Two cleavage products are generated: the extracellular (ECD) and the intracellular domains (ICD). The Trop2 ICD accumulates in the nucleus, colocalizing with β-catenin to promote prostatic intraepithelial neoplasia (PIN) and self-renewal (17). The molecular mechanisms governing the expression and activation of Trop2 are poorly understood. The current studies were undertaken to investigate the mechanisms governing Trop2 activity in prostate cancer and given the importance of Src kinase, to determine the potential role for Src in Trop2 activity.

LNCaP cell line was obtained from ATCC, the v-Src-PEC and the NeuT-PEC cell lines were previously described (18). Original cells were expanded and stored in liquid nitrogen at an early passage. During the experiments, the morphology of all cell lines was checked routinely under phase contrast microscope. For the LNCaP cell line, proliferation and AR abundance in response to DHT stimulation were tested by MTT assay and Western blot analysis. For v-Src-PEC and NeuT-PEC cell lines, the proliferation in response to Src kinase inhibitor or NeuT inhibitor was tested. v-Src or NeuT expression in these cells was checked by Western blot analysis for verification. All of the newly revived cells were treated with BM cyclins (Roche) and mycoplasma contamination was determined with Hoechst 33258 staining under high magnification fluorescent microscope routinely. DNA transfection and luciferase assays were performed as previously described (1, 18). The CBF-Luc and −3,400 cyclin D1-Luc reporter plasmids were previously described (19, 20). The Src kinase inhibitor PP1 (4-amino-5-(4-methylphenyl)-7-(t-butel)pyrazolo-d-3-4-pyrimidine; Calbiochem) and dasatinib (BMS-354825, Selleckchem), the CDK inhibitor abemaciclib (MedChem Express), palbociclib (Sigma-Aldrich), ribociclib (Selleckchem), and the EGFR inhibitor canertinib (Selleckchem) were used at the indicated doses.

Experimental procedures with mice were approved by the ethics committee of Thomas Jefferson University. Mouse ventral prostates were fixed in 4% paraformaldehyde, and then used for sectioning and hematoxylin and eosin (H&E) staining. Antibodies used for Western blot analysis and immunohistochemical staining in this study were as follows: anti-cyclin D1, anti-vinculin, anti-presenilin1 (PS1), anti-presenilin2 (PS2), anti-TACE (Santa Cruz Biotechnology), and anti-Numb (Cell Signaling Technology), anti-Notch 1 (Millipore 07-1232), anti-p-Src (Upstate 07-020, Tyr 416), anti-Src (Oncogene OP07). Anti-Trop2- ICD antibody was from Professor Owen Witte (University of California, Los Angeles, Los Angeles, CA).

FACS analysis for prostate cancer stem cells was based on prior publications (1, 19, 21–23). Before labeling, the cells were blocked with normal mouse IgG in 1:100 dilution for 30 minutes and then incubated with fluorescein isothiocyanate (FITC) or phycoerythrin (PE)-labeled rat anti-mouse Sca-1 (clone E13-161.7; Pharmingen; 1/100–1/200), PE-labeled rat anti-mouse CD133 (1:10; clone MB9-3G8; Miltenyi Biotec), PE/Cy5-labeled rat anti-human/mouse CD44 (1:200; clone IM7; BioLegend), PE/Cy5-labeled rat anti-human/mouse CD49f (1/10), and/or FITC-labeled goat anti-mouse Trop2 (FAB1122F; R&D) for 1 hour. All experiments were conducted at 4°C. Cell sorting was performed on a FACSCalibur Cell Sorter (BD Biosciences). The data were analyzed with FlowJo software (TreeStar, Inc).

v-Src-PEC cells were plated at a density of 10,000 cells/mL on ultra-low attachment Corning cell culture plates and grown in DMEM/F12 with B27, 20 ng/mL EGF, 20 ng/mL FGF, and 4 μg/mL heparin. Prostatospheres were collected by gentle centrifugation (800 rpm) after 7 to 10 days and counted under the microscope using a 96-well plate (1, 24).

The transfection of siRNA to cyclin D1 and control siRNA into the v-Src cell line, and the infection of cyclin D1 shRNA into LNCaP cells, were performed as previously described (1). pTRIPZ tet-inducible shRNA Vector, which use TurboRFP as shRNA expression reporter, was from Qiagen Biotechnology.

Cells were grown in 4-well chamber slides and were fixed with 4% paraformaldehyde in PBS for 20 minutes at room temperature. The slides were rinsed with PBS and permeated with 0.05% NP-40 in PBS. The primary antibodies were rabbit polyclonal anti-Trop2-ICD (1/200) and mouse anti-β-catenin (mouse, Santa Cruz Biotechnology; SC-7963, 1/100). The secondary antibodies used were rhodamine red X–conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories; 1/100) and Alexa Fluor 633–conjugated F (ab′) 2 fragment of goat anti-mouse immunoglobulin G (IgG; Molecular Probes; 1/250). Fluorescence imaging was acquired with a 40× objective of a Zeiss LSM510/META laser confocal microscope. ImageJ was used to quantify fluorescence intensity of the whole cell and nucleus. Thirty cells were measured for each sample and the results were shown as relative intensity per μm².

Transcript expression profiling was previously performed on the v-Src-PEC cell lines along with parental prostate epithelial cells (PEC) using Affymetrix MoGene 1.0ST microarrays and the microarray data were deposited to GEO (GSE 37428; ref. 18). Genes differentially expressed in the v-Src-PECs relative to parental PECs were compared with transcript profiles of CD133⁺ prostate cancer stem cells from a previously published study (25). The analysis is described in detail in the Supplementary Methods based on ref. 26.

The Database for Annotation, Visualization and Integrated Discovery (DAVID; ref. 27) functional annotation tool was used to analyze the genes identified in common between CD133⁺ and v-Src-PEC differentially expressed gene lists for enriched Gene Ontology biological process (GO-BP) terms (28). GO-BP terms were reported at a 10% FDR cutoff and ranked on the basis of gene count for visualization in a bar chart.

The tissue microarray (TMA) was constructed at Thomas Jefferson University Hospital. All patients had undergone radical prostatectomies. Ethical approval was obtained from the Thomas Jefferson University Institutional Ethical Review Board. A detailed description is provided in the Supplementary Methods section.

IHC staining was conducted using the cyclin D1 antibody (Thermo Fisher Scientific, RB-010-P, 1:2000), rabbit Trop-2 ICD antibody (gift from Dr. Owen Witte, dilution 1:400), Numb antibody (Abcam, ab14140, 1:250), and presenilin 2 antibody (Santa Cruz Biotechnology, sc-1456, 1:150) by a DAKO Autostainer Plus equipment with an enzyme labeled biotin–streptavidin system. The slides were scanned on the BLISS system (Bacus Laboratory) and quantified on the basis of the staining intensity and the proportion of cells stained. All comparisons of staining intensities were made at ×200 magnification.

Kaplan–Meier analysis was used to evaluate the difference in recurrence-free survival associated with high expression versus low expression of Trop2 ICD, cyclin D1, and Numb proteins, respectively, in the 126 samples that had both a clinical record and IHC staining. The Cox regression fitting proportional hazards models to censored survival data was used to evaluate the association of all three markers to the risk of recurrence. Stratification was performed recursively. On the basis of the risk score, patients were assigned to high-, medium-, low-risk groups, and the difference in recurrence-free survival was evaluated among the three groups.

Comparisons between groups were analyzed by the two-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.

To determine the functional significance of Src kinase activity in the maintenance of cellular proliferation, v-Src PEC–stable prostate cancer cell lines were analyzed. The addition of the Src inhibitor PP1 reduced cellular proliferation by 20% to 50% in a dose-dependent manner (Fig. 1A). The Src inhibitor dasatinib similarly reduced cellular proliferation by 10% (P < 0.05) to 40% (P < 0.001) in a dose-dependent manner (Fig. 1B). Phospho-Src was downregulated upon treatment with dasatinib and PP1 (Fig. 1C). Cyclin D1 promotes DNA synthesis of the murine prostate and human prostate cancer cell lines (1). The relative abundance of cyclin D1 was reduced 40% by 25 to 100 nmol/L dasatinib (Fig. 1C) associated with a reduction in cellular proliferation. PP1 reduced cyclin D1 abundance by 50% (Fig. 1C). In recent studies, Trop2, which is known to be expressed in a subpopulation of prostate basal cells with stem cell characteristics, was shown to correlate with poor prognosis in prostate cancer (11, 16, 29). Trop2 is activated by regulated intra-membrane proteolysis (RIP), a mechanism involved in processing and activation of other transmembrane proteins, including N-cadherin and E-cadherin (30). Nuclear ICD of Trop2 is found in human prostate cancer, but not in the adjacent benign tissue (17). To determine whether Src kinase activity regulates the relative abundance of Trop2 ICD, Western blot analysis was conducted with an ICD-specific antibody. The relative abundance of Trop2 ICD was reduced 50% in the presence of the Src kinase inhibitors, dasatinib, and PP1 (Fig. 1C). To examine the kinetics with which Src kinase inhibition by dasatinib reduced the abundance of the Trop2 ICD, a time course analysis was conducted. Cyclin D1 levels were reduced >50% by 24 hours, and >90% by 48 hours. The reduction in Trop2 ICD abundance was reduced 50% at 48 hours (Supplementary Fig. S1).

To examine the molecular signaling pathways activated upon v-Src transformation of prostate epithelial cells, gene expression profiling was conducted to compare v-Src–transformed PECs and parental PECs. Experiments were conducted on three separate PEC-Src lines (Fig. 2A). In previous studies, CD133⁺ cells were considered enriched for prostate cancer stem cells (31, 32). A correlative analysis was conducted examining the genes that were differentially expressed between the CD133⁺ versus CD133⁻ prostate cancer cells (25). We therefore compared the genes enriched by v-Src in PECs with the genes expressed in CD133⁺ prostate cells. The relative abundance of genes in the CD133⁺ prostate cancer stem cell signature is shown in Fig. 2A (left panel). The genes differentially expressed within v-Src prostate cancer cell lines, relative to nontransformed parental prostate epithelial cells, are shown as the differentially expressed genes (right panel; ref. 18). A substantial overlap was seen between the genes enriched in CD133⁺ prostate cells when compared with genes regulated by v-Src. Genes induced within CD133⁺ cells were also induced by v-Src, and genes repressed by CD133⁺ were repressed by v-Src (Fig. 2A). The P value for the degree of similarity of the CD133⁺ stem cell signature with v-Src–regulated genes was significant (P = 2.148 × 10⁻¹⁰; Fig. 2A).

Seventy-seven genes regulated by v-Src (77/466, ∼17%) were regulated by CD133⁺ in a common manner (P = 5.52 × 10⁻¹⁴; Fig. 2B). These studies suggest an enrichment of Src-regulated genes among the CD133 enriched genes. GO terms were used to identify gene pathways regulated upon v-Src transformation of PECs. Additional pathways enriched upon v-Src transformation of PECs, included “cell cycle, DNA damage repair, DNA metabolic process and chromosomal organization” (Fig. 2C). To determine whether CD133⁺ cells were enriched for Src kinase activity, the oncogene-transformed PECs were FACS sorted for CD133 and then characterized for Src activity using the antibody directed to activated Src (Src^p Tyr 416; Fig. 2D). CD133⁺ cells were enriched 3-fold for Src^p Tyr 416 compared with CD133⁻ cells (Fig. 2E).

Prostatosphere formation assays, a surrogate measure of prostate cancer stem cells, were performed on the v-Src-PEC cell line. Approximately 16 of 1,000 cells formed prostatospheres in v-Src-PECs, whereas in parental PECs, only 1 of 1,000 formed prostatospheres (Fig. 3A). The v-Src-PEC line–derived prostatospheres were consistently larger than the nontransformed, and only the v-Src-PEC lines gave rise to secondary prostatospheres (Fig. 3A). Treatment with the Src kinase inhibitors dasatinib or PP1 reduced the number (Fig. 3B), but not the size of the prostatospheres.

In view of the finding that v-Src transformation induced expression of genes associated with CD133⁺, a prostate stem cell marker, we examined further the association of v-Src transformation with the expression of prostatic cancer stem cell epitopes. FACS was conducted for the relative proportion of other epitope markers of prostate cancer stem cells. v-Src transformation of prostate epithelium was associated with an approximately 6-fold increase of Sca1^high cells (Fig. 3C). Similarly, the proportion of CD44⁺/CD133⁺ cells was increased 6-fold (Fig. 3D). The proportion of CD49f⁺ Sca1⁺ Trop2⁺ cells was increased 2.2-fold in v-Src–transformed PECs compared with parental PECs (Fig. 3F).

Although the total Trop2 ICD abundance was reduced by Src kinase inhibitor, the nuclear pool of Trop2-ICD is considered to be the biologically active moiety, therefore we next sought to determine whether Src kinase governs the abundance of nuclear Trop2 ICD. Comparison was made between parental and v-Src–transformed PECs by immunostaining for Trop2-ICD (red), β-catenin (green), and nucleus (DAPI, blue; Fig. 4A). The relative abundance of nuclear Trop2 ICD was enhanced 4-fold upon v-Src transformation (Fig. 4B). The addition of the Src inhibitor, either 100 nmol/L dasatinib or 10 μmol/L PP1, reduced nuclear Trop2 ICD abundance by approximately 40% (Fig. 4C and D). In contrast, β-catenin immunostaining was not significantly reduced by dasatinib.

The v-Src-PEC cell lines were derived from FVB murine prostate epithelium and can therefore be reintroduced into immunocompetent FVB mice (18). In view of the importance of the immune system in the onset and progression of prostate cancer, we examined the abundance of the Trop2 ICD in v-Src tumors implanted into immunocompetent mice in vivo. The presence of nuclear Trop2 ICD was demonstrated in the v-Src prostate tumors in vivo (Fig. 4E). The relative staining of the Trop2 ICD in the v-Src prostate tumors was increased 2-fold compared with the nontransformed murine prostate gland (Fig. 4F). The induction of Trop2 abundance assessed by Western blot analysis was increased approximately 10-fold in a series of v-Src tumors derived after implantation in FVB mice (Fig. 4G). RNA extracted from v-Src tumor and normal ventral prostate tissues was assessed by microarray analysis. The relative mRNA abundance of Trop2 was induced 3.5-fold. We examined the abundance of several other genes associated with the prostate cancer stem cell signature (33). Notch1 was induced 3.5-fold. CD44 mRNA was induced more than 5-fold (Fig. 4H); CD44 is a cell adhesion glycoprotein that participates in presentation of cytokines and associates with stem cell functions (34). The abundance of Wnt7A was induced 50% (Fig. 4H), consistent with recent studies suggesting heterotypic signaling from the interstitium maintains prostate cancer progression (35).

Previous studies in human breast cancer cells demonstrated that cyclin D1 enhances Notch1 activity through inducing γ-secretase activity (19). The γ-secretase cleavage complex components PS1 and PS2 contribute to the γ-secretase activity governing Trop2 cleavage (17). We therefore examined the potential importance of cyclin D1 in the induction of the Trop2 ICD. To determine the mechanism by which Src induces cyclin D1 abundance, we first examined the relative abundance of cyclin D1 in v-Src versus parental PECs. Cyclin D1 mRNA was induced 2-fold by v-Src transformation (18). The −3,400 bp cyclin D1 promoter fragment linked to a luciferase reporter gene was induced 1.7-fold by v-Src (Fig. 5A) and reduced 70% by dasatinib in v-Src PECs (Fig. 5B). Immunohistochemical staining for Trop2 ICD was conducted in cyclin D1^+/+ and cyclin D1^−/− mouse prostate. We observed a 4-fold decrease of nuclear Trop2 ICD in cyclin D1 knockout prostates when compared with normal wild-type prostate (Fig. 5C). To determine whether cyclin D1–mediated induction of Trop2 ICD nuclear abundance is regulated in transformed prostate cancer cells, a cyclin D1 shRNA linked to a tet-inducible red fluorescent protein (RFP) was introduced into the LNCaP prostate cancer epithelial cell line (Fig. 5D). The addition of doxycycline induced the expression system and thereby RFP (Fig. 5D). Western blot analysis demonstrated cyclin D1 shRNA reduced the relative abundance of cyclin D1 by >50%, associated with a reduction in Trop2 ICD by approximately 50% (Fig. 5E). Vinculin, used as a control for normalization of protein abundance, was unchanged (Fig. 5E). Immunofluorescence staining for nuclear Trop2 in LNCaP cells treated with cyclin D1 shRNA demonstrated the reduction in nuclear Trop2 ICD upon induction of cyclin D1 shRNA (Fig. 5F and G). CBF-1 (C-promoter binding factor-1), the mammalian homolog of Suppressor of Hairless (Drosophila) and CSL protein in C. elegans, also known as RBP-JK, is activated by the Trop2 ICD. As our studies showed v-Src enhanced Trop2 cleavage, we determined whether CBF activity was maintained by v-Src kinase activity in prostate cancer cells. CBF activity was assessed using a synthetic CBF luc reporter in v-Src-PEC cell line. The Src kinase inhibitors dasatinib and PP1 reduced CBF activity in a dose-dependent manner (Fig. 5H). Cyclin D1 siRNA reduced CBF (CBF₈luc) activity approximately 47% in v-Src-PEC cell line (Fig. 5I).

Intramembrane proteolysis of Trop2 occurs via a γ-secretase cleavage complex (which includes PS1 and PS2) and TACE (ADAM17). To determine the mechanism by which cyclin D1 enhances Trop2 ICD accumulation, we considered that cyclin D1 may increase the abundance of the Trop2 cleavage complex. Immunohistochemical staining of cyclin D1^+/+ and cyclin D1^−/− mouse prostate glands was conducted to examine the abundance of the components regulating Trop2 ICD abundance. TACE and PS2 are the key inducers of Trop2 cleavage (17). Therefore, we examined TACE and PS2 in cyclin D1^+/+ and cyclin D1^−/− prostate (Fig. 6A). The deletion of the cyclin D1 gene reduced TACE and PS2 abundance in the prostate by 50% (Fig. 6B and C). Previous studies had demonstrated that cyclin D1 represses Numb to thereby induce Notch1 activity in breast cancer cells (19). As Notch can also activate CBF, we considered the possibility that endogenous cyclin D1 may induce CBF through repression of Numb in prostate epithelium. Consistent with prior findings in the mammary gland, we found cyclin D1 gene deletion enhanced Numb abundance in the prostate in vivo (Fig. 6D). To determine whether the mRNA levels of the Trop2 cleavage complex were induced by cyclin D1, mRNA levels of the Trop2 proteolytic cleavage components were assessed in prostate tissues of cyclin D1^+/+ and cyclin D1^−/− mice by RT-PCR (Fig. 6E). Cyclin D1 reduced Numb and increased PS2 and TACE (Fig. 6E). To determine whether cyclin D1 maintains expression of the Trop2 cleavage complex in transformed prostate cells, Western blot analysis was conducted on cyclin D1–transduced v-Src PECs. Commensurate with the approximately 50% reduction in cyclin D1 abundance, Trop2 ICD abundance was reduced approximately 50%, and PS2 was reduced approximately 50% (Fig. 6F).

Cyclin D1 is known to convey kinase-dependent and -independent functions. To determine whether the induction of Trop2 by cyclin D1 was kinase dependent, we examined the effect of inhibiting cyclin D1/cdk activity using the cdk inhibitors abemaciclib, palbociclib, and ribociclib. The cdk inhibitors reduced pRb^p at 24 hours, and reduced Trop2 ICD abundance at 48 and 72 hours. At 48 hours, the reduction in Trop2 ICD was 10% with abemaciclib (500 nmol/L), whereas palbociclib (>60%) and ribociclib (>60%) was more substantial.

To examine further the specificity of the effect mediated by oncogenic Src, we examined the possibility that growth factor signaling via the EGFR may induce Trop2 ICD. We therefore tested the EGFR antagonist canertinib. Canertinib is a 3-chloro 4-fluoro 4-anilinoquinazoline compound. It is a low molecular weight irreversible pan-EGFR family TKI and has been shown to inhibit cell proliferation via restraint of ERK/MAPK in a number of different cell types. Canertinib treatment of the v-Src-PEC inhibited cell proliferation (Supplementary Fig. S2A). Canertinib treatment of the v-Src-PEC did not affect levels of Trop2 ICD, and did not affect cyclin D1 levels. These findings suggest that dissociable pathways conduct selective signaling to Trop2-ICD abundance. Together, these studies demonstrate that v-Src induces cyclin D1, and that cyclin D1 induces expression of components of the Trop2 cleavage complex, directly through increasing PS2 and TACE, and indirectly through reducing Numb, thereby enhancing the abundance of the Trop2 ICD, and CBF activity in prostate cancer cells (Supplementary Fig. S3).

To determine whether the abundance of cyclin D1, Trop2 ICD, PS2, and Numb correlate with outcome in human prostate cancer patients, an annotated TMA of patients with prostate cancer was analyzed. IHC staining was conducted of a 126 prostate cancer patient TMA. The expression of cyclin D1, Trop2 ICD, PS2, and Numb was quantified (Supplementary Fig. S4). Kaplan–Meier analysis was used to evaluate the difference in recurrence-free survival associated with high expression versus low expression of cyclin D1, Trop2 ICD, PS2, and Numb. Survival probability of cyclin D1 high expression group was approximately 60%, whereas the low expression group is more than 90%. The difference was significant (P = 0.049; Fig. 7A). The survival probability of Trop2 ICD high expression group was approximately 25%, which was significantly increased to 73% in the low expression group (P = 0.008; Fig. 7B). The change in survival probability between PS2 groups was not significant (Fig. 7C). In the low Numb expression group, the survival probability was reduced to 0% compared with 72% for the high expression group (P < 0.001; Fig. 7D). These three target genes were combined and recurrence-free survival analysis was conducted. Using the Cox model based on the risk score determined by recursive partitioning with the predictors, cyclin D1, Numb, and Trop2, patients were assigned to high, medium, low recurrence risk groups, and the recurrence-free survival curves of different groups and HRs between the groups were analyzed. Compared with the low-risk group, the HR for survival was 3-fold increased in the medium-risk group and 4.3-fold in the high-risk group (P < 0.001; Fig. 7E).

Src kinase is increased in approximately 80% of human prostate cancers (8). The current studies provide several lines of evidence that v-Src transformation of PECs enriches for prostate cancer stem cells. First, v-Src transformation induced expression of genes identified in prostate cancer stem cells. Genes associated with the prostate cancer stem cell signature (33) included Notch1, which was induced 3.5-fold. CD44 mRNA, which encodes a cell adhesion glycoprotein that participates in presentation of cytokines and associates with stem cell functions (34), was induced more than 5-fold (Fig. 4H). The abundance of Wnt7A was induced 50%, consistent with recent studies suggesting heterotypic signaling from the interstitium maintains prostate cancer progression (35). CD133⁺ cells are enriched for cancer stem cells from prostate cancer (36). Together with α2β1 integrin, CD133 is used to enrich for stem cells that have increased proliferation potential and undergo full prostatic differentiation in vivo (37–39). Approximately 17% of genes enriched in the CD133 population were also induced by oncogenic Src transformation. Second, v-Src induced the abundance of the epitopes characteristic of prostate cancer stem cells. Third, v-Src induced prostatosphere formation and Src kinase inhibitors reduced the number of prostatospheres. Fourth, the abundance of the Trop2 ICD, which is known to promote stem cell expansion (17, 19), was enhanced by Src kinase activity.

In the current studies, cyclin D1 was required for v-Src–induced Trop2 ICD accumulation. The TNFα-converting enzyme (TACE), and γ-secretase, both participate in Trop2 ICD accumulation. TACE mediates the initial proteolysis and ectodomain shedding, followed by intramembrane proteolysis carried out by the γ-secretase complex. Herein, endogenous cyclin D1 was shown to enhance expression of the Trop2 cleavage complex by increasing the abundance of TACE (ADAM17) and PS2. TACE is a member of the ADAM family of proteases. Our prior studies demonstrated that endogenous cyclin D1 maintains ADAM protease expression in the mammary gland (40). In the current studies, cyclin D1 induced the abundance of TACE in the normal prostate gland. In prostate epithelial cells in culture and in transformed prostate cancer cells, cyclin D1 also induced expression of PS2, the catalytic subunit of the γ-secretase complex that conducts the sequential intramembrane proteolysis of Trop2 (40). Thus, consistent with the finding in the mammary gland in vivo, in which cyclin D1 induced the ADAM proteases, the current studies demonstrate cyclin D1 induces expression of the ADAM17 protease (TACE) in the prostate in vivo.

In the current studies endogenous cyclin D1 augmented CBF activity in a Src kinase–dependent manner. Trop2 and Notch augment activity of the transcription factor CBF-1 (suppressor of hairless, LAG1, also known as RBP-RJ; ref. 41) and LEF-1. Notch activity is restrained by endogenous Numb. In the current studies, endogenous cyclin D1 repressed Numb in prostate cancer cells. These findings are consistent with prior studies in breast cancer epithelial cells in which cyclin D1 augmented ErbB2-induced Notch1 activity, through repression of Numb (19). These prior studies demonstrated the induction of Notch signaling by cyclin D1 (19). Using cyclin D1 knockout mice, it was demonstrated that γ-secretase cleavage of Notch1 was enhanced by endogenous cyclin D1 (19). In a reciprocal feedback, Notch is known to induce cyclin D1 expression and cyclin D1 is required for Notch-induced transformation (42), consistent with findings that Notch1 and cyclin D1 expression correlate during embryogenesis (42).

The current studies demonstrated that a reduction in Numb, an increase in Trop2 ICD and an increase in cyclin D1 abundance, each predict poor outcome in patients with prostate cancer. When combined, these three genes assigned to high, medium, or low recurrence risk groups, gave a high-risk group with a HR 4.35 times greater than the low-risk group. Although cyclin D1 protein is overexpressed in a subset of prostate cancers, correlating with poor outcome, it is the cyclin D1-mediated gene expression signature that provides additional prognostic value to the Gleason score (1, 43). The current studies are consistent with a model in which the molecular targets of cyclin D1–mediated signaling in the prostate are important predictors of poor outcome in prostate cancer.

No potential conflicts of interest were disclosed.

Conception and design: X. Ju, X. Jiao, G. Di Sante, R.G. Pestell

Development of methodology: X. Ju, X. Jiao, G. Di Sante, Z. Li, A. Hawkins

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Ju, X. Jiao, G. Di Sante, S. Deng, A. Hawkins

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Ju, X. Jiao, A. Ertel, G. Di Sante, T. Zhan, L.G. Gomella, L.R. Languino, R.G. Pestell

Writing, review, and/or revision of the manuscript: X. Ju, X. Jiao, T. Stoyanova, S. Andò, A. Fatatis, M.P. Lisanti, L.G. Gomella, R.G. Pestell

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Ju, X. Jiao, M.C. Casimiro, A. Di Rocco, A. Hawkins, T. Stoyanova

Study supervision: X. Ju, X. Jiao, M.C. Casimiro

This work was supported in part by the NIH grants R01CA070896, R01CA075503, R01CA132115, R01CA107382, R01CA086072 (R.G. Pestell), and R01CA109874 (L.R. Languino), the Sidney Kimmel Cancer Center NIH Cancer Center Core grant P30CA056036 (R.G. Pestell), generous grants from the Dr. Ralph and Marian C. Falk Medical Research Trust (R.G. Pestell), and a grant from Pennsylvania Department of Health (R.G. Pestell). M.P. Lisanti and his laboratory were supported via the resources of Thomas Jefferson University.

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.