HK1218552B - Antibody binding sites specific for egfrviii - Google Patents

Abstract

This application describes binding proteins that specifically bind to EGFRvIII and multispecific binding proteins that specifically bind to EGFRvIII and CD3. Further described is a multispecific tandem diabody that binds to EGFRvIII and CD3. Further described are highly cytotoxic EGFRvIII/CD3 bispecific tandem diabody for recruiting T cells to specifically and potently kill several types of solid tumor cancer.

Description

Antibody binding site specific for EGFRvIII

The present invention relates to a binding protein that specifically binds to EGFRvIII and a multispecific binding protein that specifically binds to EGFRvIII and CD3. The present invention further extends to a multispecific tandem diabody that binds to EGFRvIII and CD3. In particular, the present invention relates to a highly cytotoxic EGFRvIII/CD3 bispecific tandem diabody that is used to recruit T cells to effectively kill several types of solid tumor cancers.

EGFR imbalance is associated with many cancers, and small molecules and EGFR-targeted antibodies have all been successfully used in clinical practice. However, due to the extensive normal tissue expression of EGFR, the antibodies approved for clinical use and those in development all have serious side effects. The deletion mutant (deletion variant) III (EGFRvIII) of EGFR with truncated extracellular domain and ensuring ligand-independent intrinsic activity is the most common mutant form associated with oncogenic transformation. EGFRvIII is specifically expressed in cancer tissue and is associated with various solid tumor types. The restricted EGFRvIII expression on cancer cells provides the opportunity for developing a cytotoxic antibody that specifically targets cancer, protects normal tissues, and fully reduces the side effects associated with EGFR treatment.

In a first aspect of the present invention, fully human, highly specific, high-affinity EGFRvIII binding proteins are described herein. The described EGFRvIII binding proteins have the following advantages:

Figure 9 shows that EGFRvIII/CD3 of the present invention is the most common EGFRvIII binding protein of the present invention.They do not promote or promote the very low EGFRvIII internalization from the cell surface of target positive molecule, and this helps to raise cytotoxic T-cell or NK-cell.The result in Fig. 9 shows, according to the present invention, after cell is exposed to EGFRvIII conjugated proteins under all test conditions, with respect to untreated cell, EGFRvIII receptor molecules still remains on the cell surface.These results show, EGFRvIII/CD3 conjugated proteins of the present invention surprisingly do not show any internalization trend (Fig. 9).The combination of EGFRvIII specific binding proteins of the present invention may suppress internalization, rather than inducing internalization, or may promote the increase that EGFRvIII expresses, thereby causes its cell surface density to increase.

Can use affinity maturation technology (affinity maturation technique) to make the avidity of these high-affinity associated proteins be further significantly improved to obtain that EGFRvIII is had specificity, K _D in 100pM scope or lower associated proteins.Surprising fact is, these associated proteins have demonstrated the abnormal specificity in conjunction with EGFRvIII, and natural EGFR (wild-type EGFR or EGFRwt) is not had cross reactivity.Due to from 269 amino acid whose frames (in-frame) of EGFR extracellular domain (exon 2-7), therefore with respect to EGFRwt, in EGFRvIII, form new epi-position (neo-epitope), and estimate that this new epi-position is quite little: it is arranged by new amino acid (novel juxtaposition of aminoacids) (amino acid 5 is fused to amino acid 274) and a new GLY amino acid composition at the deletion site.

The goal of affinity maturation screening methods is to improve binding affinity by promoting the selection of binding proteins with reduced off-rates ( _kOFF ). However, surprisingly, the affinity matured binding proteins shown have greatly increased on-rates (kON ₎ , while the reduction in _kOFF has only a small contribution to the 100-fold increased binding of the binding proteins, with some variable antibody domains showing _KD less than 100 pM.

These unique properties make the variable antibody domains described herein that are specific for EGFRvIII particularly suitable for the development of multispecific, e.g., bispecific, multivalent, immune effector cell engaging, tumor-targeting antibody therapeutics, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies.

In a further aspect of the invention, multivalent EGFRvIII binding antibodies are provided that have two binding sites for EGFRvIII, or in the case of a multivalent bispecific antibody, in addition to two binding sites for EGFRvIII, also have a binding site for the T-cell antigen CD3 or CD16A specifically expressed on natural killer (NK) cells.

In one embodiment, use EGFRvIII is had two binding sites and comprises in addition the EGFRvIII/CD3 series connection diabody that CD3 has two binding sites, and tested their combination with recombinant EGFRvIII-or EGFR-Fc fused antigen.Under non-reducing conditions (due to complete disulfide bond, molecular structure still part retains), or under reducing conditions (protein structure complete denaturation), EGFRvIII-or EGFR-Fc fused antigen separates by SDS-PAGE, is transferred to PVDF membrane by Western blotting (Western Blotting), and uses different EGFRvIII/CD3 series connection diabody antibodies (diabodyantibody) or EGFR-to assess in conjunction with modifying (decorate) trace (Fig. 4).This embodiment shows, comprises the parental EGFRvIII binding domains that EGFRvIII has two binding sites and the EGFRvIII binding domains of affinity maturation and also keeps its selection specificity to EGFRvIII in multivalent antibody form. Under reducing and non-reducing conditions, all EGFRvIII/CD3 series connection double antibodies all can identify EGFRvIII, but all can not identify full-length wild-type EGFR under these conditions arbitrarily.This embodiment also proves, and EGFRvIII specific binding structural domain described herein can identify linear epitope, and its reactivity does not need complete three-dimensional structure.This is opposite with EGFR-in conjunction with IgG antibody cetuximab (Erbitux (Erbitux)), EGFR-in conjunction with IgG antibody cetuximab can identify wild-type EGFR and ripe EGFRvIII, but its reactivity obviously needs complete disulfide bond and conformational epitope (Fig. 4).

In another embodiment, EGFRvIII/CD3 tandem diabodies comprising parental or affinity-matured EGFRvIII specific domains in combination with different CD3 binding domains and/or having single binding domains of different orders in the tandem diabody molecule are used. The tandem diabody with the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII - _{V H} ^CD3 contains a divalent EGFRvIII binding site in the middle of the tandem diabody molecule, while the tandem diabody with the domain order V _H ^EGFRvIII -V _L ^CD3 -V _H ^CD3 -V _L ^EGFRvIII has two EGFRvIII binding sites in the outside position and two CD3 binding sites in the middle. In ELISA, the concentration-dependent binding of different EGFRvIII-domains and domain order mutants of the tandem diabody to EGFRvIII antigen, wild-type EGFR antigen, and to the concentration-dependent binding of the CD3 antigen (Fig. 6) are analyzed. All tandem diabodies containing affinity-matured EGFRvIII binding domains showed a substantial increase in binding to EGFRvIII, clearly manifested as corresponding binding curves shifting by more than two orders of magnitude to lower concentrations ( FIG. 6 ). No binding of any tandem diabody to wild-type EGFR antigen was observed.

Can select and identify anti-EGFRvIII antibody by phage display library (phase display libraries), for example, selectively combine the antigen-binding proteins of the EGFR of maturation rather than native form (native form).T cell is the effective tumor-killing effector cell that can not be raised by natural antibody.Therefore, in a further aspect of the present invention, the invention provides one group of multispecific EGFRvIII/CD3 antigen-binding proteins, particularly series connection diabody, it can raise T cell, have the binding characteristic and cytotoxicity of a wide range.Characterized the feature of the binding characteristic of antigen-binding proteins of the present invention, the cytotoxic activity of T cell mediation and the T cell activation of target mediation in one group of in vitro test.In FACS, antigen-binding proteins of the present invention shows unusual specificity to the EGFRvIII antigen in Biacore, ELISA and EGFRvIII-positive cell.Within the full concentration range of test, do not observe any conjugated protein and EGFR antigen or with the detectable combination of EGFRwt expressing cell. The most effective high-affinity series diabody shows the cytotoxicity of EC50 for F98 glioma cells and Chinese hamster ovary celIs expressing EGFRvIII and CHO cells; Remaining series diabody shows the cytotoxicity of EC50 up to about 10000pM.Also tested these series diabody pairs as the cytotoxicity of the EGFRwt ⁺ cell of the more sensitive probe of the residue of native form.When the series diabody concentration of maximum assessment was 0.5 μM, cytotoxicity was observed on EGFRwt ⁺ cell or other EGFRvIII-negative cells.It is crucial to be combined with the high-affinity of CD3 to effectively T cell recruitment, yet, when there is no EGFRvIII ⁺ target cell in vitro, as lacking measurement by proliferation, the series diabody of the present invention does not cause T cell activation, and this contributes to good preclinical safety (Figure 10).In a word, the strict specificity and high efficiency of anti-tumor cell toxicity are mediated by EGFRvIII/CD3 series diabody.

Tandem diabodies with CD3 binding _KD ranging from 1.1 nM to approximately 500 nM but relatively constant EGFRvIII binding _KD (ranging from 2.0 nM to 6.7 nM) were generated and showed good correlation between CD3 binding strength and cytotoxic potency against cells expressing EGFRvIII, with _EC50 values ranging from 25 pM to approximately 10,000 pM (Figure 11).

Although the affinity of the antigen-binding protein of the present invention for EGFRvIII has been significantly improved, the specificity is not lost.

is a trademark of Affimed Therapeutics used to represent tandem diabodies (Kipriyanov et al., 1999, J. Mol. Biol, 293: 41-56; Cochlovius et al., 2000, Cancer Res., 60: 4336-4341; Reusch et al., 2004, Int. J. Cancer, 112: 509-518, Kipriyanov, 2009, Methods Mol Biol, 562: 177-93; McAleese and Eser, 2012, Future Oncol. 8: 687-95). In the context of the present invention, TandAb and tandem diabodies are used as synonyms.

The sequence of wild-type EGFR gene and protein is known. EGFRvIII is the result of an in-frame deletion of exons 2-7 (801bp) of the wild-type EGFR gene, resulting in 267 amino acid deletions in the receptor's outer domain and a new glycine residue at the junction of exons 1 and 8. This new amino acid juxtaposition in the EGFR extracellular domain generates tumor-specific and immunogenic epitopes.

According to first aspect of the present invention, described EGFRvIII had to specific associated proteins at least, it does not cross-react with wild-type EGFR.This associated proteins preferably comprises CDR1, CDR2, CDR3 of the antibody variable heavy chain that is selected from the CDR described in the following table 1A and CDR1, CDR2 and the CDR3 of antibody variable light chain.

Table 1A: Sequence identifiers of amino acid sequences of human light and heavy chain CDRs from binding proteins that specifically bind to EGFRvIII

Therefore, an embodiment of the present invention is an EGFRvIII binding protein having at least one EGFRvIII binding site, comprising an antibody variable heavy chain domain and/or an antibody variable light chain domain; wherein the antibody variable heavy chain domain comprises a CDR1 selected from SEQ ID NO: 27, 33, 42, 46, 49, 53, a CDR2 selected from SEQ ID NO: 28, 38, 43, 50, 54, 61, 63 and a CDR3 of SEQ ID NO: 29; the antibody variable light chain domain comprises a CDR1 selected from SEQ ID NO: 30, 34, 36, 39, 44, 47, 51, 56, 59, 64, a CDR2 selected from SEQ ID NO: 31, 40, 57 and a CDR3 selected from SEQ ID NO: 32, 35, 37, 41, 45, 48, 52, 55, 58, 60, 62 and 65.

According to further embodiment, described associated proteins comprises and has the antibody variable heavy chain domain that is selected from SEQ ID NO:1,3,5,7,9,11,13,15,17,19,21 and 25 as shown in Table 1B and/or is selected from SEQ ID NO:2,4,6,8,10,12,14,16,18,20,22 and 26 as shown in Table 1B at least one EGFRvIII binding site of antibody variable light chain domain.The aminoacid sequence of three variable light chain CDR and three variable heavy chain CDR is shown in bold and is underlined in Table 1.These structural sites show that the affinity to EGFRvIII is improved, and specificity does not have loss.These antigen-binding proteins do not cross-react with wild-type EGFR antigen.

In some embodiments, the present invention relates to an antigen-binding protein of the present invention.These antigen-binding proteins of the present invention do not show or only show minimized EGFRvIII internalization when in conjunction with EGFRvIII.According to the present invention, under the test conditions according to embodiment 9, after cell is exposed to EGFRvIII-in conjunction with antibody, be greater than 80%, preferably greater than 90%, most preferably be greater than 95% EGFRvIII receptor molecules and still remain on cell surface.The protein-binding combination of EGFRvIII of the present invention may suppress internalization, rather than inducing internalization, or may promote that the expression of EGFRvIII increases, thereby causes its cell surface density to increase.

The term "binding protein" refers to an immunoglobulin derivative with antigen-binding properties, i.e., an immunoglobulin polypeptide or fragment thereof containing an antigen-binding site. The binding protein comprises the variable domains of an antibody or fragment thereof. Each antigen-binding domain is formed by an antibody, i.e., an immunoglobulin, a variable heavy chain domain (VH), and an antibody variable light chain domain (VL), which bind to the same epitope. The variable light chain domain or variable heavy chain domain of the present invention is a polypeptide comprising CDR1, CDR2, and CDR3. Preferably, the binding protein of the present invention lacks immunoglobulin constant domains. The term "binding protein" also refers to antibody fragments or derivatives, including Fab, Fab', F(ab') ₂ , Fv fragments, single-chain Fv, diabodies, tandem diabody flexibodies (WO03/025018), and tandem single-chain Fv ((scFv) ₂ ).

Table 1B: Amino acid sequences of all anti-EGFRvIII variable heavy chain domains (VH) and variable light chain domains (VL) (amino acid sequences of CDR1, CDR2, and CDR3 are shown in bold and underlined)

In a preferred embodiment, the binding protein that confers specificity for EGFRvIII comprises an antigen binding site from one of the following combinations of variable heavy chain domains and variable light chain domains shown in Table 1:

(i) SEQ ID NO: 1 and SEQ ID NO: 2,

(ii) SEQ ID NO: 3 and SEQ ID NO: 4,

(iii) SEQ ID NO: 5 and SEQ ID NO: 6,

(iv) SEQ ID NO: 7 and SEQ ID NO: 8,

(v) SEQ ID NO: 9 and SEQ ID NO: 10,

(vi) SEQ ID NO: 11 and SEQ ID NO: 12,

(vii) SEQ ID NO: 13 and SEQ ID NO: 14,

(viii) SEQ ID NO: 15 and SEQ ID NO: 16,

(ix) SEQ ID NO: 17 and SEQ ID NO: 18,

(x) SEQ ID NO: 19 and SEQ ID NO: 20,

(xi) SEQ ID NO: 21 and SEQ ID NO: 22, or

(xii) SEQ ID NO:25 and SEQ ID NO:26.

The present invention also provides not only EGFRvIII has specificity, and also has the associated proteins of at least one other functional domain.In a further embodiment, described at least one other functional domain is effector domain (effector domain)." effector domain " is that effector cell is had specific binding site, and it can stimulate or trigger cytotoxicity, engulfment, antigen presentation, cytokine release.Such effector cell is, for example but not limited to, T-cell or NK-cell.Particularly, described other effector domains comprise at least one antibody variable heavy chain domain and at least one variable light chain domain that form the antigen binding site of CD3 preferred mankind CD3.

In some embodiments, the EGFRvIII conjugated proteins of the present invention can be multispecific.Term used herein " multispecific " means to refer to that conjugated proteins of the present invention have at least two antigen binding sites to different epi-positions, and wherein at least one is used for EGFRvIII.For example, conjugated proteins can be trispecific, and have binding site and have at least one binding site to effector cell such as epi-position on CD3 or antigen to two different epi-positions and/or antigen on tumor cell.Conjugated proteins can have binding site to different epi-positions of same antigen and/or the epi-position of different antigens.Such multispecific conjugated proteins comprise single-chain diabody, (scFv) ₂ , series diabody and elastomeric antibody (referring to Le Gallet al., 1999FEBS Letts 453:164-168 and WO 03/025018).In a specific embodiment of the present invention, conjugated proteins are to EGFRvIII and CD3 bispecific.

The CD3 binding site of the multispecific binding protein of the invention can be composed of a variable heavy chain domain (SEQ ID NO: 23) and a variable light chain domain (SEQ ID NO: 24) as shown in Table 2, or a close homolog of such sequences that differs from the CDR sequences by only two amino acids, or any other fully human, humanized or non-human variable antibody domain specific for CD3 (referred to as CD3), such as, for example, a variable antibody domain specific for CD3, in particular CD3 epsilon, or any other fully human, humanized or non-human variable antibody domain, such as, for example, the variable antibody domain of UCHT1 or OKT3.

Table 2: Amino acid sequences of fully human anti-CD3 variable heavy chain domain (VH) and variable light chain domain (VL) (amino acid sequences of CDR1, CDR2 and CDR3 are shown in bold and underlined)

In addition, EGFRvIII conjugated proteins of the present invention can be multivalent.Term used herein " multivalent " means to refer to, and conjugated proteins of the present invention comprises at least two antigen binding sites.Antigen binding site can have identical or different specificity.In embodiments of the present invention, conjugated proteins is in conjunction with identical epi-position with at least two binding sites in a divalent manner.The protein-conjugated example of divalence is diabody, and the protein-conjugated example of at least tetravalence is a series connection diabody.

In a further aspect of the present invention, described EGFRvIII associated proteins and bispecific EGFRvIII and CD3 binding site are humanized or fully human, i.e. people-source. In a further aspect of the present invention, EGFRvIII associated proteins or multispecific EGFRvIII and CD3 associated proteins are dimers, i.e. comprise two polypeptides with EGFRvIII and CD3 antigen binding sites.

According to the present invention, dimerization bispecific EGFRvIII and CD3 binding proteins are provided in the form of series connection diabody.Such series connection diabody is by connecting four antibody variable binding domains (heavy chain variable domain (VH) and light chain variable domain (VL) in single gene construct (referring to McAleese and Eser, 2012, Future Oncol.8:687-95)) that can homodimerize and build.In such series connection diabody, joint (linker) length is that it prevents the intermolecular pairing of variable domain so that molecule can not fold back to form single-chain diabody on itself, but is forced to pair with the complementary domain (complementary domain) of another chain.These domains can also be arranged like this, namely make corresponding VH and VL domain pair during dimerization.After being expressed by single gene construction, two same polypeptide chains head to tail folding form the functional non-covalent homodimer (McAleese and Eser, 2012, Future Oncol.8:687-95) of about 110kDa. Despite the absence of intramolecular covalent bonds, homodimers, once formed, are highly stable, remain intact and do not revert to monomeric form.

In one embodiment of a multispecific tandem diabody, a humanized bispecific tetravalent antibody is provided that has two binding sites each for CD3 and EGFRvIII (EGFRvIII/CD3RECRUIT-) to utilize the cytotoxic capacity of T cells to treat glioblastoma (GB), prostate cancer, head and neck cancer, and other epidermal growth factor receptor mutant vIII-positive (EGFRvIII+) cancers.

In the present invention, the tandem diabody is the most widely used antibody of the present invention.Relative to traditional monoclonal antibody, tandem diabody has a large amount of properties that advantage is provided, also has less multispecific molecule.Tandem diabody is fully functional when there is no glycosylation.Tandem diabody can only contain antibody variable domains, and therefore can not have any side effect that can be relevant to Fc part.Because bispecific tandem diabody allows divalence to be bound to each in EGFRvIII and CD3, therefore, its avidity (avidity) is identical with the avidity of IgG.The size of tandem diabody is about 110kDa, is less than IgG, and this can enhance tumor infiltration.But, compared with the less bispecific form based on antibody-binding domain or non-antibody scaffold, this size is much higher than the kidney threshold of first-pass clearance (first-pass clearance), and this provides pharmacokinetic advantage. The generation and preparation of such tandem diabodies is described, for example, in Kipriyanov SM: Methods Mol. Biol. 2009; 562: 177-93, Kipriyanov SM: Methods Mol. Biol. 2003; 207: 323-33 or McAleese and Eser, 2012, Future Oncol. 8: 687-95.

Tandem diabodies are well expressed in CHO cells, enabling efficient upstream and downstream manufacturing processes to be put in place.

The EGFRvIII and CD3 bispecific tandem double antibody of the present invention is designed to specifically target EGFRvIII ⁺ tumor cells by recruiting cytotoxic T cells. Antibodies cannot directly recruit cytotoxic T cells. On the contrary, by employing (engage) the CD3 molecules specifically expressed on these cells, the tandem double antibody can make cytotoxic T cells and EGFRvIII ⁺ tumor cells cross-linked in a highly specific manner, thereby significantly increasing the cytotoxic effect of these molecules. The tandem double antibody shows strong, specific and effective cytotoxicity. Significant evidence is that T cells can play a role in controlling tumor growth. For example, the presence of cytotoxic T cells in colorectal tumors and lymph nodes from NHL patients is shown to be correlated with better clinical results. In addition, for melanoma vaccines, and antibodies for CTLA-4, a negative regulator of T-cell activation, have demonstrated the potential for the therapy of designing induction of T-cell responses. The tandem double antibody of the present invention employs cytotoxic T cells by binding surface-expressed CD3, preferably CD3ε, which forms a part of the T-cell receptor. The simultaneous binding of this tandem diabody to CD3 and EGFRvIII expressed on the surface of specific tumor cells leads to T-cell activation and mediates subsequent tumor cell lysis.

"Dimer" refers to a complex of two polypeptides. Preferably, the two polypeptides are non-covalently associated with each other, particularly under conditions where there is no covalent bond between the two polypeptides. Preferably, the bispecific dimer is a homodimer, i.e., comprises two identical polypeptides. However, a trispecific or other multispecific dimer can be a heterodimer and comprise two different polypeptides, for example, at least one binding site of the variable light chain domain is located on one polypeptide and the variable heavy chain domain is located on the other domain. The term "polypeptide" refers to a polymer of amino acid residues connected by amide bonds. Preferably, the polypeptide is an unbranched single-chain fusion protein. In this polypeptide, the variable antibody domains are connected in sequence. In addition to the variable domain N-terminus and/or C-terminus, the polypeptide may also have contiguous amino acid residues. For example, such contiguous amino acid residues may include a Tag sequence that may be beneficial to polypeptide purification, preferably at the C-terminus.

Each polypeptide of the bispecific tandem diabody contains at least four variable domains, namely the variable light chain (VL) and variable heavy chain (VH) of the CD3 binding protein and the variable light chain (VL) and variable heavy chain (VH) of the EGFRvIII binding protein. The four binding domains are connected by peptide linkers L1, L2 and L3 and can be arranged as follows from N-terminus to C-terminus:

(i)VL(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3); or

(ii) VH(CD3)-L1-VL(EGFRvIII)-L2-VH(EGFRvIII)-L3-VL(CD3); or

(iii) VL(EGFRvIII)-L1-VH(CD3)-L2-VL(CD3)-L3-VH(EGFRvIII); or

(iv) VH(EGFRvIII)-L1-VL(CD3)-L2-VH(CD3)-L3-VL(EGFRvIII).

In an embodiment of the present invention, the four variable domains are arranged as VL ^{(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3). Tandem diabodies having the domain order VLCD3-VHEGFRvIII -VLEGFRvIII-VHCD3} _{and containing} ^different _EGFRvIII ^binding _sequences , despite having different affinities for EGFRvIII, all showed very similar binding to the second specific CD3, as shown by ELISA on the CD3γε antigen ( FIG5A ). In other tandem diabody domain sequences containing EGFRvIII binding domains in the outer positions and two CD3 binding sites in the middle ( _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ), a significant improvement in EGFRvIII binding to the affinity-matured EGFRvIII binding domain was also evident, while binding of the tandem diabody middle domain to CD3 was slightly weaker and more variable between individual tandem diabodies than the first domain sequence (Figure 5B). Similar observations were made when tandem diabodies containing different EGFRvIII-binding domains were constructed using different CD3 domains (SEQ ID NOs: 23 and 24) (Figure 5C).

The length of the linker affects the flexibility of the antigen-binding tandem diabody. For example, Todorovska et al., 2001 Journal of Immunological Methods 248:47-66; Perisic et al., 1994 Structure 2:11217-1226; Le Gall et al., 2004 Protein Engineering 17:357-366 and WO94/13804 describe the effect of linker length on the formation of dimeric antigen-binding proteins.

According to the present invention, the length of the peptide linkers L1, L2, and L3 is preferably such that the domains of one polypeptide can intermolecularly bind to the domains of another polypeptide to form a dimeric antigen-binding tandem diabody. Such linkers are "short," i.e., comprised of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, or 12 amino acid residues. Thus, the linker consists of approximately 12 or fewer amino acid residues. In the case of 0 amino acid residues, the linker is a peptide bond. Such short linkers facilitate intermolecular dimerization of the two polypeptides by binding and forming a correct antigen-binding site between the antibody variable light chain domain and the antibody variable heavy chain domain of the different polypeptides. Short linkers of approximately 12 or fewer amino acid residues generally prevent intermolecular interaction between adjacent domains of the same polypeptide chain. In one embodiment of the present invention, these linkers consist of approximately 3 to approximately 10, e.g., 4, 5, or 6, consecutive amino acid residues.

Regarding the amino acid composition of these linkers, peptides are selected that do not interfere with the dimerization of the two polypeptides. For example, linkers containing glycine and serine residues generally provide protease resistance. The amino acid sequence of the linker can be optimized, for example, by phage display methods to increase antigen binding and the yield of antigen-binding polypeptide dimers. Examples of peptide linkers suitable for the tandem diabodies of the present invention are GGSGGS (SEQ ID NO: 66), GGSG (SEQ ID NO: 67), or GGSGG (SEQ ID NO: 68).

EGFRvIII associated proteins as herein described and multi-specific series connection diabody can produce by the polynucleotide of expressing coding series connection diabody polypeptide, and the polypeptide of described series connection diabody is combined to form antigen-in conjunction with series connection diabody with another identical polypeptide.Therefore, the further embodiment of the present invention is the polynucleotide of coding antigen as herein described-in conjunction with series connection diabody polypeptide, for example DNA or RNA.

Polynucleotide can be constructed by methods well known to those skilled in the art, for example, by combining the gene of at least four antibody variable domains separated by the sequence encoding peptide linkers or directly connected by peptide bonds into a single gene construct, and in bacteria or other suitable expression systems such as for example expressed in Chinese hamster ovary celI, the single gene construct is operably connected with a suitable promotor, alternatively a protein tag for detection and purification, and a suitable transcription terminator. According to employed vector systems and hosts, any number of suitable transcription and translation elements comprising formation and inducible promoters can be used. Select promoter so that it drives the expression of polynucleotides in respective host cells.

As a further embodiment of the present invention, the polynucleotide can be inserted into a vector, preferably an expression vector. The recombinant vector can be constructed according to methods known to those skilled in the art.

Various expression vector/host systems can be used to contain or express polynucleotides encoding polypeptides of the present invention. Examples of expression vectors for expression in E. coli are pSKK (Le Gall et al., J Immunol Methods. (2004) 285(1): 111-27) or pcDNA5 (Invitrogen) for expression in mammalian cells.

Thus, the antigen-binding tandem diabodies described herein can be prepared by introducing a vector encoding the polypeptides described above into host cells and culturing the cells under conditions whereby the polypeptide chains are expressed, can be isolated from the culture medium, and optionally thereafter purified.

In a further embodiment, the present invention provides a pharmaceutical composition comprising an EGFRvIII binding protein, an antigen-binding tandem diabody, a vector comprising a polynucleotide encoding the polypeptide of the antigen-binding tandem diabody or a host cell transformed by the vector, and at least one pharmaceutically acceptable carrier. The term "pharmaceutically acceptable carrier" means any carrier that does not interfere with the effectiveness of the biological activity of the component and is non-toxic to the administered patient. Suitable pharmaceutical carrier examples are known in the art and include physiological saline, phosphate buffered saline, water, emulsions such as oil/water emulsions, various types of wetting agents, sterile solutions, etc. Such carriers can be prepared by conventional methods and administered to an individual at a suitable dosage. Preferably, the composition is sterile. These compositions may also include adjuvants, such as preservatives, emulsifiers, and dispersants. The effects of microorganisms can be prevented by including various antibacterial and antifungal agents.

The skilled artisan will readily construct and obtain antigen-binding proteins such as the tandem diabodies described herein without undue burden by using well-established techniques and standard methods known in the art, see, for example, Sambrook, Molecular Cloning A Laboratory Manual, Cold Spring Harbor Laboratory (1989) N.Y.; The Protein Protocols Handbook, John M. Walker, ed., Humana Press Inc. (2002); or Antibody engineering: methods and protocols/Benny K.C. Lo, ed.; Benny K.C. II Series: Methods in molecular biology (Totowa, N.J.).

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 illustrates the position of variable sequence in the EGFRvIII specific high-affinity binding domain derived after affinity maturation.The CDR position that is different from the heavy chain and light chain variable region of parental sequence is marked with asterisk, and amino acid is highlighted with grey separately.The position with useful sudden change is highlighted with dark grey.

Fig. 2 shows that among Biacore T200, multi-circulation kinetics measures the result that different scFv antibodies are attached to recombinant EGFRvIII-Fc antigen or wild-type EGFR-Fc antigen.Sensogram is presented at the kinetics of surface plasmon resonance (SPR) response unit (RU) measured in time under 4 different scFv antibody concentration levels (86nM, 17.2nM, 3.4nM, 0.69nM).Measurement is in conjunction with 180 seconds, after this measurement dissociates.Use 1:1 to combine model analysis data.

Figure 3 shows the results of flow cytometry measurement of scFv antibody binding to (A) F98 rat glioma cells overexpressing EGFRvIII (F98 ^EGFRvIII ), (B) F98 cells overexpressing native EGFR (F98 ^EGFR ), and (C) untransfected F98 cells (F98), as well as the results of similar scFv antibody binding to (D) CHO cells overexpressing EGFRvIII (CHO ^EGFRvIII ), (E) CHO cells overexpressing native EGFR (CHO ^EGFRvIII ), and (F) untransfected CHO cells (CHO).

Fig. 4 illustrates the tandem diabody containing different EGFRvIII binding domains and is separated by SDA-PAGE and is transferred to the EGFRvIII-or natural EGFR-Fc fusion antigen (non-reducing protein or the albumen with DTT reduction) of the purification of PDVF film by western blotting.Fc fusion rotein is by disulfide bond at Fc part dimerization, but is monomer after reducing with DTT.EGFRvIII-specific antibody is bonded to EGFRvIII-fusion rotein with reduction or non-reduction conformation, but is not combined with natural EGFR.This anti-EGFR antibody cetuximab is as contrast, and identifies non-reducing EGFRvIII-Fc and non-reducing EGFR-Fc antigen, but does not identify the antigen of DTT-processing.

Figure 5 shows the concentration-dependent binding of different EGFRvIII/CD3-specific tandem diabodies to EGFRvIII-Fc, EGFR-Fc, and CD3γε recombinant antigens analyzed in ELISA. (A) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , (B) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , and (C) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies containing the CD3 domains of SEQ ID NOs: 23 and 24 and having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 . The improvement in binding of tandem diabodies containing affinity-matured EGFRvIII-binding domains to EGFRvIII was clearly detectable as a shift of the respective binding curves to lower concentrations in all formats/domain orders.

Figure 6 shows the results of multi-cycle kinetic measurements of EGFRvIII/CD3-specific tandem diabodies containing different bivalent EGFRvIII binding domains in the order of _V ^CD3 - _V ^EGFRvIII - _V ^EGFRvIII - _V ^CD3 bound to recombinant EGFRvIII-Fc antigen in a Biacore T200. Sensogram displays the kinetics of surface plasmon resonance (SPR) response units (RU) measured over time at different tandem diabody concentration levels. Binding was measured for 180 seconds, followed by dissociation. Data were analyzed using a 1:1 binding model.

Fig. 7 shows the immunohistochemistry (IHC) staining result of tissue section from (A) three individual head and neck (H＆N) cancer patients, (B) three individual glioblastoma patients, and (C) representative results from prostate cancer, mammary cancer (Her2neg) and non-small cell lung cancer (NSCLC).The EGFRvIII-specific double antibody containing EGFRvIII-specific binding domain Li3G30 of the divalent conformation of 2 kinds of different concentrations is dyed in section, and described divalent conformation is similar to the arrangement of EGFRvIII targeting tandem double antibody antibody (for example, (V _L ^CD3- ) V _H ^EGFRvIII - _{V L} ^EGFRvIII (-V _H ^CD3 )) but lacks CD3 binding part.As a control, section is dyed with the EGFR-specific IgG antibody cetuximab of identification EGFR (being present on healthy tissue and cancerous tissue) and EGFRvIII.The specificity of dyeing is confirmed by the control section hatched in the secondary reagent that does not contain EGFR-or EGFRvIII-specific primary antibody.

Fig. 8 illustrates that on EGFRvIII positive cells, compared with the tandem diabody of EGFRvIII and CD3 respectively containing the parental EGFRvIII binding domain (Li3G30) before affinity maturation, the cytotoxicity ability, effectiveness and target-dependent specificity of EGFRvIII/CD3 tandem diabody containing different affinity maturation EGFRvIII-specific binding domains: (A) ABC 470 containing EGFRvIII/CD3 tandem diabody, (B) ABC 471 containing EGFRvIII/CD3 tandem diabody, (C) ABC 472 containing EGFRvIII/CD3 tandem diabody. Using CHO cells expressing EGFRvIII (CHO ^EGFRvIII ), expressing EGFRvIII negative EGFR (CHO ^EGFR ) or untransfected CHO cells as target cells and using human PBMC as effector cells, in the analysis based on flow cytometry, antibody concentration-dependent cytotoxicity was measured. In each panel (AC), tandem diabodies containing a single EGFRvIII binding domain and CD3 domain v64, in the order _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , are shown on the left, and tandem diabodies containing a single EGFRvIII binding domain combined with a different CD3 domain, in the order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , are shown on the right.

Figure 9 shows the results of evaluating EGFRvIII receptor internalization from the cell surface when combined with EGFRvIII-specific antibodies. Cells expressing EGFRvIII were incubated for 24 hours at 37°C with varying concentrations of three different ^EGFRvIII / ^CD3 tandem diabody antibodies with the domain order V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII - _{V H} ^CD3 (i) or _{V H} EGFRvIII-V L ^CD3 -V H CD3-V _L ^EGFRvIII (iv), and/or EGFRvIII/CD3 tandem diabody antibodies containing different CD3-binding domains. After adjustment, all remaining EGFRvIII receptor molecules on the cell surface were stained with the respective tandem diabody antibodies at a saturating concentration of 10 μg/ml.

Figure 10 is presented in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) series double antibody antibody, assesses the result of cell proliferation (BrdU incoporation) in human PBMC culture.PBMC is used as positive control, and it breeds in the presence of CD3 binding antibody OKT3 or phytohemagglutinin (plant lectin phytohemagglutinin) (PHA).In the absence of EGFRvIII-positive target cells, with different EGFRvIII/CD3 series double antibodies, hatch PBMC and do not induce activation and the proliferation of PBMC.

Figure 11 illustrates that the CD3-binding affinity of EGFRvIII/CD3 tandem diabodies is positively correlated with their cytotoxic capacity. Regarding CD3-binding on CD3 ⁺ -Jurkat cells, EGFRvIII-binding on CHO ^EGFRvIII cells and cytotoxic capacity, all 20 EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) in the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII -V _H ^CD3 and different CD3-binding domain combinations with different affinities for CD3 were analyzed. The scope of CD3-binding K _D value is 1nM to approximately 500nM. For each analyzed tandem diabodies, relative to CD3-binding K _D value, EGFRvIII-binding K _D value and cytotoxic EC ₅₀ value are mapped. While the TandAbs showed only small changes in EGFRvIII-binding _KD values, the cytotoxicity _EC50s showed an almost linear increase as the CD3-binding _KD values of the tandem diabodies increased.

Figure 12 shows the in vivo proof-of-concept results from the study. Two EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII-binding domain (Li3G30) but different CD3-binding domains or domain sequences were analyzed for dose-dependent inhibition of subcutaneous F98 ^EGFRvIII xenograft tumor growth and compared with the effect of cetuximab administered at a much higher concentration. (A) Dose-dependent tumor growth inhibition caused by the ^EGFRvIII /CD3 ^v6 tandem diabody with the domain sequence _VL ^CD3 - _VH EGFRvIII- _VL ^EGFRvIII - _VH ^CD3 ; (B) Dose-dependent tumor growth inhibition caused by EGFRvIII/CD3 tandem diabodies containing different high-affinity CD3-binding domains and the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII .

The present invention is further illustrated by the following examples, but the present invention is not limited thereto.

Example 1: Discovery and affinity maturation of EGFRvIII-specific binding proteins:

Li3G30 Discovery:

The phage display library of people's scFv sequence based on IgM source is carried out two to three rounds of panning (panning) and EGFRvIII is had specific binding substances (binders) with enrichment.Before every round of panning, described library is pre-incubated with the binding substances of (deplete) wild-type form EGFR or the binding substances of fusion rotein Fc part with EGFR-Fc.On the solid phase of EGFRvIII coating and in solution, select (selection) in parallel.After two rounds and three rounds of panning, pick out monoclonal, induce soluble scFv to express and screen the extract that is combined with EGFRvIII-Fc and EGFR-Fc.

According to the present invention, the fully human high specific variable antibody domains combined with EGFRvIII have been described, and EGFRvIII is the deletion mutant of EGFR, and it is specifically expressed in tumor rather than on healthy tissue with the form of mutation.Full human EGFRvIII specific domains described herein were initially found in the phage display screening of the fully human scFv library using based on IgM source.Surprisingly, although two or three rounds of panning were carried out using exhaustion step (depletion procedure) to wild-type EGFR antigen, this should be enriched to have specific binding substances to EGFRvIII, but in ELISA, most of the scFvs of gained are all combined to two kinds of albumen (EGFRvIII and wild-type EGFR).Identified a high EGFRvIII specific sequence, called after Li3G30.Except that there is little difference in the N-terminal of VH and VL, the framework region of heavy chain and light chain and CDR1 and CDR2 are identical with sequence VH5-51 and VL3-25 of germline encoding respectively.Li3G30 was identified from the third round of liquid phase panning.

Affinity Maturation:

The library of antibody framework pair VH5-51/VL3-25 was designed using an algorithm ordered from DistributedBio. The specificity determining residues of Li3G30 were identified and included in the library design. The design includes 52 randomized CDR sites, each with a separately defined amino acid distribution. Three different loop lengths were built into the VL CDR3. Gene fragments encoding the randomized sites of VH and VL were ordered from Geneart/Lifetechnologies and synthesized using TRIM-technology. The fragments were cloned into the phage display vector pEXHAM (Schwarz et al. 2004) to a final library size of 3.7E+8 transformed E. coli cells. The library was packaged into phage particles and, through panning and screening steps, separated with improved affinity and retained variants specific for EGFRvIII. Panning was performed using the EGFRvIII-Fc antigen immobilized on a protein-bound plastic surface. The elutriation step is designed to the scFv that helps to have the EGFRvIII bonded with low dissociation rate ( _kOFF ) by implementing following steps: washing step comprises several washing buffer incubation steps that continue up to 30 minutes, to help select the scFv that slowly dissociates.Another step is based on the overnight competition with solubility EGFRvIII.In order to guarantee in the process of affinity maturation, can not select EGFR-EGFRvIII cross-reaction antibody or the antibody in conjunction with Fc, before first elutriation step, by the pre-incubation of wild-type EGFR-Fc, the library is exhausted possible wild-type EGFR binding substances.And owing to the existence of solubility EGFR in the process of the elutriation step being carried out to fixing EGFRvIII, screening scheme is unfavorable for cross-reaction binding substances.

The unique EGFRvIII-specific scFv was further tested for binding to transfected CHO and F98 cells expressing EGFRvIII using flow cytometry. ScFvs that showed binding to cells expressing wild-type EGFR and/or untransfected cells were excluded from further analysis. Initial _KD ranking using Biacore X100 was used to identify scFvs with improved affinity for EGFRvIII compared to the original clone, Li3G30. The best variants were characterized using detailed SPR measurements and fluorescence-based stability assays.

For VL CDR3, the initial library for affinity maturation comprises three kinds of different loop lengths, and known all lengths are compatible with VH5-51/VL3-25 framework.Surprisingly, in the selection process, the middle loop length is obviously preferred.Therefore, compared with the anti-EGFRvIII antibody of parent generation, the VL CDR3 of the variant of all affinity maturations has all shortened one amino acid (table 1, Fig. 1).The sequence of selected scFv is compared with the input library, and this analysis has shown 3 useful sudden changes in VL CDR3.The heavy chain of the anti-EGFRvIII antibody of parent generation and the CDR1 and CDR2 of light chain are identical with the sequence encoded separately of germline sequence VH5-51 and VL3-25.In 75% of the sites, the residue of each germline coding is reconstructed (reconstituted) or obviously preferred.Shown a useful sudden change in VH CDR1, one in VL CDR1 (Fig. 1).The amino acid distribution in the remaining randomization site is very close to the distribution in the library before selecting.

Example 2: Measurement of scFV binding to EGFRvIII and wild-type EGFR in Biacore:

Preparation of anti-hu Fc IgG CM5-chip:

Anti-hu Fc IgG CM5-chip was prepared by covalent amine coupling using an amine coupling kit (Amine-Coupling-Kit) (GE) according to the manufacturer's instructions. IgG was diluted to a concentration of 25 μg/ml in the provided immobilization buffer (10 mM sodium acetate, pH 5.0). For the coupling step, the predefined method "Amine" of the Biacore T200 control software was applied. A target level of 5000-10000 RU was achieved in all 4 flow cells. Before injecting the recombinant protein solution, the surface of the chip was activated with EDC/NHS. The remaining binding sites on the surface were blocked with ethanolamine. 1×HBS-P+ was selected as the running buffer for the entire coupling step.

Concentration range of single-chain Fv’s (scFv) used in SPR measurements:

Before SPR measurement, the ScFv concentration was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate the molar concentration of the His-tagged scFv antibody, a molecular weight of approximately 28 kD was assumed. The scFv was adjusted to the following final concentrations by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 430 nM, 86 nM, 17.2 nM, 3.44 nM, 0.688 nM, and 0 nM.

ScFv measurement conditions:

EGFRvIII-Fc and wild-type EGFR-Fc fusion proteins were diluted in 1× HBS-P+ to a concentration of 6.25 nM. Both antigens were injected at a flow rate of 10 μl/min: EGFRvIII-Fc for 75 seconds and wild-type EGFR-Fc for 90 seconds. EGFRvIII-Fc was injected into flow cell 2, and wild-type EGFR-Fc was injected into flow cell 4. Flow cells 1 and 3 were left blank.

1× HBS-P+ was used as running buffer throughout the procedure. The scFv antibody was injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. The signal measured in flow cell 3, where no antigen was captured, was subtracted from the signal in flow cell 4 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0M _MgCl2 (30 seconds, 10 μL/min). Before each measurement, four "start-up cycles" were performed without scFv. ScFv solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

The kinetics of the scFv antibodies were determined in the MCK (Multi Cycle Kinetic) measurement. To estimate the apparent binding affinity, the binding curves were evaluated using the "Kinetic" method included in the Biacore evaluation software. Using this method, all binding curves were fitted using the "Langmuir 1:1 interaction model" with the software. The results are shown in Table 3.

The affinity maturation screening step that is used is intended to improve the EGFRvIII binding affinity of EGFRvIII specific binding domain more easily by making the selection of the binding substances with the dissociation rate ( _kOFF ) of reduction.Yet, surprisingly, the affinity maturation binding substances (table 3) of gained shows the association rate ( _kON ) that significantly increases, and the reduction of _kOFF reaches 100 times of combination for the raising of scFv and only makes less contribution, and some binding substances demonstrate the _KD lower than 100pM simultaneously (table 3).Especially surprisingly, although the screening step is conducive to the scFv that selects the _kOFF reduction, the reduction of realization _KD is mainly due to the raising of _kON.This also is surprising, because the affinity maturation of antibody is relevant (Schier et al., 1996, Pini et al., 1998, Boder et al., 2000) with the reduction of _kOFF conventionally.

Table 3: Summary of scFv binding affinities ("VH" is the variable heavy chain domain of the scFv, "VL" is the variable light chain domain of the scFv). _K values, association rates (kON) and dissociation rates ( _kOFF ) of different EGFRvIII binding domains in scFv format for recombinant EGFRvIII-Fc antigen measured in a Biacore ( _SPR ) 1:1 binding model; "enhancement factor" refers to the fold change relative to the parental scFv, i.e., Li3G30 (VH SEQ ID NO: 25 and VL SEQ ID NO: 26).

Eight of the 11 affinity-matured scFvs showed at least a 5-fold improvement in _kON compared to the parental scFv; three of these had a 20-fold higher _kON . Compared to the improvement in _kON , the effect of affinity maturation on _kOFF was quite small. Only three affinity-matured scFvs showed a 4-5-fold improvement in _kOFF . Interestingly, two of these were isolated from a competitive selection method, which may be a more effective method for selecting scFvs with reduced _kOFF than prolonged incubation in wash buffer.

The average melting temperature of the selected affinity matured candidates determined by differential scanning fluorimetry (DSF) was 62.2°C, ranging from 53°C to 67.2°C.

Maintaining high specificity for EGFRvIII relative to wild-type EGFR during affinity maturation

Affinity maturation is usually accompanied by small changes in binding epitopes, which can affect specificity/cross-reactivity (Barbas et al, 1994, Parsons et al., 1996, Winkler et al., 2000). Due to the in-frame deletion of 269 amino acids in the wild-type EGFR extracellular domain (exon 2-7), an EGFRvIII new epitope relative to wild-type EGFR is formed, and it is predicted to be quite small: it is composed of new amino acids in parallel (amino acid 5 is fused to amino acid 274) and a new GLY amino acid at the deletion site. Even if it is expected that small changes in binding epitopes will quickly destroy the abnormal specificity of EGFRvIII in this case, a cross-reactive conjugate that can also identify native EGFR can be obtained.

Surprisingly, the affinity maturation described herein resulted in up to 100-fold improved binding of anti-EGFRvIII antibodies without loss of exclusive specificity for EGFRvIII (Table 3, Figure 2).

During the selection of affinity-matured binders, the binding of the only EGFRvIII-specific scFv to CHO cells expressing EGFRvIII (CHO ^EGFRvIII ) and F98 cells (F98 ^EGFRvIII ), as well as to CHO cells expressing human wild-type EGFR (CHO ^EGFR ) and F98 cells (F98 ^EGFR ), and to untransfected CHO and F98 cells were tested in flow cytometry.

Characterized in more detail 11 high-specificity anti-EGFRvIII scFv that do not have any measurable background combination with wild-type EGFR positive cell (CHO ^EGFR ) or the Chinese hamster ovary celI of non-transfected.In the best case, measured the avidity to 80pM of EGFRvIII antigen, corresponding to 100 times of raising coefficient (table 3) compared with parental scFv.Simultaneously, the anti-EGFRvIII scFv that is detected all does not demonstrate any cross reactivity (Fig. 2) with wild-type EGFR antigen.

To our knowledge, such high affinity for EGFRvIII has never been observed for a human antibody domain that binds exclusively specifically to EGFRvIII, without measurable cross-reactivity with wild-type EGFR. Safdari and colleagues recently described results for humanizing the murine EGFRvIII-binding antibody domain MR1 (Safdari et al., 2014), which was first described in its murine form by Lorimer and colleagues (Lorimer et al., 1996, Beers et al., 2000, Kuan et al., 2000). In their humanization approach, they achieved an increase in the affinity of humMR1, reaching similar picomolar _K values reported here. However, in contrast to the present invention, they did not completely eliminate the cross-reactivity of the antibody domain with wild-type EGFR, as can be clearly seen in Figure 4 of their publication (Safdari et al., 2014). Thus, after SDS PAGE and Western blotting, humMR1 was not only specific for EGFRvIII (145 kD), but also recognized EGFR (170 kD) (Safdari et al., 2014). Applicants extensively tested the specificity and cross-reactivity of the EGFRvIII binding domain of the present application before and after affinity maturation and surprisingly found no cross-reactivity signals with wild-type EGFR or purified recombinant wild-type EGFR antigen in Biacore (Figure 2), SDS PAGE and Western blotting (Figure 4), or ELISA (Figure 5). The EGFRvIII binding domain of the present application also did not exhibit any cross-reactivity with wild-type EGFR overexpressed on the cell surface of transfected CHO cells (CHO ^EGF ) or F98 glioma cells (F98 ^EGFR ) (Figure 3).

Example 3: High specificity and high affinity of EGFRvIII-specific scFv antibodies were assessed by flow cytometry for binding to F98 rat glioma cells (F98 ^EGFRvIII ) or stably transfected CHO cells (CHO ^EGFRvIII ) expressing EGFRvIII; no binding to F98 cells (F98 ^EGFR ) or untransfected F98 cells (F98) expressing native EGFR, and no binding to CHO cells (CHO ^EGFR ) or untransfected CHO cells (CHO) expressing native EGFR.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), Ficoll Paque PLUS (GE Healthcare), FITC-conjugated goat anti-human IgG (Dianova), FITC-conjugated goat anti-mouse IgG, min X (Dianova), L-glutamine (Invitrogen), _NaN3 (Carl Roth), PBS (Invitrogen), Penicillin/Streptomycin (Invitrogen), Propidium iodide (Sigma), RPMI-1640 (Invitrogen).

Cells and cell lines:

F98 rat glioma cells, F98 cells overexpressing EGFR (F98 ^EGFR ), and F98 cells overexpressing EGFRvIII (F98 ^EGFRvIII ) were purchased from the American type culture collection (ATCC) and cultured according to the recommended protocol.

F98 (ATCC CRL-2397)

F98 ^EGFR (ATCC CRL-2948)

F98 ^npEGFRvIII (ATCC CRL-2949)

Stably transfected CHO cells expressing recombinant EGFR or EGFRvIII were generated at Affimed according to the following protocol:

The gene of coding wild-type EGFR or deletion mutant EGFRvIII is synthesized by Life Technologies/GeneArt (Regensburg, Germany), and is subcloned among the mammalian expression vector pcDNA5/FRT.Expressing the stable Chinese hamster ovary celI of recombinant wild-type EGFR or EGFRvIII generates based on the host cell line Flp-In CHO (Life Technologies) of previous adaptation to suspension growth.The day before transfection, Flp-In Chinese hamster ovary celI is carried out passage culture (not having bleomycin (Zeocin)) in the HyClone CDM4CHO that is supplemented with L-glutamine, HT supplement and penicillin/streptomycin.The stable expression cell line is by using polyethyleneimine (PEI) transfection reagent, with the product expression based on pcDNA5/FRT and integration plasmid and Flp recombinase expression plasmid pOG44 (Life Technologies) cotransfection Flp-In Chinese hamster ovary celI system generates. 2.5 μg of total DNA was diluted in 50 μL of OptiMEMI medium and combined with 7.5 μg of PEI diluted in 50 μL of OptiMEMI medium. The mixture was incubated for 10 minutes and then added to 2×10 ⁶ Flp-In CHO cells suspended in 1 mL of CHO-S-SFMII medium. One day after transfection, the cells were diluted to a density of 1×10 ⁵ viable cells/mL in CHO-S-SFMII medium supplemented with 500 μg/mL hygromycin B and seeded in T75 culture flasks. Flp recombinase mediates the insertion of the Flp-In expression construct into the genome at the integrated FRT site through site-specific DNA recombination. During the selection phase, the viable cell density was measured once or twice a week, the cells were centrifuged and resuspended in fresh selection medium at a maximum density of 1×10 ⁵ viable cells/mL. After about 2-3 weeks of selection using 500 μg/mL hygromycin B, stable cells expressing EGFR or EGFRvIII were recovered and then transferred to HyClone CDM4 CHO suspension culture medium. Stable cells expressing wild-type EGFR or EGFRvIII were cryopreserved in a culture medium containing 50% ProFreeze (Lonza) and 7.5% DMSO.

To determine the density of wild-type EGFR or EGFRvIII cell surface antigens on stably transfected cell lines, a flow cytometric indirect immunofluorescence assay (QIFIKIT, Dako) was used according to the manufacturer's instructions and demonstrated that the density of wild-type EGFR on CHO ^EGFR stably transfected cells was approximately equal to the density of EGFRvIII on CHO ^EGFRvIII stably transfected cells, whereas no wild-type EGFR or EGFRvIII binding sites were detected on untransfected CHO cells.

Determination of Antibody Binding and Affinity by Flow Cytometry:

Cells and indicated scFv antibodies are started from 100 μ g/mL (about 3300nM) in 100 μ L serial dilutions in FACS buffer and are hatched on ice for 45min (minute). After washing 3 times with FACS buffer, cells are hatched on ice for 45min with 0.1mL 10 μ g/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After second washing cycle, under the same conditions as before, cells are hatched with the goat anti-mouse IgG antibody that 0.1mL 15 μ g/mL FITC is put together. As a control, in the absence of scFv, cells are hatched with anti-His IgG 13/45/31, then with the goat anti-mouse IgG antibody that FITC is put together. Then cells are washed again and are resuspended in the 0.2mL FACS buffer containing 2 μ g/mL propidium iodide to exclude dead cells. The fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer with MXP software (Beckman-Coulter, Krefeld, Germany) or a Millipore Guava EasyCyte flow cytometer with Incyte software (Merck Millipore, Schwalbach, Germany). The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany).

If analyzing with a Beckman-Coulter FC500 MPL flow cytometer, use 0.5 × 10 ⁶ cells/stain; if analyzing with a Millipore Guava EasyCyte flow cytometer, use only 0.25 × 10 ⁶ cells/stain.

KD values were calculated using the equation for single-site binding (hyperbola) using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) after subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone.

The results are shown in Figure 3.

Example 4: SDS PAGE and Western blotting analysis of the binding specificity of antibodies binding to EGFRvIII.

To detect the binding specificity of EGFRvIII/CD3 tandem diabodies containing different binding domains to EGFRvIII antigen and to demonstrate the absence of binding to wild-type EGFR antigen, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis were performed.

By recombinant DNA technology, the sequence of the extracellular domain of encoding wild-type EGFR or truncated EGFRvIII is fused to the Fc part of human IgG antibody.DNA construct is transfected into the Chinese hamster ovary celI as the wild-type EGFR-Fc or EGFRvIII-Fc of the reorganization of soluble protein and is secreted into the cell culture supernatant, uses protein A chromatography to purify the wild-type EGFR-Fc or EGFRvIII-Fc of described reorganization from described cell culture supernatant.Through the formation of intramolecular disulfide bond in the Fc part, wild-type EGFR-Fc (or EGFRvIII-Fc) fused antigen forms the dimer with two identical chains.Due to EGFRvIII-specific 267 amino acid deletions, EGFRvIII-Fc causes the Fc fusion protein of dimer form to have the size difference of about 50kD than EGFR-Fc little by about 25kD.

Equal amounts of purified wild-type EGFR-Fc (and EGFRvIII-Fc) were mixed with either non-reducing 2× SDS PAGE sample buffer or reducing 2× SDS PAGE sample buffer containing dithiothreitol (DTT) as a reducing agent. Samples containing DTT were heated at 95°C for 5 minutes and then loaded onto a 4-20% standard TGX precast SDS PAGE gel (Biorad). 1 μg of protein sample was used per lane. To separate the proteins in the gel, SDS-PAGE was run at 300V in 1× Tris/glycine/SDS buffer (Biorad) for approximately 25 minutes. Total protein was visualized in the gel using a standard stain-free molecular imaging system (Biorad). A Page Ruler unstained protein ladder (Thermo Scientific) was used as a molecular weight marker. The gel was then blotted onto a PVDF membrane using BioRad's semi-dry blotting Fastblot system. The membrane was blocked in 3% (w/v) nonfat dry milk in 1× TBS for 30 min at room temperature. The membrane was cut into small pieces, each containing a similar protein sample to be blotted. Different tandem diabodies were diluted to a concentration of 2 μg/ml in 3% (w/v) nonfat dry milk in 1× TBS and incubated with separate membrane pieces on a rocking platform for 1 hour. The anti-EGFR antibody cetuximab was used as a control at a concentration of 10 μg/ml. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS, then incubated with HRP-conjugated secondary detection antibodies PentaHIS-HRP (QIAGEN) (for detection of the tandem diabodies) or Protein L-HRP (Thermo Scientific) (for detection of cetuximab) at a dilution of 1:5000 in 3% nonfat dry milk in TBS on a rocking platform for 1 hour. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS. HRP-mediated color development on the membrane was initiated by adding 0.06% DAB + 0.02% CoCl ₂ + 0.015% H ₂ O ₂ in TBS. The reaction was terminated by adding water. The membrane was dried and scanned.

The results are shown in Figure 4.

Example 5: The binding of EGFRvIII/CD3 tandem diabodies containing different EGFRvIII binding domains to EGFRvIII antigen and CD3, as well as their binding specificity to EGFRvIII in the absence of binding to wild-type EGFR antigen, were analyzed by enzyme-linked immunosorbent assay (ELISA).

ELISA steps:

96-well ELISA plates (Immuno MaxiSorp; Nunc) were coated overnight at 4°C with EGFRvIII-Fc, wild-type EGFR-Fc, or CD3γε recombinant antigen in 100 mM carbonate-bicarbonate buffer. To achieve approximately equal antigen molar coating densities, EGFRvIII-Fc was coated at a concentration of 4 μg/ml, EGFR-Fc at a concentration of 6 μg/ml, and CD3γε at a concentration of 1.5 μg/ml. After a blocking step using 3% (w/v) skim milk powder (Merck) in PBS, 11 serial dilutions of different EGFRvIH/CD3 tandem diabodies ranging from 200 ng/μl to 6.5× ^10-6 ng/μl in PBS containing 0.3% (w/v) skim milk powder (Merck) were incubated on the plates at 25°C for 1.5 h. After incubation, the plates were washed three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20 (Serva). 50 ng/ml of Protein L-HRP was added and incubated on the plates at 25°C for 1 hour. After washing three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20, tetramethylbenzidine (TMB) substrate (Seramun) was added for detection. The reaction was terminated after approximately 2 minutes by adding 100 μl/well of 0.5 _{M H₂SO₄} . The absorbance of the wells was measured at 450 nm using a multiwell plate reader (Victor, Perkin Elmer).

Absorbance values were plotted as lines in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada).

The results are shown in FIG5 .

Example 6: Measurement of Binding of Tandem Diabodies to EGFRvIII by SPR:

Preparation of anti-hu Fc IgG CM5-chip

An anti-hu Fc IgG CM5-chip was prepared as described above (Example 2).

Concentration range of tandem diabodies used in SPR measurements:

Before SPR measurement, the concentration of the tandem diabody was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate molar concentrations, a molecular weight of 105 kD was assumed for all tandem diabodies. The tandem diabodies were adjusted to the final concentrations mentioned below by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 150 nM, 30 nM, 6 nM, 1.2 nM, 0.24 nM, 0.048 nM, and 0 nM; or 90 nM, 30 nM, 10 nM, 3.33 nM, 1.11 nM, 0.37 nM, 0.123 nM, 0.041 nM, and 0 nM, or the concentrations indicated.

Binding assay conditions for tandem diabody KD measurements

Dilute the EGFRvIII-Fc fusion protein in 1×HBS-P+ to a concentration of 6.25 nM. For EGFRvIII-Fc, inject the antigen solution at a flow rate of 10 μl/min for 40 seconds. Inject the antigen into flow cell 2. Flow cell 1 serves as the reference cell.

1× HBS-P+ was used as running buffer throughout the procedure. Serial dilutions of the diabody were injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0 M _MgCl2 (3 x 15 seconds, 10 μL/min). Before each measurement, three "priming cycles" were performed without the tandem diabody. The tandem diabody solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

Binding parameters of the tandem diabodies were determined in an MCK (Multi Cycle Kinetic) measurement. To estimate apparent binding affinity, binding curves were evaluated using the "Kinetic" method included in the Biacore X100 evaluation software. Using this method, binding curves were fitted in whole or in part using the "Langmuir 1:1 interaction model" with the software.

Binding parameters of tandem diabodies

To analyze the effect of combining two EGFRvIII binding domains into bivalent or multivalent or multispecific molecules such as EGFRvIII/CD3 tandem diabodies, the apparent affinity of EGFRvIII/CD3 tandem diabodies to EGFRvIII antigen was measured by SPR using BiacoreX100 ( FIG6 ).

For all molecules detected, the apparent affinity of the EGFRvIII specific binding domain in the EGFRvIII/tandem diabody improves, and the best combination tandem diabody has achieved a _K of 11pM. Relative to the corresponding tandem diabody containing the parental EGFRvIII binding domain, the _K of the tandem diabody containing the domain of affinity maturation reduces by up to 25 times. Relative to the _K of the measurement of the monovalent binding scFv, the improvement of _K is also up to 25 times. Interestingly, contrary to the improvement achieved by the affinity maturation driven mainly by the improved k _ON , the further improvement of binding affinity in the bivalent tandem diabody format is realized by slower k _ofOFF to a great extent (Table 4).

Therefore, the tandem diabody format provides a way to further enhance binding affinity while maintaining the specificity of the antibody binding domain.

Table 4: Summary of measured KD, association rates (kON), and dissociation rates ^{( kOFF} ) for binding of different EGFRvIII/ ^CD3 tandem diabodies containing bivalent parental or affinity matured ^EGFRvIII -specific binding domains with the domain order _{VLCD3 - VHEGFRvIII} - _VLEGFRVIII - _VHCD3 to recombinant EGFRvIII-Fc antigen in the SPR _{1:1 binding} model.

Example 7: Immunohistochemistry (IHC) staining of solid tumor tissue sections using anti-EGFRvIII bispecific diabody or cetuximab

In contrast to wild-type EGFR, which is ubiquitously expressed on healthy tissues, expression of the mutant receptor EGFRvIII is highly tumor-specific. However, the high frequency of EGFRvIII expression in glioma cells is consistent with literature descriptions, but the extent of EGFRvIII expression and its homogeneity in other tumors remain controversial.

A small immunohistochemistry (IHC) study was initiated, analyzing the reactivity of tissue sections from three cancer patients each suffering from glioblastoma (GB), head and neck (H&N) cancer, prostate cancer, Her2-negative breast cancer, and non-small cell lung cancer (NSCLC) with the EGFRvIII specific binding domains in the form of a bivalent diabody according to the present invention ( FIG. 7 ).

The present invention relates to the immunohistochemical study of the expression of EGFRvIII and the tumor specificity of EGFRvIII binding antibodies to assess different tumors.To contain the bivalent double antibody form of the EGFRvIII binding domain Li3G30 arranged with V _H ^EGFRvIII - _{V L} ^EGFRvIII , the EGFRvIII combination of EGFRvIII specific variable domains has been detected.Secondary detection is implemented by the His tag on the double antibody protein.In contrast, used natural EGFR to have specific IgG antibody cetuximab (Erbitux (Erbitux)).

All people's tissue sections are purchased from BioChain Institute, Inc., and according to supplier's specification sheets process section.Make tissue section be adapted to room temperature and hatch with DAKO peroxidase blocker, then use 10% goat serum to seal.EGFRvIII specific double antibody was hatched 1 hour with two concentration levels of 0.5 μ g/ml and 0.1 μ g/ml, then hatched and passed through Envision+HRP-DAB system detection mouse antibody (DAKO) with anti-His antibody (Dianova).According to manufacturers' specification sheets, use Klear people HRP-polymer DAB detection system (GBI Labs), implement the tissue section dyeing that carries out with control item Erbitux (Merck KGaA) with the concentration of 10g/ml.

Finally, sections were stained with hematoxylin and mounted with mountant medium (Shandon ConsulMount ^™ , Thermo Scientific).

The results are shown in FIG7 .

Two demonstrate in three glioblastoma samples with the obvious and specific reactivity of EGFRvIII specific double antibody.Surprisingly, all three head and neck cancer samples all demonstrate strong and obvious EGFRvIII positive, and similarly for prostate cancer, mammary cancer and NSCLC, obtained the specific staining (Fig. 7) of tumor tissue with EGFRvIII specific antibody described herein.

Based on these properties, the EGFRvIII binding domains described herein, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies, are excellently suited for developing multispecific, multivalent, immune effector cell engaging, tumor targeting antibody therapies.

Example 8: Evaluation of cytotoxic activity mediated by EGFRvIII/CD3 tandem diabodies in a FACS-based cytotoxicity assay

In the cytotoxicity assay based on cell, analyzed EGFRvIII/CD3 series connection diabody antibody, it contains the parental generation that is combined with different CD3 binding domains or the EGFRvIII specificity domain of affinity maturation and/or each binding domain (Fig. 8) with different orders in described series connection diabody molecule.By using the Chinese hamster ovary celI (CHO ^EGFRvIII ) of expressing EGFRvIII, the Chinese hamster ovary celI (CHO ^EGFR ) of expressing wild-type EGFR and not transfected Chinese hamster ovary celI as target cell to analyze the specificity of target-mediated, series connection diabody and the lethality of T cell mediation.

Materials (media, buffers, and reagents):

DMSO (Sigma), EasySep ^™ Human T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD4+ T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD8+ T Cell Enrichment Kit (Stem Cell Technologies), The Big Easy EasySep ^™ Magnet (Stem Cell Technologies), FCS (Invitrogen), Lymphoprep (Stem Cell), L-glutamine (Invitrogen), mlgGl FITC (ADG), CD16-FITC (MEM-154) (Thermo Fisher Scientific), mlgGl-FITC/mlgGl-PE/mlgGl-ECD (Beckman Coulter), mlgGl-PE (Beckman Coulter), mlgGl-PC5 (Beckman Coulter), mlgGl-PC7 (Beckman Coulter), CD8-FITC/CD4-PE/CD3-ECD (Beckman Coulter), CD16-PC5 (Beckman Coulter), CD19-PC7 (Beckman Coulter), CD16-FITC/CD56-PE/CD3-ECD (Beckman Coulter), CD14-PC7 (Beckman Coulter), CD33-PE (MACS Miltenyi Biotech), NaN ₃ (Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), PKH67 green fluorescent cell labeling (Linker) Midi kit (Sigma), Gammanorm human IgG (Octapharma), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

Isolation of PBMCs from buffy coats and enrichment of T cells:

On the day of blood draw, buy buffy coat from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany); Buffy coat preparation is stored at room temperature overnight, and then PBMC is separated by density gradient centrifugation. Buffy coat sample is diluted with two to three volumes of PBS, layered on Lymphoprep cushion and centrifuged at 800 × g for 25 min at room temperature w/o brake. PBMC at the interface is collected and washed 3 times with PBS, and then they are used for enrichment of T cells or flow cytometric analysis. According to the manufacturer's instructions, EasySep ^™ human T cell enrichment kit (for immunomagnetic separation of untouched human T cells) and Big Easy EasySep ^™ magnet are used to enrich T cells from the PBMC group.

Characterization of effector cells by flow cytometry:

To assess the distribution of PBMC subpopulations and the purity of enriched PBMC or T cell subsets, PBMC staining was performed.

For flow cytometric analysis, isolated PBMCs or enriched PBMC subpopulations were resuspended in PBS supplemented with 2% heat-inactivated FCS and 0.1% sodium azide (referred to as FACS buffer) and 1 mg/mL polyclonal human IgG to a density of 0.5-2 × 10 ⁶ /mL. 0.5 ml aliquots of the cell suspension were then incubated with the antibody for 15 minutes in the dark according to the manufacturer's recommended protocol.

Table 5: Pipetting scheme for characterization of PBMCs and fractionated PBMCs by flow cytometry

After incubation, samples were measured on a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) in the order 3, 4, 5, 6, 1, 2 using a protocol developed for PBMC and multicolor staining (FITC PEECD PC5 PC7; FITC PE ECD; FITC PE PC5PC7) in MXP acquisition software (Beckman-Coulter). For analysis and data plotting, CXP analysis software (Beckman-Coulter) was used.

Cytotoxicity assay:

As previously described, T cells used as effector cells were characterized by flow cytometry. The target cells used in the cytotoxicity assay were CHO cells stably transfected to overexpress EGFRvIII (CHO ^EGFRv111 ), CHO cells stably transfected to express wild-type EGFR (CHO ^EGFR ), or untransfected CHO cells (CHO). Cell lines were generated as previously described and cultured under standard conditions as previously described. For the cytotoxicity assay, target cells were collected, washed twice with RPMI 1640 medium without FCS, and resuspended in diluent C provided by the PKH67 green fluorescent cell labeling Midi kit to a density of 2×10 ⁷ /mL. The cell suspension was then mixed with an equal volume of double-concentrated PKH67-labeling solution (e.g., 1 μL PKH in 250 μL diluent C) according to the manufacturer's instructions and incubated at room temperature for 2-5 minutes with regular mixing. The staining reaction was terminated by adding an equal volume of FCS and incubating for 1 minute. After washing the labeled target cells with complete RPMI medium, the cells were counted and resuspended in complete RPMI medium to a density of 2×10 ⁵ /mL.

2×10 ⁴ target cells were then seeded with T cells at a 5:1 E:T ratio in each well of a round-bottom 96-well microplate along with the indicated antibodies in a total volume of 200 μL/well. Nine serial 1:5 dilutions, starting at 30 μg/mL, were typically tested. Natural cell death in the absence of antibody and target killing by effectors were determined in at least three replicates on each plate. Tandem diabody-mediated killing was typically determined in duplicate.

After centrifugation at 200 g for 2 minutes at room temperature, the assay plate was incubated at 37° C. in a humidified atmosphere containing 5% CO ₂ for 40-48 hours. After incubation, the culture was washed once with FACS buffer and then resuspended in 150 μL FACS buffer supplemented with 2 μg/mL PI. The absolute number of viable target cells, characterized by positive green PKH67 staining but negative PI staining, was measured using a Millipore Guava EasyCyte flow cytometer (Merck Millipore).

Based on the remaining live target cells measured, the percentage of specific cell lysis was calculated according to the following formula: [1-(number of live target cells _(sample) )/(number of live target cells _(native) )]×100%. Sigmoidal dose response curves and EC50 values were calculated by nonlinear regression/4-parameter logistic fit using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, California, USA).

Data Analysis:

Lysis values obtained for a given antibody concentration were determined in duplicate and analyzed by Sigmoidal dose-response/4-parameter logistic fit analysis using Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) and used to calculate the mean EC50 and SD of the replicates for percentage lysis.

The results are shown in FIG8 and Table 6.

The tandem diabody of all detections all demonstrates the concentration dependency to the cell of expressing EGFRvIII, high specificity, cytotoxicity and high efficiency, and the survival of EGFRvIII negative cells is not impaired.Within the whole concentration range of any tandem diabody of expressing EGFRwt-cell or non-transfected cell detection, do not observe detectable cytotoxicity.Therefore, the cytotoxicity of EGFRvIII/CD3 tandem diabody mediation is strictly target-dependent.In cytotoxicity assay, the EC 50 of the tandem diabody that contains parental EGFRvIII binding domain before affinity maturation and domain order is V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 is _25pM (Fig. 8).Corresponding tandem diabody contains the EGFRvIII binding domain that avidity improves with identical domain order (V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 ), shows enhanced cytotoxicity efficacy equally, and in the best case, reaches the EC ₅₀ (table 6) of 1.5pM. The relative increase in cytotoxic activity is more evident in the tandem diabodies with EGFRvIII specific domains located in the external position (domain order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ) (Table 6). In these tandem diabodies, up to 70-fold increased cytotoxic potency can be achieved by replacing the low-affinity EGFRvIII binding domain Li3G30 with an affinity-enhanced domain in these tandem diabodies (Figure 8, Table 6).

Table 6: to the summary of the EC50 value measured with different EGFRvIII/CD3 series connection diabodies in the cytotoxicity assay, described different EGFRvIII/CD3 series connection diabodies contain the EGFRvIII specific binding domain of affinity maturation with two different domain orders.Shown the raising coefficient relative to the series connection diabodies that contain the parental generation EGFRvIII binding domain of correspondence.

Choi et al. (Proc Natl Acad Sci USA. 2013 Jan 2; 1 10 (1): 270-5; WO2013/7185010) describes a t EGFRvIII/CD3 comparator, an EGFRvIII/CD3 bispecific T-cell engager (BiTE), which has an EGFRvIII binding site (based on the mouse MR1 domain) and a CD3 binding site (based on the anti-CD3 antibody OKT3). In a cell-based cytotoxicity assay, the antibody achieved 50% of the maximum specific lysis achieved at a concentration of about ^10-2 μg/ml, which corresponds to a molar EC50 concentration of about 200 pM for the 52kD BiTE (Figure 4C of Choi et al., 2013). Comparisons show that the cytotoxicity of the tandem diabodies described herein containing the parental, rather than affinity-matured, EGFRvIII binding domains is even about 10-fold higher than that of the EGFRvIII/CD3 BiTE. Furthermore, the tandem diabody containing the affinity-matured EGFRvIII binding domain of the present invention even exhibited 100-fold higher cytotoxicity than that of EGFRvIII/CD3 BiTE.

Example 9: EGFRvIII receptor internalization induced by EGFRvIII/CD3 tandem diabody on CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII )

To evaluate EGFRvIII regulation by EGFRvIII/CD3 tandem diabodies, CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII ) were seeded in aliquots of 2.5×10 ⁵ cells per well of a round-bottom 96-well microplate and incubated with increasing concentrations of EGFRvIII/CD3 tandem diabodies in RPMI 1640 medium supplemented with 10% FCS, 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, and 100 μg/mL sodium pyruvate (referred to herein as complete RPMI medium; all components purchased from Invitrogen) at 37° C., 5% CO ₂ , in a humidified incubator for 24 hours. Aliquots of cells were cultured in the absence of antibody to serve as control samples for detection of the highest EGFRvIII cell surface levels and as negative controls for staining performed with the secondary reagent alone.

44.After hatching, cell is washed 2 times with ice-cold phosphate buffered saline (PBS, Invitrogen) (being called FACS buffer in this article) that is supplemented with 2% heat inactivation FCS (Invitrogen) and 0.1% sodium azide (Roth, Karlsruhe, Germany), then with the antibody (this antibody is identical with the antibody that hatches with cell during adjustment incubation) of saturation concentration (10 μ g/mL) dyeing to assess the maximum antibody combination corresponding to the remaining EGFRvIII antigen on the cell surface.The aliquots of the cell that is cultivated under the situation that there is no antibody are dyed with the same antibody of saturation concentration, to measure the maximum antibody binding capacity corresponding to the maximum measurable EGFRvIII level on the cell surface.After washing 3 times with FACS buffer, with series connection double antibody treatment cell and dye with the anti-His mAb (Dia910, Dianova) of 10 μ g/mL, then with the goat anti-mouse IgG (Dianova) dyeing that puts together with 15 μ g/mL FITC. After repeated washing with FACS buffer, cells were resuspended in FACS buffer containing 2 μg/mL propidium iodide (Sigma) to exclude dead cells. Approximately 5 × 10 ³ propidium iodide-negative cells from each sample were analyzed using MXP software and an FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany). The mean fluorescence intensity (MFI) of the samples was determined using CXP software (Beckman-Coulter) or cells were analyzed using a Millipore Guava EasyCyte flow cytometer using Incyte software (Merck Millipore, Schwalbach, Germany).

After subtracting the signal obtained from cell staining with the secondary reagent alone, the mean fluorescence intensity of the measured samples was plotted as a line in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada), and the relative amount of EGFRvIII remaining on the cell surface was calculated using the fluorescence signal obtained by staining untreated cell samples at 37°C in the absence of the EGFRvIII/CD3 tandem diabody, which was set to represent 100% of cell surface EGFRvIII.

The results are shown in FIG9 .

Some bibliographic reports have described and compared deletion mutant EGFRvIII and demonstrated impaired internalization (Fenstermaker et al.2000, Han et al.2006, Grandal et al.2007, Gan et al.2009) with natural EGFR.However, about EGFRvIII specific antibody or for several disclosed reports of the antibody (for example cetuximab) also in conjunction with the wild-type EGFR of EGFRvIII, described in conjunction with rapid internalization and the degraded (Reist et al.1995, Foulon et al.2000, Kuan et al.2000, Patel et al.2007, Jutten et al.2009, Dreier et al.2012) of EGFRvIII in the cell that expresses EGFRvIII after antibody separately.

For internalization determination, use the transfection CHO ^EGFRvIII cell of overexpressing EGFRvIII.The cells of these expression EGFRvIII and the EGFRvIII/CD3 series connection double antibody of varying concentrations were hatched 24 hours at 37 ℃, described the condition in the context of other EGFRvIII binding antibodies to induce strong internalization.After this adjustment (during which antibody-receptor complex can internalize or not internalize), give all remaining EGFRvIII receptor molecules dyeing (Fig. 9) on the cell surface with the saturation concentration of 10 μ g/ml with each antibody.The result that Fig. 9 presents shows, after cell is exposed to EGFRvIII-in conjunction with series connection double antibody, under all detection conditions, the EGFRvIII receptor molecules more than 80% and up to 140% existing on untreated cell are still available (140% or more>100% general value refers to and reduces (cellular internalization) with the number of receptor on the cell surface on the contrary, and the cell processed with series connection double antibody demonstrates the increase of receptor on the cell surface.Internalization to 100% comes from the contrast with untreated cell). These results show, EGFRvIII/CD3 series connection diabody of the present invention surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may even suppress internalization and cause EGFRvIII receptor levels on the cell surface to increase.The result that Fig. 9 presents shows, with respect to untreated cell, under all detection conditions, after cell is exposed to EGFRvIII-in conjunction with antibody, at least more than 80% still remain on the cell surface.These results show, EGFRvIII/CD3 antibody of the present invention, especially EGFRvIII/CD3 series connection diabody surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may suppress internalization or promote EGFRvIII to express and increase, cause its cell surface density to increase.

Example 10: Evaluation of cell proliferation in PBMC culture in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies

To evaluate whether EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies induce T cell activation and subsequent proliferation in the absence of EGFRvIII ⁺ target cells, human PBMCs were cultured in the presence of increasing concentrations of EGFRvIII/CD3 tandem diabodies. After 5 days of incubation, proliferation was assessed in a BrdU incorporation assay.

Materials (culture media, buffers, and reagents)

FCS (Invitrogen), human serum (Sigma), Lymphoprep (Stemcell Technologies), L-glutamine (Invitrogen), OKT3 (Biolegend), PBS (Invitrogen), β-mercaptoethanol (Invitrogen), penicillin/streptomycin (Invitrogen), RPMI-1640 (Invitrogen), sodium pyruvate (Invitrogen), BrdU cell proliferation ELISA (Roche).

cell:

Buffy coats were purchased from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany) on the day of blood draw; buffy coat preparations were stored at room temperature overnight before PBMC isolation.

Isolation of PBMCs from buffy coats and enrichment of T cells:

PBMCs were isolated from the buffy coat by density gradient centrifugation. The buffy coat sample was diluted with two to three volumes of PBS, layered onto a Lymphoprep cushion, and centrifuged at 800 × g for 25 minutes at room temperature with or without braking. PBMCs at the interface were collected and washed three times with PBS before use in proliferation assays.

Proliferation assay

All assay plates were blocked for 2 hours at room temperature with RPMI1640 medium supplemented with 10% heat-inactivated human serum, 4 mM L-glutamine, 100 U/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, 1 mM sodium pyruvate, and 0.05 mM β-mercaptoethanol (referred to herein as complete RPMI medium) to prevent nonspecific binding of antibodies to plastic surfaces. After removing the blocking medium, 4 × 10 ⁵ PBMCs were seeded in each well of a flat-bottom 96-well microplate with the indicated concentrations of test and control items in complete RPMI medium in a total volume of 200 μL/well. Each sample was assayed in triplicate. IgG anti-CD3 clone OKT3 was used as a positive control. To assess spontaneous proliferation, cells were cultured in the absence of antibody in six replicates. The plates were then incubated at 37°C and 5% CO ₂ in a humidified atmosphere for 4 days. 18 hours before termination of incubation, the cell cultures were pulsed with 100 μM BrdU. BrdU incorporation was quantified using a BrdU proliferation ELISA kit according to the manufacturer's instructions. Color development was stopped by adding 100 mM _{H₂SO₄ ,} and absorbance was measured at 450 nm using a multilabel plate reader (Victor 3, Perkin Elmer). Absorbance values were analyzed and the mean and SD were plotted using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA).

The results are shown in FIG10 .

Example 11: Correlation between CD3 Binding Affinity and Cytotoxicity of EGFRvIII/CD3 Tandem Diabody

Constructed and expressed EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) and different CD3 binding domains (which have some affinity for CD3 T cell antigens) and measured binding to CD3 ⁺ Jurkat cells and EGFRvIII ⁺ CHO ^EGFRvIII cells; In addition, cytotoxicity assays were performed using CHO ^EGFRvIII cells as target cells and PBMC cells as effector cells. The order of each VH and VL (heavy and light chain) domain of all tested tandem diabodies tested in this example was the same: _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 .

Antibody binding to EGFRvIII ⁺ CHO ^EGFRvIII cells and cytotoxic activity mediated by the EGFRvIII/CD3 tandem diabody were evaluated by flow cytometry as described in previous examples.

After staining CD3 ⁺ Jurkat cells with increasing concentrations of the tandem diabody, the apparent affinity of the EGFRvIII/CD3 tandem diabody for CD3 ⁺ cells was determined by flow cytometry. KD values were calculated by nonlinear regression/hyperbola using the mean fluorescence values for each tandem diabody concentration determined by flow cytometry.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), FITC-conjugated goat anti-mouse IgGmin X (Dianova), L-glutamine (Invitrogen), NaN ₃ (Carl Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

cell:

Jurkat cells (kindly provided by Dr. G. Moldenhauer (DKFZ, Heidelberg, Germany)) were cultured in RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, and 100 μg/mL streptomycin sulfate (all components were from Invitrogen, referred to herein as complete RPMI 1640 medium).

Determination of antibody affinity to CD3+ cells by flow cytometry:

Cells were incubated on ice for 45 min with 100 μL serial dilutions of the indicated antibodies starting from 100 μg/mL (about 1000 nM) in ice-cold PBS (referred to as FACS buffer) supplemented with 2% heat-inactivated FCS and 0.1% sodium azide. After washing 3 times with FACS buffer, cells were incubated on ice for 45 min with 0.1 mL 10 μg/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After the second wash cycle, under the same conditions as before, cells were incubated with 0.1 mL 15 μg/mL FITC-conjugated goat anti-mouse IgG antibody. As a control, in the absence of a primary antibody, cells were incubated with anti-His IgG 13/45/31 and then with FITC-conjugated goat anti-mouse IgG antibody. The cells were then washed again and resuspended in 0.2 mL FACS buffer containing 2 μg/mL propidium iodide to exclude dead cells. Fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) with MXP software or a Millipore Guava EasyCyte flow cytometer (Merck Millipore, Schwalbach, Germany) with Incyte software. The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany). If the Beckman-Coulter FC500 MPL flow cytometer was used for analysis, 0.5×10 ⁶ cells/staining were used, and if the Millipore Guava EasyCyte flow cytometer was used, only 0.25×10 ⁶ cells/staining were used.

After subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone, the fluorescence intensity values and the equation (hyperbola) for single-site binding of GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) were used to calculate KD values.

For the series connection diabody of each analysis, the KD value in conjunction with EGFRvIII and cytotoxic EC50 are drawn into the function of the KD value in conjunction with CD3.When the KD value in conjunction with EGFRvIII of series connection diabody only demonstrated very little variability, the EC50 cytotoxicity value demonstrated and was accompanied by the CD3 that series connection diabody increases gradually in conjunction with KD value and almost became linear increase (Figure 11).

Example 12: Inhibition of F98 ^EGFRvIII Xenograft Tumors by EGFRvIII/CD3 (In Vivo Proof of Concept)

On day 0 (d0), nine experimental groups of immunodeficient NOD/scid mice (NOD/MrkBomTac-Prkdcscid, Taconic Denmark) were xenografted with a suspension of 4× ¹⁰ F98 ^npEGFRvIII cells mixed with 1× ¹⁰ purified human PBMCs from healthy donors via subcutaneous (sc) injection. To account for possible donor variability of PBMCs, each experimental group was subdivided into two cohorts, each receiving PBMCs from only one individual donor. 2 h after tumor cell inoculation, followed by 24 h, 48 h, 72 h, and 96 h, 100 μg, 10 μg, or 1 μg of each tandem double antibody test item was administered to the animals via the tail vein (iv). The control group received vehicle only. Starting on d0, another control was administered (ip) with 1 mg of cetuximab (Erbitux) twice a week. Table 6 summarizes the dose groups and donor allocation. Tumor volume was monitored by measuring the major and minor diameters of the tumors with calipers three times a week from day 3 to day 52. Tumor volume was calculated from the diameters according to the following formula: Volume = (minor diameter) ² x major diameter x 0.5.

Table 7: Animal dose groups

References:

Schwarz M, Rottgen P, Takada Y, Le Gall F, Knackmuss S, Bassler N, Buttner C, Little M, Bode C, Peter. Single-chain antibodies for the conformation-specific blockade of activated platelet integrin alphallbbeta3 designed by subtractive selection from naive human phage libraries. FASEB J. 2004 Nov; 18(14): 1704-6.

Schier R, McCall A, Adams GP, Marshall KW, Merritt H, Yim M, Crawford RS, Weiner LM, Marks C, Marks JD. Isolation of picomolar affinity anti-c-erbB-2 single-chain Fv by molecular evolution of the complementarity determining regions in the center of the antibody binding site. J Mol Biol. 1996.

Pini A, Viti F, Santucci A, Carnemolla B, Zardi L, Neri P, Neri D. Design and use of a phage display library. Human antibodies with subnanomolar affinity against a marker of angiogenesis eluted from a two-dimensional gel. J Biol Chem. 1998.

Boder ET, Midelfort S, Wittrup KD. Directed evolution of antibody fragments with monovalent femtomolar antigen-binding affinity. Proc Natl Acad Sci U S A. 2000 Sep 26; 97(20): 10701-5.

Barbas CF 3rd,Hu D,Dunlop N,Sawyer L,Cababa D,Hendry RM,Nara PL,Burton DR.In vitro evolution of a neutralizing human antibody to humanimmunodeficiency virus type 1 to enhance affinity and broaden strain cross-reactivity.Proc Natl Acad Sci U S A.1994 Apr 26;91(9):3809-13.

Parsons HL, Earnshaw JC, Wilton J, Johnson S, Schueler PA, Mahoney W, McCafferty J. Directing phage selections towards specific epitopes. Protein Eng. 1996 Nov; 9(l 1): 1043-9.

Winkler, Kramer A, Kiittner G, Seifert M, Scholz C, Wessner H, Schneider-Mergener J, Hohne W. Changing the antigen binding specificity by single point mutations of an anti-p24 (HIV-1) antibody. J Immunol. 2000 Oct 15; 165(8): 4505-14.

Safdari Y, Farajnia S, Asgharzadeh M, Omidfar K, Khalili M. humMRl, a highly specific humanized single chain antibody for targeting EGFRvIII. Int Immunopharmacol. 2014 Feb; 18(2): 304-10.

Lorimer IA, Keppler-Hafkemeyer A, Beers RA, Pegram CN, Bigner DD, Pastan I. Recombinant immunotoxins specific for a mutant epidermal growth factor receptor: targeting with a single chain antibody variable domain isolated byphage display. Proc Natl Acad Sci US A. 1996 Dec 10; 93(25): 14815-20.

Beers R, Chowdhury P, Bigner D, Pastan I. Immunotoxins with increased activity against epidermal growth factor receptor vlll-expressing cells produced by antibody phage display. Clin Cancer Res. 2000 Jul; 6(7): 2835-43.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Choi BD, Kuan CT, Cai M, Archer GE, Mitchell DA, Gedeon PC, Sanchez-Perez L, Pastan I, Bigner DD, Sampson JH. Systemic administration of a bispecific antibody targeting EGFRvIII successfully treats intracerebral glioma. Proc Natl Acad Sci U SA. 2013 Jan 2;110(l):270-5.

Fenstermaker RA, Ciesielski MJ. Deletion and tandem duplication of exons 2-7 in the epidermal growth factor receptor gene of a human malignant glioma. Oncogene. 2000 Sep 14; 19(39): 4542-8.

Han W, Zhang T, Yu H, Foulke JG, Tang CK. Hypophosphorylation of residue Y1045 leads to defective downregulation of EGFRvIII. Cancer Biol Ther. 2006 Oct; 5(10): 1361-8.

Grandal MV, Zandi R, Pedersen MW, Willumsen BM, van Deurs B, Poulsen HS. EGFRvIII escapes down-regulation due to impaired internalization and sorting to lysosomes. Carcinogenesis. 2007 Jul; 28(7): 1408-17. Gan HK, Kaye AH, Luwor RB. The EGFRvIII variant inglioblastoma multiforme. J Clin Neurosci. 2009 Jun; 16(6): 748-54.

Reist CJ, Archer GE, Kurpad SN, Wikstrand CJ, Vaidyanathan G, Willingham MC, Moscatello DK, Wong AJ, Bigner DD, Zalutsky MR. Tumor-specific anti-epidermal growth factor receptor variant III monoclonal antibodies: use of the tyramine-cellobiose radioiodination method enhances cellular retention and uptake in tumor xenografts. Cancer Res. 1995 Oct 1; 55(19): 4375-82.

Foulon CF, Reist CJ, Bigner DD, Zalutsky MR. Radioiodination via D-aminoacid peptide enhances cellular retention and tumor xenograft targeting of an internalizing anti-epidermal growth factor receptor variant III monoclonal antibody. Cancer Res. 2000 Aug 15; 60(16): 4453-60.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Patel D, Lahiji A, Patel S, Franklin M, Jimenez X, Hicklin DJ, Kang X. Monoclonal antibody cetuximab binds to and down-regulates constitutively activated epidermal growth factor receptor vIII on the cell surface. Anticancer Res. 2007 Sep-Oct; 27(5A): 3355-66. Jutten B, Dubois L, Li Y, Aerts H, Wouters BG, Lambin P, Theys J, Lammering G. Binding of cetuximab to the EGFRvIII deletion mutant and its biological consequences in malignant glioma cells. Radiother Oncol. 2009 Sep; 92(3): 393-8.

Dreier A, Barth S, Goswami A, Weis J. Cetuximab induces mitochondrial translocalization of EGFRvIII, but not EGFR: involvement of mitochondria in tumor drug resistance? Tumour Biol. 2012 Feb;33(l):85-94.

Antibody binding site specific for EGFRvIII

The present invention relates to a binding protein that specifically binds to EGFRvIII and a multispecific binding protein that specifically binds to EGFRvIII and CD3. The present invention further extends to a multispecific tandem diabody that binds to EGFRvIII and CD3. In particular, the present invention relates to a highly cytotoxic EGFRvIII/CD3 bispecific tandem diabody that is used to recruit T cells to effectively kill several types of solid tumor cancers.

EGFR imbalance is associated with many cancers, and small molecules and EGFR-targeted antibodies have all been successfully used in clinical practice. However, due to the extensive normal tissue expression of EGFR, the antibodies approved for clinical use and those in development all have serious side effects. The deletion mutant (deletion variant) III (EGFRvIII) of EGFR with truncated extracellular domain and ensuring ligand-independent intrinsic activity is the most common mutant form associated with oncogenic transformation. EGFRvIII is specifically expressed in cancer tissue and is associated with various solid tumor types. The restricted EGFRvIII expression on cancer cells provides the opportunity for developing a cytotoxic antibody that specifically targets cancer, protects normal tissues, and fully reduces the side effects associated with EGFR treatment.

In a first aspect of the present invention, fully human, highly specific, high-affinity EGFRvIII binding proteins are described herein. The described EGFRvIII binding proteins have the following advantages:

Figure 9 shows that EGFRvIII/CD3 of the present invention is the most common EGFRvIII binding protein of the present invention.They do not promote or promote the very low EGFRvIII internalization from the cell surface of target positive molecule, and this helps to raise cytotoxic T-cell or NK-cell.The result in Fig. 9 shows, according to the present invention, after cell is exposed to EGFRvIII conjugated proteins under all test conditions, with respect to untreated cell, EGFRvIII receptor molecules still remains on the cell surface.These results show, EGFRvIII/CD3 conjugated proteins of the present invention surprisingly do not show any internalization trend (Fig. 9).The combination of EGFRvIII specific binding proteins of the present invention may suppress internalization, rather than inducing internalization, or may promote the increase that EGFRvIII expresses, thereby causes its cell surface density to increase.

Can use affinity maturation technology (affinity maturation technique) to make the avidity of these high-affinity associated proteins be further significantly improved to obtain that EGFRvIII is had specificity, K _D in 100pM scope or lower associated proteins.Surprising fact is, these associated proteins have demonstrated the abnormal specificity in conjunction with EGFRvIII, and natural EGFR (wild-type EGFR or EGFRwt) is not had cross reactivity.Due to from 269 amino acid whose frames (in-frame) of EGFR extracellular domain (exon 2-7), therefore with respect to EGFRwt, in EGFRvIII, form new epi-position (neo-epitope), and estimate that this new epi-position is quite little: it is arranged by new amino acid (novel juxtaposition of aminoacids) (amino acid 5 is fused to amino acid 274) and a new GLY amino acid composition at the deletion site.

The goal of affinity maturation screening methods is to improve binding affinity by promoting the selection of binding proteins with reduced off-rates ( _kOFF ). However, surprisingly, the affinity matured binding proteins shown have greatly increased on-rates (kON ₎ , while the reduction in _kOFF has only a small contribution to the 100-fold increased binding of the binding proteins, with some variable antibody domains showing _KD less than 100 pM.

These unique properties make the variable antibody domains described herein that are specific for EGFRvIII particularly suitable for the development of multispecific, e.g., bispecific, multivalent, immune effector cell engaging, tumor-targeting antibody therapeutics, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies.

In a further aspect of the invention, multivalent EGFRvIII binding antibodies are provided that have two binding sites for EGFRvIII, or in the case of a multivalent bispecific antibody, in addition to two binding sites for EGFRvIII, also have a binding site for the T-cell antigen CD3 or CD16A specifically expressed on natural killer (NK) cells.

In one embodiment, use EGFRvIII is had two binding sites and comprises in addition the EGFRvIII/CD3 series connection diabody that CD3 has two binding sites, and tested their combination with recombinant EGFRvIII-or EGFR-Fc fused antigen.Under non-reducing conditions (due to complete disulfide bond, molecular structure still part retains), or under reducing conditions (protein structure complete denaturation), EGFRvIII-or EGFR-Fc fused antigen separates by SDS-PAGE, is transferred to PVDF membrane by Western blotting (Western Blotting), and uses different EGFRvIII/CD3 series connection diabody antibodies (diabodyantibody) or EGFR-to assess in conjunction with modifying (decorate) trace (Fig. 4).This embodiment shows, comprises the parental EGFRvIII binding domains that EGFRvIII has two binding sites and the EGFRvIII binding domains of affinity maturation and also keeps its selection specificity to EGFRvIII in multivalent antibody form. Under reducing and non-reducing conditions, all EGFRvIII/CD3 series connection double antibodies all can identify EGFRvIII, but all can not identify full-length wild-type EGFR under these conditions arbitrarily.This embodiment also proves, and EGFRvIII specific binding structural domain described herein can identify linear epitope, and its reactivity does not need complete three-dimensional structure.This is opposite with EGFR-in conjunction with IgG antibody cetuximab (Erbitux (Erbitux)), EGFR-in conjunction with IgG antibody cetuximab can identify wild-type EGFR and ripe EGFRvIII, but its reactivity obviously needs complete disulfide bond and conformational epitope (Fig. 4).

In another embodiment, EGFRvIII/CD3 tandem diabodies comprising parental or affinity-matured EGFRvIII specific domains in combination with different CD3 binding domains and/or having single binding domains of different orders in the tandem diabody molecule are used. The tandem diabody with the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII - _{V H} ^CD3 contains a divalent EGFRvIII binding site in the middle of the tandem diabody molecule, while the tandem diabody with the domain order V _H ^EGFRvIII -V _L ^CD3 -V _H ^CD3 -V _L ^EGFRvIII has two EGFRvIII binding sites in the outside position and two CD3 binding sites in the middle. In ELISA, the concentration-dependent binding of different EGFRvIII-domains and domain order mutants of the tandem diabody to EGFRvIII antigen, wild-type EGFR antigen, and to the concentration-dependent binding of the CD3 antigen (Fig. 6) are analyzed. All tandem diabodies containing affinity-matured EGFRvIII binding domains showed a substantial increase in binding to EGFRvIII, clearly manifested as corresponding binding curves shifting by more than two orders of magnitude to lower concentrations ( FIG. 6 ). No binding of any tandem diabody to wild-type EGFR antigen was observed.

Can select and identify anti-EGFRvIII antibody by phage display library (phase display libraries), for example, selectively combine the antigen-binding proteins of the EGFR of maturation rather than native form (native form).T cell is the effective tumor-killing effector cell that can not be raised by natural antibody.Therefore, in a further aspect of the present invention, the invention provides one group of multispecific EGFRvIII/CD3 antigen-binding proteins, particularly series connection diabody, it can raise T cell, have the binding characteristic and cytotoxicity of a wide range.Characterized the feature of the binding characteristic of antigen-binding proteins of the present invention, the cytotoxic activity of T cell mediation and the T cell activation of target mediation in one group of in vitro test.In FACS, antigen-binding proteins of the present invention shows unusual specificity to the EGFRvIII antigen in Biacore, ELISA and EGFRvIII-positive cell.Within the full concentration range of test, do not observe any conjugated protein and EGFR antigen or with the detectable combination of EGFRwt expressing cell. The most effective high-affinity series diabody shows the cytotoxicity of EC50 for F98 glioma cells and Chinese hamster ovary celIs expressing EGFRvIII and CHO cells; Remaining series diabody shows the cytotoxicity of EC50 up to about 10000pM.Also tested these series diabody pairs as the cytotoxicity of the EGFRwt ⁺ cell of the more sensitive probe of the residue of native form.When the series diabody concentration of maximum assessment was 0.5 μM, cytotoxicity was observed on EGFRwt ⁺ cell or other EGFRvIII-negative cells.It is crucial to be combined with the high-affinity of CD3 to effectively T cell recruitment, yet, when there is no EGFRvIII ⁺ target cell in vitro, as lacking measurement by proliferation, the series diabody of the present invention does not cause T cell activation, and this contributes to good preclinical safety (Figure 10).In a word, the strict specificity and high efficiency of anti-tumor cell toxicity are mediated by EGFRvIII/CD3 series diabody.

Tandem diabodies with CD3 binding _KD ranging from 1.1 nM to approximately 500 nM but relatively constant EGFRvIII binding _KD (ranging from 2.0 nM to 6.7 nM) were generated and showed good correlation between CD3 binding strength and cytotoxic potency against cells expressing EGFRvIII, with _EC50 values ranging from 25 pM to approximately 10,000 pM (Figure 11).

Although the affinity of the antigen-binding protein of the present invention for EGFRvIII has been significantly improved, the specificity is not lost.

is a trademark of Affimed Therapeutics used to represent tandem diabodies (Kipriyanov et al., 1999, J. Mol. Biol, 293: 41-56; Cochlovius et al., 2000, Cancer Res., 60: 4336-4341; Reusch et al., 2004, Int. J. Cancer, 112: 509-518, Kipriyanov, 2009, Methods Mol Biol, 562: 177-93; McAleese and Eser, 2012, Future Oncol. 8: 687-95). In the context of the present invention, TandAb and tandem diabodies are used as synonyms.

The sequence of wild-type EGFR gene and protein is known. EGFRvIII is the result of an in-frame deletion of exons 2-7 (801bp) of the wild-type EGFR gene, resulting in 267 amino acid deletions in the receptor's outer domain and a new glycine residue at the junction of exons 1 and 8. This new amino acid juxtaposition in the EGFR extracellular domain generates tumor-specific and immunogenic epitopes.

According to first aspect of the present invention, described EGFRvIII had to specific associated proteins at least, it does not cross-react with wild-type EGFR.This associated proteins preferably comprises CDR1, CDR2, CDR3 of the antibody variable heavy chain that is selected from the CDR described in the following table 1A and CDR1, CDR2 and the CDR3 of antibody variable light chain.

Table 1A: Sequence identifiers of amino acid sequences of human light and heavy chain CDRs from binding proteins that specifically bind to EGFRvIII

Therefore, an embodiment of the present invention is an EGFRvIII binding protein having at least one EGFRvIII binding site, comprising an antibody variable heavy chain domain and/or an antibody variable light chain domain; wherein the antibody variable heavy chain domain comprises a CDR1 selected from SEQ ID NO: 27, 33, 42, 46, 49, 53, a CDR2 selected from SEQ ID NO: 28, 38, 43, 50, 54, 61, 63 and a CDR3 of SEQ ID NO: 29; the antibody variable light chain domain comprises a CDR1 selected from SEQ ID NO: 30, 34, 36, 39, 44, 47, 51, 56, 59, 64, a CDR2 selected from SEQ ID NO: 31, 40, 57 and a CDR3 selected from SEQ ID NO: 32, 35, 37, 41, 45, 48, 52, 55, 58, 60, 62 and 65.

According to further embodiment, described associated proteins comprises and has the antibody variable heavy chain domain that is selected from SEQ ID NO:1,3,5,7,9,11,13,15,17,19,21 and 25 as shown in Table 1B and/or is selected from SEQ ID NO:2,4,6,8,10,12,14,16,18,20,22 and 26 as shown in Table 1B at least one EGFRvIII binding site of antibody variable light chain domain.The aminoacid sequence of three variable light chain CDR and three variable heavy chain CDR is shown in bold and is underlined in Table 1.These structural sites show that the affinity to EGFRvIII is improved, and specificity does not have loss.These antigen-binding proteins do not cross-react with wild-type EGFR antigen.

In some embodiments, the present invention relates to an antigen-binding protein of the present invention.These antigen-binding proteins of the present invention do not show or only show minimized EGFRvIII internalization when in conjunction with EGFRvIII.According to the present invention, under the test conditions according to embodiment 9, after cell is exposed to EGFRvIII-in conjunction with antibody, be greater than 80%, preferably greater than 90%, most preferably be greater than 95% EGFRvIII receptor molecules and still remain on cell surface.The protein-binding combination of EGFRvIII of the present invention may suppress internalization, rather than inducing internalization, or may promote that the expression of EGFRvIII increases, thereby causes its cell surface density to increase.

The term "binding protein" refers to an immunoglobulin derivative with antigen-binding properties, i.e., an immunoglobulin polypeptide or fragment thereof containing an antigen-binding site. The binding protein comprises the variable domains of an antibody or fragment thereof. Each antigen-binding domain is formed by an antibody, i.e., an immunoglobulin, a variable heavy chain domain (VH), and an antibody variable light chain domain (VL), which bind to the same epitope. The variable light chain domain or variable heavy chain domain of the present invention is a polypeptide comprising CDR1, CDR2, and CDR3. Preferably, the binding protein of the present invention lacks immunoglobulin constant domains. The term "binding protein" also refers to antibody fragments or derivatives, including Fab, Fab', F(ab') ₂ , Fv fragments, single-chain Fv, diabodies, tandem diabody flexibodies (WO03/025018), and tandem single-chain Fv ((scFv) ₂ ).

Table 1B: Amino acid sequences of all anti-EGFRvIII variable heavy chain domains (VH) and variable light chain domains (VL) (amino acid sequences of CDR1, CDR2, and CDR3 are shown in bold and underlined)

In a preferred embodiment, the binding protein that confers specificity for EGFRvIII comprises an antigen binding site from one of the following combinations of variable heavy chain domains and variable light chain domains shown in Table 1:

(i) SEQ ID NO: 1 and SEQ ID NO: 2,

(ii) SEQ ID NO: 3 and SEQ ID NO: 4,

(iii) SEQ ID NO: 5 and SEQ ID NO: 6,

(iv) SEQ ID NO: 7 and SEQ ID NO: 8,

(v) SEQ ID NO: 9 and SEQ ID NO: 10,

(vi) SEQ ID NO: 11 and SEQ ID NO: 12,

(vii) SEQ ID NO: 13 and SEQ ID NO: 14,

(viii) SEQ ID NO: 15 and SEQ ID NO: 16,

(ix) SEQ ID NO: 17 and SEQ ID NO: 18,

(x) SEQ ID NO: 19 and SEQ ID NO: 20,

(xi) SEQ ID NO: 21 and SEQ ID NO: 22, or

(xii) SEQ ID NO:25 and SEQ ID NO:26.

The present invention also provides not only EGFRvIII has specificity, and also has the associated proteins of at least one other functional domain.In a further embodiment, described at least one other functional domain is effector domain (effector domain)." effector domain " is that effector cell is had specific binding site, and it can stimulate or trigger cytotoxicity, engulfment, antigen presentation, cytokine release.Such effector cell is, for example but not limited to, T-cell or NK-cell.Particularly, described other effector domains comprise at least one antibody variable heavy chain domain and at least one variable light chain domain that form the antigen binding site of CD3 preferred mankind CD3.

In some embodiments, the EGFRvIII conjugated proteins of the present invention can be multispecific.Term used herein " multispecific " means to refer to that conjugated proteins of the present invention have at least two antigen binding sites to different epi-positions, and wherein at least one is used for EGFRvIII.For example, conjugated proteins can be trispecific, and have binding site and have at least one binding site to effector cell such as epi-position on CD3 or antigen to two different epi-positions and/or antigen on tumor cell.Conjugated proteins can have binding site to different epi-positions of same antigen and/or the epi-position of different antigens.Such multispecific conjugated proteins comprise single-chain diabody, (scFv) ₂ , series diabody and elastomeric antibody (referring to Le Gallet al., 1999FEBS Letts 453:164-168 and WO 03/025018).In a specific embodiment of the present invention, conjugated proteins are to EGFRvIII and CD3 bispecific.

The CD3 binding site of the multispecific binding protein of the invention can be composed of a variable heavy chain domain (SEQ ID NO: 23) and a variable light chain domain (SEQ ID NO: 24) as shown in Table 2, or a close homolog of such sequences that differs from the CDR sequences by only two amino acids, or any other fully human, humanized or non-human variable antibody domain specific for CD3 (referred to as CD3), such as, for example, a variable antibody domain specific for CD3, in particular CD3 epsilon, or any other fully human, humanized or non-human variable antibody domain, such as, for example, the variable antibody domain of UCHT1 or OKT3.

Table 2: Amino acid sequences of fully human anti-CD3 variable heavy chain domain (VH) and variable light chain domain (VL) (amino acid sequences of CDR1, CDR2 and CDR3 are shown in bold and underlined)

In addition, EGFRvIII conjugated proteins of the present invention can be multivalent.Term used herein " multivalent " means to refer to, and conjugated proteins of the present invention comprises at least two antigen binding sites.Antigen binding site can have identical or different specificity.In embodiments of the present invention, conjugated proteins is in conjunction with identical epi-position with at least two binding sites in a divalent manner.The protein-conjugated example of divalence is diabody, and the protein-conjugated example of at least tetravalence is a series connection diabody.

In a further aspect of the present invention, described EGFRvIII associated proteins and bispecific EGFRvIII and CD3 binding site are humanized or fully human, i.e. people-source. In a further aspect of the present invention, EGFRvIII associated proteins or multispecific EGFRvIII and CD3 associated proteins are dimers, i.e. comprise two polypeptides with EGFRvIII and CD3 antigen binding sites.

According to the present invention, dimerization bispecific EGFRvIII and CD3 binding proteins are provided in the form of series connection diabody.Such series connection diabody is by connecting four antibody variable binding domains (heavy chain variable domain (VH) and light chain variable domain (VL) in single gene construct (referring to McAleese and Eser, 2012, Future Oncol.8:687-95)) that can homodimerize and build.In such series connection diabody, joint (linker) length is that it prevents the intermolecular pairing of variable domain so that molecule can not fold back to form single-chain diabody on itself, but is forced to pair with the complementary domain (complementary domain) of another chain.These domains can also be arranged like this, namely make corresponding VH and VL domain pair during dimerization.After being expressed by single gene construction, two same polypeptide chains head to tail folding form the functional non-covalent homodimer (McAleese and Eser, 2012, Future Oncol.8:687-95) of about 110kDa. Despite the absence of intramolecular covalent bonds, homodimers, once formed, are highly stable, remain intact and do not revert to monomeric form.

In one embodiment of a multispecific tandem diabody, a humanized bispecific tetravalent antibody is provided that has two binding sites each for CD3 and EGFRvIII (EGFRvIII/CD3RECRUIT-) to utilize the cytotoxic capacity of T cells to treat glioblastoma (GB), prostate cancer, head and neck cancer, and other epidermal growth factor receptor mutant vIII-positive (EGFRvIII+) cancers.

In the present invention, the tandem diabody is the most widely used antibody of the present invention.Relative to traditional monoclonal antibody, tandem diabody has a large amount of properties that advantage is provided, also has less multispecific molecule.Tandem diabody is fully functional when there is no glycosylation.Tandem diabody can only contain antibody variable domains, and therefore can not have any side effect that can be relevant to Fc part.Because bispecific tandem diabody allows divalence to be bound to each in EGFRvIII and CD3, therefore, its avidity (avidity) is identical with the avidity of IgG.The size of tandem diabody is about 110kDa, is less than IgG, and this can enhance tumor infiltration.But, compared with the less bispecific form based on antibody-binding domain or non-antibody scaffold, this size is much higher than the kidney threshold of first-pass clearance (first-pass clearance), and this provides pharmacokinetic advantage. The generation and preparation of such tandem diabodies is described, for example, in Kipriyanov SM: Methods Mol. Biol. 2009; 562: 177-93, Kipriyanov SM: Methods Mol. Biol. 2003; 207: 323-33 or McAleese and Eser, 2012, Future Oncol. 8: 687-95.

Tandem diabodies are well expressed in CHO cells, enabling efficient upstream and downstream manufacturing processes to be put in place.

The EGFRvIII and CD3 bispecific tandem double antibody of the present invention is designed to specifically target EGFRvIII ⁺ tumor cells by recruiting cytotoxic T cells. Antibodies cannot directly recruit cytotoxic T cells. On the contrary, by employing (engage) the CD3 molecules specifically expressed on these cells, the tandem double antibody can make cytotoxic T cells and EGFRvIII ⁺ tumor cells cross-linked in a highly specific manner, thereby significantly increasing the cytotoxic effect of these molecules. The tandem double antibody shows strong, specific and effective cytotoxicity. Significant evidence is that T cells can play a role in controlling tumor growth. For example, the presence of cytotoxic T cells in colorectal tumors and lymph nodes from NHL patients is shown to be correlated with better clinical results. In addition, for melanoma vaccines, and antibodies for CTLA-4, a negative regulator of T-cell activation, have demonstrated the potential for the therapy of designing induction of T-cell responses. The tandem double antibody of the present invention employs cytotoxic T cells by binding surface-expressed CD3, preferably CD3ε, which forms a part of the T-cell receptor. The simultaneous binding of this tandem diabody to CD3 and EGFRvIII expressed on the surface of specific tumor cells leads to T-cell activation and mediates subsequent tumor cell lysis.

"Dimer" refers to a complex of two polypeptides. Preferably, the two polypeptides are non-covalently associated with each other, particularly under conditions where there is no covalent bond between the two polypeptides. Preferably, the bispecific dimer is a homodimer, i.e., comprises two identical polypeptides. However, a trispecific or other multispecific dimer can be a heterodimer and comprise two different polypeptides, for example, at least one binding site of the variable light chain domain is located on one polypeptide and the variable heavy chain domain is located on the other domain. The term "polypeptide" refers to a polymer of amino acid residues connected by amide bonds. Preferably, the polypeptide is an unbranched single-chain fusion protein. In this polypeptide, the variable antibody domains are connected in sequence. In addition to the variable domain N-terminus and/or C-terminus, the polypeptide may also have contiguous amino acid residues. For example, such contiguous amino acid residues may include a Tag sequence that may be beneficial to polypeptide purification, preferably at the C-terminus.

Each polypeptide of the bispecific tandem diabody contains at least four variable domains, namely the variable light chain (VL) and variable heavy chain (VH) of the CD3 binding protein and the variable light chain (VL) and variable heavy chain (VH) of the EGFRvIII binding protein. The four binding domains are connected by peptide linkers L1, L2 and L3 and can be arranged as follows from N-terminus to C-terminus:

(i)VL(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3); or

(ii) VH(CD3)-L1-VL(EGFRvIII)-L2-VH(EGFRvIII)-L3-VL(CD3); or

(iii) VL(EGFRvIII)-L1-VH(CD3)-L2-VL(CD3)-L3-VH(EGFRvIII); or

(iv) VH(EGFRvIII)-L1-VL(CD3)-L2-VH(CD3)-L3-VL(EGFRvIII).

In an embodiment of the present invention, the four variable domains are arranged as VL ^{(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3). Tandem diabodies having the domain order VLCD3-VHEGFRvIII -VLEGFRvIII-VHCD3} _{and containing} ^different _EGFRvIII ^binding _sequences , despite having different affinities for EGFRvIII, all showed very similar binding to the second specific CD3, as shown by ELISA on the CD3γε antigen ( FIG5A ). In other tandem diabody domain sequences containing EGFRvIII binding domains in the outer positions and two CD3 binding sites in the middle ( _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ), a significant improvement in EGFRvIII binding to the affinity-matured EGFRvIII binding domain was also evident, while binding of the tandem diabody middle domain to CD3 was slightly weaker and more variable between individual tandem diabodies than the first domain sequence (Figure 5B). Similar observations were made when tandem diabodies containing different EGFRvIII-binding domains were constructed using different CD3 domains (SEQ ID NOs: 23 and 24) (Figure 5C).

The length of the linker affects the flexibility of the antigen-binding tandem diabody. For example, Todorovska et al., 2001 Journal of Immunological Methods 248:47-66; Perisic et al., 1994 Structure 2:11217-1226; Le Gall et al., 2004 Protein Engineering 17:357-366 and WO94/13804 describe the effect of linker length on the formation of dimeric antigen-binding proteins.

According to the present invention, the length of the peptide linkers L1, L2, and L3 is preferably such that the domains of one polypeptide can intermolecularly bind to the domains of another polypeptide to form a dimeric antigen-binding tandem diabody. Such linkers are "short," i.e., comprised of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, or 12 amino acid residues. Thus, the linker consists of approximately 12 or fewer amino acid residues. In the case of 0 amino acid residues, the linker is a peptide bond. Such short linkers facilitate intermolecular dimerization of the two polypeptides by binding and forming a correct antigen-binding site between the antibody variable light chain domain and the antibody variable heavy chain domain of the different polypeptides. Short linkers of approximately 12 or fewer amino acid residues generally prevent intermolecular interaction between adjacent domains of the same polypeptide chain. In one embodiment of the present invention, these linkers consist of approximately 3 to approximately 10, e.g., 4, 5, or 6, consecutive amino acid residues.

Regarding the amino acid composition of these linkers, peptides are selected that do not interfere with the dimerization of the two polypeptides. For example, linkers containing glycine and serine residues generally provide protease resistance. The amino acid sequence of the linker can be optimized, for example, by phage display methods to increase antigen binding and the yield of antigen-binding polypeptide dimers. Examples of peptide linkers suitable for the tandem diabodies of the present invention are GGSGGS (SEQ ID NO: 66), GGSG (SEQ ID NO: 67), or GGSGG (SEQ ID NO: 68).

EGFRvIII associated proteins as herein described and multi-specific series connection diabody can produce by the polynucleotide of expressing coding series connection diabody polypeptide, and the polypeptide of described series connection diabody is combined to form antigen-in conjunction with series connection diabody with another identical polypeptide.Therefore, the further embodiment of the present invention is the polynucleotide of coding antigen as herein described-in conjunction with series connection diabody polypeptide, for example DNA or RNA.

Polynucleotide can be constructed by methods well known to those skilled in the art, for example, by combining the gene of at least four antibody variable domains separated by the sequence encoding peptide linkers or directly connected by peptide bonds into a single gene construct, and in bacteria or other suitable expression systems such as for example expressed in Chinese hamster ovary celI, the single gene construct is operably connected with a suitable promotor, alternatively a protein tag for detection and purification, and a suitable transcription terminator. According to employed vector systems and hosts, any number of suitable transcription and translation elements comprising formation and inducible promoters can be used. Select promoter so that it drives the expression of polynucleotides in respective host cells.

As a further embodiment of the present invention, the polynucleotide can be inserted into a vector, preferably an expression vector. The recombinant vector can be constructed according to methods known to those skilled in the art.

Various expression vector/host systems can be used to contain or express polynucleotides encoding polypeptides of the present invention. Examples of expression vectors for expression in E. coli are pSKK (Le Gall et al., J Immunol Methods. (2004) 285(1): 111-27) or pcDNA5 (Invitrogen) for expression in mammalian cells.

Thus, the antigen-binding tandem diabodies described herein can be prepared by introducing a vector encoding the polypeptides described above into host cells and culturing the cells under conditions whereby the polypeptide chains are expressed, can be isolated from the culture medium, and optionally thereafter purified.

In a further embodiment, the present invention provides a pharmaceutical composition comprising an EGFRvIII binding protein, an antigen-binding tandem diabody, a vector comprising a polynucleotide encoding the polypeptide of the antigen-binding tandem diabody or a host cell transformed by the vector, and at least one pharmaceutically acceptable carrier. The term "pharmaceutically acceptable carrier" means any carrier that does not interfere with the effectiveness of the biological activity of the component and is non-toxic to the administered patient. Suitable pharmaceutical carrier examples are known in the art and include physiological saline, phosphate buffered saline, water, emulsions such as oil/water emulsions, various types of wetting agents, sterile solutions, etc. Such carriers can be prepared by conventional methods and administered to an individual at a suitable dosage. Preferably, the composition is sterile. These compositions may also include adjuvants, such as preservatives, emulsifiers, and dispersants. The effects of microorganisms can be prevented by including various antibacterial and antifungal agents.

The skilled artisan will readily construct and obtain antigen-binding proteins such as the tandem diabodies described herein without undue burden by using well-established techniques and standard methods known in the art, see, for example, Sambrook, Molecular Cloning A Laboratory Manual, Cold Spring Harbor Laboratory (1989) N.Y.; The Protein Protocols Handbook, John M. Walker, ed., Humana Press Inc. (2002); or Antibody engineering: methods and protocols/Benny K.C. Lo, ed.; Benny K.C. II Series: Methods in molecular biology (Totowa, N.J.).

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 illustrates the position of variable sequence in the EGFRvIII specific high-affinity binding domain derived after affinity maturation.The CDR position that is different from the heavy chain and light chain variable region of parental sequence is marked with asterisk, and amino acid is highlighted with grey separately.The position with useful sudden change is highlighted with dark grey.

Fig. 2 shows that among Biacore T200, multi-circulation kinetics measures the result that different scFv antibodies are attached to recombinant EGFRvIII-Fc antigen or wild-type EGFR-Fc antigen.Sensogram is presented at the kinetics of surface plasmon resonance (SPR) response unit (RU) measured in time under 4 different scFv antibody concentration levels (86nM, 17.2nM, 3.4nM, 0.69nM).Measurement is in conjunction with 180 seconds, after this measurement dissociates.Use 1:1 to combine model analysis data.

Figure 3 shows the results of flow cytometry measurement of scFv antibody binding to (A) F98 rat glioma cells overexpressing EGFRvIII (F98 ^EGFRvIII ), (B) F98 cells overexpressing native EGFR (F98 ^EGFR ), and (C) untransfected F98 cells (F98), as well as the results of similar scFv antibody binding to (D) CHO cells overexpressing EGFRvIII (CHO ^EGFRvIII ), (E) CHO cells overexpressing native EGFR (CHO ^EGFRvIII ), and (F) untransfected CHO cells (CHO).

Fig. 4 illustrates the tandem diabody containing different EGFRvIII binding domains and is separated by SDA-PAGE and is transferred to the EGFRvIII-or natural EGFR-Fc fusion antigen (non-reducing protein or the albumen with DTT reduction) of the purification of PDVF film by western blotting.Fc fusion rotein is by disulfide bond at Fc part dimerization, but is monomer after reducing with DTT.EGFRvIII-specific antibody is bonded to EGFRvIII-fusion rotein with reduction or non-reduction conformation, but is not combined with natural EGFR.This anti-EGFR antibody cetuximab is as contrast, and identifies non-reducing EGFRvIII-Fc and non-reducing EGFR-Fc antigen, but does not identify the antigen of DTT-processing.

Figure 5 shows the concentration-dependent binding of different EGFRvIII/CD3-specific tandem diabodies to EGFRvIII-Fc, EGFR-Fc, and CD3γε recombinant antigens analyzed in ELISA. (A) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , (B) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , and (C) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies containing the CD3 domains of SEQ ID NOs: 23 and 24 and having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 . The improvement in binding of tandem diabodies containing affinity-matured EGFRvIII-binding domains to EGFRvIII was clearly detectable as a shift of the respective binding curves to lower concentrations in all formats/domain orders.

Figure 6 shows the results of multi-cycle kinetic measurements of EGFRvIII/CD3-specific tandem diabodies containing different bivalent EGFRvIII binding domains in the order of _V ^CD3 - _V ^EGFRvIII - _V ^EGFRvIII - _V ^CD3 bound to recombinant EGFRvIII-Fc antigen in a Biacore T200. Sensogram displays the kinetics of surface plasmon resonance (SPR) response units (RU) measured over time at different tandem diabody concentration levels. Binding was measured for 180 seconds, followed by dissociation. Data were analyzed using a 1:1 binding model.

Fig. 7 shows the immunohistochemistry (IHC) staining result of tissue section from (A) three individual head and neck (H＆N) cancer patients, (B) three individual glioblastoma patients, and (C) representative results from prostate cancer, mammary cancer (Her2neg) and non-small cell lung cancer (NSCLC).The EGFRvIII-specific double antibody containing EGFRvIII-specific binding domain Li3G30 of the divalent conformation of 2 kinds of different concentrations is dyed in section, and described divalent conformation is similar to the arrangement of EGFRvIII targeting tandem double antibody antibody (for example, (V _L ^CD3- ) V _H ^EGFRvIII - _{V L} ^EGFRvIII (-V _H ^CD3 )) but lacks CD3 binding part.As a control, section is dyed with the EGFR-specific IgG antibody cetuximab of identification EGFR (being present on healthy tissue and cancerous tissue) and EGFRvIII.The specificity of dyeing is confirmed by the control section hatched in the secondary reagent that does not contain EGFR-or EGFRvIII-specific primary antibody.

Fig. 8 illustrates that on EGFRvIII positive cells, compared with the tandem diabody of EGFRvIII and CD3 respectively containing the parental EGFRvIII binding domain (Li3G30) before affinity maturation, the cytotoxicity ability, effectiveness and target-dependent specificity of EGFRvIII/CD3 tandem diabody containing different affinity maturation EGFRvIII-specific binding domains: (A) ABC 470 containing EGFRvIII/CD3 tandem diabody, (B) ABC 471 containing EGFRvIII/CD3 tandem diabody, (C) ABC 472 containing EGFRvIII/CD3 tandem diabody. Using CHO cells expressing EGFRvIII (CHO ^EGFRvIII ), expressing EGFRvIII negative EGFR (CHO ^EGFR ) or untransfected CHO cells as target cells and using human PBMC as effector cells, in the analysis based on flow cytometry, antibody concentration-dependent cytotoxicity was measured. In each panel (AC), tandem diabodies containing a single EGFRvIII binding domain and CD3 domain v64, in the order _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , are shown on the left, and tandem diabodies containing a single EGFRvIII binding domain combined with a different CD3 domain, in the order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , are shown on the right.

Figure 9 shows the results of evaluating EGFRvIII receptor internalization from the cell surface when combined with EGFRvIII-specific antibodies. Cells expressing EGFRvIII were incubated for 24 hours at 37°C with varying concentrations of three different ^EGFRvIII / ^CD3 tandem diabody antibodies with the domain order V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII - _{V H} ^CD3 (i) or _{V H} EGFRvIII-V L ^CD3 -V H CD3-V _L ^EGFRvIII (iv), and/or EGFRvIII/CD3 tandem diabody antibodies containing different CD3-binding domains. After adjustment, all remaining EGFRvIII receptor molecules on the cell surface were stained with the respective tandem diabody antibodies at a saturating concentration of 10 μg/ml.

Figure 10 is presented in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) series double antibody antibody, assesses the result of cell proliferation (BrdU incoporation) in human PBMC culture.PBMC is used as positive control, and it breeds in the presence of CD3 binding antibody OKT3 or phytohemagglutinin (plant lectin phytohemagglutinin) (PHA).In the absence of EGFRvIII-positive target cells, with different EGFRvIII/CD3 series double antibodies, hatch PBMC and do not induce activation and the proliferation of PBMC.

Figure 11 illustrates that the CD3-binding affinity of EGFRvIII/CD3 tandem diabodies is positively correlated with their cytotoxic capacity. Regarding CD3-binding on CD3 ⁺ -Jurkat cells, EGFRvIII-binding on CHO ^EGFRvIII cells and cytotoxic capacity, all 20 EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) in the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII -V _H ^CD3 and different CD3-binding domain combinations with different affinities for CD3 were analyzed. The scope of CD3-binding K _D value is 1nM to approximately 500nM. For each analyzed tandem diabodies, relative to CD3-binding K _D value, EGFRvIII-binding K _D value and cytotoxic EC ₅₀ value are mapped. While the TandAbs showed only small changes in EGFRvIII-binding _KD values, the cytotoxicity _EC50s showed an almost linear increase as the CD3-binding _KD values of the tandem diabodies increased.

Figure 12 shows the in vivo proof-of-concept results from the study. Two EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII-binding domain (Li3G30) but different CD3-binding domains or domain sequences were analyzed for dose-dependent inhibition of subcutaneous F98 ^EGFRvIII xenograft tumor growth and compared with the effect of cetuximab administered at a much higher concentration. (A) Dose-dependent tumor growth inhibition caused by the ^EGFRvIII /CD3 ^v6 tandem diabody with the domain sequence _VL ^CD3 - _VH EGFRvIII- _VL ^EGFRvIII - _VH ^CD3 ; (B) Dose-dependent tumor growth inhibition caused by EGFRvIII/CD3 tandem diabodies containing different high-affinity CD3-binding domains and the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII .

The present invention is further illustrated by the following examples, but the present invention is not limited thereto.

Example 1: Discovery and affinity maturation of EGFRvIII-specific binding proteins:

Li3G30 Discovery:

The phage display library of people's scFv sequence based on IgM source is carried out two to three rounds of panning (panning) and EGFRvIII is had specific binding substances (binders) with enrichment.Before every round of panning, described library is pre-incubated with the binding substances of (deplete) wild-type form EGFR or the binding substances of fusion rotein Fc part with EGFR-Fc.On the solid phase of EGFRvIII coating and in solution, select (selection) in parallel.After two rounds and three rounds of panning, pick out monoclonal, induce soluble scFv to express and screen the extract that is combined with EGFRvIII-Fc and EGFR-Fc.

According to the present invention, the fully human high specific variable antibody domains combined with EGFRvIII have been described, and EGFRvIII is the deletion mutant of EGFR, and it is specifically expressed in tumor rather than on healthy tissue with the form of mutation.Full human EGFRvIII specific domains described herein were initially found in the phage display screening of the fully human scFv library using based on IgM source.Surprisingly, although two or three rounds of panning were carried out using exhaustion step (depletion procedure) to wild-type EGFR antigen, this should be enriched to have specific binding substances to EGFRvIII, but in ELISA, most of the scFvs of gained are all combined to two kinds of albumen (EGFRvIII and wild-type EGFR).Identified a high EGFRvIII specific sequence, called after Li3G30.Except that there is little difference in the N-terminal of VH and VL, the framework region of heavy chain and light chain and CDR1 and CDR2 are identical with sequence VH5-51 and VL3-25 of germline encoding respectively.Li3G30 was identified from the third round of liquid phase panning.

Affinity Maturation:

The library of antibody framework pair VH5-51/VL3-25 was designed using an algorithm ordered from DistributedBio. The specificity determining residues of Li3G30 were identified and included in the library design. The design includes 52 randomized CDR sites, each with a separately defined amino acid distribution. Three different loop lengths were built into the VL CDR3. Gene fragments encoding the randomized sites of VH and VL were ordered from Geneart/Lifetechnologies and synthesized using TRIM-technology. The fragments were cloned into the phage display vector pEXHAM (Schwarz et al. 2004) to a final library size of 3.7E+8 transformed E. coli cells. The library was packaged into phage particles and, through panning and screening steps, separated with improved affinity and retained variants specific for EGFRvIII. Panning was performed using the EGFRvIII-Fc antigen immobilized on a protein-bound plastic surface. The elutriation step is designed to the scFv that helps to have the EGFRvIII bonded with low dissociation rate ( _kOFF ) by implementing following steps: washing step comprises several washing buffer incubation steps that continue up to 30 minutes, to help select the scFv that slowly dissociates.Another step is based on the overnight competition with solubility EGFRvIII.In order to guarantee in the process of affinity maturation, can not select EGFR-EGFRvIII cross-reaction antibody or the antibody in conjunction with Fc, before first elutriation step, by the pre-incubation of wild-type EGFR-Fc, the library is exhausted possible wild-type EGFR binding substances.And owing to the existence of solubility EGFR in the process of the elutriation step being carried out to fixing EGFRvIII, screening scheme is unfavorable for cross-reaction binding substances.

The unique EGFRvIII-specific scFv was further tested for binding to transfected CHO and F98 cells expressing EGFRvIII using flow cytometry. ScFvs that showed binding to cells expressing wild-type EGFR and/or untransfected cells were excluded from further analysis. Initial _KD ranking using Biacore X100 was used to identify scFvs with improved affinity for EGFRvIII compared to the original clone, Li3G30. The best variants were characterized using detailed SPR measurements and fluorescence-based stability assays.

For VL CDR3, the initial library for affinity maturation comprises three kinds of different loop lengths, and known all lengths are compatible with VH5-51/VL3-25 framework.Surprisingly, in the selection process, the middle loop length is obviously preferred.Therefore, compared with the anti-EGFRvIII antibody of parent generation, the VL CDR3 of the variant of all affinity maturations has all shortened one amino acid (table 1, Fig. 1).The sequence of selected scFv is compared with the input library, and this analysis has shown 3 useful sudden changes in VL CDR3.The heavy chain of the anti-EGFRvIII antibody of parent generation and the CDR1 and CDR2 of light chain are identical with the sequence encoded separately of germline sequence VH5-51 and VL3-25.In 75% of the sites, the residue of each germline coding is reconstructed (reconstituted) or obviously preferred.Shown a useful sudden change in VH CDR1, one in VL CDR1 (Fig. 1).The amino acid distribution in the remaining randomization site is very close to the distribution in the library before selecting.

Example 2: Measurement of scFV binding to EGFRvIII and wild-type EGFR in Biacore:

Preparation of anti-hu Fc IgG CM5-chip:

Anti-hu Fc IgG CM5-chip was prepared by covalent amine coupling using an amine coupling kit (Amine-Coupling-Kit) (GE) according to the manufacturer's instructions. IgG was diluted to a concentration of 25 μg/ml in the provided immobilization buffer (10 mM sodium acetate, pH 5.0). For the coupling step, the predefined method "Amine" of the Biacore T200 control software was applied. A target level of 5000-10000 RU was achieved in all 4 flow cells. Before injecting the recombinant protein solution, the surface of the chip was activated with EDC/NHS. The remaining binding sites on the surface were blocked with ethanolamine. 1×HBS-P+ was selected as the running buffer for the entire coupling step.

Concentration range of single-chain Fv’s (scFv) used in SPR measurements:

Before SPR measurement, the ScFv concentration was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate the molar concentration of the His-tagged scFv antibody, a molecular weight of approximately 28 kD was assumed. The scFv was adjusted to the following final concentrations by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 430 nM, 86 nM, 17.2 nM, 3.44 nM, 0.688 nM, and 0 nM.

ScFv measurement conditions:

EGFRvIII-Fc and wild-type EGFR-Fc fusion proteins were diluted in 1× HBS-P+ to a concentration of 6.25 nM. Both antigens were injected at a flow rate of 10 μl/min: EGFRvIII-Fc for 75 seconds and wild-type EGFR-Fc for 90 seconds. EGFRvIII-Fc was injected into flow cell 2, and wild-type EGFR-Fc was injected into flow cell 4. Flow cells 1 and 3 were left blank.

1× HBS-P+ was used as running buffer throughout the procedure. The scFv antibody was injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. The signal measured in flow cell 3, where no antigen was captured, was subtracted from the signal in flow cell 4 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0M _MgCl2 (30 seconds, 10 μL/min). Before each measurement, four "start-up cycles" were performed without scFv. ScFv solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

The kinetics of the scFv antibodies were determined in the MCK (Multi Cycle Kinetic) measurement. To estimate the apparent binding affinity, the binding curves were evaluated using the "Kinetic" method included in the Biacore evaluation software. Using this method, all binding curves were fitted using the "Langmuir 1:1 interaction model" with the software. The results are shown in Table 3.

The affinity maturation screening step that is used is intended to improve the EGFRvIII binding affinity of EGFRvIII specific binding domain more easily by making the selection of the binding substances with the dissociation rate ( _kOFF ) of reduction.Yet, surprisingly, the affinity maturation binding substances (table 3) of gained shows the association rate ( _kON ) that significantly increases, and the reduction of _kOFF reaches 100 times of combination for the raising of scFv and only makes less contribution, and some binding substances demonstrate the _KD lower than 100pM simultaneously (table 3).Especially surprisingly, although the screening step is conducive to the scFv that selects the _kOFF reduction, the reduction of realization _KD is mainly due to the raising of _kON.This also is surprising, because the affinity maturation of antibody is relevant (Schier et al., 1996, Pini et al., 1998, Boder et al., 2000) with the reduction of _kOFF conventionally.

Table 3: Summary of scFv binding affinities ("VH" is the variable heavy chain domain of the scFv, "VL" is the variable light chain domain of the scFv). _K values, association rates (kON) and dissociation rates ( _kOFF ) of different EGFRvIII binding domains in scFv format for recombinant EGFRvIII-Fc antigen measured in a Biacore ( _SPR ) 1:1 binding model; "enhancement factor" refers to the fold change relative to the parental scFv, i.e., Li3G30 (VH SEQ ID NO: 25 and VL SEQ ID NO: 26).

Eight of the 11 affinity-matured scFvs showed at least a 5-fold improvement in _kON compared to the parental scFv; three of these had a 20-fold higher _kON . Compared to the improvement in _kON , the effect of affinity maturation on _kOFF was quite small. Only three affinity-matured scFvs showed a 4-5-fold improvement in _kOFF . Interestingly, two of these were isolated from a competitive selection method, which may be a more effective method for selecting scFvs with reduced _kOFF than prolonged incubation in wash buffer.

The average melting temperature of the selected affinity matured candidates determined by differential scanning fluorimetry (DSF) was 62.2°C, ranging from 53°C to 67.2°C.

Maintaining high specificity for EGFRvIII relative to wild-type EGFR during affinity maturation

Affinity maturation is usually accompanied by small changes in binding epitopes, which can affect specificity/cross-reactivity (Barbas et al, 1994, Parsons et al., 1996, Winkler et al., 2000). Due to the in-frame deletion of 269 amino acids in the wild-type EGFR extracellular domain (exon 2-7), an EGFRvIII new epitope relative to wild-type EGFR is formed, and it is predicted to be quite small: it is composed of new amino acids in parallel (amino acid 5 is fused to amino acid 274) and a new GLY amino acid at the deletion site. Even if it is expected that small changes in binding epitopes will quickly destroy the abnormal specificity of EGFRvIII in this case, a cross-reactive conjugate that can also identify native EGFR can be obtained.

Surprisingly, the affinity maturation described herein resulted in up to 100-fold improved binding of anti-EGFRvIII antibodies without loss of exclusive specificity for EGFRvIII (Table 3, Figure 2).

During the selection of affinity-matured binders, the binding of the only EGFRvIII-specific scFv to CHO cells expressing EGFRvIII (CHO ^EGFRvIII ) and F98 cells (F98 ^EGFRvIII ), as well as to CHO cells expressing human wild-type EGFR (CHO ^EGFR ) and F98 cells (F98 ^EGFR ), and to untransfected CHO and F98 cells were tested in flow cytometry.

Characterized in more detail 11 high-specificity anti-EGFRvIII scFv that do not have any measurable background combination with wild-type EGFR positive cell (CHO ^EGFR ) or the Chinese hamster ovary celI of non-transfected.In the best case, measured the avidity to 80pM of EGFRvIII antigen, corresponding to 100 times of raising coefficient (table 3) compared with parental scFv.Simultaneously, the anti-EGFRvIII scFv that is detected all does not demonstrate any cross reactivity (Fig. 2) with wild-type EGFR antigen.

To our knowledge, such high affinity for EGFRvIII has never been observed for a human antibody domain that binds exclusively specifically to EGFRvIII, without measurable cross-reactivity with wild-type EGFR. Safdari and colleagues recently described results for humanizing the murine EGFRvIII-binding antibody domain MR1 (Safdari et al., 2014), which was first described in its murine form by Lorimer and colleagues (Lorimer et al., 1996, Beers et al., 2000, Kuan et al., 2000). In their humanization approach, they achieved an increase in the affinity of humMR1, reaching similar picomolar _K values reported here. However, in contrast to the present invention, they did not completely eliminate the cross-reactivity of the antibody domain with wild-type EGFR, as can be clearly seen in Figure 4 of their publication (Safdari et al., 2014). Thus, after SDS PAGE and Western blotting, humMR1 was not only specific for EGFRvIII (145 kD), but also recognized EGFR (170 kD) (Safdari et al., 2014). Applicants extensively tested the specificity and cross-reactivity of the EGFRvIII binding domain of the present application before and after affinity maturation and surprisingly found no cross-reactivity signals with wild-type EGFR or purified recombinant wild-type EGFR antigen in Biacore (Figure 2), SDS PAGE and Western blotting (Figure 4), or ELISA (Figure 5). The EGFRvIII binding domain of the present application also did not exhibit any cross-reactivity with wild-type EGFR overexpressed on the cell surface of transfected CHO cells (CHO ^EGF ) or F98 glioma cells (F98 ^EGFR ) (Figure 3).

Example 3: High specificity and high affinity of EGFRvIII-specific scFv antibodies were assessed by flow cytometry for binding to F98 rat glioma cells (F98 ^EGFRvIII ) or stably transfected CHO cells (CHO ^EGFRvIII ) expressing EGFRvIII; no binding to F98 cells (F98 ^EGFR ) or untransfected F98 cells (F98) expressing native EGFR, and no binding to CHO cells (CHO ^EGFR ) or untransfected CHO cells (CHO) expressing native EGFR.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), Ficoll Paque PLUS (GE Healthcare), FITC-conjugated goat anti-human IgG (Dianova), FITC-conjugated goat anti-mouse IgG, min X (Dianova), L-glutamine (Invitrogen), _NaN3 (Carl Roth), PBS (Invitrogen), Penicillin/Streptomycin (Invitrogen), Propidium iodide (Sigma), RPMI-1640 (Invitrogen).

Cells and cell lines:

F98 rat glioma cells, F98 cells overexpressing EGFR (F98 ^EGFR ), and F98 cells overexpressing EGFRvIII (F98 ^EGFRvIII ) were purchased from the American type culture collection (ATCC) and cultured according to the recommended protocol.

F98 (ATCC CRL-2397)

F98 ^EGFR (ATCC CRL-2948)

F98 ^npEGFRvIII (ATCC CRL-2949)

Stably transfected CHO cells expressing recombinant EGFR or EGFRvIII were generated at Affimed according to the following protocol:

The gene of coding wild-type EGFR or deletion mutant EGFRvIII is synthesized by Life Technologies/GeneArt (Regensburg, Germany), and is subcloned among the mammalian expression vector pcDNA5/FRT.Expressing the stable Chinese hamster ovary celI of recombinant wild-type EGFR or EGFRvIII generates based on the host cell line Flp-In CHO (Life Technologies) of previous adaptation to suspension growth.The day before transfection, Flp-In Chinese hamster ovary celI is carried out passage culture (not having bleomycin (Zeocin)) in the HyClone CDM4CHO that is supplemented with L-glutamine, HT supplement and penicillin/streptomycin.The stable expression cell line is by using polyethyleneimine (PEI) transfection reagent, with the product expression based on pcDNA5/FRT and integration plasmid and Flp recombinase expression plasmid pOG44 (Life Technologies) cotransfection Flp-In Chinese hamster ovary celI system generates. 2.5 μg of total DNA was diluted in 50 μL of OptiMEMI medium and combined with 7.5 μg of PEI diluted in 50 μL of OptiMEMI medium. The mixture was incubated for 10 minutes and then added to 2×10 ⁶ Flp-In CHO cells suspended in 1 mL of CHO-S-SFMII medium. One day after transfection, the cells were diluted to a density of 1×10 ⁵ viable cells/mL in CHO-S-SFMII medium supplemented with 500 μg/mL hygromycin B and seeded in T75 culture flasks. Flp recombinase mediates the insertion of the Flp-In expression construct into the genome at the integrated FRT site through site-specific DNA recombination. During the selection phase, the viable cell density was measured once or twice a week, the cells were centrifuged and resuspended in fresh selection medium at a maximum density of 1×10 ⁵ viable cells/mL. After about 2-3 weeks of selection using 500 μg/mL hygromycin B, stable cells expressing EGFR or EGFRvIII were recovered and then transferred to HyClone CDM4 CHO suspension culture medium. Stable cells expressing wild-type EGFR or EGFRvIII were cryopreserved in a culture medium containing 50% ProFreeze (Lonza) and 7.5% DMSO.

To determine the density of wild-type EGFR or EGFRvIII cell surface antigens on stably transfected cell lines, a flow cytometric indirect immunofluorescence assay (QIFIKIT, Dako) was used according to the manufacturer's instructions and demonstrated that the density of wild-type EGFR on CHO ^EGFR stably transfected cells was approximately equal to the density of EGFRvIII on CHO ^EGFRvIII stably transfected cells, whereas no wild-type EGFR or EGFRvIII binding sites were detected on untransfected CHO cells.

Determination of Antibody Binding and Affinity by Flow Cytometry:

Cells and indicated scFv antibodies are started from 100 μ g/mL (about 3300nM) in 100 μ L serial dilutions in FACS buffer and are hatched on ice for 45min (minute). After washing 3 times with FACS buffer, cells are hatched on ice for 45min with 0.1mL 10 μ g/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After second washing cycle, under the same conditions as before, cells are hatched with the goat anti-mouse IgG antibody that 0.1mL 15 μ g/mL FITC is put together. As a control, in the absence of scFv, cells are hatched with anti-His IgG 13/45/31, then with the goat anti-mouse IgG antibody that FITC is put together. Then cells are washed again and are resuspended in the 0.2mL FACS buffer containing 2 μ g/mL propidium iodide to exclude dead cells. The fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer with MXP software (Beckman-Coulter, Krefeld, Germany) or a Millipore Guava EasyCyte flow cytometer with Incyte software (Merck Millipore, Schwalbach, Germany). The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany).

If analyzing with a Beckman-Coulter FC500 MPL flow cytometer, use 0.5 × 10 ⁶ cells/stain; if analyzing with a Millipore Guava EasyCyte flow cytometer, use only 0.25 × 10 ⁶ cells/stain.

KD values were calculated using the equation for single-site binding (hyperbola) using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) after subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone.

The results are shown in Figure 3.

Example 4: SDS PAGE and Western blotting analysis of the binding specificity of antibodies binding to EGFRvIII.

To detect the binding specificity of EGFRvIII/CD3 tandem diabodies containing different binding domains to EGFRvIII antigen and to demonstrate the absence of binding to wild-type EGFR antigen, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis were performed.

By recombinant DNA technology, the sequence of the extracellular domain of encoding wild-type EGFR or truncated EGFRvIII is fused to the Fc part of human IgG antibody.DNA construct is transfected into the Chinese hamster ovary celI as the wild-type EGFR-Fc or EGFRvIII-Fc of the reorganization of soluble protein and is secreted into the cell culture supernatant, uses protein A chromatography to purify the wild-type EGFR-Fc or EGFRvIII-Fc of described reorganization from described cell culture supernatant.Through the formation of intramolecular disulfide bond in the Fc part, wild-type EGFR-Fc (or EGFRvIII-Fc) fused antigen forms the dimer with two identical chains.Due to EGFRvIII-specific 267 amino acid deletions, EGFRvIII-Fc causes the Fc fusion protein of dimer form to have the size difference of about 50kD than EGFR-Fc little by about 25kD.

Equal amounts of purified wild-type EGFR-Fc (and EGFRvIII-Fc) were mixed with either non-reducing 2× SDS PAGE sample buffer or reducing 2× SDS PAGE sample buffer containing dithiothreitol (DTT) as a reducing agent. Samples containing DTT were heated at 95°C for 5 minutes and then loaded onto a 4-20% standard TGX precast SDS PAGE gel (Biorad). 1 μg of protein sample was used per lane. To separate the proteins in the gel, SDS-PAGE was run at 300V in 1× Tris/glycine/SDS buffer (Biorad) for approximately 25 minutes. Total protein was visualized in the gel using a standard stain-free molecular imaging system (Biorad). A Page Ruler unstained protein ladder (Thermo Scientific) was used as a molecular weight marker. The gel was then blotted onto a PVDF membrane using BioRad's semi-dry blotting Fastblot system. The membrane was blocked in 3% (w/v) nonfat dry milk in 1× TBS for 30 min at room temperature. The membrane was cut into small pieces, each containing a similar protein sample to be blotted. Different tandem diabodies were diluted to a concentration of 2 μg/ml in 3% (w/v) nonfat dry milk in 1× TBS and incubated with separate membrane pieces on a rocking platform for 1 hour. The anti-EGFR antibody cetuximab was used as a control at a concentration of 10 μg/ml. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS, then incubated with HRP-conjugated secondary detection antibodies PentaHIS-HRP (QIAGEN) (for detection of the tandem diabodies) or Protein L-HRP (Thermo Scientific) (for detection of cetuximab) at a dilution of 1:5000 in 3% nonfat dry milk in TBS on a rocking platform for 1 hour. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS. HRP-mediated color development on the membrane was initiated by adding 0.06% DAB + 0.02% CoCl ₂ + 0.015% H ₂ O ₂ in TBS. The reaction was terminated by adding water. The membrane was dried and scanned.

The results are shown in Figure 4.

Example 5: The binding of EGFRvIII/CD3 tandem diabodies containing different EGFRvIII binding domains to EGFRvIII antigen and CD3, as well as their binding specificity to EGFRvIII in the absence of binding to wild-type EGFR antigen, were analyzed by enzyme-linked immunosorbent assay (ELISA).

ELISA steps:

96-well ELISA plates (Immuno MaxiSorp; Nunc) were coated overnight at 4°C with EGFRvIII-Fc, wild-type EGFR-Fc, or CD3γε recombinant antigen in 100 mM carbonate-bicarbonate buffer. To achieve approximately equal antigen molar coating densities, EGFRvIII-Fc was coated at a concentration of 4 μg/ml, EGFR-Fc at a concentration of 6 μg/ml, and CD3γε at a concentration of 1.5 μg/ml. After a blocking step using 3% (w/v) skim milk powder (Merck) in PBS, 11 serial dilutions of different EGFRvIH/CD3 tandem diabodies ranging from 200 ng/μl to 6.5× ^10-6 ng/μl in PBS containing 0.3% (w/v) skim milk powder (Merck) were incubated on the plates at 25°C for 1.5 h. After incubation, the plates were washed three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20 (Serva). 50 ng/ml of Protein L-HRP was added and incubated on the plates at 25°C for 1 hour. After washing three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20, tetramethylbenzidine (TMB) substrate (Seramun) was added for detection. The reaction was terminated after approximately 2 minutes by adding 100 μl/well of 0.5 _{M H₂SO₄} . The absorbance of the wells was measured at 450 nm using a multiwell plate reader (Victor, Perkin Elmer).

Absorbance values were plotted as lines in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada).

The results are shown in FIG5 .

Example 6: Measurement of Binding of Tandem Diabodies to EGFRvIII by SPR:

Preparation of anti-hu Fc IgG CM5-chip

An anti-hu Fc IgG CM5-chip was prepared as described above (Example 2).

Concentration range of tandem diabodies used in SPR measurements:

Before SPR measurement, the concentration of the tandem diabody was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate molar concentrations, a molecular weight of 105 kD was assumed for all tandem diabodies. The tandem diabodies were adjusted to the final concentrations mentioned below by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 150 nM, 30 nM, 6 nM, 1.2 nM, 0.24 nM, 0.048 nM, and 0 nM; or 90 nM, 30 nM, 10 nM, 3.33 nM, 1.11 nM, 0.37 nM, 0.123 nM, 0.041 nM, and 0 nM, or the concentrations indicated.

Binding assay conditions for tandem diabody KD measurements

Dilute the EGFRvIII-Fc fusion protein in 1×HBS-P+ to a concentration of 6.25 nM. For EGFRvIII-Fc, inject the antigen solution at a flow rate of 10 μl/min for 40 seconds. Inject the antigen into flow cell 2. Flow cell 1 serves as the reference cell.

1× HBS-P+ was used as running buffer throughout the procedure. Serial dilutions of the diabody were injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0 M _MgCl2 (3 x 15 seconds, 10 μL/min). Before each measurement, three "priming cycles" were performed without the tandem diabody. The tandem diabody solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

Binding parameters of the tandem diabodies were determined in an MCK (Multi Cycle Kinetic) measurement. To estimate apparent binding affinity, binding curves were evaluated using the "Kinetic" method included in the Biacore X100 evaluation software. Using this method, binding curves were fitted in whole or in part using the "Langmuir 1:1 interaction model" with the software.

Binding parameters of tandem diabodies

To analyze the effect of combining two EGFRvIII binding domains into bivalent or multivalent or multispecific molecules such as EGFRvIII/CD3 tandem diabodies, the apparent affinity of EGFRvIII/CD3 tandem diabodies to EGFRvIII antigen was measured by SPR using BiacoreX100 ( FIG6 ).

For all molecules detected, the apparent affinity of the EGFRvIII specific binding domain in the EGFRvIII/tandem diabody improves, and the best combination tandem diabody has achieved a _K of 11pM. Relative to the corresponding tandem diabody containing the parental EGFRvIII binding domain, the _K of the tandem diabody containing the domain of affinity maturation reduces by up to 25 times. Relative to the _K of the measurement of the monovalent binding scFv, the improvement of _K is also up to 25 times. Interestingly, contrary to the improvement achieved by the affinity maturation driven mainly by the improved k _ON , the further improvement of binding affinity in the bivalent tandem diabody format is realized by slower k _ofOFF to a great extent (Table 4).

Therefore, the tandem diabody format provides a way to further enhance binding affinity while maintaining the specificity of the antibody binding domain.

Table 4: Summary of measured KD, association rates (kON), and dissociation rates ^{( kOFF} ) for binding of different EGFRvIII/ ^CD3 tandem diabodies containing bivalent parental or affinity matured ^EGFRvIII -specific binding domains with the domain order _{VLCD3 - VHEGFRvIII} - _VLEGFRVIII - _VHCD3 to recombinant EGFRvIII-Fc antigen in the SPR _{1:1 binding} model.

Example 7: Immunohistochemistry (IHC) staining of solid tumor tissue sections using anti-EGFRvIII bispecific diabody or cetuximab

In contrast to wild-type EGFR, which is ubiquitously expressed on healthy tissues, expression of the mutant receptor EGFRvIII is highly tumor-specific. However, the high frequency of EGFRvIII expression in glioma cells is consistent with literature descriptions, but the extent of EGFRvIII expression and its homogeneity in other tumors remain controversial.

A small immunohistochemistry (IHC) study was initiated, analyzing the reactivity of tissue sections from three cancer patients each suffering from glioblastoma (GB), head and neck (H&N) cancer, prostate cancer, Her2-negative breast cancer, and non-small cell lung cancer (NSCLC) with the EGFRvIII specific binding domains in the form of a bivalent diabody according to the present invention ( FIG. 7 ).

The present invention relates to the immunohistochemical study of the expression of EGFRvIII and the tumor specificity of EGFRvIII binding antibodies to assess different tumors.To contain the bivalent double antibody form of the EGFRvIII binding domain Li3G30 arranged with V _H ^EGFRvIII - _{V L} ^EGFRvIII , the EGFRvIII combination of EGFRvIII specific variable domains has been detected.Secondary detection is implemented by the His tag on the double antibody protein.In contrast, used natural EGFR to have specific IgG antibody cetuximab (Erbitux (Erbitux)).

All people's tissue sections are purchased from BioChain Institute, Inc., and according to supplier's specification sheets process section.Make tissue section be adapted to room temperature and hatch with DAKO peroxidase blocker, then use 10% goat serum to seal.EGFRvIII specific double antibody was hatched 1 hour with two concentration levels of 0.5 μ g/ml and 0.1 μ g/ml, then hatched and passed through Envision+HRP-DAB system detection mouse antibody (DAKO) with anti-His antibody (Dianova).According to manufacturers' specification sheets, use Klear people HRP-polymer DAB detection system (GBI Labs), implement the tissue section dyeing that carries out with control item Erbitux (Merck KGaA) with the concentration of 10g/ml.

Finally, sections were stained with hematoxylin and mounted with mountant medium (Shandon ConsulMount ^™ , Thermo Scientific).

The results are shown in FIG7 .

Two demonstrate in three glioblastoma samples with the obvious and specific reactivity of EGFRvIII specific double antibody.Surprisingly, all three head and neck cancer samples all demonstrate strong and obvious EGFRvIII positive, and similarly for prostate cancer, mammary cancer and NSCLC, obtained the specific staining (Fig. 7) of tumor tissue with EGFRvIII specific antibody described herein.

Based on these properties, the EGFRvIII binding domains described herein, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies, are excellently suited for developing multispecific, multivalent, immune effector cell engaging, tumor targeting antibody therapies.

Example 8: Evaluation of cytotoxic activity mediated by EGFRvIII/CD3 tandem diabodies in a FACS-based cytotoxicity assay

In the cytotoxicity assay based on cell, analyzed EGFRvIII/CD3 series connection diabody antibody, it contains the parental generation that is combined with different CD3 binding domains or the EGFRvIII specificity domain of affinity maturation and/or each binding domain (Fig. 8) with different orders in described series connection diabody molecule.By using the Chinese hamster ovary celI (CHO ^EGFRvIII ) of expressing EGFRvIII, the Chinese hamster ovary celI (CHO ^EGFR ) of expressing wild-type EGFR and not transfected Chinese hamster ovary celI as target cell to analyze the specificity of target-mediated, series connection diabody and the lethality of T cell mediation.

Materials (media, buffers, and reagents):

DMSO (Sigma), EasySep ^™ Human T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD4+ T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD8+ T Cell Enrichment Kit (Stem Cell Technologies), The Big Easy EasySep ^™ Magnet (Stem Cell Technologies), FCS (Invitrogen), Lymphoprep (Stem Cell), L-glutamine (Invitrogen), mlgGl FITC (ADG), CD16-FITC (MEM-154) (Thermo Fisher Scientific), mlgGl-FITC/mlgGl-PE/mlgGl-ECD (Beckman Coulter), mlgGl-PE (Beckman Coulter), mlgGl-PC5 (Beckman Coulter), mlgGl-PC7 (Beckman Coulter), CD8-FITC/CD4-PE/CD3-ECD (Beckman Coulter), CD16-PC5 (Beckman Coulter), CD19-PC7 (Beckman Coulter), CD16-FITC/CD56-PE/CD3-ECD (Beckman Coulter), CD14-PC7 (Beckman Coulter), CD33-PE (MACS Miltenyi Biotech), NaN ₃ (Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), PKH67 green fluorescent cell labeling (Linker) Midi kit (Sigma), Gammanorm human IgG (Octapharma), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

Isolation of PBMCs from buffy coats and enrichment of T cells:

On the day of blood draw, buy buffy coat from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany); Buffy coat preparation is stored at room temperature overnight, and then PBMC is separated by density gradient centrifugation. Buffy coat sample is diluted with two to three volumes of PBS, layered on Lymphoprep cushion and centrifuged at 800 × g for 25 min at room temperature w/o brake. PBMC at the interface is collected and washed 3 times with PBS, and then they are used for enrichment of T cells or flow cytometric analysis. According to the manufacturer's instructions, EasySep ^™ human T cell enrichment kit (for immunomagnetic separation of untouched human T cells) and Big Easy EasySep ^™ magnet are used to enrich T cells from the PBMC group.

Characterization of effector cells by flow cytometry:

To assess the distribution of PBMC subpopulations and the purity of enriched PBMC or T cell subsets, PBMC staining was performed.

For flow cytometric analysis, isolated PBMCs or enriched PBMC subpopulations were resuspended in PBS supplemented with 2% heat-inactivated FCS and 0.1% sodium azide (referred to as FACS buffer) and 1 mg/mL polyclonal human IgG to a density of 0.5-2 × 10 ⁶ /mL. 0.5 ml aliquots of the cell suspension were then incubated with the antibody for 15 minutes in the dark according to the manufacturer's recommended protocol.

Table 5: Pipetting scheme for characterization of PBMCs and fractionated PBMCs by flow cytometry

After incubation, samples were measured on a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) in the order 3, 4, 5, 6, 1, 2 using a protocol developed for PBMC and multicolor staining (FITC PEECD PC5 PC7; FITC PE ECD; FITC PE PC5PC7) in MXP acquisition software (Beckman-Coulter). For analysis and data plotting, CXP analysis software (Beckman-Coulter) was used.

Cytotoxicity assay:

As previously described, T cells used as effector cells were characterized by flow cytometry. The target cells used in the cytotoxicity assay were CHO cells stably transfected to overexpress EGFRvIII (CHO ^EGFRv111 ), CHO cells stably transfected to express wild-type EGFR (CHO ^EGFR ), or untransfected CHO cells (CHO). Cell lines were generated as previously described and cultured under standard conditions as previously described. For the cytotoxicity assay, target cells were collected, washed twice with RPMI 1640 medium without FCS, and resuspended in diluent C provided by the PKH67 green fluorescent cell labeling Midi kit to a density of 2×10 ⁷ /mL. The cell suspension was then mixed with an equal volume of double-concentrated PKH67-labeling solution (e.g., 1 μL PKH in 250 μL diluent C) according to the manufacturer's instructions and incubated at room temperature for 2-5 minutes with regular mixing. The staining reaction was terminated by adding an equal volume of FCS and incubating for 1 minute. After washing the labeled target cells with complete RPMI medium, the cells were counted and resuspended in complete RPMI medium to a density of 2×10 ⁵ /mL.

2×10 ⁴ target cells were then seeded with T cells at a 5:1 E:T ratio in each well of a round-bottom 96-well microplate along with the indicated antibodies in a total volume of 200 μL/well. Nine serial 1:5 dilutions, starting at 30 μg/mL, were typically tested. Natural cell death in the absence of antibody and target killing by effectors were determined in at least three replicates on each plate. Tandem diabody-mediated killing was typically determined in duplicate.

After centrifugation at 200 g for 2 minutes at room temperature, the assay plate was incubated at 37° C. in a humidified atmosphere containing 5% CO ₂ for 40-48 hours. After incubation, the culture was washed once with FACS buffer and then resuspended in 150 μL FACS buffer supplemented with 2 μg/mL PI. The absolute number of viable target cells, characterized by positive green PKH67 staining but negative PI staining, was measured using a Millipore Guava EasyCyte flow cytometer (Merck Millipore).

Based on the remaining live target cells measured, the percentage of specific cell lysis was calculated according to the following formula: [1-(number of live target cells _(sample) )/(number of live target cells _(native) )]×100%. Sigmoidal dose response curves and EC50 values were calculated by nonlinear regression/4-parameter logistic fit using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, California, USA).

Data Analysis:

Lysis values obtained for a given antibody concentration were determined in duplicate and analyzed by Sigmoidal dose-response/4-parameter logistic fit analysis using Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) and used to calculate the mean EC50 and SD of the replicates for percentage lysis.

The results are shown in FIG8 and Table 6.

The tandem diabody of all detections all demonstrates the concentration dependency to the cell of expressing EGFRvIII, high specificity, cytotoxicity and high efficiency, and the survival of EGFRvIII negative cells is not impaired.Within the whole concentration range of any tandem diabody of expressing EGFRwt-cell or non-transfected cell detection, do not observe detectable cytotoxicity.Therefore, the cytotoxicity of EGFRvIII/CD3 tandem diabody mediation is strictly target-dependent.In cytotoxicity assay, the EC 50 of the tandem diabody that contains parental EGFRvIII binding domain before affinity maturation and domain order is V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 is _25pM (Fig. 8).Corresponding tandem diabody contains the EGFRvIII binding domain that avidity improves with identical domain order (V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 ), shows enhanced cytotoxicity efficacy equally, and in the best case, reaches the EC ₅₀ (table 6) of 1.5pM. The relative increase in cytotoxic activity is more evident in the tandem diabodies with EGFRvIII specific domains located in the external position (domain order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ) (Table 6). In these tandem diabodies, up to 70-fold increased cytotoxic potency can be achieved by replacing the low-affinity EGFRvIII binding domain Li3G30 with an affinity-enhanced domain in these tandem diabodies (Figure 8, Table 6).

Table 6: to the summary of the EC50 value measured with different EGFRvIII/CD3 series connection diabodies in the cytotoxicity assay, described different EGFRvIII/CD3 series connection diabodies contain the EGFRvIII specific binding domain of affinity maturation with two different domain orders.Shown the raising coefficient relative to the series connection diabodies that contain the parental generation EGFRvIII binding domain of correspondence.

Choi et al. (Proc Natl Acad Sci USA. 2013 Jan 2; 1 10 (1): 270-5; WO2013/7185010) describes a t EGFRvIII/CD3 comparator, an EGFRvIII/CD3 bispecific T-cell engager (BiTE), which has an EGFRvIII binding site (based on the mouse MR1 domain) and a CD3 binding site (based on the anti-CD3 antibody OKT3). In a cell-based cytotoxicity assay, the antibody achieved 50% of the maximum specific lysis achieved at a concentration of about ^10-2 μg/ml, which corresponds to a molar EC50 concentration of about 200 pM for the 52kD BiTE (Figure 4C of Choi et al., 2013). Comparisons show that the cytotoxicity of the tandem diabodies described herein containing the parental, rather than affinity-matured, EGFRvIII binding domains is even about 10-fold higher than that of the EGFRvIII/CD3 BiTE. Furthermore, the tandem diabody containing the affinity-matured EGFRvIII binding domain of the present invention even exhibited 100-fold higher cytotoxicity than that of EGFRvIII/CD3 BiTE.

Example 9: EGFRvIII receptor internalization induced by EGFRvIII/CD3 tandem diabody on CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII )

To evaluate EGFRvIII regulation by EGFRvIII/CD3 tandem diabodies, CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII ) were seeded in aliquots of 2.5×10 ⁵ cells per well of a round-bottom 96-well microplate and incubated with increasing concentrations of EGFRvIII/CD3 tandem diabodies in RPMI 1640 medium supplemented with 10% FCS, 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, and 100 μg/mL sodium pyruvate (referred to herein as complete RPMI medium; all components purchased from Invitrogen) at 37° C., 5% CO ₂ , in a humidified incubator for 24 hours. Aliquots of cells were cultured in the absence of antibody to serve as control samples for detection of the highest EGFRvIII cell surface levels and as negative controls for staining performed with the secondary reagent alone.

44.After hatching, cell is washed 2 times with ice-cold phosphate buffered saline (PBS, Invitrogen) (being called FACS buffer in this article) that is supplemented with 2% heat inactivation FCS (Invitrogen) and 0.1% sodium azide (Roth, Karlsruhe, Germany), then with the antibody (this antibody is identical with the antibody that hatches with cell during adjustment incubation) of saturation concentration (10 μ g/mL) dyeing to assess the maximum antibody combination corresponding to the remaining EGFRvIII antigen on the cell surface.The aliquots of the cell that is cultivated under the situation that there is no antibody are dyed with the same antibody of saturation concentration, to measure the maximum antibody binding capacity corresponding to the maximum measurable EGFRvIII level on the cell surface.After washing 3 times with FACS buffer, with series connection double antibody treatment cell and dye with the anti-His mAb (Dia910, Dianova) of 10 μ g/mL, then with the goat anti-mouse IgG (Dianova) dyeing that puts together with 15 μ g/mL FITC. After repeated washing with FACS buffer, cells were resuspended in FACS buffer containing 2 μg/mL propidium iodide (Sigma) to exclude dead cells. Approximately 5 × 10 ³ propidium iodide-negative cells from each sample were analyzed using MXP software and an FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany). The mean fluorescence intensity (MFI) of the samples was determined using CXP software (Beckman-Coulter) or cells were analyzed using a Millipore Guava EasyCyte flow cytometer using Incyte software (Merck Millipore, Schwalbach, Germany).

After subtracting the signal obtained from cell staining with the secondary reagent alone, the mean fluorescence intensity of the measured samples was plotted as a line in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada), and the relative amount of EGFRvIII remaining on the cell surface was calculated using the fluorescence signal obtained by staining untreated cell samples at 37°C in the absence of the EGFRvIII/CD3 tandem diabody, which was set to represent 100% of cell surface EGFRvIII.

The results are shown in FIG9 .

Some bibliographic reports have described and compared deletion mutant EGFRvIII and demonstrated impaired internalization (Fenstermaker et al.2000, Han et al.2006, Grandal et al.2007, Gan et al.2009) with natural EGFR.However, about EGFRvIII specific antibody or for several disclosed reports of the antibody (for example cetuximab) also in conjunction with the wild-type EGFR of EGFRvIII, described in conjunction with rapid internalization and the degraded (Reist et al.1995, Foulon et al.2000, Kuan et al.2000, Patel et al.2007, Jutten et al.2009, Dreier et al.2012) of EGFRvIII in the cell that expresses EGFRvIII after antibody separately.

For internalization determination, use the transfection CHO ^EGFRvIII cell of overexpressing EGFRvIII.The cells of these expression EGFRvIII and the EGFRvIII/CD3 series connection double antibody of varying concentrations were hatched 24 hours at 37 ℃, described the condition in the context of other EGFRvIII binding antibodies to induce strong internalization.After this adjustment (during which antibody-receptor complex can internalize or not internalize), give all remaining EGFRvIII receptor molecules dyeing (Fig. 9) on the cell surface with the saturation concentration of 10 μ g/ml with each antibody.The result that Fig. 9 presents shows, after cell is exposed to EGFRvIII-in conjunction with series connection double antibody, under all detection conditions, the EGFRvIII receptor molecules more than 80% and up to 140% existing on untreated cell are still available (140% or more>100% general value refers to and reduces (cellular internalization) with the number of receptor on the cell surface on the contrary, and the cell processed with series connection double antibody demonstrates the increase of receptor on the cell surface.Internalization to 100% comes from the contrast with untreated cell). These results show, EGFRvIII/CD3 series connection diabody of the present invention surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may even suppress internalization and cause EGFRvIII receptor levels on the cell surface to increase.The result that Fig. 9 presents shows, with respect to untreated cell, under all detection conditions, after cell is exposed to EGFRvIII-in conjunction with antibody, at least more than 80% still remain on the cell surface.These results show, EGFRvIII/CD3 antibody of the present invention, especially EGFRvIII/CD3 series connection diabody surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may suppress internalization or promote EGFRvIII to express and increase, cause its cell surface density to increase.

Example 10: Evaluation of cell proliferation in PBMC culture in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies

To evaluate whether EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies induce T cell activation and subsequent proliferation in the absence of EGFRvIII ⁺ target cells, human PBMCs were cultured in the presence of increasing concentrations of EGFRvIII/CD3 tandem diabodies. After 5 days of incubation, proliferation was assessed in a BrdU incorporation assay.

Materials (culture media, buffers, and reagents)

FCS (Invitrogen), human serum (Sigma), Lymphoprep (Stemcell Technologies), L-glutamine (Invitrogen), OKT3 (Biolegend), PBS (Invitrogen), β-mercaptoethanol (Invitrogen), penicillin/streptomycin (Invitrogen), RPMI-1640 (Invitrogen), sodium pyruvate (Invitrogen), BrdU cell proliferation ELISA (Roche).

cell:

Buffy coats were purchased from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany) on the day of blood draw; buffy coat preparations were stored at room temperature overnight before PBMC isolation.

Isolation of PBMCs from buffy coats and enrichment of T cells:

PBMCs were isolated from the buffy coat by density gradient centrifugation. The buffy coat sample was diluted with two to three volumes of PBS, layered onto a Lymphoprep cushion, and centrifuged at 800 × g for 25 minutes at room temperature with or without braking. PBMCs at the interface were collected and washed three times with PBS before use in proliferation assays.

Proliferation assay

All assay plates were blocked for 2 hours at room temperature with RPMI1640 medium supplemented with 10% heat-inactivated human serum, 4 mM L-glutamine, 100 U/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, 1 mM sodium pyruvate, and 0.05 mM β-mercaptoethanol (referred to herein as complete RPMI medium) to prevent nonspecific binding of antibodies to plastic surfaces. After removing the blocking medium, 4 × 10 ⁵ PBMCs were seeded in each well of a flat-bottom 96-well microplate with the indicated concentrations of test and control items in complete RPMI medium in a total volume of 200 μL/well. Each sample was assayed in triplicate. IgG anti-CD3 clone OKT3 was used as a positive control. To assess spontaneous proliferation, cells were cultured in the absence of antibody in six replicates. The plates were then incubated at 37°C and 5% CO ₂ in a humidified atmosphere for 4 days. 18 hours before termination of incubation, the cell cultures were pulsed with 100 μM BrdU. BrdU incorporation was quantified using a BrdU proliferation ELISA kit according to the manufacturer's instructions. Color development was stopped by adding 100 mM _{H₂SO₄ ,} and absorbance was measured at 450 nm using a multilabel plate reader (Victor 3, Perkin Elmer). Absorbance values were analyzed and the mean and SD were plotted using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA).

The results are shown in FIG10 .

Example 11: Correlation between CD3 Binding Affinity and Cytotoxicity of EGFRvIII/CD3 Tandem Diabody

Constructed and expressed EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) and different CD3 binding domains (which have some affinity for CD3 T cell antigens) and measured binding to CD3 ⁺ Jurkat cells and EGFRvIII ⁺ CHO ^EGFRvIII cells; In addition, cytotoxicity assays were performed using CHO ^EGFRvIII cells as target cells and PBMC cells as effector cells. The order of each VH and VL (heavy and light chain) domain of all tested tandem diabodies tested in this example was the same: _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 .

Antibody binding to EGFRvIII ⁺ CHO ^EGFRvIII cells and cytotoxic activity mediated by the EGFRvIII/CD3 tandem diabody were evaluated by flow cytometry as described in previous examples.

After staining CD3 ⁺ Jurkat cells with increasing concentrations of the tandem diabody, the apparent affinity of the EGFRvIII/CD3 tandem diabody for CD3 ⁺ cells was determined by flow cytometry. KD values were calculated by nonlinear regression/hyperbola using the mean fluorescence values for each tandem diabody concentration determined by flow cytometry.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), FITC-conjugated goat anti-mouse IgGmin X (Dianova), L-glutamine (Invitrogen), NaN ₃ (Carl Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

cell:

Jurkat cells (kindly provided by Dr. G. Moldenhauer (DKFZ, Heidelberg, Germany)) were cultured in RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, and 100 μg/mL streptomycin sulfate (all components were from Invitrogen, referred to herein as complete RPMI 1640 medium).

Determination of antibody affinity to CD3+ cells by flow cytometry:

Cells were incubated on ice for 45 min with 100 μL serial dilutions of the indicated antibodies starting from 100 μg/mL (about 1000 nM) in ice-cold PBS (referred to as FACS buffer) supplemented with 2% heat-inactivated FCS and 0.1% sodium azide. After washing 3 times with FACS buffer, cells were incubated on ice for 45 min with 0.1 mL 10 μg/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After the second wash cycle, under the same conditions as before, cells were incubated with 0.1 mL 15 μg/mL FITC-conjugated goat anti-mouse IgG antibody. As a control, in the absence of a primary antibody, cells were incubated with anti-His IgG 13/45/31 and then with FITC-conjugated goat anti-mouse IgG antibody. The cells were then washed again and resuspended in 0.2 mL FACS buffer containing 2 μg/mL propidium iodide to exclude dead cells. Fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) with MXP software or a Millipore Guava EasyCyte flow cytometer (Merck Millipore, Schwalbach, Germany) with Incyte software. The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany). If the Beckman-Coulter FC500 MPL flow cytometer was used for analysis, 0.5×10 ⁶ cells/staining were used, and if the Millipore Guava EasyCyte flow cytometer was used, only 0.25×10 ⁶ cells/staining were used.

After subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone, the fluorescence intensity values and the equation (hyperbola) for single-site binding of GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) were used to calculate KD values.

For the series connection diabody of each analysis, the KD value in conjunction with EGFRvIII and cytotoxic EC50 are drawn into the function of the KD value in conjunction with CD3.When the KD value in conjunction with EGFRvIII of series connection diabody only demonstrated very little variability, the EC50 cytotoxicity value demonstrated and was accompanied by the CD3 that series connection diabody increases gradually in conjunction with KD value and almost became linear increase (Figure 11).

Example 12: Inhibition of F98 ^EGFRvIII Xenograft Tumors by EGFRvIII/CD3 (In Vivo Proof of Concept)

On day 0 (d0), nine experimental groups of immunodeficient NOD/scid mice (NOD/MrkBomTac-Prkdcscid, Taconic Denmark) were xenografted with a suspension of 4× ¹⁰ F98 ^npEGFRvIII cells mixed with 1× ¹⁰ purified human PBMCs from healthy donors via subcutaneous (sc) injection. To account for possible donor variability of PBMCs, each experimental group was subdivided into two cohorts, each receiving PBMCs from only one individual donor. 2 h after tumor cell inoculation, followed by 24 h, 48 h, 72 h, and 96 h, 100 μg, 10 μg, or 1 μg of each tandem double antibody test item was administered to the animals via the tail vein (iv). The control group received vehicle only. Starting on d0, another control was administered (ip) with 1 mg of cetuximab (Erbitux) twice a week. Table 6 summarizes the dose groups and donor allocation. Tumor volume was monitored by measuring the major and minor diameters of the tumors with calipers three times a week from day 3 to day 52. Tumor volume was calculated from the diameters according to the following formula: Volume = (minor diameter) ² x major diameter x 0.5.

Table 7: Animal dose groups

References:

Schwarz M, Rottgen P, Takada Y, Le Gall F, Knackmuss S, Bassler N, Buttner C, Little M, Bode C, Peter. Single-chain antibodies for the conformation-specific blockade of activated platelet integrin alphallbbeta3 designed by subtractive selection from naive human phage libraries. FASEB J. 2004 Nov; 18(14): 1704-6.

Schier R, McCall A, Adams GP, Marshall KW, Merritt H, Yim M, Crawford RS, Weiner LM, Marks C, Marks JD. Isolation of picomolar affinity anti-c-erbB-2 single-chain Fv by molecular evolution of the complementarity determining regions in the center of the antibody binding site. J Mol Biol. 1996.

Pini A, Viti F, Santucci A, Carnemolla B, Zardi L, Neri P, Neri D. Design and use of a phage display library. Human antibodies with subnanomolar affinity against a marker of angiogenesis eluted from a two-dimensional gel. J Biol Chem. 1998.

Boder ET, Midelfort S, Wittrup KD. Directed evolution of antibody fragments with monovalent femtomolar antigen-binding affinity. Proc Natl Acad Sci U S A. 2000 Sep 26; 97(20): 10701-5.

Barbas CF 3rd,Hu D,Dunlop N,Sawyer L,Cababa D,Hendry RM,Nara PL,Burton DR.In vitro evolution of a neutralizing human antibody to humanimmunodeficiency virus type 1 to enhance affinity and broaden strain cross-reactivity.Proc Natl Acad Sci U S A.1994 Apr 26;91(9):3809-13.

Parsons HL, Earnshaw JC, Wilton J, Johnson S, Schueler PA, Mahoney W, McCafferty J. Directing phage selections towards specific epitopes. Protein Eng. 1996 Nov; 9(l 1): 1043-9.

Winkler, Kramer A, Kiittner G, Seifert M, Scholz C, Wessner H, Schneider-Mergener J, Hohne W. Changing the antigen binding specificity by single point mutations of an anti-p24 (HIV-1) antibody. J Immunol. 2000 Oct 15; 165(8): 4505-14.

Safdari Y, Farajnia S, Asgharzadeh M, Omidfar K, Khalili M. humMRl, a highly specific humanized single chain antibody for targeting EGFRvIII. Int Immunopharmacol. 2014 Feb; 18(2): 304-10.

Lorimer IA, Keppler-Hafkemeyer A, Beers RA, Pegram CN, Bigner DD, Pastan I. Recombinant immunotoxins specific for a mutant epidermal growth factor receptor: targeting with a single chain antibody variable domain isolated byphage display. Proc Natl Acad Sci US A. 1996 Dec 10; 93(25): 14815-20.

Beers R, Chowdhury P, Bigner D, Pastan I. Immunotoxins with increased activity against epidermal growth factor receptor vlll-expressing cells produced by antibody phage display. Clin Cancer Res. 2000 Jul; 6(7): 2835-43.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Choi BD, Kuan CT, Cai M, Archer GE, Mitchell DA, Gedeon PC, Sanchez-Perez L, Pastan I, Bigner DD, Sampson JH. Systemic administration of a bispecific antibody targeting EGFRvIII successfully treats intracerebral glioma. Proc Natl Acad Sci U SA. 2013 Jan 2;110(l):270-5.

Fenstermaker RA, Ciesielski MJ. Deletion and tandem duplication of exons 2-7 in the epidermal growth factor receptor gene of a human malignant glioma. Oncogene. 2000 Sep 14; 19(39): 4542-8.

Han W, Zhang T, Yu H, Foulke JG, Tang CK. Hypophosphorylation of residue Y1045 leads to defective downregulation of EGFRvIII. Cancer Biol Ther. 2006 Oct; 5(10): 1361-8.

Grandal MV, Zandi R, Pedersen MW, Willumsen BM, van Deurs B, Poulsen HS. EGFRvIII escapes down-regulation due to impaired internalization and sorting to lysosomes. Carcinogenesis. 2007 Jul; 28(7): 1408-17. Gan HK, Kaye AH, Luwor RB. The EGFRvIII variant inglioblastoma multiforme. J Clin Neurosci. 2009 Jun; 16(6): 748-54.

Reist CJ, Archer GE, Kurpad SN, Wikstrand CJ, Vaidyanathan G, Willingham MC, Moscatello DK, Wong AJ, Bigner DD, Zalutsky MR. Tumor-specific anti-epidermal growth factor receptor variant III monoclonal antibodies: use of the tyramine-cellobiose radioiodination method enhances cellular retention and uptake in tumor xenografts. Cancer Res. 1995 Oct 1; 55(19): 4375-82.

Foulon CF, Reist CJ, Bigner DD, Zalutsky MR. Radioiodination via D-aminoacid peptide enhances cellular retention and tumor xenograft targeting of an internalizing anti-epidermal growth factor receptor variant III monoclonal antibody. Cancer Res. 2000 Aug 15; 60(16): 4453-60.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Patel D, Lahiji A, Patel S, Franklin M, Jimenez X, Hicklin DJ, Kang X. Monoclonal antibody cetuximab binds to and down-regulates constitutively activated epidermal growth factor receptor vIII on the cell surface. Anticancer Res. 2007 Sep-Oct; 27(5A): 3355-66. Jutten B, Dubois L, Li Y, Aerts H, Wouters BG, Lambin P, Theys J, Lammering G. Binding of cetuximab to the EGFRvIII deletion mutant and its biological consequences in malignant glioma cells. Radiother Oncol. 2009 Sep; 92(3): 393-8.

Dreier A, Barth S, Goswami A, Weis J. Cetuximab induces mitochondrial translocalization of EGFRvIII, but not EGFR: involvement of mitochondria in tumor drug resistance? Tumour Biol. 2012 Feb;33(l):85-94.

The present invention relates to a binding protein that specifically binds to EGFRvIII and a multispecific binding protein that specifically binds to EGFRvIII and CD3. The present invention further extends to a multispecific tandem diabody that binds to EGFRvIII and CD3. In particular, the present invention relates to a highly cytotoxic EGFRvIII/CD3 bispecific tandem diabody that is used to recruit T cells to effectively kill several types of solid tumor cancers.

EGFR imbalance is associated with many cancers, and small molecules and EGFR-targeted antibodies have all been successfully used in clinical practice. However, due to the extensive normal tissue expression of EGFR, the antibodies approved for clinical use and those in development all have serious side effects. The deletion mutant (deletion variant) III (EGFRvIII) of EGFR with truncated extracellular domain and ensuring ligand-independent intrinsic activity is the most common mutant form associated with oncogenic transformation. EGFRvIII is specifically expressed in cancer tissue and is associated with various solid tumor types. The restricted EGFRvIII expression on cancer cells provides the opportunity for developing a cytotoxic antibody that specifically targets cancer, protects normal tissues, and fully reduces the side effects associated with EGFR treatment.

In a first aspect of the present invention, fully human, highly specific, high-affinity EGFRvIII binding proteins are described herein. The described EGFRvIII binding proteins have the following advantages:

Figure 9 shows that EGFRvIII/CD3 of the present invention is the most common EGFRvIII binding protein of the present invention.They do not promote or promote the very low EGFRvIII internalization from the cell surface of target positive molecule, and this helps to raise cytotoxic T-cell or NK-cell.The result in Fig. 9 shows, according to the present invention, after cell is exposed to EGFRvIII conjugated proteins under all test conditions, with respect to untreated cell, EGFRvIII receptor molecules still remains on the cell surface.These results show, EGFRvIII/CD3 conjugated proteins of the present invention surprisingly do not show any internalization trend (Fig. 9).The combination of EGFRvIII specific binding proteins of the present invention may suppress internalization, rather than inducing internalization, or may promote the increase that EGFRvIII expresses, thereby causes its cell surface density to increase.

Can use affinity maturation technology (affinity maturation technique) to make the avidity of these high-affinity associated proteins be further significantly improved to obtain that EGFRvIII is had specificity, K _D in 100pM scope or lower associated proteins.Surprising fact is, these associated proteins have demonstrated the abnormal specificity in conjunction with EGFRvIII, and natural EGFR (wild-type EGFR or EGFRwt) is not had cross reactivity.Due to from 269 amino acid whose frames (in-frame) of EGFR extracellular domain (exon 2-7), therefore with respect to EGFRwt, in EGFRvIII, form new epi-position (neo-epitope), and estimate that this new epi-position is quite little: it is arranged by new amino acid (novel juxtaposition of aminoacids) (amino acid 5 is fused to amino acid 274) and a new GLY amino acid composition at the deletion site.

The goal of affinity maturation screening methods is to improve binding affinity by promoting the selection of binding proteins with reduced off-rates ( _kOFF ). However, surprisingly, the affinity matured binding proteins shown have greatly increased on-rates (kON ₎ , while the reduction in _kOFF has only a small contribution to the 100-fold increased binding of the binding proteins, with some variable antibody domains showing _KD less than 100 pM.

These unique properties make the variable antibody domains described herein that are specific for EGFRvIII particularly suitable for the development of multispecific, e.g., bispecific, multivalent, immune effector cell engaging, tumor-targeting antibody therapeutics, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies.

In a further aspect of the invention, multivalent EGFRvIII binding antibodies are provided that have two binding sites for EGFRvIII, or in the case of a multivalent bispecific antibody, in addition to two binding sites for EGFRvIII, also have a binding site for the T-cell antigen CD3 or CD16A specifically expressed on natural killer (NK) cells.

In one embodiment, use EGFRvIII is had two binding sites and comprises in addition the EGFRvIII/CD3 series connection diabody that CD3 has two binding sites, and tested their combination with recombinant EGFRvIII-or EGFR-Fc fused antigen.Under non-reducing conditions (due to complete disulfide bond, molecular structure still part retains), or under reducing conditions (protein structure complete denaturation), EGFRvIII-or EGFR-Fc fused antigen separates by SDS-PAGE, is transferred to PVDF membrane by Western blotting (Western Blotting), and uses different EGFRvIII/CD3 series connection diabody antibodies (diabodyantibody) or EGFR-to assess in conjunction with modifying (decorate) trace (Fig. 4).This embodiment shows, comprises the parental EGFRvIII binding domains that EGFRvIII has two binding sites and the EGFRvIII binding domains of affinity maturation and also keeps its selection specificity to EGFRvIII in multivalent antibody form. Under reducing and non-reducing conditions, all EGFRvIII/CD3 series connection double antibodies all can identify EGFRvIII, but all can not identify full-length wild-type EGFR under these conditions arbitrarily.This embodiment also proves, and EGFRvIII specific binding structural domain described herein can identify linear epitope, and its reactivity does not need complete three-dimensional structure.This is opposite with EGFR-in conjunction with IgG antibody cetuximab (Erbitux (Erbitux)), EGFR-in conjunction with IgG antibody cetuximab can identify wild-type EGFR and ripe EGFRvIII, but its reactivity obviously needs complete disulfide bond and conformational epitope (Fig. 4).

In another embodiment, EGFRvIII/CD3 tandem diabodies comprising parental or affinity-matured EGFRvIII specific domains in combination with different CD3 binding domains and/or having single binding domains of different orders in the tandem diabody molecule are used. The tandem diabody with the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII - _{V H} ^CD3 contains a divalent EGFRvIII binding site in the middle of the tandem diabody molecule, while the tandem diabody with the domain order V _H ^EGFRvIII -V _L ^CD3 -V _H ^CD3 -V _L ^EGFRvIII has two EGFRvIII binding sites in the outside position and two CD3 binding sites in the middle. In ELISA, the concentration-dependent binding of different EGFRvIII-domains and domain order mutants of the tandem diabody to EGFRvIII antigen, wild-type EGFR antigen, and to the concentration-dependent binding of the CD3 antigen (Fig. 6) are analyzed. All tandem diabodies containing affinity-matured EGFRvIII binding domains showed a substantial increase in binding to EGFRvIII, clearly manifested as corresponding binding curves shifting by more than two orders of magnitude to lower concentrations ( FIG. 6 ). No binding of any tandem diabody to wild-type EGFR antigen was observed.

Can select and identify anti-EGFRvIII antibody by phage display library (phase display libraries), for example, selectively combine the antigen-binding proteins of the EGFR of maturation rather than native form (native form).T cell is the effective tumor-killing effector cell that can not be raised by natural antibody.Therefore, in a further aspect of the present invention, the invention provides one group of multispecific EGFRvIII/CD3 antigen-binding proteins, particularly series connection diabody, it can raise T cell, have the binding characteristic and cytotoxicity of a wide range.Characterized the feature of the binding characteristic of antigen-binding proteins of the present invention, the cytotoxic activity of T cell mediation and the T cell activation of target mediation in one group of in vitro test.In FACS, antigen-binding proteins of the present invention shows unusual specificity to the EGFRvIII antigen in Biacore, ELISA and EGFRvIII-positive cell.Within the full concentration range of test, do not observe any conjugated protein and EGFR antigen or with the detectable combination of EGFRwt expressing cell. The most effective high-affinity series diabody shows the cytotoxicity of EC50 for F98 glioma cells and Chinese hamster ovary celIs expressing EGFRvIII and CHO cells; Remaining series diabody shows the cytotoxicity of EC50 up to about 10000pM.Also tested these series diabody pairs as the cytotoxicity of the EGFRwt ⁺ cell of the more sensitive probe of the residue of native form.When the series diabody concentration of maximum assessment was 0.5 μM, cytotoxicity was observed on EGFRwt ⁺ cell or other EGFRvIII-negative cells.It is crucial to be combined with the high-affinity of CD3 to effectively T cell recruitment, yet, when there is no EGFRvIII ⁺ target cell in vitro, as lacking measurement by proliferation, the series diabody of the present invention does not cause T cell activation, and this contributes to good preclinical safety (Figure 10).In a word, the strict specificity and high efficiency of anti-tumor cell toxicity are mediated by EGFRvIII/CD3 series diabody.

Tandem diabodies with CD3 binding _KD ranging from 1.1 nM to approximately 500 nM but relatively constant EGFRvIII binding _KD (ranging from 2.0 nM to 6.7 nM) were generated and showed good correlation between CD3 binding strength and cytotoxic potency against cells expressing EGFRvIII, with _EC50 values ranging from 25 pM to approximately 10,000 pM (Figure 11).

Although the affinity of the antigen-binding protein of the present invention for EGFRvIII has been significantly improved, the specificity is not lost.

is a trademark of Affimed Therapeutics used to represent tandem diabodies (Kipriyanov et al., 1999, J. Mol. Biol, 293: 41-56; Cochlovius et al., 2000, Cancer Res., 60: 4336-4341; Reusch et al., 2004, Int. J. Cancer, 112: 509-518, Kipriyanov, 2009, Methods Mol Biol, 562: 177-93; McAleese and Eser, 2012, Future Oncol. 8: 687-95). In the context of the present invention, TandAb and tandem diabodies are used as synonyms.

The sequence of wild-type EGFR gene and protein is known. EGFRvIII is the result of an in-frame deletion of exons 2-7 (801bp) of the wild-type EGFR gene, resulting in 267 amino acid deletions in the receptor's outer domain and a new glycine residue at the junction of exons 1 and 8. This new amino acid juxtaposition in the EGFR extracellular domain generates tumor-specific and immunogenic epitopes.

According to first aspect of the present invention, described EGFRvIII had to specific associated proteins at least, it does not cross-react with wild-type EGFR.This associated proteins preferably comprises CDR1, CDR2, CDR3 of the antibody variable heavy chain that is selected from the CDR described in the following table 1A and CDR1, CDR2 and the CDR3 of antibody variable light chain.

Table 1A: Sequence identifiers of amino acid sequences of human light and heavy chain CDRs from binding proteins that specifically bind to EGFRvIII

Therefore, an embodiment of the present invention is an EGFRvIII binding protein having at least one EGFRvIII binding site, comprising an antibody variable heavy chain domain and/or an antibody variable light chain domain; wherein the antibody variable heavy chain domain comprises a CDR1 selected from SEQ ID NO: 27, 33, 42, 46, 49, 53, a CDR2 selected from SEQ ID NO: 28, 38, 43, 50, 54, 61, 63 and a CDR3 of SEQ ID NO: 29; the antibody variable light chain domain comprises a CDR1 selected from SEQ ID NO: 30, 34, 36, 39, 44, 47, 51, 56, 59, 64, a CDR2 selected from SEQ ID NO: 31, 40, 57 and a CDR3 selected from SEQ ID NO: 32, 35, 37, 41, 45, 48, 52, 55, 58, 60, 62 and 65.

According to further embodiment, described associated proteins comprises and has the antibody variable heavy chain domain that is selected from SEQ ID NO:1,3,5,7,9,11,13,15,17,19,21 and 25 as shown in Table 1B and/or is selected from SEQ ID NO:2,4,6,8,10,12,14,16,18,20,22 and 26 as shown in Table 1B at least one EGFRvIII binding site of antibody variable light chain domain.The aminoacid sequence of three variable light chain CDR and three variable heavy chain CDR is shown in bold and is underlined in Table 1.These structural sites show that the affinity to EGFRvIII is improved, and specificity does not have loss.These antigen-binding proteins do not cross-react with wild-type EGFR antigen.

In some embodiments, the present invention relates to an antigen-binding protein of the present invention.These antigen-binding proteins of the present invention do not show or only show minimized EGFRvIII internalization when in conjunction with EGFRvIII.According to the present invention, under the test conditions according to embodiment 9, after cell is exposed to EGFRvIII-in conjunction with antibody, be greater than 80%, preferably greater than 90%, most preferably be greater than 95% EGFRvIII receptor molecules and still remain on cell surface.The protein-binding combination of EGFRvIII of the present invention may suppress internalization, rather than inducing internalization, or may promote that the expression of EGFRvIII increases, thereby causes its cell surface density to increase.

The term "binding protein" refers to an immunoglobulin derivative with antigen-binding properties, i.e., an immunoglobulin polypeptide or fragment thereof containing an antigen-binding site. The binding protein comprises the variable domains of an antibody or fragment thereof. Each antigen-binding domain is formed by an antibody, i.e., an immunoglobulin, a variable heavy chain domain (VH), and an antibody variable light chain domain (VL), which bind to the same epitope. The variable light chain domain or variable heavy chain domain of the present invention is a polypeptide comprising CDR1, CDR2, and CDR3. Preferably, the binding protein of the present invention lacks immunoglobulin constant domains. The term "binding protein" also refers to antibody fragments or derivatives, including Fab, Fab', F(ab') ₂ , Fv fragments, single-chain Fv, diabodies, tandem diabody flexibodies (WO03/025018), and tandem single-chain Fv ((scFv) ₂ ).

Table 1B: Amino acid sequences of all anti-EGFRvIII variable heavy chain domains (VH) and variable light chain domains (VL) (amino acid sequences of CDR1, CDR2, and CDR3 are shown in bold and underlined)

In a preferred embodiment, the binding protein that confers specificity for EGFRvIII comprises an antigen binding site from one of the following combinations of variable heavy chain domains and variable light chain domains shown in Table 1:

(i) SEQ ID NO: 1 and SEQ ID NO: 2,

(ii) SEQ ID NO: 3 and SEQ ID NO: 4,

(iii) SEQ ID NO: 5 and SEQ ID NO: 6,

(iv) SEQ ID NO: 7 and SEQ ID NO: 8,

(v) SEQ ID NO: 9 and SEQ ID NO: 10,

(vi) SEQ ID NO: 11 and SEQ ID NO: 12,

(vii) SEQ ID NO: 13 and SEQ ID NO: 14,

(viii) SEQ ID NO: 15 and SEQ ID NO: 16,

(ix) SEQ ID NO: 17 and SEQ ID NO: 18,

(x) SEQ ID NO: 19 and SEQ ID NO: 20,

(xi) SEQ ID NO: 21 and SEQ ID NO: 22, or

(xii) SEQ ID NO:25 and SEQ ID NO:26.

The present invention also provides not only EGFRvIII has specificity, and also has the associated proteins of at least one other functional domain.In a further embodiment, described at least one other functional domain is effector domain (effector domain)." effector domain " is that effector cell is had specific binding site, and it can stimulate or trigger cytotoxicity, engulfment, antigen presentation, cytokine release.Such effector cell is, for example but not limited to, T-cell or NK-cell.Particularly, described other effector domains comprise at least one antibody variable heavy chain domain and at least one variable light chain domain that form the antigen binding site of CD3 preferred mankind CD3.

In some embodiments, the EGFRvIII conjugated proteins of the present invention can be multispecific.Term used herein " multispecific " means to refer to that conjugated proteins of the present invention have at least two antigen binding sites to different epi-positions, and wherein at least one is used for EGFRvIII.For example, conjugated proteins can be trispecific, and have binding site and have at least one binding site to effector cell such as epi-position on CD3 or antigen to two different epi-positions and/or antigen on tumor cell.Conjugated proteins can have binding site to different epi-positions of same antigen and/or the epi-position of different antigens.Such multispecific conjugated proteins comprise single-chain diabody, (scFv) ₂ , series diabody and elastomeric antibody (referring to Le Gallet al., 1999FEBS Letts 453:164-168 and WO 03/025018).In a specific embodiment of the present invention, conjugated proteins are to EGFRvIII and CD3 bispecific.

The CD3 binding site of the multispecific binding protein of the invention can be composed of a variable heavy chain domain (SEQ ID NO: 23) and a variable light chain domain (SEQ ID NO: 24) as shown in Table 2, or a close homolog of such sequences that differs from the CDR sequences by only two amino acids, or any other fully human, humanized or non-human variable antibody domain specific for CD3 (referred to as CD3), such as, for example, a variable antibody domain specific for CD3, in particular CD3 epsilon, or any other fully human, humanized or non-human variable antibody domain, such as, for example, the variable antibody domain of UCHT1 or OKT3.

Table 2: Amino acid sequences of fully human anti-CD3 variable heavy chain domain (VH) and variable light chain domain (VL) (amino acid sequences of CDR1, CDR2 and CDR3 are shown in bold and underlined)

In addition, EGFRvIII conjugated proteins of the present invention can be multivalent.Term used herein " multivalent " means to refer to, and conjugated proteins of the present invention comprises at least two antigen binding sites.Antigen binding site can have identical or different specificity.In embodiments of the present invention, conjugated proteins is in conjunction with identical epi-position with at least two binding sites in a divalent manner.The protein-conjugated example of divalence is diabody, and the protein-conjugated example of at least tetravalence is a series connection diabody.

In a further aspect of the present invention, described EGFRvIII associated proteins and bispecific EGFRvIII and CD3 binding site are humanized or fully human, i.e. people-source. In a further aspect of the present invention, EGFRvIII associated proteins or multispecific EGFRvIII and CD3 associated proteins are dimers, i.e. comprise two polypeptides with EGFRvIII and CD3 antigen binding sites.

According to the present invention, dimerization bispecific EGFRvIII and CD3 binding proteins are provided in the form of series connection diabody.Such series connection diabody is by connecting four antibody variable binding domains (heavy chain variable domain (VH) and light chain variable domain (VL) in single gene construct (referring to McAleese and Eser, 2012, Future Oncol.8:687-95)) that can homodimerize and build.In such series connection diabody, joint (linker) length is that it prevents the intermolecular pairing of variable domain so that molecule can not fold back to form single-chain diabody on itself, but is forced to pair with the complementary domain (complementary domain) of another chain.These domains can also be arranged like this, namely make corresponding VH and VL domain pair during dimerization.After being expressed by single gene construction, two same polypeptide chains head to tail folding form the functional non-covalent homodimer (McAleese and Eser, 2012, Future Oncol.8:687-95) of about 110kDa. Despite the absence of intramolecular covalent bonds, homodimers, once formed, are highly stable, remain intact and do not revert to monomeric form.

In one embodiment of a multispecific tandem diabody, a humanized bispecific tetravalent antibody is provided that has two binding sites each for CD3 and EGFRvIII (EGFRvIII/CD3RECRUIT-) to utilize the cytotoxic capacity of T cells to treat glioblastoma (GB), prostate cancer, head and neck cancer, and other epidermal growth factor receptor mutant vIII-positive (EGFRvIII+) cancers.

In the present invention, the tandem diabody is the most widely used antibody of the present invention.Relative to traditional monoclonal antibody, tandem diabody has a large amount of properties that advantage is provided, also has less multispecific molecule.Tandem diabody is fully functional when there is no glycosylation.Tandem diabody can only contain antibody variable domains, and therefore can not have any side effect that can be relevant to Fc part.Because bispecific tandem diabody allows divalence to be bound to each in EGFRvIII and CD3, therefore, its avidity (avidity) is identical with the avidity of IgG.The size of tandem diabody is about 110kDa, is less than IgG, and this can enhance tumor infiltration.But, compared with the less bispecific form based on antibody-binding domain or non-antibody scaffold, this size is much higher than the kidney threshold of first-pass clearance (first-pass clearance), and this provides pharmacokinetic advantage. The generation and preparation of such tandem diabodies is described, for example, in Kipriyanov SM: Methods Mol. Biol. 2009; 562: 177-93, Kipriyanov SM: Methods Mol. Biol. 2003; 207: 323-33 or McAleese and Eser, 2012, Future Oncol. 8: 687-95.

Tandem diabodies are well expressed in CHO cells, enabling efficient upstream and downstream manufacturing processes to be put in place.

The EGFRvIII and CD3 bispecific tandem double antibody of the present invention is designed to specifically target EGFRvIII ⁺ tumor cells by recruiting cytotoxic T cells. Antibodies cannot directly recruit cytotoxic T cells. On the contrary, by employing (engage) the CD3 molecules specifically expressed on these cells, the tandem double antibody can make cytotoxic T cells and EGFRvIII ⁺ tumor cells cross-linked in a highly specific manner, thereby significantly increasing the cytotoxic effect of these molecules. The tandem double antibody shows strong, specific and effective cytotoxicity. Significant evidence is that T cells can play a role in controlling tumor growth. For example, the presence of cytotoxic T cells in colorectal tumors and lymph nodes from NHL patients is shown to be correlated with better clinical results. In addition, for melanoma vaccines, and antibodies for CTLA-4, a negative regulator of T-cell activation, have demonstrated the potential for the therapy of designing induction of T-cell responses. The tandem double antibody of the present invention employs cytotoxic T cells by binding surface-expressed CD3, preferably CD3ε, which forms a part of the T-cell receptor. The simultaneous binding of this tandem diabody to CD3 and EGFRvIII expressed on the surface of specific tumor cells leads to T-cell activation and mediates subsequent tumor cell lysis.

"Dimer" refers to a complex of two polypeptides. Preferably, the two polypeptides are non-covalently associated with each other, particularly under conditions where there is no covalent bond between the two polypeptides. Preferably, the bispecific dimer is a homodimer, i.e., comprises two identical polypeptides. However, a trispecific or other multispecific dimer can be a heterodimer and comprise two different polypeptides, for example, at least one binding site of the variable light chain domain is located on one polypeptide and the variable heavy chain domain is located on the other domain. The term "polypeptide" refers to a polymer of amino acid residues connected by amide bonds. Preferably, the polypeptide is an unbranched single-chain fusion protein. In this polypeptide, the variable antibody domains are connected in sequence. In addition to the variable domain N-terminus and/or C-terminus, the polypeptide may also have contiguous amino acid residues. For example, such contiguous amino acid residues may include a Tag sequence that may be beneficial to polypeptide purification, preferably at the C-terminus.

Each polypeptide of the bispecific tandem diabody contains at least four variable domains, namely the variable light chain (VL) and variable heavy chain (VH) of the CD3 binding protein and the variable light chain (VL) and variable heavy chain (VH) of the EGFRvIII binding protein. The four binding domains are connected by peptide linkers L1, L2 and L3 and can be arranged as follows from N-terminus to C-terminus:

(i)VL(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3); or

(ii) VH(CD3)-L1-VL(EGFRvIII)-L2-VH(EGFRvIII)-L3-VL(CD3); or

(iii) VL(EGFRvIII)-L1-VH(CD3)-L2-VL(CD3)-L3-VH(EGFRvIII); or

(iv) VH(EGFRvIII)-L1-VL(CD3)-L2-VH(CD3)-L3-VL(EGFRvIII).

In an embodiment of the present invention, the four variable domains are arranged as VL ^{(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3). Tandem diabodies having the domain order VLCD3-VHEGFRvIII -VLEGFRvIII-VHCD3} _{and containing} ^different _EGFRvIII ^binding _sequences , despite having different affinities for EGFRvIII, all showed very similar binding to the second specific CD3, as shown by ELISA on the CD3γε antigen ( FIG5A ). In other tandem diabody domain sequences containing EGFRvIII binding domains in the outer positions and two CD3 binding sites in the middle ( _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ), a significant improvement in EGFRvIII binding to the affinity-matured EGFRvIII binding domain was also evident, while binding of the tandem diabody middle domain to CD3 was slightly weaker and more variable between individual tandem diabodies than the first domain sequence (Figure 5B). Similar observations were made when tandem diabodies containing different EGFRvIII-binding domains were constructed using different CD3 domains (SEQ ID NOs: 23 and 24) (Figure 5C).

The length of the linker affects the flexibility of the antigen-binding tandem diabody. For example, Todorovska et al., 2001 Journal of Immunological Methods 248:47-66; Perisic et al., 1994 Structure 2:11217-1226; Le Gall et al., 2004 Protein Engineering 17:357-366 and WO94/13804 describe the effect of linker length on the formation of dimeric antigen-binding proteins.

According to the present invention, the length of the peptide linkers L1, L2, and L3 is preferably such that the domains of one polypeptide can intermolecularly bind to the domains of another polypeptide to form a dimeric antigen-binding tandem diabody. Such linkers are "short," i.e., comprised of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, or 12 amino acid residues. Thus, the linker consists of approximately 12 or fewer amino acid residues. In the case of 0 amino acid residues, the linker is a peptide bond. Such short linkers facilitate intermolecular dimerization of the two polypeptides by binding and forming a correct antigen-binding site between the antibody variable light chain domain and the antibody variable heavy chain domain of the different polypeptides. Short linkers of approximately 12 or fewer amino acid residues generally prevent intermolecular interaction between adjacent domains of the same polypeptide chain. In one embodiment of the present invention, these linkers consist of approximately 3 to approximately 10, e.g., 4, 5, or 6, consecutive amino acid residues.

Regarding the amino acid composition of these linkers, peptides are selected that do not interfere with the dimerization of the two polypeptides. For example, linkers containing glycine and serine residues generally provide protease resistance. The amino acid sequence of the linker can be optimized, for example, by phage display methods to increase antigen binding and the yield of antigen-binding polypeptide dimers. Examples of peptide linkers suitable for the tandem diabodies of the present invention are GGSGGS (SEQ ID NO: 66), GGSG (SEQ ID NO: 67), or GGSGG (SEQ ID NO: 68).

EGFRvIII associated proteins as herein described and multi-specific series connection diabody can produce by the polynucleotide of expressing coding series connection diabody polypeptide, and the polypeptide of described series connection diabody is combined to form antigen-in conjunction with series connection diabody with another identical polypeptide.Therefore, the further embodiment of the present invention is the polynucleotide of coding antigen as herein described-in conjunction with series connection diabody polypeptide, for example DNA or RNA.

Polynucleotide can be constructed by methods well known to those skilled in the art, for example, by combining the gene of at least four antibody variable domains separated by the sequence encoding peptide linkers or directly connected by peptide bonds into a single gene construct, and in bacteria or other suitable expression systems such as for example expressed in Chinese hamster ovary celI, the single gene construct is operably connected with a suitable promotor, alternatively a protein tag for detection and purification, and a suitable transcription terminator. According to employed vector systems and hosts, any number of suitable transcription and translation elements comprising formation and inducible promoters can be used. Select promoter so that it drives the expression of polynucleotides in respective host cells.

As a further embodiment of the present invention, the polynucleotide can be inserted into a vector, preferably an expression vector. The recombinant vector can be constructed according to methods known to those skilled in the art.

Various expression vector/host systems can be used to contain or express polynucleotides encoding polypeptides of the present invention. Examples of expression vectors for expression in E. coli are pSKK (Le Gall et al., J Immunol Methods. (2004) 285(1): 111-27) or pcDNA5 (Invitrogen) for expression in mammalian cells.

Thus, the antigen-binding tandem diabodies described herein can be prepared by introducing a vector encoding the polypeptides described above into host cells and culturing the cells under conditions whereby the polypeptide chains are expressed, can be isolated from the culture medium, and optionally thereafter purified.

In a further embodiment, the present invention provides a pharmaceutical composition comprising an EGFRvIII binding protein, an antigen-binding tandem diabody, a vector comprising a polynucleotide encoding the polypeptide of the antigen-binding tandem diabody or a host cell transformed by the vector, and at least one pharmaceutically acceptable carrier. The term "pharmaceutically acceptable carrier" means any carrier that does not interfere with the effectiveness of the biological activity of the component and is non-toxic to the administered patient. Suitable pharmaceutical carrier examples are known in the art and include physiological saline, phosphate buffered saline, water, emulsions such as oil/water emulsions, various types of wetting agents, sterile solutions, etc. Such carriers can be prepared by conventional methods and administered to an individual at a suitable dosage. Preferably, the composition is sterile. These compositions may also include adjuvants, such as preservatives, emulsifiers, and dispersants. The effects of microorganisms can be prevented by including various antibacterial and antifungal agents.

The skilled artisan will readily construct and obtain antigen-binding proteins such as the tandem diabodies described herein without undue burden by using well-established techniques and standard methods known in the art, see, for example, Sambrook, Molecular Cloning A Laboratory Manual, Cold Spring Harbor Laboratory (1989) N.Y.; The Protein Protocols Handbook, John M. Walker, ed., Humana Press Inc. (2002); or Antibody engineering: methods and protocols/Benny K.C. Lo, ed.; Benny K.C. II Series: Methods in molecular biology (Totowa, N.J.).

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 illustrates the position of variable sequence in the EGFRvIII specific high-affinity binding domain derived after affinity maturation.The CDR position that is different from the heavy chain and light chain variable region of parental sequence is marked with asterisk, and amino acid is highlighted with grey separately.The position with useful sudden change is highlighted with dark grey.

Fig. 2 shows that among Biacore T200, multi-circulation kinetics measures the result that different scFv antibodies are attached to recombinant EGFRvIII-Fc antigen or wild-type EGFR-Fc antigen.Sensogram is presented at the kinetics of surface plasmon resonance (SPR) response unit (RU) measured in time under 4 different scFv antibody concentration levels (86nM, 17.2nM, 3.4nM, 0.69nM).Measurement is in conjunction with 180 seconds, after this measurement dissociates.Use 1:1 to combine model analysis data.

Figure 3 shows the results of flow cytometry measurement of scFv antibody binding to (A) F98 rat glioma cells overexpressing EGFRvIII (F98 ^EGFRvIII ), (B) F98 cells overexpressing native EGFR (F98 ^EGFR ), and (C) untransfected F98 cells (F98), as well as the results of similar scFv antibody binding to (D) CHO cells overexpressing EGFRvIII (CHO ^EGFRvIII ), (E) CHO cells overexpressing native EGFR (CHO ^EGFRvIII ), and (F) untransfected CHO cells (CHO).

Fig. 4 illustrates the tandem diabody containing different EGFRvIII binding domains and is separated by SDA-PAGE and is transferred to the EGFRvIII-or natural EGFR-Fc fusion antigen (non-reducing protein or the albumen with DTT reduction) of the purification of PDVF film by western blotting.Fc fusion rotein is by disulfide bond at Fc part dimerization, but is monomer after reducing with DTT.EGFRvIII-specific antibody is bonded to EGFRvIII-fusion rotein with reduction or non-reduction conformation, but is not combined with natural EGFR.This anti-EGFR antibody cetuximab is as contrast, and identifies non-reducing EGFRvIII-Fc and non-reducing EGFR-Fc antigen, but does not identify the antigen of DTT-processing.

Figure 5 shows the concentration-dependent binding of different EGFRvIII/CD3-specific tandem diabodies to EGFRvIII-Fc, EGFR-Fc, and CD3γε recombinant antigens analyzed in ELISA. (A) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , (B) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , and (C) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies containing the CD3 domains of SEQ ID NOs: 23 and 24 and having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 . The improvement in binding of tandem diabodies containing affinity-matured EGFRvIII-binding domains to EGFRvIII was clearly detectable as a shift of the respective binding curves to lower concentrations in all formats/domain orders.

Figure 6 shows the results of multi-cycle kinetic measurements of EGFRvIII/CD3-specific tandem diabodies containing different bivalent EGFRvIII binding domains in the order of _V ^CD3 - _V ^EGFRvIII - _V ^EGFRvIII - _V ^CD3 bound to recombinant EGFRvIII-Fc antigen in a Biacore T200. Sensogram displays the kinetics of surface plasmon resonance (SPR) response units (RU) measured over time at different tandem diabody concentration levels. Binding was measured for 180 seconds, followed by dissociation. Data were analyzed using a 1:1 binding model.

Fig. 7 shows the immunohistochemistry (IHC) staining result of tissue section from (A) three individual head and neck (H＆N) cancer patients, (B) three individual glioblastoma patients, and (C) representative results from prostate cancer, mammary cancer (Her2neg) and non-small cell lung cancer (NSCLC).The EGFRvIII-specific double antibody containing EGFRvIII-specific binding domain Li3G30 of the divalent conformation of 2 kinds of different concentrations is dyed in section, and described divalent conformation is similar to the arrangement of EGFRvIII targeting tandem double antibody antibody (for example, (V _L ^CD3- ) V _H ^EGFRvIII - _{V L} ^EGFRvIII (-V _H ^CD3 )) but lacks CD3 binding part.As a control, section is dyed with the EGFR-specific IgG antibody cetuximab of identification EGFR (being present on healthy tissue and cancerous tissue) and EGFRvIII.The specificity of dyeing is confirmed by the control section hatched in the secondary reagent that does not contain EGFR-or EGFRvIII-specific primary antibody.

Fig. 8 illustrates that on EGFRvIII positive cells, compared with the tandem diabody of EGFRvIII and CD3 respectively containing the parental EGFRvIII binding domain (Li3G30) before affinity maturation, the cytotoxicity ability, effectiveness and target-dependent specificity of EGFRvIII/CD3 tandem diabody containing different affinity maturation EGFRvIII-specific binding domains: (A) ABC 470 containing EGFRvIII/CD3 tandem diabody, (B) ABC 471 containing EGFRvIII/CD3 tandem diabody, (C) ABC 472 containing EGFRvIII/CD3 tandem diabody. Using CHO cells expressing EGFRvIII (CHO ^EGFRvIII ), expressing EGFRvIII negative EGFR (CHO ^EGFR ) or untransfected CHO cells as target cells and using human PBMC as effector cells, in the analysis based on flow cytometry, antibody concentration-dependent cytotoxicity was measured. In each panel (AC), tandem diabodies containing a single EGFRvIII binding domain and CD3 domain v64, in the order _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , are shown on the left, and tandem diabodies containing a single EGFRvIII binding domain combined with a different CD3 domain, in the order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , are shown on the right.

Figure 9 shows the results of evaluating EGFRvIII receptor internalization from the cell surface when combined with EGFRvIII-specific antibodies. Cells expressing EGFRvIII were incubated for 24 hours at 37°C with varying concentrations of three different ^EGFRvIII / ^CD3 tandem diabody antibodies with the domain order V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII - _{V H} ^CD3 (i) or _{V H} EGFRvIII-V L ^CD3 -V H CD3-V _L ^EGFRvIII (iv), and/or EGFRvIII/CD3 tandem diabody antibodies containing different CD3-binding domains. After adjustment, all remaining EGFRvIII receptor molecules on the cell surface were stained with the respective tandem diabody antibodies at a saturating concentration of 10 μg/ml.

Figure 10 is presented in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) series double antibody antibody, assesses the result of cell proliferation (BrdU incoporation) in human PBMC culture.PBMC is used as positive control, and it breeds in the presence of CD3 binding antibody OKT3 or phytohemagglutinin (plant lectin phytohemagglutinin) (PHA).In the absence of EGFRvIII-positive target cells, with different EGFRvIII/CD3 series double antibodies, hatch PBMC and do not induce activation and the proliferation of PBMC.

Figure 11 illustrates that the CD3-binding affinity of EGFRvIII/CD3 tandem diabodies is positively correlated with their cytotoxic capacity. Regarding CD3-binding on CD3 ⁺ -Jurkat cells, EGFRvIII-binding on CHO ^EGFRvIII cells and cytotoxic capacity, all 20 EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) in the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII -V _H ^CD3 and different CD3-binding domain combinations with different affinities for CD3 were analyzed. The scope of CD3-binding K _D value is 1nM to approximately 500nM. For each analyzed tandem diabodies, relative to CD3-binding K _D value, EGFRvIII-binding K _D value and cytotoxic EC ₅₀ value are mapped. While the TandAbs showed only small changes in EGFRvIII-binding _KD values, the cytotoxicity _EC50s showed an almost linear increase as the CD3-binding _KD values of the tandem diabodies increased.

Figure 12 shows the in vivo proof-of-concept results from the study. Two EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII-binding domain (Li3G30) but different CD3-binding domains or domain sequences were analyzed for dose-dependent inhibition of subcutaneous F98 ^EGFRvIII xenograft tumor growth and compared with the effect of cetuximab administered at a much higher concentration. (A) Dose-dependent tumor growth inhibition caused by the ^EGFRvIII /CD3 ^v6 tandem diabody with the domain sequence _VL ^CD3 - _VH EGFRvIII- _VL ^EGFRvIII - _VH ^CD3 ; (B) Dose-dependent tumor growth inhibition caused by EGFRvIII/CD3 tandem diabodies containing different high-affinity CD3-binding domains and the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII .

The present invention is further illustrated by the following examples, but the present invention is not limited thereto.

Example 1: Discovery and affinity maturation of EGFRvIII-specific binding proteins:

Li3G30 Discovery:

The phage display library of people's scFv sequence based on IgM source is carried out two to three rounds of panning (panning) and EGFRvIII is had specific binding substances (binders) with enrichment.Before every round of panning, described library is pre-incubated with the binding substances of (deplete) wild-type form EGFR or the binding substances of fusion rotein Fc part with EGFR-Fc.On the solid phase of EGFRvIII coating and in solution, select (selection) in parallel.After two rounds and three rounds of panning, pick out monoclonal, induce soluble scFv to express and screen the extract that is combined with EGFRvIII-Fc and EGFR-Fc.

According to the present invention, the fully human high specific variable antibody domains combined with EGFRvIII have been described, and EGFRvIII is the deletion mutant of EGFR, and it is specifically expressed in tumor rather than on healthy tissue with the form of mutation.Full human EGFRvIII specific domains described herein were initially found in the phage display screening of the fully human scFv library using based on IgM source.Surprisingly, although two or three rounds of panning were carried out using exhaustion step (depletion procedure) to wild-type EGFR antigen, this should be enriched to have specific binding substances to EGFRvIII, but in ELISA, most of the scFvs of gained are all combined to two kinds of albumen (EGFRvIII and wild-type EGFR).Identified a high EGFRvIII specific sequence, called after Li3G30.Except that there is little difference in the N-terminal of VH and VL, the framework region of heavy chain and light chain and CDR1 and CDR2 are identical with sequence VH5-51 and VL3-25 of germline encoding respectively.Li3G30 was identified from the third round of liquid phase panning.

Affinity Maturation:

The library of antibody framework pair VH5-51/VL3-25 was designed using an algorithm ordered from DistributedBio. The specificity determining residues of Li3G30 were identified and included in the library design. The design includes 52 randomized CDR sites, each with a separately defined amino acid distribution. Three different loop lengths were built into the VL CDR3. Gene fragments encoding the randomized sites of VH and VL were ordered from Geneart/Lifetechnologies and synthesized using TRIM-technology. The fragments were cloned into the phage display vector pEXHAM (Schwarz et al. 2004) to a final library size of 3.7E+8 transformed E. coli cells. The library was packaged into phage particles and, through panning and screening steps, separated with improved affinity and retained variants specific for EGFRvIII. Panning was performed using the EGFRvIII-Fc antigen immobilized on a protein-bound plastic surface. The elutriation step is designed to the scFv that helps to have the EGFRvIII bonded with low dissociation rate ( _kOFF ) by implementing following steps: washing step comprises several washing buffer incubation steps that continue up to 30 minutes, to help select the scFv that slowly dissociates.Another step is based on the overnight competition with solubility EGFRvIII.In order to guarantee in the process of affinity maturation, can not select EGFR-EGFRvIII cross-reaction antibody or the antibody in conjunction with Fc, before first elutriation step, by the pre-incubation of wild-type EGFR-Fc, the library is exhausted possible wild-type EGFR binding substances.And owing to the existence of solubility EGFR in the process of the elutriation step being carried out to fixing EGFRvIII, screening scheme is unfavorable for cross-reaction binding substances.

The unique EGFRvIII-specific scFv was further tested for binding to transfected CHO and F98 cells expressing EGFRvIII using flow cytometry. ScFvs that showed binding to cells expressing wild-type EGFR and/or untransfected cells were excluded from further analysis. Initial _KD ranking using Biacore X100 was used to identify scFvs with improved affinity for EGFRvIII compared to the original clone, Li3G30. The best variants were characterized using detailed SPR measurements and fluorescence-based stability assays.

For VL CDR3, the initial library for affinity maturation comprises three kinds of different loop lengths, and known all lengths are compatible with VH5-51/VL3-25 framework.Surprisingly, in the selection process, the middle loop length is obviously preferred.Therefore, compared with the anti-EGFRvIII antibody of parent generation, the VL CDR3 of the variant of all affinity maturations has all shortened one amino acid (table 1, Fig. 1).The sequence of selected scFv is compared with the input library, and this analysis has shown 3 useful sudden changes in VL CDR3.The heavy chain of the anti-EGFRvIII antibody of parent generation and the CDR1 and CDR2 of light chain are identical with the sequence encoded separately of germline sequence VH5-51 and VL3-25.In 75% of the sites, the residue of each germline coding is reconstructed (reconstituted) or obviously preferred.Shown a useful sudden change in VH CDR1, one in VL CDR1 (Fig. 1).The amino acid distribution in the remaining randomization site is very close to the distribution in the library before selecting.

Example 2: Measurement of scFV binding to EGFRvIII and wild-type EGFR in Biacore:

Preparation of anti-hu Fc IgG CM5-chip:

Anti-hu Fc IgG CM5-chip was prepared by covalent amine coupling using an amine coupling kit (Amine-Coupling-Kit) (GE) according to the manufacturer's instructions. IgG was diluted to a concentration of 25 μg/ml in the provided immobilization buffer (10 mM sodium acetate, pH 5.0). For the coupling step, the predefined method "Amine" of the Biacore T200 control software was applied. A target level of 5000-10000 RU was achieved in all 4 flow cells. Before injecting the recombinant protein solution, the surface of the chip was activated with EDC/NHS. The remaining binding sites on the surface were blocked with ethanolamine. 1×HBS-P+ was selected as the running buffer for the entire coupling step.

Concentration range of single-chain Fv’s (scFv) used in SPR measurements:

Before SPR measurement, the ScFv concentration was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate the molar concentration of the His-tagged scFv antibody, a molecular weight of approximately 28 kD was assumed. The scFv was adjusted to the following final concentrations by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 430 nM, 86 nM, 17.2 nM, 3.44 nM, 0.688 nM, and 0 nM.

ScFv measurement conditions:

EGFRvIII-Fc and wild-type EGFR-Fc fusion proteins were diluted in 1× HBS-P+ to a concentration of 6.25 nM. Both antigens were injected at a flow rate of 10 μl/min: EGFRvIII-Fc for 75 seconds and wild-type EGFR-Fc for 90 seconds. EGFRvIII-Fc was injected into flow cell 2, and wild-type EGFR-Fc was injected into flow cell 4. Flow cells 1 and 3 were left blank.

1× HBS-P+ was used as running buffer throughout the procedure. The scFv antibody was injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. The signal measured in flow cell 3, where no antigen was captured, was subtracted from the signal in flow cell 4 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0M _MgCl2 (30 seconds, 10 μL/min). Before each measurement, four "start-up cycles" were performed without scFv. ScFv solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

The kinetics of the scFv antibodies were determined in the MCK (Multi Cycle Kinetic) measurement. To estimate the apparent binding affinity, the binding curves were evaluated using the "Kinetic" method included in the Biacore evaluation software. Using this method, all binding curves were fitted using the "Langmuir 1:1 interaction model" with the software. The results are shown in Table 3.

The affinity maturation screening step that is used is intended to improve the EGFRvIII binding affinity of EGFRvIII specific binding domain more easily by making the selection of the binding substances with the dissociation rate ( _kOFF ) of reduction.Yet, surprisingly, the affinity maturation binding substances (table 3) of gained shows the association rate ( _kON ) that significantly increases, and the reduction of _kOFF reaches 100 times of combination for the raising of scFv and only makes less contribution, and some binding substances demonstrate the _KD lower than 100pM simultaneously (table 3).Especially surprisingly, although the screening step is conducive to the scFv that selects the _kOFF reduction, the reduction of realization _KD is mainly due to the raising of _kON.This also is surprising, because the affinity maturation of antibody is relevant (Schier et al., 1996, Pini et al., 1998, Boder et al., 2000) with the reduction of _kOFF conventionally.

Table 3: Summary of scFv binding affinities ("VH" is the variable heavy chain domain of the scFv, "VL" is the variable light chain domain of the scFv). _K values, association rates (kON) and dissociation rates ( _kOFF ) of different EGFRvIII binding domains in scFv format for recombinant EGFRvIII-Fc antigen measured in a Biacore ( _SPR ) 1:1 binding model; "enhancement factor" refers to the fold change relative to the parental scFv, i.e., Li3G30 (VH SEQ ID NO: 25 and VL SEQ ID NO: 26).

Eight of the 11 affinity-matured scFvs showed at least a 5-fold improvement in _kON compared to the parental scFv; three of these had a 20-fold higher _kON . Compared to the improvement in _kON , the effect of affinity maturation on _kOFF was quite small. Only three affinity-matured scFvs showed a 4-5-fold improvement in _kOFF . Interestingly, two of these were isolated from a competitive selection method, which may be a more effective method for selecting scFvs with reduced _kOFF than prolonged incubation in wash buffer.

The average melting temperature of the selected affinity matured candidates determined by differential scanning fluorimetry (DSF) was 62.2°C, ranging from 53°C to 67.2°C.

Maintaining high specificity for EGFRvIII relative to wild-type EGFR during affinity maturation

Affinity maturation is usually accompanied by small changes in binding epitopes, which can affect specificity/cross-reactivity (Barbas et al, 1994, Parsons et al., 1996, Winkler et al., 2000). Due to the in-frame deletion of 269 amino acids in the wild-type EGFR extracellular domain (exon 2-7), an EGFRvIII new epitope relative to wild-type EGFR is formed, and it is predicted to be quite small: it is composed of new amino acids in parallel (amino acid 5 is fused to amino acid 274) and a new GLY amino acid at the deletion site. Even if it is expected that small changes in binding epitopes will quickly destroy the abnormal specificity of EGFRvIII in this case, a cross-reactive conjugate that can also identify native EGFR can be obtained.

Surprisingly, the affinity maturation described herein resulted in up to 100-fold improved binding of anti-EGFRvIII antibodies without loss of exclusive specificity for EGFRvIII (Table 3, Figure 2).

During the selection of affinity-matured binders, the binding of the only EGFRvIII-specific scFv to CHO cells expressing EGFRvIII (CHO ^EGFRvIII ) and F98 cells (F98 ^EGFRvIII ), as well as to CHO cells expressing human wild-type EGFR (CHO ^EGFR ) and F98 cells (F98 ^EGFR ), and to untransfected CHO and F98 cells were tested in flow cytometry.

Characterized in more detail 11 high-specificity anti-EGFRvIII scFv that do not have any measurable background combination with wild-type EGFR positive cell (CHO ^EGFR ) or the Chinese hamster ovary celI of non-transfected.In the best case, measured the avidity to 80pM of EGFRvIII antigen, corresponding to 100 times of raising coefficient (table 3) compared with parental scFv.Simultaneously, the anti-EGFRvIII scFv that is detected all does not demonstrate any cross reactivity (Fig. 2) with wild-type EGFR antigen.

To our knowledge, such high affinity for EGFRvIII has never been observed for a human antibody domain that binds exclusively specifically to EGFRvIII, without measurable cross-reactivity with wild-type EGFR. Safdari and colleagues recently described results for humanizing the murine EGFRvIII-binding antibody domain MR1 (Safdari et al., 2014), which was first described in its murine form by Lorimer and colleagues (Lorimer et al., 1996, Beers et al., 2000, Kuan et al., 2000). In their humanization approach, they achieved an increase in the affinity of humMR1, reaching similar picomolar _K values reported here. However, in contrast to the present invention, they did not completely eliminate the cross-reactivity of the antibody domain with wild-type EGFR, as can be clearly seen in Figure 4 of their publication (Safdari et al., 2014). Thus, after SDS PAGE and Western blotting, humMR1 was not only specific for EGFRvIII (145 kD), but also recognized EGFR (170 kD) (Safdari et al., 2014). Applicants extensively tested the specificity and cross-reactivity of the EGFRvIII binding domain of the present application before and after affinity maturation and surprisingly found no cross-reactivity signals with wild-type EGFR or purified recombinant wild-type EGFR antigen in Biacore (Figure 2), SDS PAGE and Western blotting (Figure 4), or ELISA (Figure 5). The EGFRvIII binding domain of the present application also did not exhibit any cross-reactivity with wild-type EGFR overexpressed on the cell surface of transfected CHO cells (CHO ^EGF ) or F98 glioma cells (F98 ^EGFR ) (Figure 3).

Example 3: High specificity and high affinity of EGFRvIII-specific scFv antibodies were assessed by flow cytometry for binding to F98 rat glioma cells (F98 ^EGFRvIII ) or stably transfected CHO cells (CHO ^EGFRvIII ) expressing EGFRvIII; no binding to F98 cells (F98 ^EGFR ) or untransfected F98 cells (F98) expressing native EGFR, and no binding to CHO cells (CHO ^EGFR ) or untransfected CHO cells (CHO) expressing native EGFR.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), Ficoll Paque PLUS (GE Healthcare), FITC-conjugated goat anti-human IgG (Dianova), FITC-conjugated goat anti-mouse IgG, min X (Dianova), L-glutamine (Invitrogen), _NaN3 (Carl Roth), PBS (Invitrogen), Penicillin/Streptomycin (Invitrogen), Propidium iodide (Sigma), RPMI-1640 (Invitrogen).

Cells and cell lines:

F98 rat glioma cells, F98 cells overexpressing EGFR (F98 ^EGFR ), and F98 cells overexpressing EGFRvIII (F98 ^EGFRvIII ) were purchased from the American type culture collection (ATCC) and cultured according to the recommended protocol.

F98 (ATCC CRL-2397)

F98 ^EGFR (ATCC CRL-2948)

F98 ^npEGFRvIII (ATCC CRL-2949)

Stably transfected CHO cells expressing recombinant EGFR or EGFRvIII were generated at Affimed according to the following protocol:

The gene of coding wild-type EGFR or deletion mutant EGFRvIII is synthesized by Life Technologies/GeneArt (Regensburg, Germany), and is subcloned among the mammalian expression vector pcDNA5/FRT.Expressing the stable Chinese hamster ovary celI of recombinant wild-type EGFR or EGFRvIII generates based on the host cell line Flp-In CHO (Life Technologies) of previous adaptation to suspension growth.The day before transfection, Flp-In Chinese hamster ovary celI is carried out passage culture (not having bleomycin (Zeocin)) in the HyClone CDM4CHO that is supplemented with L-glutamine, HT supplement and penicillin/streptomycin.The stable expression cell line is by using polyethyleneimine (PEI) transfection reagent, with the product expression based on pcDNA5/FRT and integration plasmid and Flp recombinase expression plasmid pOG44 (Life Technologies) cotransfection Flp-In Chinese hamster ovary celI system generates. 2.5 μg of total DNA was diluted in 50 μL of OptiMEMI medium and combined with 7.5 μg of PEI diluted in 50 μL of OptiMEMI medium. The mixture was incubated for 10 minutes and then added to 2×10 ⁶ Flp-In CHO cells suspended in 1 mL of CHO-S-SFMII medium. One day after transfection, the cells were diluted to a density of 1×10 ⁵ viable cells/mL in CHO-S-SFMII medium supplemented with 500 μg/mL hygromycin B and seeded in T75 culture flasks. Flp recombinase mediates the insertion of the Flp-In expression construct into the genome at the integrated FRT site through site-specific DNA recombination. During the selection phase, the viable cell density was measured once or twice a week, the cells were centrifuged and resuspended in fresh selection medium at a maximum density of 1×10 ⁵ viable cells/mL. After about 2-3 weeks of selection using 500 μg/mL hygromycin B, stable cells expressing EGFR or EGFRvIII were recovered and then transferred to HyClone CDM4 CHO suspension culture medium. Stable cells expressing wild-type EGFR or EGFRvIII were cryopreserved in a culture medium containing 50% ProFreeze (Lonza) and 7.5% DMSO.

To determine the density of wild-type EGFR or EGFRvIII cell surface antigens on stably transfected cell lines, a flow cytometric indirect immunofluorescence assay (QIFIKIT, Dako) was used according to the manufacturer's instructions and demonstrated that the density of wild-type EGFR on CHO ^EGFR stably transfected cells was approximately equal to the density of EGFRvIII on CHO ^EGFRvIII stably transfected cells, whereas no wild-type EGFR or EGFRvIII binding sites were detected on untransfected CHO cells.

Determination of Antibody Binding and Affinity by Flow Cytometry:

Cells and indicated scFv antibodies are started from 100 μ g/mL (about 3300nM) in 100 μ L serial dilutions in FACS buffer and are hatched on ice for 45min (minute). After washing 3 times with FACS buffer, cells are hatched on ice for 45min with 0.1mL 10 μ g/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After second washing cycle, under the same conditions as before, cells are hatched with the goat anti-mouse IgG antibody that 0.1mL 15 μ g/mL FITC is put together. As a control, in the absence of scFv, cells are hatched with anti-His IgG 13/45/31, then with the goat anti-mouse IgG antibody that FITC is put together. Then cells are washed again and are resuspended in the 0.2mL FACS buffer containing 2 μ g/mL propidium iodide to exclude dead cells. The fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer with MXP software (Beckman-Coulter, Krefeld, Germany) or a Millipore Guava EasyCyte flow cytometer with Incyte software (Merck Millipore, Schwalbach, Germany). The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany).

If analyzing with a Beckman-Coulter FC500 MPL flow cytometer, use 0.5 × 10 ⁶ cells/stain; if analyzing with a Millipore Guava EasyCyte flow cytometer, use only 0.25 × 10 ⁶ cells/stain.

KD values were calculated using the equation for single-site binding (hyperbola) using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) after subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone.

The results are shown in Figure 3.

Example 4: SDS PAGE and Western blotting analysis of the binding specificity of antibodies binding to EGFRvIII.

To detect the binding specificity of EGFRvIII/CD3 tandem diabodies containing different binding domains to EGFRvIII antigen and to demonstrate the absence of binding to wild-type EGFR antigen, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis were performed.

By recombinant DNA technology, the sequence of the extracellular domain of encoding wild-type EGFR or truncated EGFRvIII is fused to the Fc part of human IgG antibody.DNA construct is transfected into the Chinese hamster ovary celI as the wild-type EGFR-Fc or EGFRvIII-Fc of the reorganization of soluble protein and is secreted into the cell culture supernatant, uses protein A chromatography to purify the wild-type EGFR-Fc or EGFRvIII-Fc of described reorganization from described cell culture supernatant.Through the formation of intramolecular disulfide bond in the Fc part, wild-type EGFR-Fc (or EGFRvIII-Fc) fused antigen forms the dimer with two identical chains.Due to EGFRvIII-specific 267 amino acid deletions, EGFRvIII-Fc causes the Fc fusion protein of dimer form to have the size difference of about 50kD than EGFR-Fc little by about 25kD.

Equal amounts of purified wild-type EGFR-Fc (and EGFRvIII-Fc) were mixed with either non-reducing 2× SDS PAGE sample buffer or reducing 2× SDS PAGE sample buffer containing dithiothreitol (DTT) as a reducing agent. Samples containing DTT were heated at 95°C for 5 minutes and then loaded onto a 4-20% standard TGX precast SDS PAGE gel (Biorad). 1 μg of protein sample was used per lane. To separate the proteins in the gel, SDS-PAGE was run at 300V in 1× Tris/glycine/SDS buffer (Biorad) for approximately 25 minutes. Total protein was visualized in the gel using a standard stain-free molecular imaging system (Biorad). A Page Ruler unstained protein ladder (Thermo Scientific) was used as a molecular weight marker. The gel was then blotted onto a PVDF membrane using BioRad's semi-dry blotting Fastblot system. The membrane was blocked in 3% (w/v) nonfat dry milk in 1× TBS for 30 min at room temperature. The membrane was cut into small pieces, each containing a similar protein sample to be blotted. Different tandem diabodies were diluted to a concentration of 2 μg/ml in 3% (w/v) nonfat dry milk in 1× TBS and incubated with separate membrane pieces on a rocking platform for 1 hour. The anti-EGFR antibody cetuximab was used as a control at a concentration of 10 μg/ml. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS, then incubated with HRP-conjugated secondary detection antibodies PentaHIS-HRP (QIAGEN) (for detection of the tandem diabodies) or Protein L-HRP (Thermo Scientific) (for detection of cetuximab) at a dilution of 1:5000 in 3% nonfat dry milk in TBS on a rocking platform for 1 hour. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS. HRP-mediated color development on the membrane was initiated by adding 0.06% DAB + 0.02% CoCl ₂ + 0.015% H ₂ O ₂ in TBS. The reaction was terminated by adding water. The membrane was dried and scanned.

The results are shown in Figure 4.

Example 5: The binding of EGFRvIII/CD3 tandem diabodies containing different EGFRvIII binding domains to EGFRvIII antigen and CD3, as well as their binding specificity to EGFRvIII in the absence of binding to wild-type EGFR antigen, were analyzed by enzyme-linked immunosorbent assay (ELISA).

ELISA steps:

96-well ELISA plates (Immuno MaxiSorp; Nunc) were coated overnight at 4°C with EGFRvIII-Fc, wild-type EGFR-Fc, or CD3γε recombinant antigen in 100 mM carbonate-bicarbonate buffer. To achieve approximately equal antigen molar coating densities, EGFRvIII-Fc was coated at a concentration of 4 μg/ml, EGFR-Fc at a concentration of 6 μg/ml, and CD3γε at a concentration of 1.5 μg/ml. After a blocking step using 3% (w/v) skim milk powder (Merck) in PBS, 11 serial dilutions of different EGFRvIH/CD3 tandem diabodies ranging from 200 ng/μl to 6.5× ^10-6 ng/μl in PBS containing 0.3% (w/v) skim milk powder (Merck) were incubated on the plates at 25°C for 1.5 h. After incubation, the plates were washed three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20 (Serva). 50 ng/ml of Protein L-HRP was added and incubated on the plates at 25°C for 1 hour. After washing three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20, tetramethylbenzidine (TMB) substrate (Seramun) was added for detection. The reaction was terminated after approximately 2 minutes by adding 100 μl/well of 0.5 _{M H₂SO₄} . The absorbance of the wells was measured at 450 nm using a multiwell plate reader (Victor, Perkin Elmer).

Absorbance values were plotted as lines in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada).

The results are shown in FIG5 .

Example 6: Measurement of Binding of Tandem Diabodies to EGFRvIII by SPR:

Preparation of anti-hu Fc IgG CM5-chip

An anti-hu Fc IgG CM5-chip was prepared as described above (Example 2).

Concentration range of tandem diabodies used in SPR measurements:

Before SPR measurement, the concentration of the tandem diabody was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate molar concentrations, a molecular weight of 105 kD was assumed for all tandem diabodies. The tandem diabodies were adjusted to the final concentrations mentioned below by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 150 nM, 30 nM, 6 nM, 1.2 nM, 0.24 nM, 0.048 nM, and 0 nM; or 90 nM, 30 nM, 10 nM, 3.33 nM, 1.11 nM, 0.37 nM, 0.123 nM, 0.041 nM, and 0 nM, or the concentrations indicated.

Binding assay conditions for tandem diabody KD measurements

Dilute the EGFRvIII-Fc fusion protein in 1×HBS-P+ to a concentration of 6.25 nM. For EGFRvIII-Fc, inject the antigen solution at a flow rate of 10 μl/min for 40 seconds. Inject the antigen into flow cell 2. Flow cell 1 serves as the reference cell.

1× HBS-P+ was used as running buffer throughout the procedure. Serial dilutions of the diabody were injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0 M _MgCl2 (3 x 15 seconds, 10 μL/min). Before each measurement, three "priming cycles" were performed without the tandem diabody. The tandem diabody solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

Binding parameters of the tandem diabodies were determined in an MCK (Multi Cycle Kinetic) measurement. To estimate apparent binding affinity, binding curves were evaluated using the "Kinetic" method included in the Biacore X100 evaluation software. Using this method, binding curves were fitted in whole or in part using the "Langmuir 1:1 interaction model" with the software.

Binding parameters of tandem diabodies

To analyze the effect of combining two EGFRvIII binding domains into bivalent or multivalent or multispecific molecules such as EGFRvIII/CD3 tandem diabodies, the apparent affinity of EGFRvIII/CD3 tandem diabodies to EGFRvIII antigen was measured by SPR using BiacoreX100 ( FIG6 ).

For all molecules detected, the apparent affinity of the EGFRvIII specific binding domain in the EGFRvIII/tandem diabody improves, and the best combination tandem diabody has achieved a _K of 11pM. Relative to the corresponding tandem diabody containing the parental EGFRvIII binding domain, the _K of the tandem diabody containing the domain of affinity maturation reduces by up to 25 times. Relative to the _K of the measurement of the monovalent binding scFv, the improvement of _K is also up to 25 times. Interestingly, contrary to the improvement achieved by the affinity maturation driven mainly by the improved k _ON , the further improvement of binding affinity in the bivalent tandem diabody format is realized by slower k _ofOFF to a great extent (Table 4).

Therefore, the tandem diabody format provides a way to further enhance binding affinity while maintaining the specificity of the antibody binding domain.

Table 4: Summary of measured KD, association rates (kON), and dissociation rates ^{( kOFF} ) for binding of different EGFRvIII/ ^CD3 tandem diabodies containing bivalent parental or affinity matured ^EGFRvIII -specific binding domains with the domain order _{VLCD3 - VHEGFRvIII} - _VLEGFRVIII - _VHCD3 to recombinant EGFRvIII-Fc antigen in the SPR _{1:1 binding} model.

Example 7: Immunohistochemistry (IHC) staining of solid tumor tissue sections using anti-EGFRvIII bispecific diabody or cetuximab

In contrast to wild-type EGFR, which is ubiquitously expressed on healthy tissues, expression of the mutant receptor EGFRvIII is highly tumor-specific. However, the high frequency of EGFRvIII expression in glioma cells is consistent with literature descriptions, but the extent of EGFRvIII expression and its homogeneity in other tumors remain controversial.

A small immunohistochemistry (IHC) study was initiated, analyzing the reactivity of tissue sections from three cancer patients each suffering from glioblastoma (GB), head and neck (H&N) cancer, prostate cancer, Her2-negative breast cancer, and non-small cell lung cancer (NSCLC) with the EGFRvIII specific binding domains in the form of a bivalent diabody according to the present invention ( FIG. 7 ).

The present invention relates to the immunohistochemical study of the expression of EGFRvIII and the tumor specificity of EGFRvIII binding antibodies to assess different tumors.To contain the bivalent double antibody form of the EGFRvIII binding domain Li3G30 arranged with V _H ^EGFRvIII - _{V L} ^EGFRvIII , the EGFRvIII combination of EGFRvIII specific variable domains has been detected.Secondary detection is implemented by the His tag on the double antibody protein.In contrast, used natural EGFR to have specific IgG antibody cetuximab (Erbitux (Erbitux)).

All people's tissue sections are purchased from BioChain Institute, Inc., and according to supplier's specification sheets process section.Make tissue section be adapted to room temperature and hatch with DAKO peroxidase blocker, then use 10% goat serum to seal.EGFRvIII specific double antibody was hatched 1 hour with two concentration levels of 0.5 μ g/ml and 0.1 μ g/ml, then hatched and passed through Envision+HRP-DAB system detection mouse antibody (DAKO) with anti-His antibody (Dianova).According to manufacturers' specification sheets, use Klear people HRP-polymer DAB detection system (GBI Labs), implement the tissue section dyeing that carries out with control item Erbitux (Merck KGaA) with the concentration of 10g/ml.

Finally, sections were stained with hematoxylin and mounted with mountant medium (Shandon ConsulMount ^™ , Thermo Scientific).

The results are shown in FIG7 .

Two demonstrate in three glioblastoma samples with the obvious and specific reactivity of EGFRvIII specific double antibody.Surprisingly, all three head and neck cancer samples all demonstrate strong and obvious EGFRvIII positive, and similarly for prostate cancer, mammary cancer and NSCLC, obtained the specific staining (Fig. 7) of tumor tissue with EGFRvIII specific antibody described herein.

Based on these properties, the EGFRvIII binding domains described herein, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies, are excellently suited for developing multispecific, multivalent, immune effector cell engaging, tumor targeting antibody therapies.

Example 8: Evaluation of cytotoxic activity mediated by EGFRvIII/CD3 tandem diabodies in a FACS-based cytotoxicity assay

In the cytotoxicity assay based on cell, analyzed EGFRvIII/CD3 series connection diabody antibody, it contains the parental generation that is combined with different CD3 binding domains or the EGFRvIII specificity domain of affinity maturation and/or each binding domain (Fig. 8) with different orders in described series connection diabody molecule.By using the Chinese hamster ovary celI (CHO ^EGFRvIII ) of expressing EGFRvIII, the Chinese hamster ovary celI (CHO ^EGFR ) of expressing wild-type EGFR and not transfected Chinese hamster ovary celI as target cell to analyze the specificity of target-mediated, series connection diabody and the lethality of T cell mediation.

Materials (media, buffers, and reagents):

DMSO (Sigma), EasySep ^™ Human T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD4+ T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD8+ T Cell Enrichment Kit (Stem Cell Technologies), The Big Easy EasySep ^™ Magnet (Stem Cell Technologies), FCS (Invitrogen), Lymphoprep (Stem Cell), L-glutamine (Invitrogen), mlgGl FITC (ADG), CD16-FITC (MEM-154) (Thermo Fisher Scientific), mlgGl-FITC/mlgGl-PE/mlgGl-ECD (Beckman Coulter), mlgGl-PE (Beckman Coulter), mlgGl-PC5 (Beckman Coulter), mlgGl-PC7 (Beckman Coulter), CD8-FITC/CD4-PE/CD3-ECD (Beckman Coulter), CD16-PC5 (Beckman Coulter), CD19-PC7 (Beckman Coulter), CD16-FITC/CD56-PE/CD3-ECD (Beckman Coulter), CD14-PC7 (Beckman Coulter), CD33-PE (MACS Miltenyi Biotech), NaN ₃ (Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), PKH67 green fluorescent cell labeling (Linker) Midi kit (Sigma), Gammanorm human IgG (Octapharma), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

Isolation of PBMCs from buffy coats and enrichment of T cells:

On the day of blood draw, buy buffy coat from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany); Buffy coat preparation is stored at room temperature overnight, and then PBMC is separated by density gradient centrifugation. Buffy coat sample is diluted with two to three volumes of PBS, layered on Lymphoprep cushion and centrifuged at 800 × g for 25 min at room temperature w/o brake. PBMC at the interface is collected and washed 3 times with PBS, and then they are used for enrichment of T cells or flow cytometric analysis. According to the manufacturer's instructions, EasySep ^™ human T cell enrichment kit (for immunomagnetic separation of untouched human T cells) and Big Easy EasySep ^™ magnet are used to enrich T cells from the PBMC group.

Characterization of effector cells by flow cytometry:

To assess the distribution of PBMC subpopulations and the purity of enriched PBMC or T cell subsets, PBMC staining was performed.

For flow cytometric analysis, isolated PBMCs or enriched PBMC subpopulations were resuspended in PBS supplemented with 2% heat-inactivated FCS and 0.1% sodium azide (referred to as FACS buffer) and 1 mg/mL polyclonal human IgG to a density of 0.5-2 × 10 ⁶ /mL. 0.5 ml aliquots of the cell suspension were then incubated with the antibody for 15 minutes in the dark according to the manufacturer's recommended protocol.

Table 5: Pipetting scheme for characterization of PBMCs and fractionated PBMCs by flow cytometry

After incubation, samples were measured on a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) in the order 3, 4, 5, 6, 1, 2 using a protocol developed for PBMC and multicolor staining (FITC PEECD PC5 PC7; FITC PE ECD; FITC PE PC5PC7) in MXP acquisition software (Beckman-Coulter). For analysis and data plotting, CXP analysis software (Beckman-Coulter) was used.

Cytotoxicity assay:

As previously described, T cells used as effector cells were characterized by flow cytometry. The target cells used in the cytotoxicity assay were CHO cells stably transfected to overexpress EGFRvIII (CHO ^EGFRv111 ), CHO cells stably transfected to express wild-type EGFR (CHO ^EGFR ), or untransfected CHO cells (CHO). Cell lines were generated as previously described and cultured under standard conditions as previously described. For the cytotoxicity assay, target cells were collected, washed twice with RPMI 1640 medium without FCS, and resuspended in diluent C provided by the PKH67 green fluorescent cell labeling Midi kit to a density of 2×10 ⁷ /mL. The cell suspension was then mixed with an equal volume of double-concentrated PKH67-labeling solution (e.g., 1 μL PKH in 250 μL diluent C) according to the manufacturer's instructions and incubated at room temperature for 2-5 minutes with regular mixing. The staining reaction was terminated by adding an equal volume of FCS and incubating for 1 minute. After washing the labeled target cells with complete RPMI medium, the cells were counted and resuspended in complete RPMI medium to a density of 2×10 ⁵ /mL.

2×10 ⁴ target cells were then seeded with T cells at a 5:1 E:T ratio in each well of a round-bottom 96-well microplate along with the indicated antibodies in a total volume of 200 μL/well. Nine serial 1:5 dilutions, starting at 30 μg/mL, were typically tested. Natural cell death in the absence of antibody and target killing by effectors were determined in at least three replicates on each plate. Tandem diabody-mediated killing was typically determined in duplicate.

After centrifugation at 200 g for 2 minutes at room temperature, the assay plate was incubated at 37° C. in a humidified atmosphere containing 5% CO ₂ for 40-48 hours. After incubation, the culture was washed once with FACS buffer and then resuspended in 150 μL FACS buffer supplemented with 2 μg/mL PI. The absolute number of viable target cells, characterized by positive green PKH67 staining but negative PI staining, was measured using a Millipore Guava EasyCyte flow cytometer (Merck Millipore).

Based on the remaining live target cells measured, the percentage of specific cell lysis was calculated according to the following formula: [1-(number of live target cells _(sample) )/(number of live target cells _(native) )]×100%. Sigmoidal dose response curves and EC50 values were calculated by nonlinear regression/4-parameter logistic fit using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, California, USA).

Data Analysis:

Lysis values obtained for a given antibody concentration were determined in duplicate and analyzed by Sigmoidal dose-response/4-parameter logistic fit analysis using Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) and used to calculate the mean EC50 and SD of the replicates for percentage lysis.

The results are shown in FIG8 and Table 6.

The tandem diabody of all detections all demonstrates the concentration dependency to the cell of expressing EGFRvIII, high specificity, cytotoxicity and high efficiency, and the survival of EGFRvIII negative cells is not impaired.Within the whole concentration range of any tandem diabody of expressing EGFRwt-cell or non-transfected cell detection, do not observe detectable cytotoxicity.Therefore, the cytotoxicity of EGFRvIII/CD3 tandem diabody mediation is strictly target-dependent.In cytotoxicity assay, the EC 50 of the tandem diabody that contains parental EGFRvIII binding domain before affinity maturation and domain order is V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 is _25pM (Fig. 8).Corresponding tandem diabody contains the EGFRvIII binding domain that avidity improves with identical domain order (V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 ), shows enhanced cytotoxicity efficacy equally, and in the best case, reaches the EC ₅₀ (table 6) of 1.5pM. The relative increase in cytotoxic activity is more evident in the tandem diabodies with EGFRvIII specific domains located in the external position (domain order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ) (Table 6). In these tandem diabodies, up to 70-fold increased cytotoxic potency can be achieved by replacing the low-affinity EGFRvIII binding domain Li3G30 with an affinity-enhanced domain in these tandem diabodies (Figure 8, Table 6).

Table 6: to the summary of the EC50 value measured with different EGFRvIII/CD3 series connection diabodies in the cytotoxicity assay, described different EGFRvIII/CD3 series connection diabodies contain the EGFRvIII specific binding domain of affinity maturation with two different domain orders.Shown the raising coefficient relative to the series connection diabodies that contain the parental generation EGFRvIII binding domain of correspondence.

Choi et al. (Proc Natl Acad Sci USA. 2013 Jan 2; 1 10 (1): 270-5; WO2013/7185010) describes a t EGFRvIII/CD3 comparator, an EGFRvIII/CD3 bispecific T-cell engager (BiTE), which has an EGFRvIII binding site (based on the mouse MR1 domain) and a CD3 binding site (based on the anti-CD3 antibody OKT3). In a cell-based cytotoxicity assay, the antibody achieved 50% of the maximum specific lysis achieved at a concentration of about ^10-2 μg/ml, which corresponds to a molar EC50 concentration of about 200 pM for the 52kD BiTE (Figure 4C of Choi et al., 2013). Comparisons show that the cytotoxicity of the tandem diabodies described herein containing the parental, rather than affinity-matured, EGFRvIII binding domains is even about 10-fold higher than that of the EGFRvIII/CD3 BiTE. Furthermore, the tandem diabody containing the affinity-matured EGFRvIII binding domain of the present invention even exhibited 100-fold higher cytotoxicity than that of EGFRvIII/CD3 BiTE.

Example 9: EGFRvIII receptor internalization induced by EGFRvIII/CD3 tandem diabody on CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII )

To evaluate EGFRvIII regulation by EGFRvIII/CD3 tandem diabodies, CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII ) were seeded in aliquots of 2.5×10 ⁵ cells per well of a round-bottom 96-well microplate and incubated with increasing concentrations of EGFRvIII/CD3 tandem diabodies in RPMI 1640 medium supplemented with 10% FCS, 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, and 100 μg/mL sodium pyruvate (referred to herein as complete RPMI medium; all components purchased from Invitrogen) at 37° C., 5% CO ₂ , in a humidified incubator for 24 hours. Aliquots of cells were cultured in the absence of antibody to serve as control samples for detection of the highest EGFRvIII cell surface levels and as negative controls for staining performed with the secondary reagent alone.

44.After hatching, cell is washed 2 times with ice-cold phosphate buffered saline (PBS, Invitrogen) (being called FACS buffer in this article) that is supplemented with 2% heat inactivation FCS (Invitrogen) and 0.1% sodium azide (Roth, Karlsruhe, Germany), then with the antibody (this antibody is identical with the antibody that hatches with cell during adjustment incubation) of saturation concentration (10 μ g/mL) dyeing to assess the maximum antibody combination corresponding to the remaining EGFRvIII antigen on the cell surface.The aliquots of the cell that is cultivated under the situation that there is no antibody are dyed with the same antibody of saturation concentration, to measure the maximum antibody binding capacity corresponding to the maximum measurable EGFRvIII level on the cell surface.After washing 3 times with FACS buffer, with series connection double antibody treatment cell and dye with the anti-His mAb (Dia910, Dianova) of 10 μ g/mL, then with the goat anti-mouse IgG (Dianova) dyeing that puts together with 15 μ g/mL FITC. After repeated washing with FACS buffer, cells were resuspended in FACS buffer containing 2 μg/mL propidium iodide (Sigma) to exclude dead cells. Approximately 5 × 10 ³ propidium iodide-negative cells from each sample were analyzed using MXP software and an FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany). The mean fluorescence intensity (MFI) of the samples was determined using CXP software (Beckman-Coulter) or cells were analyzed using a Millipore Guava EasyCyte flow cytometer using Incyte software (Merck Millipore, Schwalbach, Germany).

After subtracting the signal obtained from cell staining with the secondary reagent alone, the mean fluorescence intensity of the measured samples was plotted as a line in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada), and the relative amount of EGFRvIII remaining on the cell surface was calculated using the fluorescence signal obtained by staining untreated cell samples at 37°C in the absence of the EGFRvIII/CD3 tandem diabody, which was set to represent 100% of cell surface EGFRvIII.

The results are shown in FIG9 .

Some bibliographic reports have described and compared deletion mutant EGFRvIII and demonstrated impaired internalization (Fenstermaker et al.2000, Han et al.2006, Grandal et al.2007, Gan et al.2009) with natural EGFR.However, about EGFRvIII specific antibody or for several disclosed reports of the antibody (for example cetuximab) also in conjunction with the wild-type EGFR of EGFRvIII, described in conjunction with rapid internalization and the degraded (Reist et al.1995, Foulon et al.2000, Kuan et al.2000, Patel et al.2007, Jutten et al.2009, Dreier et al.2012) of EGFRvIII in the cell that expresses EGFRvIII after antibody separately.

For internalization determination, use the transfection CHO ^EGFRvIII cell of overexpressing EGFRvIII.The cells of these expression EGFRvIII and the EGFRvIII/CD3 series connection double antibody of varying concentrations were hatched 24 hours at 37 ℃, described the condition in the context of other EGFRvIII binding antibodies to induce strong internalization.After this adjustment (during which antibody-receptor complex can internalize or not internalize), give all remaining EGFRvIII receptor molecules dyeing (Fig. 9) on the cell surface with the saturation concentration of 10 μ g/ml with each antibody.The result that Fig. 9 presents shows, after cell is exposed to EGFRvIII-in conjunction with series connection double antibody, under all detection conditions, the EGFRvIII receptor molecules more than 80% and up to 140% existing on untreated cell are still available (140% or more>100% general value refers to and reduces (cellular internalization) with the number of receptor on the cell surface on the contrary, and the cell processed with series connection double antibody demonstrates the increase of receptor on the cell surface.Internalization to 100% comes from the contrast with untreated cell). These results show, EGFRvIII/CD3 series connection diabody of the present invention surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may even suppress internalization and cause EGFRvIII receptor levels on the cell surface to increase.The result that Fig. 9 presents shows, with respect to untreated cell, under all detection conditions, after cell is exposed to EGFRvIII-in conjunction with antibody, at least more than 80% still remain on the cell surface.These results show, EGFRvIII/CD3 antibody of the present invention, especially EGFRvIII/CD3 series connection diabody surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may suppress internalization or promote EGFRvIII to express and increase, cause its cell surface density to increase.

Example 10: Evaluation of cell proliferation in PBMC culture in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies

To evaluate whether EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies induce T cell activation and subsequent proliferation in the absence of EGFRvIII ⁺ target cells, human PBMCs were cultured in the presence of increasing concentrations of EGFRvIII/CD3 tandem diabodies. After 5 days of incubation, proliferation was assessed in a BrdU incorporation assay.

Materials (culture media, buffers, and reagents)

FCS (Invitrogen), human serum (Sigma), Lymphoprep (Stemcell Technologies), L-glutamine (Invitrogen), OKT3 (Biolegend), PBS (Invitrogen), β-mercaptoethanol (Invitrogen), penicillin/streptomycin (Invitrogen), RPMI-1640 (Invitrogen), sodium pyruvate (Invitrogen), BrdU cell proliferation ELISA (Roche).

cell:

Buffy coats were purchased from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany) on the day of blood draw; buffy coat preparations were stored at room temperature overnight before PBMC isolation.

Isolation of PBMCs from buffy coats and enrichment of T cells:

PBMCs were isolated from the buffy coat by density gradient centrifugation. The buffy coat sample was diluted with two to three volumes of PBS, layered onto a Lymphoprep cushion, and centrifuged at 800 × g for 25 minutes at room temperature with or without braking. PBMCs at the interface were collected and washed three times with PBS before use in proliferation assays.

Proliferation assay

All assay plates were blocked for 2 hours at room temperature with RPMI1640 medium supplemented with 10% heat-inactivated human serum, 4 mM L-glutamine, 100 U/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, 1 mM sodium pyruvate, and 0.05 mM β-mercaptoethanol (referred to herein as complete RPMI medium) to prevent nonspecific binding of antibodies to plastic surfaces. After removing the blocking medium, 4 × 10 ⁵ PBMCs were seeded in each well of a flat-bottom 96-well microplate with the indicated concentrations of test and control items in complete RPMI medium in a total volume of 200 μL/well. Each sample was assayed in triplicate. IgG anti-CD3 clone OKT3 was used as a positive control. To assess spontaneous proliferation, cells were cultured in the absence of antibody in six replicates. The plates were then incubated at 37°C and 5% CO ₂ in a humidified atmosphere for 4 days. 18 hours before termination of incubation, the cell cultures were pulsed with 100 μM BrdU. BrdU incorporation was quantified using a BrdU proliferation ELISA kit according to the manufacturer's instructions. Color development was stopped by adding 100 mM _{H₂SO₄ ,} and absorbance was measured at 450 nm using a multilabel plate reader (Victor 3, Perkin Elmer). Absorbance values were analyzed and the mean and SD were plotted using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA).

The results are shown in FIG10 .

Example 11: Correlation between CD3 Binding Affinity and Cytotoxicity of EGFRvIII/CD3 Tandem Diabody

Constructed and expressed EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) and different CD3 binding domains (which have some affinity for CD3 T cell antigens) and measured binding to CD3 ⁺ Jurkat cells and EGFRvIII ⁺ CHO ^EGFRvIII cells; In addition, cytotoxicity assays were performed using CHO ^EGFRvIII cells as target cells and PBMC cells as effector cells. The order of each VH and VL (heavy and light chain) domain of all tested tandem diabodies tested in this example was the same: _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 .

Antibody binding to EGFRvIII ⁺ CHO ^EGFRvIII cells and cytotoxic activity mediated by the EGFRvIII/CD3 tandem diabody were evaluated by flow cytometry as described in previous examples.

After staining CD3 ⁺ Jurkat cells with increasing concentrations of the tandem diabody, the apparent affinity of the EGFRvIII/CD3 tandem diabody for CD3 ⁺ cells was determined by flow cytometry. KD values were calculated by nonlinear regression/hyperbola using the mean fluorescence values for each tandem diabody concentration determined by flow cytometry.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), FITC-conjugated goat anti-mouse IgGmin X (Dianova), L-glutamine (Invitrogen), NaN ₃ (Carl Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

cell:

Jurkat cells (kindly provided by Dr. G. Moldenhauer (DKFZ, Heidelberg, Germany)) were cultured in RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, and 100 μg/mL streptomycin sulfate (all components were from Invitrogen, referred to herein as complete RPMI 1640 medium).

Determination of antibody affinity to CD3+ cells by flow cytometry:

Cells were incubated on ice for 45 min with 100 μL serial dilutions of the indicated antibodies starting from 100 μg/mL (about 1000 nM) in ice-cold PBS (referred to as FACS buffer) supplemented with 2% heat-inactivated FCS and 0.1% sodium azide. After washing 3 times with FACS buffer, cells were incubated on ice for 45 min with 0.1 mL 10 μg/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After the second wash cycle, under the same conditions as before, cells were incubated with 0.1 mL 15 μg/mL FITC-conjugated goat anti-mouse IgG antibody. As a control, in the absence of a primary antibody, cells were incubated with anti-His IgG 13/45/31 and then with FITC-conjugated goat anti-mouse IgG antibody. The cells were then washed again and resuspended in 0.2 mL FACS buffer containing 2 μg/mL propidium iodide to exclude dead cells. Fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) with MXP software or a Millipore Guava EasyCyte flow cytometer (Merck Millipore, Schwalbach, Germany) with Incyte software. The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany). If the Beckman-Coulter FC500 MPL flow cytometer was used for analysis, 0.5×10 ⁶ cells/staining were used, and if the Millipore Guava EasyCyte flow cytometer was used, only 0.25×10 ⁶ cells/staining were used.

After subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone, the fluorescence intensity values and the equation (hyperbola) for single-site binding of GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) were used to calculate KD values.

For the series connection diabody of each analysis, the KD value in conjunction with EGFRvIII and cytotoxic EC50 are drawn into the function of the KD value in conjunction with CD3.When the KD value in conjunction with EGFRvIII of series connection diabody only demonstrated very little variability, the EC50 cytotoxicity value demonstrated and was accompanied by the CD3 that series connection diabody increases gradually in conjunction with KD value and almost became linear increase (Figure 11).

Example 12: Inhibition of F98 ^EGFRvIII Xenograft Tumors by EGFRvIII/CD3 (In Vivo Proof of Concept)

On day 0 (d0), nine experimental groups of immunodeficient NOD/scid mice (NOD/MrkBomTac-Prkdcscid, Taconic Denmark) were xenografted with a suspension of 4× ¹⁰ F98 ^npEGFRvIII cells mixed with 1× ¹⁰ purified human PBMCs from healthy donors via subcutaneous (sc) injection. To account for possible donor variability of PBMCs, each experimental group was subdivided into two cohorts, each receiving PBMCs from only one individual donor. 2 h after tumor cell inoculation, followed by 24 h, 48 h, 72 h, and 96 h, 100 μg, 10 μg, or 1 μg of each tandem double antibody test item was administered to the animals via the tail vein (iv). The control group received vehicle only. Starting on d0, another control was administered (ip) with 1 mg of cetuximab (Erbitux) twice a week. Table 6 summarizes the dose groups and donor allocation. Tumor volume was monitored by measuring the major and minor diameters of the tumors with calipers three times a week from day 3 to day 52. Tumor volume was calculated from the diameters according to the following formula: Volume = (minor diameter) ² x major diameter x 0.5.

Table 7: Animal dose groups

References:

Schwarz M, Rottgen P, Takada Y, Le Gall F, Knackmuss S, Bassler N, Buttner C, Little M, Bode C, Peter. Single-chain antibodies for the conformation-specific blockade of activated platelet integrin alphallbbeta3 designed by subtractive selection from naive human phage libraries. FASEB J. 2004 Nov; 18(14): 1704-6.

Schier R, McCall A, Adams GP, Marshall KW, Merritt H, Yim M, Crawford RS, Weiner LM, Marks C, Marks JD. Isolation of picomolar affinity anti-c-erbB-2 single-chain Fv by molecular evolution of the complementarity determining regions in the center of the antibody binding site. J Mol Biol. 1996.

Pini A, Viti F, Santucci A, Carnemolla B, Zardi L, Neri P, Neri D. Design and use of a phage display library. Human antibodies with subnanomolar affinity against a marker of angiogenesis eluted from a two-dimensional gel. J Biol Chem. 1998.

Boder ET, Midelfort S, Wittrup KD. Directed evolution of antibody fragments with monovalent femtomolar antigen-binding affinity. Proc Natl Acad Sci U S A. 2000 Sep 26; 97(20): 10701-5.

Barbas CF 3rd,Hu D,Dunlop N,Sawyer L,Cababa D,Hendry RM,Nara PL,Burton DR.In vitro evolution of a neutralizing human antibody to humanimmunodeficiency virus type 1 to enhance affinity and broaden strain cross-reactivity.Proc Natl Acad Sci U S A.1994 Apr 26;91(9):3809-13.

Parsons HL, Earnshaw JC, Wilton J, Johnson S, Schueler PA, Mahoney W, McCafferty J. Directing phage selections towards specific epitopes. Protein Eng. 1996 Nov; 9(l 1): 1043-9.

Winkler, Kramer A, Kiittner G, Seifert M, Scholz C, Wessner H, Schneider-Mergener J, Hohne W. Changing the antigen binding specificity by single point mutations of an anti-p24 (HIV-1) antibody. J Immunol. 2000 Oct 15; 165(8): 4505-14.

Safdari Y, Farajnia S, Asgharzadeh M, Omidfar K, Khalili M. humMRl, a highly specific humanized single chain antibody for targeting EGFRvIII. Int Immunopharmacol. 2014 Feb; 18(2): 304-10.

Lorimer IA, Keppler-Hafkemeyer A, Beers RA, Pegram CN, Bigner DD, Pastan I. Recombinant immunotoxins specific for a mutant epidermal growth factor receptor: targeting with a single chain antibody variable domain isolated byphage display. Proc Natl Acad Sci US A. 1996 Dec 10; 93(25): 14815-20.

Beers R, Chowdhury P, Bigner D, Pastan I. Immunotoxins with increased activity against epidermal growth factor receptor vlll-expressing cells produced by antibody phage display. Clin Cancer Res. 2000 Jul; 6(7): 2835-43.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Choi BD, Kuan CT, Cai M, Archer GE, Mitchell DA, Gedeon PC, Sanchez-Perez L, Pastan I, Bigner DD, Sampson JH. Systemic administration of a bispecific antibody targeting EGFRvIII successfully treats intracerebral glioma. Proc Natl Acad Sci U SA. 2013 Jan 2;110(l):270-5.

Fenstermaker RA, Ciesielski MJ. Deletion and tandem duplication of exons 2-7 in the epidermal growth factor receptor gene of a human malignant glioma. Oncogene. 2000 Sep 14; 19(39): 4542-8.

Han W, Zhang T, Yu H, Foulke JG, Tang CK. Hypophosphorylation of residue Y1045 leads to defective downregulation of EGFRvIII. Cancer Biol Ther. 2006 Oct; 5(10): 1361-8.

Grandal MV, Zandi R, Pedersen MW, Willumsen BM, van Deurs B, Poulsen HS. EGFRvIII escapes down-regulation due to impaired internalization and sorting to lysosomes. Carcinogenesis. 2007 Jul; 28(7): 1408-17. Gan HK, Kaye AH, Luwor RB. The EGFRvIII variant inglioblastoma multiforme. J Clin Neurosci. 2009 Jun; 16(6): 748-54.

Reist CJ, Archer GE, Kurpad SN, Wikstrand CJ, Vaidyanathan G, Willingham MC, Moscatello DK, Wong AJ, Bigner DD, Zalutsky MR. Tumor-specific anti-epidermal growth factor receptor variant III monoclonal antibodies: use of the tyramine-cellobiose radioiodination method enhances cellular retention and uptake in tumor xenografts. Cancer Res. 1995 Oct 1; 55(19): 4375-82.

Foulon CF, Reist CJ, Bigner DD, Zalutsky MR. Radioiodination via D-aminoacid peptide enhances cellular retention and tumor xenograft targeting of an internalizing anti-epidermal growth factor receptor variant III monoclonal antibody. Cancer Res. 2000 Aug 15; 60(16): 4453-60.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Patel D, Lahiji A, Patel S, Franklin M, Jimenez X, Hicklin DJ, Kang X. Monoclonal antibody cetuximab binds to and down-regulates constitutively activated epidermal growth factor receptor vIII on the cell surface. Anticancer Res. 2007 Sep-Oct; 27(5A): 3355-66. Jutten B, Dubois L, Li Y, Aerts H, Wouters BG, Lambin P, Theys J, Lammering G. Binding of cetuximab to the EGFRvIII deletion mutant and its biological consequences in malignant glioma cells. Radiother Oncol. 2009 Sep; 92(3): 393-8.

Dreier A, Barth S, Goswami A, Weis J. Cetuximab induces mitochondrial translocalization of EGFRvIII, but not EGFR: involvement of mitochondria in tumor drug resistance? Tumour Biol. 2012 Feb;33(l):85-94.

The present invention relates to a binding protein that specifically binds to EGFRvIII and a multispecific binding protein that specifically binds to EGFRvIII and CD3. The present invention further extends to a multispecific tandem diabody that binds to EGFRvIII and CD3. In particular, the present invention relates to a highly cytotoxic EGFRvIII/CD3 bispecific tandem diabody that is used to recruit T cells to effectively kill several types of solid tumor cancers.

EGFR imbalance is associated with many cancers, and small molecules and EGFR-targeted antibodies have all been successfully used in clinical practice. However, due to the extensive normal tissue expression of EGFR, the antibodies approved for clinical use and those in development all have serious side effects. The deletion mutant (deletion variant) III (EGFRvIII) of EGFR with truncated extracellular domain and ensuring ligand-independent intrinsic activity is the most common mutant form associated with oncogenic transformation. EGFRvIII is specifically expressed in cancer tissue and is associated with various solid tumor types. The restricted EGFRvIII expression on cancer cells provides the opportunity for developing a cytotoxic antibody that specifically targets cancer, protects normal tissues, and fully reduces the side effects associated with EGFR treatment.

In a first aspect of the present invention, fully human, highly specific, high-affinity EGFRvIII binding proteins are described herein. The described EGFRvIII binding proteins have the following advantages:

Figure 9 shows that EGFRvIII/CD3 of the present invention is the most common EGFRvIII binding protein of the present invention.They do not promote or promote the very low EGFRvIII internalization from the cell surface of target positive molecule, and this helps to raise cytotoxic T-cell or NK-cell.The result in Fig. 9 shows, according to the present invention, after cell is exposed to EGFRvIII conjugated proteins under all test conditions, with respect to untreated cell, EGFRvIII receptor molecules still remains on the cell surface.These results show, EGFRvIII/CD3 conjugated proteins of the present invention surprisingly do not show any internalization trend (Fig. 9).The combination of EGFRvIII specific binding proteins of the present invention may suppress internalization, rather than inducing internalization, or may promote the increase that EGFRvIII expresses, thereby causes its cell surface density to increase.

Can use affinity maturation technology (affinity maturation technique) to make the avidity of these high-affinity associated proteins be further significantly improved to obtain that EGFRvIII is had specificity, K _D in 100pM scope or lower associated proteins.Surprising fact is, these associated proteins have demonstrated the abnormal specificity in conjunction with EGFRvIII, and natural EGFR (wild-type EGFR or EGFRwt) is not had cross reactivity.Due to from 269 amino acid whose frames (in-frame) of EGFR extracellular domain (exon 2-7), therefore with respect to EGFRwt, in EGFRvIII, form new epi-position (neo-epitope), and estimate that this new epi-position is quite little: it is arranged by new amino acid (novel juxtaposition of aminoacids) (amino acid 5 is fused to amino acid 274) and a new GLY amino acid composition at the deletion site.

The goal of affinity maturation screening methods is to improve binding affinity by promoting the selection of binding proteins with reduced off-rates ( _kOFF ). However, surprisingly, the affinity matured binding proteins shown have greatly increased on-rates (kON ₎ , while the reduction in _kOFF has only a small contribution to the 100-fold increased binding of the binding proteins, with some variable antibody domains showing _KD less than 100 pM.

These unique properties make the variable antibody domains described herein that are specific for EGFRvIII particularly suitable for the development of multispecific, e.g., bispecific, multivalent, immune effector cell engaging, tumor-targeting antibody therapeutics, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies.

In a further aspect of the invention, multivalent EGFRvIII binding antibodies are provided that have two binding sites for EGFRvIII, or in the case of a multivalent bispecific antibody, in addition to two binding sites for EGFRvIII, also have a binding site for the T-cell antigen CD3 or CD16A specifically expressed on natural killer (NK) cells.

In one embodiment, use EGFRvIII is had two binding sites and comprises in addition the EGFRvIII/CD3 series connection diabody that CD3 has two binding sites, and tested their combination with recombinant EGFRvIII-or EGFR-Fc fused antigen.Under non-reducing conditions (due to complete disulfide bond, molecular structure still part retains), or under reducing conditions (protein structure complete denaturation), EGFRvIII-or EGFR-Fc fused antigen separates by SDS-PAGE, is transferred to PVDF membrane by Western blotting (Western Blotting), and uses different EGFRvIII/CD3 series connection diabody antibodies (diabodyantibody) or EGFR-to assess in conjunction with modifying (decorate) trace (Fig. 4).This embodiment shows, comprises the parental EGFRvIII binding domains that EGFRvIII has two binding sites and the EGFRvIII binding domains of affinity maturation and also keeps its selection specificity to EGFRvIII in multivalent antibody form. Under reducing and non-reducing conditions, all EGFRvIII/CD3 series connection double antibodies all can identify EGFRvIII, but all can not identify full-length wild-type EGFR under these conditions arbitrarily.This embodiment also proves, and EGFRvIII specific binding structural domain described herein can identify linear epitope, and its reactivity does not need complete three-dimensional structure.This is opposite with EGFR-in conjunction with IgG antibody cetuximab (Erbitux (Erbitux)), EGFR-in conjunction with IgG antibody cetuximab can identify wild-type EGFR and ripe EGFRvIII, but its reactivity obviously needs complete disulfide bond and conformational epitope (Fig. 4).

In another embodiment, EGFRvIII/CD3 tandem diabodies comprising parental or affinity-matured EGFRvIII specific domains in combination with different CD3 binding domains and/or having single binding domains of different orders in the tandem diabody molecule are used. The tandem diabody with the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII - _{V H} ^CD3 contains a divalent EGFRvIII binding site in the middle of the tandem diabody molecule, while the tandem diabody with the domain order V _H ^EGFRvIII -V _L ^CD3 -V _H ^CD3 -V _L ^EGFRvIII has two EGFRvIII binding sites in the outside position and two CD3 binding sites in the middle. In ELISA, the concentration-dependent binding of different EGFRvIII-domains and domain order mutants of the tandem diabody to EGFRvIII antigen, wild-type EGFR antigen, and to the concentration-dependent binding of the CD3 antigen (Fig. 6) are analyzed. All tandem diabodies containing affinity-matured EGFRvIII binding domains showed a substantial increase in binding to EGFRvIII, clearly manifested as corresponding binding curves shifting by more than two orders of magnitude to lower concentrations ( FIG. 6 ). No binding of any tandem diabody to wild-type EGFR antigen was observed.

Can select and identify anti-EGFRvIII antibody by phage display library (phase display libraries), for example, selectively combine the antigen-binding proteins of the EGFR of maturation rather than native form (native form).T cell is the effective tumor-killing effector cell that can not be raised by natural antibody.Therefore, in a further aspect of the present invention, the invention provides one group of multispecific EGFRvIII/CD3 antigen-binding proteins, particularly series connection diabody, it can raise T cell, have the binding characteristic and cytotoxicity of a wide range.Characterized the feature of the binding characteristic of antigen-binding proteins of the present invention, the cytotoxic activity of T cell mediation and the T cell activation of target mediation in one group of in vitro test.In FACS, antigen-binding proteins of the present invention shows unusual specificity to the EGFRvIII antigen in Biacore, ELISA and EGFRvIII-positive cell.Within the full concentration range of test, do not observe any conjugated protein and EGFR antigen or with the detectable combination of EGFRwt expressing cell. The most effective high-affinity series diabody shows the cytotoxicity of EC50 for F98 glioma cells and Chinese hamster ovary celIs expressing EGFRvIII and CHO cells; Remaining series diabody shows the cytotoxicity of EC50 up to about 10000pM.Also tested these series diabody pairs as the cytotoxicity of the EGFRwt ⁺ cell of the more sensitive probe of the residue of native form.When the series diabody concentration of maximum assessment was 0.5 μM, cytotoxicity was observed on EGFRwt ⁺ cell or other EGFRvIII-negative cells.It is crucial to be combined with the high-affinity of CD3 to effectively T cell recruitment, yet, when there is no EGFRvIII ⁺ target cell in vitro, as lacking measurement by proliferation, the series diabody of the present invention does not cause T cell activation, and this contributes to good preclinical safety (Figure 10).In a word, the strict specificity and high efficiency of anti-tumor cell toxicity are mediated by EGFRvIII/CD3 series diabody.

Tandem diabodies with CD3 binding _KD ranging from 1.1 nM to approximately 500 nM but relatively constant EGFRvIII binding _KD (ranging from 2.0 nM to 6.7 nM) were generated and showed good correlation between CD3 binding strength and cytotoxic potency against cells expressing EGFRvIII, with _EC50 values ranging from 25 pM to approximately 10,000 pM (Figure 11).

Although the affinity of the antigen-binding protein of the present invention for EGFRvIII has been significantly improved, the specificity is not lost.

is a trademark of Affimed Therapeutics used to represent tandem diabodies (Kipriyanov et al., 1999, J. Mol. Biol, 293: 41-56; Cochlovius et al., 2000, Cancer Res., 60: 4336-4341; Reusch et al., 2004, Int. J. Cancer, 112: 509-518, Kipriyanov, 2009, Methods Mol Biol, 562: 177-93; McAleese and Eser, 2012, Future Oncol. 8: 687-95). In the context of the present invention, TandAb and tandem diabodies are used as synonyms.

The sequence of wild-type EGFR gene and protein is known. EGFRvIII is the result of an in-frame deletion of exons 2-7 (801bp) of the wild-type EGFR gene, resulting in 267 amino acid deletions in the receptor's outer domain and a new glycine residue at the junction of exons 1 and 8. This new amino acid juxtaposition in the EGFR extracellular domain generates tumor-specific and immunogenic epitopes.

According to first aspect of the present invention, described EGFRvIII had to specific associated proteins at least, it does not cross-react with wild-type EGFR.This associated proteins preferably comprises CDR1, CDR2, CDR3 of the antibody variable heavy chain that is selected from the CDR described in the following table 1A and CDR1, CDR2 and the CDR3 of antibody variable light chain.

Table 1A: Sequence identifiers of amino acid sequences of human light and heavy chain CDRs from binding proteins that specifically bind to EGFRvIII

Therefore, an embodiment of the present invention is an EGFRvIII binding protein having at least one EGFRvIII binding site, comprising an antibody variable heavy chain domain and/or an antibody variable light chain domain; wherein the antibody variable heavy chain domain comprises a CDR1 selected from SEQ ID NO: 27, 33, 42, 46, 49, 53, a CDR2 selected from SEQ ID NO: 28, 38, 43, 50, 54, 61, 63 and a CDR3 of SEQ ID NO: 29; the antibody variable light chain domain comprises a CDR1 selected from SEQ ID NO: 30, 34, 36, 39, 44, 47, 51, 56, 59, 64, a CDR2 selected from SEQ ID NO: 31, 40, 57 and a CDR3 selected from SEQ ID NO: 32, 35, 37, 41, 45, 48, 52, 55, 58, 60, 62 and 65.

According to further embodiment, described associated proteins comprises and has the antibody variable heavy chain domain that is selected from SEQ ID NO:1,3,5,7,9,11,13,15,17,19,21 and 25 as shown in Table 1B and/or is selected from SEQ ID NO:2,4,6,8,10,12,14,16,18,20,22 and 26 as shown in Table 1B at least one EGFRvIII binding site of antibody variable light chain domain.The aminoacid sequence of three variable light chain CDR and three variable heavy chain CDR is shown in bold and is underlined in Table 1.These structural sites show that the affinity to EGFRvIII is improved, and specificity does not have loss.These antigen-binding proteins do not cross-react with wild-type EGFR antigen.

In some embodiments, the present invention relates to an antigen-binding protein of the present invention.These antigen-binding proteins of the present invention do not show or only show minimized EGFRvIII internalization when in conjunction with EGFRvIII.According to the present invention, under the test conditions according to embodiment 9, after cell is exposed to EGFRvIII-in conjunction with antibody, be greater than 80%, preferably greater than 90%, most preferably be greater than 95% EGFRvIII receptor molecules and still remain on cell surface.The protein-binding combination of EGFRvIII of the present invention may suppress internalization, rather than inducing internalization, or may promote that the expression of EGFRvIII increases, thereby causes its cell surface density to increase.

The term "binding protein" refers to an immunoglobulin derivative with antigen-binding properties, i.e., an immunoglobulin polypeptide or fragment thereof containing an antigen-binding site. The binding protein comprises the variable domains of an antibody or fragment thereof. Each antigen-binding domain is formed by an antibody, i.e., an immunoglobulin, a variable heavy chain domain (VH), and an antibody variable light chain domain (VL), which bind to the same epitope. The variable light chain domain or variable heavy chain domain of the present invention is a polypeptide comprising CDR1, CDR2, and CDR3. Preferably, the binding protein of the present invention lacks immunoglobulin constant domains. The term "binding protein" also refers to antibody fragments or derivatives, including Fab, Fab', F(ab') ₂ , Fv fragments, single-chain Fv, diabodies, tandem diabody flexibodies (WO03/025018), and tandem single-chain Fv ((scFv) ₂ ).

Table 1B: Amino acid sequences of all anti-EGFRvIII variable heavy chain domains (VH) and variable light chain domains (VL) (amino acid sequences of CDR1, CDR2, and CDR3 are shown in bold and underlined)

In a preferred embodiment, the binding protein that confers specificity for EGFRvIII comprises an antigen binding site from one of the following combinations of variable heavy chain domains and variable light chain domains shown in Table 1:

(i) SEQ ID NO: 1 and SEQ ID NO: 2,

(ii) SEQ ID NO: 3 and SEQ ID NO: 4,

(iii) SEQ ID NO: 5 and SEQ ID NO: 6,

(iv) SEQ ID NO: 7 and SEQ ID NO: 8,

(v) SEQ ID NO: 9 and SEQ ID NO: 10,

(vi) SEQ ID NO: 11 and SEQ ID NO: 12,

(vii) SEQ ID NO: 13 and SEQ ID NO: 14,

(viii) SEQ ID NO: 15 and SEQ ID NO: 16,

(ix) SEQ ID NO: 17 and SEQ ID NO: 18,

(x) SEQ ID NO: 19 and SEQ ID NO: 20,

(xi) SEQ ID NO: 21 and SEQ ID NO: 22, or

(xii) SEQ ID NO:25 and SEQ ID NO:26.

The present invention also provides not only EGFRvIII has specificity, and also has the associated proteins of at least one other functional domain.In a further embodiment, described at least one other functional domain is effector domain (effector domain)." effector domain " is that effector cell is had specific binding site, and it can stimulate or trigger cytotoxicity, engulfment, antigen presentation, cytokine release.Such effector cell is, for example but not limited to, T-cell or NK-cell.Particularly, described other effector domains comprise at least one antibody variable heavy chain domain and at least one variable light chain domain that form the antigen binding site of CD3 preferred mankind CD3.

In some embodiments, the EGFRvIII conjugated proteins of the present invention can be multispecific.Term used herein " multispecific " means to refer to that conjugated proteins of the present invention have at least two antigen binding sites to different epi-positions, and wherein at least one is used for EGFRvIII.For example, conjugated proteins can be trispecific, and have binding site and have at least one binding site to effector cell such as epi-position on CD3 or antigen to two different epi-positions and/or antigen on tumor cell.Conjugated proteins can have binding site to different epi-positions of same antigen and/or the epi-position of different antigens.Such multispecific conjugated proteins comprise single-chain diabody, (scFv) ₂ , series diabody and elastomeric antibody (referring to Le Gallet al., 1999FEBS Letts 453:164-168 and WO 03/025018).In a specific embodiment of the present invention, conjugated proteins are to EGFRvIII and CD3 bispecific.

The CD3 binding site of the multispecific binding protein of the invention can be composed of a variable heavy chain domain (SEQ ID NO: 23) and a variable light chain domain (SEQ ID NO: 24) as shown in Table 2, or a close homolog of such sequences that differs from the CDR sequences by only two amino acids, or any other fully human, humanized or non-human variable antibody domain specific for CD3 (referred to as CD3), such as, for example, a variable antibody domain specific for CD3, in particular CD3 epsilon, or any other fully human, humanized or non-human variable antibody domain, such as, for example, the variable antibody domain of UCHT1 or OKT3.

Table 2: Amino acid sequences of fully human anti-CD3 variable heavy chain domain (VH) and variable light chain domain (VL) (amino acid sequences of CDR1, CDR2 and CDR3 are shown in bold and underlined)

In addition, EGFRvIII conjugated proteins of the present invention can be multivalent.Term used herein " multivalent " means to refer to, and conjugated proteins of the present invention comprises at least two antigen binding sites.Antigen binding site can have identical or different specificity.In embodiments of the present invention, conjugated proteins is in conjunction with identical epi-position with at least two binding sites in a divalent manner.The protein-conjugated example of divalence is diabody, and the protein-conjugated example of at least tetravalence is a series connection diabody.

In a further aspect of the present invention, described EGFRvIII associated proteins and bispecific EGFRvIII and CD3 binding site are humanized or fully human, i.e. people-source. In a further aspect of the present invention, EGFRvIII associated proteins or multispecific EGFRvIII and CD3 associated proteins are dimers, i.e. comprise two polypeptides with EGFRvIII and CD3 antigen binding sites.

According to the present invention, dimerization bispecific EGFRvIII and CD3 binding proteins are provided in the form of series connection diabody.Such series connection diabody is by connecting four antibody variable binding domains (heavy chain variable domain (VH) and light chain variable domain (VL) in single gene construct (referring to McAleese and Eser, 2012, Future Oncol.8:687-95)) that can homodimerize and build.In such series connection diabody, joint (linker) length is that it prevents the intermolecular pairing of variable domain so that molecule can not fold back to form single-chain diabody on itself, but is forced to pair with the complementary domain (complementary domain) of another chain.These domains can also be arranged like this, namely make corresponding VH and VL domain pair during dimerization.After being expressed by single gene construction, two same polypeptide chains head to tail folding form the functional non-covalent homodimer (McAleese and Eser, 2012, Future Oncol.8:687-95) of about 110kDa. Despite the absence of intramolecular covalent bonds, homodimers, once formed, are highly stable, remain intact and do not revert to monomeric form.

In one embodiment of a multispecific tandem diabody, a humanized bispecific tetravalent antibody is provided that has two binding sites each for CD3 and EGFRvIII (EGFRvIII/CD3RECRUIT-) to utilize the cytotoxic capacity of T cells to treat glioblastoma (GB), prostate cancer, head and neck cancer, and other epidermal growth factor receptor mutant vIII-positive (EGFRvIII+) cancers.

In the present invention, the tandem diabody is the most widely used antibody of the present invention.Relative to traditional monoclonal antibody, tandem diabody has a large amount of properties that advantage is provided, also has less multispecific molecule.Tandem diabody is fully functional when there is no glycosylation.Tandem diabody can only contain antibody variable domains, and therefore can not have any side effect that can be relevant to Fc part.Because bispecific tandem diabody allows divalence to be bound to each in EGFRvIII and CD3, therefore, its avidity (avidity) is identical with the avidity of IgG.The size of tandem diabody is about 110kDa, is less than IgG, and this can enhance tumor infiltration.But, compared with the less bispecific form based on antibody-binding domain or non-antibody scaffold, this size is much higher than the kidney threshold of first-pass clearance (first-pass clearance), and this provides pharmacokinetic advantage. The generation and preparation of such tandem diabodies is described, for example, in Kipriyanov SM: Methods Mol. Biol. 2009; 562: 177-93, Kipriyanov SM: Methods Mol. Biol. 2003; 207: 323-33 or McAleese and Eser, 2012, Future Oncol. 8: 687-95.

Tandem diabodies are well expressed in CHO cells, enabling efficient upstream and downstream manufacturing processes to be put in place.

The EGFRvIII and CD3 bispecific tandem double antibody of the present invention is designed to specifically target EGFRvIII ⁺ tumor cells by recruiting cytotoxic T cells. Antibodies cannot directly recruit cytotoxic T cells. On the contrary, by employing (engage) the CD3 molecules specifically expressed on these cells, the tandem double antibody can make cytotoxic T cells and EGFRvIII ⁺ tumor cells cross-linked in a highly specific manner, thereby significantly increasing the cytotoxic effect of these molecules. The tandem double antibody shows strong, specific and effective cytotoxicity. Significant evidence is that T cells can play a role in controlling tumor growth. For example, the presence of cytotoxic T cells in colorectal tumors and lymph nodes from NHL patients is shown to be correlated with better clinical results. In addition, for melanoma vaccines, and antibodies for CTLA-4, a negative regulator of T-cell activation, have demonstrated the potential for the therapy of designing induction of T-cell responses. The tandem double antibody of the present invention employs cytotoxic T cells by binding surface-expressed CD3, preferably CD3ε, which forms a part of the T-cell receptor. The simultaneous binding of this tandem diabody to CD3 and EGFRvIII expressed on the surface of specific tumor cells leads to T-cell activation and mediates subsequent tumor cell lysis.

"Dimer" refers to a complex of two polypeptides. Preferably, the two polypeptides are non-covalently associated with each other, particularly under conditions where there is no covalent bond between the two polypeptides. Preferably, the bispecific dimer is a homodimer, i.e., comprises two identical polypeptides. However, a trispecific or other multispecific dimer can be a heterodimer and comprise two different polypeptides, for example, at least one binding site of the variable light chain domain is located on one polypeptide and the variable heavy chain domain is located on the other domain. The term "polypeptide" refers to a polymer of amino acid residues connected by amide bonds. Preferably, the polypeptide is an unbranched single-chain fusion protein. In this polypeptide, the variable antibody domains are connected in sequence. In addition to the variable domain N-terminus and/or C-terminus, the polypeptide may also have contiguous amino acid residues. For example, such contiguous amino acid residues may include a Tag sequence that may be beneficial to polypeptide purification, preferably at the C-terminus.

Each polypeptide of the bispecific tandem diabody contains at least four variable domains, namely the variable light chain (VL) and variable heavy chain (VH) of the CD3 binding protein and the variable light chain (VL) and variable heavy chain (VH) of the EGFRvIII binding protein. The four binding domains are connected by peptide linkers L1, L2 and L3 and can be arranged as follows from N-terminus to C-terminus:

(i)VL(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3); or

(ii) VH(CD3)-L1-VL(EGFRvIII)-L2-VH(EGFRvIII)-L3-VL(CD3); or

(iii) VL(EGFRvIII)-L1-VH(CD3)-L2-VL(CD3)-L3-VH(EGFRvIII); or

(iv) VH(EGFRvIII)-L1-VL(CD3)-L2-VH(CD3)-L3-VL(EGFRvIII).

In an embodiment of the present invention, the four variable domains are arranged as VL ^{(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3). Tandem diabodies having the domain order VLCD3-VHEGFRvIII -VLEGFRvIII-VHCD3} _{and containing} ^different _EGFRvIII ^binding _sequences , despite having different affinities for EGFRvIII, all showed very similar binding to the second specific CD3, as shown by ELISA on the CD3γε antigen ( FIG5A ). In other tandem diabody domain sequences containing EGFRvIII binding domains in the outer positions and two CD3 binding sites in the middle ( _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ), a significant improvement in EGFRvIII binding to the affinity-matured EGFRvIII binding domain was also evident, while binding of the tandem diabody middle domain to CD3 was slightly weaker and more variable between individual tandem diabodies than the first domain sequence (Figure 5B). Similar observations were made when tandem diabodies containing different EGFRvIII-binding domains were constructed using different CD3 domains (SEQ ID NOs: 23 and 24) (Figure 5C).

The length of the linker affects the flexibility of the antigen-binding tandem diabody. For example, Todorovska et al., 2001 Journal of Immunological Methods 248:47-66; Perisic et al., 1994 Structure 2:11217-1226; Le Gall et al., 2004 Protein Engineering 17:357-366 and WO94/13804 describe the effect of linker length on the formation of dimeric antigen-binding proteins.

According to the present invention, the length of the peptide linkers L1, L2, and L3 is preferably such that the domains of one polypeptide can intermolecularly bind to the domains of another polypeptide to form a dimeric antigen-binding tandem diabody. Such linkers are "short," i.e., comprised of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, or 12 amino acid residues. Thus, the linker consists of approximately 12 or fewer amino acid residues. In the case of 0 amino acid residues, the linker is a peptide bond. Such short linkers facilitate intermolecular dimerization of the two polypeptides by binding and forming a correct antigen-binding site between the antibody variable light chain domain and the antibody variable heavy chain domain of the different polypeptides. Short linkers of approximately 12 or fewer amino acid residues generally prevent intermolecular interaction between adjacent domains of the same polypeptide chain. In one embodiment of the present invention, these linkers consist of approximately 3 to approximately 10, e.g., 4, 5, or 6, consecutive amino acid residues.

Regarding the amino acid composition of these linkers, peptides are selected that do not interfere with the dimerization of the two polypeptides. For example, linkers containing glycine and serine residues generally provide protease resistance. The amino acid sequence of the linker can be optimized, for example, by phage display methods to increase antigen binding and the yield of antigen-binding polypeptide dimers. Examples of peptide linkers suitable for the tandem diabodies of the present invention are GGSGGS (SEQ ID NO: 66), GGSG (SEQ ID NO: 67), or GGSGG (SEQ ID NO: 68).

EGFRvIII associated proteins as herein described and multi-specific series connection diabody can produce by the polynucleotide of expressing coding series connection diabody polypeptide, and the polypeptide of described series connection diabody is combined to form antigen-in conjunction with series connection diabody with another identical polypeptide.Therefore, the further embodiment of the present invention is the polynucleotide of coding antigen as herein described-in conjunction with series connection diabody polypeptide, for example DNA or RNA.

Polynucleotide can be constructed by methods well known to those skilled in the art, for example, by combining the gene of at least four antibody variable domains separated by the sequence encoding peptide linkers or directly connected by peptide bonds into a single gene construct, and in bacteria or other suitable expression systems such as for example expressed in Chinese hamster ovary celI, the single gene construct is operably connected with a suitable promotor, alternatively a protein tag for detection and purification, and a suitable transcription terminator. According to employed vector systems and hosts, any number of suitable transcription and translation elements comprising formation and inducible promoters can be used. Select promoter so that it drives the expression of polynucleotides in respective host cells.

As a further embodiment of the present invention, the polynucleotide can be inserted into a vector, preferably an expression vector. The recombinant vector can be constructed according to methods known to those skilled in the art.

Various expression vector/host systems can be used to contain or express polynucleotides encoding polypeptides of the present invention. Examples of expression vectors for expression in E. coli are pSKK (Le Gall et al., J Immunol Methods. (2004) 285(1): 111-27) or pcDNA5 (Invitrogen) for expression in mammalian cells.

Thus, the antigen-binding tandem diabodies described herein can be prepared by introducing a vector encoding the polypeptides described above into host cells and culturing the cells under conditions whereby the polypeptide chains are expressed, can be isolated from the culture medium, and optionally thereafter purified.

In a further embodiment, the present invention provides a pharmaceutical composition comprising an EGFRvIII binding protein, an antigen-binding tandem diabody, a vector comprising a polynucleotide encoding the polypeptide of the antigen-binding tandem diabody or a host cell transformed by the vector, and at least one pharmaceutically acceptable carrier. The term "pharmaceutically acceptable carrier" means any carrier that does not interfere with the effectiveness of the biological activity of the component and is non-toxic to the administered patient. Suitable pharmaceutical carrier examples are known in the art and include physiological saline, phosphate buffered saline, water, emulsions such as oil/water emulsions, various types of wetting agents, sterile solutions, etc. Such carriers can be prepared by conventional methods and administered to an individual at a suitable dosage. Preferably, the composition is sterile. These compositions may also include adjuvants, such as preservatives, emulsifiers, and dispersants. The effects of microorganisms can be prevented by including various antibacterial and antifungal agents.

The skilled artisan will readily construct and obtain antigen-binding proteins such as the tandem diabodies described herein without undue burden by using well-established techniques and standard methods known in the art, see, for example, Sambrook, Molecular Cloning A Laboratory Manual, Cold Spring Harbor Laboratory (1989) N.Y.; The Protein Protocols Handbook, John M. Walker, ed., Humana Press Inc. (2002); or Antibody engineering: methods and protocols/Benny K.C. Lo, ed.; Benny K.C. II Series: Methods in molecular biology (Totowa, N.J.).

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 illustrates the position of variable sequence in the EGFRvIII specific high-affinity binding domain derived after affinity maturation.The CDR position that is different from the heavy chain and light chain variable region of parental sequence is marked with asterisk, and amino acid is highlighted with grey separately.The position with useful sudden change is highlighted with dark grey.

Fig. 2 shows that among Biacore T200, multi-circulation kinetics measures the result that different scFv antibodies are attached to recombinant EGFRvIII-Fc antigen or wild-type EGFR-Fc antigen.Sensogram is presented at the kinetics of surface plasmon resonance (SPR) response unit (RU) measured in time under 4 different scFv antibody concentration levels (86nM, 17.2nM, 3.4nM, 0.69nM).Measurement is in conjunction with 180 seconds, after this measurement dissociates.Use 1:1 to combine model analysis data.

Figure 3 shows the results of flow cytometry measurement of scFv antibody binding to (A) F98 rat glioma cells overexpressing EGFRvIII (F98 ^EGFRvIII ), (B) F98 cells overexpressing native EGFR (F98 ^EGFR ), and (C) untransfected F98 cells (F98), as well as the results of similar scFv antibody binding to (D) CHO cells overexpressing EGFRvIII (CHO ^EGFRvIII ), (E) CHO cells overexpressing native EGFR (CHO ^EGFRvIII ), and (F) untransfected CHO cells (CHO).

Fig. 4 illustrates the tandem diabody containing different EGFRvIII binding domains and is separated by SDA-PAGE and is transferred to the EGFRvIII-or natural EGFR-Fc fusion antigen (non-reducing protein or the albumen with DTT reduction) of the purification of PDVF film by western blotting.Fc fusion rotein is by disulfide bond at Fc part dimerization, but is monomer after reducing with DTT.EGFRvIII-specific antibody is bonded to EGFRvIII-fusion rotein with reduction or non-reduction conformation, but is not combined with natural EGFR.This anti-EGFR antibody cetuximab is as contrast, and identifies non-reducing EGFRvIII-Fc and non-reducing EGFR-Fc antigen, but does not identify the antigen of DTT-processing.

Figure 5 shows the concentration-dependent binding of different EGFRvIII/CD3-specific tandem diabodies to EGFRvIII-Fc, EGFR-Fc, and CD3γε recombinant antigens analyzed in ELISA. (A) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , (B) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies having the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , and (C) shows the binding signals of the EGFRvIII/CD3-specific tandem diabodies containing the CD3 domains of SEQ ID NOs: 23 and 24 and having the domain sequence _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 . The improvement in binding of tandem diabodies containing affinity-matured EGFRvIII-binding domains to EGFRvIII was clearly detectable as a shift of the respective binding curves to lower concentrations in all formats/domain orders.

Figure 6 shows the results of multi-cycle kinetic measurements of EGFRvIII/CD3-specific tandem diabodies containing different bivalent EGFRvIII binding domains in the order of _V ^CD3 - _V ^EGFRvIII - _V ^EGFRvIII - _V ^CD3 bound to recombinant EGFRvIII-Fc antigen in a Biacore T200. Sensogram displays the kinetics of surface plasmon resonance (SPR) response units (RU) measured over time at different tandem diabody concentration levels. Binding was measured for 180 seconds, followed by dissociation. Data were analyzed using a 1:1 binding model.

Fig. 7 shows the immunohistochemistry (IHC) staining result of tissue section from (A) three individual head and neck (H＆N) cancer patients, (B) three individual glioblastoma patients, and (C) representative results from prostate cancer, mammary cancer (Her2neg) and non-small cell lung cancer (NSCLC).The EGFRvIII-specific double antibody containing EGFRvIII-specific binding domain Li3G30 of the divalent conformation of 2 kinds of different concentrations is dyed in section, and described divalent conformation is similar to the arrangement of EGFRvIII targeting tandem double antibody antibody (for example, (V _L ^CD3- ) V _H ^EGFRvIII - _{V L} ^EGFRvIII (-V _H ^CD3 )) but lacks CD3 binding part.As a control, section is dyed with the EGFR-specific IgG antibody cetuximab of identification EGFR (being present on healthy tissue and cancerous tissue) and EGFRvIII.The specificity of dyeing is confirmed by the control section hatched in the secondary reagent that does not contain EGFR-or EGFRvIII-specific primary antibody.

Fig. 8 illustrates that on EGFRvIII positive cells, compared with the tandem diabody of EGFRvIII and CD3 respectively containing the parental EGFRvIII binding domain (Li3G30) before affinity maturation, the cytotoxicity ability, effectiveness and target-dependent specificity of EGFRvIII/CD3 tandem diabody containing different affinity maturation EGFRvIII-specific binding domains: (A) ABC 470 containing EGFRvIII/CD3 tandem diabody, (B) ABC 471 containing EGFRvIII/CD3 tandem diabody, (C) ABC 472 containing EGFRvIII/CD3 tandem diabody. Using CHO cells expressing EGFRvIII (CHO ^EGFRvIII ), expressing EGFRvIII negative EGFR (CHO ^EGFR ) or untransfected CHO cells as target cells and using human PBMC as effector cells, in the analysis based on flow cytometry, antibody concentration-dependent cytotoxicity was measured. In each panel (AC), tandem diabodies containing a single EGFRvIII binding domain and CD3 domain v64, in the order _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 , are shown on the left, and tandem diabodies containing a single EGFRvIII binding domain combined with a different CD3 domain, in the order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII , are shown on the right.

Figure 9 shows the results of evaluating EGFRvIII receptor internalization from the cell surface when combined with EGFRvIII-specific antibodies. Cells expressing EGFRvIII were incubated for 24 hours at 37°C with varying concentrations of three different ^EGFRvIII / ^CD3 tandem diabody antibodies with the domain order V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII - _{V H} ^CD3 (i) or _{V H} EGFRvIII-V L ^CD3 -V H CD3-V _L ^EGFRvIII (iv), and/or EGFRvIII/CD3 tandem diabody antibodies containing different CD3-binding domains. After adjustment, all remaining EGFRvIII receptor molecules on the cell surface were stained with the respective tandem diabody antibodies at a saturating concentration of 10 μg/ml.

Figure 10 is presented in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) series double antibody antibody, assesses the result of cell proliferation (BrdU incoporation) in human PBMC culture.PBMC is used as positive control, and it breeds in the presence of CD3 binding antibody OKT3 or phytohemagglutinin (plant lectin phytohemagglutinin) (PHA).In the absence of EGFRvIII-positive target cells, with different EGFRvIII/CD3 series double antibodies, hatch PBMC and do not induce activation and the proliferation of PBMC.

Figure 11 illustrates that the CD3-binding affinity of EGFRvIII/CD3 tandem diabodies is positively correlated with their cytotoxic capacity. Regarding CD3-binding on CD3 ⁺ -Jurkat cells, EGFRvIII-binding on CHO ^EGFRvIII cells and cytotoxic capacity, all 20 EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) in the domain order V _L ^CD3 -V _H ^EGFRvIII -V _L ^EGFRvIII -V _H ^CD3 and different CD3-binding domain combinations with different affinities for CD3 were analyzed. The scope of CD3-binding K _D value is 1nM to approximately 500nM. For each analyzed tandem diabodies, relative to CD3-binding K _D value, EGFRvIII-binding K _D value and cytotoxic EC ₅₀ value are mapped. While the TandAbs showed only small changes in EGFRvIII-binding _KD values, the cytotoxicity _EC50s showed an almost linear increase as the CD3-binding _KD values of the tandem diabodies increased.

Figure 12 shows the in vivo proof-of-concept results from the study. Two EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII-binding domain (Li3G30) but different CD3-binding domains or domain sequences were analyzed for dose-dependent inhibition of subcutaneous F98 ^EGFRvIII xenograft tumor growth and compared with the effect of cetuximab administered at a much higher concentration. (A) Dose-dependent tumor growth inhibition caused by the ^EGFRvIII /CD3 ^v6 tandem diabody with the domain sequence _VL ^CD3 - _VH EGFRvIII- _VL ^EGFRvIII - _VH ^CD3 ; (B) Dose-dependent tumor growth inhibition caused by EGFRvIII/CD3 tandem diabodies containing different high-affinity CD3-binding domains and the domain sequence _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII .

The present invention is further illustrated by the following examples, but the present invention is not limited thereto.

Example 1: Discovery and affinity maturation of EGFRvIII-specific binding proteins:

Li3G30 Discovery:

The phage display library of people's scFv sequence based on IgM source is carried out two to three rounds of panning (panning) and EGFRvIII is had specific binding substances (binders) with enrichment.Before every round of panning, described library is pre-incubated with the binding substances of (deplete) wild-type form EGFR or the binding substances of fusion rotein Fc part with EGFR-Fc.On the solid phase of EGFRvIII coating and in solution, select (selection) in parallel.After two rounds and three rounds of panning, pick out monoclonal, induce soluble scFv to express and screen the extract that is combined with EGFRvIII-Fc and EGFR-Fc.

According to the present invention, the fully human high specific variable antibody domains combined with EGFRvIII have been described, and EGFRvIII is the deletion mutant of EGFR, and it is specifically expressed in tumor rather than on healthy tissue with the form of mutation.Full human EGFRvIII specific domains described herein were initially found in the phage display screening of the fully human scFv library using based on IgM source.Surprisingly, although two or three rounds of panning were carried out using exhaustion step (depletion procedure) to wild-type EGFR antigen, this should be enriched to have specific binding substances to EGFRvIII, but in ELISA, most of the scFvs of gained are all combined to two kinds of albumen (EGFRvIII and wild-type EGFR).Identified a high EGFRvIII specific sequence, called after Li3G30.Except that there is little difference in the N-terminal of VH and VL, the framework region of heavy chain and light chain and CDR1 and CDR2 are identical with sequence VH5-51 and VL3-25 of germline encoding respectively.Li3G30 was identified from the third round of liquid phase panning.

Affinity Maturation:

The library of antibody framework pair VH5-51/VL3-25 was designed using an algorithm ordered from DistributedBio. The specificity determining residues of Li3G30 were identified and included in the library design. The design includes 52 randomized CDR sites, each with a separately defined amino acid distribution. Three different loop lengths were built into the VL CDR3. Gene fragments encoding the randomized sites of VH and VL were ordered from Geneart/Lifetechnologies and synthesized using TRIM-technology. The fragments were cloned into the phage display vector pEXHAM (Schwarz et al. 2004) to a final library size of 3.7E+8 transformed E. coli cells. The library was packaged into phage particles and, through panning and screening steps, separated with improved affinity and retained variants specific for EGFRvIII. Panning was performed using the EGFRvIII-Fc antigen immobilized on a protein-bound plastic surface. The elutriation step is designed to the scFv that helps to have the EGFRvIII bonded with low dissociation rate ( _kOFF ) by implementing following steps: washing step comprises several washing buffer incubation steps that continue up to 30 minutes, to help select the scFv that slowly dissociates.Another step is based on the overnight competition with solubility EGFRvIII.In order to guarantee in the process of affinity maturation, can not select EGFR-EGFRvIII cross-reaction antibody or the antibody in conjunction with Fc, before first elutriation step, by the pre-incubation of wild-type EGFR-Fc, the library is exhausted possible wild-type EGFR binding substances.And owing to the existence of solubility EGFR in the process of the elutriation step being carried out to fixing EGFRvIII, screening scheme is unfavorable for cross-reaction binding substances.

The unique EGFRvIII-specific scFv was further tested for binding to transfected CHO and F98 cells expressing EGFRvIII using flow cytometry. ScFvs that showed binding to cells expressing wild-type EGFR and/or untransfected cells were excluded from further analysis. Initial _KD ranking using Biacore X100 was used to identify scFvs with improved affinity for EGFRvIII compared to the original clone, Li3G30. The best variants were characterized using detailed SPR measurements and fluorescence-based stability assays.

For VL CDR3, the initial library for affinity maturation comprises three kinds of different loop lengths, and known all lengths are compatible with VH5-51/VL3-25 framework.Surprisingly, in the selection process, the middle loop length is obviously preferred.Therefore, compared with the anti-EGFRvIII antibody of parent generation, the VL CDR3 of the variant of all affinity maturations has all shortened one amino acid (table 1, Fig. 1).The sequence of selected scFv is compared with the input library, and this analysis has shown 3 useful sudden changes in VL CDR3.The heavy chain of the anti-EGFRvIII antibody of parent generation and the CDR1 and CDR2 of light chain are identical with the sequence encoded separately of germline sequence VH5-51 and VL3-25.In 75% of the sites, the residue of each germline coding is reconstructed (reconstituted) or obviously preferred.Shown a useful sudden change in VH CDR1, one in VL CDR1 (Fig. 1).The amino acid distribution in the remaining randomization site is very close to the distribution in the library before selecting.

Example 2: Measurement of scFV binding to EGFRvIII and wild-type EGFR in Biacore:

Preparation of anti-hu Fc IgG CM5-chip:

Anti-hu Fc IgG CM5-chip was prepared by covalent amine coupling using an amine coupling kit (Amine-Coupling-Kit) (GE) according to the manufacturer's instructions. IgG was diluted to a concentration of 25 μg/ml in the provided immobilization buffer (10 mM sodium acetate, pH 5.0). For the coupling step, the predefined method "Amine" of the Biacore T200 control software was applied. A target level of 5000-10000 RU was achieved in all 4 flow cells. Before injecting the recombinant protein solution, the surface of the chip was activated with EDC/NHS. The remaining binding sites on the surface were blocked with ethanolamine. 1×HBS-P+ was selected as the running buffer for the entire coupling step.

Concentration range of single-chain Fv’s (scFv) used in SPR measurements:

Before SPR measurement, the ScFv concentration was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate the molar concentration of the His-tagged scFv antibody, a molecular weight of approximately 28 kD was assumed. The scFv was adjusted to the following final concentrations by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 430 nM, 86 nM, 17.2 nM, 3.44 nM, 0.688 nM, and 0 nM.

ScFv measurement conditions:

EGFRvIII-Fc and wild-type EGFR-Fc fusion proteins were diluted in 1× HBS-P+ to a concentration of 6.25 nM. Both antigens were injected at a flow rate of 10 μl/min: EGFRvIII-Fc for 75 seconds and wild-type EGFR-Fc for 90 seconds. EGFRvIII-Fc was injected into flow cell 2, and wild-type EGFR-Fc was injected into flow cell 4. Flow cells 1 and 3 were left blank.

1× HBS-P+ was used as running buffer throughout the procedure. The scFv antibody was injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. The signal measured in flow cell 3, where no antigen was captured, was subtracted from the signal in flow cell 4 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0M _MgCl2 (30 seconds, 10 μL/min). Before each measurement, four "start-up cycles" were performed without scFv. ScFv solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

The kinetics of the scFv antibodies were determined in the MCK (Multi Cycle Kinetic) measurement. To estimate the apparent binding affinity, the binding curves were evaluated using the "Kinetic" method included in the Biacore evaluation software. Using this method, all binding curves were fitted using the "Langmuir 1:1 interaction model" with the software. The results are shown in Table 3.

The affinity maturation screening step that is used is intended to improve the EGFRvIII binding affinity of EGFRvIII specific binding domain more easily by making the selection of the binding substances with the dissociation rate ( _kOFF ) of reduction.Yet, surprisingly, the affinity maturation binding substances (table 3) of gained shows the association rate ( _kON ) that significantly increases, and the reduction of _kOFF reaches 100 times of combination for the raising of scFv and only makes less contribution, and some binding substances demonstrate the _KD lower than 100pM simultaneously (table 3).Especially surprisingly, although the screening step is conducive to the scFv that selects the _kOFF reduction, the reduction of realization _KD is mainly due to the raising of _kON.This also is surprising, because the affinity maturation of antibody is relevant (Schier et al., 1996, Pini et al., 1998, Boder et al., 2000) with the reduction of _kOFF conventionally.

Table 3: Summary of scFv binding affinities ("VH" is the variable heavy chain domain of the scFv, "VL" is the variable light chain domain of the scFv). _K values, association rates (kON) and dissociation rates ( _kOFF ) of different EGFRvIII binding domains in scFv format for recombinant EGFRvIII-Fc antigen measured in a Biacore ( _SPR ) 1:1 binding model; "enhancement factor" refers to the fold change relative to the parental scFv, i.e., Li3G30 (VH SEQ ID NO: 25 and VL SEQ ID NO: 26).

Eight of the 11 affinity-matured scFvs showed at least a 5-fold improvement in _kON compared to the parental scFv; three of these had a 20-fold higher _kON . Compared to the improvement in _kON , the effect of affinity maturation on _kOFF was quite small. Only three affinity-matured scFvs showed a 4-5-fold improvement in _kOFF . Interestingly, two of these were isolated from a competitive selection method, which may be a more effective method for selecting scFvs with reduced _kOFF than prolonged incubation in wash buffer.

The average melting temperature of the selected affinity matured candidates determined by differential scanning fluorimetry (DSF) was 62.2°C, ranging from 53°C to 67.2°C.

Maintaining high specificity for EGFRvIII relative to wild-type EGFR during affinity maturation

Affinity maturation is usually accompanied by small changes in binding epitopes, which can affect specificity/cross-reactivity (Barbas et al, 1994, Parsons et al., 1996, Winkler et al., 2000). Due to the in-frame deletion of 269 amino acids in the wild-type EGFR extracellular domain (exon 2-7), an EGFRvIII new epitope relative to wild-type EGFR is formed, and it is predicted to be quite small: it is composed of new amino acids in parallel (amino acid 5 is fused to amino acid 274) and a new GLY amino acid at the deletion site. Even if it is expected that small changes in binding epitopes will quickly destroy the abnormal specificity of EGFRvIII in this case, a cross-reactive conjugate that can also identify native EGFR can be obtained.

Surprisingly, the affinity maturation described herein resulted in up to 100-fold improved binding of anti-EGFRvIII antibodies without loss of exclusive specificity for EGFRvIII (Table 3, Figure 2).

During the selection of affinity-matured binders, the binding of the only EGFRvIII-specific scFv to CHO cells expressing EGFRvIII (CHO ^EGFRvIII ) and F98 cells (F98 ^EGFRvIII ), as well as to CHO cells expressing human wild-type EGFR (CHO ^EGFR ) and F98 cells (F98 ^EGFR ), and to untransfected CHO and F98 cells were tested in flow cytometry.

Characterized in more detail 11 high-specificity anti-EGFRvIII scFv that do not have any measurable background combination with wild-type EGFR positive cell (CHO ^EGFR ) or the Chinese hamster ovary celI of non-transfected.In the best case, measured the avidity to 80pM of EGFRvIII antigen, corresponding to 100 times of raising coefficient (table 3) compared with parental scFv.Simultaneously, the anti-EGFRvIII scFv that is detected all does not demonstrate any cross reactivity (Fig. 2) with wild-type EGFR antigen.

To our knowledge, such high affinity for EGFRvIII has never been observed for a human antibody domain that binds exclusively specifically to EGFRvIII, without measurable cross-reactivity with wild-type EGFR. Safdari and colleagues recently described results for humanizing the murine EGFRvIII-binding antibody domain MR1 (Safdari et al., 2014), which was first described in its murine form by Lorimer and colleagues (Lorimer et al., 1996, Beers et al., 2000, Kuan et al., 2000). In their humanization approach, they achieved an increase in the affinity of humMR1, reaching similar picomolar _K values reported here. However, in contrast to the present invention, they did not completely eliminate the cross-reactivity of the antibody domain with wild-type EGFR, as can be clearly seen in Figure 4 of their publication (Safdari et al., 2014). Thus, after SDS PAGE and Western blotting, humMR1 was not only specific for EGFRvIII (145 kD), but also recognized EGFR (170 kD) (Safdari et al., 2014). Applicants extensively tested the specificity and cross-reactivity of the EGFRvIII binding domain of the present application before and after affinity maturation and surprisingly found no cross-reactivity signals with wild-type EGFR or purified recombinant wild-type EGFR antigen in Biacore (Figure 2), SDS PAGE and Western blotting (Figure 4), or ELISA (Figure 5). The EGFRvIII binding domain of the present application also did not exhibit any cross-reactivity with wild-type EGFR overexpressed on the cell surface of transfected CHO cells (CHO ^EGF ) or F98 glioma cells (F98 ^EGFR ) (Figure 3).

Example 3: High specificity and high affinity of EGFRvIII-specific scFv antibodies were assessed by flow cytometry for binding to F98 rat glioma cells (F98 ^EGFRvIII ) or stably transfected CHO cells (CHO ^EGFRvIII ) expressing EGFRvIII; no binding to F98 cells (F98 ^EGFR ) or untransfected F98 cells (F98) expressing native EGFR, and no binding to CHO cells (CHO ^EGFR ) or untransfected CHO cells (CHO) expressing native EGFR.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), Ficoll Paque PLUS (GE Healthcare), FITC-conjugated goat anti-human IgG (Dianova), FITC-conjugated goat anti-mouse IgG, min X (Dianova), L-glutamine (Invitrogen), _NaN3 (Carl Roth), PBS (Invitrogen), Penicillin/Streptomycin (Invitrogen), Propidium iodide (Sigma), RPMI-1640 (Invitrogen).

Cells and cell lines:

F98 rat glioma cells, F98 cells overexpressing EGFR (F98 ^EGFR ), and F98 cells overexpressing EGFRvIII (F98 ^EGFRvIII ) were purchased from the American type culture collection (ATCC) and cultured according to the recommended protocol.

F98 (ATCC CRL-2397)

F98 ^EGFR (ATCC CRL-2948)

F98 ^npEGFRvIII (ATCC CRL-2949)

Stably transfected CHO cells expressing recombinant EGFR or EGFRvIII were generated at Affimed according to the following protocol:

The gene of coding wild-type EGFR or deletion mutant EGFRvIII is synthesized by Life Technologies/GeneArt (Regensburg, Germany), and is subcloned among the mammalian expression vector pcDNA5/FRT.Expressing the stable Chinese hamster ovary celI of recombinant wild-type EGFR or EGFRvIII generates based on the host cell line Flp-In CHO (Life Technologies) of previous adaptation to suspension growth.The day before transfection, Flp-In Chinese hamster ovary celI is carried out passage culture (not having bleomycin (Zeocin)) in the HyClone CDM4CHO that is supplemented with L-glutamine, HT supplement and penicillin/streptomycin.The stable expression cell line is by using polyethyleneimine (PEI) transfection reagent, with the product expression based on pcDNA5/FRT and integration plasmid and Flp recombinase expression plasmid pOG44 (Life Technologies) cotransfection Flp-In Chinese hamster ovary celI system generates. 2.5 μg of total DNA was diluted in 50 μL of OptiMEMI medium and combined with 7.5 μg of PEI diluted in 50 μL of OptiMEMI medium. The mixture was incubated for 10 minutes and then added to 2×10 ⁶ Flp-In CHO cells suspended in 1 mL of CHO-S-SFMII medium. One day after transfection, the cells were diluted to a density of 1×10 ⁵ viable cells/mL in CHO-S-SFMII medium supplemented with 500 μg/mL hygromycin B and seeded in T75 culture flasks. Flp recombinase mediates the insertion of the Flp-In expression construct into the genome at the integrated FRT site through site-specific DNA recombination. During the selection phase, the viable cell density was measured once or twice a week, the cells were centrifuged and resuspended in fresh selection medium at a maximum density of 1×10 ⁵ viable cells/mL. After about 2-3 weeks of selection using 500 μg/mL hygromycin B, stable cells expressing EGFR or EGFRvIII were recovered and then transferred to HyClone CDM4 CHO suspension culture medium. Stable cells expressing wild-type EGFR or EGFRvIII were cryopreserved in a culture medium containing 50% ProFreeze (Lonza) and 7.5% DMSO.

To determine the density of wild-type EGFR or EGFRvIII cell surface antigens on stably transfected cell lines, a flow cytometric indirect immunofluorescence assay (QIFIKIT, Dako) was used according to the manufacturer's instructions and demonstrated that the density of wild-type EGFR on CHO ^EGFR stably transfected cells was approximately equal to the density of EGFRvIII on CHO ^EGFRvIII stably transfected cells, whereas no wild-type EGFR or EGFRvIII binding sites were detected on untransfected CHO cells.

Determination of Antibody Binding and Affinity by Flow Cytometry:

Cells and indicated scFv antibodies are started from 100 μ g/mL (about 3300nM) in 100 μ L serial dilutions in FACS buffer and are hatched on ice for 45min (minute). After washing 3 times with FACS buffer, cells are hatched on ice for 45min with 0.1mL 10 μ g/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After second washing cycle, under the same conditions as before, cells are hatched with the goat anti-mouse IgG antibody that 0.1mL 15 μ g/mL FITC is put together. As a control, in the absence of scFv, cells are hatched with anti-His IgG 13/45/31, then with the goat anti-mouse IgG antibody that FITC is put together. Then cells are washed again and are resuspended in the 0.2mL FACS buffer containing 2 μ g/mL propidium iodide to exclude dead cells. The fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer with MXP software (Beckman-Coulter, Krefeld, Germany) or a Millipore Guava EasyCyte flow cytometer with Incyte software (Merck Millipore, Schwalbach, Germany). The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany).

If analyzing with a Beckman-Coulter FC500 MPL flow cytometer, use 0.5 × 10 ⁶ cells/stain; if analyzing with a Millipore Guava EasyCyte flow cytometer, use only 0.25 × 10 ⁶ cells/stain.

KD values were calculated using the equation for single-site binding (hyperbola) using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) after subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone.

The results are shown in Figure 3.

Example 4: SDS PAGE and Western blotting analysis of the binding specificity of antibodies binding to EGFRvIII.

To detect the binding specificity of EGFRvIII/CD3 tandem diabodies containing different binding domains to EGFRvIII antigen and to demonstrate the absence of binding to wild-type EGFR antigen, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis were performed.

By recombinant DNA technology, the sequence of the extracellular domain of encoding wild-type EGFR or truncated EGFRvIII is fused to the Fc part of human IgG antibody.DNA construct is transfected into the Chinese hamster ovary celI as the wild-type EGFR-Fc or EGFRvIII-Fc of the reorganization of soluble protein and is secreted into the cell culture supernatant, uses protein A chromatography to purify the wild-type EGFR-Fc or EGFRvIII-Fc of described reorganization from described cell culture supernatant.Through the formation of intramolecular disulfide bond in the Fc part, wild-type EGFR-Fc (or EGFRvIII-Fc) fused antigen forms the dimer with two identical chains.Due to EGFRvIII-specific 267 amino acid deletions, EGFRvIII-Fc causes the Fc fusion protein of dimer form to have the size difference of about 50kD than EGFR-Fc little by about 25kD.

Equal amounts of purified wild-type EGFR-Fc (and EGFRvIII-Fc) were mixed with either non-reducing 2× SDS PAGE sample buffer or reducing 2× SDS PAGE sample buffer containing dithiothreitol (DTT) as a reducing agent. Samples containing DTT were heated at 95°C for 5 minutes and then loaded onto a 4-20% standard TGX precast SDS PAGE gel (Biorad). 1 μg of protein sample was used per lane. To separate the proteins in the gel, SDS-PAGE was run at 300V in 1× Tris/glycine/SDS buffer (Biorad) for approximately 25 minutes. Total protein was visualized in the gel using a standard stain-free molecular imaging system (Biorad). A Page Ruler unstained protein ladder (Thermo Scientific) was used as a molecular weight marker. The gel was then blotted onto a PVDF membrane using BioRad's semi-dry blotting Fastblot system. The membrane was blocked in 3% (w/v) nonfat dry milk in 1× TBS for 30 min at room temperature. The membrane was cut into small pieces, each containing a similar protein sample to be blotted. Different tandem diabodies were diluted to a concentration of 2 μg/ml in 3% (w/v) nonfat dry milk in 1× TBS and incubated with separate membrane pieces on a rocking platform for 1 hour. The anti-EGFR antibody cetuximab was used as a control at a concentration of 10 μg/ml. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS, then incubated with HRP-conjugated secondary detection antibodies PentaHIS-HRP (QIAGEN) (for detection of the tandem diabodies) or Protein L-HRP (Thermo Scientific) (for detection of cetuximab) at a dilution of 1:5000 in 3% nonfat dry milk in TBS on a rocking platform for 1 hour. The membrane was washed three times for 10 minutes each with TBST (TBS + 0.1% (v/v) Tween 20) and once with TBS. HRP-mediated color development on the membrane was initiated by adding 0.06% DAB + 0.02% CoCl ₂ + 0.015% H ₂ O ₂ in TBS. The reaction was terminated by adding water. The membrane was dried and scanned.

The results are shown in Figure 4.

Example 5: The binding of EGFRvIII/CD3 tandem diabodies containing different EGFRvIII binding domains to EGFRvIII antigen and CD3, as well as their binding specificity to EGFRvIII in the absence of binding to wild-type EGFR antigen, were analyzed by enzyme-linked immunosorbent assay (ELISA).

ELISA steps:

96-well ELISA plates (Immuno MaxiSorp; Nunc) were coated overnight at 4°C with EGFRvIII-Fc, wild-type EGFR-Fc, or CD3γε recombinant antigen in 100 mM carbonate-bicarbonate buffer. To achieve approximately equal antigen molar coating densities, EGFRvIII-Fc was coated at a concentration of 4 μg/ml, EGFR-Fc at a concentration of 6 μg/ml, and CD3γε at a concentration of 1.5 μg/ml. After a blocking step using 3% (w/v) skim milk powder (Merck) in PBS, 11 serial dilutions of different EGFRvIH/CD3 tandem diabodies ranging from 200 ng/μl to 6.5× ^10-6 ng/μl in PBS containing 0.3% (w/v) skim milk powder (Merck) were incubated on the plates at 25°C for 1.5 h. After incubation, the plates were washed three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20 (Serva). 50 ng/ml of Protein L-HRP was added and incubated on the plates at 25°C for 1 hour. After washing three times with 300 μl/well of PBS containing 0.1% (v/v) Tween 20, tetramethylbenzidine (TMB) substrate (Seramun) was added for detection. The reaction was terminated after approximately 2 minutes by adding 100 μl/well of 0.5 _{M H₂SO₄} . The absorbance of the wells was measured at 450 nm using a multiwell plate reader (Victor, Perkin Elmer).

Absorbance values were plotted as lines in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada).

The results are shown in FIG5 .

Example 6: Measurement of Binding of Tandem Diabodies to EGFRvIII by SPR:

Preparation of anti-hu Fc IgG CM5-chip

An anti-hu Fc IgG CM5-chip was prepared as described above (Example 2).

Concentration range of tandem diabodies used in SPR measurements:

Before SPR measurement, the concentration of the tandem diabody was checked by A280 determination using Nandrop. The concentration (c) of the stock solution was calculated according to the following equation:

c[mg/ml]=(A280)/(2.25[cm ² /mg]×1[cm])

Different extinction coefficients were calculated using the "Expasy Protparam" prediction tool (http://web.expasy.org/protparam/). To calculate molar concentrations, a molecular weight of 105 kD was assumed for all tandem diabodies. The tandem diabodies were adjusted to the final concentrations mentioned below by serial dilution in 1×HBS-P+ buffer. The following concentrations were prepared: 150 nM, 30 nM, 6 nM, 1.2 nM, 0.24 nM, 0.048 nM, and 0 nM; or 90 nM, 30 nM, 10 nM, 3.33 nM, 1.11 nM, 0.37 nM, 0.123 nM, 0.041 nM, and 0 nM, or the concentrations indicated.

Binding assay conditions for tandem diabody KD measurements

Dilute the EGFRvIII-Fc fusion protein in 1×HBS-P+ to a concentration of 6.25 nM. For EGFRvIII-Fc, inject the antigen solution at a flow rate of 10 μl/min for 40 seconds. Inject the antigen into flow cell 2. Flow cell 1 serves as the reference cell.

1× HBS-P+ was used as running buffer throughout the procedure. Serial dilutions of the diabody were injected into all four flow cells at a flow rate of 30 μL/min for 180 seconds. The signal measured in flow cell 1, where no antigen was captured, was subtracted from the signal in flow cell 2 to correct for background binding in the binding curve. After a 540-second dissociation time, the chip was regenerated as described below.

The anti-hu Fc IgG chip was regenerated by injecting high salt buffer 3.0 M _MgCl2 (3 x 15 seconds, 10 μL/min). Before each measurement, three "priming cycles" were performed without the tandem diabody. The tandem diabody solutions were measured from the lowest to the highest concentration, including one cycle without running buffer (0 μM).

Binding parameters of the tandem diabodies were determined in an MCK (Multi Cycle Kinetic) measurement. To estimate apparent binding affinity, binding curves were evaluated using the "Kinetic" method included in the Biacore X100 evaluation software. Using this method, binding curves were fitted in whole or in part using the "Langmuir 1:1 interaction model" with the software.

Binding parameters of tandem diabodies

To analyze the effect of combining two EGFRvIII binding domains into bivalent or multivalent or multispecific molecules such as EGFRvIII/CD3 tandem diabodies, the apparent affinity of EGFRvIII/CD3 tandem diabodies to EGFRvIII antigen was measured by SPR using BiacoreX100 ( FIG6 ).

For all molecules detected, the apparent affinity of the EGFRvIII specific binding domain in the EGFRvIII/tandem diabody improves, and the best combination tandem diabody has achieved a _K of 11pM. Relative to the corresponding tandem diabody containing the parental EGFRvIII binding domain, the _K of the tandem diabody containing the domain of affinity maturation reduces by up to 25 times. Relative to the _K of the measurement of the monovalent binding scFv, the improvement of _K is also up to 25 times. Interestingly, contrary to the improvement achieved by the affinity maturation driven mainly by the improved k _ON , the further improvement of binding affinity in the bivalent tandem diabody format is realized by slower k _ofOFF to a great extent (Table 4).

Therefore, the tandem diabody format provides a way to further enhance binding affinity while maintaining the specificity of the antibody binding domain.

Table 4: Summary of measured KD, association rates (kON), and dissociation rates ^{( kOFF} ) for binding of different EGFRvIII/ ^CD3 tandem diabodies containing bivalent parental or affinity matured ^EGFRvIII -specific binding domains with the domain order _{VLCD3 - VHEGFRvIII} - _VLEGFRVIII - _VHCD3 to recombinant EGFRvIII-Fc antigen in the SPR _{1:1 binding} model.

Example 7: Immunohistochemistry (IHC) staining of solid tumor tissue sections using anti-EGFRvIII bispecific diabody or cetuximab

In contrast to wild-type EGFR, which is ubiquitously expressed on healthy tissues, expression of the mutant receptor EGFRvIII is highly tumor-specific. However, the high frequency of EGFRvIII expression in glioma cells is consistent with literature descriptions, but the extent of EGFRvIII expression and its homogeneity in other tumors remain controversial.

A small immunohistochemistry (IHC) study was initiated, analyzing the reactivity of tissue sections from three cancer patients each suffering from glioblastoma (GB), head and neck (H&N) cancer, prostate cancer, Her2-negative breast cancer, and non-small cell lung cancer (NSCLC) with the EGFRvIII specific binding domains in the form of a bivalent diabody according to the present invention ( FIG. 7 ).

The present invention relates to the immunohistochemical study of the expression of EGFRvIII and the tumor specificity of EGFRvIII binding antibodies to assess different tumors.To contain the bivalent double antibody form of the EGFRvIII binding domain Li3G30 arranged with V _H ^EGFRvIII - _{V L} ^EGFRvIII , the EGFRvIII combination of EGFRvIII specific variable domains has been detected.Secondary detection is implemented by the His tag on the double antibody protein.In contrast, used natural EGFR to have specific IgG antibody cetuximab (Erbitux (Erbitux)).

All people's tissue sections are purchased from BioChain Institute, Inc., and according to supplier's specification sheets process section.Make tissue section be adapted to room temperature and hatch with DAKO peroxidase blocker, then use 10% goat serum to seal.EGFRvIII specific double antibody was hatched 1 hour with two concentration levels of 0.5 μ g/ml and 0.1 μ g/ml, then hatched and passed through Envision+HRP-DAB system detection mouse antibody (DAKO) with anti-His antibody (Dianova).According to manufacturers' specification sheets, use Klear people HRP-polymer DAB detection system (GBI Labs), implement the tissue section dyeing that carries out with control item Erbitux (Merck KGaA) with the concentration of 10g/ml.

Finally, sections were stained with hematoxylin and mounted with mountant medium (Shandon ConsulMount ^™ , Thermo Scientific).

The results are shown in FIG7 .

Two demonstrate in three glioblastoma samples with the obvious and specific reactivity of EGFRvIII specific double antibody.Surprisingly, all three head and neck cancer samples all demonstrate strong and obvious EGFRvIII positive, and similarly for prostate cancer, mammary cancer and NSCLC, obtained the specific staining (Fig. 7) of tumor tissue with EGFRvIII specific antibody described herein.

Based on these properties, the EGFRvIII binding domains described herein, such as, for example, EGFRvIII/CD3 bispecific tandem diabodies, are excellently suited for developing multispecific, multivalent, immune effector cell engaging, tumor targeting antibody therapies.

Example 8: Evaluation of cytotoxic activity mediated by EGFRvIII/CD3 tandem diabodies in a FACS-based cytotoxicity assay

In the cytotoxicity assay based on cell, analyzed EGFRvIII/CD3 series connection diabody antibody, it contains the parental generation that is combined with different CD3 binding domains or the EGFRvIII specificity domain of affinity maturation and/or each binding domain (Fig. 8) with different orders in described series connection diabody molecule.By using the Chinese hamster ovary celI (CHO ^EGFRvIII ) of expressing EGFRvIII, the Chinese hamster ovary celI (CHO ^EGFR ) of expressing wild-type EGFR and not transfected Chinese hamster ovary celI as target cell to analyze the specificity of target-mediated, series connection diabody and the lethality of T cell mediation.

Materials (media, buffers, and reagents):

DMSO (Sigma), EasySep ^™ Human T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD4+ T Cell Enrichment Kit (Stem Cell Technologies), EasySep ^™ Human CD8+ T Cell Enrichment Kit (Stem Cell Technologies), The Big Easy EasySep ^™ Magnet (Stem Cell Technologies), FCS (Invitrogen), Lymphoprep (Stem Cell), L-glutamine (Invitrogen), mlgGl FITC (ADG), CD16-FITC (MEM-154) (Thermo Fisher Scientific), mlgGl-FITC/mlgGl-PE/mlgGl-ECD (Beckman Coulter), mlgGl-PE (Beckman Coulter), mlgGl-PC5 (Beckman Coulter), mlgGl-PC7 (Beckman Coulter), CD8-FITC/CD4-PE/CD3-ECD (Beckman Coulter), CD16-PC5 (Beckman Coulter), CD19-PC7 (Beckman Coulter), CD16-FITC/CD56-PE/CD3-ECD (Beckman Coulter), CD14-PC7 (Beckman Coulter), CD33-PE (MACS Miltenyi Biotech), NaN ₃ (Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), PKH67 green fluorescent cell labeling (Linker) Midi kit (Sigma), Gammanorm human IgG (Octapharma), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

Isolation of PBMCs from buffy coats and enrichment of T cells:

On the day of blood draw, buy buffy coat from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany); Buffy coat preparation is stored at room temperature overnight, and then PBMC is separated by density gradient centrifugation. Buffy coat sample is diluted with two to three volumes of PBS, layered on Lymphoprep cushion and centrifuged at 800 × g for 25 min at room temperature w/o brake. PBMC at the interface is collected and washed 3 times with PBS, and then they are used for enrichment of T cells or flow cytometric analysis. According to the manufacturer's instructions, EasySep ^™ human T cell enrichment kit (for immunomagnetic separation of untouched human T cells) and Big Easy EasySep ^™ magnet are used to enrich T cells from the PBMC group.

Characterization of effector cells by flow cytometry:

To assess the distribution of PBMC subpopulations and the purity of enriched PBMC or T cell subsets, PBMC staining was performed.

For flow cytometric analysis, isolated PBMCs or enriched PBMC subpopulations were resuspended in PBS supplemented with 2% heat-inactivated FCS and 0.1% sodium azide (referred to as FACS buffer) and 1 mg/mL polyclonal human IgG to a density of 0.5-2 × 10 ⁶ /mL. 0.5 ml aliquots of the cell suspension were then incubated with the antibody for 15 minutes in the dark according to the manufacturer's recommended protocol.

Table 5: Pipetting scheme for characterization of PBMCs and fractionated PBMCs by flow cytometry

After incubation, samples were measured on a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) in the order 3, 4, 5, 6, 1, 2 using a protocol developed for PBMC and multicolor staining (FITC PEECD PC5 PC7; FITC PE ECD; FITC PE PC5PC7) in MXP acquisition software (Beckman-Coulter). For analysis and data plotting, CXP analysis software (Beckman-Coulter) was used.

Cytotoxicity assay:

As previously described, T cells used as effector cells were characterized by flow cytometry. The target cells used in the cytotoxicity assay were CHO cells stably transfected to overexpress EGFRvIII (CHO ^EGFRv111 ), CHO cells stably transfected to express wild-type EGFR (CHO ^EGFR ), or untransfected CHO cells (CHO). Cell lines were generated as previously described and cultured under standard conditions as previously described. For the cytotoxicity assay, target cells were collected, washed twice with RPMI 1640 medium without FCS, and resuspended in diluent C provided by the PKH67 green fluorescent cell labeling Midi kit to a density of 2×10 ⁷ /mL. The cell suspension was then mixed with an equal volume of double-concentrated PKH67-labeling solution (e.g., 1 μL PKH in 250 μL diluent C) according to the manufacturer's instructions and incubated at room temperature for 2-5 minutes with regular mixing. The staining reaction was terminated by adding an equal volume of FCS and incubating for 1 minute. After washing the labeled target cells with complete RPMI medium, the cells were counted and resuspended in complete RPMI medium to a density of 2×10 ⁵ /mL.

2×10 ⁴ target cells were then seeded with T cells at a 5:1 E:T ratio in each well of a round-bottom 96-well microplate along with the indicated antibodies in a total volume of 200 μL/well. Nine serial 1:5 dilutions, starting at 30 μg/mL, were typically tested. Natural cell death in the absence of antibody and target killing by effectors were determined in at least three replicates on each plate. Tandem diabody-mediated killing was typically determined in duplicate.

After centrifugation at 200 g for 2 minutes at room temperature, the assay plate was incubated at 37° C. in a humidified atmosphere containing 5% CO ₂ for 40-48 hours. After incubation, the culture was washed once with FACS buffer and then resuspended in 150 μL FACS buffer supplemented with 2 μg/mL PI. The absolute number of viable target cells, characterized by positive green PKH67 staining but negative PI staining, was measured using a Millipore Guava EasyCyte flow cytometer (Merck Millipore).

Based on the remaining live target cells measured, the percentage of specific cell lysis was calculated according to the following formula: [1-(number of live target cells _(sample) )/(number of live target cells _(native) )]×100%. Sigmoidal dose response curves and EC50 values were calculated by nonlinear regression/4-parameter logistic fit using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, California, USA).

Data Analysis:

Lysis values obtained for a given antibody concentration were determined in duplicate and analyzed by Sigmoidal dose-response/4-parameter logistic fit analysis using Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) and used to calculate the mean EC50 and SD of the replicates for percentage lysis.

The results are shown in FIG8 and Table 6.

The tandem diabody of all detections all demonstrates the concentration dependency to the cell of expressing EGFRvIII, high specificity, cytotoxicity and high efficiency, and the survival of EGFRvIII negative cells is not impaired.Within the whole concentration range of any tandem diabody of expressing EGFRwt-cell or non-transfected cell detection, do not observe detectable cytotoxicity.Therefore, the cytotoxicity of EGFRvIII/CD3 tandem diabody mediation is strictly target-dependent.In cytotoxicity assay, the EC 50 of the tandem diabody that contains parental EGFRvIII binding domain before affinity maturation and domain order is V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 is _25pM (Fig. 8).Corresponding tandem diabody contains the EGFRvIII binding domain that avidity improves with identical domain order (V _L ^CD3 -V _H ^EGFRvIII - _{V L} ^EGFRvIII -V _H ^CD3 ), shows enhanced cytotoxicity efficacy equally, and in the best case, reaches the EC ₅₀ (table 6) of 1.5pM. The relative increase in cytotoxic activity is more evident in the tandem diabodies with EGFRvIII specific domains located in the external position (domain order _VH ^EGFRvIII - _VL ^CD3 - _VH ^CD3 - _VL ^EGFRvIII ) (Table 6). In these tandem diabodies, up to 70-fold increased cytotoxic potency can be achieved by replacing the low-affinity EGFRvIII binding domain Li3G30 with an affinity-enhanced domain in these tandem diabodies (Figure 8, Table 6).

Table 6: to the summary of the EC50 value measured with different EGFRvIII/CD3 series connection diabodies in the cytotoxicity assay, described different EGFRvIII/CD3 series connection diabodies contain the EGFRvIII specific binding domain of affinity maturation with two different domain orders.Shown the raising coefficient relative to the series connection diabodies that contain the parental generation EGFRvIII binding domain of correspondence.

Choi et al. (Proc Natl Acad Sci USA. 2013 Jan 2; 1 10 (1): 270-5; WO2013/7185010) describes a t EGFRvIII/CD3 comparator, an EGFRvIII/CD3 bispecific T-cell engager (BiTE), which has an EGFRvIII binding site (based on the mouse MR1 domain) and a CD3 binding site (based on the anti-CD3 antibody OKT3). In a cell-based cytotoxicity assay, the antibody achieved 50% of the maximum specific lysis achieved at a concentration of about ^10-2 μg/ml, which corresponds to a molar EC50 concentration of about 200 pM for the 52kD BiTE (Figure 4C of Choi et al., 2013). Comparisons show that the cytotoxicity of the tandem diabodies described herein containing the parental, rather than affinity-matured, EGFRvIII binding domains is even about 10-fold higher than that of the EGFRvIII/CD3 BiTE. Furthermore, the tandem diabody containing the affinity-matured EGFRvIII binding domain of the present invention even exhibited 100-fold higher cytotoxicity than that of EGFRvIII/CD3 BiTE.

Example 9: EGFRvIII receptor internalization induced by EGFRvIII/CD3 tandem diabody on CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII )

To evaluate EGFRvIII regulation by EGFRvIII/CD3 tandem diabodies, CHO cells expressing recombinant EGFRvIII (CHO ^EGFRvIII ) or F98 cells expressing recombinant EGFRvIII (F98 ^EGFRvIII ) were seeded in aliquots of 2.5×10 ⁵ cells per well of a round-bottom 96-well microplate and incubated with increasing concentrations of EGFRvIII/CD3 tandem diabodies in RPMI 1640 medium supplemented with 10% FCS, 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, and 100 μg/mL sodium pyruvate (referred to herein as complete RPMI medium; all components purchased from Invitrogen) at 37° C., 5% CO ₂ , in a humidified incubator for 24 hours. Aliquots of cells were cultured in the absence of antibody to serve as control samples for detection of the highest EGFRvIII cell surface levels and as negative controls for staining performed with the secondary reagent alone.

44.After hatching, cell is washed 2 times with ice-cold phosphate buffered saline (PBS, Invitrogen) (being called FACS buffer in this article) that is supplemented with 2% heat inactivation FCS (Invitrogen) and 0.1% sodium azide (Roth, Karlsruhe, Germany), then with the antibody (this antibody is identical with the antibody that hatches with cell during adjustment incubation) of saturation concentration (10 μ g/mL) dyeing to assess the maximum antibody combination corresponding to the remaining EGFRvIII antigen on the cell surface.The aliquots of the cell that is cultivated under the situation that there is no antibody are dyed with the same antibody of saturation concentration, to measure the maximum antibody binding capacity corresponding to the maximum measurable EGFRvIII level on the cell surface.After washing 3 times with FACS buffer, with series connection double antibody treatment cell and dye with the anti-His mAb (Dia910, Dianova) of 10 μ g/mL, then with the goat anti-mouse IgG (Dianova) dyeing that puts together with 15 μ g/mL FITC. After repeated washing with FACS buffer, cells were resuspended in FACS buffer containing 2 μg/mL propidium iodide (Sigma) to exclude dead cells. Approximately 5 × 10 ³ propidium iodide-negative cells from each sample were analyzed using MXP software and an FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany). The mean fluorescence intensity (MFI) of the samples was determined using CXP software (Beckman-Coulter) or cells were analyzed using a Millipore Guava EasyCyte flow cytometer using Incyte software (Merck Millipore, Schwalbach, Germany).

After subtracting the signal obtained from cell staining with the secondary reagent alone, the mean fluorescence intensity of the measured samples was plotted as a line in the graph using GraphPad Prism software (GraphPad Software, San Diego, CA, Canada), and the relative amount of EGFRvIII remaining on the cell surface was calculated using the fluorescence signal obtained by staining untreated cell samples at 37°C in the absence of the EGFRvIII/CD3 tandem diabody, which was set to represent 100% of cell surface EGFRvIII.

The results are shown in FIG9 .

Some bibliographic reports have described and compared deletion mutant EGFRvIII and demonstrated impaired internalization (Fenstermaker et al.2000, Han et al.2006, Grandal et al.2007, Gan et al.2009) with natural EGFR.However, about EGFRvIII specific antibody or for several disclosed reports of the antibody (for example cetuximab) also in conjunction with the wild-type EGFR of EGFRvIII, described in conjunction with rapid internalization and the degraded (Reist et al.1995, Foulon et al.2000, Kuan et al.2000, Patel et al.2007, Jutten et al.2009, Dreier et al.2012) of EGFRvIII in the cell that expresses EGFRvIII after antibody separately.

For internalization determination, use the transfection CHO ^EGFRvIII cell of overexpressing EGFRvIII.The cells of these expression EGFRvIII and the EGFRvIII/CD3 series connection double antibody of varying concentrations were hatched 24 hours at 37 ℃, described the condition in the context of other EGFRvIII binding antibodies to induce strong internalization.After this adjustment (during which antibody-receptor complex can internalize or not internalize), give all remaining EGFRvIII receptor molecules dyeing (Fig. 9) on the cell surface with the saturation concentration of 10 μ g/ml with each antibody.The result that Fig. 9 presents shows, after cell is exposed to EGFRvIII-in conjunction with series connection double antibody, under all detection conditions, the EGFRvIII receptor molecules more than 80% and up to 140% existing on untreated cell are still available (140% or more>100% general value refers to and reduces (cellular internalization) with the number of receptor on the cell surface on the contrary, and the cell processed with series connection double antibody demonstrates the increase of receptor on the cell surface.Internalization to 100% comes from the contrast with untreated cell). These results show, EGFRvIII/CD3 series connection diabody of the present invention surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may even suppress internalization and cause EGFRvIII receptor levels on the cell surface to increase.The result that Fig. 9 presents shows, with respect to untreated cell, under all detection conditions, after cell is exposed to EGFRvIII-in conjunction with antibody, at least more than 80% still remain on the cell surface.These results show, EGFRvIII/CD3 antibody of the present invention, especially EGFRvIII/CD3 series connection diabody surprisingly does not demonstrate any internalization trend (Fig. 9).The combination of EGFRvIII specific structural domain of the present invention does not induce internalization, but may suppress internalization or promote EGFRvIII to express and increase, cause its cell surface density to increase.

Example 10: Evaluation of cell proliferation in PBMC culture in the presence of EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies

To evaluate whether EGFRvIII/CD3 (or EGFRvIII/CD16A) tandem diabodies induce T cell activation and subsequent proliferation in the absence of EGFRvIII ⁺ target cells, human PBMCs were cultured in the presence of increasing concentrations of EGFRvIII/CD3 tandem diabodies. After 5 days of incubation, proliferation was assessed in a BrdU incorporation assay.

Materials (culture media, buffers, and reagents)

FCS (Invitrogen), human serum (Sigma), Lymphoprep (Stemcell Technologies), L-glutamine (Invitrogen), OKT3 (Biolegend), PBS (Invitrogen), β-mercaptoethanol (Invitrogen), penicillin/streptomycin (Invitrogen), RPMI-1640 (Invitrogen), sodium pyruvate (Invitrogen), BrdU cell proliferation ELISA (Roche).

cell:

Buffy coats were purchased from Deutsches Rotes Kreuz, DRK-Blutspendedienst Baden-Wurttemberg-Hessen (Mannheim, Germany) on the day of blood draw; buffy coat preparations were stored at room temperature overnight before PBMC isolation.

Isolation of PBMCs from buffy coats and enrichment of T cells:

PBMCs were isolated from the buffy coat by density gradient centrifugation. The buffy coat sample was diluted with two to three volumes of PBS, layered onto a Lymphoprep cushion, and centrifuged at 800 × g for 25 minutes at room temperature with or without braking. PBMCs at the interface were collected and washed three times with PBS before use in proliferation assays.

Proliferation assay

All assay plates were blocked for 2 hours at room temperature with RPMI1640 medium supplemented with 10% heat-inactivated human serum, 4 mM L-glutamine, 100 U/mL penicillin G sodium salt, 100 μg/mL streptomycin sulfate, 1 mM sodium pyruvate, and 0.05 mM β-mercaptoethanol (referred to herein as complete RPMI medium) to prevent nonspecific binding of antibodies to plastic surfaces. After removing the blocking medium, 4 × 10 ⁵ PBMCs were seeded in each well of a flat-bottom 96-well microplate with the indicated concentrations of test and control items in complete RPMI medium in a total volume of 200 μL/well. Each sample was assayed in triplicate. IgG anti-CD3 clone OKT3 was used as a positive control. To assess spontaneous proliferation, cells were cultured in the absence of antibody in six replicates. The plates were then incubated at 37°C and 5% CO ₂ in a humidified atmosphere for 4 days. 18 hours before termination of incubation, the cell cultures were pulsed with 100 μM BrdU. BrdU incorporation was quantified using a BrdU proliferation ELISA kit according to the manufacturer's instructions. Color development was stopped by adding 100 mM _{H₂SO₄ ,} and absorbance was measured at 450 nm using a multilabel plate reader (Victor 3, Perkin Elmer). Absorbance values were analyzed and the mean and SD were plotted using GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA).

The results are shown in FIG10 .

Example 11: Correlation between CD3 Binding Affinity and Cytotoxicity of EGFRvIII/CD3 Tandem Diabody

Constructed and expressed EGFRvIII/CD3 tandem diabodies containing the same EGFRvIII binding domain (Li3G30) and different CD3 binding domains (which have some affinity for CD3 T cell antigens) and measured binding to CD3 ⁺ Jurkat cells and EGFRvIII ⁺ CHO ^EGFRvIII cells; In addition, cytotoxicity assays were performed using CHO ^EGFRvIII cells as target cells and PBMC cells as effector cells. The order of each VH and VL (heavy and light chain) domain of all tested tandem diabodies tested in this example was the same: _VL ^CD3 - _VH ^EGFRvIII - _VL ^EGFRvIII - _VH ^CD3 .

Antibody binding to EGFRvIII ⁺ CHO ^EGFRvIII cells and cytotoxic activity mediated by the EGFRvIII/CD3 tandem diabody were evaluated by flow cytometry as described in previous examples.

After staining CD3 ⁺ Jurkat cells with increasing concentrations of the tandem diabody, the apparent affinity of the EGFRvIII/CD3 tandem diabody for CD3 ⁺ cells was determined by flow cytometry. KD values were calculated by nonlinear regression/hyperbola using the mean fluorescence values for each tandem diabody concentration determined by flow cytometry.

Materials (media, buffers, and reagents):

Anti-His IgG 13/45/31 (Dianova), FCS (Invitrogen), FITC-conjugated goat anti-mouse IgGmin X (Dianova), L-glutamine (Invitrogen), NaN ₃ (Carl Roth), PBS (Invitrogen), penicillin/streptomycin (Invitrogen), propidium iodide (Sigma), RPMI-1640 (Invitrogen)

cell:

Jurkat cells (kindly provided by Dr. G. Moldenhauer (DKFZ, Heidelberg, Germany)) were cultured in RPMI 1640 medium supplemented with 2 mM L-glutamine, 100 IU/mL penicillin G sodium salt, and 100 μg/mL streptomycin sulfate (all components were from Invitrogen, referred to herein as complete RPMI 1640 medium).

Determination of antibody affinity to CD3+ cells by flow cytometry:

Cells were incubated on ice for 45 min with 100 μL serial dilutions of the indicated antibodies starting from 100 μg/mL (about 1000 nM) in ice-cold PBS (referred to as FACS buffer) supplemented with 2% heat-inactivated FCS and 0.1% sodium azide. After washing 3 times with FACS buffer, cells were incubated on ice for 45 min with 0.1 mL 10 μg/mL mouse monoclonal anti-His antibody clone 13/45/31 in the same buffer. After the second wash cycle, under the same conditions as before, cells were incubated with 0.1 mL 15 μg/mL FITC-conjugated goat anti-mouse IgG antibody. As a control, in the absence of a primary antibody, cells were incubated with anti-His IgG 13/45/31 and then with FITC-conjugated goat anti-mouse IgG antibody. The cells were then washed again and resuspended in 0.2 mL FACS buffer containing 2 μg/mL propidium iodide to exclude dead cells. Fluorescence of 1×10 ⁴ living cells was measured using a Beckman-Coulter FC500 MPL flow cytometer (Beckman-Coulter, Krefeld, Germany) with MXP software or a Millipore Guava EasyCyte flow cytometer (Merck Millipore, Schwalbach, Germany) with Incyte software. The mean fluorescence intensity of the cell samples was calculated using CXP software (Beckman-Coulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany). If the Beckman-Coulter FC500 MPL flow cytometer was used for analysis, 0.5×10 ⁶ cells/staining were used, and if the Millipore Guava EasyCyte flow cytometer was used, only 0.25×10 ⁶ cells/staining were used.

After subtracting the fluorescence intensity values of cells stained with secondary and tertiary reagents alone, the fluorescence intensity values and the equation (hyperbola) for single-site binding of GraphPad Prism software (GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, CA, USA) were used to calculate KD values.

For the series connection diabody of each analysis, the KD value in conjunction with EGFRvIII and cytotoxic EC50 are drawn into the function of the KD value in conjunction with CD3.When the KD value in conjunction with EGFRvIII of series connection diabody only demonstrated very little variability, the EC50 cytotoxicity value demonstrated and was accompanied by the CD3 that series connection diabody increases gradually in conjunction with KD value and almost became linear increase (Figure 11).

Example 12: Inhibition of F98 ^EGFRvIII Xenograft Tumors by EGFRvIII/CD3 (In Vivo Proof of Concept)

On day 0 (d0), nine experimental groups of immunodeficient NOD/scid mice (NOD/MrkBomTac-Prkdcscid, Taconic Denmark) were xenografted with a suspension of 4× ¹⁰ F98 ^npEGFRvIII cells mixed with 1× ¹⁰ purified human PBMCs from healthy donors via subcutaneous (sc) injection. To account for possible donor variability of PBMCs, each experimental group was subdivided into two cohorts, each receiving PBMCs from only one individual donor. 2 h after tumor cell inoculation, followed by 24 h, 48 h, 72 h, and 96 h, 100 μg, 10 μg, or 1 μg of each tandem double antibody test item was administered to the animals via the tail vein (iv). The control group received vehicle only. Starting on d0, another control was administered (ip) with 1 mg of cetuximab (Erbitux) twice a week. Table 6 summarizes the dose groups and donor allocation. Tumor volume was monitored by measuring the major and minor diameters of the tumors with calipers three times a week from day 3 to day 52. Tumor volume was calculated from the diameters according to the following formula: Volume = (minor diameter) ² x major diameter x 0.5.

Table 7: Animal dose groups

References:

Schwarz M, Rottgen P, Takada Y, Le Gall F, Knackmuss S, Bassler N, Buttner C, Little M, Bode C, Peter. Single-chain antibodies for the conformation-specific blockade of activated platelet integrin alphallbbeta3 designed by subtractive selection from naive human phage libraries. FASEB J. 2004 Nov; 18(14): 1704-6.

Schier R, McCall A, Adams GP, Marshall KW, Merritt H, Yim M, Crawford RS, Weiner LM, Marks C, Marks JD. Isolation of picomolar affinity anti-c-erbB-2 single-chain Fv by molecular evolution of the complementarity determining regions in the center of the antibody binding site. J Mol Biol. 1996.

Pini A, Viti F, Santucci A, Carnemolla B, Zardi L, Neri P, Neri D. Design and use of a phage display library. Human antibodies with subnanomolar affinity against a marker of angiogenesis eluted from a two-dimensional gel. J Biol Chem. 1998.

Boder ET, Midelfort S, Wittrup KD. Directed evolution of antibody fragments with monovalent femtomolar antigen-binding affinity. Proc Natl Acad Sci U S A. 2000 Sep 26; 97(20): 10701-5.

Barbas CF 3rd,Hu D,Dunlop N,Sawyer L,Cababa D,Hendry RM,Nara PL,Burton DR.In vitro evolution of a neutralizing human antibody to humanimmunodeficiency virus type 1 to enhance affinity and broaden strain cross-reactivity.Proc Natl Acad Sci U S A.1994 Apr 26;91(9):3809-13.

Parsons HL, Earnshaw JC, Wilton J, Johnson S, Schueler PA, Mahoney W, McCafferty J. Directing phage selections towards specific epitopes. Protein Eng. 1996 Nov; 9(l 1): 1043-9.

Winkler, Kramer A, Kiittner G, Seifert M, Scholz C, Wessner H, Schneider-Mergener J, Hohne W. Changing the antigen binding specificity by single point mutations of an anti-p24 (HIV-1) antibody. J Immunol. 2000 Oct 15; 165(8): 4505-14.

Safdari Y, Farajnia S, Asgharzadeh M, Omidfar K, Khalili M. humMRl, a highly specific humanized single chain antibody for targeting EGFRvIII. Int Immunopharmacol. 2014 Feb; 18(2): 304-10.

Lorimer IA, Keppler-Hafkemeyer A, Beers RA, Pegram CN, Bigner DD, Pastan I. Recombinant immunotoxins specific for a mutant epidermal growth factor receptor: targeting with a single chain antibody variable domain isolated byphage display. Proc Natl Acad Sci US A. 1996 Dec 10; 93(25): 14815-20.

Beers R, Chowdhury P, Bigner D, Pastan I. Immunotoxins with increased activity against epidermal growth factor receptor vlll-expressing cells produced by antibody phage display. Clin Cancer Res. 2000 Jul; 6(7): 2835-43.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Choi BD, Kuan CT, Cai M, Archer GE, Mitchell DA, Gedeon PC, Sanchez-Perez L, Pastan I, Bigner DD, Sampson JH. Systemic administration of a bispecific antibody targeting EGFRvIII successfully treats intracerebral glioma. Proc Natl Acad Sci U SA. 2013 Jan 2;110(l):270-5.

Fenstermaker RA, Ciesielski MJ. Deletion and tandem duplication of exons 2-7 in the epidermal growth factor receptor gene of a human malignant glioma. Oncogene. 2000 Sep 14; 19(39): 4542-8.

Han W, Zhang T, Yu H, Foulke JG, Tang CK. Hypophosphorylation of residue Y1045 leads to defective downregulation of EGFRvIII. Cancer Biol Ther. 2006 Oct; 5(10): 1361-8.

Grandal MV, Zandi R, Pedersen MW, Willumsen BM, van Deurs B, Poulsen HS. EGFRvIII escapes down-regulation due to impaired internalization and sorting to lysosomes. Carcinogenesis. 2007 Jul; 28(7): 1408-17. Gan HK, Kaye AH, Luwor RB. The EGFRvIII variant inglioblastoma multiforme. J Clin Neurosci. 2009 Jun; 16(6): 748-54.

Reist CJ, Archer GE, Kurpad SN, Wikstrand CJ, Vaidyanathan G, Willingham MC, Moscatello DK, Wong AJ, Bigner DD, Zalutsky MR. Tumor-specific anti-epidermal growth factor receptor variant III monoclonal antibodies: use of the tyramine-cellobiose radioiodination method enhances cellular retention and uptake in tumor xenografts. Cancer Res. 1995 Oct 1; 55(19): 4375-82.

Foulon CF, Reist CJ, Bigner DD, Zalutsky MR. Radioiodination via D-aminoacid peptide enhances cellular retention and tumor xenograft targeting of an internalizing anti-epidermal growth factor receptor variant III monoclonal antibody. Cancer Res. 2000 Aug 15; 60(16): 4453-60.

Kuan CT, Wikstrand CJ, Archer G, Beers R, Pastan I, Zalutsky MR, Bigner DD. Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv. Int J Cancer. 2000 Dec 15; 88(6): 962-9.

Patel D, Lahiji A, Patel S, Franklin M, Jimenez X, Hicklin DJ, Kang X. Monoclonal antibody cetuximab binds to and down-regulates constitutively activated epidermal growth factor receptor vIII on the cell surface. Anticancer Res. 2007 Sep-Oct; 27(5A): 3355-66. Jutten B, Dubois L, Li Y, Aerts H, Wouters BG, Lambin P, Theys J, Lammering G. Binding of cetuximab to the EGFRvIII deletion mutant and its biological consequences in malignant glioma cells. Radiother Oncol. 2009 Sep; 92(3): 393-8.

Dreier A, Barth S, Goswami A, Weis J. Cetuximab induces mitochondrial translocalization of EGFRvIII, but not EGFR: involvement of mitochondria in tumor drug resistance? Tumour Biol. 2012 Feb;33(l):85-94.

Claims (20)

1. An EGFRvIII-binding protein having at least one EGFRvIII binding site, comprising an antibody variable heavy chain domain and an antibody variable light chain domain, wherein...

(a)

(a ₁ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 27, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 28, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 30, the antibody variable light chain domain CDR2 is SEQ ID NO: 31, and the antibody variable light chain domain CDR3 is SEQ ID NO: 32;

(a ₂ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 33, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 28, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 34, the antibody variable light chain domain CDR2 is SEQ ID NO: 31, and the antibody variable light chain domain CDR3 is SEQ ID NO: 35;

(a ₃ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 33, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 28, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 36, the antibody variable light chain domain CDR2 is SEQ ID NO: 31, and the antibody variable light chain domain CDR3 is SEQ ID NO: 37;

(a ₄ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 33, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 38, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 39, the antibody variable light chain domain CDR2 is SEQ ID NO: 40, and the antibody variable light chain domain CDR3 is SEQ ID NO: 41;

(a ₅ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 42, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 43, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 44, the antibody variable light chain domain CDR2 is SEQ ID NO: 40, and the antibody variable light chain domain CDR3 is SEQ ID NO: 45;

(a ₆ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 48, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 28, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 47, the antibody variable light chain domain CDR2 is SEQ ID NO: 40, and the antibody variable light chain domain CDR3 is SEQ ID NO: 48;

(a ₇ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 49, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 50, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 51, the antibody variable light chain domain CDR2 is SEQ ID NO: 40, and the antibody variable light chain domain CDR3 is SEQ ID NO: 52;

(a ₈ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 53, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 54, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 44, the antibody variable light chain domain CDR2 is SEQ ID NO: 40, and the antibody variable light chain domain CDR3 is SEQ ID NO: 55;

(a ₉ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 53, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 28, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 56, the antibody variable light chain domain CDR2 is SEQ ID NO: 57, and the antibody variable light chain domain CDR3 is SEQ ID NO: 58;

(a ₁₀ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 46, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 28, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 59, the antibody variable light chain domain CDR2 is SEQ ID NO: 40, and the antibody variable light chain domain CDR3 is SEQ ID NO: 60;

(a ₁₁ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 33, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 61, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 44, the antibody variable light chain domain CDR2 is SEQ ID NO: 40, and the antibody variable light chain domain CDR3 is SEQ ID NO: 62;

(a ₁₂ ) The antibody variable heavy chain domain CDR1 is SEQ ID NO: 46, the antibody variable heavy chain domain CDR2 is SEQ ID NO: 63, and the antibody variable heavy chain domain CDR3 is SEQ ID NO: 29; the antibody variable light chain domain CDR1 is SEQ ID NO: 64, the antibody variable light chain domain CDR2 is SEQ ID NO: 31, and the antibody variable light chain domain CDR3 is SEQ ID NO: 65;

or

(b) The antibody variable heavy chain domain is selected from SEQ ID NO: 1, SEQ ID NO: 3, SEQ ID NO: 5, SEQ ID NO: 7, SEQ ID NO: 9, SEQ ID NO: 11, SEQ ID NO: 13, SEQ ID NO: 15, SEQ ID NO: 17, SEQ ID NO: 19, SEQ ID NO: 21 and SEQ ID NO: 25; the antibody variable light chain domain is selected from SEQ ID NO: 2, SEQ ID NO: 4, SEQ ID NO: 6, SEQ ID NO: 8, SEQ ID NO: 10, SEQ ID NO: 12, SEQ ID NO: 14, SEQ ID NO: 16, SEQ ID NO: 18, SEQ ID NO: 20, SEQ ID NO: 22 and SEQ ID NO: 26.

2. The EGFRvIII binding protein according to claim 1, comprising a variable heavy chain domain selected from SEQ ID NO: 5, SEQ ID NO: 7 and SEQ ID NO: 25 or a variable light chain domain selected from SEQ ID NO: 6, SEQ ID NO: 8 and SEQ ID NO: 26.

3. The EGFRvIII binding protein according to claim 1 or 2, wherein, after cells are exposed to the EGFRvIII binding protein at 37°C for 24 hours in a saturated state, at least 80% of the EGFRvIII receptor molecules present on untreated cells remain usable.

4. The EGFRvIII binding protein according to claim 1 or 2, wherein the EGFRvIII binding protein comprises at least one other functional domain.

5. The EGFRvIII binding protein according to claim 4, wherein the other functional domains are effector domains.

6. The EGFRvIII binding protein according to claim 5, wherein the effector domain is T cell or NK cell specific.

7. The EGFRvIII binding protein according to claim 6, wherein the effector domain is a CD3 binding site comprising at least one antibody variable heavy chain domain and at least one variable light chain domain capable of forming a CD3 antigen binding site.

8. The EGFRvIII binding protein according to claim 7, wherein the CD3 binding site comprises a variable heavy chain domain with the amino acid sequence SEQ ID NO: 23 and a variable light chain domain with the amino acid sequence SEQ ID NO: 24.

9. The EGFRvIII binding protein according to claim 1, wherein the EGFRvIII binding protein is humanized or fully human.

10. The EGFRvIII binding protein according to claim 4, wherein the EGFRvIII binding protein is multispecific.

11. The EGFRvIII binding protein of claim 10, wherein the EGFRvIII binding protein is trispecific.

12. The EGFRvIII binding protein of claim 10, wherein the EGFRvIII binding protein is bispecific.

13. The EGFRvIII binding protein of claim 12, wherein the bispecific binding protein is a tandem biantibody specific to EGFRvIII and CD3 or a tandem biantibody specific to EGFRvIII and CD16A.

14. The EGFRvIII binding protein of claim 13, wherein in each polypeptide of the tandem biantibody, four variable chain domains are fused to each other via peptide linkers L1, L2, and L3 in the following order:

(i)VL(CD3)-L1-VH(EGFRvIII)-L2-VL(EGFRvIII)-L3-VH(CD3); or

(ii) VH(CD3)-L1-VL(EGFRvIII)-L2-VH(EGFRvIII)-L3-VL(CD3); or

(iii) VL(EGFRvIII)-L1-VH(CD3)-L2-VL(CD3)-L3-VH(EGFRvIII); or

(iv) VH(EGFRvIII)-L1-VL(CD3)-L2-VH(CD3)-L3-VL(EGFRvIII).

15. The EGFRvIII binding protein according to claim 14, wherein the peptide linkers L1, L2 and L3 consist of 12 or fewer amino acid residues.

16. The EGFRvIII binding protein of claim 15, wherein the peptide linker is GGSGGS.

17. A polynucleotide encoding the EGFRvIII binding protein of claim 1 or 14.

18. A vector comprising the polynucleotide of claim 17.

19. A host cell transformed or transfected using the vector of claim 18.

20. A pharmaceutical composition comprising: (i) the EGFRvIII binding protein of claim 13, and (ii) a pharmaceutically acceptable carrier.