WO2004033641A2 - Gene expression profiling of bladder cancer - Google Patents

Abstract

Gene expression profiling disclosed herein classified bladder tumors based on their histopathogenesis and clinical outcome. Both clusters and individual target genes were identified that would provide novel means of molecular diagnosis and outcome prediction for patients with bladder cancer.

Description

GENE EXPRESSION PROFILING OF BLADDER CANCER

Cross-reference to Related Applications

This non-provisional patent application claims benefit of priority of provisional patent applications 60/416,002 and 60/416,003, both filed October 4, 2002 and now abandoned.

Federal Funding Legend

This invention was supported in part by National Cancer Institute Grant CA-47538 and National Institutes of Health Grant NS39662. Consequently, the federal government has certain rights in this invention.

BACKGROUND OF THE INVENTION

Field of the Invention

The present invention relates generally to the field of cancer research. More specifically, the present invention provides gene expression profiling for bladder cancer.

Description of the Related Art

Bladder cancer is one of the most common malignancies in developed countries, ranking as the sixth most frequent neoplasm. Bilharzial-related bladder carcinoma (BBC) is the most common malignant neoplasm in Egypt, occurring also with a high incidence in other regions of the Middle East and East Africa. Certain clinical and pathological features of bilharzial-related bladder carcinoma are different than those described for the conventional transitional cell carcinoma, such as the high incidence of detecting squamous metaplasia and the development of squamous cell carcinoma. Transitional cell carcinoma has been classified into two groups with distinct behavior and different molecular profiles: low grade tumors (always papillary and usually superficial), and high-grade tumors (either papillary or non-papillary, and often invasive). The inactivation of both RB and p53 pathways has been shown to be required for the transformation and immortalization of uroepithelial cells, and their alterations are common and of predictive nature in clinical studies of bladder cancer. Cross-talk between these pathways and adhesion signaling, such as those generated by cadherin-catenin complexes, have been described to be involved in bladder cancer progression.

In the post genome era, and in view of the advent of high-throughput methods of molecular analysis, it is expected that specific tumor types will have distinct gene expression profiles. The elucidation of the molecular events involved in tumorigenesis and tumor progression is directly leading to the discovery and application of novel biological markers. The diagnosis and prognosis of certain neoplasms are in many cases enhanced by the use of such markers, and the marker itself may constitute a therapeutic target. There is a need in the art for methods of gene expression profiling for bladder cancer and uses thereof. The present invention fulfills this long-standing need and desire in the art.

SUMMARY OF THE INVENTION

In the present invention, bladder cancer was characterized and new targets involved in bladder tumor progression were validated using a combination of cDNA and tissue microarray technologies. This study was designed to characterize the expression profiles of nine bladder cancer cell lines (T24, J82, 5637, HT-1376, RT4, SCaBER, TCCSUP, UMUC-3, and HTl 197) using cDNA microarrays (8976 genes and ESTs). Novel targets of potential clinical relevance involved in bladder cancer progression were validated by immunohistochemistry using tissue microarrays of primary bladder tumors. Hierarchical clustering classified uroepithelial cells based on their histopathogenesis and cell cycle alterations. Keratin 10 and caveolin-1 transcripts were more abundant in tumor cells from squamous and invasive origin. Their combined expression was shown to stratify bladder rumors and define squamous differentiation. To assess the robustness of the clustering analysis, a bootstrap resampling technique was used. This grouped tumor cell lines based on their biological properties, including cell cycle and cell adhesion features. E-cadherin, zyxin, and moesin were identified as genes differentially expressed in these clusters and related to the p53, RB and INK4A status of the cell lines. Loss of these adhesion molecules was associated with stage and grade in primary tumors, and moesin expression was also associated with survival. Deregulation of cell cycle and apoptotic pathways, such as mutations or altered expression of p53, pRB and INK4A (pl6), are necessary for uroepithelial transformation. However, it appears that deregulation of cell adhesion is a common event associated with tumor progression in uroepithelial neoplasms.

Two main clusters segregating superficial from invasive transitional carcinomas were also identified that could provide prognostic information. Cytokeratin 20, neuropilin 2, p21 and p33INGl were selected among the top ranked molecular targets differentially expressed between superficial and invasive tumors and validated by immunohistochemistry with tissue microarrays. Their expression patterns were associated with pathological stage, tumor grade, and altered RB expression. Moreover, p33INGl expression levels were related to overall survival. Generation of a support vector machine algorithm revealed the relevance of WNT signaling and mitotic checkpoint alteration during bladder cancer progression. In summary, gene profiling successfully classified bladder tumors based on their histopathogenesis and clinical outcome, and identified molecular biomarkers of potential clinical significance.

Other and further aspects, features, and advantages of the present invention will be apparent from the following description of the presently preferred embodiments of the invention. These embodiments are given for the purpose of disclosure.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows the hierarchical clustering of bladder cancer cell lines by overexpressed genes displaying red to green ratios higher than 2.0 in at least one experiment obtained by standard clustering analysis. Using this approach, tumor cell lines were classified according to histopathogenesis criteria. SCaBER cells derived from a squamous carcinoma of the bladder were separated from cells derived from conventional transitional carcinomas. Tumor cell lines derived from invasive tumor samples were separated from cells obtained from metastatic (TCCSUP) and superficial (RT4) bladder tumors.

Figure 2 shows a representative immunostaining patterns of caveolin- 1 and keratin 10 in normal urothelium, squamous metaplasia, transitional and squamous bladder tumors. Caveolin-1 expression was undetectable in normal urothelium (A), and most transitional carcinomas of the bladder (E); however, caveolin-1 levels were detected in areas of squamous metaplasia (C), as well as transitional carcinomas with squamous differentiation and squamous cell carcinomas of the bladder (G). Similarly, keratin 10 expression was undetected in normal urothelium (B), and most transitional carcinomas (F); however, keratin 10 levels were identified in squamous metaplasia (D), as well as in transitional carcinomas with squamous differentiation and squamous cell carcinomas of the bladder (H). Note the different patterns of caveolin-1 and keratin 10. Caveolin-1 was expressed in basal cell layers in the squamous metaplasia and areas of squamous carcinoma, whereas keratin 10 was identified in suprabasal layers. There was a significant difference regarding the expression of these proteins in tumors with squamous differentiation versus those with pure transitional cell carcinoma features (p<0.001). (Original magnifications: A through H, x200)

Figure 3 shows the robustness of the clustering analysis, a bootstrap resampling technique was applied. First, a large number (1000 in this analysis) of copies of the data were generated using a Monte Carlo resampling technique. Each of these generated datasets was then clustered using the standard hierarchical method, namely dot product (angle) metric and ward linkage. The 1000 resulting trees were then used to build a consensus tree using the CONSENS program from the PHYLIP package. The output consists of a count at each node of the tree that represents how many of the 1000 trees had that bipartition. Nodes with values close to 1000 are more significant than others scoring lower values. The higher the number at each node of the tree, the more similar the expression patterns of the cells within clusters are. Tumor cell lines displaying similar alterations in the TP53, RB and ARF/pl6 pathways grouped together within significant clusters. In the bottom part, it is shown the logarithmic ratio of the expression of these three genes in each of the cell lines. Positive and negative numbers mean higher and lower expression of these genes respectively among the cells.

Figure 4 shows a representative immunostaining patterns of zyxin, moesin and E-cadherin in primary bladder tumors. Superficial tumors showed high levels of E-cadherin (A), zyxin (B), and moesin (C) expression. However, invasive bladder neoplasms were found to express low to undetectable levels of these proteins [E-cadherin (D), zyxin (E), and moesin (F)]. There was a significant difference regarding the expression of these proteins with histopathological stage and tumor grade (p<0.005). (Original magnifications: A, B, and C, x200; D, E, and F, x400).

Figure 5 shows the survival analysis for patients with bladder tumors stratified by moesin expression treated as a categorical variable. The Kaplan-Meier method was used to estimate overall disease-free survival; log-rank analysis was utilized to compare the curves. Detection of moesin in the membrane of tumor cells from primary tumor samples was found to be significantly associated with overall survival in this subset of 67 bladder cancer patients (median follow up time: 36 months) (ρ=0.01). Figure 6A shows the hierarchical clustering using bootstrap resampling techniques classified bladder tumors according to histopathological criteria. A tree is constructed by finding for each node the pairing that occurred most often in the 1000 separate trials displaying this count at each node of the tree. The number on each node represents how many times that samples to the right are grouped together out of a total of 1000 tries, a larger number indicates tight clustering. S: superficial bladder tumors, I: organ-confined invasive bladder tumors, I/M: invasive bladder tumors developing metastatic disease.

Figure 6B shows the Kaplan-Meier survival analysis of patients with superficial (cluster 1) and invasive (cluster 2) bladder tumors stratified by bootstrap clusters. Clusters containing the superficial and invasive tumors were significantly associated with overall survival (Log Rank p=0.0025).

Figure 6C shows a multidimensional analysis: Four groups of expression profiles were identified by factor analysis and these were consistent with the superficial (groups 1 and 3) and invasive (groups 2 and 4) clusters.

Figure 6D shows the multidimensional analysis: the expression profiles of certain superficial tumors (163, 165, 169) were more similar to some organ-confined invasive lesions.

Figures 7A-B shows epresentative examples of the staining evaluation of p33INGl between superficial and invasive bladder tumors. P33ING1 nuclear expression was high in superficial transitional carcinomas of the bladder (A). However, p33INGl expression levels were lower in invasive bladder tumors (B). There was a significant difference regarding the expression of this protein in tumors regarding stage and grade in the subset of bladder cancer patients analyzed (p<0.0005). (Original magnifications: x400).

Figure 7C shows the Kaplan-Meier survival analysis of patients with bladder tumors stratified by the expression of p33INGl, one of the biomarkers identified in the study. p33INGl was found to be significantly associated with overall survival in the subset of 69 bladder tumors (median follow-up time: 36 months)

(p=0.02)

Figure 8 shows a cluster analysis based on the generated Support Vector Machine algorithm. WNT and mitotic spindle checkpoint were revealed among the altered pathways during bladder cancer progression. Relevant genes related to several networks are highlighted: p53/apoptosis (blue), kinetochore/spindle checkpoint signaling/anaphase promoting complex/cell cycle (magenta), actin polymerization/cell polarity/cell migration (orange), and vesicle budding/vesicular transport/cell adhesion (green).

DETAILED DESCRIPTION OF THE INVENTION

Large-scale transcript profiling of individual bladder tumors using cDNA array analysis disclosed herein contributed to a biologically oriented classification of bladder cancer. The combination of cDNA and tissue microarrays facilitated validation of known and novel targets of potential clinical significance. Both clusters and individual targets provided novel means of molecular diagnosis and outcome prediction for patients with bladder cancer. Overall, gene profiling classified bladder tumors based on their histopathogenesis and clinical outcome. Two major sets of experiments were conducted. Initially, the inventors used bladder cancer cell lines and cDNA microarrays to identify differentially expressed genes between distinct histopathological tumor types and stages of the disease. In a second approach, tissue microarrays were used to validate the potential clinical significance of the targets identified by cDNA microarrays at the microanatomical detail using immunohistochemistry on clinical material. A cohort of superficial and invasive bladder neoplasms was used to evaluate the association between molecular targets and histopathological variables including stage and grade. An additional tissue microarray, containing bladder tumors with characterized p53 and pRB alterations and annotated follow-up, was used to delineate associations between molecular markers and these critical pathways, as well as with clinical outcome.

In summary, the present invention identifies clusters and/or individual target genes that would provide novel means of molecular diagnosis and outcome prediction for patients with bladder cancer. The identified genes may also be targets for therapeutic intervention in treating bladder cancer.

In the present invention, there is provided a method of diagnosis for squamous metaplasia of bladder cancer based on the expression of caveolin- 1 or keratin 10. As presented below, caveolin-1 and keratin 10 are markers of squamous differentiation. Normal urothelium and superficial conventional transitional cell carcinoma (cTCC) had undetectable levels of both caveolin-1 and keratin 10. However, areas of squamous metaplasia and carcinoma identified in bilharzial-releted bladder carcinoma, as well as areas of squamous differentiation identified in cTCC, had significant expression of both proteins.

In another aspect of the present invention, there is provided a method of diagnosis for bladder cancer based on the expression level of zyxin, E-cadherin, moesin, cytokeratin 20, neuropilin 2, p21 or p33INGl. As shown below, the expression levels of these proteins correlate with the stage and grade of bladder cancer in individual patient.

In yet another aspect of the present invention, there is provided method of discriminating between superficial and invasive bladder cancer in an individual based on differential expression of zyxin, E-cadherin, moesin or p33INGl. Data presented below indicate that superficial tumors express high levels of zyxin, E- cadherin, moesin or p33INGl, whereas invasive neoplasms express low to undetectable levels of these proteins.

The present invention also provides a method of predicting survival outcome for a bladder cancer patient based on the level of expression of p33INGl. The level of p33INGl expression was shown to correlate with survival outcome of bladder cancer, i.e. patients exhibiting a higher expression of p33INGl showed a shorter survival time than those with low expression of this protein.

The present invention is also directed to a method of discriminating between superficial and invasive bladder cancer in an individual, comprising the steps of: collecting biological samples from said individual; and determining in said samples the expression of a gene identified by an accession number selected from the group consisting of AA01 1414, AA021434, AA021464, AA028884, AA034115, AA035095, AA043806, AA074666, AA083385, AA101348, AA127058, AA132065, AA143509, AA147928, AA156863, AA165403, AA172210, AA190401, AA256462, AA279188, AA394148, AA402766, AA421518, AA424578, AA425861, AA430520, AA434068, AA446453, AA447696, AA449831, AA450227, AA450265, AA454566, AA454862, AA455150, AA455281, AA456136, AA457092, AA457162, AA457725, AA458661, AA459663, AA464152, AA464192, AA465031 , AA465378, AA465593, AA478268, AA485052, AA486313, AA486374, AA486761, AA487020, AA487223, AA487265, AA487899, AA489400, AA490047, AA490390, AA496359, AA496784, AA496948, AA504128, AA504617, AA598759, AA598815, AA620479, AA625981, AA629584, AA633757, AA669341, AA680322, AA682613, AA683085, AA705886, AA775415, AA862434, AA934762, AA935560, AI017703, H05769, H17158, H20652, H21040, H23366, H23880, H54093, H73731, H84444, H93463, H94897, H99502, N30811, N54338, N69283, N73536, N91962, R16165, R22439, R24543, R25377, R27552, R49144, R53889, R55763, R69307, R76314, R78514, T53404, T57815, T67053, T81091, T96829, W49619, W69906, W96107, AA010393, AA019591, AA024832, AA1 13339, AA1 15248, AA121704, AA134595, AA142875, AA157797, AA165400, AA284268, AA284292, AA404694, AA406603, AA421783, AA424834, AA429399, AA431 184, AA435936, AA436158, AA436871, AA443193, AA443285, AA453607, AA453748, AA454579, AA454625, AA4551 19, AA457374, AA459950, AA460365, AA463958, AA482325, AA488526, AA488645, AA489246, AA489661 , AA496780, AA504894, AA599093, AA609067, AA609134, AA621335, AA705060, AA708310, H09747, H09818, H10335, H17335, H23277, H29292, H41096, H53141, H58736, H65834, H70815, H93463, H95989, N21548, N38891, N56882, N58283, N66933, N94428, R08891, R09585, R22271, R28669, R36449, R43525, R44132, R51080, R56219, R56432, R60053, R60927, R64066, R92455, R94943, R98628, R99918, T50370, T55592, T61792, T68461, T71680, T86983, T90641, T96711, W31919, W56308, W99364, wherein said gene is differentially expressed at the mRNA level in superficial and invasive bladder cancer.

The present invention if further directed to a method for identifying the presence or absence of a squamous metaplasia of bladder cancer phenotype in a cell or cells, comprising determining the expression level of caveolin-1 or keratin 10 in said cell or cells, wherein a detectable expression level of caveolin-1 or keratin 10 in the cell or cells indicates the presence of squamous metaplasia of bladder cancer phenotype and an undetectable level of caveolin-1 or keratin 10 in said cell or cells indicates the absence of squamous metaplasia of bladder cancer phenotype.

In another aspect, the present invention provides a method of identifying the presence or absence of a squamous metaplasia of bladder cancer in an individual, comprising the steps of: collecting a biological sample from said individual; and determining the expression level of caveolin-1 or keratin 10 in said sample, wherein a detectable expression level of caveolin-1 or keratin 10 in said sample indicates the presence of said squamous metaplasia of bladder cancer and an undetectable level of caveolin-1 or keratin 10 in said sample indicates the absence of said squamous metaplasia of bladder cancer. Preferably, the expression level is a protein expression level or a nucleic acid expression level. In another aspect, the present invention provides a method of identifying the presence or absence of a bladder cancer in an individual, comprising the steps of: collecting a biological sample from said individual; and determining in said sample the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin, cytokeratin 20, neuropilin 2, p21 and p33INGl, wherein the level of expression indicates the presence or absence of a bladder cancer in the individual. In another specific aspect, this method may further comprise correlating the expression level with the stage and grade of bladder cancer in the individual. Preferably, the expression of the protein is determined at protein level or the expression of the protein is determined at nucleic acid level.

In another aspect, the present invention provides a method of discriminating between a superficial and an invasive bladder cancer in an individual, comprising the steps of: collecting a biological sample from said individual; and determining in said sample the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin and p33INGl, wherein said protein is differentially expressed in superficial and invasive bladder cancer. Preferably, the level of expression is a level of expressed protein or the level of expression is a level of nucleic acid expression.

In another aspect, the present invention provides a method of predicting survival outcome of an individual having bladder cancer, comprising the steps of: collecting biological samples from said individual; and determining in said samples the level of expression of p33INGl, wherein said level of expression correlates with survival outcome of said individual. Preferably, the expression of said protein is determined at protein or nucleic acid level. In another aspect, the present invention provides a kit for identifying the presence or absence of a squamous metaplasia of bladder cancer phenotype in a cell or cells, comprising a reagent or reagents capable of determining the expression level of caveolin-1 or keratin 10 in said cell or cells, wherein detectable expression levels of caveolin-1 or keratin 10 in said samples indicates the presence of squamous metaplasia of bladder cancer phenotype and undetectable levels of caveolin-1 or keratin 10 in said cell or cells indicates the absence of squamous metaplasia of bladder cancer phenotype. In another aspect, the present invention provides a kit for identifying the presence or absence of bladder cancer in an individual, the kit comprising a reagent or reagents capable of determining the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin, cytokeratin 20, neuropilin 2, p21 and p33INGl.

In another aspect, the present invention provides a kit for discriminating between superficial and invasive bladder cancer in an individual, the kit comprising a reagent or reagents capable of determining the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin and ρ33INGl.

In another aspect, the present invention provides a kit for predicting survival outcome of an individual having bladder cancer, the kit comprising a reagent or reagents capable of determining in the samples the level of expression of p33INGl. In all embodiments of the kit of the present invention, it is contemplated that the reagent could be an antibody. Alternatively, in all embodiments of the kit of the present invention, it is contemplated that the one reagent could be a nucleic acid. In the embodiments of the kit of the present invention, it is contemplated that the it may further compriseinstructions for correlating the level of expression with a clinical diagnosis. The following examples are given for the purpose of illustrating various embodiments of the invention and are not meant to limit the present invention in any fashion. The present examples, along with the methods, procedures, treatments, molecules, and specific compounds described herein are presently representative of preferred embodiments. One skilled in the art will appreciate readily that the present invention is well adapted to carry out the objects and obtain the ends and advantages mentioned, as well as those objects, ends and advantages inherent herein. Changes therein and other uses which are encompassed within the spirit of the invention as defined by the scope of the claims will occur to those skilled in the art.

EXAMPLE 1

Molecular Profiling of Bladder Cancer Using cDNA Microarrays Defines Histogenesis And Biological Phenotypes

Expression profiling classified nine bladder cancer cell lines under study based on histopathological characteristics of the tumors from which they were obtained. Keratin 10 and caveolin-1 expression was associated with the presence of squamous differentiation, as well as with pathological stage and tumor grade. The application of bootstrapping techniques to hierarchical clustering grouped most of the bladder cancer cell lines based on their alterations in p53 and RB pathways. Target genes identified herein, namely zyxin, E-cadherin and moesin, were associated with p53 and/or pRB alterations in the primary bladder tumors analyzed. Identified target genes obtained from high-throughput molecular profiling of cultured cells were shown to have clinical impact when validated in primary bladder tumors using tissue microarrays. These target genes were significantly associated with bladder cancer progression; and moesin provided predictive outcome information.

Cell Culture And RNA Extraction

Nine bladder cancer cell lines including T24, J82, 5637, HT1376, RT4, SCaBER, TCCSUP, UMUC-3, and HTl 197 were obtained from ATCC (Rockville, MD) and cultured under identical conditions following standard procedures. All cells were grown and harvested at 75%-90% confluence no longer than 4-6 passages in culture for the extraction of total RNA using RNeasy protocol (Qiagen, Valencia, CA). Cytospins were also prepared and later used for target validation.

Preparation of cDNA Microarrays And Image Acquisition

A set of 8976 sequence-verified human IMAGE cDNA clones, representing both known genes and expressed sequence tags (ESTs), were PCR amplified and spotted onto poly-lysine coated microscope slides using a custom robot designed and built at Albert Einstein College of Medicine (Cheung et al., 1999).

Ten μg of total RNA of each cell line was labeled with Cy5 (red) and hybridized against lOμg of total RNA of a pool containing equal RNA quantities of all of these cell lines labeled with Cy3 (green). Labeling and hybridization of cDNA to arrays was carried out as previously described (Stears et al., 2000). One duplicate or one reverse- labeling experiment was carried out for validation of expression changes of the hybridization of the cell lines. Following hybridization, slides were washed, dried and scanned by a custom-built laser scanner (Cheung et al., 1999). Intensity data were integrated with 8x oversampling (Cheung et al., 1999). Scanalyse software was used for gridding and calculation of red (R) and green (G) signal intensities (Eisen et al., 1998).

Collection And Analysis of The Data of The cDNA Microarrays

Normalization: Before any analysis, plots of the fold change versus the average intensity were examined to look for abnormalities in single-array data. It is common to plot a red versus green channel scatter plot to examine distribution of intensities; however, the inventors found that transforming to fold change versus average intensity displayed the data in a more easily viewed form. If Ired is the background subtracted red channel intensity, and Igreen is the background subtracted green intensity, then the following variables were created: R=Ired/Igreen and A= (Ired x Igreen), where R is simply the fold change ratio and A is the average intensity (the geometric mean which is equivalent to averaging the log intensity). The curvature in the scatter plot indicated a dependence of the ratio R on the overall intensity. This curve was then used to normalize the data: logIred/Igreen->log (Ired/Igreen)-c(A) where c(A) is the fit. This is equivalent to multiplying the green channel intensity (or dividing the red) by an intensity dependent normalization constant k(A) where log[(k(A)]= c(A). Optimal normalized data should be horizontal and centered at zero. Samples were normalized using this intensity-dependent normalization using the Splus function lowess (Dudoit, 2000). Normalized fold changes in gene expression were then used to further analyze and cluster the various cell lines. Cutoffs: The absolute value of the fold change (R G and G/R) had to be greater than 2.0 in at least one experiment, and the average intensity (A) need to be greater than 300. This filter reduced the number of genes from 8976 to 234. Data was filtered to select genes that had both a fold change to remove the background of mostly unchanging genes and an average intensity distinguishable from the noise of the microchip hybridization.

Clustering: The relationship among cell lines was analyzed using hierarchical clustering limiting to over-expressed genes with R G or G/R ratios higher than 2.0 in at least one experiment (Eisen et al., 1998). To assess the robustness of the clustering analysis, a bootstrap resampling technique was used to generate 1000 copies of the data set by adding Gaussian noise to the original data. The mean value of the noise was zero and the standard deviation was dependent on the average intensity of a given spot. In order to determine this intensity dependent noise value, the data from a sample replicated eight times were used. By fitting a curve to the scatter of the standard deviation of the eight replicates as a function of average log intensity, the inventors obtained a curve for the average noise as a function of intensity. This was used in the Montecarlo resampling to set the value of the standard deviation. Each of the 1000 bootstrap samples was then clustered using the hierarchical method with the dot plot product (angle) metric and ward linkage. A consensus tree was constructed using the CONSENS program (version 3.5c). This program constructs a tree by finding for each node the pairing that occurred most often in the 1000 separate trials. A graph was constructed that displays this count at each node of the tree. Nodes with values closer to 1000 are more robust than with lower ones.

Validation of The Results

Northern Blotting: Northern hybridization was performed using 10 ug of total RNA from the bladder cancer cell lines used in the analysis (see above), and probes generated from the cDNA clones (data not shown).

Tissue Samples And Tissue Microarrays: Three different bladder cancer microarrays were constructed for this study. Normal and tumor tissues were embedded in paraffin and five-μm sections were stained with hematoxylin and eosin to identify viable, moφhologically representative areas of the specimen from which needle core samples were taken using a precision instrument (Beecher Instruments, Silver Spring, MD) (Hoos et al., 2001). From each specimen triplicate tissue cores with diameters of 0.6 mm were punched and arrayed on the recipient paraffin block. Five-μm sections of these tissue array blocks were cut and placed on charged polylysine-coated slides and used for immunohistochemical analysis. Arrayed normal tissues known to express the antigens under study served were used as baseline positive controls and showed physiological expression patterns of these markers. These tissue microarrays included a total of 173 bladder primary transitional cell carcinoma tumors obtained under Institutional Review Board approved protocols. Tumor stage and grade were defined according to consensus criteria. A total of 40 superficial and 64 invasive transitional cell carcinoma tumors were analyzed in two microarrays. These tumors corresponded to 14 grade 1, 8 grade 2, and 82 grade 3 lesions. Another tissue microarray comprised a cohort of 69 bladder primary transitional cell carcinoma cases with known p53, pi 6 and pRB status, and consisted of two superficial and 67 invasive lesions. In addition, 20 cases of Bilharzial-related invasive bladder cancer (BBC) were also analyzed, including 14 squamous (S-BBC) and 6 transitional (T-BBC) carcinomas, for a total of 193 cases. These BBC lesions were also analyzed for patterns of p53 expression (see below).

Immunohistochemistry: Protein patterns of expression of identified targets were assessed at the microanatomical level for caveolin-1, keratin 10, E- cadherin, zyxin, and moesin, using both cytospin from all cell lines studied and tissue samples outlined above. Standard immunoperoxidase procedures were used for immunohistochemistry (Hoos et al., 2001). The following antibodies were used: anti- caveolin-1, mouse monoclonal IgGl at 1:1000 dilution (2.5 μg/ml) (BD Transductions Labs, Lexington, KY); anti-keratin 10, mouse monoclonal clone DC-K10 at 1:2000 (1.0 μg/ml) (Neomarkers, Fremont, CA); anti-E-cadherin, mouse monoclonal clone 36 at 1:1000 (2.5 μg/ml) (BD Transductions Labs, Lexington, KY); anti-moesin, mouse monoclonal clone 38/87 at 1:50 (4 μg/ml) with microwave pretreatment of the slides (Neomarkers, Fremont, CA); anti-zyxin, mouse monoclonal clone 21 at 1:25 (10 μg/ml) with microwave pretreatment of the slides (Transduction Labs, Lexington, KY); anti-RB, mouse monoclonal clone 3C8 at a final concentration of 1.2 μg/ml (QED Bioscience, San Diego, CA); anti-pl6, mouse monoclonal clone clone DCS- 50.1/H4 at 2.5 μg/ml (Calbiochem, Cambridge, MA); and a mouse anti-human monoclonal antibody against p53 (1 :500, Ab-2, clone 1801; Calbiochem). Staining conditions were optimized on sections from formalin-fixed, paraffin-embedded tissue controls for each antibody as specified by manufacturers. Antibody reactivity was detected using diaminobenzidine as chromogen, and sections were counterstained with hematoxylin. The primary antibody was omitted for negative controls. p53 staining was defined as negative (undetectable levels to <20 of tumor cells displaying nuclear staining) or positive (moderate to intense nuclear immunoreactivities in >20% of cells) (McShane et al., 2000). There is no consensus on the cutoffs of the immunohistochemical expression of the other markers, and thus they were analyzed as continuous variables, or taking the cutoff of 0% versus higher than 0% when they were considered as categorical.

Data analysis: All conventional transitional cell carcinoma (n = 173) were used for the analysis of association between p53 and pRB with keratin 10, caveolin-1, E-cadherin, zyxin and moesin. These cases were also utilized for evaluating marker expression versus histopathological stage and tumor grade using the non-parametric Wilcoxon-Mann-Whitney and Kruskall-Wallis tests (Tudor and Koch, 1994). The consensus value of the three representative cores from each tumor sample arrayed was used for statistical analyses. The association of keratin 10 and caveolin-1 with squamous differentiation was analyzed using the total cohort of 193 cases, including the 20 Bilharzial-related bladder tumors. Expression values were displayed as mean values accompanied of 95% confidence intervals and/or range.

The relationship of marker to outcome was evaluated using a subset of 69 conventional transitional cell carcinoma cases for which follow up was available. Overall-survival time was defined as the months elapsed between transurethral resection (two superficial lesions) or cystectomy (rest of cases) and death from disease (or the last follow-up date). Patients who were alive at the last follow-up or lost to follow-up were censored. For survival analysis, expression marker results were analyzed as continuous variables. Membrane expression of moesin was also considered as a categorical variable becasue its median expression value was zero. The association of the marker expression levels with overall survival was analyzed using the Wald test, and the log-rank test was used to examine their relationship when different cutoffs were applied (Fleming and Lin, 2000). Survival curves were plotted using standard Kaplan-Meier methodology (Kaplan and Meier, 1958). Additionally, the association of the markers with the p53 (mutation analysis) and pi 6 (mutation and polymorphism analysis) status in this subset of 69 patients was evaluated. Associations between markers were analyzed using Kendall's tau test. Histopathogenetic Categorization of Bladder Cancer Cell Lines

Hierarchical clustering of cDNA microarray experiments, based on 234 genes that showed a R/G or G/R fold ratio higher than 2 and intensities higher than 300 classified these tumor cells according to the histopathological characteristics from the tumors they were obtained from. SCaBER cells, derived from a squamous carcinoma of the bladder, were distinguished from cells derived from transitional carcinomas. Moreover, tumor cells from invasive lesions clustered together, and were separated from those cells derived from metastatic (TCCSUP) or superficial (RT4) bladder cancers (Figure 1). The complete list of 234 genes is shown in Table 1.

Caveolin-1 And Keratin 10 Are Markers of Squamous Differentiation And Are Associated With Tumor Stage And Grade

Both caveolin-1 and keratin 10 were differentially expressed among the various bladder tumor cells lines analyzed. High levels of caveolin-1 were detected in SCaBER cells, while its expression was low to undetectable in RT4 cells. The expression of Keratin 10 was high in several cell lines, and a previous report had linked its expression to squamous bladder carcinoma.

Based on these observations and availability of well-characterized antibodies to their encoded products, the inventors further explored their patterns of expression in several normal samples of human urothelium, urothelial squamous metaplasia, and the above-mentioned tissue collections, including 173 conventional transitional cell carcinoma (cTCC) and 20 bilharzial-related bladder carcinoma (BBC) (16 squamous-BBC and 4 transitional-BBC). It was found that normal urothelium and superficial cTCC had undetectable levels of both caveolin-1 and keratin 10. However, areas of squamous metaplasia and carcinoma identified in BBC, as well as areas of squamous differentiation identified in cTCC, had significant expression of both proteins. Caveolin-1 was expressed in 12 of 16 of squamous-BBC, as well as in 2 of 4 of transitional-BBC and 72 of 173 of cTCC. Keratin 10 was found in 8 of 16 of squamous-BBC, 1 of 4 of transitional-BBC, and 28 of 173 of cTCC (Figure 2). Statistical analysis of data revealed that both caveolin-1 and keratin 10 were significantly associated with identification of squamous differentiation in 49 of 193 patients (p<0.001).

Separate analysis of the 173 cTCC lesions revealed that only one of 42 superficial lesions displayed caveolin-1 in few tumor cells (approximately 3% tumor cells), while 70 of 131 invasive tumors showed caveolin-1 expression ranging from 3% to 83% tumor cells. Keratin 10 was undetectable in all superficial lesions, while 28 of 131 invasive tumors displayed immunoreactivities ranging from 3% to 70% tumor cells. All grade 1 lesions showed undetectable expression of caveolin-1 and keratin 10. Grade 2 tumors showed expression lower than 3% for caveolin-1, while in grade 3 tumors the mean number of cells showing positive expression of caveolin-1 was 13% (95%> Cl :8.9-17.2%). Keratin 10 expression was also undetectable for tumors of grade 2 and for those displaying grade 3 the mean number of cells showing positive expression of keratin 10 was 2.3% (95% Cl :0.5-4.1%). Overall, there was a statistical association between caveolin-1 expression and both tumor stage (p<0.001) and grade (p<0.001). Keratin 10 also reached a significant statistical association with tumor stage (p=0.019) and grade (p=0.018).

It is postulated that cTCC in which these products were identified harbor moφhologically unrecognizable areas of squamous differentiation. This could have important clinical implications since it has been reported that invasive bladder tumors with squamous features do not respond to MVAC treatment.

Clustering Associates Expression Profiling To Biological Phenotypes Related To p53 And RB Pathways.

A Montecarlo bootstrap method was applied to establish the robustness among the grouping of the cell lines based on the expression of 234 genes selected from the cDNA microarrays showing a R/G or a G/R fold ratio higher than 2 and intensities higher than 300. Using this approach, the analyzed tumor cells assembled based on their reported molecular alterations related to the p53 and RB signaling pathways (Figure 3). Two main clusters or groups were identified: T24, SCABER and UMUC3 (group 1) and HT1376, HTl 197 and TCCSUP (group 2). Briefly, cells that harbor TP53 mutations at exons 4 and 5, detectable pRB, and INK4A mutations (group 1) clustered together and were distinguishable from those having TP53 mutations affecting exons 7, 10, and 11, undetectable levels of pRB, and a wild-type INK4A locus (group 2). Cells with p53 mutations in exon 8 showed a higher distance from this second group.

In an attempt to identify genes related to this clustering and of potential biological significance, a search for common over- and under-expressed genes in each cluster-group and not in the other was performed. Of interest, only three known genes (zyxin, protocadherin 13, and moesin) and an EST were found differentially expressed between the two significant clusters. In order to validate these results at the protein level, immunohistochemical studies using antibodies to zyxin, moesin, and E-cadherin (a down stream product of protocadherin 13 for which antibodies are not available) were performed on cytospins from the analyzed bladder cancer cell lines. Cells from group 1 displayed lower transcript levels of protocadherin 13 and zyxin than cells from group 2. Alternatively, group 2 cells had lower transcript levels of moesin than cells from group 1.

Zyxin. E-cadherin. And Moesin Are Associated With Altered pRB Expression. TP53 Mutation Localization. Tumor Stage And Grade In Primary Bladder Cancer

In order to validate the results obtained through permutation clustering using clinical primary bladder tumors, patterns of p53 and pRB expression were assessed in the cohort of patients under study. In addition, TP53 mutation status was established in a subset of 69 cTCC cases: 37 tumors had wild-type TP53, 9 lesions harbored mutations affecting exons 4 or 5, and 23 had a mutation between exons 6 and 11 (Wikman et al., 2000). Cytoplasmic zyxin expression was significantly associated with detection of TP53 mutations affecting exons 6-11 (p=0.03). Regarding p53 and pRB expression status: 73 cases displayed a p53 positive phenotype (nuclear immunoreactivities >20% tumor cells), while the remaining cases were classified as having a p53 negative phenotype; 44 cases had undetectable pRB, while the remaining cases showed heterogeneous pRB nuclear immunoreactivities. Zyxin was observed in the cytoplasm of 102 of 173 cTCC cases, while membrane E-cadherin staining was found in 146 of these 173 lesions (Figure 4). Moesin was detected as a membrane staining in 26 of 173 cTCC (Figure 4).

No statistical association was found between expression patterns of p53 and either zyxin, E-cadherin or moesin. Nor was there any relationship among mutations, polymoφhisms or protein over-expression of pi 6 in a subset of 69 patients of whom INK4A/pl6 was available. However, there was a significant association between pRB levels and these three markers (Table 2). Furthermore, expression levels of E-cadherin, zyxin and moesin were significantly associated with tumor stage (p<0.001 for the three markers) and grade (p<0.001 for E-cadherin and zyxin, and p=0.005 for moesin).

Moesin Is A Predictive Marker In Bladder Cancer

The potential overall survival prognostic utility of caveolin-1, keratin 10, E-cadherin, moesin, and zyxin was evaluated using 69 cTCC for which clinical follow-up was available. Membrane moesin expression was associated with overall survival (p=0.01) in this subset of 69 patients (Figure 5). Ppatients presenting a positive moesin expression displayed p53>20% in 5/10, E-cadherin>0% in 10/10, zyxin>0% in 6/10, total pRb>10% in 9/10, undeφhosphorylated pRb>10% in 3/10 and pl6>0% in 2/10 of the cases. No polymoφhism or mutation in INK4a/pl6 gene was detected.

Bladder cancer comprises a variety of distinct neoplastic disorders. Transitional and squamous carcinomas are the most prevalent forms of bladder cancer. However, adenocarcinomas, small cell, and neuroendocrine tumors are also found as primary bladder tumors with a lower frequency. Identification of the prevailing and, if present, secondary histogenetic features of the tumor have significant clinical connotations, since it is well known the lack of response to certain therapeutic regimens in the context of specific tumor types. For example, squamous carcinoma of the bladder has been reported to be more resistant to radio- and chemotherapy than conventional transitional bladder tumors. Data from this study revealed a characteristic pattern of caveolin-1 and keratin 10 expression in early squamous metaplasia and squamous carcinomas in the setting of BBC. In addition, the inventors observed that the expression of caveolin-1 and keratin 10 in certain cTCC, usually identified as clusters of tumor cells heterogeneously stained within the bulk of the tumor. Furthermore, there was a significant association regarding detection of both caveolin-1 and keratin 10 in the bladder tumor samples analyzed. Thus, they may serve as markers of squamous differentiation prior to the moφhological identification of this cellular phenotype. This phenomenon might be linked to the lower response to MNAC observed in certain cTCC, harboring histologically unrecognizable areas of squamous differentiation, and may be of assistance in selecting those patients that would benefit from other therapeutic regimens. The increased expression of caveolin-1 has been related to cellular transformation and tumor progression, being associated with augmented cell signaling activity. One of the molecules that is organized and concentrated in the scaffolding domain of caveolin-1 is the epidermal growth factor receptor (EGFr). Bladder cancer cells have increased growth factor receptors, including EGFr and Her-2/Νeu proteins, and this phenomenon has been associated with tumor progression. Interestingly, data from the present analysis also links caveolin-1 expression with increased tumor pathological stage and tumor grade. A recent study has also described association with tumor grade but not with patient outcome. Whether the role of caveolin-1 as a membrane protein implicated in selective transcytosis and increased signaling, or to what extent keratinization are critical in bladder cancer progression and chemoresistance in tumors presenting squamous differentiation has to be further studied.

One of the most notable findings presented above is that significant clusters obtained by bootstrapping methods grouped the bladder cancer cell lines analyzed based on their p53 and RB pathway status. Growth control in mammalian cells is accomplished largely by the action of pRB (regulating exit from the Gl phase) and the p53 protein (triggering growth arrest or apoptotic processes in response to cellular stress). pRB and p53 serve collaborative roles in tumorigenesis as evidenced by their frequent alterations in human tumors, including bladder cancer. The mechanistic basis for this dual requirement stems, in part, from the deactivation of a p53-dependent cell suicide program that would normally be brought about as a response to unchecked cellular proliferation resulting from RB-deficiency.

Two significant major patterns arise from the bootstrapping analysis study in these cell lines based on the combined expression of 234 genes and contrasted with the molecular characterization of p53/pRB/INK4a in these cells lines already described. The combined alteration of these critical networks could support these clusters. Cells harboring TP53 mutations in the amino-terminal transactivation domain, presenting with high pRB levels, and INK4A mutations grouped together (UMUC3, SCaBER, and T24); and certain cells with TP53 mutations in the core domain, undetectable pRB levels, and wild-type INK4A locus were in the same cluster (HTl 197, HT1376, and TCCSUP). TP53 mutations in the core domain affect the ability of p53 to bind DNA and are associated with loss of the contralateral allele, completely inactivating p53 function and thus impacting on both cell cycle arrest and induction of apoptosis. However, mutations in the transactivation domain result in products that preserve to some extent p53 activities, such as DNA binding. This incomplete p53 suppressive phenotype is usually associated with other alterations in the pathway, mainly pl4ARF mutations or Hdm2 amplification/overexpression. In these cases, the inventors found detectable pRB expression, but the other prevalent mutation of the RB pathway, namely pl6/INK4A deletions, was detected in all cell lines displaying such genotype. Finally, the inventors observed that RT4 cells had a wild-type TP53, but harbored a homozygous INK4A deletion, and lacked pRB expression. These data support the working model previously reported. However, cells with mutations in exon 8 in the core domain (5637 and J82) were not included into the cluster described by the boostrap technique.

Three adhesion-related molecules, zyxin, E-cadherin, and moesin, were found to be associated with the p53/RB patterns discussed above. Those lines harboring TP53 mutations in the core domain and lacking pRB displayed low moesin transcript levels; while those lines with TP53 mutations in the transactivating domain and high levels of pRB displayed low zyxin and E-cadherin. Moreover, the associations observed in cell lines were also found and validated in primary bladder tumors. Low levels of moesin, zyxin and E-cadherin were significantly associated with advanced pathological stage and higher tumor grade, supporting their involvement in bladder cancer progression.

Alterations of E-cadherin had been described as common events in bladder cancer. The present observations regarding the association of E-cadherin with histopathological stage and tumor grade were in accordance with other previous studies. Loss of zyxin has been associated with neoplastic transformation, as it was found as a marker of acute myeloid leukemia subtype. The functional role of zyxin is not completely defined, but it appears to be involved in signaling networks established between focal adhesion plaques and the nucleus. There has been no report dealing with altered patterns of zyxin in clinical samples to date. Results from this study revealed that moesin provided prognostic information regarding poor outcome in bladder cancer. The loss of moesin has recently been reported as being associated with advanced ovarian cancer, metastatic melanoma, and lung carcinoma, suggesting that it could have a critical tumor suppressive function altered in multiple human cancers. It was also observed a predictive utility of moesin expression in bladder cancer. The inventors evaluated the alterations in p53/pRB/pl6 in these patients that showed a shorter survival. Most of them presented alterations in these molecules that may also account with their aggressive outcome.

The three adhesion-related markers (zyxin, E-cadherin, moesin) found differentially expressed in bladder cancer in the present study share a common feature: a relationship to the beta-catenin pathway. Beta-Catenin is a cytoplasmic protein that participates in the assembly of cell-cell adherens junctions by binding cadherins to the actin cytoskeleton. The cytoplasmic domain of E-cadherin interacts directly with beta-catenin. Zyxin is a cytoplasmic adherens junction protein found in complexes with alpha-actinin and actin. The association of zyxin with cadherins has also been previously reported. Moesin is a member of the ERM (ezrin, radixin and moesin) family of proteins located just beneath the plasma membranes, which are also thought to be involved in the association of actin filaments with the plasma membrane regulating cell-cell and cell-matrix adhesion. The association of ezrin with E-cadherin and beta-catenin has also been revealed by coprecipitation studies. The involvement of beta catenin in bladder cancer progression has recently been described in murine and human models, an observation supporting the present findings. Among downstream targets of beta-catenin, cyclin Dl has been shown to be critical in Gl-S cell cycle transition by phosphorylating pRB. Interestingly, the three identified markers were associated with RB gene expression in primary tumors, and this association could be attributed to alteration of cyclin Dl expression levels as reported in colorectal and desmoid tumors. Disruption of the beta-catenin signaling pathway by alterations in the physiological balance between its interactions with zyxin, E-cadherin or moesin could mechanistically account for the invasiveness potential that certain bladder cancer cell lines under study display. Other alternative mechanisms affecting this pathway include alterations of the Wnt signaling, RAS mutations, and mutations affecting the beta-catenin gene itself. Aberrant accumulation of beta-catenin in solid tumors has been also associated with mutational inactivation of the TP53 gene. Overexpression of wild- type p53, by either transfection or DNA damage, has been shown to down- regulate beta-catenin in human and mouse cells. However, whether the association between zyxin and TP53 mutations could also be related to an altered beta-catenin pathway remains to be elucidated.

The present study has identified deregulation of three adhesion molecular targets as common alterations in high-grade bladder cancer cells with different described phenotypes of cell cycle regulator genes. The loss of these cell- adhesion molecules was correlated with tumor progression in primary bladder tumors. This observation reveals the importance of the interactions among tumor cells as well as tumor cells with the surrounding stroma in cancer progression. It appears that deregulation of cell cycle and apoptotic pathways, such as mutations or altered expression of p53 and pRB, are necessary for uroepithelial transformation. However, they appear to be insufficient for bladder cancer progression. Even though linear models are a "simplification" of complex pathological events, it appears that deregulation of cell adhesion is a common event associated with tumor progression in bladder cancer, independently of the genetic alterations triggering tumorigenesis. Expression profiling has revealed the common deregulation of cell adhesion displayed in highly invasive and differentiated cells alteration of alternative adhesion pathways. The most relevant finding was that these molecular targets identified in vitro, zyxin, E-cadherin and moesin, were found critical in progression in clinical material supporting a relevant role of deregulation of cell adhesion in bladder cancer progression.

In summary, molecular profiling using cDNA microarrays clustered bladder cancer based on both histopathogenesis and biological criteria. Novel targets genes have been validated using tissue arrays containing well characterized primary tumors. Keratin 10 and caveolin-1 defined squamous differentiation, and might become useful markers to further stratify bladder tumors. E-cadherin, moesin and zyxin were associated with tumor progression, revealing the relevance of deregulation of cell adhesion in bladder cancer progression. Finally, moesin expression appeared to be a significant prognostic factor associated with patient survival.

TABLE 1 .

234 Genes Used For Hierarchical Clustering Analysis of Bladder Cancer Cell Lines cDNA Microarray

1. Non-specific cross reacting antigen

2. Nerve growth factor beta

3. Myelin basic protein

4. Membrane metallo-endopeptidase (neutral endopeptidase, enkephalinase, CALLA, CD 10)

5. Human stanniocalcin precursor (STC) mRNA, complete cds

6. Human mRNA for KIAA0269 gene, complete cds

7. Human clone 23587 mRNA sequence

8. Homo sapiens mRNA for NB thymosin beta, complete cds

9. Homo sapiens Rac3 (RAC3) mRNA, complete cds

10. Fibrillin 2

1 1. Brain-derived neurotrophic factor

12. Plasminogen activator inhibitor, type II (arginine-serpin)

13. MULTIFUNCTIONAL AMINOACYL-TRNA SYNTHETASE

14. MHC class II DQ-beta associated with DR2, DQwl protein

15. Keratin 5 (epidermolysis bullosa simplex, Dowling-Meara/Kobner/Weber-Cockayne types)

16. Human skeletal muscle LIM-protein SLIM1 mRNA, complete cds

17. Human 19.8 kDa protein mRNA, complete cds

18. Heme oxygenase (decycling) 1

19. ESTs, Weakly similar to ! ! ! ! ALU SUBFAMILY SX WARNING ENTRY ! ! ! ! [H.sapiens]

20. ESTs, Highly similar to IG ALPHA-2 CHAIN C REGION [H.sapiens]

21. ESTs

22. ESTs

23. ESTs

24. ESTs

25. ESTs, Weakly similar to F43C1.3 [C.elegans]

26. ESTs, Highly similar to ZYXIN [Gallus gallus]

27. ESTs, Highly similar to ZINC FINGER PROTEIN 42 [Homo sapiens]

28. ESTs, Highly similar to INHIBIN BETA A CHAIN PRECURSOR [Bos taurus]

29. ESTs

30. ESTs

31. ESTs

32. ESTs

33. ESTs

34. ESTs

35. ESTs

36. ESTs 37. ESTs

38. ESTs

39. ESTs

40. ESTs

41. ESTs

42. ESTs

43. ESTs

44. ESTs

45. ESTs

46. INTERFERON-INDUCIBLE PROTEIN 9-27

47. Human tumor susceptiblity protein (TSG101) mRNA, complete cds

48. Human mRNA for KIAA0075 gene, partial cds

49. Human mRNA for BST-2, complete cds

50. Human TRAF-interacting protein I-TRAF mRNA, complete cds

51. Homo sapiens mRNA for zinc finger protein FPM315, complete cds 52. H.sapiens mRNA for processing a-glucosidase I

53. Complement component 4-binding protein, beta

54. Casein kinase 2, alpha prime polypeptide

55. UDP-GLUCURONOSYLTRANSFERASE 2B4 PRECURSOR, MICROSOMAL

56. ESTs, Highly similar to RAS-RELATED PROTEIN RAB- 18A [Lymnaea stagnalis]

57. ESTs, Highly similar to 50S RIBOSOMAL PROTEIN L2 [Bacillus stearothermophilus]

58. Human osteoprotegerin (OPG) mRNA, complete cds

59. ESTs, Weakly similar to weak similarity to ribosome releasing factors [C.elegans]

60. 5' nucleotidase (CD73)

61. S 100 calcium-binding protein A9 (calgranulin B)

62. Human G protein gamma- 11 subunit mRNA, complete cds

63. ESTs

64. ESTs

65. X-LΓNKED HELICASE II

66. Protein-tyrosine kinase 7

67. Pregnancy-specific beta- 1 glycoprotein 13

68. Peripheral myelin protein 22

69. Moesin

70. Membrane component, chromosome 1, surface marker 1 (40kD glycoprotein, identified by monoclonal antibody GA733)

71. Interferon-inducible 56-KDa protein

72. INSULIN-LIKE GROWTH FACTOR BINDING PROTEIN 1 PRECURSOR

73. Human putative EPH-related PTK receptor ligand LERK-8 (Eplg8) mRNA, complete cds

74. Human clone 23589 mRNA sequence

75. Homo sapiens mRNA for aurora/IPLl -related kinase, complete cds

76. H.sapiens IL- 13Ra mRNA

77. GLYCYLPEPTIDE N-TETRADECANOYLTRANSFERASE

78. Fibronectin 1

79. Annexin VIII

80. Keratin 10 (epidermolytic hyperkeratosis; keratosis palmaris et plantaris) 81. KINESIN HEAVY CHAIN

82. Human (clone 8B 1) Br-cadherin mRNA, complete cds

83. Homo sapiens mRNA for osteoblast specific factor 2 (OSF-2os)

84. Hemopoietic cell kinase

85. H.sapiens mRNA for testican

86. ESTs, Highly similar to METALLOTHIONEIN-II [H.sapiens]

87. ESTs, Highly similar to DEVELOPMENTAL PROTEIN SEVEN IN ABSENTIA [Drosophila melanogaster]

88. ESTs

89. ESTs, Moderately similar to ninein [M.musculus]

90. ESTs, Highly similar to PROTOPORPH YRINOGEN OXIDASE [H.sapiens]

91. ESTs, Highly similar to INSULIN-INDUCED GROWTH RESPONSE PROTEIN CL-6 [Rattus norvegicus]

92. ESTs, Highly similar to GLYCOGEN PHOSPHORYLASE, LIVER FORM [Homo sapiens]

93. ESTs

94. ESTs

95. ESTs

96. ESTs

97. ESTs

98. ESTs

99. ESTs

100. ESTs

101. ESTs

102. ESTs

103. ESTs

104. ESTs

105. ESTs

106. ESTs

107. ESTs

108. Pregnancy specific beta- 1 glycoprotein 5

109. Human fibroblast growth factor homologous factor 1 (FHF-1) mRNA, complete cds

1 10. Homo sapiens mRNA for smallest subunit of ubiquinol-cytochrome c reductase, complete cds

1 1 1. HEAT SHOCK PROTEIN HSP 90-ALPHA

1 12. H.sapiens mRNA for protein-tyrosine-phosphatase (tissue type: foreskin)

1 13. GROl oncogene (melanoma growth stimulating activity, alpha)

1 14. INTERFERON-INDUCIBLE PROTEIN 1 -8U

1 15. ESTs, Weakly similar to UTROPHIN [Homo sapiens]

1 16. ESTs, Weakly similar to kruppel-related zinc finger protein [H.sapiens]

1 17. ESTs, Moderately similar to Mouse 19.5 mRNA, complete cds [M.musculus]

1 18. ESTs, Highly similar to mitogen-induced [M.musculus]

1 19. Human extracellular protein (S 1 -5) mRNA, complete cds

120. PROBABLE TRANS- 1 ,2-DIHYDROBENZENE- 1 ,2-DIOL DEHYDROGENASE

12 1. ESTs

122. ESTs

123. PREGNANCY-SPECIFIC BETA- 1 GLYCOPROTEIN D PRECURSOR 124. Human protease M mRNA, complete cds

125. Human mRNA for KIAAO 146 gene, partial cds

126. Human cellular proto-oncogene (c-mer) mRNA, complete cds

127. Human TAR RNA binding protein (TRBP) mRNA, complete cds

128. Homo sapiens mRNA expressed in osteoblast, complete cds

129. Homo sapiens clone 22 mRNA, alternative splice variant alpha- 1 , complete cds

130. Epidermal growth factor receptor pathway substrate 15

13 1. JNK ACTIVATING KINASE 1

132. Human TFIIA gamma subunit mRNA, complete cds

133. ESTs, Moderately similar to nuclear LIM interactor [M.musculus]

134. ESTs, Highly similar to GLIA DERIVED NEXIN PRECURSOR [Homo sapiens]

135. ESTs

136. ESTs

137. ESTS

138. ESTs, Weakly similar to fractionated X-irradiation-induced 29 thymoma [M.musculus]

139. ESTs

140. ESTs

141. ESTs

142. ESTS

143. ESTS

144. ESTS

145. ESTS

146. ESTS

147. ESTS

148. ESTS

149. ESTS

150. ESTs

15 1. Integrin, alpha M (complement component receptor 3, alpha; also known as CDl lb (pi 70), macrophage antigen alpha polypeptide)

152. Human pregnancy-specific beta-1 glycoprotein mRNA, complete cds

153. Human mRNA for KIAAOOOl gene, complete cds

154. Human apM2 mRNA for GS2374 (unknown product specific to adipose tissue), complete cds

155. Human RalGDS-like 2 (RGL2) mRNA, partial cds

156. Homo sapiens hCPE-R mRNA for CPE-receptor, complete cds

157. Homo sapiens creatine transporter mRNA, complete cds

158. H.sapiens mRNA for hepatocyte nuclear factor 4 gamma

159. H.sapiens mRNA for Not56-like protein

160. H.sapiens mRNA for IL 13 receptor

161. Cellular retinoic acid-binding protein [human, skin, mRNA, 735 nt]

162. Human hnRNP type A/B protein mRNA, complete cds

163. ESTs, Weakly similar to hTAFII 100 [H.sapiens]

164. ESTs, Weakly similar to GAGE-4 protein [H.sapiens]

165. ESTs, Highly similar to CARCINOEMBRYONIC ANTIGEN CGM6 PRECURSOR [Homo sapiens]

166. ESTs, Weakly similar to ovary2 [D.melanogaster] 167. ESTs, Weakly similar to T-LYMPHOCYTE MATURATION-ASSOCIATED PROTEIN [H.sapiens]

168. ESTs, Highly similar to 8A-2V protein [M.musculus]

169. H.sapiens H4/g gene for H4 histone

170. Homo sapiens breast cancer-specific protein 1 (BCSG1) mRNA, complete cds

171. Human tetracycline transporter-like protein mRNA, complete cds

172. Homo sapiens (hue) mRNA, complete cds

173. ESTs

174. Solute carrier family 9 (sodium/hydrogen exchanger), isoform 1 (antiporter, Na+/H+, amiloride sensitive)

175. NKG2-D TYPE II INTEGRAL MEMBRANE PROTEIN

176. Matrix metalloproteinase 1 (interstitial collagenase)

177. MULTIDRUG RESISTANCE- ASSOCIATED PROTEIN 1

178. Human mRNA for KIAA0365 gene, partial cds

179. Human MHC class II HL A-DR2-Dw 12 mRNA DQw 1 -beta, complete cds

180. CARTILAGE GLYCOPROTEIN-39 PRECURSOR

181. Aminolevulinate, delta-, synthase 2 (sideroblastic/hypochromic anemia)

182. Proteoglycan 1 , secretory granule

183. Prolactin

184. Human growth hormone-dependent insulin-like growth factor-binding protein mRNA, complete cds

185. Human clone A9A2BRB6 (CAC)n/(GTG)n repeat-containing mRNA

186. Human GDP-dissociation inhibitor protein (Ly-GDI) mRNA, complete cds

187. ESTs

188. ESTs

189. ESTs, Weakly similar to putative pi 50 [H.sapiens]

190. ESTs, Weakly similar to HYPOTHETICAL 41.9 KD PROTEIN IN SDS3-THS1 INTERGENIC REGION [S.cerevisiae]

191. ESTs, Moderately similar to Lasp-1 protein [H.sapiens]

192. ESTs, Highly similar to MUSCLE-CADHERIN PRECURSOR [Mus musculus]

193. ESTs

194. ESTs

195. ESTS

196. ESTS

197. ESTS

198. ESTS

199. ESTS

200. ESTs

201. ESTs

202. ESTS

203. ESTS

204. ESTS

205. ESTS

206. Pregnancy-specific beta- 1 glycoprotein 4

207. Phosphofructokinase, muscle

208. Human squamous cell carcinama of esophagus mRNA for GRB-7 SH2 domain protein, complete cds

209. Human mRNA for IgG Fc binding protein, complete cds 2 10. Human L-kynurenine hydrolase mRNA, complete cds

21 1. Human Ca2+-dependent activator protein for secretion mRNA, complete cds

212. Homo sapiens mRNA for low molecular mass ubiquinone-binding protein, complete cds

213. Homo sapiens mRNA for glutathione transferase A4-4

214. H.sapiens mRNA for laminin

215. Fatty acid binding protein 4, adipocyte

216. Cadherin 2, N-cadherin (neuronal)

217. CYSTATIN A

218. Prostacyclin-stimulating factor [human, cultured diploid fibroblast cells, mRNA, 1124 nt]

2 19. Integrin, beta 3 (platelet glycoprotein Ilia, antigen CD61)

220. ESTs, Weakly similar to putative pl50 [H.sapiens]

22 1. ESTs, Highly similar to PROHIBITIN [Homo sapiens]

222. Homo sapiens mRNA from chromosome 5q21-22, clone:357Ex

223. ESTs, Weakly similar to putative p 150 [H.sapiens]

224. ESTs, Weakly similar to T-complex protein 10A [H.sapiens]

225. ESTs, Weakly similar to Sxmlp [S.cerevisiae]

226. ESTs, Moderately similar to ! ! ! ! ALU SUBFAMILY SB2 WARNING ENTRY ! ! ! ! [H.sapiens]

227. ESTs, Moderately similar to PROTEIN-TYROSINE PHOSPHATASE [Autographa califomica nuclear polyhedrosis virus]

228. Human (clone CTG- A4) mRNA sequence

229. Interleukin 1 , alpha

230. Homo sapiens G protein beta 5 subunit mRNA, complete cds

231. Human homolog of yeast mutL (hPMS 1) gene, complete cds

232. MELANOMA- ASSOCIATED ANTIGEN 4

233. Homo sapiens mRNA for hepatocyte growth factor activator inhibitor, complete cds

234. ESTs

TABLE 2

Association Between Expression Levels of Moesin. E-Cadherin. And Zyxin With

Expression Levels of pRB or TP53.

EXAMPLE 2 Molecular Diagnosis And Outcome Prediction In Bladder Tumors By Gene Profiling Using cDNA Microarrays

This example discloses cDNA microarray analysis that demonstrates two main clusters segregating superficial from invasive transitional carcinomas. These clusters would provide prognostic information. Cytokeratin 20, neuropilin 2, p21 and p33INGl were selected among the top ranked molecular targets differentially expressed between superficial and invasive tumors and validated by immunohistochemistry with tissue microarrays. Their expression patterns were associated with pathological stage, tumor grade, and altered RB expression. Moreover, p33INGl expression levels were related to overall survival. Generation of a support vector machine algorithm revealed the relevance of WNT signaling and mitotic checkpoint alteration during bladder cancer progression. In summary, gene profiling successfully classified bladder tumors based on their histopathogenesis and clinical outcome, and identified molecular biomarkers of potential clinical significance.

Cell Lines And Tumor Samples For cDNA Analysis

Four bladder cancer cell lines: T24, J82, RT4, and HTl 197 were obtained from ATCC (Rockville, MD) and maintained following standard procedures. All cells were grown and harvested at 75%-90% confluence no longer than 4-6 passages. Total RNA of cell lines was extracted using RNeasy (Qiagen, Valencia, CA). Fifteen patients with bladder cancer were included for the expression profiling study. Specimens were collected under an IRB approved tissue procurement protocol. Bladder tumors embedded in OCT were macro-dissected to ensure a minimum of 75% of tumor cells. Total RNA from bladder tumors was isolated in two steps using TRIzol (Life Technologies, Carlsbad, CA), followed by RNeasy purification.

cDNA Microarrays Preparation And Image Acquisition

A set of 17,842 sequence- verified human IMAGE cDNA clones, representing both known genes and ESTs, were PCR amplified and spotted onto poly-lysine coated microscope slides by the Albert Einstein College of Medicine microarray facility (Cheung et al., 1999). Five micrograms of total RNA from each bladder tissue and pool of cell lines was linearly amplified using a single round (Hoos et al., 2001; Tseng et al., 2001). Amplified cRNA obtained from bladder tumors were labeled with Cy5 (red) and hybridized against amplified cRNA from the pool containing equal RNA quantities of the four cell lines labeled with Cy3 (green). Following hybridization, slides were washed, dried and scanned by an Axon automated laser scanner. GenePix software was used for gridding and signal intensities calculation (Tseng et al., 2001; Eisen et al., 1998).

Collection And Analysis of Data

Normalization: cDNA microarrays were normalized using an intensity-dependent algorithm (Tseng et al., 2001). Normalized fold changes in gene expression were then used to further analyze and cluster the bladder tumors (Eisen et al., 1998; Felsenstein, 1985). Clustering: Before clustering, the data was filtered to select genes having both significant average intensities and fold changes removing the background of unchanging genes. It was required that genes have a fold change of at least 3 (up or down) and an average intensity greater than 316 for at least 2 samples using the geometric average of the two channel intensities. This filter reduced the number of genes from 17,842 to 15,609. The set of 15 bladder tumor samples were then analyzed using the hierarchical clustering with the Ward linkage method combined with non-parametric bootstrap resampling and consensus tree building to determine the support for sample groupings (Eisen et al., 1998; Felsenstein, 1985).

Gene Ranking: Several scoring methods were applied to rank genes that could separate early-stage tumors from invasive organ-confined lesions and those developing metastatic bladder disease. Initially, the Mann- Whitney- Wilcoxon rank sum test was applied to identify genes differentially expressed between the two significant clusters (Dawson-Saunders and Trapp, 1994). Only genes showing a p value lower than 0.05 were considered for further analysis.

The method of single-variable logistic regression was applied labeling samples based on the histopathological records of the tumors analyzed in this study (Li and Yang, 2002; Xiong et al., 2001). If the log-expression level of gene j in sample i is X_j„ the sample is labeled as either one type of cancer (y_ι=l) or another (y,=0)_> the model of logistic regression is: Prob (y,=l) = 1/(1+ exp(-a_j - b_j x,,)) where "exp" is the exponential function, and a, and b_j are parameters in the model. The parameter values are obtained by the maximum likelihood estimation. Single variable logistic regression was carried out for all 15650 genes/ESTs according to the maximum likelihood. The decay of the maximum likelihood as a function of the rank is approximated a power- law function or Zipf s law (Li and Yang, 2002).

Multidimensional analysis was then performed taking the 5,616 genes providing data simultaneously in all the tumors. The gene-expression matrix (N experiments x M genes) was first pretreated. All columns (M genes) were re- normalized with the transformation z = (x-μ)lσ , that is, a new matrix (X-matrix) was created where each column has a mean equal to zero and variance equal to one (z- transformation). The X-matrix was then studied with Q-mode Factor Analysis (FA) (Reyment and Joreskog, 1996; De las Rivas et al., 2002). FA seeks finding an underlying orthogonal factor model of the X-matrix of the form: X = LF + E , where L is the loadings matrix, F is the scores matrix, and E is the residual matrix. First, loadings were obtained by scaling the eigenvector matrix (P) obtained from Principal Components Analysis (PCA): L = P L^{1 2}. The optimal dimensionality was three, explaining 76.6 % of the variance. The factors were then rotated by means of a varimax rotation (Saitou & Νei, 1987; Reyment and Joreskog, 1996), and from these rotated factors the scores matrix was generated by an ordinary (unweighted) least squares procedure: F = L^"1/2 P*'X, where P* is the rotated eigenvectors, and the prime denotes the transpose. These loading distances were clustered with a neighbor- joining (NJ) algorithm to build a dendrogram.

An estimation of the reliability of each branch was obtained by means of a jackknife bootstrap analysis. Bootstrap values were computed from selected random subsets of 75% of the genes and ESTs in the X-matrix, by reanalyzing the new resulting matrixes by FA and NJ. Relationships among samples obtained from the previous step were contrasted with the available clinical and pathological data. A supervised method (Cristiani & Shawe-Taylor, 2000) was applied in order to detect the combination of genes in the X-matrix that can optimally be able to explain these groups. Specifically, the hypeφlane in the loadings space that optimally separates the previously defined groups of samples was obtained by means of a Support Vector Machine (SVM) algorithm. The FA scores were then projected onto the characteristic vector of this hypeφlane and then were z-transformed. The sorted z-scores were used to select the important descriptors able to separate the groups. A z-score cutoff of 2 was used to obtain a subset of cDNAs with the best discriminant properties.

To obtain insights into the biochemical pathways involved in bladder tumor progression, the biological function index of the genes under study in GO was searched (Xie et al., 2002). Biological processes according to GO were available only for 1,044 genes out of the 5,616 under analysis. The inventors identified by t-test the genes differentially expressed among each of the four groups generated by FA-SVM, as well as among superficial lesions (pooling groups 1 and 2) and invasive tumors (pooling groups 3 and 4) versus the rest of the experiments. The average t-test is calculated within the different groups as a measure of enrichment of each biological process in the different groups. Only genes with t-test -values lower than 0.01 were considered for further analysis. The statistical significance of the association of groups and biological processes (GO indexes) was evaluated by means of the hypergeometric distribution. The inventors computed the probability that at least x genes (with a t-test ?-value lower than 0.01) were annotated within any given biological process in a random subset of n genes, where N denoted the total number of annotated genes (1044) in the entire dataset, and A the number of these genes with a particular annotation. These p-values were obtained according to the following equation: p(x; N,A,n) =

The following criteria were requ red for calcu ating t e significance of the biological processes. Only biological processes (GO indexes) with more than 5 members in the set were selected for further analysis, and only annotated genes with t-test p- value lower than 0.01 within each biological process were considered. It was also required that at least two genes reached this significance within a biological process. Hypergeometric p- values lower than 0.05 were considered significant, but the inventors also focused on those marginally sub-optimals, to compensate the sparsity of GO annotations and the limitation of that non-annotated genes could be included in this analysis. Finally, significant cDNAs (according to a t-test), and belonging to significant or marginally significant GO pathways (p-value<0.01) were grouped and represented using a color-graded spectrum.

Clinical Validation of The Results

Tissue Samples In Tissue Microarrays Three different bladder cancer microarrays were used in this study (Hoos et al., 2001), including a total of 173 bladder primary transitional cell carcinoma (TCC) tumors obtained under IRB approved protocol. A total of 40 superficial and 64 invasive TCC tumors were analyzed in two microarrays. These tumors corresponds to grade 1 (n=24), grade 2 (n=8) and grade 3 (n=82) lesions. The third tissue microarray comprised a cohort of 69 bladder primary TCC cases with known p53/pRB status and annotated follow-up, including two superficial and 67 invasive lesions.

Immunohistochemistry Protein patterns of expression were assessed at the microanatomical level using both cytospins from cell lines studied (data not shown) and tissue microarrays outlined above. Standard avidin-biotin immunoperoxidase procedures were applied for immunohistochemistry. The following panel of mouse monoclonal antibodies were used: np-2 (clone 54; BD Transductions Labs, Lexington, KY); cytokeratin 20 (clone Ks20.8; DAKO, Denmark); cyclin E (clone cyE05; Neomarkers, Fremont, CA); p53 (clone 1801; Calbiochem, Cambridge, MA); total pRB (clone 3C8; QED Bioscience, San Diego, CA); under-phosphorylated pRB (clone G99-549; BD Transductions Labs); ninjurin (clone 50; BD Transductions Labs); p33INGl (clone CAB1; BD Transductions Labs); and p21/WAFl (clone Ab-1, Calbiochem). Control tissues for specificity assessment were used according to manufacturers' recommendations. The inventors used a 20% cutoff for p53 staining, 10% for p21 and 25% for cyclin E. There is no consensus on the cutoffs of the immunohistochemical expression of the other markers, and thus they were analyzed as continuous variables, or taking several cutoffs when considered as categorical. Statistical Analysis All TCC (n = 173) were used for the analysis of association between p53 and pRB with np-2, cytokeratin 20, cyclin E and p21. These cases were also utilized for evaluating marker expression versus histopathological stage and tumor grade, using the non-parametric Wilcoxon-Mann- Whitney and Kruskall-Wallis tests (Dawson-Saunders and Trapp, 1994). The consensus value of the representative cores from each tumor sample arrayed was used for statistical analyses.

The inventors analyzed the relationship of the cluster analysis of the bladder tumors to which expression profiling was performed with overall survival. Additionally, the association of the markers identified in the DNA microarray analysis to outcome was also evaluated using a subset of 69 cTCC cases for which follow up was available. Overall- survival time was defined as the months elapsed between transurethral resection or cystectomy and death from disease (or the last follow-up date). Patients who were alive at the last follow-up or lost to follow-up were censored. For survival analysis, bootstrapping cluster and biomarkers were analyzed as categorical variables. The association of the marker expression levels with overall survival was analyzed using the Wald test, and the log-rank test was used to examine their relationship when different cutoffs were applied. Survival curves were plotted using Kaplan-Meier methodology. Associations between markers were analyzed using Kendall's tau b test using the SPSS statistical package (version 8.0).

Molecular Classification And Predictive Value of Hierarchical Clustering

The present analysis was carried out on the basis of two complementary sets of experiments. Initially, bladder tumors were analyzed using cDNA microarrays to identify differentially expressed genes among histopathologically distinct tumors (Table 3). The transcriptome of 15 bladder tumors was compared against a pool of four bladder cancer cell lines containing equal RNA quantities using cDNA microarrays with 17,842 known genes and expressed sequence tags (ESTs) (Cheung et al., 1999). Secondly, the potential clinical significance of the selected targets identified by cDNA microarrays was validated at the microanatomical level using immunohistochemistry on tissue microarrays containing well-characterized bladder carcinomas (Hoos et al., 2001). A cohort of superficial and invasive bladder neoplasms was used to evaluate the association between biomarkers and histopathological stage and grade (n=173). A subset of these bladder tumors (n=69), with characterized p53 and pRB alterations and clinical follow-up, was used to delineate associations between potential novel biomarkers and these cell cycle regulators as well as with patient outcome. The use of unsupervised hierarchical clustering combined with non- parametric bootstrap analysis classified primary bladder carcinomas based on their histopathological criteria. Overall, the superficial tumors clustered together and were segregated from invasive transitional carcinomas. Cases developing metastasis and displaying a shorter survival could be distinguished from others displaying a longer survival and those that had organ-confined disease. The bootstrap resampling technique was able to establish a high confidence of these clusters (Figure 6A). Patients whose tumor samples were subjected to gene profiling had a median follow- up of 12 months (mean: 14.3 months, range: 1 to 44 months) (Table 3). The superficial and the invasive clusters were significantly associated with overall survival (p=0.0025) (Figure 6B). These results revealed the diagnostic and prognostic utility of unsupervised clustering since the identified clusters were associated with histopathogenesis and overall survival.

A multidimensional analysis of 4729 genes providing expression data in all the tumors, revealed four groups of expression profiles consistent with the previously identified clusters (Figure 6C, Table 3). Interestingly, this additional analysis also revealed that the gene expression profiles of certain superficial tumors were more similar to some organ-confined invasive lesions (Figure 6D). Identification of Genes Differentially Expressed Between Superficial And Invasive Bladder Cancer

Gene identification was the next step in the study. Several scoring methods were applied to rank the genes according to their ability to separate the superficial from the invasive organ-confined and those developing metastatic bladder tumors (Tables 4-6). First, the Mann- Whitney- Wilcoxon rank sum test was applied as a standard means for gene identification between two groups. The goal was to identify genes differentially expressed between the superficial and invasive clusters. It was observed that the first 120 genes correctly classified the samples contained in each cluster (p=0.033) (Table 4). Two genes, p21 and cyclin E were selected for further study due to their participation in the p53 and RB signaling pathways, both of which are frequently altered in bladder cancer progression.

A single-gene variable logistic regression analysis was carried out as a standard classification/discrimination model to rank genes by their classification performance. The goal was to identify genes differentially expressed between superficial and the invasive non-metastatic together with invasive tumors associated with development of metastatic disease. The results demonstrated that any of the 92 top-ranked genes could differentiate superficial versus invasive tumors based on their levels of gene expression. For those genes ranked from 93 to 500 using this analysis, a maximum of 3 misclassifications could be obtained. The inventors chose to focus on the 92 genes that provided no misclassification among superficial versus invasive and those developing metastasis (Table 5). Two genes from this initial group, cytokeratin 20 and neuropilin-2 (np-2), coding for soluble proteins with potential role for tumor marker development were selected and studied by immunohistochemistry on tissue microarrays.

Finally, the inventors continued to further elucidate the genes that best characterized each of the four groups generated by the multidimensional approach (Table 6). Overall, groups 2 and 4 in this analysis included tumors with the worst clinical outcome. Since the inventors were interested in the identification of genes with potential prognostic utility, one gene from each of these groups (ninjurin and p33INGl) was selected in order to evaluate their potential prognostic value. Association of The Identified Markers With Tumor Stage. Grade. And P53/ RB Expression And Overall Survival

The potential roles of the identified target genes in diagnosing patients with superficial and invasive disease were analyzed using tissue microarrays containing transitional cell carcinoma (TCC) of different stages and grades. Overall, the results demonstrated that the expression of these molecular markers was associated with tumor stage, grade, p53/pRB expression and overall survival. Cytokeratin 20, np-2, p21, and p33INGl were differentially expressed in superficial and invasive tumors. In the subset of patients analyzed, there was a significant correlation between the expression of these proteins and tumor stage and grade (Table 7a). Levels of p33INGl expression were easily detectable in normal urothelium, and in the majority of superficial TCC, but were much lower in invasive tumors (Figure

7).

Since both p53 and RB signaling pathways are frequently altered during bladder cancer progression, the association of these biomarkers with p53 and pRB status was also evaluated. The expression of cyclin E was significantly associated with p53 expression; cytokeratin 20, np-2, p21, and p33INGl were all associated with altered pRB expression (Table 7b). A significant correlation between p33INGl expression with cyclin E and p21 was also noted. When the overall survival prognostic utility of cytokeratin 20, np-2, p21 and cyclin E was evaluated using 69 TCC for which clinical follow-up was available, it was observed that only the expression of p33INGl was significantly associated with overall survival (p=0.02). Patients displaying a higher expression of p33INGl showed a shorter survival than those with low expression of this protein (Figure 7C).

Molecular Pathways Involved In Bladder Cancer

A supervised method, based on a Support Vector Machine algorithm

(SVM), was applied to detect the combination of genes that can optimally explain the two main groups (superficial from invasive transitional carcinomas) identified by multidimensional analysis. The goal was to obtain insight about the biochemical pathways involved in tumor progression. The subset of genes with the most optimal discriminatory properties was clustered using the correlation coefficient of the original data as the similarity measure. These genes were then grouped according to the molecular pathways in which they are involved. SVM revealed that WNT signaling pathway and mitotic spindle checkpoint are among the most important networks altered during bladder cancer progression (Figure 8). Representative genes of these pathways that were top-ranked based on their z-score included FAK, zyxin or cdcl6.

TABLE 3

Clinical . Histopathological And ] Epidemiol oeical Characteristics of Bladder Cancer

Patients

Follow- Clinica

Carcino Prostat Smokin Familiar up 1

Patie B M Ag Se ma e g Cancer (months Outco nt ID T D e X TNM In situ Cancer Habit History ) me

174 1 1 67 M TISG3N0 YES YES NO NO 12 NED

160 1 1 83 M TAG1N0 YES YES NO NO 20 NED

157 1 1 61 M TISG3N0 YES NO NO NO 44 NED

134 1 1 80 M T3BG3NO YES YES YES YES 13 NED T4G3N0M

170 2 2 55 M 1 YES YES YES NK 4 DOD

168 2 2 61 M T4BN1M2 YES YES YES NK 1 DOD

169 1 3 75 F TAG3NO NO NO NO NK 41 NED

165 1 3 75 M TAG1NO NO NO NO NO 11 NED

163 1 3 65 M TAG1NO NO NO YES YES 17 NED

162 1 3 60 M TISG3NO YES YES YES YES 15 NED

141 2 4 49 F TAG3M10 NO NO YES YES 3 DOD

135 2 4 64 F T4BG3N1 YES NO YES YES 11 DOD

133 2 4 83 M T3BG3NO YES YES YES YES 1 DOD

130 2 4 72 M T3BG3N0 NO YES NO NO 13 NED

124 2 4 59 F T3AN1M0 NO NO YES NO 9 NED

BT: Bootstrap Clustering; MD: Multidimensional Grouping; TNM: Tumor Node Metastases; Clinical outcome: NED (no evidence of disease), DOD (death of disease). TABLE

120 Genes Identified By Mann- Whitney- Wilcoxon Test

Ace. Num Name

AAOl 1414 Homo sapiens fibrinogen alpha chain preproprotein (FGA) gene, complete cds, alternatively spliced.789 (0.0)

AA021434 Homo sapiens, Similar to retinal degeneration B beta, clone MGC: 14375

IMAGE:4299595, mRNA, complete cds.325 (8e-87). Human DNA sequence from clone RP5-973N23 on chromosome 6pl2.3-21.2, complete sequence 42 (0.18)

AA021464 Homo sapiens chromosome 8 clone RPl 1-86015, complete sequence. 278 (4e-72) AA028884 Homo sapiens clone RP4-647J21, complete sequence.914 (0.0)

AA034115 Homo sapiens 12q BAC RP 11 -415112 (Roswell Park Cancer Institute Human BAC Library) complete sequence.967 (0.0)

AA035095 Homo sapiens, Similar to Breakpoint cluster region protein, uterine leiomyoma, 1 ; barrier to autointegration factor, clone IMAGE:3027737, mRNA. Homo sapiens, Similar to Breakpoint cluster region protein, uterine leiomyoma, 1 ; barrier to autointegration factor, clone MGC:14564 IMAGE:4074168, mRNA, complete cds. Homo sapiens Breakpoint cluster region protein, uterine leiomyoma, 1; barrier to autointegration factor BCRP1), mRNA.779 (0.0)

AA043806 Human beta 3-endonexin mRNA, long form and short form, complete cds.884

(O.O)Homo sapiens integrin beta 3 binding protein (beta3-endonexin) (ITGB3BP), mRNA.761 (O.O)Homo sapiens nuclear receptor co-activator NRIF3 (NRIF3) mRNA, alternatively spliced, complete cds.726 (0.0)

AA074666 Human DNA sequence from clone CTB- 1189H8 on chromosome 1 , complete sequence.121 (e-25)

AA083385 Homo sapiens BTB/POZ domain containing 1 protein (BTBD1) mRNA, complete cds.575 (e-161)

AA101348 Homo sapiens, clone IMAGE:4025624, mRNA.Homo sapiens similar to dendritic cell protein (LOC63319), mRNA. 809 (O.O)Homo sapiens dendritic cell protein (GA17), mRNA.801 (0.0)

AA127058 Homo sapiens similar to RIKEN cDNA 2610103J23 gene (LOC92140), mRNA. Homo sapiens genomic DNA, chromosome 8q23, clone: KB1907C4.535 (E-150)

AA 132065 Homo sapiens chromosome 5 clone CTC-222022, complete sequence. Homo sapiens mRNA for SMAP-5, partial cds.383 (e-103)Human DNA sequence from clone RP1- 315G1 on chromosome Xq24-25. Contains a PDZ (DHR, GLGF) domain protein pseudogene, the API3 gene for apoptosis inhibitor 3 (XIAP, HILP), a putative novel gene, ESTs, STSs, GSSs and a putative CpG island, complete sequence.44 (0.15)

AA143509 Homo sapiens pyrroline-5-carboxylate synthetase (glutamate gamma-semialdehyde synthetase) (PYCS), mRNA.993 (O.O)Human DNA sequence from clone RPl 1-7D5 on chromosome 10, complete sequence.696 (0.0)

AA147928 Human DNA sequence from clone RP4-758J18 on chromosome lp36.31-36.33, complete sequence.999 (0.0)

AA156863 Homo sapiens phosphomannomutase mRNA, complete cds.1124 (0.0)

AA 165403 Homo sapiens BM-019 mRNA, complete cds. Homo sapiens acid cluster protein 33 (ACP33), mRNA.Homo sapiens GL010 mRNA, complete cds.975 (0.0)

AA172210 Homo sapiens cDNA FLJ30082 fis, clone BGGI12000839. 327 (2e-87)

AA190401 Homo sapiens, Similar to esterase 10, clone IMAGE:3350277, mRNA.396 (e-

108)Homo sapiens chromosome 22ql 1 clone pl087110, complete sequence.Homo sapiens, BH3 interacting domain death agonist, clone MGC: 15319 IMAGE:4025880, mRNA, complete cds.394 (e-107)

AA256462 Homo sapiens, nudix (nucleoside diphosphate linked moiety X)-type motif 3, clone MGC: 12752 IMAGE:4303483, mRNA, complete cds.404 (e-110)

AA279188 Homo sapiens a disintegrin and metalloproteinase domain 8 (ADAM8), mRNA.823 (O.O)Homo sapiens mRNA for transmembrane protein, complete cds.783 (0.0) AA394148

AA402766 Homo sapiens small membrane protein 1 (SMP1) gene, complete cds.565 (e- 158)H.sapiens mRNA for rhesus polypeptide (RhVI).103 (e-19)

AA421518 Homo sapiens adaptor-related protein complex 2, sigma 1 subunit (AP2S1), mRNA. 862 (0.0). Homo sapiens adaptor-related protein complex 2, sigma 1 subunit (AP2S1), transcript variant API 7, mRNA.H.sapiens mRNS for clathrin-associated protein.805 (0.0) Homo sapiens adaptor-related protein complex 2, sigma 1 subunit (AP2S1), transcript variant AP17delta, mRNA.799 (0.0)

AA424578 Homo sapiens cDNA FLJ32289 fis, clone PROST2000432, highly similar to Homo sapiens mRNA for UDP-Gal:GlcNAc galactosyltransferase.833 (0.0) Homo sapiens, UDP-Gal:betaGlcNAc beta 1,4- galactosyltransferase, polypeptide 3, clone MGC:12774 IMAGE:3677118, mRNA, complete cds.825 (0.0)

AA425861 Homo sapiens, clone MGC:27221 IMAGE:4306900, mRNA, complete cds.Homo sapiens chromosome 19 clone LLNLF-172E10, complete sequence.507 (e-141) Homo sapiens similar to peroxisomal enoyl-coenzyme A hydratase-like protein; delta3,5- delta2,4-dienoyl-CoA isomerase; peroxisomal enoyl-CoA hydratase 1 ; dienoyl-CoA isomerase (LOCI 15289), mRNA.502 (e-140)

AA430520 Homo sapiens, CDP-diacylglycerol~inositol 3-phosphatidyltransferase

(phosphatidylinositol synthase), clone MGC: 1328 IMAGE:3139134, mRNA, complete cds.844 (O.O)Homo sapiens CDP-diacylglycerol~inositol 3- phosphatidyltransferase (phosphatidylinositol synthase) (CDIPT), mRNA.837 (O.O)Homo sapiens phosphatidylinositol synthase (PIS) mRNA, complete cds. 827 (

AA434068 Homo sapiens TRF2-interacting telomeric RAP1 protein (RAP1), mRNA.809 (0.0)

AA446453 Homo sapiens similar to PREFOLDIN SUBUNIT 5 (C-MYC BINDING PROTEIN MM-1) (MYC MODULATOR 1) (H. sapiens) (LOC121342), mRNA.Homo sapiens, prefoldin 5, clone MGC:5329 IMAGE:2900793, mRNA, complete cds.Homo sapiens mRNA for MM-1 alpha, complete cds.Homo sapiens mRNA for c-myc binding protein, complete cds.797 (0.0)

AA447696 Homo sapiens CGI-127 protein (LOC51646), mRNA.739 (0.0)

AA449831 Homo sapiens, growth factor receptor-bound protein 2, clone MGC: 1737

IMAGE:3345524, mRNA, complete cds.Homo sapiens growth factor receptor-bound protein 2 (GRB2), mRNA.Homo sapiens epidermal growth factor receptor-binding protein GRB2 (EGFRBP-GRB2) mRNA sequence.492 (e-136)

AA450227 Homo sapiens proteasome (prosome, macropain) 26S subunit.non-ATPase, 4

(PSMD4), mRNA.Homo sapiens, proteasome (prosome, macropain) 26S subunit, non-ATPase, 4, clone MGC:8410 IMAGE:2820813, mRNA, complete cds.Human antisecretory factor- 1 mRNA, complete cds. 345 (9e-93)

AA450265 Homo sapiens proliferating cell nuclear antigen (PCNA), mRNA. Human DNA sequence from clone RP4-746J20 on chromosome 20. Contains the PCNA gene for proliferating cell nuclear antigen, the 5' end of the gene HSPC274, the 5' end of the CDS2 gene for CDP-diacylglycerol synthase (phosphatidate cytidylyltransferase) 2,ESTs, STSs, GSSs and three CpG islands, complete sequence.728 (0.0) Human cyclin protein gene, complete cds. 690 (0.0)

AA454566 Homo sapiens similar to putative (H. sapiens) (LOC120036), mRNA.418 (e-114) AA454862 Homo sapiens CGI-135 protein (LOC51024), mRNA.509 (e-142)

AA455150 Human chromosome 14 DNA sequence BAC C-2011M8 of library CalTech-D from chromosome 14 of Homo sapiens (Human), complete sequence.387 (e-105)

AA455281 Homo sapiens, defender against cell death 1, clone MGC:17117 IMAGE:3454611, mRNA, complete cds.934 (O.O)Human mRNA for DAD-1, complete cds.898 (0.0)

AA456136 Homo sapiens genomic DNA, chromosome 8q23, clone: KB1460A1.852 (0.0)

AA457092

AA457162 Homo sapiens similar to RIKEN cDNA 2810417J12 gene (LOCI 14984), mRNA.Homo sapiens chromosome 16, cosmid clone 399H11 (LANL), complete sequence. 353 (6E-95) AA457725 Homo sapiens chromosome 17, clone hRPC.4_G_17, complete sequence.Homo sapiens GABA(A) receptor-associated protein (GABARAP), mRNA.Homo sapiens ganglioside expression factor 2 homolog mRNA, complete cds.783 (O.O)Homo sapiens MM46 mRNA, complete cds. Homo sapiens FLC3B mRNA for MAPI light chain 3 related protein, complete cds.767 (0.0)

AA458661

AA459663 Homo sapiens peroxiredoxin 4 (PRDX4), mRNA.Homo sapiens, thioredoxin peroxidase (antioxidant enzyme), clone MGC:22910 IMAGE:4075326, mRNA, complete cds. Human antioxidant enzyme AOE37-2 mRNA, complete cds.585 (e-165)

AA464152 Human DNA sequence from clone RP 11 -502H 18 on chromosome 1 , complete sequence.702 (0.0) Homo sapiens quiescin Q6 (QSCN6), mRNA.700 (0.0) Homo sapiens bone-derived growth factor (BPGF-1) mRNA, complete cds.652 (O.O)Human chorionic gonadotropin beta subunit mRNA, 5' flank, clone pCG-beta-474.92 (4e-16)

AA464192 Homo sapiens, Similar to hypothetical protein, clone MGC:3404 IMAGE:3530647, mRNA, complete cds.Homo sapiens HSPC227 mRNA, complete cds.Human DNA sequence from clone RP13-26D14 on chromosome Xql3.2-21.1, complete sequence.848 (0.0)

AA465031 Homo sapiens, pleckstrin homology, Sec7 and coiled/coil domains 2 (cytohesin-2), clone MGC:642 IMAGE:3538580, mRNA, complete cds.Homo sapiens cytohesin-2 mRNA, complete cds.Human Sec7p-like protein mRNA, partial cds.1003 (O.O)Homo sapiens pleckstrin homology, Sec7 and coiled/coil domains 2 (cytohesin-2) (PSCD2), mRNA.987 (O.O)H.sapiens mRNA for Arno protein.967 (0.0)

AA465378 Homo sapiens cDNA FLJ33055 fis, clone TRACH1000125, highly similar to Ig delta chain.601 (e-169)Human immunoglobulin C(mu) and C(delta) heavy chain genes (constant regions).Human Ig germline delta H-chain C-region gene, C-delta-3 domain (CLL lymphocyte).418 (e-114)

AA465593 Homo sapiens proteasome (prosome, macropain) subunit, alpha type, 3 (PSMA3), mRNA. Human mRNA for proteasome subunit HC8. 787 (0.0)

AA478268 Homo sapiens, Similar to C-terminal binding protein 1, clone MGC: 12707

IMAGE:4128336, mRNA, complete cds.414 (E-113)Homo sapiens C-terminal binding protein 1 (CTBP1), mRNA.Homo sapiens phosphoprotein CtBP mRNA, complete cds.387 (E-105)

AA485052 Homo sapiens, proteasome (prosome, macropain) 26S subunit, non-ATPase, 3, clone MGC:9893 IMAGE:3868681, mRNA, complete cds. Homo sapiens clone 308 proteasome subunit p58 mRNA, complete cds.236 (4e-60)

AA486313 Homo sapiens low density lipoprotein-related protein-associated protein 1 (alpha-2- macroglobulin receptor-associated protein 1) (LRPAP1), mRNA. Human DNA sequence from cosmid L98A6, Huntington's Disease Region, chromosome 4pl6.3 contains LRPAP1 (low-density lipoprotein-associated protein- 1) and CpG islands. Human alpha-2-macroglobulin receptor-associated protein mRNA, complete cds.593 (e-167)

AA486374 Homo sapiens lysyl-tRNA synthetase mRNA, complete cds; nuclear gene for mitochondnal product; alternatively spliced.Homo sapiens lysyl-tRNA synthetase (KARS), mRNA.Human mRNA for KIAA0070 gene, partial cds.769 (0.0)

AA486761 Homo sapiens tyrosyl-tRNA synthetase (YARS), mRNA.440 (E-121)Homo sapiens EGF-like-domain, multiple 4 (EGFL4), mRNA.40 (0.83)

AA487020 Homo sapiens isoprenylcysteine carboxyl methyltransferase (ICMT), mRNA. Human DNA sequence from clone RP1-120G22 on chromosome lp36.21-36.33, complete sequence.398 (e-109)Homo sapiens MSTP098 (MST098) mRNA, complete cds.Homo sapiens prenylcysteine carboxyl methyltransferase (PCCMT) mRNA, complete cds.391 (e-106)

AA487223 Homo sapiens synovial sarcoma translocation gene on chromosome 18-like 2

(SS18L2), mRNAHomo sapiens kiaa-iso protein mRNA, complete cds..515 (E- 144)Homo sapiens, synovial sarcoma translocation gene on chromosome 18-like 2, clone MGC:22369 IMAGE:4658643, mRNA, complete cds.51 1 (E-143) AA487265 Homo sapiens KIAAO 102 gene product (KIAAO 102), mRNA.730 (0.0) Homo sapiens similar to MICROSOMAL SIGNAL PEPTIDASE 25 KDA SUBUNIT (SPC25) (LOC94390), mRNA. Human DNA sequence from clone RP5-1053E7 on chromosome lp21.1-21.3, complete sequence.722 (0.0)

AA487899 Homo sapiens, uncharacterized hematopoietic stem/progenitor cells protein MDS027, clone MGC:17655 IMAGE:3858231, mRNA, complete cds.Homo sapiens uncharacterized hematopoietic stem/progenitor cells protein MDS027 mRNA, complete cds.232 (5e-59)

AA489400 Homo sapiens proteasome (prosome, macropain) subunit, beta type, 7(PSMB7), mRNA.Human mRNA for proteasome subunit z, complete cds.878 (O.O)Homo sapiens, proteasome (prosome, macropain) subunit, beta type, 7, clone IMAGE:3912276, mRNA, partial cds.870 (0.0)

AA490047 Homo sapiens poly(rC) binding protein 1 (PCBP1), mRNA.Human alpha-CPl mRNA, complete cds.H.sapiens hnRNP-El mRNA.1011 (0.0) H.sapiens mRNA for nucleic acid binding protein sub2.3.987 (0.0)

AA490390 Homo sapiens small acidic protein (IMAGE145052), mRNA.Homo sapiens chromosome 11, clone RPl 1-4B7, complete sequence.216 (8e-54)

AA496359 Homo sapiens similar to immediate early protein (LOC95255), mRNA.Homo sapiens immediate early protein (ETR101), mRNA.Homo sapiens, Similar to kinesin family member 5B, clone MGC: 15265 IMAGE:4297793, mRNA, complete cds.208 (6e-52)

AA496784 Homo sapiens, clone IMAGE:3959959, mRNA, partial cds. Human (chromosome

3p25) membrane protein mRNA.1015 (O.O)Homo sapiens SEC13-like 1 (S. cerevisiae) (SEC13L1), mRNA.1007 (0.0)

AA496948 Homo sapiens M5-14 protein (LOC51300), mRNA.636 (e-180)

AA504128 Human DNA sequence from clone RP4-800J21 on chromosome 20 Contains ESTs, STSs, GSSs and two CpG islands. Contains the 3' part of the RAE1 gene for a homolog to RNA export protein 1 from S.pombe and the gene for the ssDNA binding protein SEB4D (HSRNASEB).n, complete sequence.648 (O.O)Homo sapiens RAE1 (RNA export 1, S.pombe) homolog (RAE1), mRNA. 630 (e-178)

AA504617 Homo sapiens RNA-binding protein (autoantigenic) (RALY), transcript variant 1 , mRNA.Homo sapiens heterogeneous nuclear ribonucleoprotein, alternate transcript (RALY) mRNA, complete cds.256 (7e-66) Homo sapiens autoantigen p542 mRNA, complete cds. 244 (3e-62)

AA598759 Homo sapiens, phosphogluconate dehydrogenase, clone MGC:8331 IMAGE:2819330, mRNA, complete cds.571 (e-160)

AA598815 Homo sapiens proteasome (prosome, macropain) subunit, alpha type, 5 (PSMA5), mRNA.H.sapiens mRNA for macropain subunit zeta.902 (0.0)

AA620479 Human DNA sequence from clone RP 11 - 165N 19 on chromosome 9, complete sequence.313 (3e-83)

AA625981 Homo sapiens FK506 binding protein 1A (12kD) (FKBP1A), transcript variant 12B, mRNA.Homo sapiens, tubulin, beta 5, clone MGC:4029 IMAGE:3617988, mRNA, complete cds.805 (0.0)

AA629584 Homo sapiens ADP-ribosylation factor 5 (ARF5), mRNA.Human ADP-ribosylation factor (hARF5) mRNA, complete cds. 668 (0.0)

AA633757 Homo sapiens splicing factor 3b, subunit 2, 145kD (SF3B2), mRNA. Human spliceosome associated protein (SAP 145) mRNA, complete cds.682 (0.0)

AA669341 Homo sapiens, unactive progesterone receptor, 23 kD, clone MGC:4004

IMAGE.2821965, mRNA, complete cds.775 (O.O)Mus musculus telomerase binding protein, p23 (Tebp-pending), mRNA.331 (4e-48)

AA680322 Homo sapiens clone RP5-855F16, complete sequence. Homo sapiens

NADH:ubiquinone oxidoreductase MLRQ subunit (NDUFA4) mRNA, complete cds.353 (3e-95)

AA682613 Homo sapiens, craniofacial development protein 1, (CFDP1), clone MGC:5126

IMAGE:3449836, mRNA, complete cds. Homo sapiens BCNT mRNA, complete cds. Homo sapiens mRNA for p97 homologous protein, partial cds.819 (0.0)

AA683085 AA705886 Homo sapiens MAX interacting protein 1 (MXI1), mRNA. Homo sapiens, Similar to

MAX-interacting protein 1, clone MGC:9999 IMAGE:3882557, mRNA, complete cds.700 (0.0)

AA775415 Homo sapiens SMT3 suppressor of mif two 3 homolog 2 (yeast) (SMT3H2), mRNA.Homo sapiens similar to SMT3 (suppressor of mif two 3, yeast) homolog 2 (LOC91930), mRNA.H.sapiens mRNA for SMT3B protein. 498 (e-138)

AA862434

AA934762 Homo sapiens, proteasome (prosome, macropain) 26S subunit, non-ATPase, 11, clone MGC:8396 IMAGE:2820583, mRNA, complete cds.Homo sapiens mRNA for 26S proteasome subunit p44.5, complete cds.710 (0.0)

AA935560 Homo sapiens relaxin 2 (H2) (RLN2), mRNA. Human DNA sequence from clone RPl 1-12D24 on chromosome 9p23-24.3 Contains the RLN1 gene encoding two isoforms of Relaxin l(Hl),the RLN2 gene encoding two isoforms of Relaxin 2 (H2), a putative novel gene, a HGM17 (high-mobility group (non-histone chromosomal) protein 17) pseudogene, four CpG island, ESTs STSs and GSSs, complete sequence.Human mRNA for prepro-relaxin H2.367 (5e-99)

AI017703 Homo sapiens, eukaryotic translation initiation factor 3, subunit 3 (gamma, 40kD), clone MGC:8431 IMAGE:2821133, mRNA, complete cds.914(0.0)Homo sapiens eukaryotic translation initiation factor 3, subunit 3 (gamma, 40kD) (EIF3S3), mRNA.Homo sapiens translation initiation factor eIF3 p40 subunit mRNA.complete cds.894 (0.0)

H05769 Homo sapiens, clone MGC:5564, mRNA, complete cds. Human DNA sequence from clone 108K11 on chromosome 6p21 Contains SRP20 (SR protein family member), Ndr protein kinase gene similar to yeast suppressor protein SRP40, EST and GSS, complete sequence.525 (e-147)

H17158 Homo sapiens ring finger protein 4 (RNF4), mRNA.448 (e-123)

H20652 Human mRNA for KIAA0069 gene, partial cds.712 (O.O)Homo sapiens ADP- ribosylation factor-like 6 interacting protein (ARL6IP), mRNA. 698 (0.0)

H21040 Human chromosome 14 DNA sequence BAC R-241 N4 of library RPCI- 11 from chromosome 14 of Homo sapiens (Human), complete sequence.381 (e-103)

H23366

H23880 Homo sapiens hypothetical protein MGC14327 (MGC14327), mRNA.551 (e-154)

H54093 Homo sapiens mRNA for KIAA1470 protein, partial cds.496 (e-138)

H73731 Homo sapiens mRNA for KIAA0601 protein, partial cds.347 (5e-93)Human DNA sequence from clone RP1-184J9 on chromosome lp35.1-36.12, complete sequence.339 (e-90)

H84444 Homo sapiens similar to CG4332 gene product (H. sapiens) (LOC 134180), mRNA.Homo sapiens cDNA FLJ 14400 fis, clone HEMBA 1003742, weakly similar to Homo sapiens cleft lip and palate transmembrane protein 1 (CLPTM1) mRNA.Homo sapiens CRR9 mRNA for cisplatin resistance related protein CRR9p, complete cds.668 (0.0)

H93463 Human DNA sequence from clone RPl 1-108F13 on chromosome 1, complete sequence.143 9e-31)

H94897 Homo sapiens glycosyltransferase AD-017 (AD-017), mRNA.696 (O.O)Human chromosome 3p21.1 gene sequence.454 (E-125)

H99502 Homo sapiens hypothetical protein MGC2941 (MGC2941), mRNA.607 (e-171)Homo sapiens, Similar to RIKEN cDNA 2410141M05 gene, clone IMAGE:4155987, mRNA, partial cds.Homo sapiens chromosome 17, clone hRPC.4_G_17, complete sequence.605 (e-170)

N30811 ESTS

N54338 Homo sapiens, clone MGC:20354 IMAGE:4548461, mRNA, complete cds.Homo sapiens mRNA full length insert cDNA clone EUROIMAGE 2176457.801 (0.0) N69283 Homo sapiens, TAR DNA binding protein, clone MGC:1433 IMAGE:3506121, mRNA, complete cds.Homo sapiens TAR DNA binding protein (TARDBP), mRNA. Homo sapiens MASP-2 gene (partial), PM-scl gene, FRAP2 gene and CDT6 gene, clone RPCI.l l-99P18.Human DNA sequence from clone RP4-635E18 on chromosome lp36.11-36.31, complete sequence.688 (0.0)

N73536 Homo sapiens hypothetical protein (BM-002), mRNA.Homo sapiens PRO3033 mRNA, complete cds.496 (e- 138)

N91962 Homo sapiens, eukaryotic translation elongation factor 1 epsilon 1, (EEFlEl),clone

MGC: 12352 IMAGE:3685030, mRNA, complete cds. Homo sapiens pi 8 protein mRNA, complete cds.Homo sapiens mRNA for pi 8 component of aminoacyl-tRNA synthetase complex, complete cds.309 (4e-82)

R 16165 Homo sapiens hypothetical PHD zinc finger protein XAP 135 gene, complete cds.Human DNA sequence from clone RPl 1-160E12 on chromosome 6, complete sequence.716 (0.0)

R22439 Homo sapiens, transmembrane protein 4, clone MGC:1545 IMAGE:3344788, mRNA, complete cds. 640 (O.O)Homo sapiens putative secreted protein ZSIG9 (ZSIG9) mRNA, complete cds.622 (e-176) Homo sapiens transmembrane protein 4 (TMEM4), mRNA.80 (2e-12)

R24543 Homo sapiens neuroepithelial cell transforming gene 1 (NET1), mRNA.Human guanine nucleotide regulatory protein (NET1) mRNA, complete cds. Homo sapiens mRNA for Rho guanine nucleotide-exchange factor, splice variant NET 1A.283 (6e-74)

R25377 Homo sapiens DEK oncogene (DNA binding) (DEK), mRNA.Homo sapiens partial unknown mRNA from drug-resistant melanoma cells, 3'UTR, clone DSM-4.Human DNA sequence from clone 298J15 on chromosome 6p22.3-23 Contains dek (putative oncogene), EST, GSS, CA repeat, CpG island, complete sequence.H.sapiens dek mRNA.492 (E-136)

R27552 Human chromosome 14 DNA sequence BAC R-354E 14 of library RPCI- 11 from chromosome 14 of Homo sapiens (Human), complete sequence.Homo sapiens mRNA for KIAA1333 protein, partial cds.428 (E-117)

R49144 Homo sapiens cDNA FLJ30169 fis, clone BRACE2000864, highly similar to

TUBULIN ALPHA-4 CHAIN.349 (7e-94)Human HALPHA44 gene for alpha-tubulin, exons 1-3.339 (e-89)

R53889 Homo sapiens high-mobility group (nonhistone chromosomal) protein 14 (HMG14), mRNA. Human non-histone chromosomal protein HMG-14 mRNA, complete cds.204 (5e-50)

R55763 Human DNA sequence from clone RP1-198I9 on chromosome 6ql2-13. Contains the gene KIAA1411, ESTs, STSs and GSSs, complete sequence.486 (e-135)

R69307 Homo sapiens leucine aminopeptidase (LOC51056), mRNA.474 (e- 131 )Human p21 (WAF1) gene, partial promoter sequence.381 (e-103)

R76314 Homo sapiens ras homolog gene family, member G (rho G) (ARHG), mRNA. 672 (Ml

R78514 Homo sapiens, Similar to CGI 1985 gene product, clone MGC:3133

IMAGE:3352960, mRNA, complete cds. Homo sapiens hypothetical protein MGC3133 (MGC3133), mRNA. Human DNA sequence from clone 197L1 on chromosome 6q24.1-25.2. Contains ESTs, GSSs and two putative CpG islands, complete sequence.686 (0.0)

AA101348 Homo sapiens, clone IMAGE:4025624, mRNA.Homo sapiens similar to dendritic cell protein (LOC63319), mRNA. 809 (O.O)Homo sapiens dendritic cell protein (GA17), mRNA.801 (0.0)

AA127058 Homo sapiens similar to RIKEN cDNA 2610103J23 gene (LOC92140), mRNA. Homo sapiens genomic DNA, chromosome 8q23, clone: KB1907C4.535 (E-150)

AA132065 Homo sapiens chromosome 5 clone CTC-222022, complete sequence. Homo sapiens mRNA for SMAP-5, partial cds.383 (e-103)Human DNA sequence from clone RP1- 315G1 on chromosome Xq24-25. Contains a PDZ (DHR, GLGF) domain protein pseudogene, the API3 gene for apoptosis inhibitor 3 (XIAP, HILP), a putative novel gene, ESTs, STSs, GSSs and a putative CpG island, complete sequence.44 (0.15) AA143509 Homo sapiens pyrroline-5-carboxylate synthetase (glutamate gamma-semialdehyde synthetase) (PYCS), mRNA.993 (O.O)Human DNA sequence from clone RPl 1-7D5 on chromosome 10, complete sequence.696 (0.0)

T53404 Homo sapiens, hypothetical protein from clone 643, clone MGC:5115 IMAGE:2984805, mRNA, complete cds.702 (0.0)

T57815 Homo sapiens similar to U5 snRNP-specific 40 kDa protein (hPrp8-binding); prp8,

U5 snRNP-specific 40 kDa protein (H.sapiens) (LOCI 27574), mRNA.328 (3e- 60)Homo sapiens U5 snRNP-specific 40 kDa protein (hPrp8-binding)(HPRP8BP), mRNA.Homo sapiens U5 snRNP-specific 40 kDa protein mRNA, complete cds.200 (6e-49)

T67053 Homo sapiens Chromosome 22ql 1.2 BAC Clone 142e2 In IGLC Region, complete sequence. Human lambda-immunoglobulin constant region complex (germline).775 (O.O)Homo sapiens cDNA FLJ32612 fis, clone STOMA2000088, highly similar to IG LAMBDA CHAIN C REGIONS.771 (O.O)Homo sapiens mRNA for immunoglobulin lambda-3 surrogate light chain, 3 exon form, 119 bp second exon.Homo sapiens mRNA for immunoglobulin lambda-3 surrogate light chain, 3 exon form, 122 bp second exon.Homo sapiens germline mRNA for immunoglobulin lambda- 1 chain constant region, Daudi cell line.737 (0.0)

T81091 Homo sapiens coatomer protein complex, subunit alpha (COP A), mRNA.531 (5e-28)

T96829 Homo sapiens similar to cyclin-E binding protein 1 (H. sapiens) (MGC 14386), mRNA.178 (3e-48)

W49619 Homo sapiens cadherin 2, type 1, N-cadherin (neuronal) (CDH2), mRNA.Homo sapiens chromosome 17, clone CTD-2023G8, complete sequence.825 (0.0)

W69906 Homo sapiens uridine monophosphate kinase (UMPK), mRNA. Homo sapiens uridine-cytidine kinase 2 (UCK2) mRNA, complete cds.373 (e-101)

W96107 Homo sapiens Sec61 gamma (SEC61G), mRNA.609 (e-172)

TABLE 5

92 Genes Identified By Single- Variable Logistic Regression

Accession NAME Number

AA010393 Homo sapiens chromosome 17, clone hRPK.214 0 1, complete sequence. 599 (e-69)

AA019591 Homo sapiens PAC clone RP5-1186P10 from 7ql l.21-q21.1, complete sequence.605 (e-170)Homo sapiens, GTF2I repeat domain-containing 1, clone MGC:9316 IMAGE:3913745, mRNA, complete cds. Homo sapiens general transcription factor 3 (GTF3) mRNA, complete cds. Homo sapiens RBAP2 (RBAP2) mRNA, complete cds. Homo sapiens putative transcription factor (WBSCR12) mRNA, complete cds. Homo sapiens muscle TFII-I repeat domain-containing protein 1 mRNA, complete cds. 414 (e-113)

AA024832 Homo sapiens mRNA; cDNA DKFZp586I0324 (from clone DKFZp586I0324). 660

(O.O)Human DNA sequence from clone RPl 1-23P11 on chromosome 13 Contains part of the GPC6 gene encoding Glypican 6, ESTs, STSs and GSSs, complete sequence.365 (2e-98) Homo sapiens leukocyte differentiation antigen (CD84) gene, partial cds.40 (1.4)

AAl 13339 Homo sapiens cDNA FLJ32537 fis, clone SMINT2000400, highly similar to Homo sapiens FRG1 mRNA.706 (0.0) Homo sapiens FSHD region gene 1 (FRG1), mRNA.601 (e-169)

AAl 15248 Homo sapiens BAC clone RPl 1-278G12 from 2, complete sequence. 862 (0.0) Homo sapiens cDNA FLJ31353 fis, clone MESAN2000264.

AA121704 Human DNA sequence from clone RPl 1-319123 on chromosome 10, complete sequence, citb 10 k 1, complete sequence. 630 (e-178)

AA134595 Homo sapiens chromosome 5 clone CTD-2353F22, complete sequence. 64 (le-7)

AAl 42875 Homo sapiens chromosome 10 clone RPl 1-176H12, complete sequence. Homo sapiens chromosome 19, cosmid R27516, complete sequence.46 (0.028)

AA157797 Homo sapiens chromosome 19, cosmid R29368, complete sequence.741 (O.O)Homo sapiens egf-like module containing, mucin-like, hormone receptor-like sequence 2 (EMR2), mRNA. Homo sapiens EGF-like module EMR2 (EMR2) mRNA, complete cds.256 (e-65)

AA165400 Homo sapiens hypothetical gene supported by AK021643 (LOC136384), 44 (0.14)Homo sapiens BAC clone RPl 1-832D10 from 7, complete sequence.

AA284268 Homo sapiens hypothetical gene supported by XM_072511 (LOC137861), mRNA.1096 (0.0) Human DNA sequence from clone RPl 1-31E23 on chromosome lq31.3-32.1 Contains STSs and GSSs, complete sequence. Human DNA sequence from clone RP3-324N14 on chromosome 6q23.1-24.3 Contains part of protein 4.1-G mRNA, STSs and GSSs, complete sequence.40 (2.2)

AA284292 H.sapiens mRNA for beta-l,4-galactosyltransferase (EC 2.4.1.22). 813 (0.0) Homo sapiens UDP-Gal:betaGlcNAc beta 1 ,4- galactosyltransferase, polypeptide 1 (B4GALT1), mRNA. 799(0.0)

AA404694 Homo sapiens PTK2 protein tyrosine kinase 2 (PTK2), mRNA. Human focal adhesion kinase (FAK) mRNA, complete cds. 747 (0.0) Homo sapiens PTK2 protein tyrosine kinase 2 (PTK2), mRNA. Homo sapiens focal adhesion kinase mRNA, complete cds. 731 (0.0)

AA406603 Human DNA sequence from clone RPl 1-475E11 on chromosome 1 , complete sequence. Homo sapiens mRNA for KIAA0761 protein, partial cds.890 (0.0) Human DNA sequence from clone 1177E19 on chromosome lp36.12-36.31. Contains the 3' part of the DNA-binding Zinc finger protein RIZ gene, ESTs, an STS, GSSs and a CpG island, complete sequence.44 (0.12) Homo sapiens fer-1-like 3, myoferlin (C. elegans) (FER1L3), mRNA. Homo sapiens myoferlin (MYOF) mRNA, complete cds, alternatively spliced.40 (1.9)

AA421783 Homo sapiens zinc finger protein 263 (ZNF263), mRNA. Homo sapiens mRNA for zinc finger protein FPM315, complete cds.791 (0.0) AA424834 Homo sapiens DC1 (DC1) mRNA, complete cds. 787 (0.0)

AA429399 Homo sapiens chromosome 1 clone RPl 1-86H7, complete sequence.593 (e-167)

Human DNA sequence from clone 283K11 on chromosome 6q23.1-24.3. Contains part of the EYA4 gene for eyes absent Drosophila) homolog 4. Contains ESTs and GSSs, complete sequence.42 (0.31)

AA431184 Human chromosome 14 DNA sequence BAC R- 1078H9 of library RPCI- 11 from chromosome 14 of homo sapiens (Human), complete sequence.579 (e-163)Homo sapiens chromosome 14 clone BAC257P13 map 14q31, complete sequence. 571 (e- 160)Homo sapiens mucin and cadherin-like (MUCDHL), transcript variant 4, mRNA. 42 (0.38)Homo sapiens MUCDHL (MUCDHL) gene, complete cds, alternatively spliced.

AA435936 Homo sapiens Xp22 bins 169-171 BAC GSHB-383H3 (Genome Systems Human BAC Library) complete sequence.496 (e-138) Homo sapiens genomic DNA, chromosome 1 lq, clone:RPl 1-29912, complete sequence.42 (0.34)

AA436158 Homo sapiens signaling adaptor protein DIP13alpha mRNA, complete cds. Homo sapiens, Similar to adaptor protein containing pH domain, PTB domain and leucine zipper motif, clone IMAGE:4295177, mRNA. Homo sapiens adaptor protein APPL mRNA, complete cds.872 (0.0)

AA436871 Homo sapiens syntaxin 3 A (STX3A), mRNA. Homo sapiens genomic DNA, chromosome 1 lq, clone:CMB9-26D16, complete sequence.617 (e-174)

AA443193 Human DNA sequence from clone RPl 1-164H16 on chromosome 6, complete sequence.662 (0.0)

AA443285 Homo sapiens hypothetical protein FLJ10769 (FLJ10769), mRNA. 777 (0.0) Homo sapiens cDNA FLJ14198 fis, clone NT2RP3002512. 769 (0.0)

AA453607 Human DNA sequence from clone RP1-180E22 on chromosome 6pl 1.2-12.3. Contains the 3' part of the gene for a novel protein with possible Calmodulin like calcium- binding domains, the gene KIAA0057, ESTs, STSs, GSSs and a putative CpG island, complete sequence.375 (e-101)

AA453748 Homo sapiens 12q24.1 BAC RPCI11-946P6 (Roswell Park Cancer Institute Human BAC Library) complete sequence.777 (O.O)Homo sapiens KE03 protein mRNA, partial cds.121 (5e-25)Homo sapiens NY-REN-25 antigen (NY-REN-25), mRNA. 115 (3e-23)

AA454579 Homo sapiens genomic DNA, chromosome 1 lq, clone:CTD-2313N18,561 (e-157)

AA454625 Homo sapiens genomic cytochrome P450, subfamily IIIA (niphedipine oxidase) (CYP3A) on chromosome 7. Homo sapiens hypothetical protein MGC5521 (MGC5521), mRNA. Homo sapiens cDNA FLJ32683 fis, clone TESTI2000120, weakly similar to Homo sapiens mRNA for zinc finger 3 (ZF3 gene).470 (e-132)

AA455119 Homo sapiens COP9 (constitutive photomorphogenic, Arabidopsis, homolog) subunit 7A (COPS7A), mRNA.1017 (0.0)

AA457374 Homo sapiens BAC clone RPl 1-57B24 from 4, complete sequence.702 (0.0). Human

DNA sequence from clone RPl 1-567B20 on chromosome 1, complete sequence. 46 (0.023)

AA459950 Homo sapiens, ribosomal protein L4, clone MGC: 15542 IMAGE:3050317, mRNA, complete cds.Human DNA sequence from clone RP4-599G15 on chromosome lpl2- 13.2, complete sequence.

AA460365 Homo sapiens ALS2CR4 mRNA, complete cds,.664 (O.O)Homo sapiens potential

LAG1 interactor mRNA, partial cds.113 (E-22)Homo sapiens neurexin Ill-alpha gene, partial cds.44 (0.082)

AA463958 Human DNA sequence from clone RP4-564M11 on chromosome lp31.1 Contains 3' end of PIGK(phosphatidylinositol glycan, class K) gene, a novel mRNA, part of a gene similar to sialyltranferase, ESTs, CA repeat, STSs and GSSs, complete sequence.345 (2e-92)

AA482325 Homo sapiens hypothetical gene supported by U18919; AL360167; BC006354;

NM_025233 (LOC95925), mRNA. Homo sapiens nucleotide binding protein (NBP), mRNA. Homo sapiens nucleotide binding protein (NBP) mRNA, complete cds.680 (0.0) Human chromosome 17q 12-21 mRNA, clone pOV-2, partial cds.664 (0.0) AA488526 Homo sapiens, Similar to nucleolar phosphoprotein pl30, clone MGC:5049

IMAGE:2900024, mRNA, complete cds. 741 (O.O)Human mRNA for KIAA0035 gene, partial cds.Homo sapiens similar to ORF (H. sapiens) (LOCI 18975), mRNA. 733 (O.O)Human DNA sequence from clone RPl 1-302K17 on chromosome 10, complete sequence.

AA488645 Homo sapiens transcriptional regulatory protein p54 mRNA, complete cds.1027 (0.0) homo sapiens NGFI-A binding protein 1 (EGR1 binding protein 1) (NAB1), mRNA. Homo sapiens NGFI-A binding protein 1 (EGR1 binding protein 1) (NAB1), mRNA. 1013 (0.0) Human transcriptional repressor (NAB1) NAB1 mRNA, complete cds.999 (0.0) Homo sapiens cell-line KG1 transcriptional regulatory protein p54 mRNA, complete cds.684 (0.0)

AA489246 Homo sapiens serine protease TADG15 mRNA, complete cds.1076 (0.0) Homo sapiens suppression of umorigenicity 14 (colon carcinoma, matriptase, epithin) (ST14), mRNA. Homo sapiens, Similar to uppression of tumorigenicity 14 (colon carcinoma, matriptase, epithin), clone IMAGE:2960020, mRNA, artial cds. Homo sapiens mRNA for prostamin, complete cds.1068 (0.0) Homo sapiens suppression oftumorigenicity 14 (colon carcinoma, matriptase, epithin) (ST14), mRNA. Homo sapiens matriptase mRNA, complete cds.1031 (0.0) Homo sapiens membrane-type serine protease 1 mRNA, complete cds. 1027 (O.O)Human SNC19 mRNA sequence.333 (e-88)

AA489661 Human DNA sequence from clone RPl 1-48715 on chromosome 10, complete sequence.799 (0.0)

AA496780 Homo sapiens, RAB7, member RAS oncogene family, clone MGC:8453 IMAGE:2821435, mRNA, complete cds. 1005 (0.0)

AA504894 Homo sapiens, Similar to RIKEN cDNA 5830420C20 gene, clone MGC: 10104

IMAGE:3898917, mRNA, complete cds.870 (0.0) Homo sapiens mannosidase, beta A, lysosomal (MANBA) gene, and ubiquitin-conjugating enzyme E2D 3 (UBE2D3) genes, complete cds.38 (6.9)

AA599093 Human DNA sequence from clone RPl 1-568G11 on chromosome 1, complete sequence.918 (O.O)Homo sapiens cDNA FLJ11866 fis, clone HEMBA1006973, highly similar to Homo sapiens rab3-GAP regulatory domain mRNA. 335 (2e-89)

AA609067 Homo sapiens cDNA FLJ25432 fis, clone TST06444. 779 (O.O)Homo sapiens chromosome 10 clone RPl 1-216P13, complete sequence.52 (4e-4)

AA609134 BAC sequence from the SPG4 candidate region at 2p21-2p22 BAC 559D11 of RPCI- 11 library from chromosome 2 of Homo sapiens (Human), complete sequence.670 (O.O)Homo sapiens baculoviral IAP repeat-containing 6 (BIRC6), mRNA. Homo sapiens ubiquitin-conjugating BIR-domain enzyme APOLLON mRNA, complete cds. Homo sapiens mRNA for KIAA1289 protein, partial cds. 238 (3e-60)

AA621335 Homo sapiens hypothetical protein FLJ11088 (FLJ11088), mRNA.747 (0.0)

AA705060 Homo sapiens similar to calcium binding protein P22 (LOCI 15833), mRNA. Homo sapiens calcium binding protein P22 (CHP), mRNA. Human calcium-binding protein chp mRNA, complete cds.513 (e-143) Human DNA sequence from clone RPl 1- 288H12 on chromosome 6 Contains the 3' part of the IGF2R (insulin-like growth factor 2 receptor) gene, a gene for an organic cation transporter protein, ESTs, STS, GSSs and CpG islands, complete sequence.Homo sapiens IGF2R gene, complete cds.367 (3e-99)

AA708310 Human DNA sequence from clone RPl 1-337C18 on chromosome 1, complete sequence. 404 (e-110)

H09747 Human DNA sequence from clone CTB-1048E9 on chromosome 22 Contains an

RPS3A (Ribosomal Protein S3A) pseudogene, the gene for a novel protein similar to ASPH (aspartate beta-hydroxylase, EC 1.14.11.16), the gene for anovel protein, ortholog of mouse tuftelin-interacting protein 10 (similar to worm C07E3.1A) two more novel genes, ESTs, STSs, GSSs and three putative CpG islands, complete sequence.567 (e-159)Homo sapiens mRNA; cDNA DKFZp761P039 (from clone DKFZp761P039); partial cds.

H09818 Homo sapiens chromosome 5 clone CTB-47B11, complete sequence.343 (4e-92)Homo sapiens hypothetical protein PRO 1331 (PRO 1331), mRNA.321 (e-85)

H10335 Homo sapiens Chromosome 1 lpl4.3 PAC clone pDJ292d23, complete sequence.595 (e-168) Homo sapiens RAD21 homolog (S. pombe) (RAD21), mRNA.40 (1.2) H17335 Homo sapiens chromosome 17 map 17q21.1, complete sequence. Homo sapiens microtubule-associated protein tau (MAPT), mRNA.525 (e-146)

H23277 Homo sapiens chromosome 2 clone RPl 1-295N18, complete sequence.547 (e-153)

Homo sapiens activin A receptor, type II (ACVR2), mRNA.Human osteosarcoma mRNA for activin typell A receptor, complete cds.

H29292 Homo sapiens similar to ecotropic viral integration site 5 (H. sapiens) (LOCI 26595), mRNA.749 (0.0) Homo sapiens EVI5 homolog mRNA, complete cds.636 (e-180)

H41096 Homo sapiens chromosome 10 clone RPl 1-369L1, complete sequence. 714 (0.0)

H53141 Human DNA sequence from clone RP1-297M16 on chromosome 6 Contains STSs and GSSs, complete sequence.96 (3e-17)

H58736 Human chromosome 14 DNA sequence BAC C-3059L23 of library CalTech-D from chromosome 14 of Homo sapiens (Human), complete sequence.587 (e-165).Homo sapiens DMR protein mRNA, complete cds.573 (e-161)

H65834 Homo sapiens chromosome 4 clone RPl 1-38904, complete sequence.50 (0.001)

H70815 Human DNA sequence from clone RPl l-571F15 on chromosome 9, complete sequence.551 (e-154)Human DNA sequence from clone RPl 1-507C10 on chromosome 6q25.2-26, complete sequence. Human DNA sequence from clone RPl 1-108L7 on chromosome 10. contains part of the gene for a novel Insulin-like growth factor binding type protein with Kazal-type serine protease inhibitor domain, the gene for a novel protein similar to rat tricarboxylate carrier, the gene for a novel PDZ (DHR, GLGF) domain protein, the gene for a novel protein similar to KIAA0552, KIAA0341 and Fugu hypothetical protein 2, the gene for a novel protein similar to Plasmodium POM1 and C. elegans F46G11.1, a putative novel gene, the EMA4G gene for semaphorin 4G and a novel gene. Contains ESTs, STSs, GSSs and seven putative CpG islands, complete sequence. Human DNA sequence from clone RPl -310013 on chromosome 20ql 1.2 Contains part of) four novel genes, a putative novel gene, ESTs, STSs, GSSs and four putative CpG islands, complete sequence. Homo sapiens DNA sequence from clone 394P21 on chromosome lp36.12-36.13. Contains the PAX7 gene, locus D1S2644, ESTs and STSs, complete sequence.107 (8e-21)

H93463 Human DNA sequence from clone RP 11 - 108F 13 on chromosome 1 , complete sequence.143 (e-31) Human DNA sequence from clone 272L16 on chromosome lq32.1-32.3. Contains the 3' end of the LAMB3 gene for Laminin, Beta 3 (Nicein, Kalinin, BM600) and a novel Rat Ca2+/Calmodulin dependent Protein Kinase LIKE gene. Contains ESTs, STSs, GSSs, genomic marker D1S491 and a ca repeat polymorphism, complete sequence.125 (3e-26)

H95989 Homo sapiens similar to embryonic seven-span transmembrane protein-like protein (H. sapiens) (LOC135428), mRNA. Human DNA sequence from clone RP1-302G2 on chromosome 6pl 1.2-21.1, complete sequence. 474 (e-131) Human 90 kD heat shock protein gene, complete cds.184 (3e-44)

N21548 Human DNA sequence from clone 301K23 on chromosome lp35.1-36.21. Contains the

5' part of a novel gene similar to predicted yeast and worm genes. Contains ESTs and GSSs, complete sequence.1017 (0.0)

N38891 Homo sapiens PAC clone RP4-701O16 from 7q33-q36, complete sequence.821 (0.0)

N56882 Homo sapiens BAC clone RP 11 -467H 10 from 7, complete sequence.811 (O.O)Homo sapiens clone RP4-802G15, complete sequence.763 (0.0)

N58283 Homo sapiens BAC clone RPl 1-273D4 from 2, complete sequence.359 (6e-97)Homo sapiens similar to ribosomal protein S2; 40S ribosomal protein S2 (H. sapiens (LOCI 39909), mRNA.40 (0.69)

N66933 Homo sapiens chromosome 19, cosmid R31237, complete sequence.369 (e-99) Homo sapiens candidate tumor suppressor pp32rl (PP32R1) gene, complete cds.135 (4e-99)

N94428 Homo sapiens E1A binding protein p300 (EP300), mRNA. 500 (e-139)

R08891 Homo sapiens KIAA0676 protein (KIAA0676), mRNA.593 (e-167)

R09585 Homo sapiens rec (LOC51201), mRNA.579 (e-163) Homo sapiens genomic DNA of

8p21.3-p22 anti-oncogene of hepatocellular colorectal and non-small cell lung cancer , segment 3/11.264 (6e-68)

R22271 Human DNA sequence from clone RPl 1-45A16 on chromosome 9q32-33.3, complete sequence.428 (e-117) R28669 Human DNA sequence from clone RP5-991C6 on chromosome 6ql4.1-15. Contains the gene for a novel protein similar to C. elegans F55A12.9 (Tr:P91086), an RPL10 (60S ribosomal protein L10) pseudogene, ESTs, STSs, GSSs and a putative CpG island, complete sequence.240 (6e-61)

R36449 Human DNA sequence from clone RP5-1169J3 on chromosome 1 lpl3, complete sequence. 597 (e-168)Homo sapiens chromosome 19 clone CTC-548K16, complete sequence.62 (4e-7)Human DNA sequence from clone RP4-609E1 on chromosome lp31.2-32.1, complete sequence. 60 (e-6)Homo sapiens BAC clone GS1-207A4 from 7pl l.2-ρ21, complete sequence. Human beta-tubulinseudogene.58 (6e-6)

R43525 Homo sapiens kinesin heavy chain member 2 (KIF2), mRNA. 515 (e-144)H.sapiens mRNA for kinesin-2.

R44132 Homo sapiens chromosome 18, clone RPl 1-756M1, complete sequence.379 (e-102)

Human DNA sequence from clone RPl 1-278E14 on chromosome 6, complete sequence.44 (0.086)

R51080 Homo sapiens hypothetical gene supported by AF086185; BC011266 (LOC93556), mRNA. 76 (3e-l l)

R56219 Human DNA sequence from clone RP 11 -88K15 on chromosome 6, complete sequence.46 (0.028)Human DNA sequence from clone 37J18 on chromosome lp36.2- 36.3.Contains a putative novel gene, ESTs and GSSs, complete sequence.44 (0.11)

R56432 Homo sapiens BAC clone RPl 1-68E19 from 2, complete sequence. 40 (1.7)Human

DNA sequence from clone RPl 1-146P20 on chromosome 10, complete sequence.Human DNA sequence from clone RPl 1-569H20 on chromosome X, complete sequence.Human DNA sequence from clone RPl 1-99J16 on chromosome 1, complete sequence.Human DNA sequence from clone RP3-377F16 on chromosome 22 Contains part of one or two novel genes, ESTs and GSSs, complete sequence.

R60053 Homo sapiens hypothetical gene supported by AF086442; AK022764; AK022851

(LOC136361), mRNA. 498 (e-138)Homo sapiens BAC clone RPl 1-511P7 from 7, complete sequence.Homo sapiens cDNA FLJ 12789 fis, clone NT2RP2001947.

R60927 Homo sapiens mRNA full length insert cDNA clone EUROIMAGE 1603443.178 (3e- 42}

R64066 Homo sapiens genomic DNA, chromosome 1 lq, clone:RPl 1-823021, complete sequence.739 (0.0)H. sapiens CpG island DNA genomic Msel fragment, clone 191e9, forward read cpgl91e9.ftlb. 48 (0.008)Homo sapiens membrane-bound aminopeptidase P (XNPEP2) gene, complete cds. 46 (0.030)

R92455 Homo sapiens, LIM protein (similar to rat protein kinase C-binding enigma), clone

MGC:2010 IMAGE:3345715, mRNA, complete cds.408 (e-11 l)Human DNA sequence from clone RPl 1-334A14 on chromosome 1, complete sequence.74 (e-10)

R94943 Human chromosome 14 DNA sequence BAC R-841020 of library RPCI- 11 from chromosome 14 of Homo sapiens (Human), complete sequence.599 (e-169)

R98628 Homo sapiens chromosome 4 clone RPl 1-397E7, complete sequence.733 (0.0) Human

DNA sequence from clone RP3-341D10 on chromosome X Contains a gene for a novel protein, part of the gene for a protein similar to ADP ribosylation factor 3, part of a gene similar to HTF9C and a CpG island, complete sequence. Homo sapiens BAC clone GS1-155M11 from 7q21-q22, complete sequence.174 (e-141) Homo sapiens genomic protocadherin alpha cluster (PCDHA@) on chromosome 5.167 (e-138)

R99918 Human DNA sequence from clone RPl 1-552113 on chromosome 10, complete sequence. 531 (e-148)

T50370 Human DNA sequence from clone RP5-1069P2 on chromosome 20 Contains the 5' end of the STK4 gene for serine/threonine kinase 4, the gene for outer mitochondrial membrane translocase HTOM34P, the gene for a novel PABPC1 (poly(A)-binding protein, cytoplasmic 1) (PABPL1) like protein. ESTs, STSs, GSSs and a putative CpG island, complete sequence. 349 9e-93)

T55592 Homo sapiens chromosome 19 clone CTD-2192J16, complete sequence.54 (5e-4)

T61792 Homo sapiens pyruvate dehydrogenase kinase 4 mRNA, 3' untranslated region, partial sequence.603 (e-160)

T68461 Homo sapiens fibrinogen, B beta polypeptide (FGB) gene, complete cds.565 (e-159) T71680 Homo sapiens BAC clone RPl 1-111H13 from 2, complete sequence.829 (0.0) Homo sapiens mitochondrial ribosomal protein L30 (MRPL30), mRNA.456 (e-126) Homo sapiens similar to testican 3 (LOCI 15443), mRNA. Homo sapiens testican 3 (HSAJ1454), mRNA.40 (1.8)

T86983 Homo sapiens chromosome 17, clone HCIT524C5, complete sequence.626 (e- 117)Human carnitine palmitoyltransferase (CPT1) mRNA, complete cds.82 (4e-13)

T90641 Human DNA sequence from clone RPl 1-337C18 on chromosome 1 , complete sequence.498 (e-138)Homo sapiens similar to SMHS2 (H. sapiens) (LOC121883), mRNA.40 (1.0)

T96711 Homo sapiens hypothetical protein FLJ14153 (FLJ14153), mRNA. Homo sapiens cDNA FLJ14153 fis, clone NT2RM1000092, weakly similar to MULTIDRUG RESISTANCE PROTEIN 2.676 (0.0) Homo sapiens mRNA for SMAP-4, complete cds.626 (e-177)

W31919 Homo sapiens chromosome 4 clone RPl 1-240A2, complete sequence.44 (0.11) Homo sapiens partial TTN gene for titin.42 (0.44)

W56308 Human chromosome 14 DNA sequence BAC R-840I19 of library RPCI-11 from chromosome 14 of Homo sapiens (Human), complete sequence.511 (e-142) Homo sapiens gastrointestinal glutathione peroxidase (GPX2) gene, complete cds. 220 (2e-55) Homo sapiens T-cell receptor alpha delta locus from bases 250472 to 501670 (section 2 of 5) of the Complete Nucleotide Sequence.40 (1.7)

W99364 Human DNA sequence from clone RPl 1-379P1 on chromosome 9, complete sequence.

686 (0.0)

TABLE 6

Four Groups Identified By Multidimensional Analysis

Group Accession Name

Group 1 R34801 Homo sapiens, clone IMAGE:4281881, mRNA. 416 (e-114)Homo sapiens, clone IMAGE:4282266, mRNA.Homo sapiens cDNA FLJ11047 fis, clone PLACE 1004510, highly similar to Homo sapiens cofactor of initiator function mRNA.

Group 1 AA416585 Hmo sapiens ACE-related carboxypeptidase ACE2 mRNA, complete cds. 761 (O.O)Homo sapiens angiotensin I converting enzyme (peptidyl-dipeptidase A)

2 (ACE2), mRNA.

Group 1 W30935 Homo sapiens cDNA: FLJ21715 fis, clone COL10287, highly similar to AF071569 Homo sapiens multifunctional calcium/calmodulin-dependent protein kinase II delta2 isoform mRNA. 730 (O.O)Homo sapiens cDNA FLJ31080 fis, clone HSYRA2001615, highly similar to Sus scrofa Calcium/calmodulin-dependent protein kinase II delta 2-subunit mRNA. 704 (O.O)Homo sapiens calcium/calmodulin-dependent protein kinase (CaM kinase) II delta (CAMK2D), mRNA. 702 (0.0)

Group 1 R89104 Human DNA sequence from clone RP1-134N8 on chromosome 20pl2.

Contains STSs, GSSs and a CpG island, complete sequence. 52 (4e-4) Group 1 AA056381 Homo sapiens CGI-04 protein (LOC51067), mRNA.476 (-132)Homo sapiens, clone MGC:22937 IMAGE:4843916, mRNA, complete cds.Homo sapiens cDNA FLJ13995 fis, clone Y79AA1002209, weakly similar to TYROSYL- TRNA SYNTHETASE (EC 6.1.1.1).

Group 1 R24258 Homo sapiens similar to protein kinase C, zeta (LOCI 13121), mRNA. 313 (3E-83)

Group 1 N74284 Homo sapiens similar to zinc finger protein (LOC90812), mRNA. 680 (0.0)

Group 1 N93470 Homo sapiens hypothetical protein FLJ10948 (FLJ10948), mRNA. 460 (E-

127) Group 1 N67578 Human chromosome 14 DNA sequence BAC R-747H7 of library RPCI-11 from chromosome 14 of Homo sapiens (Human), complete sequence. 48

(0.006)Human aquaporin-5 (AQP5) gene, exon 4 and complete cds.

Group 1 R23189 Homo sapiens, clone hRPK.l l A l, complete sequence. 42 (0.41)Human DNA sequence from clone RPl 1-421P21 on chromosome 6, complete sequence.

Group 1 AA460289 Homo sapiens BAC clone RP11-182H20 from Y, complete sequence. 617 (- 174) Homo sapiens testis transcript Y 7 (TTY7) mRNA, partial cds, alternatively spliced. 325 (E-86)

Group 1 N30621 Human DNA sequence from clone RPl 1-631 F7 on chromosome 6 Contains

STSs, GSSs and a CpG island, complete sequence. 894 (0.0) Group 1 N62696 Homo sapiens mRNA full length insert cDNA clone EUROIMAGE

288936.801 (0.0)

Group 1 N74963 Homo sapiens chromosome 5 clone RP11-546B8, complete sequence. 831 (0.0) Human DNA sequence from clone RP1-211D12 on chromosome 20ql2- 13.2 Contains the 3' end of the STK4 gene encoding serine/threonine kinase 4, the KCNS1 gene encoding the potassium voltage-gated channel, delayed- rectifier (subfamily S, member 1), a gene similar to Elafin-like protein from mouse, a putative novel gene, a putative p53 responsive gene (PRG5) a CpG island, ESTs, STSs and GSSs, complete sequence. 46 (0.27)

Group 1 AA457253 Human DNA sequence from clone 281H8 on chromosome 6q25.1-25.3. Contains up to four novel genes, one with similarity to KIAA0323 and worm C30F12.1 and another with Ubiquitin-Like protein gene SMT3 (the latter in an intron of a novel gene). Contains ESTs, STSs, GSSs, a putative CpG island and genomic marker D6S1553, complete sequence.811 (0.0) Homo sapiens mRNA for KIAA0733 protein, partial cds.

Group 2 AA620674 Homo sapiens BAC clone RPl 1-546P22 from 2, complete sequence.78 (7e- 9)Homo sapiens chromosome 10 clone RPl 1-26702, complete sequence.Human DNA sequence from clone RPl 1-498G22 on chromosome 10, complete sequence.Human DNA sequence from clone RPl 1-380120 on chromosome 9, complete sequence.

Group 2 R98047 Homo sapiens BAC clone CTB-103H13 from 7q31, complete sequence.611

(e-172)Homo sapiens full length insert cDNA clone YR42A07. Group 2 H97496 Homo sapiens chromosome 2 clone RPl 1-33708, complete sequence.420 (e-

_LL

Group 2 AA425543 Homo sapiens chromosome 8, clone RPl 1-498N9, complete sequence. 609 (e-

172) Group 3 W68127 Human DNA sequence from clone RPl 1-87H20 on chromosome 13, complete sequence.589 (E-165)Homo sapiens clone HAW1052 unknown mRNA.579 (E- 162}

Group 3 AA420993 Homo sapiens Bardet-Biedl syndrome 4 (BBS4), mRNA.720 (0.0)

Group 3 AA010182 Homo sapiens 12 BAC RPl 1-593B8 (Roswell Park Cancer Institute Human BAC Library) complete sequence. 749 (0.0)

Group 3 H72878 Human chromosome 14 DNA sequence BAC R-552K16 of library RPCI-11 from chromosome 14 of Homo sapiens (Human), complete sequence.593 (E-

167)

Group 3 AA406020 Homo sapiens interferon-stimulated protein, 15 kDa (ISG15), mRNA.833 (O.O)Human interferon-induced 17-kDa/15-kDa protein mRNA, complete cds.

Human interferon-induced 15-Kd protein (ISG) gene, exon809 (0.0)

Group 3 H28734 Homo sapiens chromosome 4 clone RPl 1-372N22, complete sequence. Homo sapiens glutamate receptor, ionotropic, AMPA 2 (GRIA2), mRNA. Human glutamate receptor 2 (HBGR2) mRNA, complete cds.555 (E-155)

Group 3 AA026167 Homo sapiens chromosome 22ql 1 clone pl43il3, complete sequence.712

_(

Group 3 N80491 Homo sapiens KIAA0630 protein (KIAA0630), mRNA.355 (2E-95)

Group 3 AA455302 Homo sapiens p33INGlb (ING1) mRNA, complete cds.276 (E-71) Homo sapiens INGl gene, exons la, lb, lc. Homo sapiens ING1 tumor suppressor, variant B (INGl) mRNA, complete cds.272 (2E-70)Homo sapiens growth inhibitory protein INGl (INGl) gene, alternatively spliced exon lb.56 (6E-6) Group 3 H19111 Human DNA sequence from clone RPl 1-149P14 on chromosome 1, complete sequence.42 (0.39)Human DNA sequence from clone RP3-404K8 on chromosome 6p22.2-22.3 Contains PRL (prolactin) gene, STSs and GSSs, complete sequence.O (1.5)

Group 3 R22579 Homo sapiens genomic sequence surrounding Notl site, clone NR5- CA3C.377 (E-102) Human DNA sequence from clone RP4-531H16 on chromosome 20pl 1.22-12. Contains the 3' end of the PCSK2 gene for proprotein convertase subtilisin/kexin type 2 (NEC2), the BFSP1 gene for beaded filament structural protein 1 (filensin), a protein 91/23 (mouse Dynein light chain, TCTEX-1 like) pseudogene, a Ubiquitin-40S Ribosomal protein S27A fusion protein pseudogene, ESTs, STSs, GSSs and a CpG island, complete Sequence.44 (0.098)

Group 3 AA054978 Homo sapiens similar to Similar to RIKEN cDNA 0710001C05 gene (H. sapiens) (LOC122961), mRNA.837 (O.O)Human DNA sequence from clone CTA-373H7 on chromosome 22ql 1.22-12.2. Contains STSs, GSSs and a CpG Island, complete sequence.502 (E-139)Homo sapiens collagen, type IV, alpha 6 (COL4A6), mRNA.42 (0.49)

Group 3 T98075 Homo sapiens chromosome 5 clone CTD-2093G19, complete sequence.400 (e-109)Human DNA sequence from clone RP4-707K17 on chromosome 20ql3.1 Contains part of the PTPRT gene encoding a protein tyrosinephosphatase receptor type T, ESTs, STSs and GSSs, complete sequence.40 (1.4)

Group 3 WO 1031 Human DNA sequence from clone RP3-341E18 on chromosome 6pl 1.2-12.3. Contains the KIAA0936 gene for a MAK-related kinase and a novel alternatively spliced gene. Contains a putative CpG island, ESTs and GSSs, complete sequence.678 (0.0)

Group 3 N52195 Homo sapiens genomic DNA, chromosome 21q, section 104/ 105. Homo sapiens genomic DNA, chromosome 6pl 1.2-12.3-ter, Ter region, clone:T1136, complete sequence.278 (3e-72)

Group 3 T98796 Homo sapiens MADS box transcription enhancer factor 2, polypeptide C (myocyte enhancer factor 2C) (MEF2C), mRNA.369 (e-99)

Group 3 R34121 Homo sapiens mRNA for KIA A0720 protein, partial cds.490 (e- 136)

Group 3 AA455172 Homo sapiens BAC clone RPl 1-92L24 from 2, complete sequence.674 (0.0) Human DNA sequence from clone RPl 1-360J12 on chromosome 9q21.12-

21.32, complete sequence.56 (2e-5)

Group 3 T60111 Homo sapiens genomic DNA, chromosome 1 lq, clone:CMB9-26D16, complete sequence.230 (2e-58)Homo sapiens fatty acid binding protein 5

(psoriasis-associated) (FABP5), mRNA. 222 (5e-56)

Group 3 W80719 Homo sapiens BAC clone RPl 1-260K18 from 4, complete sequence.339 (5e- 91)Homo sapiens excision and cross link repair protein (ERCC4) gene, complete genomic sequence.52 (2e-4)

Group 3 N30302 Homo sapiens similar to GTP-binding protein (LOC95301), mRNA.Homo sapiens, Similar to guanine nucleotide binding protein-like 1 , clone IMAGE:4098352, mRNA.Homo sapiens genomic DNA, chromosome 6p21.3, HLA Class I region, section 12/20.Homo sapiens cDNA: FLJ22738 fis, clone HUV00522, highly similar to HUMHSPR Human GTP-binding protein (HSR1) mRNA.341 (3e-91)

Group 3 AA488171 Homo sapiens mRNA for KIAA1743 protein, partial cds.757 (0.0)

Group 3 W86997 Homo sapiens chromosome 5 clone CTC-332L22, complete sequence.751 ) Group 3 AA479920 Homo sapiens cDNA FLJ31951 fis, clone NT2RP7007177, weakly similar to Homo sapiens multiple membrane spanning receptor TRC8 m.RNA.666 (0.0) Group 3 AA436454 Homo sapiens genomic sequence surrounding Notl site, clone NR1-OH7C. 515 (e-144)

Group 3 AA169814 Homo sapiens sorting nexin 2 (SNX2), mRNA.581 (e-163)

Group 3 N68327 Human DNA sequence from clone RPl 1-177H22 on chromosome 10, complete sequence.787 (0.0) Group 3 H96213 Homo sapiens similar to early development regulator 1 ; homolog of polyhomeotic 1; homolog of mouse Rae28 (H. sapiens) (LOCI 21482), mRNA.Homo sapiens early development regulator 1 (homolog of polyhomeotic 1) (EDR1), mRNA.Homo sapiens gene for polyhomeotic 1 homolog, partial cds, exon 15.519 (e-145)

Group 3 N64741 Homosapiens, slug (chicken homolog), zinc finger protein, clone MGC: 17388 IMAGE:3911047, mRNA, omplete cds. Homo sapiens zinc finger protein

SLUG (SLUG) gene, complete cds. 206 (e-50)

Group 3 AAl 64229 Homo sapiens BAC clone RPl 1-567F11 from 2, complete sequence.387 (e-

105) Group 3 AA496792 Human DNA sequence from clone RPl 1-447M12 on chromosome 9, complete sequence.799 (0.0) Group 3 R25614 Homo sapiens chromosome 5 clone CTD-2299E8, complete sequence.230 (8e- 58} .

Group 3 AA131794 Homo sapiens genomic DNA, chromosome 1 lq clone:RPl 1- 746C14,complete sequence.42 (0.22)

Group 3 H90627 Human DNA sequence from clone RPl 1-199016 on chromosome 9, complete sequence.581 (e-163) Human DNA sequence from clone RP1-28F12 on chromosome 20ql 1.22-12 Contains part of the KIAA0823 gene, ESTs, STSs and GSSs, complete sequence.100 (2e-18)

Group 3 AAl 52351 Homo sapiens PAC clone RP5-978E18 from 7p21, complete sequence.517 (e- 144}

Group 3 AA464708 Homo sapiens PNAS-7 mRNA, partial sequence.741 (0.0)

Group 3 H63518 Homo sapiens BAC clone RPl 1-270E5 from 2, complete sequence.Homo sapiens PAC clone RP5-81904 from 7q33-q35, complete sequence.42 (0.37)Homo sapiens mannose receptor, C type 1 (MRC1), mRNA.38 (5.7)

Group 3 AA457138 Homo sapiens frizzled homolog 8 (Drosophila) (FZD8), mRNA. Homo sapiens FZD8 mRNA for seven-transmembrane receptor Frizzled-8, complete cds.753 (0.0)

Group 3 AA126673 Homo sapiens chromosome 5 clone CTD-2306M10, complete sequence.813

Group 3 AA432268 Homo sapiens BAC clone GS1-96J14 from 7pl 1.2-p21, complete sequence. 884 (0.0)

Group 3 AA446013 Homo sapiens ST5 gene for suppression of tumorigenicity 5, L27a gene for ribosomal protein L27a and KIAA0298 gene. Homo sapiens gene for ribosomal protein L27A, complete cds.624 (e-176)

Group 3 H47327 Homo sapiens transforming, acidic coiled-coil containing protein 1 (TACCl), mRNA.Homo sapiens mRNA for KIAA1103 protein, partial cds.468 (e-129) Group 3 R76275 Homo sapiens ASCL3 gene, CEGP1 gene, Cl lorfl4 gene, Cl lorfl5 gene, Cl lorflό gene and Cl lorf!7 gene.159 (2e-36)

Group 3 AA609976 Homo sapiens similar to mitochondnal capsule selenoprotein (H.sapiens)

(LOCI 27476), mRNA. Homo sapiens mitochondrial capsule selenoprotein (MCSP), mRNA.890 (0.0)

Group 3 H68988 Homo sapiens f-box and leucine-rich repeat protein 5 (FBXL5), transcript variant 2, mRNA.414 (e-113 Group 3 R72174 Homo sapiens, membrane interacting protein of RGS16, clone MGC:23190 IMAGE:4855079, mRNA, complete cds. Human Chromosome 16 BAC clone CIT987SK-A-363E6, complete equenceHuman Chromosome 16 BAC clone

CIT987SK-A-363E6, complete sequence.527 (e-147)

Group 3 H99816 Homo sapiens procollagen-lysine, 2-oxoglutarate 5-dioxygenase (lysine hydroxylase) 2 (PLOD2), mRNA.529 (e-148)

Group 3 N51987 Homo sapiens chromosome 10 clone RP11-397P14, complete sequence.46

(0.031)

Goup 4 T41203 Homo sapiens chromosome 4 clone RPl 1-678H22, complete sequence.474

(E-131)

Goup 4 AA460369 Homo sapiens, clone MGC: 19524 IMAGE:4329693, mRNA, complete cds.682 (0.0)

Goup 4 H90848 Human DNA sequence from clone RPl 1-286F20 on chromosome 1 Contains part of a novel gene for KIAA1383 protein, ESTs, STSs and GSSs, complete sequence.82 (5e-13)

Goup 4 AA663435 Homo sapiens tripartite motif-containing 28 (TRIM28), mRNA. Homo sapiens, Similar to KRAB-ssociated protein 1, clone IMAGE:3831930, mRNA, partial cds. Homo sapiens, KRAB-associated protein 1, clone MGC:3849 IMAGE:2906325, mRNA, complete cds. 977 (O.O)Human nuclear corepressor KAP-1 (KAP-1) mRNA, complete cds. 973 (O.O)Human transcriptional corepressor hKAPl/TIFlB mRNA, complete cds. H.sapiens mRNA for TIFlbeta zinc finger protein. 957 (0.0)

Goup 4 R89581 Homo sapiens BAC clone RPl 1-795L3 from 2, complete sequence.98 (8e-18)

Goup 4 H73591 Homo sapiens similar to cytochrome b5 outer mitochondrial membrane precursor (H. sapiens) (LOCI 24229), mRNA. Homo sapiens cytochrome b5 outer mitochondrial membrane precursor (CYB5-M), mRNA.341 (3e-91)

Goup 4 H48105 Homo sapiens PAC clone RP4-593H12 from 7p31, complete sequence. 383 (e-104) Homo sapiens protein similar to E.coli yhdg and R. capsulatus nifR3 (PP35), mRNA. 379 (e-103) Human PP35 mRNA, complete cds. 331 (2e-88)

Goup 4 H56152 Homo sapiens BAC clone RPl 1-575M4 from 7, complete sequence.432 (e- 118 Goup 4 R98344 Homo sapiens, transmembrane 4 superfamily member 9, clone MGC:9300 IMAGE:3895933, mRNA, complete cds. Homo sapiens tetraspan NET-4 mRNA, complete cds.291 (3e-76) Homo sapiens tetraspan 5 (TSPAN-5), mRNA. Homo sapiens tetraspan TM4SF (TSPAN-5) gene, complete cds. 274

(6e-71)

Goup 4 AA481480 Homo sapiens, KIAA0255 gene product, clone IMAGE:3507918, mRNA. Human DNA sequence from clone RP5-836N17 on chromosome 20ql 1.1- 11.21 Contains part of the HCK (hemopoietic cell kinase) gene, the KIAA0255 gene, a ribosomal protein L30 pseudogene, ESTs, STSs, GSSs and CpG Islands, complete sequence.954 (0.0)

Goup 4 R02069 Homo sapiens hnRNP 2H9E mRNA, complete cds.432 (e-119) Homo sapiens, Similar to heterogeneous nuclear ribonucleoprotein H3 (2H9), clone

IMAGE:3927179, mRNA, partial cds.167 (8e-39)

Goup 4 AA971406 Homo sapiens hypothetical gene supported by D87682 (LOC136725), mRNA. Homo sapiens clone UWGC:djsl or RPl 1-16G1 from 7pl4-15, complete sequence. Human mRNA for KIAA0241 gene, partial cds. 753 (0.0)

Goup 4 R00884 Homo sapiens dihydrofolate reductase (DHFR), mRNA. Human dihydrofolate reductase gene, exon 6 and 3' flank.402 (e-110) Homo sapiens genomic DNA, chromosome 6p21.3, HLA Class I region, section 6/20. Homo sapiens genomic DNA, 237 kb segment from 6p21.3 region including HLA genes, complete sequence. 252 (e-64)

Goup 4 T61475 Homo sapiens clone 23664 and 23905 mRNA sequence. Homo sapiens cDNA: FLJ23599 fis, clone LNG15473, highly similar to AF035315 Homo sapiens clone 23664 and 23905 mRNA sequence.591 (e-166)

Goup 4 H05635 Homo sapiens chromosome 16 clone RP11-137H10, complete sequence.589

(e-166) Goup 4 AA453603 Homo sapiens orphan nuclear receptors (NR1I2) gene, complete cds, alternatively spliced.680 (0.0) Human DNA sequence from clone RPl 1-90H3 on chromosome 1, complete sequence.48 (0.006)

Goup 4 N77198 Homo sapiens chromosome 17, clone RP11-272E10, complete sequence.470 (e-130) Homo sapiens chromosome 19, BAC 273239 (CIT-B-320G13), complete sequence.153 (2e-34)

Goup 4 AA035730 Homo sapiens chromosome 5 clone CTC-202F10, complete sequence.829

(0-0) Goup 4 T86959 Homo sapiens chromosome 5 clone CTC-529P8, complete sequence.872 (0.0) Homo sapiens PAC clone RP4-547C10 from 7p21-p22, complete sequence.205 (5e-50) Homo sapiens chromosome 15 clone CTD-2306A12 map 15q21.1, complete sequence. 196 (e-47)

Goup 4 AA625806 Homo sapiens ninjurin 1 (NINJ1), mRNA. Human adhesion molecule ninjurin mRNA, complete cds.454 (e-125) Goup 4 AA489314 Homo sapiens similar to putative (LOC96645), mRNA. Homo sapiens p25 mRNA, complete cds. Homo sapiens, Similar to gp25L2 protein, clone MGC:2142 IMAGE:2967520, mRNA, complete cds.676 (0.0) Homo sapiens sulfotransferase family, cytosolic, IC, member 2 (SULT1C2), mRNA. Homo sapiens SULT1C sulfotransferase (SULT1C) mRNA, complete cds. 476 (e-

132)

Goup 4 AA432108 Homo sapiens serine racemase (SRR), mRNA. Homo sapiens cDNA FLJ13107 fis, clone NT2RP3002501, weakly similar to THREONINE DEHYDRATASE CATABOLIC (EC 4.2.1.16).979 (0.0) Homo sapiens chromosome 17 sequence from PAC RPCI-5 1037N22 map 17ql3.3 region D17S695-D17S654, complete sequence. 858 (0.0) Homo sapiens mRNA for

KIAA1401 protein, partial cds.848 (0.0)

Goup 4 AA088438 Homo sapiens chromosome 16 clone CTC-508F8, complete sequence.145 (4e- 32) TABLE 7A

Association Between The Expression of The Identified Markers With Tumor Stage

And Grade

TABLE 7B

Association Between The Expression of The Identified Markers With The Expression of p53 And pRB

* Significant also for total RB. Data from this example revealed that expression profiling segregated superficial and invasive tumors. Moreover, these clusters rendered predictive information. Multidimensional analyses supported such clustering while also identifying those superficial tumors with expression profiles similar to invasive lesions. The analytical approach undertaken identified genes that were associated with tumor stage and grade, as well as altered p53 and pRB expression. Furthermore, p33INGl showed a significant association with patient survival when validated in tissue microarrays. Overall, both clusters and individual targets showed clinical value for subtype classification and prognosis of patients with bladder cancer.

Following the segregation of tumor subtypes by gene clustering, analysis of data focused on gene identification methods and validation using tissue microarrays. Different algorithms were applied for gene identification providing distinct ranked gene lists. The Mann- Whitney- Wilcoxon test is a standard means for gene identification between two groups. In this case, p21 and cyclin E were selected for further validation. p21 was found to be associated with tumor stage and grade, in accordance with previous series.

Single-variable logistic regression is a standard classification/discrimination model to rank gene by their classification performance. np2 and cytokeratin-20, known soluble proteins, were selected for further validation, and their expression shown to be associated with tumor stage and grade, np-2 is a transmembrane receptors for semaphorins (mediators of neuronal guidance), and for several angiogenic factors, including vascular endothelial growth factor 145 (VEGF 145) and VEGF 165. It has been reported that osteosarcomas overexpressing np-2 had increased vascularity and poorer prognosis, suggesting that np-2 acts as a VEGF- amplifier in these tumors. An association between np-2 expression and tumor progression has also been reported for certain neoplasms, including prostate and lung cancer. The association of cytokeratin 20 with stage and grade had previously been reported for bladder cancer both in tissue and urine specimens. Ninjurin and p33INGl were selected among the target genes differentially expressed in the poor prognostic groups identified by multidimensional analysis. Ninjurin a nerve injury induced protein involved in neuronal growth, is known to be altered in hepatocellular carcinoma, acute lymphoblastic leukemia, and was reported to be down-regulated by p53. Nevertheless, ninjurin was not found to be significantly associated with tumor stage and grade in the cohort of bladder cancer patients analyzed. The expression of p33INGl was significantly associated with tumor stage and grade, as well as with overall patient survival. p33INGl has been reported to cooperate with p53 in blocking cell proliferation and enhancing apoptosis, and it has been ascribed as candidate tumor suppressor gene. Its expression has been reported to be involved in the progression of lymphoid tumors. Down-regulation of p33INGl was associated with pRB, p21 and cyclin E expression, but not with p53. The SVM algorithm revealed that WNT and mitotic spindle checkpoint are among the most important pathways altered during bladder cancer progression. Deregulation of WNT was observed in the present analysis. Furthermore, WNT signaling and the mitotic spindle checkpoint are related. WNT signaling induces alignment and orientation of the mitotic spindle presumably by directly targeting the cytoskeleton. Relationships between p53 and spindle checkpoint are also being established. For example, it has been described that 53BP1, a direct p53 binding protein that contains two BRCT domains and implicated in early response to DNA damage, significantly co-localizes with CENP-E to kinetochores. 53BP1 is loaded to kinetochores in prophase, before CENP-E, and is released by mid- anaphase.

The following references were cited herein: Cheung et al., Making and reading microarrays. Nat. Genet. 21(Suppl):15-19 (1999). Cristiani & Shawe-Taylor, Support Vector Machines and other kernel-based learning methods, Cambridge University Press, Cambridge, (2000).

Dawson-Saunders and Trapp, Basic & Clinical Biostatistics. 2nd edition, Norwalk,

Connecticut, Appleton & Lange (1994). De las Rivas et al., Genome Research 12:567-583 (2002).

Dudoit, Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. U.C. Berkeley Dept of Statistics Technical

Report #578 (2000). Eisen et al., Proc. Natl. Acad. Sci. USA 95:14863-14868 (1998). Felsenstein, Evolution 39:783-791 (1985).

Fleming and Lin, Biometrics 56:971-983 (2000).

Hoos et al., Am. J. Pathol. 158:1245-1251 (2001).

Kaplan and Meier, J. Am. Stat. 53:457-481 (1958). Li & Yang, How many genes are needed for a discriminant microarray data analysis, in Methods of Microarray Data Analysis, Eds Lin et al. (Kluwer Academic), pp.137-150 (2002).

McShane et al., Clin. Cancer Res. 6:1854-1864 (2000).

Reyment & Joreskog, Applied Factor Analysis in the Natural Sciences. Cambridge University Press, Cambridge, ( 1996).

Saitou & Nei, Mol. Biol. Evol. 4:406-425 (1987).

Stears et al., Physiol. Genomics 3:93-99 (2000).

Tseng et al., Nucleic Acids Res. 29:2549-2557 (2001).

Tudor and Koch, Stat. Methods Med. Res. 3:345-381 (1994). Wikman et al., Clin. Chem. 46:1555-1561 (2000).

Xie et al., Genome Res. 12:785-94 (2002).

Xiong et al., Mol. Genet. Metab. 73:239-47 (2001).

Any patents or publications mentioned in this specification are indicative of the levels of those skilled in the art to which the invention pertains.

Further, these patents and publications are incorporated by reference herein to the same extent as if each individual publication was specifically and individually indicated to be incorporated by reference.

Claims

WHAT IS CLAIMED IS:

1. A method of diagnosing squamous metaplasia of bladder cancer in an individual, comprising the steps of: collecting biological samples from said individual; and determining the expression of caveolin-1 or keratin 10 in said samples, wherein the expression of caveolin-1 or keratin 10 in said samples indicates the presence of squamous metaplasia of bladder cancer.

2. The method of claim 1, wherein expression of caveolin-1 or keratin 10 is determined at protein level.

3. The method of claim 1 , wherein expression of caveolin- 1 or keratin 10 is determined at nucleic acid level.

4. A method of diagnosing bladder cancer in an individual, comprising the steps of: collecting biological samples from said individual; and determining in said samples the level of gene expression for a protein selected from the group consisting of zyxin, E-cadherin, moesin, cytokeratin

20, neuropilin 2, p21 and p33INGl, wherein said level of expression correlates with the stage and grade of bladder cancer in said individual.

5 . The method of claim 4, wherein said gene expression is determined at protein level.

6. The method of claim 4, wherein said gene expression is determined at nucleic acid level.

7. A method of discriminating between superficial and invasive bladder cancer in an individual, comprising the steps of: collecting biological samples from said individual; and determining in said samples the level of gene expression for a protein selected from the group consisting of zyxin, E-cadherin, moesin, p21, cytokeratin 20, neuropilin-2 and p33INGl, wherein said protein is differentially expressed in superficial and invasive bladder cancer.

8. The method of claim 7, wherein said gene expression is determined at protein level.

9. The method of claim 5, wherein said gene expression is determined at nucleic acid level.

1 0. A method of predicting survival outcome of an individual having bladder cancer, comprising the steps of: collecting biological samples from said individual; and determining in said samples the level of gene expression for p33INGl or moesin, wherein said level of expression correlates with survival outcome of said individual.

1 1 . The method of claim 7, wherein said gene expression is determined at protein or nucleic acid level.

12. A method of discriminating between superficial and invasive bladder cancer in an individual, comprising the steps of: collecting biological samples from said individual; and determining in said samples the expression of a gene identified by an accession number selected from the group consisting of AA01 1414, AA021434, AA021464, AA028884, AA0341 15, AA035095, AA043806, AA074666, AA083385, AA101348, AA127058, AA132065, AA143509, AA147928, AA156863, AA165403, AA172210, AA190401, AA256462, AA279188, AA394148, AA402766, AA421518, AA424578, AA425861, AA430520, AA434068, AA446453, AA447696, AA449831, AA450227, AA450265, AA454566, AA454862, AA455150, AA455281 , AA456136, AA457092, AA457162, AA457725, AA458661, AA459663, AA464152, AA464192, AA465031 , AA465378, AA465593, AA478268, AA485052, AA486313, AA486374, AA486761, AA487020, AA487223, AA487265, AA487899, AA489400, AA490047, AA490390, AA496359, AA496784, AA496948, AA504128, AA504617, AA598759, AA598815, AA620479, AA625981, AA629584, AA633757, AA669341, AA680322, AA682613, AA683085, AA705886, AA775415, AA862434, AA934762, AA935560, AI017703, H05769, H17158, H20652, H21040, H23366, H23880, H54093, H73731, H84444, H93463, H94897, H99502, N30811, N54338, N69283, N73536, N91962, R16165, R22439, R24543, R25377, R27552, R49144, R53889, R55763, R69307, R76314, R78514, T53404, T57815, T67053, T81091, T96829, W49619, W69906, W96107, AA010393, AA019591, AA024832, AAl 13339, AAl 15248, AA121704, AA134595, AA142875, AA157797, AA165400, AA284268, AA284292, AA404694, AA406603, AA421783, AA424834, AA429399, AA431184, AA435936, AA436158, AA436871, AA443193, AA443285, AA453607, AA453748, AA454579, AA454625, AA4551 19, AA457374, AA459950, AA460365, AA463958, AA482325, AA488526, AA488645, AA489246, AA489661, AA496780, AA504894, AA599093, AA609067, AA609134, AA621335, AA705060, AA708310, H09747, H09818, H10335, H17335, H23277, H29292, H41096, H53141, H58736, H65834, H70815, H93463, H95989, N21548, N38891, N56882, N58283, N66933, N94428, R08891, R09585, R22271, R28669, R36449, R43525, R44132, R51080, R56219, R56432, R60053, R60927, R64066, R92455, R94943, R98628, R99918, T50370, T55592, T61792, T68461, T71680, T86983, T90641, T96711, W31919, W56308, W99364, wherein said gene is differentially expressed at the mRNA level in superficial and invasive bladder cancer.

13. A method for identifying the presence or absence of a squamous metaplasia of bladder cancer phenotype in a cell or cells, comprising determining the expression level of caveolin-1 or keratin 10 in said cell or cells, wherein a detectable expression level of caveolin-1 or keratin 10 in said cell or cells indicates the presence of squamous metaplasia of bladder cancer phenotype and an undetectable level of caveolin-1 or keratin 10 in said cell or cells indicates the absence of squamous metaplasia of bladder cancer phenotype.

14. A method of identifying the presence or absence of a squamous metaplasia of bladder cancer in an individual, comprising the steps of:

(a) collecting a biological sample from said individual; and

(b) determining the expression level of caveolin-1 or keratin 10 in said sample, wherein a detectable expression level of caveolin-1 or keratin 10 in said sample indicates the presence of said squamous metaplasia of bladder cancer and an undetectable level of caveolin-1 or keratin 10 in said sample indicates the absence of said squamous metaplasia of bladder cancer.

15. The method of claim 1 or 2, wherein the expression level is a protein expression level.

16. The method of claim 1 or 2, wherein the expression level is a nucleic acid expression level.

17. A method of identifying the presence or absence of a bladder cancer in an individual, comprising the steps of: collecting a biological sample from said individual; and determining in said sample the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin, cytokeratin 20, neuropilin 2, p21 and p33INGl, wherein said level of expression indicates the presence or absence of a bladder cancer in said individual.

18. The method of claim 17, further comprising correlating said expression level with the stage and grade of bladder cancer in said individual.

19. The method of claim 17 or 18, wherein expression of said protein is determined at protein level.

20. The method of claim 17 or 18, wherein expression of said protein is determined at nucleic acid level.

21. A method of discriminating between a superficial and an invasive bladder cancer in an individual, comprising the steps of: collecting a biological sample from said individual; and determining in said sample the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin and p33INGl, wherein said protein is differentially expressed in superficial and invasive bladder cancer.

22. The method of claim 21, wherein said level of expression is a level of expressed protein.

23. The method of claim 21, wherein said level of expression is a level of nucleic acid expression.

24. A method of predicting survival outcome of an individual having bladder cancer, comprising the steps of: collecting biological samples from said individual; and determining in said samples the level of expression of p33INGl, wherein said level of expression correlates with survival outcome of said individual.

25. The method of claim 24, wherein expression of said protein is determined at protein or nucleic acid level.

26. A kit for identifying the presence or absence of a squamous metaplasia of bladder cancer phenotype in a cell or cells, comprising a reagent or reagents capable of determining the expression level of caveolin-1 or keratin 10 in said cell or cells, wherein detectable expression levels of caveolin-1 or keratin 10 in said samples indicates the presence of squamous metaplasia of bladder cancer phenotype and undetectable levels of caveolin-1 or keratin 10 in said cell or cells indicates the absence of squamous metaplasia of bladder cancer phenotype.

27. A kit for identifying the presence or absence of bladder cancer in an individual, the kit comprising a reagent or reagents capable of determining the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin, cytokeratin 20, neuropilin 2, p21 and p33INGl.

27. A kit for discriminating between superficial and invasive bladder cancer in an individual, the kit comprising a reagent or reagents capable of determining the level of expression of a protein selected from the group consisting of zyxin, E-cadherin, moesin and p33INGl .

28. A kit for predicting survival outcome of an individual having bladder cancer, the kit comprising a reagent or reagents capable of determining in said samples the level of expression of p33INGl.

29. The kit according to any one of claims 26-28 wherein at least one reagent is an antibody.

30. The kit according to any one of claims 26-28 wherein at least one reagent is a nucleic acid.

31. The kit according to any one of claims 26-28 further comprising instructions for correlating said level of expression with a clinical diagnosis.