Library of antigen-binding molecules containing modified antibody variable regions

Abstract

To provide a library that consists of a plurality of antigen-binding molecules having different sequences from each other, and includes a variable region of an antibody, in which the variable region has binding activity to a first antigen and a second antigen different from the first antigen, but does not simultaneously bind to the first antigen and the second antigen.SOLUTION: A library that mainly consists of a plurality of antigen-binding molecules having different sequences from each other, in which an antigen-binding region in the antigen-binding molecule is a variable region of an antibody which can bind to the first antigen and a second antigen different from the first antigen, but does not simultaneously bind to the first antigen and the second antigen, either one of the first antigen and the second antigen is CD3, and the other antigen is a molecule expressed on the surface of a T-cell or other immunocyte.SELECTED DRAWING: None

Description

[Technical Field]

[0001] The present invention provides a population (library) of antigen-binding molecules comprising antibody variable regions that can bind to two different antigens (a first antigen and a second antigen) but do not simultaneously bind to both antigens, a method for producing the library, a method for producing bispecific antibodies comprising a common light chain variable region using the library, and a method for selecting antigen-binding molecules comprising variable regions with enhanced binding to a first antigen.The present invention also relates to antigen-binding molecules comprising antibody variable regions that can bind to three different antigens (a first antigen, a second antigen, and a fourth antigen) but do not simultaneously bind to the three antigens, pharmaceutical compositions comprising the antigen-binding molecules, and methods for producing them. [Background technology]

[0002] Antibodies have attracted attention as pharmaceuticals due to their high stability in plasma and minimal side effects (Nat. Biotechnol. (2005) 23, 1073-1078 (Non-Patent Document 1) and Eur J Pharm Biopharm. (2005) 59 (3), 389-396 (Non-Patent Document 2)). Antibodies not only bind to antigens and have agonistic or antagonistic effects, but also induce cytotoxic activity (also known as effector function) by effector cells, such as ADCC (Antibody Dependent Cytotoxicity), ADCP (Antibody Dependent Cell phagocytosis), and CDC (Complement-Dependent Cytotoxicity). Because antibodies of the IgG1 subclass in particular exhibit effector function against cancer cells, numerous antibody drugs have been developed for the cancer field.

[0003] For an antibody to exhibit ADCC, ADCP, and CDC, it is essential that the Fc region of the antibody binds to antibody receptors (FcγR) and various complement components present on effector cells such as NK cells and macrophages. In humans, the FcγR protein family has been reported to include FcγRIa, FcγRIIa, FcγRIIb, FcγRIIIa, and FcγRIIIb isoforms, and their respective allotypes have also been reported (Immunol. Lett. (2002) 82, 57-65 (Non-Patent Document 3)). Of these isoforms, FcγRIa, FcγRIIa, and FcγRIIIa contain a domain called ITAM (Immunoreceptor Tyrosine-based Activation Motif) in the intracellular domain, which transduces activation signals. On the other hand, only FcγRIIb contains a domain called ITIM (Immunoreceptor Tyrosine-based Inhibitory Motif) in its intracellular domain, which transmits inhibitory signals. It is known that all FcγRs transmit signals by being crosslinked by immune complexes (Nat. Rev. Immunol. (2008) 8, 34-47 (Non-Patent Document 4)). In fact, when an antibody exerts its effector function on cancer cells, the Fc region of multiple antibodies bound to the cancer cell membrane clusters FcγRs on the effector cell membrane, transducing an activation signal in the effector cells. This results in a cytocidal effect, but FcγR crosslinking is limited to effector cells located near the cancer cells, indicating that immune activation occurs only locally in the cancer cells (Ann. Rev. Immunol. (1988). 6, 251-81 (Non-Patent Document 5)).

[0004] Native immunoglobulins bind to antigens through their variable regions and complement receptors such as FcγR, FcRn, FcαR, and FcεR through their constant regions. FcRn, one of the binding molecules that interacts with the Fc region of IgG, binds to each heavy chain of the antibody, with one molecule of FcRn binding to each IgG antibody molecule. However, unlike FcRn, FcγR interacts with the antibody's hinge region and CH2 domain, and only one molecule of FcγR binds to each IgG antibody molecule (J. Bio. Chem., (20001) 276, 16469-16477). Furthermore, it has been shown that the binding between FcγR and an antibody Fc region is important for several amino acid residues in the antibody hinge region and CH2 domain, and for the sugar chain attached to Asn at EU numbering position 297 attached to the CH2 domain (Chem. Immunol. (1997), 65, 88-110 (Non-Patent Document 6), Eur. J. Immunol. (1993) 23, 1098-1104 (Non-Patent Document 7), Immunol. (1995) 86, 319-324 (Non-Patent Document 8)). Focusing on this binding site, various Fc region mutants with FcγR binding properties have been studied, and Fc region mutants with higher binding activity to activating FcγR have been obtained (WO2000 / 042072 (Patent Document 1), WO2006 / 019447 (Patent Document 2)). For example, Lazar et al. succeeded in increasing the binding activity of human IgG1 to human FcγRIIIa (V158) by approximately 370-fold by substituting Ser at position 239 (EU numbering), Ala at position 330, and Ile at position 332 (EU numbering) with Asn, Leu, and Glu, respectively (Proc. Natl. Acad. Sci. USA (2006) 103, 4005-4010 (Non-Patent Document 9), WO2006 / 019447 (Patent Document 2)). This variant has an approximately 9-fold higher ratio of binding activity to FcγRIIIa and FcγIIb (A / I ratio) compared to the wild-type.In addition, Shinkawa et al. succeeded in increasing the FcγRIIIa-binding activity by approximately 100-fold by deleting the fucose in the sugar chain attached to Asn at position 297 (EU numbering) (J. Biol. Chem. (2003) 278, 3466-3473 (Non-Patent Document 10)). These methods make it possible to significantly improve the ADCC activity of human IgG1 compared to that of native human IgG1.

[0005] Conventional natural IgG antibodies recognize and bind to a single epitope using their variable region (Fab), and therefore can only bind to a single antigen. However, cancer and inflammation are known to involve multiple proteins, and crosstalk between proteins can occur. For example, several inflammatory cytokines (TNF, IL1, and IL6) are known to be involved in immune diseases (Nat. Biotech., (2011) 28, 502-10 (Non-Patent Document 11)). Furthermore, activation of other receptors is known to be one of the mechanisms by which cancer acquires drug resistance (Endocr Relat Cancer (2006) 13, 45-51 (Non-Patent Document 12)). In such cases, conventional antibodies that recognize a single epitope cannot inhibit multiple proteins.

[0006] Antibodies that bind to two or more antigens with a single molecule (called bispecific antibodies) are being researched as molecules that inhibit multiple targets. By improving natural IgG-type antibodies, it is possible to give them the ability to bind to two different antigens (first and second antigens) (MAbs. (2012) Mar 1, 4(2)). Therefore, not only can a single molecule neutralize two or more antigens, but it can also enhance antitumor activity by cross-linking cytotoxic cells with cancer cells. Previously reported molecular forms of bispecific antibodies include molecules in which antigen-binding sites are added to the N- or C-terminus of the antibody (DVD-Ig and scFv-IgG), molecules in which the two Fab regions of the antibody have different sequences (common light chain bispecific antibodies and hybrid hybridomas), molecules in which one Fab region recognizes two antigens (two-in-one IgG), and molecules in which the loop region of the CH3 region serves as a new antigen-binding site (Fcab) (Nat. Rev. (2010), 10, 301-316 (Non-Patent Document 13), Peds (2010), 23(4), 289-297 (Non-Patent Document 14)). Since all bispecific antibodies interact with FcγR via the Fc region, the antibody effector function is preserved. Therefore, bispecific antibodies simultaneously bind to FcγR for all antigens they recognize and exhibit ADCC activity against cells expressing the antigen.

[0007] If all of the antigens recognized by a bispecific antibody are specifically expressed in cancer, binding to either antigen will result in cytotoxic activity against cancer cells, and thus a more effective anticancer effect can be expected than with conventional antibody drugs that recognize a single antigen. However, if any one of the antigens recognized by a bispecific antibody is expressed in normal tissues or on immune cells, cross-linking with FcγR will cause damage to normal tissues and release of cytokines (J. Immunol. (1999) Aug 1, 163(3), 1246-52 (Non-Patent Document 15)). As a result, strong side effects will be induced.

[0008] For example, catumaxomab is known as a bispecific antibody that recognizes proteins expressed on T cells and proteins expressed on cancer cells (cancer antigens). Catumaxomab has two Fab fragments that bind to a cancer antigen (EpCAM) and the CD3ε chain expressed on T cells, respectively. Catumaxomab induces cytotoxic activity by T cells through simultaneous binding of the cancer antigen and CD3ε, and induces cytotoxic activity by antigen-presenting cells such as NK cells and macrophages through simultaneous binding of the cancer antigen and FcγR. By utilizing these two cytotoxic activities, catumaxomab has been shown to be highly effective in treating malignant ascites when administered intraperitoneally, and has been approved in Europe. (Cancer Treat Rev. (2010) Oct 36(6), 458-67 (Non-Patent Document 16)) Furthermore, cases have been reported in which antibodies that react against cancer cells were produced by the administration of catumaxomab, demonstrating the induction of adaptive immunity (Future Oncol. (2012) Jan 8(1), 73-85 (Non-Patent Document 17)). Based on these results, antibodies that possess both T cell-mediated cytotoxic activity and FcγR-mediated effects on cells such as NK cells and macrophages (particularly so-called trifunctional antibodies) are attracting attention because they are expected to have strong anti-tumor effects and induce adaptive immunity.

[0009] However, trifunctional antibodies simultaneously bind to CD3ε and FcγR even in the absence of cancer antigens, and therefore crosslink CD3ε-expressing T cells with FcγR-expressing cells, resulting in the production of large amounts of various cytokines, even in an environment where cancer cells are not present. Due to this cancer antigen-independent induction of cytokine production, trifunctional antibodies are currently only administered intraperitoneally (Cancer Treat Rev. 2010 Oct 36(6), 458-67 (Non-Patent Document 16)), and systemic administration is extremely difficult due to the severe cytokine storm-like side effects (Cancer Immunol Immunother. 2007 Sep;56(9):1397-406 (Non-Patent Document 18)). Furthermore, with bispecific antibodies of the prior art, both the first antigen, a cancer antigen (EpCAM), and the second antigen, CD3ε, can bind to FcγR simultaneously, and it is therefore structurally impossible to avoid such side effects caused by the simultaneous binding of FcγR and the second antigen, CD3ε.

[0010] In recent years, improved antibodies have been provided that use an Fc region with reduced binding activity to FcγR, thereby eliciting cytotoxic activity by T cells while avoiding side effects ( WO2012 / 073985 ). However, due to their molecular structure, even such antibodies cannot bind to cancer antigens while simultaneously acting on the two immune receptors, CD3ε and FcγR. To date, no antibody has been known that can induce both T cell-mediated cytotoxicity and non-T cell-mediated cytotoxicity in a cancer antigen-specific manner while avoiding side effects. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] WO2000 / 042072 [Patent Document 2] WO2006 / 019447 [Non-patent literature]

[0012] [Non-Patent Document 1] Nat. Biotechnol. (2005) 23, 1073-1078 [Non-patent document 2] Eur J Pharm Biopharm. (2005) 59 (3), 389-396 [Non-patent document 3] Immunol. Lett. (2002) 82, 57-65 [Non-patent document 4] Nat. Rev. Immunol. (2008) 8, 34-47

Non-licensed Document 5

Non-licensed Document 6

Non-licensed Document 7

Non-licensed literature 9

Non-licensed literature 10

Non-licensed Document 11

Non-licensed Document 12

Non-licensed Document 13

Non-licensed Document 14

Non-licensed Document 15

Non-licensed Document 16

Non-licensed Document 17

Non-licensed Document 18

[0013] The present invention was made in light of these circumstances, and its objective is to provide a population (library) of antigen-binding molecules comprising variable regions of antibodies that have binding activity to two different antigens (a first antigen and a second antigen) but do not simultaneously bind to these antigens, a method for producing the library, a method for selecting or producing desired antigen-binding molecules using the library, and a method for selecting antigen-binding molecules comprising variable regions with enhanced binding to a first antigen. Another objective of the present invention is to provide antigen-binding molecules comprising variable regions of antibodies that can bind to three different antigens (a first antigen, a second antigen, and a fourth antigen) but do not simultaneously bind to the three antigens, pharmaceutical compositions comprising the antigen-binding molecules, and methods for producing them. [Means for solving the problem]

[0014] The present inventors conducted extensive research to solve the above-mentioned problems. As a result, the present inventors prepared an antigen-binding molecule comprising a variable region of an antibody having binding activity to two different antigens (a first antigen and a second antigen) but not simultaneously binding to these antigens, and a variable region that binds to a different antigen (a third antigen). By utilizing the binding activity of the antigen-binding molecule to three different antigens, the inventors succeeded in enhancing the activity of the antigen-binding molecule. Furthermore, the inventors succeeded in preparing an antigen-binding molecule that can avoid cross-linking between different cells that occurs when binding to antigens expressed on different cells, which is thought to cause side effects when conventional multispecific antigen-binding molecules are used as pharmaceuticals. Furthermore, the inventors succeeded in preparing an antigen-binding molecule library comprising variable regions of an antibody that can bind to two different antigens (a first antigen and a second antigen) but not simultaneously binding to both antigens. Furthermore, the inventors used the library to obtain (select) antigen-binding molecules with binding activity to two desired antigens, and to obtain (select) variable regions with enhanced binding to the desired antigen. Furthermore, the present inventors have succeeded in producing an antigen-binding molecule comprising the variable region of an antibody that can bind to three different antigens (a first antigen, a second antigen, and a fourth antigen) but does not bind to the three antigens simultaneously.

[0015] More specifically, the present invention relates to the following: [1] A library mainly composed of a plurality of antigen-binding molecules whose sequences differ from one another, wherein the antigen-binding regions in the antigen-binding molecules are antibody variable regions that can bind to a first antigen and a second antigen different from the first antigen, but do not simultaneously bind to the first antigen and the second antigen; wherein either the first antigen or the second antigen is CD3, and the other antigen is a molecule expressed on the surface of T cells or other immune cells. [2] The library according to [1], wherein the variable region that does not simultaneously bind to the first antigen and the second antigen is a variable region that does not simultaneously bind to the first antigen and the second antigen, each of which is expressed on different cells. [3] The library according to [1] or [2], wherein the variable region is a variable region into which at least one amino acid modification has been introduced, and the modified amino acid is an amino acid in the CDR1, CDR2, CDR3, or FR3 region of the variable region of an antibody. [4] The library according to [3], wherein the modified amino acid is an amino acid at at least one position selected from positions 31 to 35, 50 to 65, 71 to 74, and 95 to 102 (Kabat numbering) in the antibody heavy chain variable region, and positions 24 to 34, 50 to 56, and 89 to 97 (Kabat numbering) in the antibody light chain variable region. [5] The library described in any one of [1] to [4], wherein the other antigen is FcγR, TLR, lectin, IgA, an immune checkpoint molecule, a TNF superfamily molecule, a TNFR superfamily molecule, or an NK receptor molecule. [6] The library according to any one of [1] to [5], wherein the antigen-binding molecule is a fusion polypeptide of an antibody variable region and at least a portion of a viral coat protein. [7] The library according to any one of [3] to [6], wherein the at least one amino acid modification is introduced into a template sequence consisting of the heavy chain variable region sequence set forth in SEQ ID NO: 96 and / or the light chain variable region sequence set forth in SEQ ID NO: 53, and the modified amino acid is an amino acid at one or more positions selected from the following: H chain: 31, 52b, 52c, 53, 54, 56, 57, 61, 98, 99, 100, 100a, 100b, 100c, 100d, 100e, 100f, 100g (Kabat numbering); Light chain: 24, 25, 26, 27, 27a, 27b, 27c, 27e, 30, 31, 33, 34, 51, 52, 53, 54, 55, 56, 74, 77, 89, 90, 92, 93, 94, 96 (Kabat numbering). [8] The following steps (a) and (b): (a) using a variable region sequence of an antibody that binds to a first antigen as a template sequence for a library, and identifying amino acid alterations that satisfy any one or more of the following (i) to (iii): (i) The modification does not substantially change the binding ability to the first antigen; (ii) the modification does not substantially alter the ECM binding ability; and (iii) an insertion of a peptide consisting of 1 to 25 amino acid residues into the CDR1, CDR2, CDR3, or FR3 region of the heavy chain variable region; and (b) designing a library containing nucleic acids encoding the template sequence and nucleic acids encoding variable regions of the template sequence that have different sequences and that have one or more amino acid alterations identified in step (a); A method for producing the library described in any one of [1] to [7], comprising: [9] The method for producing the library according to [8], characterized in that the library is produced using the heavy chain variable region sequence set forth in SEQ ID NO: 96 and / or the light chain variable region sequence set forth in SEQ ID NO: 53 as template sequences for the library.

[10] A method for producing an antigen-binding molecule comprising a variable region that can bind to a first antigen, which is CD3, and a second antigen different from the first antigen, but does not simultaneously bind to the first antigen and the second antigen, the method comprising the following steps (a) to (c): (a) contacting the library according to any one of [1] to [7] with a second antigen; (b) recovering the antigen-binding molecule bound to the second antigen in step (a); and (c) selecting, from the population of antigen-binding molecules collected in step (b), antigen-binding molecules comprising variable regions that do not simultaneously bind to the first antigen and the second antigen.

[11] The method for producing an antigen-binding molecule of

[10] , wherein the second antigen is FcγR, TLR, lectin, IgA, an immune checkpoint molecule, a TNF superfamily molecule, a TNFR superfamily molecule, or an NK receptor molecule.

[12] A method for producing a bispecific antibody using a library mainly composed of a plurality of antigen-binding molecules having sequences different from one another, wherein the antigen-binding regions in the antigen-binding molecules are variable regions of antibodies consisting of a template sequence or antibody variable regions having at least one amino acid modification in the template sequence, and the variable regions are antibody variable regions that can bind to a first antigen and a second antigen different from the first antigen, but do not simultaneously bind to the first antigen and the second antigen, the method comprising the steps of (a) to (c) below: (a) selecting a template sequence from the library as a variable region that binds to a first antigen; (b) selecting a variable region that binds to the second antigen but not the first antigen as a variable region that binds to the second antigen, the step comprising the following steps (i) to (iv): (i) contacting the library with a second antigen; (ii) recovering the antigen-binding molecule bound to the second antigen in step (i); (iii) contacting the population of antigen-binding molecules collected in step (ii) with a first antigen; and (iv) selecting antigen-binding molecules that do not bind to the first antigen in step (iii); and (c) producing a bispecific antibody comprising a variable region that binds to the first antigen selected in step (a) and a variable region that binds to the second antigen selected in step (b).

[13] The method for producing a bispecific antibody according to

[12] , wherein the first antigen is CD3 and the second antigen is a molecule that is specifically expressed in a cancer tissue.

[14] The following steps (a) to (c): (a) contacting the library according to any one of [1] to [7] with a first antigen; (b) recovering the antigen-binding molecule bound to the first antigen in step (a); and (c) selecting antigen-binding molecules comprising a variable region with enhanced binding to the first antigen from the population of antigen-binding molecules bound to the first antigen in step (b). A method for selecting a variable region with enhanced binding to a first antigen, comprising:

[15] An antigen-binding molecule comprising an antibody variable region that can bind to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen), but does not bind to the three antigens simultaneously.

[16] The antigen-binding molecule of

[15] , further comprising a variable region that binds to a third antigen different from the three antigens.

[17] The antigen-binding molecule of

[15] or

[16] , wherein the variable regions that do not simultaneously bind to the three antigens are variable regions that do not simultaneously bind to a first antigen, a second antigen, and a fourth antigen, each of which is expressed on a different cell.

[18] The antigen-binding molecule of any one of

[15] to

[17] , further comprising an Fc region of an antibody.

[19] The antigen-binding molecule of

[18] , wherein the Fc region has a reduced FcγR-binding activity compared to the Fc region of a native human IgG1 antibody.

[20] The antigen-binding molecule of any of

[15] to

[19] , wherein the variable region of the antibody capable of binding to the three antigens is a variable region into which at least one amino acid modification has been introduced, and the modified amino acid is an amino acid in the CDR1, CDR2, CDR3, or FR3 region of the antibody variable region.

[21] The antigen-binding molecule of

[20] , wherein the modified amino acid is an amino acid at at least one position selected from positions 31 to 35, 50 to 65, 71 to 74, and 95 to 102, according to Kabat numbering, in the antibody heavy chain variable region, and positions 24 to 34, 50 to 56, and 89 to 97, according to Kabat numbering, in the antibody light chain variable region.

[22] The antigen-binding molecule of any of

[15] to

[21] , wherein any one of the first antigen, the second antigen, and the fourth antigen is a molecule specifically expressed on the surface of T cells, and the remaining two antigens are molecules expressed on the surface of T cells or other immune cells.

[23] The antigen-binding molecule of

[22] , wherein the remaining two antigens are FcγR, TLR, lectin, IgA, an immune checkpoint molecule, a TNF superfamily molecule, a TNFR superfamily molecule, or an NK receptor molecule.

[24] The antigen-binding molecule of any one of

[16] to

[23] , wherein the third antigen is a molecule that is expressed specifically in cancer tissue.

[25] A pharmaceutical composition comprising the antigen-binding molecule of any one of

[15] to

[24] and a medically acceptable carrier.

[26] A method for producing the antigen-binding molecule of any one of

[15] to

[24] , comprising steps (i) to (iv): (i) preparing a library of antigen-binding molecules comprising variable regions in which at least one amino acid in the variable regions of antibodies that bind to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen) has been modified, and in which at least one amino acid in the modified variable regions differs from each other; (ii) selecting, from the prepared library, antigen-binding molecules comprising variable regions that have binding activity to the three antigens but do not simultaneously bind to the three antigens; (iii) culturing host cells containing nucleic acids encoding the variable regions of the antigen-binding molecules selected in step (ii) to express antigen-binding molecules comprising variable regions of antibodies that can bind to the three antigens but do not simultaneously bind to the three antigens; and (iv) recovering the antigen-binding molecule from the host cell culture.

[27] The production method of

[26] , wherein the variable regions contained in the antigen-binding molecules selected in step (ii) that do not simultaneously bind to the three antigens are variable regions that do not simultaneously bind to the three antigens expressed on different cells.

[28] The production method described in

[26] or

[27] , wherein the host cells cultured in step (iii) further contain a nucleic acid encoding the Fc region of an antibody.

[29] The production method according to

[28] , wherein the binding activity of the Fc region to FcγR is reduced compared to the binding activity of the Fc region of a native human IgG1 antibody to FcγR.

[30] The production method according to any one of

[26] to

[29] , wherein the antigen-binding molecule produced is a multispecific antibody.

[31] The method of any one of

[26] to

[30] , wherein at least one modified amino acid in the variable region in step (i) is a substituted or inserted amino acid.

[32] The production method according to

[31] , wherein the number of inserted amino acids is 1 to 25.

[33] The method of any one of

[26] to

[32] , wherein the modification is an amino acid modification in the CDR1, CDR2, CDR3, or FR3 region of the antibody variable region.

[34] The method of any one of

[26] to

[33] , wherein the modification is a modification of an amino acid in a loop region.

[35] The production method according to any one of

[26] to

[33] , wherein the modification is a modification of at least one amino acid selected from among 31 to 35, 50 to 65, 71 to 74, and 95 to 102, according to the Kabat numbering, in the antibody heavy chain variable region, and 24 to 34, 50 to 56, and 89 to 97, according to the Kabat numbering, in the antibody light chain variable region.

[36] A manufacturing method described in any one of

[26] to

[35] , wherein any one of the first antigen, the second antigen, and the fourth antigen is a molecule specifically expressed on the surface of T cells, and the remaining two antigens are molecules expressed on the surface of T cells or other immune cells.

[37] The manufacturing method described in

[36] , wherein any one of the first antigen, second antigen, and fourth antigen is CD3, and the remaining two antigens are FcγR, TLR, IgA, lectin, immune checkpoint molecule, TNF superfamily molecule, TNFR superfamily molecule, or NK receptor molecule.

[38] The production method of any of

[26] to

[37] , wherein the antigen-binding molecule further comprises a variable region that binds to a third antigen different from the three antigens.

[39] The manufacturing method described in

[38] , wherein the third antigen is a molecule that is expressed specifically in cancer tissue.

[40] A method for treating cancer, comprising the step of administering the antigen-binding molecule of any one of

[15] to

[24] .

[41] The antigen-binding molecule of any one of

[15] to

[24] for use in the treatment of cancer.

[42] Use of the antigen-binding molecule of any of

[15] to

[24] in the manufacture of a cancer therapeutic agent.

[43] A process for producing a cancer therapeutic agent, comprising a step of using the antigen-binding molecule of any one of

[15] to

[24] . It will be understood by those skilled in the art that any combination of one or more of the above-described aspects is also included in the present invention, as long as there is no technical contradiction based on the technical common sense of those skilled in the art. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a conceptual diagram of an antibody that binds to a first antigen and a second antigen, but not simultaneously. [Figure 2] This is a conceptual diagram of an antibody that does not bind to two antigens simultaneously and therefore does not cross-link. [Figure 3] This is a conceptual diagram of an antibody that simultaneously binds to two antigens but does not simultaneously bind to two cells. [Figure 4] FIG. 1 is a conceptual diagram of an antibody that crosslinks cancer cells with T cells expressing a first receptor. [Figure 5]

[0023] Figure 1 is a conceptual diagram of an antibody that crosslinks cancer cells with cells expressing a second receptor. [Figure 6] A conceptual diagram of an antibody that crosslinks cancer cells and immune cells but does not crosslink immune cells themselves. [Figure 7] 1 is a graph showing the results of cell ELISA for CD3ε in CE115. [Figure 8] FIG. 1 shows the molecular form of EGFR_ERY22_CE115. [Figure 9]1 is a graph showing the TDCC results of EGFR_ERY22_CE115 (SK-pca13a). [Figure 10] 1 is a graph showing the binding of humanized CE115 to CD3ε. [Figure 11] 1 is a graph showing the results of ECL-ELISA detecting the binding of RGD-inserted CE115 to integrins. [Figure 12] 1 is a graph showing the results of ECL-ELISA detecting the binding of RGD-inserted CE115 to CD3ε. [Figure 13] 1 is a graph showing the results of ECL-ELSIA for detecting the simultaneous binding of RGD-inserted CE115 to integrin and CD3ε. The results are shown for variants that bind simultaneously. [Figure 14] 1 is a graph showing the results of simultaneous binding ECL-ELISA of RGD-inserted CE115. The results are also shown for variants that do not bind simultaneously. [Figure 15] 1 is a graph showing the results of ECL-ELISA detecting the binding of TLR2-binding peptide-inserted CE115 to TLR2. [Figure 16] 1 is a graph showing the results of ECL-ELISA detecting the binding of TLR2-binding peptide-inserted CE115 to CD3ε. [Figure 17] 1 is a graph showing the results of ECL-ELISA detecting the simultaneous binding of TLR2-binding peptide-inserted CE115 to TLR2 and CD3. [Figure 18] This is an example of a sensorgram for an antibody with a binding ratio of less than 0.8. The vertical axis represents the RU value (response) and the horizontal axis represents time. [Figure 19] FIG. 1 shows the binding of phage-displayed Fab domains to CD3ε and IL6R. [Figure 20] This figure shows the binding of phage-displayed Fab domains to CD3ε and human IgA (hIgA). NC indicates binding to a plate on which no antigen was immobilized. [Figure 21] FIG. 1 shows the binding of IgG clones to CD3ε and human IgA (hIgA). [Figure 22] FIG. 1 shows that the binding of IgG clones to human IgA is inhibited by CD3ε, and the clones cannot simultaneously bind to human IgA (hIgA) and CD3ε. [Figure 23] This figure shows the binding of phage-displayed Fab domains to CD3ε and human CD154. NC indicates binding to a plate on which no antigen was immobilized. [Figure 24] FIG. 1 shows the binding of IgG clones to CD3ε and human CD154. [Figure 25] FIG. 1 shows that the binding of IgG clones to human CD154 is inhibited by CD3ε, and the clones cannot simultaneously bind to human CD154 and CD3ε. DETAILED DESCRIPTION OF THE INVENTION

[0017] In one aspect, the present invention relates to a library mainly composed of a plurality of antigen-binding molecules having different sequences, wherein the antigen-binding regions in the antigen-binding molecules are variable regions of antibodies (hereinafter also referred to as variable regions of the antibodies of the present invention) that can bind to a first antigen and a second antigen different from the first antigen, but do not simultaneously bind to the first antigen and the second antigen. In the antigen-binding regions contained in the library of the present invention, preferably, either the first antigen or the second antigen is CD3, and the other antigen is a molecule expressed on the surface of T cells or other immune cells.

[0018] In another aspect, the present invention relates to an antigen-binding molecule comprising the variable region of an antibody that can bind to three different antigens (a first antigen, a second antigen, and a fourth antigen) but does not simultaneously bind to the three antigens. Here, "three different antigens" in the present invention refers to a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen.

[0019] As used herein, the antigens included in the term "two different antigens" are referred to as "first antigen" and "second antigen," and the antigens included in the term "three different antigens" are referred to as "first antigen," "second antigen," and "fourth antigen."

[0020] Furthermore, the present invention relates to antigen-binding molecules that further comprise a variable region that binds to a "third antigen," which is an antigen different from the "three different antigens."

[0021] In the present invention, the term "antibody variable region" generally refers to a region composed of four framework regions (FR) and three complementarity-determining regions (CDRs) sandwiched therebetween, and also includes partial sequences thereof as long as they have the activity of binding to a part or all of an antigen. Regions comprising an antibody light chain variable region (VL) and an antibody heavy chain variable region (VH) are particularly preferred. The variable regions of the antibodies of the present invention may have any sequence and may be those of antibodies of any origin, such as mouse, rat, rabbit, goat, camel, humanized antibodies obtained by humanizing these non-human antibodies, and human antibodies. A "humanized antibody," also known as a reshaped human antibody, is an antibody derived from a mammal other than a human, for example, by grafting the complementarity-determining region (CDR) of a mouse antibody onto the CDR of a human antibody. Methods for identifying CDRs are known (Kabat et al., Sequence of Proteins of Immunological Interest (1987), National Institutes of Health, Bethesda, Md.; Chothia et al., Nature (1989) 342: 877), as are general genetic recombination techniques (see European Patent Application Publication No. EP 125023 and WO 96 / 02576).

[0022] The phrase "antibody variable region" of the present invention "does not bind to a first antigen and a second antigen simultaneously" means that when the variable region of the antibody of the present invention is bound to the first antigen, it cannot bind to the second antigen, and conversely, when the variable region is bound to the second antigen, it cannot bind to the first antigen. In other words, the phrase "antibody variable region" of the present invention "can bind to a first antigen and a second antigen, but does not bind to the first antigen and the second antigen simultaneously" or "binds to a first antigen and a second antigen, but does not bind to the first antigen and the second antigen simultaneously" means that the variable region of the antibody of the present invention cannot bind to the second antigen when bound to the first antigen, but can bind to the second antigen when not bound to the first antigen, and conversely, cannot bind to the first antigen when bound to the second antigen, but can bind to the first antigen when not bound to the second antigen. Here, "not simultaneously binding to the first and second antigens" also includes not cross-linking two cells, one expressing the first antigen and the other expressing the second antigen, or not simultaneously binding to the first and second antigens expressed on separate cells. Furthermore, it also includes cases where the first and second antigens are not expressed on the cell membrane as soluble proteins, or where both are present on the same cell, and the antibody can simultaneously bind to both the first and second antigens, but cannot simultaneously bind to them when they are expressed on different cells. The variable region of such an antibody is not particularly limited as long as it retains the relevant function, but examples include variable regions in which some amino acids in the variable region of an IgG antibody have been modified to bind to the desired antigen. The amino acids to be modified are, for example, selected from the variable region of an antibody that binds to the first or second antigen, so that the amino acid modification does not abolish binding to the antigen. Here, "expressed on different cells" means that the expression occurs on separate cells, and such a combination of cells may be, for example, cells of the same type, such as a T cell and another T cell, or cells of different types, such as a T cell and an NK cell.

[0023] The phrase "the variable region of an antibody" of the present invention "does not simultaneously bind to three different antigens (the first antigen, the second antigen, and the fourth antigen)" means that the variable region of an antibody of the present invention can simultaneously bind to either (A) only one of the three different antigens (i.e., three types of antigens), or (B) a combination of any two of the three different antigens (i.e., three types of antigens). In other words, the expression "the variable region of an antibody of the present invention does not simultaneously bind to three different antigens (a first antigen, a second antigen, and a fourth antigen)" means that (A) when bound to any one of the three different antigens (three types of antigens), it cannot bind to the remaining two antigens, or (B) when bound to any two of the three different antigens (three types of antigens), it cannot bind to the remaining antigen. Specifically, (A) above means that (i) when the variable region of an antibody of the present invention is bound to the first antigen, it cannot bind to either the second or fourth antigen, (ii) when the variable region is bound to the second antigen, it cannot bind to either the first or fourth antigen, and (iii) when the variable region is bound to the fourth antigen, it cannot bind to either the first or second antigen. Specifically, (B) above means that (i) when the variable region of the antibody of the present invention is bound to the first antigen and the second antigen, the variable region cannot bind to the fourth antigen; (ii) when the variable region of the antibody of the present invention is bound to the first antigen and the fourth antigen, the variable region cannot bind to the second antigen; and (iii) when the variable region of the antibody of the present invention is bound to the second antigen and the fourth antigen, the variable region cannot bind to the first antigen. The variable regions of the antibodies of the present invention do not simultaneously bind to three different antigens, i.e., do not simultaneously bind to the first antigen, the second antigen, and the fourth antigen, and preferably do not simultaneously bind to the first antigen and the second antigen, do not simultaneously bind to the first antigen and the fourth antigen, and / or do not simultaneously bind to the second antigen and the fourth antigen. Most preferably, the variable region of an antibody of the present invention does not simultaneously bind to at least two or more antigens selected from the group consisting of a first antigen, a second antigen, and a fourth antigen (i.e., does not simultaneously bind to a combination of the first antigen and the second antigen, a combination of the first antigen and the fourth antigen, a combination of the second antigen and the fourth antigen, or a combination of the first antigen, the second antigen, and the fourth antigen). Here, "not simultaneously binding to three different antigens (first, second, and fourth antigens)" also includes not crosslinking two or more types of cells selected from the group consisting of cells expressing the first antigen, cells expressing the second antigen, and cells expressing the fourth antigen, or not simultaneously binding to two or more types of antigens selected from the group consisting of the first, second, and fourth antigens expressed on separate cells. Furthermore, it also includes cases where one or more of the three different antigens (first, second, and fourth antigens) are not expressed on the cell membrane as soluble proteins, or where two or more of the three different antigens are present on the same cell and can simultaneously bind to those antigens, but cannot simultaneously bind to those antigens when they are expressed on different cells. The variable region of such an antibody is not particularly limited as long as it has the relevant function, and examples include variable regions in which some amino acids in the variable region of an IgG antibody have been modified so as to bind to the desired antigen.

[0024] The amino acid modifications of the present invention may be used alone or in combination. When multiple compounds are used in combination, the number of compounds to be combined is not particularly limited and can be set appropriately within the range in which the object of the invention can be achieved, for example, 2 or more and 30 or less, preferably 2 or more and 25 or less, 2 or more and 22 or less, 2 or more and 20 or less, 2 or more and 15 or less, 2 or more and 10 or less, 2 or more and 5 or less, or 2 or more and 3 or less. When multiple amino acid modifications are combined, the amino acid modifications may be made only to the heavy chain variable region or light chain variable region of the antibody, or may be appropriately distributed and made to both the heavy chain variable region and the light chain variable region.

[0025] The amino acid residues to be modified (modified amino acids) can be one or more amino acid residues in the variable region, as long as the antigen-binding activity is maintained. When modifying amino acids in the variable region, it is not particularly limited, but it is preferable that the binding activity of the antibody before modification is maintained, for example, 50% or more, preferably 80% or more, and more preferably 100% or more of the binding activity compared to before modification. Furthermore, the binding activity may be increased by the amino acid modification, for example, 2-fold, 5-fold, 10-fold, etc., compared to before modification.

[0026] Preferred regions for amino acid modification include solvent-exposed regions and loop regions in the variable region. Among these, the CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numberings 31-35, 50-65, 71-74, and 95-102 in the H-chain variable region and Kabat numberings 24-34, 50-56, and 89-97 in the L-chain variable region are preferred, with Kabat numberings 31, 52a-61, 71-74, and 97-101 in the H-chain variable region and Kabat numberings 24-34, 51-56, and 89-96 in the L-chain variable region being more preferred. Furthermore, amino acids that increase antigen-binding activity may also be introduced during amino acid modification.

[0027] In the present invention, the term "loop region" refers to a region containing residues that are not involved in maintaining the β-barrel structure of immunoglobulin. In the present invention, an amino acid modification refers to any one of substitution, deletion, addition, insertion, or modification, or a combination thereof. In the present invention, an amino acid modification can be referred to as an amino acid mutation, and the terms are used interchangeably.

[0028] When substituting an amino acid residue, the purpose is to change, for example, the following points (a) to (c) by substituting another amino acid residue: (a) the backbone structure of the polypeptide in the sheet or helical structure region; (b) the charge or hydrophobicity at the target site; or (c) the size of the side chain. Amino acid residues are classified into the following groups based on common side chain properties: (1) hydrophobic: norleucine, met, ala, val, leu, ile; (2) neutral hydrophilic: cys, ser, thr, asn, gln; (3) acidic: asp, glu; (4) basic: his, lys, arg; (5) residues that affect chain orientation: gly, pro; and (6) aromatic: trp, tyr, phe.

[0029] Substitutions of amino acid residues within each of these groups are called conservative substitutions, while substitutions of amino acid residues between other groups are called non-conservative substitutions. The substitutions in the present invention may be conservative substitutions, non-conservative substitutions, or a combination of conservative and non-conservative substitutions.

[0030] Modifications to amino acid residues also include selecting a variable region capable of binding to the first and second antigens but incapable of simultaneously binding to these antigens from among those in which amino acids have been randomly modified within the variable region of an antibody that binds to the first or second antigen, so that the amino acid modification does not result in a loss of antigen-binding ability, or inserting a peptide known to have binding activity to the desired antigen into the aforementioned region. Modifications also include selecting a variable region capable of binding to the first, second, and fourth antigens but incapable of simultaneously binding to these antigens from among those in which amino acids have been randomly modified within the variable region of an antibody that binds to the first, second, or fourth antigen, so that the amino acid modification does not result in a loss of antigen-binding ability, or inserting a peptide known to have binding activity to the desired antigen into the aforementioned region. Examples of peptides known to have binding activity to the desired antigen include the peptides listed in Table 1.

[0031] [Table 1]

[0032] One embodiment of the present invention provides an antigen-binding molecule comprising an antibody variable region in which amino acids in the heavy chain variable region have been modified so that the antibody can bind to a first antigen and a second antigen different from the first antigen, but does not simultaneously bind to the first and second antigens. Another embodiment of the present invention provides an antigen-binding molecule comprising an antibody variable region in which amino acids in the heavy chain variable region have been modified so that the antibody can bind to three different antigens (a first antigen, a second antigen, and a fourth antigen), but does not simultaneously bind to all three antigens. For example, by introducing the above-mentioned amino acid modifications (substitution, deletion, addition, insertion, or modification, or a combination thereof) into the heavy chain variable region, it is possible to create an antibody variable region that can bind to a first antigen and a second antigen different from the first antigen, but does not simultaneously bind to the first and second antigens, or an antibody variable region that can bind to three different antigens (a first antigen, a second antigen, and a fourth antigen), but does not simultaneously bind to all three antigens. The heavy chain variable region is preferred as the site for introducing amino acid alterations, with more preferred regions including solvent-exposed regions and loop regions in the variable region. Of these, CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numbering 31-35, 50-65, 71-74, and 95-102 in the H chain variable region are preferred, with Kabat numbering 31, 52a-61, 71-74, and 97-101 in the H chain variable region being more preferred. Furthermore, amino acids that increase antigen-binding activity may also be introduced during amino acid alterations.

[0033] In addition to the modifications described above, the variable regions of the antibodies of the present invention may also contain known modifications. For example, modification of the N-terminal glutamine of a variable region to pyroglutamic acid by pyroglutamylation is a modification well known to those skilled in the art. Thus, when the N-terminus of the heavy chain of the antibody of the present invention is glutamine, the antibody comprises a variable region in which the glutamine has been modified to pyroglutamic acid.

[0034] Furthermore, for example, amino acid modifications may be introduced into the variable regions of these antibodies to improve antigen binding, pharmacokinetics, stability, or antigenicity. The variable regions of the antibodies of the present invention may be modified to have pH-dependent binding to the antigen, thereby enabling them to repeatedly bind to the antigen (WO / 2009 / 125825).

[0035] Furthermore, for example, amino acid modifications can be made to the variable regions of these antibodies that bind to a third antigen, such that the binding activity to the antigen changes depending on the concentration of a target tissue-specific compound (WO2013 / 180200).

[0036] Furthermore, for example, modifications of the variable region can be made to increase binding activity, improve specificity, lower pI, impart pH-dependent properties to antigen binding, improve thermal binding stability, improve solubility, stability against chemical modification, improve glycan-derived heterogeneity, avoid T cell epitopes identified using in silico prediction to reduce immunogenicity or identified by in vitro T cell assays, or introduce T cell epitopes that activate regulatory T cells (mAbs 3:243-247, 2011).

[0037] Whether the variable regions of the antibodies of the present invention are "capable of binding to the first antigen and the second antigen" can be determined using known methods. For example, it can be measured by electrochemiluminescence (ECL) (BMC Research Notes 2011, 4:281). Specifically, for example, a region of a biotin-labeled test antigen-binding molecule capable of binding to the first and second antigens, such as a minibody consisting of the Fab region, or a monovalent antibody (an antibody lacking one of the two Fab regions found in conventional antibodies), is mixed with the first or second antigen labeled with a sulfo-tag (Ru complex), and the mixture is added to a streptavidin-immobilized plate. At this time, the biotin-labeled test antigen-binding molecule binds to the streptavidin on the plate. The sulfo-tag is made to emit light, and the luminescence signal is detected using a Sector Imager 600, 2400 (MSD), or the like, to confirm binding of the first or second antigen to the aforementioned region of the test antigen-binding molecule. Measurement can also be performed by ELISA, FACS (fluorescence activated cell sorting), ALPHA screen (Amplified Luminescent Proximity Homogeneous Assay), or the BIACORE method using the surface plasmon resonance (SPR) phenomenon (Proc. Natl. Acad. Sci. USA (2006) 103 (11), 4005-4010).

[0038] Specifically, measurements can be performed using, for example, Biacore (GE Healthcare), an interaction analysis device that utilizes the surface plasmon resonance (SPR) phenomenon. Biacore includes models such as Biacore T100, T200, X100, A100, 4000, 3000, 2000, 1000, and C. Any Biacore sensor chip, such as a CM7, CM5, CM4, CM3, C1, SA, NTA, L1, HPA, or Au chip, can be used. A capture protein, such as Protein A, Protein G, Protein L, anti-human IgG antibody, anti-human IgG-Fab, anti-human L chain antibody, anti-human Fc antibody, antigen protein, or antigen peptide, that captures the antigen-binding molecule of the present invention is immobilized on the sensor chip using a coupling method such as amine coupling, disulfide coupling, or aldehyde coupling. A first or second antigen is then injected as an analyte, the interaction is measured, and a sensorgram is obtained. In this case, the concentration of the first antigen or the second antigen can be in the range of several μM to several pM depending on the strength of the interaction such as KD of the sample to be measured.

[0039] Alternatively, instead of the antigen-binding molecule, the first or second antigen can be immobilized on a sensor chip and then allowed to interact with the antibody sample to be evaluated. Whether the antibody variable region of the antigen-binding molecule of the present invention has binding activity for the first or second antigen can be determined from the dissociation constant (KD) value calculated from the sensorgram of the interaction or the degree of increase in the sensorgram before and after exposure to the antigen-binding molecule sample.

[0040] The ALPHA screen is performed using ALPHA technology, which uses two beads, donor and acceptor, based on the following principle: A luminescent signal is detected only when a molecule bound to the donor bead biologically interacts with a molecule bound to the acceptor bead and the two beads are in close proximity. A photosensitizer inside the donor bead, excited by a laser, converts surrounding oxygen into excited singlet oxygen. The singlet oxygen diffuses around the donor bead and, when it reaches a nearby acceptor bead, triggers a chemiluminescent reaction within the bead, ultimately emitting light. If the molecules bound to the donor bead and the molecules bound to the acceptor bead do not interact, the singlet oxygen produced by the donor bead does not reach the acceptor bead, and no chemiluminescent reaction occurs.

[0041] One of the substances to be observed for interaction (ligand) is immobilized on a thin gold film on a sensor chip. When light is shone from the back of the sensor chip so that it is totally reflected at the interface between the gold film and the glass, a portion of the reflected light exhibits a reduced reflection intensity (SPR signal). When the other substance to be observed for interaction (analyte) is poured over the surface of the sensor chip, binding occurs between the ligand and the analyte, increasing the mass of the immobilized ligand molecules and changing the refractive index of the solvent on the sensor chip surface. This change in refractive index shifts the position of the SPR signal (conversely, dissociation returns the signal position). The Biacore system plots this shift, i.e., the change in mass on the sensor chip surface, on the vertical axis, and displays the change in mass over time as measurement data (sensorgram). From the sensorgram, the amount of analyte bound to the ligand captured on the sensor chip surface (the change in response on the sensorgram before and after analyte interaction) can be determined. However, because the amount of binding also depends on the amount of ligand, comparisons must be made under conditions where the amount of ligand is essentially the same. Furthermore, the kinetics: association rate constant (ka) and dissociation rate constant (kd) can be determined from the sensorgram curve, and the affinity (KD) can be determined from the ratio of these constants. Inhibition assays are also suitable for use in the BIACORE method. Examples of inhibition assays are described in Proc. Natl. Acad. Sci. USA (2006) 103 (11), 4005-4010.

[0042] Whether the variable regions of the antibodies of the present invention are capable of binding to three different antigens (the first antigen, the second antigen, and the fourth antigen) can be appropriately determined by those skilled in the art using the known methods described above, for example, electrochemiluminescence (ECL) method.

[0043] Whether an antigen-binding molecule of the present invention "does not simultaneously bind to a first antigen and a second antigen" can be confirmed by first confirming that it has binding activity to the first antigen and the second antigen, then allowing either the first antigen or the second antigen to bind to an antigen-binding molecule comprising a variable region having the binding activity, and then measuring whether it has binding activity to the remaining antigen using the method described above. Alternatively, this can be confirmed by measuring whether binding of an antigen-binding molecule to either a first antigen or a second antigen immobilized on an ELISA plate or a sensor chip is inhibited by adding the other antigen to the solution.

[0044] Specifically, for example, when using the ECL method, a biotin-labeled test antigen-binding molecule is combined with a first antigen labeled with a sulfo-tag (Ru complex) and an unlabeled second antigen. If the test antigen-binding molecule can bind to both the first and second antigens but does not simultaneously bind to both the first and second antigens, a mixture of the test antigen-binding molecule and the first antigen is added to a streptavidin-coated plate and the sulfo-tag is allowed to emit light. In the absence of the unlabeled second antigen, the luminescence signal is detected, but in the presence of the second antigen, the luminescence signal decreases. The relative binding activity can be determined by quantifying this decrease in the emitted signal.

[0045] In the case of an ALPHA screen, in the absence of a competing second antigen, the test antigen-binding molecule interacts with the first antigen, generating a signal at 520-620 nm. An untagged second antigen competes with the interaction between the test antigen-binding molecule and the first antigen. Relative binding activity can be determined by quantifying the decrease in fluorescence resulting from competition. Biotinylation of polypeptides using sulfo-NHS-biotin or similar is known. GST-tagging of the first antigen can be achieved by expressing a fusion gene in-frame fusing a polynucleotide encoding the first antigen with a polynucleotide encoding GST in cells harboring an expression vector, followed by purification using a glutathione column. The resulting signal is suitably analyzed by fitting it to a one-site competition model using nonlinear regression analysis using software such as GRAPHPAD PRISM (GraphPad, San Diego). In this case, the same analysis is possible even if the second antigen is tagged and the first antigen is not tagged. Another method is fluorescence resonance energy transfer (FRET). FRET is the direct transfer of excitation energy between two adjacent fluorescent molecules due to electronic resonance. When FRET occurs, the excitation energy of the donor (a fluorescent molecule in an excited state) is transferred to the acceptor (another fluorescent molecule in close proximity to the donor). This causes the fluorescence emitted from the donor to disappear (or, more precisely, shorten its fluorescence lifetime), and instead, fluorescence is emitted from the acceptor. This phenomenon can be used to analyze whether an antibody is a dual-Fab. For example, when a first antigen fused with a fluorescent donor and a second antigen fused with a fluorescent acceptor simultaneously bind to a test antigen-binding molecule, the donor's fluorescence disappears and the acceptor emits fluorescence, resulting in a change in the fluorescence wavelength. Such an antibody is not considered to be a dual-Fab. On the other hand, if the fluorescence wavelength of the fluorescent donor bound to the first antigen does not change when the first antigen, second antigen, and test antigen-binding molecule are mixed, the test antigen-binding molecule can be said to be dual-Fab.

[0046] Alternatively, for example, a biotin-labeled test antigen-binding molecule is bound to streptavidin on donor beads, and a first antigen tagged with glutathione S-transferase (GST) is bound to acceptor beads. In the absence of a competing second antigen, the test antigen-binding molecule and the first antigen interact to generate a signal at 520-620 nm. An untagged second antigen competes with the interaction between the test antigen-binding molecule and the first antigen. Relative binding activity can be determined by quantifying the decrease in fluorescence that occurs as a result of competition. Biotinylation of polypeptides using sulfo-NHS-biotin or the like is known. GST-tagging of the first antigen can be achieved by expressing a fusion gene in which a polynucleotide encoding the first antigen and a polynucleotide encoding GST are fused in frame in cells harboring an expression vector, followed by purification using a glutathione column. The resulting signals are suitably analyzed by fitting to a one-site competition model using non-linear regression analysis using software such as GRAPHPAD PRISM (GraphPad, San Diego).

[0047] The tagging is not limited to GST, and any tag such as histidine tag, MBP, CBP, Flag tag, HA tag, V5 tag, or c-myc tag may be used. Furthermore, the binding of the test antigen-binding molecule to the donor beads is not limited to binding using the biotin-streptavidin reaction. In particular, when the test antigen-binding molecule contains Fc, a possible method is to bind the test antigen-binding molecule via an Fc-recognizing protein such as Protein A or Protein G on the donor beads.

[0048] Furthermore, when the first antigen and the second antigen are not expressed on the cell membrane as soluble proteins, or when both are present on the same cell, a protein can simultaneously bind to both the first antigen and the second antigen, but cannot simultaneously bind to them when they are expressed on different cells, measurement can also be performed using known methods. Specifically, even if the result is positive in ECL-ELISA, which detects simultaneous binding to a first antigen and a second antigen, if cells expressing the first antigen, cells expressing the second antigen, and a test antigen-binding molecule are mixed together and these three antigens do not bind simultaneously, this indicates that simultaneous binding is not possible when the antigens are expressed on different cells. For example, measurement can be performed using ECL-ELISA using cells. Cells expressing the first antigen are first immobilized on a plate, and the test antigen-binding molecule is allowed to bind to the cells, after which cells expressing the second antigen are added. By detecting using a sulfo-tagged antibody against a different antigen expressed only in cells expressing the second antigen, a signal is observed if the two antigens expressed on the two cells simultaneously bind, but no signal is observed if they do not simultaneously bind. Alternatively, measurement can be performed using the alphascreen method. When cells expressing a first antigen bound to donor beads, cells expressing a second antigen bound to acceptor beads, and a test antigen-binding molecule are mixed, a signal is observed if the two antigens expressed on the two cells bind simultaneously, but no signal is observed if they do not bind simultaneously. Alternatively, measurement is possible using interaction analysis using Octet. First, cells expressing a peptide-tagged first antigen are bound to a biosensor that recognizes the peptide tag. When interaction analysis is performed in a well containing cells expressing a second antigen and a test antigen-binding molecule, if the two antigens expressed on the two cells bind simultaneously, a large wavelength shift is observed due to the binding of the test antigen-binding molecule and the cells expressing the second antigen to the biosensor. If they do not bind simultaneously, only the test antigen-binding molecule binds to the biosensor, resulting in a small wavelength shift.

[0049] Alternatively, measurement based on biological activity rather than binding activity is also possible. For example, when cells expressing a first antigen and cells expressing a second antigen are co-cultured with a test antigen-binding molecule, if two antigens expressed on the two cells simultaneously bind to each other, they are mutually activated via the test antigen-binding molecule, and changes in activation signals such as increased downstream phosphorylation of each antigen can be detected. Alternatively, since cytokine production is induced as a result of activation, simultaneous binding to two cells can be determined by measuring the amount of cytokine production.

[0050] Whether an antigen-binding molecule of the present invention "does not simultaneously bind to three different antigens" can be determined by a person skilled in the art, for example, by confirming that the molecule has binding activity to three different antigens (the first antigen, the second antigen, and the fourth antigen) and then appropriately measuring this using the known method described above, such as electrochemiluminescence (ECL) or the like.

[0051] In the present invention, the term "Fc region" refers to a region of an antibody molecule that includes a hinge region or a portion thereof, and a fragment consisting of the CH2 and CH3 domains. The Fc region of an IgG class refers to, for example, the region from cysteine at position 226 to the C-terminus, or from proline at position 230 to the C-terminus, according to the EU numbering system (also referred to herein as the EU INDEX), but is not limited thereto. The Fc region can be suitably obtained by partially digesting an IgG1, IgG2, IgG3, or IgG4 monoclonal antibody with a protease such as pepsin, followed by re-elution of the fraction adsorbed to a Protein A or Protein G column. The protease is not particularly limited, as long as it can digest a full-length antibody to produce Fab and F(ab')2 fragments by appropriately setting the enzyme reaction conditions, such as pH. Examples of such proteases include pepsin and papain.

[0052] The "antigen-binding molecule" of the present invention is not particularly limited as long as it comprises an "antibody variable region" of the present invention, and may further comprise a peptide or protein having a length of about five amino acids or more. It is not limited to peptides and proteins derived from living organisms, and may be, for example, a polypeptide consisting of an artificially designed sequence. It may also be any of natural polypeptides, synthetic polypeptides, recombinant polypeptides, etc.

[0053] Preferred examples of antigen-binding molecules of the present invention include antigen-binding molecules comprising the Fc region of an antibody.

[0054] The "Fc region" of the present invention can be, for example, an Fc region derived from native IgG. Here, native IgG refers to a polypeptide that includes the same amino acid sequence as that of naturally occurring IgG and belongs to the class of antibodies substantially encoded by immunoglobulin gamma genes. For example, native human IgG refers to native human IgG1, native human IgG2, native human IgG3, native human IgG4, etc. Native IgG also includes naturally occurring mutants thereof. For the constant regions of human IgG1, human IgG2, human IgG3, and human IgG4 antibodies, multiple allotype sequences due to genetic polymorphisms are described in "Sequences of proteins of immunological interest," NIH Publication No. 91-3242, and any of these may be used in the present invention. In particular, for the human IgG1 sequence, the amino acid sequence at positions 356-358 (EU numbering) may be either DEL or EEM.

[0055] Examples of antibody Fc regions include IgA1, IgA2, IgD, IgE, IgG1, IgG2, IgG3, IgG4, and IgM types. The Fc region of the antibody of the present invention can be, for example, an Fc region derived from a natural human IgG antibody. The Fc region of the present invention can be, for example, an Fc region derived from a constant region of natural IgG, specifically, a constant region derived from natural human IgG1 (SEQ ID NO: 1), a constant region derived from natural human IgG2 (SEQ ID NO: 2), a constant region derived from natural human IgG3 (SEQ ID NO: 3), or a constant region derived from natural human IgG4 (SEQ ID NO: 4). The constant region of natural IgG also includes naturally occurring mutants thereof.

[0056] As the Fc region of the present invention, an Fc region with reduced binding activity to an Fcγ receptor is particularly preferred. Herein, an Fcγ receptor (sometimes referred to as an Fcγ receptor, FcγR, or Fcγ receptor) refers to a receptor that can bind to the Fc region of IgG1, IgG2, IgG3, or IgG4, and essentially refers to any member of the family of proteins encoded by the Fcγ receptor gene. In humans, this family includes, but is not limited to, FcγRI (CD64), which includes the isoforms FcγRIa, FcγRIb, and FcγRIc; FcγRII (CD32), which includes the isoforms FcγRIIa (including allotypes H131 (H type) and R131 (R type)), FcγRIIb (including FcγRIIb-1 and FcγRIIb-2), and FcγRIIc; and FcγRIII (CD16), which includes the isoforms FcγRIIIa (including allotypes V158 and F158) and FcγRIIIb (including allotypes FcγRIIIb-NA1 and FcγRIIIb-NA2), as well as any unidentified human FcγRs or FcγR isoforms or allotypes. FcγRs may be derived from any organism, including, but not limited to, humans, mice, rats, rabbits, and monkeys. Mouse FcγRs include, but are not limited to, FcγRI (CD64), FcγRII (CD32), FcγRIII (CD16), and FcγRIII-2 (CD16-2), as well as any unidentified mouse FcγRs or FcγR isoforms or allotypes. Preferred examples of such Fcγ receptors include human FcγRI (CD64), FcγRIIa (CD32), FcγRIIb (CD32), FcγRIIIa (CD16), and / or FcγRIIIb (CD16).

[0057] FcγRs are classified into activating receptors with ITAMs (immunoreceptor tyrosine-based activation motifs) and inhibitory receptors with ITIMs (immunoreceptor tyrosine-based inhibitory motifs). FcγRs are classified into activating FcγRs (FcγRI, FcγRIIa R, FcγRIIa H, FcγRIIIa, and FcγRIIIb) and inhibitory FcγRs (FcγRIIb). The polynucleotide and amino acid sequences of FcγRI are listed in NM_000566.3 and NP_000557.1, respectively; the polynucleotide and amino acid sequences of FcγRIIa are listed in BC020823.1 and AAH20823.1, respectively; the polynucleotide and amino acid sequences of FcγRIIb are listed in BC146678.1 and AAI46679.1, respectively; the polynucleotide and amino acid sequences of FcγRIIIa are listed in BC033678.1 and AAH33678.1, respectively; and the polynucleotide and amino acid sequences of FcγRIIIb are listed in BC128562.1 and AAI28563.1, respectively (RefSeq accession numbers). FcγRIIa has two genetic polymorphisms in which the 131st amino acid of FcγRIIa is substituted with histidine (H type) or arginine (R type) (J. Exp. Med., 172, 19-25, 1990). FcγRIIb has two genetic polymorphisms in which the 232nd amino acid of FcγRIIb is substituted with isoleucine (I type) or threonine (T type) (Arthritis. Rheum., 46: 1242-1254 (2002)). FcγRIIIa has two genetic polymorphisms in which the 158th amino acid of FcγRIIIa is substituted with valine (V type) or phenylalanine (F type) (J. Clin. Invest., 100(5): 1059-1070 (1997)). Furthermore, there are two types of genetic polymorphisms in FcγRIIIb: NA1 type and NA2 type (J. Clin. Invest. 85: 1287-1295 (1990)).

[0058] Whether or not the binding activity to Fcγ receptors is reduced can be confirmed by well-known methods such as FACS, ELISA format, ALPHA screen (Amplified Luminescent Proximity Homogeneous Assay), and the BIACORE method using the surface plasmon resonance (SPR) phenomenon (Proc. Natl. Acad. Sci. USA (2006) 103 (11), 4005-4010). The ALPHA screen is performed using ALPHA technology, which uses two beads, donor and acceptor, based on the following principle: A luminescent signal is detected only when a molecule bound to the donor bead biologically interacts with a molecule bound to the acceptor bead and the two beads are in close proximity. A photosensitizer inside the donor bead, excited by a laser, converts surrounding oxygen into excited singlet oxygen. The singlet oxygen diffuses around the donor bead and, when it reaches a nearby acceptor bead, triggers a chemiluminescent reaction within the bead, ultimately emitting light. If the molecules bound to the donor bead and the molecules bound to the acceptor bead do not interact, the singlet oxygen produced by the donor bead does not reach the acceptor bead, and no chemiluminescent reaction occurs.

[0059] For example, biotin-labeled antigen-binding molecules are bound to donor beads, and glutathione S-transferase (GST)-tagged Fcγ receptors are bound to acceptor beads. In the absence of competing antigen-binding molecules having mutant Fc regions, antigen-binding molecules having wild-type Fc regions interact with Fcγ receptors, generating a signal at 520-620 nm. Antigen-binding molecules having untagged mutant Fc regions compete with the interaction between antigen-binding molecules having wild-type Fc regions and Fcγ receptors. Relative binding affinity can be determined by quantifying the decrease in fluorescence that occurs as a result of competition. Biotinylation of antigen-binding molecules such as antibodies using sulfo-NHS-biotin or similar is known. Methods for tagging Fcγ receptors with GST include expressing a fusion gene in which a polynucleotide encoding the Fcγ receptor and a polynucleotide encoding GST are fused in frame in cells harboring an expression vector, followed by purification using a glutathione column. The resulting signals are suitably analyzed by fitting to a one-site competition model using non-linear regression analysis using software such as GRAPHPAD PRISM (GraphPad, San Diego).

[0060] One of the substances (ligand) whose interaction is to be observed is immobilized on a thin gold film on a sensor chip. When light is shone from the back of the sensor chip so that it is totally reflected at the interface between the gold film and the glass, a portion of the reflected light exhibits a reduced reflection intensity (SPR signal). When the other substance (analyte) whose interaction is to be observed is passed over the surface of the sensor chip, binding occurs between the ligand and the analyte, increasing the mass of the immobilized ligand molecule and changing the refractive index of the solvent on the sensor chip surface. This change in refractive index shifts the position of the SPR signal (conversely, dissociation returns the signal position). The Biacore system plots the amount of shift (i.e., the change in mass on the sensor chip surface) on the vertical axis, and displays the change in mass over time as measurement data (sensorgram). The kinetics (association rate constant (ka) and dissociation rate constant (kd)) can be calculated from the sensorgram curve, and affinity (KD) can be calculated from the ratio of these constants. Inhibition assays are also suitable for use with the BIACORE method. An example of an inhibition assay is described in Proc. Natl. Acad. Sci. USA (2006) 103 (11), 4005-4010.

[0061] As used herein, "decreased binding activity to an Fcγ receptor" means, for example, that the binding activity of a test antigen-binding molecule is 50% or less, preferably 45% or less, 40% or less, 35% or less, 30% or less, 20% or less, or 15% or less, and particularly preferably 10% or less, 9% or less, 8% or less, 7% or less, 6% or less, 5% or less, 4% or less, 3% or less, 2% or less, or 1% or less, compared to the binding activity of a control antigen-binding molecule comprising an Fc region, based on the above-mentioned analytical methods. Antigen-binding molecules having the Fc region of an IgG1, IgG2, IgG3, or IgG4 monoclonal antibody can be used as a control. The structures of the Fc regions are set forth in SEQ ID NO: 1 (RefSeq Accession No. AAC82527.1 with an A added to the N-terminus), SEQ ID NO: 2 (RefSeq Accession No. AAB59393.1 with an A added to the N-terminus), SEQ ID NO: 3 (RefSeq Accession No. CAA27268.1 with an A added to the N-terminus), and SEQ ID NO: 4 (RefSeq Accession No. AAB59394.1 with an A added to the N-terminus). Furthermore, when an antigen-binding molecule having a mutant Fc region of an antibody of a certain isotype is used as a test substance, the effect of the mutation in the mutant on Fcγ receptor-binding activity can be verified by using an antigen-binding molecule having an Fc region of the antibody of the certain isotype as a control. As described above, antigen-binding molecules having mutant Fc regions verified to have reduced Fcγ receptor-binding activity can be appropriately prepared.

[0062] Known examples of such mutants include a deletion of amino acids 231A-238S, as identified according to EU numbering (WO 2009 / 011941), C226S, C229S, P238S, (C220S) (J. Rheumatol (2007) 34, 11), C226S, C229S (Hum. Antibod. Hybridomas (1990) 1(1), 47-54), and C226S, C229S, E233P, L234V, and L235A (Blood (2007) 109, 1185-1192). Preferred examples of such an antigen-binding molecule include an Fc region in which any of the following amino acids, as specified by EU numbering, have been substituted among the amino acids constituting the Fc region of an antibody of a particular isotype: 220, 226, 229, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 264, 265, 266, 267, 269, 270, 295, 296, 297, 298, 299, 300, 325, 327, 328, 329, 330, 331, or 332. The antibody isotype from which the Fc region is derived is not particularly limited, and Fc regions derived from IgG1, IgG2, IgG3, or IgG4 monoclonal antibodies can be used as appropriate, with an Fc region derived from a naturally occurring human IgG1 antibody being preferred. For example, any of the following substitutions, specified according to EU numbering, among the amino acids constituting the Fc region of an IgG1 antibody (where the number indicates the position of the amino acid residue specified according to EU numbering, the single-letter amino acid code preceding the number indicates the amino acid residue before substitution, and the single-letter amino acid code following the number indicates the amino acid residue before substitution): (a) L234F, L235E, P331S, (b) C226S, C229S, P238S, (c) C226S, C229S, (d)C226S, C229S, E233P, L234V, L235A Alternatively, antigen-binding molecules having an Fc region in which the amino acid sequence at positions 231 to 238 has been deleted may also be used as appropriate.

[0063] Furthermore, any of the following substitutions, specified according to EU numbering, among the amino acids constituting the Fc region of an IgG2 antibody (the numbers indicate the amino acid residue positions specified according to EU numbering, the single-letter amino acid code preceding the number indicates the amino acid residue before substitution, and the single-letter amino acid code following the number indicates the amino acid residue before substitution): (e)H268Q, V309L, A330S, P331S (f)V234A (g)G237A (h) V234A, G237A (i) A235E, G237A (j) V234A, A235E, G237A Antigen-binding molecules having an Fc region modified with α-glucan may also be used appropriately.

[0064] Furthermore, any of the following substitutions, specified according to EU numbering, among the amino acids constituting the Fc region of an IgG3 antibody (where the number indicates the position of the amino acid residue specified according to EU numbering, the single-letter amino acid code preceding the number indicates the amino acid residue before substitution, and the single-letter amino acid code following the number indicates the amino acid residue before substitution): (k)F241A (l) D265A (m)V264A Antigen-binding molecules having an Fc region modified with α-glucan may also be used appropriately.

[0065] Furthermore, any of the following substitutions, specified according to EU numbering, among the amino acids constituting the Fc region of an IgG4 antibody (the numbers indicate the amino acid residue positions specified according to EU numbering, the single-letter amino acid code preceding the number indicates the amino acid residue before substitution, and the single-letter amino acid code following the number indicates the amino acid residue before substitution): (n) L235A, G237A, E318A (o)L235E (p)F234A, L235A Antigen-binding molecules having an Fc region modified with α-glucan may also be used appropriately.

[0066] Other preferred examples include antigen-binding molecules having an Fc region in which any of the amino acids at positions 233, 234, 235, 236, 237, 327, 330, and 331 specified according to EU numbering among the amino acids constituting the Fc region of a native human IgG1 antibody has been substituted with an amino acid corresponding to the corresponding EU numbering in the corresponding IgG2 or IgG4.

[0067] Other preferred examples include antigen-binding molecules having an Fc region in which one or more of the following amino acids, identified according to EU numbering among the amino acids constituting the Fc region of a native human IgG1 antibody, have been substituted with other amino acids: 234, 235, and 297. The type of amino acid present after substitution is not particularly limited, but particularly preferred are antigen-binding molecules having an Fc region in which one or more of the amino acids at positions 234, 235, and 297 have been substituted with alanine.

[0068] Other preferred examples include antigen-binding molecules having an Fc region in which any of the following amino acids, identified according to EU numbering, at position 265 of the amino acids constituting the Fc region of an IgG1 antibody, has been substituted with another amino acid. The type of amino acid present after substitution is not particularly limited, but particularly preferred are antigen-binding molecules having an Fc region in which the amino acid at position 265 has been substituted with alanine.

[0069] Furthermore, one preferred embodiment of the "antigen-binding molecule" of the present invention is a multispecific antibody comprising the variable region of the antibody of the present invention.

[0070] For the aggregation of multispecific antibodies, a technique can be applied in which undesired association of H chains is suppressed by introducing electric charge repulsion at the interface of the second constant region (CH2) or third constant region (CH3) of the antibody H chain (WO2006 / 106905). In techniques that suppress unintended association between H chains by introducing electric charge repulsion at the CH2 or CH3 interface, examples of amino acid residues that contact the interface of other H chain constant regions include the regions corresponding to residues 356 (EU numbering), 439 (EU numbering), 357 (EU numbering), 370 (EU numbering), 399 (EU numbering), and 409 (EU numbering) in the CH3 region.

[0071] More specifically, for example, an antibody comprising two types of H chain CH3 regions can be an antibody in which one to three pairs of amino acid residues selected from the pairs of amino acid residues shown in (1) to (3) below in the first H chain CH3 region have the same charge: (1) amino acid residues contained in the H chain CH3 region, which are amino acid residues at positions 356 and 439 (EU numbering), (2) amino acid residues contained in the H chain CH3 region, which are amino acid residues at positions 357 and 370 (EU numbering), and (3) amino acid residues contained in the H chain CH3 region, which are amino acid residues at positions 399 and 409 (EU numbering).

[0072] Furthermore, the antibody can be one in which a second H chain CH3 region different from the first H chain CH3 region has a set of amino acid residues selected from the sets of amino acid residues shown in (1) to (3), and one to three sets of amino acid residues corresponding to the sets of amino acid residues shown in (1) to (3) that have the same charge in the first H chain CH3 region have an opposite charge to the corresponding amino acid residues in the first H chain CH3 region.

[0073] The amino acid residues described in (1) to (3) above are close to each other when associated. Those skilled in the art can identify the positions corresponding to the amino acid residues described in (1) to (3) above for a desired H chain CH3 region or H chain constant region by homology modeling using commercially available software, and can appropriately modify the amino acid residues at those positions.

[0074] In the above-described antibody, the "charged amino acid residue" is preferably selected from amino acid residues included in either group (a) or (b) below: (a) glutamic acid (E), aspartic acid (D), (b) Lysine (K), arginine (R), and histidine (H).

[0075] In the above-mentioned antibodies, "having the same charge" means, for example, that two or more amino acid residues all have an amino acid residue included in one of the above groups (a) or (b). "Having opposite charges" means, for example, that when at least one amino acid residue among two or more amino acid residues has an amino acid residue included in one of the above groups (a) or (b), the remaining amino acid residues have an amino acid residue included in a different group.

[0076] In a preferred embodiment, the above-mentioned antibody may have the first H chain CH3 region and the second H chain CH3 region cross-linked by a disulfide bond. The amino acid residues to be modified in the present invention are not limited to those in the antibody variable region or constant region described above. Those skilled in the art can identify amino acid residues that form an interface between polypeptide mutants or heteromultimers by homology modeling using commercially available software, and can modify the amino acid residues at those sites to control association.

[0077] Other known techniques can also be used to aggregate multispecific antibodies of the present invention. Substituting an amino acid side chain in the variable region of one antibody H chain with a larger side chain (knob) and substituting an amino acid side chain in the opposing variable region of the other H chain with a smaller side chain (hole) allows the knob to be positioned in the hole, thereby efficiently facilitating aggregation of polypeptides with different amino acids that have Fc domains (WO 1996 / 027011, Ridgway JB et al., Protein Engineering (1996) 9, 617-621, Merchant AM et al., Nature Biotechnology (1998) 16, 677-681).

[0078] In addition, other known techniques can also be used to form multispecific antibodies of the present invention. By using a strand-exchange engineered domain CH3, in which a portion of the CH3 of one antibody H chain is modified with a corresponding IgA-derived sequence and the complementary portion of the CH3 of the other H chain is modified with a corresponding IgA-derived sequence, association of polypeptides with different sequences can be efficiently induced by complementary association of CH3s (Protein Engineering Design & Selection, 23; 195-202, 2010). This known technique can also be used to efficiently form desired multispecific antibodies.

[0079] Other techniques that can be used to form multispecific antibodies include antibody production techniques that utilize the association of antibody CH1 and CL, or the association of VH and VL, as described in WO2011 / 028952; techniques for producing bispecific antibodies using separately prepared monoclonal antibodies (Fab Arm Exchange), as described in WO2008 / 119353 and WO2011 / 131746; techniques for controlling the association between CH3s of antibody heavy chains, as described in WO2012 / 058768 and WO2013 / 063702; techniques for producing bispecific antibodies composed of two types of light chains and one type of heavy chain, as described in WO2012 / 023053; and a technique for producing bispecific antibodies using two bacterial cell lines that each express one half of an antibody chain composed of one H chain and one L chain, as described in Christoph et al. (Nature Biotechnology Vol. 31, pp. 753-758 (2013)). In addition to the above-mentioned assembly techniques, the CrossMab technique (Scaefer et al. (Proc. Natl. Acad. Sci. USA (2011) 108, 11187-11192)), known as an assembly technique for heterologous light chains, in which a light chain forming a variable region binding to a first epitope and a light chain forming a variable region binding to a second epitope are assembled with a heavy chain forming a variable region binding to the first epitope and a heavy chain forming a variable region binding to the second epitope, respectively, can also be used to produce the multispecific or multiparatopic antigen-binding molecules provided by the present invention. An example of a technique for producing bispecific antibodies using separately prepared monoclonal antibodies is a method in which a monoclonal antibody in which specific amino acids present in the heavy chain CH3 region have been substituted is placed under reducing conditions to promote antibody heterodimerization, thereby obtaining the desired bispecific antibody. Preferred amino acid substitution sites in this method include, for example, residues 392 and 397 (EU numbering) in the CH3 region.Furthermore, bispecific antibodies can also be produced using antibodies in which one to three pairs of amino acid residues in the first H chain CH3 region selected from the pairs of amino acid residues shown in (1) to (3) below have the same electric charge: (1) amino acid residues in the H chain CH3 region, which are at positions 356 and 439 (EU numbering), (2) amino acid residues in the H chain CH3 region, which are at positions 357 and 370 (EU numbering), and (3) amino acid residues in the H chain CH3 region, which are at positions 399 and 409 (EU numbering). Furthermore, bispecific antibodies can also be produced using antibodies containing pairs of amino acid residues selected from the pairs of amino acid residues shown in (1) to (3) in a second H chain CH3 region different from the first H chain CH3 region, wherein one to three pairs of amino acid residues corresponding to pairs of amino acid residues shown in (1) to (3) with the same electric charge in the first H chain CH3 region have the opposite electric charge to the corresponding amino acid residues in the first H chain CH3 region (WO2015 / 046467).

[0080] Even if the desired multispecific antibody cannot be efficiently formed, it is still possible to obtain the multispecific antibody of the present invention by isolating and purifying the desired multispecific antibody from the produced antibodies. For example, a method has been reported in which amino acid substitutions are introduced into the variable regions of two types of H chains to impart a difference in isoelectric point, thereby enabling the purification of two types of homoantibodies and the desired heteroantibody by ion exchange chromatography (WO2007114325). Another method for purifying heteroantibodies has been reported in which a heterodimerized antibody consisting of a mouse IgG2a H chain that binds to Protein A and a rat IgG2b H chain that does not bind to Protein A is purified using Protein A (WO98050431, WO95033844). Furthermore, by using an H chain in which the amino acid residues at positions 435 and 436 (EU numbering), which are the binding sites between IgG and Protein A, are substituted with amino acids that have different binding strengths to Protein A, such as Tyr or His, the interaction between each H chain and Protein A can be changed, and by using a Protein A column, it is possible to efficiently purify only the heterodimerized antibody.

[0081] These techniques can also be used in combination, for example, two or more techniques. These techniques can also be applied separately, as appropriate, to the two H chains to be associated. The antigen-binding molecules of the present invention may be antigen-binding molecules having the same amino acid sequence as those modified as described above, which have been separately prepared.

[0082] The amino acid sequence can be modified by various methods known in the art. These methods include, but are not limited to, site-directed mutagenesis (Hashimoto-Gotoh, T, Mizuno, T, Ogasahara, Y, and Nakagawa, M. (1995) An oligodeoxyribonucleotide-directed dual amber method for site-directed mutagenesis. Gene 152, 271-275; Zoller, MJ, and Smith, M. (1983) Oligonucleotide-directed mutagenesis of DNA fragments cloned into M13 vectors. Methods Enzymol. 100, 468-500; Kramer, W, Drutsa, V, Jansen, HW, Kramer, B, Pflugfelder, M, and Fritz, HJ (1984) The gapped duplex DNA approach to oligonucleotide-directed mutation construction. Nucleic Acids Res. 12, 9441-9456; Kramer, W, and Fritz, HJ (1984) The gapped duplex DNA approach to oligonucleotide-directed mutation construction. Nucleic Acids Res. 12, 9441-9456). Mutations can be performed by methods such as HJ (1987) Oligonucleotide-directed construction of mutations via gapped duplex DNA methods. Enzymol. 154, 350-367; Kunkel, TA (1985) Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc Natl Acad Sci U S A. 82, 488-492), PCR mutagenesis, cassette mutagenesis, etc.

[0083] Alternatively, the "antigen-binding molecule" of the present invention may be an antibody fragment that contains, in a single polypeptide chain, both the heavy and light chains that form the "antibody variable region" of the present invention, but lacks a constant region. Such antibody fragments may be, for example, diabodies (Db), single-chain antibodies, or sc(Fab')2.

[0084] Db is a dimer composed of two polypeptide chains (Holliger P et al., Proc. Natl. Acad. Sci. USA 90: 6444-6448 (1993), EP404,097, WO93 / 11161, etc.), and each polypeptide chain is linked by a linker of, for example, about 5 residues, which is so short that the light chain variable region (VL) and heavy chain variable region (VH) cannot bind to each other within the same chain. VL and VH encoded on the same polypeptide chain cannot form a single-chain variable region fragment because the linker between them is short, and instead form two antigen-binding sites by dimerizing.

[0085] An example of a single-chain antibody is sc(Fv)2. sc(Fv)2 is a single-chain antibody in which four variable regions, two VL and two VH, are linked by a linker such as a peptide linker to form a single chain (J Immunol. Methods (1999) 231 (1-2), 177-189). The two VH and VL may be derived from different monoclonal antibodies. Suitable examples include bispecific sc(Fv)2s that recognize two different epitopes present in the same antigen, as disclosed in Journal of Immunology (1994) 152 (11), 5368-5374. sc(Fv)2s can be produced by methods known to those skilled in the art. For example, they can be produced by linking scFvs with a linker such as a peptide linker.

[0086] As used herein, the antigen-binding domain constituting sc(Fv)2 may be configured as an antibody in which two VHs and two VLs are arranged in the following order, starting from the N-terminus of the single-chain polypeptide: VH, VL, VH, VL ([VH] linker [VL] linker [VH] linker [VL]). However, the order of the two VHs and two VLs is not limited to the above configuration and may be arranged in any order. For example, the following order configurations are also possible: [VL] linker [VH] linker [VH] linker [VL] [VH] linker [VL] linker [VL] linker [VH] [VH] linker [VH] linker [VL] linker [VL] [VL] linker [VL] linker [VH] linker [VH] [VL] linker [VH] linker [VL] linker [VH]

[0087] The molecular form of sc(Fv)2 is also described in detail in WO2006 / 132352, and based on these descriptions, a person skilled in the art would be able to appropriately prepare a desired sc(Fv)2 for producing the antigen-binding molecules disclosed herein.

[0088] The antigen-binding molecules of the present invention may also be conjugated with carrier polymers such as PEG or organic compounds such as anticancer drugs. Furthermore, glycosylation sequences can be inserted to suitably add sugar chains to achieve the desired effects.

[0089] The linker linking the antibody variable regions can be any peptide linker that can be introduced by genetic engineering or a synthetic compound linker (see, for example, Protein Engineering, 9 (3), 299-305, 1996), but peptide linkers are preferred in the present invention. The length of the peptide linker is not particularly limited and can be selected appropriately by those skilled in the art depending on the purpose. A preferred length is 5 amino acids or more (the upper limit is not particularly limited, but is usually 30 amino acids or less, preferably 20 amino acids or less), with 15 amino acids being particularly preferred. When sc(Fv)2 contains three peptide linkers, all of the peptide linkers may be the same length, or peptide linkers of different lengths may be used.

[0090] For example, for a peptide linker: Ser Gly·Ser Gly Gly Ser Ser Gly Gly Gly·Gly·Gly·Ser (SEQ ID NO: 5) Ser·Gly·Gly·Gly (SEQ ID NO: 6) Gly·Gly·Gly·Gly·Ser (SEQ ID NO: 7) Ser·Gly·Gly·Gly·Gly (SEQ ID NO: 8) Gly·Gly·Gly·Gly·Gly·Ser (SEQ ID NO: 9) Ser·Gly·Gly·Gly·Gly·Gly (SEQ ID NO: 10) Gly·Gly·Gly·Gly·Gly·Gly·Ser (SEQ ID NO: 11) Ser·Gly·Gly·Gly·Gly·Gly·Gly·Gly (SEQ ID NO: 12) (Gly·Gly·Gly·Gly·Ser (SEQ ID NO: 7))n (Ser·Gly·Gly·Gly·Gly (SEQ ID NO: 8)) [n is an integer of 1 or more], etc. However, the length and sequence of the peptide linker can be appropriately selected by those skilled in the art depending on the purpose.

[0091] Synthetic chemical linkers (chemical crosslinkers) are crosslinkers commonly used for crosslinking peptides, such as N-hydroxysuccinimide (NHS), disuccinimidyl suberate (DSS), bis(sulfosuccinimidyl)suberate (BS3), dithiobis(succinimidyl propionate) (DSP), dithiobis(sulfosuccinimidyl propionate) (DTSSP), ethylene glycol bis(succinimidyl succinate) (EGS), ethylene glycol bis(sulfosuccinimidyl succinate) (sulfo-EGS), disuccinimidyl tartrate (DST), disulfosuccinimidyl tartrate (sulfo-DST), bis[2-(succinimidooxycarbonyloxy)ethyl]sulfone (BSOCOES), and bis[2-(sulfosuccinimidooxycarbonyloxy)ethyl]sulfone (sulfo-BSOCOES). These crosslinkers are commercially available. When four antibody variable regions are linked, three linkers are usually required, and the same linkers may be used for all of them, or different linkers may be used.

[0092] F(ab')2 comprises two light chains and two heavy chains comprising the constant regions of the CH1 domain and a portion of the CH2 domain, such that interchain disulfide bonds are formed between the two heavy chains. The F(ab')2 that constitutes the polypeptide complexes disclosed herein can be conveniently obtained by partially digesting a full-length monoclonal antibody or other antibody containing the desired antigen-binding domain with a protease such as pepsin, followed by removal of the Fc fragment by adsorption onto a Protein A column. There is no particular limitation on the protease used, as long as it can digest a full-length antibody to produce F(ab')2 in a limited manner by appropriately setting the enzyme reaction conditions, such as pH. Examples of such proteases include pepsin and ficin.

[0093] Furthermore, the antigen-binding molecules of the present invention may contain additional modifications in addition to the amino acid modifications described above. The additional modifications may be selected from, for example, amino acid substitution, deletion, or modification, or a combination thereof. For example, the antigen-binding molecules of the present invention can be further modified as desired, provided that the intended function of the molecule is not substantially altered. For example, such mutations can be made by conservative substitution of amino acid residues. Furthermore, even if a modification alters the intended function of the antigen-binding molecules of the present invention, such modifications can be made as long as the change in function is within the scope of the objectives of the present invention.

[0094] Modifications of amino acid sequences in the present invention also include post-translational modifications. Specific examples of post-translational modifications include the addition or deletion of glycosylation. For example, when an antigen-binding molecule of the present invention has an IgG1-type constant region, the amino acid residue at position 297 (EU numbering) can be modified with a glycosylation. The glycosylation structure to be modified is not limited. In general, antibodies expressed in eukaryotic cells contain glycosylation in the constant region. Therefore, antibodies expressed in the following cells are usually modified with some form of glycosylation. Mammalian antibody-producing cells A eukaryotic cell transformed with an expression vector containing DNA encoding the antibody The eukaryotic cells shown here include yeast and animal cells. For example, CHO cells and HEK293H cells are typical animal cells for transformation with an expression vector containing antibody-encoding DNA. On the other hand, antibodies of the present invention also include those that are not glycosylated at the relevant positions. Antibodies whose constant regions are not glycosylated can be obtained by expressing a gene encoding the antibody in prokaryotic cells such as Escherichia coli.

[0095] In the present invention, a more specific example of an additional modification may be the addition of sialic acid to the sugar chain of the Fc region (MAbs. 2010 Sep-Oct;2(5):519-27.).

[0096] Furthermore, when the antigen-binding molecule of the present invention has an Fc region, for example, amino acid substitutions that improve binding activity to FcRn (J Immunol. 2006 Jan 1;176(1):346-56, J Biol Chem. 2006 Aug 18;281(33):23514-24, Int Immunol. 2006 Dec;18(12):1759-69, Nat Biotechnol. 2010 Feb;28(2):157-9, WO / 2006 / 019447, WO / 2006 / 053301, WO / 2009 / 086320) or amino acid substitutions to improve antibody heterogeneity or stability ((WO / 2009 / 041613)) may be added.

[0097] Furthermore, the term "antibody" in the present invention is used in the broadest sense and includes any antibody, such as a monoclonal antibody (including a full-length monoclonal antibody), a polyclonal antibody, an antibody mutant, an antibody fragment, a multispecific antibody (e.g., a bispecific antibody), a chimeric antibody, or a humanized antibody, as long as it exhibits the desired biological activity.

[0098] The antibody of the present invention is not limited by the type of antigen, the origin of the antibody, etc., and may be any antibody. The origin of the antibody is not particularly limited, but examples include human antibody, mouse antibody, rat antibody, and rabbit antibody.

[0099] Methods for producing antibodies are well known to those skilled in the art. For example, monoclonal antibodies may be produced by hybridoma methods (Kohler and Milstein, Nature 256:495 (1975)), recombinant methods (U.S. Patent No. 4,816,567), or isolated from phage antibody libraries (Clackson et al., Nature 352:624-628 (1991); Marks et al., J. Mol. Biol. 222:581-597 (1991)). Alternatively, antibodies may be isolated from single B cell clones (N. Biotechnol. 28(5): 253-457 (2011)).

[0100] Humanized antibodies are also called reshaped human antibodies. Specifically, humanized antibodies in which the CDRs of a non-human animal, such as a mouse antibody, are grafted onto a human antibody are well known. Common genetic recombination techniques for obtaining humanized antibodies are also known. Specifically, overlap extension PCR is a well-known method for grafting the CDRs of a mouse antibody onto human FRs.

[0101] A vector for expressing a humanized antibody can be prepared by inserting DNA encoding an antibody variable region in which three CDRs and four FRs are linked together with DNA encoding a human antibody constant region into an expression vector so that they are fused in frame. After introducing the integration vector into a host to establish recombinant cells, the recombinant cells are cultured to express the DNA encoding the humanized antibody, and the humanized antibody is produced in the cultured cells (see European Patent Publication EP 239400 and International Publication WO1996 / 002576).

[0102] If necessary, amino acid residues in the FR can be substituted so that the CDRs of the reshaped human antibody form an appropriate antigen-binding site. For example, amino acid sequence mutations can be introduced into the FR by applying the PCR method used to graft mouse CDRs onto human FRs.

[0103] Transgenic animals carrying the entire repertoire of human antibody genes (see International Publications WO1993 / 012227, WO1992 / 003918, WO1994 / 002602, WO1994 / 025585, WO1996 / 034096, and WO1996 / 033735) can be used as immunized animals, and desired human antibodies can be obtained by DNA immunization.

[0104] Furthermore, techniques for obtaining human antibodies by panning using a human antibody library are also known. For example, the V region of a human antibody is expressed on the surface of a phage as a single-chain antibody (scFv) by phage display. Phages expressing scFvs that bind to an antigen can be selected. The DNA sequence encoding the V region of a human antibody that binds to an antigen can be determined by analyzing the genes of the selected phage. After determining the DNA sequence of the scFv that binds to the antigen, the V region sequence can be fused in frame with the sequence of the C region of a desired human antibody and then inserted into an appropriate expression vector to prepare an expression vector. The expression vector is introduced into a suitable expression cell such as those listed above, and the gene encoding the human antibody is expressed to obtain the human antibody. These methods are already known (see International Publications WO1992 / 001047, WO1992 / 020791, WO1993 / 006213, WO1993 / 011236, WO1993 / 019172, WO1995 / 001438, and WO1995 / 015388).

[0105] In addition to phage display, other known techniques for obtaining human antibodies by panning using a human antibody library include techniques using cell-free translation systems, techniques for displaying antigen-binding molecules on the surface of cells or viruses, and techniques using emulsions. For example, techniques using cell-free translation systems include ribosome display, which forms a complex between mRNA and the translated protein via ribosomes by removing a stop codon, cDNA display, which covalently binds a gene sequence to a translated protein using a compound such as puromycin, and mRNA display, and CIS display, which forms a complex between a gene and the translated protein using a nucleic acid-binding protein. Furthermore, in addition to phage display, techniques for displaying antigen-binding molecules on the surface of cells or viruses include E. coli display, Gram-positive bacteria display, yeast display, mammalian cell display, and viral display. Techniques using emulsions include in vitro viral display, which involves encapsulating genes and translation-related molecules in an emulsion. These methods are already known (Nat Biotechnol. 2000 Dec;18(12):1287-92, Nucleic Acids Res. 2006;34(19):e127, Proc Natl Acad Sci US A. 2004 Mar 2;101(9):2806-10, Proc Natl Acad Sci US A. 2004 Jun 22;101(25):9193-8, Protein Eng Des Sel. 2008 Apr;21(4):247-55, Proc Natl Acad Sci US A. 2000 Sep 26;97(20):10701-5, MAbs. 2010 Sep-Oct;2(5):508-18, Methods Mol Biol. 2012;911:183-98).

[0106] The variable regions constituting the antibodies of the present invention can be variable regions that recognize any antigen.

[0107] The rest of the snowflakes are smooth It is located in the area of 17-IA, 4-1 BB, 4Dc, 6-keto-PGF1a, 8-iso-PGF2a, 8-oxo-dG, A1 Adenosine Receptor, A33, ACE, ACE-2, Activin, Activin A, Activin AB, Activin B, Activin C, Activin RIA, Activin RIA ALK-2, Activin RIB ALK-4, Activin RIIA, Activin RIIB, ADAM, ADAM10, ADAM12, ADAM15, ADAM17 / TACE, ADAM8, ADAM9, ADAMTS, ADAMTS4, ADAMTS5, Addressins, adiponectin, ADP ribosyl cyclase-1, aFGF, AGE, ALCAM, ALK, ALK-1, ALK-7, allergen, alpha1-antichemotrypsin, alpha1-antitrypsin, alpha-synuclein, alpha-V / beta-1 antagonist, amine, amylin, amyloid beta, amyloid immunoglobulin heavy chain variable region. amyloid immunoglobulin light chain variable region, Androgen, ANG, angiotensinogen, Angiopoietin ligand-2, anti-Id, antithrombinIII, Anthrax, APAF-1, APE, APJ, apo A1, apo serum amyloid A, Apo-SAA, APP, APRIL, AR, ARC, ART, Artemin, ASPARTIC, Atrial natriuretic factor, atrial natriuretic peptide, atrial natriuretic peptides A, atrial natriuretic peptides B,atrial natriuretic peptides C, av / b3 integrin, Axl, B7-1, B7-2, B7-H, BACE, BACE-1, Bacillus anthracis protective antigen, Bad, BAFF, BAFF-R, Bag-1, BAK, Bax, BCA-1, BCAM, BcI, BCMA, BDNF, b-ECGF, beta-2-microglobulin, betalactamase, bFGF, BID, Bik, BIM, BLC, BL-CAM, BLK, B-lymphocyte Stimulator (BIyS), BMP, BMP-2 (BMP-2a), BMP-3 (Osteogenin), BMP-4 (BMP-2b), BMP-5, BMP-6 (Vgr-1), BMP-7 (OP-1), BMP-8 (BMP-8a), BMPR, BMPR-IA (ALK-3), BMPR-IB (ALK-6), BMPR-II (BRK-3), BMPs, BOK, Bombesin, Bone-derived neurotrophic factor, bovine growth hormone, BPDE, BPDE-DNA, BRK-2, BTC, B-lymphocyte cell adhesion molecule, C10, C1-inhibitor, C1q, C3, C3a, C4, C5, C5a(complement 5a), CA125, CAD-8, Cadherin-3, Calcitonin, cAMP, Carbonic anhydrase-IX, carcinoembryonic antigen (CEA), carcinoma-associated antigen, Cardiotrophin-1, Cathepsin A, Cathepsin B, Cathepsin C / DPPI, Cathepsin D, Cathepsin E, Cathepsin H, Cathepsin L, Cathepsin O, Cathepsin S, Cathepsin V, Cathepsin X / Z / P, CBL, CCI, CCK2, CCL,CCL1 / I-309, CCL11 / Eotaxin, CCL12 / MCP-5, CCL13 / MCP-4, CCL14 / HCC-1, CCL15 / HCC-2, CCL16 / HCC-4, CCL17 / TARC, CCL18 / PARC, CCL19 / ELC, CCL2 / MCP-1, CCL20 / MIP-3-alpha, CCL21 / SLC, CCL22 / MDC, CCL23 / MPIF-1, CCL24 / Eotaxin-2, CCL25 / TECK, CCL26 / Eotaxin-3, CCL27 / CTACK, CCL28 / MEC, CCL3 / M1P-1-alpha, CCL3Ll / LD-78-beta, CCL4 / MIP-l-beta, CCL5 / RANTES, CCL6 / C10, CCL7 / MCP-3, CCL8 / MCP-2, CCL9 / 10 / MTP-1-gamma, CCR, CCR1, CCR10, CCR2, CCR3, CCR4, CCR5, CCR6, CCR7, CCR8, CCR9, CD1, CD10, CD105, CD11a, CD11b, CD11c, CD123, CD13, CD137, CD138, CD14, CD140a, CD146, CD147, CD148, CD15, CD152, CD16, CD164, CD18, CD19, CD2, CD20, CD21, CD22, CD23, CD25, CD26, CD27L, CD28, CD29, CD3, CD30, CD30L, CD32, CD33 (p67 proteins), CD34, CD37, CD38, CD3E, CD4, CD40, CD40L, CD44, CD45, CD46, CD49a, CD49b, CD5, CD51, CD52, CD54, CD55, CD56, CD6, CD61, CD64, CD66e, CD7, CD70, CD74, CD8, CD80 (B7-1), CD89, CD95, CD105, CD158a, CEA, CEACAM5, CFTR, cGMP, CGRP receptor, CINC, CKb8-1, Claudin18, CLC, Clostridium botulinum toxin,Clostridium difficile toxin, Clostridium perfringens toxin, c-Met, CMV, CMV UL, CNTF, CNTN-1, complement factor 3 (C3), complement factor D, corticosteroid-binding globulin, Colony stimulating factor-1 receptor, COX, C-Ret, CRG-2, CRTH2, CT-1, CTACK, CTGF, CTLA-4, CX3CL1 / Fractalkine, CX3CR1, CXCL, CXCL1 / Gro-alpha, CXCL10, CXCL11 / I-TAC, CXCL12 / SDF-l-alpha / beta, CXCL13 / BCA-1, CXCL14 / BRAK, CXCL15 / Lungkine. CXCL16, CXCL16, CXCL2 / Gro-beta CXCL3 / Gro-gamma, CXCL3, CXCL4 / PF4, CXCL5 / ENA-78, CXCL6 / GCP-2, CXCL7 / NAP-2, CXCL8 / IL-8, CXCL9 / Mig, CXCLlO / IP-10, CXCR, CXCR1, CXCR2, CXCR3, CXCR4, CXCR5, CXCR6, cystatin C, cytokeratin tumor-associated antigen, DAN, DCC, DcR3, DC-SIGN, Decay accelerating factor, Delta-like protein ligand 4, des(1-3)-IGF-1 (brain IGF-1), Dhh, DHICA oxidase, Dickkopf-1, digoxin, Dipeptidyl peptidase IV, DKl, DNAM-1, Dnase, Dpp, DPPIV / CD26, Dtk, ECAD, EDA, EDA-A1, EDA-A2, EDAR, EGF, EGFR (ErbB-1), EGF like domain containing protein 7, Elastase, elastin, EMA, EMMPRIN, ENA,ENA-78, Endosialin, endothelin receptor, endotoxin, Enkephalinase, eNOS, Eot, Eotaxin, Eotaxin-2, eotaxini, EpCAM, Ephrin B2 / EphB4, Epha2 tyrosine kinase receptor, epidermal growth factor receptor (EGFR), ErbB2 receptor, ErbB3 tyrosine kinase receptor, ERCC, EREG, erythropoietin (EPO), Erythropoietin receptor, E-selectin, ET-1, Exodus-2, F protein of RSV, F10, F11, F12, F13, F5, F9, Factor Ia, Factor IX, Factor Xa, Factor VII, factor VIII, Factor VIIIc, Fas, FcalphaR, FcepsilonRI, FcgammaIIb, FcgammaRI, FcgammaRIIa, FcgammaRIIIa, FcgammaRIIIb, FcRn, FEN-1, Ferritin, FGF, FGF-19, FGF-2, FGF-2 receptor, FGF-3, FGF-8, FGF-acidic, FGF-basic, FGFR, FGFR-3, Fibrin, fibroblast activation protein (FAP), fibroblast growth factor, fibroblast growth factor-10, fibronectin, FL, FLIP, Flt-3, FLT3 ligand, Folate receptor, follicle stimulating hormone (FSH), Fractalkine (CX3C), free heavy chain, free light chain, FZD1, FZD10, FZD2, FZD3, FZD4, FZD5, FZD6, FZD7, FZD8, FZD9, G250, Gas 6, GCP-2, GCSF,G-CSF, G-CSF receptor, GD2, GD3, GDF, GDF-1, GDF-15 (MIC-1), GDF-3 (Vgr-2), GDF-5 (BMP-14 / CDMP-1), GDF-6 (BMP-13 / CDMP-2), GDF-7 (BMP-12 / CDMP-3), GDF-8 (Myostatin), GDF-9, GDNF, Gelsolin, GFAP, GF-CSF, GFR-alpha1, GFR-alpha2, GFR-alpha3, GF-β1, gH envelope glycoprotein, GITR, Glucagon, Glucagon receptor, Glucagon-like peptide 1 receptor, Glut 4, Glutamate carboxypeptidase II, glycoprotein hormone receptors, glycoprotein IIb / IIIa (GP IIb / IIIa), Glypican-3, GM-CSF, GM-CSF receptor, gp130, gp140, gp72, granulocyte-CSF (G-CSF), GRO / MGSA, Growth hormone releasing factor, GRO-β, GRO-γ, H. pylori, Hapten (NP-cap or NIP-cap), HB-EGF, HCC, HCC 1, HCMV gB envelope glycoprotein, HCMV UL, Hemopoietic growth factor (HGF), Hep B gp120, heparanase, heparin cofactor II, hepatic growth factor, Bacillus anthracis protective antigen, Hepatitis C virus E2 glycoprotein, Hepatitis E, Hepcidin, Her1, Her2 / neu (ErbB-2), Her3 (ErbB-3), Her4 (ErbB-4), herpes simplex virus (HSV) gB glycoprotein, HGF, HGFA,High molecular weight melanoma-associated antigen (HMW-MAA), HIV envelope proteins such as GP120, HIV MIB gp 120 V3 loop, HLA, HLA-DR, HM1.24, HMFG PEM, HMGB-1, HRG, Hrk, HSP47, Hsp90, HSV gD glycoprotein, human cardiac myosin, human cytomegalovirus (HCMV), human growth hormone (hGH), human serum albumin, human tissue-type plasminogen activator (t-PA), Huntingtin, HVEM, IAP, ICAM, ICAM-1, ICAM-3, ICE, ICOS, IFN-alpha, IFN-beta, IFN-gamma, IgA, IgA receptor, IgE, IGF, IGF binding proteins, IGF-1, IGF-1 R, IGF-2, IGFBP, IGFR, IL, IL-1, IL-10, IL-10 receptors, IL-11, IL-11 receptors, IL-12, IL-12 receptors, IL-13, IL-13 receptors, IL-15, IL-15 receptors, IL-16, IL-16 receptors, IL-17, IL-17 receptors, IL-18 (IGIF), IL-18 receptors, IL-1alpha, IL-1beta, IL-1 receptors, IL-2, IL-2 receptors, IL-20, IL-20 receptors, IL-21, IL-21 receptors, IL-23, IL-23 receptors, IL-2 receptors, IL-3, IL-3 receptors, IL-31, IL-31 receptors, IL-3 receptors, IL-4, IL-4 receptors IL-5, IL-5 receptors, IL-6,IL-6 receptors, IL-7, IL-7 receptors, IL-8, IL-8 receptors, IL-9, IL-9 receptors, immunoglobulin immune complex, immunoglobulins, INF-alpha, INF-alpha receptors, INF-beta, INF-beta receptors, INF-gamma, INF-gamma receptors, IFN type-I , IFN type-I receptor, influenza, inhibin, Inhibin α, Inhibin β, iNOS, insulin, Insulin A-chain, Insulin B-chain, Insulin-like growth factor 1, insulin-like growth factor 2, insulin-like growth factor binding proteins, integrin, integrin alpha2, integrin alpha3, integrin alpha4, integrin alpha4 / beta1, integrin alpha-V / beta-3, integrin alpha-V / beta-6, integrin alpha4 / beta7, integrin alpha5 / beta1, integrin alpha5 / beta3, integrin alpha5 / beta6, integrin alphaσ (alphaV), integrin alphaθ, integrin beta1, integrin beta2, integrin beta3(GPIIb-IIIa), IP-10, I-TAC, JE, kalliklein, Kallikrein 11, Kallikrein 12, Kallikrein 14, Kallikrein 15, Kallikrein 2, Kallikrein 5, Kallikrein 6, Kallikrein L1, Kallikrein L2, Kallikrein L3, Kallikrein L4, kallistatin, KC, KDR,Keratinocyte Growth Factor (KGF), Keratinocyte Growth Factor-2 (KGF-2), KGF, killer immunoglobulin-like receptor, kit ligand (KL), Kit tyrosine kinase, laminin 5, LAMP, LAPP (Amylin, islet-amyloid polypeptide), LAP (TGF- 1), latency associated peptide, Latent TGF-1, Latent TGF-1 bp1, LBP, LDGF, LDL, LDL receptor, LECT2, Lefty, Leptin, leutinizing hormone (LH), Lewis-Y antigen, Lewis-Y related antigen, LFA-1, LFA-3, LFA-3 receptors, Lfo, LIF, LIGHT, lipoproteins, LIX, LKN, Lptn, L-Selectin, LT-a, LT-b, LTB4, LTBP-1, Lung surfactant, Luteinizing hormone, Lymphotactin, Lymphotoxin Beta Receptor, Lysosphingolipid receptor, Mac-1, macrophage-CSF (M-CSF), MAdCAM, MAG, MAP2, MARC, maspin, MCAM, MCK-2, MCP, MCP-1, MCP-2, MCP-3, MCP-4, MCP-I (MCAF), M-CSF, MDC, MDC (67 a.a.), MDC (69 a.a.), megsin, Mer, MET tyrosine kinase receptor family, METALLOPROTEASES, Membrane glycoprotein OX2, Mesothelin, MGDF receptor, MGMT, MHC (HLA-DR), microbial protein, MIF, MIG, MIP, MIP-1α, MIP-1β, MIP-3α, MIP-3β,MIP-4, MK, MMAC1, MMP, MMP-1, MMP-10, MMP-11, MMP-12, MMP-13, MMP-14, MMP-15, MMP-2, MMP-24, MMP-3, MMP-7, MMP-8, MMP-9, monocyte attractant protein, monocyte colony inhibitory factor, mouse gonadotropin-associated peptide, MPIF, Mpo, MSK, MSP, MUC-16, MUC18, mucin (Mud), Muellerian-inhibiting substance, Mug, MuSK, Myelin associated glycoprotein, myeloid progenitor inhibitor factor-1 (MPIF-I), NAIP, Nanobody, NAP, NAP-2, NCA 90, NCAD, N-Cadherin, NCAM, Neprilysin, Neural cell adhesion molecule, neroserpin, Neuronal growth factor (NGF), Neurotrophin-3, Neurotrophin-4, Neurotrophin-6, Neuropilin 1, Neurturin, NGF-beta, NGFR, NKG20, N-methionyl human growth hormone, nNOS, NO, Nogo-A, Nogo receptor, non-structural protein type 3 (NS3) from the hepatitis C virus, NOS, Npn, NRG-3, NT, NT-3, NT-4, NTN, OB, OGG1, Oncostatin M, OP-2, OPG, OPN, OSM, OSM receptors, osteoinductive factors, osteopontin, OX40L, OX40R, oxidized LDL, p150, p95, PADPr, parathyroid hormone, PARC, PARP, PBR, PBSF, PCAD,P-Cadherin, PCNA, PCSK9, PDGF, PDGF receptor, PDGF-AA, PDGF-AB, PDGF-BB, PDGF-D, PDK-1, PECAM, PEDF, PEM, PF-4, PGE, PGF, PGI2, PGJ2, PIGF, PIN, PLA2, Placenta growth factor, placental alkaline phosphatase (PLAP), placental lactogen, pl, asminogen activator inhibitor-1, platelet-growth factor, plgR, PLP, poly glycol chains of different size(e.g. PEG-20, PEG-30, PEG40), PP14, prekallikrein, prion protein, procalcitonin, Programmed cell death protein 1, proinsulin, prolactin, Proprotein convertase PC9, prorelaxin, prostate specific membrane antigen (PSMA), Protein A, Protein C, Protein D, Protein S, Protein Z, PS, PSA, PSCA, PsmAr, PTEN, PTHrp, Ptk, PTN, P-selectin glycoprotein ligand-1, R51, RAGE, RANK, RANKL, RANTES, relaxin, Relaxin A-chain, Relaxin B-chain, renin, respiratory syncytial virus (RSV) F, Ret, reticulon 4, Rheumatoid factors, RLI P76, RPA2, RPK-1, RSK, RSV Fgp, S100, RON-8, SCF / KL, SCGF, Sclerostin, SDF-1, SDF1α, SDF1β, SERINE, Serum Amyloid P, Serum albumin, sFRP-3, Shh, Shiga like toxin II, SIGIRR, SK-1, SLAM, SLPI, SMAC, SMDF, SMOH, SOD, SPARC, sphingosine 1-phosphate receptor 1, Staphylococcal lipoteichoic acid, Stat, STEAP, STEAP-II, stem cell factor (SCF), streptokinase, superoxide dismutase,syndecan-1, TACE, TACI, TAG-72 (tumor-associated glycoprotein-72), TARC, TB, TCA-3, T-cell receptor alpha / beta, TdT, TECK, TEM1, TEM5, TEM7, TEM8, Tenascin, TERT, testicular PLAP-like alkaline phosphatase, TfR, TGF, TGF-alpha, TGF-beta, TGF-beta Pan Specific, TGF-beta RII, TGF-beta RIIb, TGF-beta RIII, TGF-beta Rl (ALK-5), TGF-beta1, TGF-beta2, TGF-beta3, TGF-beta4, TGF-beta5, TGF-I, Thrombin, thrombopoietin (TPO), Thymic stromal lymphoprotein receptor, Thymus Ck-1, thyroid stimulating hormone (TSH), thyroxine, thyroxine-binding globulin, Tie, TIMP, TIQ, Tissue Factor, tissue factor protease inhibitor, tissue factor protein, TMEFF2, Tmpo, TMPRSS2, TNF receptor I, TNF receptor II, TNF-alpha, TNF-beta, TNF-beta2, TNFc, TNF-RI, TNF-RII, TNFRSF10A (TRAIL R1 Apo-2 / DR4), TNFRSF10B (TRAIL R2 DR5 / KILLER / TRICK-2A / TRICK-B), TNFRSF10C (TRAIL R3 DcR1 / LIT / TRID), TNFRSF10D (TRAIL R4 DcR2 / TRUNDD), TNFRSF11A (RANK ODF R / TRANCE R), TNFRSF11B (OPG OCIF / TR1), TNFRSF12 (TWEAK R FN14), TNFRSF12A,TNFRSF13B (TACI), TNFRSF13C (BAFF R), TNFRSF14 (HVEM ATAR / HveA / LIGHT R / TR2), TNFRSF16 (NGFR p75NTR), TNFRSF17 (BCMA), TNFRSF18 (GITR AITR), TNFRSF19 (TROY TAJ / TRADE), TNFRSF19L (RELT), TNFRSF1A (TNF Rl CD120a / p55-60), TNFRSF1B (TNF RII CD120b / p75-80), TNFRSF21 (DR6), TNFRSF22 (DcTRAIL R2 TNFRH2), TNFRSF25 (DR3 Apo-3 / LARD / TR-3 / TRAMP / WSL-1), TNFRSF26 (TNFRH3), TNFRSF3 (LTbR TNF RIII / TNFC R), TNFRSF4 (OX40 ACT35 / TXGP1 R), TNFRSF5 (CD40 p50), TNFRSF6 (Fas Apo-1 / APT1 / CD95), TNFRSF6B (DcR3 M68 / TR6), TNFRSF7 (CD27), TNFRSF8 (CD30), TNFRSF9 (4-1 BB CD137 / ILA), TNFRST23 (DcTRAIL R1 TNFRH1), TNFSF10 (TRAIL Apo-2 Ligand / TL2), TNFSF11 (TRANCE / RANK Ligand ODF / OPG Ligand), TNFSF12 (TWEAK Apo-3 Ligand / DR3 Ligand), TNFSF13 (APRIL TUE2), TNFSF13B (BAFF LEAD / TALL1 / THANK / TNFSF20), TNFSF14 (LIGHT WHO Ligand / LTg), TNFSF15 (TL1A / VEGI), TNFSF18 (GITR Ligand AITR Ligand / TL6), TNFSF1A (TNF-a Connectin / DIF / TNFSF2), TNFSF1B (TNF-b LTa / TNFSF1), TNFSF3 (LTb TNFC / p33), TNFSF4 (OX40 Ligand gp34 / TXGP1),TNFSF5 (CD40 Ligand CD154 / gp39 / HIGM1 / IMD3 / TRAP), TNFSF6 (Fas Ligand Apo-1 Ligand / APT1 Ligand), TNFSF7 (CD27 Ligand CD70), TNFSF8 (CD30 Ligand CD153), TNFSF9 (4-1 BB Ligand CD137 Ligand), TNF-α, TNF-β, TNIL-I, toxic metabolite, TP-1, t-PA, Tpo, TRAIL, TRAIL R, TRAIL-R1, TRAIL-R2, TRANCE, transferrin receptor, transforming growth factors (TGF) such as TGF-alpha and TGF-beta, Transmembrane glycoprotein NMB, Transthyretin, TRF, Trk, TROP-2, Trophoblast glycoprotein, TSG, TSLP, Tumor Necrosis Factor (TNF), tumor-associated antigen CA 125, tumor-associated antigen expressing Lewis Y related carbohydrate, TWEAK, TXB2, Ung, uPAR, uPAR-1, Urokinase, VAP-1, vascular endothelial growth factor (VEGF), vaspin, VCAM, VCAM-1, VECAD, VE-Cadherin, VE-Cadherin-2, VEFGR-1 (flt-1), VEFGR-2, VEGF receptor (VEGFR), VEGFR-3 (flt-4), VEGI, VIM, Viral antigens, VitB12 receptor, Vitronectin receptor, VLA, VLA-1, VLA-4, VNR integrin, von Willebrand Factor (vWF), WIF-1, WNT1, WNT10A, WNT10B, WNT11, WNT16, WNT2,These include WNT2B / 13, WNT3, WNT3A, WNT4, WNT5A, WNT5B, WNT6, WNT7A, WNT7B, WNT8A, WNT8B, WNT9A, WNT9B, XCL1, XCL2 / SCM-1-beta, XCL1 / Lymphotactin, XCR1, XEDAR, XIAP, and XPD.

[0108] Among the antibody variable regions contained in the antigen-binding molecules of the present invention, the "first antigen" and "second antigen" are those bound by variable regions that can bind to "two different antigens" but cannot simultaneously bind to these antigens, or the "first antigen," "second antigen," and "fourth antigen" are those bound by variable regions that can bind to "three different antigens" but cannot simultaneously bind to these antigens. Preferred examples of these include immune cell surface molecules (e.g., T cell surface molecules, NK cell surface molecules, dendritic cell surface molecules, B cell surface molecules, NKT cell surface molecules, MDSC cell surface molecules, and macrophage surface molecules), antigens that are expressed on tumor cells, tumor blood vessels, stromal cells, and the like, but are also expressed in normal tissues (integrin, tissue factor, VEGFR, PDGFR, EGFR, IGFR, MET chemokine receptor, heparan sulfate proteoglycan, CD44, fibronectin, DR5, TNFRSF, etc.). In a combination of a "first antigen" and a "second antigen" that binds to a variable region that can bind to "two different antigens" but cannot simultaneously bind to these antigens, it is preferable that either the first antigen or the second antigen is, for example, a molecule specifically expressed on T cells, and the other antigen is a molecule expressed on the surface of T cells or other immune cells. In another embodiment, it is preferable that either the first antigen or the second antigen is, for example, a molecule specifically expressed on T cells, and the other antigen is a molecule expressed on immune cells that is different from the previously selected antigen. Specific examples of molecules specifically expressed on T cells include CD3 and T cell receptors. CD3 is particularly preferred. The site of CD3 to which the antigen-binding molecule of the present invention binds may be, for example, in the case of human CD3, any epitope present in the γ chain, δ chain, or ε chain sequence that constitutes human CD3. Particularly preferred are epitopes present in the extracellular region of the ε chain of the human CD3 complex.The polynucleotide sequences of the γ, δ, and ε chains that make up CD3 are shown in SEQ ID NOs: 83 (NM_000073.2), 85 (NM_000732.4), and 87 (NM_000733.3), and their polypeptide sequences are shown in SEQ ID NOs: 84 (NP_000064.1), 86 (NP_000723.1), and 88 (NP_000724.1) (RefSeq accession numbers are shown in parentheses). Further examples of antigens include Fcγ receptors, TLRs, lectins, IgA, immune checkpoint molecules, TNF superfamily molecules, TNFR superfamily molecules, and NK receptor molecules. Furthermore, in a combination of a "first antigen," a "second antigen," and a "fourth antigen" that binds to a variable region that can bind to "three different antigens" but cannot simultaneously bind to these antigens, it is preferable that any one of the first, second, and fourth antigens is, for example, a molecule specifically expressed on T cells, and the remaining two antigens are molecules expressed on the surface of T cells or other immune cells. In another embodiment, in a combination of a "first antigen," a "second antigen," and a "fourth antigen," it is preferable that any one of the first, second, and fourth antigens is, for example, a molecule specifically expressed on T cells, and the remaining two antigens are molecules expressed on immune cells that are different from the previously selected antigen. Specific examples of molecules specifically expressed on T cells include CD3 and T cell receptors. CD3 is particularly preferred. The site of CD3 to which the antigen-binding molecules of the present invention bind may be any epitope present in the γ, δ, or ε chain sequences constituting human CD3. An epitope present in the extracellular domain of the ε chain of the human CD3 complex is particularly preferred. The polynucleotide sequences of the γ, δ, or ε chain structures constituting CD3 are shown in SEQ ID NOs: 83 (NM_000073.2), 85 (NM_000732.4), and 87 (NM_000733.3), and their polypeptide sequences are shown in SEQ ID NOs: 84 (NP_000064.1), 86 (NP_000723.1), and 88 (NP_000724.1) (RefSeq accession numbers are shown in parentheses). Furthermore, examples of the remaining two antigens include Fcγ receptors, TLRs, lectins, IgA, immune checkpoint molecules, TNF superfamily molecules, TNFR superfamily molecules, and NK receptor molecules. Furthermore, among the variable regions of antibodies contained in the antigen-binding molecules of the present invention, the "third antigen" that binds to another variable region and is different from the above-mentioned "first antigen," "second antigen," and "fourth antigen" is preferably, for example, an antigen specific to tumor cells, and includes antigens that are expressed in association with the malignant transformation of cells, as well as abnormal sugar chains that appear on the cell surface or on protein molecules when cells become cancerous. Specifically, for example, ALK receptor (pleiotrophin receptor), pleiotrophin, KS 1 / 4 pancreatic cancer antigen, ovarian cancer antigen (CA125), prostatic acid phosphate, prostate-specific antigen (PSA), melanoma-associated antigen p97, melanoma antigen gp75, high molecular weight melanoma antigen (HMW-MAA), prostate-specific membrane antigen, carcinoembryonic antigen (CEA), polymorphic epithelial mucin antigen, human milk fat globule antigen (CEA), colorectal tumor-associated antigens such as TAG-72, CO17-1A, GICA 19-9, CTA-1 and LEA, and Burkitt's lymphoma antigen-38.13, melanoma-specific antigens such as CD19, human B lymphoma antigen-CD20, CD33, ganglioside GD2, ganglioside GD3, ganglioside GM2 and ganglioside GM3, virus-induced tumor antigens such as tumor-specific transplantable cell surface antigen (TSTA), T antigen, envelope antigens of DNA tumor viruses and RNA tumor viruses, carcinoembryonic antigen α-fg such as colon CEA, 5T4 oncofetal trophoblast glycoprotein and bladder tumor carcinoembryonic antigen. differentiation antigens such as human lung cancer antigens L6 and L20, fibrosarcoma antigens, human leukemia T cell antigen-Gp37, neoglycoproteins, sphingolipids, breast cancer antigens such as EGFR (epidermal growth factor receptor), NY-BR-16, NY-BR-16 and HER2 antigen (p185HER2), polymorphic epithelial mucin (PEM), malignant human lymphocyte antigen-APO-1, differentiation antigens such as I antigen found in fetal erythrocytes, early endoderm I antigen found in adult erythrocytes, embryos before implantation, gastric cancer I (Ma), M18 and M39 found in mammary epithelium, SSEA-1, VEP8, VEP9, Myl, VIM-D5 found in bone marrow cells, D156-22 and TRA-1-85 (blood group H) found in colorectal cancer, SCP-1 found in testicular and ovarian cancer, C14 found in colon cancer, F3 found in lung cancer, AH6 found in gastric cancer, Y hapten, Ley found in embryonal carcinoma cells, TL5 (blood group A), EGF receptor found in A431 cells, and pancreatic cancer E1 series (blood group B) found in embryonal carcinoma cells, FC10.2 found in embryonal carcinoma cells, gastric cancer antigen, CO-514 (blood group Lea) found in adenocarcinoma, NS-10 found in adenocarcinoma, CO-43 (blood group Leb), G49 found in the EGF receptor of A431 cells, MH2 (blood group ALeb / Ley) found in colon cancer, 19.9 found in colon cancer, gastric cancer mucin, T5A7 found in bone marrow cells, R24 found in melanoma, 4.2, GD3, D1 found in embryonal carcinoma cells.1, OFA-1, GM2, OFA-2, GD2, and M1:22:25:8, as well as SSEA-3 and SSEA-4 found in 4- to 8-cell stage embryos, subcutaneous T-cell lymphoma antigen, MART-1 antigen, sialyl Tn (STn) antigen, colon cancer antigen NY-CO-45, lung cancer antigen NY-LU-12 variant A, adenocarcinoma antigen ART1, paraneoplastic-associated brain-testis cancer antigen (cancer neural antigen MA2, paraneoplastic neural antigen), neural carcinoma abdominal antigen 2 (NOVA2), blood cell cancer antigen gene 520, tumor-associated antigen CO-029, tumor-associated antigen M Examples include AGE-C1 (cancer / testis antigen CT7), MAGE-B1 (MAGE-XP antigen), MAGE-B2 (DAM6), MAGE-2, MAGE-4a, MAGE-4b, and MAGE-X2, cancer-testis antigen (NY-EOS-1), YKL-40, and fragments of any of the above polypeptides or modified structures thereof (such as the above-mentioned modified phosphate groups or sugar chains), EpCAM, EREG, CA19-9, CA15-3, cereal SSEA-1 (SLX), HER2, PSMA, CEA, and CLEC12A.

[0109] The antigen-binding molecules of the present invention can be produced by methods known to those skilled in the art. For example, antibodies can be produced by the following methods, but the methods are not limited to these. Many combinations of host cells and expression vectors are known for producing antibodies by introducing a gene encoding an isolated polypeptide into a suitable host. All of these expression systems can be applied to isolating the antigen-binding molecules of the present invention. When eukaryotic cells are used as host cells, animal cells, plant cells, or fungal cells can be used as appropriate. Specific examples of animal cells include the following cells: (1) Mammalian cells: CHO (Chinese hamster ovary cell line), COS (Monkey kidney cell line), myeloma (Sp2 / O, NS0, etc.), BHK (baby hamster kidney cell line), HEK293 (human embryonic kidney cell line with sheared adenovirus (Ad)5 DNA), PER.C6 cells (human embryonic retinal cell line transformed with the Adenovirus Type 5 (Ad5) E1A and E1B genes), Hela, Vero, etc. (Current Protocols in Protein Science (May 2001, Unit 5.9, Table 5.9.1)) (2) Amphibian cells: Xenopus oocytes, etc. (3) Insect cells: sf9, sf21, Tn5, etc. Antibodies can also be produced in E. coli (mAbs 2012 Mar-Apr; 4(2): 217-225.) or yeast (WO2000023579). Antibodies produced in E. coli are not glycosylated, whereas antibodies produced in yeast are glycosylated.

[0110] DNA encoding an antibody heavy chain in which one or more amino acid residues in the variable region have been substituted with other amino acids of interest, and DNA encoding an antibody light chain are expressed. DNA encoding a heavy chain or light chain in which one or more amino acid residues in the variable region have been substituted with other amino acids of interest can be obtained, for example, by obtaining DNA encoding the variable region of an antibody produced against a certain antigen using a known method and introducing appropriate substitutions so that codons encoding specific amino acids in the region now encode other amino acids of interest.

[0111] Alternatively, DNA encoding a protein in which one or more amino acid residues in the variable region of an antibody prepared against a certain antigen using a known method have been substituted with other amino acids of interest can be designed and then chemically synthesized to obtain DNA encoding a heavy chain in which one or more amino acid residues in the variable region have been substituted with other amino acids of interest. The amino acid substitution site and type are not particularly limited. Preferred regions for amino acid modification include solvent-exposed regions and loop regions in the variable region. Among these, CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numbering 31 to 35, 50 to 65, 71 to 74, 95 to 102 for the H-chain variable region, and Kabat numbering 24 to 34, 50 to 56, 89 to 97 for the L-chain variable region are preferred, and Kabat numbering 31, 52a to 61, 71 to 74, 97 to 101 for the H-chain variable region, and Kabat numbering 24 to 34, 51 to 56, 89 to 96 for the L-chain variable region are more preferred. Furthermore, the amino acid modification is not limited to substitution, but may be any one of deletion, addition, insertion, or modification, or a combination thereof.

[0112] Furthermore, DNA encoding a heavy chain in which one or more amino acid residues in the variable region have been substituted with other amino acids of interest can be prepared by dividing it into partial DNAs. Examples of combinations of partial DNAs include, but are not limited to, DNA encoding a variable region and DNA encoding a constant region, or DNA encoding a Fab region and DNA encoding an Fc region. DNA encoding a light chain can also be prepared by dividing it into partial DNAs in a similar manner.

[0113] Methods for expressing the above DNA include the following. For example, a heavy chain expression vector is constructed by incorporating DNA encoding a heavy chain variable region into an expression vector together with DNA encoding a heavy chain constant region. Similarly, a light chain expression vector is constructed by incorporating DNA encoding a light chain variable region into an expression vector together with DNA encoding a light chain constant region. These heavy and light chain genes can also be incorporated into a single vector.

[0114] When DNA encoding the antibody of interest is incorporated into an expression vector, it is incorporated into the expression vector so that expression is under the control of an expression control region, such as an enhancer or promoter. Next, host cells are transformed with this expression vector to express the antibody. In this case, an appropriate combination of host and expression vector can be used.

[0115] Examples of vectors include M13 vectors, pUC vectors, pBR322, pBluescript, pCR-Script, etc. Furthermore, for the purpose of subcloning or excision of cDNA, in addition to the above vectors, for example, pGEM-T, pDIRECT, pT7, etc. can be used.

[0116] When vectors are used to produce the antibodies of the present invention, expression vectors are particularly useful. For example, when the host is Escherichia coli such as JM109, DH5α, HB101, or XL1-Blue, it is essential that the expression vector contain a promoter that enables efficient expression in E. coli, such as the lacZ promoter (Ward et al., Nature (1989) 341, 544-546; FASEB J. (1992) 6, 2422-2427, incorporated herein by reference in its entirety), the araB promoter (Better et al., Science (1988) 240, 1041-1043, incorporated herein by reference in its entirety), or the T7 promoter. In addition to the above vectors, other such vectors include pGEX-5X-1 (Pharmacia), the "QIAexpress system" (QIAGEN), pEGFP, and pET (in this case, the host is preferably BL21, which expresses T7 RNA polymerase).

[0117] The vector may also contain a signal sequence for polypeptide secretion. When producing a polypeptide in the periplasm of E. coli, the signal sequence used may be the pelB signal sequence (Lei, SP et al., J. Bacteriol. (1987) 169, 4397, the entire contents of which are incorporated herein by reference). The vector can be introduced into host cells using, for example, the lipofectin method, the calcium phosphate method, or the DEAE-Dextran method.

[0118] In addition to E. coli expression vectors, examples of vectors for producing the polypeptides of the present invention include mammalian-derived expression vectors (e.g., pcDNA3 (Invitrogen), pEGF-BOS (Nucleic Acids. Res. 1990, 18(17), p5322, the entire contents of which are incorporated herein by reference), pEF, and pCDM8), insect cell-derived expression vectors (e.g., the "Bac-to-BAC baculovirus expression system" (GIBCO BRL), and pBacPAK8), plant-derived expression vectors (e.g., pMH1 and pMH2), animal virus-derived expression vectors (e.g., pHSV, pMV, and pAdexLcw), retrovirus-derived expression vectors (e.g., pZIPneo), yeast-derived expression vectors (e.g., the "Pichia Expression Kit" (Invitrogen), pNV11, and SP-Q01), and Bacillus subtilis-derived expression vectors (e.g., pPL608 and pKTH50).

[0119] For expression in animal cells such as CHO cells, COS cells, NIH3T3 cells, and HEK293 cells, it is essential to have a promoter necessary for intracellular expression, such as the SV40 promoter (Mulligan et al., Nature (1979) 277, 108, incorporated herein by reference in its entirety), the MMTV-LTR promoter, the EF1α promoter (Mizushima et al., Nucleic Acids Res. (1990) 18, 5322, incorporated herein by reference in its entirety), the CAG promoter (Gene. (1991) 108, 193, incorporated herein by reference in its entirety), or the CMV promoter. It is even more preferable if the vector contains a gene for selecting transformed cells (e.g., a drug resistance gene that can be distinguished by a drug (e.g., neomycin, G418, etc.)). Examples of vectors with such properties include pMAM, pDR2, pBK-RSV, pBK-CMV, pOPRSV, and pOP13. In addition, the EBNA1 protein may be coexpressed to increase the copy number of the gene, using a vector containing the replication origin OriP (Biotechnol Bioeng. 2001 Oct 20;75(2):197-203., Biotechnol Bioeng. 2005 Sep 20;91(6):670-7.).

[0120] Furthermore, to achieve stable gene expression and increase the intracellular copy number of a gene, one method involves introducing a vector (e.g., pCHOI) containing a complementary DHFR gene into CHO cells deficient in the nucleic acid synthesis pathway and amplifying the gene with methotrexate (MTX). Another method involves transforming COS cells carrying a gene expressing SV40 T antigen on their chromosomes with a vector (e.g., pcD) containing an SV40 replication origin. Replication origins derived from polyomavirus, adenovirus, bovine papillomavirus (BPV), etc. can also be used. Furthermore, to increase the gene copy number in a host cell system, the expression vector can contain a selection marker such as the aminoglycoside transferase (APH) gene, thymidine kinase (TK) gene, Escherichia coli xanthine-guanine phosphoribosyltransferase (Ecogpt) gene, or dihydrofolate reductase (dhfr) gene.

[0121] Antibodies can be recovered, for example, by culturing the transformed cells and then isolating the antibodies from the transformed cells or from the culture medium. Antibody isolation and purification can be carried out by an appropriate combination of methods such as centrifugation, ammonium sulfate fractionation, salting out, ultrafiltration, C1q, FcRn, protein A, protein G columns, affinity chromatography, ion exchange chromatography, and gel filtration chromatography.

[0122] Efficient methods for producing multispecific antibodies include the above-mentioned techniques, such as the knobs-into-holes technique (WO 1996 / 027011, Ridgway JB et al., Protein Engineering (1996) 9, 617-621, Merchant AM et al., Nature Biotechnology (1998) 16, 677-681) and the technique of suppressing undesired association of H chains by introducing charge repulsion (WO 2006 / 106905).

[0123] The present invention further provides a method for producing an antigen-binding molecule of the present invention, comprising an antibody variable region capable of binding to two different first and second antigens, but which does not simultaneously bind to the first and second antigens (first variable region), and a variable region (second variable region) that binds to a third antigen different from the first and second antigens, the method comprising the step of preparing a library of antigen-binding molecules having diverse amino acid sequences of the first variable region.

[0124] For example, the production method may include the following steps: (i) preparing a library of antigen-binding molecules comprising variable regions in which at least one amino acid in the variable region of an antibody that binds to a first antigen or a second antigen has been modified, and in which at least one of the modified amino acids in the variable region differs from one another; (ii) selecting, from the prepared library, antigen-binding molecules that have binding activity to the first antigen and the second antigen but do not bind to the first antigen and the second antigen simultaneously, and that contain variable regions; (iii) culturing a host cell comprising a nucleic acid encoding the variable region of the antigen-binding molecule selected in step (ii) and a nucleic acid encoding a variable region of an antigen-binding molecule that binds to a third antigen to express an antigen-binding molecule comprising a variable region of an antibody that can bind to the first antigen and the second antigen but does not simultaneously bind to the first antigen and the second antigen, and a variable region that binds to the third antigen; and (iv) recovering the antigen-binding molecule from the host cell culture.

[0125] In the present production method, step (ii) may be the following optional step: (v) selecting, from the prepared library, antigen-binding molecules that have binding activity to the first antigen and the second antigen but do not simultaneously bind to the first antigen and the second antigen, which are expressed on different cells.

[0126] The antigen-binding molecule used in step (i) above is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0127] The amino acids to be modified are selected, for example, from the variable region of an antibody that binds to the first or second antigen, so that the amino acid modification does not result in loss of binding to the antigen.

[0128] The amino acid modifications of the present invention may be used alone or in combination. When multiple compounds are used in combination, the number of compounds to be combined is not particularly limited, and may be, for example, 2 or more and 30 or less, preferably 2 or more and 25 or less, 2 or more and 22 or less, 2 or more and 20 or less, 2 or more and 15 or less, 2 or more and 10 or less, 2 or more and 5 or less, or 2 or more and 3 or less. When multiple amino acid modifications are combined, the amino acid modifications may be made only to the heavy chain variable region or light chain variable region of the antibody, or may be appropriately distributed and made to both the heavy chain variable region and the light chain variable region.

[0129] Preferred regions for amino acid modification include solvent-exposed regions and loop regions in the variable region. Among these, the CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numberings 31-35, 50-65, 71-74, and 95-102 for the H-chain variable region and Kabat numberings 24-34, 50-56, and 89-97 for the L-chain variable region are preferred, with Kabat numberings 31, 52a-61, 71-74, and 97-101 for the H-chain variable region and Kabat numberings 24-34, 51-56, and 89-96 for the L-chain variable region being more preferred.

[0130] Furthermore, modifying amino acid residues also includes randomly modifying amino acids in the aforementioned regions of the variable regions of an antibody that binds to the first or second antigen, or inserting a peptide known to have binding activity toward the desired antigen into the aforementioned regions. Thus, the antigen-binding molecules of the present invention can be obtained by selecting, from among the modified antigen-binding molecules, variable regions that can bind to the first and second antigens but cannot simultaneously bind to these antigens. Examples of peptides known to have binding activity toward the desired antigen include the peptides listed in Table 1 above.

[0131] Whether a variable region is capable of binding to both the first antigen and the second antigen but not being able to bind to these antigens simultaneously, and further whether a variable region is capable of simultaneously binding to both the first antigen and the second antigen when one of the first antigen and the second antigen is present on a cell and the other is present alone, when both are present alone, or when both are present on the same cell, but not being able to simultaneously bind when they are expressed on different cells, can be confirmed in the same manner using the methods described above.

[0132] The present invention further provides a method for producing an antigen-binding molecule of the present invention, which comprises an antibody variable region capable of binding to two different first and second antigens, but which does not simultaneously bind to the first and second antigens (first variable region), the method comprising the step of preparing a library of antigen-binding molecules having diverse amino acid sequences of the first variable region.

[0133] Examples of methods for producing such antigen-binding molecules include production methods comprising the following steps: (i) preparing a library of antigen-binding molecules comprising variable regions in which at least one amino acid in the variable region of an antibody that binds to a first antigen or a second antigen has been modified, and in which at least one of the modified amino acids in the variable region differs from one another; (ii) selecting, from the prepared library, antigen-binding molecules that have binding activity to the first antigen and the second antigen but do not bind to the first antigen and the second antigen simultaneously, and that contain variable regions; (iii) culturing a host cell containing a nucleic acid encoding the variable region of the antigen-binding molecule selected in step (ii) to express an antigen-binding molecule comprising a variable region of an antibody that can bind to the first antigen and the second antigen but does not simultaneously bind to the first antigen and the second antigen; and (iv) recovering the antigen-binding molecule from the host cell culture. Preferred regions for the amino acid modification include the heavy chain variable region. More preferred are solvent-exposed regions and loop regions in the variable region. Among these, the CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numberings 31-35, 50-65, 71-74, and 95-102 for the H chain variable region and Kabat numberings 24-34, 50-56, and 89-97 for the L chain variable region are preferred, with Kabat numberings 31, 52a-61, 71-74, and 97-101 for the H chain variable region and Kabat numberings 24-34, 51-56, and 89-96 for the L chain variable region being more preferred.

[0134] In the present production method, the step (ii) may be the following selective step: (v) selecting, from the prepared library, antigen-binding molecules that have binding activity to the first antigen and the second antigen but do not simultaneously bind to the first antigen and the second antigen, which are expressed on different cells.

[0135] The antigen-binding molecule used in step (i) above is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0136] The amino acids to be modified are selected, for example, from the variable region of an antibody that binds to the first or second antigen, so that the amino acid modification does not result in loss of binding to the antigen.

[0137] The amino acid modifications of the present invention may be used alone or in combination. When multiple compounds are used in combination, the number of compounds to be combined is not particularly limited, and may be, for example, 2 or more and 30 or less, preferably 2 or more and 25 or less, 2 or more and 22 or less, 2 or more and 20 or less, 2 or more and 15 or less, 2 or more and 10 or less, 2 or more and 5 or less, or 2 or more and 3 or less. When multiple amino acid modifications are combined, the amino acid modifications may be made to only one of the heavy chain variable region or light chain variable region of the antibody, or may be appropriately distributed and made to both the heavy chain variable region and the light chain variable region.

[0138] Furthermore, modifying amino acid residues also includes randomly modifying amino acids in the aforementioned regions of the variable regions of an antibody that binds to the first or second antigen, or inserting a peptide known to have binding activity toward the desired antigen into the aforementioned regions. Thus, the antigen-binding molecules of the present invention can be obtained by selecting, from among the modified antigen-binding molecules, variable regions that can bind to the first and second antigens but cannot simultaneously bind to these antigens. Examples of peptides known to have binding activity toward the desired antigen include the peptides listed in Table 1 above.

[0139] Whether a variable region is capable of binding to both the first antigen and the second antigen but not being able to bind to these antigens simultaneously, and further whether a variable region is capable of simultaneously binding to both the first antigen and the second antigen when one of the first antigen and the second antigen is present on a cell and the other is present alone, when both are present alone, or when both are present on the same cell, but not being able to simultaneously bind when they are expressed on different cells, can be confirmed in the same manner using the methods described above.

[0140] Furthermore, the present invention also includes antigen-binding molecules produced by the production method. The type and range of amino acid modifications introduced by this method are not particularly limited.

[0141] A non-limiting embodiment of the libraries of the present invention is a library in which CD3 (in the case of human CD3, the γ chain, δ chain, or ε chain constituting human CD3) is selected as the first antigen, and the library is composed of antigen-binding molecules that bind to CD3 and any second antigen.

[0142] A preferred embodiment of the library of the present invention is a library mainly composed of a plurality of antigen-binding molecules having sequences different from one another, wherein the antigen-binding regions in the antigen-binding molecules are variable regions of antibodies that can bind to a first antigen and a second antigen different from the first antigen but do not simultaneously bind to the first antigen and the second antigen, and wherein one of the first antigen and the second antigen is CD3, and the other antigen is a molecule expressed on the surface of T cells or other immune cells.

[0143] When either the first antigen or the second antigen is human CD3, the antigen-binding molecule preferably binds to the γ chain, δ chain, or ε chain that constitutes human CD3. Furthermore, the variable region is preferably a variable region that does not simultaneously bind to the first antigen and the second antigen, each of which is expressed on different cells. Here, "expressed on different cells" means that the antigens are expressed on different cells, and such a combination of cells may be, for example, cells of the same type, such as a T cell and another T cell, or cells of different types, such as a T cell and an NK cell.

[0144] The present invention further provides a method for producing an antigen-binding molecule of the present invention, which comprises antibody variable regions capable of binding to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first and second antigens), a variable region (first variable region) that does not simultaneously bind to the three antigens, and a variable region (second variable region) that binds to a third antigen different from the three antigens, the method comprising the step of preparing a library of antigen-binding molecules having diverse amino acid sequences of the first variable region.

[0145] For example, the production method may include the following steps: (i) preparing a library of antigen-binding molecules comprising variable regions in which at least one amino acid in the variable regions of antibodies that bind to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen) has been modified, and in which at least one amino acid in the modified variable regions differs from each other; (ii) selecting, from the prepared library, antigen-binding molecules comprising variable regions that have binding activity to the three antigens but do not simultaneously bind to the three antigens; (iii) culturing a host cell comprising a nucleic acid encoding the variable region of the antigen-binding molecule selected in step (ii) and a nucleic acid encoding a variable region of an antigen-binding molecule that binds to a third antigen, to express an antigen-binding molecule comprising variable regions of an antibody that can bind to the three antigens but does not simultaneously bind to the three antigens, and a variable region that binds to the third antigen; and (iv) recovering the antigen-binding molecule from the host cell culture.

[0146] In the present production method, step (ii) may be the following optional step: (v) selecting, from the prepared library, antigen-binding molecules that have binding activity to the three antigens but that contain variable regions that do not simultaneously bind to the three antigens expressed on different cells.

[0147] The antigen-binding molecule used in step (i) above is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0148] The amino acids to be modified are selected, for example, from the variable regions of an antibody that binds to the three antigens, and such amino acid modification does not result in loss of binding to the antigen.

[0149] The amino acid modifications of the present invention may be used alone or in combination. When multiple compounds are used in combination, the number of compounds to be combined is not particularly limited, and may be, for example, 2 or more and 30 or less, preferably 2 or more and 25 or less, 2 or more and 22 or less, 2 or more and 20 or less, 2 or more and 15 or less, 2 or more and 10 or less, 2 or more and 5 or less, or 2 or more and 3 or less. When multiple amino acid modifications are combined, the amino acid modifications may be made only to the heavy chain variable region or light chain variable region of the antibody, or may be appropriately distributed and made to both the heavy chain variable region and the light chain variable region.

[0150] Preferred regions for amino acid modification include solvent-exposed regions and loop regions in the variable region. Among these, the CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numberings 31-35, 50-65, 71-74, and 95-102 for the H-chain variable region and Kabat numberings 24-34, 50-56, and 89-97 for the L-chain variable region are preferred, with Kabat numberings 31, 52a-61, 71-74, and 97-101 for the H-chain variable region and Kabat numberings 24-34, 51-56, and 89-96 for the L-chain variable region being more preferred.

[0151] Furthermore, modifying amino acid residues also includes randomly modifying amino acids in the aforementioned regions of the variable regions of antibodies that bind to the three antigens, or inserting into the aforementioned regions peptides known to have binding activity toward the desired antigens. Thus, antigen-binding molecules of the present invention can be obtained by selecting, from among the modified antigen-binding molecules, variable regions that can bind to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first and second antigens) but cannot simultaneously bind to these antigens. Examples of peptides known to have binding activity toward the desired antigens include the peptides listed in Table 1 above.

[0152] Whether a variable region is capable of binding to the three antigens but is unable to simultaneously bind to these antigens, and further whether a variable region is capable of simultaneously binding to one or more of the three different antigens (first antigen, second antigen, fourth antigen) when these antigens are present singly and not on a cell, or when two or more of the three different antigens are present on the same cell, but is unable to simultaneously bind to these antigens when they are expressed on different cells, can be confirmed in the same manner using the methods described above.

[0153] Furthermore, the present invention provides methods for producing antigen-binding molecules of the present invention, which comprise antibody variable regions capable of binding to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen), but which do not simultaneously bind to the three antigens (first variable region), the method comprising the step of preparing a library of antigen-binding molecules having diverse amino acid sequences of the first variable region.

[0154] Examples of methods for producing such antigen-binding molecules include production methods comprising the following steps: (i) preparing a library of antigen-binding molecules comprising variable regions in which at least one amino acid in the variable regions of antibodies that bind to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen) has been modified, and in which at least one amino acid in the modified variable regions differs from each other; (ii) selecting, from the prepared library, antigen-binding molecules comprising variable regions that have binding activity to the three antigens but do not simultaneously bind to the three antigens; (iii) culturing host cells containing nucleic acids encoding the variable regions of the antigen-binding molecules selected in step (ii) to express antigen-binding molecules comprising variable regions of antibodies that can bind to the three antigens but do not simultaneously bind to the three antigens; and (iv) recovering the antigen-binding molecule from the host cell culture. Preferred regions for the amino acid modification include the heavy chain variable region. More preferred are solvent-exposed regions and loop regions in the variable region. Among these, the CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numberings 31-35, 50-65, 71-74, and 95-102 for the H chain variable region and Kabat numberings 24-34, 50-56, and 89-97 for the L chain variable region are preferred, with Kabat numberings 31, 52a-61, 71-74, and 97-101 for the H chain variable region and Kabat numberings 24-34, 51-56, and 89-96 for the L chain variable region being more preferred.

[0155] In the present production method, the step (ii) may be the following selective step: (v) selecting, from the prepared library, antigen-binding molecules that have binding activity to the three antigens but that contain variable regions that do not simultaneously bind to the three antigens expressed on different cells.

[0156] The antigen-binding molecule used in step (i) above is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0157] The amino acids to be modified are selected, for example, from the variable regions of an antibody that binds to the three antigens, and such amino acid modification does not result in loss of binding to the antigen.

[0158] The amino acid modifications of the present invention may be used alone or in combination. When multiple compounds are used in combination, the number of compounds to be combined is not particularly limited, and may be, for example, 2 or more and 30 or less, preferably 2 or more and 25 or less, 2 or more and 22 or less, 2 or more and 20 or less, 2 or more and 15 or less, 2 or more and 10 or less, 2 or more and 5 or less, or 2 or more and 3 or less. When multiple amino acid modifications are combined, the amino acid modifications may be made to only one of the heavy chain variable region or light chain variable region of the antibody, or may be appropriately distributed and made to both the heavy chain variable region and the light chain variable region.

[0159] Furthermore, modifying amino acid residues also includes randomly modifying amino acids in the above-mentioned regions of the variable regions of antibodies that bind to the three antigens, or inserting peptides known to have binding activity toward the desired antigens into the above-mentioned regions. Thus, the antigen-binding molecules of the present invention can be obtained by selecting variable regions that can bind to the three antigens but cannot simultaneously bind to these antigens from among the modified antigen-binding molecules. Examples of peptides known to have binding activity toward the desired antigens include the peptides listed in Table 1 above.

[0160] Whether a variable region is capable of binding to the three antigens but is unable to simultaneously bind to these antigens, and further whether a variable region is capable of simultaneously binding to one or more of the three different antigens (first antigen, second antigen, fourth antigen) when these antigens are present singly and not on a cell, or when two or more of the three different antigens are present on the same cell, but is unable to simultaneously bind to these antigens when they are expressed on different cells, can be confirmed in the same manner using the methods described above.

[0161] Furthermore, the present invention also includes antigen-binding molecules produced by the production method. The type and range of amino acid modifications introduced by this method are not particularly limited.

[0162] A non-limiting embodiment of the library of the present invention is a library in which CD3 (in the case of human CD3, the γ chain, δ chain, or ε chain constituting human CD3) is selected as the first antigen, and the library is composed of antigen-binding molecules that bind to CD3 and any second and fourth antigens.

[0163] A preferred embodiment of the library of the present invention is a library mainly composed of a plurality of antigen-binding molecules whose sequences differ from one another, wherein the antigen-binding regions in the antigen-binding molecules are variable regions of antibodies that can bind to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen) but do not simultaneously bind to the three antigens, and wherein one of the third antigens is CD3, and the remaining two antigens are molecules expressed on the surface of T cells or other immune cells.

[0164] When any one of the three antigens is human CD3, the antigen-binding molecule preferably binds to the γ chain, δ chain, or ε chain that constitutes human CD3. Furthermore, the variable regions are preferably variable regions that do not simultaneously bind to the three antigens, each of which is expressed on a different cell. Here, "expressed on different cells" means that the antigens are expressed on different cells, and such a combination of cells may be, for example, cells of the same type, such as a T cell and another T cell, or cells of different types, such as a T cell and an NK cell.

[0165] As used herein, the term "library" refers to a plurality of antigen-binding molecules or a plurality of fusion polypeptides comprising antigen-binding molecules, or a nucleic acid or polynucleotide encoding these sequences. The sequences of the plurality of antigen-binding molecules or the plurality of fusion polypeptides comprising antigen-binding molecules contained in a library are not a single sequence, but rather are antigen-binding molecules or fusion polypeptides comprising antigen-binding molecules with sequences that differ from one another.

[0166] In one embodiment of the present invention, a fusion polypeptide can be prepared between an antigen-binding molecule of the present invention and a heterologous polypeptide. In one embodiment, the fusion polypeptide can be fused to at least a portion of a viral coat protein, for example, a viral coat protein selected from the group consisting of pIII, pVIII, pVII, pIX, Soc, Hoc, gpD, pVI, and mutants thereof.

[0167] In one embodiment, the antigen-binding molecules of the present invention may be ScFv, Fab fragments, F(ab)2, or F(ab')2. In another embodiment, a library is provided that essentially consists of a plurality of fusion polypeptides, each of which has a different sequence from the other, between these antigen-binding molecules and a heterologous polypeptide. Specifically, a library is provided that essentially consists of a plurality of fusion polypeptides, each of which has a different sequence from the other, fused with these antigen-binding molecules and at least a portion of a viral coat protein, for example, selected from the group consisting of pIII, pVIII, pVII, pIX, Soc, Hoc, gpD, pVI, and mutants thereof. The antigen-binding molecules of the present invention may further comprise a dimerization domain. In one embodiment, the dimerization domain may be located between the antibody heavy or light chain variable region and at least a portion of the viral coat protein. This dimerization domain may comprise at least one dimerization sequence and / or a sequence containing one or more cysteine residues. This dimerization domain may preferably be linked to the C-terminus of the heavy chain variable region or constant region. The dimerization domain can have various structures depending on whether the antibody variable region is produced as a fusion polypeptide component with a viral coat protein component (without an amber stop codon after the dimerization domain) or whether the antibody variable region is produced primarily without a viral coat protein component (e.g., with an amber stop codon after the dimerization domain).When the antibody variable region is produced primarily as a fusion polypeptide with a viral coat protein component, bivalent display is achieved by one or more disulfide bonds and / or a single dimerization sequence.

[0168] As used herein, the term "different sequences" in the description of multiple antigen-binding molecules with different sequences means that the sequences of the individual antigen-binding molecules in the library are different from each other. In other words, the number of different sequences in the library reflects the number of independent clones with different sequences in the library, and is sometimes referred to as the "library size." In a typical phage display library, 10 6 From 10 12 By applying known techniques such as ribosome display, the library size can be increased to 10 14 However, the actual number of phage particles used in panning selection of a phage library is usually 10 to 10,000 times larger than the library size. This excess, also called the "library equivalent number," indicates that there may be 10 to 10,000 individual clones with the same amino acid sequence. Therefore, the term "different in sequence from each other" in the present invention means that the sequences of the individual antigen-binding molecules in the library, excluding the library equivalent number, are different from each other, more specifically, there may be 10 or more antigen-binding molecules with different sequences from each other. 6 From 10 14 molecules, preferably 10 7 From 10 12 molecules, more preferably 10 8 From 10 11 , particularly preferably 10 8 From 10 10 It means to exist.

[0169] Furthermore, the term "mainly consisting of" in the description of a library of the present invention consisting mainly of multiple antigen-binding molecules reflects the number of antigen-binding molecules that differ in binding activity to the first and / or second antigen (or the binding activity to the first, second, and / or fourth antigen) among the number of independent clones with different sequences in the library. Specifically, the term "mainly consisting of" refers to ... independent clones with different sequences in the library. Specifically, the term "mainly consisting of" refers to the number of antigen-binding molecules that differ in binding activity to the first, second, and / or fourth antigen (or the binding activity to the first, second, and / or fourth antigen) among the independent clones with different sequences in the library. 4Preferably, the present invention provides a method for detecting antigen-binding molecules that exhibit such binding activity. 5 More preferably, the present invention provides a library containing at least 10 antigen-binding molecules that exhibit such binding activity. 6 Particularly preferably, the present invention provides a library containing at least 10 antigen-binding molecules that exhibit such binding activity. 7 Preferably, the present invention provides a library containing at least 10 antigen-binding molecules that exhibit such binding activity. 8 The present invention provides a library in which antigen-binding molecules are present. Alternatively, this can be suitably expressed as the proportion of antigen-binding molecules that differ in binding activity for the first and / or second antigen (or the binding activity for the first, second, and / or fourth antigen) among the number of independent clones with different sequences in the library. Specifically, the present invention provides a library in which antigen-binding molecules that exhibit such binding activity account for 0.1% to 80%, preferably 0.5% to 60%, more preferably 1% to 40%, even more preferably 2% to 20%, and particularly preferably 4% to 10% of the number of independent clones with different sequences in the library. Fusion polypeptides, polynucleotide molecules, or vectors can also be expressed as the number of molecules or the proportion of all molecules, as described above. Viruses can also be expressed as the number of virus individuals or the proportion of all individuals, as described above.

[0170] The library of the present invention contains at least 1, 10, 100, 1000, 1000, 1000 or 1000 antigen-binding molecules comprising variable regions of antibodies that can bind to a first antigen and a second antigen different from the first antigen but do not simultaneously bind to the first antigen and the second antigen (or at least 1, 10, 100, 1000, 1000 or 1000 antigen-binding molecules comprising variable regions of antibodies that can bind to three different antigens but do not simultaneously bind to the three antigens). 4 molecule, 10 5 molecule, 10 6 molecule, 10 7 Numerator or 10 8It is preferable that a plurality of antigen-binding molecules comprising variable regions of an antibody that can bind to a first antigen and a second antigen different from the first antigen but does not simultaneously bind to the first antigen and the second antigen (or a plurality of antigen-binding molecules comprising variable regions of an antibody that can bind to three different antigens but does not simultaneously bind to the three antigens) is present in a number that is 10 times the number of independent clones with different sequences in the library. -7 %% to 80%, preferably 10 -6 % to 60%, more preferably 10 -5 % to 40%, 10 -4 % to 30%, 10 -3 % to 20%, 10 -2 % to 10%, 10 -1 Provides libraries that are included in % to 1%.

[0171] One embodiment of the "library mainly composed of a plurality of antigen-binding molecules whose sequences differ from one another" of the present invention is a library mainly composed of (i) variable regions of a plurality of antibodies whose sequences differ from one another, or antigen-binding molecules comprising the variable regions, or (ii) nucleic acids encoding the variable regions of a plurality of antibodies whose sequences differ from one another, or antigen-binding molecules comprising the variable regions.

[0172] Here, the multiple antigen-binding molecules with different sequences are antigen-binding molecules containing variable regions in which at least one amino acid alteration has been introduced into the template sequence of the library (e.g., antibody variable regions). Preferred regions for amino acid alteration include heavy chain variable regions and / or light chain variable regions. More preferred are solvent-exposed regions and loop regions in the variable regions. Among these, CDR1, CDR2, CDR3, FR3 regions, and loop regions are preferred. Specifically, Kabat numberings 31 to 35, 50 to 65, 71 to 74, and 95 to 102 for the H-chain variable region and Kabat numberings 24 to 34, 50 to 56, and 89 to 97 for the L-chain variable region are preferred, and Kabat numberings 31, 52a to 61, 71 to 74, and 97 to 101 for the H-chain variable region and Kabat numberings 24 to 34, 51 to 56, and 89 to 96 for the L-chain variable region are more preferred.

[0173] The antigen-binding molecule is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0174] The amino acids to be modified are selected, for example, from the variable region of an antibody that binds to the first or second antigen, and amino acids that do not abolish binding to the antigen upon amino acid modification. Alternatively, the amino acids to be modified are selected from the variable region of an antibody that binds to three different antigens, and amino acids that do not abolish binding to the three antigens upon amino acid modification.

[0175] The amino acid modifications of the present invention may be used alone or in combination. When multiple compounds are used in combination, the number of compounds to be combined is not particularly limited, and may be, for example, 2 or more and 30 or less, preferably 2 or more and 25 or less, 2 or more and 22 or less, 2 or more and 20 or less, 2 or more and 15 or less, 2 or more and 10 or less, 2 or more and 5 or less, or 2 or more and 3 or less. When multiple amino acid modifications are combined, the amino acid modifications may be made to only one of the heavy chain variable region or light chain variable region of the antibody, or may be appropriately distributed and made to both the heavy chain variable region and the light chain variable region.

[0176] Furthermore, modifying amino acid residues also includes randomly modifying amino acids in the variable region of an antibody that binds to the first or second antigen (or the variable region of an antibody that binds to the first, second, or fourth antigen), or inserting a peptide known to have binding activity toward a desired antigen into the above region. Thus, the antigen-binding molecules of the present invention can be obtained by selecting, from the modified antigen-binding molecules, a variable region that can bind to the first and second antigens but cannot simultaneously bind to these antigens (or a variable region that can bind to three different antigens but cannot simultaneously bind to the three antigens). Examples of peptides known to have binding activity toward a desired antigen include the peptides listed in Table 1 above.

[0177] Whether a variable region is capable of binding to a first antigen and a second antigen but not to these antigens simultaneously (or a variable region is capable of binding to three different antigens but not to all three antigens simultaneously), and whether a variable region is capable of simultaneously binding to both a first antigen and a second antigen when one of the first antigen and the second antigen is present on a cell and the other is present alone, when both antigens are present alone, or when both antigens are present on the same cell, but not when they are expressed on different cells (or whether a variable region is capable of simultaneously binding to three different antigens when one or more of the antigens are present alone but not on a cell, or when two or more of the three different antigens are present on the same cell, but not when they are expressed on different cells).

[0178] As used herein, "phage display" refers to a technique in which mutant polypeptides are displayed on the particle surface of phages, e.g., filamentous phage, as proteins fused to at least a portion of a coat protein. The utility of phage display lies in the rapid and efficient selection of sequences that bind to target antigens with high affinity from large libraries of randomized protein variants. Display of peptide and protein libraries on phage has been used to screen millions of polypeptides for specific binding properties. Multivalent phage display methods have been used to display small random peptides and small proteins through fusion with gene III or gene VIII of filamentous phage (Wells and Lowman, Curr. Opin. Struct. Biol. (1992) 3, 355-362, and references therein). In monovalent phage display, a protein or peptide library is fused to gene III or a portion thereof and expressed at low levels in the presence of wild-type gene III protein, such that phage particles display one or zero copies of the fusion protein. Selection is based on intrinsic ligand affinity and uses phagemid vectors, which simplify DNA manipulations, as avidity effects are reduced compared to polyvalent phage (Lowman and Wells, Methods: A Companion to Methods in Enzymology (1991) 3, 205-216).

[0179] A "phagemid" is a plasmid vector containing a bacterial origin of replication, such as ColE1, and a copy of the intergenic region of a bacteriophage. Phagemids can be derived from any known bacteriophage, such as filamentous bacteriophage and lambda-type bacteriophage. Plasmids usually also contain a selectable marker for antibiotic resistance. DNA fragments cloned into these vectors can be propagated as plasmids. When cells transfected with these vectors contain all the genes necessary for phage particle production, the replication mode of the plasmid changes to rolling circle replication, producing a single-stranded copy of the plasmid DNA and packaging phage particles. Phagemids can form infectious or non-infectious phage particles. This term also includes phagemids containing a phage coat protein gene, or a fragment thereof, linked to a gene for a heterologous polypeptide as a gene fusion, such that the heterologous polypeptide is displayed on the surface of the phage particle.

[0180] The term "phage vector" refers to a double-stranded replicative form of a bacteriophage that contains a heterologous gene and is capable of replication. The phage vector has a phage origin of replication that allows phage replication and phage particle formation. The phage is preferably a filamentous bacteriophage, such as M13, f1, fd, Pf3 phage or derivatives thereof, or a lambdoid phage, such as lambda, 21, phi80, phi81, 82, 424, 434, etc. or derivatives thereof.

[0181] An "oligonucleotide" is a short, single- or double-stranded polydeoxynucleotide chemically synthesized by known methods (e.g., solid-phase methods, such as those described in EP 266032, using phosphotriester, phosphite, or phosphoramidite chemistry, or via a deoxynucleotide H-phosphonate intermediate, as described in Froeshler et al. (Nucl. Acids. Res. (1986) 14, 5399-5407)). Other methods include the polymerase chain reaction and other autoprimer methods described below, and oligonucleotide synthesis on a solid support. All of these methods are described in Engels et al. (Agnew. Chem. Int. Ed. Engl. (1989) 28, 716-734). These methods are used if the entire nucleic acid sequence of a gene is known, or if the sequence of a nucleic acid complementary to the coding strand is available. Alternatively, if the amino acid sequence of interest is known, possible nucleic acid sequences can be conveniently inferred using known and preferred coding residues for each amino acid residue. Oligonucleotides can be purified on polyacrylamide gels or molecular sizing columns, or by precipitation methods.

[0182] The terms "fusion protein" and "fusion polypeptide" refer to a polypeptide having two portions covalently linked together, each portion having a distinct property. The property can be a biological property, such as in vitro or in vivo activity. The property can also be a single chemical or physical property, such as binding to an antigen of interest or catalysis of a reaction. The two portions can be linked directly by a single peptide bond or through a peptide linker containing one or more amino acid residues. Typically, the two portions and the linker are in the same reading frame. Preferably, the two portions of the polypeptide are derived from heterologous or different polypeptides.

[0183] The term "coat protein" refers to a protein, at least a portion of which is present on the surface of a virus particle. Functionally, a coat protein is any protein that is associated with a virus particle during virus assembly in a host cell and remains associated with the virus until the virus infects another host cell. A coat protein can be a major coat protein or a minor coat protein. A minor coat protein is a coat protein that is typically present in the outer coat of a virus, preferably present in at least about 5, more preferably at least about 7, and more preferably at least about 10 or more copies of the protein per virion. A major coat protein can be present in tens, hundreds, or even thousands of copies per virion. An example of a major coat protein is the p8 protein of filamentous phage.

[0184] As a non-limiting embodiment of the present invention, the following six methods for preparing a library are exemplified. 1. A method of inserting a peptide (this term is used to include polypeptides and proteins) that binds to a second antigen (or a peptide that binds to a second antigen and a peptide that binds to a fourth antigen) into an antigen-binding molecule that binds to a first antigen 2. A method of preparing a library in which various amino acids appear at positions that can be used to lengthen (extend) the loop in an antigen-binding molecule, and obtaining antigen-binding molecules from the library that have binding activity to an arbitrary second antigen (or an arbitrary second antigen and an arbitrary fourth antigen) using antigen-binding activity as an index. 3. A method in which an amino acid that maintains binding activity to a first antigen is identified from an antigen-binding molecule previously known to bind to a first antigen using an antibody produced by site-directed mutagenesis, and an antigen-binding molecule that has binding activity to an arbitrary second antigen (or an arbitrary second antigen and an arbitrary fourth antigen) is obtained from a library in which the identified amino acid appears, using antigen-binding activity as an index. 4. A method in which, in the method of 3, an antibody library is further constructed in which various amino acids appear at positions that allow the loop in the antigen-binding molecule to be modified (extended) to be longer, and antigen-binding molecules having binding activity to any second antigen (or any second antigen and any fourth antigen) are obtained from the library using antigen-binding activity as an index. 5. In the method of 1, 2, 3, or 4, a glycosylation sequence (e.g., NxS, NxT, where x is any amino acid other than P) is modified to include a glycosylation sequence that is recognized by a glycosylation receptor (e.g., a high-mannose type glycosylation sequence is added, which is recognized by the high-mannose receptor. High-mannose type glycosylation sequences can be obtained by adding kifunensine during antibody expression (MAbs. 2012 Jul-Aug;4(4):475-87)). 6. A method in which, in method 1, 2, 3, or 4, Cys, Lys, or a non-natural amino acid is inserted or substituted into the loop region or a region that could be modified with various amino acids to covalently add a domain that binds to a second antigen (or a domain that binds to a second antigen and a domain that binds to a fourth antigen) (a method typified by antibody drug conjugates, in which Cys, Lys, or a non-natural amino acid is covalently bonded (mAbs 6:1, 34-45; January / February 2014, WO2009 / 134891A2, Bioconjug Chem. 2014 Feb 19;25(2):351-61)). In the six library construction methods exemplified above, the sites for amino acid substitution in the antigen-binding molecules or peptide insertion into the antigen-binding molecules are preferably the Fab or variable region of the antigen-binding molecules. Preferred regions include solvent-exposed regions and loop regions in the variable regions. Of these, CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numberings 31 to 35, 50 to 65, 71 to 74, and 95 to 102 for the H-chain variable region and Kabat numberings 24 to 34, 50 to 56, and 89 to 97 for the L-chain variable region are preferred, with Kabat numberings 31, 52a to 61, 71 to 74, and 97 to 101 for the H-chain variable region and Kabat numberings 24 to 34, 51 to 56, and 89 to 96 for the L-chain variable region being more preferred.

[0185] One embodiment of the method for inserting a peptide that binds to a second antigen (or a peptide that binds to a second antigen and a peptide that binds to a fourth antigen) into an antigen-binding molecule that binds to a first antigen, as described in Method 1 above, includes inserting G-CSF as exemplified in Angew Chem Int Ed Engl. 2013 Aug 5;52(32):8295-8. In another embodiment, the inserted peptide can be obtained from a peptide-display library, but it is also possible to use the whole or part of a naturally occurring protein.

[0186] In one aspect of the present invention, there is provided a method for producing a library, comprising the steps of: The following steps (a) and (b): (a) using a variable region sequence of an antibody that binds to a first antigen as a template sequence for a library, and identifying amino acid alterations that satisfy any one or more of the following (i) to (iii): (i) The modification does not substantially change the binding ability to the first antigen; (ii) the modification does not substantially alter the ECM binding ability; and (iii) Insertion of a peptide consisting of 1 to 25 amino acid residues into the CDR1, CDR2, CDR3, or FR3 region of the heavy chain variable region or the light chain variable region; and (b) designing a library containing nucleic acids encoding the template sequence and nucleic acids encoding variable regions of the template sequence that have different sequences and that have one or more amino acid alterations identified in step (a) above; A method for producing a library, comprising: This production method can be used to produce a library consisting mainly of a plurality of antigen-binding molecules with different sequences, wherein the antigen-binding regions in the antigen-binding molecules are variable regions of antibodies that can bind to a first antigen and a second antigen different from the first antigen but do not simultaneously bind to the first antigen and the second antigen, and wherein one of the first and second antigens is CD3, and the other antigen is a molecule expressed on the surface of T cells or other immune cells. This production method can also be used to produce a library consisting mainly of a plurality of antigen-binding molecules with different sequences, wherein the antigen-binding regions in the antigen-binding molecules are variable regions of antibodies that can bind to three different antigens (a first antigen, a second antigen different from the first antigen, and a fourth antigen different from the first and second antigens) but do not simultaneously bind to the three antigens, and wherein one of the three antigens is CD3, and the remaining two antigens are molecules expressed on the surface of T cells or other immune cells.

[0187] In the present invention, the term "template sequence (for a library)" refers to an antibody amino acid sequence (e.g., an antibody variable region sequence or CDR sequence) that serves as a template sequence for preparing a library. A library of antigen-binding molecules can be prepared by using this sequence to identify modified amino acids that can be compiled into a library (e.g., as can be identified by the above-mentioned aspects (a)(i), (ii), and (iii)). For example, when the first antigen is CD3, those skilled in the art can appropriately select the variable region sequence of an antibody that binds to the CD3 antigen as the template sequence for the library.

[0188] Step (a)(i) above: To identify modified amino acids that do not substantially alter the binding ability to CD3 (to identify amino acids that maintain the binding activity to the first antigen, CD3 (in the case of human CD3, the γ chain, δ chain, or ε chain that constitutes human CD3)), for example, amino acids can be modified at sites thought to be involved in antigen binding to produce single-amino-acid-modified antibodies and then evaluated. CD3 binding of single-amino-acid-modified antibodies can be evaluated by any method known to those skilled in the art, including, for example, ELISA, FACS (fluorescence activated cell sorting), ALPHA screen (Amplified Luminescent Proximity Homogeneous Assay), and the BIACORE method, which utilizes the surface plasmon resonance (SPR) phenomenon.

[0189] To identify modified amino acids that do not substantially alter the binding ability to CD3 (to identify amino acids that maintain binding activity to CD3) in the above step (a)(i), the ratio of the binding amount of various modified antibodies to that of the antibody before modification can be used. That is, when the binding amount of the antibody before modification is X and the binding amount of the single amino acid modified antibody is Y, the value Z (ratio of binding amount) = Y / X can be used. When Z (ratio of binding amount) is 0.5 or more, 0.6 or more, 0.7 or more, 0.8 or more, 0.9 or more, preferably 0.8 or more, it can be considered that binding to the antibody before modification is maintained. An antibody library can be constructed so that amino acids that maintain these bindings are present.

[0190] Extracellular matrix (ECM) is an extracellular component present in various parts of the body. Therefore, it is known that antibodies that strongly bind to ECM exhibit poor blood kinetics (short half-life) ( WO2012093704A1 ). Therefore, it is preferable to select amino acids that do not enhance ECM binding from antibody libraries (i.e., identify modified amino acids that do not substantially alter ECM binding ability).

[0191] To identify modified amino acids that do not substantially alter ECM binding ability (select amino acids that do not enhance ECM binding) in the above step (a)(ii), ECM binding can be evaluated according to the method of Reference Example 2, for example, and the ECM binding value (ECL response) of each variant divided by the antibody ECM binding value of MRA (H chain SEQ ID NO: 57, L chain SEQ ID NO: 58) can be used. Taking into account the effect of enhancing ECM binding through multiple modifications, values of 5-fold, 6-fold, 7-fold, 8-fold, 9-fold, 10-fold, 15-fold, 20-fold, or even 30-fold can be used as effective values, with values of up to 10-fold being preferred for use in libraries. An antibody library can be constructed so that amino acids selected in this way are present.

[0192] Although not limited to this, when the peptide insertion into CDR3 is 6 amino acids, the presence of many amino acids with positive charges in their side chains in the extended loop of CDR3 enhances binding to ECM, and therefore it is preferable that no amino acids with three or more positive charges in their side chains appear in the loop.

[0193] To enhance the diversity of the library of the present invention, the library can be prepared by the above step (a)(iii): identifying modified amino acids to be inserted into the variable region (inserting peptides into the variable region). Preferred regions for peptide insertion include solvent-exposed regions and loop regions in the variable region. Among these, CDR1, CDR2, CDR3, FR3, and loop regions are preferred. Specifically, Kabat numberings 31-35, 50-65, 71-74, and 95-102 in the H-chain variable region and Kabat numberings 24-34, 50-56, and 89-97 in the L-chain variable region are preferred, with Kabat numberings 31, 52a-61, 71-74, and 97-101 in the H-chain variable region and Kabat numberings 24-34, 51-56, and 89-96 in the L-chain variable region being more preferred. The Kabat numberings 99-100 in the H-chain variable region are even more preferred. Furthermore, when modifying amino acids, amino acids that increase antigen-binding activity may also be introduced.

[0194] In a non-limiting embodiment of the present invention, the length of the inserted peptide can be 1 to 3 amino acids, 4 to 6 amino acids, 7 to 9 amino acids, 10 to 12 amino acids, 13 to 15 amino acids, 15 to 20 amino acids, or 21 to 25 amino acids, and is preferably 1 to 3 amino acids, 4 to 6 amino acids, or 7 to 9 amino acids.

[0195] The insertion site and length of peptides to enhance library diversity can be investigated by preparing molecules with inserted peptides and evaluating the CD3 binding of the molecules. Methods known to those skilled in the art can be appropriately selected for evaluation, and examples include ELISA, FACS (fluorescence activated cell sorting), ALPHA screen (Amplified Luminescent Proximity Homogeneous Assay), and the BIACORE method using surface plasmon resonance (SPR) phenomenon.

[0196] In one non-limiting embodiment of the present invention, an antibody library for obtaining antibodies that bind to CD3 and a second antigen (or CD3, a second antigen, and a fourth antigen) can be designed as follows. Step 1: Select amino acids that retain CD3 binding ability (CD3 binding ability must be 80% or more of that of the unmodified antibody) For example, a library can be created to obtain antibodies that bind to CD3 and a second antigen (or CD3, a second antigen, and a fourth antigen) so that the amino acids selected in step 1 appear.

[0197] In one non-limiting embodiment of the present invention, an antibody library for obtaining antibodies that bind to CD3 and a second antigen (or CD3, a second antigen, and a fourth antigen) can be designed as follows. Step 1: Select amino acids that retain CD3 binding ability (CD3 binding ability must be 80% or more of that of the unmodified antibody) Step 2: Insert amino acids between 99-100 (Kabat numbering) of the heavy chain CDR3 For example, by inserting amino acids into the CDR3 region in step 2 in addition to step 1, the diversity of the library can be increased, and a library can be prepared for obtaining antibodies that bind to CD3 and a second antigen (or CD3, a second antigen, and a fourth antigen).

[0198] In one non-limiting embodiment of the present invention, an antibody library for obtaining antibodies that bind to CD3 and a second antigen (or CD3, a second antigen, and a fourth antigen) can be designed as follows. Step 1: Select amino acids that retain CD3 binding ability (CD3 binding ability must be 80% or more of that of the unmodified antibody) Step 2: Select amino acids whose ECM binding is within 10-fold of that before modification compared to MRA Step 3: Insert amino acids between positions 99-100 (Kabat numbering) of the heavy chain CDR3 For example, adding step 2 to steps 1 and 3 allows amino acids that do not enhance ECM binding to be selected from those appearing in the library, but the method is not limited to this. Furthermore, even when a library is designed without going through step 2, it is possible to measure and evaluate the ECM binding of antigen-binding molecules obtained from the library.

[0199] In one non-limiting embodiment of the present invention, when the VH region CE115HA000 (SEQ ID NO: 52) is used as the template sequence for a CD3 (CD3ε)-binding antibody, examples of modified amino acids used in library design include one or more of the amino acids at positions 11, 31, 52a, 52b, 52c, 53, 54, 56, 57, 61, 72, 78, 98, 99, 100, 100a, 100b, 100c, 100d, 100e, 100f, 100g, and 101 in the heavy chain variable region according to the Kabat numbering system. Preferred, but not limited to, is the above-mentioned library using a template sequence (SEQ ID NO: 94) in which the VH region CE115HA000 (SEQ ID NO: 52) has been modified by the amino acid alterations V11L / L78I.Further preferred, but not limited to, is the above-mentioned library using a template sequence (SEQ ID NO: 95) in which the VH region CE115HA000 (SEQ ID NO: 52) has been modified by the amino acid alterations V11L / A52aD / L78I, or the above-mentioned library using a template sequence (SEQ ID NO: 96) in which the CDR3 sequence has been extended. For example, when the heavy chain variable region sequence set forth in SEQ ID NO: 96 is used as the template sequence, examples of modified amino acids used in library design include one or more of the following amino acids in the heavy chain variable region: 31, 52b, 52c, 53, 54, 56, 57, 61, 98, 99, 100, 100a, 100b, 100c, 100d, 100e, 100f, and 100g (Kabat numbering). In another embodiment, rCE115Hchain (SEQ ID NO: 97) can also be used as the template sequence for a CD3-binding antibody.

[0200] In one non-limiting embodiment of the present invention, when the VL region GLS3000 (SEQ ID NO: 53) is used as the template sequence for a CD3 (CD3ε)-binding antibody, examples of modified amino acids used in library design include one or more of the amino acids at positions 24, 25, 26, 27, 27a, 27b, 27c, 27e, 30, 31, 33, 34, 51, 52, 53, 54, 55, 56, 74, 77, 89, 90, 92, 93, 94, and 96 (Kabat numbering) contained in the light chain variable region. In another embodiment, rCE115Lchain (SEQ ID NO: 98) can also be used as a template sequence for a CD3-binding antibody.

[0201] In one non-limiting embodiment of the present invention, a heavy chain variable region sequence comprising heavy chain CDR1 (SEQ ID NO: 99), heavy chain CDR2 (SEQ ID NO: 100), and heavy chain CDR3 (SEQ ID NO: 101) can be used as a template sequence for a CD3-binding antibody. In yet another embodiment, a heavy chain CDR3 (SEQ ID NO: 102) with an inserted 6 amino acid residues can be used. Furthermore, in another embodiment, a light chain variable region sequence including light chain CDR1 (SEQ ID NO: 103), light chain CDR2 (SEQ ID NO: 104), and light chain CDR3 (SEQ ID NO: 105) can also be used as a template sequence for a CD3-binding antibody. As the framework amino acid sequences contained in the variable regions, currently known completely human framework region sequences included on websites such as IMGT (http: / / www.imgt.org / textes / IMGTrepertoire / ) can be appropriately used as germline sequences contained in the antigen-binding molecules of the present invention, but are not limited thereto.

[0202] In the present invention, designing a library includes, but is not limited to, designing a library containing multiple variants of antigen-binding molecules that contain variable regions in which amino acids at specific sites have been modified to desired amino acids, using known library technologies such as NNK and TRIM Library (Gonzalez-Muñoz A et al. MAbs 2012, Lee CV et al. J Mol Biol. 2004, Knappik A. et al. J Mol Biol. 2000, Tiller T et al. MAbs 2013).

[0203] In the present invention, "one or more amino acids" is not particularly limited to the number of amino acids, and may be two or more types of amino acids, five or more types of amino acids, ten or more types of amino acids, fifteen or more types of amino acids, or twenty types of amino acids.

[0204] In one embodiment, the present invention provides a method for selecting a variable region with enhanced binding to a first antigen, the method comprising the steps of: The following steps (a) to (c): (a) contacting the library of the present invention with a first antigen; (b) recovering the antigen-binding molecule bound to the first antigen in step (a); and (c) a method for selecting a variable region with enhanced binding to a first antigen, the method comprising the step of selecting, from the population of antigen-binding molecules that bind to the first antigen in step (b), antigen-binding molecules that contain a variable region with enhanced binding to the first antigen.

[0205] In this embodiment, the first antigen is preferably CD3, but is not particularly limited thereto.

[0206] The antigen-binding molecule used in the above selection method is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0207] In step (c) above, antigen-binding molecules comprising a variable region with enhanced binding to the first antigen can be selected by measuring binding to the first antigen using methods known to those skilled in the art, such as ELISA, FACS (fluorescence activated cell sorting), ALPHA screen (Amplified Luminescent Proximity Homogeneous Assay), or the BIACORE method utilizing surface plasmon resonance (SPR). Specifically, by comparing the binding activity of an antigen-binding molecule comprising a variable region sequence selected as a template sequence for the library with the binding activity of the antigen-binding molecule bound to the first antigen recovered in step (b), if an antigen-binding molecule is found that has higher binding activity than the antigen-binding molecule comprising the variable region sequence of the template, the antigen-binding molecule can be determined to be an antigen-binding molecule comprising a variable region with enhanced binding to the first antigen.

[0208] The library used in the above step (a) is preferably a library produced by identifying amino acids that retain 80% or more of the CD3 binding ability, but is not limited to this.

[0209] One aspect of the present invention provides a method for producing an antigen-binding molecule, comprising the steps of: A method for producing an antigen-binding molecule comprising a variable region capable of binding to a first antigen, which is CD3, and a second antigen different from the first antigen, but not simultaneously binding to the first antigen and the second antigen, the method comprising the following steps (a) to (c): (a) contacting the library of the present invention with a second antigen; (b) recovering the antigen-binding molecule bound to the second antigen in step (a); and (c) selecting, from the population of antigen-binding molecules collected in step (b), antigen-binding molecules comprising variable regions that do not simultaneously bind to the first antigen and the second antigen.

[0210] Another aspect of the present invention provides a method for producing an antigen-binding molecule, comprising the steps of: A method for producing an antigen-binding molecule comprising a variable region capable of binding to a first antigen, which is CD3, a second antigen different from the first antigen, and a fourth antigen different from the first antigen and the second antigen, but not simultaneously binding to the three antigens, the method comprising the following steps (a) to (f): (a) contacting the library of the present invention with a second antigen; (b) recovering the antigen-binding molecule bound to the second antigen in step (a); (c) selecting, from the population of antigen-binding molecules collected in step (b), antigen-binding molecules comprising a variable region that does not simultaneously bind to the first antigen and the second antigen; (d) contacting the population of antigen-binding molecules recovered in step (c) with a fourth antigen; (e) recovering the antigen-binding molecule bound to the fourth antigen in step (d); and (f) selecting, from the population of antigen-binding molecules recovered in step (e), antigen-binding molecules comprising variable regions that do not simultaneously bind to the first antigen, the second antigen, and the fourth antigen.

[0211] In these embodiments, the first antigen is preferably CD3, and the second and fourth antigens are preferably molecules expressed on the surface of T cells or other immune cells. Furthermore, the second antigen is preferably FcγR, TLR, lectin, IgA, an immune checkpoint molecule, a TNF superfamily molecule, a TNFR superfamily molecule, or an NK receptor molecule.

[0212] The antigen-binding molecule is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0213] In the present production method, step (c) may be the following optional step: (d) selecting, from the population of antigen-binding molecules recovered in step (b), antigen-binding molecules that have binding activity to the first antigen and the second antigen but contain variable regions that do not simultaneously bind to the first antigen and the second antigen, each of which is expressed on different cells. In the present manufacturing method, step (f) may be the following optional step: (g) selecting, from the population of antigen-binding molecules recovered in step (c), antigen-binding molecules that have binding activity to the first antigen, the second antigen, and the fourth antigen but contain variable regions that do not simultaneously bind to the first antigen, the second antigen, and the fourth antigen, each of which is expressed on different cells.

[0214] One aspect of the present invention provides a method for producing a bispecific antibody comprising a shared light chain variable region, comprising the following steps: A method for producing a bispecific antibody, comprising the steps of: (a) preparing a library of a plurality of antigen-binding molecules having different sequences from one another; wherein the antigen-binding regions of the antigen-binding molecules are variable regions consisting of a template sequence or variable regions having at least one amino acid modification in the template sequence; and the variable regions are variable regions of antibodies that can bind to a first antigen and a second antigen different from the first antigen, but do not simultaneously bind to the first antigen and the second antigen; and the method comprises the steps of: (a) selecting a template sequence from the library of the present invention as a variable region that binds to a first antigen; (b) selecting a variable region that binds to the second antigen but not the first antigen as a variable region that binds to the second antigen, the step comprising the following steps (i) to (iv): (i) contacting the library of the present invention with a desired second antigen; (ii) recovering the antigen-binding molecule bound to the second antigen in step (i); (iii) contacting the population of antigen-binding molecules recovered in step (ii) with a first antigen; (iv) selecting antigen-binding molecules that do not bind to the first antigen in step (iii); and (c) producing a bispecific antibody comprising a variable region that binds to the first antigen selected in step (a) and a variable region that binds to the second antigen selected in step (b).

[0215] In this embodiment, it is preferable that the first antigen is CD3 and the second antigen is a molecule that is specifically expressed in cancer tissue.

[0216] By producing a bispecific antibody using this embodiment, it is possible to efficiently obtain, for example, a shared light chain variable region of the bispecific antibody that comprises a variable region that binds to CD3 (the first antigen) and a variable region that binds to a molecule that is specifically expressed in cancer tissue (the second antigen).

[0217] The antigen-binding molecule used in step (b) above is not particularly limited as long as it contains an antibody variable region, and may be an antibody fragment such as Fv, Fab, or Fab', or an antibody containing an Fc region.

[0218] Regarding the display of fusion polypeptides, fusion polypeptides of the variable regions of antigen-binding molecules can be displayed in various forms on the surface of cells, viruses, or phagemid particles. These forms include single-chain Fv fragments (scFv), F(ab) fragments, and multivalent forms of these fragments. Multivalent forms are preferably dimers of ScFv, Fab, or F(ab'), which are referred to herein as (ScFv)2, F(ab)2, and F(ab')2, respectively. One reason why multivalent display is preferred is that it allows the identification of clones that normally have low affinity, or that they have multiple antigen-binding sites, which allows for more efficient selection of rare clones during the selection process.

[0219] Methods for displaying fusion polypeptides containing antibody fragments on the surface of bacteriophage are known in the art and are described, for example, in WO1992001047 and herein. Other related methods are described in WO1992020791, WO1993006213, WO1993011236, and WO1993019172, and those skilled in the art can use these methods as appropriate. Other known publications (H.R. Hoogenboom & G. Winter (1992) J. Mol. Biol. 227, 381-388, WO1993006213, and WO1993011236) demonstrate the identification of antibodies against various antigens displayed on the phage surface using artificially rearranged variable region gene repertoires.

[0220] When a vector is constructed for display in the form of an scFv, nucleic acid sequences encoding the light and heavy chain variable regions of the antigen-binding molecule are contained in the vector. Generally, the nucleic acid sequence encoding the heavy chain variable region of the antigen-binding molecule is fused to a viral coat protein component. The nucleic acid sequence encoding the light chain variable region of the antigen-binding molecule is linked to the heavy chain variable region of the antigen-binding molecule by a nucleic acid sequence encoding a peptide linker. The peptide linker generally contains about 5 to 15 amino acids. Optionally, other sequences encoding labels useful for purification or detection, for example, can be fused to the 3' end of the nucleic acid sequence encoding either or both of the light chain variable region of the antigen-binding molecule or the heavy chain variable region of the antigen-binding molecule.

[0221] When a vector is constructed for display in the form of F(ab), the vector contains nucleic acid sequences encoding the variable and constant regions of the antigen-binding molecule. The nucleic acid encoding the light chain variable region is fused to a nucleic acid sequence encoding the light chain constant region. The nucleic acid sequence encoding the heavy chain variable region of the antigen-binding molecule is fused to a nucleic acid sequence encoding the heavy chain constant CH1 region. Generally, the nucleic acid sequence encoding the heavy chain variable and constant regions is fused to a nucleic acid sequence encoding all or part of a viral coat protein. The heavy chain variable and constant regions are preferably expressed as a fusion with at least a portion of the viral coat protein, while the light chain variable and constant regions are expressed separately from the heavy chain viral coat fusion protein. The heavy and light chains are bound to each other, but this binding can be covalent or noncovalent. Optionally, other sequences encoding polypeptide tags useful for purification or detection, for example, can be fused to either or both the 3' end of the nucleic acid sequence encoding the light chain constant region of the antigen-binding molecule or the 3' end of the nucleic acid sequence encoding the heavy chain constant region of the antigen-binding molecule.

[0222] Regarding the introduction of the vector into a host cell, the vector constructed as described above is introduced into the host cell for amplification and / or expression. The vector can be introduced into the host cell by known transformation methods, including electroporation, calcium phosphate precipitation, etc. If the vector is an infectious particle such as a virus, the vector itself invades the host cell. The fusion protein is displayed on the surface of the phage particle by transfecting the host cell with a replicable expression vector into which a polynucleotide encoding the fusion protein has been inserted and producing phage particles by known techniques.

[0223] Replicable expression vectors can be introduced into host cells using a variety of methods. In one non-limiting embodiment, vectors can be introduced into cells using electroporation, as described in WO2000106717. Cells are optionally cultured in standard culture medium for approximately 6 to 48 hours (or until the OD at 600 nm reaches 0.6 to 0.8) at 37°C, and the culture medium is then centrifuged (e.g., decanted) to remove the culture supernatant. In the initial stage of purification, the cell pellet is preferably resuspended in a buffer (e.g., 1.0 mM HEPES, pH 7.4). The supernatant is then removed from the suspension by another centrifugation. The resulting cell pellet is resuspended in, for example, 5-20% V / V diluted glycerol. The supernatant is removed from the suspension by another centrifugation to obtain a cell pellet. The cell pellet is resuspended in water or diluted glycerol, and the final cell concentration is adjusted to the desired concentration based on the measured cell concentration of the resulting suspension.

[0224] For example, a preferred recipient cell is the electroporation-competent E. coli strain SS320 (Sidhu et al. (Methods Enzymol. (2000) 328, 333-363)). E. coli strain SS320 was prepared by mating MC1061 cells with XL1-BLUE cells under conditions sufficient to transfer the fertility episome (F' plasmid) or XL1-BLUE to the MC1061 cells. E. coli strain SS320, deposited with the ATCC (10801 University Boulevard, Manassas, Virginia), has been assigned accession number 98795. Any F' episome that allows phage replication in this strain can be used in the present invention. Suitable episomes are available from strains deposited with the ATCC or are commercially available (TG1, CJ236, CSH18, DHF', ER2738, JM101, JM103, JM105, JM107, JM109, JM110, KS1000, XL1-BLUE, 71-18, etc.).

[0225] Using a higher DNA concentration (approximately 10-fold) during electroporation improves transformation efficiency, increasing the amount of DNA transformed into host cells. Using a higher cell concentration also increases efficiency (approximately 10-fold). The increased amount of transferred DNA allows for the creation of libraries with greater diversity and a larger number of independent clones with different sequences. Transformants are typically selected by their ability to grow on antibiotic-containing media.

[0226] The present invention also provides nucleic acids encoding the antigen-binding molecules of the present invention. The nucleic acids of the present invention may be in any form, such as DNA or RNA.

[0227] Furthermore, the present invention provides a vector comprising the nucleic acid of the present invention. The type of vector can be appropriately selected by those skilled in the art depending on the host cell into which the vector is to be introduced, and for example, the vectors described above can be used.

[0228] Furthermore, the present invention relates to a host cell transformed with the vector of the present invention. The host cell can be appropriately selected by those skilled in the art, and for example, the host cells described above can be used.

[0229] The present invention also provides pharmaceutical compositions comprising the antigen-binding molecules of the present invention and a pharmaceutically acceptable carrier. The pharmaceutical compositions of the present invention can be formulated by known methods by incorporating a pharmaceutically acceptable carrier in addition to the antigen-binding molecules of the present invention. For example, they can be used parenterally in the form of a sterile solution or suspension injection in water or other pharmaceutically acceptable solutions. For example, they can be formulated by appropriately combining them with pharmacologically acceptable carriers or vehicles, specifically, sterile water, physiological saline, vegetable oils, emulsifiers, suspending agents, surfactants, stabilizers, flavoring agents, excipients, vehicles, preservatives, binders, etc., and blending them into a unit dosage form required for generally accepted pharmaceutical practice. Specific examples of carriers include light anhydrous silicic acid, lactose, crystalline cellulose, mannitol, starch, carmellose calcium, carmellose sodium, hydroxypropyl cellulose, hydroxypropyl methylcellulose, polyvinyl acetal diethylaminoacetate, polyvinylpyrrolidone, gelatin, medium-chain fatty acid triglyceride, polyoxyethylene hydrogenated castor oil 60, sucrose, carboxymethyl cellulose, corn starch, inorganic salts, etc. The amount of the active ingredient in these preparations is such that an appropriate dose within the specified range is obtained.

[0230] Sterile compositions for injection can be formulated according to common pharmaceutical practice using a vehicle such as distilled water for injection. Examples of aqueous solutions for injection include isotonic solutions containing physiological saline, glucose or other auxiliary agents, such as D-sorbitol, D-mannose, D-mannitol, and sodium chloride, which may be used in combination with an appropriate solubilizing agent, such as alcohol, specifically ethanol, polyalcohols such as propylene glycol and polyethylene glycol, and nonionic surfactants such as polysorbate 80™ and HCO-50.

[0231] Examples of oily liquids include sesame oil and soybean oil, which may be used in combination with benzyl benzoate or benzyl alcohol as a solubilizing agent. Buffers such as phosphate buffer, sodium acetate buffer, soothing agents such as procaine hydrochloride, stabilizers such as benzyl alcohol, phenol, and antioxidants may also be added. The prepared injection solution is usually filled into appropriate ampoules. Administration is preferably parenteral, specifically, injections, nasal administration, pulmonary administration, transdermal administration, and the like. Examples of injections include intravenous, intramuscular, intraperitoneal, and subcutaneous injections, which can be administered systemically or locally.

[0232] Furthermore, the administration method can be selected appropriately depending on the patient's age and symptoms. The dosage of a pharmaceutical composition containing a polypeptide or a polynucleotide encoding the polypeptide can be selected, for example, from the range of 0.0001 mg to 1,000 mg per kg of body weight per administration. Alternatively, the dosage can be selected, for example, from the range of 0.001 to 100,000 mg / body weight per patient, but is not necessarily limited to these values. The dosage and administration method vary depending on the patient's weight, age, symptoms, etc., but can be selected appropriately by those skilled in the art.

[0233] The present invention also provides methods for treating cancer, which comprise a step of administering the antigen-binding molecules of the present invention; antigen-binding molecules of the present invention for use in cancer treatment; use of the antigen-binding molecules of the present invention in the manufacture of cancer therapeutic agents; and processes for manufacturing cancer therapeutic agents, which comprise a step of using the antigen-binding molecules of the present invention.

[0234] The correspondence between the three-letter and one-letter amino acid codes used in this specification is as follows: Alanine: Ala: A, Arginine: Arg: R, Asparagine: Asn: N, Aspartic acid: Asp: D, Cysteine: Cys: C, Glutamine: Gln: Q, Glutamic acid: Glu: E, Glycine: Gly: G, Histidine: His: H, Isoleucine: Ile: I, Leucine: Leu: L, Lysine: Lys: K, Methionine: Met: M, Phenylalanine: Phe: F, Proline: Pro: P, Serine: Ser: S, Threonine: Thr: T, Tryptophan: Trp: W, Tyrosine: Tyr: Y, Valine: Val: V

[0235] It will be understood by those skilled in the art that any combination of one or more aspects described in this specification is also included in the present invention, as long as there is no technical contradiction based on the technical common sense of those skilled in the art.

[0236] All prior art documents cited in this specification are hereby incorporated by reference.

[0237] The present invention is further illustrated by, but not limited to, the following examples. [Example]

[0238] Example 1: Concept of an engineered immunoglobulin variable (Fab) region that binds to CD3 (first antigen) and another antigen (second antigen) but does not simultaneously bind to CD3 (first antigen) and the other antigen (second antigen) on different cells When immunoglobulins simultaneously bind to two or more activating FcγR molecules, or when they simultaneously bind to different antigens and activating FcγRs, cross-linking of the activating FcγRs may result in the transduction of ITAM signals from the FcγRs, leading to the activation of immune cells. As mentioned above, one IgG antibody molecule can only bind to one FcγR molecule, so only in the presence of antigens can two or more activating FcγR molecules be cross-linked, resulting in the activation of immune cells.

[0239] Furthermore, when an IgG antibody binds to an antigen via its variable region (Fab), it can simultaneously bind to one FcγR molecule via its Fc region, resulting in cross-linking between the antigen-expressing cell and an FcγR-expressing cell. Depending on the antigen-expressing cell, cross-linking of the antigen with FcγR may not be desirable. Specifically, for example, when the antigen is CD3, cross-linking of T cells with FcγR-expressing cells may result in immune activation, such as cytokine release (J. Immunol. (1999) Aug 1, 163(3), 1246-52). In such cases, alterations to the Fc region can eliminate FcγR-binding activity and prevent cross-linking of the antigen with FcγR (Advanced Drug Delivery Reviews (2006) 58, 640-656). Similarly, when the antigen of an IgG antibody is a TNFR superfamily molecule such as CD40, OX40, or CD27, or CD3 and TLRs such as TLR2, 4, 8, or 9, cross-linking via FcγR results in systemic immune activation, so simultaneous binding to these molecules expressed on different cells is undesirable.

[0240] On the other hand, while conventional multispecific antibodies can simultaneously bind to multiple antigens, simultaneous binding to multiple antigens may not be desirable depending on the antigen combination. For example, integrin αvβ3, a known adhesion molecule, is expressed on many cancer cells and on blood vessels surrounding tumors, making it a useful targeting molecule for tumor targeting (R. Haubner, PLoS Med., 2, e70(2005)). However, it is also known to be expressed on various normal cells (Thromb Haemost. 1998 Nov;80(5):726-34.). Therefore, if a multispecific antibody simultaneously binds to both CD3 and integrin αvβ3, it is possible that normal cells will be damaged by the potent cytotoxic activity of T cells.

[0241] Therefore, as a method for controlling such undesirable cross-linking reactions, a variable region (Fab) was devised in which one portion of the variable region binds to a first antigen and another portion of the variable region that is not involved in the first binding binds to a second antigen (Dual Binding Fab) (Figure 1). In this case, as shown in Figure 1, if two adjacent portions of a single variable region (Fab) are essential for binding to each antigen, binding of the first antigen inhibits binding of the second antigen, and similarly, binding of the second antigen inhibits binding of the first antigen. Therefore, it was thought that an improved antibody with such Dual Binding Fab properties would not be able to simultaneously bind to both the first and second antigens, and therefore would not undergo cross-linking reactions between the first and second antigens (Figure 2). Furthermore, if the first and second antigens are not expressed on the cell membrane as soluble proteins, or if both are present on the same cell, they can simultaneously bind to both the first and second antigens. However, if they are expressed on different cells, they do not simultaneously bind and do not cross-link the two cells. This is also considered a dual-binding Fab (Figure 3). On the other hand, the antigen that binds to the other variable (Fab) region (third antigen) is thought to cross-link with the first antigen (Figure 4) and also with the second antigen (Figure 5). The constant region of the antibody can be an Fc region that binds to FcγR, or an Fc region with reduced FcγR binding activity can be used. By utilizing the properties of such Dual Binding Fab, it is possible to further enhance cancer specificity by further endowing it with the ability to target integrins in cancer tissues in a technique that damages cancer cells expressing cancer antigens by redirecting T cells via antibodies.

[0242] Specifically, if the variable (Fab) region can be modified to form a dual-binding Fab, and the following properties can be imparted, it will be possible to create antibodies with the effects shown in Figure 1. 1. Having binding activity to a first antigen 2. Having binding activity to a second antigen 3. Does not bind to the first antigen and the second antigen simultaneously The phrase "not simultaneously binding to the first antigen and the second antigen" includes not bridging two cells, one expressing the first antigen and the other expressing the second antigen, or not simultaneously binding to the first antigen and the second antigen expressed on separate cells, and also includes cases where the first antigen and the second antigen are not expressed on the cell membrane as soluble proteins, or where, when both are present on the same cell, the antigen can simultaneously bind to both the first antigen and the second antigen, but cannot simultaneously bind to them when they are expressed on different cells.

[0243] Similarly, if the variable (Fab) region can be modified to form a Dual Binding Fab, and the following properties can be imparted, it will be possible to create antibodies with the effects shown in Figure 6. 1. Possess binding activity to a first antigen on T cells 2. Having binding activity to a second antigen on antigen-presenting cells 3. Does not bind to the first antigen and the second antigen simultaneously

[0244] Example 2: Preparation of anti-human, cynomolgus CD3ε antibody CE115 (2-1) Hybridoma production using rats immunized with human CD3 and cynomolgus monkey CD3 expressing cells SD rats (female, 6 weeks old at the start of immunization, Charles River Japan) were immunized with human CD3εγ or cynomolgus monkey CD3εγ-expressing Ba / F3 cells as follows. The first immunization was designated as day 0. On day 0, 5 x 10 7 On day 14, 5 x 10 human CD3εγ-expressing Ba / F3 cells were intraperitoneally administered together with Freund's incomplete adjuvant (Difco). 7 cynomolgus monkey CD3εγ-expressing Ba / F3 cells were intraperitoneally administered, followed by 5 x 10 7Human or cynomolgus monkey CD3εγ-expressing Ba / F3 cells were administered intraperitoneally. One week after the final CD3εγ administration (day 49), human CD3εγ-expressing Ba / F3 cells were administered intravenously as a booster. Three days later, rat spleen cells and mouse myeloma cells SP2 / 0 were fused using PEG1500 (Roche Diagnostics) according to standard methods. The fused cells, i.e., hybridomas, were cultured in RPMI1640 medium containing 10% FBS (hereinafter referred to as 10% FBS / RPMI1640).

[0245] On the day after the fusion, (1) the fused cells were suspended in a semi-solid medium (StemCells) and subjected to selective culture of hybridomas, and hybridoma colonization was carried out.

[0246] Hybridoma colonies were picked on days 9 or 10 after fusion and seeded at one colony per well into a 96-well plate containing HAT selection medium (10% FBS / RPMI 1640, 2 vol% HAT 50x concentrate (Dainippon Pharmaceutical), 5 vol% BM-Condimed H1 (Roche Diagnostics)). After 3–4 days of culture, the culture supernatant was collected from each well and the rat IgG concentration in the culture supernatant was measured. From the culture supernatants in which rat IgG was confirmed, clones producing antibodies that specifically bound to human CD3εγ were selected by cell-ELISA using either human CD3εγ-expressing Ba / F3 cells or Ba / F3 cells that do not express human CD3εγ (Figure 7). Furthermore, cross-reactivity with monkey CD3εγ was evaluated by cell-ELISA using cynomolgus monkey CD3εγ-expressing Ba / F3 cells (Figure 7).

[0247] (2-2) Preparation of anti-human, anti-monkey CD3ε chimeric antibodies Total RNA was extracted from hybridoma cells using RNeasy Mini Kits (QIAGEN), and cDNA was synthesized using the SMART RACE cDNA Amplification Kit (BD Biosciences). The antibody variable region genes were inserted into a cloning vector by PCR using the cDNA. The nucleotide sequence of each DNA fragment was determined using a BigDye Terminator Cycle Sequencing Kit (Applied Biosystems) on an ABI PRISM 3700 DNA Sequencer (Applied Biosystems) according to the method described in the accompanying instructions. The CDRs and FRs of the CE115 H chain variable region (SEQ ID NO: 13) and CE115 L chain variable region (SEQ ID NO: 14) were determined according to Kabat numbering.

[0248] The chimeric antibody H chain, which combines the above-mentioned rat antibody H chain variable region with a human antibody IgG1 chain constant region, and the chimeric antibody L chain, which combines the above-mentioned rat antibody L chain variable region with a human antibody Kappa chain constant region, were inserted into an animal cell expression vector. The CE115 chimeric antibody was expressed and purified using the constructed expression vector (Reference Example 1).

[0249] (2-3) Preparation of EGFR_ERY22_CE115 Next, we created a molecule using an IgG against a cancer antigen (EGFR) as the base scaffold, with one Fab replaced with a CD3ε-binding domain. As described above, a silent Fc with reduced binding to Fcγ receptors (FcγRs) was used as the Fc of the IgG base scaffold. The variable regions of cetuximab, cetuximab-VH (SEQ ID NO: 15) and cetuximab-VL (SEQ ID NO: 16), were used as the EGFR-binding domain. G1d, in which Gly and Lys were removed from the C-terminus of IgG1, A5, in which D356K and H435R mutations were introduced into G1d, and B3, in which K439E mutation was introduced into G1d, were used as the antibody H-chain constant region, and these were combined with Cetuximab-VH to prepare Cetuximab-VH-G1d (SEQ ID NO: 17), Cetuximab-VH-A5 (SEQ ID NO: 18), and Cetuximab-VH-B3 (SEQ ID NO: 19) according to the method of Reference Example 1. When the antibody H-chain constant region is designated as H1, the sequence corresponding to the H chain of an antibody having Cetuximab-VH in the variable region is shown as Cetuximab-VH-H1. Here, when an amino acid modification is indicated, it is shown as D356K, where the first letter (corresponding to the D in D356K) represents the alphabet when the amino acid residue before modification is expressed in one letter, the following number (corresponding to 356 in D356K) represents the EU numbering of the modified site, and the last letter (corresponding to the K in D356K) represents the alphabet when the amino acid residue after modification is expressed in one letter.

[0250] EGFR_ERY22_CE115 (Figure 8) was prepared by replacing the VH and VL domains of EGFR Fab. Specifically, a series of expression vectors into which polynucleotides encoding EGFR ERY22_Hk (SEQ ID NO: 20), EGFR ERY22_L (SEQ ID NO: 21), CE115_ERY22_Hh (SEQ ID NO: 22), and CE115_ERY22_L (SEQ ID NO: 23) were inserted were prepared by methods known to those skilled in the art, such as PCR using primers containing appropriate sequences similar to those described above.

[0251] The following combinations of expression vectors were transfected into FreeStyle293-F cells to transiently express each target molecule. ·Target molecule: EGFR_ERY22_CE115 Polypeptides encoded by polynucleotides inserted into expression vectors: EGFR ERY22 Hk, EGFR ERY22 L, CE115 ERY22 Hh, CE115 ERY22 L

[0252] (2-4) Purification of EGFR_ERY22_CE115 The resulting culture supernatant was applied to an Anti-FLAG M2 column (Sigma), washed, and then eluted with 0.1 mg / mL FLAG peptide (Sigma). The fraction containing the target molecule was applied to a HisTrap HP column (GE Healthcare), washed, and then eluted with an imidazole gradient. The fraction containing the target molecule was concentrated by ultrafiltration and then applied to a Superdex 200 column (GE Healthcare). The purified target molecule was obtained by collecting only the monomer fraction of the eluate.

[0253] (2-5) Measurement of cytotoxic activity using human peripheral blood mononuclear cells (2-5-1) Preparation of human peripheral blood mononuclear cell (PBMC) solution 50 mL of peripheral blood was collected from healthy adult volunteers using a syringe pre-filled with 100 μL of 1,000 U / mL heparin solution (Novo Heparin Injection 5,000 U, Novo Nordisk). After diluting the blood two-fold with PBS(-), the blood was divided into four equal portions and added to a Leucosep lymphocyte separation tube (Cat. No. 227290, Greiner Bio-One) pre-filled with 15 mL of Ficoll-Paque PLUS and centrifuged. After centrifugation (2,150 rpm, 10 min, room temperature), the mononuclear cell fraction was isolated. After washing the mononuclear cell fraction once with Dulbecco's Modified Eagle's Medium (Sigma, hereafter referred to as 10% FBS / D-MEM) containing 10% FBS, the cells were collected at a cell density of 4 × 10 6 The cell solution was prepared using 10% FBS / D-MEM to a concentration of 1 / mL. The cell solution thus prepared was used as a human PBMC solution in the subsequent tests.

[0254] (2-5-2) Measurement of cytotoxic activity Cytotoxic activity was evaluated by measuring the cell proliferation inhibition rate using the xCELLigence Real-Time Cell Analyzer (Roche Diagnostics). The target cells used were the SK-pca13a cell line, established by forcibly expressing human EGFR in the SK-HEP-1 cell line. SK-pca13a cells were detached from the dish and cultured at 1 × 10 4 The cells were seeded at 100 μL / well on an E-Plate 96 plate (Roche Diagnostics) so that the cells were 100 cells / well, and viable cell measurement was initiated using the xCELLigence Real-Time Cell Analyzer. The next day, the plate was removed from the xCELLigence Real-Time Cell Analyzer, and 50 μL of each antibody prepared at the appropriate concentrations (0.004, 0.04, 0.4, 4 nM) was added to the plate. After incubation at room temperature for 15 minutes, 50 μL (2 × 10) of the human PBMC solution prepared in (2-5-1) was added. 5Cells / well) were added to the plate, and the plate was reset in the xCELLigence Real-Time Cell Analyzer to begin measuring viable cells. The reaction was carried out under conditions of 5% carbon dioxide and 37°C, and the cell proliferation inhibition rate (%) was calculated using the Cell Index value 72 hours after the addition of human PBMCs, using the formula below. The Cell Index value used in the calculation was normalized so that the Cell Index value immediately before antibody addition was 1. Cell proliferation inhibition rate (%) = (AB) × 100 / (A-1) A shows the average cell index value in wells to which no antibody was added (target cells and human PBMCs only), and B shows the average cell index value in each well. The test was performed in triplicate.

[0255] When the cytotoxic activity of EGFR_ERY22_CE115 was measured using CE115 and PBMCs prepared from human blood as effector cells, extremely strong activity was observed (FIG. 9).

[0256] [Example 3] Preparation of an antibody that binds to CD3 and human integrin αvβ3 but does not bind simultaneously As shown in Figures 1 to 6, a dual-binding Fab is a molecule that binds to CD3 (first antigen) and a target antigen (second antigen) through its variable (Fab) region but does not simultaneously bind to both CD3 (first antigen) and the target antigen (second antigen). When introducing amino acid modifications into the Fab region of an antibody that binds to CD3 (first antigen) to enable binding to a second antigen, the amino acid modifications are typically introduced into both the two H or L chains. However, if modifications are introduced into both H or L chains, the two Fabs of the antibody may bind to the two antigens, respectively, and simultaneously bind to and crosslink CD3 (first antigen) and the target antigen (second antigen). Therefore, one Fab of the antibody is designed to bind to a third antigen or to nothing, and the other Fab is designed as a dual-binding Fab, preventing crosslinking between CD3 (first antigen) and the target antigen (second antigen).

[0257] (3-1) Preparation of antibodies that bind to CD3 and human integrin αvβ3 but do not bind simultaneously Integrin αvβ3, a known adhesion molecule, is expressed on many cancer cells and tumor-surrounding blood vessels, making it a useful targeting molecule for tumor targeting. However, it is also known to be expressed on a variety of normal cells (Thromb Haemost. 1998 Nov;80(5):726-34.). Therefore, we hypothesized that simultaneous binding of CD3 and integrin αvβ3 might result in the potent cytotoxicity of T cells in normal cells. Therefore, we hypothesized that if we could engineer a molecule that does not simultaneously bind to both CD3 and integrin αvβ3, we could target anti-EGFR antibody molecules to integrin αvβ3-expressing tumor cells without harming normal cells. Specifically, we sought to develop a dual-binding Fab molecule that binds to EGFR with one variable region (Fab) and to both CD3 and integrin αvβ3 with the other variable region, but does not simultaneously bind to both CD3 and integrin αvβ3.

[0258] If it can be demonstrated that "a molecule in which the Fab region binds to CD3 in the absence of integrin αvβ3 and in which the Fab region binds to integrin αvβ3 in the absence of CD3, and in which the molecule that binds to CD3 does not bind to integrin αvβ3, or in which the molecule that binds to integrin αvβ3 does not bind to CD3," it can be said that a dual binding Fab molecule with the desired properties of a dual binding Fab (i.e., capable of binding to CD3 and a second antigen, but not simultaneously binding to CD3 and a second antigen) has been created.

[0259] (3-2) Isolation of antibodies with Fab regions that bind to integrin αvβ3 As methods for obtaining dual-binding Fab molecules, two methods were considered: a method using a library and a method inserting a peptide known to have binding activity to a protein. RGD (Arg-Gly-Asp) peptide is known to be a peptide that has binding activity to integrin αvβ3. Therefore, a heterodimeric antibody was prepared according to Reference Example 1, in which one Fab contains an EGFR-binding domain and the other contains a CD3-binding domain and an integrin αvβ3-binding domain, and an RGD peptide was inserted into the heavy chain loop of CE115 (heavy chain variable region: SEQ ID NO: 13, light chain variable region: SEQ ID NO: 14), an antibody that binds to CD3ε. Specifically, a series of expression vectors were prepared in which polynucleotides encoding EGFR ERY22_Hk (SEQ ID NO: 20), EGFR ERY22_L (SEQ ID NO: 21), and CE115_ERY22_L (SEQ ID NO: 23), respectively, and polynucleotides encoding any of the following were inserted: CE115_2 ERY22_Hh (SEQ ID NO: 24, Kabat numbering 52b-53 are replaced with K and N, respectively), CE115_4 ERY22_Hh (SEQ ID NO: 25, Kabat numbering 52b-54 are replaced with S and N, respectively), CE115_9 ERY22_Hh (SEQ ID NO: 26, RGD inserted between 52a and 52b in the Kabat numbering), CE115_10 ERY22_Hh (SEQ ID NO: 27, RGD inserted between 52b and 52c in the Kabat numbering), CE115_12 ERY22_Hh (SEQ ID NO: 28, RGD inserted between 72-73 in the Kabat numbering system), CE115_17 ERY22_Hh (SEQ ID NO: 29, Kabat numbering 52b-52c replaced with K and S, respectively), CE115_47 ERY22_Hh (SEQ ID NO: 30, RGD inserted between 98-99 in the Kabat numbering system), CE115_48 ERY22_Hh (SEQ ID NO: 31, RGD inserted between 99-100 in the Kabat numbering system), CE115_49 ERY22_Hh (SEQ ID NO: 32, inserted into RGD between 100-100a in Kabat numbering). As a control, an antibody (EH240-Kn125 / EH240-Hl076 / L73; SEQ ID NOs: 33 / 34 / 35) in which an RGD (Arg-Gly-Asp) peptide was inserted into the CH3 region of the antibody reported in J. Biotech, 155, 193-202, 2011 was produced according to Reference Example 1. This molecule binds to integrin αvβ3 via the CH3 region and is thought to be able to bind to CD3 and integrin αvβ3 simultaneously.

[0260] (3-3) Confirmation of antibody binding to integrin αvβ3 The binding of molecules with RGD (Arg-Gly-Asp) peptide inserted into the Fab region to integrin αvβ3 was assessed by electrochemiluminescence (ECL) assay. Specifically, biotin-anti-human IgG Ab (Southern Biotech) diluted in TBS solution containing 0.1% BSA, 0.1 g / L calcium chloride, and 0.1 g / L magnesium chloride (referred to as diluted (+) solution), antibody solutions prepared at 5 μg / mL or 1 μg / mL, and sulfo-tagged integrin αvβ3 (R&D Systems) were incubated in a Nunc-Immuno TM MicroWell TMTwenty-five microliters of the solution was added to each well of a 96-well round plate (Nunc), mixed, and incubated overnight at 4°C to allow antibody-antigen complex formation. 150 μL of a TBS solution containing 0.5% BSA, 0.1 g / L calcium chloride, and 0.1 g / L magnesium chloride (referred to as "blocking (+) solution") was added to each well of a streptavidin plate (MSD) and incubated overnight at 4°C. After removing the blocking solution, the plate was washed three times with 250 μL of a TBS solution containing 0.1 g / L calcium chloride and 0.1 g / L magnesium chloride (referred to as "TBS (+) solution"). 75 μL of the antibody-antigen complex solution was added to each well and incubated at room temperature for 2 hours to allow the biotin-anti-human IgG Ab to bind to the streptavidin plate. After removing the antibody-antigen complex solution, the plate was washed three times with TBS(+) solution, 150 μL of READ buffer (MSD) was added to each well, and the luminescence signal of the sulfo-tag was detected using a Sector Imager 2400 (MSD).

[0261] The results are shown in Figure 11. The parent antibody EGFR ERY22_Hk / EGFR ERY22_L / CE115 ERY22_Hh / CE115_ERY22_L showed no binding activity to integrin αvβ3, whereas the antibodies EGFR ERY22_Hk / EGFR ERY22_L / CE115_2 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_4 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_9 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_10 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_12 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_17 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_47 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_48 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_49 ERY22_Hh / CE115_ERY22_L and CE115_ERY22_L were observed to bind to integrin αvβ3.

[0262] (3-4) Confirmation of antibody binding to CD3 (CD3ε) Next, we used ECL to determine whether the antibodies that bind to integrin αvβ3 via their Fab region, prepared in the previous section, retained their binding activity to CD3. Specifically, biotin-anti human IgG Ab (Southern Biotech) diluted with TBS solution containing 0.1% BSA (referred to as diluted (-) solution), antibody solutions adjusted to 5 μg / mL or 1 μg / mL, and sulfo-tagged CD3ε homodimer protein were incubated in Nunc-Immuno TM MicroWell TM25 μL of this solution was added to each well of a 96-well round plate (Nunc), mixed, and incubated overnight at 4°C to allow antibody-antigen complex formation. 150 μL of TBS solution containing 0.5% BSA (referred to as blocking (-) solution) was added to each well of a streptavidin plate (MSD) and incubated overnight at 4°C. After removing the blocking solution, the plate was washed three times with 250 μL of TBS (-) solution. 75 μL of the antibody-antigen complex solution was added to each well and incubated at room temperature for 2 hours to allow the biotin-anti-human IgG Ab to bind to the streptavidin plate. After removing the antibody-antigen complex solution, the plate was washed three times with TBS (-) solution. 150 μL of READ buffer (MSD) was added to each well, and the sulfo-tag luminescence signal was detected using a Sector Imager 2400 (MSD).

[0263] The results are shown in Figure 12. In addition to the parent antibody EGFR ERY22_Hk / EGFR ERY22_L / CE115 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_2 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_4 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_9 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_10 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_12 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_17 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_47 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_48 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_49 ERY22_Hh / Both CE115_ERY22_L were observed to bind to CD3.

[0264] (3-5) Confirmation by ECL that integrin αvβ3 and CD3 do not simultaneously bind to the Fab region From the results of the previous section, a molecule was obtained that has binding activity to both integrin αvβ3 and CD3. Next, we determined whether the Fab region prepared in the previous section simultaneously binds to CD3 (CD3ε) and integrin αvβ3.

[0265] When a molecule with an RGD (Arg-Gly-Asp) peptide inserted into the Fab region simultaneously binds to integrin αvβ3 and CD3, adding integrin αvβ3 and biotinylated CD3 to the antibody solution will bind to both antigens, allowing detection by ECL. Specifically, biotinylated human CD3ε homodimer protein diluted with diluent (+) solution, an antibody solution prepared at 10 μg / mL or 5 μg / mL, and sulfo-tagged integrin αvβ3 (R&D Systems) were incubated in a Nunc-Immuno TM MicroWell TM 25 μL of this solution was added to each well of a 96-well round plate (Nunc), mixed, and incubated overnight at 4°C to allow antibody-antigen complex formation. 150 μL of blocking (+) solution was added to each well of a streptavidin plate (MSD) and incubated overnight at 4°C. After removing the blocking solution, the plate was washed three times with 250 μL of TBS (+) solution containing 0.1 g / L calcium chloride and 0.1 g / L magnesium chloride. 75 μL of the antibody-antigen complex solution was added to each well and incubated at room temperature for 2 hours to allow the biotin-anti-human IgG Ab to bind to the streptavidin plate. After removing the antibody-antigen complex solution, the plate was washed three times with TBS (+). 150 μL of READ buffer (MSD) was added to each well, and the sulfo-tag luminescence signal was detected using a Sector Imager 2400 (MSD).

[0266] The results are shown in Figures 13 and 14. EGFR ERY22_Hk / EGFR ERY22_L / CE115_2 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_12 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_17 ERY22_Hh / CE115_ERY22_L, in which the RGD (Arg-Gly-Asp) peptide was inserted into the Fab region, simultaneously bound to integrin αvβ3 and CD3, and thus strong signals were detected in ECL measurements. On the other hand, the signals were weak in EGFR ERY22_Hk / EGFR ERY22_L / CE115_9 ERY22_Hh / CE115_ERY22_L and EGFR ERY22_Hk / EGFR ERY22_L / CE115_48 ERY22_Hh / CE115_ERY22_L (Figure 13). Furthermore, EGFR ERY22_Hk / EGFR ERY22_L / CE115_4 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_10 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_47 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_49 ERY22_Hh / CE115_ERY22_L showed almost no signal in ECL assays (Figure 14). This suggests that these antibodies do not bind to integrin αvβ3 when bound to CD3.

[0267] (3-6) Consideration that integrin αvβ3 and CD3 do not simultaneously bind to the Fab region by ECL analysis Based on these results, we were able to create an antibody with the properties of a dual-binding Fab molecule, in which a single Fab binds to both CD3 (CD3ε) and integrin αvβ3, but does not simultaneously bind to both CD3 (CD3ε) and integrin αvβ3. In this example, we conferred binding activity to the second antigen by inserting an RGD peptide, which binds to the second antigen (integrin αvβ3), into the Fab of an antibody with a variable region that binds to the first antigen, CD3, and obtained a molecule that does not simultaneously bind to both CD3 and the second antigen. Using a similar method, a peptide with binding activity to a protein, such as those exemplified in WO2006036834, can be inserted into the loop of the Fab to obtain a dual-binding Fab molecule with binding activity to any second antigen. Alternatively, peptides exhibiting binding activity to proteins can be obtained by preparing a peptide library using methods known to those skilled in the art and selecting peptides with the desired activity (Pasqualini R., Nature, 1996, 380 (6572):364-6). Furthermore, it is believed that dual-binding Fab molecules having binding activity to any second antigen can be created by using a library of antigen-binding molecules in which the Fab loops have been modified to be longer (extended), as described in Example 5. Since variable regions for a first antigen can be obtained by various methods known to those skilled in the art, it can be said that using such a library it is possible to create dual-binding Fab molecules that have binding activity to any first antigen and any second antigen but cannot simultaneously bind to the first antigen and the second antigen.

[0268] These results demonstrate that EGFR ERY22_Hk / EGFR ERY22_L / CE115_4 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_10 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_47 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_49 ERY22_Hh / CE115_ERY22_L bind to CD3 and integrin αvβ3, but do not bind to CD3 and integrin αvβ3 simultaneously. Specifically, EGFR ERY22_Hk / EGFR ERY22_L / CE115_4 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_10 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_47 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_49 ERY22_Hh / CE115_ERY22_L are molecules with dual-binding Fab, demonstrating the feasibility of creating such molecules.

[0269] [Example 4] Preparation of an antibody that binds to CD3 and human toll-like receptor 2 (TLR2) but not simultaneously (4-1) Preparation of antibodies that bind to both CD3 and human TLR2 but not simultaneously TLR2, a pattern recognition receptor, is primarily expressed on immune cells such as macrophages, dendritic cells, and B cells, and is a useful target molecule for activating immune cells. TLR2 is also expressed on normal cells, such as epithelial and endothelial cells. The simultaneous binding of a cancer antigen and CD3 recruits CD3-expressing T cells into the tumor environment, where they then kill cancer cells. The simultaneous binding of a cancer antigen and TLR2 may also recruit and activate TLR2-expressing immune cells into the tumor environment. TLR2-recruited immune cells internalize cancer cells killed by T cells, process the antigen, and present it to HLA, thereby activating T cells. This may lead to more potent T cell activation and adaptive immunity. However, the simultaneous binding of CD3 and TLR2 may result in the potent cytotoxicity of T cells, resulting in the activation of both immune cells and normal cells. Therefore, we thought that if we could create a molecule that does not simultaneously bind to CD3 and TLR2, we could recruit TLR2-expressing immune cells and normal cells without harming these cells. Specifically, we attempted to obtain a dual-binding Fab molecule that binds to EGFR with one variable region (Fab) and to the first antigen, CD3, and the second antigen, TLR2, with the other variable region, but does not simultaneously bind to both CD3 and TLR2.

[0270] If it can be demonstrated that "a molecule in which the Fab region binds to CD3 in the absence of TLR2 and in which the Fab region binds to TLR2 in the absence of CD3, and in which the molecule that binds to CD3 does not bind to TLR2, or in which the molecule that binds to TLR2 does not bind to CD3," it can be said that a dual binding Fab molecule with the desired properties of a dual binding Fab (i.e., capable of binding to CD3 and a second antigen, but not simultaneously binding to CD3 and a second antigen) has been created.

[0271] (4-2) Identification of antibodies with Fab regions that bind to TLR2 The RWGYHLRDRKYKGVRSHKGVPR peptide (SEQ ID NO: 36) is known to have binding activity to human TLR2. Therefore, a heterodimeric antibody was prepared according to Reference Example 1, in which one Fab is an EGFR-binding domain and the other Fab is a CD3-binding domain and a TLR2-binding domain, with a TRL2-binding peptide inserted into the heavy chain loop of CE115 (heavy chain variable region: SEQ ID NO: 13, light chain variable region: SEQ ID NO: 14), an antibody that binds to CD3ε. Specifically, a series of expression vectors were prepared in which polynucleotides encoding EGFR ERY22_Hk (SEQ ID NO: 20), EGFR ERY22_L (SEQ ID NO: 21), and CE115_ERY22_L (SEQ ID NO: 23) were inserted, along with polynucleotides encoding any of the following: CE115_DU21 ERY22_Hh (SEQ ID NO: 37, TRL2-binding peptide inserted between 52b and 52c in the Kabat numbering system), CE115_DU22 ERY22_Hh (SEQ ID NO: 38, TRL2-binding peptide inserted between 52b and 52c in the Kabat numbering system), CE115_DU26 ERY22_Hh (SEQ ID NO: 39, TRL2-binding peptide inserted between 72-73 in the Kabat numbering system), CE115_DU27 ERY22_Hh (SEQ ID NO: 40, TRL2-binding peptide inserted between Kabat numbering 72-73). As controls, an antibody in which a TLR2-binding peptide was added to the C-terminus of its CH3 domain (CE115_ERY22_DU42_Hh, SEQ ID NO: 41) and an antibody in which a peptide containing Cys residues at both ends of the TLR2-binding peptide was added to the C-terminus of its CH3 domain (CE115_ERY22_DU43_Hh, SEQ ID NO: 42) were produced according to Reference Example 1. This molecule binds to TLR2 via its CH3 domain and is thought to be able to bind to both CD3 and TLR2 simultaneously.

[0272] (4-3) Confirmation of antibody binding to TLR2 The binding of molecules with TLR2-binding peptides inserted into the Fab region to TLR2 was determined by electrochemiluminescence (ECL) assay. Specifically, biotin-anti-human IgG Ab (Southern Biotech) diluted with 0.1% BSA in TBS solution (referred to as diluted (-) solution), antibody solutions prepared at 5 μg / mL or 1 μg / mL, and sulfo-tagged TLR2 (abnova) were incubated in a Nunc-Immuno TM MicroWell TM 25 μL of this solution was added to each well of a 96-well round plate (Nunc), mixed, and incubated overnight at 4°C to allow antibody-antigen complex formation. 150 μL of TBS solution containing 0.5% BSA (referred to as blocking (-) solution) was added to each well of a streptavidin plate (MSD) and incubated overnight at 4°C. After removing the blocking solution, the plate was washed three times with 250 μL of TBS (-) solution. 75 μL of the antibody-antigen complex solution was added to each well and incubated at room temperature for 2 hours to allow the biotin-anti-human IgG Ab to bind to the streptavidin plate. After removing the antibody-antigen complex solution, the plate was washed three times with TBS (-) solution. 150 μL of READ buffer (MSD) was added to each well, and the sulfo-tag luminescence signal was detected using a Sector Imager 2400 (MSD).

[0273] The results are shown in Figure 15. The parent antibody EGFR ERY22_Hk / EGFR ERY22_L / CE115 ERY22_Hh / CE115_ERY22_L showed no binding activity to TLR2, whereas the following antibodies showed no binding activity to TLR2: EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU21 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU22 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU26 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU27 ERY22_Hh / CE115_ERY22_L and CE115_ERY22_L were observed to bind to TLR2.

[0274] (4-4) Confirmation of antibody binding to CD3 (CD3ε) Next, we used ECL to determine whether the antibodies that bind to TLR2 via the Fab region prepared in the previous section retained their binding activity to CD3 (CD3ε). Specifically, biotin-anti-human IgG Ab (Southern Biotech) diluted with TBS solution containing 0.1% BSA (referred to as diluted (-) solution), antibody solutions adjusted to 5 μg / mL or 1 μg / mL, and sulfo-tagged CD3ε homodimer protein were incubated in Nunc-Immuno TM MicroWell TM25 μL of this solution was added to each well of a 96-well round plate (Nunc), mixed, and incubated overnight at 4°C to allow antibody-antigen complex formation. 150 μL of TBS solution containing 0.5% BSA (referred to as blocking (-) solution) was added to each well of a streptavidin plate (MSD) and incubated overnight at 4°C. After removing the blocking solution, the plate was washed three times with 250 μL of TBS (-) solution. 75 μL of the antibody-antigen complex solution was added to each well and incubated at room temperature for 2 hours to allow the biotin-anti-human IgG Ab to bind to the streptavidin plate. After removing the antibody-antigen complex solution, the plate was washed three times with TBS (-) solution. 150 μL of READ buffer (MSD) was added to each well, and the sulfo-tag luminescence signal was detected using a Sector Imager 2400 (MSD).

[0275] The results are shown in Figure 16. In addition to the parent antibody EGFR ERY22_Hk / EGFR ERY22_L / CE115 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU21 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU22 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU26 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU27 ERY22_Hh / CE115_ERY22_L and CE115_ERY22_L were observed to bind to CD3.

[0276] (4-5) Confirmation by ECL that TLR2 and CD3 do not simultaneously bind to the Fab region From the results of the previous section, we obtained molecules that have binding activity to both TLR2 and CD3. Next, we determined whether the Fab region prepared in the previous section binds to both CD3 and TLR2 simultaneously.

[0277] When a molecule with a TLR2-binding peptide inserted into the Fab region binds to both TLR2 and CD3 simultaneously, adding TLR2 and biotinylated CD3 to the antibody solution will bind to both antigens, allowing detection by ECL. Specifically, biotinylated human CD3ε homodimer protein diluted with diluent (-) solution, antibody solution prepared at 10 μg / mL or 5 μg / mL, and sulfo-tagged TLR2 (R&D Systems) were incubated in a Nunc-Immuno TM MicroWell TM 25 μL of this solution was added to each well of a 96-well round plate (Nunc), mixed, and incubated overnight at 4°C to allow antibody-antigen complex formation. 150 μL of blocking solution was added to each well of a streptavidin plate (MSD) and incubated overnight at 4°C. After removing the blocking solution, the plate was washed three times with 250 μL of TBS(-) solution containing 0.1 g / L calcium chloride and 0.1 g / L magnesium chloride. 75 μL of the antibody-antigen complex solution was added to each well and incubated at room temperature for 2 hours to allow the biotin-anti-human IgG Ab to bind to the streptavidin plate. After removing the antibody-antigen complex solution, the plate was washed three times with TBS(-), and 150 μL of READ buffer (MSD) was added to each well. The sulfo-tag luminescence signal was detected using a Sector Imager 2400 (MSD).

[0278] The results are shown in Figure 17. EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU42 ERY22_Hh / CE115_ERY22_L and EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU43 ERY22_Hh / CE115_ERY22_L, in which a TLR2-binding peptide was added to the CH3 region, simultaneously bound to TLR2 and CD3, and thus strong signals were detected in ECL measurements. On the other hand, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU21 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU22 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU26 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU27 ERY22_Hh / CE115_ERY22_L showed almost no signal in ECL assays, suggesting that these antibodies do not bind to TLR2 when bound to CD3.

[0279] (4-6) Consideration that TLR2 and CD3 do not simultaneously bind to the Fab region using ECL Based on these results, we were able to create an antibody with the properties of a dual-binding Fab molecule, in which one Fab binds to CD3 and TLR2, but does not simultaneously bind to both CD3 and TLR2. In this example, we used an antibody with a variable region that binds to a first antigen, CD3, and inserted the RWGYHLRDRKYKGVRSHKGVPR peptide, which binds to a second antigen, TLR2, into the Fab. This conferred binding activity to the second antigen, resulting in a molecule that does not simultaneously bind to CD3 and the second antigen. Using a similar method, a peptide with binding activity to a protein, such as those exemplified in WO2006036834, can be inserted into a loop in the Fab to obtain a dual-binding Fab molecule with binding activity to any second antigen. Alternatively, peptides that exhibit binding activity to proteins can be obtained by preparing a peptide library using methods known to those skilled in the art and selecting peptides with the desired activity (Pasqualini R., Nature, 1996, 380 (6572):364-6). Furthermore, it is believed that dual-binding Fab molecules having binding activity toward any second antigen can be created by using a library of antigen-binding molecules in which the loops in the Fab have been modified to be longer (extended) as described in Example 5. Since variable regions for a first antigen can be obtained by various methods known to those skilled in the art, it can be said that the use of such a library makes it possible to create dual-binding Fab molecules that have binding activity toward any first antigen and any second antigen but cannot simultaneously bind to the first antigen and the second antigen.

[0280] These results demonstrate that EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU21 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU22 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU26 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU27 ERY22_Hh / CE115_ERY22_L bind to CD3 and TLR2, but do not bind to CD3 and TLR2 simultaneously. Specifically, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU21 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU22 ERY22_Hh / CE115_ERY22_L, EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU26 ERY22_Hh / CE115_ERY22_L, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_DU27 ERY22_Hh / CE115_ERY22_L are molecules with dual-binding Fab, demonstrating the feasibility of creating such molecules.

[0281] [Example 5] Antibody modification to generate antibodies that bind to CD3 and a second antigen (5-1) Examination of the insertion site and length of a peptide that can bind to a second antigen We attempted to obtain a dual-binding Fab molecule that binds to a cancer antigen with one variable region (Fab) and binds to a first antigen, CD3, and a second antigen with the other variable region, but does not simultaneously bind to CD3 and the second antigen. A heterodimerized antibody was prepared according to Reference Example 1, in which one Fab serves as an EGFR-binding domain and the other Fab serves as a CD3-binding domain, with a GGS peptide inserted into the heavy chain loop of CE115, an antibody that binds to CD3ε.

[0282] That is, EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE31 ERY22_Hh / CE115_ERY22_L (SEQ ID NOs: 20 / 21 / 43 / 23) in which a GGS residue is inserted between K52B and S52c in CDR2, EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE32 ERY22_Hh / CE115_ERY22_L (SEQ ID NOs: 20 / 21 / 44 / 23) in which a GGSGGS peptide (SEQ ID NO: 90) is inserted, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE33 ERY22_Hh / CE115_ERY22_L (SEQ ID NOs: 20 / 21 / 44 / 23) in which a GGSGGSGGS peptide (SEQ ID NO: 91) is inserted. CE115_ERY22_L: (SEQ ID NOs: 20 / 21 / 45 / 23) were prepared. Similarly, EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE34 ERY22_Hh / CE115_ERY22_L: (SEQ ID NOs: 20 / 21 / 46 / 23) in which GGS was inserted between D72 and D73, which is the loop site in framework 3, EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE35 ERY22_Hh / CE115_ERY22_L (SEQ ID NOs: 20 / 21 / 47 / 23) in which a GGSGGS peptide (SEQ ID NO: 90) was inserted, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE36 ERY22_Hh / CE115_ERY22_L: (SEQ ID NOs: 20 / 21 / 48 / 23) were prepared. EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE37 ERY22_Hh / CE115_ERY22_L: (SEQ ID NOs: 20 / 21 / 49 / 23) were prepared with a GGS inserted between A99 and Y100 in CDR3, EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE38 ERY22_Hh / CE115_ERY22_L (SEQ ID NOs: 20 / 21 / 50 / 23) were prepared with a GGSGGSGGS peptide inserted, and EGFR ERY22_Hk / EGFR ERY22_L / CE115_CE39 ERY22_Hh / CE115_ERY22_L: (SEQ ID NOs: 20 / 21 / 51 / 23) was prepared.

[0283] (5-2) Confirmation of binding of CE115 antibody containing GGS peptide to CD3ε The binding affinity of the various antibodies to CD3ε was confirmed using Biacore T100. Biotinylated CD3ε epitope peptides were bound to a CM5 chip via streptavidin, and the antibodies were applied as analytes to analyze the binding affinity.

[0284] The results are shown in Table 2. The binding affinity of CE35, CE36, CE37, CE38, and CE39 to CD3ε was equivalent to that of the parent antibody, CE115. This indicates that it is possible to insert peptides that bind to second antigens into these loops. Furthermore, the insertion of GGSGGSGGS into CE36 and CE39 did not reduce the binding affinity, indicating that inserting peptides of at least 9 amino acids into these sites does not affect binding to CD3ε.

[0285] [Table 2]

[0286] In other words, by using such peptide-inserted CE115 to obtain an antibody that binds to a second antigen, it was shown that it is possible to produce an antibody that can bind to CD3 and the second antigen but not simultaneously. Here, the amino acid sequence of the peptide to be inserted or substituted is randomly modified using known methods such as site-directed mutagenesis (Kunkel et al. (Proc. Natl. Acad. Sci. USA (1985) 82, 488-492)) or overlap extension PCR, and the binding activity of each variant is compared using the above-mentioned method to determine the insertion or substitution sites, as well as the types and lengths of amino acids, that will still exhibit the desired activity even after the amino acid sequence has been modified, thereby producing a library.

[0287] [Example 6] Library design for obtaining antibodies that bind to CD3 and a second antigen (6-1) Antibody library for obtaining antibodies that bind to CD3 and a second antigen (also called Dual Fab Library) The following six methods are exemplified as methods for selecting CD3 (CD3ε) as the first antigen and obtaining an antibody that binds to CD3 (CD3ε) and an arbitrary second antigen. 1. A method in which a peptide or polypeptide that binds to a second antigen is inserted into a Fab domain that binds to a first antigen (in addition to the peptide insertions shown in Examples 3 and 4, there is also a method in which G-CSF is inserted, as exemplified in Angew Chem Int Ed Engl. 2013 Aug 5;52(32):8295-8). The binding peptide or polypeptide can be obtained from a library that displays peptides or polypeptides, but it is also possible to use all or part of a naturally occurring protein. 2. A method of constructing an antibody library in which various amino acids appear at positions that allow the loop in the Fab to be modified (extended) longer, as shown in Example 5, and obtaining Fabs having binding activity to an arbitrary second antigen from the antibody library using the binding activity to the antigen as an index. 3. A method in which an antibody is created by site-directed mutagenesis from a Fab domain known to bind to CD3, amino acids that maintain binding activity to CD3 are identified, and Fabs that have binding activity to a given second antigen are obtained from an antibody library in which the identified amino acids appear, using the binding activity to the antigen as an index. 4. A method in which an antibody library is constructed in which various amino acids appear at positions that allow the loop in the Fab to be modified (extended) longer, and Fab having binding activity to an arbitrary second antigen is obtained from the antibody library using the binding activity to the antigen as an index, in the method of 3. In the method described in 5.1.2.3.4., a glycosylation sequence (e.g., NxS, NxT, where x is any amino acid other than P) is modified to include a glycosylation sequence that is recognized by a glycosylation receptor (e.g., a high-mannose type glycosylation sequence is added, which is recognized by the high-mannose receptor. High-mannose type glycosylation sequences can be obtained by adding kifunensine during antibody expression (MAbs. 2012 Jul-Aug;4(4):475-87)). A method in which Cys, Lys, or a non-natural amino acid is inserted or substituted into the loop region or a region that could be modified to various amino acids in method 6.1.2.3.4, and a domain that binds to a second antigen (a polypeptide, a sugar chain, or a nucleic acid such as a TLR agonist) is covalently attached (a method typified by antibody drug conjugates, in which Cys, Lys, or a non-natural amino acid is covalently attached; see mAbs 6:1, 34-45; January / February 2014, WO2009 / 134891A2, Bioconjug Chem. 2014 Feb 19;25(2):351-61). Using the above method, a dual-binding Fab can be obtained that binds to both the first and second antigens but does not bind to each other simultaneously. This Fab can be combined with a domain that binds to any third antigen (referred to as the other variable region and described in Example 1) by methods known to those skilled in the art, such as a common L chain, cross-mab, or Fab arm exchange method.

[0288] (6-2) Site-directed mutagenesis to generate single amino acid-modified CD3 (CD3ε) binding antibodies The template sequences selected for the CD3 (CD3ε)-binding antibody were CE115HA000 (SEQ ID NO: 52) for the VH region and GLS3000 (SEQ ID NO: 53) for the VL region. Amino acid modifications were made to each of the sites thought to be involved in antigen binding according to Reference Example 1. The H-chain constant region was pE22Hh (a sequence in which the sequence from CH1 onward of natural IgG1 was modified with L234A, L235A, N297A, D356C, T366S, L368A, and Y407V, the GK sequence at the C-terminus was deleted, and a DYKDDDDK sequence (SEQ ID NO: 89) was added; SEQ ID NO: 54), and the L-chain constant region was a Kappa chain (SEQ ID NO: 55). The modified sites are shown in Table 3. For evaluation of CD3 (CD3ε)-binding activity, a single-amino-acid-modified antibody was obtained as a one-arm antibody (an antibody lacking one of the Fab domains of natural IgG). Specifically, for the H chain modification, GLS3000 was used, in which the modified H chain was linked to the constant region pE22Hh, and Kn010G3 (the amino acid sequence from position 216 onwards of native IgG1 with the addition of C220S, Y349C, T366W, and H435R modifications; SEQ ID NO: 56) and a kappa chain linked to the 3' end. For the L chain modification, a sequence in which a kappa chain was linked to the 3' end of the modified L chain and CE115HA000 and Kn010G3, in which pE22Hh was linked to the 3' end as the H chain, were used, and these were expressed in FreeStyle293 cells and purified (using the method described in Reference Example 1).

[0289] [Table 3]

[0290] (6-3) CD3 binding evaluation of single amino acid modified antibodies The single amino acid variants constructed, expressed, and purified in (6-2) were evaluated using a BiacoreT200 (GE Healthcare). After immobilizing an appropriate amount of CD3ε homodimer protein on a sensor chip CM4 (GE Healthcare) by the amino coupling method, an appropriate concentration of antibody was injected as an analyte to allow it to interact with the CD3ε homodimer protein on the sensor chip. The sensor chip was then regenerated by injecting 10 mmol / L Glycine-HCl (pH 1.5). Measurements were performed at 25°C, and HBS-EP+ (GE Healthcare) was used as the running buffer. The dissociation constant K was calculated using a single-cycle kinetics model (1:1 binding RI=0) for the binding amount and the sensorgram obtained in the measurement. D (M) was calculated using Biacore T200 Evaluation Software (GE Healthcare).

[0291] (6-3-1) H chain modification Table 4 shows the ratio of the binding amount of various H chain variants to the unmodified antibody CE115HA000. That is, when the binding amount of an antibody containing CE115HA000 is X and the binding amount of a single amino acid modified H chain is Y, Z (ratio of binding amount) = Y / X. In this case, as shown in Figure 18, when Z is less than 0.8, the sensorgram indicates that the binding amount is very low, and the dissociation constant K D Next, the dissociation constants K of various H chain variants to CE115HA000 may not be calculated. D The (M) ratio (=KD value of CE115HA000 / KD value of variant) is shown in Table 5. When Z shown in Table 4 is 0.8 or greater, it is considered that the antibody maintains binding to the unmodified antibody CE115HA000. Therefore, an antibody library designed to contain these amino acids can be used as a Dual Fab Library.

[0292] [Table 4]

[0293] [Table 5]

[0294] (6-3-2) Light chain modification Table 6 shows the ratio of the binding amount of various modified L chains to the unmodified antibody GLS3000. That is, when the binding amount of an antibody containing GLS3000 is X and the binding amount of a single amino acid modified L chain is Y, Z (ratio of binding amount) = Y / X. In this case, as shown in Figure 18, when Z is less than 0.8, the sensorgram indicates that the binding amount is very low, and the dissociation constant K D Next, it was suggested that the dissociation constant K of various L chain variants against GLS3000 may not be calculated. D The ratios of (M) are shown in Table 7. When Z shown in Table 6 is 0.8 or greater, it is considered that the antibody maintains binding to the unmodified antibody GLS3000. Therefore, an antibody library designed to contain these amino acids can be used as a Dual Fab Library.

[0295] [Table 6]

[0296] [Table 7]

[0297] (6-4) ECM (Extracellular matrix) binding evaluation of single amino acid modified antibodies Extracellular matrix (ECM) is an extracellular component present in various parts of the body. Therefore, antibodies that strongly bind to ECM are known to have poor blood kinetics (short half-life) ( WO2012093704A1 ). Therefore, it is preferable to select amino acids that do not enhance ECM binding from antibody libraries.

[0298] Antibodies for each H chain or L chain variant were obtained by the method described in (6-2). ECM binding was then evaluated according to the method described in Reference Example 2. The ECM binding value (ECL response) for each variant was divided by the antibody ECM binding value of the MRA (H chain SEQ ID NO: 57, L chain SEQ ID NO: 58) performed on the same plate or on the same day, and the values are shown in Table 8 (H chain) and Table 9 (L chain). As shown in Tables 8 and 9, some modifications tended to enhance ECM binding. Of the values shown in Table 8 (H chain) and Table 9 (L chain), taking into consideration the effect of multiple modifications on enhancing ECM binding, values up to 10 times were considered effective and were adopted for the Dual Fab Library.

[0299] [Table 8]

[0300] [Table 9]

[0301] (6-5) Examination of peptide insertion site and length to enhance library diversity In Example 5, it was demonstrated that peptides could be inserted at various positions using GGS sequences without losing binding to CD3 (CD3ε). It was thought that if loop extension were possible in a dual Fab library, the library would contain a wider variety of molecules (also expressed as greater diversity), making it possible to obtain Fab domains that bind to a variety of second antigens. Therefore, since it was predicted that binding activity would decrease with peptide insertion, the CE115HA000 sequence was modified with V11L / D72A / L78I / D101Q to enhance binding activity to CD3ε, and pE22Hh was linked to this sequence. A GGS linker was inserted into this sequence as in Example 5 to prepare a molecule, and CD3 binding was evaluated. The GGS sequence was inserted between positions 99 and 100 on the Kabat numbering system. The antibody molecule was expressed as a one-arm antibody. Specifically, the aforementioned H chain containing a GGS linker was Kn010G3 (sequence number: 56), and the L chain was a sequence linking GLS3000 (sequence number: 53) and a Kappa sequence (sequence number: 55), and expression and purification were carried out according to Reference Example 1.

[0302] (6-6) Confirmation of binding of CE115 antibody containing GGS peptide to CD3 The binding of modified antibodies with GGS peptides inserted to CD3ε was tested using Biacore by the method described in Example 6. The results, as shown in Table 10, demonstrated that GGS linkers can be inserted into loop sites. It is particularly possible to insert a GGS linker into the H chain CDR3 region, which is important for antigen binding, and binding to CD3ε was maintained regardless of whether 3, 6, or 9 amino acids were inserted. While a GGS linker was used in this study, antibody libraries containing various amino acids other than GGS may also be used.

[0303] [Table 10]

[0304] (6-7) Examination of library insertion into H chain CDR3 using NNS bases In (6-6), insertion of 3, 6, or 9 amino acids is possible using a GGS linker. It was anticipated that a library containing 3, 6, or 9 amino acids could be created and antibodies binding to a second antigen could be isolated using conventional antibody isolation methods, such as phage display. Therefore, when inserting 6 amino acids into CDR3, we used NNS bases (which allow various amino acids to appear) to examine whether binding to CD3 would be maintained even if various amino acids appeared at the insertion site. Since decreased binding activity was predicted, primers were designed using NNS bases to insert 6 amino acids between positions 99 and 100 (Kabat numbering) in the CDR3 of the CE115HA340 sequence (SEQ ID NO: 59), which has higher CD3ε binding activity than CE115HA000. The antibody molecules were expressed as one-arm antibodies. Specifically, the H chain containing the above-mentioned modifications was Kn010G3 (SEQ ID NO: 56), and the L chain was a sequence linking GLS3000 (SEQ ID NO: 53) and the Kappa sequence (SEQ ID NO: 55). Expression and purification were carried out according to Reference Example 1. The resulting modified antibodies were evaluated for binding using the method described in (6-3). The results are shown in Table 11. It was demonstrated that binding to CD3 (CD3ε) was maintained even when various amino acids appeared at the amino acid extension site. Furthermore, whether or not nonspecific binding was enhanced was evaluated using the method described in Reference Example 2. The results are shown in Table 12. As a result, since the presence of many amino acids with positively charged side chains in the extended loop of CDR3 enhances binding to ECM, it was desirable to avoid the presence of amino acids with three or more positively charged side chains in the loop.

[0305] [Table 11]

[0306] [Table 12]

[0307] (6-7) Design and build a Dual Fab Library Based on the studies described in Example 6, an antibody library (Dual Fab Library) for obtaining antibodies that bind to CD3 and a second antigen was designed as follows. Step 1: Select amino acids that retain CD3 (CD3ε) binding ability (CD3 binding amount is 80% or more of CE115HA000) Step 2: Select amino acids whose ECM binding is within 10-fold of that before modification compared to MRA Step 3: Insert 6 amino acids between positions 99-100 (Kabat numbering) of the heavy chain CDR3 Furthermore, since the antigen-binding site of Fab is diversified even by step 1 alone, it can be used to identify antigen-binding molecules that bind to a second antigen. Furthermore, since the antigen-binding site of Fab is diversified even by steps 1 and 3 alone, it can be used to identify antigen-binding molecules that bind to a second antigen. Even when designing a library without step 2, the ECM binding of the obtained molecules can be measured and evaluated.

[0308] Based on the above, the H chain of the Dual Fab Library was diversified using the sequence of the CE115HA000 framework (FR) with the V11L / L78I mutations added as CDRs as shown in Table 13, and the L chain was diversified using the GLS3000 CDRs as shown in Table 14. These antibody library fragments can be synthesized using DNA synthesis methods known to those skilled in the art. The following Dual Fab libraries can be created: (1) a library in which the H chain is diversified as shown in Table 13 and the L chain is fixed to the original sequence GLS3000 or the L chain with enhanced CD3ε binding described in Example 6; (2) a library in which the H chain is fixed to the original sequence (CE115HA000) or the H chain with enhanced CD3ε binding described in Example 6 and the L chain is diversified as shown in Table 14; or (3) a library in which the H chain is diversified as shown in Table 13 and the L chain is diversified as shown in Table 14. The heavy chain consisted of a sequence in which the V11L / L78I mutations were added to the framework (FR) of CE115HA000, and the CDRs were diversified as shown in Table 13. The library sequence was then entrusted to DNA2.0, a DNA synthesis company, to obtain antibody library fragments (DNA fragments). The obtained antibody library fragments were amplified by PCR and inserted into a phagemid for phage display. GLS3000 was selected as the light chain. The constructed phagemid for phage display was then introduced into E. coli by electroporation, and E. coli carrying the antibody library fragments were generated.

[0309] [Table 13]

[0310] [Table 14]

[0311] [Example 7] Obtaining Fab domains that bind to CD3 and a second antigen (IL6R) from a dual Fab library (7-1) Isolation of Fab domains that bind to human IL6R Fab domains (antibody fragments) that bind to human IL6R were identified from the Dual Fab library designed and constructed in Example 6. Antibody fragments capable of binding to human IL6R were enriched using biotin-labeled human IL6R as the antigen. Phages were produced from E. coli harboring the constructed phage display phagemid. The culture medium of the E. coli in which the phage was produced was supplemented with 2.5 M NaCl / 10% PEG to precipitate the phage population. The resulting phage library solution was then diluted with TBS. Next, BSA was added to the phage library solution to a final concentration of 4% BSA. Panning was performed using a commonly used method using antigens immobilized on magnetic beads (J. Immunol. Methods. (2008) 332 (1-2), 2-9; J. Immunol. Methods. (2001) 247 (1-2), 191-203; Biotechnol. Prog. (2002) 18 (2) 212-20; Mol. Cell Proteomics (2003) 2 (2), 61-9). NeutrAvidin-coated beads (Sera-Mag SpeedBeads NeutrAvidin-coated) or Streptavidin-coated beads (Dynabeads M-280 Streptavidin) were used as magnetic beads.

[0312] Specifically, 250 pmol of biotin-labeled antigen was added to the prepared phage library solution, and the phage library solution was allowed to contact the antigen for 60 minutes at room temperature. BSA-blocked magnetic beads were added, and the antigen-phage complexes were allowed to bind to the magnetic beads for 15 minutes at room temperature. The beads were washed three times with TBST (TBS containing 0.1% Tween 20, manufactured by TaKaRa) and then twice more with 1 mL of TBS. Then, 0.5 mL of 1 mg / mL trypsin was added, and the beads were suspended at room temperature for 15 minutes. The beads were then immediately separated using a magnetic stand, and the phage solution was recovered. The recovered phage solution was added to 10 mL of E. coli strain ER2738 in the logarithmic growth phase (OD600 0.4-0.5). The E. coli cells were cultured at 37°C for 1 hour with gentle agitation to infect the phages. The infected E. coli were seeded onto a 225 mm x 225 mm plate. Phages were then recovered from the culture medium of the seeded E. coli to prepare a phage library solution. This cycle, called panning, was repeated multiple times. From the second round onward, 40 pmol of biotin-labeled antigen was used. In the fourth round of panning, phages were enriched based on their CD3 binding. Specifically, 250 pmol of biotin-labeled CD3ε peptide antigen (amino acid sequence: SEQ ID NO: 60) was added to the prepared phage library solution, and the phage library was contacted with the antigen for 60 minutes at room temperature. BSA-blocked magnetic beads were added, and the antigen-phage complexes were allowed to bind to the magnetic beads for 15 minutes at room temperature. The beads were then washed with 1 mL of TBS containing 0.1% Tween 20 and TBS. The beads were added with 0.5 mL of 1 mg / mL trypsin and suspended at room temperature for 15 minutes. The beads were then immediately separated using a magnetic stand, and the phage solution was collected. The phage collected from the trypsin-treated phage solution was added to 10 mL of E. coli strain ER2738 in the logarithmic growth phase (OD600 0.4-0.7). The E. coli was cultured at 37°C with gentle agitation for 1 hour to infect the phage.The infected E. coli was plated onto a 225 mm x 225 mm plate, and then the phage library solution was recovered by recovering the phages from the culture medium of the plated E. coli. Furthermore, to prevent multiple phages from infecting a single E. coli cell, the phage library solution prepared from E. coli cells infected with the phages recovered from the fifth panning was diluted 100,000 times and used to infect E. coli cells to obtain single colonies.

[0313] (7-2) Binding of phage-displayed Fab domains to CD3 or IL6R (phage ELISA) Phage-containing culture supernatants were collected from single E. coli colonies obtained by the above method according to a standard method (Methods Mol. Biol. (2002) 178, 133-145). The phage-containing culture supernatants, to which BSA had been added to a final concentration of 4% BSA, were subjected to ELISA as follows: StreptaWell 96 microtiter plates (Roche) were coated with 100 μL of PBS containing biotin-labeled antigen (biotinylated CD3ε peptide or biotinylated human IL6R) overnight at 4°C or for 1 hour at room temperature. After washing each well of the plate with PBST to remove the antigen, the wells were blocked with 250 μL of 4% BSA-TBS for at least 1 hour. After removing the 4% BSA-TBS, the prepared culture supernatants were added to each well, and the plate was left to stand at room temperature for 1 hour, allowing the phage-displaying antibodies to bind to the antigen present in each well. After washing each well with TBST, HRP-conjugated anti-M13 antibody (Amersham Pharmacia Biotech) diluted in TBS with a final concentration of 4% BSA was added and incubated for 1 hour. After washing with TBST, TMB single solution (ZYMED) was added to each well. The color reaction was stopped by adding sulfuric acid, and the color development was measured by measuring the absorbance at 450 nm. The results are shown in Figure 19. Clones #50 and #62 were shown to bind to CD3ε and human IL6R. In other words, using the Dual Fab Library, clones that exhibited binding to the second antigen (human IL6R in Example 7) could be selected. Further evaluations were performed to select clones that exhibited binding, which could then be converted into IgG (by linking the VH and VL sequences of the clones to human H-chain or L-chain constant regions, respectively) and their binding to CD3ε and the second antigen (human IL6R) could be evaluated. Furthermore, whether CD3ε and a second antigen (human IL6R) bind simultaneously can be examined by the methods or competitive assays described in Examples 3 and 4. In the competitive assay, for example, the binding to CD3ε is reduced in the presence of a second antigen compared to the binding of the antibody alone, indicating that the antibodies do not bind simultaneously.

[0314] [Example 8] Obtaining Fab domains that bind to CD3 and a second antigen (human IgA) from a dual Fab library (8-1) Isolation of Fab domains that bind to human IgA IgA is an antibody isotype that is abundant in the body and is known to be involved in the body's defenses in the intestine and mucous membranes, and is known to bind to FcαR (Fc alpha receptor) (J. Pathol. 208: 270-282, 2006). Fab domains (antibody fragments) that bind to human IgA were identified from the Dual Fab library designed and constructed in Example 6. Antibody fragments capable of binding to human IgA were enriched using biotin-labeled human IgA (described in Reference Example 3) as the antigen. Phages were produced from E. coli harboring the constructed phage display phagemid. The culture medium of the E. coli in which the phage was produced was supplemented with 2.5 M NaCl / 10% PEG to precipitate the phage population. The resulting phage library solution was then diluted with TBS. Next, BSA was added to the phage library solution to a final concentration of 4% BSA. Panning was performed using a commonly used method using antigens immobilized on magnetic beads (J. Immunol. Methods. (2008) 332 (1-2), 2-9; J. Immunol. Methods. (2001) 247 (1-2), 191-203; Biotechnol. Prog. (2002) 18 (2) 212-20; Mol. Cell Proteomics (2003) 2 (2), 61-9). NeutrAvidin-coated beads (Sera-Mag SpeedBeads NeutrAvidin-coated) or Streptavidin-coated beads (Dynabeads M-280 Streptavidin) were used as magnetic beads. Specifically, 250 pmol of biotin-labeled antigen was added to the prepared phage library solution, and the phage library solution was allowed to contact the antigen for 60 minutes at room temperature. BSA-blocked magnetic beads were added, and the antigen-phage complexes were allowed to bind to the magnetic beads for 15 minutes at room temperature. The beads were washed three times with TBST (TBS containing 0.1% Tween 20, manufactured by TaKaRa) and then twice more with 1 mL of TBS. Then, 0.5 mL of 1 mg / mL trypsin was added, and the beads were suspended at room temperature for 15 minutes. The beads were then immediately separated using a magnetic stand, and the phage solution was recovered. The recovered phage solution was added to 10 mL of E. coli strain ER2738 in the logarithmic growth phase (OD600 0.4-0.5). The E. coli cells were cultured at 37°C for 1 hour with gentle agitation to infect the phages. The infected E. coli was plated onto a 225 mm x 225 mm plate. Phages were then collected from the plated E. coli culture to prepare a phage library. This cycle, called panning, was repeated four times. From the second panning onwards, human IgA was used at 40 pmol.

[0315] (8-2) Binding of phage-displayed Fab domains to CD3 or human IgA Phage-containing culture supernatants were collected from single colonies of E. coli obtained by the above method according to a standard method (Methods Mol. Biol. (2002) 178, 133-145). The phage-containing culture supernatants, to which BSA had been added to a final concentration of 4% BSA, were subjected to ELISA as follows: StreptaWell 96 microtiter plates (Roche) were coated with 100 μL of PBS containing biotin-labeled antigen (biotin-labeled CD3ε peptide or biotin-labeled human IgA, Reference Example 3) overnight at 4°C or for 1 hour at room temperature. After washing each well of the plate with PBST to remove the antigen, the wells were blocked with 250 μL of 0.1x TBS / 150 mM NaCl / 0.02% Skim Milk for at least 1 hour. The 0.1x TBS / 150mM NaCl / 0.02% Skim Milk was removed, and the prepared culture supernatant was added to each well. The plate was then incubated at room temperature for 1 hour to allow the phage-displayed antibodies to bind to the antigens present in each well. After washing each well with 0.1x TBS / 150mM NaCl / 0.01% Tween 20, HRP-conjugated anti-M13 antibody (Amersham Pharmacia Biotech) diluted in 0.1x TBS / 150mM NaCl / 0.01% Tween 20 was added and incubated for 1 hour. After washing with TBST, TMB single solution (ZYMED) was added to each well. The color reaction was stopped by adding sulfuric acid, and the color development was measured by absorbance at 450 nm. The results are shown in Figure 20. As shown in Figure 20, the presence of clones that bind to both CD3 and human IgA was demonstrated, and by using the Dual Fab Library, clones that exhibit binding to the second antigen (human IgA in Example 8) could be selected.

[0316] (8-3) Binding of IgG with the isolated Fab domain to CD3 or human IgA For clones shown to bind to CD3 and human IgA in (8-2), VH fragments were amplified by PCR from E. coli containing the clone sequences using primers that specifically bind to the H chain of the Dual Fab Library. The amplified VH fragments were incorporated into an animal cell expression plasmid containing pE22Hh as described in Reference Example 1, and expressed and purified as one-arm antibodies in the same manner as in Example 6 (6-2). The clone names and H-chain sequence SEQ ID NOs are shown in Table 15. Specifically, expression and purification were carried out according to Reference Example 1 using the H chain shown in Table 15, Kn010G3 (SEQ ID NO: 56), and the L chain, a sequence formed by linking GLS3000 (SEQ ID NO: 53) and the Kappa sequence (SEQ ID NO: 55).

[0317] [Table 15]

[0318] The binding of the obtained antibody molecules having the Fab region to CD3ε and human IgA was determined by electrochemiluminescence (ECL) assay. Specifically, biotin-labeled CD3ε peptide (described in Example 7) or biotin-labeled human IgA (Reference Example 3) diluted with TBST solution (TBS supplemented with 0.1% Tween 20, manufactured by TaKaRa), an antibody solution adjusted to 2 μg / mL, and a sulfo-tagged anti-human IgG antibody (Invitrogen #628400) were added to Nunc-Immuno. TM MicroWell TM25 μL of this solution was added to each well of a 96-well round plate (Nunc), mixed, ...

Claims

1. A method for screening for an antigen-binding molecule comprising a variable region capable of binding to a first antigen, which is CD3, and a second antigen different from the first antigen, but not simultaneously binding to the first antigen and the second antigen, the method comprising the following steps (a) to (c): (a) contacting a library mainly consisting of a plurality of antigen-binding molecules having different sequences, each of the plurality of antigen-binding molecules having at least one amino acid modification introduced into a template sequence that binds to a CD3 chain, wherein the template sequence comprises SEQ ID NO: 52, 94, 96, or 59 as a heavy chain variable domain, and the one amino acid modification comprises insertion of an amino acid that binds to the second antigen into at least one position selected from a loop region, CDR1, CDR2, CDR3, or FR3 region of an antibody variable region, with the second antigen; (b) recovering the antigen-binding molecule bound to the second antigen in step (a); and (c) selecting, from the population of antigen-binding molecules recovered in step (b), antigen-binding molecules comprising variable regions that do not simultaneously bind to the first antigen and the second antigen.

2. The screening method according to claim 1, wherein at least one position selected from the loop region, CDR1, CDR2, CDR3, or FR3 region of the antibody variable region is any one of positions 31 to 35, 50 to 65, 71 to 74, and 95 to 102 according to the Kabat numbering system in the heavy chain variable region.

3. The screening method according to claim 1 or 2, wherein the template sequence binds to the CD3ε chain.

4. The screening method according to any one of claims 1 to 3, wherein step (c) is a step of selecting, from the population of antigen-binding molecules recovered in step (b), antigen-binding molecules comprising variable regions that do not simultaneously bind to the first antigen and the second antigen, each of which is expressed on different cells.

5. The screening method according to claim 1 , wherein the second antigen is a molecule expressed on the surface of a T cell or other immune cell.

6. The screening method according to any one of claims 1 to 5, wherein the second antigen is an FcγR, a TLR, a lectin, an IgA, an immune checkpoint molecule, a TNF superfamily molecule, a TNFR superfamily molecule, or an NK receptor molecule.

7. The screening method according to any one of claims 1 to 6, wherein the antigen-binding molecule is a fusion polypeptide of an antibody variable region and at least a part of a viral coat protein.

8. The screening method according to any one of claims 1 to 7, wherein the at least one amino acid modification is introduced into the template sequence, and the modified amino acid is an amino acid at any one or more positions selected from the following: H chain: 31, 52b, 52c, 53, 54, 56, 57, 61, 98, 99, 100, 100a, 100b, 100c, 100d, 100e, 100f, 100g (Kabat numbering); Light chain: 24, 25, 26, 27, 27a, 27b, 27c, 27e, 30, 31, 33, 34, 51, 52, 53, 54, 55, 56, 74, 77, 89, 90, 92, 93, 94, 96 (Kabat numbering).

9. The screening method according to any one of claims 1 to 8, further comprising the steps of: (d) contacting the library with the first antigen; (e) recovering the antigen-binding molecule bound to the first antigen in step (d); and (f) selecting, from the population of antigen-binding molecules that bound to the first antigen in step (e), antigen-binding molecules that comprise a variable region whose binding to the first antigen is enhanced compared to the template sequence.

10. A method for screening for an antigen-binding molecule, the antigen-binding molecule comprising a variable region capable of binding to a first antigen which is CD3 and a second antigen different from the first antigen but not simultaneously binding to the first antigen and the second antigen, and which is a fusion polypeptide of an antibody variable region and at least a portion of a viral coat protein, the method comprising the following steps (a) to (c): (a) contacting a library mainly consisting of a plurality of antigen-binding molecules having different sequences, each of the plurality of antigen-binding molecules having at least one amino acid modification introduced into a template sequence that binds to a CD3 chain, wherein the template sequence comprises SEQ ID NO: 13, 52, 94, 96, 97, or 59 as a heavy chain variable domain, and the one amino acid modification comprises insertion of an amino acid that binds to the second antigen into at least one position selected from a loop region, CDR1, CDR2, CDR3, or FR3 region of an antibody variable region, with the second antigen; (b) recovering the antigen-binding molecule bound to the second antigen in step (a); and (c) selecting, from the population of antigen-binding molecules recovered in step (b), antigen-binding molecules comprising variable regions that do not simultaneously bind to the first antigen and the second antigen.

11. The screening method described in claim 10, wherein at least one position selected from the loop region, CDR1, CDR2, CDR3 or FR3 region of the variable region of the antibody is any one of Kabat numbering 31 to 35, 50 to 65, 71 to 74 and 95 to 102 in the heavy chain variable region.

12. The screening method according to claim 10 or 11, wherein the template sequence binds to the CD3ε chain.

13. The screening method according to claim 10, wherein step (c) is a step of selecting, from the population of antigen-binding molecules collected in step (b), antigen-binding molecules comprising variable regions that do not simultaneously bind to the first antigen and the second antigen, each of which is expressed on different cells.

14. A screening method described in any one of claims 10 to 13, wherein the second antigen is a molecule expressed on the surface of a T cell or other immune cell.

15. A screening method described in any one of claims 10 to 14, wherein the second antigen is FcγR, TLR, lectin, IgA, an immune checkpoint molecule, a TNF superfamily molecule, a TNFR superfamily molecule, or an NK receptor molecule.

16. The method of claim 10, wherein the at least one amino acid modification is introduced into the template sequence, and the modified amino acid is an amino acid at one or more positions selected from the following: H chain: 31, 52b, 52c, 53, 54, 56, 57, 61, 98, 99, 100, 100a, 100b, 100c, 100d, 100e, 100f, 100g (Kabat numbering); Light chain: 24, 25, 26, 27, 27a, 27b, 27c, 27e, 30, 31, 33, 34, 51, 52, 53, 54, 55, 56, 74, 77, 89, 90, 92, 93, 94, 96 (Kabat numbering).

17. The screening method according to any one of claims 10 to 16, further comprising the steps of: (d) contacting the library with the first antigen; (e) recovering the antigen-binding molecule bound to the first antigen in step (d); and (f) selecting, from the population of antigen-binding molecules that bound to the first antigen in step (e), antigen-binding molecules that comprise a variable region whose binding to the first antigen is enhanced compared to the template sequence.

Images