US007615349B2

(12) United States Patent (10) Patent No.: US 7,615,349 B2 Riker et al. (45) Date of Patent: Nov. 10, 2009

(54) MELANOMA GENESIGNATURE Barrow et al., “Tumor Antigen Expression in Melanoma Varies According to Antigen and Stage.” Clin. Can. Res., vol. 12:764-771 (75) Inventors: Adam I. Riker, Mobile, AL (US); (2006). Steven Alan Enkemann, Lutz, FL (US) Bauskin et al., “Role of Macrophage Inhibitory Cytokine-1 in Tumorigenesis and Diagnosis of Cancer. Cancer Research, vol. (73) Assignees: H. Lee Moffitt Cancer Center and 66:4983-4986 (2006). Research Institute, Inc., Tampa, FL Baylin et al., “Alterations in DNA Methylation: A Fundamental (US); University of South Florida, Aspect of Neoplasia.” Adv. in Cancer Res, vol. 72: 141-196 (1998). Tampa, FL (US) Bittner et al., “Molecular classification of cutaneous malignant mela noma by expression profiling.” Nature, vol. 406:536-540 (*) Notice: Subject to any disclaimer, the term of this (2000). patent is extended or adjusted under 35 Brasseur et al. “Expression of Mage in Primary and Metastatic U.S.C. 154(b) by 0 days. Cutaneous Melanoma.” Int. J. Cancer, vol. 63:375-380 (1995). Cascinelli et al., “Sentinel Lymph Node Biopsy in Cutaneous Mela (21) Appl. No.: 11/852,102 noma: The WHO Melanoma Program Experience.” Annals of Surgi cal Oncology, vol. 7:469-474 (2000). (22) Filed: Sep. 7, 2007 Chen et al., “Decreased PITX1 in lung cancer.” Lung Cancer, vol. 55:287-294 (2007). (65) Prior Publication Data Coppola et al., “Correlation of Osteopontin Expression and US 2008/O113360A1 May 15, 2008 Pathological Stage across a Wide Variety of Tumor Histologies.” Clin. Can. Res., vol. 10:184-190 (2004). Related U.S. Application Data DeRisi et al., “Use of a cDNA microarray to analyze gene expression patterns in human cancer.” Nature Genetics, vol. 14:457-460 (1996). (60) Provisional application No. 60/824,849, filed on Sep. Dobbin et al., “Interlaboratory Comparability Study of Cancer Gene 7, 2006. Expression Analysis Using Oligonucleotide Microarrays.” Clinical Cancer Research, vol. 11:565-572 (2005). (51) Int. Cl. Eton et al., “Prognostic Factors for Survival of Patients Treated CI2O I/68 (2006.01) Systemically for Disseminated Melanoma.” Journal of Clinical (52) U.S. Cl...... 435/6 Oncology, vol. 16:1103-1111 (1998). (58) Field of Classification Search ...... 435/6 Fujiwara et al., “Isolation of a candidate tumor Suppressor gene on See application file for complete search history. 8p21.3-p22 that is homologous to an extracellular domain of the PDGF receptor beta gene.” Oncogene, vol. 10:891-895 (56) References Cited (1995). Gallagher et al., “Multiple markers for melanoma progression regu U.S. PATENT DOCUMENTS lated by DNA methylation: Insights from transcriptomic studies.” 7,056,674 B2 6, 2006 Baker et al. Carcinogenesis, vol. 26:1856-1867 (2005). 7,081,340 B2 7/2006 Baker et al. Garraway et al., “Integrative genomic anlayses identify MITF as a 7,171,311 B2 1/2007 Dai et al. lineage Survival oncogene amplified in malignant melanoma.” 7,247.426 B2 7/2007 Yakhini et al. Nature, vol. 436:117-122 (2005). 2006, O183141 A1 8/2006 Change et al. Golub et al., “Molecular Classification of Cancer: Class Discovery 2007/O154889 A1 7/2007 Wang et al. and Class Prediction by Gene Expression Monitoring.” Science, vol. 286:531-537 (1999). Haddad et al., “The Progression of Melanoma Nodal Metastasis is OTHER PUBLICATIONS Dependent on Tumor Thickness of the Primary Lesion.” Annals of Surgical Oncology, vol. 6: 144-149 (1999). Tockman et al (Cancer Res., 1992, 52:271 1s-2718s).* Hanahan et al., “The Hallmarks of Cancer. Cell, vol. 100:57-70 Greenbaum et al. (Genome Biology, 2003, vol. 4, Issue 9, pp. 117.1- (2000). 117.8).* Haqq et al., “The gene expression signatures of melanoma progres Glas et al., "Converting a breast cancer microarray signature into a sion.” Proc. Natl. Acad. Sci. U. S. A., vol. 102:6092-6097 (2005). high-throughput diagnostic test.” BMC Genomics, vol. 7:278-288 Harbig et al., “A sequence-based identification of the genes detected (2006). by probesets on the Affymetrix U133 plus 2.0 array.” Nucleic Acids Ai et al., “Epigenetic Silencing of the Tumor Suppressor Cystatin M Occurs during Breast Cancer Progression.” Cancer Research, vol. Research, vol. 33.e31 (2005). 66:7900-7909 (2006). (Continued) An et al., “Deletion of tumor Suppressor genes in Chinese non-Small cell lung cancer.” Cancer Letters, vol. 184:189-195 (2002). Primary Examiner Sean E Aeder Balch et al. “A New American Joint Committee on Cancer Staging (74) Attorney, Agent, or Firm Fish & Richardson P.C. System for Cutaneous Melanoma.” Cancer, vol. 88: 1484-1491 (2000). (57) ABSTRACT Balch et al., “Final version of the American Joint Committee on Cancer Staging System for Cutaneous Melanoma.” Journal of Clini cal Oncology, vol 19:3635-3648 (2001). Methods and compositions for evaluating gene expression in Balch et al., “Prognostic Factors Analysis of 17,600 Melanoma Patients: Validation of the American Joint Committee on Cancer melanoma samples are provided herein. Melanoma Staging System.” Joint of Clinical Oncology, vol. 19:3622-3634 (2001). 8 Claims, 6 Drawing Sheets US 7,615,349 B2 Page 2

OTHER PUBLICATIONS Warrington et al., “Comparison of human adult and fetal expression and identification of 535 housekeeping/maintenance genes.” Physiol. Herman et al., “Gene Silencing in Cancer in Association with Pro Genomics, vol. 2: 143-147 (2000). moter Hypermethylation.” N. Engl. J. Med., vol. 349:2042-2054 Weyers et al., “Classification of Cutaneous Malignant Melanoma: A (2003). Reassessment of Histopathologic Criteria for the Distinction of Dif Hodi. “Well-DefinedMelanoma Antigens as Progression Markers for ferentTypes.” Cancer, vol. 86:288-299 (1999). Melanoma: Insights into Differential Expression and Host Response Winnepenninckx et al., “Gene Expression Profiling of Primary Cuta Based on Stage.” Clin. Can. Res., vol. 12:673-678 (2006). neous Melanoma and Clinical Outcome,” J. Natl. Cancer Inst., vol. Jaeger et al., “Gene Expression Signatures for Tumor Progression, 98:472-482 (2006). Tumor Subtype, and Tumor Thickness in Laser-Microdissected Wollmann et al., “The macrophage inhibitory cytokine integrates Melanoma Tissues.” Clin. Can. Res., vol. 13:806-815 (2007). AKT/PKB and MAP kinase signaling pathways in breast cancer Jemal et al., "Cancer Statistics, 2007, C.A. Cancer J. Clin., vol. cells.” Carcinogenesis, vol. 26:900-907 (2005). 57:43-66 (2007). Zhang et al., “Aberrant promoter methylation and silencing of the Jensen et al., “Down-Regulation of Pro-Apoptotic Genes is an Early POU2F3 gene in cervical cancer.” Oncogene, vol. 25:5436-5445 Event in the Progression of Malignant Melanoma.” Annals of Surgi (2006). cal Oncology, vol. 14:1416-1423 (2007). Zhou et al., “Osteopontin Expression Correlates with Melanoma Kim et al., “Epigenomic Profiling Reveals Novel and Frequent Tar Invasion.” J. Invest. Dermatol., vol. 124:1044-1052 (2005). gets of Aberrant DNA Methylation-Mediated Silencing in Malignant Alonso et al., “A High-Throughput Study in Melanoma Identifies Glioma.” Cancer Res., vol. 66:7490-7501 (2006). Epithelial-Mesenchymal Transition as a Major Determinant of Kolfschoten et al., “A genetic screen identifies PITX1 as a suppressor Metastasis.” Cancer Research, vol. 67:3450-3460 (2007). of RAS activity and tumorigenicity.” Cell, vol. 121:849-858 (2005). Baldi et al., “Identification of genes down-regulated during mela Lerebours et al., “Fine Deletion Mapping of Chromosome 8p in noma progression: a cDNA array study.” Experimental Dermatology, Non-Small-Cell Lung Carcinoma.” Int. J. Cancer, vol. 81:854-858 vol. 12:213-218 (2003). (1999). Becker et al., “Discrimination of Melanocytic Tumors by cDNA Levy et al., “MITF. master regulator of melanocyte development and Array Hybridization of Tissues Prepared by Laser Pressure Catapult melanoma oncogene.” Trends Molecular Medicine, vol. 12:406-414 ing.” The Journal of Investigative Dermatology, vol. 122:361-368 (2006). (2004). Li et al., “Model-based analysis of oligunucleotide arrays: Expres De Wit et al., “Analysis of differential gene expression in human sion index computation and outlier detection.” Proc. Natl. Acad. Sci. melanocytic tumour lesions by custom made oligonucleotide arrays.” U.S.A., vol. 98:31-36 (2001). British Journal of Cancer, vol. 92:2249-2261 (2005). Lord et al... “Increased CDX2 and decreased PITXI homeobox gene Dooley et al., “DNA Microarrays and Likelihood Ratio expression in Barrett's esophagus and Barrett’s-associated Bioinformatic Methods: Discovery of Human Melanocyte Biomark adenocarcinoma.” Surgery, vol. 138:924-931 (2005). ers.” Pigment Cell Res., vol. 16:245-253 (2003). McGill et al., “c-Met Expression is Regulated by Mitf in the Eisen et al., "Cluster analysis and display of genome-wide expression Melanocyte Lineage.” J. Biol. Chem., vol. 281: 10365-10373 (2006). patters.” Proc. Nat. Acad. Sci. USA, vol.95:14863-14868 (1998). Momparler, “Cancer epigenetics.” Oncogene, vol. 22:6479-6483 Felicetti et al., “Role of PLZF in melanoma progression.” Oncogene, (2003). vol. 23:4567-4576 (2004). Muthusamy et al., “Epigenetic Silencing of Novel Tumor Suppres Ghosh et al., “Three-Dimensional Culture of Melanoma Cells Pro sors in Malignant Melanoma.” Cancer Res., vol. 66:11187-11 193 foundly Affects Gene Expression Profile: A High Density (2006). Oligonucleotide Array Study,” Journal of Cellular Physiology, vol. Riker, “Isolation and Culture of Melanoma Cell Lines. Cancer Cell 204:522-531 (2005). Culture. Methods and Protocols, Totowa: Humana Press, vol. 88: 93 Heyer et al., “Exploring Expression Data: Identification and Analysis 100 (2004). of Coexpressed Genes' Genome Research, vol. 9:1106-1115 (1999). Riker et al., Development and Characterization of Melanoma Cell Hoek et al., “Expression Profiling Reveals Novel Pathways in the Lines Established by Fine Needle Aspiration Biopsy: Advances in the Transformation of Melanocytes to Melanomas.” Cancer Research, Monitoring of Patients with Metastatic Melanoma, Cancer Detection vol. 64:5270-5282 (2004). and Prevention, vol. 23:387-396 (1999). Hoek et al., “Metastatic potential of melanomas defined by specific Rivenbarket al., “DNA methylation-dependent silencing of CST6 in gene expressonprofiles with no BRAF signature.” Pigment Cell Res., human breast cancer cell lines.” Laboratory Investigation, vol. vol. 19:290-302 (2006). 86: 1233-1242 (2006). Jean et al., “The expression of genes induced in melanocytes by Seitz et al., “Genetic Background of Different Cancer Cell Lines exposure to 365-nm UVA: study by cDNA arrays and real-time Influences the Gene Set Involved in Chromosome 8 Mediated Breast quantitative RT-PCR." Biochimica et Biophysica Acta, vol. 1522:89 Tumor Suppression.” Genes, & Cancer, vol. 45:612 96 (2001). 627 (2006). Mandruzzato et al., “A gene expression signature associated with Seykora et al., “Gene Expression Profiling of Melanocytic Lesions.” Survival in metastatic melanoma.” Journal of Translational Medi The American Journal of Dermatopathology, vol. 25:6-11 (2003). cine, vol. 4:50 (2006). Smith et al., “Whole-Genome Expression Profiling of the Melanoma Mirmohammadsadegh et al., Rapid Identification of Dysregulated Progression Pathway Reveals Marked Molecular Differences Genes in Cutaneous Malignant Melanoma Metastases, Cells Tissues Between Nevi?Melanoma in Situ and Advanced-Stage Melanomas.” Organs, vol. 177:119-123 (2004). Cancer Biology & Therapy, vol. 4:1018-1029 (2005). Mischiati et al., “cDNA-Array Profiling of Melanomas and Paired Talantov et al., “Novel Genes Associated with Malignant Melanoma Melanocyte Cultures,” Journal of Cellular Physiology, vol. 207:697 but not Benign Melanocytic Lesions.” Clinical Cancer Research, vol. 705 (2006). 11:7234-7242 (2005). Moretti et al., “In situ expression of transforming growth factor f3 is Trent et al., “Tumorigenicity in Human Melanoma Cell Lines Con associated with melanoma progression and correlates with Ki67, trolled by Introduction of Human ,” Science, vol. HLA-DR and B3 integrin expression.” Melanoma Research, vol. 247:568-571 (1990). 7:313-321 (1997). Van Gelder et al., “Amplified RNA synthesized from limited quanti Niezabitowski et al., Prognostic Evaluation of Cutaneous Malignant ties of heterogeneous cDNA” Proc. Natl. Acad. Sci. U.S.A., vol. Melanoma: A Clinicopathologic and Immunohistochemical Study, 87: 1663-1667 (1990). Journal of Surgical Oncology, vol. 70: 150-160 (1999). Vogl et al., Gene Expression Profile Changes between Melanoma Onkenet al., “Gene Expression Profiling in Uveal Melanoma Reveals Metastases and their Daughter Cell Lines: Implication for Vaccina Two Molecular Classes and Predicts Metastatic Death.” Cancer tion Trials, J. Invest. Dermatol., vol. 124:401–404 (2005). Research, vol. 64:7205-7209 (2004). US 7,615,349 B2 Page 3

Onken et al. "Association Between Microarray Gene Expression Väisänen et al., “Prognostic Value of MMP-2 Immunoreactive Pro Signature and Extravascular Matrix Patterns in Primary Uveal Mela tein (72 kD Type IV Collagenase) in Primary Skin Melanoma.” nomas.” American Journal of Opthalmology, vol. 140:748-749 Journal of Pathology, vol. 186:51-58 (1998). (2005). Wang et al., “Prospective Molecular Profiling of Melanoma Onken et al., “Functional Gene Expression Analysis Uncovers Phe Metastases Suggests Classifiers of Immune Responsiveness. Can notypic Switch in Aggressive Uveal Melanomas.” Cancer Research, cer Research, vol. 62:3581–3586 (2002). vol. 66:4602-4609 (2006). Wang et al., “Iterative Normalization of cDNA Microarray Data.” Pavey et al., “Microarray expression profiling in melanoma reveals a IEEE Transactions on Information Technology in Biomedicine, vol. BRAF mutation signature.” Oncogene, vol. 23:4060-4067 (2004). 6:29-37 (2002). Seftor et al., “Molecular determinants of human uveal melanoma Zuidervaartet al., “Gene expression profiling identifies tumour mark invasion and metastasis.” Clinical & Experimental Metastasis, vol. ers potentially playing a role in uveal melanoma development. Brit 19:233-246 (2002). ish Journal of Cancer, vol. 89:1914-1919 (2003). Seykora et al., “Gene Expression Profiling of Melanocytic Lesions.” Dressman et al., “Gene expression profiles of multiple breast cancer The American Journal of Dermatopathology, vol. 25:6-11 (2003). phenotypes and response to neoadjuvant chemotherapy,” Clin. Can Soikkeli et al., “Systematic Search for the best gene expression mark cer Res., 12(3):819-826 (2006). ers for melanoma micrometastasis detection.” Journal of Pathology, de Wit et al., “Analysis of differential gene expression in human vol. 213: 180-189 (2007). melanocytic tumour lesions by custom made oligonucleotide arrays.” Tschentscher et al., “Tumor Classification Based on Gene Expression British Journal of Cancer, 92:2249-2261 (2005). Profiling Shows That Uveal Melanomas with and without Riker et al., “The Gene Expression Profiles of primary and metastatic Monosomey 3 Represent Two Distinct Entities.” Cancer Research, melanoma yields a transition point of tumor progression and vol. 63:2578-2584 (2003). metastasis.” BMC Medical Genomics, 1:13 (2008). Tusher et al., “Significance analysis of microarrays applied to the ionizing radiation response.” PNAS, vol. 98:51 16-5121 (2001). * cited by examiner U.S. Patent US 7,615,349 B2

0

U.S. Patent Nov. 10, 2009 Sheet 2 of 6 US 7,615,349 B2

primaries

100%

L5 90% 80% 70% 60% 50% 40% 30% 20% 10% O%

8 8 CO in 5 2 : s s U.S. Patent Nov. 10, 2009 Sheet 3 of 6 US 7,615,349 B2

SEINESOOONO

(OWW-()06o SISAeuw Ob-L U.S. Patent US 7,615,349 B2

ebueuopio SSAeUW We JeOJOW U.S. Patent Nov. 10, 2009 Sheet 5 of 6 US 7,615,349 B2

CD

'... s.s.l., ::::----- 4|| || || || || || || || ?:s: 12 6 9. as-s-s---|--|-- 2 (2a, + i + i + O + i + -- :-

r - I - -- 1, C. 2||' ||. || || || || || || % (R- + i + -- - --

-- -. - % {Y K9(R- O A 3-* * * : * -: * : * i : * * + 8%|& ||- + || " || || || || || || -- “s ------e + - + i + i + -- + ------M S. ca % ------4.S. %%------

a 5 S2 ; 5 & is a5 s : S 3is aT fi : ) is a O 4 - O 3 c5 gig& U.S. Patent Nov. 10, 2009 Sheet 6 of 6 US 7,615,349 B2

(WIN) gASW (WIN) A933N (WWI) 90803W. (W3) WO833W (W3d) OA33N GW3d) & 33 : i

(NLN)6833N (WIN)CO833W (WO)2O833 (W3)AA33W. (NW)aA33W (WWI)39933N (WWI)-aldoN (INN) wall 33N G7-9N-WON

r r L l L L US 7,615,349 B2 1. 2 MELANOMIA GENESIGNATURE This transition in gene expression involves both increased expression levels of genes such as MAGE genes, GPR19, RELATED APPLICATIONS BCL2A1, SOX5, BUB1, and RGS20, and an even greater reduction in the expression of genes such as SPRR1A/B, This application claims the benefit of U.S. Provisional 5 KRT16/17, CD24, LOR, GATA3, MUC15, and TMPRSS4. Application No. 60/824,849, filed Sep. 7, 2006, which is The transition in gene expression also involves other genes incorporated herein by reference in its entirety. described herein. For example, the transition involves a reduction in expression of a plurality of (including all of) the GOVERNMENT SUPPORT following genes: GJB6, SPRR1A, SERPINB5, CALML5 10 (CLSP), DSC1, PKP1, CLCA2, DSG1, CDSN, LY6D, The methods and compositions described herein were LCE2B, FLG: RP1-14N1.3, KRT16, SBSN, SERPINB3, made with government support awarded by the ARMY Medi SERPINB7, KRT17, KLK7, LOR, SLURP1, LOC63928, cal Research and Material Command (MRMC) under Grant KRT15, LGALS7, CST6, SPRR1B, CNFN, TRIM29, No. DAMD17-02-2-0051. The government has certain rights EPPK1, SFN, KRT6B, DSG3, SPRR2B, DMKN, ASAH3, in the invention. 15 SERPINB13, KLK11, AADACL2, DAPL1, ABCA12, DSC3, POF1B, GATA3, LYPD3, KRT6A, EHF, PCDH21, TECHNICAL FIELD CBLC, FGFR2, SCEL, and FGFR3. For example, the transi tion involves an increase in expression of a plurality of (in Methods and compositions for evaluating tissues, e.g., cluding all of) the following genes: MAGEA3, MAGEA6, tumors, are provided herein. 2O CSAG2 (TRAG3), MAGEA12, MAGEA2, TRIM51, NRP2, MAGEA1, MSI2, GYPC, SPP1, SOX5, KIFC1, HILS1, RGS20, BUB1, IGF2BP3, FRMD5, C1orf)0, EYA4, BACKGROUND BCL2A1 SLC16A4, AKT3, CDC45L, SEC22L3, PEG 10, POPDC3, MAGEA5, GLUD2, ST6GALNAC3, SEZ6L2, In the United States, the overall incidence of melanoma is 25 DUSP4, ABCB5, RASGRF1, DUSP4, FLJ40142, BRRN1, increasing at a rate faster than any other cancer, with recent PHLDA1, MMP14, DUSP6, DPY19L1, GLUD1, estimates for lifetime risk of developing invasive melanoma LOC346615, CALU, RNF157, PRDM13, PBK, KIAA1618, at 1/49 (Jemal et al., C. A. Cancer J. Clin. 57:43-66, 2007). NEDD4L. BICD1, and RRM2. The transition may further The development of melanoma begins with the malignant involve an increase in expression of one or more of the fore transformation of normal human epithelial melanocytes 30 going genes in conjunction with a decrease in expression of (NHEM) located within the basement membrane of the skin, one or more of the previous set of genes. Additionally, a but the genetic changes associated with the progression of correlation between primary melanoma tumor thickness, as NHEM to melanoma are not well characterized (Bittneret al., measured by Breslow's depth, and the accumulation of indi Nature. 406:536-540, 2000; DeRisi et al., Nat. Genet. 14:457 vidual gene expression changes has also been discovered. The 460, 1996; Golub et al., Science. 286:531-37, 1999; Hanahan 35 genes identified as changing expression in primary cutaneous et al., Cell. 100-57-70, 2000; Seykora et al., Am. J. Dermato melanoma along the spectrum of increasing Breslows thick pathol. 25:6-11, 2003: Su et al., Nature. 406:536-540, 2000; ness, are useful markers for the existence of cells character Trent et al., Science. 247:568-571, 1990; Weyers et al., Can istic of metastatic melanoma. As further described herein, cer. 86:288-299, 1999). Similarly, the molecular mechanisms expression of the genes (e.g., five or more of the genes listed underlying further progression from a primary tumor to a 40 above, and/or five or more of the genes described in Tables. metastatic melanoma are also inadequately defined. A-D, herein) can be examined in various combinations. There is a correlation between the thickness of the primary Accordingly, in one aspect, the technology herein features melanoma and its capacity to metastasize to the draining a method of evaluating a melanoma from a patient. The lymph node basin(s) and hematogenously (Haddad et al., method includes determining expression of five or more Ann. Surg. Oncol. 6: 144-149, 1999; Cascinelli et al., Ann. As genes in a test sample from a melanoma, relative to a control, Surg. Oncol. 7:469-474, 2000). Once melanoma has metas wherein the five or more genes are selected from the genes tasized by either route, the overall survival for patients greatly listed in Table A and Table B, thereby evaluating the mela diminishes (Balchet al., Cancer. 88:3635-3648; 2001; Balch Oa. et al., J. Clin. Oncol. 19:3622-3634, 2001). Whereas patients In various embodiments, expression of at least 10 genes with thin primary tumors are cured by Surgery, patients diag- so from Tables A and B is determined, e.g., expression of at least nosed with metastatic melanoma (AJCC stage IV) have an 25, 50, 100, 250, 500, 750, 1000, 1250, or 1500 genes is overall poor prognosis, with 6 out of every 7 skin cancer determined. In various embodiments, expression of no more deaths due to metastatic melanoma (Balch et al., Cancer. than 1500, 1250, 1000, 750, 500, 250, 100, 50, or 25 genes is 88: 1484-1491, 2000; Eton et al., J. Clin. Oncol. 16:1103 determined. The at least 10 genes may be chosen in any 1111, 1998: Jemal et al., C. A. Cancer J. Clin. 57:43-66, ss combination from Tables A and B. Thus, in some embodi 2007). ments, the at least 10 genes includes five genes from Table A and five genes from Table B. Other combinations may be SUMMARY examined, such as one gene from Table A and nine genes from Table B, and so forth. In some embodiments, expression of The compositions and related methods provided herein are 60 genes from Table A or Table B is determined (e.g., expression based, in part, on the discovery of unique gene expression of at least 10 genes from Table A is determined to the exclu profiles characteristic of primary basal cell, squamous cell, sion of genes from Table B, or, alternatively, expression of at non-metastatic, and metastatic melanoma skin cancer least 10 genes from Table B is determined, to the exclusion of samples. A consistent “transition Zone' of gene expression genes from Table A). change within primary melanoma Samples was observed and 65 In various embodiments, expression of the five or more has allowed identification of gene expression profiles capable genes is determined relative to expression of the five or more of distinguishing a primary tumor from metastatic melanoma. genes in a reference set of non-metastatic cutaneous tissue US 7,615,349 B2 3 4 samples, wherein a decrease in expression of one or more of (e.g., expression of at least five genes from Table C is deter a gene of Table A, and an increase in expression of one or mined, or expression of at least five genes from Table D is more of a gene of Table B, relative to expression of the five or determined). more genes in the reference set, indicates an increased like Expression of the five or more genes can be determined lihood that the test sample is from a metastatic melanoma relative to expression of the five or more genes in a reference and/or indicates a poor prognosis. The method can further set of non-metastatic cutaneous tissue samples, wherein a include determining that the patient should undergo a treat decrease in expression of one or more of a gene of Table C, ment protocol. For example, patients for which the melanoma and an increase in expression of one or more of agene of Table sample expression is indicative of a metastatic melanoma D, relative to expression of the five or more genes in the may elect to undergo a more aggressive treatment, e.g., with 10 reference set, indicates an increased likelihood that the test interferon alpha2b, or interleukin2, Surgery to remove addi sample is a metastatic melanoma and/or indicates a poor tional tissue (e.g., addition melanoma tissue at the site from prognosis. The method can further include determining that which the original sample was obtained, or at another site, the patient should undergo a treatment protocol, based on the e.g., in a lymph node), or an experimental treatment. Patients determination of gene expression. in which expression is not indicative of a metastatic mela 15 In some embodiments expression of the five or more genes noma may elect to forgo a treatment. is compared to: (a) expression in a first reference set of non The non-metastatic cutaneous tissue samples (e.g., the ref metastatic cutaneous tissue samples, and (b) expression in a erence samples to which expression in the test sample is second reference set of metastatic melanoma tissue samples: compared) can include one or more of the following: normal wherein a greater similarity in expression of the five or more human epithelial melanocytes, primary cutaneous mela genes in the test sample to the second reference set than to the noma, basal cell carcinoma, squamous cell carcinoma, mela first reference set indicates an increased likelihood that the noma in situ, and/or thin melanoma (<1.5 mm Breslow's test sample is a metastatic melanoma. thickness). The determining expression of five or more genes in the In various embodiments, expression of the five or more test sample can include isolating RNA from the test sample, genes is compared to: (a) expression in a first reference set of 25 and detecting expression of the RNA, or detecting protein non-metastatic cutaneous tissue samples, and (b) expression expression. in a second reference set of metastatic melanoma tissue In another aspect, the technology also features kits for samples; wherein a greater similarity in expression of the five evaluating a melanoma sample. The kits include polynucle or more genes in the test sample to the second reference set otides (e.g., primers or probes) for analysis of at least 5, 10. than to the first reference set indicates an increased likelihood 30 25, 50, 75, or 100 genes from Tables C and D, wherein each that the test sample is a metastatic melanoma. oligonucleotide specifically hybridizes to one of the genes The determining expression of five or more genes in the from Tables C and D. The kits can include polynucleotides for test sample can include isolating RNA from the test sample, analysis of up to 25, 50, 75, or 100 genes from Tables Cand and detecting expression of the RNA. RNA expression can be D. detected directly or indirectly, e.g., using microarray or PCR 35 analysis. The sample can be a fixed, paraffin-embedded For example, a kit includes pairs of polynucleotides for biopsy sample, or a frozen sample. amplification of the genes from Tables C and D by PCR. In various embodiments, the determining expression of The technology also features kits for evaluating a mela five or more genes in the test sample includes reverse tran noma sample that include polynucleotides (e.g., primers or Scriptase polymerase chain reaction (RT-PCR), e.g., quanti 40 probes) for analysis of at least 250, 500, 750, 1000, 1250, or tative PCR, e.g., real time quantitative PCR. 1500 genes from Tables A and B, wherein each oligonucle In various embodiments, the determining expression of otide specifically hybridizes to one of the genes from Tables five or more genes in the test sample includes microarray A and B. The kits can include polynucleotides for analysis of analysis. up to 250, 500,270, 1000, 1250, or 1500 genes from Tables A In other embodiments, the determining expression of five 45 and B. In some embodiments, the polynucleotides are immo or more genes in the test sample includes analysis of protein bilized on a Solid Support, e.g., as in a microarray. expression, e.g., immunohistochemical analysis of The technology also features kits for evaluating protein encoded by on or more of the genes, or proteomic analysis. expression in a melanoma sample. The kit includes, for The test sample can be a test sample from a melanoma example, reagents (e.g., ) for detection of proteins having an intermediate thickness (e.g., a Breslow's thickness 50 encoded by at least 5, 10, 25, or 50 genes from Tables Cand of 1-4 mm). The test sample can be from a thin or thick D melanoma. A “sample' is any biological material obtained from an In another aspect, the technology features a method of individual. A "melanoma sample' or "melanoma tissue evaluating a melanoma from a patient, which method sample' is a sample that includes primarily melanoma cells. includes, for example, determining expression offive or more 55 "Gene’ refers to a polynucleotide sequence that comprises genes in a test sample from the melanoma, relative to a con sequences that are expressed in a cell as RNA and control trol, wherein the five or more genes are selected from the sequences necessary for the production of a transcript or genes listed in Table C and Table D, thereby evaluating the precursor. A gene expression product analyzed according to a melanoma. The method can include other features described method described herein can be encoded by a full length herein. For example, expression of at least 10, 25, 50, 75, or 60 coding sequence or by any portion of the coding sequence. 100 genes from Tables C and D can determined. Expression “Polynucleotide' refers to a polymeric form of nucleotides of no more than 100, 75, 50, 25, or 10 genes can be deter of any length, either ribonucleotides or deoxyribonucle mined. The at least 5 genes from Table C and Table D may be otides. Thus, the term includes, but is not limited to, single-, examined in any combination, such as one gene from Table C double-, or multi-stranded DNA or RNA, genomic DNA, and four genes from Table D; or four genes from Table C and 65 cDNA, DNA-RNA hybrids, or a polymer comprising purine one gene from Table D. In some embodiments, expression of and pyrimidine bases or other natural, chemically or bio genes solely from just one of Table C or Table D is determined chemically modified, non-natural, or derivatized nucleotide US 7,615,349 B2 5 6 bases, as well as polynucleotides that have been modified in and also illustrates the transitional tumor class that would order to introduce a means for attachment (e.g., to a Support most benefit from assays to detect the metastatic cell type. for use as a microarray). FIG. 2A is a graph depicting comparative quantitative PCR The descriptions herein are phrased in terms of “five or (qPCR) gene expression analysis data of putative tumor Sup more' or “ten or more genes, but the choices of five and ten pressor genes (CST6, DSC3, PITX1, POU2F3) and onco would be understood to be for the purposes of illustration and genes (GDF15, SPP-1) in PCM (n=7) and MM (n=13) are non-limiting. One may also examine expression of other samples. Relative quantitation of target gene expression for numbers such as 3, 4, 6, 7, 8, 9, 11, 12, 13, 14, 15, 16, 17, 18, each sample was, determined using the equation 2-, 19, 20, 25, and 30 genes, and so forth. In some embodiments, where GAPDH was used as the internal reference and normal the combination of genes from Tables A, B, C, and/or D which 10 skin as the calibrator. Values were Log base 10 transformed is examined includes one or more genes implicated in one or (y-axis) so that all values below zero represent a down-regu more (e.g., 2, 3, 4, 5, 6, 7, 8, or 9) of the following biological lation in gene expression and values above Zero represent an processes, as understood by one skilled in the art: kerati up-regulation in gene expression, compared to normal skin. nocyte differentiation, development, melanocyte FIG. 2B is a graph depicting results of correlative microar differentiation, cell differentiation, morphogenesis, muscle 15 ray analysis of gene expression levels in primary and meta development, nervous system development, cell adhesion, static melanoma samples compared with normal skin. The the Wnt receptor signaling pathway, cell-cell signaling, statistical differences of gene expression between primary organization and biogenesis, inflammatory or (PCM) and metastatic melanoma (MM) samples were ana immune response, cell motility and chemotaxis, electron lyzed by Wilcoxon's signed rank test; two-tailed significance transport, carbohydrate metabolism, lipid metabolism, pro level was set at C.-0.05. Compared to PCM samples (n=7), the teolysis, signal transduction, protein transport, protein bio expression levels of 4 putative tumor Suppressor genes synthesis, transcription, DNA repair, cell cycle regulation or (CST6, p<0.0001; DSC3, p<0.0001; PITX1, p=0.0043, proliferation, or apoptosis. In some embodiments, at least one POU2F3, p<0.0001) were significantly decreased in MM gene implicated in each of the above processes is examined. samples (n=40), while the expression of putative oncogenes The details of one or more embodiments of the technology 25 (GDF15, p=0.0027: SPP1, p<0.0001) were significantly are set forth in the accompanying drawings and the descrip increased in MM samples. tion below. Other features, objects, and advantages of the FIG. 3A is a table listing results of semi-quantitative. RT technology will be apparent from the description and draw PCR analysis of oncogene and tumor Suppressor gene (TSG) ings, and from the claims. All cited patents, and patent appli mRNA expression in normal human skin, NHEM, PCM and cations and references (including references to public 30 MM cell lines. A panel of 12 putative oncogenes and 6 TSG sequence database entries) are incorporated by reference in were analyzed, with the relative levels of mRNA expression their entireties for all purposes. as follows: negative band (-), faintly visible band (+/-), vis ible band (+), strongly visible band (++), N=Not Done. B-ac DESCRIPTION OF DRAWINGS tin served as the internal comparative control. The grey values 35 of PCR products of each gene were analyzed by the Alpha ase 3 software and standardized according to B- in every FIG. 1A is a graph depicting the number and functional sample. classifications of genes, lower expression (upper bar for each FIG. 3B is a set of western blots in which expression of classification) and higher expression (lower bar for each clas SPP1 (FIG. 3B, panelsa and b), DSC3, CLCA2, PDGFRL, sification) of which is associated with metastatic melanoma. 40 and (FIG. 3B, panels c, d, e, and f, respectively) FIG.1B is a heatmap depicting the relative gene expression was examined in metastatic melanoma (MM) and primary of the full spectrum of genes described herein in the non cutaneous melanoma (PCM) daughter cell lines. The blot metastatic (primary) tumors relative to the metastatic tumors depicted in FIG. 3B, panels a and b, was probed with anti (MM). The dark grey shading (appearing primarily in the SPP1 . Panels c, d, e, and f, were probed with anti upper left and lower right quadrants) corresponds to higher 45 bodies to DSC-3, CLCA2, PDGFRL, and C-tubulin (as an than average gene expression, black depicts average gene internal control), respectively. Lanes 1-3 of each blot con expression, and the light shading (appearing primarily in tained PCM samples. Lanes 4-6 of each blot contained MM lower left and upper right quadrants) depicts lower than aver cell lines. age gene expression. Average is set as the average across all the samples used in the figure. Individual rows represent 50 DETAILED DESCRIPTION individual tumor samples and individual columns depict indi vidual genes identified in the experiments described herein. Methods and compositions for evaluating tissue samples FIG. 1C is a set of graphs depicting relative expression (e.g., cutaneous tissue samples, e.g., Samples from primary levels of a subset of genes described herein in samples of melanomas) are provided herein to determine whether the normal skin, basal cell carcinoma (BCC), squamous cell car 55 samples exhibit a gene expression profile characteristic of cinoma (SCC), melanoma in situ (MIS), thin melanoma, aggressive, metastatic melanomas, or a profile characteristic melanoma of intermediate thickness (I.M.), thick primary, of non-metastatic melanomas. The ability to classify samples metastatic melanoma (Subcutaneous, lymph node and distant; with a high degree of accuracy and sensitivity facilitates MM), and melanoma cell lines derived from patients with prognosis and Subsequent clinical decisions (e.g., whether or stage IV melanoma. Genes more highly expressed in meta 60 not to undergo further Surgery or other treatment modalities). static melanoma are depicted in the upper graph and genes Accordingly, the technology provides, interalia, sets of genes whose expression is decreased in the metastatic melanoma whose expression can be examined to determine whether a cells are depicted in the lower graph. In both cases the tran cutaneous tissue sample is non-metastatic, or metastatic, as sition from a non-metastatic tissue expression level to the well as methods of analyzing expression of the gene sets, and metastatic tissue expression level occurs in the I.M. thickness 65 compositions for performing the analysis. sample set. This figure illustrates the distinct differences The technology herein features combinations of genes, between the non-metastatic and metastatic melanoma tumors e.g., combinations of genes listed in Tables A and B below, US 7,615,349 B2 7 8 combinations of genes listed in Tables C and D below. Analy the combination of measures that identify a metastatic sample sis of the expression of these genes can be performed to and a non-metastatic sample. The specific mathematical pro identify metastatic melanoma tumors (e.g., in primary cuta cess depends on the method used for measuring gene expres neous melanoma Samples, or in Samples from compound Sion, the number of genes, and the nature of the genes chosen nevi). The technology also features methods of analyzing the to participate in the assay. Based on this unique combination expression of combinations of genes. Various techniques are and the reference sample values a threshold value will be Suitable for analyzing gene expression, including those that determined (or mathematical formula) that will identify the measure RNA or protein expression. For example, a sample unknown sample as more like the metastatic samples or more from a melanoma (e.g., a primary cutaneous melanoma) is like the non-metastatic samples. One of skill would under collected and processed to obtain RNA, protein, or tissue 10 sections, to produce a test sample for analysis. The relative stand that genes that are not differentially expressed can be expression of several, dozens, 50, 100, or hundreds of genes examined in methods described herein, e.g., as a control. in the test sample is determined. The gene expression values Tables A-D set forth sets of genes, the expression of which are compared to a reference set of values derived from has been shown to correlate with metastatic melanoma. selected non-metastatic melanoma samples and metastatic 15 Tables A and C list genes whose expression is decreased in melanoma samples measured by the same assay as used to metastatic melanoma samples, relative to non-metastatic determine expression in the test sample. The values obtained samples. Tables B and D list genes whose expression is for the test sample characterize the sample as metastatic or increased in metastatic melanoma samples, relative to non non-metastatic. metastatic samples. Thus, the genes provided in these tables, In some methods, a small number of genes (e.g., a Subset of and Subsets thereof, are useful markers for metastatic mela genes from Tables C and D, e.g., 5, 10, 15, 25 genes) is Oa. selected for analysis, and expression levels are determined by In various embodiments, the technology provides a Subset a quantitative or semi-quantitative PCR method or by immu of genes from Tables A and B for evaluating a cutaneous nohistochemistry. The combination of measured values for tissue sample, sets of oligonucleotides (e.g., for use as probes the respective genes are compared to control values to deter 25 and primers) for analyzing expression of the Subsets, and mine the degree to which the test sample contains gene methods for analyzing their expression, as described in more expression values indicative of a metastatic tumor sample. detail below. The set includes, for example, at least 5, 10, 50. The test samples are, for example, Surgically collected tumors 100,250, 500, 750, or 1000 genes from Tables A and B, in any collected in a manner that preserves RNA, or alternatively proportion (e.g., 800 genes from Table A and 200 genes from fixed (e.g., formalin fixed) and embedded in paraffin prior to 30 Table B). analysis, preserved by flash freezing or fixation, and/or For example, the subset includes the genes listed in Tables treated with an RNA Stabilization Reagent. C and D. The genes in Tables C and D, and subsets thereof, are In some methods, expression of a larger number of genes is useful for evaluating a cutaneous-tissue sample, and in meth analyzed, e.g., using microarrays. Nucleic acids from the test ods for analyzing expression of the Subsets, as described in sample are hybridized to arrays under appropriate conditions, 35 arrays are scanned, and the data processed by standard meth more detail below. An exemplary set includes, for example, at ods for feature extraction and normalization in order to obtain least 5, 10, 25, 35, or 51 genes from each of Tables C and D. individual gene expression values. In these methods, a few The lists shown in Tables C and D were generated from the hundred to more than a thousand genes can be used to deter lists shown in Tables A and B by selecting the genes that mine the character of the test sample. One of several methods 40 exhibit the greatest difference in gene expression between the might be employed to identify the metastatic potential of the metastatic samples and the non-metastatic samples based on sample under investigation, based on the microarray-deter microarray analysis. The genes on these lists also represent mined gene expression values. Typically, reference samples genes, the expression or expression products of which have for metastatic melanoma and non-metastatic melanoma are been the Subject of biological investigations. analyzed in advance. The test sample is compared to the 45 Tables A-D include full length gene names (Gene descrip reference samples by classification schemes Such as cluster tion), gene symbols, GenBank accession numbers (GenBank ing, weighted Voting, principle components analysis, self ID), , gene accession numbers (Entrez, Gene ID), and organizing maps, and/or neural networks. Each of these UniGene accession numbers (UniGene ID). GenBank, schemes is essentially a mathematical system for maximizing Entrez, and UniGene records can be accessed on the World the geometric separation of classes (metastatic and non-meta 50 WideWeb at the following address: ncbi.nlm.nih.gov. These static) in multidimensional space using the individual gene records provide sequences and additional information for expression values as coordinates to plot an individual sample each gene. relative to reference samples in multidimensional space. The genes listed in Tables A-D are generally referred to Whether a method suitable for analysis of smaller or larger elsewhere herein by gene symbol. Gene symbols shown in numbers of genes is employed, the reference samples define parentheses are aliases or former designations.

TABLE A Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples.

Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 1. 24-dehydrocholesterol reductase DHCR 24 NM 014762.2 1718 HS.498.727 2. 26 serine protease P11 NM OO6025.2 89.09 HS.997 3. abhydrolase domain containing 9 ABHD9 NM O24794.1 79852 HS.156457 US 7,615,349 B2 10

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID ABI gene family, member 3 (NESH) NM O15429.2 S.47701S binding protein absent in melanoma 1 NM OO1624.1 2O2 S.486074 absent in melanoma 1-like NM O17977 55.057 S.128.738 acid phosphatase, prostate PP NM OO1099.2 55 S.433060 Acrg embryonic lethality (mouse) KOOOOO9 AA42S275 170622 S.3566OS minimal region ortholog actin binding LIM protein 1 BLIM1 NM OO10O34O7.1 3983 S.438236 10. activin A receptor, type ILA C V R2A. NM OO1616.3 92 S.47O174 11. adaptor-related protein complex 1, mu NM OO5498.3 1OOS3 S.18894 2 subunit 12. adenosine kinase NM OO1123.2 132 S.500118 13. ADMP NM 145035 165679 S.369104 14. AFG3 ATPase family gene 3-like 2 NM OO6796 10939 S.436683 (yeast) 1S. AHNAK nucleoprotein (desmoyokin) NM O24060 79026 S.378738 16. aldehyde dehydrogenase 2 family NM OOO690.2 217 S.436437 (mitochondrial) 17. aldehyde dehydrogenase 3 family, NM OOO382.2 224 S.499886 member A2 18. aldehyde dehydrogenase 3 family, NM OOO695.3 222 S.87539 member B2 19. aldehyde dehydrogenase 3 family, NM OOO691.3 218 S.531 682 member A1 20. aldo-keto reductase family 1, member NM O2O299.3 S.116724 B10 (aldose reductase) 21. aldo-keto reductase family 1, member NM OO1353 1645 S.295131 C1 (dihydrodiol dehydrogenase 1; 20 alpha (3-alpha)-hydroxysteroid dehydrogenase) 22. aldo-keto reductase family 1, member AKR1C2 NM 001354.4 1646 S.SS8319 C2 (dihydrodiol dehydrogenase 2; bile acid binding protein; 3-alpha ydroxysteroid dehydrogenase, type I) 23. kaline ceramidase ACER1 NM 133492 125981 S.352.609 24. pha-2-glycoprotein 1, Zinc AZGP1 NM OO1185 563 S.4O7861 25. pha-2-macroglobulin-like 1 A2ML.1 AKOSA908.1 144568 S.334306 26. ALS2 C-terminal like ALS2CL NM 147129.2 259173 S.517937 27. androgen receptor (dihydrotestosterone AR NM 001011645.1 367 S.496240 receptor; testicular feminization; spinal and bulbar muscular atrophy; Kennedy disease) 28. 3, node of Ranvier (ankyrin G) ANK3 NM 001149.2 288 S.499.725 29. ankyrin repeat domain 22 ANKRD22 NM 144590. 118932 S.217484 30. ankyrin repeat domain 35 ANKRD35 NM 144698.2 148741 S.133O81 31. A3 ANXA3 NM OO5139 306 S.480042 32. annexin A8 ANXA8 NM OO1630 244 S.87268 33. annexin A9 NM OO3568. 8416 S.43O324 34. aquaporin 3 NM OO4925 360 S.234.642 35. arachidonate 12-lipoxygenase NM OOO697. 239 S.422967 36. arachidonate 12-lipoxygenase, 12R NM OO1139. 242 S.136574 type 37. arachidonate lipoxygenase 3 NM 021628. S9344 S.232770 38. arginase, liver NM OOOO45.2 383 S.44O934 39. argininosuccinate synthetase NM 054012.2 445 S.SS8301 40. arrestin domain containing 4 NM 183376. 91947 S.6093 41. aryl hydrocarbon receptor nuclear NM O2O183 56938 S.434269 translocator-like 2 42. arylacetamide deacetylase(esterase) NM 001086.2 13 S.SO6908 43. arylacetamide deacetylase-like 2 DACL2 NM 207365. 344752 S.1002O6 ataxin 3 ATXN3 NM 001024631.1 4287 S.526425 45. ATPase type 13A4 ATP13A4 NM O32279.2 84239 S.56.11OO 46. ATPase type 13 A5 ATP13AS AY358.667.1 344.905 S.S31,335 47. ATPase, Ca++ transporting, type 2C, ATP2C2 NM O14861. 9914 S.6168 member 2 48. ATPase, Class 1, type 8B, member 1 NM OO5603 5205 S.418426 49. ATPase, H+ transporting, lysosomal NM OO103936.2.1 245973 S.372429 42 kDa, V1 subunit C isoform 2 SO. ATPase, H+/K+ transporting, ATP12A NM OO1676.4 479 S.147111 nongastric, alpha polypeptide S1. ATP-binding cassette, Sub-family A ABCA12 NM 173076.2 26,154 S.134585 (ABC1), member 12 US 7,615,349 B2 11 12

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 52. autism susceptibility candidate 2 AUTS2 NM O15570.1 S.21631 53. bactericidal permeability-increasing BPIL2 NM 174932 S.372939 protein-like 2 54. basonuclin 1 BNC1 NM OO1717.2 646 S.459153 55. B-box and SPRY domain containing BSPRY NM O17688.1 54836 S.49487O 56. B-cell CLL/lymphoma 11A (zinc BCL11A NM 138559.1 53335 S.370549 finger protein) 57. B-cell CLL/lymphoma 11B (zinc finger BCL11B NM 022898.1 64919 S.S10396 protein) S8. BCL2 adenovirus E1B 19 kD BNIPL AYO33OOO.1 149428 S.SS3613 interacting protein like 59. BCL2-associated athanogene BAG1 NM OO4323.3 573 S.377484 60. BCL2-like 10 (apoptosis facilitator) BCL2L10 NM O2O396.2 10017 S.283672 61. betacellulin BTC NM OO1729.1 685 S.SS83O2 62. bleomycin hydrolase BLMH NM OOO386.2 642 S.371914 63. Bochomolog (mouse) BOC NM O33254.2 91653 S.556004 64. bradykinin receptor B2 BDKRB2 NM OOO623.2 624 S.525572 65. BRG1-binding protein ELD/OSA1 ELD, OSA1 NM O2O732 57492 S.436008 66. BTB (POZ) domain containing 11 BTBD11 NM OO1017523.1 121551 S.271272 67. BTG family, member 3 BTG3 NM OO6806 109SO S.47342O 68. butyrobetaine (gamma), 2-oxoglutarate BBOX1 NM OO3986 8424 S.144845 dioxygenase (gamma-butyrobetaine hydroxylase) 1 69. cadherin-like 22 CDH22 NM O21248 64,405 S.382126 70. cadherin-like 26 CDH26 NM 021810.3 60437 S.S4973 71. calcium channel, voltage-dependent, CACNA2D3 NM O18398.2 55799 S.128594 alpha 2 delta3 subunit 72. calcium channel, voltage-dependent, CACNB4 NM OO1005746.1 785 beta 4 subunit 73. calcium/-dependent protein CAMK1D 57118 kinase ID 74. calmodulin-like 3 CALML3 NM O05185 810 75. calmodulin-like 5 CALMLS NM O17422.3 S1806 76. Small subunit 2 CAPNS2 NM O32330 84290 77. carbonic anhydrase II CA2 NM OOOO67.1 760 78. carbonic anhydrase XII CA12 NM 001218.3 771 79. carbonic anhydrase XIII CA13 NM 198584.1 377677 80. carboxypeptidase A4 CPA4 NM O16352 S1200 81. carboxypeptidase Z. CPZ NM 001014447.1 8532 82. carcinoembryonic antigen-related cell CEACAM19 BCO83499.1 56971 S.416925 adhesion molecule 19 83. carcinoembryonic antigen-related cell CEACAM6 NM 0024.83.3 S.466814 adhesion molecule 6 (non-specific cross reacting antigen) 84. Cas-Br-M (murine) ecotropic retroviral CBLC NM 012116.2 23624 S.466907 transforming sequence c 85. casein kinase 1, alpha 1 CSNK1A1 NM OO1892 1452 S.442592 86. caspase 14, apoptosis-related cysteine CASP14 NM 012114 23581 S.248226 protease 87. caspase recruitment domain family, CARD14 NM 024110 79092 S.SSO529 member 14 88. castor homolog 1, Zinc finger CASZ1 NM O17766.2 S4897 S.4398.94 (Drosophila) 89. , beta interacting protein 1 CTNNBIP1 NM O2O248.2 56998 S.4637.59 90. cathepsin L2 CTSL2 NM OO1333.2 1515 S.87417 91. CCAAT?enhancer binding protein CEBPA NM 004364.2 1 OSO S.76171 (C/EBP), alpha 92. CD1a antigen NM OO1763.1 909 S.1309 93. CD1E antigen, epolypeptide NM O3O893.1 913 S.249217 94. CD207 antigen, langerin NM O15717.2 SO489 S.199731 95. CD24 antigen (Small cell lung NM O13230.2 934 S.37S108 carcinoma cluster 4 antigen) 96. CDC-like kinase 4 NM O2O666 57396 S.406557 97. CDP-diacylglycerol synthase NM OO1263.2 1040 S.444924 (phosphatidate cytidylyltransferase) 1 98. cellular retinoic acid binding protein 2 RABP2 NM OO1878.2 1382 S.405662 99. centaurin, delta 1 ENTD1 NM O15230 116984 S.427719 100. checkpoint Suppressor 1 HES1 NM O05197 1112 S.434286 101. chemokine (C-C motif) ligand 22 CL22 NM OO2990.3 6367 S.S34347 102. chemokine (C-X3-C motif) receptor 1 X3CR1 NM OO1337 1524 S.78913 103. chemokine (C-X-C motif) ligand 14 XCL14 NM OO4887.3 9547 S.483444 104. chiltinase 3-like 2 HI3L2 NM OO1025197.1 1117 S.514840 US 7,615,349 B2 13 14

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID OS. hloride channel, calcium activated, CLCA2 NM OO6536 S.241551 amily member 2 O6. hloride channel, calcium activated, CLCA4 NM 012128.2 S.SSSO12 amily member 4 O7. hloride intracellular channel 3 CLIC3 NM 004669.2 9022 S.64746 O8. holinergic receptor, nicotinic, alpha CHRNA9 NM O17581.2 55584 S.272278 olypeptide 9 O9. hromatin modifying protein 4C CHMP4C NM 152284.3 92421 S.183861 10. hromosome 1 open reading frame 106 C1orf106 NM O18265.1 55765 S.S18997 11. hromosome 1 open reading frame 116 C1orf116 NM O23938.4 79098 S.32417 12. hromosome 1 open reading frame 161 C1orf161 NM 152367.1 126868 S.376.194 13. hromosome 1 open reading frame 172 C1orf172 NM 152365.1 126695 S.18888.1 14. Chromosome 1 open reading frame 21 C1orf21 NM O3O806.3 81 S63 S.4971.59 1S. hromosome 1 open reading frame 210 C1orf210 NM 182517.1 1494.66 S.158963 16. hromosome 1 open reading frame 42 C1orfA2 NM O19060.1 S4S44 S.1101.96 17. hromosome 1 open reading frame 46 C1orfA6 AFOOSO82 388699 S.S1642O 18. hromosome 1 open reading frame 68 C1orf68 AFOOSO81.1 5531.68 S.S47272 19. hromosome 10 open reading frame C10orf118 NM O18017.2 55088 S.159066 18 20. OOSOc O open rea ing rame 57 C10orf57 NM O25125.2 8O195 S.169982 21. OOSOc O open rea ing rame 72 C10orf72 NM 144984.1 196740 S.522928 22. OOSOc O open rea ing rame 99 C10orf 9 NM 207373.1 387695 S.298713 23. OOSOc 1 open rea ing rame 52 C11orf52 NM O80659.1 91894 S.97O13 24. OOSOc 4 open rea ing 8t C14orf116 NM O18589 55453 S. 60548 16 25. OOSOc 4 open rea ing rame 29 C14orf29 NM 181533 145447 S.271896 26. OOSOc 4 open rea ing rame 47 C14orf247 NM 152332 123O36 S.57787 27. OOSOc 4 open rea ing rame 78 C14orf78 AKO941431 113146 S.441.783 28. OOSOc 9 open rea ing rame 33 C19Crf33 NM O33520.1 S.348SS3 (IMUP) 29. hromosome 2 open reading frame 54 C2Orfs4 NM O248611 79.919 S.193745 30. hromosome 2 open reading frame 55 C2Orfss NM 207362.1 343990 S.469398 31. hromosome 20 open reading frame C20orf128 NM 178468.2 128876 S.SS4917 28 32. hromosome 20 open rea ing rame 38 C20orf58 NM O18327.1 55304 S.272242 33. hromosome 20 open rea ing rame 42 C20orf242 NM O17671.3 55612 S.472O54 34. hromosome 20 open rea ing rame 55 C20orf5S NM 031424 83541 S.S34O72 35. hromosome 20 open reading frame 74 C20orf74 AKOO2211.1 57186 S.472285 36. hromosome 21 open reading frame 34 C21orf54 NM OO1005733.1 388815 S.473394 37. hromosome 4 open reading frame 32 C4orf32 NM 152400.1 132720 S.23439 38. hromosome 5 open reading frame 27 C5orf27 NM 175616.2 202299 S.8373 39. open reading frame 4 CSOrfa. NM 016348.1 10826 S.519694 40. hromosome 6 open reading 132 C6orf132 XM 371820 647024 S.444.277 41. hromosome 6 open reading frame 143 C6orf143 NM OO1 010872.1 222584 S.25809S 42. hromosome 6 open reading frame 162 C6orf162 NM O20425 57150 S.70769 43. hromosome 8 open reading frame 47 C8orfa.7 NM 173549.1 2O3111 S.1714S5 hromosome 8 open reading frame 61 C8orf61 NM OO1034061.1 61943S S.127675 45. hromosome 9 open reading frame 111 C9Crf111 NM 152286.2 375775 S.294147 46. hromosome 9 open reading frame 3 C9orf3 NM O32823.3 84909 S.434253 47. athrin, light polypeptide (Lcb) CLTB NM O07097.2 1212 S.484241 48. audin 1 CLDN1 NM 021101.3 9076 S.43906O 49. audin 4 CLDN4 NM OO1305.3 1364 S.520942 SO. audin 8 CLDN8 NM 199328.1 9073 S.162209 S1. coagulation factor II (thrombin) F2RL1 NM O05242.3 21SO S.154299 receptor-like 1 52. coagulation factor III (thromboplastin, F3 NM OO1993.2 215.2 S.6.2192 tissue factor) 53. COBL-like 1 COBLL1 NM O14900.3 22837 S.470457 54. coiled-coil domain containing 3 CCDC3 NM O31455.2 83.643 S.49872O 55. collagen, type VII, alpha 1 COL7A1 NM 000094.2 1294 S.476.218 (epidermolysis bullosa, dystrophic, dominant and recessive) 56. collagen, type XVII, alpha 1 COL17A1 NM 130778.1 1308 S.117938 57. collagen, type XXI, alpha 1 COL21A1 NM O3O820.3 81578 S.47629 S8. contactin 1 CNTN1 NM 001843 1272 S.143434 59. cordon-bleu homolog (mouse) COBL NM O15198.2 23242 S.99141 60. CDSN NM OO1264.3 1041 S.310958 61. cornifelin CNFN NM 032488.2 84518 S.148590 62. coxsackie virus and adenovirus CXADR NM OO1338.3 1525 S.473417 receptor 63. creatine kinase, mitochondrial 1A CKMT1A NM OO1 O15001.1 S48596 S.425633 64. crumbs homolog 3 (Drosophila) CRB3 NM 174881.2 92359 S.1SO319 65. cryptochrome 2 (photolyase-like) CRY2 NM O21117.1 1408 S.532491 US 7,615,349 B2 15 16

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 66. C type lectin domain family 2, member B CLEC2B NM OO5127.2 9976 S.852O1 67. cyclic nucleotide gated channel beta 1 CNGB1 NM OO1297.1 258 S.147062 68. cylindromatosis (turban tumor CYLD NM O15247 S4O S.432993 syndrome) 69. cyStatin A (Stefin A) CSTA NM OO5213.3 475 S.518198 70. cystatin E/M CST6 NM OO1323.2 474 S.13938.9 71. cysteine rich transmembrane BMP CRIM1 NM 016441.1 51232 S.332847 regulator 1 (chordin-like) 72. cysteine?tyrosine-rich 1 CYYR1 NM 052954.2 1161.59 S.37445 73. cysteine-rich protein 1 (intestinal) CRIP1 NM OO1311.3 396 S.70327 74. cysteine-rich Secretory protein 3 CRISP3 NM 006061.1 10321 S.404466 75. cytidine deaminase CDA NM OO1785.2 978 S.466910 76. cytochrome b5 type A (microsomal) CYB5A NM OO1914.2 528 S.465413 77. cytochrome P450, family 2, Subfamily CYP2C18 NM OOO772.1 S62 S.S11872 C, polypeptide 18 78. cytochrome P450, amily 2, Subfamily CYP2E1 NM OOO773.3 571 S.12907 E, polypeptide 1 79. cytochrome P450, amily 26, Subfamily CYP26B1 NM O19885.2 56.603 B, polypeptide 1 80. cytochrome P450, amily 3, Subfamily CYP3A5 NM OOO777.2 577 A, polypeptide 5 81. cytochrome P450, amily 3, Subfamily CYP3A7 NM OOO765.2 551 S.111944 A, polypeptide 7 82. cytochrome P450, amily 39, subfamily CYP39A1 NM O16593.3 S1302 S.387367 A, polypeptide 1 83. cytochrome P450, amily 4, Subfamily CYP4F12 NM O23944.1 66002 S.131459 F. polypeptide 12 84. cytochrome P450, amily 4, Subfamily CYP4F2 NM OO1082.3 8529 S.SS8423 F. polypeptide 2 85. cytochrome P450, amily 4, Subfamily CYP4F22 NM 173483.1 126410 S.156452 F. polypeptide 22 86. cytochrome P450, amily 4, Subfamily CYP4F3 NM OOO896.1 4051 S.106242 F. polypeptide 3 87. cytochrome P450, amily 4, Subfamily CYP4X1 NM 178033.1 260293 X, polypeptide 1 88. cytokine receptor-like factor 1 CRLF1 NM OO4750.2 9244 S.114948 89. cytoplasmic polyadenylation elemen CPEB3 NM O14912.3 22849 S.131683 binding protein 3 90. eath associated protein-like 1 NM OO1017920.1 921.96 S.S9761 91. eath-associated protein kinase 2 NM O14326.3 23604 S.237886 92. ecorin NM 133505.2 634 S.156316 93. edicator of cytokinesis 9 NM O15296.1 23348 S.314413 94. efensin, beta 1 NM OO5218.3 672 S.32949 95. efensin, beta 103A NM 018661.2 SS894 S.283O82 96. eiodinase, iodothyronine, type II NM OO1007023.1 734 S.2O2354 97. DENN/MADD domain containing 2C NM 198459.2 1632.59 S.127350 98. ermcidin NM O53283.2 1171.59 S350570 99. ermokine NM O33317.2 93.099 S.41779S 2OO. esmocollin 1 NM 004948 823 S.348436 2O1. esmocollin 2 NM 004949.2 824 S.95612 2O2. esmocollin 3 NM OO1941 825 S.41690 2O3. esmoglein 1 NM OO1942.1 828 S.2633 2O4. esmoglein 3 (pemphigus vulgaris NM OO1944.1 830 S.1925 antigen) esmoplakin NM OO1008844.1 S.S19873 DIRAS family, GTP-binding RAS-like 3 NM OO4675.2 S.194.695 Discs, large homolog 2, chapsyn-110 NM OO1364.2 S.SO3453 (Drosophila) 2O8. istal-less homeo box 3 NM OO5220.2 S.134194 209. DKFZP564O0823 protein NM O15393.2 S.10546O 210. ual adaptor of phosphotyrosine and 3 NM O14395.1 S.436271 phosphoinositides 211. ual oxidase 1 UOX1 NM O17434 S.272.813 212. ual oxidase 2 UOX2 NM O14080.3 S.71377 213. ual oxidase maturation factor 1 UOXA1 NM 144565 S.356664 214. ynein, light chain, roadblock-type 2 YNLRB2 NM 130897.1 S.98849 215. ystonin S T NM O2O388.2 S.485 616 216. E74-like factor 5 (ets domain NM 001422.2 S.11713 transcription factor) 217. Early endosome antigen 1, 162 kD NM OO3566 S.403150 218. echinoderm associated ML.1 NM OO10O8707.1 S.12451 protein like 1 US 7,615,349 B2 17 18

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 219. ectonucleoside triphosphate ENTPD3 NM OO1248 956 S.441145 diphosphohydrolase 3 22O. EF-hand calcium binding domain 6 FCAB6 NM 198856.1 64800 S.368507 221. EGF-like-domain, multiple 3 GFL3 AL1343O3 1953 S.56.1.86 222. egl nine homolog 3 (C. elegans) GLN3 NM O22073.2 112399 S.13SSO7 223. EH-domain containing 2 HD2 NM 014601.2 3O846 S.32S6SO 224. electron-transferring-flavoprotein TFDH NM 004453.1 2110 S.155729 dehydrogenase 225. ELL-related RNA polymerase II, E LL2 NM 012081 22936 S.192221 elongation factor 226. ELMO domain containing 1 LMOD1 NM O18712.2 55531 S.495779 227. elongation factor RNA polymerase II LL3 NM O25165.2 80237 S.424.126 like 3 228. elongation of very long chain fatty E LOVL4 NM O22726.2 6785 S.1O1915 acids (FEN1/Elo2, SUR4/Elo3, yeast)- like 4 229. ELOVL family member 7, elongation E LOVL7 NM O24930.1 79.993 S.274256 of long chain fatty acids (yeast) 230. embryonal Fyn-associated Substrate FS NM O05864.2 10278 S.24587 231. empty spiracles homolog 2 MX2 NM 004.098.2 2018 S.202095 (Drosophila) 232. endothelial differentiation, sphingolipid E DG8 NM O30760.3 53637 S.SO1561 G-protein-coupled receptor, 8 233. endothelin 1 DN1 NM OO1955 1906 S.51.1899 234. engulfment and cell motility 3 (ced-12 LMO3 NM O24712.3 79767 S.377416 homolog, C. elegans) 235. enoyl Coenzyme A hydratase domain ECHDC2 NM 018281. 55.268 S.476319 containing 2 236. VPL NM OO1988. 2125 237. EPH receptor B6 PHB6 NMOO4445.2 2051 238. ephrin-A3 FNA3 NM OO4952.3 1944 239. epidermal growth factor receptor GFR NM 201282. 1956 (erythroblastic leukemia viral (v-erb-b) oncogene homolog, avian) 240. epidermal retinal dehydrogenase 2 NM 138969.2 195814 S.17O673 241. epiplakin 1 NM O31308. 83481 S.200412 242. epiregulin NM 001432 2069 S.115263 243. epithelial 2 NM 001424.3 2013 S.S31561 244. epithelial V-like antigen 1 NM OO5797.2 102OS S.116651 245. EPS8-like 1 NM 133180. S4869 S.438862 246. EPS8-like 2 NM O22772.2 64787 S.SSO16 247. epsin 3 NM O17957. SSO40 S.165904 248. erythrocyte membrane protein band 4.1 NM 022140 64097 S.SS3S42 ike 4A 249. erythrocyte membrane protein band 4.1 E NM 018424. S4566 S.26918O ike 4B 2SO. erythrocyte membrane protein band 4.1 E NM O20909.2 57669 S.369232 ike 5 251. ets homologous factor NM O12153.3 26298 S.SO2306 252. ets variant gene 7 (TEL2 oncogene) TV 7 NM O16135.2 51513 S.272398 253. eukaryotic elongation factor-2 kinase NM O13302.3 29904 S.498892 254. exophilin 5 NM O15065.1 23O86 S.269591 255. amily with sequence similarity 107, NM O07177.1 11170 S.8O22 member A 256. amily with sequence similarity 13, FAM13C1 NM OO1 OO1971.1 22096S S.499.704 member C1 257. amily with sequence similarity 46, FAM46B NM 052943.2 115572 S.59771 member B 258. amily with sequence similarity 62 (C2 FAM62C BCO37292.1 838SO S.477711 domain containing), member C 259. amily with sequence similarity 79, FAM79B NM 198485.1 28S386 S.338851 member B 26O. amily with sequence similarity 83, FAM83A NM 032899.4 84985 S.3798.21 member A 261. amily with sequence similarity 83, FAM83F NM 1384.35.1 113828 S.19768O member F 262. FAT tumor Suppressor homolog 2 FAT2 NM OO1447.1 21.96 S.1321.58 (Drosophila) 263. atty acid amide hydrolase 2 FAAH2 NM 174912.2 158584 S.496.205 264. atty acid binding protein 5 (psoriasis FABP5 NM OO1444.1 2171 S.S58327 associated) 26S. F-box and WD-40 domain protein 7 NM OO1013415.1 SS294 S.S19029 (archipelago homolog, Drosophila) US 7,615,349 B2 19 20

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 266. F-box protein 42 NM O18994.1 S44S5 HS.S22384 267. Fc fragment of IgE, high affinity I, NM OO2001 2205 HS.897 receptor for; alpha polypeptide 268. Fc fragment of IgG binding protein FCGBP NM OO3890.1 8857 HS.111732 269. er-1-like 3, myoferlin (C. elegans) FER1L3 NM O13451.2 26509 HSSOOS/2 270. FERM and PDZ omain containing 1 FRMPD1 NM O14907.1 22844 HS.163990 271. etuin B FETUB NM O14375.2 26998 HS.81073 272. fibroblast growth actor binding protein 1 FGFBP1 NM O05130.3 9982 HS.1690 273. fibroblast growth actor receptor 2 FGFR2 NM O22972.1 2263 HS.S33683 (bacteria-expresse dkinase, grow h factor receptor, craniofacial dysos osis 1, Crouzon syndrome, Pfeiffer syndrome, Jackson Weiss syndrome) 274. fibroblast growth factor receptor 3 NM 000142.2 2261 (achondroplasia, thanatophoric dwarfism) 275. fibronectin type III domain containing 6 FNDC6 NM 144717.2 152028 S.61232 276. fibulin 2 FBLN2 NM OO1998.2 2199 S.198862 277. FLG NM OO2016.1 2312 S.23.783 278. filaggrin 2 RP1-14N1.3 NM 001014342.1 388698 S.156124 279. FLJ41603 protein FLJ41603 NM 001001669.2 38.9337 S.2562O6 28O. orkhead box N1 FOXN1 NM OO3593.2 8456 S.1983.13 281. orkhead box Q1 FOXO1 NM O33260.2 94234 S.2974S2 282. rizzled homolog 10 (Drosophila) FZD10 NM O07 1972 11211 S.31664 283. urry homolog (Drosophila) FRY NM O23037 10129 S.390874 284. FYVE, RhoGEF and PH domain FGD6 BCO13319.1 55785 S.SO6381 containing 6 28S. G protein-coupled receptor 1 GPR1 NM O05279.2 282S S.184907 286. G protein-coupled receptor 115 GPR115 NM 153838.2 221393 S.15O131 287. G protein-coupled receptor 172B GPR172B NM O17986.2 55065 S.110 128 288. G protein-coupled receptor 87 GPR87 NM O23915.2 S3836 S.S8561 289. G protein-coupled receptor, family C, GPRC5C NM O22036.2 55890 S.446438 group 5, member C 29O. galectin-related protein HSPC159 NM 014181.1 29094 S.372208 291. gamma-aminobutyric acid (GABA) A GABRE NM 021984.2 2564 S.2278S receptor, epsilon 292. gamma-glutamyltransferase 6 homolog GGT6 NM 153338.1 124975 S.130749 (rat) 293. gap junction protein, alpha 1, 43 kDa A1 NM OOO165.2 2697 Hs.74471 (connexin 43) 294. gap junction protein, beta 2, 26 kDa NM 004004.3 27O6 S.524894 (connexin 26) 295. gap junction protein, beta 3, 31 kDa NM 024009.2 2707 S.S22561 (connexin 31) 296. gap junction protein, beta 5 (connexin NM OO5268.2 2709 S.198249 31.1) 297. gap junction protein, beta 6 (connexin NM OO6783.2 10804 S.511757 30) 298. GATA binding protein 3 3 NM OO1002295.1 2625 S.524134 299. () NM O22041.2 8139 S.112569 3OO. GIPC PDZ domain containing family, PC2 NM O17655.4 S4810 S.13852 member 2 301. gliomedin LDN BXS38105.1 342O3S S.526441 3O2. GLIS family Zinc finger 1 LIS1 NM 147193.1 148979 S.306691 3O3. glutamic pyruvate transaminase PT2 NM 133443.1 847O6 S.460693 (alanine aminotransferase) 2 3O4. glutathione peroxidase 2 NM O02083.2 2877 HS.2704 (gastrointestinal) 3OS. utathione peroxidase 3 (plasma) NM 002084.3 2878 S.386793 306. utathione S-transferaseA3 STA3 NM OOO847.3 2940 S.102484 307. utathione S-transferase A4 STA4 NM OO1512.2 2941 S.48SSS7 3O8. utathione S-transferase omega 2 STO2 NM 1832.39.1 119391 S.2O3634 3.09. ycerophosphodiester DPD2 NM O17711.2 54857 S.438712 hosphodiesterase domain containing 2 3.10. ycerophosphodiester DPD3 NM OO1031718.1 79153 S.289015 hosphodiesterase domain containing 3 3.11. ycolipid transfer protein NM 016433.3 S1228 S.381.256 312. ycoprotein Ib (platelet), beta NM 000407.4 2812 S.51741O olypeptide 313. ycosylphosphatidylinositol specific PLD1 NM OO1503.2 2822 S.512OO1 hospholipase D1 3.14. GM2 ganglioside activator NM OOO405.3 2760 S.483873 US 7,615,349 B2 21 22

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 3.15. GPI-gamma 4 GPIG4 NM 152545 153O20 S.352SS2 316. grainyhead-like 1 (Drosophila) GRHL1 NM 198182.1 2984 S.418493 317. grainyhead-like 2 (Drosophila) GRHL2 NM O24915.1 79977 S.161160 3.18. grainyhead-like 3 (Drosophila) GRHL3 NM 021180.2 57822 S.36982S 3.19. GRAM domain containing 1C GRAMD1C NM 017577.2 S4762 S.24583 32O. GRINL1A combined protein Gcom1 NM OO1018090.1 14578 S.SO841 321. growth arrest-specific 6 GAS6 NM 000820.1 262 S.3692O1 322. growth hormone regulated TBC protein 1 GRTP1 NM O24719.1 79774 S.170904 323. guanine deaminase GDA NM OO4293 S.494163 324. guanylate binding protein family, GBP6 AL703282 S.254338 member 6 325. H3 histone, family 3B (H3.3B) H3F3B NM OO5324.3 326. hairless homolog (mouse) HR NM 005144.3 327. heparan Sulfate (glucosamine) 3-O- HS3ST6 NM 0010096.06.1 sulfotransferase 6 328. hepatic leukemia factor HLF NM 002126.4 329. hepatocellular carcinoma antigen gene LOC63928 NM O22097.1 520 330. hephaestin HEPH NM O14799 98.43 331. histidine ammonia-lyase HAL NM 002108.2 3O34 332. homeo box A9 HOXA9 NM 152739.3 3205 333. homeodomain-only protein HOP NM O32495 8452S 334. homer homolog 2 (Drosophila) HOMER2 NM 004839.2 94.55 335. hook homolog 1 (Drosophila) HOOK1 NM O15888.4 S1361 S.378836 336. Htra serine peptidase 1 HTRA1 NM OO2775.3 S654 S.5O128O 337. huntingtin interacting protein 1 related HIP1R NM OO3959.1 9026 S.524.815 338. hyaluronan synthase 3 HAS3 NM 138612.1 3O38 S.85962 339. hyaluronoglucosaminidase 1 HYAL1 NM 153282.1 3373 S.75619 340. hyaluronoglucosaminidase 4 HYAL4 NM 012269.1 23553 S.28673 341. hydatidiform mole associated and HYMAI NROO2768.1 57061 S.698.009 imprinted 342. hydroxysteroid (11-beta) HSD11B2 NM 000196.2 3291 S.1376 dehydrogenase 2 343. hypothetical gene Supported by LOC441461 XM 499157.2 441461 S.1631SS BCO3O123 344. hypothetical LOC653602 LOC6536O2 XM 93.9526.1 S.433956 345. hypothetical protein DJ667H12.2 DJ667H12.2 NM O19605 56256 S.4458.35 346. hypothetical protein DKFZP761M1511 DKFZP761M1511 AKO96661.1 S4492 S.91521 347. hypothetical protein FLJ10156 FLJ101S6 NM O19013 S4478 S.404323 348. hypothetical protein FLJ10634 FLJ10634 XM 938.980.1 643.338 S.334475 349. hypothetical protein FLJ10706 FLJ10706 NM 01818.6 55732 S.443551 3SO. hypothetical protein FLJ11183 FLJ111.83 AKO26881 55785 S.17O623 351. hypothetical protein FLJ11235 FLJ11235 NM O19033 S4508 S.555952 352. hypothetical protein FLJ20003 FLJ2OOO3 NM O17615 S.258798 353. hypothetical protein FLJ21511 FL21511 NM O25087.1 S.479703 3S4. hypothetical protein FLJ22757 FLJ22757 NM O24898 79958 S.236449 355. hypothetical protein FLJ23306 FLJ23.306 NM O24530 79579 S.5890 356. hypothetical protein FLJ25217 FLJ25 217 XM 9398.93.1 6SO803 S.1534SO 357. hypothetical protein FLJ30532 FLJ30532 NM 144724 153562 S.3906O1 358. hypothetical protein FLJ32798 FLJ32798 NM 173496 143098 S.350684 359. hypothetical protein FLJ33868 FLJ33868 NM 152574 158219 S.496OS 360. hypothetical protein FLJ35880 FLJ3588O NM 153264.2 256O76 S.2O5403 361. hypothetical protein FLJ37464 FLJ37464 NM 173815.3 283848 S.346947 362. hypothetical protein FLJ39005 FLJ390OS NM 178521 284.313 S.130286 363. hypothetical protein FLJ90492 FLJ90492 NM 181783 160418 S.331268 364. hypothetical protein from clone 643 LOCS7228 NM O2O467 57228 S.2O65O1 365. hypothetical protein LOC126917 LOC126917 XM 928886.1 126917 S.466 625 366. hypothetical protein LOC130576 LOC130576 NM 177964.3 130576 S.357567 367. hypothetical protein LOC131873 LOC131873 XM 94.0071.1 131873 S.477571 368. hypothetical protein LOC144501 LOC1445O1 NM 182507 1445O1 S.14O978 369. hypothetical protein LOC147645 LOC147645 XM O85831.10 147645 S.293.236 370. hypothetical protein LOC151878 LOC151878 BCO14063.1 151878 S.68O377 371. hypothetical protein LOC196264 LOC196264 BGO371.01 196264 S.15396 372. hypothetical protein LOC283070 LOC283O70 AKO97377.1 283,070 S.376O41 373. hypothetical protein LOC2834.04 LOC2834.04 BCO391041 2834.04 S.SS6299 374. hypothetical protein LOC283666 LOC2836.66 BCO3SO94.2 2836.66 S.56.0343 375. hypothetical protein LOC283874 LOC283874 AKO97909.1 283,874 S.459699 376. hypothetical protein LOC284023 LOC284O23 XM 933997.1 284O23 S.354493 377. hypothetical protein LOC285535 LOC285535 AKO21540.1 285535 S.40O256 378. hypothetical protein LOC286440 LOC286440 AK123807.1 286440 S.348844 379. hypothetical protein LOC338.667 LOC3386.67 BCO43578.1 3386.67 S.SS8217 380. hypothetical protein LOC348938 LOC348938 XM 371777.4 348938 S.4285 381. hypothetical protein LOC646014 LOC646014 XM 928978.1 646O14 S.117853 US 7,615,349 B2 23 24

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 382. hypothetical protein MGC14128 MGC14128 NM O32899 84985 S.290881 383. hypothetical protein MGC14376 MGC14376 NM OO1001870.1 84981 S.417157 384. hypothetical protein MGC2655 MGC26SS XM 085463.6 146439 S.S13285 385. hypothetical protein MGC4248 MGC4248 NM O32333 84.293 S.3O1519 386. ICEBERG caspase-1 inhibitor CEBERG NM O21571 59082 S.S6279 387. IGF-like family member 1 GFL1 NM 198541.1 374918 S.S46SS4 388. IGF-like family member 2 GFL2 NM OO10O2915.1 14792O S.993.76 389. InaD-like (Drosophila) NADL NM 170605.2 102O7 S.478.125 390. inhibitor of DNA binding 1, dominant NM O02165.2 3397 S.SO4609 negative helix-loop-helix protein 391. inositol 1,3,4,5,6-pentakisphosphate 2 PPK NM O22755.4 64768 S.4598.96 kinase 392. inositol(myo)-1 (or 4)- NM O14214.1 3613 S.367992 monophosphatase 2 393. insulin-degrading enzyme NM 004969 3416 S.SOOS46 394. integrin, beta 4 NM OOO213.3 3691 S.3702SS 395. integrin, beta 6 NM OOO888.3 3694 S.47O399 396. interferon regulatory factor 6 NM OO6147.2 3664 S.3SS827 397. interleukin 1 family, member 10 (theta) NM 173161.1 84639 S.306974 398. interleukin 1 family, member 5 (delta) NM O12275.2 26525 S.51 6301 399. interleukin 1 family, member 7 (Zeta) NM O14439.3 271.78 S.166371 400. interleukin 1 family, member 8 (eta) NM O14438.3 271 77 S.278.909 401. interleukin 1 family, member 9 NM 019618.2 S63OO S.211238 402. interleukin 1 receptor antagonist NM 173843.1 3557 S.81134 403. interleukin 1 receptor, type I NM OOO877.2 3554 S.SS7403 404. interleukin 18 (interferon-gamma NM OO1562.2 3606 S.83.077 inducing factor) 40S. interleukin 20 receptor, alpha NM O14432.2 53832 S.445868 4O6. interleukin 22 receptor, alpha 1 NM O21258.2 58985 S.110915 4O7. involucrin NMOO5547 3713 S.157091 408. iroquois homeobox protein 1 NM O24337.3 79.192 S.424156 409. iroquois homeobox protein 3 NM O24336.1 79191 S.4992O5 410. iroquois homeobox protein 4 NM O16358.1 50805 S.196927 411. iroquois homeobox protein 5 NM O05853.4 10265 S.435730 412. jagged 1 (Alagille syndrome) NM 000214.1 182 S.224O12 413. junction UP NM OO2230.1 3728 S.514174 414. kallikrein 10 LK10 NM OO2776.3 5655 S.27.5464 415. kallikrein 11 LK11 NM OO6853.2 11012 S.57771 416. kallikrein 12 LK12 NM O19598.2 43849 S.411572 417. kallikrein 13 LK13 NM O15596.1 26085 S.165296 418. kallikrein 5 LKS NM O12427.3 2S818 S.SO915 419. kallikrein 6 (neurosin, Zyme) LK6 NM 001012964.1 5653 S.79361 420. kallikrein 7 (chymotryptic, stratum LK7 NM OO5046.2 5650 S.151254 corneum) 421. kallikrein 8 (neuropsinovasin) NM OO7196.2 112O2 S.104570 422. kallikrein 9 NM O12315.1 284,366 S.448.942 423. 1 (epidermolytic NM O06121 3848 S.8O828 hyperkeratosis) 424. (epidermolytic KRT10 NM 000421.2 3858 S.99936 hyperkeratosis; keratosis palmaris et plantaris) 425. (epidermolysis bullosa KRT14 NM OOO526.3 3861 S.3SS214 simplex, Dowling-Meara, Koebner) 426. KRT1S NM OO2275.2 3866 S.8O342 427. (focal non-epidermolytic KRT16 NM OO5557.2 3868 S.432448 palmoplantar keratoderma) 428. K RT17 NM 000422.1 3872 S.2785 429. keratin 1B RT1B BCO33.366.1 374.454 S.334989 430. keratin 23 (histone deacetylase RT23 NM O15515.3 25984 S.9029 inducible) 431. (epidermal KRT2A NM 000423.2 3849 S.707 bullosa of Siemens) 432. (epidermolysis bullosa NM 000424.2 3852 S.433845 simplex, Dowling Meara Kobner/Weber-Cockayne types) 433. keratin 5b NM 173352 196374 S.121824 434. RT6A NM OO5554 3853 S.367762 435. RT6B NM OO5555.2 3854 S.524438 436. RT 7 NM OO5556 3855 S.4115O1 437. keratin 74 RT74 NM 175053.2 121391 S.S62SS 438. keratin, hair, acidic, 1 RTHA1 NM OO2.277.2 3881 S.41696 439. keratin, hair, acidic, 2 RTHA2 NM OO2278.2 3882 S.41752 US 7,615,349 B2 25 26

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 440. keratinocyte differentiation-associated KRTDAP NM 207392.1 388533 S.112457 protein 441. KIAAO513 AAO513 NM O14732 97.64 S.3O1658 442. KIAAO514 AAO514 NM 014696.2 97.21 S.523375 443. KIAAO874 protein AAO874 NM O15208 23253 S388877 444. KIAA1117 AA1117 AKO94766.1 23O33 S.S2O246 445. KIAA1145 protein AA1145 NM O2O698 57458 S.173392 446. KIAA1217 AA1217 BX648451.1 S6243 S.445.885 447. KIAA1411 AA1411 NM O2O819 57579 S.2117OO 448. KIAA1543 AA1543 BCO2O431 57662 S.17686 449. KIAA1671 protein AA1671 AL832O19.1 853.79 S.419171 450. KIAA1912 protein AA1912 XM O55636.4 1148OO S.117136 451. KIAA2O22 AA2O22 NM OO10O8537.1 340533 S.124128 452. family member 1C F1C NM OO6612.3 10749 S.435120 453. kinesin light chain 3 LC3 BCO2S318.1 1477OO S.298079 454. KIT ligand TLG NM OOO899.3 4254 S.1048 455. kringle containing transmembrane REMEN1 NM OO1039571.1 83999 S.22933S protein 1 456. Kruppel-like factor 4 (gut) KLF4 NM OO4235.3 93.14 S.3762O6 457. Kruppel-like factor 5 (intestinal) KLFS NM OO1730.3 688 S.508234 458. Kruppel-like factor 8 KLF8 NM O07250 11279 S.411296 459. adinin 1 LAD1 NM OO5558.3 3898 S.S19035 460. LAG1 longevity assurance homolog 3 LASS3 NM 178842.2 204219 S.416099 (S. cerevisiae) 461. LAG1 longevity assurance homolog 4 LASS4 NM O24552.1 S.515111 (S. cerevisiae) 462. aminin, alpha 2 (merosin, congenital LAMA2 NM 000426.2 3908 muscular dystrophy) 463. aminin, alpha 3 LAMA3 NM 198129.1 3909 S.436.367 464. aminin, beta 3 LAMB3 NMOOO228 3914 S.497636 465. aminin, beta 4 LAMB4 NM O07356.1 22798 S.62022 466. aminin, gamma 2 LAMC2 NM OO5562.1 3918 S.S305.09 467. ate cornified envelope 1B LCE1B NM 178349.1 353132 S.375103 468. ate cornified envelope 2B LCE2B NM O14357.3 26239 S.234766 469. ate cornified envelope 3D LCE3D NM O32563.1 84648 S.244349 470. atrophilin 3 LPHN3 NM O15236.3 23284 S.28391 471. ectin, galactoside-binding, soluble, 7 LGALS7 NM OO2307.1 3963 S.558.355 (galectin 7) 472. eucine rich repeat containing 8 family, NM O25061.3 S.SO1511 member E 473. eucine rich repeat neuronal 1 LRRN1 NM O2O873.3 57633 S.163244 474. eucine-rich repeat-containing G LGR4 NM 018490.1 55366 S.SO2176 protein-coupled receptor 4 475. eucine-rich repeat-containing G NM 021636.2 59352 protein-coupled receptor 6 476. eucine-rich repeats and NM 153377.3 121227 S.253736 immunoglobulin-like domains 3 477. eukotriene B4 receptor NM 181657.1 1241 S.S2S256 478. eukotriene C4 synthase NM 145867.1 40S6 S.456 479. igand of numb-protein X1 NM 032622.1 84708 S.4077SS 480. ikely ortholog of mouse Sh3 domain NM O15677 26751 S.14736S YSC-like 1 481. LIM domain kinase 2 LIMK2 NM O05569.3 3985 S.474596 482. ines homolog 1 (Drosophila) LINS1 BCO1O363.1 5518O S.105633 483. ipocalin 2 (oncogene 24-p3) LCN2 NM O05564 3934 S.204238 484. LOC153470 LOC153470 BCO21680.1 38.9336 S.173059 485. LOC284752 LOC284752 AIOS6877 284752 S.382O7 486. LOC340813 LOC340813 AL11747S 340813 S.306344 487. LOC345926 LOC345926 XM 933651.1 644241 S.62929 488. LOC346876 LOC346876 AI224.578 346876 S.147678 489. LOC3494.96 LOC3494.96 BFO64181 349496 S.147964 490. LOC349889 LOC349889 AW589793 349889 S.224713 491. oricrin LOR NM OOO427 4014 S.25168O 492. unatic fringe homolog (Drosophila) NM 00104O168.1 3955 S.159142 493. Ly-6 neurotoxin-like protein 1 NM O23946 66004 S.15866S 494. LY6/PLAUR domain containing 3 NM O14400.2 27076 S.377.028 495. LY6/PLAUR domain containing 5 NM 182573.1 284348 S.44289 496. ymphocyte antigen 6 complex, D NM OO3695.2 8581 S.415762 497. ymphocyte antigen 6 complex, locus NM O25261.1 80740 S.241586 G6C 498. ymphocyte antigen 75 LY75 NM 002349.1 406S S.153563 499. ysosomal-associated membrane LAMP3 NM O14398.2 27074 S.518448 protein 3 US 7,615,349 B2 27 28

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID SOO. macrophage stimulating 1 receptor (c- MST1R NM OO2447.1 4486 S.517973 met-related tyrosine kinase) SO1. major histocompatibility complex, HLA-DQB2 NM 182549.1 312O class II, DQ beta 2 SO2. mal, T-cell differentiation protein 2 MAL2 NM O52886 114569 S.2O1083 SO3. mal, T-cell differentiation protein-like MALL NM O05434.3 7851 S.1850SS SO4. malic enzyme 1, NADP(+)-dependent, ME1 NM OO2395 41.99 S.14732 cytosolic 505. MAM domain containing 2 MAMDC2 NM 153267.3 256691 S.127386 SO6. mannan-binding lectin serine peptidase MASP1 NM OO1879.4 S648 S.89983 1 (C4C2 activating component of Ra reactive factor) 507. MAS-related GPR, member F MRGPRF NM 145O15.2 219928 S.11851.3 SO8. matrix metallopeptidase 28 MMP28 NM O24302.3 79148 S.380710 SO9. melanoma-derived leucine Zipper, MLZE NM O31415.2 56169 S.133244 extra-nuclear factor . membrane associated guanylate kinase, MAGI1 NM 004742.2 9223 S.651939 WW and PDZ domain containing 1 . membrane protein, palmitoylated 7 MPP7 NM 173496.2 143098 S.4991.59 (MAGUK p55 subfamily member 7) . membrane-associated ring finger MAR3 NM 178450.2 11S123 S.132441 . membrane-spanning 4-domains, NM OOO139.2 S.386.748 subfamily A, member 2 (Fe fragment of IgE, high affinity 1, receptor for: beta polypeptide) meningioma (disrupted in balanced MN1 NM 002430.2 4330 S.268515 translocation) 1 . metallophosphoesterase MPPE1 NM O23075 65258 S.154145 ... metastasis associated lung MALAT1 AK13O345.1 378938 S.1871.99 adenocarcinoma transcript 1 (non coding RNA) methyltransferase like 7A METTL7A NM O14033.3 2S840 S.288771 MICAL C-terminal like MICALCL NM O32867.1 84.953 S.128,196 microfibrillar-associated protein 3-like MFAP3L ABO14526.1 9848 S.178121 . microseminoprotein, beta MSMB NM 002443.2 4477 S.255462 . microsomal glutathione S-transferase 1 MGST1 NM O2O300.3 4257 S.3897OO . microtubule associated serine/threonine MAST4 NM 198828.1 375449 S.482329 kinase family member 4 523. microtubule-associated protein 7 MAP7 NM OO3980.3 9053 S.486548 524. mitogen-activated protein kinase kinase MAP3K9 NM 033141.2 4293 S.437214 kinase 9 525. monoamine oxidase A MAOA NM OOO240.2 4128 S.183109 526. MRS2-like, magnesium homeostasis MRS2L NM 020662.2 57380 S.S33291 factor (S. cerevisiae) 527. mucin 15 MUC15 NM 145650.2 143662 S28. muscleblind-like (Drosophila) MBNL1 NM O21038.3 4154 529. VB MYOSB XM 371116.4 4645 S30. myosin VI MYO6 NM OO4999.3 4646 531. NACHT, leucine rich repeat and PYD NALP1 AKO57464.1 22861 (pyrin domain) containing 1 532. NAD(P) dependent steroid HSPC105 NM 145168 93517 S.87779 dehydrogenase-like 533. NADH dehydrogenase (ubiquinone) 1 NDUFA4L2 NM O20142.3 S6901 S.221447 alpha Subcomplex, 4-like 2 S34. Nance-Horan syndrome (congenital NHS NM 1982.70.2 4810 cataracts and dental anomalies) 535. Nanog homeobox NANOG NM O24865.1 79.923 S.3292.96 S36. NDRG family member 4 NDRG4 NM 020465.2 65009 S.322430 537. nebullette NEBL NM OO6393 10529 S.SO2S S38. nephronectin NPNT NM OO1033047.1 255743 S.518921 539. netrin 4 NTN4 NM O21229.3 59.277 S.2O1034 S4O. neuroepithelial cell transforming gene 1 NET1 NM O05863.2 10276 S.2515S 541. neuromedin U NMU NM OO6681.1 10874 S.418367 542. neuronal guanine nucleotide exchange NGEF NM O1985.0.1 25791 S.97316 factor 543. neurotrophin 5 (neurotrophin 4/5) NTFS NM OO6179 4909 S.266902 544. NGNL6975 UNQ6975 AKO9SSSO.1 400952 S.468.368 S45. nicotinamide nucleotide NMNAT3 NM 178177.2 349565 S.2O8673 adenylyltransferase 3 546. NIPA-like domain containing 1 NPAL1 CR749.484.1 152519 S.134190 547. NIPA-like domain containing 2 NPAL2 NM O24759.1 79815 S.3094.89 S48. NSE1 NSE1 NM 145175 151354 S.2608SS US 7,615,349 B2 29 30

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 549. NUAK family, SNF1-like kinase, 1 NUAK1 NM O14840.2 98.91 S.524692 550. nuclear factor IB NFIB NM OO5596.2 4781 S.370359 551. nuclear factor of activated T-cells, NFATC3 NM OO4555.2 4775 S.341716 cytoplasmic, calcineurin-dependent 3 552. nucleosome assembly protein 1-like 2 NAP1L2 NM O21963.2 4674 S. 6618O 553. O-acyltransferase (membrane bound) OACT2 NM 138799.2 129642 S.467634 domain containing 2 SS4. odz, Odd OZ ten-m homolog 2 ODZ2 ABO32953 57451 S.155915 (Drosophila) 555. odz, Odd OZ ten-m homolog 4 ODZ.4 BF112171 26O11 S.213087 (Drosophila) 556. olfactomedin-like 2A NM 182487.1 1696.11 S.357 004 557. orphan short-chain dehydrogenase? NM 148.897.1 121214 S.38O178 reductase 558. OTU domain, ubiquitin aldehyde OTUB2 BCOO9615.1 78990 S.27881S binding 2 559. ovo-like 1 (Drosophila) OVOL1 NM 004561.2 5O17 S.134434 S60. oxoglutarate (alpha-ketoglutarate) OXGR1 NM O80818.3 27,199 S.352.218 receptor 1 S61. p21(CDKN1A)-activated kinase 6 PAK6 NM O2O168.3 56924 S.S13645 S62. p53-regulated apoptosis-inducing P53AIP1 NM 022112.1 63970 S.160953 protein 1 S63. paired-like homeodomain transcription PITX1 NM OO2653 5307 S.84136 actor 1 564. palladin, cytoskeletal associated protein PALLD NM 016081 23022 S.194431 565. palmdelphin PALMD NM O17734.2 54873 S.483993 566. par-6 partitioning defective 6 homolog PARD6G NM O32510 845S2 S.223584 gamma (C. elegans) 567. patatin-like phospholipase domain PNPLA1 NM 173676.1 28S848 S.407002 containing 1 568. PDZ domain containing 3 NM 178140.2 23037 S.481819 569. PDZK1 interacting protein 1 NM OO5764.3 10158 S.431099 570. peptidase inhibitor 3, skin-derived NM OO2638.2 5266 S.112341 (SKALP) 571. peptidyl arginine deiminase, type 1 NM O13358.1 29943 S.412941 572. peptidylprolyl isomerase C (cyclophilin NM O00943.4 S48O S.110364 C) 573. peptidylprolyl isomerase D PPID NM O05038.2 S481 S.183958 (cyclophilin D) 574. PL NM OO2705.3 S493 S.192233 575. peroxisomal biogenesis factor 3 EX3 AKO23593.1 153914 s.7277 576. PERP, TP53 apoptosis effector ERP NM 022121.2 6406S S.520421 577. PET112-like (yeast) ET112L NM 004564.1 S188 S.119316 578. Phosphatidylinositol glycan anchor GL NM OO4278 94.87 S.433422 biosynthesis, class L 579. hosphoinositide-3-kinase, class 2, NM OO4570.2 S288 S.225OO gamma polypeptide S80. hospholipase A2 receptor 1, 180 kDa NM OO1007267.1 22925 S.410477 S81. hospholipase A2, group III NM O15715.2 SO487 S.1496.23 S82. hospholipase A2, group IVB NM O05090.2 8681 S.S672S4 ( cytosolic) S83. hosphorylase kinase, alpha 1 (muscle) PHKA1 NM O02637.1 5255 S84. hosphotyrosine interaction domain PID1 NM O17933.3 55022 containing 1 585. phytanoyl-CoA dioxygenase domain PHYHD1 NM 174933.2 2S4295 S.326391 containing 1 S86. acenta-specific 2 PLAC2 NM 153375.1 257OOO S.515575 587. akophilin 1 (ectodermal PKP1 NM OO1005337.1 5317 S.4973SO dysplasiaskin fragility syndrome) S88. akophilin 2 PKP2 NM OO1005242.1 5318 S.164384 S89. akophilin 3 PKP3 NM OO7183.2 11187 S.S34395 590. asma membrane proteolipid PLLP NM O15993.1 S1090 S.20O821 (plasmolipin) 591. pleckstrinhomology domain PLEKHA1 NM OO1 OO1974.1 59338 S.2878.30 containing, family A (phosphoinositide binding specific) member 1 592. pleiomorphic adenoma gene-like 1 PLAGL1 NM OO2656.2 5325 S.444975 593. poliovirus receptor-related 4 PVRL4 NM 030916.1 81607 S.4924.90 594. potassium channel, Subfamily K, KCNK1 NM OO2.245 3775 S.376874 member 1 595. potassium channel, Subfamily K, KCNK6 NM 004823.1 9424 S.240395 member 6 US 7,615,349 B2 31 32

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 596. potassium channel, Subfamily K, KCNK7 NM O33455.1 10O89 S.175218 member 7 597. potassium inwardly-rectifying channel, KCNJ12 NM 021012.3 3768 S.23.63 subfamily J, member 12 598. potassium voltage-gated channel, Shal KCND3 NM 004980.3 3752 S.S35274 related subfamily, member 3 599. POU domain, class 2, transcription POU2F3 NM O1435.2.1 25833 S.227115 actor 3 600. POU domain, class 3, transcription POU3F3 NM OO6236.1 5455 S.248.158 actor 3 6O1. PP12104 mRNA PP12104 XM 928.053.1 643008 6O2. pre-B-cell leukemia transcription factor 1 PBX1 NM OO2585.1 5087 S.493.096 603. premature ovarian failure, 1B POF1B NM 024921 79983 S.267038 604. PRKC, apoptosis, WT1, regulator PAWR NM OO2583 5074 S.406O74 605. progastricsin (pepsinogen C) PGC NM 002630.1 5225 S.1867 606. progesterone receptor membrane PGRMC2 NM OO6320.1 10424 S.SOf 910 component 2 6O7. programmed cell death 4 (neoplastic PDCD4 NM O14456 27250 S.232543 transformation inhibitor) 608. prolactin-induced protein PIP NM O02652.2 S304 S.99949 609. proline rich Gla (G-carboxyglutamic PRRG4 NM 024081.4 79056 S.471695 acid) 4 (transmembrane) 610. prominin 2 PROM2 NM 144707.1 150696 S.469313 611. proprotein convertase Subtilisinkexin PCSKS NM O06200.2 5125 S.368542 type 5 612. prostaglandin E receptor 3 (Subtype PTGER3 NM 198712.2 5733 S.445OOO EP3) 613. prostaglandin-endoperoxide synthase 1 PTGS1 NM 080591.1 5742 S.2O1978 (prostaglandin GH synthase and cyclooxygenase) 614. protease, serine 27 PRSS27 NM 031948.3 83886 S.332878 61S. protease, serine, 12 (neurotrypsin, RSS12 NM OO3619.2 84.92 S.445857 motopsin) 616. protease, serine, 2 (trypsin 2) RSS2 NM OO2770.2 S645 S.367767 617. protease, serine, 3 (mesotrypsin) RSS3 NM O02771.2 S646 S.435699 618. protease, serine, 8 (prostasin) RSS8 NM OO2773.2 5652 S.75799 619. protein kinase (cAMP-dependent, KIA NM 181839.1 5569 S.433700 catalytic) inhibitor alpha 620. protein phosphatase 1, regulatory NM OO6663.2 10848 S.S6O721 (inhibitor) subunit 13 like 621. protein phosphatase 1, regulatory NM 030949.1 81706 S.486,798 (inhibitor) subunit 14C 622. protein phosphatase 2 (formerly 2A), NM 020416 5522 S.479069 regulatory subunit B (PR52), gamma isoform 623. protein phosphatase 2 (formerly 2A), PPP2R3A NM OO2718.3 5523 regulatory Subunit B", alpha 624. protein tyrosine phosphatase, non PTPN21 NM O07039.2 11099 S.437040 receptor type 21 625. protein tyrosine phosphatase, non PTPN3 NM OO2829.2 5774 S.436429 receptor type 3 626. protein tyrosine phosphatase, receptor PTPRF NM 002840 5792

627. tyrosine phosphatase, receptor PTPRK NM 002844.2 5796 S.155919

628. protocadherin 2 PCDH21 BCO38,799.1 92211 S.137.556 AA1775) 629. protocadheringamma Subfamily B, 7 PCDHGB7 NM O32101.1 S6099 S.36816O 630. psoriasis Susceptibility 1 candidate 2 PSORS1C2 NM O14069.1 170680 S.146824 631. PTK6 protein tyrosine kinase 6 PTK6 NM OO5975.2 5753 S.S1133 632. Purkinje cell protein 4 like 1 PCP4L1 XM 93.8798.1 654790 S.433 150 633. pyrin-domain containing protein 1 PYC1 NM 152901 260434 S.58314 634. RAB11 family interacting protein 1 RAB11FIP1 NM OO1002233.1 80223 S.191179 (class I) 63S. RAB25, member RAS oncogene RAB2S NM O2O387.1 57111 S.4913O8 amily 636. RAB27B, member RAS oncogene RAB27B NM 00416.3.3 5874 S.S.14875 amily 637. RAB3D, member RAS oncogene RAB3D NM 004283.2 95.45 S.567257 amily 638. RAB7B, member RAS oncogene NM 177403.3 338382 S.S34612 amily US 7,615,349 B2 33 34

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 639. Rap guanine nucleotide exchange RAPGEFL1 NM O16339.1 51195 S.158530 factor (GEF)-like 1 640. Rap2 binding protein 9 NM 138290 154661 S.41 1488 641. RAR-related orphan receptor A ORA NM 002943.2 6095 S.S11626 642. ras homolog gene family, member D OD NM O14578.2 29984 S.15114 643. ras homolog gene family, member V OV NM 133639 171177 S.447901 644. RAS p21 protein activator 4 SA4 NM OO6989.3 101S6 S.558443 645. RAS-like, estrogen-regulated, growth E RG NM O32918. 85004 S.199487 inhibitor 646. regulating synaptic membrane NM O14747.2 9783 S.434924 exocytosis 3 647. repetin XM 937200. 126638 S.376144 648. restin (Reed-Steinberg cell-expressed NM OO2956.2 6249 S.524809 -associated protein) 649. retinoic acid early transcript 1E NM 139165. 135250 S.51.1818 6SO. retinol dehydrogenase 12 (all-trans and DH12 NM 152443. 145226 S.415322 9-cis) 651. retinol dehydrogenase 16 (all-trans and DH16 NM OO3708.2 8608 S.134958 13-cis) 652. Rhesus blood group, B glycoprotein R HBG NM O20407. 57127 S.131835 653. Rho guanine nucleotide exchange A. RHGEF4 NM O15320.2 SO649 S.469935 factor (GEF) 4 654. Rho-guanine nucleotide exchange GNEF XM 93.2952. S.33254 factor 655. rhomboid, veinlet-like 2 (Drosophila) NM O17821.3 S4933 S.524626 656. ribonuclease, RNase A family, 7 NM O32572.2 84659 S.S2S2O6 657. ring finger protein 180 NM 178532 285671 S.98890 658. ring finger protein 39 NM 170769. 8O352 S.1211.78 659. RNA binding motif protein 35A NMOO1034915.1 S4845 S.487,471 660. RNA binding motif protein 35B NM O24939.2 80004 S.436585 661. S100 calcium binding protein A12 NM OO5621. 6283 S.19413 (calgranulin C ) 662. S100 calcium binding protein A14 OOA14 NM O2O672. 57402 S.288998 663. S100 calcium binding protein A2 s OOA2 NM OO5978.3 6273 S.S16484 664. S100 calcium binding protein A7 OOA7 NM OO2963.2 6278 S.1124.08 (psoriasin 1) 665. S100 calcium binding protein A7-like 1 S NM 176823.2 338324 S.442337 666. S100 calcium binding protein A8 NM OO2964.3 6279 S.416O73 (calgranulin A 667. S100 calcium binding protein A9 OOA9 NM OO2965.2 S.112405 (calgranulin B 668. S100 calcium binding protein P OOP NM OO5980.2 6286 S.2962 669. sciellin CEL NM O03843 8796 S.492938 670. sclerostin domain containing 1 OSTDC1 NM O15464 25928 S.25956 671. secreted LY6 PLAUR domain NM O2O427.2 57152 S.1O3SOS containing 1 672. secretoglobin, family 2A, member 2 NM 002411.1 42SO S.46452 673. secretory leukocyte peptidase inhibitor NM 003064.2 6590 S.S17070 674. Selenoprotein P. plasma, 1 NM O05410.2 6414 S.275775 675. sema domain, immunoglobulin domain NM 00418.6.2 6405 S.32981 (Ig), short basic domain, secreted, (semaphorin) 3F 676. serine (or cysteine) proteinase S ERPINB13 NM O12397 52.75 S.241407 inhibitor, clade B (ovalbumin), member 13 677. serine palmitoyltransferase, long chain S AKO75271.1 14O911 base subunit 2-like (aminotransferase 2) 678. serine peptidase inhibitor, Kazal type 5 NM OO6846.2 11 OOS S.331SSS 679. serine peptidase inhibitor, Kunitz type 1 NM OO3710.3 6692 S.2339SO 680. serine peptidase inhibitor, Kunitz type, 2 NM 021102.2 10653 S.31439 681. serine/threonine?tyrosine kinase 1 NM 018423.1 55359 S.24979 682. serpin peptidase inhibitor, clade A NM 173850.2 145264 S.99476 (alpha-1 antiproteinase, antitrypsin), member 12 683. serpin peptidase inhibitor, clade B ERPINB12 NM 080474.1 89777 S.348.541 (ovalbumin), member 12 684. serpin peptidase inhibitor, clade B ERPINB2 NM OO2575.1 5055 S.514913 (ovalbumin), member 2 685. serpin peptidase inhibitor, clade B ERPINB3 NM OO6919.1 6317 S.227948 (ovalbumin), member 3 US 7,615,349 B2 35 36

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 686. serpin peptidase inhibitor, clade B SERPINB4 NM OO2974.1 6318 S.123O3S (ovalbumin), member 4 687. serpin peptidase inhibitor, clade B SERPINB5 NM OO2639 5268 S.SS279 (ovalbumin), member 5 688. serpin peptidase inhibitor, clade B SERPINB7 NM OO104O147.1 871O S.1382O2 (ovalbumin), member 7 689. serpin peptidase inhibitor, clade B SERPINB8 NM 001031848.1 52.71 S.368.077 (ovalbumin), member 8 690. SH2 domain containing 3A NM OO5490.1 10045 S.439645 691. SH3 domain containing ring finger 2 NM 152550.2 153769 S.443728 692. SH3 domain protein D19 NM OO1009555.2 152503 S.519018 693. SH3-domain binding protein 2 NM OO3O23.2 6452 S.1676.79 694. SH3-domain GRB2-like endophilin B2 NM O20145.2 S6904 S.46O238 695. sidekick homolog 1 (chicken) NM 152744 221935 S.155959 696. signal transducing adaptor family NM 00101.3841.1 5562O S.194385 member 2 697. similar to alpha-2-glycoprotein 1, Zinc LOC646.282 ACOO4522 646.282 S.568109 698. similar to common salivary protein 1 LOC124220 NM 145252.2 124220 S.1OS887 699. Similar to RIKEN cDNA 1810006A16 LOC91862 NM O52858 91862 S.4357.64 gene 700. Similar to RIKEN cDNA 2310002J15 MGC5993.7 NM 199001.1 375,791 S.512469 gene 701. similar to WDNM1-like protein LOC645638 AA1492SO 645638 S.S61OS 702. skin aspartic protease SASP NM 152792.1 151516 S.SS6O2S 703. slit homolog 3 (Drosophila) SLIT3 NM OO3062 6586 S.129229 704. SMAD in the antisense orientation DAMS NM O22001 95.97 S.1677OO 705. SMAD, mothers against DPP homolog SMAD1 NM OO5900 4086 S.388294 1 (Drosophila) . Small breast epithelial mucin LOC118430 NM O58173 118430 S.3484.19 . Small proline-rich protein 1A SPRR1A NMOO5987.2 6698 S.4632O . Small proline-rich protein 1B (cornifin) SPRR1B NM OO3125.2 6699 S.1076 . Small proline-rich protein 2A SPRR2A NM OO5988.2 67OO S.3SSS42 . Small proline-rich protein 2G SPRR2G NM 001014291.2 67O6 S.490253 ... sodium channel, nonvoltage-gated 1 SCNN1A NM 001038.4 6337 S.130989 alpha ... sodium channel, nonvoltage-gated 1, SCNN1B NM OOO336.1 6338 S.414614 beta (Liddle syndrome) ... sodium channel, nonvoltage-gated 1, SCNN1G NM OO1039 6340 S.371727 gamma sodium channel, voltage-gated, type SCN4B NM 174934.1 6330 S.65239 IV, beta ... solute carrier family 1 (high affinity NM O05071 6511 S.S15217 aspartate? glutamate transporter), member 6 ... solute carrier family 15 (oligopeptide SLC15A1 NM O05073.1 6564 S.436893 transporter), member 1 ... solute carrier family 16 SLC16A10 NM O18593.3 117247 (monocarboxylic acid transporters), member 10 ... solute carrier family 16 SLC16A14 NM 152527.3 151473 S.SO4317 (monocarboxylic acid transporters), member 14 ... solute carrier family 18 (vesicular SLC18A2 NM OO3054.2 6571 S.369009 monoamine), member 2 720. solute carrier family 2 (facilitated SLC2A12 NM 1451.76.2 154091 S.486.508 glucose transporter), member 12 721. Solute carrier family 22 (extrancuronal NM O21977 6581 S.242721 monoamine transporter), member 3 722. Solute carrier family 24 57419 S.211252 (sodium/potassium calcium exchanger), member 3 723. solute carrier family 26, member 9 NM 134.325.1 115019 724. solute carrier family 28 (sodium NM O22127.1 64O78 coupled nucleoside transporter), member 3 725. solute carrier family 39 (zinc NM O14579.1 29986 transporter), member 2 726. solute carrier family 4, sodium SLC4A11 NM 032034.1 83959 S.1056O7 bicarbonate transporter-like, member 11 727. solute carrier family 47, member 2 NM 152908.2 1468O2 S.12683O 728. solute carrier family 5 (sodium glucose NM 000343.1 6523 S.1964 cotransporter), member 1 US 7,615,349 B2 37 38

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 729. solute carrier family 6 (amino acid SLC6A14 NM OO7231.1 11254 S.522109 transporter), member 14 730. solute carrier family 6 NM 001043.2 (neurotransmitter transporter, noradrenalin), member 2 731. SH3 domain binding protein 1 SSH3BP1 NM O05470 1OOO6 S.42710 732. spectrin, beta, non-erythrocytic 2 SPTBN2 NM OO6946.1 6712 S.26915 733. spermatogenesis associated 18 SPATA18 NM 145263.1 132671 S.S27090 homolog (rat) 734. sphingomyelin phosphodiesterase 3, SMPD3 NM O18667.2 55512 S.368421 neutral membrane (neutral sphingomyelinase II) 735. sphingosine-1-phosphate phosphotase 2 SGPP2 NM 152386.2 S.210043 736. SRY (sex determining region Y)-box SOX15 NM OO6942.1 S.95582 15 737. SRY (sex determining region Y)-box 7 SOX7 NM 031439.2 83595 S.213194 738. START domain containing 5 STARDS NM O30574.2 80765 S.S13075 739. STEAP family member 4 STEAP4 AKO268.06.1 79689 S.S21008 740. Steroid-5-alpha-reductase, alpha SRDSA1 NM OO1047.2 6715 S.SS2 polypeptide 1 (3-oxo-5alpha-steroid delta 4-dehydrogenase alpha 1) 741. Steroidogenic acute regulator STAR NM 000349.2 6770 S.S21535 742. stonin 2 STN2 NM O33104 85439 S.14248 743. stratifin SFN NM OO6142.3 2810 S.S23718 744. Sulfotransferase family 1E, estrogen SULT1E1 NM OO5420.2 6783 S.479898 preferring, member 1 745. Sulfotransferase family, cytosolic, 2B, SULT2B1 NM OO4605.2 S.369331 member 1 746. Suppression of tumorigenicity 14 ST14 NM O21978.3 6768 S.SO4315 (colon carcinoma) 747. Suprabasin SBSN NM 198538.1 3.74897 748. Surfactant, pulmonary-associated S FTPD NM OO3019.4 6441 protein D 749. Sushi domain containing 4 S USD4 NM O17982.2 55061 S.SS8826 750. synaptotagmin-like 1 S YTL1 NM O32872.1 84958 S.4691.75 751. Syndecan 1 S DC1 NM 001006946.1 6382 S.2246O7 752. Syndecan binding protein (Syntenin) 2 S DCBP2 NM O15685.3 27111 S.S16836 753. syntaxin 19 S TX19 NM OO1001850.1 415117 S.S33O86 754. Synuclein, alpha interacting protein SNCAIP NM OO5460.2 9627 S.426463 (Synphilin) 755. T-box 1 NM O05992.1 6899 S.173984 756. T-cell lymphoma invasion and NM OO3253.1 7074 S.S17228 metastasis 1 757. t-complex 11 (mouse) like 2 NM 152772.1 2SS394 S.132OSO 758. tensin 4 NM O32865.3 84.951 S.438292 759. tetra-peptide repeat homeobox-like NM 182629 34882S S.52918O 760. tetraspanin similiar to uroplakin 1 NM 130783 90.139 S.385634 761. tetratricopeptide repeat domain 12 BCO32355.1 54970 S.288772 762. tetratricopeptide repeat domain 18 NM 145170.2 118491 S.549236 763. tetratricopeptide repeat domain 22 NM O17904.1 55001 S.16230 764. tetratricopeptide repeat domain 9 XM 938197.1 23508 765. three prime repair exonuclease 2 NM O807013 11219 S.1708.35 766. thymic stromal co-transporter NM O33051.2 S7864 S.512668 767. thyroid hormone receptor, beta NM 000461.3 7O68 S.187861 (erythroblastic leukemia viral (v-erb-a) oncogene homolog2, avian) 768. TIGA1 TIGA1 NM O53000 11491.5 S.12082 769. Tight junction protein 2 (Zona TP2 NM 201629.1 9414 S.S.O.382 occludens 2) 770. transcription elongation factor A (SII), 3 TCEA3 NM 003196.1 6920 S.148105 771. transcription factor 4 TCF4 NM OO3199.1 6925 S.20O285 772. transcription factor AP-2 beta TFAP2B NM OO3221.2 7021 S.331 O2 (activating enhancer binding protein 2 beta) 773. transforming growth factor, alpha TGFA NM OO3236.1 7039 S.170009 774. transforming, acidic coiled-coil TACC2 NM OO6997 10579 S.SO12S2 containing protein 2 775. 1 (K polypeptide TGM1 NM OOO359.1 7051 epidermal type I, protein-glutamine gamma-glutamyltransferase) 776. transglutaminase 3 (E. polypeptide, TGM3 NM OO3245.2 7053 S.2022 protein-glutamine-gamma glutamyltransferase) US 7,615,349 B2 39 40

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 777. transglutaminase 5 TGM5 NM OO4245.2 9333 S.129719 778. transient receptor potential cation TRPV1 NM O18727.4 7442 S.SS841S channel, subfamily V, member 1 779. transmembrane pro ease, serine 11D TMPRSS11D NM 004262.2 94O7 S.132.195 780. transmembrane pro ease, serine 13 TMPRSS13 NM O32046.1 84OOO S.2663O8 781. transmembrane pro ease, serine 4 TMPRSS4 NM O19894.2 56649 S.161985 782. transmembrane pro ein 125 TMEM12S NM 144626.1 128.218 S.104,476 783. transmembrane pro ein 154 TMEM154 NM 15268.0.1 201799 S.51890O 784. transmembrane pro ein 16A TMEM16A NM 018043.4 55107 S.S.O.3074 785. transmembrane pro ein 16J TMEM16 NM OO1O12302.1 338440 S.5O1622 786. transmembrane pro ein 184A TMEM184A NM 152689.2 2O2915 S.556755 787. transmembrane pro ein 23 TMEM23 NM 147156.3 259230 S.386215 788. transmembrane pro ein 30B TMEM3OB NM OO1017970.1 161291 S.14618O 789. transmembrane pro ein 40 TMEM40 NM 018306.2 55.287 S.47SSO2 790. transmembrane pro ein 45A TMEM4SA NM 018004.1 55076 S.126598 791. transmembrane pro ein 45B TMEM4SB NM 138788.2 12O224 S.504301 792. transmembrane pro ein 79 TMEM19 NM O32323.1 84.283 S.3474.08 793. tripartite motif-con aining 29 TRIM29 NM 012101 236SO S.SO4115 794. tripartite motif-con aining 31 TRIM3 NM O07028 11074 S.91096 795. tripartite motif-con aining 7 TRIM7 NM 203294.1 81786 S.487412 796. trypsinogen C TRY6 NR 001296.2 154754 S.697092 797. tryptase alphabeta 1 TPSAB NM OO3294.3 7177 S.405479 798. tryptase beta 2 TPSB2 NM O24164.5 64,499 S.S92982 799. tubulin, alpha 1 (testis specific) TUBA1 NM OO6000.1 7277 S.75318 800. tuftelin 1 TUFT1 NM O2O127.1 7286 S.4.89922 8O1. tumor necrosis factor (ligand) TNFSF10 NM OO3810.2 8743 S.4782.75 superfamily, member 10 8O2. tumor necrosis factor receptor TNFRSF25 NM OO3790.2 8718 S.462529 superfamily, member 25 8O3. tumor protein D52-like 1 NM OO10O3395.1 71.64 S.2O1482 804. tumor protein p73-like NM OO3722 8626 S.137569 805. tumor-associated calcium signal NM OO2354.1 4O72 S.692 transducer 1 806. tumor-associated calcium signal NM OO2353.1 4070 S.23582 transducer 2 807. twist homolog 2 (Drosophila) TWIST2 NM O57179 117581 S.422585 8O8. ubiquitin protein ligase E3A (human UBE3A NM 000462.2 7337 S.22543 papilloma virus E6-associated protein, Angelman syndrome) Ubiquitin specific peptidase 2 USP2 NM OO4205.3 9099 S.S24085 UDPglucuronosyltransferase 1 family, NM 000463.2 S4658 S.554822 polypeptide A1 UDP-N-acetyl-alpha-D- GALNT6 NM O07210.2 11226 S.505575 galactosamine:polypeptide N acetylgalactosaminyltransferase 6 (GalNAc-T6) unc-93 homolog A (C. elegans) UNC93A NM O189742 S4346 S.145911 vasoactive intestinal peptide receptor 1 VIPR1 NM 004624.2 7433 S.3485OO vav 3 oncogene WAV3 NM OO6113.3 10451 S.267659 vestigial like 3 (Drosophila) WGLL3 NM O16206.2 389136 S.435O13 v-ets erythroblastosis virus E26 ETS2 NM OO5239.4 2114 S.S17296 oncogene homolog 2 (avian) visinin-like 1 VSNL1 NM OO33854 7447 S.444212 . V-maf musculoaponeurotic MAF NM OO5360 4094 S.134859 fibrosarcoma oncogene homolog (avian) ... WAP four-disulfide core domain 12 WFDC12 NM 080869.1 128488 S.35218O 820. WAP four-disulfide core domain 5 WFDC5 NM 145652.2 149708 S.37SO31 821. WD repeat domain 66 WDR66 BCO36233.1 1444O6 S.S.Of 125 822. wingless-type MMTV integration site WNT11 NM 004626.2 7481 S.108219 family, member 11 823. wingless-type MMTV integration site WNT3 NM O30753.3 7473 S.445.884 family, member 3 824. wingless-type MMTV integration site WNT4 NM O30761.3 S4361 S.25766 family, member 4 825. wingless-type MMTV integration site WNTSB NM O30775.2 81029 S.306051 family, member 5B 826. X Kell blood group precursor-related, XKRX NM 212559.1 402415 S.364911 X-linked 827. X102 protein X102 NM O3O879 63969 S.248065 828. Xgblood group (pseudoautosomal XG NM 175569.1 7499 S.179673 boundary-divided on the X chromosome) US 7,615,349 B2 41 42

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 829. YOD1 OTU deubiquinating enzyme 1 YOD1 NM O18566.3 554.32 S.391944 homolog (yeast) 830. Zinc and ring finger 1 NM O32268.3 84937 S.427284 831. Zinc finger and BTB domain containing 3 NM O24784.2 79842 S.SS8545 832. Zinc finger and BTB domain containing NM OO1039360.1 2015O1 S.S15388 7C 833. Zinc finger protein 165 ZNF165 NM 003447.2 7718 S.55481 834. Zinc finger protein 185 (LIM domain) ZNF185 NM O07150.1 7739 S.16622 835. Zinc finger protein 273 ZNF273 AU132789 10793 S.386264 836. Zinc finger protein 42 ZFP42 NM 174900.3 132625 s.335787 837. Zinc finger protein 662 ZNF662 NM 207404.2 389114 S.293388 838. Zinc finger protein 67 homolog ZFP67 NM O15872 S1043 S.159265 (mouse) 839. Zinc finger protein 682 ZNF682 NM O33196.1 91120 S.306.298 840. Zinc finger protein 750 ZNF750 NM O24702.1 79755 S.SS8544 841. Zinc finger, DHHC-type containing 21 ZDHHC21 NM 178566.2 340481 S.S.61951 842. simil air to cystin 1 LOC649824 XM 938.892 649824 HS.27092 843. LOC642S87 AKO91113 91170 S.301.885 844. hypo hetical gene Supported by LOC285259 AI703273 2852.59 S.1461.59 BCO3S064 845. hypo hetical gene Supported by LOC284974 AI928242 284974 S.119903 BCO2OSS4 846. hypo hetical gene Supported by LOC284.668 BE672487 284.668 S.25766 BCOO7386 847. hypo hetical gene Supported by LOC284456 AL137325 284.456 S.406781 AL137325 848. hypo hetical gene Supported by LOC284224 BCO34640 284224 S.382OOO BCO34640 849. simil air to seven transmembrane helix LOC283929 BG109230 283929 S.374278 receptor 850. simil arto Zinc finger protein 366 LOC2S4647 BE858.194 254647 S.323OS3 851. hypothetical gene Supported by LOC221495 AL117529 221495 S.406762 AL117529 852. hypothetical gene Supported by LOC221042 AL137430 221042 S.38096S AL137430 853. simil ar to p53-induced protein PIGPC1 LOC2O4288 AIO49608 204288 S.3555.17 854. hypothetical gene Supported by LOC145739 AL117S70 145739 S.375657 AL117S70 855. Uncharacterized transcript Supported AW195351 by AW195351 found within the Catenin delta 1 locus 856. Uncharacterized transcript Supported AA418.074 S.110286 by AA418.074 857. ported by sequence in AI888.057 858. ported by sequence in BG484769 342.289 S.115838 859. ported by sequence in XM 933918 646769 S.115838 860. ported by sequence in AKOOOO90 861. ported by sequence in AA86338.9 862. ported by sequence in W1353.06 S.444.277 863. ported by sequence in FOOSO82 S.S1642O 864. ported by sequence in 81445 865. ported by sequence in LOSO153 866. ported by sequence in 934364 S.145761 867. ported by sequence in W1314SO S.121070 868. ported by sequence in 832594 S.471433 869. ported by sequence in 73742 S.36288 870. ported by sequence in E674309 S.418279 871. ported by sequence in U158573 S.288926 872. ported by sequence in WO 2 2 6O 7 S.3792S3 873. ported by sequence in R 77414 S.333SS 874. ported by sequence in U155612 S.269.545 875. ported by sequence in O89783 S.123307 876. ported by sequence in L359055 S.390270 877. ported by sequence in E64S279 S.29792 878. ported by sequence in 939452 S.445064 879. ported by sequence in W974998 S.222430 880. ported by sequence in KO96998 S.29952 881. ported by sequence in WO2S141 S.432SO4 882. ported by sequence in 4 S 2 3 S S S.445872 883. ported by sequence in L137535 S.15806 884. ported by sequence in AAS16469 S.433643 885. ported by sequence in AU1475.15 S.406812 886. ported by sequence in AW974.077 S.283349 US 7,615,349 B2 43 44

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID

887. Gene Su bote by seq uence in AA8904.87 S 24598 888. Gene Su bote by seq uence in N66591 S 44O646 889. Gene Su bote by seq uence in U15146S S 224012 890. Gene Su bote by seq uence in WS17412 S. 50757 891. Gene Su bote by seq uence in U147518 892. Gene Su bote by seq uence in F111088 893. Gene Su bote by seq uence in 924O46 894. Gene Su bote by seq uence in 971.251 895. Gene Su bote by seq uence in R48254 896. Gene Su bote by seq uence in O966.34 897. Gene Su bote by seq uence in FO56273 898. Gene Su bote by seq uence in 457984 899. Gene Su bote by seq uence in G4292SS 900. Gene Su bote by seq uence in O904.87 901. Gene Su bote by seq uence in 65.9426 902. Gene Su bote by seq uence in TS1136 903. Gene Su bote by seq uence in AW291545 904. Gene Su bote by seq uence in AA687916 90S. Gene Su bote by seq uence in BF114646 906. Gene Su bote by seq uence in AWO14647 907. Gene Su bote by seq uence in W96062 908. Gene Su bote by seq uence in AI373107 909. Gene Su bote by seq uence in AA758732 910. Gene Su bote by seq uence in S900 911. Gene Su bote by seq uence in L515437 912. Gene Su bote by seq uence in 334358 913. Gene Su bote by seq uence in 458439 914. Gene Su bote by seq uence in L157448 91S. Gene Su bote by seq uence in KO223SO 916. Gene Su bote by seq uence in 97.2146 917. Gene Su bote by seq uence in 22OOS 918. Gene Su bote by seq uence in F245284 919. Gene Su bote by seq uence in O93221 920. Gene Su bote by seq uence in CO19703 921. Gene Su bote by seq uence in U146924 S. 922. Gene Su bote by seq uence in 8296OS S 436O77 923. Gene Su bote by seq uence in W301393 S 493477 924. Gene Su bote by seq uence in 88.112 S 234.478 925. Gene Su bote by seq uence in W173504 S 76704 926. Gene Su bote by seq uence in 565852 S .71947 927. Gene Su bote by seq uence in O22066 S 372209 928. Gene Su bote by seq uence in S.68845 S 324323 929. Gene Su bote by seq uence in S98831 S 434643 930. Gene Su bote by seq uence in U1594.46 S 498.954 931. Gene Su bote by seq uence in CO4O322 S 4.433O 932. Gene Su bote by seq uence in AA826931 S 4.91024 933. Gene Su bote by seq uence in BCO29440 S 374843 934. Gene Su bote by seq uence in AI950O23 S 270751 935. Gene Su bote by seq uence in AV739 182 S 1131SO 936. Gene Su bote by seq uence in AI829721 S 482381 937. Gene Su bote by seq uence in AA876179 S 1346SO 938. Gene Su bote by seq uence in K024907 S 306723 939. Gene Su bote by seq uence in WA51197 S 113418 940. Gene Su bote by seq uence in F224444 S 127274 941. Gene Su bote by seq uence in W130600 S 99.472 942. Gene Su bote by seq uence in 042373 S 132917 943. Gene Su bote by seq uence in LO38973 S 144873 944. Gene Su bote by seq uence in E6724.08 S 107708 945. Gene Su bote by seq uence in W139091 S 161158 946. Gene Su bote by seq uence in WO25O23 S 234.478 947. Gene Su bote by seq uence in ESOO942 S 170540 948. Gene Su bote by seq uence in W242920 S 1293.68 949. Gene Su bote by seq uence in AA603472 S 28456 950. Gene Su bote by seq uence in AU144382 S SO1925 951. Gene Su bote by seq uence in AL8331SO S .327631 952. Gene Su bote by seq uence in BM992214 S 314518 953. Gene Su bote by seq uence in AL3590SS S 390270 954. Gene Su bote by seq uence in AA167323 S 188682 955. Gene Su bote by seq uence in AI970797 S 1331.52 956. Gene Su bote by seq uence in AI733O37 S 129990 957. Gene Su bote by seq uence in H15396 S 107510 958. Gene Su bote by seq uence in BFOSSO60 S .3871OO 959. Gene Su bote by seq uence in N3O188 S 93739 US 7,615,349 B2 45 46

TABLE A-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples.

Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 960. Gene Supported by sequence info. AKOS6784 HS.224.056 961. Gene Supported by sequence info. AU147152 Hs.118317 962. Gene Supported by sequence info. AI341837 HS.436105 963. Gene Supported by sequence info. AIOS1950 HS.99472 964. Gene Supported by sequence info. N91149 HS.279639 965. Gene Supported by sequence info. AI191905 HS.228982 966. Gene Supported by sequence info. BFSO8208 HS.24OO74 967. Gene Supported by sequence info. AU147861 HS.188O82 968. Gene Supported by sequence info. AA779.333 HS2O158 969. Gene Supported by sequence info. AW274846 HS.19339 970. Gene Supported by sequence info. AL137616 HS.30483 971. Gene Supported by sequence info. AKO2S151 HS.268S97 972. Gene Supported by sequence info. BF961733 HS.445604 973. Gene Supported by sequence info. AIO79329 Hs.271876 974. Gene Supported by sequence info. AW2928.30 HS.43SOO1 975. Gene Supported by sequence info. BCO4O965 HS.27.0114 976. Gene Supported by sequence info. AA933O82 HS.126883 977. Gene Supported by sequence info. BCO38,784 HS.385767 978. Gene Supported by sequence info. AW3O22O7 HS.228982 979. Gene Supported by sequence info. AKO2S378 HS.4667OO 980. Gene Supported by sequence info. R97781 HS.2471SO 981. Gene Supported by sequence info. AI6SS611 HS.124863 982. Gene Supported by sequence info. BF847 120 HS.318887 983. Gene Supported by sequence info. AA461490 HS.66O72 984. Gene Supported by sequence info. BCO34636 HS.385523 985. Gene Supported by sequence info. N66393 HS.102754 986. Gene Supported by sequence info. AI637733 HS.220624 987. Gene Supported by sequence info. AAS83O38 HS.3492O7 988. Gene Supported by sequence info. AU156822 Hs.287577 989. Gene Supported by sequence info. BE222843 HS.231852 990. Gene Supported by sequence info. BCO4O965 HS.27.0114 991. Gene Supported by sequence info. AI147310 HS.14673O 992. Gene Supported by sequence info. A417988 HS.37648 993. Gene Supported by sequence info. AA82SS10 HS.124304 994. Gene Supported by sequence info. BG26OO87 HS.7956 995. Gene Supported by sequence info. AI6276.79 HS.37442O 996. Gene Supported by sequence info. AWO70877 HS.432615 997. Gene Supported by sequence info. AA922936 HS.110O39 998. Gene Supported by sequence info. AI935541 HS.127009 999. Gene Supported by sequence info. T68445 Hs.76704 000. Gene Supported by sequence info. AKO97810 HS.146493 001. Gene Supported by sequence info. AWO81982 HS.S9507 002. Gene Supported by sequence info. AI399889 HS.381,411 003. Gene Supported by sequence info. BCO39513 HS.407S75 004. Gene Supported by sequence info. ALS12727 HS.232127 005. Gene Supported by sequence info. AKO21990 HS.287.466 006. Gene Supported by sequence info. AKOOO 106 HS.272227 007. Gene Supported by sequence info. AI819863 HS.106243 008. Gene Supported by sequence info. AI743489 HS.322679 009. Gene Supported by sequence info. ALOSO2O4 HS.285.40 010. Gene Supported by sequence info. AWOO6352 HS.159643 011. Gene Supported by sequence info. BE843544 HS.444613 O12. Gene Supported by sequence info. AIT96535 HSSO4568 013. Gene Supported by sequence info. AFO86294 HS.125844 014. Gene Supported by sequence info. BG2906SO HS.134876 015. Gene Supported by sequence info. AV741130 HS.173704 O16. Gene Supported by sequence info. AI928O3S HS.282O89 017. Gene Supported by sequence info. AI674565 HS.8379 018. Gene Supported by sequence info. AIO853.38 HS.375591 019. Gene Supported by sequence info. AAO34012 HS.37648 020. Gene Supported by sequence info. AI288.186 HS.58611 021. Gene Supported by sequence info. AKO24927 HS.152423 022. Gene Supported by sequence info. BG149SS7 HS.37648 US 7,615,349 B2 47 48

TABLE B Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 023. ATP-binding cassette, Sub-family B ABCBS NM 178559.3 34O273 HS.404102 (MDR/TAP), member 5 024. abhydrolase domain containing 10 ABHD10 NM 018394.1 SS347 HS.477115 025. acyl-CoA synthetase long-chain family ACSL3 NM 004.457 2181 HS.268O12 member 3 026. alkylglycerone phosphate synthase AGPS NM OO3659.2 854O HS.S16543 027. V-akt murine thymoma viral oncogene AKT3 NM 005465.3 1OOOO HS.498.292 homolog 3 (protein kinase B, gamma) 028. anillin, actin binding protein (scraps ANLN NM 018685.2 544.43 HS-6218O homolog, Drosophila) 029. adaptor-related protein complex 1, AP1S2 NM OO3916 890S HS.121592 Sigma 2 subunit 030. apolipoprotein C-II APOC2 NM 000483.3 344 HS.75615 031. ATPase family, AAA domain ATAD2 NM 014109.2 29028 HS.370834 containing 2 O32. activating transcription factor 6 ATF6 NM OO7348 22926 HS.492740 033. ataxiatelangiectasia mutated (includes ATM NM 000051 472 HS.435561 complementation groups A, C and D) O34. ATPase, H+ transporting, lysosomal ATP6V1C1 NM OO1007254.1 528 HS.86.905 42 kDa, V1 subunit C, isoform 1 035. ATPase, Class 1, type 8B, member 2 ATP8B2 NM 020452.2 S7198 HS.4357OO 036. antizyme inhibitor 1 AZIN1 NM 148174.2 S1582 HS.459 106 037. UDP-GlcNAc:beta Gal beta-1,3-N- B3GNT1 NM OO6577.5 10678 HS.1732O3 acetylglucosaminyltransferase 1 O38. UDP-Gal:betaGlcNAc beta 1,4- B4GALT6 NM 004775.2 9331 HS.464848 galactosyltransferase, polypeptide 6 O39. bromodomain adjacent to Zinc finger BAZ1B NM O23005.2 9031 HS.488671 domain, 1B 040. BRCA2 and CDKN1A interacting BCCIP NM O16567.2 56647 HS.370292 protein 041. BCL2-related protein A1 BCL2A1 NM 004.049.2 597 HS.227817 042. bifunctional apoptosis regulator BFAR NM O16561.1 S1283 HS.43SSS6 043. bicaudal D homolog 1 (Drosophila) BICD1 NM OO1714 636 HSSOS2O2 044. baculoviral IAP repeat-containing 5 BIRCS NM 00101.2270.1 332 HS.S14527 (survivin) 045. barren homolog 1 (Drosophila) BRRN1 NM O15341.3 23397 HS.308045 046. butyrophilin, subfamily 2, member A1 BTN2A1 NM 007049.2 11120 HS.159028 047. BUB1 budding uninhibited by BUB1 NM 004336.2 699 HS.469649 benzimidazoles 1 homolog (yeast) 048. open reading frame 11 C12Orf11 NM 018164.1 55726 Hs.505.077 O49. chromosome 12 open reading frame 24 C12Orf24 NM 013300.1 29902 HS.4366.18 050. core 1 synthase, glycoprotein-N- C1GALT1 NM O2O156.1 S6913 HS.239666 acetylgalactosamine 3-beta galactosyltransferase, 1 051. chromosome 1 open reading frame 103 C1 orf103 NM 018372.3 SS791 HS.25245 052. chromosome 1 open reading frame 67 C1 Orf67 XM 290922.4 2OOO95 HS.133977 053. chromosome 1 open reading frame 90 C1orf)0 BCOO4269.1 84734 HS.18449 O54. open reading frame C20orf104 NM 016436 S1230 HS.301OSS 104 055. chromosome 20 open reading frame 30 C20orf50 NM 001009924.1 290S8 HS.472O24 056. chromosome 7 open reading frame 11 C7orf11 NM 138701.1 136647 HS.1291.59 057. open reading frame 100 C9Crf100 NM 032818.2 84904 HS.277026 O58. calumenin CALU NM 001219.2 813 HS.7753 059. cell cycle associated protein 1 CAPRIN1 NM O05898.4 4O76 HS.471818 060. cell division cycle 2, G1 to S and G2 to M CDC2 NM OO1786 983 HS.334562 O61. CDC42 effector protein (Rho GTPase CDC42EP3 NM OO6449.3 10602 HS.369574 binding) 3 062. CDC45 cell division cycle 45-like (S. cerevisiae) CDC45L NM 003504.3 8318 HS.474217 063. CDC6 cell division cycle 6 homolog CDC6 NM OO1254.3 990 HS.405958 (S. cerevisiae) O64. cell division cycle associated 1 CDCA1 NM 031423.2 83S40 HS.2345.45 065. cell division cycle associated 3 CDCA3 NM O31299.3 83461 HS.524216 066. cyclin-dependent kinase 2 CDK2 NM OO1798.2 1017 HS.19.192 067. centromere protein A, 17 kDa CENPA NM OO1809 1058 HS.1594 068. centromere protein F, 350/400ka CENPF NM 0.16343 1063 HS.497741 (mitosin) 069. centromere protein N CENPN NM 018455.3 SS839 HS.283532 070. cofilin 2 (muscle) CFL2 NM 021914.5 1073 HS.18O141 071. chromodomain helicase DNA binding CHD7 NM O17780.2 SS 636 HS2O395 protein 7 072. carbohydrate (chondroitin 4) CHST11 NM 018413.2 SOS 15 HS.17569 sulfotransferase 11 US 7,615,349 B2 49 50

TABLE B-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 073. cytoskeleton associated protein 2-like CKAP2L NM 152515.2 150468 HS.4342SO 074. CDC28 protein kinase regulatory CKS2 NM OO1827.1 1164 HS.837.58 subunit 2 O75. claspin homolog (Xenopus laevis) CLSPN NM 022111.2 63967 HS.175613 076. CSAG family, member 2 CSAG2 NM 004909.1 9598 HS.S22810 O77. cancertestis antigen 1A CTAG1A NM 13925.0.1 2461 OO H.S.SS9536 078. canceritestis antigen 1B CTAG1B NM OO1327.1 1485 H.S.S34310 079. canceritestis antigen 2 CTAG2 NM 020994.2 3O848 HS.87225 080. cyStathionase (cystathionine gamma- CTH NM 001902.4 1491 HS.19904 lyase) 081. cathepsin L-like 3 CTSLL3 L25629.1 1518 HS.418.123 O82. cortactin CTTN NM OO5231 2017 HS.301.348 083. DBF4 homolog (S. cerevisiae) DBF4 NM OO6716.3 10926 HS.208414 084. development and differentiation DDEF1 NM 018482.2 SO807 HS.10601S enhancing factor 1 085. DEAD (Asp-Glu-Ala-Asp) box DDX18 NM OO6773.3 8886 HS-3634.92 polypeptide 18 086. DEP domain containing 1 DEPDC1 NM O17779.3 SS635 HS.445098 087. dihydrofolate reductase DHFR NM 000791.3 1719 HS.464813 O88. hypothetical protein DKFZp762E1312 DKFZp762E1312 NM 018410.3 SS355 HS.S32968 089. DnaJ (Hsp40) homolog, Subfamily C, DNAJC10 NM 018981.1 S4431 HS.S16632 member 10 090. dpy-19-like 1 (C. elegans) DPY19L1 AJO11911.1 23333 HS.S10645 091. denticleless homolog (Drosophila) DTL NM 016448.1 S1514 HS.126,774 092. deltex 3 homolog (Drosophila) DTX3 NM 178502.2 1964.03 HS.32374 093. dual specificity phosphatase 4 DUSP4 NM OO1394.5 1846 HS.417962 094. dual specificity phosphatase 6 DUSP6 NM OO1946.2 1848 HS.298654 095. endothelin 3 EDN3 NM 207032.1 1908 HS.1408 O96. engrailed homolog 2 EN2 NM OO1427.2 2O2O HS-134989 O97. ets variant gene 1 ETV NMOO4956.3 2115 HS.22634 O98. exonuclease 1 EXO NM 130398.2 91.56 HS.498248 099. exosome component 3 EXOSC3 NM 016042.2 S1010 HS.493887 00. eyes absent homolog4 (Drosophila) EYA4 NM 004100 2070 HS.1024.08 01. family with sequence similarity 126, FAM126A NM 032581.2 84668 HS.85603 member A 02. family with sequence similarity 29, FAM29A NM O17645.3 S4801 HS.S33468 member A 03. family with sequence similarity 62 (C2 FAM62B NM 020728.1 S7488 HS.490795 domain containing) member B 04. F-box only protein 32 FBXO32 NM 058229 114907 HS.403933 05. FK506 binding protein 10, 65 kDa FKBP10 NM O21939.2 60681 HS.463O3S 06. hypothetical protein FLJ10781 FLJ10781 NM 018215.2 SS228 HS.839S 07. hypothetical protein FLJ11029 FLJ11029 NM 018304 SS771 HS.274448 08. hypothetical protein FLJ13236 FLJ13236 NM 024902 79962 HS.170298 09. FLJ20105 protein FLJ201OS NM 001009954.1 S4821 HS.47SS8 10. hypothetical protein FLJ30655 FLJ30655 NM 144643 132320 HS.404OOO 11. hypothetical protein FLJ30707 FLJ30707 NM 145019 22O108 HS.292590 12. FLJ36874 protein FLJ36874 NM 152716.1 219988 HS.S23698 13. fibronectin type III domain containing FNDC3B NM O22763.2 64778 HS.15943O 3B 14. FERM domain containing 5 FRMDS NM 032892.3 84978 HS.368399 15. FSH primary response (LRPR1 FSHPRH1 NM OO6733 2491 HS.318398 homolog, rat) 1 16. GA binding protein transcription GABPB2 NM O05254.4 2SS3 HS.S11316 actor, beta subunit 2 17. G antigen 4 GAGE4 NM OO1474.1 2576 HS.460641 18. G antigen 8 GAGE8 NM 012196.1 26749 HS.278606 19. UDP-N-acetyl-alpha-D- GALNT7 NM O17423.1 S1809 HS.1274O7 galactosamine:polypeptide N acetylgalactosaminyltransferase 7 (GalNAc-T7) 20. ganglioside-induced differentiation- GDAP1 NM 018972 S4332 HS.16895O associated protein 1 21. growth differentiation factor 15 GDF15 NM 004.864.1 9518 HS.S15258 22. glutaminase GLS NM O14905.2 2744 HS.116448 23. glutamate dehydrogenase 2 GLUD2 BCOOS111.1 2747 HS.368538 24. golgi transport 1 homolog B (S. cerevisiae) GOLT1B NM 016072.2 S1026 HS-62275 25. G patch domain containing 2 GPATC2 NM 01804.0.1 SS1 OS HS.42O757 26. glycerol-3-phosphate dehydrogenase 2 GPD2 NM 000408.2 2820 HS.148266 (mitochondrial) 27. G protein-coupled receptor 107 GPR107 NM 020960.3 S7720 HS.S12461 28. G protein-coupled receptor 19 GPR19 NM 006143.1 2842 HS.92.458 US 7,615,349 B2 51 52

TABLE B-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 29. general transcription factor IIH, GTF2H2 NM OO1515.2 2966 HS.191356 polypeptide 2, 44 kDa 30. general transcription factor IIH, GTF2H4 NM OO1517.4 2968 HS.485070 polypeptide 4, 52 kDa 31. G-2 and S-phase expressed 1 GTSE1 NM 016426.4 S1512 HS.386189 32. H2A histone family, member V H2AFV NM 201436.1 94239 HS.157379 33. hairyienhancer-of-split related with HEY1 NM 012258 23462 HS.234434 YRPW notif 1 34.3-hydroxyisobutyrate dehydrogenase HIBADH NM 152740.2 11112 HS.406758 35. histone linker H1 domain, spermatid- HILS1 NM 194O72.1 373861 HS.25934 specific 1 36. histone 1, H2bh HIST1H2BEH NM OO3524.2 8345 HS.2478.15 37. histone 1, H3d HIST1H3D NM 003530 8351 HS.S32144 38. hyaluronan-mediated motility receptor HMMR NM 012484.1 3.161 HS.72SSO (RHAMM) 39. HN1 like HN1L, NM 144570 90861 HS.437433 40. heterogeneous nuclear HNRPC NM OO4500 3183 HS.356139 ribonucleoprotein C (C1/C2) 41. homeo box A10 HOXA10 NM 018951.3 3206 HS.110637 42. homeo box B6 HOXB6 NM 018952 3216 HS.98428 43. homeo box B7 HOXB7 NM OO4502.3 32.17 HS.436181 44. homeo box B9 HOXB9 NM O24017 3219 HS.321142 45. homeo box D13 HOXD13 NM 000523.2 3239 HS.152414 46. heat shock 90 kDa protein 1, alpha-like 3 HSPCAL3 NM OO1040141.1 3324 HS.S23S60 47. heat shock 90 kDa protein 1, beta HSPCB NM O07355.2 3326 HSSO9736 48. heat shock 60 kDa protein 1 HSPD1 NM 002156.4 3.329 HS.113684 (chaperonin) 49. heat shock 10 kDa protein 1 HSPE1 NM 002157.1 3336 HS.SS8338 (chaperonin 10) 50. heat shock 105 kDa, 110 kDa protein 1 HSPH1 NMOO6644.2 10808 HS.36927 51. insulin-like growth factor 2 mRNA GF2BP3 NM OO6547 10643 HS.432616 binding protein 3 52. interleukin 1 receptor accessory protein L1RAP NM 00218.2.2 3556 HS.478673 53. potassium voltage-gated channel, KCNQ5 NM 019842.2 S6479 HS.98129 KQT-like subfamily, member 5 54 KIAAO101 KIAAO101 NM 014736.4 97.68 HS.81892 55. KIAAO470 KIAAO470 AKOO1664.1 9859 HS.408293 56. KIAA1618 KLAA1618 NM 020954.2 S7714 HS.S14554 57. KIAA1917 protein KLAA1917 BFOS6204 114804 HS.434389 58. kinesin family member 18A KIF18A NM O31217.2 81930 HS.301052 59. kinesin family member 4A KIF4A NM 012310.2 241.37 HS.279766 60. kinesin family member C1 KIFC1 NM OO2263.2 3833 HS.436912 61. karyopherin alpha 2 (RAG cohort 1, KPNA2 NM 002266.2 3838 HS.1595.57 importin alpha 1) 62. lactamase, beta 2 LACTB2 NM 016027.1 51110 HS.118554 63. hypothetical protein LOC144874 LOC144874 AL832853.1 144874 HS.439363 64. hypothetical protein LOC150946 LOC1SO946 AV751887 1SO946 HS.187912 65. hypothetical protein LOC152485 LOC152485 NM 178835.2 152485 HS.133916 66. LOC34661S LOC346615 XM 934971.1 645591 HS.453810 67. LOC346616 LOC346616 BCO12751.2 645591 HS.285410 68. LOC346658 LOC346658 AI143641 346658 HS.181400 69. LOC346888 LOC346888 XM 294450.5 346887 HS.127286 70. hypothetical gene LOC4.01431 LOC4O1431 NM OO10O8745.1 4.01431 HSSS6097 71. hypothetical gene Supported by LOC44O995 BCO68O85.1 44O995 HSS52645 BCO34933; BCO68085 72. PTD016 protein LOCS1136 NM 016125 S1136 HS.S31.701 73. melanoma antigen LOCS1152 XM 927576.1 644,433 HS.132526 74. hypothetical protein DKFZp434D2328 LOC91526 NM 153697 91526 HS.11571 75. hypothetical protein LOC92249 LOC92249 AKOO1756.1 92.249 HS.31532 76. peroxisomal Ion protease 2 LONP2 NM 031490.2 83752 HSSSS994 77. leucine zipper protein 5 LUZP5 AKO920O8.1 54892 HS.18616 78. mannose-6-phosphate receptor (cation M6PR NM 002355.2 4074 HS-134084 dependent) 79. MAD2 mitotic arrest deficient-like 1 MAD2L1 NM 002358.2 408S HS.28312 (yeast) 80. melanoma antigen family A, 1 (directs MAGEA1 NM 004988.3 4100 HS.72879 expression of antigen MZ2-E) 81. melanoma antigen family A, 12 MAGEA12 NM 005.367.4 4111 HS.169246 82. melanoma antigen family A, 2B MAGEA2B NM 153488.3 266.740 HS.S34597 83. melanoma antigen family A, 3 MAGEA3 NM OO5362.3 4102 HS.417816 84. melanoma antigen family A, 5 MAGEAS NM 021049.3 4104 HS.546265 85. melanoma antigen family A, 6 MAGEA6 NM O05363.2 4105 HS.441113 86. mannosidase, endo-alpha-like MANEAL NM 152496.1 14917S H.S.S34562 US 7,615,349 B2 53 54

TABLE B-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 87. microtubule-associated protein 9 MAP9 NM OO103958.0.1 79884 HS.S493.31 88. MARCKS-like 1 MARCKSL1. NM 023009.4 65108 HS.75061 89. MCM10 minichromosome MCM10 NM 018518.3 SS388 HS.198363 maintenance deficient 10 (S. cerevisiae) 90. MCM8 minichromosome maintenance MCM8 NM 032485.4 8451S HS.437582 deficient (S. cerevisiae) 91. mucolipin 2 MCOLN2 NM 153259.2 2SS231 HS.459.526 92. malic enzyme 2, NAD(+)-dependent, ME2 NM 002396.3 4200 HS.233119 mitochondrial 93. methyltransferase like 2 METTL2 NM 018396.2 SS798 HS.433213 94. milkfat globule-EGF factor 8 protein MFGE8 NM O05928.1 4240 HS.3745 95. hypothetical protein MGC11082 MGC11082 NM 032691 84777 HS.326729 96. hypothetical protein MGC5509 MGCSSO9 NM O24093 79074 HS.409606 97. muskelin 1, intracellular mediator MKLN1 NM 013255 4289 HS.145599 containing kelch motifs 98. matrix metallopeptidase 12 MMP12 NM OO2426.2 4321 HS.1695 (macrophage elastase) 99. matrix metallopeptidase 14 MMP14 NM 004995.2 4323 HS.2399 (membrane-inserted) 200. matrix metallopeptidase 8 (neutrophil MMP8 NM OO2424.1 4317 HS.161839 collagenase) 201. meiotic nuclear divisions 1 homolog MND NM 032117.2 84057 HS.294.088 (S. cerevisiae) 202. M-phase phosphoprotein 9 MPHOSPH9 NM O22782 10198 HS.445084 203. metallophosphoesterase domain MPPED2 NM 001584.1 744 HS.28979S containing 2 204. myosin regulatory light chain MRCL3 MRCL3 NM OO6471 10627 HS.233936 205. mitochondrial ribosomal protein L44 MRPL44 NM 022915.2 65080 HS2O3559 206. musashi homolog 2 (Drosophila) MSI2 NM 138962.2 124540 HS-134470 207. methylene tetrahydrofolate MTHFD2 NMOO6636 10797 HS.154672 dehydrogenase (NAD+ dependent), methenyltetrahydrofolate cyclohydrolase 208. mitochondrial translational release MTRF1 NM 004294.2 96.17 HS.382176 factor 1 209. multiple substrate lipid kinase MULK BCOO9775.1 SS750 HS.S21240 210. nuclear cap binding protein subunit 1, NCBP1 NM 0024.86 4686 HS.4392O3 80 kDa 211. nuclear cap binding protein subunit 2, NCBP2 NM OO7362 22916 HS.240770 20 kDa 212. neural precursor cell expressed, NEDD4L NM O15277.2 23327 HS.185677 developmentally down-regulated 4-like 213. NIMA (never in mitosis genea)-related NEK2 NM OO2497.2 4751 HS.153704 kinase 2 214. nicotinamide nucleotide NNT NM 012343.2 23S30 HS.482O43 transhydrogenase 215. neuropilin 2 NRP2 NM OO3872 88.28 HS.471200 216.5',3'-nucleotidase, cytosolic NTSC NM 014595 3O833 HS.672O1 217. Nudic domain containing 1 NUDCD1 NM 032869.2 84.955 Hs.558577 218. nudix (nucleoside diphosphate linked NUDT4 NM O19094 11163 HSSO 6325 moiety X)-type motif 4 219. Outer dense fiber of sperm tails 2 ODF2 NM 153437.1 4957 HS.1290SS 220. procollagen-proline, 2-oxoglutarate 4- P4HA1 NM OOO917.2 SO33 HSSOOO47 dioxygenase (proline 4-hydroxylase), alpha polypeptide I 221. poly(A) binding protein, cytoplasmic 4 PABPC4 NM OO3819.2 8761 HS.169900 (inducible form) 222. Pantigen family, member 1 (prostate PAGE1 NM 003785.3 8712 HS.128231 associated) 223. PDZ binding kinase PBK NM 018492.2 SS872 HS.104741 224. PDGFA associated protein 1 PDAP1 NM O14891.5 11333 HS.278426 225. phosphodiesterase 4B, cAMP-specific PDE4B NM OO10373.39.1 S142 HS.198072 (phosphodiesterase E4 dunce homolog, Drosophila) 226. protein disulfide isomerase family A, PDLA6 NM O05742.2 1O130 HS.2121 O2 member 6 227. paternally expressed 10 PEG10 NM O15068.3 23O89 HS.147492 228. peroxisome biogenesis factor 13 PEX13 NM 002618.2 S194 HS.368348 229. prefoldin2 PFDN2 NM 012394.3 S2O2 HS.492S16 230. phosphofructokinase, muscle PFKM NM 000289.3 S213 HS.7S160 231. 2 PFN2 NM 002628.4 S217 HS.91747 232. prohibitin PHB NM 002634.2 S245 HS.S14303 US 7,615,349 B2 55 56

TABLE B-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 233. pleckstrin homology-like domain, PHILDA1 AKO74510.1 652993 HS.82101 amily A, member 1 234. proteolipid protein 1 (Pelizaeus- PLP1 NM 000533.3 S354 HS.1787 Merzbacher disease, spastic paraplegia 2, uncomplicated) 235. phospholipid scramblase 1 PLSCR1 NM 021105.1 S359 HS.130759 236. plexin B3 PLXNB3 NM 005.393.1 S365 HS.380742 237. partner of NOB1 homolog (S. cerevisiae) PNO1 NM 020143.2 S6902 HS.262858 238. polymerase (RNA) I polypeptide B, POLR1B NM O19014.3 84172 HS.86337 28 kDa 239. polymerase (RNA) III (DNA directed) POLR3K NM 016310.2 S1728 HS.4371.86 polypeptide K, 12.3 kDa 240. popeye domain containing 3 POPDC3 NM O22361.3 642O8 HS.458336 241. POU domain, class 3, transcription POU3F2 NM O05604.2 S4S4 HS.182SOS actor 2 242. PTPRF interacting protein, binding PPFIBP1 NM OO3622 8496 HS.172445 protein 1 (liprin beta 1) 243. PR domain containing 13 PRDM13 NM 021620.2 S9336 HS.287386 244. prolyl endopeptidase-like PREPL ABOO7896.1 9581 HS.112916 245. protein kinase, cAMP-dependent, PRKAR1A NM OO2734.3 SS73 HS.28O342 regulatory, type I, alpha (tissue specific extinguisher 1) 246. phosphoribosyl pyrophosphate PRPS1 NM OO2764.2 S631 HS.S6 synthetase 1 247. proline?serine-rich coiled-coil 1 PSRC1 NM 001005290.2 84722 HS.405925 248. PTK2 protein tyrosine kinase 2 PTK2 NM O05607 S747 HS.39S482 249. protein tyrosine phosphatase-like A PTPLAD1 NM 016395.1 S1495 HS.S12973 domain containing 1 250. RAD54 homolog B (S. cerevisiae) RADS4B NM 012415.2 25788 HS.30561 251. RAD54-like (S. cerevisiae) RADS4L NM 003579.2 8438 HSS23220 252. Ras protein-specific guanine RASGRF1 NM 002891 S923 HS.459035 nucleotide-releasing factor 1 253. regulator of G-protein signalling 1 RGS1 NM OO2922.3 S996 HS.752S6 254. regulator of G-protein signalling 20 RGS2O NM 003702 8601 HS.368,733 255. Rashomolog enriched in brain RHEB NM O05614 6009 HS.283S21 256. Rho-related BTB domain containing 2 RHOBTB2 NM O15178.1 23221 HS.372688 257. rashomolog gene family, member Q RHOQ NM 012249.2 234.33 HS.S4912S 258. ribulose-5-phosphate-3-epimerase RPE NM 199229.1 61.20 HS.282260 259. Ras-related GTP binding D RRAGD NM 021244.2 S8528 HS.48S938 260. ribonucleotide reductase M2 RRM2 NM 001034 6241 HS.226390 polypeptide 261. RRN3 RNA polymerase I transcription RRN3 NM 018427.3 S4700 HS.46OO78 actor homolog (yeast) 262. RNA terminal phosphate cyclase RTCD1 NM 003729.1 8634 HS.484222 domain 1 263. Sin3-associated polypeptide, 30 kDa SAP30 NM 003864 8819 HS.413835 264. Src family associated phosphoprotein 2 SCAP2 NM OO3930.3 8935 HS2O0770 265. SEC22 vesicle trafficking protein-like SEC22L3 NM 032970.2 91.17 HS.445892 3 (S. cerevisiae) 266. SEC24 related gene family, member A SEC24A AJ131244 10802 HS.211612 (S. cerevisiae) 267. septin 2 SEPT2 NM 001008491.1 473S HS.335057 268. septin 9 SEPT9 NM O06640 108O1 HS.288094 269. Stress-associated endoplasmic SERP1 NM 014.445 27230 HS.S18326 reticulum protein 1 270. serine (or cysteine) proteinase SERPINE2 NM OO6216 S270 HS.21858 inhibitor, clade E (nexin, plasminogen activator inhibitor type 1), member 2 271. seizure related 6 homolog (mouse)-like 2 SEZ6L2 NM 012410.1 26470 HS.6314 272. sarcoglycan, epsilon SGCE NM OO3919.1 8910 HS.371.199 273. solute carrier family 16 SLC16A4 NM 004696.1 9122 HS.3S1306 (monocarboxylic acid transporters), (SL16AJ) member 4 274 solute carrier family 2 (facilitated SLC2A3 NM OO6931.1 651S HS.419240 glucose transporter), member 3 275. solute carrier family 43, member 3 SLC43 A3 NM O14096.2 2901S HS.99962 276. SMC4 structural maintenance of SMC4L1 NM OO1002799.1 10051 HS.S8992 chromosomes 4-like 1 (yeast) 277. sorting nexin 25 SNX25 NM O31953.2 83891 HS.369091 278. SRY (sex determining region Y)-box 5 SOX5 NM OO6940 666O HS.434948 279. secreted phosphoprotein 1 SPP1 NM OOO582 6696 HS.313 (osteopontin, bonesialoprotein I, early T-lymphocyte activation 1) US 7,615,349 B2 57 58

TABLE B-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples. Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 28O. sprouty-related, EVH1 domain SPR E NM 152594.1 161742 S.S25781 containing 1 281. sprouty-related, EVH1 domain SPR E NM 181784.1 200734 containing 2 282. Sprouty homolog 2 (Drosophila) SPRY2 NM 005842.2 10253 S.18676 283. signal sequence receptor, alpha SSR NM 003144.2 6745 S.114033 (translocon-associated protein alpha) 284. ST6 (alpha-N-acetyl-neuraminyl-2,3- ST6GALNAC3 NM 152996.1 256435 S.337040 beta-galactosyl-1,3)-N- acetylgalactosaminide alpha-2,6- sialyltransferase 3 285. stress 70 protein chaperone, STCH NM OO6948.4 6,782 S.352341 microsome-associated, 60 KDa 286. serine/threonine kinase 6 STK6 NM 1984.33.1 6790 287. TATA box binding protein (TBP)- TAF1A NM 005681.2 901S associated factor, RNA polymerase I, A, 48 kDa 288. transcription factor 20 (AR1) TCF2O NM 181492.1 6942 S.47SO18 289. transferrin receptor (p90, CD71) TFRC NM 003234.1 7037 S.529618 29O. TGFB-induced factor (TALE family TGIF NM 170695.2 7.050 S.373550 homeobox) 291. transketolase-like 1 TKTL1 NM 012253.2 8277 S.102866 292. transmembrane protein 38B TMEM38B NM 018112.1 55151 S.411925 293. transmembrane protein 4 TMEM4 NM O14255 10330 HS.8752 2.94. transmembrane protein 48 TMEM48 NM 018087.3 55706 S476525 295. transmembrane protein 64 TMEM64 NM OO1008495.1 1692OO S.SS6805 296. TRAF2 and NCK interacting kinase TNIK NM O15028.1 23O43 S.34024 297. Erinucleotide repeat containing 15 TNRC15 NM O15575 26058 S.334871 298. orsin A interacting protein 1 TOR1AIP1 NM O15602.2 26092 S.496459 299. TP53TG3 protein TP53TG3 NMO16212.2 241SO S.S13537 3OO. TPX2, microtubule-associated, TPX2 NM 012112.4 22974 S.24.458O homolog (Xenopus laevis) 301. translocation associated membrane TRAM1 NM O14294.3 234.71 S.491988 protein 1 3O2. tribbles homolog 2 (Drosophila) TRIB2 NM 021643.1 28.951 3O3. tripartite motif-containing 51 TRIMS1 NM 032681.1 84767 3O4. trophinin associated protein (tastin) TROAP NM 005480.2 10O24 3OS. translin TSN NM 004622.2 7247 306. etratricopeptide repeat domain 26 TTC26 NM 024926.1 79989 307. hymidylate synthetase TYMS NM 001071.1 7298 3O8. ubiquitin-conjugating enzyme E2C UBE2C NM 181800.1 11065 309. ubiquitin-conjugating enzyme E2I UBE2I NM OO3345 7329 (UBC9 homolog, yeast) 31 O. ubiquitin-conjugating enzyme E2T UBE2T NM 014176.1 29089 HS.S199 (putative) 311. ubiquitin family domain containing 1 UBFD1 NM 019116 S6061 HS.3459 312. UDP glycosyltransferase 8 (UDP UGT8 NM OO3360.2 7368 S.1441.97 galactose ceramide galactosyltransferase) 313. X antigen family, member 1 XAGE1 NM 020411.1 9503 S.112208 314. X antigen family, member 2 XAGE2 NM 130777.1 95O2 S.522654 315. Zinc finger protein 697 ZNF697 XM 371286.4 90874 S.381105 316. hypothetical protein LOC650446 LOC6SO446 XM 939537 650446 S.38O331 317. musashi homolog 2 (Drosophila) MSI2 NM 138962 124540 S.185084 3.18. neuropilin 2 NRP2 NM OO3872 8828 S.471200 319. hypothetical gene Supported by BFS16262 144705 S.44817 AFO864O9 32O. glutamate dehydrogenase 2 GLUD2 AA909218 321. hypothetical protein LOC647291 LOC647291 M OO10397.95.1 647291 322. hypothetical gene Supported by 631833 347884 S.374451 AL832403; BCO19824 323. Gene Supported by sequence info. 3863S S.391401 324. Gene Supported by sequence info. G285837 S.444.096 325. Gene Supported by sequence info. W270845 326. Gene Supported by sequence info. 810266 S.130853 327. Gene Supported by sequence info. 911318 S.224153 328. Gene Supported by sequence info. OO3S08 S.117689 329. Gene Supported by sequence info. 554O75 S.120204 330. Gene Supported by sequence info. A424969 S.146268 331. Gene Supported by sequence info. W665.538 S.479714 332. Gene Supported by sequence info. F224436 S.123294 333. Gene Supported by sequence info. KO24236 S.476469 334. Gene Supported by sequence info. E891646 S.281434 US 7,615,349 B2 59 60

TABLE B-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples.

Gene Entrez Gene description Symbol GenBank ID Gene ID UniGene ID 1335. Gene Supported by sequence info. AI936197 HS.360386 1336. Gene Supported by sequence info. AI3747S6 HS.116453 1337. Gene Supported by sequence info. AI861893 HS.129967 1338. Gene Supported by sequence info. BCO26261 HS.364642

TABLE C Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic Samples. Gene Gene description Symbol GenBank ID Entrez Gene ID gap junction protein, beta 6 (connexin 30) GB6 NM OO6783.2 10804 Small proline-rich protein 1A SPRR1A NM OO5987.2 6698 serpin peptidase inhibitor, clade B SERPINB5 NM 002639.2 5268 (ovalbumin), member 5 calmodulin-like 5 CALMLS NM O17422.3 S1806 (CLSP) desmocollin 1 DSC1 NM 024421. 1823 1 (ectodermal dysplasiaskin PKP1 NM 001005337.1 5317 fragility syndrome) chloride channel, calcium activated, family CLCA2 NM OO6536.4 member 2 desmoglein 1 DSG1 NM OO1942. 1828 corneodesmosin CDSN L2O815.1 1041 lymphocyte antigen 6 complex, locus D LY6D NM OO3695.2 8581 late cornified envelope 2B LCE2B NM O14357.3 26239 filaggrin FLG NM 002016. 2312 filaggrin 2 RP1-14N1.3 NM 001014342.1 388698 keratin 16 (focal non-epidermolytic KRT16 NM OO5557.2 3868 palmoplantar keratoderma) Suprabasin SBSN NM 198538.11 3.74897 serpin peptidase inhibitor, clade B SERPINB3 NM OO6919. 6317 (ovalbumin), member 3 serpin peptidase inhibitor, clade B SERPINB7 NM 00104.0147.1 (ovalbumin), member 7 keratin 17 KRT17 NM 000422. 3872 kallikrein 7 (chymotryptic, stratum corneum) KLK7 NM 005046.2 5650 loricrin LOR BC108290.1 4014 secreted LY6/PLAUR domain containing 1 SLURP1 NM 020427.2 57152 hepatocellular carcinoma antigen gene 520 LOC63928 NM O22097. 63928 keratin 15 KRT1S NM OO2275.2 3866 lectin, galactoside-binding, soluble, 7 (galectin LGALS7 NM 0023.07. 3963 7) cystatin E/M CST6 NM OO1323.2 1474 Small proline-rich protein 1B (cornifin) SPRR1B NM OO3125.2 6699 cornifelin CNFN NM 032488.2 84518 triparlite motif-containing 29 TRIM29 NM 058.193. 236SO epiplakin 1 EPPK1 AL137725.1 83481 stratifin SFN NM OO6142.3 2810 keratin 6B KRT6B NM OO5555.2 3854 desmoglein 3 (pemphigus vulgaris antigen) DSG3 BXS38327.1 1830 Small proline-rich protein 2B SPRR2B NM 001017418.1 6701 dermokine DMKN NM 033317.2 93.099 N-acylsphingosine amidohydrolase (alkaline ASAH3 NM 133492 125981 ceramidase) 3 serpin peptidase inhibitor, clade B SERPINB13 NM 012397.2 52.75 (ovalbumin), member 13 kallikrein 11 KLK11 NM OO6853.2 11012 arylacetamide deacetylase-like 2 AADACL2 NM 207365.1 34.4752 death associated protein-like 1 DAPL1 NM 001017920.1 921.96 ATP-binding cassette, Sub-family A (ABC1), ABCA12 NM 173076.2 26154 member 12 desmocollin 3 DSC3 NM OO1941.2 1825 premature ovarian failure, 1B POF1B NM 024921.2 79983 GATA binding protein 3 GATA3 NM OO1002295.1 2625 LY6/PLAUR domain containing 3 LYPD3 NM O14400.2 27076 keratin 6A KRT6A ALS695.11 140446 US 7,615,349 B2 61

TABLE C-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is decreased relative to non-metastatic samples. Gene Gene description Symbol GenBank ID Entrez Gene ID Ets homologous factor EHF AIf 63378 26298 protocadherin 21 PCDH21 NM 033100.1 92211 Cas-Br-M (murine) ecotropic retroviral CBLC NM 012116.2 23624 transforming sequence c fibroblast growth factor receptor 2 (bacteria- FGFR2 NM 022969.1 2263 expressed kinase, keratinocyte growth factor receptor, craniofacial dysostosis 1, Crouzon syndrome, Pfeiffer syndrome, Jackson-Weiss syndrome) Sciellin SCEL NM 144777.1 8796 fibroblast growth factor receptor 3 FGFR3 NM 000142.2 2261 (achondroplasia, thanatophoric dwarfism)

TABLED Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic Samples. Gene GenBank reference Gene description Symbol ID Entrez Gene ID melanoma antigen family A, 3 MAGEA3 NM OO5362.3 4102 melanoma antigen family A, 6 MAGEA6 NM OO5363.2 41OS CSAG family, member 2 CSAG2 NM 004909.1 9598 (TRAG3) melanoma antigen family A, 12 MAGEA12 NM OO5367.4 4111 melanoma antigen family A, 2 MAGEA2 NMOO5361.2 4101 tripartite motif-containing 51 TRIMS1 NM 032681.1 84767 neuropilin 2 NRP2 NM OO3872.2 8828 melanoma antigen family A, 1 (directs MAGEA1 NM 004988.3 41OO expression of antigen MZ2-E) musashi homolog 2 (Drosophila) MIS12 BFO2921S 124540 Glycophorin C (Gerbich blood group) GYPC H3863S 2995 Secreted phosphoprotein 1 (osteopontin, bone SPP1 NM 00104.0060.1 6696 sialoprotein I, early T-lymphocyte activation ) SRY (sex determining region Y)-box 5 SOX5 NM OO6940.4 6660 kinesin family member C1 KIFC1 NM OO2263.2 3833 histone linker H1 domain, spermatid-specific 1 HILS1 NM 194O72.1 373861 regulator of G-protein signalling 20 RGS2O BCO18,618.1 8601 BUB1 budding uninhibited by benzimidazoles BUB1 NM 004336.2 699 homolog (yeast) insulin-like growth factor 2 mRNA binding IGF2BP3 NM OO6547.2 10643 protein 3 FERM domain containing 5 FRMDS NM O32892.3 84978 chromosome 1 open reading frame 90 C1orf)0 BCOO4269.1 84734 eyes absent homolog 4 (Drosophila) EYA4 BE674583 2070 BCL2-related protein A1 BCL2A1 NM OO4049.2 597 Solute carrier family 16 (monocarboxylic acid SLC16A4 NM 004696.1 9122 transporters), member 4 v-akt murine thymoma viral oncogene AKT3 NM OO5465.3 1OOOO homolog 3 (protein kinase B, gamma) CDC45 cell division cycle 45-like (S. cerevisiae) CDC4SL NM OO3504.3 8318 SEC22 vesicle trafficking protein-like 3 (S. cerevisiae) SEC22L3 NM O32970.2 9117 paternally expressed 10 PEG10 NM O15068.3 23O89 popeye domain containing 3 POPDC3 NM O22361.3 642O8 melanoma antigen family A, 5 MAGEAS NM 021049.3 4104 glutamate dehydrogenase 2 GLUD2 BCOOS111.1 2747 ST6 (alpha-N-acetyl-neuraminyl-2,3-beta- ST6GALNAC3 NM 152996.1 256435 galactosyl-1,3)-N-acetylgalactosaminide alpha-2,6-sialyltransferase 3 seizure related 6 homolog (mouse)-like 2 SEZ6L2 NM 201575.1 26470 dual specificity phosphatase 4 DUSP4 BE222344 1846 ATP-binding cassette, Sub-family B ABCBS BCO44248.1 34O273 (MDR/TAP), member 5 Ras protein-specific guanine nucleotide- RASGRF1 NM OO2891.3 5923 releasing factor 1 dual specificity phosphatase 4 DUSP4 NM OO1394.5 1846 FLJ40142 protein FLJ4O142 BFS16262 4OOO73 barren homolog 1 (Drosophila) BRRN1 NM O15341.3 23397 US 7,615,349 B2 63 64

TABLE D-continued Genes differentially expressed in metastatic melanoma. Expression of the genes listed in this table is increased relative to non-metastatic samples. Gene GenBank reference Gene description Symbol ID Entrez Gene ID pleckstrin homology-like domain, family A, PHILDA1 NM O07350.3 22822 member 1 matrix metallopeptidase 14 (membrane- MMP14 NM OO4995.2 4323 inserted) dual specificity phosphatase 6 DUSP6 NM OO1946.2 1848 dpy-19-like 1 (C. elegans) DPY19L1 XM 371891.2 23333 glutamate dehydrogenase 1 GLUD1 AA909218 2746 LOC346615 LOC346615 XM 934971.1 645591 calumenin CALU NM 001219.2 813 ring finger protein 157 RNF157 BFOS 6204 114804 PR domain containing 13 PRDM13 NM 021620.2 S9336 PDZ binding kinase PBK NM O18492.2 55872 KIAA1618 KIAA1618 NM O20954.2 57714 neural precursor cell expressed, NEDD4L NM O15277.2 23327 developmentally down-regulated 4-like bicaudal D homolog 1 (Drosophila) BICD1 BCO1OO91.2 636 ribonucleotide reductase M2 polypeptide RRM2 NM 001034.1 6241

Gene Expression Analysis kits for RNA isolation from commercial manufacturers, such As discussed above, combinations of genes are provided 25 as Qiagen, can be used. For example, total RNA from a herein, for analysis of gene expression in cutaneous tumors sample can be isolated using Qiagen RNeasy mini-columns. (e.g., primary melanoma samples) to determine whether the Other commercially available RNA isolation kits include tumors exhibit a metastatic expression pattern. Methods for MasterPureTM Complete DNA and RNA Purification Kit analyzing gene expression include methods based on hybrid (EPICENTRETM, Madison, Wis.), and, Paraffin Block RNA ization analysis of polynucleotides, sequencing of polynucle 30 Isolation Kit (Ambion, Inc.). Total RNA from tissue samples otides, and analysis of protein expression (e.g., proteomics can be also isolated using RNA Stat-60 (Tel-Test) or by based methods). Commonly used methods are for the cesium chloride density gradient centrifugation. quantification of mRNA expression in a sample include Next, RNA is reverse transcribed into cDNA. The cDNA is northern blotting and in situ hybridization (Parker & Barnes, amplified in a PCR reaction. Two commonly used reverse Methods in Molecular Biology 106:247-283, 1999); RNAse 35 transcriptases are avian myeloblastosis virus reverse tran protection assays (Hod, Biotechniques 13:852 854, 1992); scriptase (AMV-RT) and Moloney murine leukemia virus and PCR-based methods, such as reverse transcription poly reverse transcriptase (MMLV-RT). The reverse transcription merase chain reaction (RT-PCR) (Weis et al., Trends in step is typically primed using specific primers, random hex Genetics 8:263 264, 1992). Alternatively, antibodies may be 40 amers, or oligo-dT primers, depending on the conditions and employed that can recognize specific duplexes, including desired readout. For example, extracted RNA can be reverse DNA duplexes, RNA duplexes, and DNA-RNA hybrid transcribed using a Gene Amp RNA PCR kit (Perkin Elmer, duplexes or DNA-protein duplexes. Representative methods Calif., USA), following the manufacturers instructions. The for sequencing-based gene expression analysis include Serial derived cDNA can then be used as a template in the subse Analysis of Gene Expression (SAGE), and gene expression quent PCR reaction. The PCR reaction typically employs the analysis by massively parallel signature sequencing (MPSS). 45 Taq DNA polymerase, which has a 5'-3' nuclease activity but PCR-Based Methods lacks a 3'-5' proofreading endonuclease-activity. Two oligo Combinations of genes indicative of metastatic or non nucleotide primers are used to generate an amplicon in the metastatic melanoma can be analyzed by PCR. PCR is useful PCR reaction. to amplify and detect transcripts from a melanoma sample. 50 Guidelines for PCR primer and probe design are described, Various PCR methodologies are useful for gene expression e.g., in Dieffenbach et al., “General Concepts for PCR Primer analyses. Design' in: PCR Primer, A Laboratory Manual, Cold Spring Reverse Transcriptase PCR (RT-PCR). RT-PCR is a sensi Harbor Laboratory Press, New York, 133-155, 1995: Innis tive quantitative method that can be used to compare mRNA and Gelfand, “Optimization of PCRs' in: PCR Protocols, A levels in different samples (e.g., non-metastatic and meta 55 Guide to Methods and Applications, CRC Press, London, static melanoma samples, or benign cutaneous and melanoma 5-11, 1994; and Plasterer, T. N. Primerselect: Primer and samples) to examine gene expression signatures. probe design. Methods Mol. Biol. 70:520-527, 1997. Factors To perform RT-PCR, mRNA is isolated from a sample considered in PCR primer design include primer length, melt (e.g., total RNA isolated from a human melanoma sample). ing temperature (Tm), and G/C content, specificity, comple mRNA can be extracted, for example, from frozen orarchived 60 mentary primer sequences, and 3'-end sequence. PCR prim paraffin-embedded and fixed (e.g. formalin-fixed) tissue ers are generally 17-30 bases in length, and Tm's between samples. Methods for mRNA extraction are known in the art. 50-80° C., e.g. about 50 to 7° C. are typically preferred. See, e.g., Ausubel et al., Current Protocols in Molecular For quantitative PCR, a third oligonucleotide, or probe, is Biology, John Wiley and Sons, 1997. Methods for RNA used to detect nucleotide sequence located between the two extraction from paraffin embedded tissues are disclosed, for 65 PCR primers. The probe is non-extendible by Taq DNA poly example, in Rupp and Locker, Lab Invest. 56: A67, 1987, and merase enzyme, and typically is labeled with a reporter fluo De Andres et al., BioTechniques 18:42044, 1995. Purification rescent dye and a quencher fluorescent dye. Any laser-in US 7,615,349 B2 65 66 duced emission from the reporter dye is quenched by the 1898, 2001); and high coverage expression profiling (HiCEP) quenching dye when the two dyes are located close together analysis (Fukumura et al., Nucl. Acids. Res. 31 (16) e94, as they are on the probe. During the amplification reaction, 2003). the Taq DNA polymerase enzyme cleaves the probe in a Microarrays template-dependent manner. The resultant probe fragments 5 Evaluating gene expression of a melanoma sample can also disassociate in solution, and signal from the released reporter be performed with microarrays. Microarrays permit simulta dye is free from the quenching effect of the second fluoro neous analysis of a large number of gene expression products. phore. One molecule of reporter dye is liberated for each new Typically, polynucleotides of interest are plated, or arrayed, molecule synthesized, and detection of the unduenched on a microchip Substrate. The arrayed sequences are then reporter dye provides the basis for quantitative analysis. 10 hybridized with nucleic acids (e.g., DNA or RNA) from cells RT-PCR can be performed using commercially available or tissues of interest (e.g., cutaneous tissue samples). The equipment, such as an ABI PRISM 7700TM Sequence Detec source of mRNA typically is total RNA (e.g., total RNA tion System (Perkin-Elmer-Applied Biosystems, Foster City, isolated from human melanoma samples, and normal skin Calif., USA), or Lightcycler(R) (Roche Molecular Biochemi samples). If the source of mRNA is a primary tumor, mRNA cals, Mannheim, Germany). Samples can be analyzed using a 15 can be extracted, for example, from frozen or archived paraf real-time quantitative PCR device such as the ABI PRISM fin-embedded and fixed (e.g. formalin-fixed) tissue samples. 7700TM Sequence Detection SystemTM. In various embodiments of the microarray technique, To minimize errors and the effect of sample-to-sample probes to at least 10, 25, 50, 100, 200, 500, 1000, 1250, 1500, variation, RT-PCR is usually performed using an internal or 1600 genes (e.g., genes listed in a Table herein, which standard. A Suitable internal standard is expressed at a con- 20 distinguish metastatic melanoma from other types of cutane stant level among different tissues, and is unaffected by the ous tissues) are immobilized on an array Substrate (e.g., a experimental variable. RNAs frequently used to normalize porous or nonporous solid Support, such as a glass, plastic, or patterns of gene expression are mRNAS for the housekeeping gel surface). The probes can include DNA, RNA, copolymer genes glyceraldehyde-3-phosphate-dehydrogenase sequences of DNA and RNA, DNA and/or RNA analogues, or (GAPDH) and B-actin. 25 combinations thereof. A variation of the RT-PCR technique is real time quantita In some embodiments, a microarray includes a Support tive PCR, which measures PCR product accumulation with an ordered array of binding (e.g., hybridization) sites for through a dual-labeled fluorigenic probe (i.e., TaqManTM each individual gene. The microarrays can be addressable probe). Real time PCR is compatible both with quantitative arrays, and more preferably positionally addressable arrays, competitive PCR, where internal competitor for each target 30 i.e., each probe of the array is located at a known, predeter sequence is used for normalization, and with quantitative mined position on the Solid Support such that the identity (i.e., comparative PCR using a normalization gene contained the sequence) of each probe can be determined from its posi within the sample, or a housekeeping gene for RT-PCR. For tion in the array. further details see, e.g. Held et al., Genome Res. 6:986-994, Each probe on the microarray can be between 10-50,000 1996. 35 nucleotides, e.g., between 300-1,000 nucleotides in length. Gene expression can be examined using fixed, paraffin The probes of the microarray can consist of nucleotide embedded tissues as the RNA source. Briefly, in one exem sequences with lengths: less than 1,000 nucleotides, e.g., plary method, sections of paraffin-embedded tumor tissue sequences 10-1,000, or 10-500, or 10-200 nucleotides in samples are cut (~10 um thick). RNA is extracted, and protein length. An array can include positive control probes, e.g., and DNA are removed. After analysis of the RNA concentra- 40 probes known to be complementary and hybridizable to tion, RNA repair and/or amplification steps may be per sequences in the test sample, and negative control probes, formed, if necessary, and RNA is reverse transcribed using e.g., probes known to not be complementary and hybridizable gene specific promoters followed by RT-PCR. Methods of to sequences in the test sample. examining expression in fixed, paraffin-embedded tissues, Methods for attaching nucleic acids to a surface are known. are described, for example, in Godfrey et al., J.; Molec. Diagn. 45 Methods for immobilizing nucleic acids on glass are 2: 84-91, 2000; and Spechtet. al., Am. J. Pathol. 158: 419-29, described, e.g., Schena et al. Science 270:467-470, 1995: 2001. DeRisi etal, Nature Genetics 14:457-460, 1996; Shalonet al., Another approach for gene expression analysis employs Genome Res. 6:639-645, 1996; and Schena et al., Proc. Natl. competitive PCR design and automated, high-throughput Acad. Sci. U.S.A. 93:10539-11286, 1995). Techniques are matrix-assisted laser desorption ionization time-of-flight 50 known for producing arrays with thousands of oligonucle (MALDI-TOF) MS detection and quantification of oligo otides at defined locations using photolithographic tech nucleotides. This method is described by Ding and Cantor, niques are described by Fodor et al., 1991, Science 251:767 Proc. Natl. Acad. Sci. USA 100:3059-3064, 2003. 773, 1991; Pease et al., Proc. Natl. Acad. Sci. U.S.A. 91:5022 See also the Mass ARRAY-based gene expression profiling 5026, 1994; Lockhart et al., Nature Biotechnology 14:1675, method, developed by Sequenom, Inc. (San Diego, Calif.). 55 1996; U.S. Pat. Nos. 5,578,832; 5,556,752; and 5,510,270). Additional PCR-based techniques for gene expression Other methods for making microarrays have been described. analysis include, e.g., differential display (Liang and Pardee, See, e.g., Maskos and Southern, Nuc. Acids. Res. 20:1679-1, Science 257:967-971, 1992); amplified fragment length poly 684, 1992. In principle, and as noted Supra, any type of array, morphism (iAFLP) (Kawamoto et al., Genome Res. 12:1305 for example, dot blots on a nylon hybridization membrane 1312, 1999); BeadArrayTM technology (Illumina, San Diego, 60 (see Sambrook et al., Molecular Cloning, A Laboratory Calif.; Oliphant et al., Discovery of Markers for Disease Manual, 2nd Ed., Vols. 1-3, Cold Spring Harbor Laboratory, (Supplement to Biotechniques), June 2002; Ferguson et al., Cold Spring Harbor, N.Y. (1989)) could be used. Analytical Chemistry 72:5618, 2000); BeadsArray for Detec The polynucleotide molecules to be analyzed may be from tion of Gene Expression (BADGE), using the commercially any clinically relevant source, and are expressed RNA or a available Luminex 100 LabMAP system and multiple color- 65 nucleic acid derived therefrom (e.g., cDNA or amplified RNA coded microspheres (LumineX Corp., Austin,Tex.) in a rapid derived from cDNA that incorporates an RNA polymerase assay for gene expression (Yang et al., Genome Res. 11:1888 promoter), including naturally occurring nucleic acid mol US 7,615,349 B2 67 68 ecules, as well as synthetic nucleic acid molecules. For enzyme Such as horse radish peroxidase or alkaline phos example, the test polynucleotide molecules include total cel phatase. Alternatively, unlabeled primary antibody is used in lular RNA, poly(A)+ messenger RNA (mRNA), or fraction conjunction with a labeled secondary antibody, comprising thereof, cytoplasmic mRNA, or RNA transcribed from cDNA antisera, polyclonal antisera or a monoclonal antibody spe (i.e., cFNA; see, e.g., Linsley & Schelter, U.S. patent appli cific for the primary antibody. Immunohistochemistry proto cation Ser. No. 09/411,074, filed Oct. 4, 1999, or U.S. Pat. No. 5,545,522, 5,891,636, or 5.716,785). Methods for preparing cols and kits are well known in the art and are commercially RNA are known and are described, e.g., in Sambrook et al., available. Molecular Cloning, A Laboratory Manual (2" Ed.), Vols. Proteomic methods can allow examination of global 1-3, Cold Spring Harbor Laboratory, Cold Spring Harbor, 10 changes in protein expression in a sample. Proteomic analysis N.Y., 1989. RNA can be fragmented by methods known in the typically involves separation of individual proteins in a art, e.g., by incubation with ZnCl2, to generate fragments of sample by 2-D gel electrophoresis (2-D PAGE), and identifi RNA. cation of individual proteins recovered from the gel, e.g. my Test polynucleotide molecules that are poorly expressed in mass spectrometry or N-terminal sequencing, and analysis of particular cells can be enriched using normalization tech 15 niques (Bonaldo et al., Genome Res. 6:791-806, 1996). the data using bioinformatics. Proteomics methods can be The test polynucleotides are detectably labeled at one or used alone or in combination with other methods for evalu more nucleotides. Any method known in the art may be used ating gene expression. to detectably label the polynucleotides. In various aspects, the expression of certain genes in a Nucleic acid hybridization and wash conditions are chosen cutaneous sample is detected to provide clinical information so that the test polynucleotide molecules specifically bind or (e.g., prognostic information, classification of the tumor from specifically hybridize to the complementary polynucleotide which the sample is derived as a metastatic melanoma or sequences of the array, preferably to a specific array site, non-metastatic melanoma). Thus, gene expression assays wherein its complementary nucleic acid is located. General include measures to correct for differences in RNA variability parameters for specific (i.e., stringent) hybridization condi 25 and quality. For example, an assay typically measures and tions for nucleic acids are described in Sambrook et al., Supra, incorporates the expression of certain normalizing genes, and in Ausubel et al., Current Protocols in Molecular Biol Such known housekeeping genes, e.g., GAPDH, B-actin, and ogy, Vol. 2, Current Protocols Publishing, New York, 1994. Cyp1. Alternatively, normalization can be based on the mean Typically, stringent conditions for short probes (e.g., 10 to 50 or median signal (Ct) of all of the assayed genes or a large nucleotide bases) will be those in which the salt concentration 30 is at least about 0.01 to 1.0 M at pH 7.0 to 8.3 and the Subset thereof (global normalization approach). In some temperature is at least about 30° C. Stringent conditions can embodiments, a normalized test RNA (e.g., from a patient also be achieved with the addition of destabilizing agents sample) is compared to the amount found in a metastatic such as formamide. When fluorescently labeled probes are melanoma, non-metastatic melanoma, and/or normal skin used, the fluorescence emissions at each site of a microarray 35 sample reference set. The level of expression measured in a can be detected by Scanning confocal laser microscopy or particular test sample can be determined to fall at Some per other methods (see Shalon et al., Genome Research 6:639 centile within a range observed in reference sets. 645, 1996: Schena et al., Genome Res. 6:639-645, 1996; and Kits Ferguson et al., Nature Biotech. 14:1681-1684, 1996). Sig The technology herein includes kits for evaluating gene nals are recorded and typically analyzed by computer. Meth 40 expression (e.g., RNA or protein) in melanoma samples. A ods for evaluating microarray data and classifying samples "kit' refers to a combination of physical elements, e.g., are described in U.S. Pat. No. 7,171,311. Serial Analysis of Gene Expression (SAGE) probes, including without limitation specific primers, labeled Gene expression in melanoma samples can also be deter nucleotic acid probes, antibodies, protein-capture agent(s), mined by serial analysis of gene expression (SAGE), which is 45 reagent(s), instruction sheet(s) and other elements useful to a method that allows the simultaneous and quantitative analy practice the technology described herein. These physical ele sis of a large number: of gene transcripts, without the need of ments can be arranged in any way Suitable for carrying out the providing an individual hybridization probe for each tran invention. script (see, e.g. Velculescu et al., Science. 270:484-487, 1995; A kit for analyzing protein expression can include specific and Velculescu et al., Cell 88:243-51, 1997). Briefly, a short 50 binding agents, such as immunological reagents (e.g., an sequence tag (about 10-14 nucleotides) is generated that con antibody, e.g., a labeled antibody) for detecting proteins tains Sufficient information to uniquely identify a transcript, expressed of one or more genes described herein (e.g., one or provided that the tag is obtained from a unique position within more genes from Table A, Table B, Table C, or Table D). For each transcript. Then, many transcripts are linked together to example, the kit can include an antibody that detects expres form long serial molecules, that can be sequenced, revealing 55 sion of GJB6, an antibody that detects expression of the identity of the multiple tags simultaneously. The expres SPPRR1A, and an antibody that detects expression of SER sion pattern of a population of transcripts can be quantita tively evaluated by determining the abundance of individual PINB5, in a tissue section. tags, and identifying the gene corresponding to each tag. Kits for analyzing RNA expression include, for example, a Protein Detection Methodologies 60 set of oligonucleotide probes for detecting expression of a set Immunohistochemical methods are also suitable for of genes described herein (e.g., five or more genes from Table detecting the expression of the melanoma signature genes A, Table B, Table C, or Table D). The probes can be provided described herein. Antibodies, most preferably monoclonal on a solid Support, as in an array (e.g., a microarray), or in antibodies, specific for a gene product are used to detect separate containers. The kits can include a set of oligonucle expression. The antibodies can be detected by direct labeling 65 otide primers useful for amplifying a set of genes described of the antibodies themselves, for example, with radioactive herein, e.g., to perform PCR analysis. Kits can include further labels, fluorescent labels, hapten labels such as, biotin, or an buffers, enzymes, labeling compounds, and the like. US 7,615,349 B2 69 70 EXAMPLES melanoma-associated tumor antigens. (MAGE, CSAG2), genes implicated in melanoma progression (GDF15. To identify the genes involved in the metastatic process of MMP14, SPP-1), cell cycle progression (CDK2, TYMS, melanoma, various non-metastatic primary skin cancers were BUB1), and the prevention of apoptosis (BIRC5, BCL2A1). compared to metastatic melanoma utilizing a gene microar These changes may reflect the higher growth capacity of the ray approach followed by functional validation of select metastatic tumors. Conversely, among the 997 genes with genes. Distinct gene expression changes occurring along the reduced expression in MM samples, many are implicated in spectrum of primary melanoma tumor thickness and meta keratinocyte differentiation and epidermal development, static melanoma were discovered. such as loricrin (LOR), involucrin (IVL), and keratin-5 10 (KRT5), Suggesting a loss of epidermal characteristics. These Example 1 expression changes suggest important comparative differ Gene Expression Differences Between Primary ences between non-metastatic and metastatic tumors. Cutaneous Cancer and Metastatic Melanoma Analysis of the functional classes of genes changed using 15 revealed that 15 genes associated with kerati Tumor samples were obtained from patients with primary nocyte differentiation and 32 genes involved in epidermis cutaneous melanoma (PCM), squamous cell carcinoma development were down-regulated in the metastatic samples (SCC), basal cell carcinoma (BCC) and metastatic melanoma (FIG. 1A). These losses were complemented by the increased (MM). Gene expression in the samples was examined by expression of genes involved in several cellular processes, microarray analysis as described in the Materials and Meth Such as DNA repair, protein transport, melanocyte differen ods, below. An initial training set of 23 tumors revealed 2,014 tiation, muscle development, nervous: System development Affymetrix probe sets with a greater than 2-fold difference in and carbohydrate metabolism. Table 1 further illustrates that the average gene expression level between the metastatic the magnitude of change in those genes under-expressed was melanoma (MM) and primary cutaneous cancers. This pre much greater on average than the level of change in over liminary list, consisting of 1,141 well characterized and 471 25 expressed genes. Overall, the losses in gene expression are poorly characterized human genes, indicates that a Substan tial difference exists between the metastatic tumors and the both greater in number and magnitude compared to the gains non-metastatic tissue types. The expression differences allow in gene expression in MM samples. for a relatively robust gene classification of tissue samples Another observation is that there are a larger number of into groups of metastatic samples and non-metastatic primary 30 genes with reduced expression in the metastatic tumors and tumors. All tumor samples were clustered utilizing the 2014 the degree of decrease is much greater. In otherwords, the loss probe sets and individually identified as metastatic or non of gene expression is greater than the gain of new gene metastatic based upon the characteristics of tumor samples in expression. This is consistent with the observation of dedif the same cluster. The initial set of samples comprised a train ferentiation which is believed to occur with the development of cancer. ing set for which 22 of 23 samples were correctly partitioned 35 into the cluster containing primary melanoma or the cluster The initial statistical analysis of microarray samples of containing MM samples. A single primary melanoma with a metastatic melanoma and non-metastatic cutaneous tumors Breslow's tumor thickness of 90 mm was misclassified as a leads to the conclusion that a fundamental difference exists MM sample. Two independent test sets comprised of primary between tumors containing metastatic potential and tumors and MM samples were similarly classified, utilizing the 2014 40 probe sets and hierarchical clustering. Co-clustering led to the without demonstrated metastatic potential. In the case of correct identification of 56 of 60 melanoma samples. In gen melanoma, it would appear the metastatic potential is associ eral, the misidentified samples were thick primary melano ated with a large number of changes in gene expression and mas classified as MM. Of note, several normal human skin fundamental changes in the spectrum of genes expressed. samples were analyzed and found to classify as non-meta 45 static by their gene expression profiles. Any measurement of this programmatic shift in gene expres A Subset of melanoma samples were examined in order to sion would be useful for the identification of metastatic mela generate a more comprehensive list of genes that were differ noma cells within a primary melanoma tumor. The data pre entially expressed between MM and PCM using serial analy sented in Table 1 addressed the question of whether there sis of microarrays (SAM). This analysis identified 1,352 50 were genes differentially expressed (increased or decreased) probe sets with higher expression in the metastatic samples and 2,991 probesets with higher expression in non-metastatic between primary (BCC/SCC/PCM) and metastatic cancers samples. This list was further reduced by removing probesets (metastatic melanoma). The full names of each gene (for that did not appear to have an average difference greater than named genes), gene symbol, accession number and gene 2-fold between groups. The resultant complete gene list is 55 identification for all genes>2-fold up- or down-regulated in shown above in Tables A and B, above. This final list consists metastatic melanoma are provided in Tables A and B above. of 1,667 Affymetrix probe sets that detect 247 poorly defined transcripts, 84 minimally defined genes, and 1007 well char FIG.1B is a heatmap depicting the relative gene expression of acterized human genes. From this list, 316 genes were highly the full spectrum of genes described herein in the non-meta expressed in MM compared to 1022 genes that were more 60 static tumors relative to the metastatic tumors. Darker gray highly expressed in the non-metastatic: cancers and normal shading corresponds to higher than average gene expression, skin. black corresponds to average gene expression, and light gray A subset of the full gene list is shown below in Table 1 correspond to lower than average gene expression. FIG. 1B below. This table illustrates two main trends. There is a shift in the kind of genes expressed, perhaps related to the funda 65 demonstrates that the combination of genes discovered herein mental characteristics of the cells comprising the tumors. For dramatically separates metastatic tumors from non-meta example, there is higher expression levels in MM for several static tumors. US 7,615,349 B2 71 72 trating how gene expression differences could distinguish TABLE 1. between these two classes of tissues. The transition Zone also illustrates the class of tumors for which a metastatic assess Differential gene expression between metastatic melanoma and non-metastatic cutaneous tumors ment is particularly useful. A transition appears to occur as primary melanoma tumors thicken. The thicker primary Increase in Decrease in tumors appear to contain gene expression patterns identical to metastatic metastatic melanoma melanoma the metastatic melanoma tumors. relative to relative to The apparent transition Zone of gene expression change Gene MIS BCCSCC Gene MIS, BCCSCC. 10 could represent a critical time period where many tumori symbol Thin MIS/Thin symbol Thin MISThin genic events occur or may simply reflect the outgrowth of an aggressive and/or metastatic cell phenotype. To address this MAGEA3 125 27 CALMLS 93 228 MAGEA1 24 13 DSC1 86 98 issue, a comparative analysis of gene expression patterns in MAGEA6 119 25 DSC3 64 71 primary melanomas of different Breslows thickness was per MAGEA2 57 31 PKP1 66 240 15 formed. PCM and MMsamples were compared to elucidate a MAGEA12 57 29 CLCA2 62 77 MAGEAS O 6 DSG3 60 19 possible relationship between relative gene expression pat CSAG2 76 36 DSG1 60 78 terns associated with PCM of increasing Breslow's thickness TRIMS1 51 35 LY6D 43 47 and that of MM samples. Table 2 (left columns) reveals the GDF15 30 47 SERPINB3 11 84 relative change in gene expression for a Subset of genes GYPC 8 14 SERPINB5 99 227 SPP1 5 7 SERPINB7 10 44 throughout the spectrum of primary melanoma tumors to MM KIFC1 5 3 C19Crf33 22 35 samples. RGS2O 4 14 FLG 12 12 C1orf-O 3 15 KRTS 49 62 Several genes, such as the MAGE genes, exhibited a steady BCL2A1 2 12 KRT17 11 96 and consistent increase in gene expression over the entire SOX5 5 8 KRT16 05 274 SLC16A4 2 29 KLK7 99 12 25 range of tumor thicknesses. However, a single major shift in AKT3 1 9 KLK8 23 32 expression was observed for most genes when thinner pri PEG10 1 10 KLK10 27 81 mary tumors were directly compared to thicker ones. This BUB1 4 3 KLK11 6 83 RASGRF1 8 12 LOR 95 98 was most apparent when comparing I.M. thickness to thick MMP14 8 6 LGALS7 84 89 PCM, with the majority of genes showing the greatest SPRED1 6 4 CST6 82 56 30 GPR19 6 5 TRIM29 79 119 increase in gene expression. Notable exceptions were genes CDK2 6 7 SFN 77 125 such as SPP1, HOXA10 and MMP14, for which the greatest HOXA10 3 4 ASAH3 69 56 HOXB6 4 6 GATA3 63 S4 differential increase in expression was at the comparative HOXB7 5 7 CBLC 60 64 interface between thin and I.M. thickness tumor samples. HOXB9 3 3 RAB2S 59 78 35 HE 7 16 S100A14 27 44 Other genes, such as MMP19, CTH, PDGFRL, C16orf.34 DUSP4 8 10 S100A7 57 60 and GPR19, showed the greatest comparative increase in DUSP6 8 6 S100A7L1 21 82 expression when comparing MIS to thin PCM lesions. CDC4SL 7 8 ICEBERG 52 48 CDC6 9 4 IVL 50 76 A similar phenomenon was observed for genes with RRM2 6 4 ELOVL4 38 34 40 decreased expression in primary tumors relative to more TYMS 4 3 CXCL14 36 37 advanced lesions (Table 2, right columns). Here, however, the BIRCS 4 2 FOXN1 33 34 largest proportion of the gene expression change occurred Abbreviations: between thick PCM and MM samples. Very little expression PCM, primary cutaneous melanoma, MIS, melanoma-in-situ, of (6B, 16, 17) and SPRR1 (A, B) was observed in BCC, basal cell carcinoma, 45 MM compared to all primary melanomas, including thick SCC, squamous cell carcinoma, lesions. Several genes, such as TMPRSS4, STAR, ST7L, Thin, thin melanomas <1.5 mm in Breslow's thickness. HAS3, FGFR3, CASZ1 and HR, were found to have gene expression changes at the very earliest stages of tumor thick Example 2 ening. Together, the gene expression patterns do not shift in a 50 coordinated fashion as would be expected as the result of the Identification of Gene Expression Differences outgrowth of a clonal aggressive or metastatic cell type. Between Metastatic and Non-Metastatic Tumor Rather a series of events may occur as PCM tumors thicken Samples that may influence the expression of different groups of genes ultimately leading to the fully metastatic cell type. This data The relative gene expression levels of 177 genes across the 55 spectrum of tissue samples examined is shown graphically in indicates that some gene expression changes may be indica FIG.1C. This analysis reveals a consistent level of expression tive of earlier events in the progression to full metastasis. The through all of the presumed "non-metastatic” samples (nor indicated changes in gene expression may signal that cells in mal skin, BCC, SCC, MIS. Thin). A marked change in the primary melanoma tumors are progressing to a fully meta gene expression levels is seen, beginning with the I.M. thick 60 static-state or that have already acquired the metastatic state. ness PCM (average Breslow's tumor thickness of 2.1 mm), In either case, the genes are useful markers for identifying progressively increasing or decreasing to the expression level aggressive tumors which warrant more aggressive treatment. representative of MM lesions. All of the thick PCM (average All annotated genes listed in Table 2 with a “-2 indicates Breslows thickness 19 mm) exhibited gene expression pat that any difference between tumors for each comparative terns similar to those of MM samples and daughter MM cell 65 analysis was less than 2-fold. Underlined numbers indicate lines. FIG.1C demonstrates the sharp contrast between meta the greatest change in gene expression across varying PCM static tumors and non-metastatic tumors or normal skin illus tumor thickness for each gene. US 7,615,349 B2 73 74 TABLE 2 Comparative analysis of gene expression changes in primary and metastatic melanoma. Relative increase in gene expression Relative decrease in gene expression Gene MIS to Thin IM to Thick MIS to Gene MIS to Thin IM to Thick MIS to Symbol Thin To IM Thick to Met Met Symbol Thin To IM Thick to Met Met MAGEA2 <2 8.6 9.4 3.1 30.7 SPRR1A <2 <2 <2 129.4 239.3 MAGEA3 <2 8.1 10.7 3 83.8 SPRR1B <2 <2 <2 52.8 100.8 MAGEA6 <2 7.1 9.8 2.7 103 KRT16 <2 <2 <2 68.3 195.5 MAGEA1, 2 <2 2.5 8.9 2.6 81.S KRT17 <2 2.6 <2 27.4 57.6 MAGE A1 <2 <2 11.1 <2 25.4 KRT6B <2 <2 <2 39.7 100.9 MAGE AS <2 2.2 3.9 <2 11.7 AQP3 <2 <2 3.2 6.1 43.2 MMP19 5.4 <2 <2 <2 7.4 CD24 <2 <2 2 4.4 18.7 PDGFRL 22.3 3.6 &2 <2 6.3 FLG <2 <2 3 24.1 140 C16orf54 5.4 <2 <2 <2 18.5 IVL <2 2.2 <2 13.9 84.1 CTH 3.9 <2 <2 <2 8.8 KLK7 <2 <2 6.5 8.9 128 GPR19 4.2 <2 3 <2 23.9 LGALS7 <2 <2 3.4 17.3 109.3 SPPI <2 13.1 <2 <2 44.9 LOR <2 <2 2.4 22.6 120.4 HOXA10 <2 3.3 <2 <2 3.8 RAB2S <2 <2 3.3 10.3 88.4 MMP14 <2 3.6 <2 2.1 9 SFN <2 <2 <2 10.6 24.2 AKT3 <2 <2 7.5 <2 4.3 C19Crf33 3.1 <2 4.6 13.3 220.3 BCL2A1 <2 2.2 4.8 <2 8.7 ASAH3 <2 <2 25.9 <2 60 BIRCS <2 <2 3.7 <2 3.4 KRT1S <2 <2 27.1 2.2 104.9 BUB1 <2 <2 9.4 <2 O.8 ELOVL4 <2 2.3 14.7 <2 41.9 CDC4SL <2 <2 13 <2 9.1 GATA3 2.5 <2 14.3 <2 23.5 CDK2 <2 <2 4.8 <2 8.4 MUC15 <2 2.2 11.9 <2 25 CSAG2 <2 <2 19.6 2.6 S4S SCEL <2 <2 28.8 <2 71.8 DUSP4 2 <2 5.7 <2 2.5 TP73L <2 <2 5.5 3.4 41.2 DUSP6 <2 <2 3.7 <2 O.S RORA <2 <2 6.2 2.2 26.6 GYPC <2 <2 12.2 <2 4.2 POU2F3 <2 <2 13.9 2.2 73.9 HEY1 <2 <2 5.7 <2 9.3 ICEBERG <2 2.7 6.7 4.5 32.3 KIFC1 <2 <2 10.4 <2 6 CASZ1 <2 4.8 2.8 <2 12.3 PEG10 <2 2.4 4.1 <2 1.3 HR <2 3.7 <2 <2 7.7 RASGRF1 <2 <2 5.1 <2 9.9 TMPRSS4 8.7 <2 <2 3.1 42 RGS2O 3.4 <2 9.2 <2 343 STAR 1.9 <2 <2 <2 11.1 SLC16A4 <2 <2 4.4 <2 26.4 ST7L 4.3 <2 3.6 <2 8.4 SOX5 <2 <2 12.3 <2 9.8 LTB4R 4. <2 2.3 <2 9.9 TRIMS1 <2 3.1 15.9 <2 63.6 HAS3 4.9 <2 <2 2.5 16.6 TYMS <2 <2 3.6 <2 4.7 FGFR3 3.9 <2 3.2 <2 7.6

Abbreviations: MIS, melanoma-in-situ, Thin, thin melanomas <1.0 mm in Breslows thickness, I.M., intermediate thickness between 1–4 mm, with thick melanomas >4 mm. 40 Example 3 for both comparisons, in particular for Such genes as KRT14. GJA1, S100A7(A9) and EHF. Other genes, like the mela Comparative Genomic Analysis of Normal Human noma associated antigens, MAGE A2 and TRAG and Epithelial Melanocytes to Primary and Metastatic PRAME were also found to be highly over-expressed in Melanoma Samples 45 NHEM to early primary or MM samples. Similarly, a marked Gene expression profiles of cultured NHEM were com decrease in gene expression was observed for several genes, pared to PCM and MM samples (Table 3), acknowledging the although of a lesser magnitude than seen for the over-ex inherent limitations associated with the comparisons of cul pressed genes. Several unique genes including PAEP. HES6, tured cells and freshly procured tumor samples. Large differ 50 ESDN, NR4A3, c6orf168 and BCL2A1, were under-ex ences in gene expression were observed between NHEM and pressed in NHEM compared to thin PCM. Other genes were early, non-metastatic PCM (MIS/thin lesions only) and MM also identified as under-expressed in both groups, such as samples. Concordant over-expression of genes were found CITED-1, GDF15, QPRT, OCA2, c-MET and MME. TABLE 3 Comparative gene expression levels of normal human epithelial melanocytes to thin primary cutaneous and metastatic melanoma samples Comparative Increase Comparative Decrease in Gene Expression (Fold Change in Gene Expression (Fold Change NHEMCft NHEMCft Gene NHEM cft Gene Metastatic Gene NHEMCft Metastatic Symbol MIS/Thin Symbol Melanoma Symbol MIS/Thin Gene Symbol Melanoma KRT14 6787 GA1 759 MME 106 MAP4 2O GA1 5929 SEPP1 338 CITED1 77 OCA2 10 EHF S487 KRT14 306 GDF15 64 TRIM7 7 US 7,615,349 B2 75 76 TABLE 3-continued Comparative gene expression levels of normal human epithelial melanocytes to thin primary cutaneous and metastatic melanoma samples Comparative Increase Comparative Decrease in Gene Expression (Fold Change in Gene Expression (Fold Change NHEMCft NHEMCft Gene NHEM cft Gene Metastatic Gene NHEMCft Metastatic Symbol MIS/Thin Symbol Melanoma Symbol MIS/Thin Gene Symbol Melanoma SCEL 3931 MAGEA2 301 PAEP 47 CITED1 6 CLCA2 3689 TRAG3 242 RPEL1 45 TRPM4 6 S100A7 3609 EHF 193 HES6 43 MME 5 KRTDAP 3416 S100A9 160 ESDN 37 TRAP1SO 5 DSC1 2782 S100A7 129 QPRT 35 FER1L3 4 GB6 2576 SCEL 126 OCA2 19 QPRT 4 CXCL14 2484 SLC22A3 125 RENBP 17 KLF8 4 LOR 2308 EPHA3 124 NR4A3 16 RPEL1 4 KRT6A 1989 KRTDAP 121 Statfc 16 PACE4 4 PKP1 1835 S1 OOA8 120 C6orf168 15 HPCAL1 4 SERPINB3 1778 ZIC1 119 BCL2A1 14 ACTR1A 4 S100A9 1768 CXCL14 118 NTT 13 14 MET 3 KRT1S 1545 IL18 108 PSCD3 14 RAB32 3 GATA3 1347 PRAME 102 HPCAL1 13 TYR 3 PPL 1339 MAGEA6 94.7 MET 12 RF6 3 IMUP 12SO PLCB4 88 ALS2CR3 12 GDF15 3 ICEBERG 1123 CLCA2 87 PTPLA 12 HINT3 3 KRT6B 1114 GB6 85 TBC1D7 11 SLC30A1 3 CSTA 693 MAGEA3 84 OA1 11 TRPV2 3 CST6 687 MMP19 59 TYR 11 RSN 3

Example 4 tate and pancreatic cancer. Several of the putative melanoma 30 TSGs have also been previously shown to contribute to the Identification of Putative Oncogenes and Tumor development and progression of cancer in other tumor his Suppressor Genes in Melanoma tologies. The shifts in gene expression occur at different stages of A perusal of the gene expression differences between PCM the thickening process for each of the oncogenes and TSGs and MM samples identifies numerous putative oncogenes and 35 tumor suppressor genes (TSG). Table 4 lists several onco listed in Table 4. Some of the genes show a progressive and genes and TSG previously implicated in tumor types. The steady increase or decrease in gene expression as tumors of gene with the largest increase in expression (13.2 fold) was greater thickness are compared. But for others, such as the SPP-1 or osteopontin. Although not previously identified as oncogenes SPP-1 and GDF15, and the TSGs PITX-1 and an oncogene, osteopontin expression has been shown to cor 40 CST6, the major shifts in gene expression appear to occur at relate with melanoma invasion and tumor progression (Zhou distinct but different times during the thickening of the pri et al., J. Invest. Dermatol. 124:1044-1052, 2005). The lin mary melanoma tumors. This observation strongly suggests eage-specific oncogene, MITF. previously shown to act as a that these changes may occur spontaneously but eventually master regulator of melanocyte development and a critical accumulate to contribute to the final metastatic phenotype. Survival oncogene amplified in melanoma showed a 3.7 fold 45 Table 4 contains a partial list of identified tumor oncogenes increase (Garraway et al., Nature. 436:33-35, 2005; Levy et and tumor suppressor genes (TSG's) in PCM and MM al., Trends Mol. Med. 12:406-14, 2006; McGill et al., J. Biol. samples. The fold increase represents the greatest fold change Chem. 281:10365-10373, 2006). Of the other genes, GDF15, noted throughout all comparisons of each PCM tumor thick c-Met and the HOX loci have been shown to act as possible ness to MM. The activating/Suppressive mechanism and oncogenes in breast cancer, Squamous cell lung cancer, pros affected tumor type are also identified.

TABLE 4 Differential expression of putative tumor oncogenes and Suppressor genes in melanoma Oncogenes Fold Interval of Activating Mechanisms in other Gene Increase Increase Tumor Histologies Affected Tumor Types

SPP-1 13.2 Thin to IM C-Met activation via Civ3 Breast, HCC, Prostate, receptor, Inhibition of apoptosis CRC, Head & Neck MITF 3.7 Progressive Somatic alteration via gene None, Lineage Specific increase amplification (Chri3p13-3p14) for Melanoma CITED-1 12.4 IM to Thick Activation of Stat-3, Ras MAPK Thyroid (cbp/p300 kinase signaling via Ets1, Ets2 transactivator) US 7,615,349 B2 77 78

TABLE 4-continued Differential expression of putative tumor oncogenes and Suppressor genes in melanoma GDF15 22.7 IM to Thick Lineage specific activation or Breast, CRC, Gastric, (PLAB) repression of ERK1/2; Integrator of Prostate, Pancreatic AKT pathway c-Met 14.5 Thick to Met Ras-Associated Protein CRC, Breast, Ovarian, (Rap1)/ERK/MAPK, rac1, Grb2, Pancreatic, Liver P13K, Src activation HOXLocus 2.1-5.0 Progressive Downstream activation of WT-1, AML, Breast, SCLC (A3, A10, B6, increase NFKB, NR4A3, BCI.2, p53 B7, B13) Tumor Suppressor Genes Fold Interval of Suppressor Mechanisms in other Gene Decrease Decrease Tumors Histologies Affected Tumor Types PITX-1 13.9 Thin to IM Ras Pathway (RASAL1) Barrett's Esophagus Prostate, Bladder CST6 (CST 66.7 IM to Thick Hypermethylation Breast, Glioma E/M) PDGFRL 7.3 IM to Thick Gene Deletion from Chri HCC, CRC, NSCLC 8p21.3-p22 DSC3 42.8 Progressive Hypermethylation Breast decrease POU2F3 49 Thin to IM Hypermethylation Cervical CLCA2 162 MIS/Thin to MM Hypermethylation Breast

Abbreviations: HCC, hepatocellular carcinoma, CRC, colorectal carcinoma, NSCLC, non-Small cell lung cancer, SCLC, squamous cell lung cancer, AML, acute myelogenous leukemia.

Example 5 evaluating metastatic and non-metastatic phenotypes. Those genes are included in Tables A and B, above. Validation of Select Candidate Genes by Semi- and 35 FIG. 3B depicts the expression level of several oncogenes Quantitative RT-PCR Analysis an TSGs cell lines derived from normal human melanocytes, normal skin, and cell lines derived from tumor samples. In To further validate the expression of putative TSG and general, oncogenes are more highly expressed in the cell lines oncogenes in our melanoma panel, real-time quantitative 40 derived from tumors and TSGs are more highly expressed: in polymerase chain reaction (RT-qPCR) assays were per the normal skin and normal human melanocytes consistent formed on 20 previously arrayed samples, comprised of 7 with their role as possible markers of the metastatic state. PCM and 13 mM samples. The results are depicted in FIG. These measurements were made with semi-quantitative PCR, 2A, which shows an overall decreased level of mRNA expres demonstrating that alternative measures of gene expression sion of TSG and increased mRNA expression of oncogenes 45 can distinguish the metastatic signature. compared to normal skin, which was used to calibrate the expression levels to a value of Zero for graphing purposes. Example 6 FIG. 2A also shows that the MM samples expressed decreased levels of TSGs and increased levels of oncogenes Functional Analysis of Select Candidate Genes by relative to the PCM samples. This was consistent for all PCM 50 Western Blot Analysis and Immunohistochemistry samples compared to MM, although not statistically signifi cant for comparisons across different thicknesses of PCM. To independently verify and validate the gene expression FIG. 2B shows the original microarray data normalized and changes at the protein level, protein expression of several graphed in a manner identical to the PCR data. A significant Suspected oncogenes and TSG was examined using Western correlation for all TSG and oncogenes examined was found. 55 Blot analysis. Osteopontin (SPP-1) protein expression was Utilizing semi-quantitative PCR analysis, several primary examined, both from melanoma cell lysates and conditioned and MM: daughter cell lines derived from the freshly pro cell free media derived from 2 primary and 6 MM daughter cured melanoma samples, normal skin and NHEM, were cell lines (FIG.3B, a, b). Interestingly, the protein expression examined for oncogene and TSG mRNA expression. The level of 2 subcutaneous melanoma nodules (MCC 12A, 12F) results are depicted in FIG. 3A. Several of the cell lines 60 procured from the same patient with MM differed. Similarly, exhibited high levels of expression for several of the reported two paired cell lines (MCC80a from a primary melanoma oncogenes while there was a much higher percentage of loss from a synchronous metastatic lymph node, MCC80b), not of TSG expression seen in most cell lines examined. Overall, ing a slight increase in SPP-1 protein expression in the latter. a favorable correlation was observed between the microarray Several other melanoma cell lines exhibited minimal SPP-1 results and both quantitative and semi-quantitative PCR 65 protein expression. (MCC12F, 66C, 80a and 89). Similar analysis for daughter and non-daughter primary and MM cell findings were noted between melanoma cell lysates and con lines. A subset of the genes listed in FIG. 3A are useful for ditioned cell media. US 7,615,349 B2 79 80 Analysis of suspected TSG in 3 primary and 3 metastatic brain). These MM samples were compared with 42 primary melanoma cell lines revealed a very low level of protein cutaneous cancers (16 PCM, 11 SCC, 15 BCC). PCM con expression of DSC3 in 6/6 cell lines (FIG. 3B, panel c) with sisted of 2 melanoma in situ (MIS), 2 thin melanomas (<1 4/6 (2/3 primary and 2/3 MM) cell lines expressing the pro mm), 3 intermediate-thickness melanomas (1-4 mm), and 9 tein for CLCA2 (FIG. 3B, panel d). Interestingly, PDGFRL thick melanomas (>4 mm). Additionally, 4 Samples of normal protein expression was observed in a single primary cell line, human skin and 1 sample of cultured NHEM were included. with no evidence of expression in any of the metastatic cell All MM samples were procured from patients that had failed lines (FIG.3B, panele). The cellular staining patterns of skin multiple previous therapies, ranging from single agent Inter and melanoma samples are available for viewing at the feron, single or multi-agent chemotherapy, immunotherapy Human Protein Atlas website (world wide web address at 10 or other experimental treatment options. All primary cutane proteinatlas.org/). These observations demonstrate that ous cancers were procured from previously untreated analysis of protein expression (e.g., antibody-based assays) patients. can be used as measures of gene expression for the identifi RNA isolation, purification and hybridization. A portion of cation of metastatic melanoma in primary cutaneous tumors. each cryopreserved tissue sample was dissolved in TRIZolR) The molecular analyses described herein clearly identifies 15 (Invitrogen, Carlsbad, Calif.), purified according to manufac distinct molecular profiles associated with MM which are turer's recommendations, and further purified on RNeasy different from PCM, SCC, and BCC as well as normal mel columns (Qiagen Inc., Valencia, Calif.). RNA integrity was anocytes and skin. Using the refined gene list (Tables A and verified by both gel electrophoresis and the Agilent 2100 B), the metastatic character of tumors (SCC/BCC/PCM/MM) Bioanalyzer (Agilent Technologies, Palo Alto, Calif.). A total can be classified correctly greater than 90% of the time. One of 5ug of RNA was processed using established Affymetrix major difficulty is classification of thick primary melanoma protocols for the generation of biotin-labeled cRNA and the tumors, where occasionally, these tumors appeared to have hybridization, staining, and scanning of arrays as outlined in the gene expression signature of MM. It was discovered that the Affymetrix technical manuals (Van Gelder et al., Proc. these tumors represent primary tumors which have already Natl. Acad. Sci. U.S.A. 87:1663-1667, 1990; Warrington et acquired the metastatic gene expression pattern. Detection 25 al., Physiol. Genomics. 2: 143-147, 2000). The processed methods that examine expression of the gene combinations RNA was hybridized to U133 Plus 2.0 arrays described herein allow identification of such tumors and from Affymetrix, Inc. (Santa Clara, Calif.), and Scanned on an inform Subsequent clinical decisions. Affymetrix GeneChip(R) scanner 3000 at 2.5um resolution. A In conclusion, a clear pattern of gene expression change more complete description of this process is available in was observed in the non-metastatic and metastatic samples 30 Dobbin et al., Clin. Can. Res. 11:565-572, 2005. The tissue examined. There is a clear point of transition in gene expres samples were processed in three independent groups. sion when comparing I.M. to: thick PCM, revealing specific Cell lines and tissue culture. Freshly excised melanoma groupings of genes involved in this process. Several of these samples were placed into culture media (RPMI 1640+5% genes and combinations thereofhave never before revealed as FCS) and tissue procurement and expansion of daughter cell functional or relevant in melanoma. The specific genes 35 lines was established utilizing previously published tech involved in this dynamic and fluid change in gene expression niques (Rikeret al., Can Detect and Prev,23(5):387-96, 1999; provides the basis for the determination of whether a thin, Riker AI. The isolation and culture of melanoma cell lines. In: I.M., or thick PCM has the genetic capability to metastasize Langdon S. editor. Cancer cell culture. Methods and proto and facilitate the development of an appropriate treatment cols. Totowa: Humana Press; pp. 93-100, 2004). All cell lines Strategy. 40 were split and passaged-10 times and, characterized by flow cytometry and/or cytospin preparation for cellular confirma Example 7 tion of melanoma cell purity (data not shown). The cell lines, TC077 and TC80a were derived from primary melanoma Materials and Methods samples with TC80b derived from a metastatic lymph node 45 (from the same patient). The cell lines, TC12A and TC12F, Tumor specimens. Tumor samples were Surgically pro were derived from 2 different subcutaneous melanoma nod cured from patients with primary cutaneous melanoma ules from: the same patient. There were 3 cell lines examined (PCM), squamous cell carcinoma (SCC), basal cell carci from metastatic samples, TC66C, TC72 and TC89. The noma (BCC) and metastatic melanoma (MM) over a 3 year NHEM were cultured according to the manufacturer direc period. All samples were obtained under an Investigational 50 tions. (Cambrex BioScience, Walkersville, Md.). Review Board (IRB) approved tissue procurement protocol Semi and real-time quantitative RT-PCR. First-strand (MCC#13448, IRBi 101751: PSMH990914-JM, cDNA synthesis was performed using Superscript III RT 020318-JM). Upon surgical removal of the primary mela (Invitrogen). Subsequently, the cDNA was used in semi noma, a single Surgical oncologist (A.I.R.) utilized a scalpel quantitative PCR. Each sample was normalized with B-Actin to macrodissect and procure a portion of the remaining pri 55 as an internal control, comparing each sample with mary tumor, with a similar technique utilized for grossly Alphalase(RFC image analysis software (Alpha Innotech, involved lymph nodes where the melanoma had completely San Leandro, Calif.), followed by densitometric analysis of replaced the lymph node. Samples were taken from non the integrated values for each sample. The expression levels necrotic areas of the tumor. The same process was performed of putative oncogenes and tumor Suppressor genes were ana for all distant metastases, with care taken to avoid Surround 60 lyzed by real-time quantitative RT-PCR (qPCR) using ing tissues and stroma. Assays-on-Demand. Gene Expression Assays (Applied Bio All samples were cryopreserved in liquid nitrogen and systems, Foster City, Calif.): SPP1 (osteoponin, assay ID stored within the Tissue Procurement Laboratory of the Mof Hs00167093 m1), GDF15 (growth differentiation factor 15, fitt Cancer Center, securely de-identified through a central assay ID Hs00171132 ml), PITX1 paired-like homeodomain ized database. Forty MM samples were analyzed, composed 65 transcription factor 1, assay ID Hs00267528 ml), DSC3 (des of 22 bulky, macroscopic (replaced) lymph node metastases, mocollin3, assay ID Hs00170032 ml), CST6 (cystatin E/M, 16 Subcutaneous and 2 solid organ metastases (adrenal and assay IDHs00154599), POU2F3 (POU domain, class 2, tran US 7,615,349 B2 81 82 scription factor 3, assay ID Hs00205009) and GAPDH (assay samples, this provides good Statistical confidence but does IDHs99999905 ml) as the internal standard. Utilizing nor not focus on the differences between primary melanoma and mal skin as the calibrator, the relative quantitation values of a metastatic melanoma. A second comparison was performed target template for each sample were expressed as 2^^. utilizing 6 thin primary melanoma samples in opposition to 6 Briefly, qPCR analysis was performed utilizing 40 ng of total cDNA in a 25 ul reaction volume (Applied Biosystems). randomly selected metastatic melanomas. The only non-ran QPCR was performed utilizing established techniques, with dom aspects of this sample selection were to avoid selecting all samples performed in triplicate and run on an ABI/PRISM samples in which the classifier disagreed with the patholo 7500 Sequence Detector System (Applied Biosystems). gists diagnosis and to avoid utilizing more than one sample Gene microarray analysis and bioinformatics. Affymetrix 10 from the same individual. For this comparison the median MAS 5.0 analysis Software was used to generate signal values false discovery rate threshold was set at 5%. This latter analy for all probe sets based upon a mean intensity of 500, subse sis is the preferred grouping of Samples, but because of the quently exported and iteratively normalized as a whole group Small sample size it is also more likely to generate false to create the final normalization based upon the most stable discoveries due to noise and outlier samples. Therefore the gene expression measurements across all samples (Li et al., 15 more confident gene list generated by combining the two Proc. Natl. Acad. Sci. U.S.A. 98:31-36, 2001). This process was performed for the initial group of tumor samples to analyses. The intersection of the two approaches yielded generate the list of normalization probesets that were Subse 1.352 probe sets with higher expression in the metastatic quently used to Scale all samples processed for this study to an samples and 2.991 probe sets with higher expression in non average intensity of 4000 for the normalization probesets. metastatic samples. This list was further reduced by removing Following scaling, the calculated signal values were then probe sets that did not appear to have a difference greater than used to calculate the average expression level for each gene in 2-fold on average between the two groups. each tissue type using an initial group of 23 tumor samples. Following all microarray analyses, the identified probe sets The average expression values derived from this initial set were annotated based on the sequence of the probes used on were directly compared to identify genes expressed at high 25 levels in one tumor type but not in the other samples using a the arrays (Harbig et al., Nucleic Acids Res. 33:e31, 2005). t-test and visual inspection to find highly differential expres Western blot analysis. Whole cell extracts from PCM and sion patterns. Genes highly expressed in metastatic melano MM cell lines were prepared by directly lysing cells in SDS mas but not primary melanomas, basal cell carcinomas, or sample buffer. Expression of SPP-1 protein was assessed in squamous cell carcinomas, were sought. Several genes were 30 cell lysate and serum-free conditioned medium. Briefly, initially selected that exhibited the idealized gene expression profiles. Additional candidate genes were then identified by 4x10° cells were plated in 5% FBS containing medium; 24 using Pearson’s correlation between the idealized gene hours later, the growth medium was replaced with serum-free expression patterns and all other probe sets on the arrays. medium. The conditioned media and cell lysates were har Positively correlated (re-0.7) and negatively correlated (r-0.7) 35 vested 24 hours later and resolved using a 12.5% SDS-PAGE. genes were identified and trimmed to include only those with Proteins were transferred to a PVDF membrane and probed a 2-fold or greater difference in the average gene expression with the anti-human SPP-1 mouse monoclonal antibody level between metastatic samples and non-metastatic tumors. This initial gene expression survey identified 2014 Affyme (Sigma, St. Louis, Mo.) (1:1000) followed by a secondary trix probe sets from the U133 Plus 2.0 arrays that showed 40 antibody conjugated to horseradish peroxidase (Amersham differential expression between metastatic tumor samples Biosciences, Piscataway, N.J.) and detected using chemilu and non-metastatic tumor samples. minescence (Santacruz, Biotechnology, Santa Cruz, Calif.). The 2014 probe sets identified as correlating with the meta The osteopontin band (SPP-1) was visualized at ~55-65kDa. static phenotype were used to cluster the samples. Following Daughter melanoma cell lines derived from the freshly pro normalization, as described above, the signal values were log 45 cured melanoma samples (with the exception of A375) were 2 transformed. Each probe set was then mean centered across all samples and the resulting values were input into Eisen's lysed by M-PERTM Mammalian Protein Extraction Reagent cluster. Hierarchical clustering was performed using absolute (Pierce, Rockford, Ill.) and processed according to manufac correlation and a complete linkage. Clustering was performed turer instructions. A total of 15 Jug of protein from each with various subgroups of the data or with all samples 50 experimental condition were electrophoresed on 10% SDS together and resulted in similar sample groupings. Individual PAGE and transferred to nitrocellulose membranes (Bio-Rad, samples were classified based on the class of the other Hercules, Calif.). Immunostaining was performed with the samples in the closest cluster. The complete microarray data following primary antibodies: DSC3 (Santa Cruz) 1:200; is available from the Gene Expression Omnibus (world wide web address: ncbi.nlm.nih.gov/geo/) under Accession num 55 CLCA2 (Novus Biologicals, Littleton, Colo.) 1:500; PDG ber GSET553. FRL (Novus Biologicals) 1:500; C-tubulin (Cell signaling, Serial Analysis of Microarrays (SAM) was performed in Danvers, Mass.), 1:1000. Immunocomplexes were visualized order to identify a more extensive list of differentially using an enhanced chemiluminescence (ECL) Western Blot expressed genes expressed between PCM and MM. Two ting Substrate (Pierce). The intensity of the bands were comparisons were made to generate a comprehensive and yet 60 scanned with a Fujifilm intelligent dark box II and analyzed confident list of genes that are differentially expressed between metastatic melanoma and non-metastatic melano with Fujifilm Las-1000 Lite V1.3 software. mas. In the first comparison, the metastatic melanoma A number of embodiments of the technology have been samples were opposed by all the non-metastatic samples described. Nevertheless, it will be understood that various including basal and squamous cell carcinoma and normal 65 modifications may be made without departing from the spirit skin. The false discovery rate threshold used to limit the gene and scope of the technology. Accordingly, other embodi list was 0% for this comparison. Because of the number of ments are within the scope of the following claims.

US 7,615,349 B2 93 94

- Continued aaaaaaaaaa aaaaaa. 1396

<210 SEQ ID NO 6 <211 LENGTH: 1560 &212> TYPE: DNA <213> ORGANISM: Homo sapiens <4 OO SEQUENCE: 6 CtcCctgttgt tdtggagga tigtctgcagc agcatttaaa ttctgggagg gcttggttgt 6 O

Cagcagcagc aggaggaggc agagcacagc atcgt.cggga ccagacticgt. Ctcaggc.ca.g 12 O ttgcagcctt ct cago caaa cqc.cgaccaa gcaaaactica citaccatgag aattgcagtg 18O atttgcttitt gcct coltagg cat cacctgt gccataccag ttaaacaggc tigattctgga 24 O agttctgagg aaaagcagaa totgttgtcc tictgaagaaa C caatgactt taaacaagag 3OO accct tccaa gtaagt ccaa caaa.gc.cat gacca catgg atgatatgga tigatgaagat 360 gatgatgacc atgtggacag C caggactico attgactica acgactctga tigatgtagat 42O gacact gatg attct cacca gtctgatgag tot caccatt ctdatgaatc tdatgaactg 48O gtcactgatt titcc.cacgga cct gccagca accogaagttt toacticcagt tdtcc ccaca 54 O gtag acacat atgatggc.cg aggtgatagt gtggttt atg gactgaggt C aaaatctaag 6OO aagttt cqca gacctgacat coagtaccct gatgctacag acgaggacat cacct cacac 660 atggaaag.cg aggagttgaa tigtgcatac aaggc catcc ccgttgcc.ca ggacctgaac 72 O gcgc ctitctg attgggacag cc.gtgggaag gacagtt atgaaacgagtica gctggatgac 78O cagagtgctgaaacccacag C caca agcag ticcagattat ataag.cggala agccaatgat 84 O gaga.gcaatg agcatt CC9a ttgattgat agt caggaac titt CCaaagt cagcc.gtgaa 9 OO titccacagcc atgaattitca cagc.catgaa gatatgctgg ttgtag accc caaaagtaag 96.O gaagaagata aac acctgaa attitcg tatt tot catgaat tagatagtgc atc.ttctgag O2O gtcaattaaa aggagaaaaa atacaattitc. tcactittgcatttagt caaa agaaaaaatg O8O ctittatagca aaatgaaaga gaa catgaaa tdottcttitc. tcagtttatt ggttgaatgt 14 O gtat ct attt gag totggaa ataactaatg tdtttgataa ttagtttagt ttgtggcttic 2OO atggaaactic cct gtaaact aaaagct tca gggittatgtc. tatgttcatt ctatagaaga 26 O aatgcaaact at cactgt at tittaatattt gttattotct catgaataga aattitatgta 32O gaagcaaaca aaatacttitt acc cacttaa aaa.gagaata taacatttta tdt cactata 38O atcttttgtt ttittaagtta gtgtatattt tdttgttgatt atcttitttgt gigtgtgaata 44 O aatcttittat cittgaatgta ataagaattt ggtggtgtca attgct tatt tdttitt.ccca SOO cggttgtcca gcaattaata aaa cataa.cc tttitt tactg. cctaaaaaaa aaaaaaaaaa 560

55 What is claimed is: cyte activation 1) (SPP1) (SEQID NO:6)inatest sample 1. A method of evaluating a melanoma from a patient, the comprising melanoma cells from the patient, and method comprising: comparing said mRNA levels to a set of reference levels determining mRNA levels of five or more genes selected that represent levels of the mRNAs in a sample compris from the group consisting of tripartite motif-containing 60 ing cells from a non-metastatic melanoma, 51 (TRIM51) (SEQID NO:1); Glycophorin C (Gerbich wherein an increase in expression of said five or more blood group) (GYPC) (SEQID NO:2); SRY (sex deter genes indicates an increased likelihood that the mela mining regionY)-box 5 (SOX5) (SEQID NO:3); kinesin noma is a metastatic melanoma. family member C1 (KIFC1) (SEQ ID NO:4); SEC22 2. The method of claim 1, wherein miRNA levels of all six vesicle trafficking protein-like 3 (S. cerevisiae) 65 genes are determined. (SEC22L3) (SEQID NO:5); and secreted phosphopro 3. The method of claim 1, wherein mRNA levels of the five tein 1 (osteopontin, bonesialoprotein I, early T-lympho or more genes is determined relative to mRNA levels of the US 7,615,349 B2 95 96 five or more genes in a reference set of non-metastatic cuta 5. The method of claim 1, wherein determining mRNA neous tissue samples, and wherein an increase in expression levels of five or more genes in the test sample comprises of said five or more genes, relative to expression of the five or isolating RNA from the test sample, and detecting expression more genes in the reference set, indicates an increased like of the RNA. lihood that the melanoma is a metastatic melanoma. 6. The method of claim 5, wherein determining mRNA 4. The method of claim 1, wherein mRNA levels of the five levels of five or more genes in the test sample comprises or more genes is compared to: (a) mRNA levels in a first performing reverse transcriptase polymerase chain reaction. reference set of non-metastatic cutaneous tissue samples, and 7. The method of claim 5, wherein determining mRNA (b) mRNA levels in a second reference set of metastatic levels of five or more genes in the test sample comprises melanoma tissue samples; wherein a greater similarity in 10 performing microarray analysis. mRNA levels of the five or more genes in the test sample to 8. The method of claim 1, wherein the test sample is a test the second reference set than to the first reference set indicates sample from a primary melanoma. an increased likelihood that the melanoma is a metastatic melanoma. k k k k k UNITED STATES PATENT AND TRADEMARK OFFICE Certificate Patent No. 7,615,349 B2 Patented: November 10, 2009 On petition ES issuance of a certificate for correction of inventorship pursuant to 35 U.S.C. 256, it has been found that the above identified patent, through error and without any deceptive intent, improperly sets forth theAccordingly, inventorship. it is hereby certified that the correct inventorship of spent is: Adam I. Riker, Mobile, AL (US); Steven Alan Enkemann, Lutz, FL (US); and Jaime Matta, Ponce (PR). Signed and Sealed this Fourth Day of January 2011. MISOOKYU Supervisory Patent Examiner Art Unit 1642 Technology Center 600