NZ624583B2 - Antibody molecules having specificity for human ox40 - Google Patents

Abstract

Discloses a bispecific antibody fusion protein which binds human OX40 and human serum albumin comprising: a heavy chain comprising, in sequence from the N-terminal, a first heavy chain variable domain (VH1), a CHI domain and a second heavy chain variable domain (VH2), a light chain comprising, in sequence from the N-terminal, a first light chain variable domain (VL1), a CL domain and a second light chain variable domain (VL2), wherein said heavy and light chains are aligned such that VH1 and VL1 form a first antigen binding site and VH2 and VL2 form a second antigen binding site, wherein the antigen bound by the first antigen binding site is human OX40 and the antigen bound by the second antigen binding site is human serum albumin, wherein the first variable domain of the heavy chain (VH1) comprises the sequence given in SEQ ID NO:1 for CDR-H1, the sequence given in SEQ ID NO:2 for CDR-H2 and the sequence given in SEQ ID NO:3 for CDR-H3 and the first variable domain of the light chain (Vd) comprises the sequence given in SEQ ID NO:4 for CDR-L1, the sequence given in SEQ ID NO:5 for CDR-L2 and the sequence given in SEQ ID NO:6 for CDR-L3, wherein the second heavy chain variable domain (VH2) has the sequence given in SEQ ID NO:11 and the second light chain variable domain (VL2) has the sequence given in SEQ ID NO: 12 and the second heavy chain variable domain (VH2) and second light chain variable domain (VL2) are linked by a disulphide bond, wherein the sequences are as defined in the complete specification. equence from the N-terminal, a first light chain variable domain (VL1), a CL domain and a second light chain variable domain (VL2), wherein said heavy and light chains are aligned such that VH1 and VL1 form a first antigen binding site and VH2 and VL2 form a second antigen binding site, wherein the antigen bound by the first antigen binding site is human OX40 and the antigen bound by the second antigen binding site is human serum albumin, wherein the first variable domain of the heavy chain (VH1) comprises the sequence given in SEQ ID NO:1 for CDR-H1, the sequence given in SEQ ID NO:2 for CDR-H2 and the sequence given in SEQ ID NO:3 for CDR-H3 and the first variable domain of the light chain (Vd) comprises the sequence given in SEQ ID NO:4 for CDR-L1, the sequence given in SEQ ID NO:5 for CDR-L2 and the sequence given in SEQ ID NO:6 for CDR-L3, wherein the second heavy chain variable domain (VH2) has the sequence given in SEQ ID NO:11 and the second light chain variable domain (VL2) has the sequence given in SEQ ID NO: 12 and the second heavy chain variable domain (VH2) and second light chain variable domain (VL2) are linked by a disulphide bond, wherein the sequences are as defined in the complete specification.

Description

Antibody molecules having specificity for human 0X40 The present invention relates to antibody molecules having specificity for antigenic determinants of0X40 and compositions comprising the same. The present invention also relates to the therapeutic uses ofthe antibody molecules, compositions and methods for producing said antibody molecules. 0X40 (also known as CD134, TNFRSF4, ACT35 or TXGPlL) is a member of the TNF receptor superfamily, which includes 4—1BB, CD27, CD30 and CD40. The extracellular ligand binding domain of 0X40 is composed of 3 full cysteine-rich domains (CRDs) and a partial, fourth C-terminal CRD (Bodmer er a], 2002, Trends Biochem Sci, 27, 19-26).

The ligand for 0X40 is OX40L and 3 copies of 0X40 bind to the trimeric ligand to form the OX40—OX40L complex an and Hymowitz, 2006, Structure, 14, 1321-1330). 0X40 is a membrane-bound receptor; however a soluble isoform has also been detected (Taylor and Schwarz, 2001, J.Immunol. Methods, 255, 67~72). The functional significance ofthe soluble form is presently n. 0X40 is not expressed on resting T cells, but is transiently expressed on activated T cells after ligation of the T cell receptor (TCR). The ligand for 0X40, OX40L, is a member ofthe TNF family and is expressed on ted antigen presenting cells (APC), including B cells, macrophages, endothelial cells and dendritic cells (DC). 0X40 is a major costimulatory receptor with sequential engagement of CD28 and 0X40 being required for optimal T cell eration and survival. Ligation of 0X40 on ted T cells leads to enhanced cytokine production and proliferation of both CD4+ and CD8+ T cells (Gramaglia et (21., 2000, J. Immunol, 165, 050, Bansal-Pakala et 52]., 2004, J.1mmunol, 172, 4821—425) and can contribute to both ongoing Th1 and Th2 responses (Gramaglia er al._, 1998, J. Immuno, 161, 6510—6517, Arestides er al., 2002, Eur. J. Immunol. 32, 2874-2880). 0X40 costimulation prolongs T cell survival beyond the initial effector phase of the immune response and increases the number of memory T cells through inhibition of effector T cell death.

When immune activation is excessive or uncontrolled, pathological allergy, asthma, ation, autoimmune and other d diseases may occur. Because 0X40 functions to enhance immune responses, it may exacerbate autoimmune and inflammatory diseases.

The role of OX40/OX40L interactions in models of disease has been demonstrated in 0X40 ut mice. In experimental allergic alomyelitis (EAE), a model of multiple sclerosis, less severe clinical signs of disease and reduced inflammatory infiltrate within the CNS was noted in 0X40 ut mice (Carboni et (11., 2003, J.Neuroimmunology, 145, 1-11).

Also 0X40 knockout mice primed and challenged with ovalbumin exhibit diminished lung inflammation (80 — 90% reduction in eosinophilia), d mucus production, and significantly attenuated airway hyper-reactivity r et 62]., 2001, J. Exp.Med., 193, 387- 392). Monoclonal antibodies to murine 0X40 ligand have shown ial s in the collagen-induced arthritis model ofrheumatoid tis (Yoshioka et al., 2000, Eur. J 40 Immunol., 30, 2815-2823), EAE (Nohara et al., 2001, J. l, 166, 2108-2115), non-obese diabetic (NOD) mice (Pakala et al., 2004, Eur. J. lmmunol., 34, 3039-3046), colitis in T cell restored mice trom et al., 2001, J. Immunol, 166, 6972—6981, Totsuka et al., 2003, Am.

J. Physio l. Gastrointest. Liver Physiol, 284, G595-G603) and models of lung inflammation (Salek-Ardakani er al., 2003, J. Exp. Med, 198, 315-324, Hoshino et al., 2003, Eur.J.Immunol, 33, 861-869). An antibody to human OX40L has been profiled in a model of lung inflammation in rhesus s and resulted in reduced levels of IL—5, IL—13 and effector memory T cells in iolar lavage fluid after allergen challenge (Seshasayee et al., 2007, J.

Clin.1nvest, 117, 3868-3878).

An increase in the sion of 0X40 has been noted in l autoimmune and inflammatory diseases. This includes an se in 0X40 expression on T cells isolated from the synovial fluid of rheumatoid arthritis patients (Brugnoni D et al., 1998, En]. Rheum, 37, 584-585; Yoshioka et (11., 2000, Eur. J. Immunol, 30, 2815-2823; Giacomelli R et al., 2001, Clin. Exp. Rheumatol., 19, 317—320). Similarly an increase in 0X40 expression has been noted in gastrointestinal tissue from patients with ulcerative s and Crohn’s e (Souza at (21., 1999, Gut, 45, 3; Stuber at al., 2000, Eur.J.Clin.Invest., 30, 594-599) and in active lesions of patients with multiple sclerosis (Carboni er al., 2003, J.Neuroimmunology, 145, 1- 11). OX40L can also be detected on human airway smooth muscle (ASM) and asthma patients ASM cells show greater inflammatory responses to OX40L ligation than healthy donors, indicating a role for the OX40/OX4OL pathway in asthma (Burgess 61 al., 2004, J. Allergy Clin Immunol., 113, 683-689; Burgess er al., 2005, J. Allergy Clin Immunol., 115, 302—308). It has also been reported that CD4+ T cells isolated from the peripheral blood of systemic lupus erythematosus (SLE) patients express elevated levels of0X40 which is associated with e activity han et a[., 2006, Clin. Exp. Immunol, 145, 235-242).

Given the role of 0X40 in allergy, asthma and diseases associated with autoimmunity and inflammation, one ch to therapy in these diseases is to block OX40-OX40L signalling through the use of anti—OX40L antibodies or antagonistic anti—0X40 antibodies Anti-OX40L antibodies have been described, see for example W02006/029879.

Numerous agonistic anti-0X40 antibodies have been described but very few antagonistic anti- OX40 antibodies are known. A rabbit polyclonal anti-mouse 0X40 antibody was produced by Stuber et al., 1996, J.Exp.Med, 183, 979-989 which blocks the interaction between 0X40 and OX40L. Mouse onal antibodies, 131 and 315 which bind human 0X40 were generated by Imura et al., 1996, J.Exp.Med, 195.

Fully human antagonistic antibodies have been described in WO2007/062245, the highest affinity of these antibodies had an affinity for cell surface expressed 0X40 (activated T cells) of 1 lnM.

Humanised antagonistic antibodies have been described in W02008/106116 and the dy with the best affinity for 0X40 had an affinity of 0.94nM.

Other anti—0X40 antibodies have been bed, including murine L106 (US Patent number 962) and murine ACT3S, commercially available from eBioscience.

We have previously described high y antagonistic anti-0X40 antibodies in 40 International Patent application number W02010/096418.

We have also previously described in International Patent application number W02010/035012, a novel specific antibody fusion molecule, hereinafter referred to as a Fab-dst and illustrated herein in Figure l. The same application provides useful anti-albumin binding variable regions which may be used to extend the half-life of the molecule.

In the present invention these albumin binding variable regions have been improved and combined in the Fab-dst format with the anti-0X40 antibodies described in W02010/096418.

The new bispecific molecule of the present invention has improved y in a number of in vitro and in vivo assays described herein when compared to the Fab’—PEG molecule previously described in W02010/096418. Accordingly, the present invention es a bispecific antibody fusion protein which binds both human 0X40 and human serum albumin which is suitable for use in the ent or prophylaxis of ogical ers mediated by 0X40 or associated with an increased level of 0X40.

Summary of the Invention In a first aspect, the present invention provides a bispecific antibody fusion protein which binds human 0X40 and human serum n comprising: a heavy chain comprising, in sequence from the N—terminal, a first heavy chain le domain (VH1), a CH1 domain and a second heavy chain variable domain (VH2), a light chain comprising, in sequence from the N-terminal, a first light chain variable domain (VLl), a CL domain and a second light chain variable domain (VL2), wherein said heavy and light chains are aligned such that VH1 and VLl form a first antigen binding site and VH2 and VL2 form a second antigen g site, wherein the antigen bound by the first antigen g site is human 0X40 and the antigen bound by the second antigen binding site is human serum albumin, wherein the first variable domain of the heavy chain (VH1) comprises the sequence given in SEQ ID N021 for CDR-Hl, the sequence given in SEQ ID N0:2 for CDR—HZ and the sequence given in SEQ ID N023 for CDR-H3 and the first variable domain of the light chain (VLl) comprises the sequence given in SEQ ID N024 for CDR—Ll, the sequence given in SEQ ID N025 for CDR-L2 and the ce given in SEQ ID N026 for CDR—L3, 3O wherein the second heavy chain variable domain (VH2) has the sequence given in SEQ ID N0:1l and the second light chain variable domain (V12) has the sequence given in SEQ ID NO: 12 and the second heavy chain variable domain (VH2) and second light chain variable domain (VL2) are linked by a disulphide bond.

In another aspect, the present invention provides a bispecific antibody fusion protein which binds human 0X40 and human Serum albumin, having a heavy chain sing the sequence given in SEQ ID N0:15 and a light chain sing the sequence given in SEQ ID N0: 1 6. 40 Brief Description of the Drawings Figure 1 shows a bispecific dy fusion of the present invention (Fab-dsFV format) (followed by page 3a) Figures 2—8 shows certain amino acid or DNA sequences ng to an antibody according to the disclosure Figure 9a shows binding of AlexaFluor 488 ed A26 Fab-dst to activated human CD4+0X40+ T cells Figure 9b shows binding for A26 Fab’, A26 Fab-FV and A26 Fab’-PEG in the presence of % HSA on activated human CD4+, OX40+ T cells.

Figure 10a shows the effect ofA26 Fab-dst on cytokine production from PBMC exposed to Dermatophagoides pteronyssinus allergic extract Figure 10b shows the ability ofA26 Fab—dsFV to inhibit CD4+ and CD8+ T cell proliferation in a Hu—NSG mouse model Figure 1 1 a shows inhibition of OX4OL binding to human ted CD4Jr 0X40+ T cells by A26 Fab-dsFV Figure 11b shows inhibition of OX40L binding to human activated CD4+ OX4OJr T cells by A26 Fab’, A26 Fab—dst, A26 EG and two controls.

Figure 12a shows A26 Fab-Fv inhibits a human mixed lymphocyte reaction (MLR) Figure 12b shows A26 Fab—Fv inhibits IFN—gamrna production during a human MLR Figure 13 shows A26 Fab-Fv reduces the percentage of ted (CD25+) CD4+ T cells after secondary antigen re-stimulation with Dermatophagoides pteronyssinus allergenic extract Figure 14 shows Fab-FV and Fab—PEG administered prior to cell transfer dose dependency inhibits CD4+ and CD8+ T cell proliferation in the Hu-NSG model Humanised 00026 anti-0X40 antibody, is referred to herein as A26.

The antibody fusion le of the present invention, referred to herein as a Fab-dst, is illustrated in Figure 1. In the present ion the Fab portion (comprising the first heavy and light chain variable regions and the constant domains) binds human 0X40 and the dsFV [FOLLOWED BY PAGE 4] portion (comprising the second heavy and light chain variable regions, linked by a disulphide bond) binds human serum albumin. In particular, the Fab n comprises the CDRS derived from an antagonistic anti-0X40 dy and the EV portion comprises the heavy and light chain variable s of a humanised anti-albumin antibody, and these albumin binding variable regions are linked by a disulphide bond.

Accordingly, the present invention es a bispecific antibody fusion protein which binds human 0X40 and human serum albumin comprising: a heavy chain comprising, in sequence from the N-terminal, a first heavy chain variable domain (VH1), a CH1 domain and a second heavy chain variable domain WHZ), l0 a light chain comprising, in sequence from the N-terminal, a first light chain variable domain (VLl), a CL domain and a second light chain variable domain (VLZ), wherein said heavy and light chains are d such that VH1 and VLl form a first antigen binding site and V52 and VLZ form a second antigen binding site, wherein the antigen bound by the first antigen binding site is human 0X40 and the antigen bound by the second antigen binding site is human serum albumin, in particular n the first le domain of the heavy chain (VH1) comprises the sequence given in SEQ ID NO:1 for CDR-Hl, the sequence given in SEQ ID NO:2 for CDR—H2 and the sequence given in SEQ ID N03 for CDR—H3 and the first variable domain ofthe light chain (VLl) comprises the sequence given in SEQ ID NO:4 for CDR-Ll, the sequence given in SEQ ID N025 for CDR-L2 and the sequence given in SEQ ID NO:6 for CDR—L3, n the second heavy chain le domain (VH2) has the sequence given in SEQ ID N021 l and the second light chain variable domain (VL2) has the ce given in SEQ ID NO: 12 and the second heavy chain variable domain (VH2) and second light chain variable domain 0&2) are linked by a disulphide bond.

The residues in antibody variable domains are conventionally numbered ing to system devised by Kabat er al. This system is set forth in Kabat et (11., 1987, in Sequences of Proteins of Immunological Interest, US Department of Health and Human Services, NIH, USA (hereafter “Kabat et a]. (supra)”). This numbering system is used in the present specification except where ise indicated.

The Kabat residue designations do not always correspond directly with the linear numbering ofthe amino acid residues. The actual linear amino acid sequence may contain fewer or additional amino acids than in the strict Kabat numbering corresponding to shortening of, or insertion into, a structural component, whether framework or complementarity determining region (CDR), of the basic le domain structure. The correct Kabat numbering ofresidues may be determined for a given antibody by alignment of residues of homology in the sequence of the dy with a “standar ” Kabat numbered sequence.

The CDRs ofthe heavy chain variable domain are located at residues 31-35 (CDR—Hl), 40 residues 50-65 Z) and residues 95—102 (CDR—H3) according to the Kabat numbering system. However, according to Chothia (Chothia, C. and Lesk, A.M. J. Mol. Biol, 196, 901- 2012/072325 917 (1987)), the loop equivalent to CDR—H1 extends from residue 26 to residue 32. Thus unless indicated otherwise ‘CDR—Hl ’ as employed herein is intended to refer to residues 26 to , as described by a combination of the Kabat numbering system and Chothia’s topological loop definition.

The CDRs ofthe light chain variable domain are located at residues 24—34 (CDR-L1), residues 50—56 (CDR-L2) and residues 89-97 (CDR-L3) according to the Kabat numbering system.

The bispecific fusion protein of the present invention comprises a Fab fragment of the anti—0X40 antagonistic antibody previously described in W02010/0964l 8. As used herein, the term ‘antagonistic’ describes an antibody fusion protein that is capable of inhibiting and/or neutralising the biological signalling activity of 0X40, for example by ng binding or substantially reducing binding of0X40 to 0X40 ligand and thus inhibiting the activation of 0X40.

Screening for antibodies to identify those that bind 0X40 can be performed using assays to measure binding to human 0X40 and/or assays to measure the ability to block the binding of 0X40 to its , OX40L. An example of a binding assay is an ELISA, in particular, using a fusion protein of human 0X40 and human PC, which is immobilized on plates, and employing a conjungated secondary antibody to detect anti—0X40 antibody bound to the fusion protein. An example of a blocking assay is a flow cytometry based assay measuring the blocking of 0X40 ligand fusion protein binding to 0X40 on human CD4 cells. A fluorescently labelled secondary antibody is used to detect the amount of 0X40 ligand fusion protein binding to the cell. This assay is looking for a reduction in signal as the antibody in the supernatant blocks the binding of ligand fusion n to 0X40. A further example of a ng assay is an assay where the blocking of costimulation of naive human T cells mediated by 0X40 ligand fusion protein coated to a plate is measured by measuring tritiated thymidine incorporation.

In the t invention, the le regions are humanised. Humanised antibodies (which include CDR—grafted antibodies) are antibody molecules having one or more complementarity ining regions (CDRs) from a non-human s and a framework region from a human globulin molecule (see, 6.g. US 5,585,089; W09 1/09967). It will be appreciated that it may only be necessary to er the specificity determining residues of the CDRs rather than the entire CDR (see for example, Kashmiri et al., 2005, Methods, 36, 25- 34). Humanised antibodies may ally r comprise one or more fiamework residues derived from the non-human species from which the CDRs were derived.

In the present invention the CDRs ofVHl and VLl are derived from the antibody known as A26, described in W02010/096418. ingly, in the ific antibody fusion protein ofthe present invention, the first le domain of the heavy chain (VH1) comprises the sequence given in SEQ ID N0:1 for CDR-H1, the sequence given in SEQ ID N02 or SEQ ID N0223 for CDR-H2 and the ce given in SEQ ID N023 for CDR-H3 and the first variable domain ofthe light chain (VL1) comprises the sequence given in SEQ ID N0:4 or SEQ ID 40 N0:24 for CDR—Ll, the sequence given in SEQ ID N0:5 for CDR—L2 and the sequence given in SEQ ID N026 for CDR-L3.

WO 68563 It will be appreciated that one or more amino acid substitutions, additions and/or deletions may be made to the CDRs provided by the present invention without icantly altering the ability ofthe antibody to bind to 0X40 and to lise 0X40 activity. The effect of any amino acid substitutions, additions and/or deletions can be readily tested by one skilled in the art, for example by using the methods described in W02010/0964l8, to determine 0X40 binding and inhibition of the 0X40/OX40L interaction. Accordingly, the present invention provides a bispecific antibody having specificity for human 0X40 comprising CDRH-l (SEQ ID N021), CDRH—Z (SEQ ID N0:2), CDRH—3 (SEQ ID N0z3), CDRL-l (SEQ ID N0:4), CDRL—2 (SEQ ID N0:5) and CDRL~3 (SEQ ID N026) as shown in Figure 2(0), for example in which one or more amino acids, for example I or 2 amino acids, in one or more of the CDRs has been substituted with another amino acid, such as a similar amino acid as defined herein below.

In one embodiment, a bispecific antibody fusion protein of the present invention comprises a heavy chain, wherein the first variable domain of the heavy chain ses three CDRs n the sequence of CDRH-l has at least 90% identity or similarity to the sequence given in SEQ ID N0:l, CDRH-2 has at least 90% identity or similarity to the sequence given in SEQ ID N0:2 and/or CDRH—3 has at least 90% identity or similarity to the sequence given in SEQ ID N023. In r embodiment, a bispecific antibody fusion protein of the present invention comprises a heavy chain, wherein the variable domain ofthe heavy chain ses three CDRs wherein the ce ofCDRH—l has at least 95% or 98% identity or similarity to the ce given in SEQ ID N011, CDRH—Z has at least 95% or 98% identity or similarity to the sequence given in SEQ ID N0:2 and/or CDRH-3 has at least 95% or 98% identity or similarity to the sequence given in SEQ ID N0:3.

"Identity", as used herein, indicates that at any ular position in the aligned sequences, the amino acid residue is identical between the sequences. "Similarity", as used herein, indicates that, at any particular position in the aligned sequences, the amino acid e is of a similar type between the sequences. For example, leucine may be substituted for isoleucine or valine. Other amino acids which can often be substituted for one another include but are not limited to: — phenylalanine, tyrosine and tryptophan (amino acids having aromatic side chains); - lysine, arginine and ine (amino acids having basic side chains); — aspartate and glutamate (amino acids having acidic side chains); — asparagine and glutamine (amino acids having amide side chains); and — ne and nine (amino acids having sulphur—containing side chains). Degrees of identity and similarity can be readily calculated (Computational Molecular Biology, Lesk, A.M., ed., Oxford University Press, New York, 1988; Biocomputing. Informatics and Genome Projects, Smith, D.W., ed., Academic Press, New York, 1993; er Analysis of Sequence Data, Part 1, Griffin, A.M., and Griffin, H.G., eds, Humana Press, New Jersey, 1994; Sequence Analysis in Molecular Biology, von Heinje, G., Academic Press, 1987, Sequence Analysis 40 Primer, Gribskov, M. and Devereux, J., eds., M Stockton Press, New York, 1991, the BLASTTM software available from NCBI (Altschul, SF. et (11., 1990, J. Mol. Biol. 215:403-410; Gish, W. & States, DJ. 1993, Nature Genet. 3:266—272. , T.L. et al, 1996, Meth. Enzymol. 266:131-141;A1tschu1, S.F. et £21., 1997, Nucleic Acids Res. 25:3389-3402; Zhang, J. & Madden, TL. 1997, Genome Res. 656,).

In another embodiment, a bispecific antibody fusion protein of the present invention comprises a light chain, wherein the first variable domain of the light chain comprises three CDRs wherein the sequence of CDRL-l has at least 90% identity or similarity to the sequence given in SEQ ID NOI4, CDRL-2 has at least 90% identity or similarity to the sequence given in SEQ ID N025 and/or CDRL—3 has at least 90% identity or rity to the sequence given in SEQ ID N026. In another embodiment, a bispecific antibody filsion protein of the present invention comprises a light chain, n the first variable domain of the light chain comprises three CDRs wherein the sequence of CDRL-l has at least 95% or 98% identity or similarity to the sequence given in SEQ ID N024, CDRL—2 has at least 95% or 98% identity or similarity to the sequence given in SEQ ID N025 and/or CDRL—3 has at least 95% or 98% identity or rity to the sequence given in SEQ ID N0z6.

In one embodiment the Fab portion of the bispecific antibody fusion protein provided by the present invention is a humanised or CDR—grafted dy molecule comprising one or more ofthe CDRs provided in SEQ ID N0s: l, 2, 3, 4, 5 and/or 6 (Figure 2 (c)) or variants thereof. As used herein, the term ‘CDR-grafied antibody molecule’ refers to an antibody le wherein the heavy and/or light chain contains one or more CDRs (including, if desired, one or more modified CDRs) from a donor antibody (cg. a murine monoclonal antibody) grafted into a heavy and/or light chain variable region framework of an acceptor antibody (e.g. a human antibody). For a review, see Vaughan et al, Nature hnology, lg, 535—539, 1998. In one embodiment rather than the entire CDR being transferred, only one or more ofthe specificity determining residues from any one ofthe CDRs described herein above are transferred to the human antibody framework (see for example, Kashmiri et (11., 2005, Methods, 36, . In one embodiment only the specificity determining residues from one or more of the CDRs described herein above are transferred to the human antibody framework. In another embodiment only the specificity determining residues from each of the CDRs bed herein above are erred to the human antibody ork.

When the CDRs or specificity determining residues are grafted, any appropriate acceptor variable region ork sequence may be used having regard to the class/type of the donor antibody from which the CDRs are derived, including mouse, primate and human framework regions. Suitably, the CDR—grafted antibody according to the present invention has a variable domain comprising human acceptor framework regions as well as one or more of the CDRs or city determining residues bed above. Thus, provided in one embodiment is a neutralising CDR-grafted antibody wherein the variable domain comprises human acceptor framework s and non-human donor CDRs.

Examples of human frameworks which can be used in the present invention are KOL, NEWM, REI, EU, TUR, TEI, LAY and POM (Kabat et (11., supra). For example, KOL and 40 NEWM can be used for the heavy chain, REI can be used for the light chain and EU, LAY and POM can be used for both the heavy chain and the light chain. atively, human germline sequences may be used; these are available at: http://vbase.mrc—cpe.cam.ac.uk/ In a CDR-grafted dy of the t invention, the acceptor heavy and light chains do not necessarily need to be derived from the same antibody and may, if desired, se ite chains having fiamework s derived from different chains.

A suitable framework region for the first heavy chain variable domain (VH1) of the present invention is derived from the human sub—group VH3 sequence 1—3 3-07 together with JH4. A suitable framework region for the light chain for the first light chain variable domain WLl) is derived from the human germline sub—group VKl sequence 2-1 1—02 together with 1K4.

Also, in a CDR—grafted antibody variable region of the present invention, the framework regions need not have y the same ce as those of the acceptor antibody. For instance, unusual residues may be changed to more frequently—occurring residues for that acceptor chain class or type. Alternatively, selected residues in the acceptor framework regions may be changed so that they correspond to the e found at the same position in the donor antibody (see Reichmann et (11., 1998, Nature, 332, 323-324). Such changes should be kept to the minimum necessary to recover the y of the donor antibody. A protocol for selecting residues in the acceptor framework regions which may need to be changed is set forth in WO 91/09967.

Suitably, in the first heavy chain variable region (VH1) of the present invention, if the acceptor heavy chain has the human VH3 sequence 1-3 3-07 together with JH4, then the acceptor framework regions of the heavy chain comprise, in addition to one or more donor CDRs, a donor residue at at least one of positions 37, 73, 78 or 94 (according to Kabat et al., (supra)). Accordingly, provided is a bispecific antibody fusion protein, wherein at least the residues at positions 37, 73, 78 and 94 of the first variable domain ofthe heavy chain are donor residues. ly, in the first light chain variable region (VLl) of the t invention, if the acceptor light chain has the human subgroup VKl sequence 2-1 1—02 together with 1K4, then the acceptor framework regions of the light chain comprise, in addition to one or more donor CDRs, a donor residue at at least one of positions 64 or 71. Accordingly, provided is a bispecific antibody fiision protein wherein at least the residues at positions 64 and 71 of the first variable domain of the light chain are donor residues.

Donor residues are residues from the donor antibody, i.e. the antibody from which the CDRs were originally derived.

In one ment, a bispecific dy fusion protein of the present invention comprises a heavy chain, wherein the first variable domain of the heavy chain (VH1) comprises the sequence given in Figure 2 (b) SEQ ID NO:8.

It will be appreciated that one or more amino acid, for example 1 or 2 amino acid, substitutions, additions and/or deletions may be made to the first heavy and light chain variable 40 domains, provided by the t invention, without significantly altering the ability of the antibody fusion n to bind to 0X40 and to neutralise 0X40 activity. The effect of any amino acid substitutions, additions and/or deletions can be readily tested by one skilled in the art, for example by using the methods bed in W02010/0964l8, to determine 0X40 g and ligand blocking.

In one embodiment, a bispecific antibody fusion protein of the present invention ses a heavy chain, wherein the first variable domain ofthe heavy chain comprises a sequence having at least 60% identity or similarity to the sequence given in Figure 2(b) SEQ ID N028. In one embodiment, an antibody fusion protein of the present invention comprises a heavy chain (VH1), wherein the first variable domain of the heavy chain comprises a sequence having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the sequence given in IO SEQ ID NO:8.

In one embodiment, a bispecific antibody fusion protein of the t invention comprises a light chain, n the first variable domain of the light chain (VLI) comprises the sequence given in Figure 2 (a) SEQ ID NO:7.

In another embodiment, a bispecific antibody fusion protein of the present invention ses a light chain, wherein the first variable domain of the light chain comprises a sequence having at least 60% ty or similarity to the ce given in SEQ ID NO:7. In one embodiment the antibody fusion protein of the present invention comprises a light chain, wherein the first variable domain of the light chain comprises a sequence having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the ce given in SEQ ID NO: 7.

In one embodiment a bispecific antibody fusion protein of the present invention comprises a heavy chain, n the first variable domain of the heavy chain (VH1) comprises the sequence given in SEQ ID NO:8 and a light chain, wherein the first variable domain of the light chain (VLI) comprises the sequence given in SEQ ID NO:7.

In another embodiment of the invention, the antibody fusion protein comprises a heavy chain and a light chain, wherein the first variable domain of the heavy chain comprises sequence having at least 60% identity or similarity to the ce given in SEQ ID NO:8 and the first variable domain of the light chain comprises a sequence having at least 60% identity or similarity to the sequence given in SEQ ID NO:7. Suitably, the antibody fusion protein comprises a heavy chain, wherein the first le domain of the heavy chain comprises a sequence having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the sequence given in SEQ ID NO:8 and a light chain, wherein the first variable domain of the light chain comprises a sequence having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the sequence given in SEQ ID NO:7.

In the ific antibody fusion protein of the present invention the heavy chain comprises a CH1 domain and light chain comprises a CL domain, either kappa or lambda.

In one embodiment a bispecific antibody fusion protein of the present invention comprises a heavy chain, wherein the heavy chain comprises the sequence given in SEQ ID NO:10 and a light chain, wherein the light chain comprises the sequence given in SEQ ID NO:9. 40 It will be appreciated that one or more amino acid, for example 1 or 2 amino acid, tutions, additions and/or deletions may be made to the dy variable and/or constant s provided by the present invention without cantly altering the ability of the antibody to bind to 0X40 and to neutralise 0X40 activity. The effect of any amino acid tutions, additions and/or deletions can be readily tested by one skilled in the art, for example by using the methods described in W02010096418, to determine 0X40 g and blocking of the X40L interaction.

In one ment of the invention, the antibody fusion protein comprises a heavy chain, wherein the VH1 and CH1 domains of heavy chain comprise a sequence having at least 60% identity or similarity to the sequence given in SEQ ID NO: 10. Suitably, the antibody fusion comprises a heavy chain, wherein the VH1 and CH1 domains ofthe heavy chain comprise a sequence having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the sequence given in SEQ ID NO:lO.

In one embodiment a bispecific antibody fusion molecule according to the present ion comprises a light chain comprising the sequence given in Figure 2(d), SEQ ID NO:9.

In one embodiment ofthe invention, the antibody fiasion protein comprises a light chain, wherein the VLl and the CH1 s ofthe light chain comprise a sequence having at least 60% identity or similarity to the sequence given in SEQ ID NO:9. For example, the antibody fusion protein ses a light chain, wherein the VLl and CL domains of the light chain comprise a sequence having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the sequence given in SEQ ID NO:9.

The second antigen bound by the bispecific antibody fusion n ofthe present invention is human serum albumin. This is bound by the EV portion of the Fab-dst which is made up ofthe second heavy and light chain variable domains, VH2 and VLZ. In the t invention, VH2 and VLZ are derived from one ofthe antibodies described in W02010/035012 and represent an improved, more human graft of that antibody.

In one embodiment the second heavy chain variable domain (VH2) has the sequence given in Figure 3(a) SEQ ID N021 1.

In one embodiment the second light chain variable domain (VLZ) has the sequence given in Figure 3(b) SEQ ID NO:12.

Accordingly, the present invention provides a bispecific antibody fusion protein which binds human 0X40 and human serum albumin comprising: a heavy chain comprising, in sequence from the N—terminal, a first heavy chain variable domain (V111), a CH1 domain and a second heavy chain variable domain (VH2), a light chain sing, in sequence item the N—terminal, a first light chain le domain (VLl), a CL domain and a second light chain variable domain (VL2), wherein said heavy and light chains are aligned such that VH1 and VLl form a first antigen binding site and VH2 and VLZ form a second antigen binding site, wherein the antigen bound by the first n binding site is human 0X40 and the n bound by the second antigen binding site is human serum albumin, wherein the first variable domain of the heavy chain (VH1) comprises the sequence 40 given in SEQ ID N021 for CDR-Hl, the sequence given in SEQ ID N022 for CDR-HZ and the sequence given in SEQ ID NO:3 for CDR—H3 and the first variable domain of the light chain (VLl) ses the ce given in SEQ ID N024 for , the sequence given in SEQ ID NO:5 for CDR-L2 and the sequence given in SEQ ID NO:6 for CDR—L3, wherein the second heavy chain variable domain (VH2) has the sequence given in SEQ ID NO:11 and the second light chain variable domain (VL2) has the sequence given in SEQ ID NO: 12 and the second heavy chain variable domain (VH2) and second light chain variable domain (VLZ) are linked by a disulphide bond.

Preferably the CH1 domain and the second heavy chain variable domain (VH2) are connected via a linker and the CL domain and the second light chain variable domain (VL2) are connected via linker. Any suitable peptide linker sequence may be used and these may be the same in each chain or ent. Suitable linkers have previously been described in W02010/035012 and are incorporated herein by reference. Examples of suitable linkers are shown in Figure 3 (c) and (d). In one embodiment the linker between the CH1 domain and the second heavy chain variable domain (VH2) comprises or consists of the sequence given in Figure 3 (c) SEQ ID N0:l3. In one embodiment the linker between the CH1 domain and the second heavy chain variable domain (VH2) comprises or consists of the sequence given in Figure 3 (c) SEQ ID NO:14. In one embodiment the linker between the CL domain and the second light chain variable domain (VLZ) comprises or ts of the sequence given in Figure 3(d) SEQ ID NO:14.

In one embodiment the linker in the light chain is a 15 amino acid sequence, in particular GGGGSGGGGSGGGGS (SEQ ID NO: 29).

In one embodiment the linker in the heavy chain is a 16 amino acid sequence, in particular SGGGGSGGGGTGGGGS (SEQ ID NO: 30).

In one embodiment the t invention provides a bispecific antibody fusion protein in which the heavy chain comprises or consists of the sequence given in Figure 3(e) (SEQ ID NO:15) and the light chain comprises or consists of the sequence given in Figure 3(f) (SEQ ID NO: 16).

In one embodiment ofthe invention, the ific antibody fusion protein comprises heavy chain and a light chain, wherein the heavy chain comprises a sequence having at least 60% identity or similarity to the sequence given in SEQ ID NO: 15 and the light chain ses a ce having at least 60% identity or similarity to the sequence given in SEQ ID NO:16. Generally, the antibody fusion comprises a heavy chain, wherein the heavy chain comprises a ce having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the sequence given in SEQ ID NO:15 and a light chain, wherein the light chain comprises a sequence having at least 70%, 80%, 90%, 95% or 98% identity or similarity to the sequence given in SEQ ID NO:16.

The antibody fusion molecules of the present invention suitably have a high binding y, in particular picomolar affinity for human 0X40 and nanomolar affinity for human serum albumin. Affinity may be measured using any suitable method known in the art, 40 including Surface Plasmon Resonance e.g. eTM, as described for 0X40 in WO20100964I 8 and serum albumin in WO2010/035012, using isolated natural or recombinant 0X40 or serum albumin or a suitable fusion protein/polypeptide.

In one example affinity is measured using recombinant human 0X40 extracellular domain as described in W02010/0964I8. In one example the recombinant human 0X40 ellular domain used is a dimer, for example an Fc fiision dimer. Suitably the antibody fusion molecules of the present ion have a g affinity for isolated human 0X40 of about 200pM or less. In one embodiment the antibody molecule of the present invention has a binding affinity of about 100 pM or less. In one embodiment the antibody molecule of the t invention has a binding affinity of about SOpM or less. In one embodiment the antibody IO fusion molecule of the present invention has a binding y of about 40pM or less.

The antibody fusion molecules of the present ion suitably have a high binding affinity for human 0X40 expressed on the surface of activated T cells, for example nanomolar or picomolar affinity. Affinity may be measured using any suitable method known in the art, including the method as described in W02010096418 using activated CD4w 0X40+ human T cells. In particular the antibody fusion molecules of the present invention have a binding affinity for cell surface expressed human 0X40 of about 2nM or better. In one example the antibody les ofthe present invention have a binding affinity for cell surface expressed human 0X40 of about lnM or better. In another example the antibody molecules of the present ion have a binding affinity for cell surface expressed human 0X40 of about 0.5 nM or better. In another example the antibody les of the present invention have a binding affinity for cell surface expressed human 0X40 of about 0.2 nM or better.

Suitably the antibody fusion molecules ofthe present invention have a binding affinity for isolated human serum albumin about SOnM or less. Suitably the antibody fusion les of the present invention have a binding affinity for isolated human serum n of about 20nM or less. In one ment the dy molecule of the present invention has a binding affinity of about lOnM or less. In one embodiment the antibody molecule of the present invention has a binding affinity of about SnM or less. In one ment the antibody fiision molecule ofthe present invention has a binding affinity of about 2nM or less.

The antibody fusion molecules of the t invention can bind human serum albumin and cynomologous, mouse and rat serum albumin. In one embodiment the antibody fusion protein of the present invention bind cynomologus serum albumin with an affinity of SnM or less. In one embodiment the dy fusion protein of the present invention binds mouse serum albumin with an y of SnM or less.

The antibody fiision molecules of the present invention are able to bind human 0X40 and human serum albumin simultaneously.

Advantageously, the fusion molecules of the present invention have a high affinity for 0X40 and also have a adequate half-life in Vivo to be therapeutically useful, for e the half-life is in the range 5-15 days, such as 7—11 days.

It will be appreciated that the affinity of antibody fusion protein provided by the present 40 invention for human 0X40 and/or human serum albumin may be altered using any suitable method known in the art. The present ion therefore also s to variants ofthe antibody 2012/072325 molecules of the present invention, which have an improved affinity for 0X40 or human serum albumin. Such variants can be obtained by a number of affinity maturation protocols including mutating the CDRs (Yang er al., J. Mol. Biol., m, 392-403, 1995), chain shuffling (Marks er al., Bio/Technology, m, 779-783, 1992), use ofmutator strains ofE. coli (Low er al., J. Mol.

Biol, 25—0, 359—368, 1996), DNA shuffling (Patten et al., Curr. Opin. Biotechnol, 8, 724—733, 1997), phage display (Thompson et £11., J. Mol. Biol, &, 77-88, 1996) and sexual PCR (Crameri et al., Nature, §_9_1, 288-291, 1998). n er a]. (supra) discusses these methods of affinity tion.

In one embodiment the bispecific antibody fusion molecules ofthe present invention block the interaction between 0X40 and 0X40L. Numerous assays suitable for determining the ability of an antibody to block this interaction are described in W02010/096418. In one embodiment the present invention provides an antibody fusion protein having specificity for human 0X40 which is capable of inhibiting the binding of human 0X40L (tested at a final concentration of 2ug/ml) to activated human CD4+OX40+ T cells by 50% at a concentration of less than 0.5nM. In one embodiment the human 0X40L used in the assay is natural human 0X40. In one embodiment the human 0X40 used in the assay is recombinant human 0X40.

If desired an dy for use in the present invention may be conjugated to one or more effector molecule(s). It will be appreciated that the effector molecule may comprise a single effector molecule or two or more such molecules so linked as to form a single moiety that can be attached to the antibodies of the t invention. Where it is desired to obtain an antibody fragment linked to an effector molecule, this may be ed by standard chemical or inant DNA procedures in which the antibody nt is linked either directly or via a coupling agent to the effector molecule. Techniques for conjugating such effector molecules to antibodies are well known in the art (see, Hellstrom er a]., lled Drug Delivery, 2nd Ed., Robinson et al., eds, 1987, pp. 623—53; Thorpe et (21., 1982 , Immunol. Rev., 621119—58 and Dubowchik et al., 1999, Pharmacology and Therapeutics, 83, ). Particular al procedures include, for e, those described in W0 93/06231, W0 83, WO 89/00195, WO 76 and W0 03/031581. atively, where the effector molecule is a protein or polypeptide the linkage may be ed using recombinant DNA procedures, for example as described in W0 86/01533 and EP0392745.

The term effector molecule as used herein includes, for example, antineoplastic agents, drugs, toxins, biologically active proteins, for example enzymes, other antibody or antibody fragments, synthetic or naturally occurring polymers, nucleic acids and fragments thereof e.g.

DNA, RNA and fragments thereof, radionuclides, particularly radioiodide, radioisotopes, chelated metals, nanoparticles and reporter groups such as fluorescent nds compounds which may be detected by NMR or ESR spectroscopy.

Examples of effector molecules may include cytotoxins or cytotoxic agents including any agent that is detrimental to (eg. kills) cells. Examples include combrestatins, dolastatins, epothilones, staurosporin, maytansinoids, spongistatins, rhizoxin, halichondrins, roridins, 40 hemiasterlins, taxol, cytochalasin B, gramicidin D, ethidium bromide, emetine, mitomycin, ide, side, vincristine, vinblastine, cin, doxorubicin, daunorubicin, dihydroxy anthracin dione, mitoxantrone, mithramycin, actinomycin D, 1—dehydrotestosterone, glucocorticoids, procaine, tetracaine, lidocaine, propranolol, and puromycin and analogs or homologs thereof.

Effector les also include, but are not limited to, antimetabo lites (e. methotrexate, 6—mercaptopurine, 6—thioguanine, cytarabine, 5—fluorouracil decarbazine), alkylating agents (e.g. mechlorethamine, thioepa chlorambucil, melphalan, carmustine (BSNU) and lomustine (CCNU), cyclothosphamide, busulfan, dibromomannitol, streptozotocin, mitomycin C, and cis—dichlorodiamine platinum (II) (DDP) cisplatin), anthracyclines (e.g. daunorubicin (f01merly daunomycin) and doxorubicin), antibiotics (ag. dactinomycin (formerly actinomycin), bleomycin, mithramycin, anthramycin (AMC), calicheamicins or duocarmycins), and anti-mitotic agents (eg. vincristine and Vinblastine).

Other effector molecules may e chelated radionuclides such as 1“In and 90Y, Lum, Bismuthm, Californiumm, Iridium192 and TungstenlSS/Rheniumlss; or drugs such as but not limited to, hosphocholines, topoisomerase I inhibitors, taxoids and suramin.

Other effector molecules include proteins, peptides and enzymes. Enzymes of interest include, but are not limited to, proteolytic enzymes, hydrolases, lyases, isomerases, transferases. ns, polypeptides and peptides of interest include, but are not limited to, immunoglobulins, toxins such as abrin, ricin A, pseudomonas in, or diphtheria toxin, a protein such as insulin, tumour is factor, terferon, rferon, nerve growth factor, platelet derived growth factor or tissue plasminogen activator, a thrombotic agent or an ngiogenic agent, e. g. tatin or endostatin, or, a biological response r such as a lymphokine, interleukin-1 (IL-1), interleukin-2 , granulocyte macrophage colony stimulating factor (GM—CSF), granulocyte colony ating factor (G—CSF), nerve growth factor (NGF) or other growth factor and immunoglobulins.

Other effector molecules may include detectable substances useful for example in diagnosis. es of detectable substances include various enzymes, prosthetic groups, fluorescent materials, scent materials, bioluminescent materials, radioactive nuclides, positron emitting metals (for use in positron emission tomography), and nonradioactive paramagnetic metal ions. See generally US. Patent No. 4,741,900 for metal ions which can be conjugated to antibodies for use as diagnostics. le enzymes include horseradish dase, ne phosphatase, beta-galactosidase, or acetylcholinesterase; suitable prosthetic groups include avidin, avidin and biotin; suitable fluorescent materials include umbelliferone, fluorescein, fluorescein isothiocyanate, rhodamine, dichlorotiiazinylamine fluorescein, dansyl chloride and phycoerythrin; suitable luminescent materials include luminol; suitable bioluminescent materials include luciferase, luciferin, and aequorin; and suitable radioactive nuclides include 1251, 1311’ mm and ”To.

Where the effector molecule is a polymer it may, in general, be a synthetic or a naturally ing polymer, for example an optionally substituted straight or branched chain polyalkylene, polyalkenylene or polyoxyalkylene r or a ed or unbranched 40 polysaccharide, e.g. a homo- or hetero- polysaccharide. 2012/072325 13,..

Specific optional substituents which may be present on the mentioned synthetic polymers include one or more hydroxy, methyl or methoxy groups.

Specific examples of synthetic polymers include optionally substituted straight or branched chain poly(ethyleneglycol), poly(propyleneglycol) poly(vinylalcohol) or tives thereof, especially optionally substituted poly(ethyleneglycol) such as methoxypoly(ethyleneglycol) or derivatives thereof. c naturally occurring polymers include lactose, amylose, dextran, glycogen or derivatives thereof.

“Derivatives” as used herein is intended to include reactive derivatives, for example IO thiol—selective reactive groups such as maleirnides and the like. The reactive group may be linked directly or through a linker segment to the polymer. It will be appreciated that the residue of such a group will in some instances form part of the product as the g group n the antibody fragment and the r.

The size of the polymer may be varied as desired, but will generally be in an average molecular weight range from 500Da to 50000Da, for example from 5000 to 40000Da such as from 20000 to 40000Da.

In one example suitable effector molecules may be attached through any available amino acid side-chain or terminal amino acid functional group located in the antibody filsion protein, for example any free amino, imino, thiol, hydroxyl or yl group. Such amino acids may occur naturally in the antibody fragment or may be engineered into the fragment using recombinant DNA methods (see for example US 996; US 5,667,425; WO98/25971).

The present invention also es an isolated DNA sequence encoding the heavy and/0r light chain(s) of an antibody molecule of the present invention. Suitably, the DNA sequence encodes the heavy or the light chain of an antibody molecule of the present invention.

The DNA sequence ofthe present invention may comprise synthetic DNA, for instance produced by al processing, cDNA, genomic DNA or any combination thereof.

DNA sequences which encode an antibody molecule of the present invention can be obtained by methods well known to those d in the art. For example, DNA sequences coding for part or all of the antibody heavy and light chains may be synthesised as d from the determined DNA sequences or on the basis of the corresponding amino acid sequences.

DNA coding for acceptor framework ces is widely available to those skilled in the art and can be readily synthesised on the basis of their known amino acid ces.

Standard techniques cular biology may be used to prepare DNA sequences coding for the antibody molecule of the present ion. Desired DNA sequences may be synthesised completely or in part using oligonucleotide synthesis techniques. Site-directed mutagenesis and polymerase chain reaction (PCR) techniques may be used as appropriate.

Examples of suitable sequences are provided in Figure 5 (a) SEQ ID N022l; Figure 5 (b) SEQ ID N022; Figure 6 (a) SEQ ID N023; Figure 6 (b) SEQ ID NO:24. Nucleotides 1- 40 63 in SEQ ID NO 21 and 1-63 in SEQ ID NO:23 encode the signal peptide ce OmpA which is cleaved to give an antagonistic dy fusion molecule ofthe present invention.

WO 20137068563 Z325 The present invention also provides an isolated DNA sequence encoding the heavy chain of an antibody fusion protein ofthe present invention which comprises SEQ ID NO:2I or SEQ ID NO:22. The present ion also provides an isolated DNA ce encoding the light chain ofan antibody fiision molecule of the present invention which comprises SEQ ID NO:23 or SEQ ID NO:24.

Other examples of suitable sequences are ed in Figure 7 (a) SEQ ID NO:25; Figure 7 (b) SEQ ID NO:26; Figure 8 (a) SEQ ID NO:27; Figure 6 (b) SEQ ID NOz28.

Nucleotides 1-57 in SEQ ID NO 25 and I~60 in SEQ ID NO 27 encode the signal peptide sequence from mouse antibody B723 (Whittle et al., I987, Protein Eng. 1(6) 499-505.) which is cleaved to give an antagonistic antibody fusion molecule of the present invention. The present invention also provides an isolated DNA sequence ng the heavy chain of an antibody fusion protein ofthe t invention which comprises SEQ IDN0225 or SEQ ID N0226. The present invention also provides an isolated DNA sequence ng the light chain of an antibody fusion molecule of the present invention which comprises SEQ ID NOz27 or SEQ ID NO:28.

The present invention also relates to a cloning or expression vector comprising one or more DNA sequences of the present invention. Accordingly, provided is a cloning or expression vector comprising one or more DNA sequences encoding an antibody fusion protein ofthe present invention. Suitably, the cloning or expression vector comprises two DNA sequences, ng the light chain and the heavy chain of the antibody molecule of the present invention, respectively. Suitably, a vector according to the t invention comprises the sequences given in SEQ ID N021 and SEQ ID NO:23. Nucleotides 1-63 in SEQ ID NO 21 and 1—63 in SEQ ID NO 23 encode the signal peptide sequence from OmpA.

General methods by which the vectors may be constructed, transfection methods and culture methods are well known to those skilled in the art. In this respect, reference is made to “Current Protocols in lar y”, 1999, F. M. Ausubel (ed), Wiley Interscience, New York and the Maniatis Manual produced by Cold Spring Harbor hing.

Also provided is a host cell comprising one or more cloning or expression vectors comprising one or more DNA sequences encoding an antibody fusion protein of the present invention. Any suitable host cell/vector system may be used for expression of the DNA sequences encoding the antibody molecule of the t invention. Bacterial, for example E. 0012', and other microbial systems may be used or eukaryotic, for example mammalian, host cell expression systems may also be used. Suitable mammalian host cells include CHO, myeloma or hybridoma cells.

The present invention also provides a process for the production of an antibody fusion molecule according to the present ion comprising culturing a host cell containing a vector of the present invention under conditions suitable for leading to expression ofprotein from DNA encoding the antibody molecule ofthe present invention, and isolating the antibody molecule. 40 For tion ofproducts sing both heavy and light , the cell line may be transfected with two vectors, a first vector ng a light chain polypeptide and a second PCT/EPZO12/072325 vector encoding a heavy chain polypeptide. Alternatively, a single vector may be used, the vector including sequences encoding light chain and heavy chain polypeptides.

As the dy fusion ns ofthe present ion are useful in the treatment and/or prophylaxis of a pathological condition, the present invention also provides a pharmaceutical or diagnostic composition comprising an antibody molecule of the present invention in combination with one or more of a ceutically acceptable excipient, diluent or carrier. Accordingly, provided is the use of an antibody fusion protein of the invention for the manufacture of a medicament. The composition will usually be supplied as part of a e, pharmaceutical composition that will normally include a pharmaceutically acceptable carrier. A ceutical composition of the present invention may additionally comprise a pharmaceutically-acceptable adjuvant.

The present invention also provides a process for preparation of a phamiaceutical or diagnostic composition comprising adding and mixing the antibody fusion le of the present invention together with one or more of a pharmaceutically acceptable excipient, diluent or carrier.

The antibody fusion molecule may be the sole active ingredient in the pharmaceutical or diagnostic composition or may be accompanied by other active ients including other antibody ingredients, for example anti-TNF, anti— IL-lfi, anti-T cell, anti—IFNy or anti-LPS dies, or non-antibody ingredients such as xanthines. Other suitable active ingredients include antibodies capable of inducing tolerance, for example, anti-CD3 or anti-CD4 antibodies.

In a further embodiment the antibody fusion protein or composition ing to the disclosure is employed in combination with a r pharmaceutically active agent, for example a corticosteroid (such as fluticasonoe propionate) and/or a beta—Z-agonist (such as salbutamol, salmeterol or formoterol) or inhibitors of cell growth and proliferation (such as rapamycin, cyclophosphmide, methotrexate) or ative a CD28 and /or CD40 inhibitor. In one embodiment the r is a small molecule. In another embodiment the inhibitor is an antibody specific to the target.

The pharmaceutical compositions suitably se a therapeutically effective amount ofthe antibody fusion protein ofthe ion. The term “therapeutically effective amount” used herein refers to an amount of a therapeutic agent needed to treat, ameliorate or prevent a targeted disease or condition, or to t a detectable therapeutic or tative effect. For any antibody, the therapeutically effective amount can be estimated initially either in cell culture assays or in animal models, usually in s, rabbits, dogs, pigs or es. The animal model may also be used to determine the appropriate concentration range and route of administration. Such information can then be used to determine useful doses and routes for administration in humans.

The precise therapeutically effective amount for a human subject will depend upon the severity of the disease state, the general health of the subject, the age, weight and gender of the 40 subject, diet, time and fi'equency of administration, drug combination(s), reaction ivities and tolerance/response to therapy. This amount can be determined by routine experimentation and is within the judgement of the clinician. Generally, a therapeutically effective amount will be from 0.01 mg/kg to 50 mg/kg, for example 0.1 mg/kg to 20 mg/kg. Pharmaceutical compositions may be conveniently presented in unit dose forms containing a predetermined amount ofan active agent ofthe invention per dose.

Compositions may be administered individually to a patient or may be administered in combination (e.g. simultaneously, sequentially or separately) with other agents, drugs or hormones.

The dose at which the antibody fusion molecule of the present invention is administered depends on the nature of the ion to be treated, the extent of the inflammation present and on whether the antibody molecule is being used prophylactically or to treat an existing condition.

The frequency of dose will depend on the half—life of the antibody fusion molecule and the duration of its effect. If the antibody molecule has a short half—life (e.g. 2 to 10 hours) it may be necessary to give one or more doses per day. Alternatively, if the antibody molecule has a long half life (e. g. 2 to 15 days) it may only be necessary to give a dosage once per day, once per week or even once every 1 or 2 months.

The ceutically acceptable carrier should not itself induce the production of dies harmfiil to the individual ing the composition and should not be toxic. Suitable carriers may be large, slowly lised macromolecules such as proteins, polypeptides, mes, polysaccharides, polylactic acids, polyglycolic acids, polymeric amino acids, amino acid copolyrners and inactive virus particles.

Pharmaceutically acceptable salts can be used, for example mineral acid salts, such as hydrochlorides, hydrobromides, phosphates and sulphates, or salts of c acids, such as acetates, propionates, malonates and benzoates.

Pharmaceutically able carriers in therapeutic compositions may additionally n liquids such as water, saline, glycerol and ethanol. Additionally, auxiliary substances, such as wetting or emulsifying agents or pH buffering substances, may be present in such compositions. Such carriers enable the pharmaceutical compositions to be formulated as tablets, pills, dragees, capsules, liquids, gels, syrups, slurries and suspensions, for ingestion by the patient.

Suitable forms for administration include forms suitable for parenteral stration, e. g. by ion or infusion, for example by bolus injection or continuous infusion. Where the product is for injection or infusion, it may take the form of a suspension, solution or on in an oily or aqueous e and it may contain formulatory agents, such as suspending, preservative, stabilising and/or dispersing agents. Alternatively, the antibody le may be in dry form, for reconstitution before use with an appropriate sterile liquid.

Once ated, the compositions of the invention can be administered directly to the subject. The ts to be treated can be animals. However, in one or more ments the compositions are adapted for administration to human subjects. 40 Suitably in formulations according to the present disclosure, the pH of the final formulation is not similar to the value of the isoelectric point of the antibody or fragment, for 2012/072325 example ifthe pH of the formulation is 7 then a pl of fiom 8—9 or above may be appropriate.

Whilst not wishing to be bound by theory it is thought that this may ultimately provide a final formulation with improved stability, for example the antibody or fragment remains in solution.

In one aspect advantageously the fusion molecule of the present disclosure does not have a pl which corresponds to an overall neutral molecule. This renders the molecule less susceptible to aggregation.

The pharmaceutical compositions of this invention may be administered by any number of routes including, but not limited to, oral, intravenous, intramuscular, arterial, intramedullary, intrathecal, intraventricular, ermal, transcutaneous (for example, see WO98/20734), aneous, eritoneal, asal, enteral, topical, sublingual, intravaginal or rectal routes. rays may also be used to administer the pharmaceutical compositions of the invention. lly, the therapeutic compositions may be prepared as injectables, either as liquid solutions or suspensions. Solid forms suitable for solution in, or suspension in, liquid vehicles prior to injection may also be ed.

Direct delivery of the compositions will generally be accomplished by injection, subcutaneously, eritoneally, intravenously or intramuscularly, or delivered to the interstitial space of a tissue. The compositions can also be administered into a lesion. Dosage treatment may be a single dose le or a multiple dose schedule.

It will be appreciated that the active ingredient in the composition will be an antibody molecule. As such, it will be susceptible to degradation in the gastrointestinal tract. Thus, if the composition is to be administered by a route using the gastrointestinal tract, the composition will need to contain agents which protect the antibody from degradation but which release the antibody once it has been absorbed from the gastrointestinal tract.

A thorough discussion ofpharmaceutically acceptable carriers is available in Remington's Pharmaceutical Sciences (Mack Publishing Company, NJ. 1991).

In one embodiment the formulation is provided as a formulation for topical administrations ing inhalation.

Suitable ble preparations include inhalable s, metering aerosols ning lant gases or inhalable solutions free from propellant gases. Inhalable powders ing to the disclosure containing the active substance may consist solely of the abovementioned active substances or of a mixture of the abovementioned active substances with physiologically acceptable excipient.

These inhalable powders may include monosaccharides (eg. glucose or arabinose), disaccharides (eg. lactose, saccharose, maltose), oligo- and ccharides (e. g. nes), polyalcohols (eg. sorbitol, mannitol, xylitol), salts (e. g. sodium chloride, calcium carbonate) or mixtures of these with one another. Mono- or disaccharides are suitably used, the use of lactose or glucose, particularly but not exclusively in the form of their hydrates.

Particles for deposition in the lung require a particle size less than 10 microns, such as 1-9 microns for example fiom 0.1 to 5 um, in particular from 1 to 5 um. The particle size of the 40 active ingredient (such as the antibody or fragment) is of primary importance.

The propellent gases which can be used to prepare the ble aerosols are known in the art. Suitable propellent gases are selected fiom among hydrocarbons such as n-propane, n- butane or isobutane and halohydrocarbons such as chlorinated and/or fluorinated derivatives of methane, ethane, propane, , cyclopropane or cyclobutane. The abovementioned propellent gases may be used on their own or in mixtures thereof. ularly suitable propellent gases are halogenated alkane tives selected fiom among TG 11, TG 12, TG 134a and TG227. Of the abovementioned nated hydrocarbons, TG134a (1,1,1 ,2-tetrafluoroethane) and TG227 (1,1,1,2,3,3,3—heptafluoropropane) and mixtures thereof are particularly suitable.

The propellent—gas-containing inhalable aerosols may also contain other ingredients such as cosolvents, stabilisers, e—active agents (surfactants), antioxidants, ants and means for adjusting the pH. All these ingredients are known in the art.

The propellant—gas—containing inhalable aerosols according to the invention may contain up to 5 % by weight of active substance. Aerosols according to the invention n, for example, 0.002 to 5 % by weight, 0.01 to 3 % by weight, 0.015 to 2 % by weight, 0.1 to 2 % by weight, 0.5 to 2 % by weight or 0.5 to 1 % by weight of active ingredient.

Alternatively topical administrations to the lung may also be by administration of a liquid solution or sion formulation, for example ing a device such as a nebulizer, for example, a nebulizer ted to a compressor (e.g., the Pari LC-Jet Plus(R) nebulizer connected to a Pari Master(R) compressor manufactured by Pari Respiratory Equipment, Inc, Richmond, Va.) The antibody fusion protein of the invention can be red dispersed in a solvent, e.g., in the form of a solution or a suspension. It can be suspended in an appropriate physiological solution, e.g., saline or other pharmacologically acceptable solvent or a ed solution. Buffered solutions known in the art may contain 0.05 mg to 0.15 mg disodium edetate, 8.0 mg to 9.0 mg NaCl, 0.15 mg to 0.25 mg polysorbate, 0.25 mg to 0.30 mg anhydrous citric acid, and 0.45 mg to 0.55 mg sodium citrate per 1 m1 of water so as to achieve a pH of about 4.0 to 5.0. A suspension can employ, for example, lyophilised antibody.

The therapeutic suspensions or solution formulations can also contain one or more excipients. Excipients are well known in the art and include buffers (e.g., citrate buffer, phosphate buffer, acetate buffer and bicarbonate buffer), amino acids, urea, alcohols, ascorbic acid, phospholipids, proteins (e.g., serum albumin), EDTA, sodium chloride, liposomes, mannitol, sorbitol, and glycerol. Solutions or suspensions can be encapsulated in liposomes or radable microspheres. The formulation will generally be provided in a substantially sterile form employing e manufacture processes.

This may include production and ization by filtration of the ed solvent/solution used for the formulation, aseptic suspension of the antibody in the e buffered solvent solution, and dispensing of the formulation into sterile receptacles by methods familiar to those of ordinary skill in the art. zable formulation according to the present disclosure may be provided, for example, as single dose units (e.g., sealed plastic ners or vials) packed in foil pes.

Each vial contains a unit dose in a volume, e.g., 2 mL, of so lvent/solutionbuffer.

The antibody fusion proteins disclosed herein may be suitable for delivery via nebulisation.

It is also envisaged that the antibody of the present invention may be administered by use of gene therapy. In order to achieve this, DNA sequences encoding the heavy and light chains of the antibody molecule under the control of appropriate DNA ents are introduced into a patient such that the dy chains are expressed from the DNA sequences and assembled in situ.

The present invention also provides an antibody fusion molecule (or compositions comprising same) for use in the control of inflammatory diseases, for example acute or c inflammatory e. Suitably, the antibody molecule (or compositions comprising same) can be used to reduce the inflammatory process or to prevent the inflammatory process. In one embodiment there is provided an in viva ion of activated T cells, in particular those involved in inappropriate inflammatory immune responses, for example recruited to the ty/location of such a response.

Reduction of activated T cells, as employed herein, may be a reduction, 10, 20, 30, 40, 50, 60, 70, 80, 90 or more percent in comparison to before treatment or t ent. ageously, treatment with an antibody, fragment or composition according to the present invention, may allow the reduction in the level of activated T cells, without reducing the patients general level of T cells ivated T cells). This may result in fewer side effects, and possibly prevent T cell ion in the patient.

The present invention also provides the antibody fusion molecule of the present invention for use in the treatment or laxis of a pathological disorder that is mediated by 0X40 or associated with an increased level of 0X40. The pathological ion, may, for example be selected from the group consisting of infections (viral, bacterial, fungal and parasitic), endotoxic shock associtated with infection, arthritis, rheumatoid arthritis, asthma, COPD, pelvic inflammatory disease, Alzheimer’s Disease, inflammatory bowel disease, Crohn’s disease, tive colitis, Peyronie’s Disease, coeliac disease, gallbladder disease, Pilonidal disease, peritonitis, psoriasis, vasculitis, surgical adhesions, stroke, Type 1 Diabetes, lyme disease, arthritis, meningoencephalitis, autoimmune uveitis, immune mediated inflammatory disorders ofthe central and peripheral nervous system such as multiple sclerosis, lupus (such as systemic lupus erythematosus and lupus nephritis) and Guillain—Barr syndrome, Atopic dermatitis, autoimmune hepatitis, fibrosing alveolitis, Grave’s disease, IgA nephropathy, idiopathic thrombocytopenic purpura, Meniere’s disease, pemphigus, y y cirrhosis, sarcoidosis, scleroderma, Wegener’s granulomatosis, other autoimmune ers, pancreatitis, trauma (surgery), grafi-versus-host disease, transplant rejection, heart disease including ischaemic diseases such as myocardial infarction as well as atherosclerosis, ascular 40 coagulation, bone tion, osteoporosis, osteoarthritis, periodontitis and hypochlorhydia.

In one embodiment the antibody fusion protein according to the invention is employed in the treatment of allergy, COPD, autoimmune disease, rheumatoid tis, asthma, graft versus host disease, Crohn’s disease, ulcerative colitis, type— 1 es, multiple sclerosis, Systemic lupus erythematosis, lupus nephritis, Myasthenia Gravis, s disease, transplant rejection, r’s granulomatosis, Henoch—Schonlein purpura, systemic sclerosis or viral- d lung inflammation.

In one embodiment the antibody fusion protein according to the invention is employed in the ent of a disease selected from the group consisting of allergy, COPD, autoimmune disease, rheumatoid arthritis, asthma, graft versus host disease, Crohn’s disease, ulcerative IO colitis, type- 1 diabetes, multiple sclerosis, Systemic lupus erythematosis, lupus nephritis, Myasthenia , Grave’s disease, transplant rejection, Wegener’s granulomatosis, Henoch— Schonlein purpura, ic sclerosis and viral-induced lung inflammation.

The present invention also provides an antibody fusion molecule according to the present invention for use in the ent or prophylaxis ofpain, particularly pain associated with inflammation.

In one embodiment the mechanism through which the fusion molecules of the present sure work include one or more of inhibition ofT cell proliferations or survival, enhancement ofTReg generation, reduced differentiation of B cells and/or decreased cytokine production.

The t invention firrther provides the use of an antibody fusion le or composition according to the present invention in the manufacture of a ment for the treatment or prophylaxis of a pathological disorder that is mediated by 0X40 or associated with an sed level of 0X40, in particular the pathological disorder is rheumatoid tis, asthma or COPD.

The present ion r provides the use of an antibody molecule, fragment or composition according to the present invention in the manufacture of a medicament for the treatment or prophylaxis of one or more medical indications described herein.

An antibody fusion molecule or composition of the present invention may be utilised in any therapy where it is desired to reduce the effects of 0X40 in the human or animal body. 0X40 may be circulating in the body or may be present in an undesirably high level localised at a particular site in the body, for example a site of inflammation.

In one ment the antibody fusion molecule of the present invention or a composition comprising the same is used for the control of inflammatory e, e.g. as described herein.

The present invention also provides a method of treating human or animal subjects suffering from or at risk of a disorder mediated by 0X40, the method comprising administering to the subject an effective amount of the antibody fusion molecule of the present invention, or a composition comprising the same.

In one embodiment there is provided a purified bispecific antibody fusion n which 40 binds human 0X40 and human serum albumin, in substantially purified from, in particular free or substantially free of endotoxin and/or host cell protein or DNA.

WO 68563 2012/072325 Purified form as used supra is intended to refer to at least 90% purity, such as 91, 92, 93, 94, 95, 96, 97, 98, 99% w/w or more pure.

Substantially free of endotoxin is generally intended to refer to an xin content of 1 EU per mg antibody product or less such as 0.5 or 0.1 EU per mg product. ntially free of host cell protein or DNA is generally intended to refer to host cell protein and/or DNA content 400ug per mg of antibody product or less such as lOOug per mg or less, in particular 20pg per mg, as appropriate.

The antibody fiision molecule ofthe present invention may also be used in diagnosis, for example in the in vivo diagnosis and imaging of e states ing 0X40. l0 Advantageously, the present fusion molecules are thought to be safe for administration to humans at a proper therapeutic dose, in particular because they are not superagonists and are unlikely to cause cytokine storm.

Superagonist as employed herein refers to an antibody which s T cells in the absence ofTCR engagement.

In one embodiment A26 Fab-Fv reduces the Division Index indicating that fewer cells in the population are committed to on; this effect is presumably mediated by the NK cells that are expressing 0X40. The Division Index represents the average number of cell divisions that a cell in the original tion has undergone and includes the undivided cells.

The eration Index reflects proliferation of the ding population only, and in one embodiment the inhibitory effect of A26 Fab-Fv using this measure is relatively reduced.

Comprising in the context of the present specification is intended to meaning including.

Where technically appropriate embodiments ofthe invention may be combined.

Embodiments are bed herein as comprising certain features/elements. The disclosure also extends to separate embodiments consisting or consisting essentially of said features/elements.

The present invention is further described by way of illustration only in the following examples, which refer to the anying Figures, in which: EXAMPLES Figures in detail: Figure l: A bispecific antibody fusion protein of the present invention, referred to as a Fab- dst.

Figure 2: a) Light chain V region of antibody A26 (SEQ ID NO:7) b) Heavy chain V region of antibody A26 (SEQ ID N028) c) CDRHl (SEQ ID NO: 1), CDRH2 (SEQ ID N022), CDRH3 (SEQ ID NO:3), CDRLl (SEQ ID N024), CDRL2 (SEQ ID NO:5) and CDRL3 (SEQ ID NO:6) of antibody A26. d) Light chain of antibody A26 Fab component (SEQ ID N019) 40 e) Heavy chain of antibody A26 Fab component (SEQ ID NO: 10) Figure 3 a) Heavy chain of anti-albumin FV component 645gI—I5 (SEQ ID N011 1) b) Light chain of anti-albumin Fv component 645gL4 (SEQ ID NO:12) c) Linker l (SEQ ID NO:I3) d) Linker 2 (SEQ ID NO: 14) e) Fab—dst heavy chain (SEQ ID NO:15) t) Fab—dst light chain (SEQ ID NO: l6) Figure 4 a) 645g1 heavy chain variable domain (SEQ ID NOzl7) b) 645g1 light chain variable domain (SEQ ID NO:18) l0 c) A26 Fab-dst 645ng (SEQ ID NO: 19) d) A26 Fab-dsFV 645ng (SEQ ID NO:20) Figure 5 a) DNA encoding heavy chain of the Fab—dst including OmpA leader (SEQ ID NO:2l) b) DNA encoding heavy chain ofthe FV Without OmpA leader (SEQ ID NO:22) I5 Figure 6 a) DNA encoding light chain of the Fab-dst including OmpA leader (SEQ ID NO:23) b) DNA encoding light chain of the Fab-dsFV without OmpA leader (SEQ ID NO:24) Figure 7 a) DNA encoding heavy chain of the Fab—dsFV including B72.3 leader (SEQ ID NO:25) b) DNA encoding heavy chain of the Fab-dst without B723 leader (SEQ ID NO:26) Figure 8 a) DNA encoding light chain of the FV including B723 leader (SEQ ID NO:27) b) DNA encoding light chain of the Fab-dsFV without B723 leader (SEQ ID NO:28) Figure 9a shows g ofAlexaFluor 488 labelled A26 Fab—dsFV to activated human CD4+OX40+T cells Figure 9b shows binding for A26 Fab’, A26 Fab-Fv and A26 Fab’—PEG in the presence of % HSA on activated human CD4+, OX40+ T cells Figure 10a shows the effect ofA26 Fab-dst on cytokine production from PBMC exposed to Dermatophagoz'des pteronyssinus allergic extract Figure 10b shows the ability of A26 t to inhibit CD4+ and CD8+ T cell proliferation in a Hu-NSG mouse model Figure 11a shows inhibition of OX4OL binding to human activated CD4+ OX40+ T cells by A26 Fab-dsFV Figure llb shows inhibition of OX40L binding to human activated CD4+ 0X40+ T cells by A26 Fab’, A26 Fab-dst, A26 Fab’-PEG and two controls.

Figure 12a shows A26 Fab-Fv inhibits a human missed lymphocyte reaction (MLR) Figure 12b shows A26 Fab—Fv inhibits IFN-gamma production during a human MLR Figure 13 shows A26 Fab-Fv reduces the percentage of activated (CD25+) CD4+ T cells after secondary antigen re-stimulation with ophagoz‘des yssinus 40 allergenic extract 7 _-3 Figure 14 shows Fab—Fv and Fab—PEG administered prior to cell transfer dose dependency inhibits CD4+ and CD8+ T cell proliferation in the Hu-NSG model DNA manipulations and l methods Competent E. coli strains were used for transformations and routine e growth. DNA restriction and modification enzymes were obtained from Roche Diagnostics Ltd. and New England s. Plasmid preparations were performed using Maxi Plasmid purification kits N, catalogue No. 12165). DNA sequencing reactions were performed using the ABI Prism Big Dye terminator sequencing kit (catalogue No. 4304149) and run on an ABI 3100 automated sequencer (Applied Biosystems). Data was analysed using the program Sequencher (Genecodes). Oligonucleotides were obtained from Simga or Invitrogen. Genes encoding initial on ces were constructed by an automated synthesis approach by DNA2.0, and modified to generate the grafted versions by oligonucleotide ed mutagenesis. The concentration of Fab-Fv was determined by a Protein-G based HPLC method.

EXAMPLE 1 Generation and analysis of different humanisation grafts of 645 in A26Fab—645dst We have previously described the Fab—dst dy format (Figure 1) (sometimes referred to herein simply as Fab-Fv) and a humanised anti-albumin antibody known as ‘645ngng ’ in W02010/035012. We have also previously described the generation of a humanised antagonistic anti—0X40 antibody known as ‘A26’ and a PEGylated Fab’ fiagment thereof in W02010/096418. Here we describe the generation of a new improved humanised graft of antibody ‘645 ’ known as 645dng5gL4 and the tion of a Fab-dst antibody molecule incorporating that graft in the EV component and the ‘A26’ variable regions in the Fab component. The variable regions of A26 are given in Figure 2a and b (SEQ ID NOS 7 and 8).

The variable and constant region sequences of A26 combined are given in Figure 2d and (SEQ ID NOs 9 and 10).

The sequences of 645ng and ng are given in Figure 4(a) and (b), SEQ ID NOs 17 and 18.

Where the term Fab’—PEG or A26 EG is used this refers to the A26 Fab-40K PEG’ bed in W02010/096418. 1.1. Construction of A26Fab-645dst(gH1ng) and A26Fab-645dst(gH5gL4)G4S linker plasmids The total coding region ofA26Fab-645dst(gL1) light chain (SEQ ID N0220) was cloned into a UCB mammalian expression vector under the control of the HCMV-MIE er and SV40E polyA sequence. The light chain variable region of 645dst(gL1) (SEQ ID NOzl8) was d to 645dst(gL4) (SEQ ID NO:12) by an pping PCR method. The total coding region of A26Fab-645dsFV(ng) heavy chain (SEQ ID NO:19 was cloned into a UCB mammalian expression vector under the control of the HCMV-MIE promoter and SV40E polyA ce. The heavy chain variable region of 645dst(ng) (SEQ ID NO:17) was mutated to 645dst(gH5) (SEQ ID NO:11) by an overlapping PCR method. The constructs were verified by sequencing. Both constructs ned the 3XG4S linker given in SEQ ID NO:l4, Figure 3(d). 1.2 Mammalian expression of A26Fab—645dst(gH1gL1) and A26Fab—645dst(gH5gL4) HEK293 cells were transfected with the heavy and light chain plasmids using ogen’s lO 293fectin transfection reagent according to the manufacturer’s instructions. Briefly, 25ug heavy chain plasmid and 25 ug light chain plasmid were incubated with lOOul 293fectin and l700ul o media for 20mins at RT. The e was then added to 50x106 HEK293 cells in 50ml suspension and incubated for 6 days with shaking at 37°C. After 6 days the supernatant was collected by centrifugation at lSOOxg for 10 minutes to remove the cells and then 0.22um l5 sterile filtered. 1.3 Protein-G purification of A26Fab-645dst(gH1gL1) and A26Fab-645dst(gH5gL4) The ~50ml of 0.22um filtered supernatants were concentrated to ~2ml using Amicon Ultra-l 5 concentrators with a lOkDa molecular weight cut off membrane and centrifugation at 4000xg in a swing out rotor. 1.8ml of concentrated atant was d at lml/min to a lml Gammabind Plus Sepharose (GE Healthcare) column equilibrated in ZOmM phosphate, 40mM NaCl pH7.4. The column was washed with 20mM phosphate, 40mM NaCl pH7.4 and the bound material eluted with 0.1M glycine/HCl pH2.7. The elution peak was collected and pH adjusted to ~pH7 with 2M Tris/HCl pH8.5. The pH adjusted elution was concentrated and diafiltered into 20mM phosphate, ISOmM NaCl pH7.4 using Amicon Ultra—15 concentrators with a lOkDa molecular weight cut off membrane and centrifugation at 4000xg in a swing out rotor, to a final volume of ~0.3ml. 1.4 Size exclusion analysis A26Fab-645dst(gH1ng) and A26Fab-645dst(gH5gL4) Protein-G purified samples were analysed by size exclusion HPLC. The samples were separated on a Superdex 200 10/300 GL Tricom column (GE Healthcare) developed with an isocratic nt of PBS pH7.4 at n. Peak detection was at 280nm and nt molecular weight was calculated by comparison to a standard curve of known molecular weight proteins verses elution volume. Changing the humanisation graft of the 645dst from ngng WO 68563 to gH5gL4 resulted in an increase in the percentage monomer of the expressed A26Fab- 645dst from 59% to 71% an increase of 12%, t any change in the thermal stability of the dst (data not shown) or in the affinity of binding ofthe dst to HSA (data not shown).

Example 2 2.1 BIAcore kinetics for A26 Fab-dst (645gH5gL4) g 0X40 In this and all uent examples the A26 Fab-dst 645gHSgL4 had the heavy chain sequence given in SEQ ID NO:15 (Figure 3 (e)) and the light chain sequence given in SEQ ID NO:16 (Figure 3(3) i.e. the heavy chain contained the G48, G4T, G4S linker given in SEQ ID N0213, figure 3(0).

BIA (Biamolecular ction Analysis) was performed using a BlAcore T200 (GE Healthcare). Affinipure F(ab’)2 Fragment goat anti-human IgG, F(ab')2 fragment specific (Jackson ImmunoReseareh) was immobilised on a CMS Sensor Chip via amine coupling chemistry to a capture level of z5000 response units (RUs). HBS—EP buffer (lOmM HEPES pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.05 % Surfactant P20, GE Healthcare) was used as the running buffer with a flow rate of 10 uL/min. A 10 it]. injection of A26 Fab' at 0.5ug/mL or A26Fab-dst at 1 ng/mL was used for capture by the immobilised anti-human IgG-F(ab')2.

Human 0X40 was titrated over the captured A26 at various concentrations (25nM to 1.5625nM) at a flow rate of 30 nL/min. The e was regenerated by 2 x 10 nL ion of 50 mM HCl, followed by a 5 nL injection of 5 mM NaOH at a flowrate of in.

Background subtraction g curves were analysed using the T200evaluation software (version 1.0) following standard procedures. Kinetic parameters were determined from the fitting algorithm. ka(1/Ms) kd(1/s) 2.18 i 0.38 E+05 1.00 13—05 4.68E—11 2.55 i 0.35 E+05 1.04 13-05 4.12E—11 Fab’ PEG 2.33 i 0.46 E+05 1.12 E-05 4.84E—11 Average of 4 determinations Table 1 2.2. BIAcore kinetics for A26 Fab—dst 5gL4) binding albumin BlA (Biamolecular lnteraction Analysis) was performed using a BlAcore T200 (GE Healthcare). ure F(ab')2 nt goat uman IgG, F(ab')2 fragment specific (Jackson lmmunoResearch) was immobilised on a CMS Sensor Chip via amine coupling chemistry to a capture level of#5000 response units (RUs). HBS—EP buffer (lOmM HEPES pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.05 % Surfactant P20, GE Healthcare) was used as the running buffer with a flow rate of 10 uL/min. A 10 uL injection of Fab-Fv at 0.75ug/mL was used for capture by the immobilised anti—human lgG—F(ab’)2. Human Serum Albumin (HSA), Mouse Serum albumin (MSA) and lgus Serum Albumin (CSA) was titrated over the captured Fab—FV at s concentrations (50nM to 6.25nM) at a flow rate of 30 uL/min. The e was regenerated by 2 X 10 uL injection of 50 mM HCl, followed by a 5 uL injection of mM NaOH at a flowrate of 10uL/min. Background ction binding curves were analysed using the T200evaluation software (version 10) following standard procedures. Kinetic parameters were determined from the fitting algorithm.

Table 2 ka(1/Ms) kd(1/s) .84 E+04 1.63 E—O4 2.93E-09 8.86 E+04 3.68 E-O4 4.16E-09 7.1 E+04 1.89 E-O4 2.66E—09 Average of 3 determinations 2.3 Demonstration ofA26 Fab-dst(645gH5gL4) binding 0X40 and albumin simultaneously The simultaneous binding ofhuman 0X40 and Human Serum Albumin to A26 Fab—dst assessed. The A26 Fab—dsFV construct was captured to the sensor chip surface as stated in the method for Biacore kinetics for binding A26 Fab—dsFV albumin. 50nM HAS, 25nM 0X40 or a mixed solution with final concentration of 5OnM BSA and 25nM 0X40 were titrated separately over the captured A26 Fab—dst. The binding se for the combined HSA/OX40 solution was equivalent to the sum of the responses of the independent injections. This confirms that the Fab-dsFV is capable of simultaneous binding to both human 0X40 and HSA.

Table 3 Sample Analyte Binding (RU) hOX4O 25 A26 Fab-Fv HSA 9 hOX40 + HSA 35 (34) 2.4 Cell-based affinity of A26 Fab—dst (645gH5gL4) Methods: A26 Fab-Fv binding to human activated CD4+0X40+ T cells.

PBMC were isolated by separation on a Ficoll gradient and activated with 4ug/mL PHA-L for 3 days at 37°C, 5% C02, 100% humidity. CD4+ T cells were isolated by negative selection using magnetic beads (CD4+ T cell Isolation Kit H for Human; Miltenyi Biotec). Approximately 1 x l05 cells were incubated in the presence of antibody in either Facs buffer .2% BSA/0.09% NaN3) or Facs buffer mented with 5% HSA at 4°C. The final concentration of the antibody ranged from 4811M — 0.0005nM)) . The cells were washed in PBS prior to analysis by flow cytometry using a FACScalibur (Becton Dickinson).Two titration data sets were produced in both buffer conditions, one with A26 Fab—dst and the second with an vant control Fab-FV to determine non—specific binding. The number of moles of bound antibody were calculated by using interpolated values from a standard curve generated by use of beads comprised of differing but known amounts of fluorescent dye. Geometric mean fluorescence values were ined in the flow cytometric analyses of cells and beads. Non— specific binding was subtracted from the A26 Fab-dst values and the specific g curve thus generated analysed by non-linear sion using a one-site binding equation (Graphpad Prism®) to determine the K13.

To determine the affinity ofA26 Fab-dsFV for cell surface expressed antigen, saturation binding experiments were performed using ted CD4+OX40Jr T cells, and Alexa Fluor 488—labelled A26 Fab-dsFV. Specific binding of antibody to receptor at equilibrium across a range of antibody concentrations was used to determine KD, assuming that only a very small fraction of antibody was bound to receptor at any point on the binding curve.

Equilibrium binding is described using the following equation: Receptor free + dyfrcc ’ a Receptor-Antibody kofr The rate of association of antibody with receptor = kon x [Receptor free] x [Antibody free] The rate of dissociation of receptor-antibody complex = koffX tor-Antibody] At brium, the association and dissociation rates are equal and an equation can be derived which describes the g isotherm; on a semi-log plot the g is sigmoidal. The KD is wo 2013/068563 defined by koff/ kon and can be calculated from the binding curve as the concentration at which half-maximal g occurs.

Binding of AlexaFluor488- labelled A26 Fab-Fv to activated human CD4+OX40+ T cells was measured by flow try across a S-log concentration range.

A representative binding curve for A26 Fab-Fv is shown in Figure 9A.

The mean KD value obtained on activated cells from 5 ent donors is l45pM.

A comparator g curve for A26 Fab, A26 Fab—Fv and A26 Fab-PEG is shown in Figure The graphs represents the mean of 4 or 5 ments where a different donor was used in each experiment.

PBMC were isolated by separation on a Ficoll gradient and activated with 4ug/mL PHA—L for 3 days at 37°C, 5% C02, 100% ty. Following this, CD4+ T cells were isolated by negative ion using magnetic beads (CD4+ T cell Isolation Kit 11 for Human; Miltenyi Biotec).

Approximately 1 x 105 cells were incubated in the presence of antibody in either Facs buffer (PBS/0.2% BSA/0.09% NaN3) or Facs buffer supplemented with 5% HSA, at 4°C. The final concentration of the antibody ranged from 48nM — 0.0005nM. The cells were washed in PBS prior to analysis by flow cytometry using a FACScalibur (Becton Dickinson). Titration data sets were also produced for isotype control antibodies for each A26 format to determine non specific binding. The number of moles of bound antibody was calculated by using interpolated values from a standard curve generated from beads comprised of differing but known s of fluorescent dye. Geometric mean fluorescence values were determined in the flow cytometric es of cells and beads. Non-specific binding was subtracted from the A26 Fab- Fv values and the specific binding curve thus generated analysed by non-linear regression using a one—site binding equation (Graphpad Prism®) to determine the K1).

Table 4: Mean K» values for A26 antibodies in human cell y assays . 0.145 at: 0.019 . A26, Fab-Fv (n=5) 0.096 :1: 0.017 536 (n=4) 0.230 t 0.057 0.322 i 0.089 8 A26 Fabilizégfl) 0.068 :1: 0.011 0.085 d: 0.031 Example 3: A26 Fab-Fv modulates cytokine production from PBMC exposed to Dermatophagoidespteronyssinus allergenic extract PBMC were isolated from allergic volunteers by separation on a Ficoll nt. Purified PBMC were exposed to 25 ug/mL Dermatophagoz’des pteronyssinus allergenic extract in the presence of test antibody (concentration range SOug/mL to 0.0005 ug/mL) in a final volume of ZOOnL per well in a 96-well round-bottomed plate. After 6 days tion at 37°C, 5% C02, 100% humidity, supernatants were harvested and assayed for IL-13 content by MSD. The graph in Figure 10 (a) shows representative data of 1 representative donor from 4, where the mean EC50 for inhibition of1L—13 production was 0.87nM (range from 0.6nM to 1.07nM).

Table 5: Mean ECso values for A26 Antibodies in human HDM in vitro assays EC50 values were calculatedfrom dual donor inhibition curves by non-linear regression using Graphpad Prism® software A26‘Fab-Fv (n'=4)' 0.865 $0.112 [A26 Fab’PEG (n=4)‘ 0.928 a 0.282 1.310 a 0.425 A26 Fab’ (1224) 0.335 a: 0.040 0.680 :I: 0.223 A26 Fab- Fv reduces the percentage of activated (CD255 CD4+ T cells after secondary antigen re-stimulation with Dermatophagoides pteronyssinus allergenic extract CD4+ T cells from allergic donors were ated in vitro for 7 days with 25ug/ml Dermatophagoides pteronyssinus allergenic extract (Greer) and autologous APC, in the presence of no antibody or 10ug/ml A26 Fab’PEG, A26 Fab-Fv or Ctrl Fab’ (A33 Fab’). Cells were washed and rested for 3 days and then mulated with Dermatophagoides pteronyssinus extract as previously (Figure 13). After 3 days, the cells were washed and fluorescently stained for surface CD3, CD4 and CD25. Cells were then analysed by flow cytometry on a FACS Canto flow ter (BD). Cells were gated on live lymphocytes and CD3+CD4+ expression prior to analysis. Data represents 11 = 3 donors including mean. n.s, A26 Fab—Fv compared to Ctrl Fab’ (significance ed using paired, 2 tailed T test).

Example 4: A26 Fab-Fv inhibits CD4+ and CD8+ T cell proliferation in a Hu—NSG mouse model.

Mice were dosed so with 0.03, 0.3, 3 or 30 nag/kg A26 Fab-Fv one day prior to transfer of 1 x 107 human PBMCs into the peritoneal cavity. After 14 days mice were bled by cardiac puncture under terminal aesthetic and then killed by cervical dislocation. The number of human CD4+ and CD8+ cells in the blood was then determined by FACS analysis (Figure 10 (b)). Data (n=10) is sed as means i SEM and statistical analysis is by one way ANOVA with 2O Bonferroni post test. Values represent % inhibition : SEM.

Results are shown in Figure 14.

The Hu—NSG model has demonstrated that A26 Fab—FV profoundly inhibits human T cell proliferation in vivo and supports A26 Fab-Fv as a viable eutic candidate for the tion of T cell mediated pathologies. In addition, the Fab-Fv format conferred greater efficacy at lower doses than the Fab’ PEG format. The decrease in this xeno-proliferative response of donor T cells may provide ting evidence that A26 Fab—FV could be a viable therapeutic for GVHD.

Example 5: Ligand-blocking capacity The capacity of A26 t to block the interaction between cell-surface expressed 0X40 and recombinant OX4OL was measured using a flow cytometry-based ligand ng assay.

Briefly, activated human CD4+OX40+ T cells were pre-incubated with a titration of A26 Fab- FV. Recombinant OX40L was subsequently added to the cells and allowed to bind in the presence of A26 t. The proportion of OX40L bound was then detected by flow cytometry using a labelled secondary reagent. Figurell shows an inhibition curve representing combined data from 3 independent donors and demonstrates that A26 Fab-dst is capable of completely blocking OX4OL binding. The mean ICso for inhibition of recombinant OX4OL binding was 0.44 nM (n = 3 donors).

Methods: Inhibition of OX40L binding to human activated CD4+OX40+ T cells by A26 Fab-FV PBMC were ed by separation on a Ficoll gradient and activated with 4 ug/mL PHA—L (Sigma) for 3 days at 37°C, 5% C02, 100% humidity. CD4Jr T cells were then purified from the culture by negative selection using MACS columns (Miltenyi Biotech, CD4Jr T cell ion kit ll). 2 x 105 CD4Jr T cells were incubated in the presence of A26 Fab—dst (final concentration range 10 ug/mL — 0.000056 ug/rnL (136.6 nM — 0.000765 nM)) for 30 minutes at 4°C. OX40L (biotinylated CD252 muCDS, Ancell) was added at a final concentration of 2 ug/mL and incubated for a flirther 30 s at 4°C. Cells were washed and OX40L binding detected by incubation with PE-labelled streptavadin (Jackson Immunoresearch) prior to analysis by flow cytometry using a FACS Canto (Becton Dickinson). A matched 40 binding Fab-dst was used as a l. The inhibition curve was analysed by non—linear regression (Graphpad ) to ine the IC50. An inhibition curve representing combined data from 3 independent donors is shown in Figure 11, where data points represent the mean and error bars represent SEM.

The mean EC50 for inhibition of recombinant OX40L g to 0X40 by A26 Fab-Fv 0.445 nM. In comparison, A26 Fab’PEG was slightly less potent at ligand ng (ECSO = 0.739 nM) Whereas A26 Fab’ had ally greater potency (ECso = 0.242 nM) than the Fab- FV as shown below.

Table 6: ECso values for inhibition of OX40L binding to human activated CD4+0X40Jr T cells by A26 antibodies , 0.445 a: 0.110 ., , p .

A26 335mg“. :23) 1 0.739 i 0.166 A26 Fab’ (yr—53). ' ' 0.242 :1: 0.069 EC50 values were calculatedfrom individual donor inhibition curves by non—linear regression using Graphpad Prism® software.

Example 6 Effect ofA26 Fab-Fv in onal human in vitro assays The effect of A26 Fab-Fv on OX40-OX40L dependent cellular interactions was assessed in a range of antigen-driven human cyte assays. These assays were performed in the presence of 5% human serum to ensure saturation of the albumin binding site of the Fv region, as would be predicted to occur in viva.

A26 Fab-Fv inhibits a mixed lymphocyte on The y allogeneic mixed lymphocyte reaction (MLR) is an in vitro model of alloreactive T cell activation and proliferation (Bach et al., 1964, O’Flaherty et al., 2000). Donor T cells are activated through recognition of allogeneic MHC antigens on unrelated donor stimulator PBMCs, ing in cellular eration and cytokine production (Lukacs et (21., 1993). T lymphocyte alloreaction has been shown to be driven by both the allogeneic MHC antigen and bound peptide (Sherman er (2]., 1993). The magnitude of an MLR se correlates with the degree ofMHC mis-matching between the responder—stimulator pair (Forrester er al., 2004). An MLR response results in the proliferation of cells from the ding donor and the production of both Th1 (IL-2, IFN-y and TNF-a) and Th2 (IL-4, IL—5, IL-10 and IL-13) T cell derived nes. The exact cytokine profile in an MLR is thought to be specific to the responder— stimulator pairing (Jordan et al., 2002). MLR assays have been used widely in research to study T cell activation pathways, screen immunosuppressive drugs and predict possible donor organ rejection in transplant recipients (Bromelow er al., 2001).

The effect of A26 Fab-Fv on in vitro human alloreactive T cell activation and proliferation investigated using an MLR assay essentially as described by O’Flaherty et al., 2000. PBMCs from two unrelated donors were co-cultured in the presence or absence of A26 Fab-Fv, A26 Fab’ or A26 Fab’PEG and cellular proliferation measured by 3H—thymidine incorporation. As shown in Figure 12, A26 Fab-Fv inhibited cellular proliferation in a concentration—dependent manner with an ECso value of 0.56 nM (40.9 ng/mL) and a maximal tion of 55% (n = 3 donor pairings). A26 Fab-Fv was slightly more potent than A26 Fab’PEG, which had an EC50 value of 0.88 nM, While A26 Fab’ had an ECso value of 0.25 nM as shown in Table 7:.

Human PBMCs from two unrelated donors were isolated from whole blood. Cells from donor were vated by diation to generate the stimulator population. Cells from the remaining donor formed the responder population. Stimulator and responder populations were mixed at a 1:1 ratio (1x105 cells/donor) and cultured in the presence A26 Fab’, A26 Fab—Fv or A26 Fab’PEG (0.4ng-25ug/mL) for 6 days. In-house reagent 01297.1 Fab-Fv was utilized as an isotype-matched control. Cellular proliferation was measured at day 6 by 3H- thymidine incorporation (0.5 uCi/well). Data is displayed as percentage tion relative to the responder plus stimulator response in the absence of biologic reagent, and is the combined data from three donor pairings. ECSO values were calculated using Graphpad Prism® software.

Table 7 ECso values for inhibition of human MLR proliferative response by A26 antibodies Antibody format EC50 (nM) Mean :1: S.E.M.

A26 Fab-Fv (n = 3) 0.56 :l: 0.12 A26 Fab’PEG (n = 3) 0.88 i 0.44 A26 Fab’ (n = 3) 0.25 :l: 0.06 atants from the human MLR were also analysed to investigate the effect of A26 Fab-Fv on cytokine production. As shown in Figure 12B, A26 Fab—Fv significantly inhibited production of IFN—y in the MLR by an average of 81% (n = 3 donor pairings).

Human PBMCs from two unrelated donors were isolated from whole blood. Cells from one donor were inactivated by'y-irradiation to generate the ator population. Cells from the remaining donor formed the responder population. Stimulator and responder populations were mixed at a 1:1 ratio (1x105 cells/donor) and cultured in the presence of 25 gig/ml A26 Fab’, A26 Fab-Fv or A26 Fab’PEG or controls (A33 Fab’ or CA162.01297.1) for 6 days. Supernatants were harvested and assayed for IFN-y content using an MSD assay. The percent inhibition was calculated relative to cells cultured with no antibody. The graphs ent pooled data from three donors (mean i S.E.M). ** = p<0.01; A26 Fab-FV compared with Ctrl Fab-Fv (significance measured using paired, 2-tailed T—test).

Example 7 A26 Fab—Fv binding to NK cells in a human MLR The effect of A26 Fab-Fv on NK cell division within an MLR was investigated. T lymphocyte alloreaction drives the mixed cyte response and A26 Fab—Fv profoundly inhibits T cell on and lFNy production in this system. Inhibition of NK cell division could also contribute to reduced IFNy production. Using CFSE-labelled responder cells inhibition of NK cell division was trated by FACS analysis of the dividing population (data not shown).

Two ent measures of cell division are shown. The Division Index represents the average 2012/072325 number of cell divisions that a cell in the original population has undergone and includes the undivided cells; A26 Fab-Fv reduces the Division Index indicating that fewer cells in the tion are committed to division; this effect is presumably mediated by the NK cells that are expressing 0X40. The Proliferation Index reflects proliferation of the responding population only, and the inhibitory effect of A26 Fab-Fv using this measure is vely reduced.

Example 8 Mean KD/ ECso values for A26 Fab-Fv in human in vitro assays Affinity/(n; 5) ' (onijiblocking(n,=_ 3) :0.445:l:0.110 0033 Mixed {Lymphocyte ionv - — 0.558 :l: 0121 0041 Inhibition ofProliferation;r(n [=' 3) j .

House ' DuSt > *’,'Mite , 7 4 0.865 $0.112 0.063 Inhibition ofIL-l 3 production (:i :54) ' It will of course be understood that the present invention has been described by way of example only, is in no way meant to be limiting, and that modifications of detail can be made within the scope of the claims hereinafter. Preferred features of each embodiment of the invention are as for each of the other embodiments mutatis is. All publications, including but not limited to patents and patent applications, cited in this specification are herein incorporated by reference as if each individual publication were cally and individually indicated to be incorporated by reference herein as though fully set forth.

Claims (10)

WHAT IS CLAIMED IS:

1. A bispecific antibody fiision protein which binds human 0X40 and human serum albumin comprising: a heavy chain comprising, in sequence from the N—terminal, a first heavy chain variable domain (VH1), a CHl domain and a second heavy chain variable domain (VH2), a light chain comprising, in sequence from the inal, a first light chain variable domain (VLl ), a CL domain and a second light chain variable domain (VLZ), wherein said heavy and light chains are aligned such that VH1 and VLl form a first antigen binding site and VH2 and VLZ form a second antigen binding site, 10 wherein the antigen bound by the first antigen binding site is human 0X40 and the antigen bound by the second antigen binding site is human serum albumin, wherein the first le domain of the heavy chain (VH1) comprises the sequence given in SEQ ID N021 for CDR-Hl, the sequence given in SEQ ID N02 for CDR-HZ and the sequence given in SEQ ID NO:3 for CDR-H3 and the first variable domain ofthe light chain 15 (VLl) comprises the sequence given in SEQ ID NO:4 for CDR—Ll, the sequence given in SEQ ID NO:5 for CDR-L2 and the sequence given in SEQ ID NO:6 for CDR-L3, wherein the second heavy chain variable domain (VH2) has the sequence given in SEQ ID NO:ll and the second light chain variable domain (VLZ) has the sequence given in SEQ ID NO: 12 and 2O the second heavy chain variable domain (VH2) and second light chain le domain (VL2) are linked by a disulphide bond.

2. A bispecific antibody fusion protein ing to claim 1 which antagonises the g of 0X40 to OX4OL.

3. A bispecific antibody fusion protein according to claim 1 or claim 2 in which there is a 25 peptide linker between the CH1 domain and the second heavy chain variable domain (VH2).

4. A bispecific antibody fusion protein ing to any one of claims 1 to 3 in which there is a peptide linker between the CL domain and the second light chain variable domain (VLZ).

5. An antibody fusion protein according to any one of claims 1 to 4 wherein the first heavy 30 chain variable domain (VH1) comprises the sequence given in SEQ ID N028.

6. An antibody fusion protein according to any one of claims 1 to 5 wherein the first light chain variable domain (VLl) ses the sequence given in SEQ ID N027.

7. An antibody fusion n ing to any one of claims 1 to 6, wherein the heavy chain comprises or consists of the sequence given in SEQ ID NO: 15. 35

8. An antibody fusion n according to any one of claims 1 to 7, wherein the light chain comprises or consists of the sequence given in SEQ ID NO:l6.

9. A bispecific antibody fusion protein which binds human 0X40 and human serum n, having a heavy chain comprising the sequence given in SEQ ID NO:15 and a light chain comprising the sequence given in SEQ ID NO:16. 40

10. An isolated DNA sequence encoding the heavy and/or light chain(s) of an antibody fusion protein according to any one of claims 1 to 9.