CA2300370A1 - Crystal of sm3 antibody (fragment) and recognizing epitope, its preparation, encoded data storage medium containing its coordinates and its diagnostical or medical use - Google Patents

Abstract

A process for preparing a crystal using cadmium. Structure factors or structural coordinates obtained from the crystal of SM3 antibody bound to an epitope can be used to design mimics of the antibody or epitope. Such mimics can be used in the diagnosis or therapy of cancer.

Description

0~~70 DEMANDES OU B1~EVETS VOLUMINEUX
LA PRESENTS PARTIIE DE C;ETTE DEMANDS OU CE BREVET
COMPREND PLUS D'UN TOME.
CECI EST LE "TOME ~ DE 2.
NOTE: Pour les tomes additionela, veuiilez contacter le Bureau canadien des brevets JUMBO A,PPL.I~~ATIONS/PATENTS
~f HiS SECTION OF THE APPLICATION/PATENT CONTAINS MORE
ifHAN ONE VOLUME
THIS I~~ VOLIJME ~ OF 2=
NOTE: For additional volumes..pl~ase contact the Canadian Patent Office . ~- --CRYSTAL OF SM3 ANTIBODY (FRAGMENT) AND RECOGNIZING EPITOPE, ITS PREPARATION, ENCODED DATA STORAGE MEDIUM CONTAINING ITS COORDINATES AND ITS DIAGNOSTICAL
OR MEDICAL USE
The present invention relates to a process for preparing a crystal, to the use of structural coordinates of a crystal comprising the epitope binding fragment of an antibody bound to a peptide, to mimics of the antibody and peptide, to products recognisable by mimics of the antibody or which recognise mimics of the peptide, to engineered antibodies and to their use in diagnosis or therapy.
Introduction The monoclonal antibody SM3 (secreted by the hybridoma HSM3 deposited with ECACC on 7 January 1987 under accession no. 87010701) binds to a tumour associated epitope on the epithelial mucin MC1C1 (WO-A-88/05054, WO 90/05142).
The MUC1 epithelial mucin is a transmembrane glycoprotein with the extracellular domain made up largely of exact tandem repeats of 20 amino acids, each of which contains five potential glycosylation sites (Gendler et al. 1988, 1990; see Figure 1). In breast and other carcinomas MUC1 is over-expressed and is aberrantly glycosylated making it antigenically distinct from the normally processed mucin. In breast cancers, the 0-glycans which are added are shorter (Hanish et al., 1989;
Hull et al., 1989: Lloyd et al., 1996), resulting in the exposure of epitopes which are normally masked on the core protein (Burchell and Taylor-Papadimitriou, 1993).
Such epitopes are called cryptic epitopes. The SM3 antibody was raised against MUC1 stripped of its carbohydrate (Burchell et al., 1987). SM3 recognises such a cryptic epitope, and shows high selectivity in 35. reacting specifically with the carcinoma associated mucin in more than 90~ of breast carcinomas (Girling et al., 1989). The high specificity of the SM3 antibody makes it a potentially useful tool in the diagnosis and treatment of breast cancer (Granowska et al., 1996).
The sequence of the repeating unit of the core protein of MUC1 contains doublets of threonine and serine, bounding a highly immunogenic domain (Burchell et al., 1989; see Figure 1). A three-dimensional NMR
structure for three multiple MUC1 peptide repeats reveals repeating "knob-like" structures, corresponding to the immunogenic domains which are connected by extended spacers (Fontenot et al., 1995). The epitope for SM3 has been mapped using overlapping peptides and found to correspond to just five contiguous amino acids, Pro-Asp-Thr-Arg-Pro (Burchell et al., 1989, hereafter referred to as the MUC1 epitope) lying within the "knob-like" domain and between the potential serine and threonine glycosylation sites. In vitro studies using recombinant GalNAc transferase which adds the first sugar to Ser or Thr suggest that Thr 1 Ser 12 and Thr 13 are glycosylated, while Thr 6 (within the MUC1 epitope) and Ser 2 are not (Wandall et al., 1997).
The inventors have crystallised a fragment of SM3 bound to a peptide, subjected the crystal to X-ray diffraction studies and measured the structure factors.
From the structure factors they have solved the structural coordinates. These may be used to generate new diagnostic and therapeutic materials and for other investigative purposes.
The invention includes the use of the structure factors and/or structural coordinates to identify, characterise, design or screen chemical entities, which have uses in diagnosis and therapy, as will be discussed below.
In one aspect the present invention provides the use of the structure factors and/or structural coordinates obtainable by subjecting a crystal comprising at least the epitope binding fragment of the SM3 antibody, bound to a peptide recognised by the epitope binding site of SM3 to X-ray diffraction measurements and optionally thereafter analysing the diffraction measurements to deduce structural coordinates.
The invention particularly provides a use of the structural coordinates of a moiety which comprises at least the epitope binding fragment of the SM3 antibody or a substantially similar fragment bound to a peptide such as the crystal peptide. The moiety may be whole SM3 antibody, or a Fab fragment derived from the digestion, of whole SM3 antibody by papain. The moiety may comprise a modified form of whole SM3 or a modified form of a fragment of SM3. Such modifications include the insertion or deletion of amino acids, or replacement of amino acids by other amino acids. Other chemical modifications may be made.
The peptide may comprise the sequence Pro-Asp-Thr-Arg-Pro, or any longer portion of the MUC1 tandem repeat, for instance the sequence Thr-Ser-Ala-Pro-Asp-Thr-Arg-Pro-Ala-Pro-Gly-Ser-Thr.
The invention provides the use of the structure factors and/or structural coordinates of a crystal of the Fab fragment of SM3 bound to the crystal peptide. The structure factors of such a crystal obtained as in the Examples are shown in Table 1 and the structural coordinates are shown in Tables 2a and 2b. The structural coordinates shown in each of Tables 2a and 2b are derived from the structure factors. However the coordinates shown in Table 2b have been calculated to a higher refinement and include additional protein atoms.
The invention provides the use of the structural coordinates shown in Table 2a and/or Table 2b.

The structural coordinates indicate the positions of individual atoms within the crystal and indicate the B
factor for each atom which gives some information about the mobility of the atoms. The structure factors may be used to derive additional information about the mobility of any individual atom or group of atoms within the crystal. This additional information, for instance, anisotropic B factors, may concern the direction of movement possible for each atom. The structure factors and structural coordinates thus give an indication of the available space for adjusting the position of individual atoms when designing mimics of the peptide or antibody.
Mimics of SM3 The invention provides the use of structure factors and/or structural coordinates to identify, characterise, design or screen mimics of SM3. The structural coordinates allow the epitope binding site bound to the peptide to be shown as a two dimensional representation, for example as in the LIGPLOTs of Figures 8 to 11 or a three dimensional representation by physical models or as displayed on a computer screen. Such representation can be used to design modifications of SM3. Such modifications includes modifications to increase the avidity of the epitope binding site for the bound peptide. This modification may preferentially increase the avidity of the epitope binding site for abnormally glycosylated MUC1.
The avidity may be increased by modifying the epitope binding site structure to increase the amount and number of interactions favourable to peptide binding and/or diminish unfavourable interactions between the peptide and antibody. Favourable interactions may be increased by extending the structure of the epitope binding site into spaces which are shown in the two WO 99!10379 PCT/GB98/02542 dimensional or three dimensional representations to be unoccupied or filled with water molecules. Such water molecules may include those which are shown in Figures 8 to 11.
The representations of the structures may be used in other ways to modify the structure of SM3. It is believed that SM3 may bind the peptide by an "induced fit" method which requires that conformation changes occur in the structure of SM3 during the process of binding the peptide. The SM3 binding site, or other parts of SM3 may be modified to allow such changes to occur more easily. Alternatively the mimic of SM3 may comprise the SM3 epitope binding site constrained in the conformation it adopts when it binds the peptide. The representations of the epitope binding site rnay be used to model such constraints by the putative introduction of covalent bonds between the atoms of SM3 which come close together when SM3 binds the peptide; one or more chemical linkers may be used between atoms of SM3 to constrain the epitope binding site to the required conformation, and/or one or more amino acids of SM3 may be replaced by analogues of the natural amino acids which help to constrain the conformation of the epitope binding site.
On the basis of such modelling SM3 mimics may be identified, characterised, designed or screened.
It is believed that the reason why SM3 has a low avidity for aberrantly glycosylated MUC1 is steric hindrance between residues close to or in the epitope binding site and carbohydrate moieties attached to positions 1, 12 and 13 of the peptide as shown in Figure 1. The representation of the epitope binding site bound to the peptide may be used to predict which residues of SM3 are likely to be involved in the steric hindrance.
One such residue may be Proline 5G of SM3. Such residues may be modified, replaced or deleted to decrease the steric hindrance in order to increase avidity.
Mimics of SM3 may for instance be obtained by computer modelling techniques. Using the structural coordinates a three dimensional representation of the surface of the epitope binding site bound to the peptide can be produced using Catalyst Software such as Catalyst/SHAPE, Catalyst/COMPARE, DBServer HipHop Ludi, MCSS and Hook which are available from Molecular Simulations Ltd., 240/250 The Quorum, Barnwell Road, Cambridge, England. Mimics of SM3 can be produced either by computationally identifying compounds which have a similar surface to the binding site of SM3, or by computationally designing compounds with surfaces which are likely to bind the peptide. Various methods can then be used to produce a three dimensional surface which is the same or similar to the epitope binding surface.
Based on this shape, packages such as Catalyst/SHAPE and Catalyst/COMPARE can be used to select compounds from databases which have a similar three dimensional shape to the epitope binding site. The epitope binding site surface may be described in more detail by analysis of the functional groups present to produce a pharmacophore.
Based on a pharmacophore, packages such as DBServerl and HipHop can be used to search databases for compounds whose surfaces are described by similar pharmacophores.
The databases that can be searched include ACD, NCI, Maybridge, Derwent World Index and BioByte.
Packages such as Ludi and MCSS can be used to select fragments or chemical entities from databases which can then be positioned_in a variety of orientations, or "docked" with the surface of the peptide. Once suitable chemical entities or fragments which bind sites on the surface of the peptide have been identified bridging fragments and framework structure are chosen with the correct size and geometry to support the chemical entities and fragments in the favourable orientation and location and to form the mimic of SM3.
Packages such as Hook can be used to select framework structures.
Once a candidate mimic of SM3 has been designed or selected by the above methods, the efficiency with which that mimic may bind to the peptide may be tested and optimized using computational or experimental evaluation.
Various parameters can be optimized depending on the desired result. These include, but are not limited to, specificity, avidity, on/off rates, and other characteristics readily identifiable by the skilled artisan.
The computational means may employ packages such as Catalyst/SHAPE, Catalyst/COMPARE, DBServer, HipHop, Ludi and MCSS to evaluate selected candidate SM3 mimics. The experimental means may comprise ELISA methods such as detailed below using GST-MUC1 ELISA plates or the techniques described in Bynum et al.
Thus, one may optionally make substitutions, deletions, or insertions in some of the components of the SM3 mimics in order to improve or modify the binding properties. Generally, initial substitutions are conservative, i.e. the replacement group will have approximately the same size, shape, hydrophobicity and charge as the original component. Such modified mimics can be computationally or experimentally evaluated in the same manner as the first candidate SM3 mimics, and if necessary further modifications can be made. This process of evaluating and modifying may be iterated any number of times.
The term aberrantly glycosylated MUC1 is used throughout this specification to refer to MUC1 which has a different level of glycosylation than is normally found on MUC1 from a particular tissue. The different level of _g_ glycosylation may be a decrease in the level of glycosylation. MUC1 with such decreased levels of glycosylation may be produced by a tumour cell, such as an adenocarcinoma cell for instance a cell from an epithelial tumour of colon, lung, ovary, pancreas or especially a breast tumour cell.
Mimics of SM3 preferably have a high avidity for aberrantly glycosylated MUC1. Such an avidity may be higher than that of SM3. They may have a higher selectivity towards aberrantly glycosylated MUC1 in preference to normally glycosylated MUC1 than SM3.
The avidity of the mimic of SM3 to bind aberrantly glycosylated MUC1 can be tested in an ELISA assay as detailed below:
GST-Muc-1 ELISA Method Preparation of GST-Muc-1 Batch 5 ELISA plate.
DNA coding for the seven tandem repeats of the 20 amino acid peptide epitope was cloned into a GST fusion vector, pGEX-2T (Pharmacia) in the E.Coli strain TG1 . The cells were grown in L broth (100mg/ml ampicillin) overnight and then IPTG was added to a final concentration of 0.5 mM
followed by a further incubation of 4 hr at 37 C The cells were pelleted (20min/10000rpm), washed x2 with PBSa and resuspended in 25mM TRIS pH 8, 25mM Glucose, 10 mM
EDTA containing lyzozyme (4mg/ml) and left for 30 min at RT. 5m1 of lysis buffer was then added and incubated on ice for 5 min. After centrifugation the supernatant was loaded onto a Glutathione Sepharose 4B column (Pharmacia) equilibrated in lysis buffer. The fusion protein product was eluted in PBSa/l5mM Glutathione and stored at 4 C
until required. SDS-PAGE and FPLC Superose 6 (Pharmacia) size separation showed a single species mwt 43 K.
Preparation of GST-Muc-1 Batch 5 ELISA Plates The GST-Muc-1 was diluted in PBSa to 50mg/ml and 50m1 -g-added to each well of a 96 well plate The plates were incubated overnight at 4 C, blocked with 200m1 2~
BSA/PBSa for 2 hr at RT. and stored at -70 C until required.
REAGENTS/BUFFERS
Alkaline Phosphatase -Goat anti-mouse (Fc Specific) Sigma No. A-2429 BSA - Albumin Bovine, Fraction V powder, RIA grade, Sigma No A-7888 ELISA Plates - Nunc-Immuno Plates No. 442404 ELISA Substrate Buffer - Diethanolamine buffer pH 9-8 400m1 Diethanolamine (Sigma D-2286) 24.5 mg MgCI2 BHD (Anala R) pH 9.8 with HC1 and make up to 500m1 with distilled water EDTA Ethylenediamine tetra acetic acid BDH Anala R
Glutathione Reduced ICN 101814 IPTG Isopropylthio-b-D-galactoside Promega Lysis Buffer 150mM NaCl, 16 mM Na2HP04, 4mM NaH2P04, 100mM EDTA, 1~ Triton X (BDH Anala R) 2 mM PMSF (Phenylmethylsulfonyl fluoride) (Sigma ) pNPP - p-Nitrophenyl Phosphate Sigma No. N-9389 PBSAa- Phosphate Buffered Saline (ICRF) TWEEN 20 - Polyoxyethylene (20) Sorbitan monolaurate GST-Muc-1 (a Glutathione-S-transferase fusion protein containing seven copies of the 20 amino acid tandem repeat (non-glycosylated) ELISA plates are stored @ -70°C pre blocked with 2~ BSA/PBSA. (PBSA-Phosphate Buffered Saline BSA- Bovine Serum Albumin, Fraction V powder, RIA
grade, Sigma No. A-7888.) 1. Thaw out a plate at room temperature (RT).
2. Test antibody samples are diluted in PBSA/
0.02$ Tween 20 (PBSA/T). e.g. SM3 from 100-0.001 mg/ml.
3. Wash the plate three times with PBSA/T. Add samples in a volume of 50 ml, in triplicate and leave at RT for 1 hour.
4. Remove the antibody samples carefully by aspiration and add 200 ml of PBSA/T to each well. Remove by aspiration and repeat. Then wash again twice with PBSA/T.
5. Add 50 ml of 1/2000 (diluted PBSA/T) Goat anti-mouse (Fc Specific) Alkaline Phosphatase (Sigma No.
A-24209) to each well and incubate at RT. for 1 hour.
6. Wash the plate three times with PBSA/T and then add 50m1 of substrate which is p-Dinitrophenyl Phosphate (Sigma N-2765) at lmg/ml in Diethanolamine Buffer pH 9.8, (400m1 distilled water, 98m1 diethanolamine, 24.5mg MgCl2, pH 9.8 with HC1 and make up to 500m1 with distilled water).

7. Incubate at RT. for 15 min and read the OD at 405 nm on the Labsystems Multiscan Plate reader.
The avidity of the mimic of SM3 can also be measured using methods described in Bynum et al.
These techniques may also be used to test the specificity of the mimic of SM3 by testing the avidity of the mimic for aberrantly glycosylated MUC1 and for normally glycosylation MUC1.
A histological screen using the mimic of SM3 can be performed using tumour tissue from a breast tumour and normal tissue, for example, as described in Girling et al., 1989. This can be used to determine if the mimic is specific for the aberrantly glycosylated MUC1 and therefore suitable for use in a method of diagnosing breast cancer. The mimic can also be tested against live tumour cells which are not fixed.
Cells which have MUC1 with reduced levels of glycosylation can be produced by the use of metabolic inhibitors of 0-linked chain extension, such as 0-benzylgalactosamine. Such cells can be used to study the effects of low levels of glycosylation of MUC1 on the avidity of the mimics of SM3.
Mimics of SM3 may be used in a diagnostic test to detect the presence of tumour cells in a tissue sample, for example in a histological screening. Mimics used in this manner may be labelled with a detectable label.
Alternatively agents able to specifically bind such mimics may be used to detect the presence of the mimics once the mimics have bound the aberrantly glycosylated.
The mimics of SM3 may be used in vivo for the detection of tumour cells. They may be used in tumour imaging in vivo. Generally, such mimics would be labelled with a detectable label.
The mimics of SM3 can be used in a method of therapy against cancer, particularly adenocarcinomas such as ovary, colon, lung, pancreas and breast epithelial cancers, especially breast cancers__ Mimics which are antibodies or substantially similar to antibodies or fragments of antibodies may bind to aberrantly glycosylated MUC1 on the surface of tumour cells and aid the killing of the tumour cells by recruiting the patients immune system.

Mimics of SM3 may be chemically linked to a cytotoxic agents such as a toxin or a radioisotope.
Binding of such toxin linked mimics to the tumour cells would lead to the killing of the tumour cell.
It is believed that high levels of MUC1 or aberrantly glycosylated forms of MUC1 may have an immunosuppressive effect. Tumour cells may have high levels of MUC1 on their surface and/or aberrantly glycosylated forms. Therefore in the method of therapy of the invention a mimic of SM3 could be used to bind to MUC1 in vivo and prevent or decrease its immunosuppressive effects. The types of immunosuppressive effects that may be prevented or decreased are discussed below in relation to mimics of the MUC2 epitope. Such a mimic could be administered in conjunction with an anti-tumour agent, such as an anti-tumour vaccine, and may have an adjuvant-like effect (the mimic may increase the immune response generated by the vaccine). The anti-tumour agent could be a mimic of the SM3 epitope. The mimic of SM3 could be administered at any time in relation to the administration of the anti-tumour agent, for example before, with or after the administration of the anti-tumour agent.
Thus the invention provides a mimic of SM3 for use in a method for treatment of the human or animal body by surgery or therapy or in a diagnostic method practised on the human or animal body, such as in anti-cancer therapy.
The invention also provides a product comprising a mimic of SM3 and an anti-tumour agent as a combined preparation for simultaneous, separate or sequential use in anti-cancer therapy.
The invention provides a use of a mimic of SM3 in the production of a pharmaceutical composition for use in the methods discussed above. The invention provides a pharmaceutical composition containing a mimic of SM3 and *rB

a diluent or carrier. The invention provides a method of treating or diagnosing cancer by administering to a human or non-human animal in need thereof an effective non-toxic amount of a mimic of SM3.
The production of mimics of SM3 and the methods, routes and dosages for use of the mimics of SM3 are discussed below.
Mimics of MUC1 epitope peptides The invention allows the use of the structure factors and the structural coordinates to identify, characterise or design a mimic of the MUC1 epitope peptides. Such a mimic may be used to produce a specific binding agent which has a desired association with aberrantly glycosylated MUC1. The mimic of the peptide can be used to select a specific binding agent from a library on the basis of its affinity to the mimic. Such a library may be a microbial display library, such as a phage display library. The specific binding agent may bind aberrantly glycosylated MUC1, and therefore be used in a similar manner to the mimics of SM3. The mimic of the peptide may be used in a vaccine administered to a human or animal. Such a vaccine would stimulate the production of antibodies able to bind aberrantly glycosylated MUC1. Such a vaccine could therefore be used as a therapy against cancer, for example to prevent or treat cancer.
Thus the invention provides a mimic of the MUC1 epitope, a mimic of the MUC1 epitope for use a method for treatment of the human or animal body by surgery or therapy or in a diagnostic method practised on the human or animal body, such as in anti-cancer therapy. The invention also provides a use of such a mimic in the production of a pharmaceutical composition for use in such a method. The invention provides a pharmaceutical composition containing a mimic of the MUC1 epitope and a diluent or carrier therefor. The invention provides a method of diagnosing or treating cancer by administrating an effective non-toxic amount of a mimic of the MUC1 epitope to a human or non-human animal in need thereof.
The production of mimics of the MUC1 epitope and methods, routes and dosages for use of the mimics of the MUC1 epitope are discussed below.
The MUC1 epitope mimics may also be administered to an animal preferably a laboratory mammal, such as a rodent, for instance a mouse, in order to induce an antibody response. Antibody secreting cells may be recovered from the animal and hybridoma technology may then be used to produce a hybridoma using such recovered cells. The hybridoma or any other antibody expression system would be a source of monoclonal antibodies capable of recognising the mimic and preferably the MUC1 epitope and thus have diagnostic and therapeutic utilities as for the SM3 mimics described above.
As described above it is believed that MUC1 may cause immunosuppressive effects. Thus the mimic of the epitope may be capable of causing an immunosuppressive effect, and thus may be used to cause such an effect in a human or animal. The immunosuppressive effect may be caused by the mimic decreasing the activity of T cells, such as CD4 or CD8 T cells. The T cells may be polyclonal.
The mimic may cause a decrease in (i)the proliferation of the T cells, (ii) their secretion of cytokines,(iii) their interaction with other cells, such as interaction mediated by ICAM-1 on the T cell surface, and/or(iv)their cytotoxic activity. The mimic may cause the cross-linking of a surface molecule on the T cells, such as ICAM-1. The mimic ma.y cause T cells to become anergic. The effects of the mimic may be reversible by the addition of IL-2 or anti-CD28 antibody.

Thus the mimic may be used to decrease or prevent an immune response, particularly when the response has a deleterious effect on the body. Thus the mimic may be used in the therapy (i.e. to treat or prevent) of diseases caused by autoimmune responses ~sucn as arthritis, multiple sclerosis, asthma or diabetes), allergies, inflammatory disorders or transplant rejections, such as graft versus host disease.
Thus the invention provides a mimic of the MUC1 epitope for use in therapy which prevents or decreases an immune response, and so can treat or prevent a disease caused by an immune response. The invention also provides a use of such a mimic in the production of a pharmaceutical composition for use in such a method. The invention provides a method of treating a disease caused by an immune response, such as autoimmune disease, allergy, inflammatory disorder or transplant rejection;
by administrating an effective non-toxic amount of a mimic of the MUC1 epitope to a human or non-human animal in need thereof.
As with the mimics of SM3 the two dimensional and three dimensional representations of the epitope binding site bound to the peptide mentioned previously can be used to design a mimic which is, for instance, a peptide or a derivative thereof or an analogue of a peptide comprising the sequence Pro-Asp-Thr-Arg-Pro or the sequence Thr-Ser-Ala-Pro-Asp-Thr-Arg-Pro-Ala-Pro-Gly-Ser constrained in the conformation approximately that which the crystal peptide adopts when bound to SM3.
The sequence can be constrained by the introduction of covalent bonds between atoms which come close together in the conformation adopted when bound to SM3.
Alternatively a chemical linker may be used to join such atoms together. The sequences may be constrained by cyclisation of a peptide in which the relevant sequence is present. The peptides may also be constrained by replacement of one or more amino acids by analogues of natural amino acids which cause the peptide analogue to adopt the required conformation when they are introduced.
Hydrogen bonds, such as intra-peptide hydrogen bonds may be introduced into the mimic. These may help to constrain the mimic to a particular conformation, such as the conformation adopted when bound to SM3.
The structural coordinates allow the epitope binding site bound to the crystal peptide to be shown as a two dimensional representation, for example the LIGPLOTs of Figures 8 to 11 or a three dimensional representation on a computer screen. Such representation can be used to design modifications to the MUC1 epitope which may increase its avidity for SM3, for example, by increasing the "on rate" relative to the crystal peptide and decreasing the "off rate". The avidity of the MUC1 epitope mimic for the epitope binding site may be increased using a similar strategy as used to design mimics of SM3. The mimic can be modified to increase the amount and number of favourable interactions with SM3, for example by extending the structure of the MUC1 epitope into spaces which are shown to be unoccupied or filled with water molecules. Such water molecules may include those shown in Figures 8 to 11. It is appreciated that such mimics of the MUC1 epitope which are designed to have an increased avidity with the epitope binding site of SM3 may not cause the production of antibodies with a higher avidity to aberrantly glycosylated _.MUC1 than SM3 has, however such mimics may lead to the stimulation of an increased antibody response due to their higher avidity to the antibody. The two dimensional and three dimensional representation of the epitope binding site bound to the crystal peptide can as mentioned previously be used to determine why SM3 binds the peptide with a low avidity. This information can then be used to design mimics of the MUC1 epitope which can select from a library those mimics of SM3 which have a high avidity to aberrantly glycosylated MUC1. Such mimics, when administered in a vaccine may also stimulate the production of antibodies with a higher avidity to aberrantly glycosylated MUC1 than SM3.
Such mimics of the MUC1 epitope may for example be designed by modifying the structure of the crystal peptide which binds SM3 in those areas where unfavourable interactions occur between the crystal peptide and epitope binding site which cause a lowering of avidity.
For example the structure of the MUC1 epitope may be extended in those areas where steric hinderance occurs between the carbohydrates on the aberrantly glycosylated MUC1 and the epitope binding site. Use of such a mimic may lead to the selection of specific binding agents (as described above) from a library or the stimulation of antibodies which have a reduced level of steric hinderance with the carbohydrate and therefore a higher avidity for aberrantly glycosylated MUC1.
Mimics of the MUC1 epitope may be designed by computer modelling techniques in a similar manner to the designing of mimics of SM3 using these techniques.
Again, packages such as Catalyst/SHAPE and Catalyst/COMPARE can be used to select compounds with a similar three dimensional shape to the peptide when it is bound to SM3. Packages such as DBServer and HipHop can be used to find compounds with similar pharmacophores to the peptide and packages such as Ludi, MCSS and Hook can be used to design a mimic which is predicted to bind well to the epitope binding site of SM3.
Once a candidate mimic has been designed or selected by the above methods, the efficiency with which that mimic may bind to the SM3 epitope binding site may *rB

be tested and optimized using computational or experimental evaluation. As described previously packages such as Catalyst/SHAPE, Catalyst/COMPARE, DBServer, Ludi and MCSS may be used to computationally evaluate the mimic. The mimic may be experimentally evaluated in the competition ELISA assay described below.
Various parameters can be optimized depending on the desired result. These include, but are not limited to, specificity, avidity, on/off rates, and other characteristics readily identifiable by the skilled artisan.
Thus, one may optionally make substitutions, deletions, or insertions in some of the components of the MUC1 epitope in order to improve or modify the binding properties. Generally, initial substitutions are conservative, i.e. the replacement group will have approximately the same size, shape, hydrophobicity and charge as the original component.
As with the mimic of SM3 the process of evaluation and modification can be iterated any number of times.
The mimic of the MUC1 epitope may be glycosylated.
The mimic may have an intra-peptide hydrogen bond.
This may be between the residues equivalent to Pro 4 and Thr 6 of the MUC1 epitope or between the residues equivalent to Asp 5 and Arg 7. The atoms involved in the hydrogen bond may be the equivalent of Pra 4 (atom 0), Thr 6 (atom N), Asp 5 (atom OD1) and/or Arg 7 (atom N).
The mimic of the MUC1 epitope may have higher avidity to SM3 than aberrantly glycosylated MUC1, or a low avidity or may not bind at all. The avidity of the mimic can be tested in a competition ELISA assay in which a MUC1 peptide such as aberrantly glycosylated MUC1, deglycosylated MUC1 or the fully stripped core protein or a fragment thereof is attached to the plate and SM3 is added in the presence of the mimic of the MUC1 epitope.

By measuring the ability of the mimic to inhibit the binding of the antibody to the MUCl peptide the avidity of the mimic for SM3 can be measured. Preferred mimics include those which lead to the selection of specific binding agents from a library which bind aberrantly glycosylated MUC1 with a higher avidity and/or higher selectivity than SM3. Preferred mimics of the MUC1 epitope are also those which stimulate the production of antibodies in vivo which have a higher affinity than SM3.
As discussed above specific binding agents include those which have been selected from a library, which may include antibodies, or those antibodies produced by B
cells which bind the mimic of the peptide. Preferred specific binding agents are those which bind aberrantly glycosylated MUC1 with an avidity and/or specificity higher than SM3.
The specific binding agents may be used in a diagnostic test to detect the presence of tumour cells in a tissue sample, for example in a histological screening.
Specific binding agents used in this manner may be labelled with a detectable label. Alternatively moieties able to specifically bind such mimics may be used to detect the presence of the specific binding agents once the specific binding agents have bound the aberrantly glycosylated.
The specific binding agents may be used in vivo for the detection of tumour cells. They may be used in tumour imaging in vivo. Generally, such specific binding agents would be labelled with a detectable label.
The specific binding agents may be used to prevent or decrease immunosuppression caused by MUC1.
The specific binding agents can be used in a method of therapy against cancer, particularly breast cancer.
Specific binding agents which are antibodies or substantially similar to antibodies or fragments of antibodies may bind to aberrantly glycosylated MUCl on the surface of tumour cells and aid the killing of the tumour cells by the immune system.
Specific binding agents may be chemically linked to a cytotoxic agents such as a toxin or a radioisotope.
Binding of such toxin linked specific binding agents to the tumour cells would lead the killing of the tumour cell.
Thus the invention provides a specific binding agent for use a method for treatment of the human or animal body by surgery or therapy or in a diagnostic method practised on the human or animal body, such as in anti-cancer therapy. The invention also provides a use of a specific binding agent in the production of a pharmaceutical composition for use in such a method. The invention provides a pharmaceutical composition containing a specific binding agent and a diluent or carrier therefor. The invention also provides a method of diagnosing or treating cancer by administering an effective non-toxic amount of a specific binding agent to a human or non-human animal in need thereof.
The invention also provides a product comprising a specific binding agent and an anti-tumour agent as a combined preparation for simultaneous, separate or sequential use in anti-cancer therapy.
The production of such specific binding agents and of the MUC1 epitope mimics and methods, routes and dosages for use of the agents and mimics are discussed below.
When administered to an animal or human the mimic of SM3 may lead to the production of antibodies which recognise the mimic. B cells from the animal or human may be used in conjunction with hybridoma technology to produce such antibodies as monoclonal antibodies. Such antibodies may have a epitope binding site which is similar in shape to the crystal peptide. The administration of such antibodies to an animal or human may lead to the production of antibodies which recognise aberrantly glycosylated MUC1. Therefore the antibodies which recognise SM3 may be used as a vaccine or may be used to produce a vaccine against cancer.
The antibodies or fragments thereof which recognise aberrantly glycosylated MUC1 produced in response to the anti-mimic of SM3 antibody may themselves be used in an anti-cancer treatment or in a diagnostic method. Thus the invention also provides such an antibody or fragment thereof for use in a method for treatment of the human or animal body by surgery or therapy or in a diagnostic method practised on the human or animal body, such as Iin anti-cancer therapy. The invention also provides use of such an antibody or fragment thereof in the production of a pharmaceutical composition for use in such a method.
The invention provides a pharmaceutical composition containing such an antibody or fragment thereof and a diluent or carrier therefor. The invention also provides a method of diagnosing or treating cancer by administering an effective non-toxic amount of such an antibody or fragment thereof to a human or non-human animal in need thereof.
The invention provides the use of H3 chain of SM3 as a therapeutic or diagnostic agent. The invention also provides the use of H1 chain of SM3 as a therapeutic or diagnostic agent. Thus the invention also provides such an H3 or H1 chain for use in a method for treatment of the human or animal body by surgery or therapy or in a diagnostic method practised on the human or animal body, such as in anti-cancer therapy. The invention also provides use of such an H1 or H3 chain in the production of a pharmaceutical composition for use in such a method.
The invention provides a pharmaceutical composition containing such an Hl or H3 chain and a diluent or carrier therefor. The invention also provides a method of diagnosing or treating cancer by administering an effective non-toxic amount of such an H1 or H3 chain to a human or non-human animal in need thereof.
The invention provides the use of the structure factors and/or structural coordinates to solve structural coordinates of other crystals. Such a crystal may comprise an antibody or antibody fragment. In particular the structural coordinates can be used to solve the structural coordinates of a crystal comprising an antibody which has a non-proline cis peptide bond. The cis peptide bond may be in the H3 chain of the antibody.
The invention provides the use of the structure I
factors and/or structural coordinates to engineer, design or modify an antibody. The engineering may be for the purpose of humanising the antibody. The engineering may comprise the replacement of portions of the antibody. The portions of the antibody may be replaced with portions from another protein, such as a human protein, for instance a human antibody. Two dimensional and three dimensional representations, such as those discussed previously, may be used during the engineering of the antibody to ensure that the epitope binding site of the antibody is the same or substantially similar to the antibody binding site of SM3. The engineering of the antibody may be for the purpose of increasing the contribution made by the H3 and H1 chains in the binding of the epitope.
The invention provides the use of an engineered glycine in a protein to insert a non-proline cis peptide bond into the protein. The insertion of the glycine may enable the protein to undergo conformational change, for instance in connection with a specific binding reaction where the affinity, avidity or selectivity of the specific binding reaction is modified by adoption of a cis peptide bond adjacent to the engineered glycine. The engineered glycine can be inserted using site directed mutation techniques, such as those described in Short Protocols in Molecular Biology, 3rd edition, published by John Wiley and Sons, Inc., USA. The protein may be an antibody or fragment of a antibody. The engineered glycine may be in a CDR loop of the antibody. The insertion of the engineered glycine in an antibody will generally affect the binding of the antibody to an epitope.
The invention also provides a method of producing a crystal which comprises the use of cadmium. Generally the method comprises contacting any one of the moieties to be crystallised with cadmium such as by use of a solution of a cadmium salt. The cadmium ions may be present in the solution in which the crystal is grown.
In the case where the mimic of SM3, the mimic of the MUC1 epitope or the specific binding agent comprise a peptide then such agents may be delivered to animal or human by the admimistration of a nucleic acid which encodes such a peptide. Transcription and translation or merely translation of the nucleic acid would lead to the production of the peptide in vivo. It is appreciated by the skilled artisan that the nucleic acid may be within a suitable vector, especially an expression vector, for instance a virus or an organism which is administered, such as a Vaccinia virus, for instance an attenuated Vaccinia virus or the nucleic acid may be administered in the presence of a suitable diluent or carrier as in the use of "naked DNA". Nucleic acids encoding such peptides form a further aspect of the present invention as do their uses in cancer therapy and pharmaceutical compositions containing them.
Thus the invention also provides such nucleic acids for use in a method for treatment of the human or animal body by surgery or therapy or in a diagnostic method practised on the human or animal body, such as in anti-cancer therapy. The invention also provides use of such nucleic acids in the production of a pharmaceutical composition for use in such a method. The invention provides a pharmaceutical composition containing such nucleic acids and a diluent or carrier therefor. The invention also provides a method of diagnosing or treating cancer by administering an effective non-toxic amount of such nucleic acids to a human or non-human animal in need thereof.
The mimic of SM3, mimics of the MUCl epitope, selective binding agents, nucleic acids, viruses or organisms containing nucleic acids of the invention (hereafter referred to as substances of the invention) may be formulated for clinical administration by mixing them with a pharmaceutically acceptable carrier or diluent. For example the substances can be formulated for topical, parenteral, intravenous, intramuscular, subcutaneous, or transdermal administration. They may be mixed with any vehicle, e.g. a diluent or carrier which is pharmaceutically acceptable and appropriate for the desired route of administration. The pharmaceutical carrier or diluent for injection may be, for example, a sterile or isotonic solution such as Water for Injection or physiological saline.
The dose may be adjusted to deliver an effective non-toxic amount and according to various parameters, especially according to the nature and efficacy of the substance used; the age, weight and condition of the patient to be treated; the mode of administration used;
the conditions to be treated: and the required clinical regimen. As a guide, the amount of therapeutic substance such as a polypeptide to be administered by injection will generally be from 10 to 1000ug. For instance 100 to 500ug. For in vivo diagnostic imaging purposes in particular, the dose will depend on similar factors but may need to be much larger in order to generate an appropriate image.
The routes of administration and dosages described are intended only as a guide since a skilled practitioner will be able to determine readily the optimum route of administration and dosage for any particular patient and condition.
The substances of the invention may be produced by conventional techniques well known to thosQ skilled in the relevant arts. Peptides may be synthesised de novo or produced by transcription and translation of DNA or translation of RNA in a suitable expression system by conventional methods. Nucleic acids may be produced by de novo synthesis, obtained from natural sources such as by conventional probing and cloning techniques or generated from natural sources by modification by well known methods. Antibodies may be made by suitable immunisation protocols and purified from immunised animals by conventional techniques, or they may be made by hybridomas and similar antibody secreting cell lines cultured in vitro or they may be obtained by expression from suitable nucleic acids. Hybridomas and other antibody secreting cells may be obtained, cultured and used to produce antibodies by conventional methods.
Chemical modification of such materials as peptides and antibodies may be achieved by well known methods.
Fragments of antibodies may be produced by known chemical or enzymatic digestions or by expression from suitably modified nucleic acids. New chemical entities such as mimics of the MUC1 epitope may be produced by the well known techniques of synthetic organic chemistry.
The substance of the invention does not include those peptides disclosed in WO-A-88/05054 or in WO-A-90/05142.
The present invention is illustrated by the figures of the accompanying drawings which are described below:
Brief Description of the Drawings Figure 1 shows the amino acid sequence of the MUC1 tandem repeat.
Figures 2 and 3 show experimental electron density maps and the refined atomic models of the epitope binding site of SM3 and part of the crystal peptide.
Figure 4 shows a comparison of the conformations of the unbound and bound peptide.
Figures 5 and 6 show molecular surface features of the SM3-crystal peptide interaction.
Figure 7 is a stereo view of the non-proline cis-peptide bond in CDR H3.
Figures 8, 9, 10 and 11 are LIGPLOTs of the peptide binding site.
Figure 12 shows experimental density maps of the cis- peptide conformation in H3.
Figure 13 shows a Ramachandran plot for the H3 fragment.
Figure 14 shows the main chain temperature factors of the fragment from H3.
Figure 15 shows a stereo view of the MUC1 peptide in the antibody combining site.
Figure 16 shows a stereo view of the non-proline cis- peptide bond in H3.
Detailed Description of the Drawincrs Figure 1 shows MUC1 tandem repeat sequence written from N- to C-terminal using the Internationally 'recognised 3-letter code (which is used throughout this specification). The epitope recognised by SM3 is shown in bold (Pro-Asp-Thr-Arg-Pro) with the immunodominant region indicated by the double headed arrow. The probable glycosylation sites are shown in bold italics. The peptide used for the crystallisation studies (Thr-Ser--Ala-Pro-Asp-Thr-Arg-Pro-Ala-Pro-Gly-Ser-Thr) is underlined (and is herein referred to as "the crystal peptide").
Figures 2 and 3 show experimental electron density maps and refined atomic models calculated using the a observed structure factors (30 to 1.95 A) and calculated phases after solvent modification. The maps are contoured at 1.0 a. Figure 2 shows part of the MUC1 peptide antigen showing the immunodominant region. The numbering refers to Figure 1. Figure 3 shows the cis-peptide bond in CDR H3 between residues G1y96H and G1n97H. Two water molecules are shown forming hydrogen bonds to amide nitrogens.
Figure 4 shows the unbound and bound peptide. The unbound peptide is shown on the left. The structure represents the "knob-like" region and was determined in solution by NMR; it includes the whole of one tandem repeat and parts of the two flanking repeats (Fontenot et al., 1995). The conformation of the crystal peptide as bound by SM3 is shown on the right. There appears to be a conformational transition of the "knob-like" region upon SM3 binding, which leads to a more extended peptide conformation. The main differences can be attributed to the ~ angles of AspSP and Arg7P.
Figure 5 shows a CPK model of the SM3 combining site with the individual CDR loops_labelled. The crystal peptide is shown contacting all of the CDR's except H2.
Figure 6 shows electrostatic surface potential representation of the SM3 peptide complex. The surface was calculated using GRASP (Nicholls et al., 1991). The crystal peptide is shown as a stick model.

Figure 7 shows the cis-peptide bond in H3. The interactions between the residues G1y96H-G1n97 and the crystal peptide (solid) are shown. Hydrogen bonds are dotted (drawn using MOLSCRIPT; Kraulis, 1991).
Figures 8, 9, l0.and 11 are LIGPLOTS showing the main interactions which occur between the SM3 binding site, the crystal peptide and water molecules.
KEY:
1 ~ ~ Liganid m/c ttis s3 Non-ligand residues involved in hydrophobic Ligand sideehain ~Ttrt~ contacts) -~ Hydrogen bond and its length ~ Corresponding atoms involved in hydrophobic contacts) ~ WAT = water Figure 12 shows experimental electron density maps calculated using the observed structure factors (30 to 1.95 A and calculated phases after solvent modification, contoured at 1.0 a level (2Fo-Fc) and 2.0 a (Fo-Fc).
Residue numbering as in the text. The model with this cis- peptide bond is coloured according to atom type.
(A) 2Fo-Fc difference density, calculated for the refined model with a trans G1y96H-G1n97H peptide bond. (B) 2Fo-Fc and Fo-Fc (residual negative electron density; and positive density) map, calculated for the refined model with a cis G1y96H-G1n97H peptide bond. (C) 2Fo-Fc difference electron density map around the final cis G1y96H-G1y97H fragment, showing the hydrogen bonds which appear to stabilise the cis - peptide conformation.
Figure 13 shows a Ramachandran plot for the H3 fragment refined with cis and trans___G1y96H-G1y97H peptide bond. Residues are labelled accordingly. Created by PROCHECK, Laskowski et al., 1993.
Figure 14 shows the main chain temperature factors of the fragment from H3 which contains the non-proline cis- peptide bond between G1y96H and G1n97H. Values for the refined cis - confirmation are shown as open boxes (solid line), while those for the traps are indicated with closed circles (dashed line).
Figure 15 shows a view of the MUC1 peptide in the antibody combining site. The peptide antigen is labelled and is shown in bold. SM3 Fab residues which interact with the peptide are labelled. Hydrogen bonds are shown as dotted lines with water molecules as black spheres.
Brawn by MOLSCRIPT (Kraulis, 1991).
Figure 15 shows the cis peptide bond. The interactions between the residues G1y96H-G1n97H and part of the peptide antigen (bold) are shown. Hydrogen bonds are dotted and waters are represented as black spheres, (drawn using MOLSCRIPT; Kraulis, 1991). The hydrogen bond between G1y96H (N) and the peptide antigen, as mediated via a water molecule, is lost for a traps -peptide conformation for G1y96H-G1n96H.
The invention is further illustrated by the Examples:
Example 1 Assessment of the Binding of the MUC1 epitope to SM3.
The sequence of the MUC1 peptide in the context of the MUC1 repeat is shown in Figure 1. The region of the MUC1 repeat for the co-crystallisation studies was chosen to keep the peptide as soluble as possible, while maintaining most of the tertiary structure of the knob region as based upon the NMR structure of the mucin repeats [Fontenot, J.D., Mariappan, S.V.S., Catasti, P., Domenech, N., Finn, O.J. and Gupta, G.J. Biomol. Struct.
and Dynam. 13, (1995) 245-260]. The binding of the MUC1 epitope to SM3 was assessed by inhibition in an ELISA
assay. The assay involved SM3 binding to a recombinant fusion protein containing seven copies of the MUC1 tandem repeat, a lug/ml concentration of SM3 was incubated with increasing concentrations of peptide, a 50~ inhibition was achieved at a peptide concentration of approximately 30ug/ml (Apparent Ka~24uM). In contrast a peptide of the same composition but different sequence (Ala-Arg-Pro-Thr-Gly-Thr-Ser-Asp-Pro-Thr-Pro-Ala-Ser) gave no detectable inhibition at the same concentration.
Example 2 Preparation of the SM3 Fab fragments.
SM3 at 4mg/ml in 100 mM sodium acetate pH 6.0, 3mM
EDTA, 50mM cysteine was incubated with 160 ug/ml of papain for 4 hr at 37°C. Fabs were purified by gel filtration on Superose 6 in PBS, followed by dialysis into 20mM Tris pH 8.0 and then Mono Q ion exchange chromatography. Fractions were analyzed by SDS PAGE and Fab fractions concentrated to 4 mg/ml in lOmM Tris pH
8Ø The overall yield was 22~.
Example 3 Crystallisation of the SM3 Fab fragment bound to the MUC1 epitope.
The SM3 Fab fragments were mixed with concentrated solution of the peptide at a molar ratio 1:5, incubated for 3 hours at 37°C, and then concentrated to 9.5 mg/ml.
The crystallisation trials were conducted using the hanging drop method. A wide range of conditions were tested using commercial screens, namely Hampton Research I and II [Jancarik, J. and Kim, S.H. J. App.I. Cryst. 29, (1991) 409-411 and Cudney, R., Patel, S., Weisgraber K., Newhouse, Y., and McPherson, A. Acta Cryst. D50, (1994) 414-423]. Although the initial screens did not give crystals, they indicated that polyethylene glycol (PEG) was the most promising precipitant. In subsequent trials we varied the molecular weight of PEG, buffer pH range (4-9) and effect of salts. Finally, a shower of tiny needles appeared in the drops equilibrated against a solution of PEG 9000 as a precipitant and acetate buffer, pH 5.0-7.0, with aggregated prisms appearing at higher pH, 6.5-7.5. However, all of these original crystals were too small for data collection. As seeding attempts were unsuccessful, we tried other ways to improve the crystal morphology and size, including a metal ion screen. Addition of cadmium chloride gave larger crystals which were often twinned or aggregated.
However, occasionally single crystals could be separated, one of which was used for data collection. The final conditions ware; 2.5u1 of the antibody-antigen complex mixed with 2.5u1 of the well solution containing 0.2 M
CdCl2, 19~ w/v of PEG 4000, and 0.1 M acetate buffer, pH
6Ø
Example 4 Diffraction measurements of the SM3 Fab fragment/MUC1 epitope crystal.
A monocrystal of dimensions 0.4 x 0.2 x 0.03 mm was used for the diffraction measurements at the X11 outstation of the DESY synchrotron in Hamburg. Analysis of the diffraction data showed that the crystal belongs to the monoclinic P2, space group with unit cell dimensions a=42.2, b=83.9, c=64.5 A, and (3=93.4°.
Assuming one complex molecule per asymmetric unit, the specific volume Vm 2.50 ~3/Da of protein corresponds to a solvent content of 51~ (Matthews, 1968). Since the number of available crystals was limited, data collection was carried out at 110 K. The crystal was transferred to a glycerol-enriched mother liquor (10$ of the water in mother liquor was substituted by glycerol) and then flash-frozen using standard techniques (Teng, 1990) in a stream of nitrogen gas produced by an Oxford Cryosystem.
A co~hplete dataset up to 1.95 ~1 resolution was collected using synchrotron radiation at wavelength 0.912 ~ (Table 3). The data was collected in 1.0° oscillation frames over 180° oscillation range on an l8cm MAR Research image plate from a single crystal. The frames were processed with DENZO software (Otwinowski, 1993) resulting in 111,363 observations with I>1.0 6(I). Further scaling with the CCP4 suite gave 33,093 independent reflections with Rmer9e of 6 . 9$ .
Example 5 Structure Determination and Model Refinement.
The CCP4 package [Collaborative Computational Project, No. 4 Acta Crystallogr. D50, (1994) 760-763].
was used for all crystallographic calculations unless otherwise stated. The structure of the SM3-MUC1 peptide complex was solved by the Molecular Replacement method as implemented in the program AMoRe [Navazza, J. Acta Crystallogr. A50, (1994) 157-163], using the Murine SE155-4 Fab fragment complexed with the dodecasaccharide [Cygler, M., Rose, D.R. and Bundle, D.R. Science 253, (1991) 442-445] as a search model. A consistent solution for the rotation and translation function was obtained only after removing the bound antigen, CDRs and N- and C-termini. Replacing all non-conserved residues with aianines and performing rigid body refinement with X-PLOR
. [Brunger, A.T. (1992) X-PLOR, A System for Crystallography and NMR (Yale Univ. Press, New Haven, CT) Version 3.1] resulted in a crystallographic R-factor of 40~, with R-free of 49~ as calculated against a randomly selected set of reflections (Brunger, A.T. Acta Crystallogr. D49, (1993) 24-36]. At this stage, several omit maps were calculated covering the whole molecule, and the model was manually rebuilt using the program 0 [Jones, T.A. and Kjeldgaard, M. (1994) 0 - The Manual.
Uppsala University, Sweden]. Further refinement of the model was performed with the standard X-PLOR simulated annealing protocol with resolution cut-off at 3.0 followed by manual rebuilding with 0 and Turbo-Frodo [A.
Russel and C. Cambillau, Marseille, France]. The R-factor converged to 39~ (R-free 43~) and substantial, continuously positive, electron density close to the CDRs became visible which corresponded to residues 2-10 of the peptide. The peptide was built into the density and further alternative rounds of refinement, DM solvent modification [Cowtan, K. Protein Crystallography 31, (1999) 34-38] and manual rebuilding was carried out using REFMAC as implemented in the CCP4 package and X-PLOR, gradually increasing the resolution to 1.95 l~.
Inspection of the 2Fo-Fc and Fo-Fc difference Fourier .
maps showed significant positive electron density corresponding to eight cadmium ions and a large number of solvent molecules. Several iterative refinement cycles were carried out, and the water and cadmium positions were checked after each cycle. All atoms were refined with individual isotropic B-factors. The Engh-Huber parameters [Engh, R.A. and Huber, R. (1991) Accurate bond and angle parameters for X-ray protein structure refinement. Acta Crystallogr. A47, 392-400] as used in X-PLOR were altered to allow the G1y96H-G1n97H bond to adopt a cis conformation. No electron density was visible for residues 128-133 of heavy chain as well as some terminal residues (L2, L209-212, H214, P1 and P11-13). For the same reason residues Asp4lL, G1u123L, G1n163L, GlnlH, G1u42H, Glu6lH, G1u85H, Ser134H, Ser160H, Ser172H, Asp173H, Ser203H, Lys208H and Ser2P were modelled as Alanines. The final model contains 3227 atoms including the peptide, 333 water molecules, two chlorine and eight cadmium ions and the final refinement statistics are shown in Table 4. The overall quality of the model was checked with PROCHECK program. [Laskowski, R.A., MacArthur, M.W., Moss, D.S. and Thornton, J.M. J.

Appl. Crystallogr. 26, (1993) 283-291]. Buried contact surface areas were calculated with the program PDBAREA
[Lee et al.J. Mol. Biol (1971), 55, 379-400 ]. It was noted that during the refinement and manual rebuilding procedures, some of the SM3 primary sequence was incorrect. A cDNA corresponding the SM3 Fab region was sequenced confirming the errors in the SM3 sequence.
The peptide was unambiguously built into the electron density from difference Fourier maps. Since the MUCl peptide is almost symmetrical, it was necessary to confirm the correct orientation for the peptide in the electron density. This was done by modelling the peptide backwards which resulted in a poorer fit into the density, and a higher R-Free and crystallographic R-factor after one round of refinement.
The correctness of the cis-bond was confirmed by trying to refine it as a traps-bond. This resulted in a significant increase in main chain B-factors for Gly96H
and subsequently poor quality electron density maps in this region (data not shown). Tn addition, some potential hydrogen bonds between the peptide and antibody are not possible if the peptide bond is in traps conformation. Unrestrained refinement using REFMAC
(CCP4, 1994) always resulted in a cis-conformation for the G1y96H-G1n97H peptide bond, the electron density for which is shown in Figure 3.
Example 6 The SM3 Fab fragment/MUC1 epitope structure.
The structure of the SM3-peptide antigen complex has revealed a number of unexpected..and new insights into how antibodies recognise peptide antigens. In particular CDR loop H3 contains a non-proline cis-peptide bond. The structure of the SM3 fab fragment/MUC1 epitope was checked to a resolution of 1.95 A. All of the residues of the epitope lie within favoured regions of the Ramachandran plot (data not shown) except for two residues (Thr5lL, Ser93L) on CDRs (L2 and L3), both of which are involved in interactions with the bound peptide. The individual atomic B-factors are within reasonable limits (average 16.32), showing higher values only in residues located on mobile loops or solvent-exposed fragments (data not shown). The quality of the density corresponding to nine residues of the MUC1 peptide antigen is shown in Figure 2A. The peptide residues have average atomic B-factors of 15.3Az indicating that the observed peptide is well ordered.
There are eight cadmium sites in the SM3-peptide, five with full occupancy, and three with relative occupancy of 0.75. They are mainly making contacts with the nitrogen atoms of histidines and carboxylic group of Glu or Asp residues between symmetry related molecules and/or chains (data not shown). All have distorted octahedral co-ordination completed by bound waters and chloride ions. It appears that the cadmium ions stabilise contacts between chains or molecules and contribute to the crystal packing since crystallisation in the absence of cadmium produces tiny needles, too small for diffraction experiments. The SM3-cadmium structure is an unusual example of a crystal structure in which cadmium ions were introduced directly and not simply by replacing other bound metal ions.
SM3 - MUC1 peptide antigen interactions The MUC1 peptide forms two intra-peptide hydrogen bonds: Pro 4 (atom 0) to Thr 6 (atom N) and Asp 5 (atom OD1) to Arg 7 (atom N).
The MUC1 peptide sits in an elongated groove of the SM3 antibody and is surrounded by all six CDR loops .
The peptide is anchored by the electrostatic interactions of AspSP and Arg7P and is bound to the SM3 antibody primarily by hydrophobic contacts (Table 4). All six CDRs of the SM3 antibody participate to some degree in the MUCI peptide binding, although of the residues on the H2 chain only Arg52H makes contact with the peptide, and this through a water mediated interaction. The major contribution to antigen binding comes from H1, of which three residues Asn3lH, Tyr32H, Trp33H are making contacts with seven residues of the peptide (Table 4). These interactions cover 38$ of the total CDRs surface area which is buried upon antibody-peptide binding, and are mainly hydrophobic, although there are some hydrogen bonds and salt bridges. The cis-peptide bond between G1y96H and G1n97H plays a crucial role in antigen binding I5 (Table 4). The nitrogen atom of G1y96H forms hydrogen bonds with both the carbonyl oxygen of AspSP and the OD2 carboxyl group via a water molecule. The hydrophobic part of the G1n97H side chain interacts with Pro9P which is oriented between Trp33H (H1) and Tyr32L (L1). Pro4P
also stacks against Trp9lL (L3), making Pro4P the most buried residue of the whole peptide antigen. The polar headgroup of G1n97H hydrogen bonds directly with residues of L1, L2 and AspSP. The important antigenic residue Arg7P interacts with Tyr32H (H1) and Asn3lH (H1) whilst the C-terminal residues of the peptide sit in a hydrophobic pocket surrounded by Pro56L (L2), Phe27H (H1) and Tyr102H (H3). In summary, there are a number of direct water-mediated hydrogen bonds between the peptide and antibody (Table 4). The rest of the SM3-peptide interactions are primarily hydrophobic.
Hydrophobic contacts are dominant at the SM3 antibody combining face since out of a total 185 X12 antigen-antibody contact area, 128 ~2 is covered by hydrophobic interactions (Table 5). A similarly high ' 35 ratio is also observed for the peptide antigen where 66$

of the buried surface area (total 603 ~2 ) is represented by non-polar residues. Such high ratios would suggest that hydrophobic interactions are the main driving force in SM3-peptide complex formation. The average Fab provides space for binding about 10 residues of peptide antigen, leaving the rest highly mobile and hence very often disordered and unobserved in electron density maps.
There is a clear surface complementarity between the SM3 antibody and peptide antigen. Of particular interest is the surface complementarity around Pro4P which appears to be facilitated by the presence of the cis-peptide bond within H3 (see later).
A fragment search against a library of all antibody structures deposited in the Brookhaven database (Bernstein et al., 1977) was unable to find a suitable replacement for H3. The best solution of the search, which used distances across the base of H3, placed the fragment Ca atoms at 2.6 ~ root mean squared separation from the equivalent crystal structure co-ordinates of H3.
However, it is well known that conformations of CDR H3 are difficult to classify and predict (Reczko et al., 1995; Shirai et al., 1996). In the SM3-peptide antigen structure, residues from CDRs H1 and to a lesser extent H3 are the major contributors to the binding of peptide antigen (Table 4). This is in contrast to other antibody-peptide complexes where L3 and H3 (MacCallum et al., 1996) or H2 and H3 (Stanfield and Wilson, 1993) are the main contributors to antigen binding. One obvious conclusion from this is that it would be difficult to predict, without a crystal structure of a complex, the expected contributions of each of the CDRs to peptide antigen binding.
The presence of non-proline cis-peptide bonds within proteins is very rare with an estimated 0.05$ of known protein structures containing such bonds (Stewart et al., 1990; Herzberg and Moult, 1991). To our knowledge, the SM3-peptide complex represents the first such example in a hypervariable loop region of an antibody combining site. Non-proline cis-peptide links have however, been found in the active sites of a number of hydrolases (Perrakis et al., 1994). In these examples, the cis-peptide conformation is probably involved in facilitating contact and/or recognition of the substrate (Perrakis et al., 1994). Interestingly, to IO date, all of the antibody-peptide structures contain a cis-proline residue within CDR L3, which is important for peptide-antigen binding. The reasons for the conservation of the cis-Pro is unclear, although it is, interesting that the two residues on either side of the cis-Pro, make at least one direct peptide contact and a second indirect contact, via other residue of the CDRs.
The SM3 structure however, does not contain the equivalent proline in L3, but appears to have the cis-peptide bond, albeit non-proline, on the opposite side of the antibody combining face in H3. At least two conclusions can be drawn from these observations: (1) A
cis-peptide bond within the antibody-antigen combining site is an important and common feature of peptide-antigen binding; (2) The SM3-peptide complex differs from all other antibody-peptide complexes in that the cis-peptide bond is non-proline and is located in another CDR
loop. As to the role of the cis-peptide bond, it may relate to specific interactions with the hydrophobic surface on the peptide antigen. An additional reason for this bond to be cis rather than trans, may be due to the small size of H3 resulting from the extended peptide antigen epitope running across its upper surface. A
small all-trans CDR would have less degrees of freedom to accommodate such a change.

Comparison of bound and unbound MUC1 peptide antigen A comparison of the SM3 bound MUC1 peptide with the unbound solution structure obtained from NMR studies of three MUCl repeats (Fontenot et al., 1995) is shown in Figure 4. The structure of the core region within each repeat in the solution structure is different to that of the antibody-bound peptide antigen. The overall geometry of the "knob-like" protrusion has changed from two consecutive type II (3-turns as observed in solution (Fontenot et al., 1995) to that of an essentially extended geometry when bound to the SM3 antibody. There are differences in all of the phi and psi torsion angles between the bound and unbound peptide, with the main I
differences centred on the phi angles of AspSP and Arg7P, which together make a significant percentage of the total peptide-contacts (Table 4).
This rearrangement could keep the charged head groups partially solvated on antibody binding, thereby presenting a predominantly hydrophobic surface to the SM3 antibody combining site. Thus, if the NMR structure of the MUC1 repeats represents the most prominent conformation in vivo (Fontenot et al., 1995), then the SM3 antibody induces a conformational change in the mucin molecule upon binding, at least within the region of the peptide epitope. This would suggest that SM3-peptide recognition is by an "induced fit" mechanism.

References Bernstein, F.C., Koetzle, T.F., Williams, G.J.B., Meyer, E.F., Jr., Brice, M.D., Rogers, J.R., Kennard, O., Shimanouchi, T., Tasumi, M. (1977) The protein data bank: A computer based archival file for macromolecular structures. J. Mol.
Biol.
112, 535-542 Brunger, A.T. 19992) X-PLOR, A System for Crystallography and NMR
(Yale Univ. Press, New Haven, CT) Version 3.1 Brunger, A.T. (1993) Assessment of phase accuracy by cross validation - the free R-value - methods and applications. Acts Crystallogr. D49, 24-36 Burchell, J. and Taylor-Papadimitriou, J. (1993) Effect of modification of carbohydrate side chains on the reactivity of antibodies with core-protein epitopes of the MUC1 gene product. Epithelial Cell Biol. 2, 155-162.
Burchell, J., Gendler, s., Taylor-Papadimitriou, J., Girling, A., Lewis, A., Millis, R. and Lamport, D. {1987) Development and characterisation of breast cancer 2 0 reactive monoclonal antibodies directed to the core protein of the human milk mucin. Cancer Res. 47, 5476-5482.
Burchell, J., Taylor-Papadimitriou, J., Boshell, M., Gendler, D. and Duhig, T.
(1989) A short sequence within the amino acid tandem repeat of a cancer associated 2 5 mucin, contains immunodominant epitopes. Int. J. Cancer 44, 691-696.
Chothia, C., Lesk, A.M., Tramontano, A., Levitt, M., Smith-Gill, S.J., Air, G., Sheriff, S., Padlan, E.A., Davies, D., Tulip, W.R., Colman P.M., Spinelli, S., Alzari, P.M. and Poljak, R.J. {1989) Conformations of immunoglobulin 3 0 hypervariable regions. Nature (London 342, 877-883.

Chothia, C., Lesk, A.M., Gherardi, E., Tomlinson, LM., Walter, G., Marks, J.D., Llewelyn, M.B. and Winter, G. (1992) Structural repertoire of the human VH
segments. J. Mol. Biol. 227, 799-817.
Collaborative Computational Project, No. 4 (1994) The CCP4 Suite.
Programs for crystallography. Acta Crystallogr. D50, 760-763.
Connlly, M.L. (1983) Analytical molecular surface calculation. J.A ppl.
Crystallogr. 16, 548-558 Cowtan, K. (1994) Joint CCP4 and ESF-EACBM Newsletter. Protein Crystallography 31, 34-38.
Cudney et al. (1994) Acts Cryst. D50, 414-423 Cygler, M., Rose, D.R. and Bundle, D.R. (1991) Recognition of a cell-surface oligosaccharide of pathogenic salmonella by an antibody fab fragment. Science 253, 442-445.
2 0 Dokurno, I'., Lally, J.M., Bates, P.A., Taylor-Papadimitriou, J., Band, H.A., Sanry, D. and Freemont, P.S. (1997) Crystallisation of an atitumour antibody complexed with a peptide epitope. Acts Crystallogr. D (in press).
Engh, R.A. and Huber, R. (1991) Accurate bond and angle parameters for X-ray 2 5 protein structure refinement. Acts Crystallogr. A47, 392-400.
Fontenot, J.D., Mariappan, S.V.S., Catasti, P., Domenech, N., Finn, O.J. and Gupta, G. (1995) Structure of the tumor associated antigen containing a tandemly repeated immunodominant epitope. J. Biomol. Struct. and Dynam. 13, 245-260.

Gendler, S., Taylor-Papadimitriou, J., Duhig, T., Rothbard, J. and Burchell, J.
(1988) A highly immunogenic region of a human polymorphic epithelial mucin expressed by carcinomas is made up of tandem repeats. J. Biol. Chem. 263, 12823.Gendler, S.J., Lancaster, C.A., Taylor-Papadimitriou, J., Duhig, T., Peat, N.Burchell, J., Pemberton, L., Lalani, E.-N. and Wilson, D. (1990) Molecular cloning and expression of the human tumour-associated polymorphic epithelial mucin. J. Biol. Chem. 265, 15286-15293.
Ghiara, J.B., Ferguson, D.C., Satterthwait, A.C., Dyson, H.J. and Wilson, LA.
(1997) Structure-based design of a constrained peptide mimic of the HIV-1 V3 loop neutralization site. J. Mol. Biol. 266, 31-39.
Girling, A., Bartkova, J., Burchell, J., Gendler, S., Gillett, C. and Taylor-Papadimitriou, J. (1989) A core protein epitope of the polymorphic epithelial mucin detected by the momoclonal-antibody sm-3 is selectively exposed in a range of primary carcinomas. Int. J. Cancer 43, 1072-1076.
Granowska, M., Biassoni, L., Carroll, M.J., Howell, R., Mather, S.J., Ellison, d., Granowski, A. and Britton, K.E. (1996) Breast cancer ~'"'Tc SM3 2 0 radioimmunoscintigraphyAct Oncologica 35, 319-321.
Hanisch, F.G., G. Uhlenbruk, K.J. Peter, H. Egge, J. Dabrowski, and U.
Dabrowski (1989) Structures of neutral O-linked polylactosaminoglycans on human skim milk mucins. A novel type of linearly extended poly-N-2 5 acetyllactosamine backbones with Gal beta(1-4)GlcNAc beta(1-6) repeating units. J.
Biol. Chem. 264, 872-883.
Herzberg, O. and Moult, J. (1991) Analysis of the steric strain in the polypeptide backbone of protein molecules. Proteins 11, 223-229.

Jancarik et al. (1991) J. Appl. Cryst. 24, 409-411.
Hull, S.R., A. Bright, K.L. Carraway, M. Abe, D.F. Hayes, and D.W. Kufe (1989) Oligosaccharide differences in the DF3 sialomucin antigen from normal human milk and the BT-20 human breast cancerinoma cell line. Cancer Commun. 1, 261-267.
Jones, T.A. and Kjeldgaard, M. (1994) O - The Manual. Uppsala University, Sweden.
Jentoft, N. (1990) Why are proteins O-glycosylated? Trends Biochem. Sci. 15, 294.
Kjeldsen, T., Clausen H., Hirohashi, S., Ogawa, T.,Iijima, H. (1988) Preparation and characterization of monoclonal-antibodies directed to the tumor-associated o-linked sialosyl-2-6 alpha-n-acetylgalactosaminyl (sialosyl-tn)epitope. Cancer Res. 48, 2214-2220.
Kraulis, P. (1991) MOLSCRIPT: a program to produce both detailed and 2 0 schematic plots of protein structures. J. Appl. Crystallogr., 24, 946-950.
Laskowski, R.A., MacArthur, M.W., Moss, D.S. and Thornton, J.M. (1993) PROCHECK: a program to check the stereochemical quality of protein 2 5 structures. J. Appl. Crystallogr: 26, 283-291.
Lee, B., and Richards, F.M. (1971) The interpretation of protein structures.
J. Mol.
Biol, 55 , 379-400.
30 Lloyd, K., Burchell, J., Kudryashov, V., Yin, B.W.T. and Taylor-Papadimitriou, J.
(1996) Comparison of Q-linked Carbohydrate Chains in MUC-1 Mucin from *rB

_99_ Normal Breast Epithelial Cell Lines and Breast Carcinoma Cell Lines. J Biol.
Chem., 271, 33325-33334.
Matthews et al. (1968) J. Mol. Biol. 33, 491-497.
MacCallum, R.M., Martin, A.C.R. and Thronton, J.M. (1996) Antibody-antigen interactions: contact analysis and binding site topography. J. Mol. Biol. 262, 745.
Navazza, J. (1994) AMORE: An automated package for molecular replacement.
Acta Crystallogr. A50, 157-163.
Nicholls, A., Sharp, K. and Honig, b. (1991) Protein folding and association:
insights from the interfacial and thermodynamic properties of hydrocarbons.
Proteins 11, 281-296.
Otwinowski (1993) DENZO. An oscillation data processing program for macromolecular crystallography. Yale University, New Haven, CT, USA.
2 0 Perrakis, A., Tews, L, Dauter, Z., Oppenheim, A.B., Chet, L, Wilson, K.S.
and Vorgias, C.E. 91994) Crystal structure of bacterial chitinase at 2.3~
resolution.
Structure 2, 1169-1180.
Reczko, N., Martin A.C.R., Bohr. H. and Suhai, S. (1995) Prediction of 2 5 hypervariable cdr-h3 loop structures in antibodies. Prop Eng. 8, 389-395.
Rees, A.R., Staunton, D., Webster, D.M., Searle, S.J., Henry, A.H. and Pedersen, J.T. (1994) Antibody design - beyond the natural limits. Trends in Biotech.
12, 199-206.

Shirai, H., Kidera, A and Nakamura, H. (1996) Structural classification of cdr-h3 in antibodies. Febs. Lett. 399, 1-8.
Stanfield, R.L. and Wilson, LA. (1993) X-ray crystallographic studies of antibody-peptide complexes. Immunometh. 3, 211-221.
Stewart, d.E., Sarkar, A and Wampler, J.E. (1990) Occurrence and role of cis peptide bonds in protein structures. J. Mol. Biol. 214, 253-260.
Teng (1990) J. Appl. Crystallogr. 23, 387-391.
Wandall, H.H., Hassan H., Mirgorodskaya, E., Kristensen, A.K., Roepstorff, P., Bennett, E.P., Nieklsen, P.A., Hollingsworth, MA., Burchell, J., Taylor-Papdimitriou, J. and Clausen H. (1997) Substrate specificities of three members of the human UDP-N-acetyl-a-D-galactosamine: Polypeptide -acetylgalactosaminyltransferase family, GaINAc-Tl, T2, and T3 J. Biol. Chem.
(in press).
Wallace et al.(1995) Protein Engineering. 8, 127-134. This software was used to 2 0 generate the LIGPLOT.
Wilson, LA. and Stanfield, R.L. (1994) Antibody-antigen interactions: new structures and new conformational changes. Curr. Opin. Struc~t. Biol. 4, 857-867.

Table 1 Structure factors of the SM3 fraament/crystal peptide crystal which is described in the Examples This table is shown below. The data is arranged in four separate columns as below. Each row of a column consisting of five numbers describes the results of a single reflection. Reading left to right, the first three numbers represent h, k and 1 respectively and are the crystallographic reflection indices. The fourth number represents the structure factor F(hkl) defined as the computed sum of the Fourier series for each reflection hkl. As multiple measurements are made of the same reflection a standard deviation for each structure factor has been provided as the fifth number in the row.
Tables 2a and 2b Structure !atomic coordinates) of the SM3 fraament/crystal peptide crystal which is described in the Examples These tables are shown below. The top row, called CRYST1, defines the spacegroup and crystal cell dimensions. Scale 1, Scale 2 and Scale 3 define an orthogonalisation matrix which when applied to the individual coordinates puts them into an orthogonal axis system i.e. 90 degrees between each of the axes x, y and z.
Below these rows reading left to right the second row gives the chemical symbol of the atom. OH2 represents a water molecule. The letter following indicates the position of the atom in the amino acid residue using a convention recognised by a skilled partisan. The letter refers to the equivalent Greek symbol (i.e. A for alpha, _q7_ B for beta etc.). The third column gives the identity of the residue in which the atom is present. WAT represents a water molecule. The fifth column gives the number of the amino acid in the protein. The sixth, seventh and eighth columns give the spatial position of the atom as x, y and z coordinates respectively. The ninth column shows the occupancy of that atom which will be 1.0 for present and 0.0 for absent (i.e. not observed) and the tenth column gives the B-factor which refers to the mobility.

00 3 ~ o .
DEMANDES OU BREVETS VOLUMlNEUX
LA PRESENTS PARTiE DE CETTE DEMANDS OU CE BREVET
COMPREND PLUS D'UN TOME.
CECI EST LE TOME ~ DE
NOTE: Pour les tomes additionels, veuillez cantacter le Bureau canadien des brevets JUMBO A,PPLlCATIONS/PATENTS
~f HIS SECTION Of= THE APPL1CAT10N/PATENT CONTAINS MORE
'if 1-lAN ONE VOLUME
THIS t~~ VOLUME j OF
' NOTE: For additional volumes-.phase contact the Canadian Patent Ofific~

Claims (33)

-287-

1. Use of the structure factors obtainable by subjecting a crystal comprising at least an epitope binding fragment of the SM3 antibody, bound to a peptide recognised by the epitope binding site of SM3 to X-ray diffraction measurements to identify, screen, characterise, design or modify a chemical entity.

2. Use of the structural coordinates obtainable by subjecting a crystal comprising at least the epitope binding fragment of the SM3 antibody, bound to a peptide recognised by the epitope binding site of SM3 to X-ray diffraction measurements and deducing the structural coordinates from the diffraction measurements, to identity, screen, characterise, design or modify a chemical entity.

3. Use according to claim 1 or claim 2 in which the fragment of the SM3 antibody is the Fab fragment obtainable by digesting SM3 with papain.

4. Use according to claim 1 or claim 2 in which a moiety substantially similar to a fragment of the SM3 antibody is used instead of the fragment of the SM3 antibody.

5. Use according to any one of the preceding claims in which the peptide comprises the sequence Pro-Asp-Thr-Arg-Pro.

6. Use according to claim 5 in which the sequence is Thr-Ser-Ala-Pro-Asp-Thr-Arg-Pro-Ala-Pro-Gly-Ser-Thr.

7. Use of the structure factors as shown in Table 1 to identify, screen, characterise, design or modify a chemical entity.

8. Use of the structural coordinates as shown in Table 2a or 2b to identify, screen, characterise, design or modify a chemical entity.

9. Use according to any one of the preceding claims of the structure factors or structural coordinates to identify, screen, characterise or design a mimic of SM3.

10. Use according to claim 9 in which the mimic of SM3 has a desired association with aberrantly glycosylated MUC1.

11. Use according to claim 10 in which the desired association is binding affinity with aberrantly glycosylated MUC1 which is higher than the binding affinity of SM3 for aberrantly glycosylated MUC1.

12. Use according to any one of claims 9 to 11 in which the mimic of SM3 is obtainable by modification of either whole SM3 or a fragment of SM3.

13. Use according to claim 12 in which the modification is the replacement of Proline 56 of SM3.

14. Use according to any one of the preceding claims of the structure factors or structural coordinates to identity, screen, characterise or design a mimic of the peptide epitope of MUC1 recognised by SM3.

15. Use according to claim 14 in which the mimic of the peptide is used to produce a moiety which has a desired association with aberrantly glycosylated MUC1.

16. Use according to claim 15 in which the moiety is an antibody.

17. Use according to claim 16 in which the antibody is produced in a method comprising vaccinating an animal or human with the mimic of the peptide.

18. Use according to claim 14 in which the mimic of the peptide can prevent or decrease an immune response.

19. Use according to any one of claims 14 to 18 in which the mimic of the peptide is obtainable by a chemical modification of a peptide.

20. Use accordingly to any one of claims 14 to 19 in which the mimic of the peptide is a glycosylated peptide.

21. A chemical entity which is a mimic of the peptide epitope of MUC1 recognised by SM3 and not described in any database at the date of filing of the application.

22. A mimic of SM3 not described in any database at the date of filing of the application.

23. An antibody having a conformation comprising an engineered non-proline cis peptide bond.

24. A crystal comprising at least an epitope binding fragment of the SM3 antibody bound to a peptide recognised by the epitope binding site of SM3.

25. A method of producing a crystal of claim 24 which comprises contacting the fragment of SM3 or the peptide with cadmium.

26. A machine readable data storage medium comprising a data storage material encoded with machine readable data which when read by an appropriate machine is capable of displaying a three dimensional representation of a crystal as defined in claim 24.

27. A chemical entity, a mimic of SM3, a mimic of the peptide epitope of MUC1 recognised by SM3, a moiety having a desired association with aberrantly glycosylated MUC1 or an engineered antibody as defined in any one of claims 7 to 23 for use in a method for treatment of the human or animal body by surgery or therapy or of diagnosis practised on the human or animal body.

28. Use of a chemical entity, a mimic of SM3, a mimic of the peptide epitope of MUC1 recognised by SM3, a moiety having a desired association with aberrantly glycosylated MUC1 or an engineered antibody as defined in any one of claims 7 to 23 in the production of a medicament for use in a method for anti-cancer treatment of the human or animal body by surgery or therapy or of cancer diagnosis practised on the human or animal body.

29. A method of cancer diagnosis or anti-cancer therapy comprising administering to a human or non-human animal in need thereof an effective non-toxic amount of a chemical entity, a mimic of SM3, a mimic of the peptide epitope of MUC1 recognised by SM3, a moiety having a desired association with aberrantly glycosylated MUC1 or an engineered antibody as defined in any one of claims 7 to 23.

30. A product comprising a mimic of SM3 or a specific binding agent and an anti-tumour agent as a combined preparation for simultaneous, separate or sequential use in anti-cancer therapy.

31. A mimic of the MUC1 epitope recognised by SM3 for use in a method of treating a disease caused by an immune response.

32. A method of therapy of a disease caused by an immune response comprising administering to a human or non-human animal in need thereof an effective non-toxic amount of a mimic of the MUC1 epitope recognised by SM3.

33. Use of the structure factors and/or structural coordinates as defined in any one of claims 1 to 8 to solve the structural coordinates of a crystal.