AU2011265359B2 - Antibodies directed to the deletion mutants of epidermal growth factor receptor and uses thereof - Google Patents

Abstract

C \NRPonblDCC\RR\4)550A4_l DOC.15/12/2011 This invention relates to novel antibodies, particularly antibodies directed against deletion mutants of epidermal growth factor receptor and particularly to the type III deletion 5 mutant, EGFRvIII. The embodiments also relate to human monoclonal antibodies directed against deletion mutants of epidermal growth factor and particularly to EGFRvIII. The embodiments also relate to variants of such antibodies. Diagnostic and therapeutic formulations of such antibodies, and immunoconjugates thereof, are also provided.

Description

Australian Patents Act 1990 - Regulation 3.2 ORIGINAL COMPLETE SPECIFICATION STANDARD PATENT Invention Title: "Antibodies directed to the deletion mutants of epidermal growth factor receptor and uses thereof" The following statement is a full description of this invention, including the best method of performing it known to us: P/00/ol C:\NRPortb[\DCC\RR\4054742_1 DOC - 16/12/1I I ANTIBODIES DIRECTED TO THE DELETION MUTANTS OF EPIDERMAL GROWTH FACTOR RECEPTOR AND USES THEREOF This application is a divisional of Australian Patent Application No. 2004259398, the entire content of which is incorporated herein by reference. 5 Related Aplications This Application claims priority to United States Provisional Applications Serial Number 60/562,453 filed April 15, 2004, Serial Number 60/525570, filed November 26, 2003, and Serial Number 60/483,145, filed June 27, 2003. FIELD OF THE INVENTION 10 The present embodiments relate to novel antibodies, particularly antibodies directed against deletion mutants of epidermal growth factor receptor and particularly to the type III deletion mutant, EGFRvII. The embodiments also relate to human monoclonal antibodies directed against deletion mutants of epidermal growth factor receptor and particularly to EGFRvI. The embodiments also relate to variants of such antibodies. Diagnostic and therapeutic formulations of 15 such antibodies, and inmunoconjugates thereof, are also provided. BACKGROUND OF THE INVENTION Tumor specific molecules to aid in better diagnosis and treatment of human and animal cancer have been sought since the last century. Hard evidence of tumor-specific substances, based on molecular structural data, has been difficult to provide in most types of human cancer except 20 those based on virally-induced cancer and involving molecular structures specified by the virus genome. There have been extremely few examples of tumor-specific molecules based on novel molecular structures. In the case of malignant human gliomas and other tumors potentially associated with amplification or changes in the epidermal growth factor receptor molecule, such as carcinoma of the breast and other human carcinomas, there have been no unequivocal 25 demonstrations of-structurally altered molecules with unique sequences. The epidermal growth factor receptor (EGFR) is the 170 kilodalton membrane glycoprotein product of the proto-oncogene c-erb B. The sequence of the EGFR gene is known (Ullrich et al. (1984). Human Epidermal Growth Factor Receptor cDNA Sequence and Aberrant Expression of the Amplified Gene in A431 Epidermoid Carcinoma Cells. Nature 309:418-425). The EGFR gene 30 is the cellular homolog of the erb B oncogene originally identified in avian erythroblastosis viruses (Downward et al. (1984). Close Similarity of Epidermal Growth Factor Receptor and v-erb B Oncogene Protein Sequence. Nature 307:521-527, Ullrich, et al. (1984)). Activation of this oncogene by gene amplification has been observed in a variety of human tumors (Haley et al. (1987A). The Epidermal Growth Factor Receptor Gene in: Oncogenes, Genes, and Growth Factors 35 Edited by: Guroff, G. 12th Edition. Chapter 2. pp. 40-76. Wiley, N.Y.), and in particular, those of glial origin (Libermann et al. (1985). Amplification, Enhanced Expression and Possible Rearrangement of EGF Receptor Gene in Primary Human Brain Tumours of Glial Origin. Nature -l- 313:144-147; Wong et al. (1987). Increased Expression of the Epidermal Growth Factor Receptor Gene in Malignant Gliomas is Invariably Associated with Gene Amplification. Proc. Natil. Acad. Sci. USA 84:6899-6903; Yarnazaki et al. (1988). Amplification of the Structurally and Functionally Altered Epidermal Growth Factor Receptor Gene (c-erbB) in Human Brain Tumors. 5 Molecular and Cellular Biology 8:1816-1820; Malden et al., (1988). Selective Amplification of the Cytoplasmic Domain of the Epidermal Growth Factor Receptor Gene in Glioblastorna Multiforme. Cancer Research 4:2711-2714). EGF-r has been demonstrated to be overexpressed on many types of human solid tumors. Mendelsohn Cancer Cells 7:359 (1989), Mendelsohn Cancer Biology 1:339-344 (1990), 10 Modjtahedi and Dean Int'l J. Oncology 4:277-296 (1994). For example, EGFR overexpression has been observed in certain lung, breast, colon, gastric, brain, bladder, head and neck, ovarian, kidney and prostate carcinomas. Modjtahedi and Dean Int'l J. Oncology 4:277-296 (1994). Both epidermal growth factor (EGF) and transforming growth factor-alpha (TGF-alpha.) have been demonstrated to bind to EGF-r and to lead to cellular proliferation and tumor growth. 15 One major difference between v-erb B oncogenes and the normal EGFR gene is that the viral oncogenes are amino-truncated versions of the normal receptor; they lack most of the extracytoplasmic domain but retain the transmembrane and tyrosine kinase domains (Fung et al., (1984) Activation of the Cellular Oncogene c-erb B by LTR Insertion: Molecular Basis for Induction of Erythroblastosis by Avian Leukosis Virus. Cell 33:357-368; Yamamoto et al., (1983). 20 A New Avain Erythroblastosis Virus, AEV-H Carries erbB Gene Responsible for the Induction of Both Erythroblastosis and Sarcoma. Cell 34:225-232, Nilsen et al., (1985). c-erbB Activation in ALV-Induced Erythroblastosis: Novel RNA Processing and Promoter Insertion Results in Expression of an Amino-Truncated EGF Receptor. Cell 41:719-726; Gammett et al., (1986). Differences in Sequences Encoding the Carboxy-Terminal Domain of the Epidermal Growth Factor 25 Receptor Correlate with Differences in the Disease Potential of Viral erbB Genes. Proc. Natl. Acad. Sci. USA 83:6053-6057). This results in a protein that is unable to bind epidermal growth factor (EGF) but can still phosphorylate other substrates (Gilmore et al., (1985). Protein Phosphorlytion at Tyrosine is Induced by the v-erb B Gene Product in Vivo and In Vitro. Cell 40:609-618; Kris et al., (1985). Antibodies Against a Synthetic Peptide as a Probe for the Kinase 30 Activity of the Avian EGF Receptor and v-erB Protein. Cell 40:619-625), and has led to speculation that the v-erb B proteins are oncogenic because the kinase domain is unregulated and constitutively active (Downward et al., 1984). A variety of genetic alterations can occur in viral erb B oncogenes, e.g. amino acid substitutions and deletions in the carboxy terminus of the gene. Available evidence, however, 35 argues that the amino truncation is critical to carcinogenesis. Amino truncations are a feature of all v-erb B oncogenes, including those that arise by promoter insertion or retroviral transduction (Nilsen et al., (1985). c-erbB Activation in ALV-Induced Erythroblastosis: Novel RNA Processing -2and Promoter Insertion Results in Expression of an Amino-Truncated EGF Receptor. Cell 41:719 726; Gammett et al., (1986). Differences in Sequences Encoding the Carboxy-Terminal Domain of the Epidermal Growth Factor Receptor Correlate with Differences in the Disease Potential of Viral erbB Genes. Proc. Natl. Acad. Sci. USA 83:6053-6057). 5 In contrast, carboxy-terminal deletions appear to be associated only with tumors that arise through retroviral transduction and seem to determine host range and tumor type specificity (Gammett et al., 1986; Raines et al., (1985). c-erbB Activation in Avian Leukosis Virus-Induced Erythroblastosis: Clustered Integration Sites and the Arrangement of Provirus in the c-erbB Alleles. Proc. Natl. Acad. Sci. USA 82:2287-2291). Transfection experiments with amino-truncated avian 10 c-erb B genes or chimeric viral oncogene-human EGF receptors demonstrates that this deletion is sufficient alone to create a transforming protein (Pelley et al., (1988). Proviral-Activated c-erbB is Leukemogenic but not Sarcomagenic: Characterization of a Replication-Competent Retrovirus Containing the Activated c-erbB. Journal of Virology 62: 1840-1844; Wells et al., (1988). Genetic Determinants of Neoplastic Transformation by the Retroviral Oncogene v-erbB. Proc. Natl. Acad. 15 Sci. USA 85:7597-7601). Amplification of the EGFR gene occurs in 40% of malignant human gliomas (Libermann et al., (1985) Amplification, Enhanced Expression and Possible Rearrangement of EGF Receptor Gene in Primary Human Brain Tumours of Glial Origin. Nature 313:144-147; Wong et al., (1987). Increased Expression of the Epidermal Growth Factor Receptor Gene in Malignant Gliomas is 20 Invariably Associated with Gene Amplification. Proc. Natl. Acad. Sci. USA 84:6899-6903), Rearrangement of the receptor gene is evident in many of the tumors with gene amplification. The structural alterations seem to preferentially affect the amino terminal half of the gene (Yamazaki et al., (1985). Amplification, Enhanced Expression and Possible Rearrangement of EGF Receptor Gene in Primary Human Brain Tumours of Glial Origin. Nature 313:144-147; Maiden et al., (1988). 25 Selective Amplification of the Cytoplasmic Domain of the Epidermal Growth Factor Receptor Gene in Glioblastoma Multiforme. Cancer Research 4:2711-2714), but the nature of the rearrangements had not at that time been precisely characterized in any tumor. Size variant EGFR genes and amplification have been reported in several human cancers. (Humphrey et al., (1988). Amplification and Expression of the.Epidermal Growth Factor Receptor 30 Gene in Human Glioma Xenografts. Cancer Research 48:2231-2238; Bigner et al., (1988) J. Neuropathol. Exp. Neurol., 47:191-205; Wong et al., (1987). Increased Expression of the Epidermal Growth Factor Receptor Gene in Malignant Gliomas is Invariably Associated with Gene Amplification. Proc. NatI. Acad. Sci. USA 84:6899-6903; and Humphrey et al. Amplification and expression of the epidermal growth factor receptor gene in human glioma xenografts. Cancer Res. 35 48(8):2231-8 (1988)) There had been no determination, however, of the molecular basis for the altered EGFR molecules in cells. -3- In 1989, work of Drs. Bigner and Vogelstein elucidated the sequence of a EGF receptor mutant that has become known as the type III mutant (also referred to as delta-EGFr or EGFrvI). This work is described in U.S. Patent Nos. 6,455,498, 6,127,126, 5,981,725, 5,814,317, 5,710,010, 5,401,828, and 5,212,290. 5 EGFR variants are caused by gene rearrangement accompanied by EGFR gene amplification. There are eight major variants of EGFr that are known: (i) EGFRvI lacks a majority of the extracellular domain of EGFR, (ii) EGFRvII consists of an 83 aa in-frame deletion in the extracellular domain of EGFR, (iii) EGFRvm consists of a 267 aa in-frame deletion in the extracellular domain of EGFR, (iv) EGFRvIV contains deletions in the cytoplasmic domain of 10 EGFR, (v) EGFRvV contains deletions in cytoplasmic domain of EGFR, (vi) EGFR.TDM/2-7 contains a duplication of exons 2-7 in the extracellular domain of EGFR, (vii) EGFR.TDM/1 8-25 contains a duplication of exons 18-26 in the tyrosine kinase domain of EGFR, and (viii) EGFR.TDM/18-26 contains a duplication of exons 18-26 in the tyrosine kinase domain of EGFR (Kuan et al. EGF mutant receptor vIII as a molecular target in cancer therapy. Endocr Relat Cancer. 15 8(2):83-96 (2001)). In addition, there is a second, more rare, EGFRvm mutant (EGFRvIII/A12-13) that possesses a second deletion that introduces a novel histidine residue at the junction of exons 11 and 14 (Kuan et al. EGF mutant receptor vI as a molecular target in cancer therapy. Endocr Relat Cancer. 8(2):83-96 (2001)). EGFRvlI is the most commonly occurring variant of the epidermal growth factor (EGF) 20 receptor in human cancers (Kuan et al. EGF mutant receptor vII as a molecular target in cancer therapy. Endocr Relat Cancer. 8(2):83-96 (2001)). During the process of gene amplification, a 267 amino acid deletion occurs in the extracellular domain creating a novel junction to which tumor specific monoclonal antibodies can be directed. This variant of the EGF receptor contributes to tumor progression through constitutive signaling in a ligand independent manner. EGFrVM is not 25 know to be expressed on any normal tissues (Wikstrand, CJ. et al. Monoclonal antibodies against EGFRIII are tumor specific and react with breast and lung carcinomas malignant gliomas. Cancer Research 55(14): 3140-3148 (1995); Olapade-Olaopa, EO. et al. Evidence for the differential expression of a variant EGF receptor protein in human prostate cancer. Br J Cancer. 82(1):186-94 (2000)). Yet, EGFRvIIl shows significant expression in tumor cells, e.g., 27-76% breast cancer 30 biopsies express EGFRvII (Wikstrand, CJ. et al. Monoclonal antibodies against EGFRIII are tumor specific and react with breast and lung carcinomas malignant gliomas. Cancer Research 55(14): 3140-3148 (1995); Ge H. et al, Evidence of high incidence of EGFRvII expression and coexpression with EGFR in human invasive breast cancer by laser capture microdissection and immunohistochemical analysis. Int J Cancer. 98(3):357-61 (2002)), 50-70% gliomas express 35 EGFRvIII (Wikstrand, CJ. et al. Monoclonal antibodies against EGFRIII are tumor specific and react with breast and lung carcinomas malignant gliomas. Cancer Research 55(14): 3140-3148 (1995); Moscatello, G. et al. Frequent expression of a mutant epidermal growth factor receptor in -4multiple human tumors. Cancer Res. 55(23):5536-9 (1,995)), 16% NSCL cancers express EGFRvIII (Garcia de Palazzo, IE. et al. Expression of mutated epidermal growth factor receptor by non-small cell lung carcinomas. Cancer Res. 53(14):3217-20 (1993)), 75% ovarian cancers express EGFRvBI (Moscatello, G. et al. Frequent expression of a mutant epidermal growth factor receptor in multiple 5 human tumors. Cancer Res. 55(23):5536-9 (1995)), and 68% prostate cancers express .EGFRvM (Olapade-Olaopa, EO. et al. Evidence for the differential expression of a variant EGF receptor protein in human prostate cancer. Br J Cancer. 82(1):186-94 (2000)). The deletion of 267 amino acids with a Glycine substitution creates a unique junction that may be capable of antibody targeting. Further, in view of EGFRvIII's expression in certain tumors 10 and its lack of expression in normal tissues, EGFRvIII may be an ideal target for drug targeting in tumor therapy. In particular, EGFRvIII would appear to be an ideal candidate for immunoconjugate therapy of tumors (e.g., an antibody conjugated to an antineoplastic agent or toxin). Another method of treatment of cancers which over-express EGFRvII involved the use of a tumor-specific ribozyme targeted specifically to the variant receptor which did not cleave normal EGFR. The 15 ribozyme was found to significantly inhibit breast cancer growth in athymic nude mice (Luo et al. Int. J. Cancer. 104(6):716-21 (2003)). General antibodies for the entire EGFRvIII protein have been described. See International Patent Application No. WO 01/62931 and Kuan et al. EGF mutant receptor vIII as a molecular target in cancer therapy. Endocr Relat Cancer. 8(2):83-96 (2001), Kuan et al. EGFRvm as a 20 promising target for antibody-based brain tumor therapy. Brain Tumor Pathol. 17(2):71-78 (2000), Kuan et al. Increased binding affinity enhances targeting of glioma xenografts by EGFRvM specific scFv. International Journal of Cancer. 88(6):962-969 (2000), Landry et al. Antibody recognition of a conformational epitope in a peptide antigen: Fv-peptide complex of an antibody fragment specific for the mutant EGF receptor, EGFRvIII. Journal of Molecular Biology. 25 308(5):883-893 (2001), Reist et al. Astatine-211 labeling of internalizing anti-EGFRvIII monoclonal antibody using N-succinimidyl 5-[21 lAt]astato-3-pyridinecarboxylate. Nuclear Medicine and Biology. 26(4):405-411 (1999), Reist et al. In vitro and in vivo behavior of radiolabeled chimeric anti-EGFRvM monoclonal antibody: comparison with its murine parent. Nuclear Medicine and Biology. 24(7):639-647 (1997), Wikstrand et al. Generation of anti-idiotypic 30 reagents in the EGFRvm tumor-associated antigen system. Cancer Immunology, Immunotherapy. 50(12):639-652 (2002), Wikstrand et al. Monoclonal antibodies against EGFRvU are tumor specific and react with breast and lung carcinomas malignant gliomas. Cancer Research. 55(14):3140-3148 (1995), Wikstrand et al, The class III variant of the epidermal growth factor receptor (EGFRvM): characterization and utilization as an immunotherapeutic target. J.Neurovirol. 35 4(2):148-158 (1998), Wikstrand et al. The class III variant of the epidermal growth factor receptor (EGFRvIII): characterization and utilization as an immunotherapeutic target. J.Neurovirol. 4(2):148-158 (1998), Jungbluth et al. A monoclonal antibody recognizing human cancers with -5amplification/overexpression of the human epidermal growth factor receptor. Proc Nat] Acad Sci U S A. 100(2):639-44 (2003), Mamot et al. Epidermal Growth Factor Receptor (EGFR)-targeted Immunoliposomes Mediate Specific and Efficient Drug Delivery to EGFR- and EGFRvIII overexpressing Tumor Cells. Cancer Research 63:3154-3161 (2003)). Each of these above 5 mentioned antibodies, however, possess or contain murine sequences in either the variable and/or constant regions. The presence of such murine derived proteins can lead to the rapid clearance of the antibodies or can lead to the generation of an immune response against the antibody in a patient. In addition, such antibodies are relatively low affinity, on the order of 2.2 x 10' through 1.5 x 10- , even afier affinity maturation. (Kuan et al. EGF mutant receptor vIII as a molecular target in cancer 10 therapy. Endocr Relat Cancer. 8(2):83-96 (2001)). In order to avoid the utilization of murine or rat derived antibodies, researchers have introduced human antibody function into rodents so that the rodents can produce fully human antibodies. See e.g., Mendez et al, Functional transplant of megabase human immunoglobulin loci recapitulates human antibody response in mice. Nat Genet.15(2):146-56 (1997). This approach has 15 been used in connection with the generation of successful antibodies directed against wild type EGFR. See e.g., Yang X et al. Development of ABX-EGF, a fully human anti-EGF receptor monoclonal antibody, for cancer therapy. Crit Rev Oncol Hemato 38(1):17-23 (2001); Yang X-D et al. Eradication of Established Tumors by a Fully Human Monoclonal Antibody to the Epidermal Growth Factor Receptor without Concomitant Chemotherapy. Cancer Research 59(6):1236-1243 20 (1999); and U.S. Patent No. 6,235,883. SUMMARY OF THE INVENTION In one embodiment, the invention comprises an isolated human monoclonal antibody that specifically binds to EGFRvfI and a peptide that comprises the sequence L E E K K G N Y V V T D H C (SEQ ID NO: 56). In another embodiment, the invention comprises an isolated human 25 monoclonal antibody that specifically binds to an epitope contained within a sequence comprising L E E K K G N Y V V T D H C (SEQ ID NO: 56), wherein the residues required for binding, as determined by Alanine scanning in a SPOTs array, are-selected from the group consisting of EEK, KKNYV, LEK, EKNY and EEKGN. Further embodiments include an isolated human monoclonal antibody that comprises a 30 heavy chain variable region amino sequence that is encoded by a VH3-33 gene. The heavy chain variable region amino sequence can include an amino acid sequence that is encoded by a JH4b gene, or an amino acid sequence that is encoded by a D gene that is selected from the group consisting of D6-13 and D3-9. Other embodiments include an isolated human monoclonal antibody that comprises a light 35 chain variable region amino sequence that is encoded by a A23(VK2) gene. The light chain variable region amino sequence can include an amino acid sequence that is encoded by a JK1 gene. -6- Other embodiments include an isolated antibody, or fragment thereof, that binds to EGFRvII and that comprises a heavy chain amino acid sequence selected from the group consisting of the heavy chain amino acid sequence of antibody 13.1.2, 131, 170, 150, 095, 250, 139, 211, 124, 318, 342 and 333 as identified in (SEQ ID NO: 138, 2, 4, 5, 7, 9, 10, 12, 13, 15, 16, and 5 17). The antibody can be a monoclonal antibody, a chimeric antibody, a humanized antibody or a human antibody. The antibody or fragment can be associated with a pharmaceutically acceptable carrier or diluent, and can be conjugated to a therapeutic agent. The therapeutic agent can be a toxin. The therapeutic agent can be a toxin such as DM-1, AEFP, AURISTATIN E, or ZAP. The agent can be associated with the antibody via a linker. The toxin can be associated with the 10 antibody via a secondary antibody. Further embodiments include a hybridoma cell line producing the antibody, and a transformed cell comprising a gene encoding the antibody. The cell can be, for example, a Chinese hamster ovary cell. Further embodiments include a method of inhibiting cell proliferation associated with the expression of EGFRvDI, comprising treating cells expressing EGFRvIII with an effective amount 15 of the antibody or fragment. In one emobidment, the antibody comprises a heavy chain amino acid sequence selected from the group consisting of the heavy chain amino acid sequence of antibody 13.1.2 (SEQ ID NO: 138), 131 (SEQ ID NO: 2), 170 (SEQ ID NO: 4), 150 (SEQ ID NO: 5), 095 (SEQ ID NO: 7), 250 (SEQ ID NO: 9), 139 (SEQ ID NO: 10), 211 (SEQ ID NO: 12), 124 (SEQ ID NO: 13), 318 (SEQ ID NO: 15), 342 (SEQ ID NO: 16), and 333 (SEQ ID NO: 17). The method 20 can be performed in vivo, and performed on a mammal, such as a human, who suffers from a cancer involving epithelial cell proliferation, such as a lung, colon, gastric, renal, prostate, breast, glioblastoma or ovarian carcinoma. Further embodiments include a method of killing a targeted cell. This is achievd by contacting the targeted cell with an antibody associated with a toxin. The antibody binds to a 25 peptide LEEKKGNY (SEQ ID NO: 133). In one embodiment, the antibody has a binding affinity greater than 1.3*10' 9 M to the peptide. In one embodiment the toxin is selected from AEFP, DM-1, and ZAP. In one embodiment, the antibody toxin compound is 10 fold more toxic to targeted cells than to cells without the peptide. In one embodiment, the antibody comprises a heavy chain amino acid sequence selected from the group consisting of the heavy chain amino acid sequence of 30 antibody 13.1.2 (SEQ ID NO: 138), 131 (SEQ ID NO: 2), 170 (SEQ ID NO: 4), 150 (SEQ ID NO: 5), 095 (SEQ ID NO: 7), 250 (SEQ ID NO: 9), 139 (SEQ ID NO: 10), 211 (SEQ ID NO: 12), 124 (SEQ ID NO: 13), 318 (SEQ ID NO: 15), 342 (SEQ.ID NO: 16), and 333 (SEQ ID NO: 17). In another embodiment, the antibody is associated with a toxin via a peptide linker or a second antibody. 35 Further embodiments of the invention include an isolated antibody that binds to EGFRvIII and that comprises a heavy chain amino acid sequence comprising the following complementarity determining regions (CDRs): -7- (a) CDR1 consisting of a sequence selected from the group consisting of the amino acid sequences for the CDR1 region of antibodies 13.1.2, 131, 170, 150, 095, 250, 139, 211, 124, 318, 342 and 333 as identifiedin SEQ IDNO: 138,2, 4,5, 7,9, 10, 12,13, 15, 16, and 17; (b) CDR2 consisting of a sequence selected from the group consisting of the amino acid 5 sequences for the CDR2 region of antibodies 13.1.2, 131, 170, 150, 095, 250, 139, 211, 124, 318, 342 and 333 as identified in SEQ ID NO: 138,2, 4, 5, 7,9, 10,12, 13, 15, 16, and 17; and (c) CDR3 consisting of a sequence selected from the group consisting of the amino acid sequences for the CDR3 region of antibodies 13.1.2, 131, 170, 150, 095, 250, 139, 211, 124, 318, 342 and 333 as identified in SEQ ID NO: 138, 2, 4, 5, 7, 9, 10, 12, 13, 15, 16, and 17. In one 10 embodiment, the antibody is a monoclonal antibody, a chimeric antibody, human, or a humanized antibody. In one embodiment, the antibody is associated with a pharmaceutically acceptable carrier, diluent, and/or therapeutic agent. In one embodiment, the therapeuctic agent is a toxin. In one embodiment, the the toxin is DM-1 or Auristatin E. Also included is an isolated antibody, or fragment thereof, that binds to EGFRvM and that 15 comprises a light chain amino acid sequence selected from the group consisting of the light chain amino acid sequence of antibody 13.1.2, 131, 170, 150, 123, 095, 139, 250, 211, 318, 342, and 333 as identified in SEQ ID NO: 140, 19, 20, 21, 29, 23, 25, 26, 28, 33, 31 and 32. The antibody can be a monoclonal antibody, a chimeric antibody, a humanized antibody, or a human antibody. It can be associated with a pharmaceutically acceptable carrier or diluent, or conjugated to a therapeutic 20 agent, such as a toxin, for example DM1 or AURISTATIN E. In one embodiment a hybridoma cell line or a transformed cell producing an antibody comprising a light chain amino acid sequence selected from the group consisting of the light chain amino acid sequence of antibody 13.1.2, 131, 170, 150, 123, 095, 139, 250, 211, 318, 342, and 333 as identified in SEQ ID NO: 140, 19, 20, 21, 29, 23, 25,26, 28, 33, 31 and 32 is contemplated. 25 Further embodiments include a hybridoma cell line producing such an antibody, and a transformed cell, such as a Chinese hamster ovary cell, comprising a gene encoding the antibody. Yet another embodiment includes a method of inhibiting cell proliferation associated with the expression of EGFRvIII, comprising treating cells expressing EGFRvIII with an effective amount of the antibodies or fragments described abovi. The method can be performed in vivo and 30 on a mammal, such as a human, who suffers from a cancer involving epithelial cell proliferation such as lung, colon, gastric, renal, prostate, breast, glioblastoma or ovarian carcinoma. Yet another embodiment includes an isolated antibody that binds to EGFRvIII and that comprises a light chain amino acid sequence comprising the following complementarity determining regions (CDRs): 35 (a) CDR1 consisting of a sequence selected from the group consisting of the amino acid sequences for the CDR1 region of antibodies 13.1.2, 131, 170, 150, 123, 095, 139, 250, 211, 318, 342, and 333 as identified in SEQ ID NO: 140, 19, 20, 21, 29, 23, 25, 26, 28, 33, 31 and 32; -8- (b) CDR2 consisting of a sequence selected from the group consisting of amino acid sequences for the CDRI region of antibodies 13.1.2, 131, 170, 150, 123, 095, 139, 250, 211, 318, 342, and 333 as identified in SEQ ID NO: 140, 19, 20, 21, 29, 23, 25, 26, 28, 33, 31 and 32; and (c) CDR3 consisting of a sequence selected from the group consisting of amino acid sequences 5 for the CDR1 region of antibodies 13.1.2, 131, 170, 150, 123, 095, 139, 250, 211, 318, 342, and 333 as identified in SEQ ID NO: 140, 19, 20, 21, 29, 23, 25, 26, 28, 33, 31 and 32. The antibody identified in the previous paragraph can further include a heavy chain amino acid sequence comprising the following complementarity determining regions (CDRs): (a) CDRI consisting of a sequence selected from the group consisting of the amino acid 10 sequences for the CDR1 region of antibodies 13.1.2, 131, 170, 150, 05, 250, 139, 211, 124, 318, 342 and 333 as identified in SEQ ID NO: 138, 2, 4, 5, 7, 9, 10, 12, 13, 15, 16, and 17; (b) CDR2 consisting of a sequence selected from the group consisting of the amino acid sequences for the CDR2 region of antibodies 13.1.2, 131, 170, 150, 095, 250, 139, 211, 124, 318, 342 and 333 as identified in SEQ ID NO: 138, 2, 4, 5, 7, 9, 10, 12, 13, 15, 16, and 17; and 15 (c) CDR3 consisting of a sequence selected from the group consisting of the amino acid sequences for the CDR3 region of antibodies 13.1.2, 131, 170, 150, 095, 250, 139, 211, 124, 318, 342 and 333 as identified in SEQ ID NO: 138, 2, 4, 5, 7, 9, 10, 12, 13, 15, 16, and 17. Further embodiments include a method of inhibiting cell proliferation associated with the expression of EGFRvlII, comprising treating cells expressing EGFRvIII with an effective amount 20 of the antibody or fragment described above. The method can be performed in vivo, on a mammal, such as a human, suffering from a cancer involving epithelial cell proliferation, such as lung carcinoma, breast carcinoma, head & neck cancer, prostate carcinoma or glioblastoma. Other embodiments include an isolated polynucleotide molecule comprising a nucleotide sequence encoding a heavy chain amino acid sequence, or a fragment thereof, selected from the 25 group consisting of the heavy chain amino acid sequence of antibodies 13.1.2, 131, 170, 150, 095, 250, 139, 211, 124, 318, 342, and 333 as identified in SEQ ID NO: 138, 2, 4, 5, 7, 9, 10, 12, 13, 15, 16, and 17, or an isolated polynucleotide molecule comprising a nucleotide sequence encoding a light chain amino acid sequence, or a fragment thereof, selected from the group consisting of the light chain amino acid sequence of antibodies 13.1.2, 131, 170, 150, 123, 095, 139, 250, 211, 318, 30 342, and 333, as identified in SEQ ID NO: 140, 19, 20, 21, 29, 23, 25, 26, 28, 33,31 and 32. Further embodiments include an article of manufacture comprising a container, a composition contained therein, and a package insert or label indicating that the composition can be used to treat cancer characterized by the expression of EGFRvIII, wherein the composition comprises an antibody as described above. Such cancers include a lung carcinoma, breast 35 carcinoma, head & neck cancer, prostate carcinoma or glioblastoma. Also included is an assay kit for the detection of EGFRvII in mammalian tissues or cells in order to screen for lung, colon, gastric, renal, prostate or ovarian carcinomas, the EGFRvIII being an antigen expressed by -9epithelial cancers, the kit comprising an antibody that binds the antigen protein and means for indicating the reaction of the antibody with the antigen, if present. The antibody can be a labeled monoclonal antibody, or the antibody can be an unlabeled first antibody and the means for indicating the reaction comprises a labeled second antibody that is anti-immunoglobulin. The 5 antibody that binds the antigen can be labeled with a marker selected from the group consisting of a fluorochrome, an enzyme, a Radionuclide and a radiopaque material. The antibody that binds the antigen can also bind to over-expressed wtEGFR. The kit can be used clinically for patient selection. A further embodiment includes an antibody which specifically recognizes the epitope of 10 EGFRvm containing the novel Gly residue. A further embodiment includes a protein variant of EGFRvM. The variant may have a pFLAG insert, may consist of the amino acids in SEQ ID) NO: 56, and can exist in silico. Another embodiment includes an antibody, or variant thereof, which binds to the recognition sequence EEKKGNYVVT (SEQ ID NO: 57). 15 Another embodiment includes an antibody variant that specifically binds to EGFRvm. The antibody variant can further bind to a peptide that comprises SEQ ID NO: 57. The antibody variant can have residues that interact with residues EKNY or EEKGN in the peptide. In one embodiment, the antibody variant binds to the peptide sequence ten fold more tightly than it does to a wild-type EGFR protein. In one embodiment, the antibody binds specifically binds to 20 EGFRvM and the peptide of SEQ ID NO: 56. In one embodiment, the isolated antibody or variant has a complementarity determining region comprising a deep cavity, wherein the cavity is created by CDR2 and CDR3 of the heavy chain, CDR3 of the light chain, and a small portion from CDR1 of the , light chain. In one embodiment, the isolated antibody or variant has residues 31, 37, 95 101, 143- 147, 159, 162-166, 169-171, 211-219, 221, and 223 within 5 angstroms of a binding 25 cavity. In one embodiment, the isolated antibody or variant has a complementarity determining region comprising a narrow groove, wherein the groove is created by heavy chain CDR2 and CDR3, and light chain CDRI, CDR2, and CDR3. In one embodiment, the isolated antibody or variant has residues 31, 33, 35-39, 51, 54-56, 58-61, 94-101, 144-148, 160, 163-166, 172, and 211 221 within 5 angstroms of a binding groove. In one embodiment, the isolated antibody or variant 30 has residues 31- 33, 35, 37, 55, 96-101, 148, 163, 165, 170, 172, 178, 217, and 218 within 5 angstroms of a binding groove. In one embodiment, the isolated antibody or variant has a paratope configured so that when the epitope of peptide EEKKGN (SEQ ID NO 127) binds to the paratope of the antibody, at least one bond is formed between two residues selected from the group consisting of E2 and Y172, K3 and H31, K4 and H31, N6 and D33, N6 and Y37, and N6 and K55. 35 In one embodiment, the isolated antibody or variant has a paratope configured so that when the epitope of peptide EEKKGNY (SEQ ID 131) binds to the paratope of the antibody, at least one bond is formed between two residues selected from the group consisting of K4 and Q95, K4 and -10- Q95, N6 and Q98, G5 and H31, Y7 and H31, Y7 and W165. In one embodiment, the antibody has a structure or interaction with a structure that is determined in silicon. Another embodiment provides a method for selecting variants that bind to EGFRvm with particular binding characteristics, the method comprising the use of a molecular structure to create a 5 paratope, the use of a molecular structure to create an epitope, calculating the interaction energy between the two and comparing that energy level to the energy level of the epitope and a second paratope of a mAb variant, and selecting a variant based on the differences in the energy levels. The method can further include using an interaction energy between a second variant of the paratope and the epitope to determine a third interaction energy and comparing the third interaction 10 energy and the second interaction energy to determine which variant to select. In one embodiment, the variant is created and tested for binding. Another embodiment provides a method for selecting variants that bind to EGFRvIIl with particular binding characteristics, the method comprising examining residues of an epitope which interact with a paratope, selecting important residues to create a recognition sequence, using this 15 sequence to create a EGFRvIII variant, and using the EGFRvIII variant to select the mAb variant. Another embodiment provides a method for making antibody variants to EGFRvM, said method comprising analyzing the residues of an epitope which interact with a paratope, selecting the more important residues of an epitope to create a recognition sequence, using the recognition sequence to create an EGFRvM variant, and using the EGFRvm variant to select antibody variants. 20 In one embodiment, the selection of the antibodies is achieved in silicon. In one embodiment, the selection of the antibodies through the use of the EGFRvmI variant is achieved by raising antibodies against EGFRvM variant. In the embodiment where the isolated antibody variant binds to EGFRvIII and the peptide of SEQ ID NO: 57, the antibody can further comprise a point mutation of the following: 25 Tyr172Arg, Leu99Glu, Argl0lGlu, Leu2l7Glu, Leu99Asn, Leu99His, L99T, ArglOlAsp, or some combination thereof. In one embodiment, the antibody is a monoclonal antibody, a chimeric antibody, a humanized antibody, or a human antibody. In one embodiment, the antibody or variant thereof binds to the sequence EEKKGNYVVT (SEQ ID NO: 57), and the antibody or variant has subnanomolar binding ability. 30 In a further embodiment, the antibody binds to EGFRvIII and the antibody has a paratope that binds to an epitope, and the epitope has a set of residues that interact with the paratope that include E, K, N, and Y. In one embodiment, the antibody is antibody 131. In a further embodiment, the antibody binds to EGFRvII and the antibody has a paratope that binds to an epitope that has a set of residues that interact with the paratope comprising: E, E, K, 35 G, and N. In one embodiment, the primary structure of the epitope is EEKKGNY (SEQ ID NO: 131). In one embodiment, the antibody is 13.1.2. -11- In a further embodiment, the antibody that binds to EGFRvIII and has a KD of less than 1.3*10'9 M, less than 1.0*10'9 M, or less than 500 pM. In one embodiment, the antibody is specific for SEQ ID NO: 56 compared to a wild type EGFR peptide. In one embodiment, the nonspecific binding of the antibody to the wild type EGFR peptide (SEQ ID NO: 134) is less than 10% of that 5 of the specific binding of the antibody to EGFRVIII (SEQ ID NO: 135). In one embodiment, the antibody is selected from the group consisting of 131, 139, and 13.1.2. In one embodiment, the antibody is internalized. In one embodiment, the internalization occurs for at least about 70% or at least about 80% of the antibody. In one embodiment, the variant human monoclonal antibody preferentially binds to an 10 epitope that is substantially unique to an EGFRvIII protein compared to a wild-type EGFR protein or variant thereof (SEQ ID NO: 134). In one embodiment, the variant comprises a heavy chain complementarity determining region (CDRI) corresponding to canonical class 1. In one embodiment, the variant comprises a heavy chain complementarity determining region (CDR2) corresponding to canonical class 3. In one embodiment, the variant comprises a light chain 15 complementarity determining region (CDRI) corresponding to canonical class 4. In one embodiment, the variant comprises a light chain complementarity determining region (CDR2) corresponding to canonical class 1. In one embodiment, the variant comprises a light chain complementarity determining region (CDR3) corresponding to canonical class 1. In one embodiment, the variant comprises a first heavy chain complementarity determining region (CDRI) 20 corresponding to canonical class 1, a second heavy chain complementarity determining region (CDR2) corresponding to canonical class 3, a first light chain complementarity determining region (CDRl) corresponding to canonical class 4, a second light chain complementarity determining region (CDR2) corresponding to canonical class 1; and a third light chain complementarity determining region (CDR3) corresponding to canonical class 1, wherein the complementary 25 determining regions are configured to allow the variant to bind to an epitope that is substantially unique to an EGFRvI protein as compared to a EGFR protein. In a further embodiment, a method of inhibiting cell proliferation associated with the expression of EGFRvIII is provided. The method involves treating cells expressing EGFRvIII with an effective amount of an antibody or fragment thereof, wherein said antibody or fragment thereof 30 binds to EGFRvIII, wherein said antibody is conjugated to a toxin and wherein the antibody comprises a heavy chain amino acid sequence selected from the group consisting of the heavy chain amino acid sequence of antibody 13.1.2 (SEQ ID NO: 138), 131 (SEQ ID NO: 2), 170 (SEQ D NO: 4), 150 (SEQ ID NO: 5), 095 (SEQ ID NO: 7), 250 (SEQ ID NO: 9), 139 (SEQ ID NO: 10), 211 (SEQ ID NO: 12), 124 (SEQ ID NO: 13), 318 (SEQ ID NO: 15), 342 (SEQ ID NO: 16), and 35 333 (SEQ ID NO: 17). The method can be performed in vivo, on a mammal, the mammal can be human, and can be suffering from a cancer involving epithelial cell proliferation, and the cancer -12may involve lung, colon, gastric, renal, prostate, breast, glioblastoma or ovarian carcinoma. The toxin may by DM-1, AEFP, MMAE, AURISTATIN E, or ZAP. In a further embodiment, a method of inhibiting cell proliferation of cells expressing EGFRvIII is provided. The method involves treating cells expressing EGFRvIII with an effective 5 amount of an antibody or fragment thereof, wherein said antibody is conjugated to a toxin, and wherein said antibody has a light chain amino acid sequence selected from the group consisting of the light chain amino acid sequence of antibodies 13.1.2, 131, 170, 150, 123, 095, 139, 250, 211, 342, 333, and 318 as identified in SEQ ID NOs: 19, 20, 21, 29, 23, 25, 26, 28, 33, 31 and 32, wherein said isolated polynucleotide molecule will bind a peptide with the sequence identified in 10 SEQ ID NO: 56. The method can be performed in vivo, on a mammal, the mammal can be human, and can be suffering from a cancer involving epithelial cell proliferation, and the cancer may involve lung, colon, gastric, renal, prostate, breast, glioblastoma or ovarian carcinoma. The toxin may by DM-1, AEFP, MMAE, AURISTATIN E, or ZAP. BRIEF DESCRIPTION OF THE DRAWINGS 15 FIG. I is an alignment between wild type EGFR and EGFRvIII showing the 267 amino acid deletion and G substition. FIG. 2 is a diagram of the design of the EGFRvIII PEP3 14-mer peptide. In FIG. 2A, the N-terminal sequence of EGFRvIII with amino acids LEEKK (SEQ ID NO: 58) (1-5) that are identical to the N-terminal seqence of EGFR, followed by the unique Glysine residue, followed by 20 amino acids that are identical to residues 273 through 280 in EGFR. FIG. 2B represents the amino acids of EGFR that are deleted in EGFRvIII (6-272). FIGs. 3A-L provide sequences of antibodies of the invention. For each antibody provided, a nucleotide and amino acid sequence is provided for both a heavy chain and a light chain variable region. Accordingly, four sequences are provided for every antibody listed. 25 FIG. 4 is a table comparing the 13.1.2 antibody heavy chain regions to a particular germ line heavy chain region. "-"s indicate that the amino acid residue of the hybridoma heavy chain region is the same as the germ line for that particular position. Deviation from the germline is indicated by the appropriate amino acid residue. FIG. 5 is a table comparing the 13.1.2 antibody light chain regions to a particular germ line 30 light chain region. "-"s indicate that the amino acid residue of the hybridoma light chain region is the same as the germ line for that particular position. Deviation from the germline is indicated by the appropriate amino acid residue. FIG. 6 is a table comparing various hybridoma derived antibody heavy chain regions to a particular germ line heavy chain region. "-"s indicate that the amino acid residue of the hybridoma 35 heavy chain region is the same as the germ line for that particular position. Deviation from the germline is indicated by the appropriate amino acid residue. -13- FIG. 7 is a table comparing various hybridoma derived antibody light chain regions to a particular germ line light chain region. "-"s indicate that the amino acid residue of the hybridoma light chain region is the same as the germ line for that particular position. Deviation from the germline is indicated by the appropriate amino acid residue. 5 FIG. 8 is a representative figure showing binding of recombinant EGFRvM mAbs to cells expressing EGFRvfII (NR6 cells). Diamonds represent 95, triangles represent 133, squares represent 139, "x" represent 150, asterixes represent 170, circles represent 221, lines 230, and rectangles represent 250. FIG. 9A shows FACS staining analysis for a human anti-EGFR antibody (ABX-EGF) to 10 H80. FIG. 9B shows FACS staining analysis for antibody 131 to H80. FIG. 9C shows FACS staining analysis for antibody 139 to H80. FIG. 9D shows FACS staining analysis for antibody 13.1.2 to H80. FIG. 9E shows FACS staining analysis for ABX-EGF to H1477. 15 FIG. 9F shows FACS staining analysis for antibody 131 to H1477. FIG. 9G shows FACS staining analysis for antibody 139 to H1477. FIG. 9H shows FACS staining analysis for antibody 13.1.2 to H1477. FIG. 91 shows FACS staining analysis for ABX-EGF to A549. FIG. 9J shows FACS staining analysis for antibody 131 to A549. 20 FIG. 9K shows FACS staining analysis for antibody 139 to A549. FIG. 9L shows FACS staining analysis for antibody 13.1.2 to A549. FIG. 9M is a graph displaying binding of EGFRvIII mAbs to glioblastoma cells. Filled triangles represent antibody 131 binding to H1477. Filled squares represent antibody 13.1.2 binding to H1477. Empty triangles represent antibody 131 binding to H80. Empty squares 25 represent antibody 13.1.2 binding to H80. FIG. 9N is a graph displaying the binding of EGFRvIII mAbs to human epidermoid carcinoma cell line A431. The filled squares represent antibody 13.1.2. The filled triangles represent antibody 131. FIG. 90 is a graph displaying the binding of antibody 13.1.2 to NR6 murine fibroblast cell 30 lines. The squares represent NR6. The triangles represent NR6 with with wild type EDFR. The circles represent NR6 with EGFRvIII. FIG. 9P is a graph displaying the binding of antibody 131 to murine fibroblast cell lines. The squares represent NR6. The triangles represent NR6 with wild type EGFR. The circles represent NR6 with EGFRvII. 35 FIG. 10A shows FACS staining analysis for a human anti-EGFR antibody (ABX-EGF) binding to cells expressing EGFR (A43 1). -14- FIG. 1OB shows FACS staining analysis for antibody 131 to cells expressing EGFR (A43 1). FIG. 10C shows FACS staining analysis for antibody 139 to cells expressing EGFR (A43 1). 5 FIG. 1OD shows FACS staining analysis for antibody 13.1.2 to cells expressing EGFR (A43 1). FIG. 1 A shows in vitro toxicities for EGFRvHI antibody 13.1.2 indirectly conjugated to AEFP in EGFRvfl expressing cells (H1477, circles) versus cells that do not express EGFRvI (H80, squares). 10 FIG. 1 B shows in vitro toxicities for EGFRvIII antibody 13.1.2 indirectly conjugated to DM1 in EGFRvM expressing cells (H1477, circles) versus cells that do not express EGFRvIfl (H80, squares). FIG. 11 C shows in vitro toxicities for EGFRvHI antibody 13.1.2 indirectly conjugated to ZAP in EGFRvII expressing cells (H1477, circles) versus cells that do not express EGFRvLI (H80, 15 squares). FIG. 1 D shows in vitro toxicities for EGFRvUI antibody 95 indirectly conjugated to AEFP in EGFRvIm expressing cells (H1477, circles) versus cells that do not express EGFRvIII (H80, squares). FIG. I1E shows in vitro toxicities for EGFRvIf antibody 95 indirectly conjugated to DM1 20 in EGFRvm expressing cells (H1477, circles) versus cells that do not express EGFRvIH (H80, squares). FIG. 1 IF shows in vitro toxicities for EGFRvIII antibody 95 indirectly conjugated to ZAP in EGFRvIH expressing cells (H1477, circles) versus cells that do not express EGFRvIm (H80, squares). 25 FIG. 1IG shows in vitro toxicities for EGFRvMI antibody 131 indirectly conjugated to AEFP in EGFRvII expressing cells (H1477, circles) versus cells that do not express EGFRvm (H80, squares). FIG. 11 H shows in vitro toxicities for EGFRvm antibody 131 indirectly conjugated to DM1 in EGFRvM expressing cells (H1477, circles) versus cells that do not express EGFRvM 30 (H80, squares). FIG. 111 shows in vitro toxicities for EGFRvH antibody 131 indirectly conjugated to ZAP in EGFRvM expressing cells (H1477, circles) versus cells that do not express EGFRvHI (1180, squares). FIG. 12A shows in vitro toxicities for EGFRvIfl antibody 139 indirectly conjugated AEFP 35 in EGFRvM expressing cells (H1477, circles) versus cells that do not express EGFRvm (H80, squares). -15- FIG. 12B shows in vitro toxicities for EGFRvI antibody 139 indirectly conjugated DM1 in EGFRvII expressing cells (H1477, circles) versus cells that do not express EGFRvVI (H80, squares). FIG. 12C shows in vitro toxicities for EGFRvIII antibody 139 indirectly conjugated ZAP in 5 EGFRvl expressing cells (111477, circles) versus cells that do not express EGFRvm (H80, squares). FIG. 12D shows in vitro toxicities for EGFRvII antibody 150 indirectly conjugated AEFP in EGFRvM expressing cells (H1477, circles) versus cells that do not express EGFRvH (H80, squares). 10 FIG. 12E shows in vitro toxicities for EGFRvm antibody 150 indirectly conjugated DM1 in EGFRvIII expressing cells (H1477, circles) versus cells that do not express EGFRvm (H80, squares). FIG. 12F shows in vitro toxicities for EGFRvII antibody 150 indirectly conjugated ZAP in EGFRvm expressing cells (H1477, circles) versus cells that do not express EGFRvfl (H80, 15 squares). FIG. 12G shows in vitro toxicities for EGFRvII antibody 170 indirectly conjugated AEFP in EGFRvIII expressing cells (H1477, circles) versus cells that do not express EGFRvIII (H80, squares). FIG. 12H shows in vitro toxicities for EGFRvHI antibody 150 indirectly conjugated DM1 20 in EGFRvm expressing cells (111477, circles) versus cells that do not express EGFRvmI (H80, squares). FIG. 121 shows in vitro toxicities for EGFRvII antibody 150 indirectly conjugated ZAP in EGFRvm expressing cells (111477, circles) versus cells that do not express EGFRvm (H80, squares). 25 FIG. 13A shows in vitro toxicities for antibody 211 indirectly conjugated to AEFP in EGFRvM expressing cells (H1477, circles) versus cells that do not express EGFRvU (H80, squares). FIG. 13B shows in vitro toxicities for antibody 211 indirectly conjugated to DM1 in EGFRvM expressing cells (H1477, circles) versus cells that do not express EGFRvAI (H80, 30 squares). FIG. 13C shows in vitro toxicities for antibody 211 indirectly conjugated to ZAP in EGFRvIII expressing cells (111477, circles) versus cells that do not express EGFRvIII (1180, squares). FIG. 13D shows in vitro toxicities for antibody 250 indirectly conjugated to AEFP in 35 EGFRvIII expressing cells (111477, circles) versus cells that do not express EGFRvUI (H80, squares). -16- FIG. 13E shows in vitro toxicities for antibody 250 indirectly conjugated to DM1 in EGFRvII expressing cells (11477, circles) versus cells that do not express EGFRvIII (H80, squares). FIG. 13F shows in vitro toxicities for antibody 250 indirectly conjugated to ZAP in 5 EGFRvlU expressing cells (111477, circles) versus cells that do not express EGFRvHII (1180, squares). FIG. 13G shows in vitro toxicities for antibody IgG1, a negative control, indirectly conjugated to AEFP in EGFRvHI expressing cells (H1477, circles) versus cells that do not express EGFRvHI (H80, squares). 10 FIG. 13H shows in vitro toxicities for antibody IgG1, a negative control, indirectly conjugated to DM1 in EGFRvIII expressing cells (H1477, circles) versus cells that do not express EGFRvII (180, squares). FIG. 131 shows in vitro toxicities for antibody IgG1, a negative control, indirectly conjugated to ZAP in EGFRvM expressing cells (H1477, circles) versus cells that do not express 15 EGFRvm (H80, squares). FIG. 14A is bar graph that demonstrates that EGFRvIII antibodies (13.1.2, 131, and 139) inhibit colony formation in H1477 cells in a clongenic assay when conjugated to AEFP. FIG. 14B is a bar graph that demonstrates that EGFRvM antibodies (13.1.2, 131, and 139) inhibit colony formation in H1477 cells in a clongenic assay when conjugated to DM1. 20 FIG. 15A is a graph showing in vitro toxicities of direct conjugates of anti-EGFRvIfl antibodies (13.1.2) with toxin, MMAE, in EGFRvII expressing cells (H1477, circles) versus cells that do not express EGFRviI (1180, squares). FIG. 15B is a graph showing in vitro toxicities of.direct conjugates of anti-EGFRvIII antibodies (13.1.2) with toxin, AEFP, in EGFRvIH expressing cells (111477, circles) versus cells 25 that do not express EGFRvI (H80, squares). FIG. 15C is a graph showing in vitro toxicities of direct conjugates of anti-EGFRvM antibodies (13.1.2) with toxin, DM1, in EGFRvHI expressing cells (111477) versus cells that do not express EGFRvM (H80). FIG. 16 shows the results of an in vivo animal model in which mice having an established 30 tumor xenograft were treated witli an anti-EGFRvflH (or dEGFR) antibody (13.1.2), that was directly conjugated to a toxin (DM1, MMAE, or AEFP). Filled squares represent 250 micrograms of dEGFR-DM1. Filled upward pointing triangles represent 75 micrograms of the same. Filled downward pointing triangles represents 75 micrograms of dEGFR-MMAE. Diamonds represent 250 micrograms of the same. The lighter square represents 75 micrograms of dEGFR-AEFP. The 35 empty square represents 250 micrograms of the same. The empty downward pointing triangle represents dEGFR and free DM1. The empty upward pointing antibody represents PBS. All antibodies used were 13.1.2. Arrows indicate pro-drug treatments. -17- FIG. 17 shows the molecular surface of antibody 131 structure model. The six CDRs are shaded different shades to mark their boundries. The binding cavity is located close to the center. FIG. 18 shows a structural model of the molecular surface of antibody 13.1.2. The six CDR regions are shaded and identified by number. The long groove is located approximately along 5 the vertical centerline. FIG. 19A is a possible docking model of the 13.1.2 antibody and peptide EEKKGN (SEQ ID NO: 127) complex. The CDR regions are shaded to denote boundries. FIG. 19B shows the hydrogen bonds in the docking model of the 13.1.2 antibody and peptide EEKKGN (SEQ ID NO: 127) complex. Shading of the CDR loops and residues is the same 10 as in FIG. 18. The peptide residue is numbered from the N-terminus, at the top of the figure, to the C-terminus as 1 through 6. Six hydrogen bonds are indicated by dashed lines. The six pairs of amino acids forming hydrogen bonds are: E2...Y172, K3...H31, K4...H31, N6...D33, N6...Y37, and N6...K55. FIG. 20 is a graph demonstrating a correlation between the epitope-antibody binding 15 energy and the logrithm of Kd for one of the docking models selected. FIG. 21 is a depiction of a refined docking model for the peptide-13.1.2 antibody complex. The peptide is rendered in a space-filling manner.. FIG. 22 is a depiction representing the hydrogen bonds in the refined docking model. FIG. 23 is a graph that depicts the linear fitting of antibody-antigen binding energy versus 20 the logrithm of relative affinities. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS As discussed above, EGFRvIII is a deletion mutant of EGFR in which 267 amino acids in the extracellular domain of EGFr are deleted with a single amino acid substitution of Glycine at the junction. These features are shown in a sequence alignment between wild type EGFR and 25 EGFRvM in FIG. 1, In view of the amino acid substitution of Glycine at the junction of the deletion, it becomes theoretically possible to generate antibodies to the novel epitope present in EGFRvm that is not present in wild type EGFR. Thus, a peptide for immunization and screening was designed, termed PEP3, as shown in FIG. 2 (Kuan et al. EGF mutant receptor vIII as a molecular target in cancer therapy. Endotcr Relat Cancer. 8(2):83-96 (2001)). Such 14-mer peptide 30 possesses the 5 n-terminal amino acids common to EGFRvm and wild type EGFR, the unique Glycine junction site, and 8 amino acid residues contained in the conserved sequences between wild type EGFR (corresponding to residues 273-280) and EGFRvm (corresponding to residues 7-14). In addition, glioblastoma cell and cells (5300.19 cells) transfected with the gene encoding EGFRvM were also utilized for immunization and screening (sometimes referred to herein as 35 B300.19/EGFRvM transfectants). In order to generate human antibodies against EGFRvIII, transgenic XenoMouse@ mice were immunized with combinations of glioblastoma cells/EGFRvIf, B300.19/EGFRvuI cells, and -18peptides (PEP3) directed to the junction region in the novel extracellular domain represented in EGFRvIII as compared to wild type EGFR. B cells from immunized mice were isolated and either used to produce hybridomas followed by screening for binding to EGFRvIII or used directly in screening for binding to EGFRvIII using XenoMaxTM/SLAMI" technologies (Babcook et al. A 5 novel strategy for generating monoclonal antibodies from single, isolated lymphocytes producing antibodies of defined specificities. Proc Natl Acad Sci U S A.93(15):7843-8 (1996), and U.S. Patent No. 5,627,052). Antibodies identified that bound to EGFRvfI were screened in a series of assays to ascertain specific recognition of EGFRvIII. Through this process, panels of human monoclonal antibodies that bound to and were specific for EGFRvM were generated, isolated, and 10 characterized. Subsequent epitope mapping demonstrated unique but overlapping specificities . All antibodies were further evaluated in vitro for their ability to be internalized by cells for the purpose of delivering cytotoxic drugs to cells. Antibodies demonstrating efficient drug delivery were directly conjugated with a cytotoxic drug and examined for their ability to kill tumor cells expressing EGFRvM in vitro and in vivo. These studies provide the basis for the next generation of 15 antibody drug conjugates for treating cancer in patients whose tumor harbor specific genetic lesions. Through the processes described above, panels of fully human anti-EGFRvIII antibodies were generated. Using the hybridoma approach, several antibodies, including antibody 13.1, 13.2, 13.3, and 13.4 that were positive on ELISA for binding with the PEP3, were generated with limited 20 cross-reactivity with wild type EGFR. Out of these, antibody 13.1 (and, particularly, its subclone 13.1.2) was selected for further research and development. Using the XenoMax approach a panel of antibodies, including antibody 131, 139, 250, and 095, were generated that were highly specific for binding with the pep3 oligonucleotide and had limited cross-reactivity with wild type EGFR. Of these, the 131 antibody has very interesting properties. The sequences for each of the antibodies 25 are displayed in FIGs. 4-7 (SEQ ID NO: 1-33 and 141-144). A comparison of the sequences and binding abilities of the various antibodies was made and the results are displayed in FIGs. 4-10. As can be seen in FIGs. 9A-9L, and FIGs. IOA-10D antibodies 131, 139, and 13.1.2 all demonstrated superior selectivity for EGFRvM expressing cells (H1477) as compared to ABX-EGF. Some of the results are shown in graph form in FIGs. 9M-9P, which demonstrates that at least two of the 30 antibodies, 13.1.2 and 131 demonstrated superior specificity for EGFRvM expressing cells compared to simply EGFRvM cells. Additionally, several possibile utilities for the antibodies of the current embodiment were examined; the results of which are shown in FIGs. 11-16. Finally, based on predicted structural models, variants of the antibodies were made in order to obtain antibodies with altered binding characteristics. 35 Further, antibodies of the invention are highly useful for the screening of other antibodies that bind to the same or similar epitopes. Antibodies of the invention can be utilized in cross -19competition studies for the elucidation of other antibodies that are expected to have the same or improved effects with respect to characteristics of the antigen-antibody complex that is formed. Each of the 131 antibody and the 13.1.2 possessed very high affinities for EGFRvII, were internalized well by cells, and appeared highly effective in cell killing when conjugated to toxins. 5 Intriguingly, both of the antibodies, despite having been generated in different immunizations of XenoMouse mice, and utilizing different technologies, were derived from very similar germline genes. Based upon epitope mapping work, however, each of the antibodies appears to bind to slightly different epitopes on the EGFRvlU molecule and have slightly different residues on EGFRvIII that are essential for binding. These results indicate that the germline gene utilization is 10 of importance to the generation of antibody therapeutics targeting EGFRvIII and that small changes can modify the binding and effects of the antibody in ways that allow for the further design of antibodies and other therapeutics based upon these structural findings. Antibodies that bind to the same epitope as, or compete for binding with, the 13.1.2 and 131 antibodies are highly desirable. As discussed in more detail below, through Alanine scanning 15 . on SPOTs arrays important residues for binding of certain antibodies have been elucidated. Accordingly, antibdodies that share critical binding residues are also highly desirable. Definitions Unless otherwise defined, scientific and technical terms used herein shall have the meanings that are commonly understood by those of ordinary skill in the art. Further, unless 20 otherwise required by context, singular terms shall include pluralities and plural terms shall include the singular. Generally, nomenclatures utilized in connection with, and techniques of, cell and tissue culture, molecular biology, and protein and oligo- or polynucleotide chemistry and hybridization described herein are those well known and commonly used in the art. Standard techniques are used for recombinant DNA, oligonucleotide synthesis, and tissue culture and 25 transformation (e.g., electroporation, lipofection). Enzymatic reactions and purification techniques are performed according to manufacturer's specifications or as commonly accomplished in the art or as described herein. The foregoing techniques and procedures are generally performed according to conventional methods well known in the art and as described in various general and more specific references that are cited and discussed throughout the present specification. See e.g., Sambrook et 30 al. Molecular Cloning: A Laboratory Manual (2d ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y., 1989). The nomenclatures utilized in connection with, and the laboratory procedures and techniques of, analytical chemistry, synthetic organic chemistry, and medicinal and pharmaceutical chemistry described herein are those well known and commonly used in the art. Standard techniques are used for chemical syntheses, chemical analyses, pharmaceutical 35 preparation, formulation, and delivery, and treatment of patients. The term "isolated polynucleotide" as used herein shall mean a polynucleotide of genomic, cDNA, or synthetic origin or some combination thereof, which by virtue of its origin the "isolated -20polynucleotide" (1) is not associated with all or a portion of a polynucleotide in which the "isolated polynucleotide" is found in nature, (2) is operably linked to a polynucleotide which it is not linked to in nature, or (3) does not occur in nature as part of a larger sequence. The term "isolated protein" referred to herein means a protein of cDNA, recombinant RNA, 5 or synthetic origin or some combination thereof, which by virtue of its origin, or source of derivation, the "isolated protein" (1) is not associated with proteins found in nature, (2) is free of other proteins from the same source, e.g. free of murine proteins, (3) is expressed by a cell from a different species, or (4) does not occur in nature. The term "polypeptide" is used herein as a generic term to refer to native protein, 10 fragments, or analogs of a polypeptide sequence. Hence, native protein, fragments, and analogs are species of the polypeptide genus. Preferred polypeptides in accordance with the invention comprise the human heavy chain immunoglobulin molecules and the human kappa light chain immunoglobulin molecules, as well as antibody molecules formed by combinations comprising the heavy chain immunoglobulin molecules with light chain immunoglobulin molecules, such as the 15 kappa light chain immunoglobulin molecules or lambda light chain immunoglobulin molecules, and vice versa, as well as fragments and analogs thereof. The term "naturally-occurring" as used herein as applied to an object refers to the fact that an object can be found in nature. For example, a polypeptide or polynucleotide sequence that is present in an organism (including viruses) that can be isolated from a source in nature and which 20 has not been intentionally modified by man in the laboratory or otherwise is naturally-occurring. The term "operably linked" as used herein refers to positions of components so described are in a relationship permitting them to function in their intended manner. A control sequence "operably linked" to a coding sequence is ligated in such a way that expression of the coding sequence is achieved under conditions compatible with the control sequences. 25 The term "control sequence" as used herein refers to polynucleotide sequences which are necessary to effect the expression and processing of coding sequences to which they are ligated. The nature of such control sequences differs depending upon the host organism; in prokaryotes, such control sequences generally include promoter, ribosomal binding site, and transcription termination sequence; in eukaryotes, generally, such control sequences include promoters and 30 transcription termination sequence. The term "control sequences" is intended to include, at a minimum, all components whose presence is essential for expression and processing, and can also include additional components whose presence is advantageous, for example, leader sequences and fusion partner sequences. The term "polynucleotide" as referred to herein means a polymeric form of nucleotides of 35 at least 10 bases in length, either ribonucleotides or deoxynucleotides or a modified form of either type of nucleotide. The term includes single and double stranded forms of DNA. -21- The term "oligonucleotide" referred to herein includes naturally occurring, and modified nucleotides linked together by naturally occurring, and non-naturally occurring oligonucleotide linkages. Oligonucleotides are a polynucleotide subset generally comprising a length of 200 bases or fewer. Preferably oligonucleotides are 10 to 60 bases in length and most preferably 12, 13, 14, 5 15, 16, 17, 18, 19, or 20 to 40 bases in length. Oligonucleotides are usually single stranded, e.g. for probes; although oligonucleotides may be double stranded, e.g. for use in the construction of a gene mutant. Oligonucleotides of the invention can be either sense or antisense oligonucleotides. The term "naturally occurring nucleotides" referred to herein includes deoxyribonucleotides and ribonucleotides. The term "modified nucleotides" referred to herein includes nucleotides with 10 modified or substituted sugar groups and the like. The term "oligonucleotide linkages" referred to herein includes oligonucleotides linkages such as phosphorothioate, phosphorodithioate, phosphoroselenoate, phosphorodiselenoate, phosphoroanilothioate, phosphoraniladate, phosphoroamidate, and the like. See e.g., LaPlanche et al. Nucl. Acids Res. 14:9081 (1986); Stec et al. J. Am. Chem. Soc. 106:6077 (1984); Stein et al. Nucl. Acids Res. 16:3209 (1988); Zon et al. Anti 15 Cancer Drug Design 6:539 (1991); Zon et al. Oligonucleotides and Analogues: A Practical Approach, pp. 87-108 (F. Eckstein, Ed., Oxford University Press, Oxford England (1991)); Stec et al. U.S. Patent No. 5,151,510; Uhlmann and Peyman Chemical Reviews 90:543 (1990). An oligonucleotide can include a label for detection, if desired. The term "variant" as used herein, is a polypeptide, polynucleotide, or molecule that differs 20 from the recited polypeptide or polynucleotide, but only such that the activity of the protein is not detrimentally altered. There may be variants of epitopes. There may be variants of antibodies. In a preferred embodiment, the ability of a protein variant to bind to the epitope is not detrimentally altered. In one embodiment, the protein variant can bind with 10-500% of the ability of the wild type mAb. For example, the protein variant can bind with 10%, 50%, 110%, 500%, or greater than 25 500% of the ability of the wild type mAb. In one embodiment, the range of binding abilities between 10-500% is inlcuded. Binding ability may be reflected in many ways, including, but not limited to the ka, kd, or KD of the variant to an epitope. In one preferred embodiment, the epitope is one described in the present specification. In one embodiment, variant antibodies can differ from the wild-type sequence by 30 substitution, deletion or addition of five amino acids or fewer. Such variants may generally be identified by modifying one of the disclosed polypeptide sequences, and evaluating the binding properties of the modified polypeptide using, for example, the representative procedures described herein. In another embodiment, polypeptide variants preferably exhibit at least about 70%, more preferably at least about 90% and most preferably at least about 95% identity to the identified 35 polypeptides. Preferrably, the variant differs only in conservative substitutions and/or modifications. Variant proteins include those that are structurally similar and those that are functionally equivalent to the protein structures described in the present specification. In another -22embodiment, the protein is a variant if it is functionally equivalent to the proteins described in this specification, so long as the paratope of variant is similar to the paratopes described in the specification. In one embodiment, any substance with a shape that is similar to the paratope described in FIG. 17 is a variant. In one embodiment, any substance with a shape that is similar to 5 the paratope described in FIG. 18 is a variant. In one embodiment, any substance that has a shape that is similar to the interaction surface described in FIG. 19A and 19B is a variant. In one embodiment, the antibody is a variant if the nucleic acid sequence can selectively hybridize to wild-type sequence under stringent conditions. In one embodiment, suitable moderately stringent conditions include prewashing in a solution of 5xSSC; 0.5% SDS, 1.0 mM 10 EDTA (pH 8:0); hybridizing at 50*C-65*C, 5xSSC, overnight or, in-the event of cross-species homology, at 45*C with 0.5xSSC; followed by washing twice at 65*C for 20 minutes with each of 2x, 0.5x and 0.2xSSC containing 0.1% SDS. Such hybridizing DNA sequences are also within the scope of this invention, as are nucleotide sequences that, due to code degeneracy, encode an antibody polypeptide that is encoded by a hybridizing DNA sequence.The term "selectively 15 hybridize" referred to herein means to detectably and specifically bind. Polynucleotides, oligonucleotides and fragments thereof in accordance with the invention selectively hybridize to nucleic acid strands under hybridization and wash conditions that minimize appreciable amounts of detectable binding to nonspecific nucleic acids. High stringency conditions can be used to achieve selective hybridization conditions as known in the art and discussed herein. Generally, the nucleic 20 acid sequence homology between the polynucleotides, oligonucleotides, and fragments of the invention and a nucleic acid sequence of interest will be at least 80%, and more typically with preferably increasing homologies of at least 85%, 90%, 95%, 99%, and 100%. Two amino acid sequences are homologous if there is a partial or complete identity between their sequences. For example, 85% homology means that 85% of the amino acids are identical when the two sequences 25 are aligned for maximum matching. Gaps (in either of the two sequences being matched) are allowed in maximizing matching; gap lengths of 5 or less are preferred with 2 or less being more preferred. Alternatively and preferably, two protein sequences (or polypeptide sequences derived from them of at least 30 amino acids in length) are homologous, as this term is used herein, if they have an alignment score of at more than 5 (in standard deviation units) using the program ALIGN 30 with the mutation data matrix and a gap penalty of 6 or greater. See Dayhoff, M.O., in Atlas of Protein Sequence and Structure, pp. 101-110 (Volume 5, National Biomedical Research Foundation (1972)) and Supplement 2 to this volume, pp. 1-10. The two sequences or parts thereof are more preferably homologous if their amino acids are greater than or equal to 50% identical when optimally aligned using the ALIGN program. The term "corresponds to" is used herein to 35 mean that a polynucleotide sequence is homologous (i.e., is identical, not strictly evolutionarily related) to all or a portion of a reference polynucleotide sequence, or that a polypeptide sequence is identical to a reference polypeptide sequence. In contradistinction, the term "complementary to" is -23used herein to mean that the complementary sequence is homologous to all or a portion of a reference polynucleotide sequence. For illustration, the nucleotide sequence "TATAC" corresponds to a reference sequence "TATAC" and is complementary to a reference sequence "GTATA". The following terms are used to describe the sequence relationships between two or more 5 polynucleotide or amino acid sequences: "reference sequence", "comparison window", "sequence identity", "percentage of sequence identity", and "substantial identity". A "reference sequence" is a defined sequence used as a basis for a sequence comparison; a reference sequence may be a subset of a larger sequence, for example, as a segment of a full-length cDNA or gene sequence given in a sequence listing or may comprise a completed cDNA or gene sequence. Generally, a reference 10 sequence is at least 18 nucleotides or 6 amino acids in length, frequently at least 24 nucleotides or 8 amino acids in length, and often at least 48 nucleotides or 16 amino acids in length. Since two polynucleotides or amino acid sequences may each (1) comprise a sequence (i.e., a portion of the complete polynucleotide or amino acid sequence) that is similar between the two molecules, and (2) may further comprise a sequence that is divergent between the two polynucleotides or amino acid 15 sequences, sequence comparisons between two (or more) molecules are typically performed by comparing sequences of the two molecules over a "comparison window" to identify and compare local regions of sequence similarity. A "comparison window", as used herein, refers to a conceptual segment of at least 18 contiguous nucleotide positions or 6 amino acids wherein a polynucleotide sequence or amino acid sequence may be compared to a reference sequence of at 20 least 18 contiguous nucleotides or 6 amino acid sequences and wherein the portion of the polynucleotide sequence in the comparison window may comprise additions, deletions, substitutions, and the like (i.e., gaps) of 20 percent or less as compared to the reference sequence (which does not comprise additions or deletions) for optimal alignment of the two sequences. Optimal alignment of sequences for aligning a comparison window may be conducted by the local 25 homology algorithm of Smith and Waterman Adv. Apple. Math. 2:482 (1981), by the homology alignment algorithm of Needleman and Wunsch J. Mol. Biol. 48:443 (1970), by the search for similarity method of Pearson and Lipman Proc. Natl. Acad. Sci. (U.S.A.) 85:2444 (1988), by computerized implementations of these algorithms (GAP, BESTFIT, FASTA, and TFASTA in the Wisconsin Genetics Software Package Release 7.0, (Genetics Computer Group, 575 Science Dr., 30 Madison, Wis.), Geneworks, or MacVector software packages), or by inspection, and the best alignment (i.e., resulting in the highest percentage of homology over the comparison window) generated by the various methods is selected. The term "sequence identity" means that two polynucleotide or amino acid sequences are identical (i.e., on a nucleotide-by-nucleotide or residue-by-residue basis) over the comparison 35 window. The term "percentage of sequence identity" is calculated by comparing two optimally aligned sequences over the window of comparison, determining the number of positions at which the identical nucleic acid base (e.g., A, T, C, G, U, or I) or residue occurs in both sequences to yield -24the number of matched positions, dividing the number of matched positions by the total number of positions in the comparison window (i.e., the window size), and multiplying the result by 100 to yield the percentage of sequence identity. The terms "substantial identity" as used herein denotes a characteristic of a polynucleotide or amino acid sequence, wherein the polynucleotide or amino acid 5 comprises a sequence that has at least 85 percent sequence identity, preferably at least 90 to 95 percent sequence identity, more usually at least 99 percent sequence identity as compared to a reference sequence over a comparison window of at least 18 nucleotide (6 amino acid) positions, frequently over a window of at least 24-48 nucleotide (8-16 amino acid) positions, wherein the percentage of sequence identity is calculated by comparing the reference sequence to the sequence 10 which may include deletions or additions which total 20 percent or less of the reference sequence over the comparison window. The reference sequence may be a subset of a larger sequence. Amino acids or nucleic acids with substantial identity to the wild-type protein or nucleic acid are examples of variants of the wild-type protein or nucleic acid. As used herein, the twenty conventional amino acids and their abbreviations follow 15 conventional usage. See Immunology - A Synthesis (2" Edition, E.S. Golub and D.R. Gren, Eds., Sinauer Associates, Sunderland, Mass. (1991)). Stereoisomers (e.g., D-amino acids) of the twenty conventional amino acids, unnatural amino acids such as a-, a-disubstituted amino acids, N-alkyl amino acids, lactic acid, and other unconventional amino acids may also be suitable components for polypeptides of the present invention. Examples of unconventional amino acids include: 4 20 hydroxyproline, y-carboxyglutamate, e-N,N,N-trimethyllysine, e-N-acetyllysine, 0-phosphoserine, N-acetylserine, N-formylmethionine, 3-methylhistidine, 5-hydroxylysine, a-N-methylarginine, and other similar amino acids and imino acids (e.g., 4-hydroxyproline). In the polypeptide notation used herein, the left-hand direction is the amino terminal direction and the right-hand direction is the carboxy-terminal direction, in accordance with standard usage and convention. 25 Similarly, unless specified otherwise, the left-hand end of single-stranded polynucleotide sequences is the 5' end; the left-hand direction of double-stranded polynucleotide sequences is referred to as the 5' direction. The direction of 5' to 3' addition of nascent RNA transcripts is referred to as the transcription direction; sequence regions on the DNA strand having the same sequence as the RNA and which are 5' to the 5' end of the RNA transcript are referred to as 30 "upstream sequences"; sequence regions on the DNA strand having the same sequence as the RNA and which are 3' to the 3' end of the RNA transcript are referred to as "downstream sequences". As applied to polypeptides, the term "substantial identity" means that two peptide sequences, when optimally aligned, such as by the programs GAP or BESTFIT using default gap weights, share at least 80 percent sequence identity, preferably at least 90 percent sequence identity, 35 more preferably at least 95 percent sequence identity, and most preferably at least 99 percent sequence identity. Preferably, residue positions which are not identical differ by conservative amino acid substitutions. Conservative amino acid substitutions refer to the interchangeability of -25residues having similar side chains. For example, a group of amino acids having aliphatic side chains is glycine, alanine, valine, leucine, and isoleucine; a group of amino acids having aliphatic hydroxyl side chains is serine and threonine; a group of amino acids having amide-containing side chains is asparagine and glutamine; a group of amino acids having aromatic side chains is 5 phenylalanine, tyrosine, and tryptophan; a group of amino acids having basic side chains is lysine, arginine, and histidine; and a group of amino acids having sulfur-containing side chains is cysteine and methionine. Preferred conservative amino acids substitution groups are: valine-leucine isoleucine, phenylalanine-tyrosine, lysine-arginine, alanine-valine, glutamic-aspartic, and asparagine-glutamine. Polypeptides with substantial identity can be variants. 10 Variant proteins also include proteins with minor variations. As discussed herein, minor variations in the amino acid sequences of antibodies or immunoglobulin molecules are contemplated as being encompassed by the present invention, providing that the variations in the amino acid sequence maintain at least 75%, more preferably at least 80%, 90%, 95%, and most preferably 99%. In particular, conservative amino acid replacements are contemplated. 15 Conservative replacements are those that take place within a family of amino acids that are related in their side chains. Genetically encoded amino acids are generally divided into families: (1) acidic=aspartate, glutamate; (2) basic=lysine, arginine, histidine; (3) non-polar=alanine, valine, leucine, isoleucine, praline, phenylalanine, methionine, tryptophan; and (4) uncharged polar-glycine, asparagine, glutamine, cysteine, serine, threonine, tyrosine. More preferred families 20 are: serine and threonine are aliphatic-hydroxy family; asparagine and glutamine are an amide containing family; alanine, valine, leucine and isoleucine are an aliphatic family; and phenylalanine, tryptophan, and tyrosine are an aromatic family. For example, it is reasonable to expect that an isolated replacement of a leucine with an isoleucine or valine, an aspartate with a glutamate, a threonine with a serine, or a similar replacement of an amino acid with a structurally 25 related amino acid will not have a major effect on the binding or properties of the resulting molecule, especially if the replacement does not involve an amino acid within a framework site. Whether an amino acid change results in a functional peptide can readily be determined by assaying the specific activity of the polypeptide derivative. Assays are described in detail herein. Fragments or analogs of antibodies or immunoglobulin molecules can be readily prepared by those of ordinary 30 skill in the art. Preferred amino- and carboxy-termini of fragments or analogs occur near boundaries of functional domains. Structural and functional domains can be identified by comparison of the nucleotide and/or amino acid sequence data to public or proprietary sequence databases. Preferably, computerized comparison methods are used to identify sequence motifs or predicted protein conformation domains that occur in other proteins of known structure and/or 35 function. Methods to identify protein sequences that fold into a known three-dimensional structure are known. Bowie et al. Science 253:164 (1991). Thus, the foregoing examples demonstrate that -26those of skill in the art can recognize sequence motifs and structural conformations that may be used to define structural and functional domains in accordance with the antibodies described herein. Preferred amino acid substitutions are those which: (1) reduce susceptibility to proteolysis, (2) reduce susceptibility to oxidation, (3) alter binding affinity for. forming protein complexes, (4) 5 alter binding affinities, and (4) confer or modify other physicochemical or functional properties of such analogs. Analogs can include various muteins of a sequence other than the naturally-occurring peptide sequence. For example, single or multiple amino acid substitutions (preferably conservative amino acid substitutions) may be made in the naturally-occurring sequence (preferably in the portion of the polypeptide outside the domain(s) forming intermolecular contacts. A 10 conservative amino acid substitution should not substantially change the structural characteristics of the parent sequence (e.g., a replacement amino acid should not tend to break a helix that occurs in the parent sequence, or disrupt other types of secondary structure that characterizes the parent sequence). Examples of art-recognized polypeptide secondary and tertiary structures are described in Proteins, Structures and Molecular Principles (Creighton, Ed., W. H. Freeman and Company, 15 New York (1984)); Introduction to Protein Structure (C. Branden and J. Tooze, eds., Garland Publishing, New York, N.Y. (1991)); and Thornton et at. Nature 354:105 (1991). The term "polypeptide fragment" as used herein refers to a polypeptide that has an amino terminal and/or carboxy-terminal deletion, but where the remaining amino acid sequence is identical to the corresponding positions in the naturally-occurring sequence deduced, for example, 20 from a full-length cDNA sequence. Fragments typically are at least 5, 6, 8 or 10 amino acids long, preferably at least 14 amino acids long, more preferably at least 20 amino acids long, usually at least 50 amino acids long, and even more preferably at least 70 amino acids long. The term "analog" as used herein refers to polypeptides which are comprised of a segment of at least 25 amino acids that has substantial identity to a portion of a deduced amino acid sequence. Analogs 25 typically are at least 20 amino acids long, preferably at least 50 amino acids long or longer, and can often be as long as a full-length naturally-occurring polypeptide. Both fragments and analogs are forms of variants Peptide analogs are commonly used in the pharmaceutical industry as non-peptide drugs with properties analogous to those of the template peptide. These types of non-peptide compound 30 are termed "peptide mimetics" or "peptidomimetics". Fauchere, J. Adv. Drug Res. 15:29 (1986); Veber and Freidinger TINS p.392 (1985); and Evans et al. J. Med. Chem. 30:1229 (1987). Such compounds are often developed with the aid of computerized molecular modeling. Peptide mimetics that are structurally similar to therapeutically useful peptides may be used to produce an equivalent therapeutic or prophylactic effect. Generally, peptidomimetics are structurally similar to 35 a paradigm polypeptide (i.e., a polypeptide that has a biochemical property or pharmacological activity), such as human antibody, but have one or more peptide linkages optionally replaced by a linkage selected from the group consisting of: --CH 2 NH--, --CH 2 S--, --CH 2

-CH

2 --, --CH=CH--(cis -27and trans), -COCH 2 -, -CH(OH)CH 2 -, and -CH 2 SO-, by methods well known in the art. Systematic substitution of one or more amino acids of a consensus sequence with a D-amiino acid of the same type (e.g., D-lysine in place of L-lysine) may be used to generate more stable peptides. In addition, constrained peptides comprising a consensus sequence or a substantially identical 5 consensus sequence variation may be generated by methods known in the art (Rizo and Gierasch Ann. Rev. Biochem. 61:387 (1992)); for example, by adding internal cysteine residues capable of forming intramolecular disulfide bridges which cyclize the peptide. Peptide rnimetics and peptidomimetics are both forms of variants. "Antibody" or "antibody peptide(s)" refer to an intact antibody, or a binding fragment 10 thereof that competes with the intact antibody for specific binding. Binding fragments are produced by recombinant DNA techniques, or by enzymatic or chemical cleavage of intact antibodies. Binding fragments include Fab, Fab', F(ab') 2 , Fv, and single-chain antibodies. An antibody other than a "bispecific" or "bifunctional" antibody is understood to have each of its binding sites identical. An antibody substantially inhibits adhesion of a receptor to a counterreceptor when an 15 excess of antibody reduces the quantity of receptor bound to counterreceptor by at least about 20%, 40%, 60% or 80%, and more usually greater than about 85% (as measured in an in vitro competitive binding assay). The term "epitope" includes any protein determinant capable of specific binding to an immunoglobulin or T-cell receptor or otherwise interacting with a molecule. Epitopic determinants 20 generally consist of chemically active surface groupings of molecules such as amino acids or carbohydrate or sugar side chains and generally have specific three-dimensional structural characteristics, as well as specific charge characteristics. An epitope may be "linear" or "conformational." In a linear epitope, all of the points of interaction between the protein and the interacting molecule (such as an antibody) occur linearally along the primary amino acid sequence 25 of the protein. In a conformational epitope, the points of interaction occur across amino acid residues on the protein that are separated from one another. An antibody is said to specifically bind an antigen when the dissociation constant is sl pM, preferably s 100 nM and more preferably 10 nM, and even more preferably s InM. Once a desired epitope on an antigen is determined, it is possible to generate antibodies to that epitope, e.g., using the techniques described in the present 30 invention. Alternatively, during the discovery process, the generation and characterization of antibodies may elucidate information about desirable epitopes. From this information, it is then possible to competitively screen antibodies for binding to the same epitope. An approach to achieve this is to conduct cross-competition studies to find antibodies that competively bind with one another, e.g., the antibodies compete for binding to the antigen. A high throughput process for 35 "binning" antibodies based upon their cross-competition is described in International Patent Application No. WO 03/48731. As will be appreciated by one of skill in the art, practically anything to which an antibody can specificially bind could be an epitope. An epitope can -28comprises those residues to which the antibody binds. In one embodiment, the epitope is the EGFRvIII epitope. In a more preferred embodiment, the epitope is that described in Example 4 of this specification. In one embodiment, the epitope is the epitope described in Example 14. In one embodiment, the epitope comprises the sequence LEEKKGNYVVTD (SEQ ID NO: 59). In one 5 embodiment, the epitope comprises the sequence EEKKGNYVVT (SEQ ID NO: 57). In one embodiment, the epitope comprises the sequence EKNY (SEQ ID NO: 60). In one embodiment, the epitope comprises the sequence EEKGN (SEQ ID NO: 61). One of skill in the art will appreciate that these need not be actually assembled in this order on a single peptide, rather, these are the residues that form the eptiope which interacts with the paratope. As will be appreciated by 10 one of skill in the art, the space that is occupied by a residue or side chain that creates the shape of a molecule helps to determine what an epitope is. Likewise, any functional groups associated with the epitope, van der Waals interactions, degree of mobility of side chains, etc. can all determine what an epitope actually is. Thus an epitope may also include energetic interactions. The term "paratope" is meant to describe the general structure of a binding region that 15 determines binding to an epitope. This structure influences whether or not and in what manner the binding region might bind to an epitope. Paratope can refer to an antigenic site of an antibody that is responsible for an antibody or fragment thereof, to bind to an antigenic determinant. Paratope also refers to the idiotope of the antibody, and the complementary determining region (CDR) region that binds to the epitope. In one embodiment, the paratope is the region of the antibody that is LI 20 10, L2 30, L3 50, Hi 20, H2 40, and H3 60 in FIG. 17. In one embodiment, the paratope is the region of the antibody that comprises the CDR sequences in Example 16 for LI, L2, L3, H1, H2, and H3. In one embodiment, the paratope is the region of the antibody that is L1 110, L2 130, L3 150, H1 120, H2 140, and H3 160 in FIG. 18, In one embodiment, the paratope is the region of the antibody that comprises the CDR sequences in Example 18 for Li, L2, L3, HI, H2, and H3. In one 25 embodiment, the paratope comprises the sequences listed in Example 18. In one embodiment, the paratope comprises the residues that interact with the epitope, as shown in FIG. 19A and FIG. 19B. The solid black structure is the peptide structure. In one embodiment, the paratope comprises residue Tyr172Arg of the 13.1.2 mAb. In one embodiment, the paratope of the 13.1.2 mAb comprises at least one residue selected from the group consisting of: Tyr 172Arg, Argl01Glu, 30 Leu99Asn, Leu99His, ArglOlAsp, Leu2l7G]n, Leu99Thr, Leu2I7Asn, Argl01Gln, and Asn35Gly. As will be appreciated by one of skill in the art, the paratope of any antibody, or variant thereof, can be determined in the manner set forth by the present application. Residues are considered "important" if they are predicted to contribute 10% of the binding energy. In one embodiment, residues are considered "important" if they are predicted to contribute 2% of the binding energy. In 35 one embodiment, residues are considered "important" if they are predicted to contribute 50% of the binding energy. In one embodiment, residues are considered "important" if they are predicted to -29interact with the surface of the epitope, or the surface of the paratope. In one embodiment, residues are considered "important" if changing the residue results in a loss in binding. The terms "specifically" or "preferrentially" binds to, or similar phrases are not meant to denote that the antibody exclusively binds to that epitope. Rather, what is meant is that the 5 antibody, or variant thereof, can bind to that epitope, to a higher degree than the antibody binds to at least one other substance to which the antibody is exposed to. In one embodiment, the specifically binding antibody will bind to the EGFRvm protein with an affinity greater than (more tightly, or lower KD) it will to the EGFR protein. For example, the specifically binding antibody will bind more tightly by at least a minimal increase to 1, 1-2, 2-5, 5-10, 10-20, 20-30, 30-50, 50 10 70, 70-90, 90-120, 120-150, 150-200, 200-300, 300-500, 500-1000 percent or more. The shorthand of amino acid, number, amino acid, e.g., Leu217GIn, denotes a mutation at the numbered amino acid, from the first amino acid, to the second amino acid. Thus, Tyrl72Arg would mean that while the wild type protein has a tyrosine at position 172, the mutant has an arginine at position 172. 15 The term "agent" is used herein to denote a chemical compound, a mixture of chemical compounds, a biological macromolecule, or an extract made from biological materials. "Mammal" when used herein refers to any animal that is considered a mammal. Preferably, the mammal is human. Digestion of antibodies with the enzyme, papain, results in two identical antigen-binding 20 fragments, known also as "Fab" fragments, and a "Fc" fragment, having no antigen-binding activity but having the ability to crystallize. Digestion of antibodies with the enzyme, pepsin, results in the a F(ab') 2 fragment in which the two arms of the antibody molecule remain linked and comprise two-antigen binding sites. The F(ab') 2 fragment has the ability to crosslink antigen. "Fv" when used herein refers to the minimum fragment of an antibody that retains both 25 antigen-recognition and antigen-binding sites. These fragments can also be considered variants of the antibody. "Fab" when used herein refers to a fragment of an antibody which comprises the constant domain of the light chain and the C141 domain of the heavy chain. The term "mAb" refers to monoclonal antibody. 30 The description of XenoMax method generated antibody sequences is coded as follows: "AB"-referring to antibody, "EGFRvII"-referring to antibody's binding specificity, "X" referring to XenoMouse mouse derived, "G 1 "-referring to IgG1 isotype or "G2" referring to IgG2 isotype, the last three digits refer to the single cell number from which the antibody was derived, for example: AB- EGFRvII -XGl-095 would be an antibody with binding specificity to EGFRvfII 35 from XenoMouse mouse of a IgG1 isotype and cell number 95. -30- The term "SC" refers to single cell and a particular XenoMax method derived antibody may be referred to as SC followed by three digits, or just three digits, referring to the single cell number from which the antibody was derived herein. The description of hybridoma derived antibody sequences is coded as follows: "AB" 5 referring to antibody, "EGFRvII"-refers to the antibody's binding specificity, "X" refers to XenoMouse mouse derived, "Gl"-refers to IgGI isotype or "G2" refers to IgG2 isotype, "K" refers to kappa, "L' refers to lambda. The-last three digits referring to the clone from which the antibody was derived, for example: AB-EGFRvfl-XGlK-13.l.2 "Label" or "labeled" as used herein refers to the addition of a detectable moiety to a 10 polypeptide, for example, a radiolabel, fluorescent label, enzymatic label chemiluminescent labeled or a biotinyl group. Radioisotopes or radionuclides may include 'H, 1 4 C, "N, "S, 9Y, "Tc, "'In, "sL 1n, fluorescent labels may include rhodamine, lanthanide phosphors or FITC and enzymatic labels may include horseradish peroxidase, P-galactosidase, luciferase, alkaline phosphatase. The term "pharmaceutical agent or drug" as used herein refers to a chemical compound or 15 composition capable of inducing a desired therapeutic effect when properly administered to a patient. Other chemistry terms herein are used according to conventional usage in the art, as exemplified by The McGraw-Hill Dictionary of Chemical Terms (Parker, S., Ed., McGraw-Hill, San Francisco (1985)). As used herein, "substantially pure" means an object species is the predominant species 20 present (i.e., on a molar basis it is more abundant than any other individual species in the composition), and preferably a substantially purified fraction is a composition wherein the object species comprises at least about 50 percent (on a molar basis) of all macromolecular species present. Generally, a substantially pure composition will comprise more than about 80 percent of all macromolecular species present in-the composition, more preferably more than about 85%, 90%, 25 95%, 99%, and 99.9%. Most preferably, the object species is purified to essential homogeneity (contaminant species cannot be detected in the composition by conventional detection methods) wherein the composition consists essentially of a single macromolecular species. The term "patient" includes human and veterinary subjects. The term "SLAM* Technology" refers to the "Selected Lymphocyte Antibody Method" 30 (Babcook et al., Proc. Natl. Acad. Sci. USA, i93:7843-7848 (1996), and Schrader, US Patent No. 5,627,052). The term "XenoMaxTM" refers to'the use of SLAM Technology with XenoMouse* mice (as described below). Antibody Structure 35 The basic antibody structural unit is known to comprise a tetramer. Each tetramer is composed of two identical pairs of polypeptide chains, each pair having one "light" (about 25 kDa) and one "heavy" chain (about 50-70 kDa). The amino-terminal portion of each chain includes a -31variable region of about 100 to 110 or more amino acids primarily responsible for antigen recognition. The carboxy-terminal portion of each chain defines a constant region primarily responsible for effector function. Human light chains are classified as kappa and lambda light chains. Heavy chains are classified as mu, delta, gamma, alpha, or epsilon, and define the 5 antibody's isotype as IgM, IgD, IgG, IgA, and IgE, respectively. Within light and heavy chains, the variable and constant regions are joined by a "J" region of about 12 or more amino acids, with the heavy chain also including a "D" region of about 10 more amino acids. See generally, Fundamental Immunology Ch. 7 (Paul, W., ed., 2nd ed. Raven Press, N.Y. (1989)). The variable regions of each light/heavy chain pair form the antibody binding site. 10 Thus, an intact antibody has two binding sites. Except in bifunctional or bispecific antibodies, the two binding sites are the same. The chains all exhibit the same general structure of relatively conserved framework regions (FR) joined by three hyper variable regions, also called complementarity determining regions or CDRs. The CDRs from the two chains of each pair are aligned by the framework regions, enabling 15 binding to a specific epitope. From N-terminal to C-terminal, both light and heavy chains comprise the domains FRI, CDRI, FR2, CDR2, FR3, CDR3 and FR4. The assignment of amino acids to each domain is in accordance with the definitions of Kabat Sequences ofProteins ofImmunological Interest (National Institutes of Health, Bethesda, Md. (1987 and 1991)), or Chothia & Lesk J. Mol. Biol. 196:901-917 (1987); Chothia et al. Nature 342:878-883 (1989). 20 A bispecific or bifunctional antibody is an artificial hybrid antibody having two different heavy/light chain pairs and two different binding sites. Bispecific antibodies can be produced by a variety of methods including fusion of hybridomas or linking of Fab' fragments. See, e.g., Songsivilai & Lachmann Clin. Exp. Immunol. 79: 315-321 (1990), Kostelny et al. J. Immunol. 148:1547-1553 (1992). Production of bispecific antibodies can be a relatively labor intensive 25 process compared with production of conventional antibodies and yields and degree of purity are generally lower for bispecific antibodies. Bispecific antibodies do not exist in the form of fragments having a single binding site (e.g., Fab, Fab', and Fv). In addition to the general structural aspects of antibodies, the more specific interaction between the paratope and the epitope may be examined through structural approaches. In one 30 embodiment, the structure of the CDRs form a paratope, through which an antibody is able to bind to an epitope. The structure of such a paratope may be determined in a number of ways. Traditional structural examination approaches may be used, such as NMR or x-ray crystalography. These approaches may examine the structure of the paratope alone, or while it is bound to the epitope. Alternatively, molecular models may be generated in silico. A structure can be generated 35 through homology modeling, aided with a commercial package, such as InsightII modeling package from Accelrys (San Diego, CA). Briefly, one can use the sequence of the antibody to be examined to search against a database of proteins of known structures, such as the Protein Data Bank. After -32one identifies homologous proteins with known structures, these homologous proteins are used as modeling templates. Each of the possible templates can be aligned, thus producing structure based sequence alignments amoung the templates. The sequence of the antibody with the unknown structure can then be aligned with these templates to generate a molecular model for the antibody 5 with the unknown structure. As will be appreciated by one of skill in the art, there are many alternative methods for generating such structures in silico, any of which may be used. For instance, a process similar to the one described in Hardman et al., issued U.S. Pat. No. 5,958,708 employing QUANTA (Polygen Corp., Waltham, Mass.) and CHARM (Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S. and Karplus, M., 1983, J. Comp. Chem,. 10 4:187) may be used. Not only is the shape of the paratope important in determining whether and how well a possible paratope will bind to an epitope, but the interaction itself, between the epitope and the paratope is a source of great information in the design of variant antibodies. As appreciated by one of skill in the art, there are a variety of ways in which this interaction can be studied. One way is to 15 use the structural model generated, perhaps as described above, and then to use a program such as InsightlI (Accelrys, San Diego, CA), which has a docking module, which, among other things, is capable of performing a Monte Carlo search on the conformational and orientational spaces between the paratope and its epitope. The result is that one is able to estimate where and how the epitope interacts with the paratope. In one embodiment, only a fragment, or variant, of the epitope 20 is used to assist in determining the relevant interactions. In one embodiment, the entire epitope is used in the modeling of the interaction between the paratope and the epitope. As will be appreciated by one of skill in the art, these two different approaches have different advantages and disadvantages. For instance, using only a fragment of the epitope allows for a more detailed examination of the possible variations of each side chain, without taking huge amounts of time. On 25 the other hand, by using only a fragment of the epitope, or simply the epitope instead of the entire protein, it is possible that the characteristics of the epitope fragment may not be the same as the characteristics for the whole epitope, thus possibly increasing the risk of being mislead during the computational modeling. In one embodiment, both approaches are used to a limited extent, in order to cross check the results. In a preferred embodiment, if a variant of an epitope is used, it will be 30 optimized so that the variant of the epitope comprises the most important residues of the epitope. The identity of the most important residues can be determined in any number of ways, for instance as described in Examples 4 and 14 of the present specification. Through the use of these generated structures, one is able to determine which residues are the most important in the interaction between the epitope and the paratope. Thus, in one 35 embodiment, one is able to readily select which residues to change in order to alter the binding characteristics of the antibody. For instance, it may be apparent from the docking models that the side chains of certain residues in the paratope may sterically hinder the binding of the epitope, thus -33altering these residues to residues with smaller side chains may be beneficial. One can determine this in many ways. For example, one may simply look at the two models and esitmate interactions based on functional groups and proximety. Alternatively, one may perform repeated pairings of epitope and paratope, as described above, in order to obtain more favorable energy interactions. 5 One can also determine these interactions for a variety of variants of the antibody to determine alternative ways in which the antibody may bind to the epitope. One can also combine the various models to determine how one should alter the structure of the antibodies in order to obtain an antibody with the particular characteristics that are desired. The models determined above can be tested through various techniques. For example, the 10 interaction energy can determined with the programs discussed above in order to determine which of the variants to further examine. Also, coulumbic and van der Waals interactions are used to determine the interaction energies of the epitope and the variant paratopes. Also site directed mutagenesis is used to see if predicted changes in antibody structure actually result in the desired changes in binding characteristics. Alternatively, changes may be made to the epitope to verify that 15 the models are correct or to determine general binding themes that may be occurring between the paratope and the epitope. The above methods for modeling structures can be used to determine what changes in protein structure will result in particular desired characteristics of an antibody. These methods can be used to determine what changes in protein structure will not result in the desired characteristics. 20 As will be appreciated by one of skill in the art, while these models will provide the guidance necessary to make the antibodies and variants thereof of the present embodiments, it may still be desirable to perform routine testing of the in silico models, perhaps through in vitro studies. In addition, as will be apparent to one of skill in the art, any modification may also have additional side effects on the activity of the antibody. For instance, while any alteration predicted to result in 25 greater binding, may induce greater binding, it may also cause other structural changes which might reduce or alter the activity of the antibody. The determination of whether or not this is the case is routine in the art and can be achieved in many ways. For Example, the activity can be tested through an ELISA test, as in Example 21. Alternatively, the samples can be tested through the use of a surface plasmon resonance device. 30 Antibodies Binding, and Variant Antibodies for Superior Binding In one embodiment, the models described above are used to increase the binding ability of the antibody to its epitope. The antibody can bind to the epitope more readily, and thus have a higher association constant (k.). Alternatively, the antibody may dissociate from the epitope slower, and thus have a lower dissociation constant (kd), or the KD of the epitope-paratope 35 interaction can be smaller in value, thus making the extent of the binding between the epitope and paratope higher. -34- In some embodiments, the variant antibodies are designed to bind with the opposite characteristics. That is, the antibodies do not bind as tightly or perhaps as quickly. In other embodiments, the variant antibodies are not different in their KD from the wild type antibodies, but the variant antibodies are more specific for a particular epitope. This may 5 mean that the paratopes of the designed antibodies have a lower risk of binding to other epitopes. The antibodies can have other characteristics that have been altered. For example, a variant may be more immune to nonspecific antibody binding or may stay solvated in solution even when the antibody is present in high concentrations. Such a variant may be present in the discussed antibodies. For instance, while the higher concentrations of some variant antibodies examined 10 below resulted in slower binding components in Biacore experiments, for instance 13.1.2 mAb, certain variants did not exhibit this slower component, even at the same concentrations, L217N-2. 1, for example. The variants predicted by the models determined above can be created and then tested to determine if they actually bind with the desired characteristics. Mutants with a greater total 15 interaction energy with the epitope can be selected for further testing. The interaction energy can be determined in a number of ways, one of which is described above. These variants can be tested in a number of ways. Exemplary options include and are not limited to KinExA (e.g., Lackie, Issued Pat. No. 5,372,783, Dec 13, 1994) (Sapidyne Instruments Inc., ID, Boise), surface plasmon resonance (SPR)(e.g., BIACORET" Biacore, Inc., Pistcataway, 20 N.J.), stopped-flow fluorescence, resonant mirror, and fluorescence polarization. Many of these options are able to not only record the data, but also provide ready means for fitting the data to various theoretical curves and thus determine the k., kd, and Kr, as well as other properties. It is important to note that the fitting of these curves to the resulting data is not without the possibility for some variation. Because of this, the relevant association, dissociation, and equilibrium 25 constants can be looked at, not only through these curve fitting mechanisms, but also in direct comparison with each other, and in light of the knowledge of one of skill in the art. Human Antibodies and Humanization of Antibodies Human antibodies avoid some of the problems associated with antibodies that possess murine or rat variable and/or constant regions. The presence of such murine or rat derived proteins 30 can lead to the rapid clearance of the antibodies or can lead to the generation of an immune response against the antibody by a patient. In order to avoid the utilization of murine or rat derived antibodies, fully human antibodies can be generated through the introduction of human antibody function into a rodent so that the rodent produces fully human antibodies. The ability to clone and reconstruct megabase-sized human loci in YACs and to introduce 35 them into the mouse germline provides a powerful approach to elucidating the functional components of very large or crudely mapped loci as well as generating useful models of human disease. Furthermore, the utilization of such technology for substitution of mouse loci with their -35human equivalents could provide unique insights into the expression and regulation of human gene products during development, their communication with other systems, and their involvement in disease induction and progression. An important practical application of such a strategy is the "humanization" of the mouse 5 humoral immune system. Introduction of human immunoglobulin (Ig) loci into mice in which the endogenous Ig genes have been inactivated offers the opportunity to study the mechanisms underlying programmed expression and assembly of antibodies as well as their role in B-cell development. Furthermore, such a strategy could provide an ideal source for production of fully human monoclonal antibodies (mAbs)--an important milestone towards fulfilling the promise of 10 antibody therapy in human disease. Fully human antibodies are expected to minimize the immunogenic and allergic responses intrinsic to mouse or mouse-derivatized mAbs and thus to increase the efficacy and safety of the administered antibodies. The use of fully human antibodies can be expected to provide a substantial advantage in the treatment of chronic and recurring human diseases, such as inflammation, autoimmunity, and cancer, which require repeated antibody 15 administrations. One approach towards this goal was to engineer mouse strains deficient in mouse antibody production with large fragments of the human Ig loci in anticipation that such mice would produce a large repertoire of human antibodies in the absence of mouse antibodies. Large human Ig fragments would preserve the large variable gene diversity as well as the proper regulation of 20 antibody production and expression. By exploiting the mouse machinery for antibody diversification and selection and the lack of immunological tolerance to human proteins, the reproduced human antibody repertoire in these mouse strains should yield high affinity antibodies against any antigen of interest, including human antigens. Using the hybridoma technology, antigen-specific human mAbs with the desired specificity could be readily produced and selected. 25 This. general strategy was demonstrated in connection with our generation of the first XenoMouse mouse strains, as published in 1994. (See Green et al. Nature Genetics 7:13-21 (1994)) The XenoMouse strains were engineered with yeast artificial chromosomes (YACs) containing 245 kb and 190 kb-sized germline configuration fragments of the human heavy chain locus and kappa light chain locus, respectively, which contained core variable and constant region sequences. Id. The 30 human Ig containing YACs proved to be compatible with the mouse system for both rearrangement and expression of antibodies and were capable of substituting for the inactivated mouse Ig genes. This was demonstrated by their ability to induce B-cell development, to produce an adult-like human repertoire of fully human antibodies, and to generate antigen-specific human mAbs. These results also suggested that introduction of larger portions of the human Ig loci containing greater 35 numbers of V genes, additional regulatory elements, and human Ig constant regions might recapitulate substantially the full repertoire that is characteristic of the human humoral response to infection and immunization. The work of Green et al. was recently extended to the introduction of -36greater than approximately 80% of the human antibody repertoire through introduction of megabase sized, germline configuration YAG fragments of the human heavy chain loci and kappa light chain loci, respectively. See Mendez et al. Nature Genetics 15:146-156 (1997) and U.S. patent application Ser. No. 08/759,620, filed Dec. 3, 1996. 5 The production of the XenoMouse mice is further discussed and delineated in U.S. Patent Application Serial Nos. 07/466,008, filed January 12, 1990, 07/610,515, filed November 8, 1990, 07/919,297, filed July 24, 1992, 07/922,649, filed July 30, 1992, filed 08/031,801, filed March 15,1993, 08/112,848, filed August 27, 1993, 08/234,145, filed April 28, 1994, 08/376,279, filed January 20, 1995, 08/430, 938, April 27, 1995, 08/464,584, filed June 5, 1995, 08/464,582, filed 10 June 5, 1995, 08/463,191, filed June 5, 1995, 08/462,837, filed June 5, 1995, 08/486,853, filed June 5, 1995, 08/486,857, filed June 5, 1995, 08/486,859, filed June 5, 1995, 08/462,513, filed June 5, 1995, 08/724,752, filed October 2, 1996, and 08/759,620, filed December 3, 1996 and U.S. Patent Nos. 6,162,963, 6,150,584, 6,114,598, 6,075,181, and 5,939,598 and Japanese Patent Nos. 3 068 180 B2, 3 068 506 B2, and 3 068 507 B2. See also Mendez et al. Nature Genetics 15:146-156 15 (1997) and Green and Jakobovits J. Exp. Med. 188:483-495 (1998). See also European Patent No., EP 0 463 151 BI, grant published June 12, 1996, International Patent Application No., WO 94/02602, published February 3, 1994, International Patent Application No., WO 96/34096, published October 31, 1996, WO 98/24893, published June 11, 1998, WO 00/76310, published December 21, 2000, WO 03/47336. 20 In an alternative approach, others, including GenPharm International, Inc., have utilized a "minilocus" approach. In the minilocus approach, an exogenous Ig locus is mimicked through the inclusion of pieces (individual genes) from the Ig locus. Thus, one or more VH genes, one or more DH genes, one or more JH genes, a mu constant region, and a second constant region (preferably a gamma constant region) are formed into a construct for insertion into an animal. This approach is 25 described in U.S. Patent No. 5,545,807 to Surani et al. and U.S. Patent Nos. 5,545,806, 5,625,825, 5,625,126, 5,633,425, 5,661,016, 5,770,429, 5,789,650, 5,814,318, 5,877,397, 5,874,299, and 6,255,458 each to Lonberg and Kay, U.S. Patent No. 5,591,669 and 6,023.010 to Krimpenfort and Berns, U.S. Patent Nos. 5,612,205, 5,721,367, and 5,789,215 to Berns et al., and U.S. Patent No. 5,643,763 to Choi and Dunn, and GenPharm International U.S. Patent Application Serial Nos. 30 07/574,748, filed August 29, 1990, 07/575,962, filed August 31, 1990, 07/810,279, filed December 17, 1991, 07/853,408, filed March 18, 1992, 07/904,068, filed June 23, 1992, 07/990,860, filed December 16, 1992, 08/053,13 1, filed April 26, 1993, 08/096,762, filed July 22, 1993, 08/155,301, filed November 18, 1993, 08/161,739, filed December 3, 1993, 08/165,699, filed December 10, 1993, 08/209,741, filed March 9, 1994. See also European Patent No. 0 546 073 BI, International 35 Patent Application Nos. WO 92/03918, WO 92/22645, WO 92/22647, WO 92/22670, WO 93/12227, WO 94/00569, WO 94/25585, WO 96/14436, .WO 97/13852, and WO 98/24884 and U.S. Patent No. 5,981,175. See further Taylor et al., 1992, Chen et al., 1993, Tuaillon et al., 1993, -37- Choi et al., 19,93, Lonberg et al., (1994), Taylor et al., (1994), and Tuaillon et al., (1995), Fishwild et al., (1996). Kirin has also demonstrated the generation of human antibodies from mice in which, through microcell fusion, large pieces of chromosomes, or entire chromosomes, have been 5 introduced. See European Patent Application Nos. 773 288 and 843 961. Xenerex Biosciences is developing a technology for the potential generation of human antibodies. In this technology, SCIOD mice are reconstituted with human lymphatic cells, e.g., B and/or T cells. Mice are then immunized with an antigen and can generate an immune response against the antigen. See U.S. Patent Nos. 5,476,996, 5,698,767, and 5,958,765. 10 Human anti-mouse antibody (HAMA) responses have led the industry to prepare chimeric or otherwise humanized antibodies. While chimeric antibodies have a human constant region and a murine variable region, it is expected that certain human anti-chimeric antibody (HACA) responses will be observed, particularly in chronic or multi-dose utilizations of the antibody. Thus, it would be desirable to provide fully human antibodies against EGFRvUI in order to vitiate concerns and/or 15 effects of HAMA or HACA response. Antibody Therapeutics As discussed herein, the function of the EGFRvIII antibody appears important to at least a portion of its mode of operation. By function, it is meant, by way of example, the activity of the EGFRvIII antibody in operation with EGFRvII. Accordingly, in certain respects, it may be 20 desirable in connection with the generation of antibodies as therapeutic candidates against EGFRvIII that the antibodies be capable of fixing complement and recruiting cytotoxic lymphocytes thus participating in CDC and ADCC. There are a number of isotypes of antibodies that are capable of the same, including, without limitation, the following: murine IgM, murine IgG2a, murine IgG2b, murine IgG3, human IgM, human IgG1, human IgG3, and human IgA. 25 Also, it may be desirable in connection with the generation of antibodies as therapeutic candidates against EGFRvIII that the antibodies be capable of activating antibody-dependent celluclar cytotoxicity (ADCC), through engagement of Fc receptors on effectors cells such as natural killer (NK) cells. There are a number of isotypes of antibodies that are capable of ADCC, including, without limitation, the following: murine IgG2a, murine IgG2b, murine IgG3, human IgGl, and 30 human IgG3. 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 the antibody can be isotype switched thereafter using conventional techniques that are well known in the art. Such techniques include the use of direct recombinant techniques (see e.g., U.S. Patent No. 4,816,397) and cell-cell fusion techniques (see e.g., U.S. Patent Nos. 5,916,771 and 6,207,418), among others. 35 In the cell-cell fusion technique, a myeloma or other cell line is prepared that possesses a heavy chain with any desired isotype and another myeloma or other cell line is prepared that -38possesses the light chain. Such cells can, thereafter, be fused and a cell line expressing an intact antibody can be isolated. By way of example, certain anti-EGFRvI antibodies discussed herein are human anti. EGFRvIII IgGI antibodies. If such antibody possessed desired binding to the EGFRvIII molecule, 5 it could be readily isotype switched to generate a human IgM, human IgG3, or human IgGA while still possessing the same variable region (which defines the antibody's specificity and some of its affinity). Such molecules, including IgGI, would then be capable of fixing complement and participating in CDC, and, if comprising and IgGI or IgG3 constant region, such molecules would also be capable of participating in antibody-dependent cellular cytotoxicity (ADCC) through 10 recruiting cytotoxic lymphocytes. Accordingly, as antibody candidates are generated that meet desired "structural" attributes as discussed above, they can generally be provided with at least certain of the desired "functional" attributes through isotype switching. Design and Generation of Other Therapeutics 15 Based on the activity of the antibodies that are produced and characterized herein with respect to EGFRvIII, the design of other therapeutic modalities beyond antibody moieties is facilitated. Such modalities include, without limitation, advanced antibody therapeutics, such as bispecific antibodies, immunotoxins, and radiolabeled therapeutics, generation of peptide therapeutics, gene therapies, particularly intrabodies, antisense therapeutics, and small molecules. 20 In connection with the generation of advanced antibody therapeutics, where complement fixation and recruitment of cytoxic lymphocytes is a desirable attribute, it is possible to enhance cell killing through the use of bispecifics, immunotoxins, or radiolabels, for example. For example, in connection with bispecific antibodies, bispecific antibodies can be generated that comprise (i) two antibodies one with a specificity to EGFRvIII and another to a 25 second molecule that are conjugated together, (ii) a single antibody that has one chain specific to EGFRvflI and a second chain specific to a second molecule, or (iii) a single chain antibody that has specificity to EGFRvIII and the other molecule. Such bispecific antibodies can be generated using techniques that are well known for example, in connection with (i) and (ii) see e.g., Fanger et al. Immunol Methods 4:72-81 (1994) and Wright and Harris, supra. and in connection with (iii) see 30 e.g., Traunecker et al. Int. J. Cancer (Suppl.) 7:51-52 (1992). In each case, the second specificity can be made to the Fc chain activation receptors, including, without limitation, CD 16 or CD64 (see e.g., Deo et al. 18:127 (1997)) CD3 (Micromet's BiTE technology) or CD89 (see e.g., Valerius et al. Blood 90:4485-4492 (1997)). Bispecific antibodies prepared in accordance with the foregoing would be likely to kill cells expressing EGFRvIII, and particularly those cells in which the 35 EGFRvIII antibodies of the invention are effective. In connection with immunotoxins, antibodies can be modified to act as immunotoxins utilizing techniques that are well known in the art. See e.g., Vitetta Immunol Today 14:252 (1993). -39- See also U.S. Patent No. 5,194,594. In connection with the preparation of radiolabeled antibodies, such modified antibodies can also be readily prepared utilizing techniques that are well known in the art. See e.g., Junghans et al. in Cancer Chemotherapy and Biotherapy 655-686 (2d edition, Chafner and Longo, eds., Lippincott Raven (1996)). See also U.S. Patent Nos. 4,681,581, 5 4,735,210, 5,101,827, 5,102,990 (RE 35,500), 5,648,471, and 5,697,902. Each of immunotoxins and radiolabeled molecules would be likely to kill cells expressing EGFRvIII, and particularly those cells in which the antibodies described herein are effective. The antibodies can be designed to bind more quickly, or to dissociate more slowly from the epitope, and thus the antibodies themselves can be designed therapeutics. The altered 10 characterisitics of the antibodies can be used, for example, in the administration of a therapeutic to a patient. Therapeutic Inmunoconjugates As will be appreciated, antibodies conjugated to drugs, toxins, or other molecules (immunoconjugates or immunotoxins) are highly useful in the targeted killing of cells that express a 15 molecule that can be specifically bound by a specific binding molecule, such as an antibody. As discussed above, EGFRvIII is not known to be expressed on any normal tissues. Further, EGFRvII shows significant expression in numerous human tumors. Accordingly, EGFRvIlI is a highly attractive molecule for targeting with an immunotoxin. Many reports have appeared on the attempted specific targeting of tumor cells with 20 monoclonal antibody-drug conjugates (Sela et al. in Immunoconjugates 189-216 (C. Vogel, ed. 1987); Ghose et al, in Targeted Drugs 1-22 (E. Goldberg, ed. 1983); Diener et al, in Antibody Mediated Delivery Systems 1-23 (J. Rodwell, ed. 1988); Pietersz et al, in Antibody Mediated Delivery Systems 25-53 (J. Rodwell, ed. 1988); Bumol et al, in Antibody Mediated Delivery Systems 55-79 (J. Rodwell, ed. 1988). Cytotoxic drugs such as methotrexate, daunorubicin, 25 doxorubicin, vincristine, vinblastine, melphalan, mitomycin C, and chlorambucil have been conjugated to a variety of murine monoclonal antibodies. In some cases, the drug molecules were linked to the antibody molecules through an intermediary carrier molecule such as serum albumin (Garnett et al. Cancer Res. 46:2407-2412 (1986); Ohkawa et al. Cancer Immumol. Immunother. 23:81-86 (1986); Endo et al. Cancer Res. 47:1076-1080 (1980)), dextran (Hurwitz et al. Apple. 30 Biochem. 2:25-35 (1980); Manabi et al. Biochem. Pharmacol. 34:289-291 (1985); Dillman et al. Cancer Res. 46:4886-4891 (1986); Shoval et al. Proc. Natl. Acad. Sci. 85: 8276-8280 (1988)), or polyglutamic acid (Tsukada et al. J. Natl. Canc. Inst. 73:721-729 (1984); Kato et al. J. Med. Chem. 27:1602-1607 (1984); Tsukada et al. Br. J. Cancer 52:111-116 (1985)). A wide array of linker technologies has been employed for the preparation of such 35 immunoconjugates and both cleavable and non-cleavable linkers have been investigated. In most cases, the full cytotoxic potential of the drugs could only be observed, however, if the drug' molecules could be released from the conjugates in unmodified form at the target site. -40- One of the cleavable linkers that has been employed for the preparation of antibody-drug conjugates is an a'cid-labile linker based on cis-aconitic acid that takes advantage of the acidic environment of different intracellular compartments such as the endosomes encountered during receptor mediated endocytosis and the lysosomes. Shen and Ryser introduced this method for the 5 preparation of conjugates of daunorubicin with macromolecular carriers (Biochem. Biophys. Res. Commun, 102:1048-1054 (1981)). Yang and Reisfeld used the same technique to conjugate daunorubicin to an anti-melanoma antibody (J. Natl. Canc. Inst. 80:1154-1159 (1988)). Recently, Dillman et al. also used an acid-labile linker in a similar fashion to prepare conjugates of daunorubicin with an anti-T cell antibody (Cancer Res. 48:6097-6102 (1988)). 10 An alternative approach, explored by Trouet et al. involved linking daunorubicin to an antibody via a peptide spacer arm (Proc. Natl. Acad. Sci. 79:626-629 (1982)). This was done under the premise that free drug could be released from such a conjugate by the action of lysosomal peptidases. In vitro cytotoxicity tests, however, have revealed that antibody-drug conjugates rarely 15 achieved the same cytotoxic potency as the free unconjugated drugs. This suggested that mechanisms by which drug molecules are released from the antibodies are very inefficient. In the area of immunotoxins, conjugates formed via disulfide bridges between monoclonal antibodies and catalytically active protein toxins were shown to be more cytotoxic than conjugates containing other linkers. See, Lambert et al. J. Biol. Chem. 260:12035-12041 (1985); Lambert et al. in 20 Immunotoxins 175-209 (A. Frankel, ed. 1988); Ghetie et al. Cancer Res. 48:2610-2617 (1988). This was attributed to the high intracellular concentration of glutathione contributing to the efficient cleavage of the disulfide bond between an antibody molecule and a toxin. Despite this, there are only a few reported examples of the use of disulfide bridges for the preparation of conjugates between drugs and macromolecules. Shen et al. described the conversion of methotrexate into a 25 mercaptoethylamide derivative followed by conjugation with poly-D-lysine via a disulfide bond (J. Biol. Chem. 260:10905-10908 (1985)). In addition, a report described the preparation of a conjugate of the trisulfide-containing toxic drug calicheamycin with an antibody (Menendez et al. Fourth International Conference on Monoclonal Antibody Immunoconjugates for Cancer, San Diego, Abstract 81 (1989)). Another report described the preparation of a conjugate of the 30 trisulfide-containing toxic drug calicheamycin with an antibody (Hinman et al, 53 Cancer Res. 3336-3342 (1993)). One reason for the lack of disulfide linked antibody-drug conjugates is the unavailability of cytotoxic drugs that bear a sulfur atom containing moiety that can be readily used to link the drug to an antibody via a disulfide bridge. Furthermore, chemical modification of existing drugs is difficult 35 without diminishing their cytotoxic potential. Another major drawback with existing antibody-drug conjugates is their inability to deliver a sufficient concentration of drug to the target site because of the limited number of targeted -41antigens and the relatively moderate cytotoxicity of cancerostatic drugs like methotrexate, daunorubicin and vincristine. In order to achieve significant cytotoxicity, linkage of a large number of drug molecules either directly to the antibody or through a polymeric carrier molecule becomes necessary. However such heavily modified antibodies often display impaired binding to the target 5 antigen and fast in vivo clearance from the blood stream. Maytansinoids are highly cytotoxic drugs. Maytansine was first isolated by Kupchan et al. from the east African shrub Maytenus serrata and shown to be 100 to 1000 fold more cytotoxic than conventional cancer chemotherapeutic agents like methotrexate, daunorubicin, and vincristine (U.S. Pat. No. 3,896,111). Subsequently, it was discovered that some microbes also produce 10 maytansinoids, such as maytansinol and C-3 esters of maytansinol (U.S. Pat. No. 4,151,042). Synthetic C-3 esters of maytansinol and analogues of maytansinol have also been reported (Kupchan et al. J. Med. Chem. 21:31-37 (1978); Higashide et al. Nature 270:721-722 (1977); Kawai et al. Chem. Pharm. Bull. 32:3441-3451 (1984)). Examples of analogues of maytansinol from which C-3 esters have been prepared include maytansinol with modifications on the aromatic 15 ring (e.g. dechloro) or at the C-9, C-14 (e.g. hydroxylated methyl group), C-15, C-18, C-20 and C 4,5. The naturally occurring and synthetic C-3 esters can be classified into two groups: (a) C-3 esters with simple carboxylic acids (U.S. Pat. Nos. 4,248,870; 4,265,814; 4,308,268; 4,308,269; 4,309,428; 4,317,821; 4,322,348; and 4,331,598), and 20 (b) C-3 esters with derivatives of N-methyl-L-alanine (U.S. Pat. Nos. 4,137,230; 4,260,608; 5,208,020; and Chem. Pharm. Bull. 12:3441 (1984)). Esters of group (b) were found to be much more cytotoxic than esters of group (a). Maytansine is a mitotic inhibitor. Treatment of L1210 cells in vivo with maytansine has been reported to result in 67% of the cells accumulating in mitosis. Untreated control cells were 25 reported to demonstrate a mitotic index ranging from between 3.2 to 5.8% (Sieber et al. 43 Comparative Leukemia Research 1975, Bibl. Haemat. 495-500 (1976)). Experiments with sea urchin eggs and clam eggs have suggested that maytansine inhibits mitosis by interfering with the formation of microtubules through the inhibition of the polymerization of the microtubule protein, tubulin (Remillard et al. Science 189:1002-1005 (1975)). 30 In vitro, P388, L1210, and LY5178 murine leukemic cell suspensions have been found to be inhibited by maytansine at doses of 104' to 10' .mu.g/.mu.1 with the P388 line being the most sensitive. Maytansine has also been shown to be an active inhibitor of in vitro growth of human nasopharyngeal carcinoma cells, and the human acute lymphoblastic leukemia line CEM was reported inhibited by concentrations as low as 104 mg/ml (Wolpert-DeFillippes et al. Biochem. 35 Pharmacol. 24:1735-1738 (1975)). In vivo, maytansine has also been shown to be active. Tumor growth in the P388 lymphocytic leukemia system was shown to be inhibited over a 50- to 100-fold dosage range which -42suggested a high therapeutic index; also significant inhibitory activity could be demonstrated with the L1210 mouse leukemia system, the human Lewis lung carcinoma system and the human B-16 melanocarcinoma system (Kupchan, Ped. Proc. 33:2288-2295 (1974)). Current methods of conjugation of maytansinoids with cell binding agents (such as 5 antibodies) involve two reaction steps. A cell binding agent, for example an antibody, is first modified with a cross-linking reagent such as N-succinimidyl pyridyldithiopropionate (SPDP) to introduce dithiopyridyl groups into the antibody (Carlsson et al. Biochem. J. 173:723-737 (1978); U.S. Pat. No. 5,208,020). In a second step, a reactive maytansinoid having a thiol group, such as DM1 (formally termed N 2 ' -deacetyl-N 2 ' -( 3 -mercapto-1-oxopropyl)-maytansine, as the starting 10 reagent., is added to the modified antibody, resulting in the displacement of the thiopyridyl groups in the modified antibodies, and the production of disulfide-linked cytotoxic maytansinoid/antibody conjugates (U.S. Pat. No. 5,208,020). A one-step process for conjugation of maytansinoids is described in U.S. Patent No. 6,441,163. Maytansinoid-based immunotoxin technology is available from Immunogen Corporation (Cambridge, MA). 15 Another important toxin technology is based upon auristatin toxins. Auristatins are derived from Dolastatin 10 that was obtained from the Indian Ocean sea hare Dolabella, as a potent cell growth inhibitory substance. See U.S. Patent Nos. 4,816,444 and 4,978,744. With respect to other Dolastatins, see also U.S. Patent Nos. 4,414,205 (Dolastatin-1, 2, and 3), 5,076,973 (Dolastatin-3), 4,486,414 (Dolastatin-A and B),; 4,986,988 (Dolastatin-13), 5,138,036 (Dolastatin-14), and 20 4,879,278 (dolastatin-15). Isolated and synthesized by Dr. Pettit and colleagues at the University of Arizona, a variety of auristatine derivatives have been tested and shown to be highly toxic to cells. See Pettit et al. Antineoplastic agents 337. Synthesis of dolastatin 10 structural modifications. Anticancer Drug Des. 10(7):529-44 (1995), Woyke et al. In vitro activities and postantifungal effects of the potent dolastatin 10 structural modification auristatin PHE. Antimicrobial Agents and 25 Chemotherapy. 45:3580-3584 (2001), Pettit et al. Specific activities of dolastatin 10 and peptide derivatives against Cryptococcus neoformans. Antimicrobial Agents and Chemotherapy. 42:2961 2965 (1998), WoykeThree-dimensional visualization of microtubules during the Cryptococcus neoformans cell cycle and the effects of auristatin PHE on microtubule integrity and nuclear localization. Submitted, Antimicrobial Agents and Chemotherapy. 30 Recently, additional auristatin derivatives have been developed that appear quite effective when delivered as payloads on antibodies. For example monomethyl auristatin E (MMAE) has been shown as a potent toxin for tumor cells when conjugated to tumor specific antibodies. Doronina et al. Development of potent monoclonal antibody auristatin conjugates for cancer therapy. Nature Biotechnology. (2003) (available online), Francisco et al. cAC10-vcMMAE, an 35 anti-CD30-monomethyl auristatin E conjugate with potent and selective antitumor activity. Blood. (2003) May 8 [Epub ahead of print]. Epub 2003 Apr 24 (available online). In addition to the toxicity of the auristatin molecule, research has shown that peptide-linked conjugates are more -43stable , and, thus, more specific and less toxic to normal tissues than other linker technologies in buffers and plasma. Doronina et al. Development of potent monoclonal antibody auristatin conjugates for cancer therapy. Nature Biotechnology. (2003) (available online), Francisco et al. cACl0-vcMMAE, an anti-CD30-monomethyl auristatin E conjugate with potent and selective 5 antitumor activity. Blood. (2003) May 8 [Epub ahead of print].' Epub 2003 Apr 24 (available online). Such linkers are based on a branched peptide design and include, for example, mAb valine-citrulline-MMAE and mAb-phenylalanine-lysine-MMAE conjugates. Doronina et al. Development of potent monoclonal antibody auristatin conjugates for cancer therapy. Nature Biotechnology. (2003) (available online), Francisco et al. cACI0-vcMMAE, an anti-CD30 10 monomethyl auristatin E conjugate with potent and selective antitumor activity. Blood. (2003) May 8 [Epub ahead of print]. Epub 2003 Apr 24 (available online). Such designs and conjugation techniques are described, for example, by King et al. Monoclonal antibody conjugates of doxorubicin prepared with branched peptide linkers: inhibition of aggregation by methoxytriethyleneglycol chains. J Med Chem. 45(19):4336-43 (2002) and Dubowchik et al. 15 Cathepsin B-sensitive dipeptide prodrugs. 2. Models of anticancer drugs paclitaxel (Taxol), mitomycin C and doxorubicin. Bioorg Med Chem Lett. 8(23):3347-52 (1998). Auristatin E-based immunotoxin technology based upon the foregoing is available from Seattle Genetics Corporation (Seattle, WA). There are a large number of novel microtubule effecting compounds obtained from natural 20 sources-extracts, and semisynthetic and synthetic analogs that appear to possess potential as toxins for the generation of immunoconjugates. (see the website at newmedinc "dot" com). Such molecules and examples of drug products utilizing them, include the following: Colchicine-site Binders (Curacin), Combretastatins (AVE806, Combretastatin A-4 prodrug (CA4P), Oxi-4503), Cryptophycins (LY355703), Discodermolide, Dolastatin and Analogs (Auristatin PHE, Dolastatin 25 10, ILX-651, Symplostatin 1, TZT-1027), Epothilones (BMS-247550, BMS-310705, EP0906, KOS-862, ZK-EPO), Eleutherobin, FR182877, Halichondrin B (E7389), Halimide (NPI-2352 and NPI-2358), Hemiasterlins (HTI-286), Laulimalide, Maytansinoids ("DM1")(Bivatuzumab mertansine, Cantuzumab mertansine, huN901-DMI/BB-10901TAP, MLN591DM1, My9-6-DMI, Trastuzumab-DMI), PC-SPES, Peloruside A, Resveratrol, S-allylmercaptocysteine (SAMC), 30 Spongistatins, Vitilevuamide, Molecular Motor-Kinesins (SB-715992), Designed Colchicine-Site Binders (A-289099, A-293620/A-318315, ABT-751/E7010, D-24851/D-64131, ZD6126), Other Novel Spindle Poisons (2-Methoxyestradiol (2-ME2), Bezimidazole Carbamates (ANG 600 series, Mebendazole), CP248/CP461, HMN-214, R440, SDX-103, T67/T607). Further, additional marine derived toxins are reviewed in Mayer, A.M.S. Marine Pharmacology in 1998: Antitumor and 35 Cytotoxic Compounds. The Pharmacologist. 41(4):159-164 (1999). -44- Therapeutic Administration and Formulations A prolonged duration of action will allow for less frequent and more convenient dosing schedules by alternate parenteral routes such as intravenous, subcutaneous or intramuscular injection. 5 When used for in vivo administration, antibody formulations described herein should be sterile. This is readily accomplished, for example, by filtration through sterile filtration membranes, prior to or following lyophilization and reconstitution. Antibodies ordinarily will be stored in lyophilized form or in solution. Therapeutic antibody compositions generally are placed into a container having a sterile access port, for example, an intravenous solution bag or vial having 10 an adapter that allows retrieval of the formulation, such as a stopper pierceable by a hypodermic injection needle. The route of antibody administration is in accord with known methods, e.g., injection or infusion by intravenous, intraperitoneal, intracerebral, intramuscular, intraocular, intraarterial, intrathecal, inhalation or intralesional routes, or by sustained release systems as noted below. 15 Antibodies are preferably administered continuously by infusion or by bolus injection. An effective amount of antibody to be employed therapeutically will depend, for example, upon the therapeutic objectives, the route of administration, and the condition of the patient. Accordingly, it is preferred for the therapist to titer the dosage and modify the route of administration as required to obtain the optimal therapeutic effect. Typically, the clinician will 20 administer antibody until a dosage is reached that achieves the desired effect. The progress of this therapy is easily monitored by conventional assays or by the assays described herein. Antibodies as described herein can be prepared in a mixture with a pharmaceutically acceptable carrier. Therapeutic compositions can be administered intravenously or through the nose or lung, preferably as a liquid or powder aerosol lyophilizedd). Composition can also be 25 administered parenterally or -subcutaneously as desired. When administered systemically, therapeutic compositions should be sterile, pyrogen-free and in a parenterally acceptable solution having due regard for pH, isotonicity, and stability. These conditions are known to those skilled in the art. Briefly, dosage formulations of the compounds are prepared for storage or administration by mixing the compound having the desired degree of purity with physiologically acceptable 30 carriers, excipients, or stabilizers. Such materials are non-toxic to the recipients at the dosages and concentrations employed, and include buffers such as TRIS HCI, phosphate, citrate, acetate and other organic acid salts; antioxidants such as ascorbic acid; low molecular weight (less than about ten residues) peptides such as polyarginine, proteins, such as serum albumin, gelatin, or immunoglobulins; hydrophilic polymers such as polyvinylpyrrolidinone; amino acids such as 35 glycine, glutamic acid, aspartic acid, or arginine; monosaccharides, disaccharides, and other carbohydrates including cellulose or its derivatives, glucose, mannose, or dextrins; chelating agents -45such as EDTA; sugar alcohols such as mannitol or sorbitol; counterions such as sodium and/or nonionic surfactants such as TWEEN, PLURONICS or polyethyleneglycol. Sterile compositions for injection can be formulated according to conventional pharmaceutical practice as described in Remington's Pharmaceutical Sciences (18'h ed, Mack 5 Publishing Company, Easton, PA, 1990). For example, dissolution or suspension of the active compound in a vehicle such as water or naturally occurring vegetable oil like sesame, peanut, or cottonseed oil or a synthetic fatty vehicle like ethyl oleate or the like may be desired. Buffers, preservatives, antioxidants and the like can be incorporated according to accepted pharmaceutical practice. 10 Suitable examples of sustained-release preparations include semipermeable matrices of solid hydrophobic polymers containing the polypeptide, which matrices are in the form of shaped articles, films or microcapsules. Examples of sustained-release matrices include polyesters, hydrogels (e.g., poly(2-hydroxyethyl-methacrylate) as described by Langer et al., J. Biomed Mater. Res., (1981) 15:167-277 and Langer, Chem. Tech., (1982) 12:98-105, or poly(vinylalcohol)), 15 polylactides (U.S. Pat. No. 3,773,919, EP 58,481), copolymers of L-glutanic acid and gamma ethyl-L-glutamate (Sidman et al., Biopolyiners, (1983) 22:547-556), non-degradable ethylene-vinyl acetate (Langer et al., supra), degradable lactic acid-glycolic acid copolymers such as the LUPRON Depot m (injectable microspheres composed of lactic acid-glycolic acid copolymer and leuprolide acetate), and poly-D-(-)-3-hydroxybutyric acid (EP 133,988). 20 While polymers such as ethylene-vinyl acetate and lactic acid-glycolic acid enable release of molecules for over 100 days, certain hydrogels release proteins for shorter time periods. When encapsulated proteins remain in the body for a long time, they may denature or aggregate as a result of exposure to moisture at 37*C, resulting in a loss of biological activity and possible changes in immunogenicity. Rational strategies can be devised for protein stabilization depending on the 25 mechanism involved. For example, if the aggregation mechanism is discovered to be intermolecular S-S bond formation through disulfide interchange, stabilization may be achieved by modifying sulfhydryl residues, lyophilizing from acidic solutions, controlling moisture content, using appropriate additives, and developing specific polymer matrix compositions. Sustained-released compositions also include preparations of crystals of the antibody 30 suspended in suitable formulations capable of maintaining crystals in suspension. These preparations when injected subcutaneously or intraperitoneally can produce a sustain release effect. Other compositions also include liposomally entrapped antibodies. Liposomes containing such antibodies are prepared by methods known per se: U.S. Pat. No. DE 3,218,121; Epstein et al., Proc. Natl. Acad. Sci. USA, (1985) 82:3688-3692; Hwang et al., Proc. Natl. Acad. Sci. USA, (1980) 35 77:4030-4034; EP 52,322; EP 36,676; EP 88,046; EP 143,949; 142,641; Japanese patent application 83-118008; U.S. Pat. Nos. 4,485,045 and 4,544,545; and EP 102,324. -46- The dosage of the antibody formulation for a given patient will be determined by the attending physician taking into consideration various factors known to modify the action of drugs including severity and type of disease, body weight, sex, diet, time and route of administration, other medications and other relevant clinical factors. Therapeutically effective dosages may be 5 determined by either in vitro or in vivo methods. An effective amount of the antibody to be employed therapeutically will depend, for example, upon the therapeutic objectives, the route of administration, and the condition of the patient. Accordingly, it is preferred that the therapist titer the dosage and modify the route of administration as required to obtain the optimal therapeutic effect. A typical daily dosage might 10 range from about 0.001 mg/kg to up to 100 mg/kg or more, depending on the factors mentioned above. Typically, the clinician will administer the therapeutic antibody until a dosage is reached that achieves the desired effect. The progress of this therapy is easily monitored by conventional assays or as described herein. It will be appreciated that administration of therapeutic entities in accordance with the 15 compositions and methods herein will be administered with suitable carriers, excipients, and other agents that are incorporated into formulations to provide improved transfer, delivery, tolerance, and the like. A multitude of appropriate formulations can be found in the formulary known to all pharmaceutical chemists: Remington's Pharmaceutical Sciences (198' ed, Mack Publishing Company, Easton, PA (1990)), particularly Chapter 87 by Block, Lawrence, therein. These 20 formulations include, for example, powders, pastes, ointments, jellies, waxes, oils, lipids, lipid (cationic or anionic) containing vesicles (such as Lipofectin'"), DNA conjugates, anhydrous absorption pastes, oil-in-water and water-in-oil emulsions, emulsions carbowax (polyethylene glycols of various molecular weights), semi-solid gels, and semi-solid mixtures containing carbowax. Any of the foregoing mixtures may be appropriate in treatments and therapies in 25 accordance with the present invention, provided that the active ingredient in the formulation is not inactivated by the formulation and the formulation is physiologically compatible and tolerable with the route of administration. See also Baldrick P. "Pharmaceutical excipient development: the need for preclinical guidance." Regul. Toxicol. Phannacol. 32(2):210-8 (2000), Wang W. "Lyophilization and development of solid protein pharmaceuticals." Int. J. Phann. 203(1-2):1-60 30 (2000), Charman WN "Lipids, lipophilic drugs, and oral drug delivery-some emerging concepts." J Pharn Sci .89(8):967-78 (2000), Powell et al. "Compendium of excipients for parenteral formulations" PDA JPharm Sci Technol. 52:238-311 (1998) and the citations therein for additional information related to formulations, excipients and carriers well known to pharmaceutical chemists. Preparation of Antibodies 35 Antibodies, as described herein, were prepared through the utilization of the XenoMouse@ technology, as described below. Such mice, then, are capable of producing human immunoglobulin molecules and antibodies and are deficient in the production of murine immunoglobulin molecules -47and antibodies. Technologies utilized for achieving the same are disclosed in the patents, applications, and references disclosed herein. In particular, however, a one embodiment of transgenic production of mice and antibodies therefrom is disclosed in U.S. Patent Application Serial No. 08/759,620, filed December 3, 1996 and International Patent Application Nos. WO 5 98/24893, published June 11, 1998 and wo 00/76310, published December 21, 2000.. See also Mendez et al. Nature Genetics 15:146-156 (1997). Through use of such technology, fully human monoclonal antibodies to a variety of antigens can be produced. In one embodiement, XenoMouse@ lines of mice are immunized with an antigen of interest (e.g. EGFRvII), lymphatic cells are recovered (such as B-cells) from the mice 10 that expressed antibodies, and such cells are fused with a myeloid-type cell line to prepare immortal hybridoma cell lines, and such hybridoma cell lines are screened and selected to identify hybridoma cell lines that produce antibodies specific to the antigen of interest. Provided herein are methods for the production of multiple hybridoma cell lines that produce antibodies specific to EGFRvIm. Further, provided herein are characterization of the antibodies produced by such cell lines, 15 including nucleotide and amino acid sequences of the heavy and light chains of such antibodies. Alternatively, instead of being fused to myeloma cells to generate hybridomas, the antibody produced by recovered cells, isolated from immunized XenoMouse@ lines of mice, are screened further for reactivity against the initial antigen, preferably EGFRvIfl protein. Such screening includes ELISA with EGFRvIII protein, in vitro binding to NR6 M cells stably expressing full 20 length EGFRvm and internalization of EGFRvII receptor by the antibodies in NR6 M cells. Single B cells secreting antibodies of interest are then isolated using a EGFRvm-specific hemolytic plaque assay (Babcook et al., Proc. Nati. Acad. Sci. USA, i93:7843-7848 (1996)). Cells targeted for lysis are preferably sheep red blood cells (SRBCs) coated with the EGFRvIII antigen. In the presence of a B cell culture secreting the immunoglobulin of interest and complement, the 25 formation of a plaque indicates specific EGFRvIII-mediated lysis of the target cells. The single antigen-specific plasma cell in the center of the plaque can be isolated and the genetic information that encodes the specificity of the antibody is isolated from the single plasma cell. Using reverse transcriptase PCR, the DNA encoding the variable region of the antibody secreted can be cloned. Such cloned DNA can then be further inserted into a suitable expression vector, preferably a vector 30 cassette such as a pcDNA, more preferably such a pcDNA vector containing the constant domains of immunglobulin heavy and light chain. The generated vector can then be transfected into host cells, preferably CHO cells, and cultured in conventional nutrient media modified as appropriate for inducing promoters, selecting transformants, or amplifying the genes encoding the desired sequences. Herein, we describe the isolation of multiple single plasma cells that produce antibodies 35 specific to EGFRvIII. Further, the genetic material that encodes the specificity of the anti EGFRvIII antibody is isolated, introduced into a suitable expression vector that is then transfected into host cells. -48- B cells from XenoMouse mice may be also be used as a source of genetic material from which antibody display libraries may be generated. Such libraries may be made in bacteriophage, yeast or in vitro via ribosome display using ordinary skills in the art. Hyperimmunized XenoMouse mice may be a rich source from which high-affinity, antigen-reactive antibodies may be isolated. 5 Accordingly, XenoMouse mice hyperimmunized against EGFRvIII may be used to generate antibody display libraries from which high-affinity antibodies against EGFRvII may be isolated. Such libraries could be screened against the pep3 oligopeptide and the resultingly derived antibodies screening against cells expressing EGFRvl to confirm specificity for the natively display antigen. Full IgG antibody may then be expressed using recombinant DNA technology. 10 See e.g., WO 99/53049. In general, antibodies produced by the above-mentioned cell lines possessed fully human IgG1 or IgG2 heavy chains with human kappa light chains. In one embodiment, the antibodies possessed high affinities, typically possessing Kd's of from about 10- through about 10-" M, when measured by either solid phase and solution phase. In other embodiments the antibodies possessed 15 lower affinities, from about 10'6 through about 104 M. As appreciated by one of skill in the art, antibodies in accordance with the present embodiments can be expressed in cell lines other than hybridoma cell lines. Sequences encoding particular antibodies can be used for transformation of a suitable mammalian host cell. Transformation can be by any known method for introducing polynucleotides into a host cell, 20 including, for example packaging the polynucleotide in a virus (or into a viral vector) and transducing a host cell with the virus (or vector) or by transfection procedures known in the art, as exemplified by U.S. Patent Nos. 4,399,216, 4,912,040, 4,740,461, and 4,959;455. The transformation procedure used depends upon the host to be transformed. Methods for introduction of heterologous polynucleotides into mammalian cells are well known in the art and include 25 dextran-mediated transfection, calcium phosphate precipitation, polybrene mediated transfection, protoplast fusion, electroporation, encapsulation of the polynucleotide(s) in liposomes, and direct microinjection of the DNA into nuclei. Mammalian cell lines available as hosts for expression are well known in the art and include many immortalized cell lines available from the American Type Culture Collection 30 (ATCC), including but not limited to Chinese hamster ovary (CHO) cells, HeLa cells, baby hamster kidney (BHK) cells, monkey kidney cells (COS), human hepatocellular carcinoma cells (e.g., Hep G2), and a number of other cell lines. Cell lines of particular preference are selected through determining which cell lines have high expression levels and produce antibodies with constitutive EGFRvIII binding properties. -49- EXAMPLES The following examples, including the experiments conducted and the results achieved are provided for illustrative purposes only and are not to be construed as limiting upon the present invention. 5 The strategy for generating EGFRvm-specific antibodies initially involved immunization of XenoMouse mice with combinations of antigens (peptide, various soluble proteins, antigen expressing cells) followed by isolation of antibody producing cells, either as through fusions to produce hybridomas or isolation of B cell cells through the XenoMax"/SLAM"" technology. Antibody producing cells were subjected to a primary screen for specificity by ELISA and a 10 secondary screen for cell surface binding by FMAT and/or FACS. Internalization assays were then conducted to identify antibodies that would be useful for drug delivery. Affinities of the antibodies were measured. Certain antibodies were selected for epitope mapping. In addition, certain antibodies were selected for in vitro and in vivo tests to analyze the'efficacy of such antibodies for treatment of cancers. 15 EXAMPLE 1 Antiaen Prepartion A. EGFRyIM PEP3-KLH Antigen Preparation In connection with Example 2, the 14-mer human EGFRvHI PEP3 (L E E K K G N Y V V T D H C (SEQ ID NO: 56)) peptide was custom synthesized by R&D Systems. The PEP3 peptide 20 was then conjugated to keyhole limpet hemocyanin (KLH), as follows: EGFRvI PEP3 (200 mcg) (R&D) was mixed with 50 mcg of keyhole limpet hemocyanin (KLH; Pierce, Rockford, IL) to a final volume of 165 mcl using distilled water. 250 mcl of conjugation buffer (0.1M MES, 0.9M NaCI, pH 4.7) was added and EGFRvIH PEP3 and KLH were crosslinked by the addition of 25 mcl of 10 mg/ml stock solution of 1-ethyl- 3

-[

3 -dimethylaminopropyl]carbodiimide hydrochloride 25 (EDC, Pierce, Rockford, IL). Conjugate was incubated for 2 hours at room temperature and the unreacted EDC was removed by centrifugation through a I kDa filter (Centrifugal filter; Millipore, Bedford, MA) using PBS pH 7.4. In connection with Example 3, the 14-mer human EGFRvIH PEP3 (L E E K K G N Y V V T D H d (SEQ ID NO: 56)) peptide was custom synthesized. The PEP3 peptide was then 30 conjugated to KLH, as follows: human EGFRvIII PEP3 (200 mcg) was mixed with 50 mcg of keyhole limpet hemocyanin (KLH; Pierce, Rockford, IL) to a final volume of 165 mcl using distilled water. 250 mcl of conjugation buffer (0.1M MES, 0.9M NaCI, pH 4.7) was added and EGFRvlI PEP3 and KLH were crosslinked by the addition of 25 mcl of 10 mg/ml stock solution of I-ethyl-3-[ 3 -dimethylaminopropyl]carbodiimide hydrochloride (EDC, Pierce, Rockford, IL). 35 Conjugate was incubated for 2 hours at room temperature and the unreacted EDC was removed by centrifugation through a 1 kDa filter (Centrifugal filter; Millipore, Bedford, MA) using PBS pH 7.4. -50- B. B390,19/EGFRvIII Transfectants In order to prepare the B300.19/EGFRvIII transfectants, wild type EGFR was initially cloned from A431 cells and EGFR gene was modified to code for EGFRvIII to delete the codons encoding residues 6-273, with a codon encoding a Glycine residue created at the junction of the 5 deletion. The deletion occurs within the codons surrounding the deletion GTI' (Valine) and CGT (Arginine), such that the resulting codoil after the deletion is GGT (Glycine). (Wikstrand et al. J Neurovirol. 4(2):148-58 (1998)) 1. Cloning of wild type EGFR Construct: PolyA+mRNA was extracted from A431 (ATCC) cells usingMicro-fast RNA kit 10 (Invitrogen, Burlington, ON). Total cDNA was synthesized from polyA+ mRNA with random pdN6 primers and M-MuLV reverse transcriptase (NEB, New England Biolabs, Beverly, Mass.). A 2.3kb PCR product was amplified from A431 cDNA with the following primers: sense 5'-GGATCrCGAGCCAGACCGGAACGACAGGCCACCTC-3'; (SEQ ID NO: 62) anti-sense 5'-CGGATCTCGAGCCGGAGCCCAGCACTIGATCTT-3' (SEQ ID NO: 15 63) using Pfu DNA polymerase. The PCR product was digested with Xhol, gel purified and ligated into plasmid pWBFNP (see International Patent Application No. WO 99/45031) linearized with XhoI to yield plasmid Wt EGFR/pWBFNP. 20 2. Generation of EGFRvII Construct: PCR products amplified from plasmid Wt-EGFR/pWBFNP template with primer pairs C13659/C29538 and C29539/C14288 (BioSource International), in which the C29538 and C29539 were phosphorylated with T4 Polynucleotide kinase (NEB, New England Biolabs, Beverly, Mass.): C13659: 5'-CGGATGAATI'CCCAGACCGGACGACAGGCCACCTC-3'(Sense) (SEQ 25 ID NO: 64); C29538: 5'-CTCITITCCTCCAGAGCC-3'(Anti-Sense) (SEQ ID NO: 65); C29539: 5'-GTAAT'ATGTGGTGACAGATC-3'(Sense) (SEQ ID NO: 66); C14288:5'-CGGATCTCGAGCrCAAGAGAGCTrGGTTGGGAGCT-3'(Anti- Sense) (SEQ ID NO: 67). 30 were ligated to introduce a deletion in the sequence encoding amino acids 6 through 273 of the EGFR extracellular domain and subcloned into expression vector pWBDHFR2 (see International Patent Application No. WO 99/45031). A 232 bp fragment representing the 5'end of the deletion was generated with primer pair C13659/C29538 from Wt-EGFR/pWBFNP template amplified with Pfu polymerase (NEB, New 35 England Biolabs, Beverly, Mass.). The PCR product was digested with EcoRi (NEB, New England Biolabs, Beverly, Mass.) and gel purified. A 1273 bp fragment representing the 3'end of the deletion was generated with primer pair C29539/C4288 from Wt-EGFR/pWBFNP and the -51template amplified with Pfu polymerase. The PCR product was digested with Xhol (NEB, New England Biolabs, Beverly, Mass.) and gel purified. Fragments were ligated into EcoRl/Xhol digested pWBDHFR2 with T4 DNA ligase (NEB, New England Biolabs, Beverly, Mass.) to yield construct EGFRvII/pWBDHFR 5 The intracellular domain of EGFR was introduced into the resulting construct as follows: A 1566bp DraiIXhoI fragment was isolated from plasmid Wt-EGFR/pWBFNP and ligated into DraII/XhoI digested EGFRvm/pWBDHFR to yield EGFRvm-FL/pWBDHFR. 3. Transfection of B300.19 cells with EGFRvM-FL/pWBDHFR: B300.19 cells (8x10 6 ) were used per transfection in 700 pl DMEM/HI medium. 20 pig 10 EGFRvM-FL/pWBDHFR and 2 tg CMV-Puro plasmid DNA were added. Cells were electroporated at 300 volts/960uF with Bio-Rad Gene Pulser. Following electroporation, cells were cooled on ice for 10 minutes and, thereafter, 10 ml non-selection medium (DMEM/HI Glucose, 10% FBS, 50 pM BME, 2mM L-Glutanine, 100 units Penicillin-G/ml, 100 units MCG Streptomycin/ml) was added. Cells were incubated for 48hrs at 37"C 7.5% CO 2 . 15 Following incubation, cells were split into selection medium (DMEM/HI Glucose, 10% FBS, 2 mM L-Glutamine, 50 pM BME, 100 units Penicillin-G/ml, 100 units MCG Streptomycin/mI, 2ug/ml puromycin) at 2x10 4 , 0.4x10 4 ' and 0.08x10 4 cells/ well in 96 well plate and were selected in selection medium for 14 days to generate stable clones. Puro resistant clones were stained with E752 mAb (an anti-EGFR antibody, described in Yang et al., Crit Rev Oncol 20 Hematol., 38(1):17-23 (2001)) and goat anti-human IgG PE then analyzed on FACS Vantage (Becton Dickinson). C. Construction of EGFRvm-RbFc Expression Constructs. In order to generate the EGFRvII rabbit Fc fusion, protein, we first constructed a vector containing DNA encoding rabbit Fc. This was ligated with DNA encoding EGFRvfl. This 25 approach is described in more detail below: I. Construction of RbFc/pcDNA3. IHygro: Primers 1322/867 (below) were used to amplify a 721bp fragment encoding the Hinge CH2-CH3 domain of rabbit IgG. #1322 (sense): 5'-GGTGGCGGTACCTGGACAAGACCGTTGCG-3' (SEQ ID NO: 68) 30 #867 (antisense): 5'-ATAAGAATGCGGCCGCTCATITACCCGGAGAGCGGGA-3' (SEQ ID NO: 69) The resulting PCR product was digested with KpnI and NotI, gel purified and ligated into KpnI/NotI digested pcDNA3.1(+)/Hygro (Invitrogen, Burlington, ON) to yield plasmid RbFc/pcDNA3.1 Hygro. 35 -52- 2. Construction of EGFRvI--RbFc/pCEP4: Primers 1290/1293 (below) were used to amplify an 1165bp product from EGFRvI FL/pWBDHFR plasmid template with Pfu polymerase #1290 (sense): 5'-CTACTAGCTAGCCACCATGCGACCCTCCGGGA-3' (SEQ ID 5 NO: 70) #1293 (anti-sense): 5'-CGGGGTACCCGGCGATGGACGGGATC-3' (SEQ ID NO: 71) The PCR product was digested with NheI and KpnI, gel purified and ligated into Nhel/KpnI digested RbFc/pcDNA3.1 Hygro to yield plasmid EGFRvIH-RbFc/pcDNA3.lHygro. A 2170 bp SnaBL/XhoI fragment was isolated from EGFRvIH-RbFc/pcDNA3.lHygro and 10 subcloned into SnaBI/XhoI digested pCEP4 (Invitrogen, Burlington, ON) to yield plasmid EGFRvHII-RbFc/pCEP4. 3. Generation of 293F EGFRvImI-RbFc stable Cell Lines: Plasmid EGFRvHI-RbFc/pCEP4 was introduced into 293F cells (Gibco, Grand Island, NY) by Calcium Phosphate transfection, as follows: one day prior to transfection, lxl 0' 293F cells were 15 plated on a gelatin coated 100mm tissue culture petridish and incubated at 5% C02, 37*C. Cells were fed with 10ml of fresh non-selective media (DMEM/F12, 10% FBS, 2mM L-Glutamine, 100U/ml Penicillin G, I00U/ml MCG Streptomycin) 2-3 hours before transfection. Transfection reagents were prepared in a microfuge tube, as follow: 10 ig of DNA (EGFRvUI-RbFc/pCEP4) was mixed with 62pil of 2M Calcium Phosphate and deionized water to make the final volume 20 50041. In another tube pipette 500tl of 2XHBS is drawn and used to transfer the transfection reagents. The solution in the tube pipette was added to the cells drop by drop, while maintaining proper pH by leaving cells in a 5% C02 incubator until transfection was performed. 15-20 hours after transfection, cells were washed with PBS and feed with 10ml of fresh 293F non-selective 25 media. Expressing cells were harvested with trypsin 48-72 post-transfection and cells were plated at 0.08x10 4 cells/well in a 96 well plate in 293F selective media (DMEM/F12, 10% FBS, 2mM L Glutamine, 100U/ml Penicillin G, 100U/ml MCG Streptomycin, 250ug/ml Hygromycin) for 14 days. Hygromycin resistent clones were screened by ELISA using anti-EGFR antibody E763 (US 30 Patent No. 6,235,883) as the capture antibody at lug/il and detecting with a goat anti-rabbit IgG HRPO (CalTag) at 1:100 dilution. D. Conjugation of EGFRvIII PEP3 to OVA via Maleimide Conjugation The EGFRvmlfpeptide-OVA used for titration of antibodies (Example 3) was produced as follows: 35 207 gg of EGFRvm PEP3 was reduced using pre-weighed DTT from Pierce (#20291). One vial of 7.7mg of pre-weighed DTT was dissolved using 100 tL of de-ionized water. The DTr -53stock was added to the EGFRvMI PEP3. The volume of the reaction was brought to 600 pL using PBS pH 7.4. The reaction was rotated for 30 minutes at room temperature. A GlO column was prepared by weighing out 5 grams of GIO sephadex beads and adding 40 mL of PBS, mixing and leaving at room temperature for 10 minutes, and then centrifuging the 5 beads at 1000 rpm for 10 minutes. The supernatant was removed and an additional 20 mL of PBS was added. The beads were centrifuged at 1000 rpm for 10 minutes. The supernatant was removed and enough PBS added to make a 50% slurry of G1O sephadex beads. 5 mL of the 50% slurry mixture was added to a 5mL spin column and the column was placed in a 14mL polypropylene tube. The column was centrifuged at 1000 rpm for 3 minutes. Another 3 mL of PBS was added 10 and the column was cenrifuged again at 1000 rpm for 3 minutes. The polypropylene tube was replaced with a new tube and the columns were now ready to use. DTT was removed from the reduced peptide. After the 30 minute reaction time for reducing the peptide, 300 pL of the reduced peptide was added per column. The column was centrifuged at 1000rpm for 3 minutes. An additional 250 pL of PBS was added to each column and 15 centrifuged again at 1000 rpm for 3 minutes. The reduced peptide was collected in the 14 mL polypropylene tube. The reduced peptide was conjugated to maleimide activated OVA and collected in an eppendorf tube. 2 mg of the maleimide activated OVA was dissolved (Pierce: 77126, Rockford IL) with maleimide conjugation buffer to make a 10 mg/mL stock. 414 gg of the maleimide activated 20 OVA was added to the reduced peptide in the eppendorf tube. 500 pL of the maleimide conjugation buffer was added to the reaction. The reaction was allowed to incubate for 2 hours at room temperature and then 2mg of cysteine was added to quench any active maleimide groups that might have been'present. The cysteine was allowed to react for 30 additional minutes at room temperature: The conjugate was then washed with a 10K centrifugal column 3 times using IX PBS 25 pH 7.4. This removed any free peptide that did not conjugate to the OVA and free cysteine. The conjugate was removed from the centrifugal column using gel loading tips and transferred to an eppendorf tube. Finally, the conjugate was brought to the desired concentration using IX PBS pH 7.4. The conjugate produced had a molar ratio of 14.5:1 (peptide:OVA) EXAMPLE 2 30 Production of anti-EGFRvlII Antibodies Through Hybridoma Generation Eight XenoMouse mice that produce antibodies with a gamma-I constant region (XenoMouse GI mice) were immunized on day 0 and boosted on days 11, 21, 32, 44 and 54 for this protocol and fusions were performed on day 58. All immunizations were conducted via subcutaneous administration at the base of tail plus intraperitoneal administartion for all injections. 35 The day 0 immunization was done with 1.5 x 107 B300.19/EGFRvIII transfected cells (Example IA) suspended in pyrogen free DPBS admixed 1:1 v/v with complete Freunds adjuvant (CFA) (Sigma, St. Louis, MO). Boosts on days 11, 21, and 32 were done with 1.5 x 107 -54- B300.19/EGFRvIII transfected cells in DPBS admixed 1:1 v/v with incomplete Freunds adjuvant (IFA) (Sigma, St. Louis, MO). The boosts on day 44 was done with 5 ig of the PEP3 (EGFRvIII peptide) - KLH conjugate (Example 1) in DPBS admixed 1:1 v/v with [FA and final boost, on day 54, was done with 5ug PEP3 (EGFRvII[ peptide) - KLH conjugate in DPBS without adjuvant. 5 On day 58, mice were euthanized, and then inguinal and Lumbar lymph nodes were recovered. Lymphocytes were released by mechanical disruption of the lymph nodes using a tissue grinder then depleted of T cells by CD90 negative selection. The fusion was performed by mixing washed enriched B cells and non-secretory myeloma P3X63Ag8.653 cells purchased from ATCC, cat. # CRL 1580 (Keaney et al, J. Immunol. 123:1548-1550 (1979)) at a ratio of 1:1. The cell 10 mixture was gently pelleted by centrifugation at 800 g. After complete removal of the supernatant, the cells were treated with 2-4 mL of Pronase solution (CalBiochem, cat. # 53702; 0.5 mg/mi in PBS) for no more than 2 minutes. Then, 3-5 ml of FBS was added to stop the enzyme activity and the suspension was adjusted to 40 ml total volume using electro cell fusion solution, ECFS (0.3M Sucrose, Sigma, Cat# S7903, 0. 1mM Magnesium Acetate, Sigma, Cat# M2545, 0.1 mM Calcium 15 Acetate, Sigma, Cat# C4705 (St. Louis, MO)). The supernatant was removed after centrifugation and the cells washed by resuspension in 40 ml ECFS. This wash step was repeated and the cells again were resuspended in ECFS to a concentration of 2x10 6 cells/ml. Electro-cell fusion was performed using a fusion generator, model ECM2001, Genetronic, Inc., San Diego, CA. The fusion chamber size used was 2.0 ml, and using 20 the following instrument settings: Alignment condition: voltage: 50 v, time: 50 s, Membrane breaking at: voltage: 3000 v, time: 30 ps, Post-fusion holding time: 3 s. After fusion, the cells were re-suspended in DMEM (JRH Biosciences),15% FCS (Hyclone), containing HAT, and supplemented with L-glutamine, pen/strep, OPI (oxaloacetate, pyruvate, bovine insulin) (all from Sigma, St. Louis, MO) and IL-6 (Boehringer Mannheim) for culture at 37 "C and 10% CO 2 in air. 25 Cells were plated in flat bottomed 96-well tissue culture plates at 4x10 4 cells per well. Cultures were maintained in HAT (hypoxanthine, aminopterin and thymidine) supplemented media for 2 weeks before transfer to HT (hypoxanthine and thymidine) supplemented media. Hybridomas were selected for by survival in HAT medium and supernatants were screened for antigen reactivity by ELISA. The ELISA format entailed incubating supernatants on antigen coated plates (EGFRvIll 30 peptide-OVA coated plates and wild type EGFr peptide-OVA coated plates as a counter screen) and detecting EGFRvIII-specific binding using horseradish peroxidase (HRP) labeled mouse anti human IgG (see Table 2.1). -55- TABLE_2.1 Plate.Weli Hybridoma 1" OD 2nd OD fusion late muEGFr EGFr 13.2 DI0 13.1 4.034 2.653 0.051 13.3 C12 13.2 3.829 2.443 0.049 13.3 Fil 13.3 3.874 1.081 0.049 13.6 B11 13.4 .3.322 1.311 0.052 Clones Plate OD #1 OD #2 cloningplate muEGFr EGFr R67..I T 13.1.1 0.5c/w D2 2.614 2.586 0.042 13.1.2 0.5c/w F5 2.248 1.272 0.041 As will be observed, at least four antigen specific hybridomas were detected: 13.1, 13.2, 13.3, and 13.4. These hybridomas that were positive in the ELISA assay EGFRvIII specificity were 5 confirmed by FACS on stably transfected 300.19 cells expressing EGFRvII versus 300.19 untransfected parental cells. Cloning was performed on selected antigen-positive wells using limited dilution plating. Plates were visually inspected for the presence of single colony growth and supernatants from single colony wells then screened by antigen-specific ELISAs and FACS confirmation as described 10 above. Highly reactive clones were assayed to verify purity of human gamma and kappa chain by multiplex ELISA using a Luminex instrument. Based on EGFRvIII specificity in the ELISA and FACS assay, Clone 13.1.2 was selected as the most promising candidate for further screening and analysis. The nucleotide and amino acid sequences of the heavy and light chains of 13.1.2 antibody are shown in FIG. 3L and SEQ ID NO: 137 and 139 for heavy and light chain nucleic acids and 138 15 and 140 for heavy and light chain amino acid sequences.. In addition, a comparison of the 13.1.2 heavy chain and light chain sequences with the germline sequence from which they were derived as shown in FIGs 4 and 5. EXAMPLE 3 Antibody Generation Through Use of XenoMax Technology 20 Immunization of XenoMouse animals Human monoclonal antibodies against human EGFRvIII were developed by sequentially immunizing XenoMouse mice that produce antibodies with a gamma-1 constant region (XenoMouse G1 mice), XenoMouse mice that produce antibodies with gamma-2 constant regions -56- (XenoMouse XMG2 mice), and XenoMouse mice that produce antibodies with a gamma-4 constant region (XenoMouse G4 mice). To generate mAbs by through XenoMax technology, cohorts of XenoMouse GI and XMG2 mice were immunized with EGFRvII PEP3 (Example IA) and EGFRvIll-expressing 5 300.19 cells (Example IB) or with bacterially expressed extracellular domain of EGFRvm protein (EGFRvIII-ECD) (Dr. Bigner, Duke University) and EGFRVIII-expressing 300.19 cells or with EGFRvII-Rabbit Fc fusion protein (EGFRvlII-RbFc) (Example IC) and EGFRvMI-expressing 300.19 cells or with EGFRvIII-RbFc only via foot pad (FP), or via base of the tail by subcutaneous injection and intraperitoneum (BIP). 10 For footpad immunizations, the initial immunization was with or without 10 X 10' EGFRvII-expressing 300.19 cells and with or without 10 lg of EGFRvIII PEP3 or EGFRvIII-ECD or EGFRvIII-RbFc mixed 1:1 v/v with Titermax gold (Sigma, Oakville, ON) per mouse. The subsequent boosts were performed with half of the amount of immunogen used in the initial immunization. The first four boosts were done by taking the immunogen mixed with alum (Sigma, 15 Oakville, ON), adsorbed overnight, per mouse as shown in the Table 3.1 below. This was followed by one injection with the respective immunogen in Titermax gold, one injection with alum and then a final boost of the immunogen in PBS as shown in Table 3.1. In particular, animals were immunized on days 0, 3, 7, 10, 14, 17, 21 and 24. The animals were bled on day 19 to obtain sera and determine the titer for harvest selection. The animals were harvested on Day 28. 20 Table 3.1 Footpad immunization schedule Group # 1 2 3 4 5 6 7 a # of animals 5 5 5 5 5 5 5 5 Mouse strain XMG2 XM3C-3 XMG2 XM3C-3 XMG2 XM3C-3 XMG2 XM3C-3 Boost Adjuvant Immunogen Immunogen Immunogen immunogen 1 Titermax EGFRvllI-300.19 EGFRvIII-300.19 EGFRvIII-300.19 EGFRvill-RbFc gold cells + PEP3- cells + EGFRvill- cells + EGFRvIll KLH ECD RbFc 2" Alum EGFRvill-300.19 EGFRvIll-300.19 EGFRvIll-300.19 EGFRvll-RbFc cells cells cells 3" Alum PEP3-KLH EGFRvill-ECD EGFRvIll-ECD EGFRvIll-RbFc Alum EGFRvII-300.19 EGFRvIll-300.19 EGFRvII-300.19 EGFRvIll-RbFc cells cells cells 51 Alum PEP3-KLH EGFRvill-ECD EGFRvill-ECD EGFRvIll-RbFc 6" Titermax EGFRvill-300.19 EGFRvill-300.19 EGFRvII-300.19 EGFRvIll-RbFc gold cells cells cells Alum PEP3-KLH EGFRvIll-ECD EGFRvIll-ECD EGFRvIll-RbFc 8 PBS EGFRvill-300.19 EGFRvIII-300.19 EGFRvIll-300.19 EGFRvIll-RbFc cells + PEP3- cells + EGFRvIII- cells + EGFRvill KLH ECD RbFc I Harvest , -57- The initial BIP immunization with the respective immunogen, as described for the footpad immunization, was mixed 1:1 v/v with Complete Freund's Adjuvant (CFA, Sigma, Oakville, ON) per mouse. Subsequent boosts were made first with the immunogen respectively, mixed 1:1 v/v with Incomplete Freund's Adjuvant (IFA, Sigma, Oakville, ON) per mouse, followed by a final 5 boost in PBS per mouse. The animals were immunized on days 0, 14, 28, 42, 56, and day 75 (final boost) as shown in Table 3.2 below. The animals were bled on day 63 to obtain sera and determine the titer for harvest selection. The animals were harvested on Day 78. Table 3.2 10 Bip Immunization schedule Group 9 10 11 12 13 14 15 16 # of animals 5 5 5 5 5 5 5 5 Mouse strain XMG2XM3C-3 XMG2 XM3C-3 XMG2 XM3C-3 XMG2 XM3C-3 Boost Adjuvant Immunogen Immunogen Immunogen Immunogen 1" CFA EGFRvII-300.19 EGFRvill-300.19 EGFRvlIlI-300.19 EGFRvIll-RbFc cells + PEP3- cells + EGFRvlIl- cells + EGFRvIll KLH ECD RbFc 2"" IFA EGFRvII-300.19 EGFRvIll-300.19 EGFRvill-300.19 EGFRvIll-RbFc cells cells cells 3'a IFA PEP3-KLH EGFRvill-ECD EGFRvIll-ECD EGFRvill-RbFc 4 IFA EGFRvII-300.19 EGFRvill-300.19 EGFRvIll-300.19 EGFRvill-RbFc ___ cells cells cells 5 IFA PEP3-KLH EGFRvIll-ECD EGFRvill-ECD EGFRvill-RbFc 6' PBS EGFRvill-300.19 EGFRvII-300.19 EGFRvill-300.19 EGFRvIll-RbFc cells + PEP3- cells + EGFRvIll- cells + EGFRvill KLH ECD RbFc Harvest Selection of animals for harvest By titer determination Anti-hEGFRvIII antibody titers were determined by ELISA. EGFRvll-RbFc (2.5 Rg/ml) or a control RbFc (2 pg/ml) or EGFRvlIlpeptide-OVA (2 pg/ml) (Example 1) or control OVA (4 15 pg/ml) were coated onto Costar Labcoat Universal Binding Polystyrene 96-well plates (Coming, Acton, MA) overnight at four degrees. The solution containing unbound antigen was removed and the plates were treated with UV light (365nm) for 4 minutes (4000 microjoules). The plates were washed five times with dH20. Sera from the EGFRvIII immunized XenoMouse@ animals, or narve XenoMouse@ animals, were titrated in 2% milk/PBS at 1:2 dilutions in duplicate from a 1:100 20 initial dilution. The last well was left blank. The plates were washed five times with dH 2 0. A goat anti-human IgG Fc-specific horseradish peroxidase (HRP, Pierce, Rockford, IL) conjugated antibody was added at a final concentration of 1 pg/mL for 1 hour at room temperature. The plates were washed five times with dH 2 O. The plates were developed with the addition of TMB chromogenic substrate (Gaithersburg, MD) for 30 minutes and the ELISA was stopped by the 25 addition of 1 M phosphoric acid. The specific titer of individual XerioMouse@ animals was -58determined from the optical density at 450 nm and is shown in Tables 3.3 and 3.4. The titer represents the reciprocal dilution of the serum and therefore the higher the number the greater the humoral immune response to hEGFRvIII. For the mice immunized via base of the tail by subcutaneous injection and intraperitoneum, 5 the titre was determined exactly as above except the plates were coated with EGFRvIII-RbFc (2.0 pg/ml) or a control RbFc (2.5 pg/ml). Table 3.3 EGFRviII Immunization Mouse EGFRvm Control peptide- OVA Group (site and Strain Duse -RbFc @ RbFc @ OVA coated at Immunogen) and sex 2.Sug/ml. 2.0ug/ml. coated at 4.0 sg/ml 2.0 pg/mi FP 0748-1 330 1 13549 <100 EGFRvll- 0748-2 237 7635 <100 300.19 cells + 0748-3 109 9824 <100 1 EGFRvIl XMG2 0748-4 714 8014 <100 PEP3-KLH 0748-5 165 9421 <100 (see lmr. Sched.) Naive <100 .n/a n/a FP 0741-1 388 347 <100 EGFRvll- 0741-2 327 240 <100 300.19 cells + XM3C- 0741-3 385 330 <100 2 EGFRvIIl 3 0741-4 589 227 <100 PEP3-KLH 0741-5 273 626 <100 (see Imm. Sched.) Naive <100 n/a n/a FP 0749-1 552 <100 <100 EGFRvill- 0749-2 477 <100 <100 300.19 cells + 0749-3 100 <100 <100 3 EGFRvIll- XMG2 0749-4 100 <100 <100 ECD (see Imm. 0749-5 1631 <100 <100 Sched.) NaYve 100 n/a n/a FP 0742-1 372 <100 <100 EGFRvIll- 0742-2 745 <100 <100 4 300.19 cells + XM3C- 0742-3 484 <100 <100 EGFRvlll- 3 0742-4 530 <100 <100 ECD (see Imm. 0742-5 270 <100 <100 Sched.) Naive 100 n/a n/a FP 0750-1 5399 175 <100 <100 EGFRvll- 0750-2 3072 151 <100 <100 300.19 cells + 0750-3 >6400 358 <100 <100 EGFRvll- 0750-4 5845 196 <100 <100 RbFc (see 0750-5 5770 196 <100 <100 Imm. Sched.) Naive 100 100 n/a n/a FP 0743-1 1220 <100 <100 <100 EGFRvll- 0743-2 1183 <100 <100 <100 6 300.19 cells + XM3C- 0743-3 645 <100 <100 <100 EGFRvill- 3 0743-4 759 <100 <100 <100 RbFc (see 0743-5 1260 <100 <100 <100 Imm. Sched.) Naive 100 <100 n/a n/a -59- EGFRvIII Immunization Mouse Mouse EGFRvIII Control peptide- OVA (site and Strain IDs -RbFc @ RbFc @ OVA coated at Immunogen) and sex 2.5ug/ml. 2.Oug/ml. coated at 4.0 sg/ml 2.0 tg/mil 0745-1 1897 <100 <100 <100 FP 0745-2 >6400 323 <100 <100 7 EGFRvIll- XMG2 0745-3 1225 <100 <100 <100 RbFc (see 0745-4 4047 <100 <100 <100 Imm. Sched.) 0745-5 852 <100 <100 <100 Naive 100 <100 n/a n/a 0744-1 362 <100 <100 <100 FP 0744-2 807 <100 <100 <100 EGFRvIll- XM3C- 0744-3 479 <100 <100 <100 RbFc (see 3 0744-4 631 <100 <100 <100 Imm. Sched.) 0744-5 1112 <100 <100 <100 I I Naive 100 <100 n/a n/a All the XenoMouse animals from group 5 and XenoMouse animals 0743-5 from group 6 from Table 3.3 were selected for XenoMax harvests based on the serology. 5 Table 3.4 EGFRvII Group Immunization Mouse Mouse EGFRvIII Control peptide- OVA (site and Strain I.Ds -RbFc @ RbFc @ OVA coated at Immunogen) and sex 2.0ug/ml. 2.5ug/ml. coated at 4.0 pig/ml 2.0 pg/mI BIP 0695-1 2921 >128000 472 EGFRvIll- 0695-2 2219 30504 379 300.19 cells + 0695-3 4609 >128000 608 9 EGFRvill XMG2 0695-4 >6400 >128000 368 PEP3-KLH 0695-5 1580 19757 269 (see 1mm. Sched.) Naive <100 n/a 242 BIP 0700-1 <100 EGFRvill- 0700-2 <100 300.19 cells + XM3C- 0700-3 >6400 10 EGFRvill 3 0700-4 5342 PEP3-KLH 0700-5 >6400 (see Imm. Sched.) Naive <100 BIP 0696-1 <100 561 240' EGFRvIll- 0696-2 <100 788 326 300.19 cells + 0696-3 <100 604 266 EGFRvIll- 0696-4 143 444 263 ECD (see 1mm. 0696-5 <100 303 254 Sched.) Naive <100 242 12 BIP XM3C- 0702-1 358 EGFRvill- 3 0702-2 469 300.19 cells + 0702-3 401 EGFRvill- 0702-4 >6400 ECD (see Inun. 0702-5 >6400 -60- EGFRvII Group Immunization Mouse Mouse EGFRvII Control peptide- OVA (site and Strain I.Ds -RbFc @ RbFc @ OVA coated at Immunogen) and sex 2.0ug/nil. 2.5ug/mI. coated at 4.0 pg/mi 2.0 pg/ml Sched.) Naive <100 BIP 0694-1 >6400 >6400 250 243 EGFRvill- 0694-2 >6400 >6400 296 309 13 600.19 cells + XMG2 0694-3 >6400 >6400 736 605 EGFRvill- 0694-4 >6400 >6400 739 1111 RbFc (see 0694-5 3710 >6400 517 465 Imm. Sched.) Naive <100 >6400 242 BIP 0703-1 2740 >6400 EGFRvill- 0703-2 408 >6400 14 300.19 cells + XM3C- 0703-3 1406 >6400 EGFRvIll- 3 0703-4 1017 >6400 RbFc (see 0703-5. 403 >6400 Imm. Sched.) Naive <100 >6400 0697-1 >6400 >6400 340 348 BIP 0697-2 >6400 >6400 642 1793 15 EGFRvllI- XMG2 0697-3 6242 >6400 319 246 RbFc (see 0697-4 1766 >6400 133 <100 bmm. Sched.) 0697-5 >6400 >6400 685 448 NaYve <100 >6400 243 242 0701-1 592 >6400 BIP 0701-2 1118 >6400 16 EGFRvIll- XM3C- 0701-3 >6400 >6400 RbFc (see 3 0701-4 <100 <100 Imm. Sched.) 0701-5 n/a n/a I Nave <100 >6400 XenoMouse animals (0695-1, 0695-3 and 0695-4) were selected for harvests based on the serology data in Table 3.4. Selection of B cells, 5 B-cells from the above-discussed animals were harvested and cultured. Those secreting EGFRvIII-peptide specific antibodies were isolated as described in Babcook et al., Proc. Natl. Acad. Sci. USA, 93:7843-7848 (1996). ELISA was used to identify primary EGFRvm-peptide OVA -specific wells. About 5 million B-cells were cultured from XenoMouse animals in 245 96 well plates at 500 or 150 or 50 cells/well, and were screened on EGFRvIII-peptide-OVA to identify 10 the antigen-specific wells. About 515 wells showed ODs significantly over background, a representative sample of which are shown in Table 3.5. -61- Table 3.5 Total Positives above cutoff OD of: # of ____J__ plates 0,0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.5 2.0 2.5 3.0 3.5 cansera 12 11 6 81 56 49 45 38 32 29 26 25 18 11 4 1 0 cells/ 52 4 well ___ Sigma 500 13 12 77 19 13 11 99 80 73 58 53 49 21 9 5 0 cells/ 1 48 3 5 9 7 0 well Sigma c15ls 20 0 04 8 91 67 55 47 45 36 33 19 9 5 2 0 well t 45 4 27 75 37 25~ 21 1 151311 105829 Total 45 20 11 4 3 7 1 3 2 2 5 7 244 of EGFRvIII-peptide-OVA-Elisa positive wells of OD > 0.5 were screened again on 5 EGFRvIII-peptide-OVA and on OVA to confirm that they were EGFRvIII-peptide specific. A representative example of these results is shown in Table 3.6. Table 3.6 1' EGFRvM 2' EGFRvml OVA Plate Well peptide-OVA peptide-OVA OD OD OD 121 G 1 0.7534 1.4065 0.135 121 A 7 1.3472 2.1491 0.126 121 D 8 0.6743 0.4179 0.153 121 E 8 2.0415 2.6965 0.148 121 H 10 0.8611 0.4288 0.159 121 C 12 2.1455 2.6443 0.140 4 122 H 1 1.8890 2.5987 0.116 4 122 H 5 0.5943 0.8321 0.157 122 F 8 0.6834 0.7715 0.145 10 Limited antigen assay and analysis The limited antigen analysis is a method that affinity-ranks the antigen-specific antibodies present in B-cell culture supernatants relative to all other antigen-specific antibodies. In the -62presence of a very low coating of antigen, only the highest affinity antibodies should be able to bind to any detectable level at equilibrium. (See, e.g., International Patent Application No. WO 03/48730) EGFRvIII peptide-OVA was coated to plates at three concentrations; 7.5 ng/ml, 1.5 ng/ml 5 and 0.03 ng/ml for overnight at 40 C on 96-well Elisa plates. Each plate was washed 5 times with dH 2 0, before 50ul of 1% milk in PBS with 0.05% sodium azide were added to the plate, followed by 4 pl of B cell supernatant added to each well. After 18 hours at room temperature on a shaker, the plates were again washed 5 times with dH 2 0. To each well was added 50ul of Gt anti-Human (Fc)-HRP at 1 pg/ml. After I hour at room temperature, the plates were again washed 5 times with 10 dH 2 0 and 50 gi of TMB substrate were added to each well. The reaction was stopped by the addition of SOuL of IM phosphoric acid to each well and the plates were read at wavelength 450nm and the results shown in Table 3.7. Table 3.7 High Limited Ag Antigen Culture Well (1.0sg/mi) Plate 0.03ng/ml 1.5nglml 7.5ngmlI OD Rank OD Rank OD Rank 133 B 2 0.7670 1 1.189 54 1.871 95 2.050 124 G 12 0.7400 2 1.895 1 3.101 1 3.463 145 C 1 0.715 3 1.552 7 2.671 10 3.194 f29 G 10 0.6720 4 1.367 22 2.692 8 2.977 186 B 6 0.657 5 1.842 2 2.859 3 3.411 143 F 12 0.653 6 1.677 3 2.741 6 3.156 136 E 3 0.6340 7 1.468 15 2.683 9 3.280 137 C 11 0.595 8 1.582 5 2.94 2 3.444 139 A 11 0.582 9 1.374 19 2.282 47 2.255 174 F 1 0.573 10 1.577 6 2.775 4 2.364 15 The results generated from limited antigen analysis were compared to the total OD obtained in high antigen assay. A relative ranking of affinity was done by taking the ratio of the OD obtained in limited antigen assay Vs that obtained in high antigen assay. Antibodies with higher 20 ratio will have the highest affinity. Table 3.7 shows the sample of B-cell culture supernatants that were ranked based on limited antigen assay OD (for the lowest antigen plating concentration of 0.03 ng/ml) Vs the high antigen assay OD. Native cell binding assay by FMAT EGFRvIII peptide-OVA-Elisa positive well supernatants were analyzed for their ability to 25 bind to the native form of EGFRvm stably expressed on NR6 cells (NR6 M cells) (See, Batra et al. Epidermal growth factor ligand-independent, unregulated, cell-transforming potential of a naturally -63occurring human mutant EGFRvlI gene. Cell Growth Differ. 6(10):1251-9 (1995)). NR 6 M cells were seeded at 8000 cells per well and incubated over night in 96 well FMAT plates. Media was then removed leaving 15 Il in the well. 15 pl B-cell culture supernatants were added and 15 il anti-human IgG Fc CyS-at 1 pg/ml final concentration added to wells. It is then left incubated at 4* 5 C for 2 hours. The cells were washed with 150 pl PBS, and fixed before reading on FMAT. The results were expressed as total fluorescent intensity (Table 3.8). Human anti-EGFRvUIl mAb 13.1.2 was used as a positive control starting at 1 pg/ml final concentration and negative control was PK 16.3.1 at the same concentration. 134 of the 244 samples tested bound to NR6M cells of which 62 had a total fluorescence of greater than 8000. 6 of these 134 binders were false positives. 10 The same type of native binding assay was done on NR6 Wt cells (NR6 cells expressing EGF receptor) (See Batra et al. Epidermal growth factor ligand-independent, unregulated, cell transforming potential of a naturally occurring human mutant EGFRvMI gene. Cell Growth Differ. 6(10):1251-9 (1995)) to eliminate the binding is due to binding to Wt receptor (Table 3.8). ABX EGF was used as a positive control and PK 16.3.1 at the same concentration was used as a negative 15 control antibody. 3 out the 134 NR6 M binders were binding strongly to NR6 Wt cells. 190 of the 244 wells bound EGFRvUI peptide in Elisa were also bound to the native form on cells. Examples are given in Table 3.8. Table 3.8 FMAT FMAT , , native native Plate VIApep 2' VA-pep- OVA OD binding binding to NR6 M to NR6 Wt calls cells 174 F 1 2.4945 3.0308 0.1900 138373 1668 187 A 4 1.5337 1.2085 0.1920 128626 202459.8 132 D 8 0.8555 1.2070 0.1649 109379 0 142 C 11 2.2889 2.8194 0.2239 94944 0 129' A 7 2.1501 2.8208 0.1515 84024 0 127 E 1 2.6923 3.1986 0.1219 82031 0 124 G 12 3.2929 3.5634 0.1455 73080 0 141 C 6 0.7512 1.2567 0.1547 60816 814.5 173 C 1 2.5728 2.5714 0.2134 58702 2523.4 128 G 9 0.6293 0.7483 0.1520 49631 0 129 H 6 2.9370 3.0952 0.2582 0 0 183 E 11 2.3450 2.7717 0.1050 0 0 20 In Table 3.8, supernatant from well 187A4 is identified as a Wt binder and 141C6 was a false positive for NR6 M cells binding. Wells 129H6 and 183E II are strong peptide binders with no native binding. 25 -64- Interalization assay The top 60 native binding B cell culture supernatants were further assayed for their ability to internalize the receptor. NR6 M cells were seeded at 8000 cells/well into 96 well FMAT plates and incubated overnight. Media was removed and 10-15 il B-Cell culture supernatant in a total 5 volume of 30 til media, in duplicate was added. Next, 15 pl of secondary antibody (SS Alexa 647 anti-hquman IgG Fab at 1.5 pg/mI final concentration) was added and the mixture was incubated on ice for 1 hr. An irrelevant B-Cell Culture supernatant was used to see the effect of the culture media. Human anti-EGFRvIII mAb 13.2.1 was used as a positive control starting at I pg/ml (final concentration) and negative control was PK 16.3.1 (human anti-KLH IgG2 antibody) at the same 1O concentration. After incubation, the cells were washed with cold PBS, 50 pl media was added to all of the wells, one of the duplicates were incubated at 37 *C for 30 mins while the other duplicate remained on ice. After the incubations media was removed, 100ul of cold 50 mM glutathione was added to the set incubated at 37 *C and 100 RI of cold media added to the other set, both sets were then left on ice for 1 hr. The cells were then washed with 100 pl cold PBS and then fixed with 1% 15 paraformaldehyde and read in FMAT. The results were expressed as % internalized, calculated as total fluorescence in the presence of glutathione/ total fluorescence in the absence of glutathione X 100. Representative information is given in Table 3.9. Table 3.9 No With . No With internalized, Well no. glutathlone glutathione (ntrnluzed FLIxcount FLIxcount (glut+/g ut-) 124 C9 1877 1394 74.3% 124 G12 26465 9959 37.6% 125 H1 14608 3686 25.2% 125 D10 2342 1236 52.8% 127 El 15059 1318 8.7% 127 B9 12444 7109 57.1% 127 El 6623 0 0.0% 128 G9 10071 1851 18.4% 129 A7 27648 8708 31.5% 130 B4 4558 4354 95.5% 131 H5 9258 2656 28.7% 132 D8 35820 13293 37.1% 133 F9 9773 3621 37.0% 136 F10 2392 0 0.0% 137 G6 5104 1021 20.0% 137 G1O 3451 0 0.0% 20 -65- EGFRvIII-specific Hemolytic Plaque Assay. A number of specialized reagents were needed to conduct this assay. These reagents were prepared as follows. 1. Biotinylation of Sheep red blood cells (SRBC). S RBCs were stored in RPMI media as a 5 25% stock. A 250 pl SRBC packed-cell pellet was obtained by aliquoting 1.0 ml of SRBC to a fresh eppendorf tube. The SRBC were pelleted with a pulse spin at 8000 rpm (6800 ref) in microfuge, the supernatant drawn off, the pellet re-suspended in 1.0 ml PBS at pH 8.6, and the centrifugation repeated. The wash cycle was repeated 2 times, then the SRBC pellet was transferred to a 15-ml falcon tube and made to 5 ml with PBS pH 8.6. In a separate 50 ml falcon 10 tube, 2.5 mg of Sulfo-NHS biotin was added to 45 ml of PBS pH 8.6. Once the biotin had completely dissolved, the 5 ml of SRBCs were added and the tube rotated at RT for 1 hour. The SRBCs were centrifuged at 3000rpm for 5 min and the supernatant drawn off. The biotinylated SRBCs were transferred to an eppendorf tube and washed 3 times as above but with PBS pH 7.4 and then made up to 5 ml with immune cell media (RPMI 1640) in a 15 ml falcon tube (5% B 15 SRBC stock). Stock was stored at 4o C until needed. 2. Streptavidin (SA) coating of B-SRBC. 1 ml of the 5% B-SRBC stock was transferred into a fresh eppendorf tube. The B-SRBC cells were washed 3 times as above and resuspended in 1.0 ml of PBS at pH 7.4 to give a final concentration of 5% (v/v). 10 pl of a 10 mg/ml streptavidin (CalBiochem, San Diego, CA) stock solution was added and the tube mixed and rotated at RT for 20 20 min. The washing steps were repeated and the SA-SRBC were re-suspended in Iml PBS pH 7.4 (5% (v/v)). 3. EGFRvII coating of SA-SRBC. The SA-SRBCs were coated with biotinylated EGFRvmpetide-OVA at 10 pig/ml, mixed and rotated at RT for 20 min. The SRBC were washed twice with 1.0 ml of PBS at pH 7.4 as above. The EGFRvIII-coated SRBC were re-suspended in 25 RPMI (+10%FCS) to a final concentration of 5% (v/v). 4. Determination of the quality of EGFRvIIpeptide-SRBC by immunofluorescence (IF). 10 pl of 5% SA-SRBC and 10 pl of 5% EGFRvIII peptide-coated SRBC were each added to a separate fresh 1.5 rn eppendorf tube containing 40ul of PBS. A control human anti-EGFRvII antibody was added to each sample of SRBCs at 45 gg/ml. The tubes were rotated at RT for 25 30 min, and the cells were then washed three times with 100 ji of PBS. The cells were re-suspended in 50 pil of PBS and incubated with 40 mcg/mL Gt-anti Human IgG Fc antibody conjugated to Alexa488 (Molecular Probes, Eugene, OR). The tubes were rotated at RT for 25 min, and then washed with 100 jil PBS and the cells re-suspended in 10 ji PBS. 10 pl of the stained cells were spotted onto a clean glass microscope slide, covered with a glass coverslip, observed under 35 fluorescent light, and scored on an arbitrary scale of 0-4. 5. Preparation of plasma cells. The contents of a single microculture well previously identified by various assays as containing a B cell clone secreting the immunoglobulin of interest -66were harvested. Using a 100-1000 p1 pipetman, the contents of the well were recovered by adding 37C RPMI (10% FCS). The cells were re-suspended by pipetting and then transferred to a fresh 1.5 ml eppendorf tube (final vol. approx 500-700 p1). The cells were centrifuged in a microfuge at 2500 rpm (660 rcf) for 1 minute at room temperature, and then the tube was rotated 180 degrees and spun 5 again for 1 minute at 2500 rpm. The freeze media was drawn off and the immune cells resuspended in 100 pl RPNfl (10% FCS), then centrifuged. This washing with RPMI (10% FCS) was repeated and the cells re-suspended in 60 pl RPMI (10% FCS) and stored on ice until ready to use. 6. Micromanipulation ofplasma cells. Glass slides (2 x 3 inch) were prepared in advance with silicone edges and allowed to cure overnight at RT. Before use, the slides were treated with 10 approx. Sul of SigmaCoat (Sigma, Oakville, ON) wiped evenly over glass surface, allowed to dry and then wiped vigorously. To a 60 p1 sample of cells was added 60 pl each of EGFRvflIpeptide coated SRBC (5% v/v stock), 4x guinea pig complement (Sigma, Oakville, ON) stock prepared in RPMI (10% FCS), and 4x enhancing sera stock (1:150 in RPMI with 10% FCS). The mixture was spotted (10-15 RI) onto the prepared slides and the spots covered with undiluted paraffin oil. The 15 slides were incubated at 37* C for a minimum of 45 minutes. The EGFRvIII-specific plasma cells were identified from plaques and rescued by micromanipulation (see Table 3.10). Table 3.10 Well ID Single Cell Number Total number of Single cells picked 124 G 12 EGFRvill-SCX-105-116 (LL) 12 129 A 7 EGFRvIll -SCX-117-128 (DM) 12 174 F I EGFRvlli -SCX-129-137 (DM) 9 182 A 5 EGFRvill -SCX-138-149 (LL); 162-169 (OP) 20 125 D 10 EGFRvII -SCX-170-181 (DM); 194-201 (LL) 20 127 B 9 EGFRviI -SCX-182-193 (LL); 202-209 (OP) 20 190 D 7 EGFRvill -SCX-210-229 (LL) 20 130 B 4 EGFRvIll -SCX-230-249 (LL) 20 138 D 2 EGFRviI -SCX-250-269 (LL) 20 145 C 1 EGFRvill -SCX-80-92 (DM) 13 172 B 12 EGFRvll -SCX-93-104 (LL) 12 187 A 4 EGFRvlli -SCX-270-281 (LL) 12 173 C I EGFRvIll -SCX-282-293 (BC) 12 127 E I EGFRvillI -SCX-294-305 (LL) 12 142 C 11 EGFRvII -SCX-306-317 (LL) 12 -67- Well ID Single Cell Number Total number of Single cells picked 141 A 10 EGFRvll -SCX-318-329 (BC) 12 132 D 8 EGFRvllI -SCX-330-341 (LL) 12 124 D 4 EGFRvIll -SCX-342-349 (BC) 8 Single cell PCR, Cloning, Expression, Purification and Characterization of Recombinant anti EGFRvUI Antibodies. The genes encoding the variable regions were rescued by RT-PCR on the single 5 micromanipulated plasma cells. mRNA was extracted and reverse transcriptase PCR was conducted to generate cDNA. The cDNA encoding the variable heavy and light chains was specifically amplified using polymerase chain reaction. The human variable heavy chain region was cloned into an IgGI expression vector. This vector was generated by cloning the constant domain of human IgG1 into the multiple cloning site of pcDNA3. 1+/Hygro (Invitrogen, Burlington, 10 ON). The human variable light chain region was cloned into an IgK expression vector. These vectors were generated by cloning the constant domain of human IgK into the multiple cloning site of pcDNA3.l+/Neo (Invitrogen, Burlington, ON). The heavy chain and the light chain expression vectors were then co-lipofected into a 60 mm dish of 70% confluent human embryonal kidney 293 cells and the transfected cells were allowed to secrete a recombinant antibody with the identical 15 specificity as the original plasma cell for 24-72 hours. The supernatant (3 mL) was harvested from the HEK 293 cells and the secretion of an intact antibody was demonstrated with a sandwich ELISA to specifically detect human IgG (Table 3.11). Specificity was assessed through binding of the recombinant antibody to EGFRvIII using ELISA (Table 3.11). 20 Table 3.11 mAb ID Cell # _______Titer m __b___D__Cel ____#_ Total antibody Antigen binding 129A7 SC- EGFRvII -XG1-123/124 >1:64 >1:64 138D2 SC- EGFRvIll -XG1-250 >1:64 >1:64 174F1 SC- EGFRvIll -XGI-131 >1:64 >1:64 182A5 SC- EGFRvill -XG1-139 >1:64 >1:64 190D7 SC- EGFRvill -XG1-211 >1:64 >1:64 125D10 SC- EGFRvill -XG2-170 >1:64 >1:64 182D5 SC- EGFRvIll -XG2-150 >1:64 >1:64 141A10 SC- EGFRvIll -XG1-318 1:64 1:64 132D8 SC- EGFRvII -XG1-333 >1:64 >1:64 124D4 SC- EGFRvIll -XG1-342 >1:64 >1:64 -68- The secretion ELISA tests were performed as follows. For Ab secretion, 2 pg/mL of Goat anti-human IgG H+L and for antigen binding, 1.5 ig/ml of EGFRvmI-Rab Ig Fc fusion protein was coated onto Costar Labcoat Universal Binding Polystyrene 96 well plates and held overnight at four degrees. The plates were washed five times with dH 2 0. Recombinant antibodies were titrated 1:2 5 for 7 wells from the undiluted minilipofection supernatant. The plates were washed five times with dH 2 0. A goat anti-human IgG Fc-specific HRP-conjugated antibody was added at a final concentration of 1 pg/mL for 1 hour at RT for the secretion plates and binding plates detected with ilpg/ml Rb anti Hu Fc for 1 hour at room temperature. The plates were washed five times with dH 2 O. The plates were developed with the addition of TMB for 30 minutes and the ELISA was 10 stopped by the addition of I M phosphoric acid. Each ELISA plate was analyzed to determine the optical density of each well at 450 nm. Sequencing and sequence analysis The cloned heavy and light chain cDNAs were sequenced in both directions and analyzed to determine the germline sequence derivation of the antibodies and identify changes from germline 15 sequence. Such sequences are provided in FIGs. 3A-3K and (SEQ ID NO: 34-55). A comparison of each of the heavy and light chain sequences and the germline sequences from which they are derived is provided in FIGs 4-7. In addition, the sequence of the hybridoma derived 13.1.2 antibody is compared to its germline sequence in FIGs. 4 and 5. As will be appreciated from the discussion herein, each of the 131 antibody and the 13.1.2 20 antibody possess very high affinities for EGFRvII, are internalized well by cells, and appear highly effective in cell killing when conjugated to toxins. Intriguingly, each of the antibodies, despite having been generated in different immunizations of XenoMouse mice, and utilizing different technologies, each are derived from very similar germline genes. Based upon epitope mapping work (described herein), each of the antibodies, however, appear to bind to slightly different 25 epitopes on the EGFRvM molecule and have slightly different residues on EGFRvm that are essential for binding. These results indicate that the germline gene utilization is of importance to generation of antibody therapeutics targeting EGFRvIII and that small changes can modify the binding and effects of the antibody in ways that allow further design of antibody and other therapeutics based upon these structural findings. 30 Binding of Anti-EGFRvM mAbs to native EGFRvIII expressed on cells In this example, binding of anti-EGFRvm antibodies to NR6 M cells was measured. Specifically, unquantitated supernatants of XenoMax derived IgGl recombinant antibodies were assayed for their ability to bind to NR6 M and NR6 WT cells. Cells were seeded at 10000 / well and incubated overnight at 37 C in FMAT 96 well plates. Media was removed and 40 pl mini lipo 35 supernatant (titrated down) was added, the cells were incubated on ice for 1 hr. The human 13.1.2 EGFRvmH antibodies and ABX EGF (E7.6.3, U.S. Patent No. 6,235,883) antibodies were added as positive controls. The PK 16.3.1 antibody was used as a negative control. The cells were washed -69with Cold PBS, secondary antibody was added (SS Alexa antihuman IgG Fc) at 1 pg/ml, 40 pl/well and incubated on ice for 1 hr. The cells were then washed with Cold PBS and fixed and read by FMAT. All antibodies were tested for specificity for binding by counter screening against NR6 WT cells. 5 Purification of Recombinant Anti-EGFRvIII Antibodies. For larger scale production, heavy and light chain expression vectors (2.5 gg of each chain/dish) were lipofected into ten 100 mm dishes that were 70% confluent with HEK 293 cells. The transfected cells were incubated at 37*C for 4 days, the supernatant (6 mL) was harvested and replaced with 6 mL of fresh media. At day 7, the supernatant was removed and pooled with the 10 initial harvest (120 mL total from 10 plates). Each antibody was purified from the supernatant using a Protein-A Sepharose (Amersham Biosciences, Piscataway, NJ) affinity chromatography (1 mL). The antibody was eluted from the Protein-A column with 500 mcL of 0.1 M Glycine pH 2.5. The eluate was dialyzed in PBS, pH 7.4 and filter-sterilized. The antibody was analyzed by non reducing SDS-PAGE to assess purity and yield. Concentration was also measured by UV analysis 15 at OD 250. Internalization of EGFRvJII receptor by recombinant anti-EGFRvAII mAbs XenoMax derived IgG I recombinant antibodies were expressed, purified and quantitated as described previously. Antibodies were further assayed for their ability to internalize the EGFRvII receptor in NR6 M cells. 250,000 NR6 M cells were incubated with primary antibody (SC95, 20 SC131, SC133, SC139, SCl50, SC170, SC211, SC230, SC250 and human 13.1.2 as a control) at 0.25 pg/ml, 7 mins on ice in 96 well v-bottomed plate in triplicate. The cells were washed with cold 10% FCS in PBS and secondary antibody (SS Alexa antihuman IgG Fab) at 3 pg/ml Fab was added and incubated for 7 mins on ice. The cells were washed with cold 10% FCS in PBS once and then resuspended in cold media. Next, two sets of the triplicate were incubated at 37 *C and 25 the remaining set was incubated at 4 *C for I hr. After that the cells incubated at 4 *C and one set of the cells incubated at 37 *C were treated with glutathione (as previously mentioned) for 1 hr on ice. Then the cells were washed and resuspended in 100 pl of cold 1% FCS in PBS and analyzed by FACS. The % internalization was calculated from the geometric mean obtained from the FACS analysis [(mean at 37 *C with glutathione - mean at 4 *C with glutathione) / (mean at 37 *C without 30 glutathione - mean at 4 C with glutathione)]. NA means that a FACS analysis was performed but the data was not provided in Table 3.12. -70- Table 3.12 FACS Geometric mean mAb Without With With glutathione % Internalization glutathione glutathlone 4 *C 37 *C 37 0C 13.1.2 22.12 19.19 5.38 82.5% sc95 22.56 17.75 5.13 72.4% scl31 NA NA NA 72% sc133 23.39 18.63 6.24 72.2% sc139 22.64 19.23 4.88 80.8% sc150 20.29 7.78 4.66 20.0% sc170 19.97 7.75 4.67 20.1% sc211 20.76 8.23 4.78 21.6% sc230 20.68 7.97 5.02 18.8% sc250 24.13 8.07 4.84 16.7% 13.1.2 is an antibody that was generated through hybridoma generation (Example 2) that was directed against the EGFRvH epitope previously and was used as a positive control in this 5 experiment. These results in Table 3.12 demonstrate the presence of two subsets of antibodies, those that are efficiently internalized (70-80%) and those that are not (22% or less). Example 4 Epitope MaDing Of Human Anti EGFRvfi Antibodies In order to determine the epitopes to which certain of the antibodies of the present invention 10 bound, the epitopes of 6 human and 3 murine monoclonal antibodies (mabs) against EGFRvI were mapped using synthetic peptides derived from the specific EGFRvM peptide sequence. The antibodies mapped were the human hybridoma derived anti-EGFRvII 13.1.2 antibody, the human XenoMax derived anti-EGFRvM 131, 139, 250, 095, and 211 antibodies and the murine anti EGFRvlII HIO, Y10, and B9 antibodies (from Dr. D. Bigner, Duke University). 15 The approach that was used was a custom SPOTs peptide array (Sigma Genosys) to study the molecular interaction of the human anti-EGFrVm antibodies with their peptide epitope. SPOTs technology is based on the solid-phase synthesis of peptides in a format suitable for the systematic analysis of antibody epitopes. Synthesis of custom arrayed oligopeptides is commerically available from Sigma-Genosys. A peptide array of overlapping oligopeptides derived from the amino-acid 20 sequence of the EGFr VIII variant was ordered from Sigma-Genosys. A series of nine 12-mer peptides were synthesized as spots on polypropylene membrane sheets. The peptide array spanned residues 1- 20 of the EGFrVM sequence, representing the deletion of amino acids 6-273 in the extracellular domain of wtEGFr, and the generation of a glycine (G) residue at the junction point. Each consecutive peptide was offset by 1 residue from the -71previous one, yielding a nested, overlapping library of arrayed oligopeptides. The membrane carrying the 9 peptides was reacted with 9 different anti EGFrVIII antibodies (1 gg/ml). The binding of the mAbs to the membrane-bound pptides was assessed by an enzyme-linked immunosorbent assay using HRP-conjugated secondary antibody followed by enhanced 5 chemiluminescence (ECL). The array utilized is shown in Table 4.1. Table 4.1 Spot Array Sequence: 1 . ALEEKKGNYVVT (SEQ ID NO:72) 10 2. LEEKKGNYVVTD (SEQ ID NO: 59) 3. EEKKGNYVVTDH (SEQ ID NO: 73) 14. EKKGNYVVTDHG (SEQ ID NO: 74) 5. KKGNYVVTDHGS (SEQ ID NO: 75) 6. KGNYVVTDHG SC (SEQ ID NO: 76) 15 7. GNYVVTDHGSCV (SEQ ID NO: 77) 8. NYVVTDHGSCVR (SEQ bD NO: 78) 9. YVVTDHGSCVRA (SEQ ID NO: 79) In addition, functional epitopes were mapped by combinatorial Alanine scanning. In this 20 process, a combinatorial Alanine-scanning strategy was used to identify amino acids in the EGFrVII peptide necessary for interaction with anti-EGFRvIII mAbs. To accomplish this, a second set of SPOTs arrays was ordered for Alanine scanning. A panel of variants peptides with alanine substitutions in each of the 12 residues was scanned as above. Spot #1, the unmutated sequence, is a positive control for antibody binding. The array utilized is shown in Table 4.2. 25 Table 4.2 Alanine Scanning Array: 1. LEEKKGNYVVTD (SEQ ID NO: -59) 2. AEEKKGNYVVTD (SEQ ID NO: 80) 30 3. LAEKKGNYVVTD (SEQ ID NO: 81) 4. LEAKKGNYVVTD (SEQ ID NO: 82) 5. LEEAKGNYVVTD (SEQ ID NO: 83) 6. LEEKAGNYVVTD (SEQ ID NO: 84) 7. LEEKKANYVVTD (SEQ ID NO: 85) 35 8. LEEKKGAYVVTD (SEQ ID NO: 86) 9. LEEKKGNAVVTD (SEQ ID NO: 87) 10. LEEKKGNYAVTD (SEQ ID NO: 88) 11. LEEKKGNYVATD (SEQ ID NO: 89) 12. LEEKKGNYVVAD (SEQ ID NO: 90) 40 13. LEEKKGNYVVTA (SEQ ID NO: 91) -72- Epitopes of all 9 mAbs to the human EGFrVM were mapped and identified by SPOTs procedure. All 9 antibodies were reactive with the peptides. The results obtained with 3 murine antibodies and 6 XenoMouse mouse derived human antibodies are presented in Table 4.3. Highlighted residues are those which we mutated to alanine and abrogated binding by the test 5 antibody. These are therefore relevant residues for binding to the antibody. Table 4.3 EGFR AT C V K K C PR N Y V V T DH G SIC V RIA EQID NO: 92 EGFRvIll L EJE K KGNY VVT D HIGISIC V RIAI(SEQID NO: 93 ) 13.1.2 E E K KGNYVVT I--------(SEQ ID NO: 94) 131 E EKKGNYVVT I V -T- (SEQ ID NO: 94) 139 L EJE K K GIN Y V V T D (SEQ ID NO: 95) 250 L E EKK G N Y VV T D (SEQID NO: 95) 095 Y V VT D H (SEQ ID NO: 96) 211 1Y-VV T D (SEQ ID NO: 97) H10 Y V V T D (SEQ ID NO: 97) Y10 E EEK K G N Y (SEQ ID NO: 98) B9 GNYVVT _SEQ ID NO: 99) The shaded amino acids shown in Table 4.3 are the most relevant residues in the eiptope for 10 antibody recognition. The minimal lengths of epitopes of all ten of the mAbs were precisely mapped using peptides of overlapping sequences, and the tolerance for mAb binding to mutated epitopes was determined by systematically replacing each residue in the epitope with Alanine. In Table 4.4, additional characteristics of the antibodies are summarized. Specifcially, a subset of the antibodies were tested for their binding of to lysates of tumor cell lines in Western 15 plates of polyacrylarnide gel electrophoresis under either non-reducing or reducing conditions. Purifed recombinant protein is also included. Antibodies binding in both reducing and non reducing conditions suggest that the epitope is linear. Sample identifications: EGFRvM - the rabbit Fc fusion protein H1477 - H80 human tumor cell line transfected with EGFRvmII expression contruct. 20 These cells express both EGFR and EGFRvII. EGFR - purified wild-type EGFR protein A431 - human tumor cell line expressing only wild-type EGFR A549 - human tumor cell line expressing only wild-type EGFR H80 - human tumor cell line expressing only wild-type EGFR 25 EGFR Biacore - mAbs were tested in Biacore for binding to purifed EGFR as a highly sensitive test for specificity Table 4.4 -73- EGFRvIll rEGFRvlI H1477 H1477 EGFR EGFR Western Western EGFRviI Western Western pep3 Western Western mAb (native) (reduced) FACS (native) (reduced) KirExA (native) (reduced) 13.1.2 + + + + + 25- 131 + + + + + 0.05pM - 139 ? + + ND ND ND ND ND 095 + + + ND ND ND ND ND 211 + + + ND ND ND ND ND 250 + + + ND ND ND ND ND MAb A431 A431 A549 A549 H80 H80 EGFR A431 Western Western A549 Western Western H80 Western Western Blacore FACS (native) (reduced) FACS (native) (reduced) FACS (native) (reduced) 13.1.2 - - - - - - - - - 131 - ++ - - + - - - - 139 N.D. ND ND ND ND ND ND ND ND ND 095 - ND ND ND ND ND ND ND ND ND 211 - ND ND ND ND ND ND ND ND ND 250 - ND ND ND ND ND ND ND ND ND 5 The results showed that most of these mAbs have essentially the same binding specificity, seven of the mAbs were shown to bind specifically to the EGFrVI variant, while 2 mAbs cross reacted with wildtype EGFr (murine H10 and human 211) in Western blots of purified protein and in lysate of A431 cells. Note, however, that while antibody 211 binds to both native and reduced purified EGFRvM in Western blots, it binds slightly more strongly to the non-reduced protein. In 10 tests against a lysate of A431 cells, antibody 211 binds strongly to a band of the size of wild-type EGFR in the non-reduced sample but there is no signal in the reduced sample. This suggests that the binding of antibody 211 is due to a conformational epitope present in wild-type EGFR and represented differently in the EGFRvIII variant. The epitopes of 5 of the mAbs are within residues 2-12 spanning the EGFRvIII variant specific Glycine residue, whereas the epitope of 4 of the mAbs 15 (including HIO and 211) spans residues 7-16 which are common to the EGFRvM and wildtype EGFr. Antibody 131 binds to A431 and A549 cells in FACS. These cells are apparently negative for expression of EGFRVIII while positive for EGFR expression. Antibody 131 does not bind to non-reduced or non-reduced purified EGFR or to reduced or non-reduced lysates of A43 and A549 cells in Westerns suggesting that antibody 131 may be binding to a variant of EGFR expressed on 20 the cell surface of some human tumor cell lines. This variant would be sensitive to denaturation. -74- EXAMPLE 5 Characterization of Secificity of Anti-EGFRvI Antibodies in Vitro The specificity of the purified antibodies was ascertained by performing FACS analysis on NR6 cells that were transfected with either wild type or mutant EGFR. Cells were incubated on ice 5 with 5 Rg/ml of the respective antibody for 1 hr, washed in FACs buffer and subsequently incubated with PE-conjugated goat anti-human IgG. EXAMPLE 6 Cross-reactivity with amplified EGFR Antibodies directed to variant EGF receptors have been shown to cross-react with subsets 10 of wild type EGF receptors on cells in which gene amplification has occurred (Johns et al., Int. J. Cancer. 98: 398, 2002). To determine whether the human EGFRVIII antibodies identified had similar properties, they were tested for their ability to recognize wild type EGF receptors on a variety of cells in culture. Antibodies were incubated with the indicated cell lines at 4*C. After washing in FACS buffer, a secondary antibody conjugated with phyoerythrin was added and the 15 incubation was continued. All cell lines analyzed expressed wild type EGFR. A subset of wild type EGFRs was recognized by the antibody XGI-131 on both A431 and SF-539 cells but not on A498 or SKRC-52 cells. Another antibody to EGFRvM, 13.1.2, did not recognize this subset of wild type EGFRs. When considered together these data indicate that only a subset of antibodies directed to the mutant EGFRvfI are able to recognize wild type EGFRs on the surface of cells. The 20 ability of certain antibodies directed to mutant EGFRvm to recognize a subpopulation of wild type EGF receptors is not dependent on total EGFR density but likely represents a novel conformational epitope that is unique to tumor cells. The ability of antibodies directed to EGFRvM to cross-react with subpopulations of wild type receptors may be determined by both the specific epitope within the junction of the mutant receptor and the affinity of the antibody for this unique epitope (See the 25 results of the epitope mapping and affinity determination section herein). EXAMPLE 7 Characterization of Siecificity of Anti-EGFRvIII Antibodies In vitro: Binding of the Antibodies to Cell Lines The specificity of the purified antibodies was ascertained by performing FACS analysis on 30 a panel of cell lines. H80, a human glioblastoma line, and H1477 (H80-EGFRvI) that expresses high levels of EGFRvIII, A431, a human epidermoid carcinoma line, and A549, a human lung carcinoma cell line were used as the cell lines. All cell lines were from Dr. Bigner except A431 and A549, which were from ATCC (Rockville, MD, U.S.A.). Cells were incubated on ice with 10 pg/ml of the respective antibody for 30 min., washed in FACS buffer and subsequently incubated 35 with PE-conjugated goat anti-human IgG from Jackson immunoResearch (West Grove, PA, U.S.A.). In FIGs. 9A-9L and 10A-10D, the darkened histogram indicates cells stained with an irrelevant IgG, the outlined, or white histogram, represents the staining of the relevant antibodies. -75- The anti-EGFRvIII antibodies 13.1.2, 131 and 139 bind to the EGFRvIII protein on the transfected cell lines. A gcaph summarizing some of the results is displayed in FIGs. 9M-9P. Antibodies directed to variant EGF receptors have been shown to cross-react with subsets of wild type EGF receptors on cells in which gene amplification has occurred (Johns et al., Int. J. 5 Cancer. 98: 398, 2002). In this example, A431 and A549 stained by XGI-131 and XGI-139. FIG. 10B and FIG. 10 C show that 131 and 139 have certain cross reactivity with the wild type EGFR instead ofjust recognizing a subset of population in H80, A431 and A549 line. However, this cross reactivity is only at 10% of the level of ABX-EGF (E7.6.3) staining on these cell lines. The results are provided in FIGs 9A-9P and 10A-10D. 10 Antibodies directed to cell surface antigens can be used as delivery vehicles that specifically transport drugs or toxins into cells. If the antibody stimulates internalization of antigen, the drug or toxin can result in death of the cell, perhaps after the drug or toxin is cleaved from the antibody. Such a mechanism can be utilized to specifically kill tumor cells in animals and in patients. One way to select antibodies that can deliver drugs to cells is through secondary 15 cytotoxicity assays. In these assays the primary antibody binds to the cell surface and a secondary antibody that is conjugated with a drug or toxin is added. If the primary antibody stimulates antigen internalization, the secondary antibody will be co-intemalized and upon cleavage of the drug or toxin result in cell killing. EXAMPLE 8 20 Secondary Cytotoxicity Assays In the following studies, EGFRvIII-specific antibodies were used to direct toxins conjugated secondary antibodies into glioblastoma cell line (1180) and EGFRvIII transfected glioblastoma cell line (H1477). Mouse anti-human IgG (cat # 555784) from Pharmingen (BD Biosciences Pharmingen) were conjugated with toxins AEFP (Seattle Genetics, Inc.) and 25 maytansine (DM1, Immunogen Inc.) to generate mah-AEFP (murine anti-human IgG-AEFP and mah-DMl (murine anti-human IgG-DM1). A saporin conjugated goat anti-human IgG, Hum-ZAP (TM, cat # IT-22-250, affinity-purified goat anti-human IgG-saporin) is from Advanced Targeting Systems (San Diego, CA, U.S.A.). H80 and H1477 cells were plated out in 96-well plates with 1000 cells in 100 sl growth medium per well. After 24 hours, primary antibodies were mixed with 30 conjugated secondary antibodies at 1:3, serially diluted at 1:5 over 6 wells. 100 ;d of diluted primary and toxin secondary antibody mixtures were added into wells of cells at final starting concentrations of 0.1 gg/ml of primary antibodies and 0.3 pg/ml of secondary antibodies. The plate was allowed to continue to culture for three days. On the fourth day, CellTiter-Glo reagents (cat #G7571) from Promega (Madison, WI, U.S.A.) were added and luminescence was read. FIGs. IIA 35 11I, 12A-121, and 13A-13I demonstrate the results from this experiment. Hum-ZAP mediated antigen specific killings in H1477 (filled circle) compared to H80 (filled square) in most EGFRvIII specific mAbs tested. MAbs XG1-131 and XGI-139 generated antigen specific secondary killings -76with mah-AEFP, at less extent with mah-DMI. Among the antibodies tested, XG1-131 performed at least one log better than 13.1.2, XGl-095, XG1-139, XGI-150, XG1-170, XG1-250 and XG1 211. IgG1 was used as a negative control and antigen positive cells (H1477) were compared to antigen negative cells (H80). 5 The amount of specific killing required can vary depending upon the particular use. In one embodiment, any reduction in possibly cancerous cells is sufficient. For example, a reduction of 0 1, 1-5, 5-10, 10-20, 20-30, 30-40, 40-50, 50-60, 60-70, 70-80, 80-90, 90-95, 95-99, or 100% of the target cells will be sufficient. In another embodiment, the desired reduction in target cell number is also a function of the nonspecific lethality of the antibody combination. For example, 10 antibody/toxin combinations that only have a 10 % decrease in target cell number may be sufficient, if there is very little nonspecific targeting and lethality by the antibody. For example, the antibody toxin combination kills less than 10% of a non-target population. Again, the particular amount will depend on the particular need and situation. Particularly useful are antibodies that are highly selective for a target cell (e.g., H1477) and bind well to the target cell, or proteins associated with 15 the cell. In one embodiment, the target is the EGFRvIII protein, or a fragment thereof. In one embodiment, antibodies that are human, or humanized, efficient at being internalized, specific to the EGFRvm protein or fragments thereof, associate tightly with the EGFRvM protein or fragment, and are associated with an effective toxin, are taught from these examples. EXAMPLE 9 20 Secondary Cytotoxicity Clononenic Assays In addition to the secondary Cytotoxicity assays, EGFRvIII-specific antibodies were tested in the clonogenic assays. Specific EGFRvm antibodies direct toxin conjugated secondary antibodies into EGFRvIII transfected glioblastoma cell line (H1477), toxins are released inside the cells and eventually reduced the cell's ability to proliferate to form colonies. Thus, the application 25 of these EGFRvm antibody-toxins generated reduced number of clones when cells were re-plated after primary and secondary toxin antibody treatments. In this example, H80 and H1477 cells were plated out in 6 well plates at 30,000 cells per well and incubated overnight. The primary antibody and secondary toxin antibody were mixed at a ratio of 1:3. This this antibody mixture was added into the proper wells at a final concentration of primary antibody at 0.5 pg/ml and secondary toxin 30 antibody at 1.5 ig/ml. This was the incubated at 37 "C overnight. After incubation, the toxin mixture was disposed of properly and thecells were detached from the wells with 1X trypsin solution. The cells were counted and plated at 200 cells per well into new 6-well plates. Triplicates wells were plated for each treatment group. These plates were incubated in a 37 *C incubator for 2 3 weeks until the colonies formed and could be identified by eye or under a microscope. The 35 medium was aspirated and SM Methylene Blue in methanol was added for 1 hour. The plate was rinsed in water and the colonies were counted. FIG. 14A and FIG. 14B show the results from this -77experiment. As can be seen, mab-AEFP secondary toxin antibody inhibited colony formation with the three EGFRvIII antibodies tested. EXAMPLE 10 Anti-EGFRvII antibody (13.1.2) Direct Conjugates in the Cytotoxicity Assays 5 Further evidence that an antibody is capable of delivering a drug to a tumor cells is provided by direct conjugation of the antibody with a cytotoxic drug. In the following example EGRvm antibody 13.1.2 was directly conjugated with auristatin E MMAE and AEFP conjugates via peptide linkers (both MMAE and AEFP are available from Seattle Genetics and described above) and and maytansine (DM1) was conjugated with a thiol linker (DM1 is available from 10 Immunogen and described above). Upon addition of the conjugate to cells that expressed the EGFRvm antigen, H1477, specific cytoxicity was observed. Cells that did not express the antigen, H80, were only killed when exposed to very high concentrations of the antibody. Results from this experiment are shown in FIGs. 15A-15C. Direct conjugation of the EGFRvIII antibodies with the drugs or toxins is a particularly 15 advantageous method for therapeutic use. Thus, this initial experiment showed that such conjugates do result in specific killing of EGFRvMl- expressing cells. EXAMPLE 11 In vivo Anti-EGFRYII Antibodies Characterization, An optional method to determine if an antibody is capable of delivering a cytotoxic drug to 20 a cell is to evaluate the effect of the conjugated antibody on the growth of human tumors in vivo. This example presents one such method. H1477 glioblastoma cells were cultured in vitro, harvested by trypsinization and subsequently embedded in Matrigel as explained below. Five million cells were injected subcutaneously into female nude mice and tumors allowed to develop until they reached a size of approximately 0.5 cm. At this time the animals were randomized into 25 groups and treatment with the indicated concentration of conjugated antibody intravenously every 4 days was initiated. The results in FIG. 16 demonstrate that antibody 13.1.2 can cause regression of glioblastoma tumors when conjugated with maytansine (dEGFR- DM1) or auristatin E (dEGFR MMAE). If the antibody is administered with an equivalent amount of unconjugated drug, (Group2), there is no effect on tumor growth proving that targeting the tumor cells in vivo requires 30 the antibody conjugation with the toxin. The animal model used above was developed by injecting H1477 cell xenografts into nude mice. Various amounts of the cells were injected with or without MATRIGEL into 8 week old nu/nu female mice and the tumor implantation over days analyzed. From this analysis, the number of cells that would allow for an appropriately sized tumor was identified as 5 million cells in 35 MATRIGEL for approximately 22 days. Group G8 was included as a control to show that the killing was antibody specific. Group G7 was included as a negative control. Thus a protocol was developed for the toxin study as follows: -78- Day 1: tumor implantation of 5 million cells in MATRIGEL into 8 week-old nu/nu female mice. Day 22: antibody pro-drug treatments every 4 days, via I.V., as shown in Table 11.1 5 Table 11.1 Group Number Treatment of Mice Gi 8 13.1.2-DM1 250 g every 4 days, via I.V. G2 8 13.1.2-DMl 75 pg every 4 days, via I.V. G3 9 13.1.2-MMAE 75 pg every 4 days, via I.V. G4 8 13.1.2-MMAE 250 ptg every 4 days, via I.V. GS 9 13.1.2-AEFP 75 jig every 4 days, via I.V. G6 8 13.1.2-AEFP 250 pg every 4 days, via I.V. G7 8 PBS every 4 days, via I.V. G8 9 13.1.2 (unconjugated) 250 Ag + Maytansine 4 g The results are shown in FIG. 16 with the arrows indicating the addition of drug. Groups G1, G6, and G4 showed effective killing. Group G3 showed a lesser amount of killing. Groups G8 10 and G7 showed no killing. Certain toxicity may have been observed in the high dose vc-AEFP group, Group G8. These animals received 2 treatments at 250 pg and 1 treatment at 125 pg. EXAMPLE 12 Expression of EGFRvM in Cancer Patients/Human Tumors The expression of EGFRvIII on human tumors was determined by staining frozen tissue 15 sections from a variety of cancer patients with a combination of 2 murine monoclonal antibodies (B9, IgG1 and Y1O, IgG2 (Dr. Bigner, Duke University)) known to bind specifically to EGFRvIL. The same sections were stained with isotype matched control antibodies. A summary of the staining results obtained from all patient samples is presented in Table 12.1. 1 20 Table 12.1 Tumor e Sample Size (N) EGFRv111>+ | EGFRvII+ Glioblastoma 8 100% 100% Breast Cancer 100 31% 24% NSCL cancer 51 47% 39% Head & neck Cancer 21 42% 38% Prostate Cancer 22 4.5% 4.5% EGFRvII>+: include all tumors that express EGFRvII EGFRvIII>++: include only those tumors that express at least 10% or more EGFRvIII -79- The expression was found primarily on the cell membrane and/or cytoplasm. Significant numbers of breast (31%), NSCL (47%), and head & neck (42%) cancer specimens stained positively for EGFRvIII. In certain instances, in order to obtain high quality IHC staining, the use of two antibodies can be better than the use of one antibody. Frozen tissue specimens were superior 5 over fixed tissues. As appreciated by one of skill in the art, it may be advantageous to test patients before using therapeutic antibodies to ensure that the tumor which is being treated expresses EGFRvIII. EXAMPLE 13 In vivo Anti-EGFRvII Antibodies Characterization. 10 The method of Example 11 will be used to treat lung cancer and gliomas. This will be broadly examined by producing animal models. Animal models for glioblastoma and lung cancer are developed as follows: lung cancer cells that express wt-EGFR are transfected with EGFRvII. The cells are injected into the lungs of nu/nu mice and tumors allowed to progress to a comparable stage to that above. Anti-EGFRvUI conjugates will then be injected intravenously as above every I 15 to 10 days as needed. The size and prevention or suppression of continued growth of these cancer cells will then be monitored, to determine the effectiveness of these Anti-EGFRvIII antibodies and antibody-toxins combinations. As appreciated by one of skill in the art, this can be done for any of the antibodies disclosed herein. EXAMPLE 14 20 Functional Characterization of Epitopes by Substitional Analyses In order to further resolve the identity of those amino acid residues that are indispensable for binding within the EGFRvIII epitope, substitutional analyses of the amino acids in the epitope peptides were performed. The starting point was the sequence that was derived from Example 4, LEEKKGNYVVTD (SEQ ID NO 59). In this example each amino acid of the mapped epitope 25 was substituted one-at-a-time by all 20 L-amino acids, thus, all possible single site substitution analogs were synthesized and screened to provide detailed information on the mode of peptide binding. Discrete substitution patterns were identified for mAbs 131 and 13.1.2. The results from the substitutions are summarized in Table 14.1. 30 Table 14.1 mAbs Recognition sequence 131 EEKKGNYVVT (SEQIDNO:57) 13.1.2 EEKKGNYVVT (SEQIDNO:57) It appears that for mAb 13.1.2, 5 residues are important for binding (bold), while only 4 residues are essential for the binding of mAb 131. The rest of the residues were replaced by various amino acids without significant loss of binding. Although the 131 and 13.1.2 epitopes are identical -80by sequence and length, the binding pattern for each appears different. Binding of mAb 131 is strongly dependent on the residues EKNY (SEQ ID NO: 60). On the other hand, the data revealed that residues EEKGN (SEQ ID NO: 61) are involed in binding of mAb 13.1.2. Example 15 5 mAbs Chain Shuffling Heavy and light chains of mAbs 131 and 13.1.2 were shuffled and transiently transfected into 293T cells. Seventy-two hours later supernatants were collected and assayed for secretion and binding to EGFrVIII antigen by ELISA. The results demonstrated that antibodies derived from expression of 131 heavy chain 10 with 13.1.2 kappa chain, and vice versa were expressed well but binding activity was reduced by 75% probably due to the different binding pattern of these two mAbs to EGFrVIII antigen. (data not shown). This demonstrates the difference between the two paratopes of the 131 and 13.1.2 mAbs, again suggesting that the structural characteristics of the epitope selected for between the two mAbs are different. 15 Example 16 Molecular Modeling of 131 and its Paratope This example demonstrates how three-dimensional structures can be generated for the proteins of the embodiments. The three-dimensional structure model of the variable region of antibody 131 was generated through a homology modeling approach using the InsightIl 20 modeling package from Accelrys (San Diego, CA). The model was built from the variable region sequences described below, Table 16.1. The residue numbering starts with the light chain amino acids, and continues to heavy chain amino acids. Table 16.1 25 Light chain variable region DTVMTQTPLSSHVTLGQPASISC (SEQ ID NO: 100) RSSQSLVHSDGNTYLS (CDR1) (SEQ ID NO: 10 1) WLQQRPGPPRLLIY (SEQ ID NO: 102) RISRRFS (CDR2) (SEQ ID NO: 103) 30 GVPDRFSGSGAGTDFTLEISRVEAEDVGVYYC (SEQ ID NO: 104) MQSTHVPRT (CDR3) (SEQ ID NO: 105) FGQTKVEIK (SEQ ID NO: 106) Heavy chain variable region 35 QVQLVESGGGVVQSGRSLRLSCAASGFTFR (SEQ ID NO: 107) NYGMH (CDR]) (SEQ ID NO: 108) WVRQAPGKGLEWVA (SEQ ID NO: 109) 40 - 81 - VIWYDGSDKYYADSVRG (CDR2) (SEQ ID NO: 110) RFTISRDNSKNTLYLQMNSLRAEDTAVYYCAR (SEQ ID NO: 111) DGYDILTGNPRDFDY (CDR 3) (SEQ ID NO: 112) WGQGTLVTVSS (SEQ ID NO: 113) 5 Antibody 131 sequences were used to search against the Protein Data Bank to identify homologous antibodies and their structures. Based on the homologous antibodies' sequence similarity to the 131 antibody, several structures were selected. The structures selected for modeling samples from the Protein Data Bank had the Protein Data Bank identifications of 10 1 HEZ, 2H I P, I AQK, I DQL, I MF2 and 1 FLR. These template structures were then aligned by superposition and used to generate structure-based sequence alignments among the templates. The sequences of antibody 131's variable region were then aligned to the template sequences. The structure and sequence alignments were used to generate the molecular model for the variable region of the 131 antibody. The sequence for CDRI, light chain was: 15 RSSQSLVHSDGNTYLS (SEQ ID NO 101). The sequence for CDR2, light chain was: RISRRFS (SEQ ID NO 103). The sequence for CDR3, light chain was: MQSTHVPRT (SEQ ID NO 105). The sequence for CDR1, heavy chain was: NYGMH (SEQ ID NO 108). The sequence for CDR2, heavy chain was: VIWYDGSDKYYADSVRG (SEQ ID NO 110). The sequence for CDR3, heavy chain was: DGYDILTGNPRDFDY (SEQ ID NO 112). 20 The interaction surface for antibody 131 was calculated from the structure model and shown in FIG. 17. The various CDRs are identified as follows: LI (light CDR1) 10, HI (heavy CDR]) 20, L2 30, H2 40, L3 50 and H3 60. A prominent feature on the predicted antibody 131 interaction surface is a deep cavity. The cavity is mainly surrounded by heavy chain CDR2, CDR3 and light chain CDR3, with a small portion contributed by light chain CDRI. 25 The cavity is probably the binding pocket. Within 5 Angstroms of the binding cavity are residues 31, 37, 95-101, 143-147, 159, 162-166, 169-171, 211-219, 221 and 223. These residues are likely to comprise the paratope and make key contacts in the binding of EGFRvIII epitope. It is also likely that the residues provide important structural features to the binding site in general. 30 Example 17 Site-directed mutagenesis confirming the model for antibody 131 This example demonstrates one method by which models that suggest residues that are important in binding may be tested. The Example also results in several antibody variants. Antibody variants of the 131 clone were generated by single residue mutations introduced to 35 the heavy and the light chain of mAb 131. These variants were then analyzed to determine how the altered side chains from the point mutation contributed to antigen binding. Changes were made in the heavy and light chains of mAb 131. On the heavy chain L216 was changed by site directed mutagenesis to R. On the light chain, V99 was changed to F. Both - 82 mutations affected the expression and secretion of the variant antibodies compared to the wildtype sequence. Both mutations resulted in a loss of binding of the mAb variant to the EGFRvlU antigen. This demonstrates L216 and V99 probably have significant contacts with the EGFRvIH antigen since substitutions of these residues to R and F respectively resulted in reduced activity. Of course, 5 it is always an option that these substitutions are disruptive to the antibody's general structure. Example 18 Molecular Modeling of 13.1.2 and its Paratope The three-dimensional structure model of the variable region of the 13.1.2 antibody was generated through homology modeling approach with the InsightII modeling package from 10 Accelrys (San Diego, CA). The model was built from the variable region sequences, shown below in Table 18.1, using the published x-ray crystal structures as templates. Table 18.1 15 Light chain variable region (1-113) DIVMTQTPLSSPVTLGQPASISC (SEQ ID NO: 114) RSSQSLVHSDGNTYLS (CDRI) (SEQ ID NO: 101) WLHQRPGQPPRLLIY (SEQ ID NO: 115) KISNRFS (CDR2) (SEQ ID NO: 116). 20 GVPDRFSGSGAGTAFTLKISRVEAEDVGVYYC (SEQ ID NO: 117) MQATQLPRT (CDR3) (SEQ ID NO: 118) FGQGTKVEIKR (SEQ ID NO: 119) Heavy chain variable region (114-234) 25 QVQLVESGGGVVQPGRSLRLSCAASGFTFS (SEQ ID NO: 120) SYGMH (CDRI) (SEQ ID NO: 121) WVRQAPGKGLEWVA (SEQ ID NO: 122) VIWYDGSNKYYVDSVKG (CDR2) (SEQ ID NO: 123) RTISRDNSKNTLYLQMNSLRAEDTAVYYCAR (SEQ ID NO: 124) 30 DGWQQLAPFDY (CDR3) (SEQ ID NO: 125) WGQGTLVTVSSA (SEQ ID NO: 126) The sequence for CDR1, light chain was: RSSQSLVHSDGNTYLS (SEQ ID NO: 101). The sequence for CDR2, light chain was: KISNRFS (SEQ ID NO: 116). The sequence for CDR3, 35 light chain was: MQATQLPRT (SEQ ID NO: 118). The sequence for CDR1, heavy chain was: SYGMH (SEQ ID NO: 121). The sequence for CDR2, heavy chain was: -83- VIWYDGSNKYYVDSVKG (SEQ ID NO: 123). The sequence for CDR3, heavy chain was: DGWQQLAPFDY (SEQ ID NO: 125). Antibody 13.1.2 sequences were used to search the Protein Data Bank to identify homologous antibodies. The structures with the Protein Data Bank identifications of 1HEZ, 2HIP, 5 8FAB and 1AQK were selected as modeling templates, based on their sequence similarity to antibody 13.1.2. The template structures were aligned by superposition and used to generate structure-based sequence alignments among the templates. The sequences of the variable regions of the 13.1.2 antibody were then aligned to the template sequences. The structure and sequence alignments were used to generate the molecular model for antibody 13.1.2 variable region. 10 The interaction surface was calculated for the model and is shown in FIG. 18. A major feature of the 13.1.2 model is a long and narrow groove on the surface of the CDR region. The groove is outlined by heavy chain CDR2 140, CDR3 160, and light chain CDRl 110, CDR2 130 and CDR3 150. One end of the groove touches the rest of light chain CDR3 150, and the other end opens to the wider area of heavy chain CDR3 160 near the heavy chain-light chain interface. The 15 groove is probably the binding pocket for the antigen. Within 5 Angstroms of the binding groove are residues 31, 33, 35-39, 51, 54-56, 58-61, 94-101, 144-148, 160, 163-166, 172, and 211-221. These residues are likely to comprise the paratope for the binding of EGFRvIII epitope. It is also likely that the residues provide important structural features to the binding site in general. Example 19 20 Docking Models of a Peptide to an Antibody The epitope mapping studies in Example 14 revealed that the relevant amino acids required for binding of the epitope to the paratope of 13.2.1 mAb reside in the six-residue peptide EEKKGN (SEQ ID. NO: 127). Therefore, docking models of this six-residue peptide complexed to the CDR region of 13.1.2 structure model were generated. First, a model of the peptide EEKKGN (SEQ ID 25 NO: 127) was produced. This was done, similarly as described before, except this time using the x ray crystal structure of 1181, as identified in the Protein Data Bank, as the template. Next, this peptide structure was manually placed into the long groove to form an initial assembly complex. A Monte Carlo search in the conformational and orientational spaces were then automatically performed with the Docking module in InsightII. The peptide conformation was allowed to be 30 flexible by giving each of the Phi, Psi and Chi angles full rotational freedom. During the docking process, the residues within 5 Angstroms of the binding groove were allowed to move while the other residues of the antibody were fixed. The plausible configurations found by Monte Carlo search were subjected to simulated annealing and energy minimization to reach the final complex structure models. For each docking model obtained, the interaction energy between the antibody 35 and the peptide was calculated with the Discover_3 module of InsightIl package. The interaction energies for all docking models were assessed and the model with the strongest overall antibody peptide interaction was examined and is shown in FIG. 19A and 19B. -84- In this docking model, there are six hydrogen bonds between peptide EEKKGN (SEQ ID NO: 127) and antibody 13.1.2, as shown in FIG. 19B. The peptide residue number is labeled from N-terminus to the C-terminus as 1 through 6. Six hydrogen bonds are indicated by green dashed lines. The six pairs of amino acids forming hydrogen bonds are: E2.. .Y1.72, K3... H3 1, K4...H3 1, 5 N6... D33, N6...Y37, and N6... K55. In this docking model, the peptide is bound to the groove in an extended p-strand-conformation. Residues in the peptide alternately face the solvent and the antibody surface. The residues facing the binding groove with the most significant contacts to the antibody are E2, K4 and N6. This indicates that these three residues may be important to peptide binding, consistent with the epitope mapping results. The interaction energies for each of the six 10 peptide residues with the 13.1.2 paratope was calculated with the Discover_3 module and the results are shown in Table 19.1. Table 19.1 shows the interaction energies for each of the six peptide residues with the 13.1.2 paratope. All energies are in the unit of kcal/mol. The residues with the strongest interaction energies are in the order of N6, K4 and E2, confirming that these residues are key contributors on the antigen side in the antibody-antigen 15 interaction, again consistent with experimental data. These data provided strong evidence to support the docking model. In this embodiment, the paratope is defined as the residues within 5 Angstroms of the docked peptide. The 20 residues comprising the paratope as so defined are residues 31-33, 35, 37, 55, 96-101, 148, 163, 165, 170, 172, 178 and 217-218. To evaluate, on an individual residue basis, the contribution of each of these residues of the antibody in the antibody 20 antigen interaction, the interaction energy between the paratope residues and the peptide EEKKGN (SEQ ID NO: 127) was calculated for each of the above 20 residues. The results are listed in Table. 19.2. Table 19.2 shows the interaction energies for each of the 20 paratope residues with the peptide EEKKGN (SEQ ID NO:127). All energies are in the unit of kcal/mol. The residues with the strongest interaction energies with the peptide are Lys55 and His31, followed by Tyr172, 25 Ala96, Asp33, Tyr37, Leu99, Thr97, Gln98, Lys178 and Asn170. Table 19.1 Peptide Residue Coulumbic VdW Total El -2.013 -3.738 -5.751 E2 -10.661 -0.617 -11.278 K3 -9.816 -0.493 -10.309 K4 -11.123 -0.968 -12.091 G5 -1.241 -1.468 -2.709 N6 -16.504 -0.181 -16.685 30 -85- Table 19,2 13.1.2 Residue Coulumbic VdW Total His3l -12.835 3.033 -9.801 Ser32 2.857 -1.062 1.794 Asp33 -4.181 -0.698 -4.879 Asn35 0.253 -1.009 -0.756 Tyr37 -2.058 -2.463 -4.521 Lys55 -14.363 1.568 -12.794 Ala96 -6.077 0.896 -5.182 Thr97 -2.739 -1.431 -4.171 Gin98 -2.542 -1.548 -4.09 Leu99 -1.507 - -2.779 -4.286 Pro100 0.439 -0.379 0.061 ArgI01 3.992 -0.549 3.443 His148 0.101 -0.083 0.018 VaI163 -0.104 -0.237 -0.342 Trp165 1.358 -1.122 0.236 Asn170 -2.102 -0.487 -2.589 Tyr172 -8.7 0.896 -7.804 Lys178 -3.614 -0.03 -3.644 Leu217 0.761 -1.426 -0.664 Ala218 -0.071 -0.281 -0.352 Example 20 Rational Design for Affmity-1nproved Antibodies 5 This Example demonstrates how the docking model can be used as the basis of rational design for affinity-improved antibodies by site-directed mutagenesis. Each of the 13.1.2 paratope residues was mutated to all 19 other amino acids in silico, resulting in a total of 19x20 or 380 virtual mutants. The mutation was done by residue replacement followed by 50 steps of energy minimization to account for any local conformational changes that could be induced by the side 10 chain change. The interaction energy between the whole peptide and the whole paratope was calculated for each mutant. Mutants with a total interaction energy stronger than the wild type 13.1.2 could potentially have a higher affinity for the peptide EEKKGN (SEQ ID NO: 127), and perhaps even the whole EGFRvIII protein. These mutants mostly have stronger coulumbic interactions than the wild type 13.1.2, but some of them have weaker van der Waals (VdW) 15 interactions than the wild type antibody. Considering that in the wild type 13.1.2 antibody, the VdW interacting energy is -9.689 kcal/mol, mutants with VdW interaction energy weaker than -8.5 kcal/mol were filtered out. The rest of the mutants that have stronger, total interaction energy, than the wild type 13.1.2 are listed in Table 20.1. The wild type data are listed at the bottom for comparison. All energies are in the units of kcal/mol. 20 -86- Table 20,1 Mutant Coulumbic VdW Total Tyrl72Arg -93.004 -8.702 -101.706 Leu99GIu -79.897 -8.506 -88.403 Arg1lGlu -77.984 -8.833 -86.817 Leu2l7GIu -75.124 -8.998 -84.123 Leu99Asn -73.337 -9.894 -83.231 Leu99HIs -73,631 -9.008 -82.639 ArglOlAsp -71.983 '-9.877 -81.861 Leu2l7GIn -70.263 -9.795 -80.058 Leu99Thr -69.882 -10.153. -80.035 Gln98Glu -70.651 -9.257 -79.908 Leu2l7Asn -70.989 -8.769 -79.758 Arg1OlGin -69.432 -10.164 -79.596 Leu2l7Asp -69.934 -9.643 -79.578 Asn35GIy -69.016 -10.191 -79.207 Tyr172His -69.312 -9.509 -78.820 VaIl163Asn -68.841 -9.944 -78.784 Tyrl72Asn -68.896 -9.871 -78.767 Ala2l8Lys -70.024 -8.570 -78,594 Asn35Arg -68.989 -9.604 -78.593 Trpl65Lys -69.578 -8.766 -78.344 Trpl65Arg -68.814 -9.216 -78.030 Leu99Tyr -67.052 -10.464 -77.517 Tyrl72Thr -68.146 -9.225 -77.371 Ala96Thr -67.534 -9.623 -77.158 Ala96Ser -67.222 -9.822 -77.045 ProlOOTrp -67.399 -9.496 -76.894 Leu2l7Ser -66.676 -10.133 -76.810 Ser32lle -66.700 -10.077 -76.777 Tyrl72Ser -67.588 -9.146 -76.734 His3lGlu -67.070 -9.461 -76.531 Leu217Tyr -65.605 -10.726 -76.331 VaIl163His -67.236 -9.064 -76.300 His148Ser -66.780 -9.495 -76.274 HIsl48Val -66.634 -9.629 -76.263 Hisl48AIa -66.770 -9.473 -76.243 Hisl48GIy -66.762 -9.456 -76.217 Hisl48Thr -66.700 -9.508 -76.209 Leu99Ser -66.126 -10.006 -76.132 ProlOOAsp -66.153 -9.787 -75.940 Trpl65Glu -66.665 -9.267 -75.932 Hisl48Asn -66.010 -9.889 -75.899 ProlOOGin -65.873 -9.871 -75.745 Leu2l7Thr -66.045 -9.672 -75.717 Ser32Val -65.845 -9.854 -75.699 Ser32Pro -65.807 -9.813 -75.620 ProlOOGly -65.841 -9.774 -75.615 Pro1OAla -65.889 -9.712 -75.601 Ser32Ala -65.497 -10.089 -75.586 Ser32Thr -65.723 -9.861 -75.584 -87- Mutant Coulumbic VdW Total Ala218Thr -66.054 -9.505 -75.560 ProlOOSer -65.831 -9.699 -75.530 Vail163GIy -65.993 -9.536 -75.529 GIn98Thr -66.162 -9.277 -75.438 ProlOOMet -65.811 -9.602 -75.412 Ser32Met -66.252 -9.153 -75.406 Ser32GIy -65.509 -9.891 -75.399 ProlOOAsn -65.729 -9.655 -75.384 Tyr37Phe -66.253 -9.020 -75.272 Vail163AIa -65.713 -9.543 -75.255 Leu2l711e -65.479 -9.759 -75.238 Wild type 13.1.2 -65.517 -9.689 -75.205 The mutants listed in Table 20.1 could be candidates for engineering of affinity-improved antibodies. For the top 14 candidates in the list, per residue contributions on the antigen side and on the antibody side were further analyzed to examine the impact of the proposed modifications. The 5 10 mutants selected for in vitro site-directed mutagenesis were Tyrl72Arg, ArglOlGlu, Leu99Asn, Leu99His, Arg1OAsp, Leu2l7GIn, Leu99Thr, Leu2l7Asn, ArgIOGIn and Asn35Gly. The results can be seen in Example 21. Example 21 Site-directed mutagenesis confirming the model for 13.1.2 10 This example demonstrates one method by which the above models, which suggest residues that are important in binding, can be tested. The Example also results in several antibody variants. Antibody variants of 13.1.2 were generated by single residue mutations introduced into the heavy and the light chains of the 13.1.2 mAb. The variants were analyzed to determine the contribution that the various side chains had in antigen binding. A list of the mutations introduced by site 15 directed mutagenesis are summarized in Table 21.1. Table 21.1 Chain Mutation I Light chain (CDR3) ArglOlAsp 2 Light chain (CDR3) ArglOGln 3 Light chain (CDR3) ArglOlGlu 4 Light chain (CDRI) Asn35Gly 5 Heavy chain (CDR3) Leu217Asn 6 Heavy chain (CDR3) Leu2l7GIn 7 Light chain (CDR3) Leu99Asn 8 Light chain (CDR3) Leu99His 9 Light chain (CDR3) leu99Thr 10 heavy chain (CDR2) Tyrl72Arg Each of the 10 mutations in Table 21.1 was introduced into the heavy or light chain of the 20 13.1.2 mAb. Each mutated chain was then transfected with the complementary wild-type chain in -88- 293 cells. Supernatants were then tested for expression and secretion of human IgG antibodies, and for binding to EGFrVM antigen. The results, as determined by an ELISA, are summarized in Table 21.2. Table 21.2 Mutation Binding Energ Expression Binding I Arg101Asp -81.861 Yes No 2 ArgI01GIn -79.596 Yes No 3 Arg101Glu -86.817 Yes No 4 Asn35Gly -79.207 Yes Yes 5 Leu2l7Asn -79.758 Yes Yes 6 Leu2l7GIn -80.058 Yes Yes 7 Leu99Asn -83.231 Yes Yes 8 Leu99His -82.639 Yes Yes 9 Leu99Thr -80.035 Yes Yes 10 Tyrl72Arg -101.706 Yes Yes 11 WT -75.205 Yes Yes 5 Example 22 Preparation of EGFRvIII/pFLAG variant construct This example demonstrates how a variant to EGFRvIII can be made. A 1092bp fragment 10 encoding the extracellular domain of EGFRvm was generated with primer pairs 9712 and 9713 (Qiagen, Valencia, CA): Primer # 9712: 5'-ataaaagcttctggaggaaaagaaaggtaatta-3' (sense) (SEQ ID NO 128) Primer # 9713: 5'-TTATTGGTACCTCAGGCGATGGACGGGATCTTA- 3' (antisense) (SEQ ID NO 129) 15 from plasmid template EGFRvM-rbIgG/pCEP4 (as described above) amplified using Pfu DNA polymerase enzyme (Stratagene, La Jolle, CA). Primer # 9712 introduced a HindIII site and primer # 9713 introduced a KpnI site. The PCR product was column purified (Qiagen column purification kit, Valencia, CA) digested with Hindill and KpnI (NEB, New England Biolabs, Beverly, Mass.) and gel purified (Qiagen gel purification kit, Valencia, CA). Fragments were ligated with T4 DNA 20 Ligase (NEB, New England Biolabs, Beverly, Mass.) into pFLAG-CMV-I (Sigma, St. Louis, MO) linearized with HindIll and KpnI (NEB, New England Biolabs, Beverly, Mass.). The resulting vector was designated EGFRvm./pFLAG-CMV- 1 # 1. Example 23 Preparation of EGFRvIm/pFLAG Recombinant Protein 25 This example demonstrates how a variant EGFRvI protein can be made. First, 500 Jig of EGFRvIIIpFLAG-CMV-l# I plasmid DNA was resuspended in 25 ml of Opti-MEMI (Invitrogen, Burlington, ON) and combined with 500 il of 293fectin (Invitrogen, Burlington, ON) resuspended in 25 ml of Opti-MEMI. The mixture was incubated for 20 min at room temperature then mixed with 293T cells (1x10') prepared in IL 293 FreeStyle media (Invitrogen, Burlington, ON), -89supplemented with 2% FBS and 50 gg/ml G418 (Invitrogen, Burlington, ON). Cells are grown for 7 days at 37*C in 8% CO 2 with shaking at 125 rpm. EGFRvIII-FLAG fusion protein purification was carried out with Anti-FLAG M2 Affinity Chromatography kit (Sigma, St. Louis, MO) according to the manufacture's protocol. 5 Monomeric fusion protein was produced as follows. First, purified protein (1508 sg), was reduced with DI' in a final concentration of 10 mM for 30 minutes at 55 *C. Then IAA (iodoacetic acid) (Sigma, St. Louis, MO) was added to 22 mM and incubated 15 minutes at room temperature in the dark then dialyzed against PBS at 4 0C in 7k MWCO dialysis cassettes (Pierce, Rockford, Ill.). 10 Examples 24-30 Binding Studies of Antibody Variants The following examples involve Biacore experiments (surface plasmon resonance) and KinExA experiments. These examples demonstrate how one can test the various antibodies and variants thereof produced by the above examples to determine if they have the desired binding 15 characteristics. All of the variants examined were variants in the 13.1.2 background. Instrumentation. All surface plasmon resonance experiments were performed using Biacore 2000 optical biosensors (Biacore, Inc., Piscataway, NJ). All Kinetic Exclusion Assays were performed using a KinExA 3000 instrument (Sapidyne Instruments, Inc., Boise, ID). 20 Reagents. Pep-3, NH 2 -LEEKKGNYVVTDHG-OH (MW = 1590 Da) (SEQ ID NO: 130), was custom synthesized and purchased from Anatech, Inc. (San Jose, CA). All mAbs were prepared in-house. The antigen EGFRvIIlpflag (iodoacetic acid reacted in order to block aggregation through free sulfhdryl groups), MW 39,907, was prepared in-house. Bovine serum albumin (BSA) fraction V 25 (#BP1605-100) was purchased from Fisher Scientific (Pittsburgh, PA). All other general reagents were purchased from Sigma-Aidrich, Inc (St. Louis, MO). All antigen and mAb samples for Biacore and KinExA analysis were prepared in vacuum degassed HBS-P buffer (0.01 M HEPES, 0.15 M NaCl, 0.005% surfactant P-20, Biacore Inc., Uppsala, Sweden) containing 100 lig/mL BSA. Biacore amine-coupling reagents, 1-ethyl-3-(3 30 dimethylaminopropyl) carbodiimide (EDC), N-hydroxysuccinimide (NHS), and ethanolamine were purchased from Biacore, Inc. Biacore surface regeneration was with a 12 second pulse of 26 mM NaOH for the pep-3/mAb 131 experiment. All other mAbs dissociated to baseline within 20 minutes. Research grade CM5 biosensor chips were purchased from Biacore, Inc. The KinExA detection antibody was Cy5-labeled goat anti-human IgG, Fey specific 35 (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, #109-175-008) and was diluted 1000-fold in HEPES buffer (0.01 M HEPES, 0.15 M NaCl, pH 7.2) from a 0.5 mg/mL stock (1 X PBS, pH 7.4). The solid phase particles used for the KinExA experiments were NHS-activated -90- Sepharose 4 Fast Flow beads (Pharmacia Biotech AB, Uppsala, Sweden, #17-0906-01). Prior to reacting the sepharose beads with antigen, a bead stock aliquot of 1.5 mL in a microcentrifuge tube was spun down and washed at least six times with cold deionized H 2 0. After rinsing the beads once with sodium carbonate buffer (0.05 M, pH 9.3), antigen (-40 Ig) in sodium carbonate buffer 5 was added to the sepharose beads. The sepharose/antigen tube was rocked overnight at 4*C. After rocking, the sepharose was spun and rinsed twice with I M Tris buffer, pH 8.3. The antigen-coated beads were then rocked for 1 hour at room temperature in I M Tris buffer with 2% BSA. Biacore Measurements. Standard EDC/NHS and carbohydrate coupling was used to covalently immobilize mAbs to 10 a CM5 sensor chip. To minimize mass transport and crowding mAbs were immobilized at levels that gave a maximum antigen binding response (R.) of no more than 50 - 100 RU. A reference flow cell on each chip was activated and blocked with no mAb immobilization to serve as a control. All Biacore kinetic experiments were conducted at 23"C. For each experiment, a series of six to eight antigen concentrations (starting with 1.01 pM pep-3) was prepared using 2-fold 15 dilutions. Antigen samples were randomly injected over the biosensor surface in triplicate at 100 pL/min. Several buffer blanks were injected intermittently over the course of an experiment for double referencing. Each pep-3 concentration and blank were injected for 90 seconds. Dissociation was followed for 13 to 180 minutes. Dissociation data for pep-3 binding to mAb 131 were acquired by alternating three additional injections of 251 nM pep-3 with three additional blank injections 20 and following the dissociation phase for 3-4 hours. All Biacore sensorgrams were processed using Scrubber software (Version 1. If, BioLogic Software, Australia). Sensorgrams were first zeroed on the y-axis and then x-aligned at the beginning of the injection. Bulk refractive index changes were removed by subtracting the reference flow cell responses. The average response of all blank injections was subtracted from all 25 analyte and blank sensorgrams to remove systematic artifacts between the experimental and reference flow cells. CLAMP biosensor data analysis software (Version 3.40, BioLogic Software, Australia) was used to determine k. and k, from the processed data sets. Data from all flow cells were globally fit to a 1:1 bimolecular binding model that included a mass transport term. For several of the mAbs the injections corresponding to the first or second concentration of the pep-3 30 series were excluded in the nonlinear kinetic fit where it was obvious that the sensorgrams were not described well by a 1:1 interaction model. The KD was calculated from the quotient kAk. KinExA Equilibrium Measurements. All KinExA experiments were conducted at room temperature (-23*C). For all equilibrium experiments, antigen was serially diluted into solutions having a constant mAb binding site 35 concentration. For the first 10 titration points the dilutions were 2-fold and the 1 1' and 12 ' serial dilutions were 10-fold. The sample flow rate for all experiments was 0.25 mL/min and the labeling antibody flow rate was 0.5 mL/min. Antigen/antibody samples were then allowed to reach -91equilibrium, which took -48-72 hr to reach. For the pep-3/mAb 131 KinExA experiment the starting concentration of pep-3 in the KD-controlled titration was 352 nM and the constant [mAb binding site] = 219 pM; for the mAb-controlled titration the starting [pep-3] = 251 nM and the [mAb binding site] = 11 nM. During the KD-controlled experiment with pep-3/mAb 131, 1.25 mL 5 of each sample was drawn through the flow cell. A sample volume of 250 pL was analyzed for the antibody-controlled experiment. Two or three replicates of each sample were measured for all equilibrium experiments. The equilibrium titration data were fit in a dual curve analysis to a 1:1 binding model using KinExA software (Version 2.4, Sapidyne Instruments). The EGFRvlIpflag/mAb 131 complex was studied with KinExA under KD-controlled 10 conditions only. The starting [EGFRvlIlpflag] was 198 nM and the [mAb binding site) was 150 pM. A sample volume of 1 mL was drawn through the flow cell. Duplicate measurements were collected for all samples. The equilibrium titration data were fit in a dual curve analysis to a 1:1 binding model using KinExA software (Version 2.4, Sapidyne Instruments). See Example 28 below for results and predicted equilibrium constant. 15 For the KinExA titrations of the EGFRvIllpflag/mAb 13.1.2 complex the starting concentration of EGFRvM was 5.26 pM (mAb-controlled), 230.1 nM (Ko-controlled) and [mAb binding site] = 9.59 nM (mAb-controlled), 498 pM (KD-controlled). During the KD-controlled experiment, 1.30 mL of each sample was drawn through the flow cell. A sample volume of 250 j1L was analyzed for the antibody-controlled experiment. Two or three replicates of each sample were 20 measured for all equilibrium experiments. The equilibrium titration data were fit in a dual curve analysis to a 1:1 binding model using KinExA software (Version 2,4, Sapidyne Instruments). Example 24 In vitro determination of binding constants for antibodies The binding kinetics of the wild type mAb 131 was observed by using a Surface Plasmon 25 Resonance (SPR) instrument from Biacore. The KD was very low, 380 pM, owing to the very slow kd and a rapid k.. Estimates for the other kinetic parameters, derived from curve fitting, were k.=2.246*10' and kd=8.502*104. In one embodiment, improved or variant antibodies with improved kinetics are taught. By improved kinetics, it is meant that one of the kinetic elements of antibody binding to an epitope is 30 superior to the same element in previously known antibodies for the same epitope. For example, an antibody that binds to pep-3 with a KD of greater (in binding ability) than 1.3*10*9 M would be an improved antibody. As such, antibodies with a KD of less than 500 rM, 500-300 nM, 300-100nM, 100-1nM, 1.3 rM, 1.3 nM to 1000 pM, 1000 pM to 900 pM, 900-5OOpM, 500-400 pM, 400-300 pM, 300-100 pM, 100-50 pM, 50-1 pM, or smaller K 0 are contemplated. 35 -92- Example 25 In vitro determination of binding constants for antibodies Similar to Example 24, the binding kinetics of mAbl3.1.2 to Pep-3 (EGFRvII epitope) were examined. The estimated KD was 67nM, but varied slightly between experiments. Estimates, 5 for the other kinetic parameters, derived from curve fitting, were kg-2.835*10' and kd=0.01922. Example 26 In vitro determination of binding constants for antibodies Similar to Example 24, the binding kinetics of mAb 095 to Pep-3 (EGFRvIl epitope) were examined. The estimated KD was 66nM. Estimates, for the kinetic parameters, derived from curve 10 fitting, were k.=1.491 * 1 0 ' and k,=9.927* 10-. Example 27 In vitro determination of binding constants for antibodies Similar to Example 24, the binding kinetics of mAb 139 to Pep-3 (EGFRvIII epitope) were examined. The estimated K 0 was 290nM. Estimates, for the kinetic parameters, derived from 15 curve fitting, were k.=10328 and kd=2.981*10- 3 . Example 28 In vitro determination of binding constants for antibodies In order to more fully analyze the binding characterisitics of the antibodies, KinExA experiments were performed to determine the binding characterisitics of the mAb 131. The KD 20 determined from a dual curve analysis was 1.74*10-'0. In a KinExA experiment, the KD for EGFRvlIlpflag to mAb 131 was 6.266* 10-". Example 29 In vitro determination of binding constants for variant antibodies In order to more fully analyze the binding characterisitics of the 13.1.2 antibodies, a 25 KinExA experiment was performed to deteremine the binding characterisitics of the mAb 13.1.2. The K 0 determined from a dual curve analysis was 7.538*10-". Additionally, the antigen in this example was the EGFRvIllpflag variant and was reacted with iodoacetic acid (IAA). Example 30 Comparison of Biacore results and KinExA results 30 The results of the previous Examples and the KinExA tests are presented in Table 30.1 below. Numbers in parentheses in Table 30.1 are 95% confidence intervals. "ND," means not determined and "*" denotes binding to EGFRvIflpflag (iodoacetic acid reacted), instead of pep-3. As is evidenced by the rate constants, mAb 131 appears to have the greatest association constant and the lowest dissociation constant, thus giving mAb 131 the lowest K 0 . 35 -93- TABLE 30.1 MAb K. (Ms') KS KD (nM) KinExA K (nM) 131 2.25 x 10u 8.50 x 10 0.380 0.174 (0.0627 on EGFRvIlIpflag) 13.1.2 2.10 (0.58) x 105 0.016 (0.003) 75 (14) 0.75 (on EGFRvIIlpflag (IAA reacted)) 095 1.49 x 10* 9.90 x 10 66 ND 139 1.03x10 2.98x10 - 290 ND Example 31 In vitro determination of binding constants for L99T-5.3 variant antibodies 5 The binding kinetics of mAb L99T-5.3 to Pep-3 (EGFRvI epitope) were examined. The first step was to immobilize 5,600 resonance units (RU) to 8,000 RU of mAb L99T-5.3 to two flow cells (Fc) of a CM5 sensor chip and 5,600 resonance units (RU) to 8,000 RU of mAb 13.1.2 to one Fc using standard EDC/NHS coupling chemistry. This surface density yielded a binding signal with pep-3 of less than 100 RU. Two CM5 sensor chips were used in total to immobilize both 10 mAbs. With the previously collected data, this produced a total of 5 independent experiments for both antibodies that allows the 95% confidence intervals to be calculated. Biacore 2000 optical biosensors were used for all studies. Next, pep-3 was flowed across the mAb immobilized biosensor surfaces. The starting concentration of pep-3 was 1.25 iM, which was followed with eight two-fold serial dilutions in 15 randomized triplicate injections. Blank injections were run every sixth sample throughout the injection series for double referencing purposes. Finally, the biosensor data was processed with Scrubber and the data was fit to curves utilizing Clamp with a 1:1 interaction model with a term included for mass transport. The high concentration injections, 1.25 pM, were excluded from the kinetic fits because it was apparent that 20 the data was not consistent with a 1:1 interaction model. Most likely, this deviation is caused by non-specific interactions occun-ing at high concentrations of pep-3. All the kinetic data fit a 1:1 interaction model satisfactorily. The estimated KD varied from 54-70 nM. Estimates, for the other kinetic parameters, which also varied slightly between runs, were k.=2.238*10' and kd=0.01217. 25 Examples 32-38 Examples 32-3 8 further examined the binding kinetics of the variant mAbs through the use of a Biacore device. The first step in these examples involved the immobilization of 5,600 resonance units (RU) to 8,000 RU of each mAb tested to one flow cell (Fc) of a CM5 sensor chip using standard EDC/NHS coupling chemistry. This surface density yielded a binding signal with 30 pep-3 of less than 100 RU. Three CM5 sensor chips were used in total to immobilize all mutant mAbs with a unique mAb immobilized to each flow cell. MAb 13.1.2 was included on one flow -94cell for two out of the three CM5 sensor chips. Biacore 2000 optical biosensors were used for all studies. Next, pep-3 was run across the mAb immobilized biosensor surfaces. The starting concentration of pep-3 was 4.98 pM, followed by eight to eleven two-fold serial dilutions in 5 randomized duplicate or triplicate injections. Blank injections were run every sixth sample throughout the injection series for double referencing purposes. Finally, the biosensor data was processed with Scrubber and fitted utilizing Clamp with a 1:1 interaction model with a term included for mass transport. Some high concentration injections (4.98 - 1.25 j±M), depending upon the mAb and its affinity, were excluded from the kinetic fits 10 when it was apparent that the data was not consistent with a 1:1 interaction model. Most likely, this deviation is caused by non-specific interactions occurring at high concentrations of pep-3. All the kinetic data fit a 1:1 interaction model. Example 32 In vitro determination of binding constants for L2170-10. 1 variant antibodies 15 The binding kinetics of mAb L217Q-10.1 to Pep-3 (EGFRvIII epitope) were examined. The estimated KD was 92 nM. Estimates, for the other kinetic parameters, derived from curve fitting, were k.=2.04* 10' and k=0.0 1885. - Example 33 In vitro determination of binding constants for L217N-2.1 variant antibodies 20 Similar to Example 32, the binding kinetics of mAb L217N-2.1 to Pep-3 (EGFRvII epitope) were examined. The estimated KD was 185 nM. Estimates, for the other kinetic parameters, derived from curve fitting, were k.=2.198* 10' and kd 1 =0.04069. Example 34 In vitro determination of binding constants for N35G-3.1 variant antibodies 25 Similar to Example 32, the binding kinetics of mAb N35G-3.1 to Pep-3 (EGFRvIII epitope) were examined. The estimated KD was 204 nM. Estimates, for the other kinetic parameters, derived from curve fitting, were k.=1.497* 10' and k,=0.030 5 7 . Example 35 In vitro determination of binding constants for variant antibodies 30 Similar to Example 32, the binding kinetics of mAb L99H-9.2 to Pep-3 (EGFRvI epitope) were examined. The estimated KD was 395 nM. Estimates, for the other kinetic parameters, derived from curve fitting, were k.=83390 and kd=0.0329 3 . Example 36 In vitro determination of binding constants for variant antibodies 35 Similar to Example 32, the binding kinetics of mAb Y172R-1.2 to Pep-3 (EGFRvIfl epitope) were examined. The estimated KD was 927 nM. Estimates, for the other kinetic parameters, derived from curve fitting, were k.=82237 and ki=0.07622. -95- Example 37 In vitro determination of binding constants for variant antibodies Similar to Example 32, the binding kinetics of mAb L99N-4.1 to Pep-3 (EGFRvIII epitope) 5 were examined. The estimated KD was 1.4 p.M. MAb L99N-4.1 was fit using a steady-state (equilibrium) binding model in order to determine the K 0 because the kinetics were too fast to be fitted. Example 38 Comparison of 13.1.2 with designed variants 10 As can be seen in Table 38.1 a mAb with improved binding characteristics was developed. The 95% confidence intervals are shown in parentheses. L99T-5.3 exhibited an enhanced c., a decreased kd, and thus a slower K 0 overall. While statistically there appears to be little if any significant difference in the equilibrium dissociation constants and kinetic rate constants of Pep-3 binding to mAbs 13.1.2 and L99T-5.3 (at the 95% confidence interval), there still seems to be an 15 intuitive bias for a marginal increase in affinity for Pep-3 binding to L99T-5.3. Moreover, when the same biosensor chip was used, L99T-5.3 seemed to always have a higher affinity than 13.1.2. Table 38.1 MAb ka (Ms-) kd (S-) Ko (nM) 13.1.2 2.10 (0.58) x 10" 0.016 (0.003) 75 (14) L99T-5.3 2.16 (0.12) x 10" 0.013 (0.001) 60(10) L217Q-10.1 2.04 x 10 0.019 92 L217N-2.1 2.20 x 1 0 0.040 185 N35G-3.1 1.50 x 0.030 204 L99H-9.2 8.34 x 104 0.033 395 Y172R-1.2 8.22 x 104 0.076 927 L99N-4.1 ND ND 1,400* 20 Additional Docking Models and Methods of Selecting Models and Predicting Binding Affinity In other embodiments, the examples described above can be performed with various length peptides rather than just peptides that are 6 amino acids in length, as long as the key binding residues are included in the peptide. For example, instead of the six amino acid peptide, EEKKGN (SEQ ID NO: 127), a seven amino acid peptide, EEKKGNY (SEQ ID NO: 131) can be used. Any 25 size peptide for the epitope can be used. In other embodiments, the peptide is selected from the following peptides: LEEKKGNYVVTDHC (SEQ ID NO: 56), LEEKKGNYVVTD (SEQ ID NO: 59), LEEKKGNYVVT (SEQ ID NO:132), and EEKKGNYVVT (SEQ ID NO:57). Any sized peptide between the short fragments disclosed herein, to the full length peptide, or variants thereof, can be used. 30 As appreciated by one of skill in the art, the presence of additional amino acids can alter the manner in which the peptide binds to the antibody. Not only does the presence of the additional -96amino acid allow for alternative and additional bonds to be formed between the peptide and the antibody, but the additional amino acid can change the structure of the peptide and the structure of the antibody upon binding of the peptide with the antibody. Thus, in one embodiment, various lengths of the epitope peptide, e.g. EEKKGN (SEQ ID NO: 127) and EEKKGNY (SEQ ID NO: 5 131), can be examined for binding properties and binding optimization. Not only will the longer fragments of the peptide provide an accurate depiction of the peptide-antibody interaction for longer segments of the peptide, but an examination of the changes in binding strength and the residues involved in binding will allow additional information concerning longer peptides to be extrapolated from the data. 10 In addition, and perhaps complementary to the testing of longer peptide fragments, additional filtering steps can be performed on the various docking models in order to select a refined docking model. An additional filtering step can allow one to filter through numerous docking models to find those that are consistent with available experimental data. In one embodiment, the filter is based on fine-resolution epitope mapping data, e.g., the 15 experimentally characterized individual residue binding profile, which could be correlated with the computed binding energy profile for each amino acid in the peptide. The binding energy profile of a seven amino acid peptide, for example, can be used to select docking models that contain similar binding energy profiles. A binding energy profile is an assignment of the binding energy of each amino acid in a peptide to the particular amino acid to create a profile of the peptide in terms of 20 each amino acid's binding energy in that model. For example, in one docking model, given a peptide comprising amino acids A and B, where A has a binding energy of -S and B has a binding energy of -20, one would have a profile of Al (at -5) and B2 (at -20). This profile could be used as a filter to select other docking models. For example, the use of this binding energy profile as a filter or "template" would result in other docking models being selected if the peptide in the 25 candidate model had a relatively low value attributed to position A, and a relatively high (larger negative, higher absolute value) value attributed to position B. In an alternative embodiment, the template requires additional limitations; for example, that the value at position B is four fold higher than the value at position A. One can compare the binding energy profile template with the profiles of the peptide in the 30 other docking models in a variety of ways. If the binding energy profile template is similar to the desired binding energy profile, then the filter can be used to pick out favorable docking models for further examination. If the binding energy profile template is dissimilar to the desired binding energy profile, then the filter can be used to eliminate unfavorable docking models. In one embodiment, the filtering process includes a template with both favorable and unfavorable binding 35 energies and the filter is used to both select and exclude docking models. As appreciated by one of skill in the art, there are many possible different binding energy profiles, and thus many different binding energy profile templates that can be used depending upon the situation. -97- In one embodiment, one can define a binding energy profile template as a template that has a series of relatively high binding energies at particular positions in the peptide. In a preferred embodiment, the binding energy profile template, and the binding energy profile selected by the template, will have relatively high binding energy at position 2, 4, or 6 of the peptide, EEKKGNY 5 (SEQ ID NO: 131). In another embodiment, the binding energy profile template will have a relatively high binding energy at'positions 2, 4, and 6 of the peptide EEKKGNY (SEQ UD NO: 131). In another embodiment, the binding energy profile template will have a relatively low binding energy attributed to position 3 of the peptide EEKKGNY (SEQ ID NO: 131). In the above discussion, the positions are assigned as follows: El, E2, K3, K4 G5, N6, Y7. 10 In one embodiment, the filtering process first involves a comparison of the binding energies at K3 and K4. Docking models that result in a relatively higher binding energy for K4 compared to K3 are selected, while docking models that result in a lower binding energy for K4 compared to K3 are filtered out. Thus, by "relatively high," it is meant that the binding energy for K4 is greater (more negative value, larger absolute value) than K3. Next, the docking models are again filtered 15 through the binding energy profile template, this time, those binding models with relatively higher energies at positions 2, 4, and 6 are selected for, while the other models can be removed. Thus, by "relatively high," it is meant that the binding energy at positions 2, 4, and 6 are higher (more negative value, larger absolute value) than the lowest binding energy in the peptide. Thus, in this embodiment, the binding energy profile template could be summarized as follows: El can be any 20 value, E2 should be greater than the lowest value, K3 should be less than K4, K4 should be greater than the lowest value, G5 can be any value, N6 should be greater than the lowest value, Y7 can be any value. Thus, El, G5, and Y7 could be any value, as long as at least one (or K3) is lower than at least one of E2, K4, and N6. In another embodiment, "relatively high" can be set to a standard value as determined through modeling or experimentation. In one embodiment, that the docking 25 models pass the first filter is more important than the docking model pass the second filtering step. As appreciated by one of skill in the art, one need not perform these two steps sequentially and they can be performed simultaneously. Additionally, these profile templates for filtering through results will vary depending upon the peptide, the antibody, and the binding conditions. One of skill in the art, given the present 30 disclosure, especially with reference to Example 14, could determine the appropriate binding energy profile template. For example, as shown in Table 14.1, there are several possible important residues for peptide binding both in the 131 and in the 13.1.2 antibody. In the 131 mAb, positions E2, K4, N6, and Y7 are important for the particular peptide tested. In the 13.1.2 mAb, positions El, E2, K4, G5, and N6 are important for the particular peptide tested. Those residues that are 35 important can be residues involved in the creation of a binding energy profile template. As clear from the discussion below, the binding energy profile template in Example 39 appears to be different from that suggested by an analysis of Example 14. Example 39 is a less stringent version -98of the template that allows more models to pass through the screening step. If one wanted to reduce the number of models that made it through the screening step, one could further add requirements concerning El and G5. The following example demonstrates both the use of a longer peptide, how it can alter the 5 results demonstrated above, what such changes can mean, as well as demonstrating the use of one of the above filters for selecting particular docking models. I Example 39 Epitope-Antibody Docking Model for a Seven Amino Acid Peptide This example demonstrates the generation of a set of docking models for a seven-residue 10 peptide complexed to the CDR region of 13.1.2 structure model. Additionally, this example demonstrates methods for selecting one docking model over another docking model. First, a structural model for the seven-residue peptide EEKKGNY (SEQ ID NO: 131) was built in an extended conformation and energy minimized with Discover 3 module in InsightIl modeling package. Next, the peptide structure was manually placed into the combining site to form 15 an initial assembly. A Monte Carlo search in the translational and rotational spaces was then automatically performed with relaxed energy constraints in the Docking module in InsightU. During the docking process, the residues within 5 Angstroms of the binding groove were allowed to move while other antibody residues were fixed. The peptide was constrained to within 5 Angstroms of the starting position. The plausible configurations found by the Monte Carlo search 20 were followed by simulated annealing and energy minimization to reach the final complex structure models. A total of 63 docking models were obtained. For each docking model, the interaction energy between the antibody and the individual residue in the peptide was calculated with Discover_3. The profile of individual residue contribution in epitope-antibody binding was inspected to select the docking models that are 25 consistent with the fine-resolution epitope mapping data, i.e., "the binding energy profile template." 19 out of 63 docking models passed this check. A typical individual residue binding energy profile is shown in Table 39.1. Consistent with the epitope mapping data, in Example 14, the binding energy for K4 is prominent, and those for N6 and E2 are large. This binding energy profile template placed particular emphasis on the fact that K4 is greater than K3. This binding energy 30 profile also placed an emphasis on the requirement that E2, K4, and N6 are relatively large. In other words, the binding energies of E2, K4, and N6 were not the lowest (least negative or smallest absolute value) binding energies in the peptide. -99- Table 39.1. Binding energy profile for individual residue in the seven-residue peptide to antibody 13.1.2 is consistent with epitope mapping data in Example 14. El E2 K3 K4 G5 N6 Y7 Total -10.97 -19.34 -13.46 -24.26 -10.1 -18.19 -15.15 -111.45 5 For the 19 models that passed the filter based on the binding energy profile, epitope antibody binding energetics simulations were performed on each of the seven mutants with affinity data (TyrI72Arg, Example 36; Leu217Asn, Example 33; Leu217Gin, Example 32; Asn35Gly, Example 34; Leu99Asn, Example 37; Leu99His, Example 35; and Leu99Thr, Example 31). Since the extent of electrostatic interaction in this complex had to be approximated, a number of different 10 dielectric constants were used in a series of calculations. The mutation was done with residue replacement followed by 30-100 steps of energy minimization to account for any local conformational changes that are induced by the side chain change. For each docking model, the interaction energy between the seven-residue peptide and the whole antibody was calculated for each mutant for the selected parameters. For each set of 8 binding energies (7 mutants plus the wild 15 type), a linear fitting procedure was done on each set of the binding data, in comparison with the logarithm of Kd. A correlation coefficient was calculated for each linear fitting. The best correlation was obtained for one model, the model with the data described in Table 39.1, with a dielectric constant 1 *r and 50 steps of energy minimization. The epitope-antibody binding energies for this model are shown in Table 39.2. The correlation coefficient was 0.80 with all the data. As 20 the KD for Leu99Asn was not measured with high degree of accuracy, see Example 37 above, a separate linear fitting was performed excluding the data for Leu99Asn. An excellent correlation coefficient of 0.91 was obtained, as shown in FIG. 20. The refined docking model is thus well represented by the above selected model. The model with space-filling peptide is shown in FIG. 21, and the hydrogen bonds are shown in FIG. 22. L3 150 is the lower section and H3 160 is the 25 upper section on FIG. 22. H2 140 is to the right of the peptide binding area. The peptide itself is placed into the binding site with B1 being positioned at the top of the page in a light shade, down through K3, K4, G5, N6, and Y7, progressively getting darker. The antibody residues involved in hydrogen bonding are shown in FIG. 22. The model produced from this example demonstrates that there are seven hydrogen bonds: K4...Q95, K4...Q95, N6...Q98, G5...H31, Y7...H31, and 30 Y7 ... W165. -100- Table 39.2. Simulation.of epitope-antibody binding energetics, in comparison with logarithm of Kd. mutant coulumbic vdw total Ln(Kd) 172Arg -19.103 -27.962 -47.065 -13.891 217Asn -19.003 -28.715 -47.718 -15.503 217Gin -18.977 -28.73 -47.707 -16.201 35Gly -19.095 -28.431 -47.526 -15.405 99Asn -18.719 -28.778 -47.497 (-13.479) 99His -18.837 -28.719 -47.556 -14.744 -99Thr -19.155 -28.704 -47.859 -16.475 WT -18.981 -28.728 -47.708 -16.269 As can be seen from the model selected in Example 39, which is represented in FIG. 21, the 5 docking model revealed some unexpected results. One interesting result is that while residues E2, K4, and N6 are important residues in the binding of the peptide as a whole, not all of these amino acids are modeled as involved in forming H-bonds with the antibody. It appears that K4 is involved in the formation of two H-bonds, both with Q95, which is consistent with K4's importance in the binding energy profile and profile template. It also appears that N6 is modeled to bond to Q98; 10 however, in this particular model, E2 does not appear to form H-bonds in the model. One interesting trend that is consistent is that each of the key residues from the binding energy profile template (e.g., E2, K4, and N6) are mostly buried and thus in close contact with the antibody binding groove. Thus, this docking model selection can account for the fact that these key residues are important because of their close interaction with the antibody. Additionally, it is possible that 15 El is involved in a hydrogen bond with W214. Example 39 also demonstrates that the above described method results in a strong correlation between binding energy and KD, suggesting that models created by this method will also allow optimization or at least a prediction of the KD of the antibody-peptide complex. As can be seen from a comparison of Example 39 and Example 19, there are some residues 20 that are important between the two models, some residues that appear only in the seven amino acid docking model, as well as some residues that do not appear to be as important in the seven amino acid docking model. For example, the seven peptide epitope appears to create H bonds between K4...Q95, K4...Q95, N6...Q98, G5...H3 1, Y7... H3 1, and Y7.. .W165. On the other hand, the six peptide epitope appears to create H bonds between E2.. .Y172, K3 ... H31, K4...H31, N6...D33, 25 N6...Y37, and N6... K55. As can be seen from the above data, both the six and the seven amino acid peptide models emphasize the importance of H 31, as both models involve H31 forming two hydrogen bonds with the peptide. While there are other possible trends between the two data sets, it also appears that many of the binding interactions have changed from the six amino acid model to -101the seven amino acid model. However, these examples demonstrate that variations due to epitope size can be detected with these models and thus the scaling up from shorter to longer epitope peptides should not be problematic in light of the present disclosure. The presence of amino acids that consistently demonstrate their importance in various binding models allows one to bias the 5 importance of the various interactions accordingly so that shorter peptide models can be more representative of longer peptide interactions. As appreciated by one of skill in the art, any of the above discussion or examples concerning the six amino acid peptide, EEKKGN (SEQ ID NO: 127), can also be applied towards the seven amino acid peptide, EEKKGNY (SEQ ID NO: 131), or any longer peptide. For instance, 10 Example 20 can be repeated with the information from Example 39 for rational design for affinity improved antibodies by site-directed mutagenesis. Furthermore, Example 21 can be repeated, using the results of Example 39, following an attempt of rational design for affinity-improved antibodies by site-directed mutagenesis to test any new antibodies derived from Example 20. In one embodiment, the results from Example 39 are used to redefine the interaction area 15 between the antibody and the peptide. For example, the paratope, for EEKKGNY (SEQ ID NO: 131), can be defined as including the other residues on the antibody that are predicted to interact with the peptide, for example, residue 95. Alternatively, as in Example 19, the paratope can be defined as all residues within 5 Angstroms of the docked peptide. In silico affinity maturation in different proteins 20 Antibody affinity maturation has been successfully done in vitro in a number of different studies. Typically, randomized mutant libraries need to be constructed by molecular biology methods and selection/screening assays need to be developed to enrich the clones with good binding capability. Selected variants then need to be purified to determine affinities. This process requires a series of lengthy and laborious experiments. The following example demonstrates that it 25 is possible to accurately predict affinity maturation through in silicon selection utilizing an antibody antigen complex structure alone. Example 40 In silicon Affinity Maturation through Antibody-Antigen Binding Energetics Simulations This Example demonstrates that in silicon antibody-antigen binding energetics simulations 30 can be used for affinity maturation. In particular, this example demonstrates .that the binding kinetics of a Fab-12 (IgG form known as rhuMAb VEGF) to VEGF (vascular endothelial growth factor) can be predicted through the above described in silicon process. The crystal structure of the VEGF-Fab complex used was located in the PDB database with the accession number lBJI, at a resolution of 2.4 angstroms. Published experimental affinity data 35 for a series of mutants of an anti-VEGF Fab were used to test the concept. The 3-D coordinates of the VEGF-Fab structure were used for carrying out in silicon mutation for the following mutants: H97Y, S100aT, T28D, 28D31H, 28D31H97Yl00aT, N31H, Y53W, 71173K, 71V73V. The -102affinity data were obtained from the paper by Chen, Y et al., (J Mol Biol., 293(4):865-81 (1999)). The energetics simulations were carried out between the various VEGF-Fab mutants, as described in Example 39. The results are listed in Table 40.1. The results from this example demonstrate that a significant correlation between the binding energy and affinity ranldng was obtained through this 5 process. The linear fitting of the binding energy versus logarithm of the relative affinity is shown in FIG. 23. The correlation coefficient of -0.91 indicates that the in silicon simulation accurately captures the detailed interaction at the atomic level. Table 40.1. Antibody-antigen binding energy simulation compared with affinity data. 10 Sequence Relative Ln(Relative Kabat Number Number Affinity Affinity) BindingEnergy H97Y 101Y 14 2.639 -59.065 S100aT 105T 1.9 0.642 -57.465 T28D 28D 1.4 0.336 -57.647 28D31H 28D31H 3.1 1.131 -57.699 28D31H97YI00a T 28D31H101Y105T 20 2.996 -59.518 N31H 31H 3.6 1.281 -57.724 Y53W 54W 1.3 0.262 -57.504 71173K 72174K 0.9 -0.105 -57.158 71V73V 72V74V 0.3 -1.204 -57.314 WT WT 1 0.000 -57.404 As is clear from the Examples above, the simulation can be extrapolated to identify higher affinity mutants without the use of in vitro experimentation. Additionally, it is clear that this 15 approach is useful for different antibodies and for different peptides. This methodology can be generally applied to perform affinity maturation in silicon, using only a high-resolution antibody antigen complex structure. In one embodiment, this use of in silicon affinity maturation will save tremendous amounts of time and resource. Example 41 20 Determination of canonical classes of antibodies Chothia, et al have described antibody structure in terms of "canonical classes" for the hypervariable regions of each immunoglobulin chain (J. Mol. Biol. 1987 Aug 20;196(4):901-17). The atomic structures of the Fab and VL fragments of a variety of immunoglobulins were analyzed to determine the relationship between their amino acid sequences and the three-dimensional -103structures of their antigen binding sites. Chothia, et al. found that there were relatively few residues that, through their packing, hydrogen bonding or the ability to assume unusual phi, psi or omega conformations, were primarily responsible for the main-chain conformations of the hypervariable regions. These residues were found to occur at sites within the hypervariable regions and in the 5 conserved beta-sheet framework. By examining sequences of immunoglobulins having unknown structure, Chothia, et al show that many immunoglobuins have hypervariable regions that are similar in size to one of the known structures and additionally contained identical residues at the sites responsible for the observed conformation. Their discovery implied that these hypervariable regions have conformations close to those 10 in the known structures. For five of the hypervariable regions, the repertoire of conformations appeared to be limited to a relatively small number of discrete structural classes. These commonly occurring main-chain conformations of the hypervariable regions were termed "canonical structures". Further work by Chothia, et al. (Nature. 1989 Dec 21-28;342(6252):877-83) and others (Martin, et al. J Mol Biol. 1996 Nov 15;263(5):800-15) confirmed that that there is a small 15 repertoire of main-chain conformations for at least five of the six hypervariable regions of antibodies. Some of the antibodies described above were analyzed to determine the canonical class for each of the antibody's complementarity determining regions (CDRs). As is known, canonical classes have only been assigned for CDR1 and CDR2 of the antibody heavy chain, along with 20 CDRI, CDR2 and CDR3 of the antibody light chain. The table below (41.1) summarizes the results of the analysis. The Canonical Class data is in the form of *HCDRl-HCDR2-LCDR1 LCDR2-LCDR3, wherein "HCDR" refers to the heavy chain CDR and "LCDR" refers to the light chain CDR. Thus, for example, a canonical class of 1-3-2-1-5 refers to an antibody that has a HCDR1 that falls into canonical class 1, a HCDR2 that falls into canonical class 3, a LCDR1 that 25 falls into canonical class 2, a LCDR2 that falls into canonical class 1, and a LCDR3 that falls into canonical class 5. TABLE 41.1 H1-H2-L1 mAb L2-L3 139 1-3-2-1-1 250 1-3-2-1-1 123 1-3-4-1-1 131 1-3-4-1-1 13_1_2 1-3-4-1-1 211 1-3-4-1-1 318 1-3-4-1-1 333 1-3-4-1-1 342 1-3-4-1-1 95 3-1-4-1-1 150 3-Y-4-1-1 170 3-Y-4-1-1 -104- Each CDR (except for H3) was assigned to a canonical structure if it satisfies the length requirement and matches the key residues defined in the canonical class. The amino acids defined for each antibody can be found, for example, in the articles by Chothia, et al. referred to above. -105- Equivalents [04121 The foregoing description and Examples detail certain preferred embodiments of the invention and describes the best mode contemplated by the inventors. It will be appreciated, however, that no matter how detailed the foregoing may appear in text, the 5 invention may be practiced in many ways and the invention should be construed in accordance with the appended claims and any equivalents thereof [0413] Throughout this specification and the claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" and "comprising", will be understood to imply the inclusion of a stated integer or step or group of 10 integers or steps but not the exclusion of any other integer or step or group of integers or steps. 104141 The reference in this specification to any prior publication (or information derived from it), or to any matter which is known, is not, and should not be taken as an acknowledgment or admission or any form of suggestion that that prior publication (or information derived from it) or known matter forms part of the common general knowledge in 15 the field of endeavour to which this specification relates. -106-

Claims (8)

1. An isolated human monoclonal antibody that comprises a heavy chain polypeptide and a light chain polypeptide, wherein the antibody is conjugated to a toxin 5 selected from the group consisting of: Maytansinoid (DM-1), saporin, Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E, wherein at least one of the heavy or light chain polypeptides comprises an amino acid sequence that is at least 90% identical to the amino acid sequence selected from the group consisting of: SEQ ID NO: 2, SEQ ID NO: 19, SEQ ID NO: 142, SEQ ID NO: 144, and any combination thereof. 10 2. The antibody of Claim 1, wherein the toxin is selected from the group consisting of: Maytansinoid (DM-l) and saporin.

3. The antibody of Claim 1, wherein the toxin comprises Maytansinoid (DM 1).

4. The antibody of Claim 1, wherein the toxin is selected from the group 15 consisting of: Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E.

5. The isolated human monoclonal antibody of Claim 1, wherein the heavy chain polypeptide comprises the amino acid sequence of SEQ ID NO: 2.

6. The isolated human monoclonal antibody of Claim 1, wherein the light chain polypeptide comprises the amino acid sequence of SEQ ID NO: 142. 20 7. The isolated human monoclonal antibody of Claim 1, wherein the antibody binds to a peptide that consists of the sequence LEEKKGNYVVTDHC (SEQ ID NO: 56).

8. The isolated human monoclonal antibody of Claim 1, wherein the heavy chain polypeptide comprises an amino acid sequence that is at least about 95% identical to the amino acid sequence of SEQ ID NO: 2 and the light chain polypeptide comprises an amino 25 acid sequence that is at least about 95% identical to the amino acid sequence of SEQ ID NO:

19. 9. The isolated human monoclonal antibody of Claim 1, wherein the heavy chain polypeptide comprises at least one of the following amino acid sequences: SEQ ID NO: 108, SEQ ID NO: 110, and SEQ ID NO: 112. 30 10. The isolated human monoclonal antibody of Claim I, wherein the heavy chain polypeptide comprises all three of the following amino acid sequences: SEQ ID NO: 108, SEQ ID NO: 110, and SEQ ID NO: 112. 11. The isolated human monoclonal antibody of Claim 1, wherein the light chain polypeptide comprises at least one of the following amino acid sequences: SEQ ID NO: 35 101, SEQ ID NO: 103, and SEQ ID NO: 105. 12. The isolated human monoclonal antibody of Claim 1, wherein the light chain polypeptide comprises all three of the following amino acid sequences: SEQ ID NO: 101, SEQ ID NO: 103, and SEQ ID NO: 105. 13. The isolated human monoclonal antibody of Claim 1, wherein the heavy 40 chain polypeptide comprises an amino acid sequence that is at least about 95% identical to the amino acid sequence of SEQ ID NO: 142 and the light chain polypeptide comprises an amino - 107 - acid sequence that is at least about 95% identical to the amino acid sequence of SEQ ID NO:

144. 14. The isolated human monoclonal antibody of Claim 1, wherein the heavy chain polypeptide comprises at least one of the following amino acid sequences: SEQ ID NO: 5 121, SEQ ID NO: 123, and SEQ ID NO: 125. 15. The isolated human monoclonal antibody of Claim 1, wherein the heavy chain polypeptide comprises all three of the following amino acid sequences: SEQ ID NO: 121, SEQ ID NO: 123, and SEQ ID NO: 125. 16. The isolated human monoclonal antibody of Claim 1, wherein the light 10 chain polypeptide comprises at least one of the following amino acid sequences: SEQ ID NO: 101, SEQ ID NO: 116, and SEQ ID NO: 118. 17. The isolated human monoclonal antibody of Claim 1, wherein the light chain polypeptide comprises all three of the following amino acid sequences: SEQ ID NO: 101, SEQ ID NO: 116, and SEQ ID NO: 118. 15 18. An isolated human monoclonal antibody that binds to EGFRvIII comprising: a heavy chain polypeptide comprising the following complementarity determining regions (CDRs): a heavy chain CDRI that is a CDR1 in SEQ ID NO: 142; a heavy chain CDR2 that is a CDR2 in SEQ ID NO: 142; a heavy chain CDR3 that is a CDR3 in SEQ ID 20 NO: 142; a light chain polypeptide comprising the following CDRs: a light chain CDR1 that is a CDR1 in SEQ ID NO: 144; a light chain CDR2 that is a CDR2 in SEQ ID NO: 144; and a light chain CDR3 that is a CDR3 in SEQ ID NO: 144; and a toxin conjugated to the antibody, wherein the toxin is selected from the group 25 consisting of: Maytansinoid (DM-l), saporin, Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E. 19. The antibody of Claim 18, wherein the toxin is selected from the group consisting of: Maytansinoid (DM-1) and saporin. 20. The antibody of Claim 18, wherein the toxin comprises Maytansinoid 30 (DM-I). 21. The antibody of Claim 18, wherein the toxin is selected from the group consisting of: Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E. 22. The isolated human monoclonal antibody of Claim 18, wherein the heavy 35 chain polypeptide comprises all three of the following amino acid sequences: SEQ ID NO: 121, SEQ ID NO: 123, and SEQ ID NO: 125. 23. The isolated human monoclonal antibody of Claim 18, wherein the light chain polypeptide comprises all three of the following amino acid sequences: SEQ ID NO: 101, SEQ ID NO: 116, and SEQ ID NO: 118. 40 24. An isolated human monoclonal antibody that binds to EGFRvIII - 108 - comprising: a heavy chain polypeptide comprising the following complementarity determining regions (CDRs): a heavy chain CDRl that is a CDRI in SEQ ID NO: 2; a heavy chain CDR2 that is a CDR2 in SEQ ID NO: 2; a heavy chain CDR3 that is a CDR3 in SEQ ID NO: 2; 5 a light chain polypeptide comprising the following CDRs: a light chain CDR] that is a CDRI in SEQ ID NO: 19; a light chain CDR2 that is a CDR2 in SEQ ID NO: 19; and a light chain CDR3 that is a CDR3 in SEQ ID NO: 19; and a toxin conjugated to the antibody, wherein the toxin is selected from the group consisting of: Maytansinoid (DM-1), saporin, Auristatin E toxin (AEFP), monomethyl 10 auristatin E (MMAE), and Auristatin E. 25. The antibody of Claim 24, wherein the toxin is selected from the group consisting of: Maytansinoid (DM- 1) and saporin. 26. The antibody of Claim 24, wherein the toxin comprises Maytansinoid (DM-1). 15 27. The antibody of Claim 24, wherein the toxin is selected from the group consisting of: Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E. 28. The isolated human monoclonal antibody of Claim 24, wherein the heavy chain polypeptide comprises all three of the following amino acid sequences: SEQ ID NO: 108, SEQ ID NO: 110, and SEQ ID NO: 112. 20 29. The isolated human monoclonal antibody of Claim 24, wherein the light chain polypeptide comprises all three of the following sequences: SEQ ID NO: 101, SEQ ID NO: 103, and SEQ ID NO: 105. 30. An isolated human monoclonal antibody that binds to an epitope on an EGFRvIII protein, wherein said antibody binds to the amino acid sequence 25 LEEKKGNYVVTDHC (SEQ ID NO: 56), wherein the second lysine in SEQ ID NO: 56 is part of said epitope, and wherein the antibody is conjugated to a toxin selected from the group consisting of: Maytansinoid (DM-1), saporin, Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E. 31. The antibody of Claim 30, wherein the toxin is selected from the group 30 consisting of: Maytansinoid (DM-1) and saporin. 32. The antibody of Claim 30, wherein the toxin comprises Maytansinoid (DM- 1). 33. The antibody of Claim 30, wherein the toxin is selected from the group consisting of: Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E. 35 34. An isolated human monoclonal antibody that binds to an EGFRvIII protein, wherein said antibody competes for binding to EGFRvIII with the antibody of Claim 1, and wherein the antibody is conjugated to a toxin selected from the group consisting of: Maytansinoid (DM-1), saporin, Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E. 40 35. The antibody of Claim 34, wherein the toxin is selected from the group -109- consisting of: Maytansinoid (DM- 1) and saporin. 36. The antibody of Claim 34, wherein the toxin comprises Maytansinoid (DM-1). 37. The antibody of Claim 34, wherein the toxin is selected from the group 5 consisting of: Auristatin E toxin (AEFP), monomethyl auristatin E (MMAE), and Auristatin E. 38. A method of killing a targeted cell comprising contacting the targeted cell with at least one of the antibodies from Claim 1 to Claim 37. 39. The method of Claim 38, wherein the method is performed in vivo. 10 40. The method of Claim 39, wherein the method is performed on mammal. 41. The method of Claim 40, wherein the mammal is a human. 42. The method of Claim 40, wherein the mammal suffers from a cancer involving epithelial cell proliferation. 43. The method of Claim 42, wherein the cancer comprises a lung carcinoma, 15 breast carcinoma, head & neck cancer, prostate carcinoma or glioblastoma. 44. The method of Claim 38, wherein the method is performed In vitro. 45. The method of Claim 38, wherein said antibody associated with said toxin is at least ten fold more toxic to said targeted cell than to cells that do not express said peptide. 46. The method of Claim 38, wherein the antibody is associated with the toxin 20 via a peptide linker. 47. The method of Claim 38, wherein the antibody is associated with the toxin via a second antibody. 48. Use of at least one of the antibodies of Claims I to 37, in the preparation of a medicament for the treatment of cancer cells expressing Epidermal Growth Factor 25 Receptor viii (EGFRvII). 49. The use of Claim 42, wherein said cancer cells are epithelial cells. 50. The use of Claim 42, wherein said cancer cells are selected from the group consisting of lung, colon, gastric, renal, prostate, breast, glioblastoma or ovarian cancer cells. 30 51. An isolated human monoclonal antibody according to any one of claims I to 37, or a method of killing a targeted cell according to any one of claims 38 to 47, or the use according to any one of claims 48 to 50, substantially as hereinbefore described with reference to the Figures and/or Examples. -110-