EZH2基因前列腺癌加剧有关

生物谷

Nature 419, 624 - 629 (2002); doi:10.1038/nature01075 The polycomb group protein EZH2 is involved in progression of prostate cancer SOORYANARAYANA VARAMBALLY*†, SARAVANA M. DHANASEKARAN*†, MING ZHOU*, TERRENCE R. BARRETTE*, CHANDAN KUMAR-SINHA*, MARTIN G. SANDA‡§, DEBASHIS GHOSH, KENNETH J. PIENTA‡§¶, RICHARD G. A. B. SEWALT#, ARIE P. OTTE #, MARK A. RUBIN*‡§ & ARUL M. CHINNAIYAN*‡§ * Department of Pathology, University of Michigan Medical School, Ann Arbor, Mic higan 48109, USA ‡ Department of Urology, University of Michigan Medical School, Ann Arbor, Michigan 48109, USA Department of Biostatistics, University of Michigan Medical School, Ann Arbor, Michigan 48109, USA ¶ Department of Internal Medicine, University of Michigan Medical School, A nn Arbor, Michigan 48109, USA § Comprehensive Cancer Center, University of Michigan Medical School, Ann Arbor , Michigan 48109, USA # Swammerdam Institute for Life Sciences, BioCentrum Amsterdam, University of Am sterdam, 1018 TV Amsterdam, The Netherlands † These authors contributed equally to this work Correspondence and requests for materials should be addressed to A.M.C. (e-mail: arul@umich.edu). Prostate cancer is a leading cause of cancer-related death in males and is secon d only to lung cancer. Although effective surgical and radiation treatments exis t for clinically localized prostate cancer, metastatic prostate cancer remains e ssentially incurable. Here we show, through gene expression profiling1, that the polycomb group protein enhancer of zeste homolog 2 (EZH2)2, 3 is overexpressed in hormone-refractory, metastatic prostate cancer. Small interfering RNA (siRNA) duplexes4 targeted against EZH2 reduce the amounts of EZH2 protein present in p rostate cells and also inhibit cell proliferation in vitro. Ectopic expression o f EZH2 in prostate cells induces transcriptional repression of a specific cohort of genes. Gene silencing mediated by EZH2 requires the SET domain and is attenu ated by inhibiting histone deacetylase activity. Amounts of both EZH2 messenger RNA and EZH2 protein are increased in metastatic prostate cancer; in addition, c linically localized prostate cancers that express higher concentrations of EZH2 show a poorer prognosis. Thus, dysregulated expression of EZH2 may be involved i n the progression of prostate cancer, as well as being a marker that distinguish es indolent prostate cancer from those at risk of lethal progression. Perturbation of the transcriptional 'memory' of a cell can lead to severe develo pmental defects5, 6. Lack of differentiation, or anaplasia, is a hallmark of can cer that results from normal cells 'forgetting' their cellular identity. Thus, i t is not surprising that dysregulation of the transcriptional maintenance system can lead to malignancy6-8. Two groups of proteins have been implicated in maint aining homeotic gene expression and include polycomb (PcG) and trithorax (trxG) group proteins9, 10. PcG proteins act in large complexes and are thought to repr ess gene expression, whereas trxG proteins are operationally defined as antagoni sts of PcG proteins and thus activate gene expression6, 9. In human malignancies , PcG and trxG proteins have been found to be dysregulated mainly in cells of ha ematopoietic origin11-13. EZH2 is the human homologue of the Drosophila protein Enhancer of Zeste (E(z))2, which genetic data define as a PcG protein with addit ional trxG properties10. E(z) and EZH2 contain a SET domain, a highly conserved domain found in chromatin-associated regulators of gene expression that often mo dulate cell growth pathways14. We previously reported gene expression profiles of benign prostate, clinically l ocalized prostate cancer and metastatic prostate cancer1. To characterize genes that specifically mark the molecular transition from organ-confined prostate can cer to metastatic prostate cancer, we implemented a statistical technique called significance analysis of microarrays (SAM)15. This approach identified 55 genes that were significantly upregulated (Fig. 1a) and 480 genes that were significa ntly downregulated (Supplementary Information) in metastatic prostate cancer rel ative to localized prostate cancer (at a false discovery rate (FDR) of 0.9% on m icroarrays containing 10,000 genes). Notably, the gene at the top of the 'list' of upregulated genes in metastatic prostate cancer was EZH2, with a d-score15 of 4.58 and a gene-specific FDR of 0.0012 (see Supplementary Information for the c omplete SAM analysis). Fig. 1b summarizes the gene expression of EZH2 in 74 pros tate tissue specimens analysed on DNA microarrays1. The EZH2 transcript was sign ificantly increased in metastatic prostate cancer with respect to clinically loc alized prostate cancer (Mann–Whitney test, P = 0.001) and benign prostate (P = 0.0001). Figure 1 Overexpression of EZH2 in metastatic hormone-refractory prostate cance r (MET). Full legend High resolution image and legend (70k) To validate the DNA microarray results, we carried out polymerase chain reaction with reverse transcription (RT–PCR) on 18 prostate samples and cell lines. As expected, the levels of EZH2 mRNA were increased in malignant relative to benign prostate samples (Fig. 1c). We also analysed tissue extracts by immunoblotting. The amounts of EZH2 protein were markedly increased in metastatic prostate cancer relative to localized prostate cancer or benign prostate (Fig. 1d). Notably, EED, a PcG protein that forms a complex with EZH2 (refs 16, 17), did not show similar protein dysregulation. We evaluated the expression of EZH2 protein in a wide range of prostate tissues (n = 1,023 tissue microarray elements) to determine the extent of its expression in situ (Fig. 2a, b). Of these samples, 400 were from 23 individuals who died o f hormone-refractory metastatic prostate cancer18. When highly expressed, EZH2 w as observed mainly in the nucleus (Fig. 2a, right), as suggested previously11. T he intensity of EZH2 staining increased from benign, prostatic atrophy, prostati c intraepithelial neoplasia, to clinically localized prostate cancer, with a res pective median staining intensity of 1.5 (standard error (s.e.) 0.04, 95% confid ence interval (CI) 1.4–1.6), 1.6 (s.e. 0.2, 95%CI 1.3–2.0), 2.3 (s.e. 0.2, 95% CI 1.9–2.7) and 2.6 (s.e. 0.1, 95%CI, 2.5–2.7), respectively (Fig. 2b). The st rongest EZH2 expression was in metastatic prostate cancer, with a median stainin g intensity of 3.1 (s.e. 0.03, 95%CI 3.0–3.2). There was a statistically significant difference in EZH2 staining intensity between benign prostate tissue and localized prostate cancer (analysis of variance (ANOVA) post-hoc analysis mean difference 1.0, P < 0.0001). Likewise, metastatic prostate cancer had significantly higher expression of EZH2 than did clinically localized prostate cancer (ANOVA post-hoc analysis mean difference 0.5, P < 0.0001). These findings suggested that as prostate neoplasia progresses there is a trend towards increased expression of EZH2 protein. They also suggested that EZH2 concentrations might indicate how aggressive an individual's prostate cancer is, given that the highest expression was observed in hormone-refractory, metastatic prostate cancer. Figure 2 Amounts of EZH2 protein correlate with the aggressiveness of prostate cancer. Full legend High resolution image and legend (52k) We therefore examined whether EZH2 protein could be used to predict clinical out come in men treated with surgery for clinically localized prostate cancer. Using a previously validated tissue microarray19, we examined 278 specimens from 64 i ndividuals with clinical follow-up. To test the utility of EZH2 as a potential t issue biomarker for prostate cancer, we examined the clinical outcome of these 6 4 cases, taking into account clinical and pathological parameters. Clinical fail ure was defined as either an increase of 0.2 ng ml-1 prostate-specific antigen ( PSA) or recurrence of disease after prostatectomy (for example, development of m etastatic disease). By Kaplan–Meier analysis (Fig. 2c), an EZH2 staining intensity of 3 or higher was associated significantly with clinical failure in 31% (10/32) of individuals (log rank P = 0.03), whereas a staining intensity of less than 3 was found in 9% (3/32) of individuals with clinical failure. There was no significant correlation between EZH2 amounts and Gleason score, tumour stage or surgical margin status. Notably, there was a significant (P = 0.048) albeit weak (Pearson coefficient = 0.33) correlation between EZH2 protein and proliferation index in situ, as assessed by the Ki-67 labelling index (Supplementary Information). Multivariable Cox hazards regression analysis showed that EZH2 protein expression (3 versus <3) was the best predictor of clinical outcome with a recurrence ratio of 4.6 (95%CI 1.2?C17.1, P = 0.02), which was significantly better than surgical margin status, ma ximum tumour dimension, Gleason score and pre-operative PSA (Supplementary Infor mation). Thus, monitoring the amounts of EZH2 protein in prostate specimens may provide additional prognostic information that is not discernible with current c linical and pathology parameters alone. To examine the role of EZH2 in prostate cancer progression, we used RNA interfer ence (RNAi) to disrupt EZH2 expression in transformed prostate cells in vitro. W e designed duplexes of 21-nucleotide RNA (siRNAs) to target EZH2 (ref. 4) and te sted them on the transformed androgen-responsive prostate cell line RWPE20, 21 a nd the metastatic prostate cancer cell line PC3. After 48 h of transfection with siRNA duplexes, quantification of the amounts of endogenous EZH2 protein showed a specific downregulation of protein in prostate cell lines (Fig. 3a). Sense or antisense oligonucleotides, as well as unrelated siRNA duplexes, did not affect the amounts of EZH2 (Fig. 3a, middle and right), verifying the specificity of t he siRNA approach. Figure 3 A role for EZH2 in prostate cell proliferation. Full legend High resolution image and legend (50k) We next examined the phenotype of the prostate cells with reduced EZH2 expressio n. RWPE cell counts taken 48 h after transfection with siRNA showed a marked (62 %) inhibition of cell growth caused by the EZH2 siRNA duplex, whereas the corres ponding sense and antisense EZH2 oligonucleotides or unrelated duplexes showed m inimal inhibition (Fig. 3b). The PC3 prostate cancer cell line showed a similar growth inhibition by EZH2 siRNA, indicating that these findings were not specifi c to the RWPE cell line (Fig. 3b). The effect of EZH2 siRNA on prostate cell gro wth was monitored at time points ranging from 24 to 120 h (Fig. 3c), during whic h over 80% growth inhibition was observed. Using a commercially available cell proliferation reagent WST-1, which measures mitochondrial dehydrogenase activity, we observed a decrease in cell proliferati on in cells transfected with the EZH2 siRNA duplex but not with the unrelated du plexes (Fig. 3d). In the time frame considered (48 h), RNAi of EZH2 did not indu ce apoptosis, as assessed by propidium iodide staining of nuclei or cleavage of poly(adenosine diphosphate)-ribosylpolymerase (data not shown). Consistent with this, the broad-spectrum caspase inhibitor, zVAD-fmk, failed to attenuate the in hibition of cell proliferation induced by EZH2 siRNA (Fig. 3d). Notably, flow cy tometric analysis of prostate cells treated with EZH2 siRNA showed cell-cycle ar rest in the G2/M phase (Fig. 3e). Unrelated control siRNA duplexes failed to ind uce a similar dysregulation of the cell cycle. Together, these observations indi cate that EZH2 may be involved in the proliferation of prostate cells by mitigat ing the G2/M transition. To understand further the functional role of EZH2 in prostate cells, we generate d an epitope-tagged version of wild-type EZH2 and a deletion mutant of EZH2 lack ing the conserved SET domain14 in the eukaryotic expression vector, pcDNA3 (Fig. 4a, b). We also designed an 'inducible' version22, 23 of EZH2 by creating a fus ion protein with a modified murine oestrogen receptor (ER; Fig. 4a, b). Figure 4 EZH2 functions as a transcriptional repressor in prostate cells. Ful l legend High resolution image and legend (209k) PcG proteins have been proposed to mediate their functions by repressing target genes2, 5. To test this hypothesis, we transiently transfected RWPE prostate cel ls with wild-type EZH2 and monitored alterations in global gene expression using DNA microarrays. The RNA from the transfected cell line was labelled with one f luorescent dye, whereas the paired reference sample was labelled with a second d istinguishable fluorescent dye. By making direct comparisons between 'gene'-tran sfected cell lines and cells lines transfected with control vector, we emphasize d the molecular differences between the samples and obviated the need for comple te transfection efficiency. When EZH2 was overexpressed in RWPE cells or SUM149 breast carcinoma cells, ther e was a consistent repression of a cohort of genes (Fig. 4c, d). When compared w ith vector-transfected cells, the only gene that was upregulated significantly i n EZH2-transfected cells was EZH2 itself (Fig. 4c). The EZH2-mediated transcript ional repression was dependent on an intact SET domain (Fig. 4c), because deleti on of this domain did not produce a repressive phenotype and in some cases 'de-r epressed' genes. EZH2 interacts with histone deacetylase 2 (HDAC2) through the E ED protein24, and in our system EZH2-mediated gene silencing was dependent on HD AC activity, because the commonly used HDAC inhibitor trichostatin A (TSA) compl etely abrogated the effects of EZH2 (Fig. 4c). Thus, EZH2 function requires both an intact SET domain and endogenous HDAC activity. To identify genes that are significantly repressed by EZH2, we compared cells tr ansfected with wild-type EZH2 with cells transfected with EZH2SET. We found that 163 genes were consistently repressed, whereas no genes were activated at an FD R of 0.0045 (Fig. 4d, and Supplementary Information). The list of genes that wer e repressed indicates several intriguing associations, including EZH2-mediated r epression of specific PcG proteins, transcription factors and cell-cycle regulat ors (Supplementary Information). In summary, we have identified the association of EZH2 with advanced prostate ca ncer by gene expression profiling of tumour specimens from individuals who died of metastatic disease. Measuring EZH2 amounts (in prostate cancer specimens) mig ht have potential as a molecular determinant of prostate cancer progression and metastasis. In addition, we have shown that EZH2 has a role in mediating cell pr oliferation and transcriptional repression in prostate cells. Overall, our study indicates that dysregulation of the transcriptional memory machinery may contri bute to the lethal progression of prostate cancer and may provide a potential me chanism for the constellation of genes repressed in metastatic disease. Methods SAM analysis of prostate cancer gene expression We carried out SAM analysis by c omparing gene expression profiles of metastatic prostate cancer samples with tho se of clinically localized prostate cancer samples from our previous work1. Gene s were analysed using Cluster25, implementing average linkage hierarchical clust ering of genes, and the output (Supplementary Information) was visualized by Tre eview25. RT–PCR We carried out RT–PCR amplifications with 1 µg of total RNA isola ted from the indicated prostate tissues and cell lines. Primer sequences are giv en in the Supplementary Information. Immunoblot analysis Prostate tissue extracts were separated by SDS–PAGE and blotted onto nitrocellulose membranes. Antibodies against EZH2 and EED (ref. 17) and a polyclonal antibody against -tubulin (Santa Cruz) were used at 1:1,000 dilution for immunoblot analysis. Tissue microarray analysis The clinically stratified prostate cancer tissue micr oarrays used in this study have been described1, 19, 26. Tissues were from the r adical prostatectomy series at the University of Michigan and from the Rapid Aut opsy Program, which are both part of Michigan Prostate SPORE Tissue Core. We obt ained Institutional Review Board approval to procure and analyse the tissues use d in this study. We evaluated EZH2 protein expression on a wide range of prostate tissue to deter mine the intensity and extent of expression in situ. Immunohistochemistry was ca rried out on three high-density tissue microarrays containing samples of benign prostate, prostatic atrophy, prostatic intraepithelial neoplasia, clinically loc alized prostate cancer and metastatic prostate cancer. We used standard biotin–avidin complex immunohistochemistry to evaluate EZH2 protein expression using an antibody against EZH2. Protein expression was scored as negative (score = 1), weak (2), moderate (3) and strong (4). Four replicate tissue cores were sampled from each of the selected tissue types. EZH2 expression was evaluated in a blind manner separately by M.A.R. and M.Z. using a validated web-based tool26, 27, and the median value of all measurements from a single individual was used for subsequent analysis. Staining assessment was highly reproducible between the two pathologists with a value of 0.73. Clinical outcomes analysis To assess individual variables for risk of recurrence , we used Kaplan–Meier survival analysis and created a univariate Cox proportional hazards model. PSA recurrence was defined as greater than 0.2 ng ml-1 after radical prostatectomy. Covariates included Gleason sum, preoperative PSA, maximum tumour dimension, tumour stage and surgical margin status. To assess the influence of several variables simultaneously, including EZH2 protein expression, we developed a final multivariate Cox proportional hazards model of statistically significant covariates. Statistical significance in univariate and multivariate Cox models was determined by Wald's test. A P value of less than 0.05 was considered statistically significant. EZH2 constructs Myc-tagged EZH2–pCMV was a gift from T. Jenuwein. The Myc–EZH2 fragment was subcloned into the expression vector pCDNA3 (Invitrogen). The EZH2 –ER and EZH2SET constructs are described in the Supplementary Information. RNAi We chemically synthesized 21-nucleotide sense and antisense RNA oligonucleo tides (Dharmacon) and annealed them to form duplexes. The EZH2 siRNA was targete d to the region corresponding to residues 85–106 of human EZH2 (NM004456). Control siRNA duplexes targeted luciferase, lamin and AMACR (NM014324). The human transformed prostate cell lines RWPE20 and PC3 were plated at 2 105 cells per well in a 12-well plate for immunoblot analysis, cell counts and FACS analysis, and at 1.5 104 cell per well in a 96-well plate for WST-1 proliferation assays. Twelve hours after plating, the cells were transfected with 60 pmol of siRNA duplex, sense or antisense oligonucleotides using oligofectamine (Invitrogen). We carried out a second identical transfection 24 h later. At certain time points after the first transfection, the cells were lysed for immunoblot analysis and treated with trypsin for estimating cell numbers or FACS analysis. For cell counts at 96 and 120 h, the cells were treated with trypsin and replated in 6-well dishes 64 h after the first transfection. Cell proliferation assays Cell proliferation was determined by a colorimetric as say of cell viability that is based on the cleavage of the tetrazolium salt WST- 1 by mitochondrial dehydrogenases (Roche). The absorbance of the formazan dye fo rmed, which correlates with the number of metabolically active cells in the cult ure, was measured at 450 nm 1 h after adding the reagent. Cell counts were estim ated by treating the cells with trypsin, followed by analysis on a Coulter cell counter at specified time points. Flow cytometric analysis Trypsin-treated cells were washed with PBS and fixed in 70% ethanol overnight for FACS analysis. Before staining with propidium iodide, the cells were washed again with PBS and analysed by flow cytometry. Microarray analysis of EZH2-transfected cells Initial testing of the transient t ransfection analysis system (data not shown) showed that overexpression of TNFR1 (p55) induced similar expression profiles to those observed after incubating ce lls with TNF28. Samples expressing the EZH2SET mutant were compared with those e xpressing EZH2 using the SAM analysis package (ref. 15 and Supplementary Informa tion). Supplementary information accompanies this paper. Received 13 May 2002;accepted 31 July 2002 References 1. Dhanasekaran, S. M. et al. Delineation of prognostic biomarkers in prostate cancer. Nature 412, 822-826 (2001) | Article | PubMed | 2. Laible, G. et al. Mammalian homologues of the Polycomb-group gene Enhancer of zeste mediate gene silencing in Drosophila heterochromatin and at S. cerevisiae telomeres. EMBO J. 16, 3219-3232 (1997) | Article | PubMed | 3. Satijn, D. P. & Otte, A. P. Polycomb group protein complexes: do different co mplexes regulate distinct target genes? Biochim. Biophys. Acta 1447, 1-16 (1999) | Article | PubMed | 4. Elbashir, S. M. et al. Duplexes of 21-nucleotide RNAs mediate RNA interferenc e in cultured mammalian cells. Nature 411, 494-498 (2001) | Article | PubMed | 5. Jacobs, J. J. & van Lohuizen, M. Cellular memory of transcriptional states by Polycomb-group proteins. Semin. Cell. Dev. Biol. 10, 227-235 (1999) | Article | PubMed | 6. Francis, N. J. & Kingston, R. E. Mechanisms of transcriptional memory. Nature Rev. Mol. Cell Biol. 2, 409-421 (2001) | Article | PubMed | 7. Jacobs, J. J., Kieboom, K., Marino, S., DePinho, R. A. & van Lohuizen, M. The oncogene and Polycomb-group gene bmi-1 regulates cell proliferation and senesce nce through the ink4a locus. Nature 397, 164-168 (1999) | Article | PubMed | 8. Jacobs, J. J. et al. Bmi-1 collaborates with c-Myc in tumorigenesis by inhibi ting c-Myc-induced apoptosis via INK4a/ARF. Genes Dev. 13, 2678-2690 (1999) | Ar ticle | PubMed | 9. Mahmoudi, T. & Verrijzer, C. P. Chromatin silencing and activation by Polycom b and trithorax group proteins. Oncogene 20, 3055-3066 (2001) | Article | PubMed | 10. LaJeunesse, D. & Shearn, A. E(z): a polycomb group gene or a trithorax group gene? Development 122, 2189-2197 (1996) | PubMed | 11. Raaphorst, F. M. et al. Coexpression of BMI-1 and EZH2 polycomb group genes in Reed-Sternberg cells of Hodgkin's disease. Am. J. Pathol. 157, 709-715 (2000) | PubMed | 12. van Lohuizen, M. et al. Identification of cooperating oncogenes in Eµ- myc transgenic mice by provirus tagging. Cell 65, 737-752 (1991) | PubMed | 13. Haupt, Y., Alexander, W. S., Barri, G., Klinken, S. P. & Adams, J. M. Novel zinc finger gene implicated as myc collaborator by retrovirally accelerated lymp homagenesis in Eµ-myc transgenic mice. Cell 65, 753-763 (1991) | PubMed | 14. Jenuwein, T., Laible, G., Dorn, R. & Reuter, G. SET domain proteins modulate chromatin domains in eu- and heterochromatin. Cell. Mol. Life Sci. 54, 80-93 (1 998) | Article | PubMed | 15. Tusher, V. G., Tibshirani, R. & Chu, G. Significance analysis of microarrays applied to the ionizing radiation response. Proc. Natl Acad. Sci. USA 98, 5116- 5121 (2001) | Article | PubMed | 16. van Lohuizen, M. et al. Interaction of mouse polycomb-group (Pc-G) proteins Enx1 and Enx2 with Eed: indication for separate Pc-G complexes. Mol. Cell. Biol. 18, 3572-3579 (1998) | PubMed | 17. Sewalt, R. G. et al. Characterization of interactions between the mammalian polycomb-group proteins Enx1/EZH2 and EED suggests the existence of different ma mmalian polycomb-group protein complexes. Mol. Cell. Biol. 18, 3586-3595 (1998) | PubMed | 18. Rubin, M. A. et al. Rapid ('warm') autopsy study for procurement of metastat ic prostate cancer. Clin. Cancer Res. 6, 1038-1045 (2000) | PubMed | 19. Perrone, E. E. et al. Tissue microarray assessment of prostate cancer tumour proliferation in African-American and white men. J. Natl Cancer Inst. 92, 937-9 39 (2000) | Article | PubMed | 20. Webber, M. M., Bello, D., Kleinman, H. K. & Hoffman, M. P. Acinar differenti ation by non-malignant immortalized human prostatic epithelial cells and its los s by malignant cells. Carcinogenesis 18, 1225-1231 (1997) | Article | PubMed | 21. Bello, D., Webber, M. M., Kleinman, H. K., Wartinger, D. D. & Rhim, J. S. An drogen responsive adult human prostatic epithelial cell lines immortalized by hu man papillomavirus 18. Carcinogenesis 18, 1215-1223 (1997) | Article | PubMed | 22. Littlewood, T. D., Hancock, D. C., Danielian, P. S., Parker, M. G. & Evan, G . I. A modified oestrogen receptor ligand-binding domain as an improved switch f or the regulation of heterologous proteins. Nucleic Acids Res. 23, 1686-1690 (19 95) | PubMed | 23. Juin, P., Hueber, A. O., Littlewood, T. & Evan, G. c-Myc induced sensitizati on to apoptosis is mediated through cytochrome c release. Genes Dev. 13, 1367-13 81 (1999) | PubMed | 24. van der Vlag, J. & Otte, A. P. Transcriptional repression mediated by the hu man polycomb-group protein EED involves histone deacetylation. Nature Genet. 23, 474-478 (1999) | Article | PubMed | 25. Eisen, M. B., Spellman, P. T., Brown, P. O. & Botstein, D. Cluster analysis and display of genome-wide expression patterns. Proc. Natl Acad. Sci. USA 95, 14 863-14868 (1998) | Article | PubMed | 26. Manley, S., Mucci, N. R., De Marzo, A. M. & Rubin, M. A. Relational database structure to manage high-density tissue microarray data and images for patholog y studies focusing on clinical outcome: the prostate specialized program of rese arch excellence model. Am. J. Pathol. 159, 837-843 (2001) | PubMed | 27. Bova, G. S. et al. Web-based tissue microarray image data analysis: initial validation testing through prostate cancer Gleason grading. Hum. Pathol. 32, 417 -427 (2001) | Article | PubMed | 28. Kumar-Sinha, C., Varambally, S., Sreekumar, A. & Chinnaiyan, A. M. Molecular cross-talk between the TRAIL and interferon signaling pathways. J. Biol. Chem. 277, 575-585 (2001) | Article | Acknowledgements. This study was made possible by tissues donated by patients wi th metastatic prostate cancer enrolled in the University of Michigan, Rapid Auto psy Program funded by the Specialized Program of Research Excellence (SPORE) in Prostate Cancer at the National Cancer Institute. We thank J. Wei for clinical d ata collection; K. Hamer for preparing polycomb antibodies; R. Kunkel for figure preparation; J. Harwood and M. LeBlanc for technical assistance; A. Menon for s equence verification; and C. Ingold and G. Tueckmantel for database assistance. A.M.C. is a Pew Foundation Scholar. This work is supported in part by grants fro m the NIH (A.M.C.), CaPCURE (A.M.C.) and the Michigan SPORE in Prostate Cancer ( K.P., M.A.R., A.M.C. and M.G.S.).