EA042472B1 - TRIPPECIFIC BOUND PROTEINS AND METHODS FOR THEIR APPLICATION - Google Patents

Description

cross reference

This application claims the priority of U.S. Provisional Application No. 62/305,088, filed March 8, 2016; U.S. Provisional Application No. 62/165,833, filed May 22, 2015; and U.S. Provisional Application No. 62/165, 153, filed May 21, 2015, all applications are incorporated herein by reference in their entirety.

Sequence listing

This application contains a listing of sequences, which was presented in electronic form in ASCII format and is fully incorporated into the present description by reference. The mentioned ASCII copy, created on May 17, 2016, is called 47517_701_601_SL.txt and is 128512 bytes in size.

State of the art

Selective destruction of a single cell or cells of a certain type is often desirable in various clinical situations. For example, the main goal of cancer therapy is the specific destruction of tumor cells while leaving healthy cells and tissues intact and undamaged. One such therapy is to induce an immune response against the tumor in order to cause immune effector cells such as natural killer (NK) cells or cytotoxic T lymphocytes (CTL) to attack and destroy tumor cells.

The essence of the invention

The present invention relates to a trispecific antigen binding protein, pharmaceutical compositions thereof, as well as nucleic acids, recombinant expression vectors, and host cells for producing such trispecific antigen binding proteins, and methods of use in the treatment of diseases, disorders, or conditions. In one aspect, the present invention provides trispecific antigen-binding proteins, said proteins including:

(a) a first domain (A) that specifically binds to human CD3;

(b) a second domain (B), which is a domain that serves to lengthen the half-life; and (c) a third domain (C) that specifically binds to the target antigen, the domains being linked in the order H ₂ N-(A)-(B)-(C)-COOH, H ₂ N-(A)- (C)-(B)-COOH, H ₂ N-(B)-(A)-(C)-COOH, H2N(B)-(C)-(A)-COOH, H ₂ N-(C) -(B)-(A)-COOH or H ₂ N-(C)-(A)-(B)-COOH using linkers L1 and L2.

In certain aspects, the present invention also relates to trispecific antigen-binding proteins, and said proteins include:

(a) a first domain (A) that specifically binds to human CD3;

(b) a second domain (B), which is a domain that serves to lengthen the half-life; and (c) a third domain (C) that specifically binds to the target antigen, the domains being linked in the order H ₂ N-(a)-(C)-(B)-COOH, H ₂ N-(B)- (A)-(C)-COOH or H ₂ N-(C)-(B)-(A)-COOH using linkers L1 and L2.

In certain aspects, the present invention also relates to trispecific antigen-binding proteins, and said proteins include:

(a) a first domain (A) that specifically binds to human CD3;

(b) a second domain (B), which is a domain that serves to lengthen the half-life; and (c) a third domain (C) that specifically binds to the target antigen, the domains being linked in the order H2N-(A)-(B)-(C)-COOH, H2N-(A)-(C)- (B)-COOH, H2N-(B)-(A)-(C)-COOH, H2N(B)-(C)-(A)-COOH, H2N-(C)-(B)-(A) -COOH or H2N-(C)-(A)-(B)-COOH using linkers L1 and L2, and where the first domain binds to human CD3 with a KD greater than 100 nM.

In certain aspects, the present invention also relates to trispecific antigen-binding proteins, and said proteins include:

(a) a first domain (A) that specifically binds to human CD3;

(b) a second domain (B), which is a domain that serves to lengthen the half-life; and (c) a third domain (C) that specifically binds to the target antigen, the domains being linked in the order H ₂ N-(A)-(B)-(C)-COOH, H ₂ N-(A)- (C)-(B)-COOH, H ₂ N-(B)-(A)-(C)-COOH, H2N(B)-(C)-(A)-COOH, H2N-(C)-( B)-(A)-COOH or H2N-(C)-(A)-(B)-COOH using linkers L1 and L2, and where the protein has a molecular weight of less than 55 kDa. In certain aspects, the present invention also relates to trispecific antigen-binding proteins, and said proteins include:

(a) a first domain (A) that specifically binds to human CD3;

(b) a second domain (B), which is a domain that serves to lengthen the half-life; and (c) a third domain (C) that specifically binds to the target antigen, wherein the data

- 1 042472 domains are linked in the order H ₂ N-(A)-(B)-(C)-COOH, H2N-(A)-(C)-(B)-COOH, H2N-(B)-(A) -(C)-COOH, H ₂ N(B)-(C)-(A)-COOH, H2N-(C)-(B)-(A)-COOH or H2N-(C)-(A)- (B)-COOH via L1 and L2 linkers, and wherein B comprises a single domain antibody that binds to serum albumin.

Various embodiments of trispecific antigen-binding proteins are also provided in the present invention, contemplated for any aspect of the present invention, alone or in combination. In some embodiments, the first domain comprises a variable light chain and a variable heavy chain, each of which is capable of specifically binding to human CD3. In some embodiments, the variable light chain is a λ, (lambda) light chain. In some embodiments, the variable light chain is a k (kappa) light chain. In some embodiments, the first domain contains a single-chain variable fragment (scFv) specific for human CD3. In some embodiments, the implementation of the first domain is specific for CD3ε (epsilon). In some embodiments, the implementation of the first domain is specific for CD3δ (delta). In some embodiments, the first domain is specific for CD3y (gamma). In some embodiments, the first domain contains complementary determining regions (CDRs) selected from the group consisting of muromonab-CD3 (OKT3), telixumab (TRX4), teplizumab (MGA031), vizilizumab (Nuvion), SP34, X35 , VIT3, VMA030 (BW264/56), CLB-T3/3, CRIS7, YTH12.5, F111-409, CLB-T3.4.2, TR-66, WT32, SPv-T3b, 11D8, XIII-141, XIII- 46, XIII-87, 12F6, T3/RW2-8C8, T3/RW2-4B6, OKT3D, M-T301, SMC2, F101.01, UCHT-1 and WT-31. In some embodiments, the first domain is humanized or human. In some embodiments, the first domain has a CD3 binding KD on CD3-expressing cells of 1000 nM or less. In some embodiments, the first domain has a CD3 binding KD on CD3 expressing cells of 100 nM or less. In some embodiments, the first domain has a CD3 binding KD on CD3-expressing cells of 10 nM or less. In some embodiments, the first domain is cross-reactive with cynomolgus CD3. In some embodiments, the implementation of the first domain contains the amino acid sequence provided in the present invention.

In some embodiments, the implementation of the second domain binds human serum albumin. In some embodiments, the second domain comprises an scFv, a variable heavy domain (VH), a variable light domain (VL), a single domain antibody, a peptide, a ligand, or a small molecular weight group. In some embodiments, the implementation of the second domain contains scFv. In some embodiments, the implementation of the second domain contains a VH domain. In some embodiments, the implementation of the second domain contains a VL domain. In some embodiments, the implementation of the second domain contains a single domain antibody. In some embodiments, the implementation of the second domain contains a peptide. In some embodiments, the implementation of the second domain contains a ligand. In some embodiments, the implementation of the second domain contains a small molecular weight group.

In some embodiments, the third domain comprises an scFv, a VH domain, a VL domain, a non-Ig domain, a ligand, a nottin, or a small molecular weight group that specifically binds to the target antigen. In some embodiments, the third domain is specific for a cell surface molecule. In some embodiments, the third domain is specific for a tumor antigen.

In some embodiments, the L1 and L2 linkers are peptide linkers. In some embodiments, the L1 and L2 linkers are independently composed of about 20 or fewer amino acid residues. In some embodiments, L1 and L2 linkers are each independently selected from (GS) _n (SEQ ID NO: 49), (GGS) n (SEQ ID NO: 50), (GGGS) n (SEQ ID NO: 51), ( GGSG)n (SEQ ID NO: 52), (GGSGG)n (SEQ ID NO: 53) or (GGGGS)n (SEQ ID NO: 54), where n is 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10. In some embodiments, the L1 and L2 linkers are each independently (GGGGS)4 (SEQ ID NO: 55) or (GGGGS) ₃ (SEQ ID NO: 56). In some embodiments, the L1 and L2 linkers are chemical linkers.

In some embodiments, the implementation of the first domain is at the N-terminus of the protein. In some embodiments, the implementation of the second domain is at the N-terminus of the protein. In some embodiments, the implementation of the third domain is at the N-terminus of the protein. In some embodiments, the first domain is located at the C-terminus of the protein. In some embodiments, the implementation of the second domain is at the C-terminus of the protein. In some embodiments, the implementation of the third domain is at the C-terminus of the protein.

In some embodiments, the molecular weight of the protein is less than about 80 kD. In some embodiments, the implementation of the molecular weight of the protein is from about 50 to about 75 kDa. In some embodiments, the molecular weight of the protein is less than about 50 kDa. In some embodiments, the molecular weight of the protein is less than about 40 kDa. In some embodiments, the molecular weight of the protein is from about 20 kD to about 40 kD. In some embodiments, the protein has a half-life of at least about 50 hours. In some embodiments, the protein has a half-life of at least

- 2 042472 at least about 100 hours In some embodiments, the implementation of the protein is characterized by increased penetration into tissues compared to IgG against the same target antigen. The present invention also relates, in another aspect, to polynucleotides encoding trispecific antigen binding proteins according to any of the above embodiments. In another aspect of the present invention, vectors are provided that contain the described polynucleotides. In another aspect of the present invention, host cells transformed with the described vectors are provided.

In yet another aspect provided by the present invention, pharmaceutical compositions are provided comprising a trispecific antigen binding protein according to any of the above embodiments, a polynucleotide encoding a trispecific antigen binding protein according to any of the above embodiments, a vector containing the described polynucleotides, or a host cell transformed a vector according to any of the above embodiments, and a pharmaceutically acceptable carrier.

The present invention also provides methods for producing a trispecific antigen-binding protein according to any of the aspects and embodiments of the present invention, said method comprising culturing a host transformed or transfected with a vector containing a nucleic acid sequence encoding any of the trispecific antigen-binding protein in the present invention, under conditions allowing carry out the expression of the protein and the isolation and purification of the resulting protein from the culture.

The present invention also relates to methods for treating or improving a proliferative disease, a neoplastic disease, an inflammatory disease, an immunological disorder, an autoimmune disease, an infectious disease, a viral disease, allergic reactions, parasitic reactions, graft-versus-host diseases, or host-versus-graft diseases, comprising administering a trispecific antigen-binding protein according to any of the above embodiments to a subject in need of such treatment or enhancement. In some embodiments, the subject is a human. In some embodiments, the method further comprises administering an agent in combination with a trispecific antigen-binding protein as described herein.

Inclusion by reference.

All publications, patents, and patent applications mentioned in this specification are incorporated herein by reference to the same extent as if each individual publication, patent, or patent application were specifically and individually designated for inclusion by reference.

Brief description of the figures

New features of the invention are set forth in full detail in the appended claims. A more complete understanding of the features and advantages of the present invention will be obtained with reference to the following detailed description, which sets forth illustrative embodiments that use the principles of the present invention, and the accompanying drawings.

Fig. 1 is a schematic representation of an exemplary trispecific antigen binding protein, wherein the protein has a constant core element containing an anti-CD3ε single chain variable fragment (scFv) and an anti-HSA heavy chain variable region; and a target-binding variable domain, which may be a VH, scFv, non-Ig binding element, or ligand.

Fig. 2 is a schematic representation of additional exemplary trispecific antigen binding proteins designed for optimal tissue penetration. In FIG. 2, left, shows by way of illustration a trispecific antigen binding protein containing single domain antibody fragments in all domains. In FIG. 2, middle, shows by way of illustration a trispecific antigen-binding protein including nottin that binds to a target antigen. In FIG. 2, right, shows by way of illustration a trispecific antigen-binding protein containing a natural ligand that binds to a target antigen.

Fig. 3 is a schematic representation of the attachment of a small molecular weight binding element to a trispecific antigen binding protein. The trispecific antigen-binding protein contains a sortase recognition sequence as the domain that binds the target antigen. After the protein is incubated with sortase and the low molecular weight binding element attached to the glycine residue, sortase ligates or conjugates the low molecular weight binding element to the recognition site. The figure shows the LPETGG sequence as SEQ ID NO: 60 and the LPETG sequence as SEQ ID NO: 57.

In FIG. 4 shows a schematic representation of six different ways in which the three domains of these trispecific antigen binding molecules can be arranged.

In FIG. 5 compares the ability of BiTE molecules (EGFR-targeted BiTE from Lutterbuese et al., 2007. PNAS 107: 12605-12610 and PSMA-targeted BiTE pasotuxizumab) with the ability of anti-EGFR and PSMA-targeted, VH domain-containing trispecific molecules to induce primary human T- cells to kill tumor cells.

- 3 042472

In FIG. 6 shows that all six possible configurations of the trispecific molecule containing the EGFR-targeted VH domain can induce T cells to kill the NCI-1563 human tumor cell line. The experiment was carried out in the absence (on the left of the figure) and the presence (on the right of the figure) of human serum albumin with EGFR-targeted BiTE as a positive control.

In FIG. 7 evaluates the ability of five possible configurations of a trispecific molecule containing a PSMA-targeted VH domain to induce T cells to kill the 22Rv1 human tumor cell line. The experiment was carried out in the absence (on the left of the figure) and the presence (on the right of the figure) of human serum albumin with PSMA-targeted BiTE as a positive control. The activity of a PSA-targeted trispecific molecule with a PSMA-targeted scFv was also shown.

In FIG. 8 shows that trispecific molecules can consist of a constant core element containing an anti-CD3ε single chain variable fragment (scFv) and an anti-HSA heavy chain variable region; and a target-binding variable domain, which may be scFv.

In FIG. 9 shows that trispecific molecules that use a finomer, as opposed to an antibody derived domain for tumor targeting, can induce T cells to kill tumor cells.

Fig. 10 shows that when anti-EGFR trispecific molecules redirect T cells to kill human CaPan2 tumor cells (panel A), T cells are activated and produce the cytokines TNF-α (panel B) and IFNy (panel C), depending on doses of the trispecific molecule.

In FIG. 11 shows that when anti-PSMA trispecific molecules redirect T cells to kill human 22Rv1 tumor cells (panel A), T cells are activated and produce the cytokines TNF-α (panel B) and IFNy (panel C), depending on doses of the trispecific molecule.

In FIG. 12 shows that anti-MSLN trispecific molecules can migrate through Matrigel faster than conventional antibodies.

In FIG. 13 shows phage titration for biotin-CD3ε and biotin-HSA.

Detailed description of the invention

The present invention relates to trispecific antigen-binding proteins, pharmaceutical compositions thereof, as well as nucleic acids, recombinant expression vectors and host cells for producing such trispecific antigen-binding proteins. The present invention also relates to methods of using the disclosed trispecific antigen-binding proteins in the prevention and/or treatment of diseases, conditions and disorders. Trispecific antigen-binding proteins are able to specifically bind to a target antigen as well as CD3 and a half-life extension domain, such as the human serum albumin (HSA) binding domain. In FIG. 1 shows one non-limiting example of a trispecific antigen binding protein.

In one aspect of the present invention, the trispecific antigen binding proteins comprise a domain (A) that specifically binds to CD3, a domain (B) that specifically binds to human serum albumin (HSA), and a domain (C) that specifically binds to a target antigen. The three domains in trispecific antigen-binding proteins are in any order. Thus, the order of the domains of the trispecific antigen-binding proteins is assumed to be:

H ₂ N-(A)-(B)-(C)-COOH,

H ₂ N-(A)-(C)-(B)-COOH,

H ₂ N-(B)-(A)-(C)-COOH,

H ₂ N-(B)-(C)-(A)-COOH,

H ₂ N-(C)-(B)-(A)-COOH or

H ₂ N-(C)-(A)-(B)-COOH.

In some embodiments, the trispecific antigen-binding proteins are in the order of H2N-(A)-(B)-(C)-COOH domains. In some embodiments, trispecific antigen-binding proteins have the order of H ₂ domains N-(A)-(C)-(B)-COOH. In some embodiments, trispecific antigen-binding proteins have the order of H2N-(B)-(A)-(C)-COOH domains. In some embodiments, trispecific antigen-binding proteins have the order of H ₂ domains of N-(B)-(C)-(A)-COOH. In some embodiments, the trispecific antigen binding proteins have the order of H2N-(C)-(B)-(A)-COOH domains. In some embodiments

- 4 042472 trispecific antigen-binding proteins have the order of domains H2N-(C)-(A)-(B)-COOH.

Trispecific antigen-binding proteins described in the present invention, optionally contain a polypeptide having the sequence described in table. 6 or 7 (SEQ ID NO: 1-48) and subsequences thereof. In some embodiments, the trispecific antigen-binding protein comprises a polypeptide having at least 70% to 95% or more homology with the sequence described in Table 1. 6 or 7 (SEQ ID NO: 1-48). In some embodiments, the trispecific antigen-binding protein comprises a polypeptide having at least 70%, 75%, 80%, 85%, 90%, 95% or more homology with the sequence described in Table 1. 6 or 7 (SEQ ID NO: 1-48). In some embodiments, the implementation trispecific antigen-binding protein has a sequence containing at least part of the sequence described in table. 6 or 7 (SEQ ID NO: 1-48). In some embodiments, the implementation trispecific antigen-binding protein contains a polypeptide containing one or more of the sequences described in table. 6 or table. 7 (SEQ ID NO: 1-48).

The trispecific antigen-binding proteins of the present invention are designed to provide specific targeting to cells expressing a target antigen by recruiting cytotoxic T cells. This improves efficiency compared to ADCC (antibody dependent cell-mediated cytotoxicity), which uses full-length antibodies directed against a single antigen and is unable to directly recruit cytotoxic T cells. In contrast, by recruiting CD3 molecules expressed specifically on these cells, trispecific antigen-binding proteins can cross-link cytotoxic T cells with cells expressing the target antigen with a high degree of specificity, thereby directing the cytotoxic potential of the T cell towards the target cell. The trispecific antigen binding proteins described in the present invention attract cytotoxic T cells through binding to cell surface expressed CD3 proteins that form part of the TCR. Simultaneous binding of several trispecific antigen-binding proteins to CD3 and a target antigen expressed on the surface of specific cells causes T cell activation and mediates subsequent lysis of the specific cell expressing the target antigen. Thus, trispecific antigen-binding proteins are expected to effect a strong, specific and efficient killing of target cells. In some embodiments, the trispecific antigen-binding proteins described herein stimulate cytotoxic T cells to kill target cells in order to eliminate pathogenic cells (eg, tumor cells, virus- or bacterial-infected cells, autoreactive T cells, etc.). In some of these embodiments, cells are selectively removed, thereby reducing the likelihood of toxic side effects. In other embodiments, the same polypeptides can be used to enhance the elimination of endogenous cells, such as B- or T-lymphocytes in an autoimmune disease or hematopoietic stem cells (HSC) in stem cell transplantation, in order to obtain a therapeutic effect .

The trispecific antigen binding proteins described in the present invention confer additional therapeutic advantages over traditional monoclonal antibodies and other smaller bispecific molecules. In general, the efficacy of recombinant proteins as drugs depends to a large extent on the intrinsic pharmacokinetics of the protein itself. One such advantage in the present invention is that the trispecific antigen-binding proteins described in the present invention have an extended pharmacokinetic half-life due to the presence of a domain that provides an extended half-life, such as a domain specific for HSA. In this regard, the trispecific antigen-binding proteins described herein have an extended serum half-life of about two, three, about five, about seven, about 10, about 12, or about 14 days in some embodiments. This is in contrast to other binding proteins, such as BiTE or DART molecules, which have relatively shorter half-lives. For example, in the case of the CD19xCD3 bispecific scFv-scFv BiTE fusion molecule, continuous delivery of the drug by intravenous infusion (IV) is required due to its short half-life. Longer intrinsic half-lives of trispecific antigen-binding proteins solve this problem, thereby providing increased therapeutic potential, such as low dose pharmaceuticals, reduced intermittent administration, and/or new pharmaceutical compositions. The trispecific antigen-binding proteins described in the present invention are also optimally sized to improve tissue penetration and tissue distribution. Larger sizes limit or prevent penetration or distribution of the protein into target tissues. The trispecific antigen-binding proteins of the present invention avoid this by being small in size, which allows improved tissue penetration and distribution. Accordingly, the trispecific antigen-binding proteins described herein, in some embodiments, are

- 5 042472 leniya have a size of from about 50 to about 80 kDa, from about 50 to about 75 kDa, from about 50 to about 70 kDa, or from about 50 to about 65 kDa. Thus, the size of trispecific antigen-binding proteins has an advantage over IgG antibodies, which are about 150 kDa, and BiTE and DART diabody molecules, which are about 55 kDa, but do not have an extended half-life and are therefore rapidly cleared. through the kidneys. In other embodiments, the trispecific antigen-binding proteins described herein are of optimal size to improve tissue penetration and distribution. In these embodiments, the trispecific antigen binding proteins are designed to be as small as possible while maintaining specificity for their targets. Accordingly, in these embodiments, the trispecific antigen binding proteins described herein are about 20 kD to about 40 kD, or about 25 to about 35 kD, to about 40 kD, to about 45 kD, to about 50 kD, to about 55 kD. kDa, up to about 60 kDa, up to about 65 kDa. In some embodiments, the trispecific antigen binding proteins described herein are about 50, 49, 48, 47, 46, 45, 44,43, 42, 41, 40 kDa, about 39 kDa, about 38 kDa, about 37 kDa, about 36 kDa, about 35 kDa, about 34 kDa, about 33 kDa, about 32 kDa, about 31 kDa, about 30 kDa, about 29 kDa, about 28 kDa, about 27 kDa, about 26 kDa, about 25 kDa, about 24 kDa, about 23 kDa, about 22 kDa, about 21 kDa, or about 20 kDa. As an example, a small size approach is to use single domain antibody (sdAb) fragments for each of the domains. For example, a particular trispecific antigen binding protein has an anti-CD3 sdAb, an anti-HSA sdAb, and an sdAb for the target antigen. This reduces the size of an exemplary trispecific antigen binding protein to less than 40 kDa. Thus, in some embodiments, the trispecific antigen-binding protein domains are single-domain antibody (sdAb) fragments. In other embodiments, the trispecific antigen-binding proteins described herein include small molecule entity (SME) binding elements for HSA and/or target antigen. SME binding elements are small molecules averaging 500 to 2000 Da and are attached to trispecific antigen binding proteins by known methods such as sortase-mediated ligation or conjugation. In these cases, one of the domains of the trispecific antigen binding protein is a sortase recognition sequence, eg LPETG (SEQ ID NO: 57). To attach an SME binding element to a trispecific antigen binding protein with a sortase recognition sequence, the protein is incubated with sortase and an SME binding element, whereby the sortase adds the SME binding element to the recognition sequence. Known SME-binding elements include MIP-1072 and MIP-1095, which bind to prostate-specific membrane antigen (PSMA) specific to the prostate. In other embodiments, the domain that binds to the target antigen of the trispecific antigen-binding proteins described herein comprises a nottin peptide for binding to the target antigen. Nottins are disulfide-stabilized peptides with a cystine knot scaffold and have an average size of about 3.5 kDa. It is proposed that nottins bind to certain tumor molecules, such as fibronectin and the VEGF receptor. In other embodiments, the domain that binds to the target antigen of the trispecific antigen-binding proteins described herein comprises a natural receptor ligand, such as B-cell activating factor (BAFF/BLyS)). Another feature of the trispecific antigen-binding proteins described in the present invention is that they are a single polypeptide design with a flexible linkage between domains. This allows for easy production and fabrication of trispecific antigen-binding proteins since they can be encoded by a single cDNA molecule that can be easily incorporated into a vector. In addition, since the trispecific antigen-binding proteins described in the present invention are a monomeric single polypeptide chain, there are no problems with chain pairing, nor the need for dimerization. The trispecific antigen-binding proteins described herein are expected to reduce the tendency to aggregate, unlike other published molecules such as the bispecific protein with Fc-gamma immunoglobulin domains.

In the trispecific antigen-binding proteins described herein, the domains are linked by internal linkers L1 and L2, with L1 linking the first and second domain of the trispecific antigen-binding proteins and L2 linking the second and third domains of the trispecific antigen-binding proteins. Linkers L1 and L2 have an optimized length and/or amino acid composition. In some embodiments, the L1 and L2 linkers are the same length and amino acid composition. In other embodiments, L1 and L2 are different. In some embodiments, the L1 and/or L2 internal linkers are short, i. consist of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 amino acid residues. Thus, in some cases, internal linkers are composed of about 12 or fewer amino acid residues. In the case of 0 amino acid residues, the internal linker is a peptide bond. In some embodiments, internal L1 linkers and/or

- 6 042472

L2 are long, i.e. consist of 15, 20 or 25 amino acid residues. In some embodiments, these internal linkers consist of about 3 to about 15, eg 8, 9, or 10 contiguous amino acid residues. With regard to the amino acid composition of the L1 and L2 internal linkers, the peptides are selected with properties that confer flexibility on trispecific antigen binding proteins, do not interfere with binding domains, and are also resistant to protease cleavage. For example, glycine and serine residues typically confer protease resistance. Examples of internal linkers suitable for binding domains in trispecific antigen binding proteins include, but are not limited to (GS) _n (SEQ ID NO: 49), (GGS) _n (SEQ ID NO: 50), (GGGS) _n (SEQ ID NO: 51), (GGSG)n (SEQ ID NO: 52), (GGSGG)n (SEQ ID NO: 53) or (GGGGS)n (SEQ ID NO: 54), where n is 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10. In one embodiment, the L1 and/or L2 internal linker is (GGGGS) ₄ (SEQ ID NO: 55) or (GGGGS) ₃ (SEQ ID NO: 56) .

CD3 binding domain.

The specificity of the T cell response is mediated by antigen recognition (displayed in the context of the large histocompatibility complex, MHC) by the TCR complex. Within the TCR, CD3 is a protein complex that includes a CD3y (gamma) chain, a CD3δ (delta) chain, and two CD3ε (epsilon) chains that are present on the cell surface. CD3 associates with the α (alpha) and β (beta) chains of the TCR, as well as the CD3Z (zeta) chain, which together form the complete TCR. Clustering of CD3 on T cells, for example by immobilized anti-CD3 antibodies, results in T cell activation similar to but independent of T cell receptor specificity.

In one aspect, the trispecific antigen-binding proteins described herein comprise a domain that specifically binds to CD3. In one aspect, the trispecific antigen-binding proteins described herein comprise a domain that specifically binds to human CD3. In some embodiments, the trispecific antigen-binding proteins described herein include a domain that specifically binds to CD3y. In some embodiments, the trispecific antigen-binding proteins described herein comprise a domain that specifically binds to CD3δ. In some embodiments, the trispecific antigen-binding proteins described herein comprise a domain that specifically binds to CD3ε.

In other embodiments, the trispecific antigen-binding proteins described herein comprise a domain that specifically binds to a TCR. In some cases, the trispecific antigen-binding proteins described in the present invention contain a domain that specifically binds the α-chain of the TCR. In some cases, the trispecific antigen binding proteins described in the present invention contain a domain that specifically binds the β chain of the TCR. In some embodiments, the CD3-binding domain of the trispecific antigen-binding proteins described herein exhibits not only strong CD3-binding affinity for human CD3, but also excellent cross-reactivity with the corresponding cynomolgus monkey CD3 proteins. In some instances, the CD3-binding domain of the trispecific antigen-binding proteins is cross-reactive with cynomolgus CD3. In certain cases, the ratios of KD values for human CD3 and CD3 cynomolgus monkey (human:cynomolgous KD) range from 5 to 0.2.

In some embodiments, the CD3 binding domain of the trispecific antigen binding protein can be any domain that binds to CD3, including, but not limited to, domains from a monoclonal antibody, a polyclonal antibody, a recombinant antibody, a human antibody, a humanized antibody. In some cases it is advantageous that the CD3 binding domain be derived from the same species in which the trispecific antigen binding protein will ultimately be used. For example, for use in humans, it may be advantageous for the CD3 binding domain of the trispecific antigen binding protein to contain human or humanized residues from the antigen binding domain of an antibody or antibody fragment.

Thus, in one aspect of the present invention, the antigen binding domain comprises a humanized or human antibody or antibody fragment, or a mouse antibody or mouse antibody fragment. In one embodiment, a humanized or human anti-CD3 binding domain comprises one or more (e.g., all three) light chain complementarity determining region 1 (LC CDR1), light chain complementarity determining region 2 (LC CDR2), and light chain complementarity determining region 3 ( LC CDR3) of a humanized or human anti-CD3 binding domain as described herein, and/or one or more (e.g., all three) heavy chain complementarity determining region 1 (HC CDR1), heavy chain complementarity determining region 2 (HC CDR2 ) and the heavy chain complementarity determining region 3 (HC CDR3) of the humanized or human anti-CD3 binding domain described in the present invention, e.g. humanized or human anti-CD3

- 7 042472 binding domain containing one or more, eg all three, LC CDRs and one or more, eg all three, HC CDRs.

In some embodiments, the humanized or human anti-CD3 binding domain comprises a humanized or human light chain variable region specific for CD3, wherein the human CD3 specific light chain variable region contains human or non-human light chain CDR sequences in a human light chain framework region. In certain instances, the light chain framework is the λ (lambda) light chain framework. In other cases, the light chain framework is a k (kappa) light chain framework.

In some embodiments, the humanized or human anti-CD3 binding domain comprises a humanized or human heavy chain variable region specific for CD3, wherein the heavy chain variable region specific for CD3 comprises human or non-human heavy chain CDR sequences in a human heavy chain framework region. In certain cases, the complementarity-determining regions of the heavy chain and/or light chain are derived from known anti-CD3 antibodies, such as e.g. SP34, X35, VIT3, VMA030 (BW264/56), CLB-T3/3, CRIS7, YTH12.5, Fl 11-409, CLBТЗ.4.2, TR-66, WT32, SPv-T3b, 11D8, XIII-141, XSh-46, XSh-87, 12F6, T3/RW2-8C8, T3/RW24B6, OKT3D, M-T301, SMC2, F101.01, UCHT-1 and WT-31.

In one embodiment, the anti-CD3 binding domain is a single chain variable fragment (scFv) containing a light chain and a heavy chain with the amino acid sequence provided herein. As used herein, a single chain variable fragment or scFv refers to an antibody fragment containing a light chain variable region and at least one antibody fragment containing a heavy chain variable region, wherein the light and heavy chain variable regions are adjacently linked via a short flexible polypeptide linker, and capable of being expressed as a single polypeptide chain, and wherein the scFv retains the specificity of the intact antibody from which it is derived. In one embodiment, the anti-CD3 binding domain comprises: a light chain variable region comprising an amino acid sequence having at least one, two, or three modifications (e.g., substitutions), but no more than 30, 20, or 10 modifications (e.g., , substitutions) amino acid sequence of the light chain variable region provided in the present invention, or a sequence having 95 to 99% identity with the amino acid sequence provided in the present invention; and/or a heavy chain variable region comprising an amino acid sequence having at least one, two, or three modifications (e.g., substitutions) but no more than 30, 20, or 10 modifications (e.g., substitutions) in the amino acid sequence of the heavy chain variable region provided in the present invention, or a sequence having 95 to 99% identity with the amino acid sequence provided in the present invention. In one embodiment, the humanized or human anti-CD3 binding domain is scFv and the light chain variable region containing the amino acid sequence described herein is fused to the heavy chain variable region containing the amino acid sequence described herein by scFv linker. The light chain variable region and the heavy chain variable region of the scFv may be located, for example, in any of the following directions: light chain variable region-scFv-linker-heavy chain variable region or heavy chain variable region-scFv-linker-light chain variable region.

In some cases, scFvs that bind to CD3 are generated according to known methods. For example, scFv molecules can be made by linking the VH and VL regions together using flexible polypeptide linkers. The scFv molecules contain an scFv linker (eg, a Ser-Gly linker) with an optimized length and/or amino acid composition. Accordingly, in some embodiments, the length of the scFv linker is such that the VH or VL domain can be linked intermolecularly to another variable domain to form a CD3 binding site. In certain embodiments, such scFv linkers are short, i. consist of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 or 12 amino acid residues. Thus, in some cases, scFv linkers are composed of about 12 or fewer amino acid residues. In the case of 0 amino acid residues, the scFv linker is a peptide bond. In some embodiments, these scFv linkers consist of about 3 to about 15, eg, 8, 9, or 10 contiguous amino acid residues. With regard to the amino acid composition of the scFv linkers, peptides have been selected that confer flexibility, do not interfere with variable domains, and also allow interchain folding to combine two variable domains to form a functional CD3 binding site. For example, scFvlinkers containing glycine and serine residues typically confer protease resistance. In some embodiments, the linkers in the scFv contain glycine and serine residues. amino acids

- 8 042472 The sequence of scFv linkers can be optimized, for example, using phage display techniques to improve CD3 binding and produce more scFv. Examples of peptidic scFv linkers suitable for linking a light chain variable domain and a heavy chain variable domain in scFv include, but are not limited to (GS) _n (SEQ ID NO: 49), (GGS) _n (SEQ ID NO: 50) , (GGGS)n (SEQ ID NO: 51), (GGSG)n (SEQ ID NO: 52), (GGSGG)n (SEQ ID NO: 53) or (GGGGS) _n (SEQ ID NO: 54), in where n is 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10. In one embodiment, the linker in the scFv can be (GgGgS) ₄ (SEQ ID NO: 55) or (GGGGS) ₃ (SEQ ID NO: 56). Changing the length of the linker can maintain or increase activity, resulting in higher performance in activity assays.

In some embodiments, the CD3-binding domain of the trispecific antigen-binding protein has an affinity for CD3 on CD3-expressing cells with a KD of 1000 nM or less, 500 nM or less, 200 nM or less, 100 nM or less, 80 nM or less, 50 nM or less, 20 nM or less, 10 nM or less, 5 nM or less, 1 nM or less, or 0.5 nM or less. In some embodiments, the CD3-binding domain of the trispecific antigen-binding protein has an affinity for CD3ε, γ, or δ with a Kd of 1000 nM or less, 500 nM or less, 200 nM or less, 100 nM or less, 80 nM or less, 50 nM or less, 20 nM or less, 10 nM or less, 5 nM or less, 1 nM or less, or 0.5 nM or less. In other embodiments, the CD3 binding domain of the trispecific antigen binding protein has low affinity for CD3, i. about 100 nM or higher.

Affinity for binding to CD3 can be determined, for example, by the ability of the trispecific antigen-binding protein itself or its CD3-binding domain to bind to CD3 coated on the assay plate; displayed on the microbial cell surface; in solution; etc. The binding activity of the trispecific antigen-binding protein itself or its CD3-binding domain of the present invention towards CD3 can be analyzed by immobilizing the ligand (e.g., CD3) or the tri-specific antigen-binding protein itself or its CD3-binding domain on a bead, support, cage, etc. . Agents in an appropriate buffer may be added and binding partners may be incubated for a specified period of time at a given temperature. After washes to remove unbound material, bound protein can be released using, for example, SDS, high pH buffers, and the like, and analyzed, for example, using Surface Plasmon Resonance (SPR)).

A domain that prolongs the half-life.

The present invention relates to domains that prolong the half-life of an antigen-binding domain. Such domains are contemplated to include, but are not limited to, HSA binding domains, Fc domains, small molecules, and other half-life prolonging domains known in the art.

Human serum albumin (HSA) (molecular weight ~67 kDa) is the most abundant protein in plasma, present at about 50 mg/mL (600 µM) and has a half-life of about 20 days in the human body. HSA serves to maintain blood plasma pH, contributes to the maintenance of colloidal arterial pressure, functions as a carrier for many metabolites and fatty acids, and also serves as a major protein for plasma drug transport.

Non-covalent association with albumin lengthens the half-life of short-lived proteins. For example, recombinant fusion of the albumin-binding domain to a Fab fragment resulted in 25-fold and 58-fold in vivo clearance and a 37-fold and 26-fold prolongation of the half-life when intravenously administered to mice and rabbits, respectively, compared with administration of the Fab fragment alone. In another example, when insulin is acylated with fatty acids to improve binding to albumin, a long-lasting effect was observed when administered subcutaneously in rabbits or pigs. Taken together, the data from these studies demonstrate an association between albumin binding and sustained release.

In one aspect of the present invention, the trispecific antigen-binding proteins described herein include a domain that provides an extension of the half-life, for example, a domain that specifically binds to HSA. In some embodiments, the HSA-binding domain of the trispecific antigen-binding protein can be any domain that binds to HSA, including, but not limited to, domains from a monoclonal antibody, a polyclonal antibody, a recombinant antibody, a human antibody, a humanized antibody. In some embodiments, the HSA binding domain is single chain variable fragments (ScFv), a single domain antibody such as a heavy chain variable domain (VH), a light chain variable domain (VL), and a variable domain (VHH) from a camel-derived single domain antibody, a peptide, ligand or low molecular weight group specific for HSA. In certain embodiments, the HSA binding domain is a single domain antibody. In other embodiments, the HSA binding domain is a peptide. IN

- 9 042472 other embodiments, the implementation of the HSA-binding domain is a small molecular weight group. The HSA-binding domain of the trispecific antigen-binding protein is contemplated to be sufficiently small and not greater than 25 kDa, not greater than 20 kDa, not greater than 15 kDa, or not greater than 10 kDa, in some embodiments. In some cases, the HSA-binding domain is 5 kDa or less if it is a peptide or a small molecular weight group.

The domain that provides an extension of the half-life of the trispecific antigen-binding protein provides a change in the pharmacodynamics and pharmacokinetics of the trispecific antigen-binding protein itself. As mentioned above, a domain that provides an extension of the half-life, prolongs the half-life. The half-life-prolonging domain also alters pharmacodynamic properties, including altering the distribution, penetration, and diffusion of the trispecific antigen-binding protein in tissue. In some embodiments, the half-life extension domain provides improved tissue targeting (including tumors), tissue distribution, tissue penetration, intra-tissue diffusion, and increased efficiency, compared to a protein without the half-life extension domain. . In one embodiment, the therapeutic methods effectively and efficiently use a reduced amount of trispecific antigen-binding protein, resulting in a reduction in side effects, such as a reduction in cellular non-tumor cytotoxicity.

Additionally, the binding affinity of the domain that provides the lengthening of the half-life, can be chosen so as to correspond to a certain half-life in a particular trispecific antigen-binding protein. Thus, in some embodiments, implementation of the present invention, the domain that provides an extension of the half-life, has a high binding affinity. In other embodiments, the implementation of the domain, providing an extension of the half-life, has an average binding affinity. In other embodiments, the half-life extension domain has low or marginal binding affinity. Examples of binding affinity include KD concentrations of 10 nM or less (high), ranging from 10 nM to 100 nM (medium), and greater than 100 nM (low). As mentioned above, binding affinities for HSA are determined using known methods, such as surface plasmon resonance (SPR).

The binding domain of the target antigen.

In addition to the described CD3 and half-life extension domain, the trispecific antigen-binding proteins described in the present invention also contain a domain that binds to a target antigen. The target antigen is involved in the disease, disorder or condition and/or associated with the disease, disorder or condition. In particular, the target antigen is associated with a proliferative disease, a neoplastic disease, an inflammatory disease, an immunological disorder, an autoimmune disease, an infectious disease, a viral disease, an allergic reaction, a parasitic reaction, a graft-versus-host disease, or a host-versus-graft disease. In some embodiments, the target antigen is a tumor antigen expressed on a tumor cell. Alternatively, in some embodiments, the target antigen binds to a pathogen, such as a virus or bacterium. In some embodiments, the target antigen is a cell surface molecule such as a protein, lipid, or polysaccharide. In some embodiments, the target antigen is on a tumor cell, a virus-infected cell, a bacterially infected cell, an injured red blood cell, an arterial plaque cell, or a fibrous tissue cell.

The design of the trispecific antigen-binding proteins described in the present invention allows the target antigen binding domain to be versatile in that the target antigen binding domain can be any type of binding domain, including, but not limited to, domains from a monoclonal antibody. , polyclonal antibody, recombinant antibody, human antibody, humanized antibody. In some embodiments, the target antigen binding domain is single chain variable fragments (scFv), a single domain antibody such as a heavy chain variable domain (VH), a light chain variable domain (VL), and a variable domain (VHH) from a camel-derived single domain antibody . In other embodiments, the target antigen binding domain is a non-Ig binding domain, ie. mimetic antibody such as anticalins, affilins, affitel molecules, affimers, affitins, alphabodies, avimers, DARPin ankyrin repeat proteins, finomers, Kunitz domain peptides, and monobodies. In other embodiments, the target antigen binding domain is a ligand or peptide that binds to or associates with the target antigen. In other embodiments, the target antigen binding domain is nottin. In other embodiments, the target antigen binding domain is a small molecular weight group.

Modifications of trispecific proteins.

- 10 042472

Trispecific antigen-binding proteins described in the present invention include derivatives or analogues, in which:

(i) an amino acid is substituted with an amino acid residue that is not the amino acid residue encoded by the genetic code, (ii) the mature polypeptide is fused to another compound such as polyethylene glycol, or (iii) additional amino acids are fused to the protein, such as a leader sequence or a secretory sequence, or protein purification sequence.

Representative modifications include, but are not limited to, acetylation, acylation, ADP-ribosylation, amidation, flavin covalent attachment, heme covalent attachment, nucleotide or nucleotide derivative covalent attachment, lipid or lipid derivative covalent attachment, phosphatidylinositol covalent attachment, crosslinking, cyclization, formation disulfide bonding, demethylation, covalent crosslinking, cystine formation, pyroglutamate formation, formylation, gamma carboxylation, glycosylation, GPI anchor formation, hydroxylation, iodination, methylation, myristoylation, oxidation, proteolytic processing, phosphorylation, prenylation, racemization, selenoylation, sulfation, transfer RNA mediated addition of amino acids to proteins, such as arginylation and ubiquitination.

Modifications are made anywhere in the trispecific antigen-binding proteins described in the present invention, including the peptide backbone, amino acid side chains, and amino or carboxyl ends. Some common peptide modifications that are useful for modifying trispecific antigen-binding proteins include glycosylation, lipid attachment, sulfation, gamma-carboxylation of glutamic acid residues, hydroxylation, blocking of an amino or carboxyl group or both groups in a polypeptide, via covalent modification, and ADP- ribosylation.

Polynucleotides encoding trispecific antigen-binding proteins.

The present invention also relates, in some embodiments, to polynucleotide molecules encoding the trispecific antigen binding protein described in the present invention. In some embodiments of the present invention, the polynucleotide molecules are presented as a DNA construct. In other embodiments, the polynucleotide molecules are provided as a messenger RNA transcript.

Polynucleotide molecules are constructed by known methods, for example, by combining genes encoding three binding domains, either separated by peptide linkers or, in other embodiments, directly linked by a peptide bond, into a single genetic construct operably linked to a suitable promoter and optionally a suitable transcription terminator, and their expression in bacteria or other suitable expression systems such as CHO cells. In embodiments where the target antigen binding domain is a small molecular weight group, the polynucleotides contain genes encoding a CD3 binding domain and a half-life prolonging domain. In embodiments where the half-life prolonging domain is a small molecular weight group, the polynucleotides contain genes encoding domains that bind to CD3 and the target antigen. Depending on the vector system and host used, any number of suitable transcription and translation elements may be used, including constitutive and inducible promoters. The promoter is chosen such that it drives the expression of the polynucleotide in the appropriate host cell.

In some embodiments, the polynucleotide is inserted into a vector, preferably an expression vector, which is a further embodiment. This recombinant vector can be constructed in accordance with known methods. Vectors of particular interest include plasmids, phagemids, phage derivatives, viruses (eg, retroviruses, adenoviruses, adeno-associated viruses, herpes viruses, lentiviruses, and the like), and cosmids. Various vector/host expression systems can be used to serve as a carrier and be used to express a polynucleotide encoding a polypeptide of the described trispecific antigen binding protein. 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, trispecific antigen-binding proteins as described herein, in some embodiments, are obtained by introducing a vector encoding a protein as described above into a host cell and culturing said host cell under conditions under which the protein domains are expressed, can be isolated and, if necessary, further purified.

pharmaceutical compositions.

In some embodiments, the present invention provides pharmaceutical compositions comprising the trispecific antigen binding protein described in the present ISO.

- 11 042472, a vector containing a polynucleotide encoding a polypeptide of trispecific antigen-binding proteins, or a host cell transformed with this vector, and at least one pharmaceutically acceptable carrier. The term pharmaceutically acceptable carrier includes, but is not limited to, any carrier that does not interfere with the effectiveness of the biological activity of the ingredients and that is not toxic to the patient to whom it is administered. Examples of suitable pharmaceutical carriers are well known in the art and include phosphate buffered saline solutions, water, emulsions such as oil/water emulsions, various types of wetting agents, sterile solutions, and the like. Such carriers may be prepared by conventional means and may be administered to the subject at a suitable dose. Preferably the compositions are sterile. These compositions may also contain adjuvants such as preservatives, emulsifiers and dispersants. Prevention of the action of microorganisms can be ensured by the inclusion of various antibacterial and antifungal agents. In some embodiments of the pharmaceutical composition, the trispecific antigen binding protein described in the present invention is encapsulated in nanoparticles. In some embodiments of the present invention, the nanoparticles are fullerenes, liquid crystals, liposomes, quantum dots, superparamagnetic nanoparticles, dendrimers, or nanorods. In other embodiments of the pharmaceutical composition, the trispecific antigen binding protein is attached to liposomes. In some cases, the trispecific antigen binding protein is conjugated to the surface of the liposomes. In some cases, the trispecific antigen binding protein is encapsulated within the envelope of the liposome. In some cases, the liposome is a cationic liposome.

Trispecific antigen binding protein described in the present invention, is considered for use as a drug. The introduction is carried out in various ways, for example, by intravenous, intraperitoneal, subcutaneous, intramuscular, topical or intradermal administration. In some embodiments, the route of administration depends on the type of therapy and the type of compound contained in the pharmaceutical composition. The dosage regimen will be determined by the attending physician and other clinical factors. Dosing for any patient depends on many factors, including patient size, body surface area, age, sex, particular compound to be administered, time and route of administration, type of therapy, general health, and other drugs administered concomitantly. An effective dose refers to amounts of the active ingredient that are sufficient to affect the course and severity of the disease, resulting in the reduction or remission of such pathology, and can be determined using known methods.

Methods of treatment.

In some embodiments, the present invention provides methods and uses for stimulating the immune system of an individual in need thereof, comprising administering a trispecific antigen-binding protein as described herein. In some instances, administration of the trispecific antigen-binding protein of the present invention induces and/or maintains cytotoxicity to a cell expressing the target antigen. In some instances, the cell expressing the target antigen is a cancer or tumor cell, a virus-infected cell, a bacterially infected cell, an autoreactive T or B cell, damaged red blood cells, arterial plaque, or fibrous tissue.

The present invention also relates to methods and uses for the treatment of a disease, disorder or condition associated with a target antigen, comprising administering to an individual in need of it, trispecific antigen-binding protein described in the present invention. Diseases, disorders, or conditions associated with the target antigen include, but are not limited to, viral infection, bacterial infection, autoimmune diseases, transplant rejection, atherosclerosis, or fibrosis. In other embodiments, the disease, disorder, or condition associated with the target antigen is a proliferative disease, a neoplastic disease, an inflammatory disease, an immunological disorder, an autoimmune disease, an infectious disease, a viral disease, an allergic reaction, a parasitic reaction, a graft-versus- host or host-versus-graft disease. In one embodiment, the disease, disorder, or condition associated with the target antigen is cancer. In one case, the cancer is a hematological cancer. In another case, the cancer is a cancer with a solid tumor.

As used herein, in some embodiments, treatment or treated or treated refers to therapeutic treatment, the purpose of which is to slow down (reduce) an undesirable physiological condition, disorder or disease, or to obtain beneficial or desired clinical results. For the purposes described in the present invention, beneficial or desirable clinical results include, but are not limited to, relief of symptoms; reducing the degree of the condition, disorder or disease; stabilization

- 12 042472 regulation (ie lack of deterioration) of the level of the condition, disorder or disease; delaying the onset or slowing the progression of the condition, disorder or disease; a radical improvement in the condition, disorder or disease state; and remission (partial or complete), whether detectable or undetectable, or alleviation or amelioration of a condition, disorder or disease. Treatment includes identifying a clinically significant response without excessive levels of side effects. Treatment also includes prolonging survival compared to expected survival if no treatment was received. In other embodiments, treating or treating or treated refers to preventive measures whose purpose is to delay the onset or reduce the severity of an undesired physiological condition, disorder, or disease, for example, in a person who is predisposed to the disease (for example, a person who carries a genetic marker for a disease such as breast cancer).

In some embodiments of the methods described herein, the trispecific antigen binding protein is administered in combination with an agent for the treatment of a particular disease, disorder, or condition. Means include, but are not limited to, therapies including antibodies, small molecules (e.g., chemotherapy drugs), hormones (steroid, peptide, and the like), radiotherapy (gamma rays, x-rays, and/or targeted delivery of radioisotopes, microwave ovens, UV radiation and the like), gene therapy (eg antisense drugs, antiretroviral therapy and the like), and other immunotherapies. In some embodiments, trispecific antigen-binding proteins are administered in combination with antidiarrheals, antiemetics, analgesics, opioids, and/or non-steroidal anti-inflammatory drugs. In some embodiments, trispecific antigen-binding proteins are administered before, during, or after surgery.

Some definitions.

As used in the present invention, half-life is used in its usual sense as described in Goodman and Gillman's The Pharmaceutical Basis of Therapeutics 21-25 (Alfred Goodman Gilman, Louis S. Goodman, and Alfred Gilman, eds., 6th ed. 1980 ). Briefly, this term is intended to cover a quantitative measure of the drug elimination period. Excretion of most drugs is exponential (i.e. follows first order kinetics) as drug concentrations typically do not reach those required to saturate the elimination process. The rate of an exponential process can be expressed by its rate constant k, expressing the relative change per unit time, or by its half-life t _1/2 , the time required for 50% completion of the process. The data units of the two constants are the time unit ^-1 and the time unit, respectively. The first order rate constant and the reaction half time are related simply (kxt _1/2 =0.693) and can be interchanged as appropriate. Because first-order elimination kinetics dictate that a constant fraction of drug is lost per unit time, the plot of the logarithm of drug concentration versus time is linear at any time after the initial distribution phase (i.e., after drug uptake and distribution are complete) . The half time period for drug withdrawal can be accurately determined from such a schedule.

Examples

Example 1 Construction of an exemplary trispecific antigen-binding protein for CD20.

Generation of scFv CD3 binding domain.

The canonical sequence of the human CD3ε chain has UniProt database access number P07766. The canonical sequence of the human CD3 chain has UniProt accession number P09693. The canonical sequence of the human CD3δ chain has accession number in the UniProt database P043234. Anti-CD3ε, CD3y, or CD3δ antibodies are prepared using known techniques such as affinity maturation. When mouse anti-CD3 antibodies are used as a starting material, humanization of mouse anti-CD3 antibodies is desirable for clinical settings where mouse-specific residues can induce the production of human-anti-mouse antigen (HAMA) in subjects who receive as treatment trispecific antigen-binding protein described in the present invention. Humanization is accomplished by grafting the CDRs from a mouse anti-CD3 antibody onto the appropriate acceptor frameworks derived from human germline immunoglobulin sequences, optionally including other modifications of the CDRs and/or frameworks. As provided in the present invention, the numbering of residues in antibodies and antibody fragments follows the Kabat numbering system (Kabat E. A. et al., 1991; Chothia et al., 1987).

Human or humanized anti-CD3 antibodies are therefore used to generate scFv sequences for the CD3 binding domains of the trispecific antigen binding protein. DNA sequences encoding human or humanized VL and VH domains are obtained and the codons for the constructs are optionally optimized for expression in

- 13 042472 cells from Homo sapiens. The order in which the VL and VH domains appear in scFv varies (i.e., VL-VH or VH-VL orientation), and three copies of G ₄ S (SEQ ID NO: 58) or G ₄ S (SEQ ID NO: 58) (G ₄ S) ₃ subunits (SEQ ID NO: 56) link variable domains to create an scFv domain. Plasmid constructs containing anti-CD3 scFv can have additional Flag, His or other affinity tags and are electroporated into HEK293 or other suitable human or mammalian cell lines and purified. Validation assays include FACS binding analysis, Proteon kinetic analysis, and staining of CD3-expressing cells.

Generation of the scFv CD20 binding domain CD20 is one of the cell surface proteins present on B lymphocytes. The CD20 antigen is found in normal and malignant pre-B and mature B-lymphocytes, including those in more than 90% of B-cell non-Hodgkin's lymphomas (NHL)). The antigen is absent in hematopoietic stem cells, activated B-lymphocytes (plasma cells), and in normal tissue. As such, several antibodies have been described, mostly of murine origin: 1F5, 2B8/C2B8, 2H7 and 1H4.

The CD20 scFv binding domain is generated similarly to the method described above for generating the CD3 scFv binding domain.

Cloning of DNA expression constructs encoding a trispecific antigen-binding protein.

Ahtu-CD3 scFv domains, which are used to construct a trispecific antigen binding protein in combination with an anti-CD20 scFv domain and an HSA binding domain (eg, a peptide or VH domain), are domains organized as shown in FIG. 1. For expression of a trispecific antigen-binding protein in CHO cells, the coding sequences for all protein domains are cloned into an expression vector system in mammalian cells. Briefly, the gene sequences encoding the CD3 binding domain, the HSA binding domain, and the CD20 binding domain, along with the peptide sequences, are separately synthesized and subcloned using L1 and L2 linkers. The resulting constructs are then ligated together in the order CD20 binding domain - L1 - CD3 binding domain - L2 - HSA binding domain, to produce the final construct. All expression constructs are designed to contain coding sequences for an N-terminal signal peptide and a C-terminal hexahistidine (6xHis) tag (SEQ ID NO: 59) to facilitate protein secretion and purification, respectively.

Expression of trispecific antigen-binding proteins in stably transfected CHO cells.

An expression system was used in CHO cells (FLP-In®, Life Technologies), a derivative of Chinese hamster ovary CHO-Ki cells (ATCC, CCL-61) (Kao and Puck, Proc. Natl. Acad Sci USA 1968;60(4): 1275-81). Adherent cells are subcultured according to standard cell culture protocols provided by Life Technologies.

To adapt to growth in suspension, cells are removed from tissue culture flasks and placed in serum-free medium. Cells adapted to growth in suspension are subjected to cryopreservation in a medium with 10% DMSO.

Recombinant CHO cell lines stably expressing secreted trispecific antigen-binding proteins are generated by transfection of suspension-adapted cells. When selecting with the antibiotic hygromycin B, viable cell densities are measured twice a week and the cells are centrifuged and resuspended in fresh selection medium at a maximum density of 0.1 x 106 viable cells/ml. Cell populations stably expressing trispecific antigen-binding proteins are recovered after 2-3 weeks of selection, and the cells are transferred to standard culture medium in shake flasks.

Expression of recombinant secreted proteins is confirmed by gel electrophoresis or protein flow cytometry. Stable cell populations are subjected to cryopreservation in a medium containing DMSO.

The trispecific antigen-binding proteins are produced in 10 day old fed-batch cultures of stably transfected CHO cell lines by secretion into the cell culture supernatant. Cell culture supernatants are harvested after 10 days with culture viability typically >75%. Samples are taken from production cultures every day and assessed for cell density and viability. On the day of cell collection, cell culture supernatants are purified by centrifugation and vacuum filtration prior to further use. Protein expression titers and product integrity in cell culture supernatants are analyzed by SDS-PAGE.

Purification of trispecific antigen-binding proteins.

The trispecific antigen-binding proteins are purified from the CHO cell culture supernatant according to a two-step procedure. The constructs are subjected to affinity chromatography in the first step followed by preparative size exclusion chromatography (SEC) on Superdex 200 in the second step. The samples are buffered and concentrated

- 14 042472 by ultrafiltration to a typical concentration of >1 mg/ml. Purity and homogeneity (typically >90%) of final samples are assessed by SDS-PAGE under reducing and non-reducing conditions, followed by immunoblotting with anti-HSA or anti-idiotypic antibody and analytical SEC, respectively. Purified proteins are stored in aliquots at -80°C until use.

Example 2 Determination of antigen affinity by flow cytometry.

The trispecific antigen-binding proteins of Example 1 were tested for their binding affinity to human CD3+ and CD20+ cells and cynomolgus CD3+ and CD20+ cells.

CD3+ and CD20+ cells are incubated with 100 µl serial dilutions of the trispecific antigen-binding proteins from Example 1. After washing three times in FACS buffer, cells are incubated with 0.1 ml of 10 µg/ml mouse monoclonal anti-idiotypic antibody in the same buffer for 45 min on ice. After the second wash cycle, the cells are incubated with 0.1 ml of 15 μg/ml FITC-conjugated goat anti-mouse IgG antibodies under the same conditions as before. As a control, cells are incubated with anti-His IgG followed by incubation with FITC-conjugated goat anti-mouse IgG antibodies without trispecific antigen-binding proteins. The cells are then washed again and resuspended in 0.2 ml FACS buffer containing 2 μg/ml propidium iodide (PI) to exclude dead cells. Fluorescence of 1x10 ⁴ live cells is measured using a Beckman-Coulter FC500 MPL flow cytometer using MXP software (BeckmanCoulter, Krefeld, Germany) or a Millipore Guava EasyCyte flow cytometer using Incyte software (Merck Millipore, Schwalbach, Germany). Average fluorescence intensities cell samples are calculated using CXP software (BeckmanCoulter, Krefeld, Germany) or Incyte software (Merck Millipore, Schwalbach, Germany). After subtracting the fluorescence intensity values of cells stained with only secondary and tertiary reagents, the values are then used to calculate KD values with the equation for monosite binding (hyperbolas) using GraphPad Prism non-linear curve fitting software (version 6.00 for Windows, GraphPad Software, La Jolla , California USA).

CD3 binding affinity and cross-reactivity was assessed by titration and flow cytometry experiments on Jurkat CD3+ cells and the cynomolgus monkey CD3 ⁺ HSC-F cell line (JCRB, cat:JCRB1164). Binding to CD20 and cross-reactivity is assessed on human tumor cell lines expressing CD20+. The cross-reactivity KD ratio is calculated using KD values determined on CHO cell lines expressing either recombinant human antigens or recombinant cynomolgus monkey antigens.

Example 3 Cytotoxicity Assay.

The trispecific antigen binding protein of Example 1 was evaluated in vitro for its mediation of T-cell dependent cytotoxicity to target cells expressing CD20+.

Fluorescently labeled CD20+ REC-1 cells (mantle cell lymphoma line, ATCC CRL3004) are incubated with isolated random donor PBMCs or with CB15 T cells (standardized T cell line) as effector cells in the presence of the trispecific antigen binding protein from Example 1. After incubation for 4 hours at 37° C. in a humidified incubator, the release of fluorescent dye from target cells into the supernatant is determined in a spectrofluorimeter. Target cells incubated without the trispecific antigen-binding protein of Example 1 and target cells completely lysed with the addition of saponin at the end of incubation serve as negative and positive controls, respectively. Based on the measured remaining living target cells, calculate the percentage of specific cell lysis according to the following formula

[1-(number of live targets (sample)/number of live targets (spontaneous))]x100%.

Sigmoid dose-response curves and EC ₅₀ values are calculated by non-linear regression/fitting of a 4-parameter logistic model using GraphPad software. The lysis values obtained for a given antibody concentration are used to calculate sigmoidal dose-response curves using a 4-parameter logistic model fit analysis using Prism software.

Example 4 Pharmacokinetics of trispecific antigen binding proteins.

The trispecific antigen binding protein of Example 1 was evaluated for half-life in animal studies.

The trispecific antigen-binding protein is administered to cynomolgus monkeys as a 0.5 mg/kg intramuscular bolus injection. Another group of cynomolgus monkeys receive a protein comparable in size to the CD3 and CD20 binding domains, but without HSA binding. The third and fourth groups receive a protein with CD3 and HSA binding domains and a protein with CD20 and HSA binding domains, respectively, both of which are comparable in size to the trispecific antigen binding protein. Each test group consists of 5 monkeys. Serum samples are collected at the mentioned times.

- 15 042472 times, serially diluted, and the concentration of proteins is determined using an ELISA for binding to CD3 and/or CD20.

Pharmacokinetic analysis is performed using plasma concentrations of the test sample. Group average plasma data for each test sample follows a multi-exponential profile plotted against time post-dose. The data are fitted using a standard two-compartment bolus model with first order rate constants for the distribution and elimination phases. The general equation for a better approximation of data for intravenous administration is c(t)=Ae' ^at +Be' ^e t, where c(t) is the plasma concentration at time t,

A and B are the intercepts on the y-axis, and a and β are the apparent first order rate constants for the distribution and clearance phases, respectively, the α-phase is the initial phase of clearance and reflects the distribution of the protein in the entire extracellular fluid of the animal, while as the second or β-phase portion of the decay curve represents true plasma clearance.

Methods for approximating such equations are well known in the art. For example,

A=D/V(a-k21)/(a-p),

B=D/V(e-k21)/(a-e), and a and β (for α>β) are the roots of the quadratic equation r2+(k12+k21 +k10)r+k21k10=0 using the calculated parameters V=volume of distribution , k10=excretion rate, k12=transfer rate from compartment 1 to compartment 2 and k21=transfer rate from compartment 2 to compartment 1, and

D=administered dose.

Data Analysis: Graphs of concentration versus time profiles were plotted using Kaleidagraph (KaleidaGraph™ V. 3.09 Copyright 1986-1997. Synergy Software. Reading, Pa.). Reported values less than reportable (LTR) are not included in the PK analysis and are not plotted. Pharmacokinetic parameters are determined by compartmental analysis using WinNonlin software (WinNonlin® Professional V. 3.1 WinNonlin™ Copyright 1998-1999 Pharsight Corporation. Mountain View, California). Pharmacokinetic parameters are calculated as described in Ritschel WA and Kearns GL, 1999, IN: Handbook Of Basic Pharmacokinetics Including Clinical Applications, ^5th edition, American Pharmaceutical Assoc, Washington, DC

The trispecific antigen binding protein of Example 1 is expected to have improved pharmacokinetic parameters, such as increased half-life, compared to proteins lacking the HSA binding domain.

Example 5 Xenograft tumor model.

The trispecific antigen binding protein of Example 1 was evaluated in a xenograft model.

Female immunodeficient NOD/SCID mice are sublethally irradiated (2 Gy) and subcutaneously inoculated with 4x106 Ramos RA1 cells in the right flank. When the tumors reach a size of 100 to 200 mm ³ , the animals are divided into 3 treatment groups. Groups 2 and 3 (8 animals each) are injected intraperitoneally with 1.5x107 activated human T cells. Three days later, group 3 animals are then treated with a total of 9 intravenous doses of 50 μg of the trispecific antigen-binding protein from Example 1 (qdx9d). Groups 1 and 2 are treated with vehicle only. Body weight and tumor volume are determined within 30 days.

Animals treated with the trispecific antigen-binding protein of Example 1 are expected to have a statistically significant reduction in tumor growth compared to the corresponding vehicle-treated controls.

Example 6 Proof-of-Concept Clinical Study Protocol for Administering the Trispecific Antigen Binding Protein of Example 1 to Patients with B-Cell Lymphoma.

This study is a Phase I/II clinical study to investigate the trispecific antigen-binding protein of Example 1 as a treatment for patients with B-cell lymphoma.

Research results.

Primary: The maximum tolerated dose of the trispecific antigen-binding protein from Example 1.

Secondary: determine if the in vitro response is a trispecific antigen-binding protein

- 16 042472 from example 1 associated with clinical response.

Phase I

The maximum tolerated dose (MTD) will be determined in the phase I section of the study.

1.1. The maximum tolerated dose (MTD) will be determined in the phase I section of the study.

1.2. Eligible patients will be entered into the trispecific antigen-binding protein study of Example 1.

1.3. The goal is to determine the highest dose of the trispecific antigen-binding protein of Example 1 that can be administered safely without serious or unmanageable side effects in participants. The dose given will depend on the number of participants who were previously enrolled in the study and how well the dose was tolerated. Not all participants will receive the same dose.

Phase II

2.1. In the next phase II section, participants will receive an MTD to determine if treatment with the trispecific antigen-binding protein therapy of Example 1 will result in at least a 20% response rate. The primary outcome for phase II is to determine whether therapy with the trispecific antigen-binding protein of Example 1 results in a clinical response in at least 20% of patients (burst response, minor response, partial response, or complete response).

The right to participate in the study.

Histologically confirmed, newly diagnosed aggressive B-cell lymphoma according to the current World Health Organization Classification, from 2001 to 2007.

Any stage of the disease.

Treatment with R-CHOP or R-CHOP as treatment regimens (±-graft).

Age >18 years.

The general condition of the patient according to the Karnofsky performance status > 50% or the general condition of the patient according to the ECOG scale from 0 to 2.

Life expectancy >6 weeks.

Example 7 Methods for Assessing Binding and Cytotoxic Activities of Trispecific Antigen Binding Molecules.

Getting protein.

Trispecific molecule sequences were cloned into the mammalian expression vector PCDNA 3.4 (Invitrogen), preceded by a leader sequence and followed by a 6x Histidine Tag (SEQ ID NO: 59). Expi293F cells (Life Technologies A14527) were maintained in suspension in flasks for optimal growth ( Thomson) at 0.2 to 8x1E6 cells/ml in Expi293 growth medium. Purified plasmid DNA was transfected into Expi293 cells according to the protocols of the Expi293 Expression System Kit (Life Technologies, A14635) and kept for 4 to 6 days post transfection. The conditioned media were partially purified by affinity and demineralization chromatography. The trispecific proteins were subsequently further purified by ion exchange, or alternatively concentrated in Amicon Ultra centrifugal filtration devices (EMD Millipore), loaded onto Superdex 200 (GE Healthcare) size exclusion chromatography medium, and eluted in neutral buffer containing vehicles. Collected fractions and final purity were assessed by SDS-PAGE and analytical SEC.

Affinity measurements.

The affinities of all binding domain molecules were measured by bilayer interferometry using an Octet instrument.

Affinities for PSMA were measured by loading human PSMA-Fc protein (100 nM) onto anti-human IgG-Fc biosensors for 120 s, followed by a 60 s baseline, after which bindings were measured by incubating the probe tip in a dilution series of trispecific molecules for 180 s, followed by dissociation for 50 s. Affinities for EGFR and CD3 were measured by loading human EGFR-Fc protein or human CD3-Flag-Fc protein, respectively (100 nM), onto anti-human IgG-Fc biosensors for 120 s, followed by a 60 s baseline, followed by binding was measured by incubating the probe tip in a dilution series of trispecific molecules for 180 s, followed by dissociation for 300 s. Affinity for human serum albumin (HSA) was measured by loading biotinylated albumin onto streptavidin biosensors, then following the same kinetic parameters as for CD3 affinity measurements. All steps were performed at 30° C. in 0.25% casein in PBS.

Cytotoxicity assays.

The T-cell dependent cellular cytotoxicity (TDCC) assay in human cells is used to measure the ability of T-cell-attracting activators, including trispecific molecules, to direct T-cells to destroy the tumor.

- 17 042472 cells (Nazarian et al., 2015. J Biomol Screen. 20:519-27). In this assay, T cells and cells of a target cancer cell line were mixed together at a ratio of 10:1 in a 384-well plate and various amounts of T-cell recruiting activator were added. After 48 hours, the T cells were washed away, leaving target cells attached to the plate that were not killed by the T cells. CellTiter-Glo® Luminescent Cell Viability Assay (Promega) was used to quantify remaining viable cells.

Tests for cytokines.

AlphaLISA assays (Perkin Elmer) for TNF-alpha and interferon-gamma are used to provide evidence that T cells are activated by trispecific molecules in the presence of target cells. For this assay, primary human T cells and human tumor cells are incubated in the presence of test molecules as described in the cytotoxicity assays. After 48 hours of incubation, 2 microliter aliquots of assay supernatants are analyzed according to the manufacturer's instructions.

diffusion tests.

A layer of Matrigel (Matrigel) (75 μl) was added to 24-well Transwell inserts (0.4 μm), after which PBS (100 and 1025 μl, respectively) was added to the upper and lower chambers and equilibrated overnight at 4°C. 100 pmol IgG or Fab (goat anti-human Fc, Jackson Immuno Research) or trispecific molecules were added to the upper chamber and the diffusion of each molecule into the lower chamber was quantified over time by ELISA analysis for each specific molecule. IgG and Fab were captured with donkey anti-goat IgG (Jackson Immunoresearch) that was immobilized on ELISA plates and detected with horseradish peroxidase-conjugated donkey anti-goat IgG (Jackson Immunoresearch) and TMB development. The trispecific molecules were captured in human serum albumin (Athens Research & Technology) which was immobilized on ELISA plates and detected by horseradish peroxidase conjugated anti-His antibody (GenScript) and TMB development.

Relative diffusion at each time point was calculated as: (concentration in the lower chamber at time=t)/(concentration in the upper chamber at time=t).

Statistically significant differences in diffusion between the IgG molecule and Fab or trispecific molecules were identified using an unpaired t-test.

Example 8 Affinity measurements for trispecific molecules directed against EGFR.

The affinities of the three binding domains in the anti-EGFR molecule were measured by bilayer interferometry using an Octet instrument and the corresponding data are summarized in Table 1. 1.

Trispecific molecules in which the EGFR binding domain is located at the N-terminus of the molecule showed significantly higher affinities for EGFR compared to trispecific molecules that contained the EGFR binding domain in the center or in the C-terminal position. Similarly, trispecific molecules containing an albumin-binding domain at the N-terminus also showed higher affinity for HSA than those containing albumin in the middle or in the C-terminal positions. In contrast, all trispecific molecules showed very similar affinities for human CD3, regardless of the position of the binding domain within the trispecific molecule.

Example 9 Affinity measurements for trispecific molecules directed against PSMA.

The affinities of the three binding domains in the anti-PSMA molecules were measured by bilayer interferometry using the Octet instrument and the corresponding data are summarized in Table 1. 2.

Trispecific molecules containing an albumin-binding domain at the N-terminus had higher affinities for HSA than those containing an albumin-binding domain in the middle or at the C-terminal positions. In contrast, the position of the CD3 binding domain does not affect affinity for its target. Similarly, the position of the PSMA-binding domain had little effect on affinity, with all trispecific molecules having affinities for human PSMA that differed from each other by no more than 3-fold.

Example 10 Cytotoxicity assays with trispecific molecules.

The trispecific molecules were tested in T cell dependent cellular cytotoxicity (TDCC) assays for their ability to induce primary human T cells to kill human tumor cells in a target tumor dependent manner.

Trispecific molecules containing single-domain antibody-derived anti-EGFR or PSMA tumor-targeting domains can induce potent cell killing by a mechanism comparable to that of bispecific T cell engagers (BiTE), see FIG. 5. Six anti-EGFR trispecific molecules with a single domain anti-EGFR antibody (see FIG. 4) and a trispecific molecule containing anti-EGFR scFv were tested in TDCC assays using the NCI-1563 human lung adenocarcinoma cell line. For comparison, EGFR BiTE was included in each assay (Lutterbuese et al., 2007. PNAS 107: 12605-12610.). It has been demonstrated that all 7 configurations are directed

- 18 042472 anti-EGFR trispecific molecules effectively kill target cells (see representative data in Tables 3 and 4 and Figures 6 and 8) with activity similar to that of EGFR BiTE. A TDCC assay was also performed with the addition of 15 mg/ml human serum albumin to evaluate the effect of albumin binding on the TDCC activity of trispecific molecules. As expected, the potential efficacy of EGFR BiTE, which lacks an albumin-binding domain, was similar in the absence or presence of albumin. The potential efficiencies of the trispecific molecules were reduced in the presence of albumin, but the magnitude of the reduction depended on the configuration of the molecule. The configurations whose potency was reduced at least in the presence of albumin were EGFR-scFv:C:A and E:A:C (anti-EGFR-scFv:anti-CD3EscFv:anti-ALB-sdAb and anti-EGFR -sdAb:anti-ALB-sdAb:anti-CD3E-scFv).

To demonstrate that the results obtained for anti-EGFR trispecific molecules can be applied to all trispecific molecules, five anti-PSMA trispecific molecules with single-domain anti-PSMA antibody and a trispecific molecule containing anti-PSMA scFv were tested in a TDCC assay. using the human prostate carcinoma epithelial cell line 22Rv1. For comparison, PSMA BiTE (pasotuxizumab) was included in the analysis. Typical results can be found in Table. 5 and in FIG. 7. Most of the anti-PSMA trispecific molecules had similar activity to PSMA BiTE in the TDCC assay, except for the trispecific molecule with the configuration A:C:P (anti-PSMA-sdAb:anti-CD3E-scFv:anti-ALB-sdAb), data trispecific molecules were also tested in a TDCC assay containing 15 mg/ml human serum albumin to evaluate the effect of albumin binding on the TDCC activity of trispecific molecules. As expected, the potential activity of PSMA BiTE, which lacks an albumin-binding domain, was similar in the absence or presence of albumin. The potential efficiencies of the trispecific molecules were reduced in the presence of albumin, but the magnitude of the reduction depended on the configuration of the molecule. The configurations whose potency was reduced at least in the presence of albumin were P:A:C (anti-PSMA-sdAb:anti-ALB-sdAb:antiCD3E-scFv). Trispecific molecules described in the present invention can use different modalities for targeted delivery to tumor target cells. In FIG. 5, 6 and 7 show trispecific molecules with sdAb-derived domains directed against a tumor, and FIG. 7 and 8 show that trispecific molecules with an scFv-derived tumor-binding domain can work equally well. In FIG. 9 demonstrates that the anti-tumor domain is not limited to antibody-derived constructs such as sdAbs and scFvs, but that non-immunoglobulin domains can also work. In this example, a 7 kDa finomer specific for Her2 was used to redirect quiescent human T cells to kill human ovarian cancer cells.

Example 11 Determination of Cytokine Production with Trispecific Molecules.

In order to show that the trispecific molecules tested in the present invention did indeed activate T cells and redirect those T cells to kill tumor cells, the production of the cytokines TNFa and IFNy was measured in parallel with the cell killing activity of T cells. , since T cells produce these cytokines when they are activated. As shown in FIG. 10 and 11, four tested trispecific molecules directed against EGFR and PSMA stimulated the production of TNFa and interferon-γ; with potential activity similar to their cell killing activity. The data obtained are consistent with the statement that trispecific molecules activate T cells by participating in the recruitment of target cells.

Example 12 Diffusion Tests

The trispecific molecules analyzed in the present invention are smaller in size than conventional IgG molecules and hence are expected to diffuse faster and penetrate tissue better than monoclonal antibodies. In order to evaluate this property, a diffusion/migration through Matrigel assay was developed. For this purpose, Transwell assay plates were coated with Matrigel, a gelatinous protein mixture resembling the complex extracellular environment found in many tissues. Trispecific molecules, full length IgG or Fab fragments were added to the upper chamber. After eight and 12 hours, the lower chamber was assessed for the number of macromolecules capable of migrating through the matrix. As shown in FIG. 12, trispecific molecules migrate at both time points at a rate much faster than full-length IgG molecules.

Example 13 Detection of anti-CD3 scFv variants with different affinities for human CD3ε.

Characterization of the original anti-CD3ε phage.

The original anti-CD3ε showed good binding to biotin-CD3-ε and low binding to biotin-HSA (FIG. 13).

Anti-CD3ε scFv phage libraries.

A library with single substitutions has been provided for the heavy chain CDR1, CDR2

- 19 042472 heavy chain, heavy chain CDR3, light chain CDR1, light chain CDR2 and light chain CDR3. The residues were changed one by one by mutagenesis.

Selection of clones and determination of binding affinity.

Libraries with single substitutions were bound to biotinylated hu-CD3ε, washed, eluted and counted. Biotinylated cynoCD3 (cynomolgus macaque) was used as a first round selection target and washed for 4 h after combinatorial binding of phages from two independent libraries (~2x selection). Biotinylated hu-CD3 (human) was used as a second round selection target and washed for 3 hours after both libraries were bound (<2x selection). The PCR inserts from the second round of selection were subcloned into the pcDNA3.4 His6 expression vector. 180 clones were selected and the DNA was purified, sequenced and transfected into Expi293. A panel of sixteen clones with an affinity range for human CD3ε was selected to more accurately determine K _D (Table 6).

In table. 1 shows the affinities of trispecific molecules containing a single domain antibody directed against EGFR for three target antigens.

Abbreviations in the table: E=anti-EGFR single domain antibody, C=anti-CD3E scFv, A=anti-albumin single domain antibody.

Table 1. Affinity

hu-EGFR hu-CD3 HSA Trispecific configuration K _D (nM) K _D (nM) K _D (nM) E:S:A 0.4 4.7 22.2 E:A:S 0.8 4.7 17.7 S:E:A 44.8 4.0 17.9 S:A:E 54.5 4.2 17.2 A:E:S 48.3 4.5 4.1 A:S:E 49.1 3.7 3.8

In table. 2 shows the affinities of trispecific molecules containing a single domain antibody directed against PSMA for three target antigens. Abbreviations in the table: P=antiPSMA single domain antibody, C=anti-CD3E scFv, A=anti-albumin single domain antibody.

Table 2. Affinity

hu-PSMA hu-CD3 HSA Trispecific configuration K _D (nM) K _D (nM) K _D (nM) R:S:A 16.7 3.6 24.0 R:A:S 31.6 4.1 21.0 S:A:R 51.0 4.2 21.7 A:R:S 25.0 2.1 3.5 A:S:R 39.7 2.7 3.5

In table. 3 shows the potential activities of trispecific molecules containing a single domain antibody directed against EGFR in cell killing assays. EC50 values are presented as molar concentrations. Abbreviations in the table: E=anti-EGFR single domain antibody, C=anti-CD3E scFv, A=anti-albumin single domain antibody.

Table 3

Protein EU50 (M) EC50 with HSA (M) Multiplicity of change E:S:A 1,ZOE-12 4.5E-11 35.4 E:A:S 1.40E-12 1.70E-11 12.3 S:E:A 5.60E-12 1D0E-10 20.4 S:A:E 5.50E-12 2.00E-10 36.2 A:E:S 6.90E-12 5.60E-10 81.5 A:S:E 6D0E-12 2.80E-10 45.5 EGFR BiTE 1.50E-12 2,ZOE-12 1.5

In table. 4 shows the potential activities of trispecific molecules containing an anti-EGFR scFv antibody and a BiTE molecule in cell killing assays. EC50 values are presented as molar concentrations. Abbreviations in the table: E=anti-EGFR single domain antibody, C=anti-CD3E scFv, A=anti-albumin single domain antibody.

- 20 042472

Table 4

Protein EU50 (M) EU50 with HSA (Μ) Multiplicity of change PSMA: scFv:C:A 1.60Ε-12 1,ZOE-11 7.8 PSMA BiTE 1,ZOE-12 1.70Ε-12 1.3

In table. 5 shows the potential activities of trispecific molecules containing a single domain antibody directed against PSMA in cell killing assays. EC50 values are presented as molar concentrations. Abbreviations in the table: P=anti-PSMA single domain antibody, C=anti-CD3E scFv, A=anti-albumin single domain antibody.

Table 5

Protein EC50(Μ) EC50 with HSA (M) Multiplicity of change R:S:A Ts70E-10 2.35E-09 14.2 R:A:S 5.90Ε-11 2.23E-10 3.8 S:A:R 2.50E-10 D23E-08 49.6 A:R:S 9D0E-11 4.02E-09 44 A:S:R He is active He is active PSMA: scFv:C:A 5.80E-10 2.00E-09 3.5 PSMA BiTE l,30E-10 6.56E-11 0.5

In table. 6 shows the binding affinities of scFv CD3e phage libraries.

Table 6

anti-SOS scFv K _D (nM) human CD3e kon(l/Mc) kdiss(1/s) Kd(hM) CD3e macaque kon(l/Mc) kdiss(1/s) Macaque/human ratio wild type 4.4 4.71E+05 2.07 3.9 4.63 1.83+05 0.9 2B2 3.8 6.08E+05 2.32 3.5 5.57E+05 1.93+05 0.9 9F2 4.1 3.61E+05 1.33 3.4 3.38E+05 1.05+05 0.8 5A2 4.3 5.66E+05 2.36 4.2 4.75E+05 1.93+05 1.0 6A2 4.7 5.22E+05 2.48 4.9 4.56E+05 2.22+05 1.0 2D2 6.4 5.27E+05 3.38 6.6 4.71E+05 3.09+05 1.0 3F2 8.0 7.04E+05 5.02 6.6 7D2E+05 4.38+05 0.8 2E4 14.4 4D6E+05 5.99 13.2 4.04E+05 5.32+05 0.9 2H2 16.0 5.87E+05 9.06Е-03 16.0 5.25E+05 8.37+05 1.0 10V2 17.9 4.90E+05 8.74Е-03 16.6 4.93E+05 8.15+05 0.9 1A2 19.9 5.99E+05 1D9E-02 17 5.31E+05 9.03+05 0.9 1С2 36.8 6.63E+05 2.44Е-02 thirty 6.69E+05 1.97+05 0.8 2A4 46.3 3.64E+05 D66E-02 43.4 3.53E+05 1.53+05 0.9 10E4 49.8 5.2E+05 2.60E-02 46.8 5.08E+05 2.38+05 0.9 8A5 109 7.46E+05 8D0E-02 103 7.23E+05 7.44+05 0.9 2G5 117 9.94E+05 1D5E-01 115 9.64E+05 1.11+05 1.0 1G4 132.9 1.67E+05 2.20E-02 133.7 D64E+05 2.19+05 1.0

Table 7. Sequences

SEQ Π) NO: Design Reduction Subsequence 1 Three specific aEGFR:aCD3:a Albumin E:S:A EVQLVESGGGLVQAGGSLRLSCAASGRTF SSYAMGWFRQAPGKEREFWAINWSSGST YYADSVKGRFTISRDNAKNTMYLQMNSL KPEDTAVYYCAAGYQINSGNYNFKDYEY DYWGQGTQVTVS SGGGGSGGGSEVQLVE SGGGLVQPGGSLKLS S AASGFTFNKYAMN WVRQAPGKNYGLEWVARIKYA

- 21 042472

DS VKDRFTISRDD SKNT AYLQMNNLKTED TAVYYCVRHGNFGNSYISYWAYWGQGTL VTVSSGGGGSGGGGSGGGGSQTWTQEPS LTVSPGGTVTLTCGSSTGAVTSGNYPNWV QQKPGQAPRGLIGGTKFLAPGTPARFSGSL LGGKAALTLSGVQPEDEAEYYCVLWYSN RWVFGGGTKLTVLGGGGSGGGSEVQLVE SGGGLVQPGNSLRLSCAASGFTFSSFGMS WVRQAPGKGLEWVSSISGSGSDTLYADSV KGRFTISRDNAKTTLYLQMNSLRPEDTAV YYCTIGGSLSRS S QGTLVTVS SHHHHHH 2 Three specific aEGFR: aAlbumin: aCD3 E:S:A EVQLVESGGGLVQAGGSLRLSCAASGRTF SSYAMGWFRQAPGKEREFWAINWSSGST Y Y A D S VKGRFTISRDN A KNTM Y LQMNS L KPEDTAVYYCAAGYQINSGNYNFKDYEY DYWGQGTQVTVS SGGGGSGGGSEVQLVE SGGGLVQPGNSLRLSCAASGFTFSSFGMS WVRQAPGKGLEWVSSISGSGSDTLYADSV KGRFTISRDNAKTTLYLQMNSLRPEDTAV YYCTIGGSLSRS S QGTLVTVS S GGGGS GGG SEVQLVESGGGLVQPGGSLKLSCAASGFTF NKYAMNWVRQAPGKGLEWVARIRSKYN NYATYYADSVKDRFTISRDDSKNTAYLQM NNLKTEDTAVYYCVRHGNFGNSYISYWA YWGQGTLVTVSSGGGGSGGGGSGGGGSQ TWTQEPSLTVSPGGTVTLTCGSSTGAVTS GNYPNWVQQKPGQAPRGLIGGTKFLAPGT PARFSGSLLGGKAALTLSGVQPEDEAEYY CVLWYSNRWVFGGGTKLTVLHHHHHH 3 Three specific aCD3: aEGFR: aAlbumin S:E:A EVQLVESGGGLVQPGGSLKLSCAASGFTF NKYAMNWVRQAPGKGLEWVARIRSKYN NYATYYADSVKDRFTISRDDSKNTAYLQM NNLKTEDTAVYYCVRHGNFGNSYISYWA YWGQGTLVTVSSGGGGSGGGGSGGGGSQ TWTQEPSLTVSPGGTVTLTCGSSTGAVTS GNYPNWVQQKPGQAPRGLIGGTKFLAPGT PARFSGSLLGGKAALTLSGVQPEDEAEYY CVLWYSNRWVFGGGTKLTVLGGGGSGGG SEVQLVESGGGLVQAGGSLRLSCAASGRT FSSYAMGWFRQAPGKEREFVVAINWSSGS TYYADSVKGRFTISRDNAKNTMYLQMNSL KPEDTAVYYCAAGYQINSGNYNFKDYEY

-22042472

DYWGQGTQVTVS SGGGGSGGGSEVQLVE SGGGLVQPGNSLRLSCAASGFTFSSFGMS WVRQAPGKGLEWVSSISGSGSDTLYADSV KGRFTISRDNAKTTLYLQMNSLRPEDTAV YYCTIGGSLSRS SQGTLVTVS SHHHHHH* 4 Three specific aSNZ: aAlbum in: aEGFR S:A:E EVQLVESGGGLVQPGGSLKLSCAASGFTF NI<YAMNWVRQAPGI<GLEWVARIRSI<YN NYATYYADSVKDRFTISRDDSKNTAYLQM NNLKTEDTAVYYCVRHGNFGNSYISYWA YWGQGTLVTVSSGGGGSGGGGSGGGGSQ TWTQEPSLTVSPGGTVTLTCGSSTGAVTS GNYPNWVQQKPGQAPRGLIGGTKFLAPGT PARFSGSLLGGKAALTLSGVQPEDEAEYY CVLWYSNRWVFGGGTKLTVLGGGGSGGG SEVQLVESGGGLVQPGNSLRLSCAASGFTF SSFGMSWVRQAPGKGLEWVSSISGSGSDT LYADSVKGRFTISRDNAKTTLYLQMNSLR PEDTAVYYCTIGGSLSRSSQGTLVTVSSGG GGSGGGSEVQLVESGGGLVQAGGSLRLSC AASGRTFS SYAMGWFRQ APGKEREFWAI NWSSGSTYYADSVKGRFTISRDNAKNTMY LQMNSLKPEDTAVYYCAAGYQINSGNYNF KDYEYDYWGQGTQVTVS SHHHHHH 5 Three specific a Albumin: aEGFR: aCD3 A:E:S EVQLVESGGGLVQPGNSLRLSCAASGFTFS SFGMSWVRQAPGKGLEWVSSISGSGSDTL YADSVKGRFTISRDNAKTTLYLQMNSLRP EDTAVYYCTIGGSLSRSSQGTLVTVSSGGG GSGGGSEVQLVESGGGLVQAGGSLRLSCA ASGRTFSSYAMGWFRQAPGKEREFWAIN WSSGSTYYADSVKGRFTISRDNAKNTMYL QMNSLKPEDTAVYYCAAGYQINSGNYNF KDYEYDYWGQGTQVTVSSGGGGSGGGSE VQLVESGGGLVQPGGSLKLSCAASGFTFN I<YAMNWVRQAPGI<GLEWVARIRSI<YNN YATYYADSVKDRFTISRDDSKNTAYLQMN NLKTEDTAVYYCVRHGNFGNSYISYWAY WGQGTLVTVSSGGGGSGGGGSGGGGSQT WTQEPSLTVSPGGTVTLTCGS STGAVTSG NYPNWVQQKPGQAPRGLIGGTKFLAPGTP ARFSGSLLGGKAALTLSGVQPEDEAEYYC VLWYSNRWVFGGGTKLTVLHHHHHH* 6 Trispecifi- A:S:E EVQLVESGGGLVQPGNSLRLSCAASGFTFS

- 23 042472

calc a Albumin: aCD3:aEGFR SFGMSWVRQAPGKGLEWVSSISGSGSDTL YADSVKGRFTISRDNAKTTLYLQMNSLRP EDTAVYYCTIGGSLSRSSQGTLVTVSSGGG GSGGGSEVQLVESGGGLVQPGGSLKLSCA ASGFTFNKYAMNWVRQAPGKGLEWVARI RSKYNNYATYYADSVKDRFTISRDDSKNT AYLQMNNLKTEDTAVYYCVRHGNFGNSY ISYWAYWGQGTLVTVSSGGGGSGGGGSG GGGSQTWTQEPSLTVSPGGTVTLTCGSST GAVTSGNYPNWVQQKPGQAPRGLIGGTKF LAPGTPARFSGSLLGGKAALTLSGVQPEDE AEYYCVLWYSNRWVFGGGTKLTVLGGGG SGGGSEVQLVESGGGLVQAGGSLRLSCAA SGRTFSSYAMGWFRQAPGKEREFWAINW SSGSTYYADSVKGRFTISRDNAKNTMYLQ MNSLKPEDTAVYYC AAGYQINS GNYNFK DYEYDYWGQGTQVTVSSHHHHHH* 7 EGFR BiTE DILLTQSPVILSVSPGERVSFSCRASQSIGTN IHWYQQRTNGSPRLLIKYASESISGIPSRFS GSGSGTDFTLSINSVESEDIADYYCQQNNN WPTTFGAGTKLELKGGGGSGGGGSGGGG SQVQLKQSGPGLVQPSQSLSITCTVSGFSLT NYGVHWVRQSPGKGLEWLGVIWSGGNTD YNTPFTSRLSINKDNSKSQVFFKMNSLQSN DTAIYYCARALTYYDYEFAYWGQGTLVT VSASGGGGSEVQLVESGGGLVQPGGSLKL SCAASGFTFNKYAMNWVRQAPGKGLEWV ARIRSKYNNYATYYADSVKDRFTISRDDS KNTAYLQMNNLKTEDTAVYYCVRHGNFG NSYISYWAYWGQGTLVTVSSGGGGSGGG GSGGGGSQTWTQEPSLTVSPGGTVTLTC GSSTGAVTSGNYPNWVQQKPGQAPRGLIG GTKFLAPGTPARFSGSLLGGKAALTLSGVQ PEDEAEYYCVLWYSNRWVFGGGTKLTVL HHHHHH 8 EGFR- scFv:C:A EGFR- scFv:C:A DILLTQSPVILSVSPGERVSFSCRASQSIGTN IHWYQQRTNGSPRLLIKYASESISGIPSRFS GSGSGTDFTLSINSVESEDIADYYCQQNNN WPTTFGAGTKLELKGGGGSGGGGSGGGG SQVQLKQSGPGLVQPSQSLSITCTVSGFSLT NYGVHWVRQSPGKGLEWLGVIWSGSNTD YNTPFTSVRLSINKFFKNSKSQ

-24042472

DTAIYYCARALTYYDYEFAYWGQGTLVT VSASGGGGSEVQLVESGGGLVQPGGSLKL SCAASGFTFNKYAMNWVRQAPGKGLEWV ARIRSKYNNYATYYADSVKDRFTISRDDS KNTAYLQMNNLKTEDTAVYYCVRHGNFG NSYISYWAYWGQGTLVTVSSGGGGSGGG GSGGGGSQTWTQEPSLTVSPGGTVTLTC GSSTGAVTSGNYPNWVQQKPGQAPRGLIG GTKFLAPGTPARFSGSLLGGKAALTLSGVQ PEDEAEYYCVLWYSNRWVFGGGTKLTVL GGGGSGGGSEVQLVESGGGLVQPGNSLRL SCAAS GFTFS SFGMS WVRQ APGKGLEWVS SISGSGSDTLYADSVKGRFTISRDNAKTTL YLQMNSLRPEDTAVYYCTIGGSLSRSSQGT LVTVSSHHHHHH 9 Three specific aPSMA:aCD3: aAlbumin R:S:A EVQLVESGGGLVQPGGSLTLSCAASRFMIS EYSMHWVRQAPGKGLEWVSTINPAGTTD YAESVI<GRFTISRDNAI<NTLYLQMNSLI<P EDTAVYYCDGYGYRGQGTQVTVSSGGGG SGGGSEVQLVESGGGLVQPGGSLKLSCAA SGFTFNKYAMNWVRQAPGKGLEWVARIR SKYNNYATYYADSVKDRFTISRDDSKNTA YLQMNNLKTEDTAVYYCVRHGNFGNSYIS YWAYWGQGTLVTVSSGGGGSGGGGSGG GGSQTWTQEPSLTVSPGGTVTLTCGSSTG AVTSGNYPNWVQQKPGQAPRGLIGGTKFL APGTPARFSGSLLGGKAALTLSGVQPEDEA EYYCVLWYSNRWVFGGGTKLTVLGGGGS GGGSEVQLVESGGGLVQPGNSLRLSCAAS GFTFS SFGMS WVRQ APGKGLEWVS SISGS GSDTLYADSVKGRFTISRDNAKTTLYLQM NSLRPEDTAVYYCTIGGSLSRSSQGTLVTV SSHHHHHH 10 Three specific aPSMA: aAlbumin :aC D3 R:A:S EVQLVESGGGLVQPGGSLTLSCAASRFMIS EYSMHWVRQAPGKGLEWVSTINPAGTTD YAESVKGRFTISRDNAKNTLYLQMNSLKP EDTAVYYCDGYGYRGQGTQVTVSSGGGG SGGGSEVQLVESGGGLVQPGNSLRLSCAA SGFTFS SFGMS WVRQ APGKGLEWVS SIS GS GSDTLYADSVKGRFTISRDNAKTTLYLQM NSLRPEDTAVYYCTIGGSLSRSSQGTLVTV

- 25 042472

SSGGGGSGGGSEVQLVESGGGLVQPGGSL KLSCAASGFTFNKYAMNWVRQAPGKGLE WV ARI RS К YNN Y AT Y Y AD S VKDRFTISRD DSKNTAYLQMNNLKTEDTAVYYCVRHGN FGNSYISYWAYWGQGTLVTVSSGGGGSG GGGSGGGGSQTWTQEPSLTVSPGGTVTL TCGSSTGAVTSGNYPNWVQQKPGQAPRG LIGGTKFLAPGTPARFSGSLLGGKAALTLS GVQPEDEAEYYCVLWYSNRWVFGGGTKL TVLHHHHHH eleven Three specific aCD3: aAlbumin: aPSMA S:A:R EVQLVESGGGLVQPGGSLKLSCAASGFTF NI<YAMNWVRQAPGI<GLEWVARIRSI<YN NYATYYADSVKDRFTISRDDSKNTAYLQM NNLKTEDTAVYYCVRHGNFGNSYISYWA YWGQGTLVTVSSGGGGSGGGGSGGGGSQ TWTQEPSLTVSPGGTVTLTCGSSTGAVTS GNYPNWVQQKPGQAPRGLIGGTKFLAPGT PARFSGSLLGGKAALTLSGVQPEDEAEYY CVLWYSNRWVFGGGTKLTVLGGGGSGGG SEVQLVESGGGLVQPGNSLRLSCAASGFTF SSFGMSWVRQAPGKGLEWVSSISGSGSDT LYADSVKGRFTISRDNAKTTLYLQMNSLR PEDTAVYYCTIGGSLSRSSQGTLVTVSSGG GGSGGGSEVQLVESGGGLVQPGGSLTLSC AASRFMISEYSMHWVRQAPGKGLEWVSTI NPAGTTDYAESVKGRFTISRDNAKNTLYL QMNSLKPEDTAVYYCDGYGYRGQGTQVT VSSHHHHHH 12 Three specific aAlbumin: aPSMA: aCD3 A:R:S EVQLVESGGGLVQPGNSLRLSCAASGFTFS SFGMSWVRQAPGKGLEWVSSISGSGSDTL YADSVKGRFTISRDNAKTTLYLQMNSLRP EDTAVYYCTIGGSLSRSSQGTLVTVSSGGG GSGGGSEVQLVESGGGLVQPGGSLTLSCA ASRFMISEYSMHWVRQAPGKGLEWVSTIN PAGTTDYAESVKGRFTISRDNAKNTLYLQ MNSLKPEDTAVYYCDGYGYRGQGTQVTV SSGGGGSEVQLVESGGGLVQPGGSLKLSC AASGFTFNKYAMNWVRQAPGKGLEWVA RIRSKYNNYATYYADSVKDRFTISRDDSK NTAYLQMNNLKTEDTAVYYCVRHGNFGN SYIS YWAYWGQGTLVTVS SGGGGSGGGG

- 26 042472

SGGGGSQTWTQEPSLTVSPGGTVTLTCGS STGAVTSGNYPNWVQQKPGQAPRGLIGGT KFLAPGTPARFSGSLLGKAALTLSGVQPE DEAEYYCVLWYSNRWVFGGGTKLTVLHH HHHH* 13 Three specific aAlbumin: aCD3:aPSMA A:S:R EVQLVESGGGLVQPGNSLRLSCAASGFTFS SFGMSWVRQAPGKGLEWVSSISGSGSDTL YADSVKGRFTISRDNAKTTLYLQMNSLRP EDTAVYYCTIGGSLSRSSQGTLVTVSSGGG GSGGGSEVQLVESGGGLVQPGGSLKLSCA ASGFTFNKYAMNWVRQAPGKGLEWVARI RSKYNNYATYYADSVKDRFTISRDDSKNT AYLQMNNLKTEDTAVYYCVRHGNFGNSY ISYWAYWGQGTLVTVSSGGGGSGGGGSG GGGSQTWTQEPSLTVSPGGTVTLTCGSST GAVTSGNYPNWVQQKPGQAPRGLIGGTKF LAPGTPARFSGSLLGGKAALTLSGVQPEDE AEYYCVLWYSNRWVFGGGTKLTVLGGGG SGGGSEVQLVESGGGLVQPGGSLTLSCAA SRFMISEYSMHWVRQAPGKGLEWVSTINP AGTTDYAESVKGRFTISRDNAKNTLYLQM NSLKPEDTAVYYCDGYGYRGQGTQVTVS SHHHHHH* 14 Three specific aPSMAscFv:aCD3: aAlbumin PSMA- scFv:C:A QVQLVESGGGLVKPGESLRLSCAASGFTFS DYYMYWVRQAPGKGLEWVAIISDGGYYT YYSDIIKGRFTISRDNAKNSLYLQMNSLKA EDTAVYYCARGFPLLRHGAMDYWGQGTL VT VS S GGGGS GGGGS GGGGSDIQMTQ SPS SLSASVGDRVTITCKASQNVDTNVAWYQQ KPGQAPKSLIYSASYRYSDVPSRFSGSASG TDFTLTISSVQSEDFATYYCQQYDSYPYTF GGGTKLEIKSGGGGSEVQLVESGGGLVQP GGSLKLS С A A SGFTFNK Y A MN WVRQ A PG KGLEWVARIRSKYNNYATYYADSVKDRF TISRDDSKNTAYLQMNNLKTEDTAVYYCV RHGNFGNSYISYWAYWGQGTLVTVSSGG GGSGGGGSGGGGSQTWTQEPSLTVSPGG TVTLTCGS STGAVTS GNYPNWVQQKPGQ APRGLIGGTKFLAPGTPARFSGSLLGGKAA LTLSGVQPEDEAEYYCVLWYSNRWVFGG GTKLTVLGGGGSGGGSEVQLVESGGGLVQ

- 27 042472

PGNSLRLSCAASGFTFSSFGMSWVRQAPG KGLEWVSSISGSGSDTLYADSVKGRFTISR DNAKTTLYLQMNSLRPEDTAVYYCTIGGS LSRSSQGTLVTVSSHHHHHH 15 PSMA BiTE QVQLVESGGGLVKPGESLRLSCAASGFTFS DYYMYWVRQAPGKGLEWVAIISDGGYYT YYSDIIKGRFTISRDNAKNSLYLQMNSLKA EDTAVYYCARGFPLLRHGAMDYWGQGTL VT VS S GGGGS GGGGS GGGGSDIQMTQ SPS SLSASVGDRVTITCKASQNVDTNVAWYQQ KPGQAPKSLIYSASYRYSDVPSRFSGSASG TDFTLTISSVQSEDFATYYCQQYDSYPYTF GGGTKLEIKSGGGGSEVQLVESGGGLVQP GGSLKLS С A A SGFTFNK Y A MN WVRQ APG KGLEWVARIRSKYNNYATYYADSVKDRF TISRDDSKNTAYLQMNNLKTEDTAVYYCV RHGNFGNSYISYWAYWGQGTLVTVSSGG GGSGGGGSGGGGSQTVVTQEPSLTVSPGG TVTLTCGS STGAVTS GNYPNWVQQKPGQ APRGLIGGTKFLAPGTPARFSGSLLGGKAA LTLSGVQPEDEAEYYCVLWYSNRWVFGG GTKLTVLHHHHHH 16 Three specific Her2- Finomer:aCB 3: a Albumin GVTLFVALYDYTSYNTRDLSFHKGEKFQIL RMEDGVWWEARSLTTGETGYIPSNYVAP VDSIQGGGGSGGGSEVQLVESGGGLVQPG GSLKLSCAASGFTFNKYAMNWVRQAPGK GLEWVARIRSKYNNYATYYADSVKDRFTI SRDDSKNTAYLQMNNLKTEDTAVYYCVR HGNFGNS YIS YWAYWGQGTLVTVS SGGG GSGGGGSGGGGSQTWTQEPSLTVSPGGT VTLTCGSSTGAVTSGNYPNWVQQKPGQAP RGLIGGTKFLAPGTPARFSGSLLGGKAALT LSGVQPEDEAEYYCVLWYSNRWVFGGGT KLTVLGGGGSGGGSEVQLVESGGGLVQPG NSLRLSCAASGFTFSSFGMSWVRQAPGKG LEWVSSISGSGSDTLYADSVKGRFTISRDN AKTTLYLQMNSLRPEDTAVYYCTIGGSLS RS S QGTLVT VS SHHHHHH* 17 Three specific AM: S EVQLVESGGGLVQPGNSLRLSCAASGFTFS SFGMSWVRQAPGKGLEWVSSISGSGSDTL YADSVKGRFTISRDNAKTTLYLQMNSLRP

- 28 042472

aCD3: aAlbumin: aMSLN EDTAVYYCTIGGSLSRSSQGTLVTVSSGGG GSGGGSQVQLVQSGGGLVQPGGSLRLSCA ASDFDFAAYDMSWVRQAPGQGLEWVAIIS HDGIDKYYDDSVKGRFTISRDNSKNTLYL QMNTLRAEDTATYQCLRLGAVGQGTLVT VSSGGGGSGGGSEVQLVESGGGLVQPGGS LKLSCAASGFTFNKYAMNWVRQAPGKGL EWVARIRSKYNNYATYYADSVKDRFTISR DDSKNTAYLQMNNLKTEDTAVYYCVRHG NFGNSYISYWAYWGQGTLVTVSSGGGGS GGGGSGGGGSQTWTQEPSLTVSPGGTVT LTCGSSTGAVTSGNYPNWVQQKPGQAPR GLIGGTKFLAPGTPARFSGSLLGGKAALTL SGVQPEDEAEYYCVLWYSNRWVFGGGTK LTVLHHHHHH 18 Three specific aAlbumin: aMSLN: aCD3 AM: S EVQLVESGGGLVQPGNSLRLSCAASGFTFS SFGMSWVRQAPGKGLEWVSSISGSGSDTL YADSVKGRFTISRDNAKTTLYLQMNSLRP EDTAVYYCTIGGSLSRSSQGTLVTVSSGGG GSGGGSQVQLVQSGGGLVQPGGSLRLSCA ASDFDFAAYDMSWVRQAPGQGLEWVAIIS HDGIDKYYDDSVKGRFTISRDNSKNTLYL QMNTLRAEDTATYQCLRLGAVGQGTLVT VSSGGGGSGGGSEVQLVESGGGLVQPGGS LKLSCAASGFTFNKYAMNWVRQAPGKGL EWVARIRSKYNNYATYYADSVKDRFTISR DDSKNTAYLQMNNLKTEDTAVYYCVRHG NFGNSYISYWAYWGQGTLVTVSSGGGGS GGGGSGGGGSQTWTQEPSLTVSPGGTVT LTCGSSTGAVTSGNYPNWVQQKPGQAPR GLIGGTKFLAPGTPARFSGSLLGGKAALTL SGVQPEDEAEYYCVLWYSNRWVFGGGTK LTVLHHHHHH 19 Three specific aAlbumin: aCD3: aMSLN ACM EVQLVESGGGLVQPGNSLRLSCAASGFTFS SFGMSWVRQAPGKGLEWVSSISGSGSDTL YADSVKGRFTISRDNAKTTLYLQMNSLRP EDTAVYYCTIGGSLSRSSQGTLVTVSSGGG GSGGGSEVQLVESGGGLVQPGGSLKLSCA ASGFTFNKYAMNWVRQAPGKGLEWVARI RSKYNNYATYYADSVKDRFTISRDDSKNT AYLQMNNLKTEDTAVYYCVRHGNFGNSY ISYWAYWGQGTLVTVSSGGGGSGGGGSG GGGSQTWTQEPSLTVSPGGTVTLTCGSST GAVTSGNYPNWVQQKPGQAPRGLIGGTKF LAPGTPARFSGSLLGGKAALTLSGVQPEDE AEYYCVLWYSNRWVFGGGTKLTVLGGGG SGGGSQVQLVQSGGGLVQPGGSLRLSCAA SDFDFAAYDMSWVRQAPGQGLEWVAIISH DGIDKYYDDSVKGRFTISRDNSKNTLYLQ MNTLRAEDTATYQCLRLGAVGQGTLVTV SSHHHHHH*

-29042472

Table 8. Sequences

SEQID NO: Connecting element Name Chain Subsequence 20 CD3 AntihuCD3EscFv DIKLQQSGAELARPGASVKMSCKTSGYTFTRY TMHWVKQRPGQGLEWIGYINPSRGYTNYNQK FKDKATLTTDKSSSTAYMQLSSLTSEDSAVYY CARYYDDHYCLDYWGQGTTLTVSSVEGGSGG SGGSGGSGGVDDIQLTQSPAIMSASPGEKVTM TCRASSSVSYMNWYQQKSGTSPKRWIYDTSK VASGVPYRFSGSGSGTSYSLTISSMEAEDAATY YCQQWS SNPLTFGAGTKLELK 21 CD3 AntihuCD3E Variable severe QVQLVESGGGVVQPGRSLRLSCAASGFKFSGY GMHWVRQAPGKGLEWVAVIWYDGSKKYYV DSVKGRFTISRDNSKNTLYLQMNSLRAEDTAV YYC ARQMGYWHFDLWGRGTL VT VS S 22 CD3 AntihuCD3E Variable light EIVLTQSPATLSLSPGERATLSCRASQSVSSYLA WYQQKPGQAPRLLIYDASNRATTGIPARFSGSG SGTDFTLTISSLEPEDFAVYYCQQRSNWPPLTF GGGTKVEIK 23 CD3 AntihuCD3E Variable severe EVQLLESGGGLVQPGGSLRLSCAASGFTFSSFP MAWVRQAPGKGLEWVSTISTSGGRTYYRDSV KGRFTISRDNSKNTLYLQMNSLRAEDTAVYYC AKFRQ YSGGFD YWGQGTLVT VS S 24 CD3 AntihuCD3E Variable light DIQLTQPNSVSTSLGSTVKLSCTLSSGNIENNY VHWYQLYEGRSPTTMIYDDDKRPDGVPDRFS GSIDRS SNS AFLTIHN VAIEDEAIYFCHS Y VS SF NVFGGGTKLTVLR 25 CD3 AntihuCD3EscFv DVQLVQSGAEVKKPGASVKVSCKASGYTFTR YTMHWVRQAPGQGLEWIGYINPSRGYTNYAD SVKGRFTITTDKSTSTAYMELSSLRSEDTATYY CARYYDDHYCLDYWGQGTTVTVSSGEGTSTG SGGSGGSGGADDIVLTQSPATLSLSPGERATLS CRASQSVSYMNWYQQKPGKAPKRWIYDTSKV ASGVPARFSGSGSGTDYSLTINSLEAEDAATYY CQQWS SNPLTF GGGTKVEIK 26 CD3 AntihuCD3E (humanized OCTZ) Variable severe QVQLVQSGGGVVQPGRSLRLSCKASGYTFTRY TMHWVRQAPGKGLEWIGYINPSRGYTNYNQK VKDRFTISRDNSKNTAFLQMDSLRPEDTGVYF CARYYDDHYCLDYWGQGTPVTVSS 27 CD3 AntihuCD3E (humanized Variable light DIQMTQSPSSLSASVGDRVTITCSASSSVSYMN WYQQTPGKAPKRWIYDTSKLASGVPSRFSGSG SGTDYTFTISSLQPEDIATYYCQQWSSNPFTFG QGTKLQITR

-30042472

th OKTZ) 28 CD3 CD3 binding element Variable severe EVQLLESGGGLVQPGGSLRLSCAASGFTFSTY AMNWVRQAPGKGLEWVSRIRSKYNNYATYY ADSVKGRFTISRDDSKNTLYLQMNSLRAEDTA VYYCVRHGNFGNSYVSWFAYWGQGTLVTVS S 29 CD3 CD3 binding element Variable light QAVVTQEPSLTVSPGGTVTLTCGSSTGAVTTS NYANWVQEKPGQAFRGLIGGTNKRAPGTPAR FSGSLLGGKAALTLSGAQPEDEAEYYCALWYS NLWVFGGGTKLTVL thirty CD3 CD3 binding element Variable severe EVQLVESGGGLVQPGRSLRLSCAASGFTFDDY TMHWVRQAPGKGLEWVSGISWNSGSIGYADS VKGRFTISRDNAKKSLYLQMNSLRAEDTALYY CAKDNSGYGHYYYGMDVWGQGTTVTVAS 31 CD3 CD3 binding element Variable light AEIVMTQSPATLSVSPGERATLSCRASQSVSSN L AWYQQKPGQ APRLLIYGASTRATGIPARF SG SGSGTEFTLTISSLQSEDFAVYYCQHYINWPLT FGGGTKVEIK 32 CD3 CD3 binding element Variable severe QVQLQQSGAELARPGASVKMSCKASGYTFTR STMHWVKQRPGQGLEWIGYINPSSAYTNYNQ KFKDKATLTADKSSSTAYMQLSSLTSEDSAVY YCASRQVHYDYNGFPYWGQGTLVTVSS 33 CD3 CD3 binding element Variable light QVVLTQSPAIMSAFPGEKVTMTCSASSSVSYM NWYQQKSGTSPKRWIYDSSKLASGVPARFSGS GSGTSYSLTISSMETEDAATYYCQQWSRNPPTF GGGTKLQITR 34 CD3 CD3 binding element Variable severe EVKLLESGGGLVQPKGSLKLSCAASGFTFNTY AMNWVRQAPGKGLEWVARIRSKYNNYATYY ADSVKDRFTISRDDSQSILYLQMNNLKTEDTA MYYCVRHGNFGNSYVSWFAYWGQGTLVTVS A 35 CD3 CD3 binding element Variable light QAVVTQESALTTSPGETVTLTCRSSTGAVTTSN YANWVQEKPDHLFTGLIGGTNKRAPGVPARFS GSLIGDKAALTITGAQTEDEAIYFCALWYSNL WVFGGGTKLTVLG 36 CD3 humanized scFv EVQLVESGGGLVQPGGSLRLSCAASGFTFNTY AMNWVRQAPGKGLEWVGRIRSKYNNYATYY ADSVKGRFTISRDDSKNTLYLQMNSLRAEDTA VYYCVRHGNFGNSYVSWFAYWGQGTLVTVS SGGGGSGGGGSGGGGSQAVVTQEPSLTVSPGG Т VTLTCGS STGAVTTSNYANWVQQKPGQAPR GLIGGTNKRAPGVPARFSGSLLGGKAALTLSG AQPEDEAEYYCALWYSNLWVFGGGTKLTVL 37 CD3 CD3 binding QVQLQQSGAELARPGASVKMSCKASGYTFTR YTMHWVKQRPGQGLEWIGYINPSRGYTNYNQ KFKDKATLTTDKSSSTAYMQLSSLTSEDSAVY

- 31 042472

element YC ARYYDDHYSLDYWGQGTTLT VS S AKTTPD IVLTQSPAIMSASPGEKVTMTCSASSSVSYMN WYQQKSGTSPKRWIYDTSKLASGVPAHFRGS GSGTSYSLTISGMEAEDAATYYCQQWSSNPFT FGSGTKLEINRADTAAAGSHHHHHH 38 HSA VH domain only EVQLLESGGGLVQPGGSLRLSCAASGFTFSKY WMS WVRQ APGKGLEWVS SIDFMGPHT YYAD SVKGRFTISRDNSKNTLYLQMNSLRAEDTAVY YCAKGRTSMLPMKGKFDYWGQGTLVTVSS 39 HSA VH domain only EVQLLESGGGLVQPGGSLRLSCTASGFTFDEY NMSWVRQAPGKGLEWVSTILPHGDRTYYADS VKGRFTISRDNSKNTLYLQMNSLRAEDTAVYY WITH AKQDPLYRFD YWGQGTL VT VS S 40 HSA VL domain only DIQMTQSPSSLSASVGDRVTITCRASQKIATYL NWYQQKPGKAPKLLIYRSSSLQSAVPSRFSGS GSGTVFTLTISSLQPEDFATYYCQQTYAVPPTF GQGTKVEIKR 41 HSA VL domain only DIQMTQSPSSLSASVGDRVTITCRASQSISSYLN WYQQKPGKAPKLLIYRNSPLQSGVPSRFSGSG SGTDFTLTISSLQPEDFATYYCQQTYRVPPTFG QGTKVEIKR 42 HSA MSA21 QVQLQESGGGLVQPGGSLRLSCEASGFTFSRF GMTWVRQAPGKGVEWVSGISSLGDSTLYADS VKGRFTISRDNAKNTLYLQMNSLKPEDTAVYY CTIGGSLNPGGQGTQVTVSS 43 HSA Non-natural consensus albumin binding domains LKEAKEKAIEELKKAGITSDYYFDLINKAKTVE GVNALKDEILKA 44 HSA anti- ALBFAB Varkabelnaya heavy EVQLLESGGGLVQPGGSLRLSCAVSGIDLSNY AINWVRQAPGKCLEWIGIIWASGTTFYATWAK GRFTISRDNSKNTVYLQMNSLRAEDTAVYYCA RTVPGYSTAPYFDLWGQGTLVTVSS 45 HSA anti- ALBFAB Variable light DIQMTQSPSSVSASVGDRVTITCQSSPSVWSNF LSWYQQKPGKAPKLLIYEASKLTSGVPSRFSGS GSGTDFTLTISSLQPEDFATYYCGGGYSSISDTT FGCGTKVEIKRT 46 HSA Only HSAVH AVQLVESGGGLVQPGNSLRLSCAASGFTFRSF GMSWVRQAPGKEPEWVSSISGSGSDTLYADSV KGRFTISRDNAKTTLYLQMNSLKPEDTAVYYC TIGGSLSRSSQGTQVTVSS 47 HSA Only HSAVH EVQLVESGGGLVQPGNSLRLSCAASGFTFSSFG MSWVRQAPGKGLEWVSSISGSGSDTLYADSV KGRFTISRDNAKTTLYLQMNSLRPEDTAVYYC TIGGSLSRSSSQGTLVTVSS 48 HSA Only HSAVH AVQLVESGGGLVQGGGSLRLACAASERIFDLN LMGWYRQGPGNERELVATCITVGDSTNYADS VKGRFTISMDYTKQTVYLHMNSLRPEDTGLY YCKIRRTWHSELWGQGTQVTVSS

While the present invention has shown and described the preferred embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are given by way of example only. Numerous variations, modifications and substitutions will be apparent to those skilled in the art without departing from the spirit of the invention. It should be understood that various alternatives to the embodiments of the invention described in the present invention may be used in the practice of the invention. It is intended that the following claims define the scope of the present invention, and that methods and constructions within the scope of these requirements and their equivalents will thus be covered.

-

Claims (20)

CLAIM 1. A trispecific antigen-binding protein, wherein the trispecific antigen-binding protein contains: (a) a first domain (A) that contains a single chain variable fragment (scFv) that specifically binds to human CD3; (b) a second domain (B) that contains a single domain antibody (sdAb) that binds human serum albumin; and (c) a third domain (C) that contains an sdAb that specifically binds to a tumor target antigen that is EGFR, PSMA, HER2, or MSLN, wherein the trispecific antigen binding protein has an affinity for binding to the tumor target antigen of 200 nM or less; moreover, the domains are connected in the order H ₂ N-(C)-(B)-(A)-COOH or using linkers L1 and L2; wherein the L1 and L2 linkers are independently (GS) _n (SEQ ID NO: 49), (GGS) _n (SEQ ID NO: 50), (GGGS) n (SEQ ID NO: 51), (GGSG) n (SEQ ID NO: 52) or (GGSGG)n (SEQ ID NO: 53), (GGGGS)n (SEQ ID NO: 54), where n is 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10, and wherein the trispecific antigen binding protein is less than 60 kDa. 2. The trispecific antigen binding protein of claim 1, wherein the first domain comprises complementarity determining regions (CDRs) selected from the group consisting of muromonab-CD3 (OKT3), telixumab (TRX4), teplizumab (MGA031), vizilizumab (Nuvion), SP34, X35, VIT3, VMA030 (BW264/56), CLB-T3/3, CRIS7, YTH12.5, F111-409, CLB-T3.4.2, TR-66, WT32, SPv-T3b, 11D8, XIII-141 , XIII-46, XIII-87, 12F6, T3/RW2-8C8, T3/RW2-4B6, OKT3D, M-T301, SMC2, F101.01, UCHT-1 and WT-31. 3. The trispecific antigen binding protein of claim 1, wherein the first domain is humanized or human. 4. The trispecific antigen binding protein of claim 1, wherein the L1 and L2 linkers are each independently (GGGGS) ₄ (SEQ ID NO: 55) or (GGGGS) ₃ (SEQ ID NO: 56). 5. The trispecific antigen binding protein of claim 1, wherein the first domain is specific for CD3ε (epsilon). 6. The trispecific antigen binding protein of claim 1, wherein the first domain is cross-reactive with cynomolgus CD3. 7. The trispecific antigen binding protein of claim 1, wherein the third domain specifically binds to EGFR. 8. The trispecific antigen binding protein of claim 1, wherein the third domain specifically binds to PSMA. 9. The trispecific antigen binding protein of claim 1, wherein the third domain specifically binds to HER2. 10. The trispecific antigen binding protein of claim 1, wherein the third domain specifically binds to the MSLN. 11. The trispecific antigen binding protein of claim 1, wherein the trispecific antigen binding protein has a half-life of at least about 50 hours. 12. Pharmaceutical composition for the treatment of a proliferative disease of an individual, containing: (i) trispecific antigen-binding protein according to any one of paragraphs. 1-11, and (ii) a pharmaceutically acceptable carrier. 13. The use of a trispecific antigen binding protein according to any one of claims 1 to 11 for the manufacture of a medicament for the treatment of cancer in an individual. 14. Use according to claim 13, wherein the cancer is a hematological cancer. 15. Use according to claim 13, wherein the cancer is a cancer with a solid tumor. 16. The use of claim 15 wherein the solid tumor cancer is breast cancer, lung cancer, prostate cancer, ovarian cancer, or pancreatic cancer. 17. Use according to claim 13, wherein the drug is to be administered in combination with chemotherapy, hormonal therapy, radiotherapy, gene therapy or immunotherapy. 18. Use according to claim 13, wherein the drug is to be administered in combination with an antidiarrheal agent, an antiemetic, an analgesic, an opioid, and/or a non-steroidal anti-inflammatory agent. 19. The use of a trispecific antigen-binding protein according to any one of claims 1 to 6 for the manufacture of a medicament for the treatment of a disease, disorder or condition associated with PSMA in an individual, wherein said (c) third domain (C) contains an sdAb that specifically binds to a tumor the target antigen, which is PSMA. 20. The use of a trispecific antigen binding protein according to any one of claims 1 to 6 for the manufacture of a medicament for the treatment of a disease, disorder or condition associated with -