WO2004081051A1

WO2004081051A1 - Bispecific antibodies

Info

Publication number: WO2004081051A1
Application number: PCT/GB2004/001026
Authority: WO
Inventors: Stephen Peter Young
Original assignee: The University Of Birmingham
Priority date: 2003-03-12
Filing date: 2004-03-11
Publication date: 2004-09-23
Also published as: GB0305702D0

Abstract

The present invention discloses a bispecific antibody (BAb) comprising two antibodies, each of which has a binding specificity to a different epitope situated on the surface of a target structure. Each of said antibodies has a relatively low binding affinity for its respective epitope. The BAbs produced according to the present invention have much lower affinity for cross-reactive non-target tissue due to the lower affinity of the MAbs used to produce them. These BAbs still provide high avidity for target tissue due to the cumulative nature of the binding interactions.

Description

BISPEOFIC ANTIBODIES

The present invention relates to the production of bispecific antibodies providing increased specificity toward target tissues over non-target tissues. The production of antibodies having improved specificity for target tissues and cells and with reduced cross-reactivity to non-target tissues has long been recognised as desirable by the medical profession.

The production of monoclonal antibodies (MAbs) through the fusion of B cells raised against a specific antigen with myeloma cells to produce hybridomas is well known. Each hybridoma produces a different monoclonal antibody to the specific antigen. Generally, those antibodies with the highest affinity for the antigen are selected as these give the strongest binding interaction, the rest of the antibodies being discarded. A monoclonal antibody has specificity for a single epitope of an antigen, but is normally bivalent (i.e. has two binding interactions) for this epitope providing improved binding characteristics.

More recently, bispecific antibodies have been developed in which two monoclonal antibodies having specificities for different antigens are coupled together (EP 0468637A). It has been suggested that these bispecific antibodies may be useful in increasing the relative amount of antibody-linked agent delivered to target tissues in vivo compared to the amount of such agent delivered to non-target tissues (EP 0331034B). The individual monoclonal antibodies having differing specificities are generally chosen because of their strong binding affinity for their respective epitopes and are cross-linked to each other to produce the bispecific antibody. The bispecific antibody provides some selectivity, binding more strongly to cells where both epitopes are expressed compared to those where only one is expressed, and may even have reduced affinity for cells expressing a single epitope when compared to the monoclonal antibody (EP 0468637). However, these bispecific antibodies still bind to cells expressing only one of the epitopes, resulting in cross-reactivity to non-target cells.

It is an object of the present invention to provide an improved, biologically useful, bispecific antibody.

According to a first aspect of the present invention there is provided a bispecific antibody comprising two antibodies, each of said antibodies having a binding specificity to a different epitope situated on the surface of a target structure, wherein each of said antibodies has a relatively low binding affinity for its respective epitope.

EP 0468637 discloses a method of in vivo targeting specific cells using high affinity commercially available or hybridoma produced monoclonal antibodies to produce bifunctional antibodies. These bifunctional antibodies provide some selectivity for target cells by providing greater avidity to target cells than cross-reactive non-target cells leading, due to the dynamism of the system, to more bispecific antibody being bound to the target than non-target at equilibrium. However, due to the high affinity of the starting MAbs for their specific epitopes, the BAb disclosed in EP 0468637 still has a relatively high affinity for cross-reactive non- target tissue. The BAbs produced according to the present invention have much lower affinity for cross-reactive non-target tissue due to the lower affinity of the MAbs used to produce them. These BAbs still provide high avidity for target tissue due to the cumulative nature of the binding interactions.

The binding affinity of the antibodies is measured in terms of the affinity constant, which is the concentration of an antibody required to cause 50% saturation of a specific antigen. The lower the concentration required, the greater the binding affinity. As used herein, relatively low binding affinity is defined as an affinity constant of at least 10^"8 M. Conventionally, monoclonal antibodies are typically selected with an affinity constant of 10^"11M or lower.

As used herein, the term "antibody" in the context of the antibodies which comprise the BAbs of the present invention includes an intact antibody molecule, an antibody fragment, for example Fab, Fab' or F(ab')2, or functional equivalents of the antibody/antibody fragment whether naturally occurring, genetically engineered or chemically synthesised.

The target structure may be a cell, collection of cells, bodily tissue or organ.

The bispecific antibody may also comprise a linker and/or spacer to which the antibodies are attached. Said linker/spacer may be a polymer (preferably non-biodegradable under physiological conditions) such as polyethylene glycol (PEG). Advantageously, the presence of the linker/spacer permits binding of epitopes which may be in different domains on the surface of the target structure.

In a preferred embodiment of the present invention, the affinity constant of each of the antibodies comprising the bispecific antibody is at least 10^"7 M, more preferably, from 10^"7 M to 10^"4 M, and most preferably from 10^~7 M to 10^"6 M. It will be understood that each of the antibodies comprising the bispecific antibody may have a different affinity constant for its specific epitope.

The bispecific antibody may comprise a therapeutic or diagnostic agent which is desirably targeted to said target structure. Examples of suitable classes of agents include, drugs (especially cytotoxic drugs), pro-drugs, T- cells or other immune system cells (e.g. an antibodies) or target markers, including fluorescent, luminescent and radioactive markers, liposomes, complement, inhibitory RNA (RNAi) or suitable plasmids encoding for RNAi sequences, and retroviral vectors coding sequences to "knockdown" or "knock-in" genes of the target structure.

Specific therapeutic agents which may be used include, but are not limited to, ^13l l, ⁶⁷Cu, ⁹⁷Ru and ¹⁸⁸Re, anti-cancer drugs such as nitrogen mustards, cytotoxic antibiotics such as bleomycin, cis- diaminodichloroplatinum, alkylating agents, antimetabolites such as 5- fluorouracil, toxins, for example mycotoxins such as trichothecenes, ricin or abrin and biological response modifiers, for example interferons, interleukins or TNF.

Suitable diagnostic agents include, but are not limited to, ⁷⁶Br, ¹⁸F or ¹²³l or NMR imaging contrast agents.

The therapeutic or diagnostic agent may be covalently or otherwise linked to the remainder of the bispecific antibody, said linking being at one or both antibodies and/or at the linker/spacer when present. The combination of an antibody with a therapeutic/diagnostic agent is often referred to as an "immuno-conjugate". Methods of producing immuno- conjugates to target pre-selected sites in the body are well known (US 4801688, US 4845200, Ghose et a/, [1983] Methods in Enzymol. 93, 280-333.), Garnett [2001] Advanced Drug Delivery Reviews 53, 1 71-21 6 and Trail and Bianchi [1999] Current Opinion in Immunology 1 1 , 584- 588.

Alternatively or in addition, the bispecific antibody may comprise a receptor for a therapeutic/diagnostic agent.

It will be understood that said receptor may be an Fc antibody fragment . The Fc fragment can serve as a receptor for complement and cells expressing Fc receptors such as neutrophils and macrophages.

Preferably, monoclonal antibodies isolated from cells produced during the primary immune response are used to produce the bispecific antibody. It will be understood that by primary immune response it is meant the immune response to an initial exposure of antibody-producing cells to an antigen. However, it is also contemplated that, depending upon the particular antigen used, it may be necessary to provide a second exposure of the antibody-producing cells to obtain antibodies having an appropriate binding affinity.

In a preferred embodiment, B-cells activated during the primary immune response caused by the first exposure of the cells to an antigen are isolated and fused with myeloma cells to produce hybridomas. The fusion of somatic cells to produce MAbs was first described by Kohler and Milstein in 1975 and provides a relatively simple method of producing and screening large numbers of MAbs to a single antigen. Usually, the hybridomas created by the cell fusion experiment will produce a number of antibody specificities against the same antigen (i.e. antibodies specific to different regions of the same antigen), each antibody having a different affinity. The production of a panel of antibodies is advantageous as some isotypes are better suited to the enzymatic reactions used in the chemical construction of BAbs than others.

It will be understood that the isolated antibody producing cells from the primary immune response are B-cells or plasma cells. They are B-cells or plasma cells which are likely to produce IgM or IgG antibodies. IgM antibodies are the first class of antibodies to appear after exposure to an antigen and generally have lower affinity for their specific epitopes than immunoglobulins produced later in the immune response or upon secondary exposure to an antigen. The bispecific antibody may be produced by any known method such as chemical cross-linking, or cross-linking using anti-lgG Fc antibodies followed by covalent cross-linking. They may also be produced biologically by fusing two antibody producing hybridoma cell lines together. In cases where the antibodies are linked to a linker/spacer, the linker/spacer may participate in the cross linking reaction to produce the bispecific antibody.

It will be understood that the invention is equally applicable to quadrivalent or bivalent BAbs derived from bivalent or monovalent antibodies respectively.

According to a second aspect of the present invention there is provided a method of selecting monoclonal antibodies for use in production of a bispecific antibody according to the first aspect, comprising screening a large number of hybridomas which produce antibodies specific for each epitope of interest by direct measurement of their binding affinity for their specific epitope, for example, by surface plasmon resonance, so allowing antibodies having differing affinities for the specific epitopes of interest to be selected.

According to the third aspect of the present invention there is provided a method of treating a patient comprising, administering to a patient a therapeutical ly effective amount of a BAb in accordance with the first aspect, said BAb having a receptor for a therapeutic agent, in conjunction with prior, concomitant or subsequent administration of the therapeutic agent. Alternatively, the method may comprise administering to a patient an immu no-conjugate, said immuno-conjugate comprising a BAb which includes a therapeutic agent in accordance with said first aspect.

Administration of said therapeutical ly effective amount of a BAb may be by any known route e.g. by intravenous, intramuscular, or subcutaneous injection, topical administration as an ointment, salve, cream or tincture, oral administration as a tablet, capsule, suspension or liquid and nasally as a spray (e.g. aerosol). The preferred route of administration is intravenous injection.

In each case, said BAb may be in admixture with one or more excipients, carriers, pH regulators, flavourings, colourings, preservatives, or other commonly used additives in the field of pharmaceuticals as appropriate for the mode of administration.

According to the fourth aspect of the present invention there is provided a method of diagnosing a disorder comprising, contacting an immuno- conjugate comprising a bispecific antibody and a diagnostic agent according to the first aspect of the present invention with a tissue, or contacting a bispecific antibody having a receptor for the diagnostic agent according to the first aspect of the present invention with a tissue prior to, concomitantly with or subsequently to addition of the diagnostic agent to said tissue.

Preferably, said tissue is an isolated tissue sample. Embodiments of the present invention will now be described, by way of example only.

Methods

Immunisation protocol

Standard protocols for the production of monoclonal antibodies in rats or mice are used.

Antigen, in the form of 2 x 10⁶ to 5 x 10¹⁰ cells or 1 to 50 μg protein or peptide in normal saline is prepared for each animal to be immunized.

The antigen is drawn into a sterile 1- to 2 ml glass syringe with a Luer-Lock tip, which is then connected to a 3-way stopcock. A volume of complete Freund's adjuvant (CFA) (Sigma-Aldrich, Poole, UK) equal to the antigen volume is drawn into another syringe and connected to the antigen- containing syringe. Emulsification of the antigen and CFA is achieved by discharging the antigen into. the CFA, then discharging back and forth until a thickened mixture results. To test whether the emulsion is stable, a drop is placed in water (a stable emulsion will not disperse). The CFA/antigen emulsion is drawn into one syringe and a sterile 20-G needle attached, the emulsion is then injected intraperitoneally into the animals using < 0.2 ml per mouse, 0.5 to 1 ml per rat, or 0.2 to 0.4 ml per hamster. After 10 to 14 days the animals are boosted with the same dose of antigen. For cell fusions plan ned for 3 days after boosting, the boost immunization is done with antigen alone in aqueous solution, or intact cells in suspension. For fusions after a longer period, the boost is done with antigen emulsified in incomplete Freund's adjuvant (IFA).

After 14 days the spleens of some animals are removed for generation of hybridomas. At this stage there is a greater likelihood of isolating low affinity antibodies. However, the response to a particular antigen may be very poor, and so other animals receive a booster immunisation in incomplete adjuvant and spleens are removed after a further 14 days. It may be necessary to repeat this process, depending on the nature of the antigen.

Hybridoma fusion

Standard protocols (Kohler and Milstein (1975) Nature 256, 495-497) are used to create the hybridomas from the spleen cells and these are grown in selection medium prior to screening. One week before fusion, expansion of the SP2/0-Ag14 myeloma cell line (European Collection of Cell Cultures, Salisbury, UK) (the fusion partner cell line) is begun in complete DMEM/HEPES/ pyruvate in 175cm³ flasks containing 100 ml of medium. By the day cell fusion is to be performed, the following total number of myeloma cells must be available: for mouse spleen, 1 x 10⁸ cells; hamster spleen, 2 x 10⁸; and rat spleen, 5-10 x 10⁸ cells at approximately 0.5 x 10⁶ cells per ml. Two mouse or hamster spleens, or one rat spleen, provides enough cells for the fusion.

One day before fusion, all reagents and media are prepared, particularly 50% PEG and the SP2/0-Ag14 myeloma cells are split into fresh complete DMEM-10/HEPES/pyruvate medium, so that the cells are growing vigorously to ensure optimal fusions. The boosted animal(s) is/are sacrificed and the spleen(s) aseptically harvested. The spleen(s) is/are transferred into a sterile 100 mm-diameter petri dish filled with 10 ml sterile complete serum-free DMEM, placed in a laminar flow hood, and teased into a single-cell suspension by squeezing with angled forceps or by chopping with fine-tipped dissecting scissors. Debris is removed and the cells dispersed further by passage through a fine-mesh metal screen. The spleen cell suspension is transferred to a sterile 50 ml conical centrifuge tube filled with sterile complete serum-free DMEM, centrifuged for 5 min at 1500 rpm (500 x g) at room temperature and the supernatant discarded. The red blood cells (RBC) are lysed by re-suspending the pellet in 5 ml ammonium chloride solution (0.02 M Tris-CI, pH 7.2 0.14 M NH4CI ) and letting them stand for 5 min at room temperature. 45 ml sterile complete serum-free DMEM is added, and cells centrifuged as above. The cells are washed a further two times in 50 ml sterile complete serum-free DMEM. While the spleen cells are being washed, the SP2/0-Ag14 myeloma cells are harvested by centrifuging at 1500 rpm (500 x g) for 5 min at room temperature and washed three times in DMEM. The spleen and myeloma cells are separately re-suspended in 10 ml complete serum-free DMEM and the cells in each cell suspension counted and their viability assessed using a haemocytometer and trypan blue exclusion. The SP2/0-Ag 14 myeloma and spleen cells are mixed at a 1 :1 ratio in a 50 ml conical centrifuge tube which is then filled with complete serum-free DMEM, and the cell mixture centrifuged for 5 min at 500 x g at room temperature. While the cells are in the centrifuge, three 37°C double-beaker water baths are prepared in a laminar flow hood by placing a 400 m l beaker containing 1 00 ml of 37°C water into 600 ml beaker containing 75 to 100 ml of 37°C water. The tubes of pre-warmed 50% PEG solution and complete serum-free DMEM are placed into two of the 37°C water baths in the hood. The supernatant from the mixed-cell pellet is aspirated and discarded and the tube placed in the third double-beaker water bath in the laminar flow hood. Using a 1 ml pipette, 1 ml of pre-warmed 50% PEG is added to the mixed-cell pellet drop-by-drop over 1 m in, with stirring after each drop, the m ixture is then stirred for an additional minute. Using a clean pipette, 1 ml of pre-warmed complete serum- free DMEM is added to the cell mixture drop-by-drop over 1 min, stirring after each drop. This is repeated once with an additional 1 ml of pre-warmed complete serum-free DMEM. 7 ml pre-warmed complete serum-free DMEM is then added drop-by-drop over 2 to 3 min. The cell mixture is then centrifuged for 5 min at 500 x g, at room. temperature. While the cells are being centrifuged, the beaker water baths are re- warmed to 37°C and placed in the laminar flow hood, complete DMEM/HEPES/pyruvate is placed in the beaker in one water bath. The supernatant from the centrifuged fused cells is discarded and the tubes , placed in the second beaker water bath. 10 ml pre-warmed complete DMEM/HEPES/pyruvate is forcefully discharged with a clean 10 ml pipette on to the cell pellet. This process is repeated until the volume of pre- warmed complete DMEM-20 added means that the cells are at 16.1 x 10⁶ cells/ml. Any cell clumps are disrupted with the pipette tip, and 10 ml of cell suspension is gently aspirated with a 10 ml pipette. 2 drops (100 to 125 μl) of suspension is added to each well of a 96-well flat-bottom plate (continue until entire suspension is plated) which is incubated overnight in a humidified 37°C, 5% CO2 incubator. After one day of incubation, the wells are checked under an inverted microscope. If the wells are seeded with the appropriate number of cells, there will be a nearly confluent monolayer of highly viable cells on the bottom of the well and obvious clumps of cells. 2 drops of complete DMEM/HEPES/ pyruvate/HAT (DMEM medium containing 10 mM HEPES , 1 mM pyruvate, and 1 % 1 00x HAT (hypoxanthine/aminopterin/thym idine)) is added to each well with a 10 ml pipette and the plate placed in a humidified 37°C, 5% CO2 incubator. On days 2, 3, 4, 5, 7, 9, and 1 1 , half the volume of each well is aspirated using a sterile, short Pasteur pipette and the cells fed by adding 2 drops of complete DMEM/HEPES/ pyruvate/HAT from a 10 ml pipette to each wel l, the plates are then returned to a humidified 37°C, 5% CO2 incubator. On day 14, the feeding protocol is repeated except that complete DMEM/HEPES/ pyruvate/HT is used. On day 1 5 and subsequently, the wel ls are fed usi ng complete DMEM/HEPES /pyruvate without HAT or HT.

Screening for low affinity antibodies using Surface Plasmon Resonance

The hybridoma supernatants are screened for specific antibodies by directly measuring their binding affinity for antigen. Supernatants are removed from the multi-well plates in which the hybridomas are grown, concentrated and injected over a BIAcore chip (BlAcore, Stevenage, U K), which has been coupled with purified antigen.

For purified antigen, preliminary pH scouting is done to determine optimal pH of the electrostatic interaction of the antigen with the BIAcore CM5 chip surface (carboxymethyl groups). Antigen at 10 μg/ml is diluted in a series of buffers at pH 7.5, 7.0, 6.5, 6.0, 5.5 and 5.0 and pumj over the chip at a rate of 5 μl/min. The pH for optimal binding is selected from this series.

Based on the initial pH scan, the concentration of the antigen for coupling is adjusted to aim for three different levels of coupling: 1000, 5000 and 10,000 RU (response units). Antigen is immobilized onto the chips by amide coupling. 1 -(3-dimethylaminopropyl)-3-ethylcarbodiimide (EDC) (0.4 M) and N-hydroxysuccinimide (NHS) (0.1 M) solution are mixed at a ratio of 1 :1 and 35 μl injected over the CM5 chip at a rate of 5 μl/min to activate the surface. Antigen at 10 μg/ml in the selected pH coupling buffer is then injected at 2 μl/min until the required RU coupling is achieved. 50 μl of 1 M ethanolamine pH 8.0 is then injected to block remaining activated groups on the chip surface. This is repeated for the other two tracks to achieve the range of coupled antigen concentrations required. A carbodiimide /ethanolamine-treated blank for the first track is included as a control. The IgG sample is run over the tracks in order of increasing antigen coupled to avoid mass transport limits with the kinetic data.

Supernatants derived from the hybridomas are assayed for Ig content and isotype using a standard ELISA technique. To be able to detect the low affinity antibodies, the supernatants are concentrated to 30 μg/ml i.e. approximately 200 nM IgG. This concentration is high enough to observe binding, even if it is low affinity, in the μM range. This is done using 96 well plate format vacuum filter apparatus. Assays are performed with a mobile phase at a flow rate of 10 μl/min using 20 mM HEPES, 150 mM NaCI, 3.4 mM EDTA, 0.05% v/v surfactant P20 (pH 7.4). Supernatants in 96-well plates are pumped at 5μl/min over the antigen-coupled chip for 2 minutes to allow the binding reaction to occur, after which buffer alone is pumped over for 4 min to allow the dissociation to be observed. Assessment of the overall binding affinity can readily be made by comparing the dissociation rates.

Antibody is cleared from the antigen by an injection of low pH buffer (10 mM glycine pH 3.0) for 1 minute, followed by a wash in pH 7.5 Hepes buffer, the next antibody is then injected.

An initial estimate of affinity can be made from the on and off kinetics by curve fitting of the data. Those hybridomas having an affinity in a suitable range are selected for expansion into a larger scale culture. After this expansion the affinity is re-assessed using the BIAcore method. A titration of the Ig's at 30, 15, 7.5, 3.75, 1.875, 0.93, 0.47, 0.23, 0.12, 0.06 μg/ml (200, 100, 50, 25, 12.5, 6.25, 3.1 , 1.5, 0.75, 0.37 nM) is then done to give a better estimate of the binding affinity.

If purified antigens are not available, then crude preparations, cell membranes or whole cells may be coupled to the BIAcore chip and used in a similar fashion.

Cloning by limiting dilution

Once suitable hybridomas have been provisionally identified, they are re- cloned to make certain that they are definitely monoclonal. The candidate hybridoma lines are resuspended in their wells and the numbers and viability assessed using a haemocyto meter and Trypan blue. 1 0 m l of cel ls at 50 viable cel ls/m l and 1 0 m l at 5 viable cel ls/m l are prepared in cloning/expansion medium. A 96-well plate is seeded with the cel l suspensions at 200 μl/well and Incubated for 7 to 10 days in a humidified 37°C, 5% CO, incubator. The optimum dilution for monoclonal growth is identified by determining the number of wells that showed growing hybridomas. Wells are inspected for monoclonality with an inverted microscope by looking for tight single clusters of cells before the cells are fed. The supernatants are re-screened using ELISA and BIAcore as above.

A second re-cloning is done, after the desired clone has been expanded as above, at lower cell density by seeding two new 96-well plates at 0.3 cells/well (60 viable cells in 40 ml cloning medium). The screening process is repeated as above and the re-cloned hybridoma expanded.

The re-cloned cells. are weaned on to complete DMEM-10/HEPES/ pyruvate by splitting the cells 1 :2 every day for 3 days with complete DMEM-IO/HEPES/pyruvate.

Purification of antibodies

Individual hybridomas are gradually expanded in DMEM/HEPES/ pyruvate containing fetal bovine serum, which has been passed through a protein A-Sepharose affinity column to clear any residual immunoglobul ins from it. Supernatants from the hybridoma are col lected and Ig purified by passing the medium through a protein G-Sepharose column and eluting the bound antibody with 0.1 M glycine pH 2.0 and rapidly neutralizing with I M Tris pH 8.0.

Hybridoma-hybridoma fusions

The method for producing bispecific antibodies by fusing individual hybridomas of different specificities was originally described by Milstein (Milstein and Cuella 1983 Nature 305 537-540) and this approach is applied here to make the bi-specific low affinity hybridomas. Methods for screening for successful mixed hybrids have more recently been developed which involve differentially fluorescently labelling of the individual hybridomas, before fusion, and then sorting by flow cytometry of the cells expressing both labels (Shi et al. 1991 J.Immunol. Methods 141 (2):165-1 75.). This eliminates non-fused cells, non-mixed hybrids and dead cells. This method is used to prepare the low affinity bi-specific hybridomas before screening and characterisation of the secreted antibodies as below.

Cross linking of antibodies

Hetero-bifunctional antibodies are produced by chemical crosslinking or by crosslinking using anti-lgG Fc antibodies, followed by covalent crosslinking. Once the value of a particular coupled pair has been established the hybridomas are used as a source of material for further manipulation such as humanisation or other structural manipulations. Polyethylene glycol (PEG) derivitisation.

PEGylation of proteins has been widely reported and can give rise to products with greatly prolonged blood half-lives. Several derivitized PEG preparations are available commercially which can be coupled to proteins using standard mild cross-linking conditions using either amino or sulphydryl groups. Whole antibody molecules or derived Fab or Fab2 fragments can be used for different purposes.

Preparation of F(ab') and F(ab')2 fragments

To IgG (20mg/ml) in 0.1 M acetate buffer pH 4.5 is added Pepsin- Sepharose (2mg enzyme per 100mg of IgG). After mixing and incubation overnight at 37°C, the pepsin-Sepharose is centrifuged out, the supernatant taken to pH 7.4 and passed into a Protein A-Sepharose column. This binds undigested IgG and the Fc fragments released by the digestion. After dialysis against PBS, any remaining unfragmented or small material is removed by chromatography on a Sephacryl S200 gel column. The purity is checked by SDS page analysis and the Fab' fragments can be made by reducing with dithiothreitol (DTT). The protein is equilibrated with Tris-HCI, degassed and made 20mM with DTT. After an hour, the DTT is rapidly removed by centrifugation on a column of Sephadex G25 and is used immediately to re-form heterodimeric F(ab')2 complexes.

To one of the Fab' proteins equilibrated in 50 mM sodium acetate, pH 5.3 containing 0.5 mM EDTA at 4°C is added a half volume of 12 mM oPDM (o-phenylenedimaleimide) dissolved in cold dimethylformamide (Glennie J Immunol (1987) 139 2367). After 30 min the maleimidated protein is separated from other solutes using a Sephadex G-25 equilibrated in the 50 mM sodium acetate, pH 5.3, buffer containing 0.5 mM EDTA. This is then added immediately to the second antibody component of the heterodimer and concentrated to approximately 5 mg/ml by centrifugal concentration. After incubation for 18 hr the pH of the reaction mixture is adjusted to 8 using 1 M Tris-HC1 , pH 8.0 before reducing with mercaptoethanol at a final concentration of 20 mM for 30 min at 30°C and blocking any free SH groups by alkylating with 25 mM iodoacetamide. Finally the bispecific F(ab')2 is separated using Sephacryl S200 in 0.2 M Tris- HC1 , 10 mM EDTA, pH 8.0.

Coupling of bifunctional F(ab')2

A 4 mg/ml solution of F(ab')2 is equilibrated in 50 mM sodium phosphate buffer (pH 8.0). Five milliliters of the protein solution is mixed with the 3400 kDa- PEG (Shearwater Polymers, Huntsville, AL) reagent:protein at a molar ratio of 3:1 . The reaction is carried out in a 15 ml polypropylene tube (Falcon), and the PEG is added while vortexing the sample at low speed.

Coupling of bifunctional Fab'

Branched derivitized PEGs are also available to which individual Fab' fragments can be coupled to yield products with variable spacing between the hetero Fab' fragments, which may have benefits when binding to widely spaced antigens on cells. The Fab' fragments are mixed at a ratio of 1 :1 in phosphate buffer pH 6.5 and the bifunctional maleimide branched PEG mPEG(MAL)2 (Shearwater Polymers, Huntsville, AL) is added. The reactivity with SH groups is greater than that with amino groups at this pH and so should orient the Fab' and leave the antigen recognition site intact. The sample is then incubated at room temperature while shaking on a rotator for 30 min. The extent of modification was evaluated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The complete reaction mixture was passed through an ion exchange column to remove any unreacted PEG and purification of singly or doubly PEGylated species. This will contain a mixture of heterobifunctional and homobifunctional molecules which will be separated using sequential affinity chromatography through columns with each bound antigen. Only the complexes retained by both columns will contain the required heterobifunctional PEG-Ab complexes.

Oligonucleotide crosslinking

Another approach to cross-linking which overcomes the mixed population of homo- and hetero- bifunctional molecules produced is to make use of the specificity of interactions of oligo-nucleotides (Chaudri FEBS Letts (1999) 450; 23). In this two complementary 15 residues oligomers are synthesised by standard procedures, one of which is biotinylated at the 5' end and both are derivitized to introduce amino groups at the 3' end. These are reacted in the dark for 30 min at 20³C with a 100-fold molar excess of the crosslinking agent N-hydroxysuccinimidyl(4- iodoacetyl)aminobenzoate (SIAB, Pierce and Warriner, Chester, UK) in 0.1 M bicarbonate buffer, pH 9.0. Activated oligonucleotide is separated from unreacted crosslinker by gel filtration on a Sephadex G-25 PD10 column (Pharmacia Biotech), equilibrated with 0.1 M phosphate buffer, pH 7.5, containing 10 mM ethylenediamine tetraacetic acid (EDTA). A 7- fold molar excess of the activated oligonucleotide in 0.1 M phosphate buffer, pH 7.5, 10 mM EDTA is incubated with the Fab' overnight at 4°C. The Fab'-oligonucleotide conjugate is purified by size exclusion chromatography using Sephacryl S200 chromatography. The oligonucleotide-labelled Fab' fragments are then mixed at a ration 1 :1 and left overnight at 4°C for binding to occur. Heterodimers are separated from unbound monomers by gel permeation chromatography on Sephacryl S200 columns.

Example

A one-to-one mixture of two low affinity antibodies is made and, if the ^■ hybridomas are derived from mouse, one half mole equivalent of a rat anti-mouse IgG Fc region is added. After 1 hr incubation on ice the mixture is passed though a sizing column of Sephacryl S-300 equilibrated in 20 mM HEPES 150 mM NaCl to separate full trimeric complexes of rat Ig coupled to two mouse IgG molecules (450,000 mw) which elutes first from the column.

These complexes are then chemically cross-linked to stabilize them by diluting the solution 10-fold in 20 mM dimethyl pimelimidate dihydrochloride dissolved in 0.2 M triethanolamine, which is pH readjusted to 8.2. This is mixed for 1 hour at room temperature and centrifugal ly concentrated on a 30,000MW cutoff Centricon filter (Millipore (UK), Ltd, Watford, UK), which also removes the crosslinking agent. The solution is made to 20 mM in ethanolamine, pH 8.2 and left for 5 min at room temperature to block remaining cross-linkers. The final product is then diluted in Hepes Saline and concentrated again on a 30,000MW cutoff Centricon filter.

Selection of hetero-bifunctional antibodies from the mixture.

Many of the complexes formed contain homo-bifunctional antibodies from which the hetero-bifunctional antibodies need to be isolated. These can be separated using affinity chromatography.

Two separate affinity columns are made by coupling the individual antigens recognized by the antibodies to CNBr-Sepharose. The antibody mixture is pumped on to the column containing the first antigen, at 4°C to maximize the binding affinity and washed through with equilibration buffer (20 mM HEPES 0.15M NaCI pH 7.5) until no further buffer is eluted. The antibodies are eluted using 100 mM glycine buffer pH 3.0. This eluate is concentrated on a 30,000MW cutoff Centricon filter, re- equilibrated to pH 7.5 and applied to an affinity column loaded with the second antigen. Antibody binding to this column is collected in the same way. This results in an antibody preparation that will recognize both antigens, and so must contain the required heterobifunctional antibody.

This approach is applicable to antibodies that have intermediate affinities for the individual antigens, but may not work if the affinity is towards the lower end of the scale, since the binding to the column will not be firm enough to allow adequate washing away of unbound antibody. In these cases an affinity column made using both antigens, crude antigen preparations or even coupled whole cell membranes is made. The mixed antibody is applied and the column washed as before. The low affinity antibodies recognizing the individual antigens are not well retained on the column and so are eluted. Only those antibodies recognizing both antigens are retained. These are eluted by stepwise decreases in the pH of the elution buffer containing glycine, to be certain that only the high avidity complexes are isolated.

Characterization of avidity of heterobifunctional antibodies.

Having purified the antibody as above binding characteristics are derived using a BIAcore chip to which both antigens have been coupled. Using the methods described above an effective binding constant or "avidity" is derived. If antigens are present on cells an estimate of binding can be achieved using flow cytometry by labelling cells with the heterobifunctional complex, followed by a FITC labeled second antibody. A more accurate effective binding constant can be derived by radiolabelling of the antibody with ¹²⁵l and performing cell binding and Scatchard analysis of the binding data.

Screening of phage display libraries for low affinity antibody fragments.

The methods described above can also be used to screen for low affinity Fab fragments or components thereof expressed in bacteria by using a phage display approach. Combinations of these low affinity fragments made by coupling, for example, to a larger carrier molecule such as PEG can also be useful.

Claims

1. A bispecific antibody comprising two antibodies, each of said antibodies having a binding ^"specificity to a different epitope situated on the surface of a target structure, wherein each of said antibodies has a relatively low binding affinity for its respective epitope.

2. The bispecific antibody as claimed in claim 1 , wherein each of the antibodies of which the bispecific antibody is comprised has an affinity constant of at least 10^"8M.

3. The bispecific antibody as claimed in claim 1 or 2, wherein each of said antibodies is selected from an intact antibody molecule, an antibody fragment, or functional equivalents of the antibody/antibody fragment which may be naturally occurring, genetically engineered or chemically synthesised.

4. The bispecific antibody as claimed in claim 3, wherein said antibody fragment is selected from an Fab, Fab' or F(ab')2.

5. The bispecific antibody as claimed in any preceding claim, wherein the target structure is a cell, collection of cells, bodily tissue or organ.

6. The bispecific antibody as claimed in any preceding claim, wherein the bispecific antibody comprises a linker and/or spacer to which the antibodies are attached.

7. The bispecific antibody as claimed in claim 6, wherein said linker/spacer is a polymer.

8. The bispecific antibody as claimed in any preceding claim, wherein the bispecific antibody comprises a therapeutic or diagnostic agent targeted to said target structure.

9. The bispecific antibody as claimed in claim 8, wherein said therapeutic or diagnostic agent is a drug, pro-drug, T-cell or other immune system cell or target marker, such as a fluorescent, luminescent or radioactive marker, liposome, complement, inhibitory RNA or plasmid encoding for an inhibitory RNA sequence, or retroviral vector coding a sequence to "knock-down" or "knock-in" genes of the target structure.

10. The bispecfic antibody as claimed in claim 8, wherein the therapeutic agent is selected from ¹³¹ l, ⁶⁷Cu, ⁹⁷Ru and ¹⁸⁸Re, anti-cancer drugs such as nitrogen mustards, cytotoxic antibiotics such as bleomycin, cis-diaminodichloroplatinum, alkylating agents, antimetabolites such as 5- fluorouracil, toxins, for example mycotoxins such as trichothecenes, ricin or abrin and biological response modifiers, for example interferons, interleukins or TNF.

1 1. The bispecific antibody as claimed in claim 8, wherein the diagnostic agent is selected from ⁷⁶Br, ¹⁸F or ¹²³l or NMR imaging contrast agents.

12. The bispecific antibody as claimed in any one of claims 8 to 1 1 . wherein the therapeutic or diagnostic agent is covalently or otherwise linked to the remainder of the bispecific antibody, said linking being at one or both antibodies and/or at the linker/spacer when present.

13. The bispecific antibody as claimed in any preceding claim, wherein the bispecific antibody comprises a receptor for a therapeutic or diagnostic agent.

14. The bispecific antibody as claimed in claim 13, wherein the receptor is an Fc antibody fragment .

15. A method of producing a bispecific antibody in accordance with any one of claim 1 to 14 comprising joining together the two antibodies having a binding specificity to a different epitope situated on the surface of the target structure.

16. The method of claim 15, wherein the two antibodies are joined by chemical cross-linking, or cross-linking using anti-lgG Fc antibodies followed by covalent cross-linking.

17. The method as claimed in claim 15, wherein the antibodies are each produced from a hybridoma of a B-cell activated during a primary immune response caused by the first exposure of the cells to an antigen fused with a myeloma cell.

18. A method of producing a bispecific antibody as claimed in any one of claims 1 to 14, comprising fusing two hybridoma cell lines, each cell line producing one of the antibodies.

19. A method of selecting monoclonal antibodies for use in the method of claims 15 to 17, comprising screening a large number of hybridomas which produce antibodies specific for each epitope of interest by direct measurement of their binding affinity for their specific epitope.

20. The method as claimed in claim 19, wherein said direct measurement is by surface plasmon resonance.

21. A method of treating a patient comprising, administering to a patient a therapeutically effective amount of (i) a bispecific antibody in accordance with claim 13 or 14, said bispecific antibody having a receptor for a therapeutic agent, in conjunction with prior, concomitant or subsequent administration of the therapeutic agent or (ii) a bispecific antibody in accordance with any one of claims 8 to 1 1 , said bispecific antibody comprising the therapeutic agent.

22. A method of diagnosing a disorder comprising, contacting (i) a bispecific antibody in accordance with any one of claims 8 to 1 1 , the bispecific antibody comprising the diagnostic agent with a tissue, or (ii) contacting a bispecific antibody in accordance with claim 13 or 14 the bispecific antibody having a receptor for the diagnostic agent with a tissue prior to, concomitantly with or subsequently to addition of the diagnostic agent to said tissue.

3. The method as claimed in claim 22, wherein said tissue is an isolated tissue sample.