AU2002365978A1

AU2002365978A1 - Tumor specific monoclonal antibodies

Info

Publication number: AU2002365978A1
Application number: AU2002365978A
Authority: AU
Inventors: Jeffry D Watkins
Original assignee: Applied Molecular Evolution Inc
Current assignee: Applied Molecular Evolution Inc
Priority date: 2001-11-19
Filing date: 2002-11-19
Publication date: 2003-06-10
Also published as: EP1446416A1; WO2003044036A1; CN1615315A; EP1446416A4; CA2467020A1; BR0214288A; JP2005517388A; MXPA04004658A

Description

WO 03/044036 PCT/US02/37134 1 TUMOR SPECIFIC MONOCLONAL ANTIBODIES BACKGROUND OF THE INVENTION The present invention relates generally to cancer and, more specifically, to human monoclonal 5 antibodies for the treatment and diagnosis of cancer. Cancer is one of the leading causes of death in the United States. Each year, more than half a million Americans die from cancer, and more than one million are newly diagnosed with the disease. In cancer, neoplastic 10 cells escape from their normal growth regulatory mechanisms and proliferate in an uncontrolled fashion, leading to the development of a malignant tumor. Tumor cells can metastasize to secondary sites if treatment of the primary tumor is either not complete or not initiated 15 before substantial progression of the disease. Early diagnosis and effective treatment of malignant tumors is therefore essential for survival. The current methods for treating cancer include surgery, radiation therapy and chemotherapy. A major 20 problem with each of these treatments is their lack of specificity for cancer cells. For instance, surgical removal of the tumor is often incomplete. Even a few residual neoplastic cells can be lethal, as they can rapidly proliferate and metastasize to other sites. 25 Radiation and chemotherapy also have serious limitations. These therapies target all growing cells of the body, including both normal and neoplastic cells. Due to their toxicity to normal tissues, the amount of radiation or chemotherapeutic agent that can be safely used is often WO 03/044036 PCT/USO2/37134 2 inadequate to kill all neoplastic cells. Additionally, their toxicity to normal tissues is manifested by unpleasant side effects, including nausea and hair loss, that severely reduce the quality of life for the cancer 5 patient undergoing these treatments. Clearly, a more selective and effective means of treating cancer is needed. Monoclonal antibodies are homogeneous preparations of immunoglobulin proteins that specifically 10 recognize and bind to regions, or epitopes, of their corresponding antigens. Neoplastic cells selectively express antigens which are not present on normal cells. Thus, monoclonal antibodies can be produced that are directed against tumor-specific antigens. Such 15 tumor-specific antigens can be linked to therapeutic moieties that kill or arrest the growth of neoplastic cells. In addition, monoclonal antibodies can be linked to diagnostic moieties that allow the imaging of neoplastic cells. Thus, monoclonal antibodies directed 20 against antigens selectively expressed by tumor cells compared to normal cells can be beneficially used for the early detection and effective treatment of cancer. Most current immunotherapeutic strategies for cancer have been of limited utility due to their reliance 25 on mouse monoclonal antibodies. Mouse monoclonal antibodies can be produced easily and in virtually unlimited quantities using hybridoma technology. However, when administered to humans, they can be recognized as foreign by the human immune system and be 30 neutralized before exerting their therapeutic effect on the diseased tissue. Furthermore, the murine immune WO 03/044036 PCT/USO2/37134 3 system often preferentially recognizes immunodominant epitopes of normal human antigens present on tumor cells. Thus, human tumor-specific antigens often fail to generate therapeutically beneficial murine antibodies. 5 Human monoclonal antibodies can overcome both of these limitations. Most importantly, human monoclonal antibodies are not as immunogenic as murine antibodies. Therefore, tumor-specific human monoclonal antibodies will be able to more effectively target and eliminate 10 neoplastic cells. Furthermore, the human immune system is less likely to generate antigens against epitopes present on normal cells, increasing the odds of generating and successfully identifying tumor-specific antigens. Additionally, the repertoire of the human 15 immune system is different from that of the mouse, containing potentially novel antibody specificities. Current procedures to produce tumor-specific human monoclonal antibodies have generally started with lymphocytes obtained from tumor-bearing patients. These 20 procedures rely on the stimulation and expansion of tumor-reactive lymphoctyes in vivo. These procedures are seriously limited by the narrow range of antigen specificities of activated B-cells of cancer patients. As it is clearly not possible to immunize individuals in 25 vivo with tumor cells, as one can with mice, it has not been possible to readily generate tumor-specific human monoclonal antibodies to any given antigen or tumor cell type. Procedures for generating tumor-specific antibodies of any desired specificity would be very 30 beneficial for effective immunotherapy and immunodiagnosis.

WO 03/044036 PCT/USO2/37134 4 Thus, there exists a need for improved tumor-specific human monoclonal antibodies for the therapy and diagnosis of cancer. The present invention satisfies this need and provides related advantages as 5 well. SUMMARY OF THE INVENTION The invention provides an isolated human monoclonal antibody or functional fragment thereof, including a complementary determining region having 10 substantially the amino acid sequence of a CDR selected from the group consisting of SEQ ID NOs:10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44, wherein the human monoclonal antibody or functional fragment thereof specifically binds a neoplastic cell or 15 antigen thereof. The invention also provides an isolated nucleic acid, encoding a human monoclonal antibody or functional fragment thereof, including a nucleotide sequence encoding substantially the amino acid sequence of a CDR 20 encoded by a nucleotide sequence selected from the group consisting of SEQ ID NOs:9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43 and 45. A method for reducing neoplastic cell proliferation is also provided. The method consists of 25 administering an effective amount of a human monoclonal antibody or functional fragment of the invention. Also provided is a method of detecting a neoplastic cell in a sample. The method consists of contacting a cell with a monoclonal antibody or functional fragment of the WO 03/044036 PCT/US02/37134 5 invention and detecting the specific binding of the human monoclonal antibody or functional fragment to the sample. The presence or increased level compared to a normal cell of the monoclonal antibody or functional fragment 5 indicates the presence of a neoplastic cell in the sample. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 shows the binding of LH11238 and LH13 antibodies to the surface of live H3464 cells. Figure 1A 10 shows fluorescent activated cell sorting (FACS) analysis using LH11238 antibody. Figure lB shows FACS analysis using LH13 antibody. Figure 2 shows that LH13 antigen is secreted by H3396 cells. Figure 2A shows that conditioned medium 15 from H3396 cells (closed circles) competes for binding of LH13 antibody to fixed cell monolayers. Figure 2B shows that LH13 antigen is secreted by H3396 cells and binds to culture dishes. Figure 3 shows the purification of LH13 antigen 20 by anion exchange chromatography on a Q Sepharose column. Figure 4 shows that LH13 antigen is susceptible to trypsin and endoglycosidase-F/peptide-N-lycosidase F treatment. Figure 5 shows characterization of recombinant 25 LH13, LH11238 and variant Fabs in an ELISA format. Panel A shows results for LH13 antibody (open circles), the HCDR3 variant S97N (filled circles), the LCDR3 variant WO 03/044036 PCT/tUS02/37134 6 R91Y (open squares), and the combinatorial mutant 4H7 (filled squares) each titrated against partially enriched LH13 antigen and detected with goat anti-human kappa alkaline phosphatase conjugate. Panel B shows results for 5 the LH11238 antibody (open circles) and the LCDR3 mutants Q89W (open squares), N93C (open triangles), N94C (filled squares), and P95aR (filled circles) each titrated against fixed monolayers of the H3396 tumor cell line and detected with goat anti-human kappa alkaline phosphatase. 10 DETAILED DESCRIPTION OF THE INVENTION The invention is directed to tumor-specific human monoclonal antibodies and functional fragments thereof. The human monoclonal antibodies and functional fragments thereof of the invention specifically bind to 15 neoplastic cells compared to normal cells, and can be used for selectively targeting tumors. These antibodies are human in origin, and are unlikely to generate an immune response upon administration to a human subject. Therefore, they can be conjugated to cytotoxic or 20 cytostatic agents and used to selectively target cancer cells for the elimination of tumors. The tumor-specific human monoclonal antibodies can also be used in diagnostic procedures to identify neoplastic cells. Early detection of cancer greatly increases the chances 25 of an individual surviving the disease. In one embodiment, the invention provides methods for generating hybridomas producing tumor-specific human monoclonal antibodies. Normal human splenocytes are immunized in vitro with tumor cells or 30 tumor cell membranes in a mixed lymphocyte reaction.

WO 03/044036 PCT/US02/37134 7 Such immunized splenocytes are then immortalized to produce hybridomas providing an unlimited supply of tumor-specific human monoclonal antibodies. Using normal cells as the source of lymphocytes greatly enlarges the 5 repertoire of different tumor-specific antibodies that can be obtained for the treatment of cancer. Additionally, the type of cell or cell membrane used as the antigen in the method of the invention can be varied as needed to efficiently produce antibodies for different 10 human therapeutic and diagnostic applications. In another embodiment, the invention is directed to nucleic acids encoding human tumor-specific monoclonal antibodies. The nucleic acids can be used to express the encoded human antibodies or fragments 15 thereof. Additionally, the encoding nucleic acids can be recombinantly engineered to produce modified human antibodies or functional fragments which exhibit higher affinity or higher selectivity for tumor cells or to augment other functional characteristics of the encoded 20 antibodies. Such modified antibodies can additionally be constructed to contain other therapeutically advantageous modifications, such as enhanced association with cytotoxic agents or increased stimulation of the immune system. 25 In a further embodiment, the invention is directed to antigens recognized by the tumor-specific human monoclonal antibodies. The antigens can be used for cancer diagnostic procedures and to develop specific cytotoxic reagents for cancer therapy. Tumor-specific 30 antigens can also be used as a vaccine and administered to individuals at risk of cancer to develop an effective WO 03/044036 PCT/US02/37134 8 immune response against neoplastic cells. The nucleic acids encoding the tumor-specific antigens can be used as probes in diagnostic procedures, or modified by recombinant methods to develop specific inhibitors of the 5 antigen. The basic structure of an immunoglobulin or antibody molecule consists of two identical light chains and two identical heavy chains, which associate non-covalently and can also be linked by disulfide bonds. 10 Each heavy and light chain contains an amino-terminal variable region of about 110 amino acids and constant sequences in the remainder of the chain. The variable region includes several hypervariable regions, or complementarity-determining regions (CDRs) that form the 15 antigen-binding site of the antibody molecule and determine its specificity. On either side of the CDRs of the heavy and light chains is the framework region, a relatively conserved sequence of amino acids that anchors and orients the CDRs. The constant region consists of 20 one of five heavy chain sequences (p, y, 5, a, or e) and one of two light chain sequences (K or X). The heavy chain constant region sequences determine the isotype of the antibody and the effector functions of the molecule. As used herein, the term "human monoclonal 25 antibody" is intended to mean a monoclonal antibody comprising substantially the same heavy and light chain CDR amino acid sequences as found in a particular human immunoglobulin. An amino acid sequence which is substantially the same as a heavy or light chain CDR 30 exhibits a considerable amount or extent of sequence identity when compared to a reference sequence and WO 03/044036 PCT/USO2/37134 9 contributes favorably to specific binding of an antigen bound specifically by an antibody having the reference sequence. Such identity is definitively known or recognizable as representing the amino acid sequence of 5 the particular human monoclonal antibody. Substantially the same heavy and light chain CDR amino acid sequence can have, for example, minor modifications or conservative substitutions of amino acids so long as the ability to bind a particular antigen is maintained. The 10 term "human monoclonal antibody" is intended to include a monoclonal antibody with substantially human CDR amino acid sequences produced, for example, by recombinant methods, by lymphocytes or by hybridoma cells. As used herein, the term "substantially the 15 same" when used in reference to an amino acid sequence is intended to mean an amino acid sequence that has sufficient structural identity or similarity to a reference amino acid sequence to be considered by one skilled in the art to have a specific function of the 20 polypeptide or fragment thereof encoded by the reference amino acid sequence. The term can include primary structural identity or similarity which is also referred to in the art as sequence identity or sequence similarity, respectively. An amino acid sequence that is 25 substantially the same as a reference sequence of an antibody, or functional fragment thereof can have a primary structure that is at least 70% identical to a reference sequence including, for example, sequences that are at least 80%, at least 83%, at least 85%, at least 30 90%, at least 95%, at least 97% or at least 98% identical to a reference sequence. The term can include tertiary structural identity or similarity, where tertiary WO 03/044036 PCT/USO2/37134 10 structure is understood to refer to the three-dimensional structure of a functionally active antibody or fragment thereof. A specific function included in the term can be any biological activity that is specific to the 5 polypeptide or fragment thereof encoded by the reference amino acid sequence including, for example, specific binding to a neoplastic cell or antigen thereof. A nucleic acid sequence that is substantially the same as a reference sequence includes one that 10 encodes the same polypeptide amino acid sequence. Nucleic acid sequences that are different from each other but encode identical amino acid sequences are commonly referred to in the art as having silent differences due to degeneracy of the genetic code. The term can include 15 a full sequence or any portion thereof such as a particular codon. Substantially identical amino acid or nucleic acid sequences can be identified using methods described below. As used herein, the term "CDR" is intended to 20 mean the non-contiguous antigen combining sites found within the variable region of both heavy and light chain polypeptides. This particular region has been described by Kabat et al., U.S. Dept. of Health and Human Services, "Sequences of Proteins of Immunological Interest" (1983) 25 and by Chothia et al., J. Mol. Biol. 196:901-917 (1987) and additionally by MacCallum et al., J. Mol. Biol. 262:732-745 (1996), which are incorporated herein by reference, where the definitions include overlapping or subsets of amino acid residues when compared against each 30 other. Nevertheless, application of either definition to refer to a CDR of an antibody or functional fragment WO 03/044036 PCT/US02/37134 11 thereof is intended to be within the scope of the term as defined and used herein. The appropriate amino acid residues which encompass the CDRs as defined by each of the above cited references are set forth below in Table I 5 as a comparison. The exact residue numbers which encompass a particular CDR will vary depending on the structure of the CDR. Those skilled in the art can routinely determine which residues comprise a particular CDR given the variable region amino acid sequence of the 10 antibody. Those skilled in the art can compare two or more antibody sequences by defining regions or individual amino acid positions of the respective sequences with the same CDR definition as described in further detail below. TABLE I: CDR Definitions 15 Kabat' Chothia 2 MacCallum VH CDR1 31-35 26-32 30-35 VH CDR2 50-65 52-56 47-58 VH CDR3 95-102 95-102 93-101 VL CDR1 24-34 24-34 30-36 20 VL CDR2 50-56 50-56 46-55 VL CDR3 89-97 89-97 89-96 1 Residue numbering follows the nomenclature of Kabat et al., supra 2 Residue numbering follows the nomenclature of Chothia et 25 al., supra Residue numbering follows the nomenclature of MacCallum et al., supra As used herein, the term "functional fragment", 30 when used in reference to a human monoclonal antibody, is WO 03/044036 PCT/US02/37134 12 intended to refer to a portion of the monoclonal antibody which is capable of specifically binding an antigen that is specifically bound by the reference antibody. A functional activity can also be, for example, an effector 5 function provided by an antibody constant region. Human monoclonal antibody functional fragments include, for example, individual heavy or light chains and fragments thereof, such as VL, VH and Fd; monovalent fragments, such as Fv, Fab, and Fab'; bivalent fragments such as 10 F(ab') 2 ; single chain Fv (scFv); Fc fragments and CDR domains. Such terms are described in, for example, Harlow and Lane, Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, New York (1989); Molec. Bioloqgy and Biotechnology: A Comprehensive Desk Reference 15 (Myers, R.A. (ed.), New York: VCH Publisher, Inc.); Huston et al., Cell Biophysics, 22:189-224 (1993); Plickthun and Skerra, Meth. Enzymol., 178:497-515 (1989) and in Day, E.D., Advanced Immunochemistry, Second Ed., Wiley-Liss, Inc., New York, NY (1990), which are 20 incorporated herein by reference. The term functional fragment is intended to include, for example, fragments produced by protease digestion or reduction of a human monoclonal antibody and by recombinant DNA methods known to those skilled in the art. 25 The term "VL fragment," as used herein, refers to a fragment of the light chain of a human monoclonal antibody which includes all or part of the light chain variable region, including the CDRs. A VL fragment can further include light chain constant region sequences. 30 The term "Fd fragment," as used herein, refers to a fragment of the heavy chain of a human monoclonal WO 03/044036 PCT/US02/37134 13 antibody which includes all or part of the heavy chain variable region, including the CDRs. An Fd fragment can further include heavy chain constant region sequences. The term "Fv fragment," as used herein, refers 5 to a monovalent antigen-binding fragment of a human monoclonal antibody, including all or part of the variable regions of the heavy and light chains, and absent of the constant regions of the heavy and light chains. The variable regions of the heavy and light 10 chains include, for example, the CDRs. For example, an Fv fragment includes all or part of the amino terminal variable region of about 110 amino acids of both the heavy and light chains. The term "Fab fragment," as used herein, refers 15 to a monovalent antigen-binding fragment of a human monoclonal antibody that is larger than an Fv fragment. For example, an Fab fragment includes the variable regions, and all or part of the first constant domain of the heavy and light chains. Thus, a Fab fragment 20 additionally includes, for example, amino acid residues from about 110 to about 220 of the heavy and light chains. The term "Fab' fragment," as used herein, refers to a monovalent antigen-binding fragment of a 25 human monoclonal antibody that is larger than a Fab fragment. For example, a Fab' fragment includes all of the light chain, all of the variable region of the heavy chain, and all or part of the first and second constant domains of the heavy chain. For example, a Fab' fragment WO 03/044036 PCT/US02/37134 14 can additionally include some or all of amino acid residues 220 to 330 of the heavy chain. The term "F(ab') 2 fragment," as used herein, refers to a bivalent antigen-binding fragment of a human 5 monoclonal antibody. A F(ab') 2 fragment includes, for example, all or part of the variable regions of two heavy chains and two light chains, and can further include all or part of the first constant domains of two heavy chains and two light chains. 10 One skilled in the art knows that the exact boundaries of a fragment of a human monoclonal antibody can be variable, so long as the fragment maintains a functional activity. Using well-known recombinant methods, one skilled in the art can engineer a nucleic 15 acid to express a functional fragment with any endpoints desired for a particular application. As used herein, the term "label" is intended to mean a moiety that can be attached to a human monoclonal 20 antibody or other molecule of the invention. Moieties can be used, for example, for therapeutic or diagnostic procedures. Therapeutic labels include, for example, moieties that can be attached to a molecule of the 25 invention and used to reduce the uncontrolled proliferation or viability of a neoplastic cell. A label which can decrease cell proliferation or viability can be, for example, a cytotoxic or cytostatic agent, growth factor, cell death receptor agonist or immune modulator.

WO 03/044036 PCT/USO2/37134 15 Diagnostic labels include, for example, moieties which can be detected by analytical methods. Analytical methods include, for example, qualitative and quantitative procedures. Qualitative analytical methods 5 include, for example, immunohistochemistry and indirect immunofluorescence. Quantitative analytical methods include, for example, immunoaffinity procedures such as radioimmunoassay, ELISA or FACS analysis. Analytical methods also include both in vitro and in vivo imaging 10 procedures. Specific examples of diagnostic labels that can be detected by analytical means include enzymes, radioisotopes, chromophores, fluorochromes, chemiluminescent markers, and biotin. A label can be attached directly to a molecule 15 of the invention, or be attached to a secondary binding agent that specifically binds a molecule of the invention. Such a secondary binding agent can be, for example, a secondary antibody. A secondary antibody can be either polyclonal or monoclonal, and of human, rodent 20 or chimeric origin. As used herein, the term "cytotoxic or cytostatic agent" is intended to mean an agent which reduces the viability or proliferative potential of a cell. Such agents can be attached, for example, to a 25 human monoclonal antibody or other molecule of the invention and used to target cells or tissues. The targeted cells and tissues can include, for example, neoplastic cells and tumors. Examples of targeted cells and tissues include those derived from breast, lung and 30 ovarian tissue. Cytotoxic or cytostatic agents can function in a variety of ways to reduce cell viability or WO 03/044036 PCT/US02/37134 16 proliferation. Such functions include, for example, inhibiting DNA synthesis, inhibiting cell division, inducing apoptosis, or inducing non-apoptotic cell killing. Specific examples of cytotoxic and cytostatic 5 agents include pokeweed antiviral protein, abrin, ricin and each of their A chains, doxorubicin, cisplastin, Iodine-131, Yttrium-90, Rhenium-188, Bismuth-212, Taxol, 5-Fluorouracil VP-16, Bleomycin, methotrexate, vindesine, adriamycin, vincristine, vinblastine, BCNU, mitomycin and 10 cyclophosphamide and certain cytokines such as TNF-a and TNF-0. Thus, cytotoxic or cytostatic agents can include, for example, radionuclides, chemotherapeutic drugs, proteins and lectins. As used herein, the term "specific binding" is 15 intended to mean a selective interaction of a human monoclonal antibody or functional fragment thereof with an antigen. For such an interaction to be selective, a human monoclonal antibody will not substantially bind, or can be made to not substantially bind, to markers other 20 than the particular antigen. Specific binding can include, for example, association constants (Ka) of about l0 s

M

- 1 or higher. Thus, specific binding can include an association constant that is at least about 1 x 106 M - 1 , 1 x 107 M 1 , or 1 x 108 M - 1 . 1 x 109 M -1 , 1 x 1010 M - 1 , 25 1 x 10" M - or 1 x 1012 M

-

. Specific binding can also include, for example, high avidity interactions. As used herein, the term "cancer" refers to a pathological condition characterized by the presence of neoplastic cells. Neoplastic cells are cells that 30 exhibit an abnormal morphological or proliferative phenotype. Such cells are characterized by, for example, WO 03/044036 PCT/USO2/37134 17 anchorage independent cell growth, proliferation in reduced-serum medium, and loss of contact inhibition. Such cells are also characterized by, for example, abnormal new growth of tissue, such as a tumor, 5 angiogenic vasculature, and invasion into surrounding tissue. Neoplastic cells can also metastasize from a primary tumor to other sites in the body. For example, a tumor of the breast, lung or ovary can metastasize to other organs yet still be recognizable as being comprised 10 of breast, lung or ovarian cells. Thus, the term "tumor" or "cancer" in reference to breast, lung or ovary is intended to include metastases of these tumors to other organs of the body. 15 As used herein, the term "effective amount" is intended to mean the amount of a molecule of the invention which can reduce proliferation of neoplastic cells. The actual amount considered to be an effective amount for a particular application can depend, for 20 example, on such factors as the affinity, avidity, stability, bioavailability or selectivity of the molecule, the moiety attached to the molecule, the pharmaceutical carrier and the route of administration. Effective amounts can be determined or extrapolated using 25 methods known to those skilled in the art. Such methods include, for example, in vitro assays with cultured cells or tissue biopsies and credible animal models. As used herein the term "isolated," when used in the reference to an antibody, means separated from one 30 or more compound that is found with the antibody in nature or in a synthetic reaction used to produce the antibody including, for example, a reagent, precursor or WO 03/044036 PCT/US02/37134 18 other reaction product. An isolated agent can also include a substantially pure agent. The term can include naturally occurring molecules such as products of biosynthetic reactions or synthetic molecules. 5 The invention provides a human monoclonal antibody or functional fragment having at least one CDR with substantially the amino acid sequence of a CDR of SEQ ID NO:2 or SEQ ID NO:4. The invention also provides a human monoclonal antibody or functional fragment having 10 at least one CDR with substantially the amino acid sequence of a CDR of SEQ ID NO:6 or SEQ ID NO:8. The invention further provides human monoclonal antibodies or functional fragments thereof produced by the hybridoma cell lines H1140, H2420 and H935. The 15 hybridoma cell lines H1140, H2420 and H935 are also provided. The human monoclonal antibodies produced by the hybridoma cell lines LH11238, LH13, H1140, H2420 and H935 all exhibit specific binding to neoplastic cells as 20 compared to normal cells and, therefore, are tumor specific human monoclonal antibodies. In particular, the human monoclonal antibodies of the invention all selectively bind breast carcinoma cells and show relatively little binding to normal fibroblasts. For 25 example, the LH11238 antibody specifically binds to an antigen present on the surface and lysosomal compartments of breast and ovarian carcinoma cells, as compared to normal fibroblasts, peripheral blood lymphocytes, melanoma cells or lung carcinoma cells. The LH13 30 antibody specifically binds a product produced by breast, WO 03/044036 PCT/USO2/37134 19 lung and ovarian carcinoma cells, as compared to normal fibroblasts and melanoma cells. The human monoclonal antibodies produced by hybridoma lines H1140, H2420, H935 and LH13 are of the 5 IgM isotype and X light chain class, whereas human monoclonal antibodies produced by hybridoma line LH11238 is of the IgM isotype and K light chain class. Further properties of the tumor-specific human monoclonal antibodies are described below in the Examples. 10 The nucleotide sequence encoding the heavy chain variable region (VH) of the human monoclonal antibody produced by LH11238 cell line has been determined and is designated SEQ ID NO:1. The VH amino acid sequence of the human monoclonal antibody produced 15 by LH11238 cell line is designated SEQ ID NO:2. The nucleotide sequence encoding the light chain variable region (VL) of the human monoclonal antibody produced by LH11238 cell line has also been determined and is designated SEQ ID NO:3. The VL amino acid sequence of 20 the human monoclonal antibody produced by LH11238 cell line is designated SEQ ID NO:4. The nucleotide sequence encoding the heavy chain variable region (VH) of the human monoclonal 25 antibody produced by LH13 cell line has been determined and is designated SEQ ID NO:5. The VH amino acid sequence of the human monoclonal antibody produced by LH13 cell line is designated SEQ ID NO:6. The nucleotide sequence encoding the light chain variable region (VL) of 30 the human monoclonal antibody produced by LH13 cell line has also been determined and is designated SEQ ID NO:7.

WO 03/044036 PCT/USO2/37134 20 The VL amino acid sequence of the human monoclonal antibody produced by LH13 cell line is designated SEQ ID NO:8. The invention further provides an isolated 5 human monoclonal antibody or functional fragment thereof, having at least one CDR with substantially the amino acid sequence of a CDR of a sequence selected from the group consisting of SEQ ID NOs:10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44, wherein the 10 human monoclonal antibody or functional fragment thereof specifically binds a neoplastic cell or antigen thereof. Variants of the human monoclonal antibodies produced by the hybridoma cell lines of the invention, including, for example, the LH13 and LH11238 cell lines, 15 can exhibit specific binding to neoplastic cells as compared to normal cells, and are, therefore, functional tumor-specific human monoclonal antibodies. Functional variants of the human monoclonal antibodies produced by the hybridoma cell lines LH13 and LH11238 and methods for 20 producing them are described in Example VIII and Table 5. Functional variants of the human monoclonal antibody produced by the LH13 cell line that specifically bind to an antigen of a neoplastic cell are listed in Table 5 and include, for example, the antibody fragment 25 produced by clone S97G, S97T or S97N each of which has an unmodified VL and a VH modified at position 101 of the sequence set forth in SEQ ID NO:6, the modification being at position 97 of HCDR3 according to the numbering system of Kabat et al. The antibody fragment produced by clone 30 S97G has a glycine at position 97 of Kabat HCDR3 as set WO 03/044036 PCT/USO2/37134 21 forth in the VH amino acid sequence of SEQ ID NO:10 which is encoded by the nucleotide sequence set forth in SEQ ID NO:9. The antibody fragment produced by clone S97T has a threonine at position 97 of Kabat HCDR3 as set forth in 5 the VH amino acid sequence of SEQ ID NO:12 and is encoded by the nucleotide sequence set forth in SEQ ID NO:11. The antibody fragment produced by clone S97N has an HCDR3 with an asparagine at position 97 of Kabat HCDR3 as set forth in the VH amino acid sequence of SEQ ID NO:14 and 10 is encoded by the nucleotide sequence set forth in SEQ ID NO:13. Other functional variants of the human monoclonal antibody produced by the LH13 cell line also include the antibody fragment produced by clone R91Y or 15 R91F each of which has an unmodified VH and a VL modified at position 90 of the sequence set forth in SEQ ID NO:8, the modification being at position 91 of LCDR3 using the numbering system of Kabat et al. The antibody fragment produced by clone R91Y has a tyrosine at position 91 of 20 Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:16 which is encoded by the nucleotide sequence set forth in SEQ ID NO:15. The antibody fragment produced by clone R91F has an LCDR3 with a phenylalanine at position 91 of Kabat LCDR3 as set forth in the VL 25 amino acid sequence of SEQ ID NO:18 which is encoded by the nucleotide sequence set forth in SEQ ID NO:17. A functional variant of the human monoclonal antibody produced by the LH13 cell line can be modified at position 98 of the VL sequence set forth in SEQ ID 30 NO:8 while retaining an unmodified VH as exemplified by the antibody fragments produced by clones V97Y and V97F, WO 03/044036 PCT/US02/37134 22 the modification being at position 97 of LCDR3 using the numbering system of Kabat et al. The antibody fragment produced by clone V97Y has a tyrosine at position 97 of Kabat LCDR3 as set forth in the VL amino acid sequence of 5 SEQ ID NO:20 which is encoded by the nucleotide sequence set forth in SEQ ID NO:19. The antibody fragment produced by clone V97F has an LCDR3 with a tyrosine at position 97 of Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:22 which is encoded by the 10 nucleotide sequence set forth in SEQ ID NO:21. Combinatorial variants of the human monoclonal antibody produced by the LH13 cell line that specifically bind to an antigen of a neoplastic cell are also listed in Table 5 and include, for example, the antibody 15 fragment produced by clone 4D5, 4E2, 4H7, 4G11 or 3G4. The antibody fragment produced by clone 4D5 is modified to have a threonine at position 97 of Kabat HCDR3 and a tyrosine at position 91 of Kabat LCDR3 compared to the human monoclonal antibody produced by the LH13 cell line. 20 The amino acid sequence of the 4D5 VH is set forth in SEQ ID NO:12 and is encoded by the nucleotide sequence set forth in SEQ ID NO:11 and the VL amino acid sequence is set forth in SEQ ID NO:16 and encoded by the nucleotide sequence set forth in SEQ ID NO:15. 25 The combinatorial variant produced by clone 4E2 is modified to have a threonine at position 97 of Kabat HCDR3, a tyrosine at position 91 of Kabat LCDR3, and a phenylalanine at position 97 of Kabat LCDR3 compared to the human monoclonal antibody produced by the LH13 cell 30 line. The amino acid sequence of the 4E2 VH is set forth in SEQ ID NO:12 and is encoded by the nucleotide sequence WO03/044036 PCT/US02/37134 23 set forth in SEQ ID NO:11 and the VL amino acid sequence is set forth in SEQ ID NO:24 and encoded by the nucleotide sequence set forth in SEQ ID NO:23. The combinatorial variant produced by clone 4H7 5 is modified to have a threonine at position 97 of Kabat HCDR3, a phenylalanine at position 91 of Kabat LCDR3, and a phenylalanine at position 97 of Kabat LCDR3 compared to the human monoclonal antibody produced by the LH13 cell line. The amino acid sequence of the 4H7 VH is set forth 10 in SEQ ID NO:12 and is encoded by the nucleotide sequence set forth in SEQ ID NO:11 and the VL amino acid sequence is set forth in SEQ ID NO:26 and encoded by the nucleotide sequence set forth in SEQ ID NO:25. The combinatorial variant produced by clone 15 4G11 is modified to have a phenylalanine at position 91 of Kabat LCDR3, and a phenylalanine at position 97 of Kabat LCDR3 compared to the human monoclonal antibody produced by the LH13 cell line. The amino acid sequence of the 4G11 VH is set forth in SEQ ID NO:6 and is encoded 20 by the nucleotide sequence set forth in SEQ ID NO:5 and the VL amino acid sequence is set forth in SEQ ID NO:26 and encoded by the nucleotide sequence set forth in SEQ ID NO:25. The combinatorial variant produced by clone 3G4 25 is modified to have a tyrosine at position 91 of Kabat LCDR3, and a phenylalanine at position 97 of Kabat LCDR3 compared to the human monoclonal antibody produced by the LH13 cell line. The amino acid sequence of the 3G4 VH is set forth in SEQ ID NO:6 and is encoded by the nucleotide 30 sequence set forth in SEQ ID NO:5 and the VL amino acid WO 03/044036 PCT/USO2/37134 24 sequence is set forth in SEQ ID NO:24 and encoded by the nucleotide sequence set forth in SEQ ID NO:23. Functional variants of the human monoclonal antibody produced by the LH11238 cell line that bind to a 5 neoplastic cell are listed in Table 5 and include, for example, the antibody fragment produced by clone Q89L, Q89G, Q89V, Q89F or Q89W each of which has an unmodified VH and a VL modified at position 97 of the sequence set forth in SEQ ID NO:4, the modification being at position 10 89 of LCDR3 according to the numbering system of Kabat et al. The antibody fragment produced by clone Q89L has a luecine at position 89 of Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:28, which is encoded by the nucleotide sequence set forth in SEQ ID NO:27. 15 The antibody fragment produced by clone Q89G has an LCDR3 with a glycine at position 89 of Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:30, which is encoded by the nucleotide sequence set forth in SEQ ID NO:29. The antibody fragment produced by clone Q89V has 20 an LCDR3 with a valine at position 89 of Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:32, which is encoded by the nucleotide sequence set forth in SEQ ID NO:31. The antibody fragment produced by clone Q89F has an LCDR3 with a phenylalanine at position 89 of 25 Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:34, which is encoded by the nucleotide sequence set forth in SEQ ID NO:33. The antibody fragment produced by clone Q89W has a tryptophon at position 89 of Kabat LCDR3 as set forth in the VL amino acid sequence of 30 SEQ ID NO:36, which is encoded by the nucleotide sequence set forth in SEQ ID NO:35.

WO 03/044036 PCT/USO2/37134 25 Other functional variants of the human monoclonal antibody produced by the LH11238 cell line also include the antibody fragment produced by clone P95aF or P95aR each of which has an unmodified VH and a 5 VL modified at position 103 of the sequence set forth in SEQ ID NO:4, the modification being at position 95 of LCDR3 according to the numbering system of Kabat et al. The antibody fragment produced by clone P95aF has a phenylalanine at position 95 of Kabat LCDR3 as set forth 10 in the VL amino acid sequence of SEQ ID NO:42 which is encoded by the nucleotide sequence set forth in SEQ ID NO:41. The antibody fragment produced by clone P95aR has a phenylalanine at position 95 of Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:44 which 15 is encoded by the nucleotide sequence set forth in SEQ ID NO:43 or SEQ ID NO:45. A functional variant of the human monoclonal antibody produced by the LH11238 cell line can be modified within LCDR3 at position 100 of the VL sequence 20 set forth in SEQ ID NO:4 while retaining an unmodified VH as exemplified by the antibody fragment produced by clone N93C. The antibody fragment produced by clone N93C has a cysteine at position 93 of Kabat LCDR3 as set forth in the VL amino acid sequence of SEQ ID NO:38 and encoded by 25 the nucleotide sequence set forth in SEQ ID NO:37. As shown for the antibody fragment produced by clone N94C, a functional variant of the human monoclonal antibody produced by the LH11238 cell line can retain an unmodified VH and be modified to have a cysteine at 30 position 101 of the VL amino acid sequence set forth in SEQ ID NO:40 (position 94 of Kabat LCDR3) and encoded by the nucleotide sequence set forth in SEQ ID NO:39.

WO 03/044036 PCT/USO2/37134 26 The hybridomas producing the tumor-specific human monoclonal antibodies of the invention were generated by in vitro immunization of human spleen cell cultures with breast carcinoma cells. Briefly, a mixed 5 lymphocyte reaction (MLR) was established by co-culturing single cell suspensions isolated from allogeneic human spleens. Tumor-reactive lymphocytes were subsequently enriched by incubating MLR cultures either with monolayers or with enriched plasma membranes of breast 10 carcinoma cells. In order to provide a permanent source of human monoclonal antibodies of the invention, immunized lymphocytes were immortalized either by fusion with the K6H6/B5 heteromyeloma cell line or by transformation with EBV followed by fusion with K6H6/B5 15 cells. The particular source of antigen and the immortalization procedures used to generate each of the hybridoma cell lines of the invention are described more fully below in the Examples. The tumor-specific human monoclonal antibodies 20 of the invention can also be generated by methods known to those skilled in the art. These methods include, for example, in vivo and in vitro enrichment of tumor-reactive lymphocytes. For example, an individual with a breast, lung or ovarian tumor can possess 25 lymphocytes which express antibodies that specifically bind tumor-specific antigens, including, for example, LH13, LH11238, H1140, H2420 or H935 antigens. Such lymphocytes can be isolated, for example, from the peripheral blood or from the spleen of the patient, and 30 immortalized as described below.

WO 03/044036 PCT/USO2/37134 27 Methods are also well known in the art for in vitro enrichment of tumor-reactive human lymphocytes, using tumor-specific antigens. The source of antigen can be, for example, substantially pure antigen, tumor cells 5 or tumor cell fractions. A substantially pure antigen can be prepared by any of the methods well known to those skilled in the art, including, for example, chromatography, electrophoretic separation and immuno-isolation. 10 An antigen useful for preparing human monoclonal antibodies of the invention can be, for example, neoplastic cells. The neoplastic cells can be, for example, cells obtained directly from tumors, cultured primary tumor cells or established cell lines. 15 Such cells can originate from any organ, tissue or fluid of the body, including, for example, breast, lung or ovary. The cancer cells can be untreated, fixed or growth-arrested. The fixation can be by any number of methods known to those skilled in the art, including, for 20 example, chemical fixation. Useful chemicals for fixation include, for example, paraformaldehyde, glutaraldehyde, methanol, or acetone. The cells can alternatively be growth-arrested using cytostatic agents. A specific example of a cytostatic agent is mitomycin C. 25 An antigen useful for preparing human monoclonal antibodies of the invention can can also be a fraction of tumor cells. The tumor cell fraction can be, for example, cellular membranes, cytoplasmic contents, or nuclei. Methods for cell fractionation are well known in 30 the art. An antigen for preparing monoclonal antibodies of the invention can also be an antigen secreted by tumor WO 03/044036 PCT/USO2/37134 28 cells. Such an antigen can be prepared, for example, by isolation of conditioned medium or cell matrix of tumor cells using procedures known in the art. An antigen prepared as described above can be 5 used to stimulate human lymphocytes to generate the tumor-specific human monoclonal antibodies of the invention. Human lymphocytes can be obtained, for example, from the peripheral blood of live individuals, or from the spleen of individuals who are deceased or 10 undergoing surgery. Lymphocytes can be cultured with antigen directly. Alternatively, lymphocytes can be cultured with antigen in a mixed lymphocyte reaction in the presence of allogeneic lymphocytes. Appropriate culturing conditions for a particular antigen and 15 lymphocyte source can readily be determined by those skilled in the art. If desired, antigen-primed lymphocytes enriched by in vivo or in vitro stimulation as described above can be immortalized by any of a number of procedures known to 20 those skilled in the art. Immortalization provides a permanent source of tumor-specific human monoclonal antibodies. Immortalization of lymphocytes can be accomplished by, for example, fusion with an immortal cell line. Such immortal cell lines useful for cell 25 fusion can be, for example, human myeloma cells or human lymphoblastoid B cell lines. Fusion partners can also be rodent myeloma cells or human:rodent heteromyeloma cell lines. The heteromyeloma cell line can be, for example, the human:mouse heterohybridoma cell line K6H6/B5. 30 Alternatively, antigen-primed lymphocytes can be immortalized by viral transformation, using, for example, WO 03/044036 PCT/USO2/37134 29 viruses. A useful virus for immortalization by viral transformation is EBV. Antigen-primed lymphocytes can also be immortalized by viral transformation followed by fusion. Viral transformation followed by fusion can be, 5 for example, EBV transformation followed by fusion with the K6H6/B5 cell line. Culture conditions for lymphocyte fusion or viral transformation can readily be determined by those skilled in the art. Immortalized lymphocytes can be screened for 10 the production of human monoclonal antibodies that specifically bind to human tumor cells using immunoassays known to those skilled in the art. Such immunoassays include, for example, quantitative and qualitative immunoassays. Qualitative immunoassays include, for 15 example, precipitin methods, agglutinin reactions, immunohistochemistry, immunofluorescence, immunoblotting and immunoprecipitation. Quantitative immunoassays include, for example, immunoaffinity methods such as radioimmunoassay, FACS analysis, and ELISA analysis. 20 ELISA analysis can be direct, sandwich or competitive. Immunoassays can be direct, using, for example, a labeled human monoclonal antibody. Such methods can alternatively be indirect, using, for example, a labeled anti-human secondary antibody. The label can be, for 25 example, a fluorescent label, an enzyme, a radioisotope, or biotin. Detection can be by spectrophotometric, radiographic or chemiluminescent means, depending on the immunoassay. Such methods can also be used to screen for an improved antibody or functional fragment thereof 30 having increased affinity for a neoplastic cell or antigen thereof.

WO 03/044036 PCT/US02/37134 30 An improved antibody or functional fragment thereof can also be identified in a screen for association rate with a neoplastic cell or antigen thereof. Specifically, an improved antibody or 5 functional fragment thereof can be identified as one that has an increase in association rate compared to the parent antibody from which it is derived. Using such a screen an antibody or functional fragment thereof having improved therapeutic potency due to increased association 10 rate can be distinguished from a binding polypeptide that has an increased Ka for the same antigen due to a decreased dissociation rate which is not correlated with therapeutic potency. In this regard, those skilled in the art will recognize that according to the relationship 15 KA = ko/koff an increased KA can be due to increased k.n or decreased kf or both. An association rate can be determined in any non-equilibrium mixture including, for example, one formed by rapidly contacting an antibody or functional 20 fragment thereof with an antigen or by rapidly changing temperature of a solution containing the binding partners. A non-equilibrium mixture can be a pre equilibrium mixture. A pre-equilibrium mixture can be formed, for example, by contacting a soluble antibody or 25 functional fragment thereof and soluble antigen in a condition where the amount of total antigen and total antibody or functional fragment thereof in the detection chamber are constant. Measurements of association rates in pre-equilibrium mixtures can be made in formats 30 providing rapid mixing of antibody or functional fragment thereof with antigen and rapid detection of changing properties of the antibody or functional fragment thereof WO 03/044036 PCT/USO2/37134 31 or antigen on a timescale of milliseconds or faster. Stopped flow and rapid quench flow instruments such as those described below provide a convenient means to measure non-equilibrium kinetics. The association rate 5 can also be measured in non-equilibrium mixtures including, for example, solutions containing insoluble species of antibody or functional fragment thereof or solutions containing variable concentrations of total antigen or total antibody or functional fragment thereof. 10 Measurement of an association rate in a non-equilibrium mixture can be made in formats providing attachment of an antigen to a surface and continuous flow of a solution containing the antibody or functional fragment thereof over the surface, or vice-versa, combined with rapid 15 detection of changing properties of the antibody or functional fragment thereof, antigen or surface such that measurements are made on a timescale of milliseconds or faster. Formats for measuring association rates in pre 20 equilibrium mixtures include, for example, stopped flow kinetic instruments and rapid quench flow instruments. A stopped flow instrument can be used to push solutions containing an antibody or functional fragment thereof and antigen from separate reservoirs into a mixing chamber 25 just prior to passage into a detection cell. The instrument can then detect a change in one or more of the above described properties to monitor progress of the binding event. A rapid quench flow instrument can be used to rapidly mix a solution containing an antibody or 30 functional fragment thereof with a solution containing an antigen followed by quenching the binding reaction after a finite amount of time. A change in one or more of the WO 03/044036 PCT/USO2/37134 32 above described properties can then be detected for quenched mixtures produced by quenching at different times following mixing. Quenching can be performed for example by freezing or addition of a chemical quenching 5 agent so long as the quenching step does not inhibit detection of the property relied upon for measurement of binding rate. Thus, a rapid quench instrument can be useful, for example, in situations where spectroscopic detection is not convenient. A variety of instruments 10 are commercially available from vendors such as KinTek Corp. (State College, PA) and Hi-Tech Scientific (Salisbury, UK). Formats for measuring association rates in non equilibrium mixtures include, for example, surface 15 plasmon resonance and evanescent wave instruments. Surface plasmon resonance and evanescent wave technology utilize an antigen or an antibody or functional fragment thereof attached to a biosensor surface. A solution containing the respective binding partner is passed over 20 the biosensor surface and the change in refractive index of the solution that occurs at the surface of a chip upon binding can be measured in a time dependent fashion. For example, surface plasmon resonance is based on the phenomenon which occurs when surface plasmon waves are 25 excited at a metal/liquid interface. Light is directed at, and reflected from, the side of the surface not in contact with sample, and SPR causes a reduction in the reflected light intensity at a specific combination of angle and wavelength. Biomolecular binding events cause 30 changes in the refractive index at the surface layer, which are detected as changes in the SPR signal. The binding event can be either binding association or WO 03/044036 PCT/USO2/37134 33 disassociation between a receptor-ligand pair. The changes in refractive index can be measured essentially instantaneously and therefore allows for determination of the individual components of an affinity constant. More 5 specifically, the method enables accurate measurements of an association rate constant (kon) and disassociation rate constant (koff). Surface plasmon resonance instruments are available in the art including, for example, the BIAcore instrument, IBIS system, SPR-CELLIA system, 10 Spreeta, and Plasmon SPR and evanescent wave technology is available in the Iasys system as described, for example, in Rich and Myszka, Curr. Opin. Biotech. 11:54 61 (2000). The association rate can be determined by 15 measuring a change in a property of a ligand or binding polypeptide at one or more discreet time intervals during the binding event using, for example, the methods described above. Measurements determined at discreet time intervals during the binding event can be used to 20 determine a quantitative measure of association rate or a relative measure of association rate. Quantitative measures of association rate can include, for example, an association rate value or k.n value. Quantitative values of association rate or kon can be determined from a 25 mathematical or graphical analysis of a time dependent measurement. Such analyses are well known in the art and include algorithms for fitting data to a sum of exponential or linear terms or algorithms for computer simulation to fit data to a binding model as described 30 for example in Johnson, Cur. Opin. Biotech. 9:87-89 (1998).

WO 03/044036 PCT/US02/37134 34 Association rates can be determined from mixtures containing insoluble species or variable concentrations of total ligand or total binding polypeptide using mathematical and graphical analyses 5 such as those described above if effects of mass transport are accounted for in the reaction. One skilled in the art can account for mass transport by comparing association rates under conditions having similar limitations with respect to mass transport or by 10 adjusting the calculated association rate according to models available in the art including, for example those described in Myszka et al., Biophys. J. 75:583-594 (1998). The tumor-specific antigen sample used for 15 screening monoclonal antibodies for tumor reactivity need not be the same as the antigen sample used to immunize the human lymphocytes. The antigen used in screening can be, for example, substantially purified antigen, live or fixed tumor cell monolayers, live or fixed tumor cell 20 suspensions, fractions of tumor cells, or sections of tumor biopsies, depending on the assay procedure employed. The tumor cells can be, for example, human breast, ovarian or lung carcinoma cells. The tumor-specific human monoclonal antibodies 25 and functional fragments of the invention do not bind, bind only minimally, or can be made to bind only minimally to normal cell antigens. Normal cells or fractions of normal cells can be used as controls for screening human monoclonal antibodies. Such normal cells 30 can be, for example, live or fixed normal tissues or cultured cell lines. Cultured normal cell lines that can WO 03/044036 PCT/USO2/37134 35 be used as controls include, for example, human fibroblasts and peripheral blood lymphocytes. Although any normal cell can be used as a control, the selection of a particular control will be based, in part, on the 5 specificity of the particular tumor-specific monoclonal antibody which is being screened. For example, if a human tumor-specific monoclonal antibody is produced and screened against a carcinoma cell, then one type of normal cell which can be used as a comparison is a normal 10 epithelial cell culture or cell line. Similarly, if a human tumor-specific monoclonal antibody is produced and screened against a sarcoma cell, then one type of normal cell which can be used as a comparison is a normal fibroblast cell culture or cell line. A normal cell 15 culture or cell line from the same tissue type or from the same individual can also provide a normal control. Those skilled in the art will know what is an appropriate type of normal cell to be used as a control to determine specific binding to a particular type of tumor cell. 20 Tumor-specific human monoclonal antibodies or functional fragments thereof can be purified and quantitated for use in immunodiagnostic and immunotherapeutic procedures. Such purification methods 25 are well known to those skilled in the art and depend on the source of human monoclonal antibodies and the particular application. Purification methods can include, for example, precipitation, electophoresis, chromatography, and immunoaffinity purification. The 30 purified antibodies can be quantitated in comparison with known standard controls, using, for example, spectrophotometry or immunoassays known in the art.

WO 03/044036 PCT/USO2/37134 36 To further characterize tumor-specific human monoclonal antibodies, their class and subclass can also be determined by immunoassays that measure the presence of individual heavy and light chain polypeptides. Such 5 immunoassays include ELISA assays and are known to those skilled in the art. The invention further provides functional fragments of the tumor-specific human monoclonal antibodies of the invention. A functional fragment of a 10 human monoclonal antibody maintains a biological activity, such as specific binding or an effector function. A functional fragment can therefore be beneficially used for the detection and treatment of cancer. 15 Functional fragments include fragments with substantially the same heavy and light chain variable regions as a tumor-specific human monoclonal antibody of the invention. For example, functional fragments include fragments wherein at least one of the CDR sequences 20 consist of substantially the same amino acid sequence as the CDR sequences of a tumor-specific human monoclonal antibody of the invention. Such functional fragments include, for example, VL, Fd, Fv, Fab, Fab', F(ab') 2 , Fc and CDR fragments. For example, a functional fragment 25 could have one or more of the three CDR sequences of the VL, or one or more of the three CDR sequences of the VH, or a combination of VL and VH CDRs of a human monoclonal antibody of the invention. The appropriate number and combination of VH and VL CDR sequences can be determined 30 by those skilled in the art depending on the desired WO 03/044036 PCT/US02/37134 37 affinity and specificity and the intended use of the functional fragment. Functional fragments of human monoclonal antibodies of the invention can readily be produced and 5 isolated using methods well known to those skilled in the art. Such methods include, for example, proteolytic methods, recombinant methods and chemical synthesis. Proteolytic methods for the isolation of functional fragments use human monoclonal antibodies as starting 10 material. Enzymes suitable for proteolysis of human monoclonal antibodies include, for example, papain, pepsin and elastin. The appropriate enzyme can be readily chosen by one skilled in the art, depending on, for example, whether monovalent or bivalent fragments are 15 required. For example, papain cleavage results in two monovalent Fab' fragments that bind antigen and an Fc fragment. Pepsin cleavage, for example, results in a bivalent F(ab') 2 fragment. A F(ab') 2 fragment of the invention can further be reduced using, for example, DTT 20 or S-mercaptoethanol to produce two monovalent Fab' fragments. Functional fragments produced by proteolysis can be purified by affinity and column chromatographic procedures. For example, undigested antibodies and Fc 25 fragments can often be removed by binding to protein A. Additionally, functional fragments can be purified by virtue of their charge and size, using, for example, ion-exchange and gel filtration chromatography. Such methods are well known to those skilled in the art.

WO 03/044036 PCT/USO2/37134 38 Recombinant methods for producing functional fragments of human monoclonal antibodies begin with the isolated nucleic acid of desired regions of the immunoglobulin heavy and light chains. Such regions can 5 include, for example, all or part of the variable region of the heavy and light chains. Such regions can particularly include the CDRs of the heavy and light chains. The invention provides an isolated nucleic acid 10 encoding a human monoclonal antibody or functional fragment thereof, having a nucleotide sequence encoding substantially the amino acid sequence of at least one CDR encoded by a nucleotide sequence selected from the group consisting of SEQ ID NOs:l, 3, 5, 7, 9, 11, 13, 15, 17, 15 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43 and 45. The invention further provides an isolated nucleic acid encoding a variable region domain, having a nucleotide sequence encoding substantially the amino acid sequence of the CDRs encoded by a nucleotide sequence 20 selected from the group consisting of SEQ ID NOs:l, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43 and 45. Also provided is an isolated nucleic acid encoding a CDR, having substantially the amino acid sequence of a CDR encoding by SEQ ID NOs:l, 3, 25 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43 and 45. A nucleic acid encoding a human monoclonal antibody or functional fragment of the invention can be produced using methods known to those skilled in the art. 30 One useful procedure for isolating such DNA begins with cDNA which can be reverse-transcribed from RNA of WO 03/044036 PCT/US02/37134 39 hybridoma cells that produce a tumor-specific human monoclonal antibody. Methods for cDNA synthesis are well known in the art. A cDNA encoding a functional fragment of a heavy or light chain can be amplified using, for 5 example, the polymerase chain reaction (PCR). The PCR technology is the subject matter of United States Patent Nos. 4,683,195, 4,800,159, 4,754,065, and 4,683,202, all of which are incorporated by reference herein. Suitable primers for PCR are known and can be determined by those 10 of skill in the art using conserved sequences which flank the particular functional fragment of a heavy or light chain. Suitable PCR conditions are known and can also similarly be determined by those skilled in the art. A nucleic acid encoding a functional fragment 15 of a human monoclonal antibody of the invention can also be directly synthesized by methods of oligonucleotide synthesis known in the art. Alternatively, smaller fragments can be synthesized and joined to form a larger functional fragment using recombinant methods known in 20 the art. Nucleic acids encoding a functional fragment of a human monoclonal antibody of the invention can be cloned into a suitable expression vector and expressed in 25 a suitable host. A suitable vector and host cell system can allow, for example, co-expression and assembly of functional fragments of the heavy and light chains. Suitable systems for the expression of antibody fragments can be determined by those skilled in the art and 30 include, for example, M13 phage immunoexpression vectors. Recombinant functional fragments of the invention can be substantially purified using methods known in the art, WO 03/044036 PCT/USO2/37134 40 and which depend on the particular vector and host expression system used. Isolated nucleic acids encoding tumor-specific human monoclonal antibodies or functional fragments can 5 also be engineered to produce antibodies with optimal properties such as affinity, selectivity, avidity, stability or bioavailability. Such modifications can include, for example, addition, deletion, or substitution of amino acid residues or substitution of a D-amino acid 10 or amino acid mimetic, so long as the fragment maintains a functional activity such as, for example, antigen binding specificity. In addition, the invention provides distinct libraries of LH13 and LH11238 variants. An LH13 library 15 can contain variants having at least one amino acid alteration in a Kabat light chain CDR3 (LCDR3) corresponding to positions 88-98 of SEQ ID NO:8, variants having at least one amino acid alteration in a Kabat heavy chain CDR3 (HCDR3) corresponding to positions 20 99-107 of SEQ ID NO:6. An LH11238 library can contain variants having at least one amino acid alteration in a Kabat LCDR3 corresponding to positions 97-106 of SEQ ID NO:4. Although the libraries exemplified above are based on the numbering system of Kabat et al., those 25 skilled in the art will be able to produce similar libraries based on any of the CDR definition including for example those of Chothia et al. or MacCallum et al using the numbering systems set forth previously. A library of the invention can contain variants having at 30 least one amino acid alteration in a region corresponding one or more of the CDRs of LH13 or LH11238 including WO 03/044036 PCT/USO2/37134 41 HCDR1, HCDR2, HCDR3, LCDR1, LCDR2 or LCDR3 according to the numbering systems set forth previously. A library of LH13 or LH11238 variants can be used to screen for antibodies or functional fragments having increased 5 antigen or neoplastic cell binding activity as set forth in further detail below. A library that is sufficiently diverse to contain an antibody variant or fragment thereof with enhanced binding affinity can be prepared by a variety of 10 methods well known in the art. Those skilled in the art will know what size and diversity of the library is necessary or sufficient for the intended purpose. For example, a library of antibody or antibody fragment variants can be prepared that contains each of the 19 15 essential amino acids not found in the parent antibody at each of the positions in one or more CDR and the resultant population screened for variants with enhanced binding activity. Alternatively, a focused library can be 20 prepared utilizing the structural, biochemical and modeling information relating to antibodies as described herein. It is understood that any information relevant to the determination or prediction of residues or regions important for the binding activity or structural 25 integrity or stability of an antibody of the invention can be useful in the design of a focused library. Thus, the antibody variants that make up the library can contain amino acid alterations at one or more amino acid positions located in a region or at a particular position 30 determined or predicted to be important for binding activity or structural integrity or stability of an WO 03/044036 PCT/USO2/37134 42 antibody that binds a neoplastic cell or antigen thereof. A focused library of antibody variants provides the advantage of decreasing the number of variants that need to be screened in order to identify a variant with 5 enhanced binding activity or structural stability. Combinatorial approaches to synthesizing and screening focused libraries can further provide advantages in the speed and efficiency of identifying improved antibody variants or fragments thereof. As 10 demonstrated in Example VIII, multiple amino acid changes in the LH13 variants were additive. The additivity of amino acid changes in other unrelated monoclonal antibodies has been reported previously as described, for example, in Yelton et al., J. Immunol. 155:1994-2004 15 (1995), Wu et al., supra (1998), and Wu et al., supra (1999). The identification of additive amino acid combinations by synthesis of a combinatorial library based on single amino acid changes discovered from HCDR3 and LCDR3 libraries permitted rapid enhancement of the 20 antibody affinity in a very efficient manner. As demonstrated in Example VIII, the two step affinity maturation of LH13 was accomplished with the synthesis of only 416 distinct protein variants consisting of 171 HCDR3 mutants and 209 LCDR3 mutants in 25 a first step followed by 36 combinatorial variants in step two. In contrast, total randomization of the three CDR residues demonstrated to influence activity, LCDR3 S97, HCDR3 R91, and HCDR3 V97, would have required the expression of a library containing 193, or 6,859, 30 variants. Thus, the combinatorial approach described in Example VIII provided stepwise improvement in affinity WO 03/044036 PCT/USO2/37134 43 that captured additivity of independent mutations is an efficient method of simplifying the affinity maturation process. Those skilled in the art will recognize from the results of Example VIII that the affinity of an 5 antibody of the invention such as an enhanced class-switched variant can be further improved through the synthesis of additional LH13 libraries corresponding to HCDRI and HCDR2 and LCDRI and LCDR2. Moreover, the beneficial mutations identified from these libraries can 10 be used in a combinatorial fashion to identify improved antibody or antibody fragment variants having increased affinity for a neoplastic cell or antigen thereof. Individual residues or regions important for binding of an antibody of the invention to a neoplastic 15 cell or antigen thereof can be determined or predicted through a variety of methods known in the art and can be used to focus the synthesis of a library of antibody or antibody fragment variants. Structural comparison of sequences (primary structures) or three dimensional 20 structures (tertiary structures) or a combination of both can be made between antibodies that bind the same antigen or similar antigens to identify individual residues or regions for mutation. For example, structural modeling based on antibody tertiary structure can reveal the 25 topological, electrostatic, hydrophobic or hydrophilic environment of a binding site or particular CDR involved in antigen binding. Comparison of shared properties of binding sites for two or more antibodies that bind a common antigen or structurally similar antigen can be 30 used to identify positions or regions for mutagenesis. Sequence comparison methods can be used to align sequences and identify positions in two or more WO 03/044036 PCT/USO2/37134 44 antibodies having residues that are homologous or share similar steric, electrostatic, hydrophobic or hydrophilic properties. A combination of primary and tertiary structure comparisons can be used to identify positions 5 to be altered. For example, positions located apart from each other in primary structure which are important for binding can be identified in a tertiary structure analysis and then emphasized in a primary sequence for use in querying a database of antibody primary sequences 10 or comparing to one or more other antibody primary sequence. Methods for comparing primary sequence structure which can be used to determine that two sequences are substantially the same are well known in 15 the art. For example, one method for determining if two sequences are substantially the same is BLAST, Basic Local Alignment Search Tool, which can be used according to default parameters as described by Tatiana et al., FEMS Microbial Lett. 174:247-250 (1999) or on the 20 National Center for Biotechnology Information web page at ncbi.nlm.gov/BLAST/. BLAST is a set of similarity search programs designed to examine all available sequence databases and can function to search for similarities in amino acid or nucleic acid sequences. A BLAST search 25 provides search scores that have a well-defined statistical interpretation. Furthermore, BLAST uses a heuristic algorithm that seeks local alignments and is therefore able to detect relationships among sequences which share only isolated regions of similarity 30 including, for example, protein domains (Altschul et al., J. Mol. Biol. 215:403-410 (1990)).

WO 03/044036 PCT/USO2/37134 45 In addition to the originally described BLAST (Altschul et al., supra, 1990), modifications to the algorithm have been made (Altschul et al., Nucleic Acids Res. 25:3389-3402 (1997)). One modification is Gapped 5 BLAST, which allows gaps, either insertions or deletions, to be introduced into alignments. Allowing gaps in alignments tends to reflect biologic relationships more closely. For example, gapped BLAST can be used to identify sequence identity within similar domains of two 10 or more polypeptides. A second modification is PSI BLAST, which is a sensitive way to search for sequence homologs. PSI-BLAST performs an initial Gapped BLAST search and uses information from any significant alignments to construct a position-specific score matrix, 15 which replaces the query sequence for the next round of database searching. A PSI-BLAST search is often more sensitive to weak but biologically relevant sequence similarities. A second resource that can be used to determine 20 if two sequences are substantially the same is PROSITE, available on the world wide web at ExPASy. PROSITE is a method of determining the function of uncharacterized polypeptides translated from genomic or cDNA sequences (Bairoch et al., Nucleic Acids Res. 25:217-221 (1997)). 25 PROSITE consists of a database of biologically significant sites and patterns that can be used to identify which known family of polypeptides, if any, the new sequence belongs. Using this or a similar algorithm, a polypeptide that is substantially the same as another 30 polypeptide can be identified by the occurrence in its sequence of a particular cluster of amino acid residues, which can be called a pattern, motif, signature or WO 03/044036 PCT/USO2/37134 46 fingerprint, that is substantially the same as a particular cluster of amino acid residues in a reference polypeptide including, for example, those found in similar domains. PROSITE uses a computer algorithm to 5 search for motifs that identify polypeptides as family members. PROSITE also maintains a compilation of previously identified motifs, which can be used to determine if a newly identified polypeptide is a member of a known family. 10 Sequence comparison can include a full sequence of a gene, cDNA or expressed products thereof or can include one or more particular regions thereof. A particular region can be identified by visual inspection of a sequence alignment to identify regions of relatively 15 high homology or similarity. Those regions can be cross referenced with structural data to find correlations between a particular structural domain and region of homology, as described above. A structural model of a reference antibody or fragment thereof can also be used 20 in an algorithm that compares polypeptide structure including, for example, SCOP, CATH, or FSSP which are reviewed in Hadley and Jones, Structure 7:1099-1112 (1999) and regions having a particular fold or conformation used as a region for sequence comparison to 25 a second polypeptide to identify substantially similar regions. In addition to structural modeling of antibodies, biochemical data can be used to determine or predict positions or regions of an antibody that are 30 important for binding a neoplastic cell or antigen thereof and these positions or regions targeted in WO 03/044036 PCT/USO2/37134 47 preparing a focused library of antibody variants. In this regard, the characterization of a parent antibody or variants thereof with respect to binding rate constant (kon), dissociation rate constant (koff) or equilibrium 5 binding affinity constant (Kd or K) is useful for identifying regions important for binding to a neoplastic cell or antigen thereof. To generate a library of antibody variants distinct types of information can be used alone or 10 combined to determine or predict a region of an amino acid sequence that is important for binding to a neoplastic cell or antigen thereof. For example, information based on structural modeling and biochemical analysis, such as comparison of affinities for variants 15 having changes at a structurally conserved position, can be combined to determine a region of an amino acid sequence of an antibody important for binding activity. Because information obtained by a variety of methods can be combined to predict regions important for binding, one 20 skilled in the art will appreciate that the regions themselves represent approximations rather than strict confines. As a result, a library of antibody variants can be altered at positions outside of the regions determined or predicted to directly interact with an 25 antigen. For example, framework regions can be important for structural stability of an antibody or fragment thereof or can influence the binding affinity by long range or through space interactions that influence binding site properties. Similarly, a variant of an 30 antibody or fragment thereof that binds a neoplastic cell or antigen thereof can have amino acid alterations WO 03/044036 PCT/USO2/37134 48 outside of the CDRs or other regions identified herein as directly involved in binding. It is further understood that the number or location of amino acid positions predicted to be 5 important for binding activity can vary based on the predictive methods used and the structures compared. For example, as described above, different CDR definitions can be used to compare antibody sequences. Those skilled in the art will understand that the same CDR definition 10 can be used to identify a region or position to be compared between two or more antibody structures. For example, two or more antibodies can be compared with respect to the Kabat et al. definition of CDR3 as described in Example VIII and shown in Table V. Those 15 skilled in the art can determine an appropriate CDR definition for use in evaluating an antibody based on the primary or tertiary structure of the antibody. Similarly, a CDR definition to be used for comparing two or more antibody structures can be determined based on 20 initial inspection of each antibody structure and identification of the definition which best fits all structures to be compared. Other factors can also be considered in choosing an appropriate CDR definition including, for example, functional properties of a parent 25 antibody or variants thereof. Those skilled in the art can perform separate comparisons of the same antibodies using different CDR definitions and thereby identify regions or positions to be altered according to the frequency with which the regions or positions are found 30 to be similar or homologous in the separate comparisons.

WO 03/044036 PCT/USO2/37134 49 Methods for preparing libraries containing diverse populations of various types of molecules such as peptides, peptoids and peptidomimetics are well known in the art (see, for example, Ecker and Crooke, 5 Biotechnology 13:351-360 (1995), and Blondelle et al., Trends Anal. Chem. 14:83-92 (1995), and the references cited therein; see, also, Goodman and Ro, Peptidomimetics for Drug Design, in "Burger's Medicinal Chemistry and Drug Discovery" Vol. 1 (ed. M.E. Wolff; John Wiley & Sons 10 1995), pages 803-861, and Gordon et al., J. Med. Chem. 37:1385-1401 (1994)). Where a molecule is a peptide, protein or fragment thereof, the molecule can be produced in vitro directly or can be expressed from a nucleic acid, which can be produced in vitro. Methods of 15 synthetic peptide chemistry are well known in the art. A library of antibody variants can be produced, for example, by constructing a nucleic acid expression library encoding antibody variants. Methods for producing such libraries are well known in the art (see, 20 for example, Sambrook et al., Molecular Cloning: A Laboratory Manual, 2nd ed., Cold Spring Harbor Press, Plainview, New York (1989); Sambrook et al., Molecular Cloning: A Laboratory Manual, 3rd ed., Cold Spring Harbor Press, Plainview, New York (2001); Ausubel et al. 25 (Current Protocols in Molecular Biology (Supplement 47), John Wiley & Sons, New York (1999)). A library of nucleic acids encoding antibody variants can be composed of DNA, RNA or analogs thereof. A library containing RNA molecules can be constructed, for example, by 30 synthesizing the RNA molecules chemically.

WO 03/044036 PCT/USO2/37134 50 The generation of a library of nucleic acids encoding antibody or antibody fragment variants can be by any means desired by the user. Those skilled in the art will know what methods can be used to generate libraries 5 of nucleic acids encoding antibody or antibody fragment variants. For example, libraries of antibody variants can be generated by mutagenesis of a nucleic acid encoding a parent antibody such as LH13 or LH11238 using methods well known to those skilled in the art (Sambrook 10 et al., supra (1989); Sambrook et al., supra (2001); Ausubel et al., supra (1999)). A library of nucleic acids encoding antibody or antibody fragment variants of the invention can be randomized to be sufficiently diverse to contain nucleic acids encoding every possible 15 naturally occurring amino acid at each amino acid position of one or more CDR. Alternatively, a library of nucleic acids can be prepared such that it contains nucleic acids encoding every possible naturally occuring amino acid at each amino acid only at positions located 20 within a region of a CDR predicted or determined to be important for binding to a neoplastic cell or antigen thereof, as described in Example VIII. One or more mutations can be introduced into a nucleic acid molecule encoding an antibody or antibody 25 fragment variant to yield a modified nucleic acid molecule using, for example, site-directed mutagenesis (see Wu (Ed.), Meth. In Enzymol. Vol. 217, San Diego: Academic Press (1993); Higuchi, "Recombinant PCR" in Innis et al. (Ed.), PCR Protocols, San Diego: Academic 30 Press, Inc. (1990), each of which is incorporated herein by reference). Such mutagenesis can be used to introduce a specific, desired amino acid alteration. Thus, WO 03/044036 PCT/USO2/37134 51 distinct libraries containing amino acid alterations in one or more of the regions determined to be important for binding activity as well as a single library containing mutations in several or all of the regions can be 5 prepared. The efficient synthesis and expression of libraries of antibody or antibody fragment variants using oligonucleotide-directed mutagenesis can be accomplished as previously described by Wu et al., Proc. Natl. Acad. 10 Sci. USA, 95:6037-6042 (1998); Wu et al., J. Mol. Biol., 294:151-162 (1999); and Kunkel, Proc. Natl. Acad. Sci. USA, 82:488-492 (1985). Oligonucleotide-directed mutagenesis is a well-known and efficient procedure for systematically introducing mutations, independent of 15 their phenotype and is, therefore, ideally suited for directed evolution approaches to protein engineering. To perform oligonucleotide-directed mutagenesis a library of nucleic acids encoding the desired mutations is hybridized to single-stranded uracil-containing template 20 of the wild-type sequence. The methodology is flexible, permitting precise mutations to be introduced without the use of restriction enzymes, and is relatively inexpensive if oligonucleotides are synthesized using codon-based mutagenesis. 25 Amino acid substitutions can be introduced by mutating nucleic acid codons encoding the particular amino acid using methods known in the art. Single or multiple codons can be varied, so long as the fragment retains a functional activity. Rapid methods for making 30 and screening multiple simultaneous changes are well known within the art and can be used to produce a library WO 03/044036 PCT/US02/37134 52 of encoding nucleic acids which contain all possible or all desired changes and then expressing and screening the library for human monoclonal antibodies or fragments which retain function. Such methods include, for 5 example, codon based mutagenesis, synthesis of stochastic oligonucleotides and partially degenerate oligonucleotide synthesis. Codon based mutagenesis is the subject matter of U.S. Patent Nos. 5,264,563 and 5,523,388 and is 10 advantageous for the above procedures since it allows for the production of essentially any and all desired frequencies of encoded amino acid residues at any and all particular codon positions within an oligonucleotide. Such desired frequencies include, for example, the truly 15 random incorporation of all twenty amino acids or a specified subset thereof as well as the incorporation of a predetermined bias of one or more particular amino acids so as to incorporate a higher or lower frequency of the biased residues compared to other incorporated amino 20 acid residues. Synthesis of stochastic oligonucleotides and partially degenerate oligonucleotide synthesis can similarly be used for producing and screening for functionally equivalent amino acid changes. However, due 25 to the degeneracy of the genetic code, such methods can incorporate redundancies at a desired amino acid position (see, for example, U.S. Patent No. 5,723,323). Stochastic oligonucleotide synthesis includes the coupling of all four nucleotides at each nucleotide 30 position within a codon. Other stochastic methods of synthesis also exist which can result in degenerate or WO 03/044036 PCT/USO2/37134 53 partially degenerate oligonucleotides or oligonucleotides which encode completely random amino acid sequences (see, for example, U.S. Patent No. 5,723,323). Partially degenerate oligonucleotide synthesis 5 is the coupling of equal portions of all four nucleotides at the first two nucleotide positions, for example, and equal portions of two nucleotides at the third position. Both of these latter synthesis methods can be found described in, for example, Cwirla et al., Proc. Natl. 10 Acad. Sci. USA 87:6378-6382, (1990) and Devlin et al., Science 249:404-406, (1990). A modification to the above-described codon-based synthesis for producing a diverse number of variant sequences can similarly be employed for the 15 production of the libraries of antibody or antibody fragment variants described herein. This modification is based on the two vessel method described above which biases synthesis toward the parent sequence and allows the user to separate the variants into populations 20 containing a specified number of codon positions that have random codon changes. Briefly, this synthesis is performed by continuing to divide the reaction vessels after the synthesis of each codon position into two new vessels. 25 After the division, the reaction products from each consecutive pair of reaction vessels, starting with the second vessel, is mixed. This mixing brings together the reaction products having the same number of codon positions with random changes. Synthesis proceeds by 30 then dividing the products of the first and last vessel WO 03/044036 PCT/US02/37134 54 and the newly mixed products from each consecutive pair of reaction vessels and redividing into two new vessels. In one of the new vessels, the parent codon is synthesized and in the second vessel, the random codon is 5 synthesized. For example, synthesis at the first codon position entails synthesis of the parent codon in one reaction vessel and synthesis of a random codon in the second reaction vessel. For synthesis at the second codon position, each of the first two reaction vessels is 10 divided into two vessels yielding two pairs of vessels. For each pair, a parent codon is synthesized in one of the vessels and a random codon is synthesized in the second vessel. When arranged linearly, the reaction products in the second and third vessels are mixed to 15 bring together those products having random codon sequences at single codon positions. This mixing also reduces the product populations to three, which are the starting populations for the next round of synthesis. Similarly, for the third, fourth and each remaining 20 position, each reaction product population for the preceding position are divided and a parent and random codon synthesized. Following the above modification of codon-based synthesis, populations containing random codon changes at 25 one, two, three and four positions as well as others can be conveniently separated out and used based on the need of the individual. Moreover, this synthesis scheme also allows enrichment of the populations for the randomized sequences over the parent sequence since the vessel 30 containing only the parent sequence synthesis is similarly separated out from the random codon synthesis.

WO 03/044036 PCT/USO2/37134 55 This method can be used to synthesize a library of nucleic acids encoding antibody variants having amino acid alterations in one or more CDR predicted to be important for binding to a neoplastic cell or antigen 5 thereof. Alternatively, a library of nucleic acids encoding antibody or antibody fragment variants can also be generated using gene shuffling. Gene shuffling or DNA shuffling is a method for directed evolution that 10 generates diversity by recombination (see, for example, Stemmer, Proc. Natl. Acad. Sci. USA 91:10747-10751 (1994); Stemmer, Nature 370:389-391 (1994); Crameri et al., Nature 391:288-291 (1998); Stemmer et al., U.S. Patent No. 5,830,721, issued November 3, 1998). Gene 15 shuffling or DNA shuffling is a method using in vitro homologous recombination of pools of selected mutant genes. For example, a pool of point mutants of a particular gene can be used. The genes are randomly fragmented, for example, using DNase, and reassembled by 20 PCR. If desired, DNA shuffling can be carried out using homologous genes from different organisms to generate diversity (Crameri et al., supra, 1998). The fragmentation and reassembly can be carried out in multiple rounds, if desired. The resulting reassembled 25 genes constitute a library of antibody or antibody fragment variants that can be used in the invention compositions and methods. For certain therapeutic and diagnostic applications it may be preferable to use antibodies or 30 fragments thereof with the same antigen specificity but with different isotypic or allotypic determinants. Such WO 03/044036 PCT/US02/37134 56 antibodies could have, for example, decreased immunogenicity, increased stability, or more optimal effector functions. Thus, functional fragments of the invention can include those obtained by cloning the CDR 5 sequences of a human monoclonal antibody of the invention into different framework regions. Such different framework regions can be obtained from different species, different human individuals, or different heavy or light chain classes from the same or different individual. 10 Such CDR grafting methods are well known to those skilled in the art. An example of grafting CDR sequences from an IgM immunoglobulin to an IgG framework region is described in Example VIII. As demonstrated in Example VIII, in vitro affinity maturation can permit the 15 improvement of virtually any antibody, including low affinity IgMs with essentially germline sequence. Functional activity of antibody or antibody fragment variants of the invention can be evaluated by, for example, methods described above for determining the 20 immunoreactivity of human monoclonal antibodies. Particularly useful methods for determining functional activity of fragments include competitive radioimmunoassay and competitive ELISA assay. Such methods can be used to screen for variants having 25 improved affinity for a neoplastic cell or antigen thereof. The methods can be used at different stringency depending upon the number of variants screened. Stringency can be also adjusted based on the number of cycles or iterations through which the screen has 30 progressed. For example, low stringency can be used in early screening steps and the stringency increased in later screening steps. Use of low stringent affinity WO 03/044036 PCT/US02/37134 57 criteria in screening antibody libraries can be used to broaden the diversity being explored without creating larger libraries or screening a greater number of clones and may lead to the discovery of novel antibodies from 5 previously screened Ig repertoires. The present invention provides pharmaceutical compositions containing a human monoclonal antibody or functional fragment of the invention and a pharmaceutical carrier. Such compositions can be used to administer a 10 human monoclonal antibody or fragment to reduce the proliferation or viability of neoplastic cells. Such compositions can also be used to detect neoplastic cells. Suitable pharmaceutical carriers for the methods of the invention are well known and include, for 15 example, aqueous solutions such as physiologically buffered saline, and other solvents or vehicles such as glycols, glycerol, oils or injectable organic esters. A pharmaceutical carrier can contain a physiologically acceptable compound that acts, for example, to stabilize 20 or increase the solubility of a pharmaceutical composition. Such a physiologically acceptable compound can be, for example, a carbohydrate, such as glucose, sucrose or dextrans; an antioxidant, such as ascorbic acid or glutathione; a chelating agent; a low molecular 25 weight protein; or another stabilizer or excipient. Pharmaceutical carriers, including stabilizers and preservatives, are described, for example, in Martin, Remington's Pharm. Sci., 15th Ed. (Mack Publ. Co., Easton, 1975), which is incorporated herein by reference. 30 All of the above-described pharmaceutical carriers and media can be what is termed in the art pharmaceutical WO 03/044036 PCT/US02/37134 58 grade which means that they are of sufficient purity and quality for use in humans and are distinguishable from comparable reagents in research grade formulations. Those skilled in the art will know that the 5 choice of the pharmaceutical medium and the appropriate preparation of the composition will depend on the intended use and mode of administration. The invention provides a variable heavy chain domain or functional fragment thereof having CDRs with 10 substantially the CDR sequences of the CDRs of a sequence selected from the group consisting of SEQ ID NOs:2, 6, 10, 12, and 14. The invention further provides a variable light chain domain or functional fragment thereof having the CDRs with substantially the CDR 15 sequences of the CDRs of a sequence selected from the group consisting of SEQ ID NOs:4, 8, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44. The invention also provides a CDR or functional fragment thereof having substantially the amino acid sequence of a CDR selected 20 from the group consisting of SEQ ID NOs:2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44. A CDR or functional fragment thereof of the 25 invention can be produced by methods known in the art and as described above. For example, a CDR of the invention can be produced by recombinant means or chemical synthesis. A CDR of the invention can be advantageously used, for example, to generate anti-idiotypic antibodies 30 that selectively bind the tumor-specific human monoclonal antibodies of the invention. Methods for producing and WO 03/044036 PCT/US02/37134 59 using anti-idiotypic antibodies for diagnostic and therapeutic purposes are well known in the art. An antibody fragment of the invention can be a variable heavy chain domain or variable light chain 5 domain. The fragment can be expressed in a recombinant system using methods described herein or a variety of other methods known in the art. For example, the expression of variable heavy chain domains having affinities for a protein antigen similar to those 10 expected of monoclonal antibodies for protein antigens are known in the art as described, for example, in Ward et al., Nature 341:544-546 (1989) and Williams et al., Proc Natl. Acad. Sci. USA 86:5537-5541 (1989). A fragment of the invention can be modified to produce 15 multivalent binding fragments, for example, by incorporation of a cysteine residue that can form an intermolecular crosslink between two such modified fragments as described for variable heavy chain domains in williams et al., supra (1989). A CDR fragment having 20 binding activity can also be produced, for example, by expressing the CDR domain such that it is conformationaly constrained. Expression of a conformationaly constrained CDR formed by addition of two cysteines capable of forming a disulfide bridge and having specificity for the 25 antigen of the parent antibody is described, for example, in Ditzel et al., The J. Immunol. 157:739-749 (1996). An antibody fragment of the invention can be modified by any of a variety of additions, deletions or substitutions of amino acids or other moieties such as labels as 30 described herein.

WO 03/044036 PCT/USO2/37134 60 The invention further provides a method of reducing neoplastic cell proliferation by administering to the cell an effective amount of a human monoclonal antibody or functional fragment produced by a cell line 5 selected from the group consisting of H1140, H2420 and H935. Also provided is a method of reducing neoplastic cell proliferation by administering to the cell an effective amount of a human monoclonal antibody or functional fragment having at least one CDR with 10 substantially the amino acid sequence of a CDR of a sequence selected from the group consisting of SEQ ID NOs:2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44. A human monoclonal antibody or functional fragment used for reducing 15 neoplastic cell proliferation can further include a label, such as a cytotoxic or cytostatic agent, and can be combined with a pharmaceutical carrier. An effective amount of an antibody or 20 functional fragment thereof for reducing neoplastic cell proliferation is known or can readily be determined by one skilled in the art using in vitro methods or credible animal models. In vitro methods can include, for example, determining an effective amount of a composition 25 for reducing neoplastic cell growth or neoplastic cell metastasis. A neoplastic cell used in an in vitro method for assaying reduction in growth or metastasis can be, for example, a tumor cell line or an ex vivo culture of a tumor. The cell line or tumor can be, for example, of 30 breast, lung or ovarian tissue in origin. An effective amount for inhibiting neoplastic cell growth can be, for example, an effective amount for inhibiting DNA synthesis, inhibiting cell division, inducing apoptosis, WO 03/044036 PCT/US02/37134 61 inducing non-apoptotic killing, or inhibiting angiogenesis. An effective amount for inhibiting metastasis of a neoplastic cell can be, for example, an amount effective for inhibiting cell motility, cell 5 migration, cell attachment, cell invasion or cell proliferation. An effective amount of an antibody or functional fragment thereof for reducing neoplastic cell proliferation can also be determined from xenografts of 10 human tumors in rodents. The rodent can be, for example, a rat or a mouse. The mouse can be, for example, normal or immunocompromised. An immunocompromised mouse can be, for example, a nude mouse or a scid mouse. Such species are well known in the art and can be obtained from 15 commercial sources. Human cancer cells can be introduced into an animal by a number of routes, including subcutaneously, intraveneously and intraperitoneally. Following establishment of a tumor, the animals can be treated with different doses of a molecule of the 20 invention, and tumor mass or volume can be determined. Efficacy can be measured as partial or complete regression of the tumor. An effective dose for treating cancer results in more partial and complete regressions of tumors. 25 An effective amount of a molecule of the invention can be determined by one skilled in the art and will depend on such factors as age, body weight, sex and medical condition of the individual, and the particular route of administration of the therapeutic agent. Useful 30 routes of administration of a composition of the invention for treating cancer include, for example, WO 03/044036 PCT/USO2/37134 62 intramuscular, intratumoral, intravascular, intraperitoneal, subcutaneous or intranasal routes. The efficacy of a particular treatment in cancer patients can be determined by one skilled in the 5 art. For example, in vivo or in vitro diagnostic methods, such as those described below, can be used to determine that a tumor has regressed or been eliminated following treatment. Additionally, normal prognostic indicators, such as survival and increased quality of 10 life for the cancer patient, can be used. The invention provides a method of detecting neoplastic cells by contacting a sample with a human monoclonal antibody or functional fragment produced by a cell line selected from the group consisting of H1140, 15 H2420 and H935, and detecting the specific binding of the human monoclonal antibody or functional fragment to the sample, wherein the presence or increased level compared to a normal cell of the human monoclonal antibody or functional fragment indicates the presence of or 20 predisposition to cancer. The invention also provides a method of detecting neoplastic cells by contacting a sample with a human monoclonal antibody or functional fragment having substantially the amino acid sequence of a CDR of a sequence selected from the group consisting of 25 SEQ ID NOs:2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44, and detecting the specific binding of the human monoclonal antibody or functional fragment to the sample, wherein the presence or increased level compared to a normal cell of the human 30 monoclonal antibody or functional fragment indicates the presence of or predisposition to cancer.

WO 03/044036 PCT/USO2/37134 63 As used herein, the term "sample" is intended to mean any biological fluid, cell, tissue, organ or portion thereof, that includes or potentially includes neoplastic cell. A biological fluid can be, for example, 5 blood or lymph. A tissue can be, for example, breast, ovary, or lung. The sample can be an in vivo or in vitro sample. An in vitro sample can be, for example, a histologic section, a specimen obtained by biopsy, or cells that are placed in or adapted to tissue culture. A 10 sample can be prepared by methods known in the art suitable for the particular format of the detection method. A sample can be contacted with a human monoclonal antibody or functional fragment of the 15 invention and specific binding of the human monoclonal antibody to the sample can be detected. Such contacting can be in vivo or in vitro, as determined by one skilled in the art depending on the format of the detection method used. Specific binding of the human monoclonal 20 antibody to the sample can be determined by immunoassays as described above. Such immunoassays include, for example, both in vivo and in vitro immunoassays. In vivo immunoassays include, for example, radioimaging. Such a method involves contacting a sample within an individual 25 with a monoclonal antibody of the invention, and detecting specific binding by, for example, radiographic imaging. In vitro immunoassays include both qualitative and quantitative assays, such as, for example, immunohistochemistry, immunofluorescence, 30 radioimmunoassay, FACS analysis, immunoblotting, immunoprecipitation and ELISA analysis, as described above.

WO 03/044036 PCT/US02/37134 64 The determination that neoplastic cells are present can be made by determining that a specifically bound human monoclonal antibody or functional fragment of the invention is present or is at an increased level 5 compared to a normal cell. As described above, one skilled in the art would be able to determine an appropriate normal cell to use for comparison with a particular type of neoplastic cell. The invention further provides a substantially 10 pure human tumor-specific antigen. The term "tumor-specific antigen" is intended to mean an antigen which is preferentially expressed by human tumor cells. The term "preferentially expressed by human tumor cells" is intended to mean that a tumor-specific antigen is 15 expressed by human tumor cells and is expressed at a substantially lower level by normal human cells. Such tumor cells that express a tumor-specific antigen of the invention can be, for example, of breast, lung or ovarian tissue origin. 20 A human tumor-specific antigen of the invention is specifically reactive with a human monoclonal antibody of the invention. Such an antigen can specifically react with, for example, a human monoclonal antibody or 25 functional fragment produced by LH11238, LH13, H1140, H2420 or H935 hybridoma cell lines. Such an antigen can also specifically react with a human monoclonal antibody or functional fragment having at least one CDR with substantially the amino acid sequence of a CDR of a 30 sequence selected from the group consisting of SEQ ID NOs:2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44 or with a human WO 03/044036 PCT/USO2/37134 65 monoclonal antibody or functional fragment having at least one CDR with substantially the amino acid sequence of a CDR of SEQ ID NO:6 or SEQ ID NO:8. The antigen reactive with LH11238 human monoclonal antibody or 5 variants of fragments thereof is a protein present on the cell surface and in lysosomal compartments of breast and ovarian cancer cells as compared to normal human fibroblasts, peripheral blood lymphocytes, melanoma cells or lung carcinoma cells. The antigen reactive with LH13 10 human monoclonal antibody is a secreted glycoprotein produced by breast, lung and ovarian cancer cells as compared to normal fibroblasts or melanoma cells. Further properties of LH11238 and LH13 antigens are described in the Examples. 15 Human tumor-specific antigens of the invention can beneficially be used for the treatment and diagnosis of cancer. For example, such antigens can be used to generate additional binding agents that specifically bind the human tumor-specific antigen for use in therapeutic 20 and diagnostic procedures. Such binding agents can, for example, inhibit or stimulate the function of a tumor-specific antigen, or modulate the immune system, such that a neoplastic cell is growth-arrested or killed. Such binding agents can also be conjugated to a label, 25 such as, for example, a cytotoxic or cytostatic agent, that causes the death or arrest of the tumor cells. Useful agents that bind to a tumor-specific antigen of the invention include, for example, ligands, receptor antagonists and antibodies. 30 A substantially pure tumor-specific antigen can also be used in the treatment of cancer by vaccinating a WO 03/044036 PCT/US02/37134 66 patient having cancer, or at risk of developing cancer with an effective amount of the antigen. Following vaccination, the immune system of the individual will be able to prevent, reduce the proliferation of, or 5 eliminate neoplastic cells expressing such an antigen. A substantially pure tumor-specific antigen of the invention can also be beneficially used in methods for detecting binding of a tumor-specific binding agent, such as a human monoclonal antibody or functional 10 fragment of the invention. For example, such an antigen can be used in an immunoassay such as competitive ELISA. An appropriate starting material for isolating a substantially pure antigen of the invention can be identified by, for example, screening a panel of human 15 tumor cells, using immunoaffinity procedures well known in the art. For example, as described in Example II, ELISA analysis can be used to determine a useful cell source of antigen for subsequent purification. A useful cell source of antigen can be, for example, breast, 20 ovarian, or lung carcinoma. A particularly useful starting material for the purification of a tumor-specific antigen is the H3396 breast carcinoma cell line. An appropriate method for isolating a 25 substantially pure antigen depends on the cellular localization of the antigen. For example, an antigen of the invention can be predominantly expressed in the secreted medium, cell surface membrane, vesicular membranes, cytoplasm,' or nucleus of the neoplastic cell. 30 Cellular localization of an antigen can be determined by, WO 03/044036 PCT/USO2/37134 67 for example, immunoassays well known in the art. For example, as described in Examples IV and VI, indirect immunofluorescence and ELISA analysis can be used to establish the predominant localization of an antigen. 5 Purification of an antigen can be monitored by immunoaffinity procedures known in the art. Particularly useful methods of monitoring purification include, for example, ELISA analysis and immunoblotting. Tumor-specific antigens can be purified from a 10 particular source by biochemical procedures well known in the art. For example, purification can include centrifugation, chromatographic methods, electrophoretic methods and immunoaffinity methods, and can be chosen by one skilled in the art depending on the characteristics 15 of a particular antigen. Centrifugation procedures can be used to concentrate or enrich for a subcellular fraction which contains an abundant amount of the antigen. Such subcellular fractionation procedures are well known in the art. Chromatographic methods are also 20 well known in the art, and include methods of separating an antigen from contaminants based on its size or differential affinity for particular resins. Such resins can include, for example, size exclusion resins, ion exchange resins, and lectin columns. As described in 25 Example VI, a particularly useful resin for purification of an antigen of the invention is Q SEPHAROSE FAST FLOW® agarose resin. Electrophoretic methods are also well known in the art, and include one and two-dimensional electrophoresis through, for example, acrylamide or 30 agarose gels. Immunoaffinity procedures are also well known in the art and include compounds conjugated to a WO 03/044036 PCT/US02/37134 68 human monoclonal antibody of the invention. Useful compounds for conjugating an antibody for immunoaffinity purification of a tumor-specific antigen include chromatographic resins and Protein A. 5 Thus, following well-known biochemical procedures, one skilled in the art can readily isolate a substantially pure tumor-specific antigen of the invention for use in therapeutic and diagnostic procedures. 10 Substantially purified tumor-specific antigens of the invention can also be prepared from nucleic acids encoding tumor-specific antigens of the invention by recombinant methods known to those skilled in the art. The invention provides isolated nucleic acids 15 encoding human tumor-specific antigens. Nucleic acids encoding human tumor-specific antigens can be isolated by methods known to those skilled in the art. Such methods include, for example, using monoclonal antibodies of the invention to screen an expression library. Other methods 20 include, for example, screening a cDNA or genomic library using degenerate oligonucleotides as hybridization probes. The sequence of such a degenerate oligonucleotide can be determined by microsequencing an isolated tumor-specific antigen of the invention or 25 fragment thereof. Other methods known to those skilled in the art for producing a nucleic acid of a tumor-specific antigen include, for example, the polymerase chain reaction (PCR), using degenerate oligonucleotide primers obtained WO 03/044036 PCT/USO2/37134 69 from amino acid sequence of a tumor-specific antigen of the invention. Desired sequences can be amplified exponentially starting from as little as a single gene copy by means of PCR. 5 The above described methods are known to those skilled in the art and are described, for example, in Sambrook et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, New York (1992) and the various references cited therein and in Ansubel et al., 10 Current Protocols in Molecular Biology, John Wiley and Sons, Baltimore, MD (1989); and in Harlow et al., Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, New York (1989). These references and the publications cited therein are hereby expressly 15 incorporated by reference. It is understood that modifications which do not substantially affect the activity of the various embodiments of this invention are also included within the definition of the invention provided herein. 20 Accordingly, the following examples are intended to illustrate but not limit the present invention. EXAMPLE I Production of Tumor-Specific Human Monoclonal Antibodies This Example shows production of hybridomas 25 that secrete human monoclonal antibodies specifically reactive with tumor cells. A procedure for immunizing normal lymphocytes in vitro with tumor cells or cell membranes prior to immortalization is described. This procedure allowed enrichment of lymphocytes producing WO 03/044036 PCT/USO2/37134 70 novel tumor-reactive human monoclonal antibodies. These human monoclonal antibodies will be useful for cancer immunotherapeutic and immunodiagnostic procedures. Lymphocytes were prepared as follows. Spleen 5 tissue was isolated from accident victims, fragmented, and forced through a no. 50 mesh wire screen (Bellco, Vineland, NJ). The cells were collected by centrifugation at 250 x g for 10 min and RBC were removed by ammonium chloride lysis. The remaining cells were 10 washed, resuspended in freezing medium (40% RPMI, 50% FCS and 10% DMSO) at a concentration of 100 to 300 x 106 cells/ml, frozen in 1.5-ml aliquots, and stored in liquid nitrogen. Both adherent cells and lymphocytes were isolated by this procedure. 15 A mixed lymphocyte reaction (MLR) was then established as follows. Frozen single-cell splenocyte preparations (described above) from two different donors were thawed by gentle shaking at 37 0 C, washed twice with RPMI, and collected by centrifugation for 10 min at 250 x 20 g. After the second wash, 3 x 106 splenocytes from each of the sources were combined in 2 ml of RPMI containing 1.5 mM HEPES, pH 7.4 (Fisher Scientific), 10% FBS (HyClone), 2 mM L-glutamine, non-essential amino acids, 1 mM sodium pyruvate, and 100 pg/ml gentamicin sulfate and placed in 25 a 24-well tissue culture dish. This MLR, together with antigenic stimulation (described below), endogenously produced the required lymphokines. MLR cultures were then stimulated with either mitomycin-treated H3396 tumor cells, plasma membrane 30 preparations from H3396 cells, or paraformaldehyde-fixed WO 03/044036 PCT/USO2/37134 71 H3922 tumor cells. H3922 and H3396 are cultured cell lines established from metastases of human breast adenocarcinoma, which were explanted and maintained in culture. Each cell line was derived from a different 5 explant. Mitomycin-treated H3396 cells were prepared as follows. H3396 cells were plated in 24-well culture dishes, grown to confluency, and treated for 12-15 h with 0.1 ig/ml mitomycin C. This concentration arrested cell 10 division in the tumor cell lines for approximately seven days. Following the 12-15 h incubation, the mitomycin C-containing media was removed and the cells were washed three times with 2 ml of phosphate buffered saline (PBS). Paraformaldeyde-fixed H2922 tumor cells were 15 prepared by incubating cells for 15 min at 25 0 C in 2% paraformaldehyde in PBS after removing the culture medium. The cells were then washed four times with PBS prior to use. Plasma membranes from H3396 cells were isolated 20 as follows. Ten confluent 150-mm dishes of H3396 cells were each rinsed twice with 10 ml of ice-cold (Tris-buffered saline) TBS and harvested in 2 ml/150-mm dish of (Tris saline) TS containing 1 ig/ml leupeptin, 1 pg/ml pepstatin, and 2 mM AEBSF. The cells were broken 25 in a 15 ml Dounce homogenizer with 40 strokes using a type A pestle. The lysate was centrifuged at 800 x g at 40 C for 5 min to remove unbroken cells and nuclei. The supernatant was saved and the pellet was resuspended in 0.5 volumes of TS buffer, homogenized and centrifuged at 30 800 x g. The supernatant was combined with the first WO 03/044036 PCT/USO2/37134 72 supernatant and centrifuged at 10,000 x g at 40 C for 2 h. The supernatant was discarded and the pellet was resuspended in 2 ml of water. A 2 ml Dounce and type B pestle was used to resuspend the membrane pellet. Phase 5 separation of membranes in a 6.4% polymer system was performed on ice by mixing 2.56 ml of 20% dextran, 1.28 ml of 40% polyethylene glycol, 0.20 ml of 0.2 M potassium phosphate, pH 7.2, 0.8 ml of 1 M sucrose, and 2.16 ml of water. To this mixture 1 ml of membranes were added and 10 the tube was inverted 20 times end over end. The phases were separated by centrifugation at 800 x g at 4 0 C for 5 min. The upper phase was drawn off and mixed with the lower phase recovered from a blank (water) sample. Likewise, the lower phase, including the material at the 15 interface, was mixed with the upper phase recovered from a blank sample. Both phases were mixed, inverted 20 times end over end, and separated by centrifugation at 800 x g at 40 C for 5 min. The material recovered from the upper phase of both samples was combined, the volume 20 was adjusted to 21 ml with TBS, and the membranes were collected by centrifugation at 100,000 x g at 40 C for 2 h. The supernatant was discarded and the pellet was resuspended with a 2 ml Dounce homogenizer with a B pestle in 1-2 ml of TS buffer containing the protease 25 inhibitors. The membranes were stored at 40 C for 12 h, at which time the insoluble material was separated from the supernatant by centrifugation at 800 x g at 40 C for 10 min. The pellet was suspended a second time in 1-2 ml of TS buffer containing protease inhibitors and stored as 30 a suspension at 40 C. The supernatant is referred to as fraction 1 while the resuspended particulate material is referred to as fraction 2. Plasma membranes in fraction 1 were enriched greater than 10-fold as determined by WO 03/044036 PCT/USO2/37134 73 measuring 5'-nucleotidase or phosphodiesterase activity. The preparations had minimal succinate-dependent cytochrome C reductase (mitochondria), or NADPH-dependent cytochrome C reductase (endoplasmic reticulum) activity. 5 MLR cultures were immunized in vitro by incubation for three days with one of the following: monolayers of mitomycin C-treated H3396 cells, monolayers of paraformalehyde-fixed H3922 cells, 5 mg of plasma membrane fraction 1 from H3396 cells, or with 10 mg of 10 plasma membrane fraction 2 from H3396 cells. In vitro immunized lymphocytes were immortalized by either of two alternative methods. In the first method, lymphocytes were fused with K6H6/B5 heteromyeloma cells. K6H6/B5 cells were maintained in 15 spinner culture in a 1:1 mixture of RPMI and Iscoves's modification of DMEM, supplemented with 10% FCS, 1% nonessential amino acids, 2 mM glutamine, and 1 mM sodium pyruvate. Approximately 4 x 106 lymphocytes were combined with 2 x 106 log phase K6H6/B5 cells and washed twice with 20 RPMI. One ml of a mixture of 35% polyethylene glycol (approximate m.w. 1450) and 7.5% DMSO in RPMI was added over 1 min, then gradually diluted with RPMI to 4 ml and then diluted with RPMI supplemented with 10% FCS to 16 ml. The lymphocyte concentration was then adjusted to 25 5 x 104 cells/ml, and 0.2 ml of cells were seeded in 96-well cell culture dishes in hybridoma medium (RPMI supplemented with 10% FCS, 1% nonessential amino acids, 2 mM glutamine, 1 mM sodium pyruvate, 15 mM HEPES, pH 7.4 and 0.1 mg/ml gentamycin) containing HAT (13.6 pg/ml 30 hypoxanthine, 3.8 ug/ml thymidine, 1 pg/ml azaserine) and 1.0 pM ouabain, to select against unfused cells.

WO 03/044036 PCT/USO2/37134 74 Hybridomas were screened for reactivity as described below, and clones of interest were expanded. Using this one-step immortalization method, 10-50 hybridomas were produced per 106 lymphocytes fused; 5 however, only 5% of the interesting clones were stable through three rounds of cell culturing. Hybridoma clones H1140, H2420 and H935 were obtained by this immortalization procedure. In an effort to improve clonal stablility, a 10 second, two-step method for immortalizing in vitro immunized lymphocytes was evaluated, involving first transforming cells with EBV and then fusing clones of interest with K6H6/B5 heteromyeloma cells. Equal volumes of lymphocytes (8 x 10' cells/ml) and EBV-transformed 15 1A2/C7 cells (2 x 106 cells/ml) were combined and the total volume was doubled by the addition of hybridoma medium. Hypoxanthine, thymidine and azaserine were added such that their final concentrations were 13.6 pg/ml, 3.8 ug/ml, and 1 pg/ml, respectively. Two hundred 20 microliters of the mixture of cells was plated per well in 96-well cell culture dishes to yield a final total of 4000 lymphocytes and 1 x l05 1A2/C7 cells per well. The cells were fed with hybridoma medium-HAT every three days and assayed for antibody production (described below) 25 after 2 weeks when colonies of cells were visible. Transformation of lymphocytes with EBV generated a higher percentage of antibody-secreting clones than was obtained from fusions of lymphocytes with the heteromyeloma line (50-100 antibody-secreting clones 30 per 106 lymphocytes). However, lymphoblastoid clones generally secreted lower levels of antibody (less than 10 WO 03/044036 PCT/USO2/37134 75 pg/ml) and, as was observed with the hybridomas, displayed poor long-term stability (approximately 5% of the initial clones of interest were stable through three rounds of expansion). To address these limitations, 5 tumor-reactive lymphoblastoid clones were fused with K6H6/B5 cells as described above, prior to multiple rounds of expansion. Formation of hybridomas from lymphoblastoid 10 cells improved the stability such that 40% of the initial lymphoblastoid clones were stable through three rounds of expansion following fusion. The combination of EBV transformation followed by fusion with a heteromyeloma cell line resulted in a higher frequency of 15 immortalization of relevant lymphocytes than was achieved utilizing either approach alone. In addition, these clones generally secreted greater than 20 pg/ml of antibody. Hybridoma clones LH11238 and LH13 were produced by this two-step immortalization method. 20 The immunization and immortalization conditions used to produce hybridoma cell lines H1140, H2420, H935 LH11238 and LH13 are summarized in Table 2.

WO 03/044036 PCT/USO2/37134 76 TABLE 2. Immunization and immortalization conditions used to generate hybridoma cell lines producing tumor specfific human monoclonal antibodies. Hybridoma Immunization Immortalization 5 H1140 5 mg H3396 plasma Fusion with K6H6/B5 membrane fraction 1 cells H2420 5 mg H3396 plasma Fusion with K6H6/B5 membrane fraction 1 cells H935 confluent monolayer of Fusion with K6H6/B5 mitomycin C-treated cells H3396 cells LH11238 10 mg H3396 plasma EBV transformation membrane fraction 2 and fusion with K6H6/B5 cells LH13 monolayer of 4 x 104 EBV transformation paraformaldehyde-fixed and fusion with H3922 cells K6H6/B5 cells 10 Culture supernatants from immortalized lymphocytes were initially screened for reactivity against monolayers of live primary tumor cells using an ELISA assay. This ensured that reactive antibodies recognized surface antigens, and also avoided artifacts 15 associated with screening fixed cells. Tumor cells were plated in 96-well cell culture dishes at a cell density sufficient to produce 90-95% confluent monolayers 12-24 h later. Just prior to use the media was removed and 50 pl of supernatant from either hybridomas or EBV transformed 20 cells was added to the wells and incubated at 40 C for 2 h. Control wells (background) were incubated with 50 jil of fresh hybridoma media.

WO 03/044036 PCT/USO2/37134 77 Following incubation, supernatants were aspirated and the cells were gently rinsed three times with 200 p2l of PBS. The cells were then incubated for 1-2 h with 50 yl of goat anti-human ig (H + L) alkaline 5 phosphatase conjugate which had been diluted 1000-fold in 1% BSA-PBS. The detection antibody was aspirated and the cells were gently rinsed four times as described above. The plates were developed for 1 h at 250 C by the addition of 50 p1 of 10 mM phenolphthalein monophosphate 10 in 0.2 M 2-amino-2-methyl-l-propanol, 0.5 M Tris, pH 10.2 with 0.1% sodium azide. The reaction was terminated by the addition of 50 il of 30 mM Tris, pH 10.2, with 15 mM EDTA and the absorbance at 560 nm was determined. The background value for this assay was generally A560<0.060. 15 Clones with absorbances 4-fold or greater above background were selected for further characterization. Antibodies that displayed reactivity with H3396 cells were also analyzed for binding to fixed HF235 normal human fibroblasts. HF235 cells were fixed 15 min 20 at 25 0 C in 2% paraformaldehyde in PBS after removing the culture medium. The cells were then washed four times with PBS prior to use. The background value for the ELISA with fixed fibroblasts was generally A560<0.080. Minimal reactivity with the fibroblasts was defined as 25 binding less than 2-fold above background. In order to determine the isotype of the antibodies produced by the hybridomas, microtiter plates were coated with 0.5 pg/ml either goat anti-human IgM or goat anti-human IgG. Antibody binding was detected with 30 goat anti-human Ig alkaline phosphate conjugate. Development of the assay was terminated when supernatant WO 03/044036 PCT/US02/37134 78 yielded a positive signal with one of the capture antibodies. Likewise, the light chain class was determined by capturing the antibody with the appropriate heavy chain reagent and detecting with goat anti-human y 5 or K chain-specific alkaline phosphatase conjugate. Immunoglobulin quantitation was performed similarly, except that supernatants containing unknown quantities of Igs were serially diluted until reactivity was undetectable. Ig concentrations were calculated from 10 values of dilutions that fell within the linear range of the standard curve, defined by standard samples of purified polyclonal human IgM, used in the range of 0.01 to 2.0 pg/ml. Immunoglobulins were precipitated from LH11238 15 hybridoma supernatant by centrifugation at 800 x g for 10 min and clarification by filtration through a 0.45 im cellulose acetate filter. The supernatant was placed in dialysis tubing (Mr cut-off 12-14 kDa) and concentrated two- to four-fold by coating the tubing with Aquacide 20 (Calbiochem) and placing it at 40 C for 6-8 h. Excess Aquacide was removed, the dialysis bag rinsed, and the sample was dialyzed 12-24 h versus multiple changes of ice-cold distilled water. The precipitated antibody was collected by centrifugation at 10,000 x g for 30 min. 25 The supernatant was removed and the pellets were resuspended in a minimal volume of warm 10-fold concentrated PBS. Following solubilization of most of the pellet, the buffer concentration was adjusted to PBS and the insoluble material was removed by centrifugation 30 at 10,000 x g for 30 min. Antibodies H1140, H2420 and WO 03/044036 PCT/USO2/37134 79 H935 could similarly be precipitated in low ionic strength buffer. The LH13 antibody was not precipitated in low ionic strength buffer. Therefore, in order to obtain 5 concentrated antibody the volume of the clarified supernatant was reduced with an Amicon apparatus utilizing a YM-30 membrane. The antibody was concentrated greater than 50-fold by this approach. 10 Five percent of the hybridomas screened (372/7216) bound tumor surface antigens greater than four-fold above background. Of these 372 clones, 55 (15%) produced antibodies that did not bind fibroblasts significantly. Although numerous antibody-secreting 15 hybridomas were isolated from control MLRs that had not been incubated with tumor cells or membrane fractions, none of these antibodies displayed preferential tumor reactivity. Approximately 20% (11/55) of hybridomas producing antibodies that displayed tumor specificity 20 were stable through multiple expansions in culture. Five of these clones secreted greater than 20 pg/ml antibody. The properties of these clones are summarized in Table 2. All five antibodies were of the IgM isotype and displayed a range of immunoreactivities as determined by assaying 25 on fixed monolayers of H3396 cells. At least one antibody was generated from each of the culture conditions and immortalization procedures employed. The two antibodies displaying the greatest immunoreactivity, LH13 and LH11238, were characterized further (below). 30 The properties of the tumor-specific human monoclonal antibodies produced by hybridoma cell lines WO 03/044036 PCT/USO2/37134 80 H1140, H2420, H935 LH11238 and LH13 are summarized in Table 3. TABLE 3: Characteristics of tumor-specific human monoclonal antibodies. 5 Hybridoma Isotype/ Secretion Immuno- Precip Light chain Level Reactivity itation class (Pg/ml) a H1140 IgM/X 40 1.0 + H2420 IgM/X 63 1.3 + H935 IgM/h 36 1.4 + LH11238 IgM/K 29 5.7 + 10 LH13 IgM/A 26 341 aSecretion level is typical value obtained from supernatant of a terminal culture blmmunoreactivity (AA560/[(pg Ig)(min)]) determined on confluent paraformaldehyde-fixed H3396 monolayers and 15 normalized to immunoreactivity of H1140 cPrecipitation denotes ability (+) or inability (-) to preciptitate antibody in low ionic strength buffer In summary, this example shows that an unstimulated human immune repetoire (splenocytes from 20 non-tumor-bearing patients) can be used to generate human monoclonal antibodies reactive with novel tumor antigens. This was achieved through the stimulation of MLR cultures with either whole tumor cells or membrane fractions derived from tumor cells. The specificity of the antigen 25 stimulation was demonstrated by the generation of tumor-specific antibodies as well as by the absence of WO 03/044036 PCT/USO2/37134 81 tumor-specific antibodies from cultures not treated with tumor cells or membrane fractions. EXAMPLE II Immunoreactivities of LH13 and LH11238 antibodies 5 This Example shows the immunoreactivities of LH13 and LH11238 antibodies with a panel of human tumor cells. In order to use LH13 and LH11238 antibodies for immunodiagnostic and immunotherapeutic purposes, the 10 range of tumor types expressing the corresponding antigens needs to be identified. Tumor cell lines are representative of the corresponding tumor type, and normal fibroblasts are representative of non-neoplastic tissues. The tumor cell lines were derived from 15 melanomas, lung carcinomas, ovarian carcinomas and breast carcinomas. The human monoclonal antibodies of the invention were tested for immunoreactivity with both human cancer cell lines and normal fibroblasts. H3396, H3464, H3477 and H3922 are cultured cell 20 lines established from metastases of human breast adenocarcinoma, which were explanted and maintained in culutre. H2981 and H2987 are cultured cell lines established from human lung carcinomas, which were explanted and maintained in culture. H3639 and H3723 are 25 culture cell lines established from human ovarian carcinomas, which were explanted and maintained in culture. Each cell line was derived from a different explant.

WO 03/044036 PCT/USO2/37134 82 Immunoreactivities were determined by the ELISA procedure described above, by incubating a broad range of antibody concentrations with fixed monolayers of tumor cells. Antibody binding was measured with saturating 5 quantities of detect antibody. Antibody binding was measured as the AA560/[(pg Ig)(min)] in the linear range of the assay. The immunoreactivities of tumor-specific human monoclonal antibodies against a panel of tumor and normal cells are presented in Table 4. 10 TABLE 4: Immunoreactivity of human monoclonal antibodies LH13 and LH11228 against tumor and normal cells. Immunoreactivity (AA560/ [(g Ig) (min) I) Cell Line Description LH13 LH11238 15 HF285 normal fibroblast 0 0 H2669 melanoma 0 0 H3774 melanoma 0 0 H3396 breast carcinoma 1.136 0.019 H3464 breast carcinoma 0.507 0.022 20 H3477 breast carcinoma 0.046 0.008 H3922 breast carcinoma Oa 0 H2981 lung carcinoma 0.584 0 H2987 lung carcinoma 0 0 H3639 ovarian carcinoma Oa 0.011 25 H3723 ovarian carcinoma 2.602 0.026 aExhibited substantial immunoreactivity upon permeabilization of cells with 0.1% digitonin WO 03/044036 PCT/USO2/37134 83 As shown in Table 4, LH13 displayed a broad cross-reactivity on the panel of tumor cells with high (H3723 and H3396), intermediate (H2981 and H3464), and low (H3477) immunoreactivities. The LH13 antigen was 5 absent or present at undetectable levels on intact normal fibroblasts (HF285), melanomas (H2669 and H3774), and with several of the intact carcinomas tested (H3922, H2987, and H3639). The lack of reactivity with H3922 cells was particularly surprising because the LH13 10 antibody was isolated from lymphocytes which had been stimulated with H3922 cells. To examine this more closely each of the cell lines was permeabilized with 0.1% digitonin and their reactivity with LH13 was re-examined. Two of the cell lines which were completely 15 negative when intact cells were assayed, H3922 and H3639, bound the LH13 antibody under conditions in which the antibody had access to intracellular compartments. Furthermore, H2981 cells, which had an intermediate LH13 immunoreactivity when assayed under non-permeabilizing 20 conditions were highly reactive under permeabilizing conditions (immunoreactivity greater than 7.0). The data obtained for immunoreactivities in the presence and absence of 0.1% digitonin could not be compared directly, however, due to differential cell loss from the cell 25 culture dishes during the incubations. LH11238 displayed a similar reactivity profile, though the magnitude of its immunoreactivity was always much less than that of the LH13 antibody. Reactivity was observed with intact H3396, H3464, H3477, H3639, and 30 H3723. Other cells tested were negative, including the fibroblast and melanoma cell lines. Based on the immunoreactivity determinations made for each of the cell WO 03/044036 PCT/USO2/37134 84 lines with LH13 and LH11238 it appears unlikely that these antibodies recognize the same epitope. For instance, LH13 displayed moderate immunoreactivity with the H2981 lung carcinoma line while no binding of LH11238 5 to this cell line was detected. Likewise, LH11238 bound intact H3639 ovarian carcinoma cells, even though binding of LH13 could not be detected under identical incubation conditions. EXAMPLE III 10 Binding Activity of LH13 and LH11238 Human Monoclonal Antibodies This Example shows the binding activity of human monoclonal antibodies LH13 and LH11238 with normal and tumor cells. 15 In order to determine the immunoreactivity of human monoclonal antibodies with live tumor cells and normal cells, flow cytometry (FACS analysis) was used. H3464 breast tumor cells were chosen because they expressed both LH13 and LH11238 antigens by ELISA 20 analysis and were readily isolated as a single cell suspension. Normal peripheral blood lymphocytes were also examined as representative of normal tissues. Facs analysis additionally permits examination of heterogeneity of a cell population with respect to 25 antigen expression. H3464 cells were removed from culture dishes with trypsin or Versene (EDTA), washed with PBS, resuspended at 2 X 106 cells/ml in tumor media, and allowed to recover 2 h at 370 C. Antibodies were WO 03/044036 PCT/USO2/37134 85 incubated at 5.0 ug/ml with 1 x 106 tumor cells in 50

A

1 total volume for 30 min on ice. The cells were washed once with 1.0 ml of ice-cold PBS, incubated with 2 Ag/ml FITC-labeled goat anti-human IgM diluted in 1% BSA-PBS 5 for 30 min on ice, and washed once more with 1.0 ml of ice-cold PBS. Antibody binding to cells was analyzed with a Becton Dickinson FACSort. Consistent with the ELISA results, H3464 cells incubated with LH11238 and FITC-labeled antihuman IgM 10 displayed a shifted staining pattern relative to an irrelevant control human IgM (Figure 1A). Greater than 90% of the cells tested bound LH11238 antibody, though the broad range of fluorescent intensities observed was consistent with heterogeneous expression levels of the 15 antigen on these tumor cells. Similar FACS staining profiles were also observed with H935, H2420, and H1140. Surprisingly, LH13 antibody, which was 23-fold more reactive than LH11238 on H3464 cells as determined by ELISA (Table 4) displayed little shift (Figure lB). 20 To determine if the LH13 antigen was particularly sensitive to the proteolytic conditions used to isolate the tumor cells for analysis (trypsin), cells were allowed longer recovery periods following isolation or non-proteolytic cell isolation methods were utilized, 25 such as Versene (EDTA) release. Neither of these approaches resulted in significantly greater staining of the tumor cells. Facs analysis of normal human cells, such as peripheral blood lymphocytes, was negative for each of WO 03/044036 PCT/US02/37134 86 the antibodies tested, consistent with the ELISA results obtained with normal fibroblast cells. EXAMPLE IV Characterization of LH11238 AntiQen 5 This Example describes the subcellular characterization of the LH11238 antigen. Both the ELISA screening and the FACS analysis of live cells described above demonstrated the expression of LH11238 antigen on the plasma membrane of tumor cells. 10 However, these approaches did not examine the distribution of the antigen on intracellular structures of carcinoma-derived cells. To examine the intracellular localization of LH11238 antigen, immunofluorescence analysis was used. 15 Monolayers of H3464 cells were seeded on 12-mm round coverslips (No. 1 thickness, 0.06-0.13 mm thick) one day prior to use and were fixed with 2% paraformaldehyde in PBS for 15 min at 250 C. The cells were rinsed twice with PBS and incubated with 50 Ail/ml 20 antibody diluted either in 1% BSA-PBS (non-permeabilized) or in 1% BSA-PBS containing 0.1% digitonin (permeabilized) for 2 h at 40 C. The cells were rinsed twice with PBS and were then incubated in the dark with FITC-labeled goat anti-human IgM diluted 1:500 in the 25 same buffer as the primary antibody. The cells were rinsed 4 times with PBS and the coverslips were mounted in Fluoromount-G (Southern Biotechnology). Cells were visualized with an Olympus microscope equipped with WO 03/044036 PCT/US02/37134 87 epifluorescent optics and an Olympus Splan 40X (NA 0.70) objective lens. Non-permeabilized, paraformaldehye-fixed H3464 cells stained with the LH11238 antibody displayed surface 5 staining consistent with localization to the plasma membrane. Consistent with the FACS analysis, greater than 70% of fixed H3464 cells bound LH11238. When H3464 cells were permeabilized to allow access to intracellular structures, punctate, peri-nuclear staining was observed. 10 LH11238 binding to H3464 cells was specific as control incubations with irrelevant, isotype and concentration-matched human antibodies did not stain intact or 0.1% digitonin permeabilized cells. The punctate, peri-nuclear staining observed on permeabilized 15 cells suggested localization of the antigen in the lysosomes. To verify this, cells were stained with antibodies to CD63, a known lysosomal glycoprotein. Incubation of H3464 cells with antibody to CD63 stained intracellular structures similar to those labeled with 20 LH11238, consistent with a lysosomal localization of intracellular LH11238. Based on these results it was concluded that the LH11238 antigen is present both on the plasma membrane and in the lysosomes of H3464 cells. 25 EXAMPLE V Internalization of LH11238 AntiQen This Example shows that LH11238 antigen is internalized from the plasma membrane to lysosomal compartments.

WO 03/044036 PCT/USO2/37134 88 Immunofluorescence experiments described above indicated that LH11238 antigen was present on both the cell surface and in lysosomes of tumor cells. The dual localization could result from LH11238 antigen being 5 co-expressed in lysosomes and the cell surface, or from being internalized to endosomal/lysosomal compartments. In order to determine whether LH11238 antigen was internalized, the procedure for immunofluorescent localization was modified. 10 Monolayers of live H3464 cells were chilled on ice for 30-60 min to completely inhibit endocytosis. The cells were then washed, incubated with LH11238 antibody, washed again, and incubated with FITC-labeled goat anti-human IgM antibody. The cells were maintained at 15 4 0 C throughout all of the preceeding steps to ensure complete inhibition of endocytosis. The cells were then shifted to 370 C with pre-warmed cell culture media. At various intervals the cells were shifted back to 40 C and fixed with 2% paraformaldehyde. Initially, diffuse 20 surface staining was observed, identical to that observed with non-permeabilized fixed cells. Following 10 min at 370 C, clustering of the LH11238 antibody at multiple sites on the cell surface was observed. With longer incubation times, LH11238 first localized to a few sites 25 on the cell surface, and then began to internalize. These staining patterns were not observed if cells were incubated with irrelevant primary antibodies, incubated with secondary antibody only, or if the cells were maintained at 40 C for the duration of the experiment. 30 These results are consistent with LH11238 binding and being internalized by H3464 cells.

WO 03/044036 PCT/USO2/37134 89 In summary, the LH11238 antibody was first characterized as a surface antigen based on ELISA analysis of intact cells, FACS analysis of live cells, and immunolocalization using non-permeabilized cells. 5 However, further immunolocalization studies with permeabilized cells demonstrated that a portion of this antigen was also localized to a punctate, peri-nuclear compartment. Similar staining was observed with cells incubated with anti-CD63 antibody, a lysosomal protein. 10 Consistent with the dual localization of the LH 11238 antigen on plasma membranes and in the lysosomes, it was demonstrated that antibody bound to the surface of live cells was internalized. The dual localization of tumor antigens to the plasma membrane and lysosomes has been 15 observed previously. For instance, the BR96 antibody binds the lysosomal-associated membrane protein lamp-1 on the surface of intact tumor cells, even though lamp-1 is normally an integral membrane protein predominantly located in the lysosomal compartment. In addition, the 20 secretion of elevated levels of soluble lysosomal proteins, such as cathespins B, D, and L from tumor cells has also been documented. A portion of the secreted cathespin D is bound by the mannose 6-phosphate receptor and internalized. At present, it is unclear whether 25 LH11238 is soluble or an integral membrane protein. However, the antigen is found in the soluble fraction of a TX-114 phase separation of cell extracts, consistent with LH11238 being membrane-associated as opposed to an integral membrane protein. 30 WO 03/044036 PCT/USO2/37134 90 EXAMPLE VI Characterization of LH13 Antigen This example shows the localization of the LH13 antigen. 5 The LH13 antibody exhibited significant immunoreactivity when assayed by ELISA on fixed monolayers of carcinoma-derived cell lines, as described above. However, FACS analysis suggested that LH13 bound only poorly to cell surfaces. Further ELISA analysis 10 indicated that LH13 antigen was predominantly present in a compartment that was only accessible to antibody when the cells were treated with 0.1% digitonin. To further examine the expression of LH13 antigen in either subcellular compartments or in the secreted medium, 15 immunofluorescence analysis and direct ELISA analysis were used. To examine the subcellular localization of LH13 antigen, immunofluorescence was performed on paraformaldehyde-fixed H3464 cells as described above. 20 Incubation of LH13 antibody with unpermeabilized H3464 cells (as described above using LH11238 antibody) resulted in weak staining of the cell surface, consistent with the slight shift observed by FACS analysis. Incubation of LH13 antibody with permeabilized H3464 25 cells by immunofluorescence resulted in no detectable staining. Possibly, little LH13 is associated with intracellular structures in H3464 cells or intracellular forms of LH13 were not recognized by the antibody in this cell line. 30 WO 03/044036 PCT/USO2/37134 91 These results could also indicate that LH13 antigen was predominantly secreted. To observe secreted LH13 antigen, a constant amount of LH13 antibody (0.01 pg/ml)was diluted into increasing amounts of culture 5 media which had been removed from confluent monolayers of H3396 cells (conditioned media). Conditioned media was then assayed for binding to fixed H3396 monolayers. A modest reduction (approximately 15%) in the binding of the LH13 antibody to the monolayer was observed with 10 increasing concentrations of conditioned media (Figure 2A), consistent with LH13 antigen being present in the media. No reduction in the binding of the LH13 antibody to the monolayer was observed with increasing concentrations of conditioned media from H3922 cells. 15 To observe LH13 antigen in conditioned media of H3396 cells, a direct ELISA format was used. Conditioned media was incubated in a 96-well cell culture dish for 10-12 h at 40 C, removed, and the washed wells were subsequently assayed for LH13 antibody binding. The 20 antigen present in conditioned media obtained from H3396 cells bound to the culture dish and was readily detected (Figure 2B). The specificity of antibody binding was demonstrated by incubating LH13 antibody with cell culture wells that had been pre-treated with: (1) 25 conditioned media from a tumor line expressing undetectable quantities of LH13 (H3922), (2) fresh media (media), or (3) nothing (blank). None of these pre-incubation conditions resulted in detectable LH13 binding to the well (Figure 2B). 30 In summary, characterization of the LH13 antigen demonstrated that the majority of this antigen is WO 03/044036 PCT/USO2/37134 92 found in conditioned media. Based on this observation, it was concluded that LH13 is secreted from tumor cells. EXAMPLE VII Purification and Further Characterization of LH13 antigen 5 This example describes the purification and further characterization of LH13 antigen. As described above, the LH13 antigen is secreted from breast tumor cells and displays a surprisingly high affinity for cell culture dishes. 10 These observations suggested that LH13 antigen could be readily purified from conditioned H3396 media using its binding to culture dishes as a means of monitoring its purification. Once the LH13 antigen is purified, its susceptibility to various reagents could readily be 15 determined to further characterize its properties. As a first step in enriching the LH13 antigen, the H3396 cell line was adapted to grow in serum-free media supplemented with minimal protein [Iscove's media supplemented with TCH (Celox, defined serum replacement), 20 2mM L-glutamine, non-essential amino acids, and 1 mM sodium pyruvate]. Cells grown under these conditions secreted comparable levels of the antigen as grown in the presence of serum. Conditioned media was collected, pooled, and 250 ml was diluted to 1 liter with water to 25 reduce the ionic strength. The diluted media was applied at 3 ml/min to a 11 cm x 1.5 cm column of Q SEPHAROSE FAST FLOW agarose resin which had been equilibrated with 0.25-fold concentrated (0.25X) PBS. The majority of the antigen bound to the column as demonstrated by the lack WO 03/044036 PCT/US02/37134 93 of reactivity of the flow-through fraction. After washing the column with 200 ml of 0.25X PBS, protein was eluted at a flow rate of 6 ml/min with a 90 min linear gradient from 100% 0.25X PBS to 70% 0.5 M NaCl in 0.25X 5 PBS. Column fractions were assayed for protein content with the BCA protein assay using BSA as the standard. Column fractions were assayed for LH13 antigen as follows. Each fraction was diluted 10-fold into water 10 and 50 pl/well was transferred to a sterile 96-well cell culture dish for 12 h at 40 C. The plate was washed twice with PBS and incubated with 10 Ag/ml LH13 antibody diluted in 1% BSA-PBS for 2 h at 370 C. The plate was then washed four times with PBS, incubated with goat 15 anti-human lg alkaline phosphatase conjugate diluted 1000-fold into 1% BSA-PBS for 1 h at 250 C, and was developed as described above. Most of the protein eluted between 150 mM and 250 mM NaCl (Figure 3, closed circles), while the 20 majority of the LH13 antigen eluted above 250 mM NaCl (Figure 3, open circles). The specific activity (ELISA signal/[protein]) of the peak antigen fraction was 200 fold greater than the starting material. As a control, LH13 antigen was diluted with a range of NaCl 25 concentrations (50 mM to 400 mM) to determine if the ionic strength affected binding to the cell culture dishes. The range of NaCl concentrations employed for binding and eluting LH13 from the column did not affect antigen binding to the cell culture dish. Therefore, the 30 direct ELISA accurately reflects the distribution of antigen in the various column fractions.

WO 03/044036 PCT/US02/37134 94 Column fractions were further resolved by electrophoresis on 4-20% SDS-polyacrylamide gradient gels. Coomassie blue staining revealed a single major protein in the most reactive fractions, but Western blots 5 of the various column fractions failed to identify this as the LHl3-reactive antigen. To further characterize the LH13 antigen, its susceptibility to a variety of treatments was examined 10 using the direct ELISA method. LH13 antigen was coated on cell culture dishes, as described above, and incubated under a variety conditions. LH13 antigen was incubated with 5 mg/ml trypsin at 370 C for 30 min. Alternatively, LH13 antigen was denatured by treatment with 10% SDS + 2% 15 P-mercaptoethanol at 370 C for 30 min. prior to treatment with 2 U/ml sialidase (Oxford GlycoSystems), 400 U/ml endoglycosidase-F/peptide-N-glycosidase F (Endo F, PNGase F)(Oxford GlycoSystems), 60 mU/ml endo-a-N acetylgalactosaminidase (O-glycanase)(Oxford 20 GlycoSystems), or 2 U/ml sialidase plus 60 mU/ml endo-a N-acetylgalactosaminidase (Sialidase, O-glycanase) at 370 C for 24 h. The culture dishes were washed three times with PBS and assessed for LH13 antibody binding as described above. The effect of each treatment was 25 compared to control samples which were treated under identical conditions with buffer alone (Figure 4). Treatment of LH13 antigen with 10% SDS ± 2% B-mercaptoethanol, sialidase, or endo-a-N 30 acetylgalactosaminidase (O-glycanase) had little effect on subsequent binding of antibody. The resistance of the antigen to treatment with sialidase and O-glycanase was not affected by denaturing the antigen prior to treatment WO 03/044036 PCT/US02/37134 95 nor by simultaneous treatment with both glycosidases. Binding of antibody in the direct ELISA was reduced by treatment with trypsin (79%, Figure 4) or with endoglycosidase-F/peptide-N-glycosidase F (36%, not 5 shown). The binding of antibody to LH13 denatured with 10% SDS + 2% -mercaptoethanol prior to treatment with endoglycosidase-F/peptide-N-glycosidase F was further diminished, reducing binding by 75% as compared to untreated samples (Figure 4). 10 Due to the broad range of specificities of endoglycosidase-F/peptide-N-glycosidase F it was not possible to reach specific conclusions regarding the structure of the carbohydrate. However, binding of LH13 15 antibody was completely unaffected by treatment with sialidase, which removes N- or O-acyl, non-reducing terminal sialic acids. Treatment with endo-u-N acetylgalactosaminidase, which removes Gal31-3GalNAc associated with serine or threonine, or the combination 20 of sialidase and endo-a-N-acetylgalactosaminidase also did not affect antibody binding. Trypsin sensitivity of antibody binding to antigen suggested that the epitope is associated with a protein component. In summary, LH13 antigen was purified more than 25 200-fold from H3396 serum-free conditioned medium, using anion exchange chromatography. Preliminary studies indicated that purified LH13 antigen is excluded from a Superdex 75 gel filtration column, consistent with the antigen being larger than 70 kDa. Treatment of LH13 30 antigen bound to cell culture plates with a variety of reagents provided evidence that both carbohydrate and protein components are present, and are involved in WO 03/044036 PCT/USO2/37134 96 either binding of the epitope to LH13 antibody or binding LH13 antigen to tissue culture plates. At present, it is not possible to distinguish between these two possibilities. 5 EXAMPLE VIII In vitro Antibody Maturation This example demonstrates cloning of antibody variable regions, synthesis of Fab libraries containing variants of the cloned variable regions, and screening to 10 identify variants with improved affinity. DNA sequencing of LH13 and LH11238 demonstrated that the H and L chains of both antibodies were highly homologous to human germline sequences. The LH11238 VH is identical to germline sequence DP-63 (VH4.21) while 15 the LH13 VH is identical to DP-10/hvl051 (Vhl-69). Furthermore, the LH11238 VL is identical to germline sequence DPK21/humkv328h5 while the LH13 VL is highly homologous to DPL16/VL3.1, containing a single nucleotide change at the V-J junction. The high degree of homology 20 between the antibodies and germline sequences demonstrated that the antibodies probably did not undergo appreciable affinity maturation in the cell cultures. In order to rapidly enhance the physical attributes of the LH13 and LH11238 antibodies through protein engineering, 25 the antibodies were cloned into a bacterial expression system, as set forth below. Antibody VH and VL variable regions were cloned as follows. Total RNA was isolated from 5 x 10' cells of each clone and fist strand cDNA was prepared. The H chain WO 03/044036 PCT/US02/37134 97 variable regions were amplified by PCR using the following set of 5' primers homologous to human signal sequences and a 3' primer (5'- AGACGAGGGGGAAAAGGGTT-3', SEQ ID NO:47) corresponding to human CH1: 5 ATGGAGTTTGGGCTGAGCTGG (SEQ ID NO:48), ATGGACTGGACCTGGAG(G/C)(A/T/G)TC (SEQ ID NO:49), ATGAA(A/G)CA(C/T)CTGTGGTTCTT (SEQ ID NO:50), ATGGGGTCAACCGCCATCCTC (SEQ ID NO:51), ATGGGATGGAGCTGTATCATC (SEQ ID NO:52), 10 ATGTCTGTCTCCTTCCTCATC (SEQ ID NO:53), and ATG(A/G)AC(C/A)TACTTTGTT(G/C)C (SEQ ID NO:54). The kappa light chain variable regions were amplified using a set of 5' primers encoding signal sequence (CT(C/T)CT(G/C)(G/T)(G/T)(G/C)CTCCTGCT(A/G)CTCTGG, SEQ ID 15 NO:55 and CT(C/T)CT(G/C)(G/T)(G/T)(G/C)CT(C/G)CT(G/A)(C/G/A/T)T(A/G )CTCTGG, SEQ ID NO:56) and a 3' primer corresponding to constant region amino acids 117-122 (CATCAGATGGCCGGGAAGAT, SEQ ID NO:57). Similarly, the 20 lambda light chain variable regions were amplified using a set of 5' primers encoding signal sequence (ATG(A/G)CCTG(C/G)(A/T)C(C/T)CCTCTC(C/T)T(C/T)CT(C/G)(A/T )(C/T)C, SEQ ID NO:58) and a 3' primer corresponding to constant region amino acids 115-124 25 (CTCCTCAGAGGAGGGCGGGAACAGAGTGAC, SEQ ID NO:59). Following PCR amplification, DNA fragments were purified on an agarose gel and cloned into the pCR2.1 vector (Invitrogen, Carlsbad CA). Multiple clones of each sample were sequenced on both strands by the fluorescent 30 dideoxynucleotide termination method using MI3 forward and reverse primers (Perkin-Elmer, Foster City, CA). A slash between nucleotides at a particular position in a WO 03/044036 PCT/USO2/37134 98 sequence is intended to indicate an equal molar mixture of those nucleotides at that position. In order to convert the IgMs to more useful 5 therapeutic mAbs the isotypes were switched from IgM to IgGl by grafting the H chain variable regions directly to a yl constant region. In vitro class switching was done to increase control over antibody effector functions and to exploit better-characterized expression, production, 10 and purification methodologies for therapeutic mAbs. The cloned VH and VL regions of the LH11238 antibody were cloned into the phage vector M13IX104CS (described in Wu et al., J. Mol. Biol. 294:151-162 (1999) and Kristensson et al., Vaccines 95, Cold Springs Harbor Laboratory 15 Press, Cold Springs Harbor, NY pp39-43 (1995) in frame to human IgG CH1 and kappa CL sequences by hybridization mutagenesis (as described in Rosok et al., J. Biol. Chem. 271:22611-22618 (1996)). The LH13 antibody was cloned into the same vector in which the kappa CL had been 20 replaced with lambda CL sequence. In order to rapidly improve the affinity of the antibodies in vitro, while minimizing the potential of altering the epitope specificity, libraries of HCDR3 and LCDR3 variants closely related to the parent sequence 25 were synthesized and screened. There are multiple mechanisms in vivo by which diversity is introduced to a greater extent in HCDR3 and LCDR3 than in the other CDRs of the antibody, consistent with these regions serving a key role in antigen recognition. Moreover, during 30 previous in vitro affinity maturation of unrelated mAbs, the greatest number of beneficial mutations were typically located in HCDR3 and LCDR3. Therefore, the WO 03/044036 PCT/USO2/37134 99 initial libraries were focused in the third CDR of the H and L chains and the variants altered from the parental sequence by a single amino acid change. Using the numbering system of Kabat et al. 5 supra, the residues chosen for mutagenesis of the LH13 CDRs were: Asn89 through Val97 in L chain CDR3 (LCDR3) and Glu95 through Tyrl02 in H chain CDR3 (HCDR3) and the residues chosen for mutagenesis of the LH11238 CDRs were: Gln89-Thr97 in LCDR3 and Glu95-Tyr102 in HCDR3. The 10 initial libraries and the combinatorial LH13 library based on the beneficial LCDR3 and HCDR3 mutations were synthesized as described in Wu et al., Proc. Natl. Acad. Sci. USA 95:6037-6042 (1998) and are shown in Table 5.

WO 03/044036 PCT/USO2/37134 100 TABLE 5: Seqruence Anti- Library Clone Heavy Light body Chain Chain 5 CDR3 CDR3 LH13 HCDR3 Wild- E D S S G W Y H Y type S97G G 10 S97T T S97N N LCDR3 Wild- N S R D S S G N P V V type R91Y Y 15 R91F F V97Y Y V97F F Combin- 4D5 T Y atorial 20 4E2 T Y F 4H7 T F F 4011 F F 3G4 Y F LH11238 LCDR3 Wild- QQ Y N N W P P Y T 25 type Q89L L Q89G G Q89V V Q89F F 30 Q89w W N93C C N94C C P95aF F P95aR R WO 03/044036 PCT/US02/37134 101 The antibody variants in these focused libraries each contained a single mutation, and all 20 amino acids were introduced at each residue of both CDRs. Thus, the LH11238 libraries contained 228 (HCDR3) and 190 5 (LCDR3) distinct non-wild-type variants while the LH13 libraries contained 171 (HCDR3) and 209 (LCDR3) distinct variants. Fabs were expressed in E. coli XLl-blue cells (Stratagene) as described in Wu et al, supra (1999) and 10 Fabs were isolated from the periplasmic space and quantitated as described in Watkins et al., Anal. Biochem. 253:37-45 (1997). Soluble Fabs were screened in an ELISA format. The binding of soluble LH13 Fab to partially purified LH13 antigen was detected with goat 15 anti-human lambda chain alkaline phosphatase conjugate. LH13 antigen was partially purified as described in Example VII. LH11238 Fab binding was determined using fixed H3396 tumor cells as described in Watkins et al., supra (1997). 20 Bacterially expressed LH13 and LH11238 Fabs displayed relatively low reactivity in the ELISA as shown in Figure 5, (open circles) indicating that the antibodies had not undergone significant antibody maturation in vivo. The results further are consistent 25 with the observation that the binding of IgM molecules is driven largely by multivalency. Screening of the focused libraries identified multiple variants of each antibody that contained single amino acid changes and displayed as much as 100-fold greater binding than the wild-type 30 antibodies as indicated by comparison of open circles with open squares in Figures 5A and 5B. Beneficial WO 03/044036 PCT/tUS02/37134 102 mutations in the LH13 antibody were identified in HCDR3 and LCDR3. As shown in Figure 5A, clone S97N (filled circles) and clone R91Y (open squares) had enhanced binding compared to the LH13 parent. Multiple beneficial 5 mutations were also identified in LCDR3 of LH11238 including clone N94C (filled squares), clone N93C (open triangles) and clone P95aR (filled circles) as shown in Figure 5B. Variants displaying enhanced binding were 10 sequenced to identify the mutations that resulted in higher affinity and are also listed in Table 5. The LH13 variant displaying the highest binding activity (Figure 5A, open squares) contained an R91Y mutation in LCDR3 and had a Kd of about 1.8 nM. The LH11238 mutant displaying 15 the greatest increase in affinity (Figure 5B, open squares) was the result of a Q89W mutation in LCDR3 and had a Kd of about 11 nM. Limited screening of the libraries identified seven LH13 variants and nine LH11238 variants that displayed greater than 2-fold enhanced 20 binding all of which are listed in Table 5. Beneficial mutations of LH13 CDRs were identified at S97 of HCDR3 and R91 and V97 of LCDR3 while the improved LH11238 variants contained beneficial changes at Q89, N93, N94, and P95a of the LCDR3. These results indicated that 25 single amino acid changes in LCDR3 of LH13 increased the affinity by greater than 50 fold while single amino acid changes in LCDR3 of LH11238 increased the affinity greater than 35 fold. A combinatorial library of the beneficial 30 mutations identified in both HCDR3 and LCDR3 of LH13 was synthesized to combine individual mutations and further WO 03/044036 PCT/USO2/37134 103 enhance antibody affinity. The library contained every possible combination of the individual mutations or wild-type amino acid at each of the three CDR positions (S97 of HCDR3 and R91 and V97 of LCDR3), resulting in a 5 library that contained 36 distinct variants. Combinatorial clones displaying higher affinity than the variants containing single amino acid changes were readily identified as shown by comparison of combinatorial mutant 4H7 (filled squares) with LCDR3 10 variant R91Y (open squares) or HCDR3 variant S97N (filled circles) in Figure 5A. Combinatorial mutant 4H7 had a Kd of about 1.1 nM. DNA sequencing of the combinatorial clones demonstrated that all contained at least two mutations, as shown in Table 5, and two contained three 15 mutations (clones 4E2 and 4H7). All five combinatorial clones displayed greater binding activity than any of the variants containing a single mutation. This example demonstrates affinity maturation of LH13 accomplished by synthesis of only 416 distinct 20 protein variants in two steps. As set forth above, the first step consisted of screening 171 HCDR3 mutants and 209 LCDR3 mutants followed by a second step in which 36 combinatorial variants were screened. In contrast, total randomization of the three CDR residues demonstrated to 25 influence activity in this study, LCDR3 S97, HCDR3 R91, and HCDR3 V97, would have required the expression of a library containing 193, or 6,859, variants. Thus, the results described here demonstrate that stepwise improvement in affinity captured additivity of 30 independent mutations and further provided an efficient method of affinity maturation.

WO 03/044036 PCT/USO2/37134 104 Throughout this application various publications have been referenced. The disclosures of these publications in their entireties are hereby incorporated by reference in this application in order to 5 more fully describe the state of the art to which this invention pertains. Although the invention has been described with reference to the disclosed embodiments, those skilled in the art will readily appreciate that the specific 10 experiments detailed are only illustrative of the invention. It should be understood that various modifications can be made without departing from the spirit of the invention. Accordingly, the invention is limited only by the following claims.

Claims

1. An isolated human monoclonal antibody or functional fragment thereof, comprising a complementary determining region having substantially the amino acid 5 sequence of a CDR selected from the group consisting of SEQ ID NOs:10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44, wherein said human monoclonal antibody or functional fragment thereof specifically binds a neoplastic cell or antigen thereof. 10

2. The isolated human monoclonal antibody of claim 1, wherein said functional fragment is selected from the group consisting of Fv, Fab, Fab', or F(ab') 2 .

3. The isolated human monoclonal antibody or functional fragment of claim 1, further comprising a 15 label.

4. The isolated human monoclonal antibody or functional fragment of claim 3, wherein said label comprises a cytotoxic or cytostatic agent.

5. A pharmaceutical composition, comprising 20 the isolated human monoclonal antibody or functional fragment of claim 1 and a pharmaceutical carrier. WO 03/044036 PCT/USO2/37134 106

6. The isolated human monoclonal antibody or functional fragment of claim 1, comprising three complementary determining regions having substantially the amino acid sequences of the three CDRs set forth in a 5 sequence selected from the group consisting of SEQ ID NOs:10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36, 38, 40, 42 and 44.

7. The isolated human monoclonal antibody or functional fragment of claim 1, further comprising a 10 physiologically acceptable compound.

8. An isolated CDR or functional fragment thereof, comprising substantially the amino acid sequence of a CDR selected from the group consisting of SEQ ID NOs:10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 15 36, 38, 40, 42 and 44 wherein said CDR or functional fragment thereof specifically binds a neoplastic cell or antigen thereof.

9. An isolated nucleic acid, encoding a human monoclonal antibody or functional fragment thereof, 20 comprising a nucleotide sequence encoding substantially the amino acid sequence of a CDR encoded by a nucleotide sequence selected from the group consisting of SEQ ID NOs:9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43 and 45. 25

10. The isolated nucleic acid of claim 9, wherein said functional fragment is selected from the group consisting of Fv, Fab, Fab', or F(ab') 2 . WO 03/044036 PCT/USO2/37134 107

11. The isolated nucleic acid of claim 9, comprising a nucleotide sequence encoding substantially the amino acid sequences of the three CDRs set forth in a sequence selected from the group consisting of SEQ ID 5 NOs:9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43 and 45.

12. An isolated nucleic acid encoding a CDR or functional fragment thereof, comprising a nucleotide sequence encoding substantially the amino acid sequence 10 of a CDR encoded by a nucleotide sequence selected from the group consisting of SEQ ID NOs:9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43 and 45.

13. A method of reducing neoplastic cell proliferation, comprising administering to said cell an 15 effective amount of the human monoclonal antibody or functional fragment of claim 1.

14. The method of claim 13, wherein said neoplastic cell is selected from the group consisting of breast cancer, lung cancer and ovarian cancer cells. WO 03/044036 PCT/US02/37134 108

15. A method of detecting a neoplastic cell in a sample, comprising: (a) contacting a sample with the human monoclonal antibody or functional fragment of claim 1; 5 and (b) detecting the specific binding of said human monoclonal antibody or functional fragment to said sample, wherein the presence or increased level compared to a normal cell of said human monoclonal antibody or 10 functional fragment indicates the presence of a neoplastic cell in said sample.

16. The method of claim 15, wherein said neoplastic cell is selected from the group consisting of breast cancer, ovarian cancer and lung cancer cells. 15