Purpose: Advanced melanoma is a highly drug-refractory neoplasm representing a significant unmet medical need.We sought to identify melanoma-associated cell surface molecules and to develop aswell as preclinically test immunotherapeutic reagents designed to exploit such targets. Experimental Design and Results: By transcript profiling, we identified glycoprotein NMB (GPNMB) as a gene that is expressed by most metastatic melanoma samples examined. GPNMB is predicted to be a transmembrane protein, thus making it a potential immunotherapeutic target in the treatment of this disease. A fully human monoclonal antibody, designated CR011, was generated to the extracellular domain of GPNMB and characterized for growth-inhibitory activity against melanoma.The CR011monoclonal antibody showed surface staining of most melanoma cell lines by flow cytometry and reacted with a majority of metastatic melanoma specimens by immunohistochemistry. CR011alone didnot inhibit the growthofmelanoma cells. However,when linked to the cytotoxic agent monomethylauristatin E (MMAE) to generate the CR011-vcMMAE antibody-drug conjugate, this reagent now potently and specifically inhibited the growth of GPNMB-positivemelanoma cells in vitro. Ectopic overexpression and small interfering RNAtransfection studies showed that GPNMBexpression is both necessary and sufficient for sensitivity to low concentrations of CR011-vcMMAE. In a melanoma xenograft model, CR011-vcMMAE induced significant dose-proportional antitumor effects, including complete regressions, at doses as low as1.25 mg/kg. Conclusion:These preclinical results support the continued evaluation of CR011-vcMMAE for the treatment of melanoma. Melanoma is a common neoplasm and its incidence is increasing worldwide at a dramatic rate (1). Melanoma accounts for only 4% of skin cancer cases yet causes f79% of all skin cancer deaths. In 2004, an estimated 55,100 Americans were diagnosed with melanoma and f7,910 would die of the disease (2). An increasing frequency of newly diagnosed melanomas, ranging from 3% to 8% annually, has also been observed worldwide (3, 4). Therapeutic options for patients with late-stage melanoma presenting with regional and/or distant metastases are limited. Dacarbazine is the only cytotoxic drug currently approved by the Food and Drug Administration for the treatment of stage IV metastatic melanoma, with a response rate of <15% and a median response duration of 4 to 5 months (5). The majority of polychemotherapy regimens failed to show significant survival benefits (6), nor did the use of adjuvant therapeutic agents such as IFN-a and interleukin 2, which pose severe toxicity (7, 8). The poor efficacy and adverse side effects of available therapies has led to a considerable interest in the development of alternative therapies, such as monoclonal antibodies (mAb), for the treatment of metastatic melanoma (9). Recent advances in genetic engineering have significantly decreased antibody immunogenicity and increased antibody half-life (10, 11). Antibody-based therapeutics, such as Rituxan, Herceptin, and Avastin, have recently enjoyed clinical success in the treatment of some hematopoietic malignancies and solid tumors. Although antibodies that target tumor or its vasculature may be useful in an unconjugated form, it is sometimes advantageous to couple a tumor-targeting antibody to an isotope (e.g., Zevalin and Bexxar) or to a cytotoxic compound (e.g., Mylotarg). This strategy allows for the selective delivery of cytotoxic agents to the tumor with the goal of reducing the toxicity that is often associated with the systemic administration of cytotoxic agents while preserving or enhancing the antitumor activity of these agents. Our genome-wide transcript expression profiling, coupled with a systems biology analysis of human melanoma clinical Cancer Therapy: Preclinical Authors’ Affiliations: CuraGen, Branford, Connecticut; Abgenix, Fremont, California; and Seattle Genetics, Bothell,Washington Received 9/16/05; revised12/9/05; accepted12/14/05. The costs of publication of this article were defrayed in part by the payment of page charges.This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section1734 solely to indicate this fact. Note: K.F.Tse andM. Jeffers contributed equally to this work. Requests for reprints:William J. LaRochelle, CuraGen Corporation, 322 East Main Street, Branford, CT 06405. Phone: 203-871-4288; Fax: 203-315-3301; E-mail: wlarochelle@curagen.com. F2006 American Association for Cancer Research. doi:10.1158/1078-0432.CCR-05-2018 www.aacrjournals.org Clin Cancer Res 2006;12(4) February15, 2006 1373 Research. on May 29, 2017. © 2006 American Association for Cancer clincancerres.aacrjournals.org Downloaded from specimens and cell lines, led to the identification of a tumorassociated protein, called glycoprotein NMB (GPNMB), as a potential target that can be exploited for the treatment of melanoma. GPNMB is predicted to be a 560-amino-acid type I transmembrane protein with closest homology (26% amino acid identity) to the melanocyte/melanoma–specific protein, pMEL17 (12). The normal function of human GPNMB is unknown, and orthologues have been isolated from mouse (DC-HIL; ref. 13), rat (Osteoactivin; ref. 14), and quail (QNR-71; ref. 15). Previous investigations have associated GPNMB expression and function with cancer. GPNMB was first identified as a gene that was differentially expressed among melanoma cell lines with high and low metastatic potential (12) and was subsequently identified as a candidate glioma tumor marker due to its high transcript expression in this tumor type and restricted normal tissue distribution (16). GPNMB expression has also been described in liver cancer, squamous cell lung carcinoma, and soft tissue tumors (17–19). Moreover, ectopic expression of GPNMB in cancer cells increased their in vitro invasiveness and promoted their metastasis in vivo (17, 20). Finally, GPNMB was shown to interact with the surface of endothelial cells (13), a finding that may have implications for GPNMB-expressing melanoma cell transendothelial migration and metastasis. To explore the potential utility of GPNMB as a target for melanoma therapy, fully-human mAbs were generated to this protein. The lead mAb, CR011, was characterized and coupled to the dolastatin-10-related cytotoxic drug monomethylauristatin E (MMAE), a potent inhibitor of mitotic spindle formation (21). The resulting antibody-drug conjugate, designated CR011-vcMMAE, was evaluated for growth-inhibitory activity on melanoma cell lines in vitro and for activity against melanoma xenografts in vivo . The results presented in this study suggest that GPNMB represents a promising target for the identification and treatment of advanced melanoma and that CR011-vcMMAE is worthy of continued therapeutic evaluation. Materials andMethods Cell lines and transfections. M14, UACC-257, and LOXIMVI cell lines were obtained from the National Cancer Institute (Bethesda, MD) and all others from the American Type Culture Collection (Manassas, VA). Cells were maintained in DMEM or RPMI containing 10% fetal bovine serum and penicillin-streptomycin. To establish stable cell lines overexpressing GPNMB, HEK293 cells were transfected with either control vector (pcDNA3.1-V5-His) or this vector containing full-length GPNMB, using LipofectAMINE (Invitrogen, Carlsbad, CA) according to the protocol of the manufacturer. Following selection in medium containing G418 (0.8 mg/mL), individual clones were selected and propagated. Small interfering RNA (siRNA) was used to inhibit GPNMB expression in SK-Mel-2 cells. Cells were transfected with 50 nmol/L of siGENOME SMART pool reagents (Dharmacon, Inc., Chicago, IL), designed to specifically target GPNMB, or siRNA to thymidylate synthase as a negative control, using the OligofectAMINE transfection reagent (Invitrogen) following the instructions of the manufacturer. Reverse transcription-PCR and real-time quantitative PCR. Total RNA was isolated using the RNeasy kit with a DNase digestion step (Qiagen, Inc., Valencia CA). Reverse transcription-PCR (RT-PCR) was done using the OneStep RT-PCR kit (Qiagen) as follows. Reverse transcription: 50jC for 45 minutes and 95jC for 15 minutes for one cycle. PCR: 1 minute at 95jC, 1 minute at 50jC, and 2 minutes at 72jC for 30 cycles with final extension for 10 minutes at 72jC. Products were separated on a 2% agarose/0.33% low melting point agarose gel and visualized by ethidium bromide staining. The integrity of each RNA sample was verified via RT-PCR with primers designed to amplify glyceraldehyde-3-phosphate dehydrogenase. The primers used for amplification are as follows (5V-3V): Real-time quantitative PCR analysis was done with an ABI Prism 7700 Sequence Detection System using TaqMan reagents (PE Applied Biosystems, Foster City, CA). Equal quantities of normalized RNAs were used as a template in PCR reactions for 40 cycles with GPNMB-specific primers to obtain threshold cycle (CT) values. The primers used for amplification are as follows (5V-3V): Forward-TCAATGGAACCTTCAGCCTTA Reverse-GAAGGGGTGGGTTTTGAAG Probe-TET-CTCACTGTGAAAGCTGCAGCACCAG-TAMRA Production and purification of recombinant human GPNMB extracellular domain protein. Oligonucleotide primers were designed to amplify the cDNA encoding the GPNMB extracellular domain (GPNMB-ECD) using a human fetal brain cDNA template. The forward primer included an in-frame BamHI site and the reverse primer contained an in-frame SalI restriction site. The primers used for amplification are as follows (5V-3V): Forward-GGATCCAAACGATTTCATGATGTGCTGGGCAATGAA Reverse-GTCGACCGAGGCTGGGTCTCTGTCAGGAACAGAAAT The PCR product was cloned into the pCR2.1-Topo vector (Invitrogen). The cDNA insert was verified by sequencing and subcloned into the BamHI/XhoI sites of pCEP4 (Invitrogen), which was modified by inserting the murine Ign secretion signal upstream, and a V5-His tag downs

The 3;8 chromosomal translocation, t(3;8) (p14.2;q24.1), was described in a family with classical features of hereditary renal cell carcinoma,]Previous studies demonstrated that the 3p14.2 breakpoint interrupts the fragile histidine triad gene (FHIT) in its 5' noncoding region. However, evidence that FHIT is causally related to renal or other malignancies is controversial. We now show that the 8q24.1 breakpoint region encodes a 664-aa multiple membrane spanning protein, TRC8, with similarity to the hereditary basal cell carcinoma/segment polarity gene, patched. This similarity involves two regions of patched, the putative sterol-sensing domain and the second extracellular loop that participates in the binding of sonic hedgehog, In the 3;8 translocation, TRC8 is fused to FHIT and is disrupted within the sterol-sensing domain. In contrast, the FHIT coding region is maintained and expressed. In a series of sporadic renal carcinomas, an acquired TRC8 mutation was identified. By analogy to patched, TRC8 might function as a signaling receptor and other pathway members, to be defined, are mutation candidates in malignant diseases involving the kidney and thyroid.

Loss of heterozygosity (LOH) involving 3p occurs in many carcinomas but is complicated by the identification of four distinct homozygous deletion regions. One putative target, 3p14.2, contains the common fragile site, FRA3B, a hereditary renal carcinoma-associated 3;8 translocation and the candidate tumor suppressor gene, FHIT. Using a approximately 300 kb comsid/lambda contig, we identified homozygous deletions in cervix, breast, lung and colorectal carcinoma cell lines. The smallest deletion (CC19) was shown not to involve FHIT coding exons and no DNA sequence alterations were present in the transcript. We also detected discontinuous deletions as well as deletions in non-tumor DNAs, suggesting that FHIT is not a selective target. Further, we demonstrate that some reported FHIT aberrations represent normal splicing variation. DNA sequence analysis of 110 kb demonstrated that the region is high in A-T content, LINEs and MER repeats, whereas Alu elements are reduced. We note an intriguing similarity in repeat sequence composition between FRA3B and a 152 kb segment from the Fragile-X region. We also identified similarity between a FRA3B segment and a small polydispersed circular DNA. In contrast to the selective loss of a tumor suppressor gene, we propose an alternative hypothesis, that some putative targets including FRA3B may undergo loss as a consequence of genomic instability. This instability is not due to DNA mismatch repair deficiency, but may correlate in part with p53 inactivation.

By applying the 'recognition mask' strategy to 300 mammalian sequences containing NotI sites we demonstrated that 5' ends of genes are highly enriched in NotI sites. A NotI linking clone NL2-252 (D3S1678) containing transferrin receptor (TFRC) gene was used as an initial point for chromosomal jumping. One of the jumping clones, J21-045 traverses 210 kbp and links NL2-252 to NL26 (D3S1632), a NotI linking clone containing highly polymorphic sequences. The TFRC gene was mapped to 3q29, close to the telomeric marker D3S2344, by linkage analysis, a panel of hybrid cell lines, GeneBridge 4 panel and FISH. Clone NLM-007 (D3S4302) was found to contain ras-homologous gene RAB7. By FISH and a panel of hybrid cell lines this gene was mapped to 3q21. This region is of particular interest due to frequent rearrangements in different types of leukemia. Clone L2-081 (D3S4283) containing new member of ubiquitin-specific proteases (HAUSP gene) was localized in 3p21 inspiring further investigation of involvement of this gene in development of lung and renal carcinomas.

Loss of chromosome 3p is a critical event in the pathogenesis of lung cancer. Overlapping homozygous 3p21.3 deletions in lung cancer cell lines involving GNAI2 were characterized and found to involve a region of genomic instability. A new widely expressed Semaphorin, H.SemaIV, was isolated from the GNAI2 deletion region. Reduced H.SemaIV expression allowed identification of additional cell lines with submicroscopic or larger deletions of the locus which occurred in a heterogeneous manner. We also demonstrate the presence of a distinct 3p21.3 homozygous deletion region, adjacent to the DNA mismatch repair gene, hMLH1, and identified deletions in direct tumors. This appears to represent one of the first demonstrations of homozygous deletions affecting 3p in direct lung tumors.

Despite several lines of evidence suggesting that common chromosomal fragile sites are biologically important as hot spots for recombination, their structure remains unknown. We showed previously that the plasmid pSV2neo preferentially integrates into bands containing fragile sites in cells transfected under conditions of fragile site induction. Here we report the isolation and characterization of the DNA sequences from two such independent integrations into 3p14.2, a common fragile site (FRA3B). These FRA3B region sequences were shown to lie within a 1330-kb YAC, 850A6, approximately 350 kb telomeric of the breakpoint of t(3;8), a constitutional rearrangement. The two integration sites are 10 kb apart, but each integration is associated with a deletion. We have constructed a partial genomic contig of the integration sites and deleted regions spanning approximately 85 kb. Analysis of the DNA sequences immediately surrounding the plasmid integrations revealed no known coding sequences or repeat structures resembling the (CGG)nmotif characteristic of the rare fragile sites. In addition, by Southern blotting analysis, none of the phage clones isolated from the FRA3B region were found to contain CGG repeats. Fluorescencein situhybridization analysis of genomic clones from this contig to metaphase cells induced to express breaks demonstrated hybridization adjoining the chromosome breaks, and occasionally the hybridization signal spanned the break. The results imply that breakage occurs at variable positions within a large region (at least on the order of 85 kb). Together, these data suggest that the structure of FRA3B differs from that of rare fragile sites.

To identify by positional cloning the putative tumor-suppressor genes on the short arm of human chromosome 3 that are involved in the initiation or progression of several human malignancies, we have developed human mouse somatic cell hybrids and NotI linking libraries. The somatic cell hybrids contain either the intact human chromosome 3 or its derivatives as the only human genetic material in rodent background. The somatic cell hybrid panel defines five chromosomal regions on the short arm and two chromosomal regions on the long arm of chromosome 3. Two hundred sixty-one NotI linking probes from three independently constructed linking libraries were assigned to the seven chromosomal regions. The somatic cell hybrid panel and the regionally localized NotI linking probes should facilitate the construction of genetic linkage and physical maps to identify various tumor-suppressor and disease-related genes not only on the chromosome 3p, but on the entire chromosome.

Human chromosome band 3p14 contains two tightly linked cytogenetic markers of broad interest, FRA3B and the t(3;8) breakpoint associated with hereditary renal cell carcinoma (RCC). The common fragile site at 3p14.2 (FRA3B) is the most sensitive site on normal human chromosomes to breakage when DNA replication is perturbed by aphidicolin or folate stress. The t(3;8)(p14.2;q24.1) translocation segregates with RCC in a large family and could mark the location of a tumor suppressor gene involved in renal cancers. In studies aimed at positional cloning of FRA3B and the t(3;8) breakpoint, we have used multicolor fluorescence in situ hybridization analysis (FISH) on metaphase spreads and interphase nuclei to order 14 yeast artificial chromosomes (YACs) in 3p14. The YACs used in this study were identified by a group of unordered lambda clones that had been previously localized to the 3p14 region and mapped proximal or distal to the t(3;8) breakpoint. FISH analysis was used to order the YACs and to map them in relation both to the t(3;8) translocation breakpoint and to FRA3B induced on normal chromosomes by treatment with aphidicolin. YACs that closely flanked both the t(3;8) translocation breakpoint and the fragile site were identified. A YAC walk from the closest distal YAC allowed the identification of a 1.3-Mb YAC derived from the CEPH large insert YAC library that spans both the FRA3B and the t(3;8) breakpoint. The order of the YACs and cytogenetic landmarks in 3p14 is cen-(126E1/230B9)-181H6-B15-D20F4-258B7-++ +280D2-70E12-168A8- 403B2-143C5-413C6-468B10-[850A6/t(3;8)/ FRA3B]-74B2.(ABSTRACT TRUNCATED AT 250 WORDS)

Loss or inactivation of a gene on the short arm of chromosome 3 may contribute to the genesis of renal cell carcinoma. A gene that corresponds to the most frequently lost RFLP site (D3F15S2) is expressed in a variety of human tissues, and at a particularly high level in the kidney. Its expression is markedly reduced in renal cell carcinoma. A database search showed that the gene product is closely related to or identical with acylpeptide hydrolase. The nucleotide identity between the rat acylpeptide hydrolase and the human gene at D3F15S2 is 88%, compatible with normal species differences. It is therefore likely that the human gene product is acylpeptide hydrolase. The renal cell carcinoma is then associated with a decrease of acylpeptide hydrolase activity. The gene may represent a tumor suppressor gene, whose loss contributes to the development of renal cell carcinoma. It might be speculated that it could act e.g. by affecting the activity of a small acetylated growth factor. Alternatively, its decreased expression may merely reflect the impairment of differentiation in RCC, compared to normal kidney. Loss of a linked but irrelevant gene by the 3p deletion is another possibility.

Deletions of the short arm of chromosome 3 and associated allele losses have been reported in the majority of sporadic renal cell carcinomas (RCC). On the basis of the combined cytogenetic and molecular data, it is reasonable to assume that a putative RCC locus, which contributes to tumor development by its loss, is located telomerically of the D3F15S2 site. Using H3E4, a D3F15S2-specific probe, we have isolated a cDNA clone (cl.4-2), and a sequence comparison revealed that the cDNA clone corresponds to the human acyl-peptide hydrolase gene. The gene is fairly universally expressed, but in RCC biopsies its expression is severely reduced, compared to the normal kidney. Cl.4-2 was used for in situ hybridization on metaphase chromosomes prepared from an Epstein-Barr virus (EBV) transformed lymphoblastoid cell line, derived from a t(3;8) (p14.2;q24.1) carrying member of the RCC family described by Cohen et al. in 1979 (N Engl J Med: 301:592-595). Carriers of this translocation regularly develop RCC by middle age. We now report that D3F15S2 is localized on the telomeric side of the constitutional breakpoint, in 3p21. The region of 3p affected by this familial translocation is thus not identical with the region of 3p most frequently deleted in sporadic RCC.