FIELD OF THE INVENTION

The present invention relates to a bispecific antibody, that specifically binds to both 90K and one of Endosialin and HER-3, a production method therefor, and novel use thereof in the treatment of hyperproliferative diseases such as tumors and metastases thereof. In particular, the invention relates to the use of bispecific antibodies able to inhibit the adhesive processes of tumor cells and the angiogenesis in tumors such as breast cancer, ovarian cancer, lung cancer, gastrointestinal cancer, melanoma, lymphoma and other tumors.

DESCRIPTION OF RELATED ART

Cancer is a disease characterized by a series of somatic changes affecting the structure and/or expression of oncogenes and tumor suppressor genes. It is well known that tumor growth beyond diameters of 1 -2 mm depends on formation of new blood vessels, a process known as angiogenesis, as well as on transformation of stromal fibroblasts and extracellular matrix proteins ¹ . In vitro and in vivo studies have demonstrated that tumor stroma and vasculature are characterized by a different expression of proteins and receptors if compared to the normal counterparts. Thereby, an approach to get better specificity to treat cancer or/and neoangiogenesis is the use of antibodies that can target specific antigens expressed in cancer or neo-endothelial cells or precursors that are not expressed or are expressed at a lower level on normal cells. These targets can be exploited using antibodies to kill antigen-bearing cells by inhibiting the biological activity of the antigen or by delivering immuno- or radio-conjugates that, when reach the antigen bearing cells, specifically kill these target cells.

Lectin galactoside-binding soluble 3 binding protein (LGALS3BP, also called 90k protein) is a heavily glycosylated secreted molecule that was identified in 1986 by lacobelli et al.

[2] in the culture medium of human breast cancer cells. Document WO 93/16180 characterizes the 90K antigen, and discloses the deposit number of the anti-90K mAb-SP- 2. The protein contains several carbohydrate chains and is organized in different functional domains [2-4] Ozaki et al. (Biochemical and Biophysical Research Communications, 317 (2004), pp. 1089-1095) discloses the identification of antigenic epitopes recognized by 90K/Mac-2 binding protein-specific cytotoxic T lymphocytes. 90K is detectable inside the cells from which it is secreted. Outside the cell, the protein is detectable in the ECM [5]. 90K is present in biological fluids, including blood, saliva, breast milk and tears, at a concentration of few microgram/ml [4, 6, 7]

Experimental evidences indicate that 90K plays a role in the adhesive processes of tumor cells. For example, the addition of a certain amount of human recombinant 90K to human melanoma cell line A375, maintained in a culture flask as unicellular suspension, determines an increase of cell-cell adhesion (named homotypic adhesion), leading to formation of multicellular aggregates [8]. This effect, which is due to the ability of 90K to bind residues of galectin-3 and galectin-1 harbored on the membrane of adjacent melanoma cells [8, 9], may be relevant during the metastatic spread of tumors. In fact, as mentioned above, tumor cells that detach from the primary tumor and enter blood vessels and/or lymphatic vessels can survive longer if they adhere all together, forming multicellular aggregates [1].

Studies have reported that 90K is able to specifically bind some ECM proteins, including collagen, fibronectin and laminin [5, 10]. Ulmer et al. (Journal of Cellular Biochemistry, 98:1351-1366 ) discloses that 90K/Mac-2 binding protein contributes to colon cancer progression by modulating tumor cell adhesion to extracellular proteins. In addition, different types of tumor cells start "spreading" as soon as they establish a contact with 90K, similarly to what is observed when cells adhere to the ECM protein laminin [5]. The cellular receptor of 90K responsible for adhesion and spreading has been identified, as beta-1 integrin [5].

Tumor cell adhesion to ECM proteins not only favors the processes of migration and cell diffusion, but also preserves cells against apoptosis (also called programmed cell death) caused by antiblastic drugs [1 1 , 12] Our group observed that when lymphoma tumor cells are maintained in a flask coated with 90K, a binding between the protein and the beta-1 integrin of lymphoma cells takes place and, as a consequence, cells become resistant to the action of antiblastic drugs, as indicated by the reduction of apoptotic cell rate [13]. The protective effect of 90K against antiblastic drug induced apoptosis can explain the poor response to chemotherapy and the reduced survival observed in patients affected by lymphoma with elevated blood levels of 90K [13-15].

International application WO 2010/097825 A1 discloses the use of an anti-90 K antibody i.e. monoclonal antibody SP-2, for prevention and treatment of tumors and their metastases. The anti-90K antibody SP-2 is able to recognize a conformational epitope between residues 107 and 435 of the amino acid sequence of the 90K protein. This document emphasizes that monoclonal antibody SP-2 inhibits the pro-adhesive function of 90K. The murine hybridoma cell line from which SP-2 is purified, was deposited by Stefano lacobelli at the DSMZ (Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH), Mascheroder Weg 1 B D-3300 Braunschweig, Germany under the Budapest Treaty, accession number DSM ACC2116, on Feb. 5, 1993, and at the C.N. CM. (Collection Nationale de Cultures de Microorganismes), Pasteur Institute of Paris, France, accession number 1-1083. SP-2 antibody produced by hybridoma DSM ACC 21 16. The DSM ACC 21 16 monoclonal antibody SP-2 is produced according to the procedures described by Kohler and Milstein, but it may be produced also according to the recombinant DNA technique, using the specific nucleotide sequence of SP-2 or a part thereof. SP-2 (MP-1959) is a murine monoclonal antibody recognizing 90K a glycoprotein secreted in large amounts by the majority of tumor cells, which plays an important role in cell-cell and cell-extracellular matrix adhesion and invasion. Recently, data have been presented that 90K functions critically as a pro-angiogenic factor through a dual mechanism, i.e by induction of tumor VEGF and stimulation of endothelial cell tubulogenesis, which is inhibited by SP-2 (Piccolo et al., J Mol Med 91 : 83-94, 2013).

All together, experimental and clinical data indicate that 90K, as a result of its ability to promote adhesion, plays a role in tumor growth and progression. For this reason, the manufacture of agents able to inhibit the pro-adhesive function of this protein is useful for prevention and/or treatment of cancer.

A further target for cancer therapy is the cell membrane protein Endosialin. Endosialin [16- 18], is a highly restricted 165-kDa cell surface glycoprotein expressed by tumor pericytes and fibroblasts in a broad range of human cancers but not detected in the respective cell types in many normal tissues. The Endosialin cDNA encodes a type I membrane protein of 757 amino acids with a predicted molecular mass of 80.9 kDa. Bioinformatic evaluation classifies Endosialin as a C-type lectin-like protein, composed of a signal leader peptide, five globular extracellular domains (including a C-type lectin domain, one domain with similarity to the Sushi/ccp/scr pattern, and three EGF repeats), followed by a mucin-like region, a transmembrane segment, and a short cytoplasmic tail. Carbohydrate analysis shows that the Endosialin core protein carries abundantly sialylated, O-linked oligosaccharides and is sensitive to O-sialoglycoprotein endopeptidase, placing it in the group of sialomucin-like molecules. Endosialin was demonstrated to interact with proteins of the extracellular matrix (Fibronectin, Collagen I) [19] mediating cell adhesion and migration; another important Endosialin interactor is the tumor secreted protein, 90K [20], a protein involved in cell adhesion and migration, acting also as a pro-angiogenic factor

[21].

The tumor vascular marker Endosialin/TEM1 is emerging as an attractive molecule for diagnostics and therapeutics because of its expression across the stroma of many human tumors, the low to absent expression in normal tissues, and accessibility from the vascular circulation. Smaller scFv constructs have also been reported for Endosialin targeting of drug-delivery vehicles [22] or diagnostics for fluorescence imaging techniques [23].

Endosialin is broadly expressed in human cancer [24] Its frequency, extent, and intensity vary among cancer subtypes as well as among individual tumors within subtypes. Endosialin was detected in almost all sarcoma suggesting that the protein is a very frequent feature of sarcoma. In sarcoma, Endosialin was detected in several cellular compartments including malignant sarcoma cells, stromal cells, and vasculature. Sarcoma subtypes with the greatest frequency, extent, and intensity of Endosialin expression and potentially the most promising therapeutic potential were synovial sarcoma, fibrosarcoma, malignant fibrous histiocytoma (MFH), liposarcoma, and osteosarcoma. In addition to sarcoma, high Endosialin expression rate was observed in vasculature of carcinomas, with bladder cancer emerging as an outstanding carcinoma subtype for Endosialin expression. The restriction of Endosialin expression in carcinomas to vasculature and stromal has implications for potential Endosialin-directed therapeutics, which could be expected to have an antiangiogenic or vascular-disrupting mechanism of action. In contrast, in sarcomas, an Endosialin-targeted therapeutic could have both a direct anticancer effect on malignant sarcoma cells, and an indirect anticancer effect due to antiangiogenic and/or vascular disrupting effects. Furthermore, for tumors expressing Endosialin directly by cancer cells, a diagnostic assay that measures the intensity of Endosialin expression in malignant tissues would assist in selecting patients that could benefit from an anti-Endosialin therapy. Thus, Endosialin holds potential value both as a biomarker for certain human cancers, like sarcoma [24-25] and as a targeted therapeutic agent.

Maia. et al [26] reported that the cytoplasmic domain of Endosialin is a key regulator of tumor growth and that tumor growth of mice lacking this domain are significantly reduced, if compared to the response in CD248WT/WT mice. In addition, they found that Endosialin present in fibroblasts expressing the cytoplasmatic domain of Endosialin also had impaired PDGF-BB-induced migration. Tomkowicz B et al [27] demonstrated that Endosialin mediates proliferation of primary human pericytes through a PDGF (platelet derived growth factor) receptor signaling pathway. Normal pericytes expressing high levels of Endosialin were able to proliferate, to respond to PDGF stimulation by phosphorylating both the PDGF receptor and the MAPK Erk1/2, and to induce the expression of the immediate early transcription factor c-Fos. In Endosialin knocked-down pericytes, PDGF-induced proliferation, Erk1/2 phosphorylation, and c-Fos expression were significantly impaired. These results indicated that Endosialin controls proliferation of human pericytes together with PDGF pathway and suggest that targeting this protein could represent a novel modality for mitigating tumor angiogenesis and suppressing cancer.

International application WO 2017/134234 describes the generation of an antibody that specifically recognizes and binds Endosialin. The antibody has the ability to become internalized in Endosialin expressing cells and to block the activation of MAPK in PDGF stimulated human pericytes. The antibody is able to block angiogenesis induced by 90K, a known Endosialin interactor and to inhibit tumor growth alone and in combination with 1959, a humanized antibody against 90K in human osteosarcoma xenograft. Furthermore, upon conjugation of the humanized version of the anti-Endosialin antibody with a duocarmycin derivative, the resulting ADC displays potent and antigen dependent in vitro tumor cell cytotoxicity and effective antitumor efficacy in vivo.

Altogether, experimental and clinical data indicate that Endosialin plays an essential role in tumor progression and angiogenesis, suggesting that agents targeting Endosialin could be useful as therapeutic and diagnostic tools for some cancers [28-32]

ErbB3 receptor, also known as HER-3, belongs to the epidermal growth factor receptor tyrosine kinase family (ErbB). This family of receptors consists of four members: ErbB1 (HER1 ), ErbB2 (HER2), ErbB3 (HER-3) and ErbB4 (HER4). Many studies have suggested a critical role for ErbB receptors in cell survival, proliferation and differentiation, as well as in malignant transformation [36, 37] The signal transduction mediated by tyrosine kinase receptors is complex and involves the interaction with two categories of ligands: epidermal growth factor (EGF) and EGF-like ligands (e.g. TGFa and amphiregulin), Neuregulin (NRG), also defined Heregulin (HRG) or Neu Differentiation Factor (NDF). Ligand binding to ErbB receptors induces the formation of receptor homo- and heterodimers and activation of the intrinsic kinase domain, resulting in phosphorylation on specific tyrosine residues within the cytoplasmic tail. These phosphorylated residues serve as docking sites for a range of proteins, the recruitment of which leads to the activation of intracellular signalling pathways. Generally, heterodimerization is preferred over homodimerization; ErbB2 is the preferred heterodimerization partner of the other ErbB receptors, including ErbB1 (activated by EGF or EGF-like ligands), and ErbB3 and ErbB4 (activated by neuregulin, NRG). The two major signaling pathways activated by ErbB receptors are Ras-Raf- MAPK and PI3K-AKT pathways. [38-40]

ErbB2 gene is amplified in 20 to 30% of breast cancers and is correlated with a poor prognosis. In the same way, ErbB3 receptor has also been shown to be overexpressed in breast cancer patients. High levels of expression of both ErbB2 and ErbB3 receptors are associated with an aggressive biology of tumor. In fact, upon NRG stimulation, ErbB2/ ErbB3 heterodimers deliver the most potent and long-lasting proliferative intracellular signal among the possible combinations of pairs of ErbB family members [38-41] Several studies have suggested an important role of ErbB3 receptor in progression of many human tumor types, such as prostate cancer, melanoma, and gastric carcinoma.

International application WO 2012/052230 discloses the anti-HER-3 antibody MP-RM-1 , deposited at Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ) on 15 October 2009 and designated DSM ACC3018. The antibody is found to be able to induce tumor regression.

All together, experimental and clinical data indicate that HER-3 plays an essential role in tumor development and progression, suggesting that agents targeting ErbB3 could provide a novel and promising approach toward the treatment of some cancers. [42-54]

In spite of scientific progress and introduction into clinical practice of new chemotherapeutic agents and targeted therapies, cancer remains a disease difficult to cure, responsible for about 13% of deaths worldwide [33-35]. Tumor cells often upregulate different growth-promoting receptors that can act either independently or crosstalk intra- cellularly. Targeting of one receptor by a mono-specific antibody may result in resistance which is associated with the upregulation of alternative receptors as well as pathway switching (Kontermann (2012) mAbs 4:2, 182-197), indicating that it may be beneficial to target or block multiple cell surface antigens (targets) on a tumor cell.

In light of the above considerations, it is clear that new, more effective, and possibly less toxic anti-tumor treatments are needed. In the present invention, novel bispecific antibodies have been developed, that are significantly more effective in the treatment of cancer than previously known monospecific antibodies and even combinations of monospecific antibodies.

Thus, according to a first aspect, the present invention provides a bispecific antibody comprising a first antigen binding domain that binds to 90K and a second antigen binding domain that binds Endosialin or HER-3. In some embodiments, the bispecific antibody further comprises a third antigen binding domain. For example, a trispecific antibody comprising antigen binding domains for 90K, Endosialin and HER-3 can be provided.

The bispecific antibody provided in the present invention preferably comprises a variable domain fragment of an antibody that specifically binds to human 90K and a variable domain fragment of an antibody that specifically binds to human Endosialin or a variable domain fragment of an antibody that specifically binds to HER-3.

Preferably, in the bispecific antibody, a heavy chain variable domain in the variable domain fragment of the antibody that specifically binds to human 90K is adjacent to a heavy chain variable domain in the variable domain fragment of the antibody that specifically binds to human Endosialin or HER-3, or a light chain variable domain in the variable domain fragment of the antibody that specifically binds to human 90K is adjacent to a light chain variable domain in the variable domain fragment of the antibody that specifically binds to human Endosialin or HER-3.

Preferably, the bispecific antibody has, from an N-terminus to a C-terminus, an arrangement of: light chain variable region and heavy chain variable region of the antibody that specifically binds to human 90K, and then heavy chain variable region and light chain variable region of the antibody that specifically binds to human Endosialin or HER-3, or heavy chain variable region and light chain variable region of the antibody that specifically binds to human 90K, and then light chain variable region and heavy chain variable region of the antibody that specifically binds to human Endosialin or HER-3.

The variable domain of the 90K antibody can be derived from the variable domain of any known 90K antibody, preferably from the sequence of the monoclonal antibody 1959. 1959 is a humanized anti-90K antibody of the isotype lgG1 that is characterized by a heavy chain variable region including the CDR sequences of the SEQ ID NOs. 1 -3 and light chain variable region comprising the CDR sequences of SEQ ID NOs. 4-6. The variable domain of the Endosialin antibody can be derived from the variable domain of any known Endosialin antibody, preferably from the sequence of the monoclonal antibody E-8.3. E-8.3 is a humanized anti-Endosialin antibody of the isotype lgG1 that is characterized by a heavy chain variable region including the CDR sequences of the SEQ ID NOs. 7-9 and light chain variable region comprising the CDR sequences of SEQ ID NOs. 10-12.

The variable domain of the HER-3 antibody can be derived from the variable domain of any known HER-3 antibody, preferably from the sequence of the monoclonal antibody EV20. EV20 is a humanized anti-HER-3 antibody of the isotype lgG1 that is characterized by a heavy chain variable region including the CDR sequences of the SEQ ID NOs. 13-15 and light chain variable region comprising the CDR sequences of SEQ ID NOs. 16-18.

In a particularly preferred embodiment of the invention, the bispecific antibody comprises an lgG1-type antibody targeting the first antigen (90K) fused with a scFv fragment targeting the second antigen (Endosialin or HER-3). Alternatively, the bispecific antibody may comprise an lgG1 -type antibody targeting the second antigen (Endosialin or HER-3)) fused with a scFv fragment targeting the first antigen (90K).

A further aspect of the invention is a nucleic acid molecule encoding the bispecific antibody, optionally in operative linkage to an expression control sequence.

A further aspect of the invention is a host, in particular a recombinant cell which comprises the nucleic acid molecule. The cell may be used for the preparation of the antibody.

Still a further aspect of the invention is pharmaceutical composition comprising the antibody, the nucleic acid molecule or the host, optionally, together with a pharmaceutical acceptable carrier.

Still a further aspect of the invention is a method for the prevention or treatment of neoplastic diseases and cancer.

DETAILED DESCRIPTION OF THE INVENTION

The present invention provides a bispecific antibody, which comprises a protein functional domain targeting a first antigen 90K and a protein functional domain targeting a second antigen which is selected from Endosialin and HER-3. The protein functional domain targeting the first antigen 90K is operably linked to the protein functional domain targeting the second antigen Endosialin or HER-3, while their respective spatial structures are maintained and their respective physiological activities are retained. The protein functional domain targeting the first antigen 90K and the protein functional domain targeting the second antigen Endosialin or HER-3 can be fused together directly without affecting their respective functions. Furthermore, the protein functional domain targeting the second antigen may be linked to the N terminus or C terminus of the protein functional domain targeting the first antigen 90K directly or with an additional spacer such as a linker therebetween. Moreover, both the heavy chain variable domain and the light chain variable domain of the protein functional domain targeting the first antigen 90K may be located at the N terminus of the protein functional domain targeting the first antigen 90K; and both the heavy chain variable domain and the light chain variable domain of the protein functional domain targeting the second antigen Endosialin or HER-3 may be located at the N terminus of the protein functional domain targeting the second antigen Endosialin or HER-3.

Preferably, the variable domain fragment of the antibody that specifically binds to human 90K comprises a heavy chain variable region comprising the CDR sequences as shown in SEQ ID NOs: 1-3 and a light chain variable region comprising the CDR sequences as shown in SEQ ID NOs: 4-6. The variable domain fragment of the antibody that specifically binds to human Endosialin preferably comprises a heavy chain variable region comprising the CDR sequences as shown in SEQ ID NOs: 7-9 and a light chain variable region comprising the CDR sequences as shown in SEQ ID NO: 10-12. The variable domain fragment of the antibody that specifically binds to human HER-3 preferably comprises a heavy chain variable region comprising the CDR sequences as shown in SEQ ID NOs: 13-15 and a light chain variable region comprising the CDR sequences as shown in SEQ ID NOs: 16-18.

In a particularly preferred embodiment of the invention, the bispecific antibody comprises an lgG1-type antibody targeting the first antigen (90K) fused with a scFv fragment targeting the second antigen (Endosialin or HER-3). Alternatively, the bispecific antibody may comprise an lgG1 -type antibody targeting the second antigen (Endosialin or HER-3)) fused with a scFv fragment targeting the first antigen (90K).

In a particularly preferred embodiment, the anti-Endosialin antibody E-8.3 is fused with the anti-90K antibody 1959, such that a fusion is made between the whole lgG1 of one antibody and the scFv of the second antibody. For example, 1959-scFv (VH-VL) can be fused to the C-terminal of the heavy chain of E-8.3 Igl 1. An especially preferred anti- 90K/anti-Endosialin-antibody with this arrangement comprises a heavy chain amino acid sequence as shown in SEQ ID NO:19 and a light chain amino acid sequence as shown in SEQ ID NO:20. Alternatively, E-8.3-scFv (VH-VL) can be fused to the C-terminal of the heavy chain of 1959-lgG1. These constructs are illustrated in Figure 1. An especially preferred anti-90K/anti-Endosialin-antibody with this arrangement comprises a heavy chain amino acid sequence as shown in SEQ ID NO:21 and a light chain amino acid sequence as shown in SEQ ID NO:22.

In another preferred embodiment, the anti-HER-3 antibody EV20 is fused with the anti- 90K antibody 1959, such that a fusion is made between the whole lgG1 of one antibody and the scFv of the second antibody. For example, 1959-scFv (VH-VL) can be fused to the C-terminal of the heavy chain of EV20 lgG1. Alternatively, EV20-scFv (VH-VL) can be fused to the C-terminal of the heavy chain of 1959-lgG1.

“ Bispecific" in the term“bispecific antibody’ as used herein refers to specifically targeting two different antigens at the same time. In the present invention, the two different antigens are 90K and one of Endosialin and HER-3, respectively. The bispecific antibody of the present invention is capable to inhibit to a greater extent tumor angiogenesis if compared to the effect of the single antibody treatments. The bispecific antibody at the same time inhibits adhesive processes of tumor cells and inhibits angiogenesis of tumor cells and pericytes.

The term “antibody’ particularly refers to molecules comprising at least one immunoglobulin heavy chain and at least one immunoglobulin light chain. Each heavy and light chain may comprise a variable and a constant domain. The antigen binding site may be formed from the variable domains of a heavy and light chain. A variable region (also referred to as variable domain) comprises complementarity determining regions (CDRs), e.g. a CDR1 , a CDR2 and a CDR3 region and framework regions (FRs) flanking the CDRs. The term“complementarity determining region" is readily understood by the skilled person (see for example Harlow and Lane (eds.), Antibodies: A Laboratory Manual, CSHL press, Cold Spring Harbor, N.Y., 1988) and refers to the stretches of amino acids within the variable domain of an antibody that primarily make contact with the antigen and determined antibody specificity. This region is also known as the hypervariable region.

The invention also encompasses fragments of antibodies, e.g. portions of the above- mentioned bispecific antibodies which comprise at least two antigen binding sites, one of 90K and the other for Endosialin or HER-3. Examples of antibody fragments include Fab fragments, Fab' fragments, F(ab')2 fragments, Fv fragments, diabodies, scFv fragments, single chain antibody molecules, small modular immunopharmaceuticals (SMIPs), affibodies, avimers, nanobodies, domain antibodies and other fragments as long as they exhibit the desired capability of binding to 90K and one of Endosialin and HER-3. For a review of certain antibody fragments see Hudson et al., Nat. Met. 9: 129-134 (2003).

A bispecific antibody according to the preferred embodiments of the present invention comprises the whole IgG of one antibody and a scFv fragment of the other antibody.

" Avimer " relates to a multimeric binding protein or peptide engineered using, for example, in vitro exon shuffling and phage display. Multiple binding domains are linked, resulting in greater affinity and specificity compared to single epitope immunoglobulin domains.

" Nanobody " or single domain antibody relates to an antibody fragment consisting of a single monomeric variable antibody domain.

"Affibody" molecules are small high affinity proteins being engineered to bind specifically to a large number of target proteins.

“ Diabodies " are antibody fragments with two antigen binding sites that may be bivalent or bispecific. See for example Hudson et al., (2003). Single-chain antibodies are antibody fragments comprising all or a portion of the heavy chain variable domain or all, or a portion of the light chain variable domain of an antibody. Antibody fragments can be made by various techniques including but not limited to proteolytic digestion of an intact antibody as well as production by recombinant hosts (e.g. E. coli or phage) as described herein.

Techniques for making multispecific, in particular bispecific antibodies include but are not limited to recombinant co-expression of two immunoglobulin heavy chain-light chain pairs having different specificities and“knob in hole” engineering. Multispecific antibodies may also be made by engineering electrostatic steering effects for making antibody Fc- heterodimeric molecules; crosslinking two or more antibodies or fragments; using leucine zippers to produce bispecific antibodies; using“ diabody” technology for making bispecific antibodies and using single-chain Fv and preparing trispecific antibodies as described. Engineered antibodies with three or more functional antigen binding sites including “octopus antibodies" are also included herein. In certain embodiments, amino acid sequence variants of the antibodies provided herein are contemplated as long as they exhibit the desired capability of binding to 90K and at least one of Endosialin and HER-3. For example, it may be desirable to improve the binding affinity and/or other biological properties of the antibody. Amino acid sequence variants of an antibody may be prepared by introducing appropriate modifications into the nucleotide sequence encoding the antibody, or by peptide synthesis. Such modifications include, for example, deletions from and/or insertions into and/or substitutions of residues within the amino acid sequences of the antibody. Any combination of deletion, insertion and substitution can be made to arrive at the final construct, provided that the final construct possesses the desired characteristics, e.g. antigen binding.

The term“bind’ or“ binding” of an antibody means an at least temporary interaction or association with or to a target antigen, i.e. 90K, Endosialin and HER-3, comprising fragments thereof containing an epitope. For 90K, a particular binding epitope is a conformational epitope between residues 107 and 435 of the amino acid sequence of the 90K protein. For Endosialin, a particular binding epitope is between amino acids 477-488 of human Endosialin.

In certain embodiments, a bispecific antibody provided herein has a dissociation constant (Kd) for 90K, Endosialin, and/or HER-3 of < 1 mM, < 100 nM, < 10 nM, < 1 nM, < 0.1 nM, < 0.01 nM, or < 0.001 nM (e.g. 10 ⁸ M or less, e.g. from 10 ⁸ M to 10 ¹³ M, e.g. 10 ⁹ M to 10 ¹³ M).

In one embodiment, Kd is measured by a radio-labeled antigen binding assay (Radioimmunoassay, RIA) performed with the Fab version of an antibody of interest and its antigen.

According to another embodiment, Kd is measured using surface plasmon resonance assays with immobilized antigen. According to a preferred embodiment of the present invention, the antibodies are human monoclonal antibodies directed against an epitope of human 90K and against an epitope of human Endosialin or human HER-3 as described herein.

The antibody may be any antibody of natural and/or synthetic origin, e.g. an antibody of mammalian origin. Preferably, the constant domain -if present- is a human constant domain. The variable domain is preferably a mammalian variable domain, e.g. a humanized or a human variable domain. Antibodies according to the invention are preferably monoclonal antibodies. In particular, antibodies of the present invention are preferably recombinant murine antibodies, chimeric, humanized or fully human antibodies, or fragments thereof.

Monoclonal antibodies may be produced by any suitable method such as that of Kohler and Milstein [55] or by recombinant DNA methods. Monoclonal antibodies may also be isolated from phage antibody libraries using techniques described in Clackson et al [56].

According to a preferred aspect of the invention, the antibodies of the invention are humanized antibodies, in particular fully human antibodies.

Humanized forms of the antibodies may be generated according to the methods known in the art such as chimerization or CDR grafting. Alternative methods for the production of humanized antibodies are well known in the art and are described in, e.g., EP-A1 0 239 400 and WO 90/07861. Human antibodies can also be derived by in vitro methods. Suitable examples include but are not limited to phage display, yeast display, and the like.

According the present invention "chimeric antibody” relates to antibodies comprising polypeptides from different species, such as, for example, mouse and human. The production of chimeric antibodies is described, for example, in WO 89/09622.

The antibody of the invention may be preferably of the lgG1 , lgG2, lgG3, lgG4, IgM, lgA1 , lgA2, IgAsec, IgD, and IgE antibody-type. It will be appreciated that antibodies that are generated need not initially possess such an isotype but, rather the antibody as generated can possess any isotype and that the antibody can be isotype-switched.

The antibodies or antibody fragments of the invention are optionally deimmunized for therapeutic purposes.

It will be apparent to those skilled in the art that the antibodies of the invention can be further coupled to other moieties for, e.g., drug targeting and imaging applications. Antibodies coupled to other moieties are also called "antibody conjugates". Coupling may be conducted chemically after expression of the antibody or antigen to site of attachment or the coupling product may be engineered into the antibody or antigen of the invention at the DNA level. For diagnostic purposes, the antibody or antibody fragment of the invention may be labelled, i.e. coupled to a labelling group. Suitable labels include radioactive labels, fluorescent labels, suitable dye groups, enzyme labels, chromogenes, chemiluminescent groups, biotinyl groups, predetermined polypeptide epitopes recognized by a secondary reporter etc. Preferably, the labels are covalently bound to the antibody.

Those labelled antibodies or antibody fragments (also referred to as “antibody conjugates") may in particular be used in immunohistochemistry assays or for molecular imaging in vivo.

For therapeutic purposes, the antibody or antibody fragment of the invention may be conjugated with an effector group, in particular a therapeutic effector group such as a cytotoxic agent or a radioactive group agent.

The bispecific antibody of the present invention may optionally be coupled to a labeling group and/or to an effector group, preferably a therapeutic group. According to a preferred aspect of the invention, the antibody is linked to a paramagnetic, radioactive or fluorogenic ion that is detectable upon imaging. This type of antibody is particularly suitable for diagnostic use.

According to another aspect of the invention, the antibody is linked to an anticellular agent, preferably in the form of anti-mitotic or DNA damaging agents capable of killing or suppressing the growth or cell division of tumor cells. The anticellular agent may, for example, comprise a chemotherapeutic agent, radioisotope or cytotoxin. Examples of anticellular agents comprise an antimetabolite, an anthracycline, a vinca alkaloid, an antibiotic, an alkylating agent or a plant-, fungus- or bacteria-derived toxin. An exemplary DNA damaging agent that may be linked to the antibody of the invention is a Minor Grove Binder duocarmycin derivative. Cytotoxins suitable to be linked to the antibody of the invention may, for example, comprise an A chain toxin, a ribosome inactivating protein, a- sarcin, aspergillin, restrictocin, a ribonuclease, diphtheria toxin or Pseudomonas exotoxin. Further, the cytotoxin may comprise deglycosylated ricin A chain.

Labelling groups or effector groups may be attached by linkers (spacer arms) of various lengths to reduce potential steric hindrance. Effector groups may be also attached directly to the antibody. According to another aspect, the present invention relates to a nucleic acid molecule encoding the bispecific antibody of the invention or fragment thereof or a nucleic acid capable of hybridizing thereto under stringent conditions. The nucleic acid molecule of the invention encoding the above-described antibody, antibody fragment or derivative thereof may be, e.g. DNA, cDNA, RNA or synthetically produced DNA or RNA or recombinantly produced chimeric nucleic acid molecule comprising any of those nucleic acid molecules either alone or in combination. The nucleic acid molecule may also be genomic DNA corresponding to the entire gene or a substantial portion thereof or to fragments and derivatives thereof. The nucleotide sequence may correspond to the naturally occurring nucleotide sequence or may contain single or multiple nucleotide substitutions, deletions or additions. In a particular preferred embodiment of the present invention, the nucleic acid molecule is a cDNA molecule.

According to the present invention, an isolated nucleic acid molecule of the present invention is particularly selected from the group consisting of:

(a) a nucleic acid sequence encoding an antibody, antibody fragment or a derivative thereof as disclosed herein,

(b) a nucleic acid sequence complementary to any of the sequences in (a); and

(c) a nucleic acid sequence capable of hybridizing to (a) or (b) under stringent conditions.

The term "hybridizing under stringent conditions" means that two nucleic acid fragments hybridize with one another under standardized hybridization conditions as described for example in Sambrook et al., "Expression of cloned genes in E. coli" in Molecular Cloning: A laboratory manual (1989), Cold Spring Harbor Laboratory Press, New York, USA. Such conditions are for example hybridization in 6.0xSSC at about 45° C. followed by a washing step with 2.0xSSC at 50° C, preferably 2.0xSSC at 65°C, or 0.2xSSC at 50°C, preferably 0.2xSSC at 65°C.

Another aspect of the invention relates to a vector comprising a nucleic acid molecule of the invention. Said vector may be, for example, a phage, plasmid, viral or retroviral vector. Retroviral vectors may be replication competent or replication defective. Preferably, the vector of the invention is an expression vector wherein the nucleic acid molecule is operatively linked to one or more control sequences allowing the transcription and optionally expression in prokaryotic and/or eukaryotic host cells. The invention further relates to a host comprising the vector of the invention. Said host may be a prokaryotic or eukaryotic cell or a non-human transgenic animal. The polynucleotide or vector of the invention which is present in the host may either be integrated into the genome of the host or it may be maintained extra chromosomally.

The host can be any prokaryotic or eukaryotic cell, such as a bacterial, insect, fungal, plant, animal, mammalian or, preferably, human cell. Preferred fungal cells are, for example, those of the genus Saccharomyces, in particular those of the species S. cerevisiae.

The invention additionally relates to a method for the preparation of an antibody, comprising culturing the host of the invention under conditions that allow synthesis of said antibody and recovering said antibody from said culture.

A further aspect of the present invention relates to a pharmaceutical composition comprising the bispecific antibody of the invention or a fragment thereof, the nucleic acid molecule, the vector, the host of the invention or an antibody obtained by a method of the invention. The term "composition" as employed herein comprises at least one compound of the invention. Preferably, such a composition is a therapeutic/pharmaceutical or a diagnostic composition.

The diagnostic composition of the invention may be used for assessing the onset or the disease status of a cancer.

The composition preferably comprises a pharmaceutically acceptable carrier, diluent and/or excipient.

Examples of suitable pharmaceutical carriers, excipients and/or diluents are well known in the art and include phosphate buffered saline solutions, water, emulsions, such as oil/water emulsions, various types of wetting agents, sterile solutions etc. Compositions comprising such carriers, excipients and/or diluents can be formulated by well-known conventional methods.

Administration of the suitable compositions may be effected by different ways, e.g., by intravenous, intraperitoneal, subcutaneous, intramuscular, topical, intradermal, intranasal or intra-bronchial administration. Preferred is an intravenous, intramuscular and/or subcutaneous administration. These pharmaceutical compositions can be administered to the subject at a suitable dose. The dosage regimen can be determined by the attending physician and clinical factors.

The compositions of the invention may be administered locally or systemically. Preparations for parenteral administration include sterile aqueous or non-aqueous solutions, suspensions, and emulsions. Examples of non-aqueous solvents are propylene glycol, polyethylene glycol, vegetable oils such as olive oil, and injectable organic esters such as ethyl oleate. Aqueous carriers include water, alcoholic/aqueous solutions, emulsions or suspensions, including saline and buffered media. Parenteral vehicles include sodium chloride solution, Ringer's dextrose, dextrose and sodium chloride, lactated Ringer's, or' fixed oils. Intravenous vehicles include fluid and nutrient replenishers, electrolyte replenishers (such as those based on Ringer's dextrose), and the like. Preservatives and other additives may also be present such as, for example, antimicrobials, anti-oxidants, chelating agents, and inert gases and the like. Furthermore, the pharmaceutical composition of the invention may comprise further agents depending on the intended use of the pharmaceutical composition.

According to an especially preferred embodiment the composition comprises a further active agent, such as a further antibody or antibody fragment.

Preferably the composition of the invention is used in combination with at least one further antineoplastic agent. Said combination is effective, for example, in inhibiting abnormal cell growth. Many antineoplastic agents are presently known in the art. In general the term includes all agents that are capable of prevention, alleviation and/or treatment of hyperproliferative disorders, especially cancer.

Preferably the antineoplastic agent is selected from the group consisting of antibodies, small molecules, antimetabolites, alkylating agents, topo-isomerase inhibitors, microtubule-targeting agents, kinase inhibitors, protein synthesis inhibitors, immuno- therapeutics, hormones or analogs thereof.

Specific examples of antineoplastic agents which can be used in combination with the antibodies provided herein include, for example, chemotherapeutic agents such as Paclitaxel, Anthracyclines, Fluoropirimidine, vinca alkaloids, platinum salts, in particular capecitabine, daunorubicin, daunomycin, dactinomycin, doxorubicin, epirubicin, idarubicin, esorubicin, bleomycin, mafosfamide, ifosfamide, cytosine arabinoside, bis- chloroethylnitrosurea, busulfan, mitomycin C, actinomycin D, mithramycin, prednisone, hydroxyprogesterone, testosterone, tamoxifen, dacarbazine, procarbazine, hexamethylmelamine, pentamethylmelamine, mitoxantrone, amsacrine, chlorambucil, methylcyclohexylnitrosurea, melphalan, cyclophosphamide, 6-mercaptopurine, 6- thioguanine, cytarabine (CA), 5-azacytidine, hydroxyurea, deoxycoformycin, 4- hydroxyperoxycyclophosphor-amide, 5-fluorouracil (5-FU), 5-fluorodeoxyuridine (5-FUdR), methotrexate (MTX), colchicine, taxol, vincristine, vinblastine, etoposide, trimetrexate, teniposide, cisplatin and diethylstilbestrol (DES).

The compositions of the invention may be administered in combination with a further therapeutic composition comprising an active agent as described above and/or irradiation and/or radiotherapy.

According to a preferred embodiment, the compositions of the invention are for the use in treating and/or preventing diseases wherein inhibition of angiogenesis is helpful, in particular neoplastic diseases or cancer. The compositions may also be used for the manufacture of a medicament for treating and/or preventing diseases wherein inhibition of angiogenesis is helpful, in particular neoplastic diseases or cancer.

Tumors to be treated are for example breast cancer, ovarian cancer, lung cancer, gastrointestinal cancer, melanoma, lymphoma and metastases thereof.

The invention further relates to a method of treating a disease wherein the antibody of the invention is administered to a mammal and wherein said disease is cancer or any other disease wherein inhibition of angiogenesis is helpful.

The present invention will be now described, according to its preferred embodiments, with particular reference to the enclosed figures.

Figure 1 illustrates bispecific antibody constructs obtained by fusing anti-human

Endosialin E-8.3 with anti-human 90K, wherein the lgG1 of one antibody is fused to the scFv of the other antibody. REFERENCES

1. Updyke TV, Nicolson GL. Malignant melanoma cell lines selected in vitro for

increased homotypic adhesion properties have increased experimental metastatic potential. Clin Exp Metastasis 1986;4:273-284.

2. lacobelli S, Arno E, D'Orazio A, Coletti G. Detection of antigens recognized by a novel monoclonal antibody in tissue and serum from patients with breast cancer. Cancer Res 1986;46:3005-10.

3. Koths K, Taylor E, Halenbeck R, Casipit C,Wang A. Cloning and characterization of a human Mac-2-binding protein, a newmember of the superfamily defined by the macrophage scavenger receptor cysteine-rich domain, J Biol Chem

1993;268:14245-9.

4. Ullrich A, Sures I, D.'Egidio M, Jallal B, Powell TJ, Herbst R, Dreps A, Azam M, Rubinstein M, Natoli C, et al. The secreted tumor-associated antigen 90K is a potent immune stimulator. J Biol Chem 1994;269:18401 -7.

5. Sasaki T, Brakebusch C, Engel J, Timpl R. Mac-2 binding protein is a cell- adhesive protein of the extracellular matrix which self-assembles into ring-like structures and binds b1 integrins, collagens and fibronectin. EMBO J

1998;17:1606-13.

6. D'Ostilio N, Sabatino G, Natoli C, Ullrich A, lacobelli S. 90K (Mac-2 BP) in human milk, Clin Exp Immunol 1996;104:543-6.

7. lacobelli S, Arno E, Sismondi P, Natoli C, Gentiloni N, Scambia G, Giai M, Cortese P, Panici PB, Mancuso S. Measurement of a breast cancer associated antigen detected by monoclonal antibody SP-2 in sera of cancer patients. Breast Cancer Res T reat 1988; 1 1 : 19-30.

8. Inohara H, Akahani S, Koths K, Raz A. Interactions between galectin-3 and Mac-2- binding protein mediate cell-cell adhesion. Cancer Res 1996;56:4530-4.

9. Grassadonia A, Tinari N, lurisci I, Piccolo E, Cumashi A, Innominato P, D'Egidio M, Natoli C, Piantelli M, lacobelli S. 90K (Mac-2 BP) and galectins in tumor progression and metastasis. Glycoconj J 2004;19:551 -6.

10. Ulmer TA, Keeler V, Andre S, Gabius HJ, Loh L, Laferte S. The tumor-associated antigen 90K/Mac-2-binding protein secreted by human colon carcinoma cells enhances extracellular levels of promatrilysin and is a novel substrate of matrix metalloproteinases-2, -7 (matrilysin) and -9: Implications of proteolytic cleavage. Biochim Biophys Acta 2010, Vol.1800; 3:336-343. 11. Damiano JS, Cress AE, Hazlehurst LA, Shtil AA, Dalton WS. Cell adhesion mediated drug resistance (CAM-DR): role of integrins and resistance to apoptosis in human myeloma cell lines. Blood 1999;93:1658.

12. Sethi T, Rintoul R, Moore S, et al. Extracellular matrix proteins protect small cell lung cancer cells against apoptosis: a mechanism for small cell lung cancer growth and drug resistance in vivo. Nat Med 1999;5:662.

13. Fornarini B, D'Ambrosio C, Natoli C, Tinari N, Silingardi V, lacobelli S. Adhesion to 90K (Mac-2 BP) as a mechanism for lymphoma drug resistance in vivo. Blood 2000;96:3282-5.

14. Rea A, Calmieri G, Tinari N, Natoli C, Tagliaferro P, Morabito A, Grassadonia A, Bianco AR, lacobelli S. 90K is a serum marker of poor prognosis in non-Hodgikin's lymphoma patients. Oncol Rep 1994;723-5.

15. Zhang DS, Jiang WQ, Li S, Zhang XS, Mao H, Chen XQ, Li YH, Zhan J,Wang FH.

Predictive significance of serum 90K/Mac-2BP on chemotherapy response in non- Hodgkin's lymphoma. Ai Zheng 2003; 22:870-3.

16. Christian S, Ahorn H, Koehler A, Eisenhaber F, Rodi HP, Garin- Chesa P, Park JE, Rettig WJ, Lenter MC. J Biol Chem. 2001 Mar 9;276(10):7408-14. Epub 2000 Nov 17. Molecular cloning and characterization of endosialin, a C-type lectin-like cell surface receptor of tumor endothelium.

17. Rettig WJ, Garin-Chesa P, Healey JH, Su SL, Jaffe EA, Old LJ. Proc Natl Acad Sci U S A. 1992 Nov 15;89(22): 10832-6. Identification of endosialin, a cell surface glycoprotein of vascular endothelial cells in human cancer.

18. Teicher BA. Int J Oncol. 2007 Feb; 30(2):305-12. Newer vascular targets:

endosialin (review).

19. Brian Tomkowicz, Katherine Rybinski, Brian Foley, Wolfgang Ebel, Brad Kline, Eric Routhier, Philip Sass, Nicholas C. Nicolaides, Luigi Grasso, and Yuhong Zhou Proc Natl Acad Sci U S A. 2007 Nov 13; 104(46): 17965-17970. Interaction of endosialin/TEM1 with extracellular matrix proteins mediates cell adhesion and migration.

20. Renate Becker, Martin C Lenter, Tobias Vollkommer, Anja M Boos, Dennis Pfaff, Hellmut G Augustin, Sven Christian. FASEB J 2008 Aug 19;22(8):3059-67. Tumor stroma marker endosialin (Tem1 ) is a binding partner of metastasis-related protein Mac-2 BP/90K.

21. Piccolo E, Tinari N, Semeraro D, Traini S, Fichera I, Cumashi A, La Sorda R,

Spinella F, Bagnato A, Lattanzio R, D'Egidio M, Di Risio A, Stampolidis P, Piantelli M, Natoli C, Ullrich A, lacobelli S. J Mol Med (Berl). 2013 Jan;91 (1 ):83-94.

LGALS3BP, lectin galactoside- binding soluble 3 binding protein, induces vascular endothelial growth factor in human breast cancer cells and promotes

angiogenesis.

Marty C, Langer-Machova Z, Sigrist S, Schott H, Schwendener RA, Ballmer-Hofer K. Isolation and characterization of a scFv antibody specific for tumor endothelial marker 1 (TEM1 ), a new reagent for targeted tumor therapy. Cancer Lett. 2006; 235:298-308.

Zhao A,Nunez-Cruz S, Li C, Coukos G, Siegel DL, Scholler N. Rapid isolation of high-affinity human antibodies against the tumor vascular marker Endosialin TEM1 , using a paired yeast- display/secretory scFv library platform. J Immunol Methods. 2011 ;363:221-232.

Rouleau C, Curiel M, Weber W, Smale R, Kurtzberg L, Mascarello J, Berger C, Wallar G, Bagley R, Honma N, Hasegawa K, Ishida I, Kataoka S, Thurberg BL, Mehraein K, Horten B, Miller G, Teicher BA. Endosialin protein expression and therapeutic target potential in human solid tumors: sarcoma versus carcinoma. Clin Cancer Res. 2008 Nov 15;14(22):7223-36.

Bagley RG. Endosialin: from vascular target to biomarker for human sarcomas. Biomark Med. 2009 Oct; 3(5):589-604.

Maia M, DeVriese A, Janssens T, Moons M, Lories RJ, Tavernier J, Conway EM. CD248 facilitates tumor growth via its cytoplasmic domain. BMC Cancer. 2011 May 8; 11 :162.

Tomkowicz B, Rybinski K, Sebeck D, Sass P, Nicolaides NC, Grasso L, Zhou Y. Endosialin/TEM-1/CD248 regulates pericyte proliferation through PDGF receptor signaling. Cancer Biol Ther. 2010 Jun 1 ;9(1 1 ):908-15.

Yang J, Zhang W. New molecular insights into osteosarcoma targeted

therapy. Curr Opin Oncol. 2013 Jul; 25(4):398-406.

Rouleau C, Sancho J, Campos-Rivera J, Teicher BA. Endosialin expression inside populations in human sarcoma cell lines. Oncol Lett. 2012 Feb; 3(2):325-329. Rouleau C, Smale R, Sancho J, Fu YS, Kurtzberg L, Weber W, Kruger A, Jones C, Roth S, Bormann C, Dunham S, Krumbholz R, Curiel M, Wallar G, Mascarello J, Campos-Rivera J, Horten B, Schmid S, Miller G, Teicher BA. Endosialin: a novel malignant cell therapeutic target for neuroblastoma. Int J Oncol.

20110ct;39(4):841.

Rouleau C, Smale R, Fu YS, Hui G, Wang F, Hutto E, Fogle R, Jones CM, Krumbholz R, Roth S, Curiel M, Ren Y, Bagley RG, Wallar G, Miller G, Schmid S, Horten B, Teicher BA. Endosialin is expressed in high grade and advanced sarcomas: evidence from clinical specimens and preclinical modeling. Int J Oncol. 2011 Jul;39(1 ):73-89. 32. Carson-Walter EB, Winans BN, Whiteman MC, Liu Y, Jarvela S, Haapasalo H, Tyler BM, Huso DL, Johnson MD, Walter KA. Characterization of TEM1/endosialin in human and murine brain tumors. BMC Cancer. 2009 Nov 30; 9:417.

33. A Jemal, R Siegel, E Ward et al. Cancer Statistics, 2009. CA Cancer J Clin 59:

225-49 (2009).

34. Mehlen P, Puisieux A. Metastasis: a question of life or death. Nat Rev Cancer;

6:449-58 (2006).

35. Chambers AF, Groom AC, MacDonald IC. Dissemination and growth of cancer cells in metastatic sites. Nat Rev Cancer 2:563- 72 (2002).

36. N.E. Hynes , MacDonald G. ErbB receptors and signaling pathways in cancer.

Curr Opin Cell Biol 21 (2):177-84 (2009).

37. T.Holbro, G.Civenni, N.E. Hynes, The ErbB receptors and their role in cancer

progression. Exp Cell Res 284, 99-1 10 (2003).

38. Browne BC, O'Brien N, Duffy et al. HER-2 signaling and inhibition in breast cancer.

Curr Cancer Drug Targets 9(3): 419-38(2009).

39. Harari D, Yarden Y. Molecular mechanisms underlying ErbB2/ERBB2 action in breast cancer. Oncogene 19 (53) : 6102-14 (2000).

40. Alimandi M, Wang LM, Bottaro D et al. Epidermal growth factor and betacellulin mediate signal transduction through co-expressed ErbB2 and ErbB3 receptors. EMBO J 16(18): 5608-17 (1997).

41. Alimandi M, Romano A, Curia MC, Muraro R et al. Cooperative siganling of ErbB3 and ErbB2 in neoplastic transformation and human mammary carcinomas.

Oncogene 10(9) : 1813-21 (1995).

42. N.E. Hynes, Targeting ERBB receptors in cancer. Recent Results Cancer Res.

172, 45-57 (2007).

43. N.E. Hynes, H.A. Lane, ERBB receptors and cancer: the complexity of targeted inhibitors. Nat Rev Cancer 5, 341 -54 (2005).

44. J.Baselga and S.M. Swain, Novel anticancer targets: revisiting ERBB2 and

discovering ERBB3. Nat.Rev.Cancer. 9,463-75, (2009).

45. B. Tanner, D Hasenclever, K Stern et al. ErbB3 predicts survival in Ovarian

cancer. J.CIin.Onco. 24, 17-23 (2006).

46. M.Reschke, D.Mihic-Probst, E.H. van der Horst et al. ERBB3 is a determinant for poor prognosis in Melanoma. Clin. Cane. Res. 14, 5188- 97 (2008).

47. M. Soler, F Manicni, O.Meca-Cortes et al. ERBB3 is requie for the maintenance of neuregulin-depndent and -independent attributes of malignant progression in prostate cancer cells. Int.J. Cancer 125,2565- 75 (2009). 48. D.F. Stem, ERBB3/ERBB3 and ERBB2/ERBB2 duet in mammamry development and breast cancer. J Mammary Gland Biol Neoplasia. 13, 215-23 (2008).

49. J.A. Engelman, K. Zejnullahu, T Mitsudomi et al. MET amplification leads to

Gefitinib resistance in lung cancer by activating ERBB3 signalling. Science 316, 1039-43 (2007).

50. L.M. Weiner, M.V. Dhodapkar and S. Ferrone. Monoclonal antibodies for cancer immunotherapy.

51. X. Huang, L. Gao, S. Wang et al. Heterotrimerization of the growth factor receptors ErbB2. ErbB3, and Insulin-like growth factor-l receptor in breast cancer cells resistant to Herceptin. Cane. Res.

52. T.Holbro, R. Beerli, F. Maurer et al. The ErbB2/ErbB3 heterodimer functions a san oncogenic unit: ErbB2 requires ErbB3 to drive breast tumor cell proliferation.

53. Y Yarden. The EGFR family and its ligands in human cancer: signaling

mechanisms and therapeutic opportunities.

54. M.R. Freeman. HER-2/ERBB3 heterodimers in prostate cancer: Whither HER1/

EGFR? Cancer Cell 6:427-428 (2004).

55. Kohler G, Milstein C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 1975; 256:495-497.

56. Clackson T, Hoogenboom HR, Griffiths AD, Winter G. Making antibody fragments using phage display libraries. Nature 1991 ; 15;352(6336):624-8.