AU6883700A

AU6883700A - Melanoma associated antigen (hmw-maa) defined by a monoclonal antibody

Info

Publication number: AU6883700A
Application number: AU68837/00A
Authority: AU
Inventors: Thomas N. Brodin; Lennart G. Ohlsson; M. Jesper Tordsson
Original assignee: Active Biotech AB
Current assignee: Active Biotech AB
Priority date: 1999-08-24
Filing date: 2000-08-22
Publication date: 2001-03-19
Anticipated expiration: 2020-08-22
Also published as: AU762240B2; CA2381099A1; EP1206702A1; SE9902989D0; WO2001014884A1; NZ517132A

Description

WO01/14884 PCT/SEOO/01597 1 Melanoma associated antigen (HMW-MAA) defined by a monoclonal antibody. The present invention is related to an epitope of high molecular weight melanoma associated antigen (HMW MAA), a binding structure and antibodies recognizing target structures of said antigen, pharmaceutical com 5 positions and vaccines as well as methods of using said epitopes and antibodies in diagnostic and therapeutic methods related to human malignant diseases. BACKGROUND OF THE INVENTION The incidence of malignant melanoma has increased 10 more than two-fold during the last 10 years. Most criti cal for a successful treatment of patients suffering from this disease is the point of time for diagnosis, since early stages of malignant melanoma can be efficiently cured with surgery, while the survival is poor for 15 patients advanced into disseminated disease. Melanoma is one of the most immunologically active solid tumors which suggest immunological modalities as adjuvant therapy after surgery of more advanced stage primary tumors. Cytotoxic T cells which specifically lyse tumor cells in 20 vitro and antibodies against tumor associated antigens can be detected in melanoma patients. Single cases demon strating natural regression of the primary tumor suggest involvement of host-mediated mechanisms in melanoma malignancies. 25 Immunotherapeutic approaches include methods meant to augment or alter particularly the T-cell based anti tumor response, and methods which use monoclonal anti bodies raised against melanoma associated antigens for targeted delivery of effector molecules. The high mole 30 cular weight melanoma associated antigen, HMW-MAA, is homogeneously and strongly expressed in a high frequency of melanoma cell lines and tissue specimens and has a limited normal tissue distribution. The antigen repre sents a well documented immunological target molecule for WO01/14884 PCT/SEOO/01597 2 diagnosis and immunotherapy of malignant melanoma using naked antibodies. HMW-MAA (High Molecular Weight Melanoma Associated Antigen) is also named MCSP (Melanoma Chon droitin Sulphate Proteoglycan), MPG (Melanoma Proteo 5 glycan), or NG2 (corresponding antigen in rat; sometimes this term is also used for the human variant of the antigen). The use of anti-HMW-MAA antibodies primarily for the delivery of potent effector molecules may offer an 10 improved therapeutic efficacy, provided that systemic and normal tissue toxicity is not substantially increased due to the introduction of such molecules. As an example, to target an immune attack against tumor cells, tumor-reactive superantigens have been gene 15 tically engineered by the construction of a fusion pro tein between the superantigen SEA and a tumor cell-bind ing Fab fragment. This fusion protein redirects T cell cytotoxicity and promote release of tumoricidal cytokines in the microenvironment of tumor cells recognized by the 20 Fab. To reduce systemic toxicity the superantigen moiety was mutated in a region important for MHC II binding, SEA(D227A). The use of different antibody specificities in the fusion proteins permitted lysis of various cell types, demonstrating the general applicability of super 25 antigens for tumor therapy. The antibody gene to be fused with the SEA gene can be cloned from murine antibody producing hybridoma cells. However, murine antibody Fabs may elicit neutralizing human anti-mouse antibodies when injected in man. The 30 diversity of antibody specificities generated in this way is also limited to those human tumor-associated antigens that demonstrate immunodominance in the mouse. To circum vent these difficulties phage display of antibody frag ments offers a powerful tool to select the specific anti 35 bodies and their encoding genes. Antibody phage libraries from a variety of immunized animal species, and naive and semisynthetic sources of WO01/14884 PCT/SE00/01597 3 1010 and beyond can be examined efficiently by the anti gen driven selection principle. Phage selection methods based on complex antigens such as cell suspensions have been described. Recently, a method to select antibody 5 phage using tumor tissue sections was described. Taken together these methods enable the identification of novel antibody specificities against both in vitro and in vivo expressed antigens. BRIEF SUMMARY OF THE INVENTION 10 The present invention is based on the identification of new epitopes on the high molecular weight melanoma associated antigen. Thus, the invention provides in a first aspect an epitope of high molecular weight melanoma associated 15 antigen (HMW-MAA), said epitope a) having the ability of being specifically blocked by and to specifically block a binding structure which recognizes HMW-MAA, b) being displayed on the surface of human cells but 20 not in cells or tissue from Macaca fascicularis, and c) having a normal tissue distribution pattern which is more restricted than that of other known epitopes of HMW-MAA. In one embodiment of said epitope the binding struc 25 ture is labeled and the binding thereof is inhibitable by an unlabeled form of said binding structure and not by other binding structures, and not inhibiting the binding of other binding structures having other specificities. In another embodiment of said epitope, said binding 30 structure comprises one or more of the CDR (complementa rity determining region) sequences comprising the amino acids number 23-33, 49-55, 88-98 and the amino acids number 158-162, 177-193, 226-240 of the amino acid sequence shown in SEQ ID NO: 2, the amino acids number 35 22-32, 48-54, 87-97 and the amino acids number 157-161, 176-192, 225-239 of the amino acid sequence shown in SEQ WO01/14884 PCT/SE00/01597 4 ID NO: 4, or other binding structures having essemtially the same epitope specificity. By "essentially the same specificity" is meant that the binding of a binding structure structure (e.g. an 5 antibody) to the epitope can only be blocked (binding inhibited) by the same binding structure or other binding structures having essentially the same specificity, and not by other binding structures having other specificites. Furthermore, the binding structure having 10 essentially the same specificity should not block the binding of other binding structures having other binding specificites. The binding of the binding structure (e.g. antibody) and other binding structures of essentially the same 15 specificity could also be blocked (binding inhibited) by a soluble form of the epitope, but not by other epitopes. In a further embodiment of said epitope, said bind ing structure is an antibody. In a still further embodiment of said epitope, said 20 antibody comprises the variable region of a light chain comprising essentially the amino acids number 1-109 of the amino acid sequence shown in SEQ ID NO: 2, and the variable region of a heavy chain comprising essentially the amino acids number 128-251 of the amino acid sequence 25 shown in SEQ ID NO: 2. In yet another embodiment of said epitope, said antibody comprises the variable region of a light chain comprising essentially the amino acids number 1-108 of the amino acid sequence shown in SEQ ID NO: 4, and the 30 variable region of a heavy chain comprising essentially the amino acids number 127-250 of the amino acid sequence shown in SEQ ID NO: 4. In another aspect of the invention, a further char acteristic feature of said epitope is that it is display 35 ed less in human uterus and kidneys than other epitopes of the HMW-MAA.

WO 01/14884 PCT/SEOO/01597 5 In a further aspect of the invention, there is pro vided an anti-idiotype of said epitope, which anti-idio type is specifically blocked by and specifically blocks a binding structure having essemtially the same binding 5 specificity for said epitope. In another aspect of the invention, there is provid ed a vaccine composition comprising as an active principle an epitope as defined above, or an anti idiotype of said epitope as defined above. 10 A further aspect of the invention relates to a bind ing structure which recognizes an epitope of the high molecular weight melanoma associated antigen (HMW-MAA) and is of an organic-chemical nature. Examples of such binding structures are antibodies 15 and so called peptidomimetics. Said peptidomimetics are small organic-chemical drug-like molecules which can be derived from a polypetide, such as an antibody, by determination of the three-dimensional structure of e.g.the CDR loops in an antibody combining site. By such 20 an elucidation of the bioactive conformation of a peptide loop in an immunoglobulin VL domain, so-called peptidomimetics based on organic-chemical molecules can be identified/constructed (6, 7, 8). A still further aspect of the invention is related 25 to an antibody having a binding structure which recognizes a target structure of high molecular weight melanoma associated antigen (HMW-MAA), and which comprises a) the CDR sequences in the light chain comprising 30 essentially the amino acids number 23-33 (CDR1), 49-55 (CDR2) and 88-98 (CDR3) of the amino acid sequence shown in SEQ ID No. 2, and the CDR sequences in the heavy chain comprising essentially the amino acids number 158-162 (CDR1), 177-193 (CDR2) and 226-240 (CDR3) of the amino 35 acid sequence shown in SEQ ID NO: 2, or b) the CDR sequences in the light chain comprising essentially the amino acids number 22-32 (CDR1), 48-54 WO01/14884 PCT/SEOO/01597 6 (CDR2), 87-97 (CDR3) of the amino acid sequence shown in SEQ ID NO: 4, and the CDR sequences in the heavy chain comprising essentially the amino acids number 157-161 (CDR1), 176-192 (CDR2), 225-239 (CDR3) of the amino acid 5 sequence shown in SEQ ID NO: 4, or other binding structures having essemtially the same epitope specificity. In an embodiment thereof, said antibody is phage selected. 10 In another embodiment of said antibody the CDR sequences are of Macaca fascicularis origin. In another embodiment of said antibody, the CDR sequences have an identity of at least 89% to corre sponding CDR sequences of human origin. 15 In another embodiment said antibody has low immuno genicity or non-immunogenicity in humans. A still further embodiment of said antibody is an antibody which has been derivatised by genetically linking to other polypeptides, or by chemical conjugation 20 to organic or non-organic chemical molecules, or by oligo- or multimerisation. Also contemplated is an antibody, which is genetically linked or chemically conjugated to cytotoxic polypeptides or to cytotoxic organic or non-organic chemical molecules, or to biolo 25 gically active molecules, or to immune activating mole cules. Said antibody may also be changed to increase the affinity thereof, or to increase the production yield thereof, or to influence the pharmacokinetic properties 30 thereof, or to give new pharmacokinetic properties thereto. In a further embodiment, said antibody is labeled and the binding thereof is inhibitable by an unlabeled form of said antibody and not by other binding struc 35 tures, and not inhibiting the binding of other binding structures having other specificities.

WO01/14884 PCT/SE00/01597 7 In one of its aspects, the present invention relates to a pharmaceutical composition comprising as an active principle an antibody as defined above. In a further aspect, the invention relates to a 5 method of in vitro histopathological diagnosis and prog nosis of human malignant desease, in which a sample is contacted with an antibody as defined above and an indi cator. According to some embodiments, said method comprises 10 tumor typing, tumor screening, tumor diagnosis and prognosis, or monitoring premalignant conditions. In a still further aspect, the present invention relates to a method for in vitro diagnosis and prognosis of human malignant disease, whereby concentrations in 15 bodily fluids of an antigen, comprising an epitope as defined above or an anti-idiotype of said epitope as defined above, is assayed. In yet another aspect, the present invention relates to a method for in vitro diagnosis and prognosis of human 20 malignant disease, whereby concentrations in bodily fluids of an antibody as defined above is assayed. In still another aspect, the present invention relates to a method for in vitro diagnosis and prognosis of human malignant disease, whereby concentrations in 25 bodily fluids of a complex of a) an antigen comprising an epitope as defined above or an anti-idiotype of said epitope as defined above, and b) an antibody as defined above is assayed. In a further aspect, the present invention relates 30 to a method for in vivo diagnosis and prognosis of human malignant disease, whereby the localisation of an anti body as defined above to tumor deposits in a human sub ject is determined. In a further embodiment, said antibody is adminis 35 tered to the subject before the determination. In another embodiment, said antibody is accumulated in tumor depo sits.

WO01/14884 PCT/SE00/01597 8 In a further embodiment, said method is quantita tive. A still further aspect of the present invention is related to a method for therapy of human malignant dis 5 ease, whereby an antibody as defined above is adminis tered to a human subject. In an embodiment of said method, said antibody has been changed by being genetically linked to molecules giving the combined molecule changed pharmacokinetic properties. 10 In a further embodiment, said antibody has been changed by being derivatised. The invention also encompasses DNA sequences coding for said amino acid sequences, which DNA sequences are shown in SEQ ID NOs: 1 and 3. 15 DETAILED DESCRIPTION OF THE INVENTION The high molecular weight melanoma associated anti gen, HMW-MAA, has been demonstrated to be useful for diagnosis and treatment of malignant melanoma (2,3). Murine monoclonal antibodies (mAbs) generated towards 20 different epitopes of this cell-surface molecule localize efficiently to metastatic lesions in patients with dis seminated disease. The use of a phage selected antibody directed to this molecule for targeting of bacterial superantigens (SAgs) and cytotoxic T cells to melanoma 25 cells has now been explored. Macaca fascicularis monkeys were immunized with a crude suspension of metastatic melanoma. A strong sero logical response towards HMW-MAA was demonstrated, indi cating its role as an immunodominant molecule in the 30 primate. By phage selection using melanoma cells and tissue sections of a recombinant antibody phage library generated from lymph node mRNA, several clones producing monoclonal scFv antibody fragments reacting with HMW-MAA were identified. One of these scFv fragments, denoted 35 K305, was transferred to a Fab format and evaluated as the tumor targeting moiety for superantigen-based immu notherapy. It binds with high affinity to a unique human- WO01/14884 PCT/SEOO/01597 9 specific epitope on the HMW-MAA, and demonstrates more restricted crossreactivity with normal smooth muscle cells than previously described murine mAbs. The K305 Fab was fused to the superantigen Staphylococcal Enterotoxin 5 A (D227A), mutated to reduce its MHC Class II binding affinity and demonstrated redirection of T cell cytotoxi city to melanoma cells in vitro. In SCID mice carrying human melanoma tumors, engraftment of human lymphoid cells followed by the K305 Fab-SEA(D227A) fusion protein, 10 induced HMW-MAA-specific tumor growth reduction. The phage selected K305 antibody demonstrates pro perties including high affinity and selectivity support ing its use as a therapeutic antibody in conjunction with T cell activating superantigens. 15 An antibody phage library derived from a tumor immunized Macaca fascicularis monkey was used for phage selection using melanoma cells and tissue sections. The approach was based on previous reports that primate anti body repertoires have been used successfully for the pro 20 duction of selective immune sera to tumor-associated antigens. Several monoclonal Macaca fascicularis anti bodies were selected which were specific for the HMW-MAA. One of these antibodies, here denoted K305, which together with K461 recognize a unique and selectively 25 expressed epitope on the human HMW-MAA molecule, was fused as a Macaca fascicularis Fab fragment to SEA(D227A). T cell lysis of melanoma cells and in vivo tumor reduction mediated by this antibody are demonstrat ed suggesting its applicability for immunotherapy of 30 malignant melanoma. The high molecular weight melanoma associated anti gen, HMW-MAA, has been demonstrated to be useful for diagnosis and treatment of malignant melanoma. Murine monoclonal antibodies (mAbs) generated towards different 35 epitopes of this cell-surface molecule localize effi ciently to metastatic lesions in patients with dissemi nated disease.

WO01/14884 PCT/SE00/01597 10 Now, the use of a phage selected antibody directed to this molecule for targeting of bacterial superantigens (SAgs) and cytotoxic T cells to melanoma cells has been explored. 5 Macaca fascicularis monkeys were immunized with a crude suspension of metastatic melanoma. A strong sero logical response towards HMW-MAA was demonstrated, indi cating its role as an immunodominant molecule in the primate. 10 By phage selection using melanoma cells and tissue sections of a recombinant antibody phage library generat ed from lymph node mRNA, several clones producing mono clonal scFv antibody fragments reacting with HMW-MAA were identified. One of these scFv fragments, here called 15 K305, was transferred to a Fab format and evaluated as the tumor targeting moiety for superantigen-based immu notherapy. It binds with high affinity to a unique human specific epitope on the HMW-MAA, and demonstrates more restricted crossreactivity with normal smooth muscle 20 cells than previously described murine mAbs. The K305 Fab was fused to the superantigen Staphylococcal Enterotoxin A (D227A), mutated to reduce its MHC Class II binding affinity and demonstrated redirection of T cell cyto toxicity to melanoma cells in vitro. In SCID mice carry 25 ing human melanoma tumors, engraftment of human lymphoid cells followed by the K305 Fab-SEA(D227A) fusion protein, induced HMW-MAA-specific tumor growth reduction. The phage selected K305 antibody demonstrates pro perties including high affinity and selectivity suppor 30 ting its use as a therapeutic antibody in conjunction with T cell activating superantigens. The invention is further illustrated in the follow ing nonlimiting experimental part of the description.

WO01/14884 PCT/SE00/01597 11 EXPERIMENTS Materials and Methods Animals and immunization procedures Severe Combined Immunodeficient (SCID) female mice 5 (C.B-17) were obtained from Bommice, Ry, Denmark and kept under pathogen-free conditions. The animals were accli matized to laboratory conditions (20 ± 1 0 C, 50 + 5% rela tive humidity) for about one week before commencement of the experiments. The animals were subjected to visual 10 inspection on arrival and housed in Macrolone cages (III) with 10 animals in each cage. After randomizing, the animals were used for the experiments. The mice were fed sterile pelleted rodent diet from Special Diets Services, Essex, UK. Sterile water was always available ad libitum. 15 The mice were 8-12 weeks of age when used in this study and the body weight ranged between 17.0-26.0 g with a mean value (±SEM) of 20.6±0.15 g. Macaca fascicularis monkeys (two individuals) were kept and immunized at the Swedish Institute for Infec 20 tious Decease Control (SIIDC), Stockholm. The monkeys were immunized s.c. with 2 ml of a crude mechanical suspension of 43 different metastatic melanoma tissue biopsies in PBS with 10 % normal Macaca fascicularis serum mixed with alun adjuvant. Booster doses were given 25 day 21, 35, and 49. The immune serum was collected day 56. The large number of melanoma samples was used to select for an antibody response to commonly expressed melanoma associated antigens. Tissues and cells 30 Human tumors and normal tissue samples and periphe ral blood mononuclear cells (PBM) from blood-donors were obtained from Lund University Hospital and Malm6 General Hospital, Sweden. The PBM were isolated by density cen trifugation over a Ficoll-Paque cushion (Pharmacia, 35 Uppsala, Sweden). The human melanoma cell lines FM3 and FM55 were a kind gift from Dr. Jesper Zeuthen while FMEX and G361 were from the American Type Tissue Culture WO01/14884 PCT/SE00/01597 12 Colection (ATCC). The cells were cultured in RPMI 1640 medium (Gibco, Middlesex, UK) supplemented with 2 mM L-glutamin (Gibco, Middlesex, UK), 0.01 M HEPES (Bio logical Industries, Israel), 1 mM NaHCO 3 (Biochrom KG, 5 Berlin, Germany), 0.1 mg/ml gentamycin sulfate (Biolo gical Industries, Kibbutz Beit Haemek, Israel), 1 mM sodium pyruvate (JRH Biosciences Industries, USA) and 10% heat inactivated fetal bovine serum (Gibco Middlesex, UK).The cells were repeatedly tested for Mycoplasma con 10 tamination with Gene-Probe (San Diego, CA) Mycoplasma T. C. test, San Diego, CA. Selection of the phagemid library The vector system, immunological tag, and construc tion of a scFv library from an immunized monkey and the 15 selection of this library using melanoma tissue sections has been described previously. In this study tissue sec tions and cell suspensions were used for the phage selec tions. For the tissue-based selections, cryosections were 20 air-dried on slides, fixed in acetone at -20 0 C for 10 min and rehydrated in 20% fetal bovine serum (FBS) in 50 mM Tris pH 7.6/150 mM NaCl (TBS) for lh at room temperature. Library phage suspensions, 1010-1011 in 100 Al 20% FBS, were preabsorbed on sections of normal spleen or small 25 bowel over the night and then transferred to slides with melanoma tissue sections for incubation at 4 0 C over the night. The slides were washed 6x10 min by gentle agita tion in 40 ml TBS in 50 ml Falcon tubes, 2x5 min in 50mM Tris pH 7.6/1M Nacl and 2 times in PBS. Phage were eluted 30 with 300 Al 0.1 M triethylamine for 15 min and neutra lized with 150 pl 1M Tris pH 7.4. The washings and elu tion were performed at room temperature. For cell selections, library phage suspensions, 109-1010 in 100 p1l PBS/1%BSA, were incubated with 3 35 million cells for lh on ice. The cells were washed 3 times including a 10 min incubation period using 2 ml PBS/1% BSA for each wash. The phage were eluted from the WO01/14884 PCT/SE00/01597 13 cells with 50 il 33 pg/ml Genenase for 15 min. The cells were removed after centrifugation and 250 il 1% BSA-PBS added to the supernatant. The phage titers were deter mined by infecting bacteria and counting colony forming 5 units using the bacterial strain E.Coli DH5aF' Phage rescue and culture of the phage library were according to standard techniques. Production of SEA(D227A) fusion proteins An expression plasmid vector, carrying a kanamycin 10 resistance gene and the lac promoter was constructed for cassette insertion of scFv genes to obtain in frame fusion with a flexible spacer of 18 residues followed by the mutant D227A of the superantigen Staphylococcal Enterotoxin A (SEA). A SpeI site 3' to the phoA promoter 15 and the XhoI site following the scFv were used to excise the signal peptide scFv encoding fragment from the phagemid vector. Single colonies of E.coli UL635 trans formed by electroporation with the expression vector containing the inserted scFv-SEA(D227A) genes, were 20 transferred to 96 Micro well plates (Nunc, Denmark) and grown for 17 h at 30 0 C in 2xYT medium supplemented with 70 Ag/ml kanamycin and 2% glucose. Small aliquots (5 l) were transferred to plates with fresh medium containing antibiotics but without glucose, and cultured at 30 0 C for 25 17 h. The Micro well plates were centrifuged and 100 AIl of the supernatants were transferred to new plates with an equal volume/well of 1% BSA. Additional induction using IPTG did not increase production of soluble fusion protein in this system (data not shown). The fusion pro 30 teins were quantified in a sandwich type ELISA using rabbit anti-SEA antibodies as capture and biotinylated anti-SEA Ig as detector antibodies. The production of the K305 Fab-SEA(D227A) and the recombinant control fusion protein C215 Fab-SEA(D227A) was performed as previously 35 described.

WO01/14884 PCT/SEOO/01597 14 EXAMPLES EXAMPLE 1 Immunization of Macaca fascicularis monkeys produced a high serum titer against human melanoma cells and HMW-MAA 5 Two Macaca fascicularis monkeys were immunized day 0, 21, 35, and 49 with alun precipitated crude suspen sions of human melanoma tumors. The immune sera collected on day 56 were tested for reactivity against melanoma cells and against the high molecular weight melanoma 10 associated antigen (HMW-MAA). The immune serum of both animals collected day 56 but not the preimmune serum bound to surface antigens on FM3 human melanoma cells in flow cytometry (equally for the two monkeys). Even at a dilution of 1:40,000 a significant but low shift in 15 flourescence could be detected compared to the preimmune serum at the same dilution (Fig 1A). A double determinant assay was used to analyse serum reactivity against the human HMW-MAA. The immune serum could be diluted 25,000 times before reactivity reached the background level of 20 the preimmune serum (1B). No reactivity was found when the melanoma extract or the capture antibody was omitted or when using a capture antibody (mAb C215) against an epithelial antigen not present in the melanoma extract. HMW-MAA specific scFv antibodies could be identified 25 from a Macaca fascicularis phage library selected against melanoma cells and tissue sections. The scFv antibody phage library was selected against cryosections of human tissues and melanoma cell lines. The yield of library phage (counted as colony forming 30 units) increased substantially after three repeated selections on cells or tissue sections. The yields were as follows for the three conseqetive rounds: i) FM3 melanoma cells, 6.8x10 - 6 , 9.3x10 4 and l.1x10 - ii) M55M1 melanoma cells, 2.0x10

-

, 1.5x10 4 and 5.4x10 -2 iii) 35 adsorption on spleen tissue sections followed by positive selection on melanoma tissue sections, 1.6x10 -6 , 4.6x10 - 7 and 1.3x10-3 iv) adsorption on small bowel sections and WO01/14884 PCT/SE00/01597 15 positive selection on FM3 melanoma cells, 1.3x10 -4 , 3.5x10 -6 and 1.3x10 -3. Soluble protein from about 200 individual scFv anti body clones were produced from each of the selected lib 5 raries from the last two selection rounds. These anti bodies were screened for binding to melanoma tissue sec tions and melanoma cell lines using immunohistochemistry and flow cytometry. A secondary antibody directed against the immunological tag, ATPAKSE, was used for the detec 10 tion in these assays. These screening systems permitted a preliminary evaluation of melanoma reactivity. Positive clones were then recloned in fusion with SEA(D227A). The latter formate permitted both a sufficiently high produc tion yield in culture supernatants and a more sensitive 15 detection on tissue sections using a rabbit antisera against SEA. Alternatively to the recloning of individual scFv:s, an aliquote of the selected library was recloned into the SEA(D227A) vector for a direct screening in this format. Several scFv-SEA(D227A) antibody-fusion proteins 20 including K305 and K461 (tissue selections on spleen and melanoma) were identified which had an immunohistochemi cal staining pattern similar to the pattern described for murine antibodies against the HMW-MAA (see below). All antibodies were identified from the second selection 25 round, demonstrating a lower frequency of binders than the third. EXAMPLE 2 Immunoprecipitaton of a 250 kDa chondroitinase sensitive protein by phage selected antibodies 30 FM3 cells were iodinated by the use of a chloramide method as previously described. Briefly, an amount of 50 pg 1,3,4,6-tetrachloro-3a-diphenylglycoluril desolved in methylene cloride was airdried in a glass tube. FM3 cells, 13 million in 0.5 ml PBS and 0.5 mCi carrier-free 35 125 - were incubated in the tube for 10 min on ice with mild agitation. After two washes in PBS the cells were incubated in KI (50 pg/ml) in PBS, for 20 min on ice to WO01/14884 PCT/SE00/01597 16 reduce unspecific incorporation of 125 - and finally washed again two times in PBS. The iodinated cells were lysed in 2 ml sucrose-NP40 buffer (0.25 M sucrose/50 mM Tris/l.5 mM MgC12 /10 mM KC1/1% NP40/protease inhibitor 5 (Boehringer Mannheim, Germany) for 30 min on ice. The lysate was diluted by adding 15 ml PBS and ultracentri fuged lh at 100 000 g and 4 0 C. For immunoprecipitation 10 Ag primary antibodies were incubated with 1 ml lysate for 1 h followed by 100 Ag rabbit-anti-SEA antibodies 10 which were incubated for 20 min and finally with 100 yil Sepharose protein G beads for 20 minutes. The beads were washed two times with 50mM Tris/500 mM NaCl/0.1% NP40. SDS-PAGE using a 4% polyacrylamide gel and protein Western blotting were performed according to standard 15 methods. Autoradiography using film (Amersham) was per formed for 1 week at -70 0 C. The phage selected antibodies (scFv/Fab -SEA(D227A) fusion proteins) were used for immunoprecipitation using an extract of 125I surface labeled FM3 melanoma cells. 20 The molecular weights of the precipitates were determined by SDS-PAGE. Three antibodies precipitated antigens with a molecular weight which distributed as a smear from a distinct 250 kDa band. Digestion of the precipitate with chondroitinase resulted in a single 250 kDa band with 25 increased intensity. The molecular weight and the sensi tivity to chondroitinase support that the antigens recog nized are identical to the HMW-MAA precipitated in paral lell by the murine mAb 9.2.27 (Fig.2). EXAMPLE 3 30 Epitope mapping of the selected antibodies A Nunc Maxisorb plate (Nunc, Denmark) was coated with 35 .l/well lg/ml mAb 9.2.27 or control antibody o/n 4 0 C. The incubation fluid was removed and non-specific binding sites were blocked lh at room temperature (RT) in 35 3% low fat milk (Semper, Sweden) which was removed before incubation for 2 h at RT with the NP-40 extract of mela noma cells diluted 1/10 in 3% milk. The plate was washed WO01/14884 PCT/SEOO/01597 17 4 times in PBS-0.05% tween 20 (PBST) after this and the following steps. The scFv or Fab-SEA(D227A) fusion pro teins at 30 pl/well diluted to lpg/ml or Macaca fascicu laris serum diluted 1/50-1/100,000 were incubated 1 h RT. 5 Secondary antibody, 1 pg/ml biotinylated rabbit anti-SEA antibodies or rabbit anti human Ig, was incubated for lh RT and finally extavidin-HRP (Sigma, St. Louis, MO) 1/1000 for 30 min RT. The colour reaction was developed using 3.3 -diaminobenzidine (Sigma) according to the 10 instructions of the supplier and read at OD405 nm. The binding of selected antibodies was epitope mapped relative to each other and to murine anti-HMW-MAA mAbs. The antibodies, Fab K305 and scFv K461 fused to SEA(D227A) were produced by fermentation and purified by 15 affinity chromatography using immobilized polyclonal rabbit anti-SEA Ig (results not shown). The yield from 1 liter fermentor cultures of the fusion proteins were 35 mg (K305) and 118 mg (K461), respectively. HMW-MAA pro teoglycan in the melanoma extract was captured with mAb 20 9.2.27. The binding of biotinylated fusion proteins to the captured antigen was competed with using non-biotiny lated fusion proteins. Alternatively, K305 Fab-SEA(D227A) was used as the capture reagent and murine antibodies were used to inhibit detection with the biotinylated 25 fusion proteins. K305 inhibited K461 (and vice versa) but not another antibody, K458. The murine anti-HMW-MAA mAbs TP41.2 or 763.74 did not inhibit K305 and K461. The K458 antibody was found to be inhibited by the mAb TP41.2 which supports its recognition of an overlapping epitope 30 on the HMW-MAA. Thus, in this assay binding to two dif ferent epitopes on the HMW-MAA could be verified, defined by K305/K461 and K458 (Fig.3). EXAMPLE 4 Affinity determination on cells and tissue sections of 35 K305 Fab-SEA(D227A) The K305 Fab-SEA(D227A) fusion protein was iodinated with 2 mCi Na 125I to 80 pg sample according to the iodo- WO01/14884 PCT/SE00/01597 18 gen method. Cryosections (8mm) were air-dried on multi well slides, fixed in acetone at -20 0 C for 10 min and rehydrated in 20% FBS. Iodinated K305 Fab-SEA(D227A) in 20% FBS was added in two-fold dilution series to the sec 5 tions and incubated lh. The sections were washed four times with TBS and dried before quantifying bound radio activity of individual wells in a gamma counter. For determination of the affinity using cells, 30 000 FM3 cells/sample were incubated with iodinated K305 Fab 10 SEA(D227A) at 100 pl/tube in a series of two-fold dilu tion in 1% BSA for lh and then washed three times in PBS before measuring bound activity. Scatchard plots were constructed using the measured values of bound and added radioactivity. 15 Iodinated antibody K305 Fab-SEA(D227A) was added in dilution series to FM3 cells, or tissue sections of human malignant melanomas and small bowel (to determine smooth muscle reactivity). Fig. 4 demonstrates the binding to melanoma tissue. 20 The affinity determined in this experiment was 1.6 nm and the number of binding sites per melanoma tissue section (8 ym thick and approximately 3x4 mm wide) was 4.7x10 9 . The results from three additional experiments per formed on melanoma tissue sections were 0.67, 0.93 and 25 1.4 nM with 2.2x10 , 2.9x10 9 and 6.4x109 sites per sec tion. The affinity was also measured on small bowel tis sue sections (smooth muscle reactivity) in four experi ments, which resulted in 0.82, 0.93, 1.1 and 1.3 nM. In one experiment the cell based system was used giving an 30 affinity of 1.6 nM and 300 000 binding sites per cell. EXAMPLE 5 Tumor and normal tissue reactivity of the fusion proteins All tissues were snap-frozen in isopentane pre chilled in liquid nitrogen. Cryosections (8mm) were air 35 dried on slides, fixed in acetone at -20 0 C for 10 min and rehydrated in 20% foetal bovine serum (FBS). Endogenous biotin was blocked with avidin for 15 min and then with WO01/14884 PCT/SE00/01597 19 biotin for 15 min, both of which was deluted 1/6 (Vector Laboratories, Burlingame, CA). If not otherwise speci fied, primary antibodies were incubated in a concentra tion of 5 yg/ml for scFv-SEA(D227A) fusion proteins and 5 1g/ml for mouse monoclonal antibodies. Affinity purified and biotinylated rabbit anti-SEA antibodies, 5 Ag/ml, and biotinylated rabbit anti-mouse immunoglobulins (DAKO A/S, Denmark), 1/300 of the commer cial stock solution, were incubated 30 min followed by 10 StreptABComplex HRP (DAKO A/S, Denmark), 1/110 of the commercial stock solution, diluted in 50 mM Tris pH 7.6 (TBS), for 30 min. Between all steps the sections were washed 3 times in TBS. Antibodies, avidin and biotin were all diluted in 20% FBS (fetal bovine serum) in TBS. The 15 staining reaction was developed for 8 min in 0.5 mg/ml 3,3 -diaminobenzidine tetrahydrochloride (Sigma) dissolv ed in Tris pH 7.6 with 0.01 percent H 2 0 2 . After 10 min counterstaining in 0.5 % methyl green, the slides were rinsed for 10 min in tap water and gradually dehydrated 20 in 70-99 % ethanol and xylene before mounting in DPX medium (Sigma). The mouse antibodies specific for HMW-MAA, TP41.2 and 763.74, were a kind gift from Dr. Soldano Ferrone and mAb 9.2.27 was obtained from PharMingen, San Diego, CA. 25 The phage selected antibody K305 and the murine anti-HMW-MAA antibody mAb 9.2.27 strongly and homogen ously stained melanoma cells in seven of eight biopsies of human malignant melanomas surgically removed from different patients. The predominant normal tissue reacti 30 vity found was a weak staining of smooth muscle, e.g in the gut, and a strong staining of a fraction of blood vessel walls (the smooth muscle layer) in all tissues. Normal tissue reactivity also included a subpopulation basal cells of the epidermis. The K305 antibody showed 35 very weak staining of the uterine smooth muscle cells as compared to mAb 9.2.27. Some blood vessels in the uterus were strongly stained with both antibodies. The murine WO01/14884 PCT/SEOO/01597 20 mAb 9.2.27 also stained cells in the glomerulus, probably mesangial cells, in contrast to K305 (Fig. 5). Renal blood vessels but not renal tubules were stained by mAb 9.2.27 and K305 (verified by staining of smooth muscle 5 actine and cytokeratin). None of the phage selected anti bodies, K305, and K461 bound to Macaca fascicularis tis sues expressing HMW-MAA in contrast to mAb 9.2.27 that demonstrated clear staining of this tissue (not shown). EXAMPLE 6 10 K305 Fab-SEA(D227A) mediated superantigen-antibody dependent cellular cytotoxicity Redirected T cell cytotoxicity (superantigen anti body dependent cellular cytotoxicity, SADCC) was measured in a standard 4 h chromium-release assay employing 15 51 Cr-labeled FM3 or G361 cells as target cells and human T cells as effector cells as previously described. Per cent specific lysis was calculated as 100 x cpm experimental release - cpm background release 20 cpm total release - cpm background release Two human melanoma target cell lines were used to demonstrate redirection of T cell cytotoxicity to tumor cells by the K305 Fab-SEA(D227A) fusion protein 25 (Fig. 6A). Dose dependent cellular cytotoxicity could also be achieved with SEA(D227A) alone or fused to irre levant antibodies (due to a weak interaction with MHC Class II expressed by some melanoma cell lines, e.g. the FM3 line). However, this cytotoxicity was demonstrated at 30 a hundred-fold higher concentration and with a lower plateau value compared to the antibody targeted K305 Fab SEA(D227A) cytotoxicity (Fig. 6A). The targeted cytotoxi city was as expected independent of superantigen presen tation on MHC class II, since specific cytotoxicity was 35 also demonstrated for MHC II negative G361 human melanoma cells (Fig. 6B).

WO01/14884 PCT/SE00/01597 21 EXAMPLE 7 Immunotherapy of established FM3 tumors in SCID mice with K305 Fab-SEA(D227A) SCID mice were injected I.P. with 3 x 106 FM3 cells 5 in 0.2 ml vehicle (PBS-1%Balb/c mouse serum). The mice were injected I.P. with 3 x 106 PBM in 0.2 ml vehicle 4 days later. One to two hours after injection of lymphoid cells all mice were injected I.V. with 100 ug test sub stance in 0.2 ml vehicle or vehicle alone. Two additional 10 I.V. injections of the test substance were given at three days intervals. The mice were sacrificed by cervical dislocation on day 49 and the number of tumors and the tumor mass was determined. Tumors with a weight of less than 5 mg were estimated as 2 mg, tumors with a weight of 15 more than 5 mg and less than 10 mg as 7 mg and tumors larger than 10 mg with the actual weight. All tumors larger than 1 mg were counted. Each treatment cohort con tained five to seven mice to permit comparison with other treatment cohorts treated simultaneously with the same 20 batch of effector cells. Statistical significance was determined by the Mann-Whitney U test using the program Sigma Stat. The ability of the fusion protein K305 Fab SEA(D227A) to mediate therapy of FM3 tumors grown intra 25 peritoneally (I.P.) in SCID mice for 4 days was investi gated. The therapy was initiated simultaneously to inocu lation of human PBM i.p. (humanized SCID model) and com pared to vehicle (PBS) and to a control fusion protein, C215 Fab-SEA(D227A), with an irrelevant Fab moiety. A 30 dramatic and statistically significant reduction of the tumor mass and number was achieved by the K305 Fab SEA(D227A) fusion protein as compared to the PBS control. Importantly, the effect was significantly higher also compared to the irrelevant fusion protein (Fig. 7), thus 35 demonstrating the dependence of specific antibody-mediat ed targeting for the therapeutic effect.

WO01/14884 PCT/SE00/01597 22 EXAMPLE 8 Sequencing of antibodies Briefly, 1.0 pmole/reaction plasmid DNA and 10 pmol/reaction primer DNA were used for automatic sequenc 5 ing on a PerkinElmer/Applied Biosystem model 373 A. The primers annealed in the lac promoter and in the SEAD(227A) gene, 39 bp upstreem of the Spel and 94 bp downstream of the SalI insertion site of the scFv gene. The annealing temperatures were 60.1 0 C and 64.9 0 C, 10 respectively. K305 and K461 scFv were sequenced at Pharmacia and Upjohn AB, Stockholm, Sweden, using the same sequencing equipment. The antibodies fall into groups based on the sequence similarity, K305 and K461 belonging to the same 15 group. Comparisons with the closest human germline sequence of the light (lambda) and heavy chain demon strated more than 89 % sequence identity on the nucleic acid level (Table 1). All of the antibodies had heavy chains with close homology to the VH3 family of human 20 germline segments. Human lambda 3 germline family homo logues were represented by the K305 and K461 antibodies. Table 1. Identity to human germline segments Antibody VH (% identity) VH family V (% identity) V4 family K305 DP-47 (89.1) VH3 humlv 318 (91.7) V?3 K461 DP-58 (90.1) " humlv 318 (90.3) V3 The epitope mapping data which suggest that K305 and K461 bind to overlapping epitopes, are supported by the 25 high sequence similarity between the two antibodies (see SEQ ID NO: 2 och SEQ ID NO: 4). DISCUSSION The identification and characterization of phage se lected primate antibodies directed to two distinct epi 30 topes on HMW-MAA is disclosed. There is also described the therapeutic use of a selected antibody for super- WO01/14884 PCT/SE00/01597 23 antigen and T cell dependent growth inhibition of mela noma in humanized SCID mice. Commonly, tumor associated antigens are defined by murine monoclonal antibodies. However, therapeutic use of 5 murine mAbs frequently leads to the development of human anti-mouse antibodies (HAMA) in patients, which poten tially neutralize the effector functions and enhance serum clearance rate. Human monoclonal antibodies should ideally be used, but have by various reasons been diffi 10 cult to produce by the hybridoma technique. Chimeric antibodies with human constant domains and antibodies with murine CDRs grafted on human Vh/Vl frameworks often demonstrate reduced neutralizing HAMA responses. However responses to V region determinants have been demon 15 strated. Another approach is to use non-human primates for the generation of therapeutic antibodies (4). Primate antibodies should not elicit a response of neutralizing antibodies in man due to their high homology with human. 20 Specific immune responses to human tumor-associated anti gens in non-human primates could potentially be less restricted than in man and the reactivity to common spe cies specific tissue antigens should be much reduced as compared to murine responses. The present identification 25 of primate antibodies to human specific (Macaca fascicu laris negative) HMW-MAA epitopes including the clone K305 demonstrating a restricted reactivity with uterine smooth muscle and kidney glomeruli as compared to murine mAbs, supports this hypothesis. To be able to identify novel 30 antibodies to epitopes on cell surface expressed tumor associated antigens, intact cells or suspensions of cells and tissues should preferentially be used as the immu nogen. A suspension of pooled metastatic melanoma tissue in alun adjuvant were used to immunize the Macaca fasci 35 cularis monkeys. The immune serum was found to be strong ly reactive both with antigens expressed on the surface of human melanoma cells and almost equally strongly and WO01/14884 PCT/SEOO/01597 24 specifically with the human HMW-MAA. This suggests that the HMW-MAA in the melanoma suspension was strongly immu nodominant to the primate, supported by the finding of high anti-HMW MAA serum anti-body titer in a chimpanzee 5 immunized with purified HMW-MAA. The scFv antibody phage library was selected on human melanoma tissue sections and melanoma cell lines to generate enriched libraries against both the authentic in vivo phenotype and against cell surface expressed anti 10 gens. A major advantage of this technology compared to the hybridoma technology is that the identification of novel antibodies and the cloning of their genes is simul taneously performed. This selection of genes encoding tumor reacting antibodies permitted an immediate con 15 struction of superantigen fusion proteins to be evaluated for tumor targeting properties and therapeutic efficacy. The scFv format was suitable for the selection procedure and for the evaluation of antibody specificities by screening on tissue sections and cell lines. However, due 20 to partial dimer formation of some library selected scFv:s (results not shown) and the lower stability of scFv fragments, Fab-SEA fusion proteins of selected clones were constructed, fermentor cultured and purified. These were needed for detailed immunohistochemical stu 25 dies, for estimation of true binding affinity and for in vivo experiments. The K305 Fab-SEA(D227A) fusion protein demonstrated a high binding affinity. The number of binding sites per cell differed less than two-fold from a number of Fab's 30 generated from papain cleaved murine mAbs against the HMW-MAA. Affinity for tissue expressed antigen in mela noma biopsy sections was similar. The number of available binding sites in a tissue section could also be calcu lated, the result of which could be used for estimation 35 of the in vivo expression level of a therapeutic target. The targeting efficacy of antibodies to solid tumors have WO01/14884 PCT/SE00/01597 25 been suggested to be dependent of the antibody affinity and to peak between defined affinity thresholds. An antibody to the c-erb B-2 oncogene product with nanomolar affinity (l.0xl0 -9 ) had a higher tumor to blood 5 ratio and tumor retention as compared to a very high affinity (l.5x10 -11 , reduced off-rate) and low affinity (3.2x10 -7 ) variant with the same specificity (Adams GP, personal communication). It is therefore suggested that the selected K305 antibody should be within the optimal 10 affinity range for targeting to the tumor. Higher affi nity variants to K305 (K461) may have existed and been selected from the library, indicated by the dose-inhibi tion curves and possibly by differences in the strenght of signal in the immunoprecipitation data (Fig. 2,3). 15 However, for scFv's, dimer formation varies and divalent formation could influence overall affinity. It is suggested that the selected scFv:s are clus tered in three distinct HMW-MAA epitope specificities, of which two can be clearly linked to HMW-MAA reactivity. 20 K305 and K461 should recognize a previously not described epitope specificity, since they could not be blocked by existing mAbs, and their pattern of distribution of tis sue reactivity, e.g. between man and primate, was diffe rent. 25 The K305 Fab-SEA(D227A) fusion protein had a very weak reactivity (if any) to human glomeruli and human uterine smooth muscle cells. This pattern is seen over a wide concentration range and is not due to a different sensitivity of the two detection formats. The K305 epi 30 tope may therefore be expressed on a HMW-MAA variant with a more restricted normal tissue distribution. The exis tence of heterogeneity of the HMW-MAA molecule has been discussed previously (5). The in vitro inhibition of melanoma cell growth and 35 adhesion to endothelial basement membranes suggested a therapeutic potential for naked anti-HMW-MAA mAbs. How ever, in clinical trials using radiolabeled mAb only WO01/14884 PCT/SE00/01597 26 minor and transient therapeutic results were observed, despite high levels of specific uptake. Antibody mediated targeting of more potent effector molecules should then be considered. 5 Superantigens have the capacity to induce T cell dependent tumor cell killing when targeted by antibodies to the tumor site. Thus, this concept benefit from the use of antibody defined epitopes on tumor associated antigens for the retargeting of the T cell arm of the 10 immune system proven to have the capability to erradicate tumor cells. For naked and radiolabeled anti-HMW-MAA mAbs little toxicity in normal tissues expressing the antigen has been reported in patients even at high antibody protein 15 and radioactivity doses (2). More potent effector mole cules such as the superantigen SEA(D227A) are needed to demonstrate the accessibility of this molecule in normal tissues and the potential to improve therapeutic efficacy without accompanying targeted organ related toxicity. 20 This study demonstrates that immune antibody reper toires of non-human primates can comprise monoclonal antibody specificities potentially useful for tumor the rapy. Furthermore, phage selection of this repertoire using whole tumor cells and tissue sections followed by 25 efficient derivation of potent effector fusion proteins can be efficiently used to identify and isolate tumor therapeutic antibodies.

WO01/14884 PCT/SEOO/01597 27 REFERENCES 1. Pluschke G, Vanek M, Evans A, Dittmar T, Schmid P, Itin P, Filardo EJ, Reisfeld RA (1996) Molecular clon ing of a human melanoma-associated chondroitin sulfate 5 proteoglycan. Proc Natl Acad Sci 93:9710 2. Siccardi AG, Burragi GL, Natali PG, Scassellati GA, Viale G, Ferrone S (1990) European multicentre study on melanoma immunoscintigraphy by means of 99 mTc-labelled monoclonal antibody fragments, the european multicentre 10 group. Eur J Nucl Med 16:317 3. Mittelman A, Chen GZ, Wong GY, Liu C, Hirai S, Ferrone S (1995) Human high moelcular weight-melanoma associated antigen mimicry by mouse anti-idiotypic mono clonal antibody MK2-23: modulation of the immunogenicity 15 in patients with malignant melanoma. Clin Cancer Res 1:705-13 4. Logdberg L, Kaplan E, Drelich M, Harfeldt E, Gunn H, Ehrlich P, Dottavio D, Lake P, Ostberg L (1995) Primate antibodies to components of the human immune 20 system. J Med Primal. 23:285-97 5. Ziai MR, Imberti L, Nicotra MR, Badaracco G, Segatto O, Natali PG, Ferrone S (1987) Analysis with monoclonal antibodies of the molecular and cellular heterogeneity of human high molecular weight melanoma 25 associated antigen. Cancer Res 67:2474 6. Zhao B, Larry R, Helms LR, DesJarlais RL, Abdel Meduid SS, Wetzel R (1995) A paradigm for drug discovery using a conformation from the crystal structure of a presentation scaffold. Nature Structural Biology, 2:1131 30 1137 7. Dougall WC, Petersen NC, Greene MI (1994) Antibody-structure-based design of pharmacological agents. TIBTECH 12:372-379 8. Kovari LC, Momany C, Rossman MG (1995) The use of 35 antibody fragments for crystallization and structure determinations. Structure 3:1291-1293.

WO01/14884 PCT/SE00/01597 28 LEGENDS TO FIGURES Fig. 1. A) Flow cytometry signal demonstrating strong reactivity against human FM3 melanoma cells of the immune Macaca fascicularis serum collected at day 56 (0) 5 compared to the preimmune serum (0). B) Specific serum reactivity against the HMW-MAA in a double determinant assay. The HMW-MAA was captured from a NP40 extract of FM3 melanoma cells with the monoclonal antibody 9.2.27 and then detected with the immune serum (D) or the 10 preimmune serum (l). As further negative controls either the irrelevant murine mAb C215 (A) or no antibody (0) was used for capturing or the melanoma extract was omitted (X). Fig. 2. Immunoprecipitation of HMW-MAA with phage 15 selected specificities from an 1% NP40 extract of FM3 melanoma cells, surface labeled with 125I. The molecular weight of the precipitates were determined by SDS-PAGE using a 4 % gel under reducing conditions. A) Precipitation with the K305 Fab-SEA(D227A) fusion 20 protein. Enzymatic cleavage with chondroitinase (+) era dicated the high molecular weight product and increased the amount of the 250 kDa band. Precipitates using the murine mAb 9.2.27 recognizing HMW-MAA or a murine nega tive control mAb C2 are shown for comparison. B) The 25 SEA(D227A) fusion proteins K305, K460 and K461 precipi tated proteins sensitive to chondroitinase, of identical molecular weight. Fig. 3. Epitope mapping of scFv/Fab-SEA(D227A) fusion proteins. The HMW-MAA was captured by mAb 9.2.27 30 from the melanoma extract and detected with biotinylated scFv/Fab-SEA(D227A) fusion proteins. Binding to HMW-MAA of the biotinylated fusion proteins of A) K305 Fab B) K461 scFv and C) K458 scFv inhibited by the non-biotiny lated fusion proteins K305 Fab (0), K461 scFv (@), K458 35 scFv (A) or with the murine anti-HMW-MAA mAbs TP41.2 (0) or 763.74 (M) . The fusion protein K305 Fab and K461 scFv inhibited each other and K458 scFv was inhibited by the WO01/14884 PCT/SEOO/01597 29 murine mAb TP41.2. Neither of the fusion proteins were inhibited by the mAb 763.74. No specific signal was achieved using the biotinylated negative control C215 Fab-SEA(D227A) reagent (not shown). 5 Fig. 4. Scatchard plot using tissue sections of metastatic melanoma for the determination of the antibody affinity. The determined affinity of the K305 Fab SEA(D227A) antibody in this experiment was 1.6 nM and the number of binding sites per section (3x4 mm) was 4.7x10 9 . 10 Fig. 5. Immunoperoxidase staining of human tissues by the HMW-MAA specific fusion protein K305 Fab SEA(D227A), in A-E, and by mAb 9.2.27, in F-H. A) Strong and homogenous staining of a malignant melanoma biopsy. B) Staining of the glomerular arteriol (AR) but not with 15 in the glomerulus in contrast to the staining of glomeru lus by mAb 9.2.27 in F). K305 Fab-SEA(D227A) staining of the uterine blood vessels but no/weak staining of the uterine smooth muscle in both high and low concentration, 20 pg/ml in C), and 0.08pg/ml in D). MAb 9.2.27 staining 20 of uterine smooth muscle and blood vessels, 0.02 pg/ml in G) and 0.005 pg/ml in H). The bar in D) is 100 pm. Fig. 6. Superantigen antibody dependent cellular cytotoxicity (SADCC) against human melanoma cells . A SEA reactive T cell line established from human PBL stimulat 25 ed with SEA wild type (12 pM) was added to the melanoma cells in an effector to target ratio of 30:1. a) Lysis of MHCII+/HMW-MAA+/C215-/C242 FM3 melanoma cells with K305 Fab-SEA(D227A) (A) was efficient in a 100-fold lower concentration and reached a higher maximum 30 cytotoxicity than SEA(D227A) (X) alone or the irrelevant fusion proteins C215 Fab-SEA(D227A) (40) and C242 Fab SEA(D227A) (0). b) The superantigen mediated lysis directed by the antigen specific antibody moiety was MHCII independent, since MHCII-/HMW-MAA+/C215 G361 human 35 melanoma could be lysed by K305 Fab-SEA(D227A) (A) but not by the fusion protein C215 Fab-SEA(D227A) (0).

WO01/14884 PCT/SE00/01597 30 Fig. 7. Statistically significant growth reduction of established FM3 tumors grown i SCID mice treated with K305 Fab-SEA(D227A). SCID mice with FM3 tumors grown I.P for 4 days were grafted with human PBM and 2h later 5 treated by I.V. injection of 100 ug of the test sub stance. The test substance treatment was given three times with three days intervals. The tumor weight and number were cacultated day 49. A: PBM + vehicle (1% Balb/c mouse serum in PBS), B: PBM + K305 Fab-SEA(D227A), 10 C: PBM + control C215 Fab-SEA(D227A).

Claims

1. An epitope of high molecular weight melanoma 5 associated antigen (HMW-MAA), said epitope a) having the ability of being specifically blocked by and to specifically block a binding structure which recognizes said epitope on HMW-MAA, b) being displayed on the surface of human cells but 10 not cells from Macaca fascicularis, and c) having a normal tissue distribution pattern which is more restricted than that of other known epitopes of HMW-MAA.

2. An epitope according to claim 1, wherein the 15 binding structure is labeled and the binding thereof is inhibitable by an unlabeled form of said binding struc ture and not by other binding structures, and not inhi biting the binding of other binding structures having other specificities. 20 3. An epitope according to claim 1, wherein said binding structure comprises one or more of the CDR (complementarity determining region) sequences comprising the amino acids number 23-33, 49-55, 88-98 and the amino acids number 158-162, 177-193, 226-240 of the amino acid 25 sequence shown in SEQ ID NO: 2, the amino acids number

22-32, 48-54, 87-97 and the amino acids number 157-161,

176-192, 225-239 of the amino acid sequence shown in SEQ ID NO: 4, or other binding structures having essemtially the same epitope specificity. 30 4. An epitope according to claim 1, wherein said binding structure is an antibody. 5. An epitope according to claim 4, wherein said antibody comprises the variable region of a light chain comprising essentially the amino acids number 1-109 of 35 the amino acid sequence shown in SEQ ID NO: 2, and the variable region of a heavy chain comprising essentially WO01/14884 PCT/SE00/01597 32 the amino acids number 128-251 of the amino acid sequence shown in SEQ ID NO: 2. 6. An epitope according to claim 4, wherein said antibody comprises the variable region of a light chain 5 comprising essentially the amino acids number 1-108 of the amino acid sequence shown in SEQ ID NO: 4, and the variable region of a heavy chain comprising essentially the amino acids number 127-250 of the amino acid sequence shown in SEQ ID NO: 4. 10 7. An epitope according to anyone of claims 1-6, which epitope is displayed less in human uterus and kidneys than other epitopes of the HMW-MAA. 8. An anti-idiotype of an epitope as defined in any one of claims 1-7, which anti-idiotype is specifically 15 blocked by and specifically blocks a binding structure having essemtially the same binding specificity for said epitope. 9. A vaccine composition comprising as an active principle an epitope as defined in anyone of claims 1-7, 20 or an anti-idiotype of said epitope as defined in claim 8. 10. A binding structure which recognizes an epitope of the high molecular weight melanoma associated antigen (HMW-MAA) and is of organic-chemical nature. 25 11. A binding structure according to claim 10, which is based on a structure of one or more of the CDR (complementarity determining region) sequences comprising the amino acids number 23-33, 49-55, 88-98 and the amino acids number 158-162, 177-193, 226-240 of the amino acid 30 sequence shown in SEQ ID NO: 2, the amino acids number 22-32, 48-54, 87-97 and the amino acids number 157-161, 176-192, 225-239 of the amino acid sequence shown in SEQ ID NO: 4. 12. An antibody having a binding structure which 35 recognizes a target structure of high molecular weight melanoma associated antigen (HMW-MAA), and which comprises WO01/14884 PCT/SEOO/01597 33 a) the CDR sequences in the light chain comprising essentially the amino acids number 23-33 (CDR1), 49-55 (CDR2) and 88-98 (CDR3) of the amino acid sequence shown in SEQ ID No. 2, and the CDR sequences in the heavy chain 5 comprising essentially the amino acids number 158-162 (CDR1), 177-193 (CDR2) and 226-240 (CDR3) of the amino acid sequence shown in SEQ ID NO: 2, or b) the CDR sequences in the light chain comprising essentially the amino acids number 22-32 (CDR1), 48-54 10 (CDR2), 87-97 (CDR3) of the amino acid sequence shown in SEQ ID NO: 4, and the CDR sequences in the heavy chain comprising essentially the amino acids number 157-161 (CDR1), 176-192 (CDR2), 225-239 (CDR3) of the amino acid sequence shown in SEQ ID NO: 4, or other binding 15 structures having essemtially the same epitope specificity. 13. An antibody according to claim 12, which is phage selected. 14. An antibody according to claim 12, wherein the 20 sequences are of Macaca fascicularis origin. 15. An antibody according to claim 12, wherein the sequences have an identity of at least 89% to correspond ing sequences of human origin. 16. An antibody according to claim 12, which has low 25 immunogenicity or non-immunogenicity in humans. 17. An antibody according to claim 12, which has been derivatised by genetically linking to other polypeptides, or by chemical conjugation to organic or non-organic chemical molecules, or by oligo- or 30 multimerisation. 18. An antibody according to claim 12, which is genetically linked or chemically conjugated to cytotoxic polypeptides or to cytotoxic organic or non-organic chemical molecules. 35 19. An antibody according to claim 12, which is genetically linked or chemically conjugated to biologi cally active molecules. WO01/14884 PCT/SE00/01597 34 20. An antibody according to claim 12, which is genetically linked or chemically conjugated to immune activating molecules. 21. An antibody according to claim 12, which has 5 been changed to increase the affinity thereof. 22. An antibody according to claim 12, which has been changed to increase the production yield thereof. 23. An antibody according to claim 12, which has been changed to influence the pharmacokinetic properties 10 thereof. 24. An antibody according to claim 12, which has been changed to give new pharmacokinetic properties thereto. 25. An antibody according to claim 12, which is 15 labeled and the binding thereof is inhibitable by an unlabeled form of said antibody and not by other binding structures, and not inhibiting the binding of other bind ing structures having other specificities. 26. A pharmaceutical composition comprising as an 20 active principle an antibody as defined in anyone of claims 12-25. 27. A method of in vitro histopathological diagnosis and prognosis of human malignant desease, whereby a sample is contacted with an antibody as defined in anyone 25 of claims 12-25 and an indicator. 28. A method according to claim 27, which method comprises tumor typing. 29. A method according to claim 27, which method comprises tumor screening. 30 30. A method according to claim 27, which method comprises tumor diagnosis and prognosis. 31. A method according to claim 27, which method comprises monitoring premalignant conditions. 32. A method for in vitro diagnosis and prognosis of 35 human malignant disease, whereby concentrations in bodily fluids of an antigen comprising an epitope, as defined in WO01/14884 PCT/SEOO/01597 35 anyone of claims 1-7, or an anti-idiotype of said epitope, as defined in claim 8, is assayed. 33. A method for in vitro diagnosis and prognosis of human malignant disease, whereby concentrations in bodily 5 fluids of an antibody as defined in anyone of claims 12 25 is assayed. 34. A method for in vitro diagnosis and prognosis of human malignant disease, whereby concentrations in bodily fluids of a complex of a) an antigen comprising an epi 10 tope, as defined in anyone of claims 1-7, or an anti idiotype of said epitope, as defined in claim 8, and b) an antibody, as defined in anyone of claims 12-25, is assayed. 35. A method for in vivo diagnosis and prognosis of 15 human malignant disease, whereby the localisation of an antibody, as defined in anyone of claims 12-25, to tumor deposits in a human subject is determined. 36. A method according to claim 35, whereby said antibody is administered to the subject before the deter 20 mination. 37. A method according to claim 35, whereby said antibody is accumulated in tumor deposits. 38. A method according to any one of claims 35-37, which is quantitative. 25 39. A method for therapy of human malignant disease, whereby an antibody as defined in anyone of claims 12-25 is administered to a human subject. 40. A method according to claim 39, whereby said antibody has been changed by being genetically linked to 30 molecules giving the combined molecule changed pharmaco kinetic properties. 41. A method according to claim 39, whereby said antibody has been changed by being derivatised.