AU2011213720B2 - Methods of diagnosing or treating prostate cancer using the ERG gene, alone or in combination with other over or under expressed genes in prostate cancer - Google Patents

Abstract

The present invention relates to oncogenes or tumor suppressor genes, as well as other genes, involved in prostate cancer and their expression products, as well as derivatives and s analogs thereof. Provided are therapeutic compositions and methods of detecting and treating cancer, including prostate and other related cancers. Also provided are methods of diagnosing and/or prognosing prostate cancer by determining the expression level of at least one prostate cancer-cell-specific gene, including, for example, the ERG gene or the LTF gene alone, or in combination with at least one of the AMACR gene and the DD3 gene. 2780598_1 (GHMMtr) P82730AU.I

Description

AUSTRALIA Patents Act 1990 COMPLETE SPECIFICATION Standard Patent Applicant: The Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc. Invention Title: METHODS OF DIAGNOSING OR TREATING PROSTATE CANCER USING THE ERG GENE, ALONE OR IN COMBINATION WITH OTHER OVER OR UNDER EXPRESSED GENES IN PROSTATE CANCER The following statement is a full description of this invention, including the best method for performing it known to us: METHODS OF DIAGNOSING OR TREATING PROSTATE CANCER USING THE ERG GENE, ALONE OR IN COMBINATION WITH OTHER OVER OR UNDER EXPRESSED GENES IN PROSTATE CANCER 5 GOVERNMENT INTEREST [001] The invention described herein may be manufactured, licensed, and used for governmental purposes without payment of royalties thereon. 10 FIELD OF THE INVENTION [002] The present invention relates to oncogenes, tumor suppressor genes, as well as other genes, and their expression products, involved in prostate cancer, as well as derivatives and analogs thereof. The invention further relates to therapeutic compositions and methods of detecting, diagnosing, and treating cancer, including prostate and other related cancers. 15 [003] The entire disclosure in the complete specification of our Australian Patent Application No. 2005245815 is by this cross-reference incorporated into the present specification. 20 BACKGROUND OF THE INVENTION [004] Prostate cancer (CaP) is the most common malignancy in American men and second leading cause of cancer mortality (Landis et al. (1999) Cancer J. Clin., 49:8-3 1; Jemal et al. (2004) Cancer J Clin 54:8-29). The molecular determinants in the development and progression of this disease are poorly understood. In recent years, there have been intensive 25 investigations of molecular genetics of the CaP. To date, however, oncogene, tumor suppressor gene, or other gene alterations common to most CaPs have not been found. Alterations of tumor suppressors such as p53, PTEN and p27, or oncogenes such as BCL2, HER2 and C-MYC associate with only small subsets of primary CaP, with more frequent association observed in advanced CaP. 30 [005] Current clinical parameters, including serum Prostate Specific Antigen (PSA), tumor stage, and Gleason score are routinely used as risk factors at the time of diagnosis, but have limited application to identify patients at a greater risk for developing aggressive CaP. Approximately 30-40% of patients treated with radical prostatectomy for localized CaP have 35 been found to have microscopic disease that is not organ-confined and a significant portion of these patients relapse. (Singh et al., Cancer Cell (2000) 1:203-209; la 2780598_1 (GHMattm)P62730.AU. I Henshell et at., Can. Res. (2003) 63: 4196-4203). Therefore, discovery of novel biomarkers or gene expression patterns defining CaP onset and progression is crucial in predicting patients with greater risk to develop aggressive CaP. [006] CaP-specific genetic alterations have been the subject of intensive research 5 by several investigations in the past five years (Srikatan et al., In Prostate Cancer, Diagnosis and Surgical Treatment (2002) Springer-Verlag, 25-40; Karan et al., Int. J. Can. (2003) 103:285-293; Augustus et al., In Molecular Pathology of Early Cancer (1999) IOS press: 321-340; Moul et al., Clin Prostate Cancer (2002) 1:42-50; Lalani et al., Cancer and Mets Rev (1997) 16: 29-66; Issacs et al., Epidemiol Rev (2001) 23:36-41; Ozen et al., 10 Anticancer Res (2000) 20:1905-1912; Morton et al., 3 Natl Med Assoc (1998) 90:S728 731). Promising leads both in biology and translational research areas are beginning to emerge from recent genomics and proteomics technology, as well as traditional approaches. However, the inherent heterogeneity of CaP has hampered the molecular characterization of CaP. 15 [007] One of the challenges in studying molecular alterations in human cancers, including prostate tumors, is to define the relative contributions of genetic alterations in epithelial and non-epithelial components of the target organ in the process of tumorigeneis. Despite advances in technology, changes in human CaP-specific epithelial and stromal cell-associated gene expression are still not well understood. 20 [008] Despite recent advances in the identification of molecular alterations associated with certain prostate cancers, the heterogeneous nature of prostate tissue has hindered the identification of genetic targets common to all, or at least the vast majority of, prostate cancers. The complexity and heterogeneity of prostate cancer has also hindered the identification of targets that allow differentiation between clinically aggressive and 25 non-aggressive cancers at the time of diagnosis. Therefore, there remains a need to identify molecular alterations specific for a pathologically defined cell population that can provide important clues for optimal diagnosis and prognosis, and help to establish individualized treatments tailored to the molecular profile of the tumor. [009] Citation of references herein shall not be construed as an admission that 30 such references are prior art to the present invention. 2 [009a] It is to be understood that, if any prior art publication is referred to herein, such reference does not constitute an admission that the publication forms a part of the common general knowledge in the art, in Australia or any other country. SUMMARY OF THE INVENTION [09b] A first aspect provides a method of diagnosing or prognosing prostate cancer in a subject, comprising: a) measuring the expression level of an ERG polypeptide in a biological sample from the subject; and b) correlating the expression level of the ERG polypeptide with the presence of prostate cancer in the subject or a higher predisposition of the subject to develop prostate cancer, wherein the expression level of the ERG polypeptide is correlated with longer PSA recurrence free survival, well and moderate tumor differentiation, a pathologic T stage of pT2 or lower, or a negative surgical margin status. [09c] A second aspect provides a method of identifying prostate cancer in a subject, comprising: a) providing a prostate tissue sample from the subject; b) contacting the prostate tissue sample with an antibody that binds to a human ERG polypeptide, wherein the antibody is a monoclonal antibody that binds to the ERG 1, ERG2, and ERG3 polypeptides; c) measuring the level of expression of the human ERG polypeptide in the prostate tissue sample by detecting the amount of binding between the antibody and the human ERG polypeptide in the prostate tissue sample; d) comparing the level of expression of the ERG polypeptide in the prostate tissue sample to the level of expression of the ERG polypeptide in a control sample; and e) identifying prostate cancer in the prostate tissue sample from the subject if the level of expression of the ERG polypeptide in the prostate tissue sample is increased as compared to the expression level of the ERG polypeptide in the control sample. 520954_1 (GHMaters) PS2730AU.1 18-Mar-14 [09d] A third aspect provides a monoclonal antibody, wherein the monoclonal antibody binds to an epitope formed by amino acids 275-322 of the polypeptide encoded by a nucleic acid comprising the sequence of SEQ ID NO: 1. [09e] A fourth aspect provides a monoclonal antibody, wherein the monoclonal antibody binds to an epitope formed by the amino acid sequence of SEQ ID NO:14. [010] Disclosed herein are methods and kits for detecting cancer, in particular prostate cancer. These methods and kits can be used to detect (either qualitatively or quantitatively) nucleic acids or proteins that serve as cancer markers. For example, the expression of the 3a 5209501 (GHMbIters) P673D.AU.1 18-Mar.I4 prostate cancer-cell-specific gene ERG, when detected in a biological sample from a subject, either alone or in combination with other cancer markers, including the expression of other prostate cancer-cell-specific genes, can be used to indicate the presence of prostate cancer in the subject or a higher predisposition of the subject to develop prostate cancer. Detecting ERG 5 expression, alone or in combination with the expression of any gene identified in Tables 1-6, can thus be used to diagnose or prognose cancer, particularly prostate cancer. [011] The method for detecting the expression of one or more prostate cancer cell specific genes, such as ERG, AMACR, and LTF or the DD3 gene, in a biological sample, may comprise: 10 (a) combining a biological sample with at least a first and a second oligonucleotide primer under hybridizing conditions, wherein the first oligonucleotide primer contains a sequence that hybridizes to a first sequence in a target sequence from a prostate cancer cell specific gene, such as ERG (SEQ ID NO:1), AMACR (SEQ ID NO:3), and/or LTF (SEQ ID NO:5) and/or DD3 (SEQ ID NO:4), and the second oligonucleotide primer contains a sequence 15 that hybridizes to a second sequence in a nucleic acid strand complementary to the target sequence, wherein the first sequence does not overlap with the second sequence, (b) adding at least one polymerase activity to produce a plurality of amplification products when the target sequence is present in the biological sample, (c) adding an oligonucleotide probe that hybridizes to at least one amplification !0 product of the target sequence, and (d) detecting whether a signal results from hybridization between the oligonucleotide probe and the at least one amplification product, wherein detection of the signal indicates the expression of a prostate cancer cell-specific gene in the biological sample. [012] The method preferably comprises detecting the expression of the following 25 combinations of genes: 1) ERG and AMACR; 2) ERG and DD3; and 3) ERG, AMACR and DD3. In another embodiment, the method comprises detecting LTF and one or more of ERG, AMACR and DD3. Expression of these genes can also be detected by measuring ERG, AMACR or LTF polypeptides in the biological sample. [0131 The biological sample is preferably a prostate tissue, blood, or urine sample. 30 Detecting a signal resulting from hybridization between the oligonucleotide probe and the at 4 354942_1 (GHMabters) P62730AU.1 4-Sep-12 least one amplification product can be used to diagnose or prognose cancer, particularly prostate cancer. [014] The oligonucleotide probe may be optionally fixed to a solid support. When detecting ERG expression in a biological sample, the oligonucleotide probe, first oligonucleotide 5 primer, and second oligonucleotide primer, each comprise a nucleic acid sequence that is capable of hybridizing under defined conditions (preferably under high stringency hybridization conditions, e.g., hybridization for 48 hours at 65*C in 6X SSC followed by a wash in 0.1 X SSX at 50'C for 45 minutes) to SEQ ID NO: I. Thus, the oligonucleotide probe, first oligonucleotide primer, and second oligonucleotide primer can include, for example, SEQ ID NO: I itself, or a 0 fragment thereof or a sequence complementary thereto. Preferably the oligonucleotide probe, first oligonucleotide primer, or second oligonucleotide primer is a fragment of SEQ ID NO:I having at least about 15, at least about 20, or at least about 50 contiguous nucleotides of SEQ ID NO: I or a sequence complementary thereto. When detecting ERG expression, the target sequence is preferably a fragment of SEQ ID NO: I. Probes, primers, and target sequences can [5 be similarly derived from other genes of interest, such as DD3 (SEQ ID NO:4), and other prostate cancer-cell-specific genes, including, for example, AMACR (SEQ ID NO:3) and LTF (SEQ ID NO:5). [015] The method of diagnosing or prognosing prostate cancer may comprise: measuring the expression level (e.g. mRNA or polypeptide) of an over expressed prostate !0 cancer cell-specific gene, such as ERG and/or AMACR, and/or the DD3 gene in a biological sample, and correlating the expression level of the ERG, AMACR, and/or DD3 gene with the presence of prostate cancer or a higher predisposition to develop prostate cancer in the subject. [016] The method of diagnosing or prognosing prostate cancer may comprise: 25 measuring the expression level (e.g. mRNA or polypeptide) of an under expressed prostate cancer cell-specific gene, such as LTF in a biological sample, and correlating the expression level of the LTF gene with the presence of prostate cancer or a higher predisposition to develop prostate cancer in the subject. [017] The skilled artisan will understand how to correlate expression levels or patterns of 30 the desired genes with the presence of prostate cancer or a higher predisposition to develop 5 3854942_1 (GHMatters) P82730.AU.1 4-Sep-12 prostate cancer. For example, the expression levels can be quantified such that increased or decreased expression levels relative to a control sample or other standardized value or numerical range indicate the presence of prostate cancer or a higher predisposition to develop prostate cancer. 5 [018] The increased or decreased expression levels in the methods of the invention may be measured relative to the expression level of the prostate cancer cell-specific gene or polypeptide in normal, matched tissue, such as benign prostate epithelial cells from the same subject. Alternatively, the expression level of a gene or polypeptide may be measured relative to the expression of the gene or polypeptide in other noncancerous samples from the subject or in 10 samples obtained from a different subject without cancer. Expression of a gene may also be normalized by comparing it to the expression of other cancer-specific markers. For example, in prostate cancer, a prostate-cell specific marker, such as PSA, can be used as a control to compare and/or normalize expression levels of other genes, such as ERG, LTF, DD3, and/or AMACR. By way of example, the method of diagnosing or prognosing prostate cancer comprises 15 measuring the expression levels of the ERG, DD3 and/or AMACR gene and diagnosing or prognosing prostate cancer, where an increased expression level of the ERG, DD3, and/or AMACR gene of at least two times as compared to the control sample indicates the presence of prostate cancer or a higher predisposition in the subject to develop prostate cancer. Conversely, by way of example, in such a method of diagnosing or prognosing prostate cancer, a decreased 20 expression of the LTF gene of at least two times as compared to the control sample indicates the presence of prostate cancer or a higher predisposition in the subject to develop prostate cancer. [019] The expression levels of prostate cancer cell-specific genes (e.g., mRNA or polypeptide expression) can be detected according to the methods described herein or using any other known detection methods, including, without limitation, immunohistochemistry, Southern 25 blotting, Northern blotting, Western blotting, ELISA, and nucleic acid amplification procedures, including but not limited to PCR, transcription-mediated amplification (TMA), nucleic acid sequence-based amplification (NASBA), self-sustained sequence replication (3SR), ligase chain reaction (LCR), strand displacement amplification (SDA), and Loop-Mediated Isothermal Amplification (LAMP). 6 3654942_1 (GHMaUets) P62730AU.1 4-Sep-12 [020] Disclosed herein is a method of determining a gene expression pattern in a biological sample, where the pattern can be correlated with the presence or absence of tumor cells, particularly prostate tumor cells. For example, ERG is detected in combination with other prostate cancer cell-specific genes (identified in Tables 1-6), including AMACR and/or LTF, to 5 obtain expression profiles from biological samples. The expression profiles of these prostate cancer-cell-specific genes are useful for detecting cancer, particularly prostate cancer. ERG can also be detected in combination with DD3, with or without other prostate cancer-cell-specific genes, such as AMACR and/or LTF, to obtain expression profiles from biological samples. These expression profiles are also useful for detecting cancer, particularly prostate cancer. 10 Increased levels of ERG, AMACR, and/or DD3 in a biological sample indicate the presence of prostate cancer or a higher predisposition in the subject to develop prostate cancer. Decreased levels of LTF in a biological sample indicate the presence of prostate cancer or a higher predisposition in the subject to develop prostate cancer. [021] Disclosed herein is a method of determining a gene expression pattern in a 15 biological sample, where the pattern can be used to indicate or predict the pathologic stage of cancer, particularly prostate cancer. For example, the gene expression pattern can be used to indicate or predict a moderate risk prostate cancer or a high risk prostate cancer or to predict whether the prostate cancer is progressing or regressing or in remission. The gene expression pattern can also be used as a prognostic indictor of disease-free survival following radical 20 prostatectomy. In a particular embodiment, gene expression patterns are derived from the expression level of the ERG gene, alone or in combination with other prostate cancer-cell specific genes (identified in Tables 1-6), including AMACR and LTF, or DD3. [022] Kits for detecting cancer, particularly prostate cancer, are also disclosed. These kits comprise a nucleic acid probe, such as the ones described herein, that hybridizes to a 25 prostate cancer-cell-specific gene. In one embodiment the nucleic acid probe hybridizes to SEQ ID NO:l (ERG) or the complement thereof under defined hybridization conditions (preferably under high stringency hybridization conditions, e.g., hybridization for 48 hours at 65'C in 6X SSC followed by a wash in 0.1 X SSX at 50'C for 45 minutes) and includes SEQ ID NO: I, itself, or a fragment of SEQ ID NO:1 having at least about 15, at least about 20, or at least about 50 30 contiguous nucleotides of SEQ ID NO: I or a sequence complementary thereto. In a particular 7 3654942_1 (GHMatters) P62730.AU 1 4-Sep-12 embodiment, the probe selectively hybridizes to the ERGI and ERG2 isoforms but not to ERG isoforms 3-9. In another embodiment, the probe selectively hybridizes to the ERGI isoform but not to ERG isoforms 2-9. The nucleic acid probe may be optionally fixed to a solid support. [023] The kit may also contain at least one additional nucleic acid probe that hybridizes 5 (preferably high stringency hybridization conditions, e.g., hybridization for 48 hours at 65*C in 6X SSC followed by a wash in 0. 1X SSX at 50'C for 45 minutes) to DD3 (SEQ ID NO:4) or a gene identified in Tables 1-6, including for example, AMACR (SEQ ID NO:3) or LTF (SEQ ID NO:5). In one embodiment, the kit comprises a first oligonucloetide probe capable of hybridizing to SEQ ID NO: I (ERG) or a sequence complimentary thereto under conditions of L0 high stringency and at least one other oligonucleotide probe capable of hybridizing to SEQ ID NO:3 (AMACR) or a sequence complimentary thereto, or to SEQ ID NO:4 (DD3) or a sequence complementary thereto, or to a gene identified in Tables 1-6 under conditions of high stringency. In a related embodiment, the kit having an ERG and AMACR probe further comprises a third oligonucleotide probe capable of hybridzing to SEQ ID NO:4 (DD3) or a sequence 15 complementary thereto. The kits described herein may optionally contain an oligonucleotide probe capable of hybridizing to SEQ ID NO:5 (LTF) or a sequence complementary thereto under conditions of high stringency. [024] The kits may further comprise a first oligonucteotide primer and a second oligonucleotide primer, where the first oligonucleotide primer contains a sequence that 0 hybridizes to a first sequence in SEQ ID NO:1, and the second oligonucleotide primer contains a sequence that hybridizes to a second sequence in a nucleic acid strand complementary to SEQ ID NO: 1, wherein the first sequence does not overlap with the second sequence. The first and second oligonucleotide primers are capable of amplifying a target sequence of interest in SEQ ID NO:1. Similarly, the kits can further comprise first and second oligonucleotide primers derived 25 from DD3 (SEQ ID NO:4) or a prostate cancer-cell-specific gene, including, for example AMACR (SEQ ID NO:3) or LTF (SEQ ID NO:5). [025] Disclosed herein are therapeutic methods of treating cancer, in particular prostate cancer. [026] Also disclosed are screening methods for identifying compounds that modulate 30 expression of a CaP-cell-specific gene, such as ERG, in prostate cancer cells. 8 3854942.1 (GHMaUers) P62730 AUA 4-Sep-12 [027] The present invention is based in part on the identification of gene expression signatures that correlate with a high risk of CaP progression. Over expression or under expression of specific genes are predictive of tumor progression. Also disclosed are genes, such as the ERG gene, and analogs of specific genes that can be used alone or in combination with 5 DD3 or other CaP-cell-specific genes, such as AMACR or LTF, to function as diagnostic and prognostic targets for cancer, particularly prostate tumors. Also disclosed are genes, such as the ERG gene, and analogs of specific genes that can be used alone or in combination as therapeutic targets for cancer, in particular prostate tumors. [028] Also disclosed are diagnostic kits comprised of an anti-CaP-cell-specific gene 10 antibody, for example, an anti-ERG gene antibody, which is optionally, detectably labeled. A kit is also provided that comprises nucleic acid primer sequences and/or a nucleic acid probe capable of hybridizing under defined conditions (preferably high stringency hybridization conditions, e.g., hybridization for 48 hours at 65*C in 6X SSC followed by a wash in 0.1X SSX at 50'C for 45 minutes) to an ERG nucleic acid. The kits may also contain an anti-DD3 gene 15 antibody or a second anti-CaP-cell-specific gene antibody, such as an anti-AMACR or anti-LTF gene antibody, or a second set of nucleic acid primer sequences and/or a nucleic acid probe capable of hybridizing under defined conditions to the DD3 gene or another CaP-cell-specific gene, such as the AMACR or LTF gene. [029] The disclosed CaP-cell-specific genes, such as ERG, can be used alone or in 20 combination as biomarkers of cancer, and in particular, prostate cancers and other related diseases, as targets for therapeutic intervention, or as gene therapy agents. [030] Also disclosed are treatment of disorders of hyperproliferation (e.g., cancer, benign tumors) by administering compounds that modulate expression of the specific genes. [031] Methods of screening cancer cells, and in particular) prostate cancer cells, for 25 specific gene expression signatures, including ERG gene expression signatures, alone or in combination with DD3 gene expression signatures or other CaP-cell-specific gene expression signatures, such as AMACR or LTF, are disclosed. [032] Additional features of the invention will be set forth in part in the description following, and in part will be understood from the description, or may be learned by practice of 30 the invention. 9 3654942_1 (GHMaters) P02730AU.1 4-Sep-12 [033] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention as claimed. 5 BRIEF DESCRIPTION OF THE FIGURES [034] Figure 1: Relative expression level of ERG (A), AMACR (B), GSTPI (C), and LTF (D) in matched tumor and benign prostate epithelial cells analyzed by QRT-PCR (TaqMan) X-axis: CaP patients analyzed (1-20); Y-axis: Expression ratio between tumor versus benign laser capture microdissection (LCM) sample pairs. 10 [035] Figure 2: Identification of genes by a distance based MDS and weighted analysis that discriminates between cancerous and benign tissue. (A) Two-dimensional MDS plot elucidating discrimination of 18 tumor samples and 18 benign samples. (B) Hierarchical clustering dendrogram with two major clusters of 18 tumor samples in the right cluster and 18 benign samples in the left cluster. 15 [036] Figure 3: A distance based MDS and weighted gene analysis using the tumor over benign ratio (or fold change) data for the identification of genes that can 9a 3654942_1 (GHManers)P62730AU.1 4-Sep-12 discriminate between high risk CaP and moderate risk CaP. (A) A supervised MDS analysis of 18 samples (9 samples from high risk group and 9 samples from moderate risk group) that ranks the genes according to their impact on minimizing cluster volume and maximizing center-to-center inter cluster distance. (B) Hierarchical clustering of the first 5 55 genes of the top 200 obtained by the MDS analysis. Genes and samples are arranged as ordered by cluster and treeview. Expression of each gene in each sample is obtained by the tumor over benign ratio or fold change (T/N). Dendrogram at the top of the cluster shows two major clusters, 9 samples of the MR groups in the right cluster and 9 samples of the HR groups in the left cluster. (C) Two-dimensional MDS plot of 18 CaP tumor 10 epithelia that shows the differentiation between the high risk group (9 tumor epithelia) and moderate risk group (9 tumor epithelia) on the basis of the impact of the rank of the genes that discriminate between the HR and MR groups. (D) Hierarchical clustering dendrogram with two major clusters of 9 samples of the MR groups in the left cluster and 8+1 samples of the HR groups in the right cluster. (E) Two-dimensional MDS plot of 18 15 CaP benign epithelia that shows the discrimination between the high risk group (9 benign epithelia) and moderate risk group (9 benign epithelia) samples. [037] Figure 4: In silico validation: the discriminatory potential of the genes that we obtained from our supervised MDS analysis on two independent data sets (Welsh et al. 2001, Singh et al. 2002). Two-dimensional MDS plot that shows the discrimination 20 between 7 tumor epithelia of the high risk group and 7 tumor epithelia of the moderate risk group using data from Welsh et al. (A), as well as discrimination between 4 tumor epithelia of the high risk group and 5 tumor epithelia of the moderate risk group using data from Singh et al. (B). [038] Figure 5: Combined gene expression analysis of ERG, AMACR and DD3 25 in tumor and benign prostate epithelial cells of 55 CaP patients. The graphs represent patient distribution by tumor versus benign gene expression ratios according to five gene expression categories: 1) "Up:" greater than 2 fold over expression in tumor compared to benign; 2) "Down:" less than 0.5 fold under expression in tumor compared to benign; 3) "Same:" no significant difference (0.5 to 2 fold); 4) "No expr.:" no detectable gene 30 expression; and 5) "Other:" collectively defines patients with expression category 2, 3 and 4 for the indicated genes (i.e., other than category 1). (A) ERG Expression. (B) 10 AMACR Expression. (C) DD3 Expression. (D) ERG or AMACR Expression. (E) ERG or DD3 Expression. (F) ERG, AMACR, or DD3 Expression. [039] Figure 6: Map of ERG 1 and ERG2 isoforms with probe and primer locations. The numbered boxes represent exons, the darker boxes after exon 16 are the 3' non-coding exon 5 regions. Translational start and stop codons are indicated by star and pound signs, respectively. The locations of the Affymetrix probe set (213541_s_at), the TaqMan probes, the traditional RT PCR primers, and the in situ hybridization probe are indicated. [040] Figure 7: Correlation of ERG] expression and PSA recurrence-free survival. Kaplan-Meier analysis of correlation with post-prostatectomy PSA recurrence-free survival was 10 performed on 95 CaP patients having detectable levels of ERGI mRNA by real time QRT-PCR (TaqMan). Kaplan-Meier survival curves were stratified by the following ERG] expression categories: 1) greater than 100 fold over expression; 2) 2-100 fold over expression; and 3) less than 2 fold over expression or under expression of ERGI in the prostate tumor cells. The p value was 0.0006. 15 [041] Figure 8: In situ hybridization images in 7 CaP patients were analysed by the Open-Lab image analysis software (Improvisation, Lexington, MA) coupled to a microscope via a cooled digital camera (Leica Microsystems, Heidelburg, Germany). Density (OD) values for tumor (dark columns) and benign (light columns) epithelium are shown on the Y axis, and patients 1-7 are shown in the X axis. Patient No. 7 was added as a control with no significant 20 ERGI expression difference between tumor and benign cells by QRT-PCR (TaqMan). Statistical analysis was performed with the SPSS software package. [042] Figure 9: ERGI is represented as a modular structure. The two conserved regions namely SAM-PNT Domain (Protein/RNA interaction domain) and ETS Domain (Interaction with DNA) are shaded. 25 DETAILED DESCRIPTION OF THE INVENTION Definitions [042a] In the claims which follow and in the description of the invention, except where the context requires otherwise due to express language or necessary implication, the word 30 "comprise" or variations such as "comprises" or "comprising" is used in an inclusive sense, i.e. 11 3654942_1 (GHMatters) P62730 AU.1 4-Sep-12 to specify the presence of the stated features but not to preclude the presence or addition of further features in various embodiments of the invention. [043] The term "CaP-cell-specific gene", or "prostate cancer-cell-specific gene", refers to a gene identified in Tables 1-6. The definition further encompasses CaP-cell-specific gene 5 analogs, e.g., orthologues and homologues, and functionally equivalent a 3654942_1 (GHMtter) P62730AU1 4-Sep-12 fragments of CaP-cell-specific genes or their analogs, the expression of which is either upregulated or downregulated in prostate cancer cells. [044] The term "CaP-cell-specific gene expression signature" refers to the pattern of upregulation or downregulation of product expression as measured by the Affymetrix 5 GeneChip assay described in Example 1, the QRT-PCR assay described in Example 2, or any other quantitative expression assay known in the art. [045] The term "ERG" refers to the ERG gene or ERG cDNA or mRNA described herein, and includes ERG isoforms, such as ERGI and ERG2. The cDNA sequence of the ERG] gene is published in GenBank under the accession number 10 M21535. The cDNA sequence of the ERG2 gene is published in GenBank under the accession number M17254. [046] The term "AMACR" refers to the AMACR gene or AMACR cDNA or mRNA described herein, and includes AMACR isoforms. The cDNA sequence of the AMACR gene is published in GenBank under the accession number NM_014324. 15 [047] The term "DD3" refers to the DD3 gene or DD3 cDNA or mRNA described herein, and includes DD3 isoforms. The cDNA sequence of the DD3 gene is published in GenBank under the accession number AF 103907 and is also disclosed in WO 98/45420 (1998). Although DD3 was originally used to describe a fragment of exon 4 of the prostate cancer antigen 3 (PCA3) gene, the term, as used in herein, is not so 20 limited. DD3 is intended to refer to the entire DD3 gene or cDNA or mRNA, which in the art is also commonly referred to as PCA3. [048] The term "LTF" refers to the LTF gene or LTF cDNA or mRNA described herein and includes LTF isoforms. The cDNA sequence of the LTF gene is published in GenBank under the accession number NM_002343. 25 [049] The term "polypeptide" is used interchangeably with the terms "peptide" and "protein" and refers to any chain of amino acids, regardless of length or posttranslational modification (e.g., glycosylation or phosphorylation), or source (e.g., species). [050] The phrase "substantially identical," or "substantially as set out," means 30 that a relevant sequence is at least 70%, 75%, 80%, 85%. 90%, 95%, 97, 98, or 99% 12 identical to a given sequence. By way of example, such sequences may be allelic variants, sequences derived from various species, or they may be derived from the given sequence by truncation, deletion, amino acid substitution or addition. For polypeptides, the length of comparison sequences will generally be at least 20, 30, 50, 100 or more amino acids. 5 For nucleic acids, the length of comparison sequences will generally be at least 50, 100, 150, 300, or more nucleotides. Percent identity between two sequences is determined by standard alignment algorithms such as, for example, Basic Local Alignment Tool (BLAST) described in Altschul et al. (1990) J. Mol. Biol., 215:403-410, the algorithm of Needleman et al. (1970) J. Mol. Biol., 48:444-453, or the algorithm of Meyers et al. 10 (1988) Comput. Apple. Biosci., 4:11-17. [051] The terms "specific interaction," "specific binding," or the like, mean that two molecules form a complex that is relatively stable under physiologic conditions. The term is also applicable where, e.g., an antigen-binding domain is specific for a particular epitope, which is carried by a number of antigens, in which case the specific binding 15 member carrying the antigen-binding domain will be able to bind to the various antigens carrying the epitope. Specific binding is characterized by a high affinity and a low to moderate capacity. Nonspecific binding usually has a low affinity with a moderate to high capacity. Typically, the binding is considered specific when the affinity constant K, is higher than 106 M', more preferably higher than 107 M-, and most preferably 108 M-. If 20 necessary, non-specific binding can be reduced without substantially affecting specific binding by varying the binding conditions. Such conditions are known in the art, and a skilled artisan using routine techniques can select appropriate conditions. The conditions are usually defined in terms of concentration of antibodies, ionic strength of the solution, temperature, time allowed for binding, concentration of non-related molecules (e.g., serum 25 albumin, milk casein), etc. The term "detectably labeled" refers to any means for marking and identifying the presence of a molecule, e.g., an oligonucleotide probe or primer, a gene or fragment thereof, or a cDNA molecule. Methods for labeling a molecule are well known in the art and include, without limitation, radioactive labeling (e.g., with an isotope such as 32 P, 3S, or 1251) and nonradioactive labeling (e.g., fluorescent and 30 chemiluminescent labeling). [052] The term "modulatory compound" is used interchangeably with the term "therapeutic" as used herein means any compound capable of "modulating" either CaP 13 cell-specific gene expression at the transcriptional, translational, or post-translational levels or modulating the biological activity of a CaP-cell-specific polypeptide. The term "modulate" and its cognates refer to the capability of a compound acting as either an agonist or an antagonist of a certain reaction or activity. The term modulate, therefore, 5 encompasses the terms "activate" and "inhibit." The term "activate," for example, refers to an increase in the expression of the CaP-cell-specific gene or activity of a CaP-cell specific polypeptide in the presence of a modulatory compound, relative to the activity of the gene or the polypeptide in the absence of the same compound. The increase in the expression level or the activity is preferably at least about 10%, 20%, 30%, 40%, 50%, 10 60%, 70%, 80%, 90%, or higher. Analogously, the term "inhibit" refers to a decrease in the expression of the CaP-cell-specific gene or activity of a CaP-cell-specific polypeptide in the presence of a modulatory compound, relative to the activity of the gene or the polypeptide in the absence of the same compound. The decrease in the expression level or the activity is preferably at least about 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 15 or higher. The expression level of the CaP-cell-specific gene or activity of a CaP-cell specific polypeptide can be measured as described herein or by techniques generally known in the art. [053) The term "treatment" is used interchangeably herein with the term "therapeutic method" and refers to both therapeutic treatment and 20 prophylactic/preventative measures. Those in need of treatment may include individuals already having a particular medical disorder as well as those who may ultimately acquire the disorder. [054] The term "isolated" refers to a molecule that is substantially free of its natural environment. Any amount of that molecule elevated over the naturally occurring 25 levels due to any manipulation, e.g., over expression, partial purification, etc., is encompassed with the definition. With regard to partially purified compositions only, the term refers to an isolated compound that is at least 50-70%, 70-90%, 90-95% (w/w), or more pure. [055] The term "effective dose," or "effective amount," refers to that amount of 30 the compound that results in amelioration of symptoms in a patient or a desired biological outcome, e.g., inhibition of cell proliferation. The effective amount can be determined as described in the subsequent sections. 14 [056] The terms "polynucleotide," "oligonucleotide," "nucleic acid," and "DNA" are used interchangeably herein and refer to deoxyribonucleic acid (DNA), and, where appropriate, ribonucleic acid (RNA). The term should also be understood to include nucleotide analogs, and single or double stranded polynucleotides. Examples of 5 polynucleotides include, but are not limited to, plasmid DNA or fragments thereof, viral DNA or RNA, anti-sense RNA, etc. The term plasmidd DNA" refers to double stranded DNA that is circular. [057] As used herein the term "hybridization under defined conditions," or "hybridizing under defined conditions," is intended to describe conditions for 10 hybridization and washes under which nucleotide sequences that are significantly identical or homologous to each other remain bound to each other. The conditions are such that sequences, which are at least about 6 and more preferably at least about 20, 50, 100, 150, 300, or more nucleotides long and at least about 70%, more preferably at least about 80%, even more preferably at least about 85-90% identical, remain bound to each other. The 15 percent identity can be determined as described in Altschul et al. (1997) Nucleic Acids Res., 25: 3389-3402. [058] Appropriate hybridization conditions can be selected by those skilled in the art with minimal experimentation as exemplified in Ausubel et al. (2004), Current Protocols in Molecular Biology, John Wiley & Sons. Additionally, stringent conditions 20 are described in Sambrook et al. (2001) Molecular Cloning: A Laboratory Manual, 3 'd ., Cold Spring Harbor Laboratory Press. A nonlimiting example of defined conditions of low stringency is as follows. Filters containing DNA are pretreated for 6 hours at 40*C in a solution containing 35% formamide, 5x SSC, 50 mM Tris-HCl (pH 7.5), 5 mM EDTA, 0.1% PVP, 0.1% Ficoll, 1% BSA, and 500 pg/ml denatured salmon sperm DNA. 25 Hybridizations are carried out in the same solution with the following modifications: 0.02% PVP, 0.02% Ficoll, 0.2% BSA, 100 gg/ml salmon sperm DNA, 10% (wt/vol) dextran sulfate, and 5-20 x 106 cpm 32 P-labeled probe is used. Filters are incubated in hybridization mixture for 18-20 hours at 40*C, and then washed for 1.5 hours at 55*C in a solution containing 2x SSC, 25 mM Tris-HCl (pH 7.4), 5 mM EDTA, and 0.1% SDS. The 30 wash solution is replaced with fresh solution and incubated an additional 1.5 hours at 60*C. Filters are blotted dry and exposed for autoradiography. Other conditions of low 15 stringency well known in the art may be used (e.g., as employed for cross-species hybridizations). [059] A non-limiting example of defined conditions of high stringency is as follows. Prehybridization of filters containing DNA is carried out for 8 hours to overnight 5 at 65*C in buffer composed of 6x SSC, 50 mM Tris-HCI (pH 7.5), 1 mM EDTA, 0.02% PVP, 0.02% Ficoll, 0.02% BSA, and 500 jig/ml denatured salmon sperm DNA. Filters are hybridized for 48 hours at 65"C in the prehybridization mixture containing 100 ig /ml denatured salmon sperm DNA and 5-20 x 106 cpm of 32 P-labeled probe. Washing of filters is done at 37"C for 1 hour in a solution containing 2x SSC, 0.01% PVP, 0.01% 10 Ficoll, and 0.01% BSA. This is followed by a wash in 0.1x SSC at 50*C for 45 minutes. Other conditions of high stringency well known in the art may be used. An oligonucleotide hybridizes specifically to a target sequence under high stringency conditions. [060] The term "solid support" means a material that is essentially insoluble 15 under the solvent and temperature conditions of the assay method, comprising free chemical groups available for joining an oligonucleotide or nucleic acid. Preferably, the solid support is covalently coupled to an oligonucleotide designed to directly or indirectly bind a target nucleic acid. When the target nucleic acid is an mRNA, the oligonucleotide attached to the solid support is preferably a poly-T sequence. A preferred solid support is 20 a particle, such as a micron- or submicron-sized bead or sphere. A variety of solid support materials are contemplated, such as, for example, silica, polyacrylate, polyacrylamide, a metal, polystyrene, latex, nitrocellulose, polypropylene, nylon or combinations thereof. More preferably, the solid support is capable of being attracted to a location by means of a magnetic field, such as a solid support having a magnetite core. Particularly preferred 25 supports are monodisperse magnetic spheres (i.e., uniform size.+-.about 5%). [061] The term "detecting" or "detection" means any of a variety of methods for determining the presence of a nucleic acid, such as, for example, hybridizing a labeled probe to a portion of the nucleic acid. A labeled probe is an oligonucleotide that specifically binds to another sequence and contains a detectable group which may be, for 30 example, a fluorescent moiety, a chemiluminescent moiety (such as an acridinium ester (AE) moiety that can be detected chemiluminescently under appropriate conditions (as described in U.S. Pat. No. 5,283,174)), a radioisotope, biotin, avidin, enzyme, enzyme 16 substrate, or other reactive group. Other well know detection techniques include, for example, gel filtration, gel electrophoresis and visualization of the amplicons, and High Performance Liquid Chromatography (HPLC). As used throughout the specification, the term "detecting" or "detection" includes either qualitative or quantitative detection. 5 [062] The term "primer" or "oligonculeotide primer" means an oligonucleotide capable of binding to a region of a target nucleic acid or its complement and promoting nucleic acid amplification of the target nucleic acid. Generally, a primer will have a free 3' end that can be extended by a nucleic acid polymerase. Primers also generally include a base sequence capable of hybridizing via complementary base interactions either directly 10 with at least one strand of the target nucleic acid or with a strand that is complementary to the target sequence. A primer may comprise target-specific sequences and optionally other sequences that are non-complementary to the target sequence. These non complementary sequences may comprise a promoter sequence or a restriction endonuclease recognition site. 15 CaP-Cell-Specific Gene Expression Signature Identification [063] The present invention is based in part on the identification and validation of consistent CaP epithelial cell specific gene expression signatures. These gene expression signatures define patients with CaP who are at risk to develop advanced disease by identifying genes and pathways in prostate epithelial cells that differentiate between 20 aggressive and non-aggressive courses of cancer development. Two patient groups were selected, a high risk (HR) group having, for example, PSA recurrence, Gleason score 8-9, T3c stage, seminal vesicle invasion, poor tumor differentiation, and a moderate risk (MR) group having, for example, no PSA recurrence, Gleason score 6-7, T2a-T3b stage, no seminal vesicle invasion, well or moderate tumor differentiation. The two patient groups 25 were matched for known risk factors: age, race, and family history of CaP. LCM derived epithelial cells from tumor and normal prostate of the two patient groups were compared by GeneChip analyses, as described in the following Example 1. Results were validated using quantitative reverse transcriptase PCR (QRT-PCR), as described in the following Example 2. The group of genes identified and validated as having the highest association 30 with aggressive or non-aggressive CaP can be used to reliably determine the likely course of CaP progression. 17 [064] Strikingly, one of the most consistently over expressed genes in CaP cells identified in this study was the ERG (ETS related gene) oncogene, a member of the ETS transcription factor family. (Oikawa et al., Gene (2003) 303:11-34; Sharrocks, AD, Nat Rev Mol Cell Biol (2001) 2(11):827-37; Hart et al., Oncogene (1995) 10(7):1423-30; Rao 5 et al., Science (1987) 237(4815): 635-639). Two isoforms of ERG, ERGI and ERG2, are over expressed with the highest frequency. The ERGI coding sequence (with start and stop codons underlined) is publicly available through GenBank under the accession number M21535, as follows: 1 gaattccctc caaagcaaga caaatgactc acagagaaaa aagatggcag aaccaagggc 10 61 aactaaagcc gtcaggttct gaacagctgg tagatgggct ggcttactga aggacatgat 121 tcagactgtc ccggacccag cagctcatat caaggaactc tcctgatgaa tgcagtgtgg 181 ccaaaggcgg gaagatggtg ggcagcccag acaccgttgg gatgaactac ggcagctaca 241 tggaggagaa gcacatgcca cccccaaaca tgaccacgaa cgagcgcaga gttatcgtgc 301 cagcagatcc tacgctatgg agtacagacc atgtgcggca gtggctggag tgggcggtga 15 361 aagaatatgg ccttccagac gtcaacatct tgttattcca gaacatcgat gggaaggaac 421 tgtgcaagat gaccaaggac gacttccaga ggctcacccc cagctacaac gccgacatcc 481 ttctctcaca tctccactac ctcagagaga ctcctcttcc acatttgact tcagatgatg 541 ttgataaagc cttacaaaac tctccacggt taatgcatgc tagaaacaca gatttaccat 601 atgagccccc caggagatca gcctggaccg gtcacggcca ccccacgccc cagtc9aaag 20 661 ctgctcaacc atctccttcc acagtgccca aaactgaaga ccagcgtcct cagttagatc 721 cttatcagat tcttggacca acaagtagcc gccttgcaaa tccaggcagt ggccagatcc 781 agctttggca gttcctcctg gagctcctgt cggacagctc caactccagc tgcatcacct 841 gggaaggcac caacggggag ttcaagatga cggatcccga cgaggtggcc cggcgctggg 901 gagagcggaa gagcaaaccc aacatgaact acgataagct cagccgcgcc ctccgttact 25 961 actatgacaa gaacatcatg accaaggtcc atgggaagcg ctacgcctac aagttcgact 1021 tccacgggat cgcccaggcc ctccagcccc accccccgga gtcatetctg tacaagtacc 1081 cctcagacct cccgtacatg ggctcctatc acgcccaccc acagaagatg aactttgtgg 1141 cgccccaccc tccagccctc cccgtgacat cttccagttt ttttgctgcc ccaaacccat 1201 actggaattc accaactggg ggtatatacc ccaacactag gctccccacc agccatatgc 30 1261 cttctcatct gggcacttac tactaaagac ctggcggagg cttttcccat cagcgtgcat 1321 tcaccagccc atcgccacaa actctatcgg agaacatgaa tcaaaagtgc ctcaagagga 1381 atgaaaaaag ctttactggg gctggggaag gaagccgggg aagagatcca aagactcttg 1441 ggagggagtt actgaagtct tactgaagtc ttactacaga aatgaggagg atgctaaaaa 1501 tgtcacgaat atggacatat catctgtgga ctgaccttgt aaaagacagt gtatgtagaa 35 1561 gcatgaagtc ttaaggacaa agtgccaaag aaagtggtct taagaaatgt ataaacttta 1621 gagtagagtt tgaatcccac taatgcaaac tgggatgaaa ctaaagcaat agaaacaaca 1681 cagttttgac ctaacatacc gtttataatg ccattttaag gaaaactacc tgtatttaaa 1741 aatagtttca tatcaaaaac aagagaaaag acacgagaga gactgtggcc catcaacaga 1801 cgttgatatg caactgcatg gcatgtgctg ttttggttga aatcaaatac attccgtttg 40 1861 atggacagct gtcagctttc tcaaactgtg aagatgaccc aaagtttcca actcctttac 1921 agtattaccg ggactatgaa ctaaaaggtg ggactgagga tgtgtataga gtgagcgtgt 1981 gattgtagac agaggggtga agaaggagga ggaagaggca gagaaggagg agaccaggct 2041 gggaaagaaa cttctcaagc aatgaagact ggactcagga catttgggga ctgtgtacaa 2101 tgagttatgg agactcgagg gttcatgcag tcagtgttat accaaaccca gtgttaggag 45 2161 aaaggacaca gcgtaatgga gaaagggaag tagtagaatt cagaaacaaa aatgcgcatc 2221 tctttctttg tttgtcaaat gaaaatttta actggaattg tctgatattt aagagaaaca 2281 ttcaggacct catcattatg tgggggcttt gttctccaca gggtcaggta agagatggcc 2341 ttcttggctg ccacaatcag aaatcacgca ggcattttgg gtaggcggcc tccagttttc 2401 ctttgagtcg cgaacgctgt gcgtttgtca gaatgaagta tacaagtcaa tgtttttccc 50 2461 cctttttata taataattat ataacttatg catttataca ctacgagttg atctcggcca 2521 gccaaagaca cacgacaaaa gagacaatcg atataatgtg gccttgaatt ttaactctgt 2581 atgcttaatg tttacaatat gaagttatta gttcttagaa tgcagaatgt atgtaataaa 2641 ataagcttgg cctagcatgg caaatcagat ttatacagga gtctgcattt gcactttttt 18 2701 tagtgactaa agttgcttaa tgaaaacatg tgctgaatgt tgtggatttt gtgttataat 2761 ttactttgtc caggaacttg tgcaagggag agccaaggaa ataggatgtt tggcacccaa 2821 atggcgtcag cctctccagg tccttcttgc ctcccctcct gtcttttatt tctagcccct 2881 tttggaacag gaaggacccc ggggtttcaa ttggagcctc catatttatg cctggaagga 5 2941 aagaggccta tgaagctggg gttgtcattg agaaettcta gttcagcacc tggtcacaaa 3001 tcacccttaa ttctgctatg attaaaatac atttgttgaa cagtgaacaa gctaccactc 3061 gtaaggcaaa ctgtattatt actggcaaat aaagcgtcat ggatagctgc aatttctcac 3121 tttaca (SEQ ID NO:1) [065) Nucleotides 195-1286 of SEQ ID NO:1 represent the coding sequence of 10 SEQIDNO:1. [066] The ERG2 coding sequence is publicly available through GenBank under the accession number M17254, as follows (with start and stop codons underlined): 1 gtccgcgcgt gtccgcgccc gcgtgtgcca gcgcgcgtgc cttggccgtg cgcgccgagc 61 cgggtcgcac taactccctc ggcgccgacg gcggcgctaa cctctcggtt attccaggat 15 121 ctttggagac ccgaggaaag ccgtgttgac caaaagcaag acaaatgact cacagagaaa 181 aaagatggca gaaccaaggg caactaaagc cgtcaggttc tgaacagctg gtagatgggc 241 tggcttactg aaggacatga ttcagactgt cccggaccca gcagctcata tcaaggaagc 301 cttatcagtt gtgagtgagg accagtcgtt gtttgagtgt gcctacggaa cgccacacct 361 ggctaagaca gagatgaccg cgtcctcctc cagcgactat ggacagactt ccaagatgag 20 421 cccacgcgtc cctcagcagg attggctgtc tcaaccccca gccagggtca ccatcaaaat 481 ggaatgtaac cctagccagg tgaatggctc aaggaactct cctgatgaat gcagtgtggc 541 caaaggcggg aagatggtgg gcagcccaga caccgttggg atgaactacg gcagctacat 601 ggaggagaag cacatgccac ccccaaacat gaccacgaac gagcgcagag ttatcgtgcc 661 agcagatcct acgctatgga gtacagacca tgtgcggcag tggctggagt gggcggtgaa 25 721 agaatatggc cttccagacg tcaacatctt gttattccag aacatcgatg ggaaggaact 781 gtgcaagatg accaaggacg acttccagag gctcaccccc agctacaacg ccgacatcct 841 tctctcacat ctccactacc tcagagagac tcctcttcca catttgactt cagatgatgt 901 tgataaagcc ttacaaaact ctccacggtt aatgcatgct agaaacacag atttaccata 961 tgagcccccc aggagatcag cctggaccgg tcacggccac cccacgcccc agtcgaaagc 30 1021 tgctcaacca tctccttcca cagtgcccaa aactgaagac cagcgtcctc agttagatcc 1081 ttatcagatt cttggaccaa caagtagccg ccttgcaaat ccaggcagtg gccagatcca 1141 gctttggcag ttcctcctgg agctcctgtc ggacagctcc aactccagct gcatcacctg 1201 ggaaggcacc aacggggagt tcaagatgac ggatcccgac gaggtggccc ggcgctgggg 1261 agagcggaag agcaaaccca acatgaacta cgataagctc agccgcgccc tccgttacta 35 1321 ctatgacaag aacatcatga ccaaggtcca tgggaagcgc tacgcctaca agttcgactt 1381 ccacgggatc gcccaggccc tccagcccca ccccccggag tcatctctgt acaagtaccc 1441 ctcagacctc ccgtacatgg gctcctatca cgcccaccca cagaagatga actttgtggc 1501 gccccaccct ccagccctcc ccgtgacatc ttccagtttt tttgctgccc caaacccata 1561 ctggaattca ccaactgggg gtatataccc caacactagg ctccccacca gccatatgcc 40 1621 ttctcatctg ggcacttact actaaagacc tggcggaggc ttttcccatc agcgtgcatt 1681 caccagccca tcgccacaaa ctctatcgga gaacatgaat caaaagtgcc tcaagaggaa 1741 tgaaaaaagc tttactgggg ctggggaagg aagccgggga agagatccaa agactcttgg 1801 gagggagtta ctgaagtctt actacagaaa tgaggaggat gctaaaaatg tcacgaatat 1861 ggacatatca tctgtggact gaccttgtaa aagacagtgt atgtagaagc atgaagtctt 45 1921 aaggacaaag tgccaaagaa agtggtctta agaaatgtat aaactttaga gtagagtttg 1981 aatcccacta atgcaaactg ggatgaaact aaagcaatag aaacaacaca gttttgacct 2041 aacataccgt ttataatgcc attttaagga aaactacctg tatttaaaaa tagtttcata 2101 tcaaaaacaa gagaaaagac acgagagaga ctgtggccca tcaacagacg ttgatatgca 2161 actgcatggc atgtgctgtt ttggttgaaa tcaaatacat tccgtttgat ggacagctgt 50 2221 cagctttctc aaactgtgaa gatgacccaa agtttccaac tcctttacag tattaccggg 2281 actatgaact aaaaggtggg actgaggatg tgtatagagt gagcgtgtga ttgtagacag 2341 aggggtgaag aaggaggagg aagaggcaga gaaggaggag accaggctgg gaaagaaact 2401 tctcaagcaa tgaagactgg actcaggaca tttggggact gtgtacaatg agttatggag 2461 actcgagggt tcatgcagtc agtgttatac caaacccagt gttaggagaa aggacacagc 55 2521 gtaatggaga aagggaagta gtagaattca gaaacaaaaa tgcgcatctc tttctttgtt 19 2581 tgtcaaatga aaattttaac tggaattgtc tgatatttaa gagaaacatt caggacctca 2641 tcattatgtg ggggctttgt tctccacagg gtcaggtaag agatggcctt cttggctgcc 2701 acaatcagaa atcacgcagg cattttgggt aggcggcctc cagttttcct ttgagtcgcg 2761 aacgctgtgc gtttgtcaga atgaagtata caagtcaatg tttttccccc tttttatata 5 2821 ataattatat aacttatgca tttatacact acgagttgat ctcggccagc caaagacaca 2881 cgacaaaaga gacaatcgat ataatgtggc cttgaatttt aactctgtat gcttaatgtt 2941 tacaatatga agttattagt tcttagaatg cagaatgtat gtaataaaat aagcttggcc 3001 tagcatggca aatcagattt atacaggagt ctgcatttgc acttttttta gtgactaaag 3061 ttgcttaatg aaaacatgtg ctgaatgttg tggattttgt gttataattt actttgtcca 10 3121 ggaacttgtg caagggagag ccaaggaaat aggatgtttg gcaccc (SEQ ID NO:2) [067] Nucleotides 257-1645 of SEQ ID NO:2 represent the coding sequence of SEQ ID NO:2. [068] Validation by QRT-PCR (TaqMan) in microdissected tumor and benign prostate epithelial cells of 20 CaP patients confirmed a consistent, tumor associated over 15 expression of ERG isoforms ERGI and/or ERG2 in 95% of patients (19 of 20) (Figure IA). As a quality test and comparison, the expression of AMACR, a recently identified CaP tissue marker (Rubin et al, JAMA (2002) 287:1662-1670; Luo et al., Cancer Res (2002) 62: 2220-2226), and of GSTP1, a gene known to have decreased expression in CaP (Nelson et al., Ann N Y Acad Sci (2001) 952: 135-144), was also determined (Figure IB 20 and IC). Robust over expression in CaP cells of 95% of the patients, similarly to ERG, was observed for AMACR, while the GSTP1 expression was significantly decreased in the tumor cells of each CaP patient, confirming the high quality of the tumor and benign LCM specimens and the reliability of the QRT-PCR. [069] Recently a detailed mapping of the chromosomal region (21q22.2-q22.3) 25 containing the ERG gene, as well as a complete exon-intron structure with 9 alternative transcripts (or isoforms) has been described. (Owczarek et al., Gene (2004) 324: 65-77). The probes on the Affymetrix GeneChip used in our initial discovery of consistent ERG over expression in CaP, as well as the TaqMan probe designed for the validation experiment, recognize a region specific to the ERG 1 and 2 isoforms only. 30 [070] Both ERG and ETS are proto-oncogenes with mitogenic and transforming activity. (Sharrocks, AD, Nat Rev Mol Cell Biol (2001) 2(11):827-37; Seth et al., Proc Nat] Acad Sci USA (1989) 86:7833-7837). Deregulation or chromosomal reorganization of ERG is linked to Ewing sarcoma, myeloid leukemia and cervical carcinoma. (DeAlva et al., Int J Surg Pathol (2001) 9: 7-17; Simpson et al., Oncogene (1997) 14: 2149-2157; 35 Shimizu et al., Proc Natl Acad Sci USA (1993) 90:10280-284; Papas, et al., Am J Med Genet Suppl. (1990) 7:251-261). ETS2 has been implicated in CaP, but it is over 20 expressed only in a small proportion of CaP specimens. (Liu et al., Prostate (1997) 30:145 53; Semenchenko, et al., Oncogene (1998) 17:2883-88). ERG over expression without amplification of DNA copy number was recently reported in acute myeloid leukemia. (Balduc et al., Proc. Natl. Acad. Sci. USA (2004) 101:3915-20). Gavrilov et al., Eur J 5 Cancer (2001) 37:1033-40 examined the expression of various transcription factors, including several proteins from the ETS family, in a very limited number of high-grade prostate cancer samples. Antibodies against the ETS family proteins, Elf-I and Fli-1, caused intense staining of most of the high-grade prostate cancer samples. In contrast, ERG protein, while being detected in the noncancerous endothelial cells (microvessels in 10 the stroma) of most samples tested, was detected in only a minority of the high-grade prostate cancers. ETS family proteins have a variety of expression patterns in human tissues. (Oikawa et al., Gene (2003) 303:11-34). ERG is expressed in endothelial tissues, hematopoietic cells, kidney, and in the urogenital tract. ERG proteins are nuclear transcription factors that form homodimers, as well as heterodimers with several other 15 members of the ETS family of transcription factors. (Carrere et al., Oncogene (1998) 16(25): 3261-68). A negative crosstalk observed between ERG and estrogen receptor (ER-alpha) may be relevant in urogenital tissues, where both transcription factors are expressed. (Vlaeminck-Guillem et al., Oncogene (2003) 22(50):8072-84). The present invention is based in part upon the surprising discovery that ERG is over expressed in the 20 majority of CaP specimens analyzed, indicating that this oncogene plays a role in prostate tumorigenesis, most likely by modulating transcription of target genes favoring tumorigenesis in prostate epithelium. [071] The present invention is further based in part upon the over expression of the AMACR gene in prostate cancer epithelium. The cDNA sequence of the AMACR is 25 publicly available through GenBank under the accession numbers NM_014324 and AF047020. The sequence (with start and stop codons underlined) corresponding to accession number NM_014324 is as follows: 1 gggattggga gggcttcttg caggctgctg ggctggggct aagggctgct cagtttcctt 61 cagcggggca ctgggaagcg ccatggcact gcagggcatc tcggtcgtgg agctgtccgg 30 121 cctggccccg ggcccgttct gtgctatggt cctggctgac ttcggggcgc gtgtggtacg 181 cgtggaccgg cccggctccc gctacgacgt gagccgcttg ggccggggca agcgctcgct 241 agtgctggac ctgaagcagc cgcggggagc cgccgtgctg cggcgtctgt gcaagcggtc 301 ggatgtgctg ctggagccct tccgccgcgg tgtcatggag aaactccagc tgggcccaga 361 gattctgcag cgggaaaatc caaggcttat ttatgccagg ctgagtggat ttggccagtc 35 421 aggaagcttc tgccggttag ctggccacga tatcaactat ttggctttgt caggtgttct 481 ctcaaaaatt ggcagaagtg gtgagaatcc gtatgccccg ctgaatctcc tggctgactt 541 tgctggtggt ggccttatgt gtgcactggg cattataatg gctctttttg accgcacacg 21 601 cactggcaag ggtcaggtca ttgatgcaaa tatggtggaa ggaacagcat atttaagttc 661 ttttctgtgg aaaactcaga aattgagtct gtgggaagca cctcgaggac agaacatgtt 721 ggatggtgga gcacctttct atacgactta caggacagca gatggggaat tcatggctgt 781 tggagcaata gaaccccagt tctacgagct gctgatcaaa ggacttggac taaagtctga 5 841 tgaacttccc aatcagatga gcatggatga ttggccagaa atgaagaaga agtttgcaga 901 tgtatttgca gagaagacga aggcagagtg gtgtcaaatc tttgacggca cagatgcctg 961 tgtgactccg gttctgactt ttgaggaggt tgttcatcat gatcacaaca aggaacgggg 1021 ctcgtttatc accagtgagg agcaggacgt gagcccccgc cctgcacctc tgctgttaaa 1081 caccccagcc atcccttctt tcaaaaggga tcctttcata ggagaacaca ctgaggagat 10 1141 acttgaagaa tttggattca gCcgCgaaga gatttatcag cttaactcag ataaaatcat 1201 tgaaagtaat aaggtaaaag ctagtctcta acttccaggc ccacggctca agtgaatttg 1261 aatactgcat ttacagtgta gagtaacaca taacattgta tgcatggaaa catggaggaa 1321 cagtattaca gtgtcctacc actctaatca agaaaagaat tacagactct gattctacag 1381 tgatgattga attctaaaaa tggttatcat tagggctttt gatttataaa actttgggta 15 1441 cttatactaa attatggtag ttattctgcc ttccagtttg cttgatatat ttgttgatat 1501 taagattctt gacttatatt ttgaatgggt tctagtgaaa aaggaatgat atattcttga 1561 agacatcgat atacatttat ttacactctt gattctacaa tgtagaaaat gaggaaatgc 1621 cacaaattgt atggtgataa aagtcacgtg aaacagagtg attggttgca tccaggcctt 1681 ttgtcttggt gttcatgatc tccctctaag cacattccaa actttagcaa cagttatcac 20 1741 actttgtaat ttgcaaagaa aagtttcacc tgtattgaat cagaatgcct tcaactgaaa 1801 aaaacatatc caaaataatg aggaaatgtg ttggctcact acgtagagtc cagagggaca 1861 gtcagtttta gggttgcctg tatccagtaa ctcggggcct gtttccccgt gggtctctgg 1921 gctgtcagct ttcctttctc catgtgtttg atttctcctc aggctggtag caagttctgg 1981 atcttatacc caacacacag caacatccag aaataaagat ctcaggaccc cccagcaagt 25 2041 cgttttgtgt ctccttggac tgagttaagt tacaagcctt tcttatacct gtctttgaca 2101 aagaagacgg gattgtcttt acataaaacc agcctgctcc tggagcttcc ctggactcaa 2161 cttcctaaag gcatgtgagg aaggggtaga ttccacaatc taatccgggt gccatcagag 2221 tagagggagt agagaatgga tgttgggtag gccatcaata aggtccattc tgcgcagtat 2281 ctcaactgcc gttcaacaat cgcaagagga aggtggagca ggtttcttca tcttacagtt 30 2341 gagaaaacag agactcagaa gggcttctta gttcatgttt cccttagcgc ctcagtgatt 2401 ttttcatggt ggcttaggcc aaaagaaata tctaaccatt caatttataa ataattaggt 2461 ccccaacgaa ttaaatatta tgtcctacca acttattagc tgcttgaaaa atataataca 2521 cataaataaa aaaa (SEQ ID NO:3) [072] Nucleotides 83-1231 of SEQ ID NO:3 represent the coding sequence of 35 AMACR. [073] The present invention is further based in part upon the over expression of the DD3 gene in prostate cancer epithelium, The cDNA sequence of the DD3 gene is publicly available through GenBank under the accession number AF103907. The sequence corresponding to accession number AF103907 is as follows: 40 1 acagaagaaa tagcaagtgc cgagaagctg gcatcagaaa aacagagggg agatttgtgt 61 ggctgcagcc gagggagacc aggaagatct gcatggtggg aaggacctga tgatacagag 121 gaattacaac acatatactt agtgtttcaa tgaacaccaa gataaataag tgaagagcta 181 gtccgctgtg agtctcctca gtgacacagg gctggatcac catcgacggc actttctgag 241 tactcagtgc agcaaagaaa gactacagac atctcaatgg caggggtgag aaataagaaa 45 301 ggctgctgac tttaccatct gaggccacac atctgctgaa atggagataa ttaacatcac 361 tagaaacagc aagatgacaa tataatgtct aagtagtgac atgtttttgc acatttccag 421 cccctttaaa tatccacaca cacaggaagc acaaaaggaa gcacagagat ccctgggaga 481 aatgcccggc cgccatcttg ggtcatcgat gagcctcgcc ctgtgcctgg tcccgcttgt 541 gagggaagga cattagaaaa tgaattgatg tgttccttaa aggatgggca ggaaaacaga 50 601 tcctgttgtg gatatttatt tgaacgggat tacagatttg aaatgaagtc acaaagtgag 661 cattaccaat gagaggaaaa cagacgagaa aatcttgatg gcttcacaag acatgcaaca 721 aacaaaatgg aatactgtga tgacatgagg cagccaagct ggggaggaga taaccacggg 781 gcagagggtc aggattctgg ccctgctgcc taaactgtgc gttcataacc aaatcatttc 22 841 atatttctaa ccctcaaaac aaagctgttg taatatctga tctctacggt tccttctggg 901 cccaacattc tccatatatc cagccacact catttttaat atttagttcc cagatctgta 961 ctgtgacctt tctacactgt agaataacat tactcatttt gttcaaagac ccttcgtgtt 1021 gctgcctaat atgtagctga ctgtttttcc taaggagtgt tctggcccag gggatctgtg 5 1081 aacaggctgg gaagcatctc aagatctttc cagggttata cttactagca cacagcatga 1141 tcattacgga gtgaattatc taatcaacat catcctcagt gtctttgccc atactgaaat 1201 tcatttccca cttttgtgcc catectcaag acctcaaaat gtcattccat taatatcaca 1261 ggattaactt ttttttttaa cctggaagaa ttcaatgtta catgcagcta tgggaattta 1321 attacatatt ttgttttcca gtgcaaagat gactaagtcc tttatccctc ccctttgttt 10 1381 gatttttttt ccagtataaa gttaaaatgc ttagccttgt actgaggctg tatacagcac 1441 agcctctccc catccctcca gccttatctg tcatcaccat caacccctcc cataccacct 1501 aaacaaaatc taacttgtaa ttccttgaac atgtcaggac atacattatt ccttctgcct 1561 gagaagctct tccttgtctc ttaaatctag aatgatgtaa agttttgaat aagttgacta 1621 tcttacttca tgcaaagaag ggacacatat gagattcatc atcacatgag acagcaaata 15 1681 ctaaaagtgt aatttgatta taagagttta gataaatata tgaaatgcaa gagccacaga 1741 gggaatgttt atggggcacg tttgtaagcc tgggatgtga agcaaaggca gggaacctca 1801 tagtatctta tataatatac ttcatttctc tatctctatc acaatatcca acaagctttt 1861 cacagaattc atgcagtgca aatccccaaa ggtaaccttt atccatttca tggtgagtgc 1921 gctttagaat tttggcaaat catactggtc acttatctca actttgagat gtgtttgtcc 20 1981 ttgtagttaa ttgaaagaaa tagggcactc ttgtgagcca ctttagggtt cactcctggc 2041 aataaagaat ttacaaagag ctactcagga ccagttgtta agagctctgt gtgtgtgtgt 2101 gtgtgtgtgt gagtgtacat gccaaagtgt gcctctctct cttgacccat tatttcagac 2161 ttaaaacaag catgttttca aatggcacta tgagctgcca atgatgtatc accaccatat 2221 ctcattattc tccagtaaat gtgataataa tgtcatctgt taacataaaa aaagtttgac 25 2281 ttcacaaaag cagctggaaa tggacaacca caatatgcat aaatctaact cctaccatca 2341 gctacacact gcttgacata tattgttaga agcacctcgc atttgtgggt tctcttaagc 2401 aaaatacttg cattaggtct cagctggggc tgtgcatcag gcggtttgag aaatattcaa 2461 ttctcagcag aagccagaat ttgaattccc tcatctttta ggaatcattt accaggtttg 2521 gagaggattc agacagctca ggtgctttca ctaatgtctc tgaacttctg tccctctttg 30 2581 tgttcatgga tagtccaata aataatgtta tctttgaact gatgctcata ggagagaata 2641 taagaactct gagtgatatc aacattaggg attcaaagaa atattagatt taagctcaca 2701 ctggtcaaaa ggaaccaaga tacaaagaac tctgagctgt catcgtcccc atctctgtga 2761 gccacaacca acagcaggac ccaacgcatg tctgagatcc ttaaatcaag gaaaccagtg 2821 tcatgagttg aattctccta ttatggatgc tagcttctgg ccatctctgg ctctcctctt 35 2881 gacacatatt agcttctagc ctttgcttcc acgactttta tcttttctcc aacacatcgc 2941 ttaccaatcc tctctctgct ctgttgcttt ggacttcccc acaagaattt caacgactct 3001 caagtctttt cttccatccc caccactaac ctgaatgcct agacccttat ttttattaat 3061 ttccaataga tgctgcctat gggctatatt gctttagatg aacattagat atttaaagct 3121 caagaggttc aaaatccaac tcattatctt ctctttcttt cacctccctg ctcctctccc 40 3181 tatattactg attgcactga acagcatggt ccccaatgta gccatgcaaa tgagaaaccc 3241 agtggctcct tgtggtacat gcatgcaaga ctgctgaagc cagaaggatg actgattacg 3301 cctcatgggt ggaggggacc actcctgggc cttcgtgatt gtcaggagca agacctgaga 3361 tgctccctgc cttcagtgtc ctctgcatct cccctttcta atgaagatcc atagaatttg 3421 ctacatttga gaattccaat taggaactca catgttttat ctgccctatc aattttttaa 45 3481 acttgctgaa aattaagttt tttcaaaatc tgtccttgta aattactttt tcttacagtg 3541 tcttggcata ctatatcaac tttgattctt tgttacaact tttcttactc ttttatcacc 3601 aaagtggctt ttattctctt tattattatt attttctttt actactatat tacgttgtta 3661 ttattttgtt ctctatagta tcaatttatt tgatttagtt tcaatttatt tttattgctg 3721 acttttaaaa taagtgattc ggggggtggg agaacagggg agggagagca ttaggacaaa 50 3781 tacctaatgc atgtgggact taaaacctag atgatgggtt gataggtgca gcaaaccact 3841 atggcacacg tatacctgtg taacaaacct acacattctg cacatgtatc ccagaacgta 3901 aagtaaaatt taaaaaaaag tga (SEQ ID NO:4) [074] The DD3 gene appears to represent a non-coding nucleic acid. Therefore, no start and stop codons have been indicated. 55 [075] The present invention is further based in part upon the under expression of the LTF gene in prostate cancer epithelium. The cDNA sequence of the lactotransferrin 23 (LTF) gene is publicly available through GenBank under the accession number NM_002343. The sequence (with start and stop codons underlined) corresponding to accession number NM_002343 is as follows: 1 agagccttcg tttgccaagt cgcctccaga ccgcagacat gaaacttgtc ttcctcgtcc 5 61 tgctgttcct cggggccctc ggactgtgtc tggctggccg taggaggagt gttcagtggt 121 gcgccgtatc ccaacccgag gccacaaaat gcttccaatg gcaaaggaat atgagaaaag 181 tgcgtggccc tcctgtcagc tgcataaaga gagactcccc catccagtgt atccaggcca 241 ttgcggaaaa cagggccgat gctgtgaccc ttgatggtgg tttcatatac gaggcaggcc 301 tggcccccta caaactgcga cctgtagcgg cggaagtcta cgggaccgaa agacagccac 10 361 gaactcacta ttatgccgtg gctgtggtga agaagggcgg cagctttcag ctgaacgaac 421 tgcaaggtct gaagtcctgc cacacaggcc ttcgcaggac cgctggatgg aatgtcccta 481 tagggacact tcgtccattc ttgaattgga cgggtccacc tgagcccatt gaggcagctg 541 tggccaggtt cttctcagcc agctgtgttc ccggtgcaga taaaggacag ttccccaacc 601 tgtgtcgcct gtgtgcgggg acaggggaaa acaaatgtgc cttctcctcc caggaaccgt 15 661 acttcagcta ctctggtgcc ttcaagtgtc tgagagacgg ggctggagac gtggctttta 721 tcagagagag cacagtgttt gaggacctgt cagacgaggc tgaaagggac gagtatgagt 781 tactctgccc agacaacact cggaagccag tggacaagtt caaagactgc catctggccc 841 gggtcccttc tcatgccgtt gtggcacgaa gtgtgaatgg caaggaggat gccatctgga 901 atcttctccg ccaggcacag gaaaagtttg gaaaggacaa gtcaccgaaa ttccagctct 20 961 ttggctcccc tagtgggcag aaagatctgc tgttcaagga ctctgccatt gggttttcga 1021 gggtgccccc gaggatagat tctgggctgt accttggctc cggctacttc actgccatcc 1081 agaacttgag gaaaagtgag gaggaagtgg ctgcccggcg tgcgcgggtc gtgtggtgtg 1141 cggtgggcga gcaggagctg cgcaagtgta accagtggag tggcttgagc gaaggcagcg 1201 tgacctgctc ctcggcctcc accacagagg actgcatcgc cctggtgctg aaaggagaag 25 1261 ctgatgccat gagtttggat ggaggatatg tgtacactgc aggcaaatgt ggtttggtgc 1321 ctgtcctggc agagaactac aaatcccaac aaagcagtga ccctgatcct aactgtgtgg 1381 atagacctgt ggaaggatat cttgctgtgg cggtggttag gagatcagac actagcctta 1441 cctggaactc tgtgaaaggC aagaagtcct gccacaccgc cgtggacagg actgcaggct 1501 ggaatatccc catgggcctg ctcttcaacc agacgggctc ctgcaaattt gatgaatatt 30 1561 tcagtcaaag ctgtgcccct gggtctgacc cgagatctaa tctctgtgct ctgtgtattg 1621 gcgacgagca gggtgagaat aagtgcgtgc ccaacagcaa cgagagatac tacggctaca 1681 ctggggcttt ccggtgcctg gctgagaatg ctggagacgt tgcatttgtg aaagatgtca 1741 ctgtcttgca gaacactgat ggaaataaca atgaggcatg ggctaaggat ttgaagctgg 1801 cagactttgc gctgctgtgc ctcgatggca aacggaagcc -tgtgactgag gctagaagct 35 1861 gccatcttgc catggecccg aatcatgccg tggtgtctcg gatggataag gtggaacgcc 1921 tgaaacaggt gttgctccac caacaggcta aatttgggag aaatggatct gactgcccgg 1981 acaagttttg cttattccag tctgaaacca aaaaccttct gttcaatgac aacactgagt 2041 gtctggccag actccatggc aaaacaacat atgaaaaata tttgggacca cagtatgtcg 2101 caggcattac taatctgaaa aagtgctcaa cctcccccct cctggaagcc tgtgaattcc 40 2161 tcaggaagta aaaccgaaga agatggccca gctccccaag aaagcctcag ccattcactg 2221 cccccagctc ttctccccag gtgtgttggg gccttggcct cccctgctga aggtggggat 2281 tgcccatcca tctgcttaca attccctgct gtcgtcttag caagaagtaa aatgagaaat 2341 tttgttgata ttctctcctt aaaaaaaaaa aaaaaaaaaa aaaaaaaaaa (SEQ ID NO:5) 45 [076] Nucleotides 39-2171 of SEQ ID NO:5 represent the coding sequence of LTF. [077] LTF is a non-heme iron binding glycoprotein and a member of the transferring gene family. Bowman et al., Adv. Genet. 25:1-38 (1988); Park et al., Proc. Natl. Acad. Sci. U.S.A., 82:3149-53 (1985). The concentration of LTF in human prostate 50 is hormone dependent and its expression is regulated by estrogen. van Sande et al., Urol. 24 Res., 9(5):241-44 (1981); Teng et al., Biochem. CellBiol., 80:7-16 (2002); Teng et al., Mol. Human Reproduction., 8, (1):58-67 (2002). LTF has also been implicated in certain cancers. For example, bovine LTF inhibits colon, esophagus, lung, and bladder carcinomas in rats. Tsuda et al., Biochem. Cell Biol., 80:131-136 (2002); Tsuda et al., 5 Biofactors., 12(1-4):83-8 (2000); Tsuda et al., Biofactors., 12(1-4):83-8 (2000); Tsuda et al., Mutat Res., 462(2-3):227-33 (2000). In a study published over 20 years ago, van Sande et al., Urol. Res. 9:241-244 (1981), examined lactoferrin protein levels in human benign prostatic hypertrophy samples. They also detected low levels of lactoferrin protein in 3 carcinoma samples. However, we are the first to report the consistent and significant 10 under expression of LTF mRNA in prostate cancer epithelial cells from a large number of patient samples. The observed under expression of LTF mRNA in such a statistically significant sample size indicates that under expression of LTF is a useful diagnostic marker for prostate cancer. [078] In one experiment, when screened using the Affymetrix GeneChip, CaP 15 tumor cells exhibited upregulated AMACR expression in comparison to matched benign cells. In this studied patient cohort (n=73), AMACR was upregulated in tumor compared to matched benign prostate epithelium in 89.04% of the patients (65 of 73), while ERG was upregulated in 78.08% (57 of 73). When these two markers were combined, we observed a 100% CaP detection rate (under the criteria that at least one marker was 20 upregulated) in the studied patient cohort (73 of 73). These data indicate that the combination of ERG and AMACR screening provides a highly accurate tool for CaP detection. [079] In another experiment, 96.4% of patients showed upregulation of either the ERG or AMACR gene in laser microdissected matched tumor and benign prostate 25 epithelial cells from 55 CaP patients (Figure 5). Similarly, 96.4% of patients showed upregulation of either the ERG or DD3 gene (Figure 5). When the expression data for the ERG, AMACR, and DD3 genes was combined, 98.2% of the CaP patients showed upregulation of at least one of the three genes in tumor cells (Figure 5). Thus, the combination of ERG, AMACR, and DD3 screening also provides a highly accurate tool 30 for CaP detection. [080) In yet another experiment, validation by QRT-PCR (TaqMan) in microdissected tumor and benign prostate epithelial cells of 20 CaP patients confirmed a 25 consistent, tumor associated under expression of LTF in 100% of patients (20 of 20) (Figure ID). Further validation studies by QRT-PCR in microdissected tumor and benign prostate epithelial cells of 103 CaP patients were consistent with the initial results, showing tumor associated under expression in 76% of patients (78 of 103). 5 Diagnostic Uses [081) In one embodiment, the present invention comprises a method of CaP diagnosis comprising screening biological samples for CaP-cell-specific gene expression signatures. In particular, the invention comprises screening for at least one of the CaP cell-specific genes listed in Tables 1-6, particularly the ERG gene, the AMACR gene, the 10 LTF gene or a combination of the ERG gene and the AMACR genes. The invention also comprises methods of diagnosing CaP comprising screening biological samples for expression of the ERG and DD3 genes, or a combination of the ERG, DD3, and AMACR genes. [082] In a further embodiment, the present invention comprises a method of CaP 15 diagnosis comprising screening biological samples for CaP-cell-specific gene expression signatures using methods known in the art, including, for example, immunohistochemistry, ELISA, in situ RNA hybridization, and any oligonucleitde amplification procedure known or later developed, including PCR (including QRT-PCR), transcription-mediated amplification (TMA), nucleic acid sequence-based amplification 20 (NASBA), self-sustained sequence replication (3SR), ligase chain reaction (LCR), strand displacement amplification (SDA), and Loop-Mediated Isothermal Amplification (LAMP). See, e.g., Mullis, U.S. Patent No. 4,683,202; Erlich et al., U.S. Patent No. 6,197,563; Walker et al., Nucleic Acids Res., 20:1691-1696 (1992); Fahy et al., PCR Methods and Applications, 1:25-33 (1991); Kacian et al., U.S. Patent No. 5,399,491; 25 Kacian et al., U.S. Patent No. 5,480,784; Davey et al., U.S. Patent No. 5,554,517; Birkenmeyer et al., U.S. Patent No. 5,427,930; Marshall et al., U.S. Patent No. 5,686,272; Walker, U.S. Patent No. 5,712,124; Notomi et al., European Patent Application No. 1 020 534 Al; Dattagupta et al., U.S. Patent No. 6,214,587; and HELEN H. LEE ET AL., NUCLEIC ACID AMPLIFICATION TECHNOLOGIES: APPLICATION TO DISEASE DIAGNOSIS (1997). Each 30 of the foregoing amplification references is hereby incorporated by reference herein. In particular, the invention comprises generating antibodies to CaP-cell-specific genes, 26 including ERG, AMACR, LTF, and DD3 for use in a immunohistochemistry assay. Other known diagnostic assays may be used to detect gene expression. [083] In a specific embodiment, the present invention comprises a method of diagnosing CaP comprising screening biological samples for expression of the ERG and 5 AMACR genes, the ERG and DD3 genes, or the ERG, AMACR, and DD3 genes, or the LTF gene using methods known in the art, including, for example, immunohistochemistry, ELISA, in situ hybridization, PCR (including QRT-PCR), transcription-mediated amplification (TMA), nucleic acid sequence-based amplification (NASBA), self-sustained sequence replication (3SR), ligase chain reaction (LCR), strand displacement 10 amplification (SDA), and Loop-Mediated Isothermal Amplification (LAMP). [084] ERG, LTF, or AMACR polypeptides, their fragments or other derivatives, or analogs thereof, may be used as immunogens in order to generate antibodies that specifically bind such immunogens. Such antibodies include, but are not limited to, polyclonal, monoclonal, chimeric, single chain and Fab fragments. In a specific 15 embodiment, antibodies to a human ERG, LTF or AMACR protein are produced. Antibodies can then be used in standard diagnostic assays to detect the protein produced by the desired gene. [085] Various procedures known in the art may be used for the production of polyclonal antibodies to an ERG, LTF, or AMACR protein or derivative or analog. In a 20 particular embodiment, rabbit polyclonal antibodies to an epitope of a ERG, LTF, or AMACR protein can be obtained. For the production of antibody, various host animals can be immunized by injection with the native ERG, LTF, or AMACR protein, or a synthetic version, or derivative (e.g., fragment) thereof, including but not limited to rabbits, mice, rats, etc. Various adjuvants may be used to increase the immunological 25 response, depending on the host species, and including, but not limited to, Freund's (complete and incomplete), mineral gels such as aluminum hydroxide, surface active substances such as lysolecithin, pluronic polyols, polyanions, peptides, oil emulsions, keyhole limpet hemocyanins, dinitrophenol, and potentially useful human adjuvants such as BCG (bacille Calmette-Guerin) and corynebacterium parvum. 30 [086] For preparation of monoclonal antibodies directed toward a ERG, LTF, or AMACR protein sequence or analog thereof, any technique, which provides for the 27 production of antibody molecules by continuous cell lines in culture may be used. For example, the hybridoma technique originally developed by Kohler et al (1975) Nature, 256:495-497, as well as the trioma technique, the human B-cell hybridoma technique (Kozbor et al. (1983) Immunology Today, 4:72), and the EBV-hybridoma technique to 5 produce human monoclonal antibodies (Cole et al. (1985) Monoclonal Antibodies and Cancer Therapy, Alan R. Liss, Inc., pp. 77-96). According to the invention, human antibodies may be used and can be obtained by using human hybridomas (Cote et al. (1983) Proc. Nat]. Acad. Sci. U.S.A., 80:2026-2030) or by transforming human B cells with EBV virus in vitro (Cole et al. (1985) Monoclonal Antibodies and Cancer Therapy, 10 Alan R. Liss, pp. 77-96). According to the invention, techniques developed for the production of chimeric antibodies (Morrison et al. (1984) Proc. Natl. Acad. Sci. U.S.A., 81:6851-6855; Neuberger et al. (1984) Nature, 312:604-608; Takeda et al. (1985) Nature, 314:452-454) by splicing the genes from a mouse antibody molecule specific for ERG, LTF, or AMACR together with genes from a human antibody molecule of appropriate 15 biological activity can be used; such antibodies are within the scope of this invention. [087] Techniques described for the production of single chain antibodies (U.S. Patent No. 4,946,778) can be used to produce ERG-, LTF-, or AMACR-specific single chain antibodies. An additional embodiment of the invention utilizes the techniques described for the construction of Fab expression libraries (Huse et al. (1989) Science, 20 246:1275-1281) to allow rapid and easy identification of monoclonal Fab fragments with the desired specificity for ERG, LTF or AMACR proteins, derivatives, or analogs. [088] Antibody fragments which contain the idiotype of the molecule can be generated by known techniques. For example, such fragments include but are not limited to: the F(ab')2 fragment which can be produced by pepsin digestion of the antibody 25 molecule; the Fab' fragments which can be generated by reducing the disulfide bridges of the F(ab') 2 fragment, the Fab fragments which can be generated by treating the antibody molecule with papain and a reducing agent, and Fv fragments, including single chain Fv (scFv) fragments. [089] In the production of antibodies, screening for the desired antibody can be 30 accomplished by techniques known in the art, e.g., ELISA. For example, to select antibodies that recognize a specific domain of a ERG, LTF, or AMACR protein, one may 28 assay generated hybridomas for a product which binds to a ERG, LTF, or AMACR fragment containing such domain. [090] A second aspect of the invention provides for use of the expression profiles resulting from these methods in diagnostic methods, including, but not limited to, 5 characterizing the treatment response to any therapy, correlating expression profiles with clinico-pathologic features, distinguishing indolent prostate cancers from those with a more aggressive phenotype (e.g. moderate risk versus high risk), analyzing tumor specimens of patients treated by radical prostate surgery to help define prognosis, screening candidate genes for the development of a'polynucleotide array for use as a blood 10 test for improved prostate cancer detection, and identifying further genes that may serve as biomarkers for response to treatment to screen drugs for the treatment of advanced prostate cancer, [091] As will be readily appreciated by persons having skill in the art, the ERG, LTF, DD3, and/or the AMACR nucleic acid sequences described herein can easily be 15 synthesized directly on a support, or pre-synthesized polynucleotide probes may be affixed to a support as described, for example, in U.S. Patent Nos. 5,744,305, 5,837,832, and 5,861,242, each of which is incorporated herein by reference. [092] Such arrays may be used to detect specific nucleic acid sequences contained in a target cell or sample, as described in U.S. Patent Nos. 5,744,305, 5,837,832, 20 and 5,861,242, each of which is incorporated herein by reference. More specifically, in the present invention, these arrays may be used in methods for the diagnosis or prognosis of prostate cancer, such as by assessing the expression profiles of genes, in biological samples. In a preferred embodiment, computer models may be developed for the analysis of expression profiles. Moreover, such polynucleotide arrays are useful in methods to 25 screen drugs for the treatment of advanced prostate cancer. In these screening methods, the polynucleotide arrays are used to analyze how drugs affect the expression of the ERG, LTF, AMACR, and/or DD3 genes. Therapeutic Uses [093] The invention provides for treatment or prevention of various diseases and 30 disorders by administration of a therapeutic compound (termed herein "therapeutic"). "Therapeutics" include but are not limited to: ERG or LTF proteins and analogs and 29 derivatives (including fragments) thereof (e.g., as described herein above); nucleic acids encoding the ERG or LTF proteins, analogs, or derivatives; ERG or LTF antisense nucleic acids, ERG or LTF dominant negative mutants, siRNA against ERG or LTF, ERG or LTF antibodies and ERG or LTF agonists and antagonists. ERG or LTF agonists and 5 antagonists, including small molecules, can be identified using the methods disclosed in this application or any standard screening assay to identify agents that modulate ERG or LTF expression or function, particularly in prostate cancer cells. For example, ERG or LTF expression or function can be readily detected, e.g., by obtaining a biological sample from a patient, e.g., a tissue sample (e.g., from biopsy tissue), a blood sample, or a urine 10 sample, and assaying it in vitro for mRNA or protein levels, structure and/or activity of the expressed ERG or LTF mRNA or protein. Many methods standard in the art can be employed, including but not limited to, kinase assays, immunoassays to detect and/or visualize ERG or LTF protein (e.g., Western blot, immunoprecipitation followed by SDS PAGE, immunocytochemistry, etc.) and/or hybridization assays to detect ERG or LTF 15 expression by detecting and/or visualizing ERG or LTF mRNA (e.g., Northern assays, dot blots, in situ hybridization, PCR (including RT-PCR), TMA, NASAB, 3SR, LCR, SDA, LAMP, etc.). Hyperproliferative Disorders [094] Disorders involving hyperproliferation of cells are treated or prevented by 20 administration of a therapeutic that antagonizes (reduces or inhibits) ERG function or expression or enhances LTF function or expression. In certain embodiments, ERG function is inhibited by use of ERG antisense nucleic acids. The present invention provides the therapeutic or prophylactic use of nucleic acids of at least 10, 15, 100, 200, 500, 1000, 1500, 2000, or 2500 contiguous nucleotides in antisense to any of the ERG 25 nucleotides described herein. In a particular embodiment, the ERG antisense nucleic acid comprises at least 10, 15, 100, 200, 500, 1000, 1500, 2000, or 2500 contiguous nucleotides in antisense orientation to the ERG nucleotide sequence. An ERG "antisense" nucleic acid as used herein refers to a nucleic acid capable of hybridizing under defined conditions to a portion of an ERG nucleic acid by virtue of some sequence complementarity. The 30 antisense nucleic acid may be complementary to a coding and/or noncoding region of an ERG nucleic acid. Such antisense nucleic acids have utility as therapeutics that inhibit 30 ERG function, and can be used in the treatment or prevention of disorders as described herein. [095] The antisense nucleic acids of the invention can be oligonucleotides that are double-stranded or single-stranded, RNA or DNA or a modification or derivative thereof, 5 which can be directly administered to a cell, or which can be produced intracellularly by transcription of exogenously, introduced coding sequences. [096] The dominant negative mutants of the invention can be produced by expression plasmids containing a nucleic acid encoding a non-functional domain of ERG, such as the DNA binding domain of ERG. These expression plasmids can be introduced 10 into a target cell or tissue and can induce tumor growth inhibition and apoptosis by acting as a dominant negative form against the wild-type ERG transcription factors influencing cell hyperproliferation (Oikawa, Cancer Sci (2004), 95:626-33). [097] RNA interference can be achieved using siRNA against the ERG gene. The siRNA is a short double stranded RNA molecule of about 18-25 nucleotides that 15 comprises a nucleotide sequence complementary to a region of the target gene. The siRNA can be introduced into a target cell or tissue, for example using an expression plasmid, where it interferes with the translation of the ERG gene. RNA interference techniques can be carried out using known methods as described, for example, in published U.S. Patent Applications 20040192626, 20040181821, and 20030148519, each 20 of which is incorporated by reference. [098] Therapeutics which are useful according to this embodiment of the invention for treatment of a disorder may be selected by testing for biological activity in promoting the survival or differentiation of cells. For example, in a specific embodiment relating to cancer therapy, including therapy of prostate cancer, a therapeutic decreases 25 proliferation of tumor cells. These effects can be measured as described in the Examples or using any other method standard in the art. [099] In specific embodiments, malignancy or dysproliferative changes (such as metaplasias and dysplasias), or hyperproliferative disorders, are treated or prevented in the prostate. 30 [0100] The therapeutics of the invention that antagonize ERG activity can also be administered to treat premalignant conditions and to prevent progression to a neoplastic or 31 malignant state, including but not limited to those disorders described herein, such as prostate cancer. Gene Therapy [0101] In a specific embodiment, nucleic acids comprising a sequence encoding an 5 ERG or LTF protein or functional derivative thereof, are administered to promote ERG or LTF function, by way of gene therapy. Alternatively, nucleic acids comprising an antisense ERG sequence are administered to antagonize ERG expression or function. Gene therapy refers to therapy performed by the administration of a nucleic acid to a subject. [0102] Any of the methods for gene therapy available in the art can be used 10 according to the present invention. For specific protocols, see Morgan (2001) Gene Therapy Protocols, 2 "d ed., Humana Press. For general reviews of the methods of gene therapy, see Goldspiel et al. (1993) Clinical Pharmacy, 12:488-505; Wu et al. (1991) Biotherapy, 3:87-95; Tolstoshev (1993) Ann. Rev. Pharmacol. Toxicol., 32:573-596; Mulligan (1993) Science, 260:926-932; and Morgan et al. (1993) Ann. Rev. Biochem., 15 62:191-217; May (1993) TIBTECH, 11(5):155-215). Methods commonly known in the art of recombinant DNA technology which can be used are described in Current Protocols in Molecular Biology (2004), Ausubel et al., eds., John Wiley & Sons, NY; and Kriegler (1990) Gene Transfer and Expression, A Laboratory Manual, Stockton Press, NY. [0103] In one embodiment, the therapeutic comprises an ERG or LTF nucleic acid 20 or antisense ERG nucleic acid that is part of a vector. In particular, such a nucleic acid has a regulatory sequence, such as a promoter, operably linked to the ERG or LTF coding region or antisense molecule, said promoter being inducible or constitutive, and, optionally, tissue-specific. In another particular embodiment, a nucleic acid molecule is used in which the ERG or LTF coding sequences and any other desired sequences are 25 flanked by regions that promote homologous recombination at a desired site in the genome, thus providing for intrachromosomal expression of the ERG or LTF nucleic acid (Koller et al. (1989) Proc. Nat]. Acad. Sci. U.S.A., 86:8932-8935; Zijlstra et al. (1989) Nature, 342:435-438). [0104] In a specific embodiment, the nucleic acid to be introduced for purposes of 30 gene therapy comprises an inducible promoter operably linked to the desired nucleic acids, 32 such that expression of the nucleic acid is controllable by the appropriate inducer of transcription. [0105] Delivery of the nucleic acid into a patient may be either direct, in which case the patient is directly exposed to the nucleic acid or nucleic acid-carrying vector, or 5 indirect, in which case, cells are first transformed with the nucleic acid in vitro, then transplanted into the patient. These two approaches are known, respectively, as in vivo or ex vivo gene therapy. [0106] In a specific embodiment, the nucleic acid is directly administered in vivo, where it is expressed to produce the encoded product. This can be accomplished by any of 10 numerous methods known in the art, e.g., by constructing it as part of an appropriate nucleic acid expression vector and administering it so that it becomes intracellular, e.g., by infection using a defective or attenuated retroviral or other viral vector (see U.S. Patent No. 4,980,286, which is incorporated herein by reference), or by direct injection of naked DNA, or by use of microparticle bombardment (e.g., a gene gun; Biolistic, DuPont), or 15 coating with lipids or cell-surface receptors or transfecting agents, encapsulation in liposomes, microparticles, or microcapsules, or by administering it in linkage to a peptide which is known to enter the nucleus, by administering it in linkage to a ligand subject to receptor-mediated endocytosis (see, e.g., Wu et al. (1987) J. Biol. Chem., 262:4429-4432). In another embodiment, a nucleic acid-ligand complex can be formed in which the ligand 20 comprises a fusogenic viral peptide to disrupt endosomes, allowing the nucleic acid to avoid lysosomal degradation. In yet another embodiment, the nucleic acid can be targeted in vivo for cell-specific uptake and expression, by targeting a specific receptor (see, e.g., PCT Pubs. WO 92/06180; WO 92/22635; W092/20316; W093/14188; WO 93/20221). Alternatively, the nucleic acid can be introduced intracellularly and incorporated within 25 host cell DNA for expression, by homologous recombination (Koller et al. (1989) Proc. Natl. Acad. Sci. U.S.A., 86:8932-8935; Zijlstra et al. (1989) Nature, 342:435-438). [0107] In a specific embodiment, a viral vector that contains an ERG or LTF nucleic acid is used. For example, a retroviral vector can be used (see, Miller et al. (1993) Meth. Enzymol., 217:581-599). These retroviral vectors have been modified to delete 30 retroviral sequences that are not necessary for packaging of the viral genome and integration into host cell DNA. The ERG or LTF nucleic acid to be used in gene therapy is cloned into the vector, which facilitates delivery of the gene into a patient. More detail 33 about retroviral vectors can be found in Boesen et al. (1994) Biotherapy, 6:291-302, which describes the use of a retroviral vector to deliver the MDRL gene to hematopoietic stem cells in order to make the stem cells more resistant to chemotherapy. Other references illustrating the use of retroviral vectors in gene therapy are: Clowes et al. (1994) J. Clin. 5 Invest., 93:644-651; Kiem et al. (1994) Blood, 83:1467-1473; Salmons et al. (1993) Hum. Gene Ther., 4:129-141; and Grossman et al. (1993) Curr. Opin. Gen. Devel., 3:110-114. [0108] Adenoviruses are other viral vectors that can be used in gene therapy. Adenoviruses are especially attractive vehicles for delivering genes to respiratory epithelia. Adenoviruses naturally infect respiratory epithelia where they cause a mild 10 disease. Other targets for adenovirus-based delivery systems are liver, the central nervous system, endothelial cells, and muscle. Adenoviruses have the advantage of being capable of infecting non-dividing cells. Kozarsky et al. (1993, Curr. Opin. Gen. Devel., 3:499 503) present a review of adenovirus-based gene therapy. Bout et al. (1994, Hum. Gene Ther., 5:3-10) demonstrated the use of adenovirus vectors to transfer genes to the 15 respiratory epithelia of rhesus monkeys. Other instances of the use of adenoviruses in gene therapy can be found in Rosenfeld et al. (1991) Science, 252:431-434; Rosenfeld et al. (1992) Cell, 68:143-155; and Mastrangeli et al. (1993) J. Clin. Invest., 91:225-234. [0109] Adeno-associated virus (AAV) has also been proposed for use in gene therapy (Walsh et al. (1993) Proc. Soc. Exp. Biol. Med., 204:289-300). 20 [0110] Another approach to gene therapy involves transferring a gene to cells in tissue culture by such methods as electroporation, lipofection, calcium phosphate mediated transfection, or viral infection. Usually, the method of transfer includes the transfer of a selectable marker to the cells. The cells are then placed under selection to isolate those cells that have taken up and are expressing the transferred gene. Those cells are then 25 delivered to a patient. [0111] In this embodiment, the nucleic acid is introduced into a cell prior to administration in vivo of the resulting recombinant cell. Such introduction can be carried out by any method known in the art, including but not limited to transfection, electroporation, microinjection, infection with a viral or bacteriophage vector containing 30 the nucleic acid sequences, cell fusion, chromosome-mediated gene transfer, microcell mediated gene transfer, spheroplast fusion, etc. Numerous techniques are known in the art 34 for the introduction of foreign genes into cells (see, e.g., Loeffler et al. (1993) Meth. Enzymol., 217:599-618; Cohen et al. (1993) Meth. Enzymol., 217:618-644; Cline (1985) Pharmac. Ther., 29:69-92) and may be used in accordance with the present invention, provided that the necessary developmental and physiological functions of the recipient 5 cells are not disrupted. The technique should provide for the stable transfer of the nucleic acid to the cell, so that the nucleic acid is expressible by the cell and preferably heritable and expressible by its cell progeny. [0112] The resulting recombinant cells can be delivered to a patient by various methods known in the art. In one preferred embodiment, epithelial cells are injected, e.g., 10 subcutaneously. In another embodiment, recombinant skin cells may be applied as a skin graft onto the patient. Recombinant blood cells (e.g., hematopoietic stem or progenitor cells) may be administered intravenously. The amount of cells envisioned for use depends on the desired effect, patient state, etc., and can be determined by one skilled in the art. [0113] Cells into which a nucleic acid can be introduced for purposes of gene 15 therapy encompass any desired, available cell type, and include, but are not limited to, epithelial cells, endothelial cells, keratinocytes, fibroblasts, muscle cells, hepatocytes, T lymphocytes, B lymphocytes, monocytes, macrophages, neutrophils, eosinophils, megakaryocytes, granulocytes; various stem or progenitor cells, in particular hematopoietic stem or progenitor cells, e.g., as obtained from bone marrow, umbilical cord 20 blood, peripheral blood, fetal liver, etc. In certain embodiments, the cells used for gene therapy are autologous to the patient. [0114] In one embodiment, an ERG or LTF nucleic acid or antisense molecule is introduced into the cells such that it is expressible by the cells or their progeny, and the recombinant cells are then administered in vivo for therapeutic effect. In a specific 25 embodiment, stem or progenitor cells are used. Any stem and/or progenitor cells which can be isolated and maintained in vitro can potentially be used in accordance with this embodiment of the present invention. Such stem cells include, but are not limited to, hematopoietic stem cells (HSC), stem cells of epithelial tissues such as the skin and the lining of the gut, embryonic heart muscle cells, liver stem cells (PCT Pub. WO 94/08598), 30 and neural stem cells (Stemple et al. (1992) Cell, 71:973-985). 35 [0115] Epithelial stem cells (ESCs) or keratinocytes can be obtained from tissues such as the skin and the lining of the gut by known procedures (Rheinwald (1980) Meth. Cell Bio., 21A:229). In stratified epithelial tissue such as the skin, renewal occurs by mitosis of stem cells within the germinal layer, the layer closest to the basal lamina. Stem 5 cells within the lining of the gut provide for a rapid renewal rate of this tissue. ESCs or keratinocytes obtained from the skin or lining of the gut of a patient or donor can be grown in tissue culture (Rheinwald (1980) Meth. Cell Bio., 21A:229; Pittelkow et al. (1986) Mayo Clinic. Proc., 61:771). If the ESCs are provided by a donor, a method for suppression of host versus graft reactivity (e.g., irradiation, drug or antibody 10 administration to promote moderate immunosuppression) can also be used. [0116] With respect to hematopoietic stem cells (HSC), any technique which provides for the isolation, propagation, and maintenance in vitro of HSC can be used in this embodiment of the invention. Techniques by which this may be accomplished include (a) the isolation and establishment of HSC cultures from bone marrow cells isolated from 15 the future host, or a donor, or (b) the use of previously established long-term HSC cultures, which may be allogeneic or xenogeneic. Non-autologous HSC may be used in conjunction with a method of suppressing transplantation immune reactions of the future host/patient. In a particular embodiment, human bone marrow cells can be obtained from the posterior iliac crest by needle aspiration (see, e.g., Kodo et al. (1984) J. Clin. Invest., 20 73:1377-1384). In one embodiment, the HSCs can be made highly enriched or in substantially pure form. This enrichment can be accomplished before, during, or after long-term culturing, and can be done by any techniques known in the art. Long-term cultures of bone marrow cells can be established and maintained by using, for example, modified Dexter cell culture techniques (Dexter et al. (1977) J. Cell Physiol., 91:335) or 25 Witlock-Witte culture techniques (Witlock et al. (1982) Proc. Natil. Acad. Sci. U.S.A., 79:3608-3612). Pharmaceutical Compositions and Administration [0117] The invention further provides pharmaceutical compositions comprising an effective amount of an ERG or LTF therapeutic, including ERG or LTF nucleic acids 30 (sense or antisense) or ERG or LTF polypeptides of the invention, in a pharmaceutically acceptable carrier, as described below. 36 [0118] Compositions comprising an effective amount of a polypeptide of the present invention, in combination with other components such as a physiologically acceptable diluent, carrier, or excipient, are provided herein. The polypeptides can be formulated according to known methods used to prepare pharmaceutically useful 5 compositions. They can be combined in admixture, either as the sole active material or with other known active materials suitable for a given indication, with pharmaceutically acceptable diluents (e.g., saline, Tris-HCl, acetate, and phosphate buffered solutions), preservatives (e.g., thimerosal, benzyl alcohol, parabens), emulsifiers, solubilizers, adjuvants and/or carriers. Suitable formulations for pharmaceutical compositions include 10 those described in Remington's Pharmaceutical Sciences, 16 'h ed., Mack Publishing Company, Easton, PA, 1980. [0119) In addition, such compositions can be complexed with polyethylene glycol (PEG), metal ions, or incorporated into polymeric compounds such as polyacetic acid, polyglycolic acid, hydrogels, dextran, etc., or incorporated into liposomes, 15 microemulsions, micelles, unilamellar or multilamellar vesicles, erythrocyte ghosts or spheroblasts. Such compositions will influence the physical state, solubility, stability, rate of in vivo release, and rate of in vivo clearance, and are thus chosen according to the intended application. [0120] The compositions of the invention can be administered in any suitable 20 manner, e.g., topically, parenterally, or by inhalation. The term "parenteral" includes injection, e.g., by subcutaneous, intravenous, or intramuscular routes, also including localized administration, e.g., at a site of disease or injury. Sustained release from implants is also contemplated. One skilled in the art will recognize that suitable dosages will vary, depending upon such factors as the nature of the disorder to be treated, the 25 patient's body weight, age, and general condition, and the route of administration. Preliminary doses can be determined according to animal tests, and the scaling of dosages for human administration is performed according to art-accepted practices. [0121] Compositions comprising nucleic acids of the invention in physiologically acceptable formulations, e.g., to be used for gene therapy are also contemplated. In one 30 embodiment, the nucleic acid can be administered in vivo to promote expression of the encoded protein, by constructing it as part of an appropriate nucleic acid expression vector and administering it so that it becomes intracellular as described in other sections herein. 37 [0122] Various delivery systems are known in the art and can be used to administer a therapeutic of the invention. Examples include, but are not limited to encapsulation in liposomes, microparticles, microcapsules, recombinant cells capable of expressing the therapeutic, receptor-mediated endocytosis (see, e.g., Wu et al. (1987) J. 5 Biol. Chem., 262:4429-4432), construction of a therapeutic nucleic acid as part of a retroviral or other vector, etc. Methods of introduction include, but are not limited to, intradermal, intramuscular, intraperitoneal, intravenous, subcutaneous, intranasal, epidural, and oral routes. The compounds may be administered by any convenient route, for example by infusion or bolus injection, by absorption through epithelial or mucocutaneous 10 linings (e.g., oral mucosa, rectal and intestinal mucosa, etc.) and may be administered together with other biologically active agents. Administration can be systemic or local. In addition, it may be desirable to introduce the pharmaceutical compositions of the invention into the central nervous system by any suitable route, including intraventricular and intrathecal injection; intraventricular injection may be facilitated by an intraventricular 15 catheter, for example, attached to a reservoir, such as an Ommaya reservoir. Pulmonary administration can also be employed, e.g., by use of an inhaler or nebulizer, and formulation with an aerosolizing agent. [0123] In a specific embodiment, it may be desirable to administer the pharmaceutical compositions of the invention locally to the area in need of treatment; this 20 may be achieved by, for example, local infusion during surgery, topical application, e.g., in conjunction with a wound dressing after surgery, by injection, by means of a catheter, a suppository, an implant, wherein the said implant is of a porous, non-porous, or gelatinous material, including membranes, such as sialastic membranes, or fibers. In one embodiment, administration can be by direct injection at the site (or former site) of a 25 malignant tumor or neoplastic or pre-neoplastic tissue. [0124] In another embodiment, the therapeutic can be delivered in a vesicle, in particular a liposome (see Langer (1990) Science, 249:1527-1533; Treat et al. (1989) in Liposomes in the Therapy of Infectious Disease and Cancer, Lopez-Berestein et al., eds., Liss, New York, pp. 353-365; Lopez-Berestein, ibid., pp. 317-327. In yet another 30 embodiment, the therapeutic can be delivered in a controlled release system. In one embodiment, a pump may be used (see Langer, supra; Sefton (1987) CRC Crit. Ref. Biomed. Eng., 14:201; Buchwald et al. (1980) Surgery, 88:507; Saudek et al. (1989) New 38 Engl. J. Med., 321:574). In another embodiment, polymeric materials can be used (see Medical Applications of Controlled Release, Langer et al., eds., CRC Pres., Boca Raton, FL, 1974; Controlled Drug Bioavailability, Drug Product Design and Performance, Smolen et al., eds., Wiley, New York, 1984; Ranger et al. (1983) J. Macromol. Sci. Rev. 5 Macromol. Chem., 23:61; see also Levy et al. (1985) Science, 228:190; During et al. (1989) Ann. Neurol., 25:351; Howard et al. (1989) J. Neurosurg., 71:105. In yet another embodiment, a controlled release system can be placed in proximity of the therapeutic target, i.e., the brain, thus requiring only a fraction of the systemic dose (see, e.g., Goodson (1984) in Medical Applications of Controlled Release, supra, vol. 2, pp. 115 10 138). Other controlled release systems are discussed in the review by Langer (1990, Science, 249:1527-1533). Diagnosis and Screening [0125] ERG, LTF, and/or AMACR proteins, analogues, derivatives, and fragments thereof, and antibodies thereto; ERG, LTF, DD3, and/or AMACR nucleic acids (and their 15 complementary and homologous sequences) and antibodies thereto, including anti-ERG, anti-DD3, anti-LTF and/or anti-AMACR antibodies, have uses in diagnostics. Such molecules can be used in assays, such as immunoassays, to detect, prognose, diagnose, or monitor various conditions, diseases, and disorders affecting ERG, LTF, DD3, and/or AMACR expression, or monitor the treatment thereof, particularly cancer, and more 20 particularly prostate cancer. In particular, such an immunoassay is carried out by a method comprising contacting a sample derived from an individual with an anti-ERG, anti-LTF, anti-DD3, and/or anti-AMACR antibody (directed against either a protein product or a nucliec acid) under conditions such that specific binding can occur, and detecting or measuring the amount of any specific binding by the antibody. In one 25 embodiment, such binding of antibody, in tissue sections, can be used to detect aberrant ERG, LTF, DD3, and/or AMACR localization or aberrant (e.g., high, low or absent) levels of ERG, LTF, DD3, and/or AMACR. In a specific embodiment, antibody to ERG, LTF, DD3, and/or AMACR can be used to assay in a biological sample (e.g., tissue, blood, or urine sample) for the presence of ERG, LTF, DD3, and/or AMACR where an aberrant 30 level of ERG, LTF, DD3, and/or AMACR is an indication of a diseased condition, such as cancer, including, for example, prostate cancer. 39 [0126] Any biological sample in which it is desired to detect an oligonucloetide or polypeptide of interest can be used, including tissue, cells, blood, lymph, semen, and urine. The biological sample is preferably derived from prostate tissue, blood, or urine. The tissue sample comprises cells obtained from a patient. The cells may be found in a 5 prostate tissue sample collected, for example, by a prostate tissue biopsy or histology section, or a bone marrow biopsy. The blood sample can include whole blood, plasma, serum, or any derivative thereof, including, for example, circulating cells, such as prostate cells, isolated from the blood sample, or nucleic acid or protein obtained from the isolated cells. Blood may contain prostate cells, particularly when the prostate cells are cancerous, 10 and, more particularly, when the prostate cancer metastasizes and is shed into the blood. Similarly, the urine sample can be whole urine or any derivative thereof, including, for example, cells, such as prostate cells, obtained from the urine. [0127] The immunoassays which can be used include, but are not limited to, competitive and non-competitive assay systems using techniques such as Western blots, 15 radioimmunoassays, ELISA, immunoprecipitation assays, immunodiffusion assays, agglutination assays, complement-fixation assays, immunoradiometric assays, fluorescent immunoassays, protein A immunoassays, to name but a few. [0128] ERG, LTF, DD3, and/or AMACR genes and related nucleic acid sequences and subsequences, including complementary sequences, can also be used in hybridization 20 assays. ERG, LTF, DD3, and/or AMACR nucleic acid sequences, or subsequences thereof comprising about at least 8, 15, 20, 50, 100, 250, or 500 nucleotides can be used as hybridization probes. Hybridization assays can be used to detect, prognose, diagnose, or monitor conditions, disorders, or disease states associated with aberrant changes in ERG, LTF, DD3, and/or AMACR expression and/or activity as described above. In particular, 25 such a hybridization assay is carried out by a method comprising contacting a sample containing nucleic acid with a nucleic acid probe capable of hybridizing under defined conditions (preferably under high stringency hybridization conditions, e.g., hybridization for 48 hours at 65*C in 6X SSC followed by a wash in 0.1X SSX at 50*C for 45 minutes) to an ERG, LTF, DD3, and/or AMACR nucleic acid, and detecting (i.e, measuring either 30 qualitatively or quantitatively) the degree of the resulting hybridization. As described herein, any nucleic acid amplification procedure, including, PCR/RT-PCR, TMA, 40 NASBA, 3SR, LCR, SDA, and LAMP can be used to detect the presence of the ERG, LTF, DD3 and/or AMACR gene and/or the level of its mRNA expression. [0129] In some applications, probes exhibiting at least some degree of self complementarity are desirable to facilitate detection of probe:target duplexes in a test 5 sample without first requiring the removal of unhybridized probe prior to detection. Molecular torch probes are a type of self-complementary probes that are disclosed by Becker et al., U.S. Patent No. 6,361,945. The molecular torch probes disclosed Becker et al. have distinct regions of self-complementarity, referred to as "the target binding domain" and "the target closing domain," which are connected by a joining region and 10 which hybridize to one another under predetermined hybridization assay conditions. When exposed to denaturing conditions, the complementary regions (which may be fully or partially complementary) of the molecular torch probe melt, leaving the target binding domain available for hybridization to a target sequence when the predetermined hybridization assay conditions are restored. And when exposed to strand displacement 15 conditions, a portion of the target sequence binds to the target binding domain and displaces the target closing domain from the target binding domain. Molecular torch probes are designed so that the target binding domain favors hybridization to the target sequence over the target closing domain. The target binding domain and the target closing domain of a molecular torch probe include interacting labels (e.g., luminescent/quencher) 20 positioned so that a different signal is produced when the molecular torch probe is self hybridized as opposed to when the molecular torch probe is hybridized to a target nucleic acid, thereby permitting detection of probe:target duplexes in a test sample in the presence of unhybridized probe having a viable label or labels associated therewith. [0130] Another example of detection probes having self-complementarity are the 25 molecular beacon probes disclosed by Tyagi et al. in U.S. Patent No. 5,925,517. Molecular beacon probes include nucleic acid molecules having a target complement sequence, an affinity pair (or nucleic acid arms) holding the probe in a closed conformation in the absence of a target nucleic acid sequence, and a label pair that interacts when the probe is in a closed conformation. Hybridization of the target nucleic 30 acid and the target complement sequence separates the members of the affinity pair, thereby shifting the probe to an open confirmation. The shift to the open confirmation is 41 detectable due to reduced interaction of the label pair, which may be, for example, a fluorophore and quencher, such as DABCYL and EDANS. (0131] By way of example, ERG, LTF, AMACR, or DD3 hybridization probes can comprise a nucleic acid having a contiguous stretch of at least about 8, 15, 20, 50, 100, 5 250, 500, 750, 1000, 1250, or 1500 contiguous nucleotides of SEQ ID NO:1, SEQ ID NO:2, SEQ ID NO:3, SEQ ID NO:4, or SEQ ID NO:5 or a sequence complementary thereto. Such contiguous fragments of SEQ ID NO:1, SEQ ID NO:2, SEQ ID NO:3, SEQ ID NO:4, or SEQ ID NO:5 may also contain at least one mutation so long as the mutant sequence retains the capacity to hybridize to SEQ ID NO: 1, SEQ ID NO:2, SEQ ID NO: 10 3, SEQ ID NO:4, or SEQ ID NO:5 under low or high stringency conditions (preferably under high stringency hybridization conditions, e.g., hybridization for 48 hours at 65*C in 6X SSC followed by a wash in 0.1X SSX at 50*C for 45 minutes). [0132] In specific embodiments, diseases and disorders involving hyperproliferation of cells, such as cancers, including, for example, prostate cancer, can be 15 diagnosed, or their suspected presence can be screened for, or a predisposition to develop such disorders can be predicted, by detecting levels of the ERG, LTF, and/or AMACR protein, ERG, DD3, and/or AMACR RNA, or ERG, DD3, and/or AMACR functional activity, or by detecting mutations in ERG, DD3, LTF and/or AMACR RNA, DNA, or protein (e.g., translocations in ERG, LFT, DD3, or AMACR nucleic acids, truncations in 20 the ERG, LFT, DD3, or AMACR gene or protein, changes in nucleotide or amino acid sequence relative to wild-type ERG, LTF, DD3, or AMACR) that cause increased or decreased expression or activity of ERG, LTF, DD3, and/or AMACR. By way of example, levels of ERG, LTF, and/or AMACR protein can be detected by immunoassay, levels of ERG, LTF, DD3, and/or AMACR mRNA can be detected by hybridization 25 assays (e.g., Northern blots, dot blots, or any nucleic acid amplification procedure, including, PCR/RT-PCR, TMA, NASBA, 3SR, LCR, SDA, and LAMP), translocations and point mutations in ERG, LTF, DD3, and/or AMACR nucleic acids can be detected by Southern blotting, RFLP analysis, any nucleic acid amplification procedure, including, PCR/RT-PCR, TMA, NASBA, 3SR, LCR, SDA, LAMP, sequencing of the ERG, LTF, 30 DD3, and/or AMACR genomic DNA or cDNA obtained from the patient, etc. [0133] In one embodiment, levels of the ERG, DD3, LTF and/or AMACR mRNA or protein in a subject sample are detected or measured and compared to the mRNA or 42 protein expression levels of the corresponding gene in a control sample or to a standard numerical value or range. For example, increased expression levels of ERG, DD3, and/or AMACR or decreased levels of LTF, relative to a matched, normal tissue sample, indicate that the subject has a malignancy or hyperproliferative disorder, including, for example, 5 prostate cancer, or a predisposition to develop the same. Other appropriate controls include other noncancerous samples from the subject, samples obtained from a different subject without cancer, or other cancer-specific markers. For example, in prostate cancer, a prostate-cell specific marker, such as PSA, can be used as a control to compare and/or normalize expression levels of other genes, such as ERG, LTF, DD3, and/or AMACR. In 10 one embodiment, a method of diagnosing cancer, such as prostate cancer, comprises obtaining a biological sample from a subject (e.g., a tissue sample (e.g., from biopsy tissue), a blood sample, or a urine sample), determining the expression level of a ERG, LTF, DD3, and/or AMACR gene and/or ERG, LTF, DD3, and/or AMACR activity in the samples, and diagnosing or prognosing cancer in said subject. In further embodiments, the 15 expression level of the ERG, LTF, DD3, and/or AMACR gene and/or ERG, LTF, DD3, and/or AMACR activity is determined by Southern blotting, Northern blotting, Western blotting, ELISA, any nucleic acid amplification procedure, including, PCR/RT-PCR, TMA, NASBA, 3SR, LCR, SDA, and LAMP, or other techniques as described herein or known in the art. Without limiting the instant invention, increased or decreased 20 expression of at least two times, as compared to the control sample indicates the presence of prostate cancer or a higher predisposition to developing prostate cancer. [0134] Another aspect of the invention provides a means for monitoring a response to "hormonal therapy" by evaluating the expression profiles of the ERG gene, alone or in combination with the AMACR and/or DD3 genes and/or LTF genes, and correlating these 25 profiles with the clinical signs of the disease. [0135] Kits for diagnostic use are also provided. A kit comprises an anti-ERG gene antibody or an antibody directed against the ERG protein and/or an anti-AMACR gene antibody or an antibody directed against the AMACR protein and/or an anti-DD3 gene antibody and/or and an anti-LTF gene antibody or an antibody directed against the 30 LTF protein,which can be optionally detectably labeled. A kit is also provided that comprises a nucleic acid probe capable of hybridizing under defined conditions (preferably under high stringency hybridization conditions, e.g., hybridization for 48 hours 43 at 65*C in 6X SSC followed by a wash in 0.1X SSX at 50*C for 45 minutes) to ERG, LTF, DD3, and/or AMACR nucleic acid. In a specific embodiment, a kit comprises at least a pair of primers (e.g., each in the size range of at least about 6, 17, 30, or 60 nucleotides) that are capable of priming amplification, by any nucleic acid amplification 5 procedure (including e.g., PCR/RT-PCR, TMA, NASBA, 3SR, LCR, SDA, LAMP), of the ERG, LTF, DD3, and/or AMACR gene or a fragment thereof. A kit can comprise a predetermined amount of a purified ERG, LTF, DD3, and/or AMACR protein or nucleic acid for use, e.g., as a standard or control. The kit can also comprise one or more components for detecting the nucleic acid probe, including components described herein 10 or known in the art. [0136] In one embodiment, the kit comprises a nucleic acid that hybridizes under defined conditions (and preferably under conditions of high stringency, e.g., hybridization for 48 hours at 65*C in 6X SSC followed by a wash in O.IX SSX at 50*C for 45 minutes) with at least one gene chosen from those genes identified in Tables 1-6 or the DD3 gene, 15 and is affixed to a support, alone, or in combination with other nucleic acids. For example, an ERG and/or LTF nucleic acid can be affixed to the support, with or without other nucleic acids. In a specific embodiment, the support comprises at least an ERG nucleic acid and an AMACR nucleic acid or at least an ERG nucleic acid and a DD3 nucleic acid. In another embodiment, the support comprises at least an ERG nucleic acid, 20 an AMACR nucleic acid, and a DD3 nucleic acid. This support can be used as part of a kit for detecting cancer, such as prostate cancer. These kits can further comprise at least a pair of primers (e.g., each in the size range of at least about 6, 17, 30, or 60 nucleotides) that are capable of priming amplification, by any nucleic acid amplification procedure (including e.g., PCR/RT-PCR, TMA, NASBA, 3SR, LCR, SDA, LAMP), of the ERG, 25 LTF, DD3, and/or AMACR gene or a fragment thereof. EXAMPLES Example 1: Screening of CaP cell-specific gene expression signatures using Affymetrix GeneChip Patient Selection 30 [0137] Specimens were obtained under an IRB-approved protocol from patients treated by radical prostatectomy (RP) at Walter Reed Army Medical Center (WRAMC). From over 300 patients two groups were selected which had prostate tumors with either 44 moderate (MR) or high risk (HR) of disease progression after RP. The HR group had PSA recurrence, Gleason score 8-9, T3c stage, seminal vesicle invasion, and poorly differentiated tumor cells; the MR group had no PSA recurrence, Gleason score 6-7, T2a T3b stage, no seminal vesicle invasion, and well to moderately differentiated tumor cells. 5 LCM compatible specimens were selected from age and race matched HR or MR patients with no family history of CaP. Tissue Samples and Laser-Capture Microdissection [0138] Normal and cancer cells were laser capture microdissected (LCM) from OCT embedded and Hematoxylin-eosin (H&E) stained frozen prostate sections of radical 10 prostatectomy specimens (2000 laser shots for one sample). Laser capture microdissection (LCM) facilitates the isolation of morphologically defined, homogenous cell populations from complex tissues by selectively adhering the cells of interest to a transparent film with focused pulses of low energy infrared laser under a microscope. Emmert-Buck et al., Science (1996); 274(5289): 921-922; Schutz et al., Nat Biotechnol (1998) 16(8): 737-742. 15 RNA Extraction and 77-based linear RNA amplification [0139] Total RNA was isolated from the LCM samples with the MicroRNA kit (Stratagene, La Jolla, CA), quantified using RiboGreen dye (Molecular Probes, Eugene, OR) and VersaFluor fluorimeter (BioRad, Hercules, CA), and quality tested by RT-PCR using NKX3.1 and GAPDH primers. Linear RNA amplification was performed using 20 RiboAmp RNA amplification kit (Arcturus, Mountain View, CA). Precisely, 2 nanograms of total RNA from LCM derived epithelial cells of normal as well as tumor tissue from each patient was used for the first round of amplification. During the second round of amplification after cDNA synthesis and purification the samples were biotinylated during in vitro transcription which was used for the GeneChip analysis. 25 Gene Chip Analysis [0140] Linearly amplified aRNA was hybridized to high-density oligonucleotide human genome array (HG U133A array) (Affymetrix, Santa Clara, CA, USA). The array contains 22,283 probe sets, about 18,000 of which represent well annotated genes, while the remainder represent various expressed sequence tags (EST) and hypothetical genes. 30 Biotinylation was carried out using aRNA by in vitro transcription using MEGA script T7 in vitro Transcription Kit (Ambion, Austin, TX, USA) cDNA and biotinylated UTP and biotinylated CTP (ENZO, Farmingdale, NY, USA)(34). The biotin labeled cRNA was 45 purified using the QIAGEN RNeasy spin columns (QIAGEN, Valencia, CA) following the manufacturer's protocol. The biotin labeled cRNA was fragmented in a 40 pt reaction mixture containing 40 mM Tris-acetate, pH 8.1, 100 mM potassium acetate, and 30 mM magnesium acetate incubated at 94*C for 35 minutes and then put on ice. 5 Hybridization, Staining and Scanning of the GeneChip [0141] The biotin labeled and fragmented aRNA was hybridized to the HG U133A array. Briefly, a 220 pl hybridization solution consisting of: IM NaCl, 10 mM Tris pH 7.6, 0.005% Triton X-100, 50 pM control Oligo B2 (5' bioGTCAAGATGCTACCGT'CAG 3') (SEQ ID NO:6) (Affymetrix); the control cRNA 10 cocktail of: Bio B (150 pM), Bio C (500 pM), Bio D (2.5 nM) and Cre X (10 nM) (American Type Tissue Collection, Manassas,VA and Lofstrand Labs, Gaithersburg, MD), 0.1 mg/mi herring sperm DNA and 0.05 pg/Il of the fragmented labeled sample cRNA was heated to 95*C for 35 min., cooled to 40*C and clarified by centrifugation. Hybridization was at 42*C in a rotisserie hybridization oven (Model 320, Affymetrix) at 15 60 rpm for 16 hours. Following hybridization, the GeneChip arrays were washed 10 times at 25 0 C with 6X SSPE-T buffer (1 M NaCl, 0.006 M EDTA, and 0.06 M Na 3

PO

4 , 0.005% Triton X-100, pH 7.6) using the automated fluidics station protocol. GeneChip arrays were incubated at 50*C in 0.5X SSPE-T, 0.005% Triton X-100 for 20 minutes at 60 rpm in the rotisserie oven. GeneChip arrays were stained for 15 minutes at room temperature and at 20 60 rpm, with streptavidin phycoerythrin (Molecular Probes, Inc., Eugene, OR) stain solution at a final concentration of 10 pg/ml in 6X SSPE-T buffer and 1.0 mg/ml acetylated bovine serum albumin (Sigma). GeneChip arrays were washed twice at room temperature with 6X SSPE-T buffer, and then were scanned with the HP GeneArray Scanner (Hewlett-Packard, Santa Clara, CA) controlled by GeneChip 3.1 Software 25 (Affymetrix). Example 2: Analysis of GeneChip Results by Supervised Multi-Dimensional Scaling (MDS) Image Analysis and Data Collection [0142] Affymetrix GeneChip Microarray Analysis Software, version 3.1 and 30 Affymetrix Micro DB and Data Mining Tool version 2.0 (Affymetrix), Microsoft Excel 2000 (Microsoft, Seattle, WA) and Statistica version 4.1 (Stat Soft, Inc., Tulsa, OK) were used. In the Affymetrix system, the average difference fluorescence is the average of the 46 difference between every perfect match probe cell and its control mismatch probe cell and is directly related to the level of expression of a transcript. A comparative file indicates the relative change in abundance (fold change) for each transcript between a baseline and an experimental sample. For further detail and advanced bioinformatic analysis we used the 5 Microarray Data Analysis software from NHGRI and the GeneSpring software (Silicon Genetics, CA). Data Analysis [0143] For clustering analysis, National Human Genome Research Institute (NHGRI) Microarray Data Analysis software was used, which partitioned the samples of 10 the high risk and moderate risk groups into well-separated and homogeneous groups based on the statistical behavior of their genes expression. To achieve the objective of clustering each of the groups, all pair-wise similarities between samples were evaluated, and then grouped via the average linkage algorithm. Pearson correlation coefficient or Euclidean distance were typically used to quantify the similarity. Unsupervised hierarchical and or 15 non hierarchical clustering was also performed using the same distance matrix. [0144] Using a matrix of Euclidean distance measurements from complete pair wise comparison of all the prostate specimens, a multidimensional scaling (MDS) method was performed using an implementation of MDS in the MATLAB package to determine the overall similarities and dissimilarities in gene expression profiles. A weighted gene 20 analysis was performed to generate a subset of genes statistically significant in separating the high risk group from the moderate risk group. [0145] Briefly, for two different groups e.g., epithelium of high risk tumor and epithelium of moderate risk tumor with a given number of samples 25 and 25, the discriminative weight for each gene is determined by the formula: w = dB/(kidwi + k 2 dw 2 + 25 a); where da is the Euclidean distance between two groups (center-to-center or between cluster Euclidean distance), dw, is the average Euclidean distance among all the epithelial samples of high risk group, dw 2 is average Euclidean distance among all the epithelial samples of moderate risk group, ki = 25/(25 +25), k 2 = 25/(25 + 25), and a is a small constant to ensure the denominator is never equal to zero. Genes were ranked according 30 to their w values. Genes with high w values created greater separation between groups and denser compaction within the group. In other words, the subset of genes with high w values have the most discriminative power to differentiate a high risk group from a 47 moderate risk group and vice versa. Sample labels were randomly permuted and the w value was computed again for each gene to test the statistical significance of the discriminative weights. Genes with the most significant expression differences were selected by p-values. A hierarchical clustering algorithm was used to verify the predictor 5 model obtained from the supervised MDS analysis. [0146] From this analysis, specific genes were identified whose expression signature in tumor tissue varied from their expression signature in benign matched tissue. Genes with a p-value of not more than 0.05 were selected and ranked by p-value, as shown in Tables 1-6. 10 In Silico Validation: [0147] We have tested the discriminatory potential of the genes that we obtained from our analysis on some independent data sets. Affymetrix oligonucleotide GeneChip Hum95Av2 data were obtained from Welsh et al. 2001, Singh et al. Genes from these data bases that correspond with the genes of our discriminatory list were selected and their 15 tumor specific expression intensities and/or tumor over normal ratio were used for an MDS analysis as described above in the data analysis section. MDS plots were obtained depicting the discriminatory capability of the genes on the independent data sets. 48 Table 1 The first 50 genes obtained from the supervised MDS analysis of tumor versus benign tissues of all the high risk and moderate risk CaP patients, ranked by p-value. (T vs B In All 18 Samples) 5 No. a Common Name of Genes Description of Genes Map p-Valu Expression Accesion 41UIRegulation Tumor Benign I. AF047020 AMACR Alpha-methylacyl-CoA racemase Sp13.2-q 1. 1 0 Up Down 2. NM_002343 LTF Lactotransferrin 3q21-q23 0 Down Up 3. NM_002275 KRT15 Keratin 15 17q21 0.000001 Down Up 4. BC000915 CUM[, CLP36. CLP-36 PDZ and UM domain I (elfin) l0q22-q26.3 0.000001 Down Up 5. X90579 CYP3A5 Cytochrome P450. subfamily 3A. 7 0.000001 Down Up polypeptide 5 6. NM003671 CDC14BI, CDCl4B2 H. sapiens CDCl4 cell division cycle 14 9q22.2-q22.31 0.000005 Down Up homolog B 7. A1424243 CEGPI H. sapiens cDNA clone 11 0.000005 Down Up IMAGE:2094442 8. NM_022370 Rbig I Hypothetical protein FU2 1044 similar I I q24.2 0.000009 Down Up to Rbigl 9. A1356398 ZFP36L2 TISDHUMAN P47974T1SIID 2 0.000018 Down Up PROTEIN 10. NM005213 STFISTFA Cystatin A steinn A) 3q21 0.000018 Down Up HI. NM_006394 RIG Regulatedingliona lipl5.1 0.000018 Down Up 12. AP275945 EVAI Epithelial V-like antigen I I Iq23.3 0.000018 Down Up 13. NM_020186 DCII DCII protein 7q2t.3 0.000018 Up Down 14. Al922538 TMEMI Transmembrane protein I 21 0.000018 Down Up 15. NM_014863 BRAG. KIAA0598 B cell RAG associated protein 10q26 0.000018 Down Up 16. A1669229 RARRES) HomosapienscDNAclone 3q25.33 0.000036 Down Up IMAGE:2315074 17. NM_006017 ACI33. CD133 Prominin (mouse)-like I 4 p1 5 .33 0.000036 Down Up 18. NM..004503 HOXC6 HomeoboxC6 12qI2-q13 0.000036 Up Down 19. NM_005084 PAFAH, LDL-PLA2 Phospholipase A2. group VII 6 p 2 l.2-pl2 0.000036 Up Down 20. NM..001511 MGSA, CXCLI.SCYBI GROI oncogene 4q21 0.000071 Down Up 21. 80054844 ARHE H.sapienscDNAclone 2q23.3 0.000071 Down Up IMAGE:3441573 22. NM_007191 WIF-1 Wnt inhibitory factor-I 12q14.2 0.000071 Down Up 23. X99268 TWIST Twist (Drosophila) hormolog 7p21.2 0.000071 Up Down 24. A1826799 EFEMPI EXTRACELLULAR PROTEIN S I-5 2pl6 0.000071 Down Up PRECURSOR 25. NM_001018 RPSIS Ribosomal protein S15 19p13.3 0.000071 Up Down 26. AV711904 LYZ Lysozyme (renal amyloidosis) 0.000071 Down Up 27. A1433463 MME NEPRILYSIN (HUMAN) 3q25.1-q25.2 0.000071 Down Up 28. BE908217 ANXA2 H. sapiens cDNA clone 15q21-q22 0.000071 Down Up IMAGE:3902323 29. NM_000441 PDS, DFNB4 Solute carrier family 26, member 4 7q31 0.000071 Down Up 30. BC003068 SLC19AI Solute carrier family 19, member I 21q22.3 0.000071 Up Down 31. NM_005950 MTI Metallothionein I G 16q13 0.000071 Down Up 32. NM_013281 FLRT3 Fibronetin leucine rich transmembrane 20pI l 0.000071 Down Up protein 3 33. A1351043 ESTs H. sapiens cDNA clone 21 0.000145 Up Down IMAGE-1948310 * 34. NM..001099 PAP Acid phosphatase, prostate 3q21-q23 0.000145 Down Up 35. NM_006113 VAV3 Vav 3 oncogene tpI3.1 0.000145 Down Up 36. NM_005980 Sl00P S100 calcium-binding protein P 4pI6 0.000145 Down Up 37. NM_000165 GJAI Gap junction protein, alpha 1, 43kD 6q21-q23.2 0.000145 Down Up (connexin 43) 38. NM_003897 DIP2. IEXI, PRGI Immediate early response 3 ' 6p21.3 0.000145 Down Up 39. BC001388 ANX2. UP2, CALIH Annexin A2 15q2t-q22 0.000145 Down Up 40. BC003070 HDR, MGC5445 GATA-binding protein 3 l0p I5 0.000145 Down Up 41. NM020139 LOC56898 Oxidoreductase UCPA 4 0.000145 Down Up 42. AK002207 KIAA0610 KIAA0610 protein 13 0.000145 Down Up 43. NM000574 CR,TC. CD55 Decay accelerating factor for 1q32 0.000145 Down Up complement 44. NM_006926 SP-A2, COLEC5 Surfactani. pulmonary-associated 10q22.q23 0.000145 Up Down protein A2 45. U37546 API2, MIHC, CIAP2 Baculovirul lAP repeal-containing 3 Ilq22 0.000145 Down Up 46. AU148057 DKK3 H. sapiens cDNAclone lIpter-p5.5 0.000145 Down Up MAMMA1002489 47. NM_002600 DPDE4, PDEIVB Phosphodiesterase 48, cAMP-specific 1p31 0.000145 Down Up 48. S59049 BL34, IER I. IR20 Regulator of 0-protein signalling I 1q31 0.0003 Down Up 49. NM_001275 CGA. CgA Chromogrunin A (parthyroid secretory l4q32 0.0003 Down Up protein 1) 50. AL575509 ETS2 H. sapiens cDNA clone CSOD1059YP21 21q22.2 0.0003 Down Up 49 Table 2. The first 50 genes from the supervised MDS analysis of tumor over benign (T/B) tissues ratio (Fold Change) of the high risk versus moderate risk CaP patients, ranked by p-value.: (T/B Fold Change In HR vs MR) No Genbank Common Name of Genes Description of Genes Map p-Value accession 1. NM004522 KINN, NKHC. NKHC2. NKHC-2 Kinesin family member 5C 2q23.3 0.00011 2. 103198 GNA13 Guanine nucleotide binding protein G (K), alpha subunit Ip13 0.000981 3. NM018010 HIPPI, FUI0147 Hypothetical protein FU10147 3q 13.13 0.003257 4. NM005479 FRATI Frequently rearranged in advanced T-cell lymphomas 10q23.33 0.004964 5. NM_021795 SAPI ELK4, ETS-domain protein (SRF accessory protein 1) 1q32 0.004964 6. NM_003113 LEUS, RFP2 Nuclear antigen Spl00 2q37.1 0.004964 7. NM_002053 GBPI Guanylate binding protein I, interferon-inducible, 67kD lp 22

.

1 0.004964 8. AF064092 GSA, GSP, GPSA, GNAS I, Guanine nucleotide regulatory protein 20ql3.2-ql 3.3 0.007579 9. BC003070 HDR. MGC5445 GATA-binding protein 3 1Op15 0.007579 10. NM_012245 SKIP. NCOA-62 SKI-interacting protein 14q24.3 0.007579 I. NM_015895 LOC51053 Geminin 6 p22.2 0.007579 12. AA083478 TRIM22 Stimulated trans-acting factor (50 kDa) I1 0.007579 13. NM_000100 PME, CST6. EPM 1. STFB Cystatin B (stefin B) 21q22.3 0.007579 14. NM_003031 SIAHI Seven in absentia (Drosophila) homolog I 16ql2 0.007579 15. NM-003407 TTP, GOS24, TIS I, NUP475 Zinc finger protein 36, C3H type, homolog (mouse) 19q13.1 0.007579 16. BF979419 ESTs ESTs, Highly similar to 60S ribosomal protein 13A 19q13.33 0.007579 [H.sapiens) 17. NM_021038 MBNL Muscleblind (Drosophila)-like 3q25 0.007579 18. NM_014454 PA26 P53 regulated PA26 nuclear protein 6q21 0.007579 19. BC004399 DEME-6 DEME-6 protein l p 32

.

3 0.007579 20. NM_018490 LGR4 G protein-coupled receptor 48 1 ipl4-pl3 0.007579 21. NM.004328 BCS, BCSI, h-BCS, Hr.6719 BCSI (yeast homolog)-like 2q33 0.007579 22. D87445 KIAA0256 KIAA0256 gene product 15 0.007579 23. NM_006326 NIFIE14 Homo sapiens seven transmembrane domain protein, 19q13.12 0.007579 mRNA 24. D83077 TI3 Tetratricopeptide repeat domain 3 Xql3.1 0.007579 25. NM_006732 GOS3 FBJ murine ostcosarcoma viral oncogene homolog B 19q 13.32 0.007579 26. NM_003760 El F4G3 Eukaryolic translation initiation factor 4 gamma, 3 I pter-p36.13 0.007579 27. NM_004905 AOP2 Anti-oxidant protein 2 lq24.1 0.01159 28. NM_018439 IMPACT Hypothetical protein IMPACT 18 0.01159 29. 8C000629 DARS Aspartyl-tRNA synthetase 2q2l.2 0.01159 30. AK002064 DKFZP564A2416 DKFZP564A2416 protein 2 0.01159 31. NM_013387 HSPCO51 Ubiquinol-cytochrome c reductase complex (7.2 kD) 22 0.01159 32. AA135522 K1AA0089 Homo sapiens KIAA0089 mRNA sequence. 3 0.01159 33. NM_015545 KIAA0632 KIAA0632 protein 7q22.1 0.01159 34. NM_005767 P2Y5 Purinergic receptor familyy A group 5) 13q14 0.01159 35. BC003682 G25K. CDC42Hs Cell division cycle 42 (GTP-binding protein, 25kD) I p36.1 0.01159 36. NM_005053 RAD23A RAD23 (S. cerevisiae) homolog A 19p13.2 0.017805 37. A1672541 IPW Human non-translated mRNA sequence. ISql-1q12 0.017805 38. AK023938 H. sapiens cDNA FU13g16 clone SELECTED MODEL ORGANISM PROTEIN 2q37.3 0.017805 SIMILARITIES 39. NM_000062 CIIN, CINH. Cl-INH Serine (or cysteine) proteinase inhibitor, clade C (Cl I lql2-q 13.1 0.017805 inhibitor) 40. AA576961 PHLDA I Pleckstrin honology-like domain, familyA, member I 12ql5 0.017805 41. A1796269 NBSI, ATV, NIBRIN H. sapiens cDNA similar to Cell Cycle Regulatory Protein Sq21-q24 0.017805 P95. 42. NM_000016 ACADM Acyl-Coenzyme A dehydrogenase. C-4 to C-12 straight lp31 0.017805 chain 43. A1867102 KIAA0906, NUP210, gp210 Nuclear pore membrane glycoprotein 210 3p 2 5.2-p25.1 0.017805 44. A1263909 ARHB, RHOB, RHOH6 Oncogene RHO6; Aplysia RAS-related homolog 6 2pter-p12 0.017805 45. NM_016021 NCUBEI Non-canonical ubquitinconjugatingenxyme I 6 0.017805 46. NM_012192 TIM98.TIMIOB Fracture callus I (rat) homolog I Ipl5.5-pl5.3 0.017805 47. NM_025087 FU21511 Hypothetical protein FUJ21511 4 0.017805 48. NM_014959 CARD8, CARDINAL. Tumor up-regulated CARD-containing antagonist of 19q13.33 0.017805 K1AA0955 caspase 9 49. AA923354 MAOA Monoamine oxidase A. Xpl1.4-p 11.3 0.017805 50. NM_021964 ZNF148 Zinc finger protein 148 (pHZ-52) 3q21 0.017805 51. NM-001674 ATF3 Activating transcriptionfactor3 1q32.3 0.017805 50 Table 3. The first 50 genes obtained from the supervised MDS analysis of tumor versus benign tissues of all the high risk CaP patients, ranked by p-value. (T vs N Intensities of 9 HR) 5 Common No. Genbank Name of Description of Genes Map p.Value Expression Accession G:9Regulation Genes Tumor Benign 1. U65585 HLA-DR1B Major histocompatibility complex, class 11, OR beta 6p21.3 0.00002 Down Up 1 2. NM_002053 GBP1 Guanylate binding protein 1, interferon-Inducible, 1 p22.1 0.000078 Down Up 3. NM_021983 HLA-DRB4 Major histocompatibility complex, class 11, DR beta 6 0.000076 Down Up 4 4. A1424243 CEGP1 Homo sapiens cDNA clone IMAGE:2094442 11 0.000102 Down Up 5. NM_002343 LTF Lactotransferrin 3q21-q23 0.000138 Down Up 6. NM_014575 SCHIP-1 Schwannomin-interacting protein 1 3q26.1 0.000257 Down Up 7. BC001 169 ESD Esterase D/formylglutathione hydrolase 13q14.1. 0.000357 Up Down q14.2 8. BF970427 UGCG UDP-glucose ceramide glucosyftransferase 9 0.000357 Down Up 9. NM_002Z75 KRT15 Keratin 15 17q21 0.000495 Down Up 10. AU148057 DKK3 H. sapiens cDNA clone MAMMA1002489 11pter- 0.000495 Down Up p15.5 11. A1922538 TMEMI Transmembrane protein 1 21 0.000689 Down Up 12. NM_004481 ( GALNAC-T2 UDP- GaINAc transferase 2 1q41 -q42 0.000689 Down Up 13. BC003070 HDR, GATA-binding protein 3 10p15 0.00097 Down Up MGC5445 14. BF979419 ESTs, similar H. sapiens 60S Ribosomal protein L13A 0.00097 Up Down to RPL13A 15. 6054844 ARHE H. sapiens cDNA clone IMAGE:3441573 2q23.3 0.00097 Down Up 16. L42024 HLA-B Major histocompatibility complex, class I, B 6p21.3 0.00138 Down Up 17. AL545982 CCT2 H. sapiens cDNA clone CSD1023YD15 12q15 0.001992 Up Down 18. NM_001993 TF, TFA, Coagulation factor liI (thromboplastin, tissue factor) 1p22-p21 0.001992 Up Down CD142 19. NM_004198 CHRNA6 Cholinergic receptor, nicotinic, alpha polypeptide 6 8p11.1 0.001992 Down Up 20. AV711904 LYZ Lysozyme (renal amyloidosis) 12q15 0.001992 Down Up 21, NM_013387 HSPCO51 Ublquinol-cytochrome c reductase complex (7.2 22 0.001992 Up Down kD) 22. AW514210 HLA-F HLA CLASS I HISTOCOMPATIBILITY ANTIGEN, 6p21.3 0.001992 Down Up FA 23. NM_005032 PLS3 Plastin 3 (T isoform) Xq24 0.002894 Down Up 24. NM003407 TTP, GOS24, Zinc finger protein 36, C3H type, homolog (mouse) 19q13.1 0.002894 Down Up NUP475 25. NM_000165 GJA1 Gap junction protein, alpha 1, 43kD (connexin 43) 6q21- 0.002894 Down Up q23.2 26. AF275945 EVA1 Epithelial V-like antigen 1 1 1q23.3 0.002894 Down Up 27. NM002450 MT1 Metallothoneln 1L 16q13 0.002894 Down Up 28. NM_005950 MT1 Metallothloneln 1G 16q13 0.002894 Down Up 29. NM.006994 BTN3A3 Butyrophlin, subfamily 3. member A3 6p21.33 0.002894 Down Up 30. A1049962 KIAA0191 H. sapiens cDNA clone IMAGE:1700970 1 0.002894 Down Up 31. X99268 TWIST Twist (Drosophila) homolog 7p21.2 0.002894 Up Down 32. NM_016021 NCUBEI Non-canonical ubquitln conjugating enzyme 1 6 0.002894 Up Down 33. NM.016205 SCDGF Platelet derived growth factor C 4Q32 0.002894 Up Down 34. A1681120 RANBP2 H. sapiens cDNA clone IMAGE:2272403 2Q11-q13 0.004205 Up Down 35. NM_000574 CR, TC. CD55 Decay accelerating factor for complement 1q32 0.004205 Down Up 36. NM_01 4937 KIAA0966 Sac domain-containing Inositol phosphatase 2 1 0q26.13 0.004205 Down Up 37. NM_005213 STF1, STFA Cystalin A (stafin A) 3q21 0.004205 Down Up 38. NM_005952 MTI Melallothionein IX 16q13 0.004205 Down Up 39. AF130095 FN1 Fibroneclin 1 2q34 0.004205 Down Up 40. BE568219 PDE8A H. sapiens cDNA clone IMAGE:3683966 15q25.1 0.004205 Up Down 41. D50925 STK37, PAS-serine/threonine kinase 2q37.3 0.004205 Down Up PASKIN, 42. NM_006113 VAV3 Vav 3 oncogene 1p13.1 0.004205 Down Up 43. NM_001018 RPS15 Rlbosomal protein S15 19p13.3 0.006189 Up Down 44. NM021038 MBNL Muscleblind (Drosophila)-like 3q25 0.006189 Down Up 45. NM_012323 U-MAF V-maf musculoaponeurotic fibrosarcoma, protein F 22q13.1 0.006189 Down Up 46. NM005138 SCD1L SCO (cytochrome oxidase deficient, yeast) 22q13.33 0.006189 Down Up homolog 2 47. AF188779 KIAA0959 RaIGDS-like gene 1q25.2 0.006189 Down Up 48. D26054 FBP Fructose-1,6-bisphosphatase 1 9q22.3 0.006189 Up Down 49. U37546 APi2, MIHC, Baculoviral IAP repeat-contalning 3 11q22 0.006189 Down Up HIAP1 50. AB046845 SMURF1 E3 ubiquitin ligase SMURF1 7q21.1- 0.006189 Down Up q31.1 51 Table 4. The first 50 genes obtained from the supervised MDS analysis of tumor versus benign tissues of all the moderate risk CaP patients, ranked by p-value: (T vs N Intensities of 9 MR) 5 N. Gnak Common Name Expression No. Genbn oGen Description of Genes Map p-Value Regulation Tumor Benign I. NM_0i4324 AMACR Alpha-methylacyl-CoA racemase 5pl3.2- 0 Up Down qll.I 2. NM_006457 ENH UM protein (similar to rat protein kinase C-binding 4q22 0.000009 Up Down enigma) 3. A1351043 ESTs H. sapiens cDNA clone IMAGE:1948310 21 0.00001I Up Down 4. Ai433463 MME H. sapiens cDNA clone similar to NEPRILYSIN 3q25. I - 0.000028 Down Up (HUMAN) q25.2 5. BE256479 HSPDI H. sapiens cDNA clone IMAGE:3352031 12p13.31 0.000037 Up Down 6. NM_015900 PS-PLA I Phosphatidylserine-specific phospholipase A I alpha 3q 13.13. 0.000083 Up Down q13.2 7. NM_002343 LTF Lactotransfenrin 3q2t-q23 0.000083 Down Up 8. NM_001099 PAP Acid phosphatase, prostate 3q21-q23 0.000083 Down Up 9. Ti5991 CHRM3 IB2413 Infant brain, Bento Scares Homo sapiens cDNA q4I-q44 0.00011 Up Down 10. NM_005084 PAFAH Phospholipase A2, group Vi1 6 p 2 1.

2 -p1 2 0.00011 Up Down It. NM_004503 HOXC6 Honeobox C6 I2qI2-q13 0.00011 Up Down 12. N74607 AQP3 H. sapiens cDNA clone IMAGE:296424 9pl3 0.000149 Down Up 13. BC003068 SLC19AI Solute carrier family 19 folatee transponer), member I 2iq22.3 0.000149 Up Down 14. M21535 ERG (ets-related ERG v-eta crythroblastosis virus E26 oncogene like 21q22.3 0.000149 Up Down gene) (avian) 15. NM_013451 MYOF, Fer-I (C.elegans)-like 3 (myoferlin) i0q24 0.0002 Down Up 16. NM_006017 AC133,CD133 Prominin(mouse)-likC I 4 pl5.

33 0.0002 Down Up 17. BESS0599 CACNAID H. sapiens cDNA clone IMAGE:3220210 3pl4.3 0.0002 Up Down 18. U22178 PSP57, PSP94 Microseminoprotein, beta. l091 1.2 0.0002 Down Up 19. NM015865 JK. UTI., UTE Solute carrier family 14 (urea transporter), member i i8qi I-q]2 0.000275 Down Up 20. NM-000441 PDS, DENB4 Solute carrier family 26. member 4 7q3i 0.000275 Down Up 21. AA877789 MYO6 H. sapiens cDNA clone IMAGE:1161091 6qi3 0.000275 Up Down 22. A1356398 ZFP36L2 H. sapiens cDNA clone IMAGE:2028039 2 0.000275 Down Up 23. BCOO915 CUM1, CLP36 PDZ and UM domain I (elfin) 10q22- 0.000275 Down Up q26.3 24. NM_000286 PEX12 Peroxisomalbiogenesis factor 12 17ql1.2 0.000275 Up Down 25. NM_003671 CDC14BI, Homo sapiens CDC14 cell division cycle 14 homolog B 9q22.2. 0.000386 Down Up CDC14B2, (S. cerevisiae) (CDC14B), transcript variant 1. mRNA q22.31 26. NM016545 SBB148 Inmediate early response 5 lq24.3 0.000386 Down Up 27. NM_002443 PSP57. PSP94 Microseminoprotein, beta- 10qI 1.2 0.000386 Down Up 28. NM004999 DFNA22 Myosin Vi 6q13 0.000386 Up Down 29. X99268 TWIST Twist (Drosophila) homolog 7p21.2 0.000386 Up Down 30. NM_023009 MACMARCKS Macrophage myristoylated alanine-rich C kinase i p34.3 0.000386 Up Down substrate 31. A1721219 TRAF3 as68bi l.xI Barstead colon HPLRB7 Homo sapiens 14q32.33 0.000547 Down Up cDNA clone IMAGE:2333853 3', mRNA sequence. 32. NM_001584 DI IS302E Chromosome I I open reading frame 8 ilpI3 0.000547 Down Up 33. NM_018846 SBB126 SBB126 protein 7pl5.3 0.000547 Up Down 34. M87771 BEK, KGFR, Fibroblast growth factor receptr 2 10q26 0.000547 Down Up 35. AF275945 EVAI Epithelial V-like antigen I I Iq23.3 0.000547 Down Up 36. A1791860 ESTs H. sapiens cDNA clone IMAGE:10 11110 0.000547 Up Down 37. BC001282 NHC High-mobility group (nonlhistone chromosomal) protein 6p2 .3 0.000547 Down Up 17-like 3 38. NM_002015 FKHI, FKHR Forthead box 01A (rhabdomyosarcoma) 13q14.1 0.000547 Down Up 39. X15306 NF-H H.sapiens NF-H gene, exon I (and joined CDS). 22q12.2 0.000547 Down Up 40. BE965029 EST H. sapiens cDNA clone IMAGE:3886131 11 0.000775 Up Down 41. NM_002275 KRT15 Keratin 15 17q2i 0.000775 Down Up 42. NM_001511 MGSA. CXCLI GRO oncogene 4q21 0.000775 Down Up 43. NM_005213 STFI, STFA Cystatin A (stefin A) 3q21 0.000775 Down Up 44. NM.007191 WIF-I Wnt inhibitory factor- I 12q14.2 0.000775 Down Up 45. H15129 MEIS3 EPIDERMAL GROWTH FACTOR-LIKE CRIPTO 17 0.000775 Down Up PROTEIN 46. AW452623 EST H. sapiens cDNA clone IMAGE:3068608 13 0.000775 Up Down 47. X90579 EST H.sapiens DNA for cyp related pseudogene 7 0.000775 Down Up 48. BC001388 ANX2, Annexin A2 15q21-q22 0.001116 Down Up ANX2LA 49. NM_014863 BRAG, B cell RAG associated protein 10q26 0.001116 Down Up 50. NM_ 021076 NEFH Neurofilament, heavy polypeptide (200kD) 22q12.2 0.001116 Down Up 52 Table 5. Top 50 Upreaulated Genes in All the 18 Samples (HR and MR) obtained from Tumor over Benign (T/B) ratio. Genbank T/N No ID Ratio Common Name of Genes Description Map 1. AF047020 39.86910 AMACR Alpha-methylacyl-CoA racemase 5p13.2-q 1.1 2. M54886 20.86411 LOC51334 MesenchymalstemcellproteinDSC54 5p13.1 3. AF070581 19.07263 ESTs Homo sapiens cDNA clone IMAGE:1948310 21 4. NM.014324 18.04841 TRG'i T cell receptor gamma locus 7p15-pI4 5. NM_001669 15.98177 NPY NeuropeptideY 7p15.1 6. NM_018360 13.34037 HOXC6 HomeoboxC6 12ql2-ql3 7. AF092132 9.588665 IMPD2 IMP(inosinemonophosphate)dehydrogenase2 3p21.2 8. NM_023067 7.712272 HSPC028 HSPC028 protein 7p21.2 9. NM_014439 7.031155 LTBPI Latent transforming growth factor beta binding protein I 2 p22-p2l 10. A1613045 6.739595 GMF Glia maturation factor, beta 14q22.1 11. AB051446 6.563991 DSC2 HUMAN Q02487 DESMOCOLLIN 2A/2B PRECURSOR 18q]2.1 12. NM_005342 6.442383 TRG.TCRG Tcell receptor gamma locus 7 p 1 5 -pl4 13. D87012 6.327042 PAWR H. sapiens cDNA clone IMAGE: 1950862 12q21 14. NM_018221 6.098105 SNX2 Sorting nexin 2 Sq23 15. NM_005 114 5.769173 HS3STI Hepuran sulfate (glucosamine)-3-O-sulfotransferase I 11 16. NM_022831 5.624385 RA70, SAPS, SKAP55R Srm family associated phosphopmciin 2 7p21-pl5 17. NM_014324 5.621786 TRG,TCRG Tcell receptor gamma locus 7 pl 5 -pl4 18. NM_006820 5.550019 BICDI BicaudalD(Drosophila)homolog I 12pl.2-p1l.1 19. NM_005574 5.454622 FOLHI Folate hydrolase (prostate-specific membrane antigen) I I IpI 1.2 20. AL365343 5.451875 KIAA0615 Homo sapiens mRNA for KIAA0615 protein, complete cds. 16q 11.2 21. NM.022580 5.318270 TBCE Tubulin-specific chaperone I q42.3 22. AK022765 5.315669 CLDN8 Claudin 8 21 23. Al)67173 5.272626 P21. NSGI, D4S234 Neumn-specific protein 4p16.3 24. NM_006220 5.180025 SHMT2 HomsapienscDNA clonelMAGE:2676158 12ql2-ql4 25. AL133600 5.146792 ANK2 Homo sapiens cDNA clone by03aO8 4q25-q27 26. AY009108 5.097967 PSM PROSTATE-SPECIFIC MEMBRANE ANTIGEN 2 (HUMAN) 27. AL035603 5.076761 FL10907 Ribonuclease 6 precursor 6q27 28. NM_014017 5.058610 MAPBPIP Mitogen-activated protein-binding protein-interacting 13 protein 29. BP247098 5.030722 PHLP, Phosducin-like 9q12-q13 DKFZp564M1863 30. U62296 4.992345 GOLPH2 Golgi phosphoprotein 2 9 31. AF130082 4.988912 EST Homo sapiens clone FLC1492 PRO3121 mRNA. complete cds 32. NM.020373 4.969535 C8orf4 Chromosome 8 open reading frame 4 8 33. U90030 4.873056 BICDI Bicaudal D homolog I (Drosophila) 6 34. NM_021071 4.821960 K1AA0426 KJAA0426 gene product 6p22.2-p21.3 35. NM.030817 4.753895 KIAA1157 KIAA1157 protein 12qI3.3-q14.1 36. NM_019844 4.700642 HPRT, HGPRT Hypoxanthine phosphoribosyhransferase I Xq26.1 37. NM.004721 4.689246 RPL29 Ribosomal protein L29 3p2l.3-p21.2 38. NM_004866 4.669274 EP2, EEF-2 Eukaryotlic translation elongation factor 2 19pter-q12 39. NM_014501 4.610132 BGN Biglycan Xq28 40. NM_020655 4.575193 SDC2 Syndecan 2 (hepnras sulfate proteoglycan 1, fibroglycan) 8q22-q23 41. NM.006716 4.557526 ASK Activator of S phase kinase l9p13.l1 42. NM_002968 4.541752 FOLH I Folate hydrolase (prostate-specific membrane antigen) I I q14.3 43. X06268 4.539479 NCUBEI Non-canonical ubquitin conjugating enzyme I 6 44. AK021609 4.520464 PTH2, PTEN2, Phosphatase and tensin homolog (mutated in multiple 9p2l PSIPTEN advanced cancers 1), pseudogene I 45. NM_00I 133 4.479513 TCTEX I L T-complex-associated-testis-expressed I-like Xp2I 46. D38491 4.477160 KJAA0461, POGZ, Pogo transposable element with ZNF domain, KIAA0461 lq21.2 protein 47. NM_006426 4.385531 DDX26 Deleted in cancer 1; RNA helicase HDB/DICEl 13ql4.12 q14.2 48. AW058148 4.347362 SPHAR S-phase response (cyclin-related) Iq42.l l-q42.3 49. U55209 4.293919 MYO7A myosin VIlA (Usher syndrome IB) 4 50. NM_004610 4.275521 K1AA0634,ASTN2 Astrouactin2 9q33.1 5 53 Table 6. Top 35 Downrequlated Genes in All the 18 Samples (HR and MR) obtained from Tumor over Benign (T/B) ratio. No. Genbank T/N Ratio Common Name of the Description Map ID Genes I. X90579 0.181138 CYP3A5 Cytochrome P450, family 3, subfamily A, 7 polypeptide 5 2. NM_005213 0.198502 STF1, STFA Cystatin A (stefin A) 3q21 3. NM_005864 0.254524 EFSI, HEFS Signal transduction protein (SH3 containing) 14q1 1.2-q12 4. X15306 0.291665 NF-H H.sapiens NF-H gene, exon I (and joined CDS). 22q12.2 5. BE908217 0.319347 ANXA2 Annexin A2 15q2I-q22 6. BC001388 0.320110 ANX2, UP2, ANX2LA Annexin A2 15q21.q22 7. U22178 0.326560 PSP57, PSP94, PSP-94 Microseminoprotein. beta- 10qi 1.2 8. NM_002443 0.338948 PSP57.PSP94,PSP-94 Microseminoproleinbeta- 10ql.2 9. NM_021076 0.359039 NEPH Neurofilament, heavy polypeptide (200kD) 22q 12.2 10. A1433463 0.360636 MME, CD10, NEP, Neprilysin 3q25.I-q25.2 CALLA I1. AF275945 0.366939 EVA I Epithelial V-like antigen I 1 q23.3 12. NM_002343 0.370305 LTF Lactotransferrin 3q21-q23 13. NM_013451 0.378555 MYOF. KIAA 1207 Fer-l (C.elegans)-like 3 (myoferlin) 10q24 14. NM..001584 0.385272 239FB. DI lS302E Chromosome I I open reading frame 8 1 lpl3 15. AL390736 0.391520 BA2093 19.1, GW 112 GW 112( differentially expressed in hematopoietic lineages ) 16. NM_000441 0.392117 PDS, DFNB4 Solute carrier family 26, member 4 7q31 17. AL031602 0.399115 ESTs EST 1p34.1-35.3 18. NM_004039 0.399796 ANXA2 Annexin A2 15q21-q22 19. NM_001546 0.402261 ID4 DNA binding inhibitor protein of ID-4 6p22-p21 20. NM_001099 0.406234 PAP Acid phosphatase, prostate 3q2).q23 21. X57348 0.422692 9112 H.sapiens mRNA (clone 9112). 1 p35.2 22. NM_020139 0.440648 LOC56898 Oxidoreductase UCPA 4 23. AU148057 0.444528 DKK3, REIC Dickkopf related protein-3 precursor (Dkk-3) I lpter-pl5.5 (Dickkopf-3) (hDkk.3) 24. BF059159 0.446108 ROBOI, DUTTI, SAX3 Roundabout, axon guidance receptor. homolog I 3p12 (Drosophila) 25. BC001120 0.448109 MAC2, GALBP, MAC-2, Lectin. galactoside-binding, soluble, 3 (galectin 3) 14q21-q22 26. N74607 0.451123 AQP3 Aquqporin 3 9pl3 27. NM_013281 0.454835 FLRT3 Fibronectin leucine rich transmembrane protein 3 20p 11 28. NM_000700 0.456566 ANXI, LPCI Annexin Al 9ql2-q21.2 29. X57348 0.458169 9112 H.sapiens mRNA (clone 9112). 1p35.2 30. A1356398 0.467028 ZFP36L2, ERF-2, TISI ID EGF-respons factor 2 2 31. AF016266 0.467787 DRS. TRAILR2, Tumor necrosis factor receptor superfamily, member 8p22-p21 TRICK2A. 10b 32. S59049 0.467913 BL34, IER 1, IR20 Regulator of G-protein signalling I 1q31 33. NM_000165 0.470393 GJAI Gap junction protein, alpha l, 43kD (connexin 43) 6q21-q23.2 34. A1826799 0.471081 EFEMPI, DRAD, FBNL EGF-CONTAINNO FIBULIN-LIKE EXTRACELLULR 2pl6 MATRIx ProTN I 35. AL575509 0.476538 ETS2 V-ets erythmblastosis virus E26 oncogene homolog 2 21q22.2 (avian) 5 Classification between Tumor and Benign Prostate Epithelium: [0148] A class prediction analysis using distance based Multi Dimensional Scaling (MDS) was used to determine expression differences between tumor and benign epithelial cells in 18 patients with radical prostatectomy. All the genes that meet a minimum level 10 of expression were included in the analysis. We used the normalized intensities of all the 18 tumor and 18 normal samples for a class prediction analysis by distance based MDS to determine differentiation between tumor and benign tissue specific gene expression profile among all the 18 patients. Using a matrix of Pearson correlation coefficients from the complete pair-wise comparison of all the experiments we observed a significant overall 15 difference in gene expression pattern between the tumor and benign tissue as displayed as a two-dimensional MDS plot in Figure 2A. The position of the each tumor and benign samples is displayed in the MDS plot in two-dimensional Euclidean space with the 54 distance among the samples reflecting correlation among the samples in each individual group (distance within the cluster) and as well as reflecting distinct separation between the two groups (center-to center distance) (Figure 2A). The MDS plot was obtained from the top 200 genes obtained by 10,000 permutations of the tumor and benign intensities of 5 4566 genes. Out of these 200 genes that define the tumor specific alteration of gene expression, 53 genes had higher expression in the tumor samples and the remaining 147 genes had higher expression in the benign samples. A partial list of genes that distinctly discriminate the tumor and benign samples from all the 18 patients is shown in Table 1. We also performed a hierarchical clustering analysis using the 200 discriminatory genes. 10 The hierarchical clustering algorithm resulted in a hierarchical dendrogram that identified two major distinct clusters of 16 tumor samples and 17 benign samples (Figure 2B). Classification of CaP into HR and MR Groups Using the Ratio of Tumor over Benign Gene Expression Intensities [0149] We used the tumor over benign gene expression intensity ratio (T/B ratio) 15 (Figure 3A) from the HR (9 patients) and MR (9 patients) groups for a class prediction analysis using distance based MDS method to determine if the 18 patients can be differentiated into the two patient groups. Pathological and clinical features of the 18 tumors used in our study were clearly distinguishable between the HR and MR groups. We observed a significant overall difference in expression pattern between the HR and 20 MR groups. The distance between the samples reflects both the extent of correlation within each individual selected group (distance within the cluster) as well as distinct separation between the two selected groups (center-to-center distance) (Figure 3A). The MDS plot obtained from top 200 genes by 10,000 permutations of the 4868 genes based on the T/B ratio is shown in Figure 3A. Out of the top 200 genes of the MDS analysis 25 135 were over expressed in the HR group and 65 genes were over expressed in the MR group, The top 50 genes with best p-values identified by the T/B ratio based MDS analysis discriminating the HR and MR groups are listed in Figure 3B. The approach we used for the interpretation of discrimination between the HR and MR groups was empirical. The 'weighted list' (Figure 3B) of individual genes whose variance of change 30 across all the tumor samples defines the boundary of a given cluster to predict a class that correlates with the pathological and clinical features of CaP. We also performed a hierarchical clustering to verify the results of the MDS analysis and also to test the potential of those 200 genes to predict class/group (HR and or MR) using another 55 approach of analysis. The resulting hierarchical dendrogram of T/B ratio demonstrates that 9 samples of the HR group formed a very distinct and tight cluster, as did the 9 samples of MR group (Figure 3B). Classification of CaP into HR and MR groups based on Gene Expression Intensities in 5 Tumor Cells [0150] MDS analysis was used to determine differentiation among 18 patients into HR and MR groups. An overall difference in tumor specific expression between the HR and MR groups is displayed as a two-dimensional MDS plot (Figure 3C). The MDS plot obtained from 10,000 permutations of the gene expression intensities of 4115 genes from 10 the tumor samples of 18 patients differentiated them into HR and MR groups based on the selected top 200 genes (Figure 3C). Out of this 200 genes, 94 had higher expression in the HR groups and the remaining 106 genes had higher expression in the MR groups. We performed a hierarchical clustering analysis using the 200 discriminatory genes obtained from the supervised MDS analysis. The resulting hierarchical dendrogram of 18 tumor 15 samples demonstrates that 9 tumor samples of the HR group and 9 tumor samples of the MR group were separated into two tight clusters. (Figure 3D). The approach we utilized on the basis of the linear correlation of global gene expression in Figure 3 to obtain 'gene cluster' interpretation to discriminate the HR and MR groups was empirical. Genes that discriminate the HR and MR groups are shown in Table 7. 20 56 Table 7. Top 17 genes analysis based on T/B fold change of HR vs MR groups Gene Bank ID Conunon Name Description Map p-Value HR MR Absent Positive I NM004522 KINN. NKHC Kinesin family member SC 2q23.3 .0001 Up Down 4 30% 3 NM.018010 4IPPI, FLJUI D ypothetical protein FU 10147 3913.13 .0033 Up Down 3 56% 10 NM_012245 SKIP. NCOA- KI-interacting protein 14q24.3 10076 Up )own 2 2.80% I NM_015895 LOC51053 eminin 6p22.2 10076 Up Oown 2 71.40% 14 NM.003031 IAHI ven in absentia (Drosophila) homolog I 16q12 10076 Up )own 3 36% 42 NMOO016 ACADM cyl-Coenzyme A dehydrogenase, C-4 to C-121p3l 10178 Jp )own I 75% rightt 7 YM_025087 F1J21511 ypothetical protein FU21511 4 .0178 up )own I % 17 M.021038 MBNL Muscleblind (Drosophila)-like 3q25 10076 own Jp 2 25 4M006732 30S3 MJ murine osteosarcona viral oncogene homolog B 19q 13.32 .0076 Down Up 3 P3% 51 NM_001674 ATF3 ctivating transcription factor 3 lq32.3 10178 Down Jp 3 100% 7 M_002053 3BPI Juanylate binding protein I, interferon-inducible 'p 22

.

1 .005 Down Jp 3% 7KD 15 NM.003407 CTP, GOS24 inc finger protein 36, C3H type, homolog (mouse) 19q13.1 10076 Down Jp 1 52% 26 NM_003760 ElF403 ukaryotic translation initiation factor 4 gamma, 3 lpter-p3 .0076 DownJp 10% 38 AK023938 omo sapien ELECTED MODEL ORGANISM PROTEIN q37.3 .0178 wnJp 0% IM LARITIES 45 M_016021 CUBE) on-canonical ubquitin conjugating enzyme 1 10178 3wn Jp ? % 5 M021795 API LK4, ETS-domain protein (SRF accessory protein lq32 .005 Jp )own 4 0% 18 M_014454 PA26 53 regulated PA26 nuclear protein 3921 .0076 Jp )own 4 3% 5 Classification of CaP into High Risk and Medium Risk Groups based on Gene Expression Intensities in Benign Prostate Epithelial [0152] We used a similar MDS and Cluster analysis as in the tumor versus tumor 10 sample gene expression intensities for the normalized intensities of 9 benign samples of HR group and 9 benign samples of MR group for a class prediction. Strikingly the MDS plot of the benign samples depicted distinct separation between the HR and MR groups (Figure 3E). We observed a significant overall difference in expression pattern between the HR and MR groups. The MDS plot obtained from the top 200 genes by 10,000 15 permutations of the 3358 genes from the benign versus benign intensities (Figure 3E). Out of this 200 genes 61 were over expressed in benign samples of the HR groups and the remaining 139 genes were over expressed in the MR groups. The 'weighted list' of individual genes whose variance of expression alteration across all the normal samples depicts the capability of a given cluster to predict classification, The hierarchical 20 clustering algorithm identified a similar major cluster of the 9 benign samples of the HR group and a cluster of 9 benign samples of the MR group. 57 [0153] The weighted gene analysis by distance based supervised multidimensional scaling method we used, (depicted in Figure 3A, 3C, and 3E) utilizing the gene expression ratio of tumor and benign intensities, gene expression intensities of tumor samples and as well as normal for obtaining a 'weighted list' of individual genes, whose variance of 5 change across all the tumor and benign samples distinctly delineate the boundary of a given cluster, to predict a class that correlates with the pathological and clinical features of CaP. Independent In silico Cross Validation [0154] In silico analysis for the predicted classifier was carried out using two 10 independent data sets. The HR and MR groups were selected on the basis of Gleason score as that was the only criterion available for these data. At least 200 genes were extracted from all the MDS analysis (see methods for detail description). This subset of 200 classifier genes were found in the data of Welsh et al. 2001 and Sing et al. 2002. Exactly similar MDS analysis (p<0.001 as measured by 10,000 permutation testing) as 15 described above was performed using the expression intensities of these 200 genes from Welsh and Singh data. MDS analysis using tumor over benign ratio of as low as 50 genes from the subset of 200 genes from Welsh data (Figure 4A) as well as Singh data (Figure 4B) clearly separated samples from HR group and samples from MR group. Thus, this observation elucidates that the differential expression profile of this small set of genes can 20 be used to predict the identity or class or group of unknown prostate cancer samples on the basis of their clinico-pathological features. The outcome of this analysis depicts that the expression profile of this small number of genes is conserved across the independent data sets. Validation of GeneChip Results by Real-Time PCR 25 [0155] To further validate the expression alterations of genes identified by GeneChip analysis with an indicated biological relevance to prostate cancer, primers and probes were obtained for real-time PCR analysis using AMACR and GSTPl. These genes were chosen for validation purposes because it has been reported previously by several investigators that AMACR is elevated and GSTP1 decreased in CaP. Each sample 30 demonstrated a unique pattern of down-regulation of GSTP1 gene in 18 of 20 samples as well as up-regulation of AMACR (Figure 1) the other two samples did show significant change (fold change less than 1.5). 58 [0156] One ng of total RNA samples from paired tumor and normal specimens was reverse-transcripted using Omnisensecript RT-kit (Qiagene, Valencia, CA) according to the manufacturer's protocol. [0157] Quantitative g6ne expression analysis was performed using TaqMan Master 5 Mix Reagent and an ABI prism 7700 Sequence Detection System (PE Applied Biosystems Foster, CA). All sets of primer and probe for tested genes were Assays-on-Demand Gene expression products obtained from PE Applied Biosystems. The expression of house keeping gene, GAPDH was simultaneously analyzed as the endogenous control of same batch of cDNA, and the target gene expression of each sample was normalized to 10 GAPDH. For each PCR run, a master-mix was prepared on ice with I x TaqMan Master Mix, 1x target gene primer/probe and lx GAPDH primer/probe. Two microliters of each diluted cDNA sample was added to 28pl of PCR master-mix. The thermal cycling conditions comprised an initial denaturation step at 95*C for 10 minutes and 50 cycles at 95*C for 15 seconds and 60*C for 1 minute. RNA samples without reverse transcription 15 were included as the negative control in each assay. All assays were performed in duplicate. Results were plotted as average CT (threshold cycle) of duplicated samples. The relative gene expression level was presented as "Fold Change" of tumor versus matched normal cells, which is calculated as: Fold change - 2 crnonnacTtunor, where A CT means normalized CT value of target genes to GAPDH. 20 Example 3: Distinguishing Between ERG1 and ERG2 Isoforms [0158] The Affymetrix GeneChip probe set (213541_s_at) and TaqMan probes used in the experiments described above recognize a region specific for both ERGI and ERG2 isoforms (Figure 6), but exclude isoforms 3 to 9. Although other primers and probes could be used, by way of example, TaqMan primers and probe recognizing both 25 ERGl and ERG2, but not other ERG isoforms were as follows: Fwd primer: 5'- AGAGAAACATTCAGGACCTCATCATTATG -3' (SEQ ID NO:7) Reverse primer: 5'- GCAGCCAAGAAGGCCATCT -3' (SEQ ID NO:8) Probe: 5'-T'GTTCTCCACAGGGT-3' (SEQ ID NO:9) The probe has the reporter dye, 6-FAM, attached to the 5' end and TAMRA attached to 30 the 3' end. The 3'-TAMRA effectively blocks extension during PCR. 59 [0159] To further distinguish between these two ERG isoforms, the expression of the ERGI and ERG2 isoforms were tested in PC3 cells and in normal prostate tissue (pooled prostate RNA from 20 men, Clontech), as well as in microdissected tumor and normal prostate epithelial cells from 5 CaP patients (data not shown). Only ERGI was 5 expressed in the prostate cells and in PC3 cells. ERG2 expression was not detectable. A TaqMan QRT-PCR probe and primers were designed that specifically recognize only the ERGI isoform (Figure 6). Although other primers and probes could be used, by way of example, we designed TaqMan primers and probes recognizing only the ERG 1 isoform as follows: 10 Forward primer: 5'-CAGGTCCITCTTGCCTCCC-3' (SEQ ID NO:10) Reverse primer: 5'-TATGGAGGCTCCAATTGAAACC-3' (SEQ ID NO:l 1) Probe: 5'-TGTCTTTTATTTCTAGCCCCTTITGGAACAGGA -3'(SEQ ID NO:12). The probe has the reporter dye, 6-FAM, attached to the 5' end and TAMRA attached to the 3' end. The 3'-TAMRA effectively blocks extension during PCR. 15 [0160] ERGI expression was determined in 228 RNA specimens from microdissected matched tumor and benign prostate epithelial cells of 114 CaP patients. Overall, 62.4% of the 114 CaP patients analyzed had significant over expression of ERG 1 isoform in their tumor cells (i.e., greater than 2 fold ERG1 expression in tumor versus benign cells), while 16.6% of CaP patients had no detectable ERGI expression, 15.0% had 20 under expression of ERGI (less than 0.5 fold difference in ERGI expression in tumor versus benign cells), and 6.0% had no significant difference (0.5 to 2 fold difference in ERG1 expression between tumor versus benign cells). [0161] In a further study, ERG expression was analyzed in 82 CaP patients. Using the TaqMan primers and probes discussed above, we observed tumor-associated over 25 expression of ERGI (isoform 1 only) and ERG (isoforms 1 and 2) in 63.4% and 72.0% of the patients, respectively. Therefore, ERGI isoform specific expression may actually reflect an underestimate of the overall ERG expression in CaP. 60 Example 4: Correlation of ERG1 Expression With Various Clinico-Pathologic Features [0162] Since the ERGI tumor versus benign expression ratio data did not have normal distribution, the Wilcoxon Rank Sum Test was used to analyze its relationship 5 with various clinico-pathologic features, as shown in Table 8. Table 8 Relationship of ERG) expression ratios in tumor versus benign prostate epithelial cells with patient clinical factors Mean scores of Clinical N Median of ERG] ERG) fold P factors fold changes changes PSA recurrence No 75 142.2 52.19 0.0042 Yes 20 1.2 32.30 Tumor 0.0020 Differentiation Well & 40 362.3 57.62 Moderate Poor 54 13.9 40.00 Pathologic T 0.0136 stage pT2 38 502.0 53.45 pT3-4 52 33.5 39.69 Margin status 0.0209 Negative 64 197.0 52.55 Positive 31 20.4 38.61 Seminal 0.2555 vesicle Negative 82 106.7 49.28 Positive 13 6.9 39.92 Race 0.0086 Caucasian 73 172.1 52.08 African 22 3.8 34.45 American Family history 0.3887 No 70 106.7 49.46 Yes 25 4.8 43.92 Diagnostic PSA (ng/nl) <= 4 13 101.3 57.15 0.1801 >4- 10 62 112.1 48.03 > 10 19 20.5 39.16 Gleason sum 0.2923 <7 33 112.0 52.06 =7 45 118.1 47.16 > 7 16 21.0 39.06 61 [0163] As shown in Table 8, 95 CaP patients with detectable ERG] expression were analyzed by Wilcoxonr ank sum test. N represents the number of CaP patients falling into the indicated clinical factor category. Significant p values (<0.05) are in bold face. 5 [0164] We also found a significant correlation of high ERGI over expression with Caucasian over African American ethnicity (p = 0.0086) (Table 8). To further explore the correlation with PSA recurrence, Kaplan-Meier survival analysis was performed based on three patient groups: 1) CaP patients with tumor versus benign ERGI expression ratio of less than 2 fold; 2) CaP patients with tumor versus benign ERGI expression ratio of 2 10 100 fold; and 3) CaP patients with tumor versus benign ERG 1 expression ratio of greater than 100 fold (Figure 7). The results show that patients with higher ERGI over expression in their prostate tumor tissue had significantly longer PSA recurrence-free survival (log rank test, P=0.0006) (Figure 7). The 36-months PSA recurrence-free survival for patients with less than 2 fold ERGI expression ratio (n = 24) was 54.4%, while for patients with 15 greater than 100 fold ERGI expression ratio (n = 47) it was 87.7%. From a univariate COX proportional hazard ratio regression analysis for PSA recurrence-free time using ERGI tumor/benign cells expression ratio, race, diagnostic PSA, Gleason sum, pathologic T stage, margin status, and seminal vesicle invasion status, we found that five of these variables (ERG 1 tumor/benign cells expression ratio, Gleason sum, pathologic T stage, 20 margin status, seminal vesicle invasion) had a significant p value (Table 9). 62 Table 9 Correlation of clinical parameters and ERGI expression ratios In tumor versus benign prostate epithelal cells with PSA recurrence-free time after radical prostatectomy Factors Crude Hazard Ratio (95% CI) P ERGI fold changes 0.0024 2 - 100 fold vs. < 2 fold 0.291 (0.093 - 0.915) 0.0347 > 100 fold vs. < 2 fold 0.173 (0.060 - 0.498) 0.0011 Race Caucasian vs. African American 1.092 (0.395 - 3.016) 0.8657 Diagnostic PSA 0.8723 > 4-10 vs. <= 4 0.976 (0.275 - 3.468) 0.9705 >10 vs. <= 4 1.285 (0.307 - 5.378) 0.7313 Gleason Sum 0.0001 7 vs. 2-6 1.574 (0.393 - 6.296) 0.5215 8-10 vs. 2-6 9.899 (2.752 - 35.610) 0.0004 Pathologic T stage pT3/4 vs. pT2 6.572 (1.517 - 28.461) 0.0118 Margin status Positive vs. Negative 2.825 (1.169 - 6.826) 0.0210 Seminal Vesicle Positive vs. Negative 3.792 (1.487 - 9.672) 0.0053 5 [0165] In Table 9, crude hazard ratios with 95% confidence interval are shown for ERG1 fold change (tumor versus benign) and six clinical parameter categories in a univariate COX proportional hazard ratio analysis. Significant p values are in bold face. The multivariate COX proportional hazard ratio regression analysis of the significant 10 variables from the univariate analysis shows that ERG1 overexpression (greater than 100 fold vs. less than 2 fold: p = 0.0239, RR = 0.274, overall p value 0.0369), and Gleason sum (Gleason 8-10 vs. Gleason 2-6: p = 0.0478, RR = 4.078, overall p value 0.0148) are independent predictors of PSA recurrence after radical prostatectomy (Table 10). These results demonstrate that the status of ERGI expression ratios (tumor vs. benign) in radical 15 prostatectomy specimens carries a predictive value for patient prognosis. 63 Table 10 Factors Crude Hazard Ratio (95% Cl) P ERG 1 fold changes 0.0369 2 - 100 fold versus < 2 fold 0.320 (0.097 - 1.059) 0.0620 > 100 fold versus < 2 fold 0.274 (0.089 - 0.843) 0.0239 Gleason Sum 0.0148 7 versus 2 - 6 0.948 (0.223 - 4.033) 0.9424 8-10 versus 2 - 6 4.078 (1.014 - 16.401) 0.0478 Pathologic T stage PT3/4 versus pT2 3.306 (0.636-17.177) 0.1550 Margin status Positive versus Negative 1.116 (0.421-2.959) 0.8254 Seminal Vesicle Positive versus Negative 1.308 (0.466-3.670) 0.6098 [0166] ERGI expression in prostate tumor tissue showed highly significant association with longer PSA recurrence free survival (p = 0.0042), well and moderately differentiated grade (p = 0.0020), lower pathologic T stage (p = 0.0136), and negative 5 surgical margin status (p = 0.0209), suggesting that ERG1 over expression in tumor cells is generally higher in less aggressive CaP than in more aggressive CaP (Table 8). [0167] The ERGI over expression in tumor cells identified by GeneChip analysis and verified by real time QRT-PCR assays was further validated by in situ hybridization. Based on the real time QRT-PCR data, 6 patients with high ERGI over expression in their 10 tumor cells (and as a control one patient with no ERGI over expression) were selected for in situ hybridization and quantitative image analysis in a blinded fashion. As expected, in each case the in situ expression data confirmed the over expression of ERGI in the tumor epithelial cells (Figure 8). Example 5: Generation and Characterization of ERG antibody 15 Cloning of ERG] into Tetracycline regulated mammalian expression vectors: [0168] ERGI cDNA was subcloned into tetracycline-regulated mammalian expression vectors (pTet-off, EC1214A). The constructs generated include, pTet-off ERGI (sense), pTet-off-ERG1 (antisense), pTet-off-FlagERG1 (sense) and pTet-off FlagERGI (antisense). Originally, ERGI construct in a riboprobe vector pGEM was 64 obtained from Dr. Dennis K. Watson, Medical University of South Carolina. The constructs were verified by dideoxy sequencing and agarose gel analysis. Generation of polyclonal ERG antibody: [0169] Antibodies against ERG were generated using peptide antigens derived 5 from the full length ERGI coding sequence. The epitope for the antigen were carefully selected such that the antibody recognizes specifically ERG1/2/3 and not other members of the ETS family (Figure 9). The following peptides, having the highest hydrophilicity (-1.26 and -0.55) and antigenicity in the desired region, were used to generate antibodies: [01701 Peptide M-50-mer: 10 CKALQNSPRLMHARNTDLPYEPPRRSAWTGHGHPTPQSKAAQPSPSTVPK-[NH 2 ] (SEQ ID NO:13) [0171] Peptide C-49-mer: CDFHGIAQALQPHPPESSLYKYPSDLPYMGSYHAHPQKMNFVAPHPPAL (SEQ ID NO:14) 15 [0172] Cysteine was added to each peptide for conjugation. Peptide M is amidated at the C-terminal residue because it is an internal pepetide. [0173] The synthesis of the peptide epitopes and the immunization of rabbits were carried out in collaboration with Bio-Synthesis Inc. Two rabbits were immunized for each of the two epitopes. Bleeds obtained post immunization were collected and tested. 20 Subsequently, bleeds from one of the rabbits from each epitope were affinity purified using SulfoLink kit (Pierce) and were verified by immunoblot analysis. Characterization of polyclonal ERG antibody by immunoblot analysis: [0174] To characterize the affinity purified antibody, we transiently transfected HEK-293 (Human embryonic kidney cell line, ATCC, Manassas, VA) with ERGI 25 constructs pTet-off-ERG1 (sense) and pTet-off-FlagERG1 (sense) using Lipofectamine reagent (Invitrogen, Carlsbad, CA) as per manufacturers instructions. HEK-293 that were not transfected with the plasmid served as a transfection control. The cells were harvested 48 hours post-transfection and processed for immunoblot analysis. Expression of ERGI following transfection was determined by immunoblotting using the affinity purified 65 polyclonal antisera generated against the unique M- and C-ERG epitopes described above. Endogenous ERGI expression was not detected in non-transfected HEK-293 cells. However, the ERG antibodies detected ERGI expression in HEK-293 cells transfected with the various ERGI constructs. Tetracycline (2 ug/ml) abolished ERGI expression in 5 both tetracycline-regulated constructs, pTet-off-ERG1 (sense) and pTet-off-FlagERG1 (sense). The M2-Flag antibody specifically recognized only the Flag-tagged ERGI protein. Example 6: Combined Expression of ERG, AMACR, and DD3 Genes in Prostate Tumors 10 [0175] The strikingly high frequency of ERG over expression in CaP cells led to a comparison of ERG expression with two other genes, AMACR and DD3, that are also over expressed in CaP cells. We have evaluated quantitative gene expression features of AMACR and DD3, along with the ERG gene, in laser imicrodissected matched tumor and benign prostate epithelial cells from 55 CaP patients. 15 (0176] Although other primers and probes can be used, by way of example, we designed the following TaqMan primers and probe recognizing the DD3 gene: Forward primer: 5'-CACATTTCCAGCCCCTTTAAATA-3' (SEQ ID NO:15) Reverse primer: 5'-GGGCGAGGCTCATCGAT-3' (SEQ ID NO:16) Probe: 5'-GGAAGCACAGAGATCCCTGGGAGAAATG-3' (SEQ ID NO:17). 20 The probe has the reporter dye, 6-FAM, attached to the 5' end and TAMRA attached to the 3' end. The 3'-TAMRA effectively blocks extension during PCR. [0177] AMACR TaqMan primers and probe were purchased from Applied Biosystems. [0178] AMACR and DD3 showed upregulation in tumor cells of 78.2% and 87.3% 25 of CaP patients, respectively (Figure 5). ERG over expression in tumor cells was detected in 78.2% of the same group of CaP patients (Figure 5). Comparative expression analysis revealed that when the AMACR and ERG expression data are combined, 96.4% of the CaP patients showed upregulation of either of the two genes in tumor cells (Figure 5). Similarly, the combination of the ERG and DD3 expression data improved the cancer 30 detection power of either of the genes to 96.4% (Figure 5). When combining the 66 expression data from all the three genes, 98.2% of the CaP patients showed upregulation of at least one of the three genes in tumor cells (Figure 5). Thus, screening for ERG gene expression, alone, or in combination with other genes that are over expressed in CaP, such as AMACR and DD3, provides a new, powerful diagnostic and prognostic tool for CaP. 5 Example 7: Under Expression of LTF in Malignant Prostate Epithelium [0179] One of the most consistently under expressed genes in CaP cells was LTF (Table 1). Validation by QRT-PCR (TaqMan) in LCM-derived tumor and benign prostate epithelial cells confirmed a consistent, tumor associated LTF under expression in 100% of CaP cells tested (Figure ID). As a quality control, the expression of AMACR, a recently 10 identified CaP tissue marker, and of GSTPI, a gene showing commonly reduced or absent expression in CaP (Nelson et al., Ann. N.Y. Acad. Sci., 952:135-44 (2001)), was also determined (Figures IB and IC, respectively). Robust under expression similar to LTF, was observed for GSTP1, while the increased expression of AMACR was noted in 95% of the tumor cells tested, confirming the high quality of the tumor and benign LCM 15 specimens and the reliability of the QRT-PCR. In a further study, LTF expression was analyzed by QRT-PCR in in microdissected tumor and benign prostate epithelial cells of 103 CaP patients. The results were consistent with the initial results, showing tumor associated under expression in 76% of patients (78 of 103). [0180] LTF under expression was also validated at the protein level with anti-LTF 20 goat polyclonal antibody (Santa Cruz, CA, sc-14434) using Western blot analysis on protein lysates and immunohistochemistry techniques. Hematotoxylin-eosin (H&E) and LTF staining was performed on tissue samples from 30 CaP patients by immunocytochemical analysis. In 30 of 30 (100%) cases, benign epithelial cells adjacent to tumor cells were highly positive for LTF, whereas, on average, less than 10% of 25 prostate tumor cells revealed LTF positive cytoplasmic staining. [0181] The specification is most thoroughly understood in light of the teachings of the references cited within the specification which are hereby incorporated by reference. The embodiments within the specification provide an illustration of embodiments of the invention and should not be construed to limit the scope of the invention. The skilled 30 artisan readily recognizes that many other embodiments are encompassed by the invention. 67

Claims (27)

1. A method of diagnosing or prognosing prostate cancer in a subject, comprising: a) measuring the expression level of an ERG polypeptide in a biological sample from the subject; and b) -correlating the expression level of the ERG polypeptide with the presence of prostate cancer in the subject or a higher predisposition of the subject to develop prostate cancer, wherein the expression level of the ERG polypeptide is correlated with longer PSA recurrence free survival, well and moderate tumor differentiation, a pathologic T stage of pT2 or lower, or a negative surgical margin status.

2. The method of claim 1, wherein the expression level of the ERG polypeptide is measured by an immunoassay using an antibody that binds to the ERG polypeptide.

3. The method of claim 1 or claim 2, wherein the ERG polypeptide is a human ERG polypeptide.

4. The method of claim 2 or claim 3, wherein the antibody is generated using a fragment of the human ERG polypeptide as an immunogen.

5. The method of any one of claims 2 to 4, wherein the antibody binds to the ERGI, ERG2, and ERG3 polypeptides.

6. The method of any one of claims 2 to 5, wherein the antibody is a monoclonal antibody.

7. The method of any one of claims 2 to 6, wherein the antibody binds to an epitope formed by amino acids 275-322 of the polypeptide encoded by a nucleic acid comprising the sequence of SEQ ID NO:1. 68 520954801 (GHMaflers) P2730.AUA 18-Mar-14

8. The method of any one of claims 2 to 6, wherein the antibody binds to an epitope formed by amino acids 118-166 of the polypeptide encoded by a nucleic acid comprising the sequence of SEQ ID NO:1.

9. The method of any one of claims I to 8, wherein the biological sample is chosen from a tissue sample, a blood sample, or a urine sample.

10. The method of any one of claims I to 9, wherein the biological sample is a prostate tissue sample.

11. The method of any one of claims I to 10, further comprising measuring the expression levels of an AMACR polypeptide and correlating the expression levels of the ERG polypeptide and the AMACR polypeptide with the presence of prostate cancer in the subject or a higher predisposition of the subject to develop prostate cancer.

12. The method of claim 11, wherein the AMACR polypeptide is encoded by a nucleic acid comprising the sequence of SEQ ID NO. 3.

13. The method of any one of claims I to 12, wherein the expression level of the ERG polypeptide in the biological sample is compared to the expression level of the ERG polypeptide in a control sample and wherein increased expression of the ERG polypeptide in the biological sample as compared to the expression of the ERG polypeptide in the control sample indicates the presence of prostate cancer in the subject or a higher predisposition of the subject to develop prostate cancer.

14. The method of claim 13, wherein the control sample is a noncancerous biological sample from the subject.

15. The method of claim 11 or claim 12, wherein the expression levels of the ERG polypeptide and the AMACR polypeptide in the biological sample are compared to the 69 5209546J (rHMtrs) P6230AU.1 184-W-14 expression levels of the ERG polypeptide and the AMACR polypeptide in a control sample and wherein increased expression of the ERG polypeptide and the AMACR polypeptide in the biological sample as compared to the expression of the ERG polypeptide and the AMACR polypeptide in the control sample indicates the presence of prostate cancer in the subject or a higher predisposition of the subject to develop prostate cancer.

16. The method of any one of claims I to 10, wherein the expression level of the ERG polypeptide is used to indicate or predict the pathologic stage of prostate cancer.

17. The method of claim 2, wherein the immunoassay is an immunohistochemistry assay, a Western blot, a radioimmunoassay, an ELISA, an immunoprecipitation assay, an immunodiffusion assay, an agglutination assay, a complement-fixation assay, an immunoradiometric assay, a fluorescent immunoassay, or a protein A immunoassay.

18. The method of any one of claims I to 17, wherein the ERG polypeptide is an ERGI or ERG2 polypeptide.

19. A method of identifying prostate cancer in a subject, comprising: a) providing a prostate tissue sample from the subject; b) contacting the prostate tissue sample with an antibody that binds to a human ERG polypeptide, wherein the antibody is a monoclonal antibody that binds to the ERG], ERG2, and ERG3 polypeptides; c) measuring the level of expression of the human ERG polypeptide in the prostate tissue sample by detecting the amount of binding between the antibody and the human ERG polypeptide in the prostate tissue sample; d) comparing the level of expression of the ERG polypeptide in the prostate tissue sample to the level of expression of the ERG polypeptide in a control sample; and e) identifying prostate cancer in the prostate tissue sample from the subject if the level of expression of the ERG polypeptide in the prostate tissue sample is increased as compared to the expression level of the ERG polypeptide in the control sample. 70 S2085461 (GHMagers) P62730AU.1 19-Mar-14

20. The method of claim 19, wherein the antibody binds to an epitope formed by amino acids 275-322 of the polypeptide encoded by a nucleic acid comprising the sequence of SEQ ID NO:1.

21. The method of claim 19, wherein the antibody binds to an epitope formed by the amino acid sequence of SEQ ID NO:14.

22. The method of any one of claims 19 to 21, wherein the control sample is a noncancerous biological sample from the subject.

23. The method of claim 22, wherein the noncancerous biological sample from the subject is normal prostate tissue comprising benign prostate epithelial cells.

24. The method of any one of claims 19 to 23, wherein the ERG polypeptide is an ERGI or ERG2 polypeptide.

25. A monoclonal antibody, wherein the monoclonal antibody binds to an epitope formed by amino acids 275-322 of the polypeptide encoded by a nucleic acid comprising the sequence of SEQ ID NO:1.

26. A monoclonal antibody, wherein the monoclonal antibody binds to an epitope formed by the amino acid sequence of SEQ ID NO:14.

27. The method of claim 1 or claim 19, or the antibody of claim 25 or claim 26, substantially as hereinbefore described with reference to the examples and figures. 71

52096460.. (GHv~ate,,) P6730.AL.1 l5-Mat-14