ANTI-CD3 ANTIBODIES, ANTI-CD123 ANTIBODIES AND BISPECIFIC ANTIBODIES SPECIFICALLY BINDING TO CD3 AND/OR CD123

Abstract

The present invention concerns antibody-like binding protein specifically binding to CD3 and binding specifically to at least one further antigen, for example CD123. The present invention also concerns antibody-like binding protein specifically binding to CD123 and binding specifically to at least one further antigen. The invention further concerns anti-CD3 antibodies and anti-CD123 antibodies. The invention also relates to pharmaceutical compositions comprising the antibody-like binding protein, anti-CD3 antibodies or anti-CD123 antibodies of the invention, and their use to treat cancer. The invention further relates to isolated nucleic acids, vectors and host cells comprising a sequence encoding said antibody-like binding protein, anti-CD3 or anti-CD123 antibody and the use of said anti-CD123 antibody as a diagnostic tool.

Claims

1: A method of treating a disease or disorder in a subject in need thereof, comprising: administering to the subject a therapeutically effective amount of an antibody-like binding protein that specifically binds to CD3ε and CD123 comprising two polypeptide chains that form two antigen-binding sites, wherein the first polypeptide chain comprises a structure represented by the formula [I]:
V.sub.D1-L.sub.1-V.sub.D2-L.sub.2-C.sub.L  [I] and the second polypeptide chain comprises a structure represented by the formula [II]:
V.sub.D3-L.sub.3-V.sub.D4-L.sub.4-C.sub.H1  [II] wherein: V.sub.D1 is a light chain variable domain of a first immuoglobulin; V.sub.D2 is a light chain variable domain of a second immunoglobulin; V.sub.D3 is a heavy chain variable domain of the second immunoglobulin; V.sub.D4 is a heavy chain variable domain of the first immunoglobulin: C.sub.L is a light chain constant domain of an immunoglobulin; C.sub.H1 is a C.sub.H1 heavy chain constant domain of an immunoglobulin; and L.sub.1, L.sub.2, L.sub.3, and L.sub.4 are amino acid linkers; wherein the first polypeptide chain and the second polypeptide chain form a cross-over light chain-heavy chain pair; and wherein V.sub.D1 comprises a CDR1-L comprising an amino acid sequence as set forth in SEQ ID NO: 378, a CDR2-L comprising an amino acid sequence of WAS, and a CDR3-L comprising an amino acid sequence as set forth in SEQ ID NO: 379; V.sub.D2 comprises a CDR4-L comprising an amino acid sequence as set forth in SEQ ID NO: 142, a CDR5-L comprising an amino acid sequence of KVS, and a CDR6-L comprising an amino acid sequence as set forth in SEQ ID NO: 11: V.sub.D3 comprises a CDR1-H comprising an amino acid sequence as set forth in SEQ ID NO: 6, a CDR2-H comprising an amino acid sequence as set forth in SEQ ID NO: 7, and a CDR3-H comprising an amino acid sequence as set forth in SEQ ID NO: 8; and V.sub.D4 comprises a CDR4-H comprising an amino acid sequence as set forth in SEQ ID NO: 381, a CDR5-H comprising an amino acid sequence as set forth in SEQ ID NO: 384, and a CDR6-H comprising an amino acid sequence as set forth in SEQ ID NO: 382.

2: The method of claim 1, wherein: L.sub.1 comprises an amino acid sequence as set forth in SEQ ID NO: 389: L.sub.2 comprises an amino acid sequence as set forth in SEQ ID NO: 389; and L.sub.3 and L.sub.4 are 0 amino acids in length.

3: The method of claim 1, wherein: V.sub.D1 comprises an amino acid sequence as set forth in SEQ ID NO: 385: V.sub.D2 comprises an amino acid sequence as set forth in SEQ ID NO: 141; V.sub.D3 comprises an amino acid sequence as set forth in SEQ ID NO: 138; V.sub.D4 comprises an amino acid sequence as set forth in SEQ ID NO: 383; C.sub.L comprises an amino acid sequence as set forth in SEQ ID NO: 310; and C.sub.H1 comprises an amino acid sequence as set forth in SEQ ID NO: 313.

4: The method of claim 1, wherein: (a) the first polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 388, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 388, which comprises a CDR1-L comprising an amino acid sequence as set forth in SEQ ID NO: 378, a CDR2-L comprising an amino acid sequence WAS, a CDR3-L comprising an amino acid sequence as set forth in SEQ ID NO: 379, a CDR4-L comprising an amino acid sequence as set forth in SEQ ID NO: 142, a CDR5-L comprising an amino acid sequence of KVS, and a CDR6-L comprising an amino acid sequence as set forth in SEQ ID NO: 11; and (b) the second polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 390, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 390, which comprises a CDR1-H comprising an amino acid sequence as set forth in SEQ ID NO: 6, a CDR2-H comprising an amino acid sequence as set forth in SEQ ID NO: 7, a CDR3-H comprising an amino acid sequence as set forth in SEQ ID NO: 8, a CDR4-H comprising an amino acid sequence as set forth in SEQ ID NO: 381, a CDR5-H comprising an amino acid sequence as set forth in SEQ ID NO: 384, and a CDR6-H comprising an amino acid sequence as set forth in SEQ ID NO: 382.

5: The method of claim 1, wherein the second polypeptide chain further comprises a Fc domain (“Fc”).

6: The method of claim 1, wherein the first polypeptide chain further comprises a Fc domain (“Fc”).

7: A method of treating a disease or disorder in a subject in need thereof, comprising: administering to the subject a therapeutically effective amount of an antibody-like binding protein that specifically binds to CD3ε and CD123 comprising two polypeptide chains that form two antigen-binding sites, wherein the first polypeptide chain comprises a structure represented by the formula [IV]:
V.sub.D1-L.sub.1-V.sub.D2-L.sub.2-C.sub.L-L.sub.5-Fc.sub.2  [IV] and the second polypeptide chain comprises a structure represented by the formula [III]:
V.sub.D3-L.sub.3-V.sub.D4-L.sub.4-C.sub.H1-Fc  [III] wherein: V.sub.D1 is a light chain variable domain of a first immunoglobulin; V.sub.D2 is a light chain variable domain of a second immuoglobulin; V.sub.D3 is a heavy chain variable domain of the second immunoglobulin; V.sub.D4 is a heavy chain variable domain of the first immunoglobulin; C.sub.L is a light chain constant domain of an immunoglobulin; C.sub.H1 is a C.sub.H1 heavy chain constant domain of an immunoglobulin; Fc is the immunoglobulin hinge region and CH.sub.2, CH.sub.3 immunoglobulin heavy chain constant domains of a first immunoglobulin; Fc.sub.2 is the immunoglobulin hinge region and CH.sub.2, CH.sub.3 immunoglobulin heavy chain constant domains of a second immunoglobulin; and L.sub.1, L.sub.2, L.sub.3, L.sub.4, and L.sub.5 are amino acid linkers; wherein the first polypeptide chain of formula [IV] and the second polypeptide chain of formula [III] form a cross-over light chain-heavy chain pair; and wherein V.sub.D1 comprises a CDR1-L comprising an amino acid sequence as set forth in SEQ ID NO: 378, a CDR2-L comprising an amino acid sequence of WAS, and a CDR3-L comprising an amino acid sequence as set forth in SEQ ID NO: 379; V.sub.D2 comprises a CDR4-L comprising an amino acid sequence as set forth in SEQ ID NO: 142, a CDR5-L comprising an amino acid sequence of KVS, and a CDR6-L comprising an amino acid sequence as set forth in SEQ ID NO: 11; V.sub.D3 comprises a CDR1-H comprising an amino acid sequence as set forth in SEQ ID NO: 6, a CDR2-H comprising an amino acid sequence as set forth in SEQ ID NO: 7, and a CDR3-H comprising an amino acid sequence as set forth in SEQ ID NO: 8; and V.sub.D4 comprises a CDR4-H comprising an amino acid sequence as set forth in SEQ ID NO: 381, a CDR5-H comprising an amino acid sequence as set forth in SEQ ID NO: 384, and a CDR6-H comprising an amino acid sequence as set forth in SEQ ID NO: 382.

8: The method of claim 7, wherein: L.sub.1 comprises an amino acid sequence as set forth in SEQ ID NO: 389: L.sub.2 comprises an amino acid sequence as set forth in SEQ ID NO: 389; and L.sub.3, L.sub.4, and L.sub.5 are 0 amino acids in length.

9: The method of claim 7, wherein: Fc comprises an amino acid sequence as set forth in SEQ ID NO: 394; and Fc.sub.2 comprises an amino acid sequence as set forth in SEQ ID NO: 392.

10: The method of claim 7, wherein: V.sub.D1 comprises an amino acid sequence as set forth in SEQ ID NO: 385: V.sub.D2 comprises an amino acid sequence as set forth in SEQ ID NO: 141; V.sub.D3 comprises an amino acid sequence as set forth in SEQ ID NO: 138; V.sub.D4 comprises an amino acid sequence as set forth in SEQ ID NO: 383; C.sub.L comprises an amino acid sequence as set forth in SEQ ID NO: 310; and C.sub.H1 comprises an amino acid sequence as set forth in SEQ ID NO: 313.

11: The method of claim 7, wherein: (a) the first polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 391, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 391, which comprises a CDR1-L comprising an amino acid sequence as set forth in SEQ ID NO: 378, a CDR2-L comprising an amino acid sequence WAS, a CDR3-L comprising an amino acid sequence as set forth in SEQ ID NO: 379, a CDR4-L comprising an amino acid sequence as set forth in SEQ ID NO: 142, a CDR5-L comprising an amino acid sequence of KVS, and a CDR6-L comprising an amino acid sequence as set forth in SEQ ID NO: 11; and (b) the second polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 393, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 393, which comprises a CDR1-H comprising an amino acid sequence as set forth in SEQ ID NO: 6, a CDR2-H comprising an amino acid sequence as set forth in SEQ ID NO: 7, a CDR3-H comprising an amino acid sequence as set forth in SEQ ID NO: 8, a CDR4-H comprising an amino acid sequence as set forth in SEQ ID NO: 381, a CDR5-H comprising an amino acid sequence as set forth in SEQ ID NO: 384, and a CDR6-H comprising an amino acid sequence as set forth in SEQ ID NO: 382.

12: A method of treating a disease or disorder in a subject in need thereof, comprising: administering to the subject a therapeutically effective amount of an antibody-like binding protein that binds specifically to CD3ε and CD123 comprising two polypeptide chains that form two antigen-binding sites, wherein a first polypeptide has a structure represented by the formula [I]:
V.sub.D1-L.sub.1-V.sub.D2-L.sub.2-C.sub.L  [I] and a second polypeptide has a structure represented by the formula [II]:
V.sub.D3-L.sub.3-V.sub.D4-L.sub.4-C.sub.H1  [II] wherein: V.sub.D1 is a light chain variable domain of a first immunoglobulin; V.sub.D2 is a light chain variable domain of a second immuoglobulin; V.sub.D3 is a heavy chain variable domain of the second immunoglobulin; V.sub.D4 is a heavy chain variable domain of the first immunoglobulin; C.sub.L is a light chain constant domain of an immunoglobulin; C.sub.H1 is a C.sub.H1 heavy chain constant domain of an immunoglobulin; L.sub.1, L.sub.2, L.sub.3, and L.sub.4 are amino acid linkers; and wherein the first polypeptide and the second polypeptide form a cross-over light chain-heavy chain pair; and wherein V.sub.D1 consists of the amino acid sequence as set forth in SEQ ID NO: 385, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 385, which comprises a CDR1-L of sequence SEQ ID NO: 378, a CDR2-L of sequence WAS, and a CDR3-L of sequence SEQ ID NO: 379: V.sub.D2 consists of the amino acid sequence as set forth in SEQ ID NO: 141, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 141, which comprises a CDR1-L of sequence SEQ ID NO: 142, a CDR2-L of sequence KVS, and a CDR3-L of sequence SEQ ID NO: 11; V.sub.D3 consists of the amino acid sequence as set forth in SEQ ID NO: 138, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 138, which comprises a CDR1-H of sequence SEQ ID NO: 6, a CDR2-H of sequence SEQ ID NO: 7, and a CDR3-H of sequence SEQ ID NO: 8; and V.sub.D4 consists of the amino acid sequence as set forth in SEQ ID NO: 383, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 383, which comprises a CDR1-H of sequence SEQ ID NO: 381, a CDR2-H of sequence SEQ ID NO: 384, and a CDR3-H of sequence SEQ ID NO: 382.

13: The method of claim 12, wherein: (a) the first polypeptide consists of the amino acid sequence as set forth in SEQ ID NO: 388, which comprises V.sub.D1 of amino acid sequence SEQ ID NO: 385, L.sub.1 of amino acid sequence SEQ ID NO: 389, V.sub.D2 of amino acid sequence SEQ ID NO: 141, L.sub.2 of amino acid sequence SEQ ID NO: 389, and C.sub.L of amino acid sequence SEQ ID NO: 310: or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 388, wherein the three CDRs of amino acid sequences SEQ ID NO: 378, WAS, and SEQ ID NO: 379 of V.sub.D1, and the three CDRs of amino acid sequences SEQ ID NO: 142, KVS, and SEQ ID NO: 11 of V.sub.D2 are unaltered; and (b) the second polypeptide consists of the amino acid sequence as set forth in SEQ ID NO: 390, which comprises V.sub.D3 of amino acid sequence SEQ ID NO: 138, L.sub.3 is 0 amino acids, V.sub.D4 of amino acid sequence SEQ ID NO: 383, L.sub.4 is 0 amino acids, and C.sub.H1 of amino acid sequence SEQ ID NO: 313, or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 390, wherein the three CDRs of amino acid sequences SEQ ID NO: 381, SEQ ID NO: 384, and SEQ ID NO: 382 of V.sub.D4, and the three CDRs of amino acid sequences SEQ ID NO: 6, SEQ ID NO: 7, and SEQ ID NO: 8 of V.sub.D3 are unaltered.

14: The method of claim 12, wherein the polypeptide of formula [II] further comprises a Fc domain and has a structure represented by formula [III]:
V.sub.D3-L.sub.3-V.sub.D4-L.sub.4-C.sub.H1-Fc  [III] wherein the polypeptide of formula [I] further comprises a Fc domain (Fc.sub.2) and has a structure represented by formula [IV]:
V.sub.D1-L.sub.1-V.sub.D2-L.sub.2-C.sub.L-L.sub.5-Fc.sub.2  [IV] wherein: Fc is the immunoglobulin hinge region and CH.sub.2, CH.sub.3 immunoglobulin heavy chain constant domains of an immunoglobulin, Fc.sub.2 is the immunoglobulin hinge region and CH.sub.2, CH.sub.3 immunoglobulin heavy chain constant domains of an immunoglobulin, and L.sub.5 is an amino acid linker; and wherein: (a) the polypeptide according to formula [IV] consists of the amino acid sequence as set forth in SEQ ID NO: 391, which comprises V.sub.D1 of amino acid sequence SEQ ID NO: 385, L.sub.1 of amino acid sequence SEQ ID NO: 389, V.sub.D1 of amino acid sequence SEQ ID NO: 141, L.sub.2 of amino acid sequence SEQ ID NO: 389, C.sub.L of amino acid sequence SEQ ID NO: 310, L.sub.5 which is 0 amino acids, and Fc.sub.2 of amino acid sequence SEQ ID NO: 392; or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 391, wherein the three CDRs of amino acid sequences SEQ ID NO: 378, WAS, and SEQ ID NO: 379 of V.sub.D1, and the three CDRs of amino acid sequences SEQ ID NO: 142, KVS, and SEQ ID NO: 11 of V.sub.D2 are unaltered; and (b) the polypeptide according to formula [III] consists of the amino acid sequence as set forth in SEQ ID NO: 393, which comprises V.sub.D3 of amino acid sequence SEQ ID NO: 138, L.sub.3 is 0 amino acids, V.sub.D4 of amino acid sequence SEQ ID NO: 383, L.sub.4 is 0 amino acids, C.sub.H1 of amino acid sequence SEQ ID NO: 313, and Fc of amino acid sequence SEQ ID NO: 394: or an amino acid sequence that is at least 85% identical to the amino acid sequence as set forth in SEQ ID NO: 393, wherein the three CDRs of amino acid sequences SEQ ID NO: 381, SEQ ID NO: 384, and SEQ ID NO: 382 of V.sub.D4, and the three CDRs of amino acid sequences SEQ ID NO: 6, SEQ ID NO: 7, and SEQ ID NO: 8 of V.sub.D3 are unaltered; and wherein the polypeptide of formula [IV] and the polypeptide of formula [III] form a cross-over light chain-heavy chain pair.

15: A method of treating a disease or disorder in a subject in need thereof, comprising: administering to the subject a therapeutically effective amount of an antibody-like binding protein comprising two polypeptide chains that form two antigen-binding sites, wherein: a first polypeptide chain comprises: a CDR1-L comprising an amino acid sequence as set forth in SEQ ID NO: 378; a CDR2-L comprising an amino acid sequence WAS: a CDR3-L comprising an amino acid sequence as set forth in SEQ ID NO: 379; a CDR4-L comprising an amino acid sequence as set forth in SEQ ID NO: 142: a CDR5-L comprising an amino acid sequence of KVS; and a CDR6-L comprising an amino acid sequence as set forth in SEQ ID NO: 11: and wherein a second polypeptide chain comprises: a CDR1-H comprising an amino acid sequence as set forth in SEQ ID NO: 6: a CDR2-H comprising an amino acid sequence as set forth in SEQ ID NO: 7; a CDR3-H comprising an amino acid sequence as set forth in SEQ ID NO: 8; a CDR4-H comprising an amino acid sequence as set forth in SEQ ID NO: 381; a CDR5-H comprising an amino acid sequence as set forth in SEQ ID NO: 384; and a CDR6-H comprising an amino acid sequence as set forth in SEQ ID NO: 382.

16: The method of claim 1, wherein the disease or disorder is cancer or a pathological immune response.

17: The method of claim 16, wherein the cancer is a hematological cancer associated with CD123 expression.

18: The method of claim 17, wherein the hematological cancer is a leukemia, a malignant lymphoproliferative condition, a blastic plasmacytoid dendritic cell neoplasm (BPDCN), systemic mastocytosis, or a lymphoma.

19: The method of claim 18, wherein the leukemia is acute myelogenous leukemia, chronic myelogenous leukemia, acute lymphoid leukemia, chronic lymphoid leukemia, hairy cell leukemia, or myelodysplasia syndrome.

20: The method of claim 19, wherein the leukemia is acute myelogenous leukemia.

21: The method of claim 18, wherein the lymphoma is multiple myeloma, non-Hodgkin's lymphoma, Burkitt's lymphoma, small cell-follicular lymphoma, or large cell-follicular lymphoma.

22: The method of claim 16, wherein the pathological immune response comprises an autoimmune disease, a transplantation-related disease, or an inflammation-associated disease.

23: The method of claim 22, wherein the autoimmune disease is Crohn's disease, ulcerative colitis, or type I diabetes.

24: The method of claim 22, wherein the transplantation-related disease is graft-versus-host disease (GVHD).

25: The method of claim 1, wherein the antibody-like binding protein is comprised within a pharmaceutical composition that further comprises a pharmaceutically acceptable carrier.

26: The method of claim 1, wherein the antibody-like binding protein induces T-cell mediated toxicity of CD123 positive tumor cells within the subject.

27: The method of claim 1, wherein the antibody-like binding protein induces an immunosuppressive effect in the subject.

28: A method of treating a disease or disorder in a subject in need thereof, comprising: administering to the subject a therapeutically effective amount of an antibody-like binding protein that specifically binds to CD3ε and CD123 comprising two polypeptide chains that form two antigen-binding sites, wherein a first polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 388 and a second polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 390 or the first polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 391 and the second polypeptide chain comprises an amino acid sequence as set forth in SEQ ID NO: 393.

29: An isolated antibody that binds to the extracellular domain of human CD3ε protein comprising: a) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence GFX.sub.1X.sub.2X.sub.3X.sub.4AW (SEQ ID NO: 331), wherein X.sub.1 is T or S, X.sub.2 is F or V, X.sub.3 is S or T and X.sub.4 is N, K, L, or Y, or any combination thereof; a CDR2-H comprising amino acid sequence IKX.sub.1X.sub.2X.sub.3NX.sub.4YX.sub.5T (SEQ ID NO: 332), wherein X.sub.1 is A or D, X.sub.2 is K or R, X.sub.3 is S or A, X.sub.4 is N or S, and X.sub.5 is A or E, or any combination thereof; and a CDR3-H comprising amino acid sequence TWRHYYSSHTMDA (SEQ ID NO: 69), or RALTYYGYKRDAMDG (SEQ ID NO: 129), or RX.sub.1X.sub.2X.sub.3YX.sub.4X.sub.5X.sub.6X.sub.7X.sub.8X.sub.9X.sub.10X.sub.11DX.sub.12 (SEQ ID NO: 333), wherein X.sub.1 is Y, G, or A, X.sub.2 is V, T, or L, X.sub.3 is H, N, Y, or Q, X.sub.4 is G, R, or A, X.sub.5 is F or V or no amino acid, X.sub.6 is R or no amino acid, X.sub.7 is F, S, or I or no amino acid, X.sub.8 is F, L, N, M, Y, S, A, or G, X.sub.9 is Y, A, K, S, N, T, F, or L, X.sub.10 is A, P, G, or T, X.sub.11 is M, L, F, or S, and X.sub.12 is A, V, or Y, or any combination thereof; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence QX.sub.1LX.sub.2HX.sub.3NGX.sub.4TY (SEQ ID NO: 334) wherein X.sub.1 is R or S, X.sub.2 is V or E, X.sub.3 is N, D, or T, and X.sub.4 is N or Y, or any combination thereof; and a CDR2-L comprising amino acid sequence KVS; and a CDR3-L comprising an amino acid sequence GQGX.sub.1X.sub.2YPFT (SEQ ID NO: 335) wherein X.sub.1 is T, A or S and X.sub.2 is H, E or Q, or any combination thereof; or b) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 30 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 30 by one amino acid substitution; a CDR2-H comprising amino acid sequence SEQ ID NO: 31 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 31 by one or more amino acid substitutions; and a CDR3-H comprising amino acid sequence SEQ ID NO: 32 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 32 by one amino acid substitution; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 34 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 34 by one amino acid substitution; a CDR2-L comprising amino acid sequence RDD or an amino acid sequence differing from the amino acid sequence RDD by one amino acid substitution; and a CDR3-L comprising amino acid sequence SEQ ID NO: 35 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 35 by one amino acid substitution; or c) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 50 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 50 by one amino acid substitution; a CDR2-H comprising amino acid sequence SEQ ID NO: 51 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 51 by one or more amino acid substitutions; and a CDR3-H comprising amino acid sequence SEQ ID NO: 52 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 52 by one amino acid substitution; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 54 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 54 by one amino acid substitution; a CDR2-L comprising amino acid sequence NAN or an amino acid sequence differing from the amino acid sequence NAN by one amino acid substitution; and a CDR3-L comprising amino acid sequence SEQ ID NO: 55 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 55 by one amino acid substitution; or d) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 90 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 90 by one amino acid substitution; a CDR2-H comprising amino acid sequence SEQ ID NO: 91 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 91 by one or more amino acid substitutions; and a CDR3-H comprising amino acid sequence SEQ ID NO: 32 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 32 by one amino acid substitution; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 93 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 93 by one amino acid substitution; a CDR2-L comprising amino acid sequence GAS or an amino acid sequence differing from the amino acid sequence GAS by one amino acid substitution; and a CDR3-L comprising amino acid sequence SEQ ID NO: 94 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 94 by one amino acid substitution; or e) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 96 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 96 by one amino acid substitution; a CDR2-H comprising amino acid sequence SEQ ID NO: 97 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 97 by one or more amino acid substitutions; and a CDR3-H comprising amino acid sequence SEQ ID NO: 98 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 98 by one amino acid substitution; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 100 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 100 by one amino acid substitution, a CDR2-L comprising amino acid sequence NTN or an amino acid sequence differing from the amino acid sequence NTN by one amino acid substitution; and a CDR3-L comprising amino acid sequence SEQ ID NO: 101 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 101 by one amino acid substitution; or f) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 103 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 103 by one amino acid substitution; a CDR2-H comprising amino acid sequence SEQ ID NO: 104 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 104 by one or more amino acid substitutions; and a CDR3-H comprising amino acid sequence SEQ ID NO: 105 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 105 by one amino acid substitution; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: by one amino acid substitution; a CDR2-L comprising amino acid sequence KVS or an amino acid sequence differing from the amino acid sequence KVS by one amino acid substitution; and a CDR3-L comprising amino acid sequence SEQ ID NO: 11 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 11 by one amino acid substitution; or g) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 116 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 116 by one amino acid substitution; a CDR2-H comprising amino acid sequence SEQ ID NO: 117 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 117 by one or more amino acid substitutions; and a CDR3-H comprising amino acid sequence SEQ ID NO: 118 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 118 by one amino acid substitution; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 100 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 100 by one amino acid substitution; a CDR2-L comprising amino acid sequence VTN or an amino acid sequence differing from the amino acid sequence VTN by one amino acid substitution; and a CDR3-L comprising amino acid sequence SEQ ID NO: 120 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 120 by one amino acid substitution; or h) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 122 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 122 by one amino acid substitution; a CDR2-H comprising amino acid sequence SEQ ID NO: 123 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 123 by one or more amino acid substitutions; and a CDR3-H comprising amino acid sequence SEQ ID NO: 124 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 124 by one amino acid substitution; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 126 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 126 by one amino acid substitution; a CDR2-L comprising amino acid sequence RDD or an amino acid sequence differing from the amino acid sequence RDD by one amino acid substitution; and a CDR3-L comprising amino acid sequence SEQ ID NO: 127 or an amino acid sequence differing from the amino acid sequence SEQ ID NO: 127 by one amino acid substitution.

30: The isolated antibody according to claim 29, comprising: a) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 6, a CDR2-H comprising amino acid sequence SEQ ID NO: 7, and a CDR3-H of amino acid sequence SEQ ID NO: 8; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10 or SEQ ID NO: 142, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or b) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 14, and a CDR3-H comprising amino acid sequence SEQ ID NO: 15; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 17 or SEQ ID NO: 184, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or c) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 19, and a CDR3-H comprising amino acid sequence SEQ ID NO: 20; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 22, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or d) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 24, a CDR2-H comprising amino acid sequence SEQ ID NO: 19, and a CDR3-H comprising amino acid sequence SEQ ID NO: 25; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 27; a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 28; or e) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 30, a CDR2-H comprising amino acid sequence SEQ ID NO: 31, and a CDR3-H comprising amino acid sequence SEQ ID NO: 32: and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 34, a CDR2-L comprising amino acid sequence ROD, and a CDR3-L comprising amino acid sequence SEQ ID NO: 35; or f) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 38; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 28; or g) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 41; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 17, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or h) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 44; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or i) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 47; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or j) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 50, a CDR2-H comprising amino acid sequence SEQ ID NO: 51, and a CDR3-H comprising amino acid sequence SEQ ID NO: 52; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 54, a CDR2-L comprising amino acid sequence NAN, and a CDR3-L comprising amino acid sequence SEQ ID NO: 55; or k) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 57, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 58; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 28; or l) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 61; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or m) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 64, a CDR2-H comprising amino acid sequence SEQ ID NO: 65, and a CDR3-H comprising amino acid sequence SEQ ID NO: 47: and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 67, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 28; or n) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 69; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 71; or o) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 84; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 17, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or p) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 75, a CDR2-H comprising amino acid sequence SEQ ID NO: 76, and a CDR3-H comprising amino acid sequence SEQ ID NO: 77; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or q) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 80, a CDR2-H comprising amino acid sequence SEQ ID NO: 76, and a CDR3-H comprising amino acid sequence SEQ ID NO: 81; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or r) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 84; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or s) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 47: and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 88; or t) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 90, a CDR2-H comprising amino acid sequence SEQ ID NO: 91, and a CDR3-H comprising amino acid sequence SEQ ID NO: 32; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 93, a CDR2-L comprising amino acid sequence GAS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 94; or u) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 96, a CDR2-H comprising amino acid sequence SEQ ID NO: 97, and a CDR3-H comprising amino acid sequence SEQ ID NO: 98; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 100, a CDR2-L comprising amino acid sequence NTN, and a CDR3-L comprising amino acid sequence SEQ ID NO: 101; or v) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 103, a CDR2-H comprising amino acid sequence SEQ ID NO: 104, and a CDR3-H comprising amino acid sequence SEQ ID NO: 105; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or w) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 80, a CDR2-H comprising amino acid sequence SEQ ID NO: 19, and a CDR3-H comprising amino acid sequence SEQ ID NO: 108; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or x) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 37, and a CDR3-H comprising amino acid sequence SEQ ID NO: 111; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 113, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 114; or y) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 116, a CDR2-H comprising amino acid sequence SEQ ID NO: 117, and a CDR3-H comprising amino acid sequence SEQ ID NO: 118; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 100, a CDR2-L comprising amino acid sequence VTN, and a CDR3-L comprising amino acid sequence SEQ ID NO: 120; or z) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 122, a CDR2-H comprising amino acid sequence SEQ ID NO: 123, and a CDR3-H comprising amino acid sequence SEQ ID NO: 124; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 126, a CDR2-L comprising amino acid sequence ROD, and a CDR3-L comprising amino acid sequence SEQ ID NO: 127; or aa) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 13, a CDR2-H comprising amino acid sequence SEQ ID NO: 19, and a CDR3-H comprising amino acid sequence SEQ ID NO: 129; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 10, a CDR2-L comprising amino acid sequence KVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 11; or bb) a heavy chain variable domain comprising a CDR1-H comprising amino acid sequence SEQ ID NO: 103, a CDR2-H comprising amino acid sequence SEQ ID NO: 104, and a CDR3-H comprising amino acid sequence SEQ ID NO: 105; and a light chain variable domain comprising a CDR1-L comprising amino acid sequence SEQ ID NO: 133, a CDR2-L comprising amino acid sequence LVS, and a CDR3-L comprising amino acid sequence SEQ ID NO: 134.

31: An antibody-like binding protein that binds specifically to human CD3ε comprising two polypeptide chains that form two antigen-binding sites, wherein a first polypeptide has a stricture represented by the formula [I]:
VD-L.sub.1-V.sub.D1-L.sub.2-C.sub.L  [I] and a second polypeptide has a structure represented by the formula [II]:
V.sub.D3-L.sub.3-V.sub.D4-L.sub.4-C.sub.H1  [II] wherein: V.sub.D1 is a heavy or light chain variable domain of a first immunoglobulin; V.sub.D2 is a heavy or light chain variable domain of a second immuoglobulin; V.sub.D3 is a heavy or light chain variable domain of the second immunoglobulin; V.sub.D4 is a heavy or light chain variable domain of the first immunoglobulin; C.sub.L is a light chain constant domain of an immunoglobulin; C.sub.H1 is a C.sub.H1 heavy chain constant domain of an immunoglobulin; L.sub.1, L.sub.2, L.sub.3, and L.sub.4 are amino acid linkers; wherein the first polypeptide and the second polypeptide form a cross-over light chain-heavy chain pair; and wherein V.sub.D1 and V.sub.D4, or V.sub.D2 and V.sub.D3 comprise a heavy chain variable domain and a light chain variable domain of an antibody according to claim 29.

32: The antibody-like binding protein of claim 31, wherein at least one of polypeptide [III] and polypeptide [IV] further comprises an Fc domain.

Description

FIGURES

[1286] FIG. 1: Sequence alignments of the VH regions of the so-called “20G6-F3”, “4B4-D7”, “4E7-C9”, “18F5-H10”, “11D7-C3”, “11H3-E5”, “13H2-C2”, “13C1-F6”, “1E6-C9”, “10F4-C10”, “10E6-G6”, “18G9-H11”, “11F3-B9”, “12G3-E8”, “5B1-G2”, “16F8-A7”, “11F9-F8”, “20E5-F10”, “20B5-F10”, “3H6-D2” anti-CD3 antibodies.

[1287] FIG. 2: Sequence alignments of the VL regions of so-called “20G6-F3”, “4B4-D7”, “4E7-C9”, “18F5-H10”, “11D7-C3”, “11H3-E5”, “13H2-C2”, “13C1-F6”, “1E6-C9”, “10F4-C10”, “10E6-G6”, “18G9-H11”, “11F3-B9”, “12G3-E8”, “5B1-G2”, “16F8-A7”, “11F9-F8”, “20E5-F10”, “20B5-F10”, “3H6-D2” anti-CD3 antibodies.

[1288] FIG. 3: Sequence alignments of the VH and VL regions of the so-called “1E1-G5”, “6D6-B8”, “8B11-B7”, “, “9F6-G3” anti-CD123 antibodies.

[1289] FIG. 4: Sequence alignments of the VH and VL regions of so-called “6C10-C4”, “9B8-G6”, “9D7-C8” anti-CD123 antibodies.

[1290] FIGS. 5 and 7: Fully human CODV-Fab-TL1 “hz20G6xhz7G3” IV Q3d in presence of human T cells inhibits Molm13 tumor growth in whole body at all tested doses.

[1291] FIGS. 6 and 8: Fully human CODV-Fab-TL1 “hz20G6xhz7G3” IV Q3d in presence of human T cells is associated with tumor regression in long bones at all tested doses.

[1292] FIGS. 9 and 11: Fully human CODV-Fab “hz20G6xhz7G3” IV in presence of human T cells inhibited tumor growth.

[1293] FIGS. 10 and 12: Fully human CODV-Fab “hz20G6xhz7G3” IV in presence of human T cells was associated with tumor regression in long bones.

[1294] FIGS. 13 and 15: Fully human CODV-Fab “hz20G6xhz7G3” CIP in presence of human T cells inhibited whole body tumor growth at dosage of 0.13 nmol/Kg/day or above.

[1295] FIGS. 14 and 16: Fully human CODV-Fab “hz20G6xhz7G3” CIP in presence of human T cells inhibited tumor growth in long bones at dosage of 0.13 nmol/Kg/day or above.

[1296] FIG. 17: Diagrammatic representation of the structure of the CODV-Ig, CODV-Fab-TL and CODV-Fab-OL (further showing LALA mutations (when Fc of IgG1 backbone is used) and Knob-into-Hole mutations).

EXAMPLES

Example 1: Antibody Generation

[1297] 1.1 Construction of hCD3ε/δ-hFc Fusion Expression Plasmid (CD3ed-Fc)

[1298] Using cDNA containing plasmids as a template, human and Macaca fascicularis CD3ε and CDδ fusion proteins were generated, as described herein below in detail, in reading frame with heavy chain constant region including the hinge region, CH2 and CH3 domains of human immunoglobulin IgG additionally carrying a 8×His or Strep-II tag for optional tandem purification.

[1299] Using human genomic DNA as template, human CD3ε and human CDδ subunit extracellular domains were amplified, including the signal sequence. The resulting amplified cleaved and purified PCR products were combined by ligation PCR and ligated into mammalian expression vector pXL by InFusion method using NheI and HindIII site. Each subunit was cloned on one plasmid. The sequence of the resulting mature human CD3ε His-tagged Fc-fusion protein is herein disclosed under SEQ ID NO: 3. Amino acids 1 to 104 of SEQ ID NO: 3 correspond to amino acids 23 to 126 of the wild-type full-length human CD3ε (herein disclosed under SEQ ID NO: 1, available in Uniprot database under accession number P07766) protein and thus the extracellular domain of human CD3ε.

[1300] Using cynomolgus monkey genomic DNA as template, Macaca fascicularis CD3ε and CD3δ extracellular domains were amplified, including the signal sequence. The resulting amplified cleaved and purified PCR products were combined by ligation PCR and ligated into mammalian expression vector pXL by InFusion method using NheI and HindIII. Each subunit was cloned on one plasmid. The resulting sequences for mature Macaca fascicularis CD3ε Fc-fusion protein is disclosed under SEQ ID NO: 4. Amino acids 1 to 95 of SEQ ID NO: 3 correspond to amino acids 23 to 117 of the full-length Macaca fascicularis CD3ε protein and thus comprises the extracellular domain of wild-type full-length Macaca fascicularis CD3ε (herein disclosed under SEQ ID NO: 2, available in Uniprot database under accession number Q95L.sub.15). The cloned fusion protein further contains one Alanine to Valine exchange at the amino acid position 35 in comparison to amino acid position 57 of the wild-type sequence.

[1301] 1.2 Expression and Purification of Human and Cyno CD3ed-Fc

[1302] Freestyle HEK293 cells growing in F17 serum free suspension culture (Life) were transiently transfected with the expression plasmid. Co-transfection of both plasmids representing the CD3ε and CD3δ extracelullar domain (ECD) subunit were performed using Cellfectin transfection reagent (Life). The cells were cultured at 37° C. for 7 days. The culture supernatant containing recombinant protein was harvested by centrifugation and was clarified by filtration (0.22 μm).

[1303] For purification, the Fc-fusion protein variants were captured on protein A matrix (GE) and were eluted by pH shift. After polishing the protein by size exclusion chromatography (SEC) using a Superdex 200 (GE) and a final ultrafiltration concentration step the protein was used for further assays.

[1304] The human heterodimer was additionally applied on His-Trap column (GE) after capture on protein A and desalted. The eluted protein was applied to a Strepavidin collumn (GE) and eluted with d-desthiobiotin before final pollishing by SEC using a Superdex 200 (GE). This strategy was used to isolate heterodimers from homodimers.

[1305] 1.3 Generation of Human/Cynomolgous Monkey Cross-Reactive Anti-CD3 Antibodies

[1306] Human and Macaca fascicularis CD3ε and CD3 δ cDNAs were cloned into Aldevron proprietary immunization vectors (pB8 and VV8) respectively, and used for genetic immunization of rats. Rats of immunization group MR12-266 (“CD3-cyno”) were initially immunized with human CD3ε and CD3δcDNAs, followed by further two immunizations with a mixture of human and Macaca fascicularis CD3ε and CD3 δcDNAs. The immune serum was taken at day 24 of the immunization protocol, after 4 genetic applications (IS24d-4). Sera, diluted in PBS 1% BSA, were tested by flow cytometry using mammalian cells transiently transfected with the target cDNAs in co-transfection experiments to obtain human and Macaca fascicularis CD3ε and CD3δ TCR complexes. In addition the immune sera were tested on the following cell lines: Jurkat E6-1 (expressing human TCR), Jurkat-RT-T3.5 (negative for TCR) and cyno HSC-F (expressing cyno TCR); no negative cell line for the cyno TCR was available. A goat anti-rat IgG R-phycoerythrin conjugate (Southern Biotech, #3030-09) at 10 pg/ml was used as a secondary antibody.

[1307] Specific reactivity of the immune sera especially against cells transfected with combinations of CD3ε and CD3δcDNAs could be detected in the immunized animals when compared to cells transfected with an irrelevant cDNA. The same is valid for the test on the TCR-positive Jurkat cell line (E6-1), when compared to the TCR-negative cell line (RT-T3.5), much lesser signals but a still significant signals were detected on the cynomologues HSC-F cell line (see table 1).

TABLE-US-00095 TABLE 1 Detection of the specific reactivity of the immune sera against TCR- complex in each animal by flow cytometry. Data is presented as geometric means of the relative fluorescence intensities (rfu). pXL-CD3e- pXL-CD3e- hum + pXL- cyno + pXL- Jurkat- Jurkat- cyno CD3d-hum CD3d-cyno pB1_irrelevant E6-1 RT-T3.5 HSC-F w/o 5 5 4 7 10 6 primary ab OKT3 — — — 858 25 — FN18 — — 4 7 10 1256 Mouse-anti-HA 6 6 — — — — Rat-anti-myc 9 7 1033   10 14 9 rat #1 1:1000 290 273 5 186 36 55 1:5000 79 86 5 65 21 20 rat #2 1:1000 499 401 5 390 55 75 1:5000 147 110 4 126 28 25 rat #3 1:1000 239 247 5 108 30 31 1:5000 89 92 4 32 17 9 rat #4 1:1000 280 283 5 176 37 73 1:5000 97 110 4 55 22 23 rat #5 1:1000 208 137 4 99 31 11 1:5000 61 65 4 47 19 7

[1308] Rats of immunization group MR12-265 (“CD3-hum”) were co-immunized with human CD3ε and CD3δ cDNAs cloned into the corresponding expression vectors. The immune serum was taken at day 24 of the immunization protocol, after 4 genetic applications (IS24d-4). Sera, diluted in PBS 1% BSA, were tested by flow cytometry using mammalian cells transiently transfected with the above mentioned target cDNAs in co-transfection experiments to obtain human and Macaca fascicularis CD3ε and CD3δ TCR complexes. A goat anti-rat IgG R-phycoerythrin conjugate (Southem Biotech, #3030-09) at 10 μg/ml was used as a secondary antibody. In addition the immune sera were tested on the following cell lines: Jurkat E6-1 (expressing human TCR), Jurkat-RT-T3.5 (negative for TCR) and cyno HSC-F (expressing cyno TCR); no negative cell line for the cyno TCR was available. A goat anti-rat IgG R-phycoerythrin conjugate (Southem Biotech, #3030-09) at 10 μg/ml was used as a secondary antibody.

[1309] Specific reactivity of the immune sera especially against cells transfected with combinations of CD3ε and CD3δ cDNAs could be detected in the immunized animals when compared to cells transfected with an irrelevant cDNA. The same is valid for the test on the TCR-positive Jurkat cell line (E6-1), when compared to the TCR-negative cell line (RT-T3.5), but much lesser signals (however in the positive rats still significant) were detected on the cyno HSC-F cell line (see table 2)

TABLE-US-00096 TABLE 2 Detection of the specific reactivity of the immune sera against TCR- complex in each animal by flow cytometry. Data is presented as geometric means of the relative fluorescence intensities (rfu). pXL-CD3e- pXL-CD3e- hum + pXL- cyno + pXL- Jurkat- Jurkat- cyno CD3d-hum CD3d-cyno pB1_irrelevant E6-1 RT-T3.5 HSC-F w/o 5 5 4 7 10 6 primary ab OKT3 — — — 858 25 — FN18 — — 4 7 10 1256 Mouse-anti-HA 6 6 — — — — Rat-anti-myc 9 7 1033   10 14 9 rat #1 1:1000 230 132 5 170 33 50 1:5000 78 40 4 54 19 15 rat #2 1:1000 216 136 6 216 48 32 1:5000 75 40 4 66 24 11 rat #3 1:1000 73 54 5 56 25 7 1:5000 25 20 4 23 15 6 rat #4 1:1000 70 56 5 41 25 8 1:5000 25 18 4 15 15 6 rat #5 1:1000 65 37 5 42 20 7 1:5000 24 17 4 17 14 6

[1310] Rats with positive sera were scarified and B cells were fused with mouse myeloma cells. The resulting hybrndomas were screened on HEK293 cells transfected with human or cynomoligus CD3ε and CD3δ expression plasmids, on Jurkat E6.1 (CD3+) and Jurkat T3.5 (CD3-) by flow cytometry. Supernatant of Hybridoma clones was used to assess by surface plasmon resonance single point kinetic against human and Macaca fascicularis CD3ε/δ complex by fixing the analyte at 25 nM (data presented in table 3).

TABLE-US-00097 TABLE 3 Detection of the specific reactivity of different hybridoma clone supernatants by flow cytometry and Biacore analysis. Supernatants were tested on different cell lines (Jurkat E6-1, Jurkat-RT-T3.5 and cyno HSC-F). Biacore analysis was performed against CD3e/d complexes from human and cynomolgous monkey respectively. Biacore human CD3ε/ Biacore cyno CD3ε/ Sample-ID δ [RU] at 25 nM δ [RU] at 25 nM 3G5-E10 27 3 18G9-H11 7 −7 12D2-E5 27 5 6C9-C9 26 32 11F9-F8 11.5 16 8H2-F3 32.5 40 4B4-D7 11 14 4E7-C9 2.5 10 13C1-F6 17 15.5 11D7-C3 17 10 13H2-C3 16 7 10E6-G6 10.5 0 12G3-E8 24 17 11F3-B9 22 17 20E3-B5 11 5 10F4-C10 22 27 11H3-E5 7 9 18F5-H10 n.a. n.a. 18H11-F10 35 55 20E5-F10 12 8 3H6-D2 9.5 7 8C2-F7 4.5 3 5B1-G2 1 0.5 7B7-F10 28 42 20B5-F10 1 9.5 16D3-E4 −2 −2 1E6-B8 n.a. n.a. 16F8-A7 17.5 22.5 11A8-D6 3E8-G1 20G6-F3 9D7-B5 n.a. n.a. 9G5-G10 17A11-D3

[1311] Positive clones were expanded and respective cDNAs for variable heavy and light chains were isolated by RT-PCR. VH and VL sequences were cloned into expression vectors in fusion with either the human CH1, IGHG1-backbone or kappa chain in order to express Fab-fragments as well as full IgGs.

[1312] 1.4 Expression of IgGs and Fab-fragments

[1313] The expression plasmids encoding the heavy and light chain of the IgGs and Fab-fragments were propagated in E. Coli NEB 10-beta (DH10B derivative). Plasmids used for transfection were prepared from E. Coli using the QIAGEN Plasmid Plus Kit (Cat. No: 12991).

[1314] HEK 293-FS cells growing in Freestyle Medium (Invitrogen) were transfected with indicated LC and HC plasmids encoding the heavy chains and light chains using 293fectin (Invitrogen) transfection reagent as described by the manufacturer. Cells were cultivated at 37° C. in a Kuhner ISF1-X shaking incubator at 110 rpm with 8% C02. After 7 days of cultivation cells were removed by centrifugation, 10% Vol/Vol 1M Tris HCl pH 8.0 was added and the supernatant was filtered via a 0.2NM bottle top filter to remove particles. CODV-IgG1 constructs were purified by affinity chromatography on Protein A columns (HiTrap Protein A HP Columns, GE Life Sciences). After elution from the column with 0.1M Citrat, pH 3.0, the CODV-IgG1 constructs were desalted using HiPrep 26/10 Desalting Columns, formulated in PBS (Gibco 14190-136).

[1315] Bispecific CODV-Fab constructs were purified by HisTrap High Performance columns (GE Healthcare, Cat. No: 17-5248-02). After elution from the column (Elution buffer 20 mM sodium phosphate, 0.5 M NaCl, 500 mM imidazole, pH 7.4), the protein containing fractions were pooled and desalted using HiPrep 26/10 Desalting Columns, formulated in PBS (Gibco 14190-136).

[1316] To separate monomers from aggregates a high resolution fractionation step in PBS (Gibco 14190-136) for both constructs, the CODV-IgG and the CODV-Fab fragment, was performed, using a HiLoad Superdex 200 26/60 320 ml column (GE Healthcare Cat. No: 29-9893-36). Monomeric fractions were pooled and concentrated up to 1 mg/ml, using Vivaspin 20 centrifugation columns (VS2002 Sartorius Stedim biotech) and filtered using a 0.22 μm membrane (Millex® Syringe Filters SLGV033RS). Protein concentration was determined by measurement of absorbance at 280 nm. Each batch was analyzed by SDS-PAGE under reducing and non-reducing conditions to determine the purity and molecular weight of each subunit and of the monomer.

[1317] 1.5 Assessment of Affinities of the Anti-CD3 Antibodies

1.5.1 Assessment of Affinities to Both Human and Macaca fascicularis CD3ε/δ

[1318] Binding affinities of anti-CD3 binding Fabs or CODV-Fabs were measured by surface plasmon resonance (SPR) using a Biacore3000 instrument (GE Healthcare). Assay buffer was HBS-EP (BR-1001-88, GE Healthcare). Capture of CD3ε/δ-Fc fusion proteins was achieved using the human antibody capture kit (GE Healthcare). The capture antibody was coupled to CM5 chips (BR-1001-88, GE Healthcare) to approx. 12.000 RU using the amine coupling kit (BR-100-50, GE Healthcare). The CD3εδ-Fc fusions proteins were captured at 10 μl/min to approx. 70 RU to yield Rmax values of 30 RU. Binding kinetics with the anti-CD3 Fabs or CODV-Fabs was measured at 30 μl/min for 240 sec. and 600 sec. for association and dissociation phase, respectively. Twofold dilutions of Fabs from 3 to 400 nM in assay buffer were used. All Fab concentrations were run in duplicate together with duplicate buffer blanks for double referencing. Regeneration of the capture surface was performed with a 1 min injection of 3M MgCl2 solution at 30 μl/min. For data analysis the BIAevaluation software v.4.1 (GE Healthcare) was used. Data were fit globally using a 1:1 Langmuir model with mass transfer.

[1319] Measurement of binding affinities of anti-CD3 IgGs and CODV-Fc proteins was performed analogous to the binding assay for Fabs and CODV-Fabs with the exception of the capture antibody. In this case the His Capture Kit (28-9950-56, GE Healthcare) was used to capture human CD3-Fc protein via His-tag. For binding assay of with Macaca fascicularis CD3-Fc the Strep-MAB classic antibody (2-1507-001, IBA) was used as the capture antibody. In this case the regeneration solution was 10 mM Glycine buffer pH2.0.

TABLE-US-00098 TABLE 4 Affinities of selected CD3 antibodies against human and Macaca fascicularis CD3ε/δ complexes measured by Biacore. Clone ka kd Rmax KD Ratio CD3εδ (Fab fragment) (1/Ms) (1/s) (RU) (M) Chi2 KD(cy)/KD(hu) human 12D2-E5 5.4E+06 6.2E−03 35 1.2E−09 0.34 4 cyno 12D2-E5 steady state steady state 8 4.8E−09 0.129 human 10E6-G6 1.4E+05 2.0E−03 37 1.4E−08 0.218 1 cyno 10E6-G6 1.9E+05 2.4E−03 18 1.3E−08 0.218 human 4E7-C9 8.9E+04 1.6E−03 29 1.7E−08 0.244 1 cyno 4E7-C9 9.8E+04 1.5E−03 20 1.5E−08 0.232 human 12G3-E8 1.1E+05 1.1E−03 30 9.3E−09 0.307 1 cyno 12G3-E8 1.7E+05 1.8E−03 15 1.0E−08 0.349 human 5B1-G2 6.0E+04 3.9E−04 36 6.5E−09 0.218 1 cyno 5B1-G2 3.4E+04 1.5E−04 20 4.3E−09 2.1 human 18F5-H10 2.0E+05 1.8E−04 28 8.9E−10 0.235 1 cyno 18F5-H10 1.4E+05 1.3E−04 15 8.9E−10 0.224 human 18G9-H11 2.5E+05 1.1E−03 31 4.4E−09 0.975 6 cyno 18G9-H11 5.1E+04 1.2E−03 18 2.4E−08 0.617 human 4B4-D7 1.1E+05 2.9E−03 22 2.7E−08 0.147 1 cyno 4B4-D7 1.1E+05 2.7E−03 12 2.4E−08 0.188 human 1E6-B8 9.3E+04 3.9E−03 26 4.1E−08 0.157 1 cyno 1E6-B8 1.0E+05 4.2E−03 15 4.2E−08 0.122 human 13H2-C3 1.7E+05 4.9E−03 20 2.9E−08 0.317 1 cyno 13H2-C3 1.8E+05 5.9E−03 12 3.2E−08 0.166 human 20G6-F3 3.5E+04 2.7E−04 21 7.7E−09 0.208 1 cyno 20G6-F3 2.7E+04 2.2E−04 15 8.2E−09 0.184 human 11F9-F8 7.7E+04 4.0E−04 22 5.2E−09 0.137 1 cyno 11F9-F8 8.3E+04 3.6E−04 13 4.4E−09 0.109 human 20E5-F10 9.9E+04 4.3E−04 25 4.3E−09 0.125 1 cyno 20E5-F10 1.2E+05 3.4E−04 14 2.9E−09 0.104 human SP-34 1.1E+06 3.0E−03 34 2.7E−09 2.3E−01 5 cyno SP-34 2.2E+05 2.8E−03 34 1.3E−08 4.7E−01 SP-34 was used as IgG
1.5.2 Binding of Anti-CD3 Antibodies to huCD3ε, huCD3δ, and huCD3ε/δ Expressed on the Surface of HEK293F Cells by Flow Cytometry

[1320] To analyze binding of antibodies to human CD3ε and human CD3δ expressed on the surface of cells, HEK293F cells were transfected with both constructs either alone or in co-transfection and signals were measured by flow cytometry. For transfection procedure FuGENE HD Transfection Reagent (Promega, #E2311) was used according to manufacturer's protocol.

[1321] HEK293F cells were seeded in Freestyle293 medium (Gibco) at 6E6 cells per tube in 50 ml Cellstar Cellreactor tubes with filter (Greiner bio-one). Transfections were done according to the FuGENE protocol. Complex preparation was done in OptiMEM without phenol red (Gibco) at ratio 3:1 (Protocol for transfection of 293F cells grown in 8,000 μl of medium in T-25 flasks using a FuGENE®HD:DNA ratio of 3.0:1, http://www.promega.com/techserv/tools/FugeneHdTool/default.aspx).

[1322] Cells were incubated on a shaker at 37° C. and 5% CO2. At day one to three after transfection cells were harvested and binding of antibodies was analyzed by flow cytometry.

[1323] Antibodies for staining were seeded at 1 μg in 50 μl per well Stain Buffer with FBS (BD Pharmingen) in 96-well U-bottom suspension culture plates (Greiner bio-one). Harvested transfected cells were resuspended in Stain Buffer with FBS and were added at 50 μl per well to the antibodies. Cells were incubated at 4° C. in the dark for 30 min and were washed twice. 0.5 μg secondary antibody Goat F(ab′)2 Anti-Human IgG-FITC (Beckman Coulter, #732598) or Goat F(ab′)2 Anti-Human kappa-PE (Southern Biotech, #206209), respectively, in combination with 0.5 μg 7-AAD per well was added in 100 μl Stain Buffer with FBS. Cells were incubated at 4° C. in the dark for 15 min and were washed twice. For measurement, cells were resuspended in 200 μl Stain Buffer with FBS. Cells were measured using the MACSQuant (Miltenyi Biotec) or LSRII (BD) flow cytometer, respectively. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of 7-AAD negative single cells positive for antibody staining (data presented in table 5).

1.5.3 Binding of anti-CD3 Fabs to huCD3ε/δ and huCD3ε/γ by SPR

[1324] Binding was tested by SPR using a BIAcore3000 instrument run with HBS-EP buffer. Recombinant huCD3 proteins (ε/δ (PB01226), ε/γ (PB01225)) were captured at 10 μl/min via Fc-tag by anti-human Fc capture antibody MAB1302 (Millipore) immobilized on a CM5 sensor chip. Anti-CD3 Fabs were used as analytes at 100 nM with association and dissociation times of 240 sec and 300 sec, respectively at 30 μl/min. After each cycle the surfaces were regenerated by a 2 min pulse of 10 mM glycine buffer pH2.5.

[1325] When only huCD3δ was expressed on the surface of HEK293F cells no signal could be detected by flow cytometry. In contrast, nearly all antibodies could bind to cells transfected with huCD3ε exclusively or in co-transfection with huCD3δ indicating that huCD3ε is necessary as epitope. In Biacore assays binding to huCD3ε was shown irrespective of whether the δ or the γ chain was used for the recombinant protein suggesting that huCD3ε is sufficient as antigen. The antibody 12D2 exceptionally bound only to huCD3ε when a co-chain was present. There may be an indirect effect of co-expression of the g or d chain regarding the conformational structure of the protein to display the epitope for this antibody. The same effect was shown for the published antibody OKT3. This antibody is described to interact with a conformational epitope formed after association of huCD3ε with huCD3δ or γ, respectively (Salmeron et al., 1991, The Journal of Immunology). It was also shown that it binds to the huCD3ε subunit exclusively (Kjer-Nielsen et al., 2004, PNAS). Because of the similar behavior of 12D2 and OKT3 an interaction with huCD3ε is supposable for 12D2. Taken all together, huCD3ε seems to be the antigenic structure for all analyzed antibodies (data presented in table 5).

TABLE-US-00099 TABLE 5 Affinities of selected CD3 antibodies against huCD3δ, huCD3ε, huCD3ε/δ and huCD3ε/γ complexes measured by Flow cytometry and Biacore. Flow cytometry % antibody positive transfected cells (n = 3-6) Biacore huCD3ε huCD3ε/δ RU of Fab binding at 100 nM Clone huCD3δ (+/−SEM) (+/−SEM) huCD3ε/δ huCD3ε/γ 10E6 n.b. 96.3 +/− 2.2 93.3 +/− 4.7 11 8 11F9 n.b. 97.0 +/− 1.5 96.0 +/− 2.5 10 7 12D2 n.b. n.b. 96.3 +/− 2.2 11 5 12G3 n.b. 96.3 +/− 2.2 96.7 +/− 1.9 10 7 13H2 n.b. 93.3 +/− 3.3 93.0 +/− 4.2 9 6 18F5 n.b.  70.6 +/− 11.9 70.4 +/− 9.9 13 7 18G9 n.b. 95.0 +/− 2.5 97.0 +/− 1.5 10 3 1E6 n.b. 87.7 +/− 4.4 91.7 +/− 4.9 10 7 20E5 n.b. 96.7 +/− 1.9 95.3 +/− 3.2 11 8 20G6 n.b.  68.6 +/− 12.1  67.4 +/− 10.4 7 4 4B4 n.b.  71.5 +/− 10.8 69.5 +/− 8.7 11 3 4E7 n.b.  69.0 +/− 12.3  66.4 +/− 10.0 8 6 5B1 n.b. 97.0 +/− 1.5 97.0 +/− 1.5 10 7 OKT3 n.b. n.b. 70.2 +/− 9.6 4 3 n.b. = no binding (<20% of binding of huCD3e + huCD3d)

[1326] 1.6 Binding of CD3 Fab to Human T Cells

[1327] The binding capacity of the CD3-Fabs was determined by flow cytometry. Primary human T cells were used as target cells. Therefore, peripheral blood mononuclear cells (PBMCs) were isolated from 200 ml peripheral blood of healthy donors treated with EDTA by Ficoll density centrifugation. 15 ml Histopaque (Sigma-Aldrich) was preloaded on a 50 ml Leucosep-Tube (Greiner bio-one). Blood was diluted with autoMACS Rinsing Buffer+1% BSA (Miltenyi Biotec) and loaded on the membrane of a total of ten prepared tubes. Tubes were centrifuged without brake for 10 min at 1000 xg. PBMCs were collected and washed with autoMACS Rinsing Buffer+1% BSA three times. Finally, PBMCs were resuspended in autoMACS Running Buffer (Miltenyi Biotec) for isolation of T lymphocytes by autoMACSpro technology using the Pan T Cell isolation Kit (Miltenyi Biotec) according to manufacturer's instructions. Purity of separated T cells was analyzed by MACSQuant flow cytometry using the human 7-Color Immunophenotyping Kit (Miltenyi Biotec). Isolated T cells were resuspended in Stain Buffer with FBS (BD Pharmingen) and 1E5 cells in 100 μl per well were seeded in 96-well U-bottom suspension culture plates (Greiner bio-one). Fab antibodies were diluted 1:3 in serial in PBS (Invitrogen) and 5 μl each were added to the cells at a final maximum concentration of 30000 ng/ml. The assay was incubated for 45 min at 4° C. Cells were washed twice with Stain Buffer with FBS and 1 μg secondary antibody Goat F(ab′)2 Anti-Human kappa-FITC (Beckman Coulter, #732621) per well was added in 100 μl Stain Buffer with FBS per well. The assay was incubated for 20 min at 4° C. and washed twice afterwards. Cells were resuspended in 150 μl Stain Buffer with FBS per well and were measured using the MACSQuant (Miltenyi Biotec) or LSRII (BD) flow cytometer. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of cells positive for antibody binding. Cells treated only with the secondary but no primary antibody were used to set the gates. EC50 Curves were calculated by XLfit (Algorithm 205), EC50 values were calculated as inflection point of the slope (data are shown in table 6).

TABLE-US-00100 TABLE 6 Affinities of CD3 Fab to human T cells measured by Flow cytometry. Presented are mean EC50 values calculated from curves. Antibody Binding to human T cells EC50 [nM] mean +/− SEM 12D2-E5-Fab 3.8 +/− 0.2 4B4-D7-Fab 15.3 +/− 3.0 1E6-C9-Fab 18.7 +/− 3.5 10E6-G6-Fab 4.7 +/− 0.9 4E7-C9-Fab 6.2 +/− 0.8 12G3-E8-Fab 4.4 +/− 0.3 5B1-G2-Fab 11.1 +/− 3.6 18F5-H10-Fab 4.1 +/− 0.0 18G9-H11-Fab 1.7 +/− 0.5 13H2-C2-Fab 14.3 +/− 0.9 20G6-F3-Fab 14.0 (n = 1) 11F9-F8-huFab 11.6 (n = 1) 20E5-F10-huFab 8.2 (n = 1)

[1328] 1.7 Safety of CD3 Fab

[1329] 1.7.1 Safety of CD3 Fab measured by CD25+ and CD69+ expression on human T cells The effect of CD3 Fab antibodies on activation status of T cells as safety read out was analyzed by flow cytometry based detection of the expression of activation marker CD25 and CD69 on the surface of primary human T cells.

[1330] Isolated primary human T lymphocytes were resuspended in RPMI+ GlutaMAX I (Gibco)+10% FCS (Invitrogen) and 2.5E5 cells were seeded in 96-well U-bottom suspension culture plates (Greiner bio-one) in 100 μl per well.

[1331] 5 μl Fab CD3 antibodies were added to the cells at a final concentration of 30 000 ng/ml. The assay was incubated for 20 h at 37° C. in 5% C02.

[1332] After incubation time cells were spun down and stained for 15 min at 4° C. in 100 μl Stain Buffer with FBS (BD Pharmingen) per well with following labeled antibodies: CD25-V450, CD69-APC

[1333] Cells were washed twice after staining, resuspended in 150 μl Stain Buffer with FBS, and 5000 cells were measured using the LSRII (BD) flow cytometer. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of CD25pos and CD69pos T cells (table 7).

TABLE-US-00101 TABLE 7 Safety of CD3 Fab measured by CD25+ and CD69+ expression on human T cells CD25+ % Activation of CD69+ % Activation of human T cells human T cells normalized to PBS normalized to PBS C = 30000 ng/ml C = 30000 ng/ml Antibody n = 1 n = 1 11H3-E5-Fab 0.7 0.5 12D2-E5-Fab - SEC 3.2 11.1 4B4-D7-Fab 2.5 5.4 1E6-B8/B9-C9-Fab 1.8 3.2 10E6-G6-Fab 0.8 2.6 4E7-C9-Fab 0.4 0.1 12G3-E8-Fab 0.0 1.5 5B1-G2-Fab 0.4 4.2 6F4-D10-Fab 2.1 6.4 18F5-H10-Fab 0.0 0.1 18G9-H11-Fab 0.7 2.2 13H2-C2-Fab 0.5 0.4 13C1-F6-Fab 0.4 0.2 1E6-B8/B9-C9-Fab 0.6 1.7 20G6-F3-Fab 0.2 3.2 1E6-B8/B9-C9-Fab 2.5 7.8 1E6-B8/B9-C9-Fab 0.1 1.0 11F9-F8-huFab 1.0 0.8 20E5-F10-huFab 1.5 1.3

1.7.2 Safety of CD3 Fab Measured by CD4+/CD69+, CD4+/CD25+, CD8+/CD69+ and CD8+/CD25+ Expression on Human T Cells

[1334] The effect of CD3 Fab antibodies on activation status of T cells as safety read out was analyzed by flow cytometry based detection of the expression of activation marker CD25 and CD69 on the surface of primary human T cells. Isolated primary human T lymphocytes were resuspended in RPMI+ GlutaMAX I (Gibco)+10% FCS (Invitrogen) and 2.5E5 cells were seeded in 96-well U-bottom suspension culture plates (Greiner bio-one) in 50 μl per well. Either T cells exclusively were tested and wells were filled-up with 50 μl RPMI+ GlutaMAX I+10% FCS, or target cells (i.e. THP-1 cell line) were added at 2.5E4 cells per well in 50 μl RPMI+ GlutaMAX I+10% FCS. Bispecific antibodies were diluted 1:3 in serial in PBS (Invitrogen) and 5 μl each were added to the cells at a final maximum concentration of 30 000 ng/ml. The assay was incubated for 20 h at 37° C. in 5% CO2. After incubation time cells were spun down and stained for 15 min at 4° C. in 100 μl Stain Buffer with FBS (BD Pharmingen) per well with following labeled antibodies: CD4-PE, CD8-APC-Cy7, CD25-APC, CD69-PE-Cy7. As Fluorescence Minus One (FMO) control activated T cells were stained as described above but CD25 was replaced by its isotype (Isotype APC-IG1k) in one tube and CD69 was replaced by its isotype (Isotype PE-Cy7-IG1k) in a second tube. Cells were washed twice after staining, resuspended in 150 μl Stain Buffer with FBS, and 5000 cells were measured using the LSRII (BD) flow cytometer. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of CD4posCD25pos, CD4posCD69pos, CD8posCD25pos, and CD8posCD69pos T cells. Gates were set according to FMO controls (see table 8).

TABLE-US-00102 TABLE 8 Safety of CD3 Fab measured by CD4+/CD69+, CD4+/CD25+, CD8+/CD69+ and CD8+/CD25+ expression on human T cells CD4+/CD69+ CD8+/CD69+ CD4+/CD25+ CD8+/CD25+ % Activation % Activation % Activation % Activation normalized to normalized to normalized to normalized to PBS PBS PBS PBS C = 100 nM C = 100 nM C = 100 nM C = 100 nM Antibody mean +/− SEM mean +/− SEM mean +/− SEM mean +/− SEM 20G6-F3-Fab 0.1 +/− 0.1 0.0 +/− 0.0 0.4 +/− 0.1 0.0 +/− 0.1 4B4-D7-Fab 3.2 +/− 1.1 0.3 +/− 0.3 1.3 +/− 0.6 0.0 +/− 0.0

Example 2: CD123 Sequences

[1335] 2.1 Construction of CD123 (IL3RA)-hFc Fusion Expression Plasmids (CD123-Fc)

[1336] Using cDNA containing plasmids as a template, human and Macaca fascicularis CD123 fusion proteins were generated in reading frame with heavy chain constant region including a GS-linker (used in Macaca protein), the hinge region, CH2 and CH3 domains of human immunoglobulin IgG additionally carrying a Strep-II Tag (only in human protein version).

[1337] Using human genomic DNA as template, human CD123 (IL3RA) extracellular domain was amplified, including the signal sequence. The resulting amplified cleaved and purified PCR products were combined by ligation PCR and ligated into mammalian expression vector pXL by InFusion method using NheI and HindIII site. The sequence of the resulting mature human CD123 Strep-II tagged Fc-fusion protein is disclosed under SEQ ID NO: 196. Amino acids 1 to 284 correspond to the amino acids 22 to 305 of the full-length wild-type human CD123 protein (herein disclosed under SEQ ID NO: 194, available from the NCBI database under the accession number NP_002174.1) and thus the extracellular domain of human CD123.

[1338] To clone Macaca fascicularis CD123cDNA was made from blood of a Macaca fascicularis population. Using this isolated cDNA as template, Macaca CD123 (IL3ra) extracellular domain was amplified, including the signal sequence. The resulting amplified cleaved and purified PCR products were combined by ligation PCR and ligated into mammalian expression vector pXL by InFusion method using NheI and HindIII. The sequence of the resulting mature human CD123 Strep-II tagged Fc-fusion protein is disclosed under SEQ ID NO: 197. Amino acids 1 to 284 correspond to the amino acids 22 to 305 of the full-length wild-type Macaca fascicularis CD123 protein (herein disclosed under SEQ ID NO: 195, available from the NCBI database under the accession number NP_002174.1) and thus the extracellular domain of human CD123.

[1339] 2.2 Expression and Purification of Human and Macaca fascicularis CD123-Fc

[1340] Freestyle HEK293 cells growing in F17 serum free suspension culture (Life) were transiently transfected with the expression plasmid. Transfection was performed using Cellfectin transfection reagent (Life) The cells were cultured at 37° C. for 7 days. The culture supernatant containing recombinant protein was harvested by centrifugation and was clarified by filtration (0.22 μm).

[1341] For purification the Fc-fusion protein variants were captured on protein A matrix (GE) and eluted by pH shift. After polishing the protein by SEC in PBS using a Superdex 200 (GE) and a final ultrafiltration concentration step, the protein was used for further assays.

[1342] 2.3 Assessment of Affinities to Both Human and Macaca Fascicularis CD123Rat IgGs from Hybridoma

[1343] Screening of anti-CD123 rat IgGs for binding affinities to human CD123 and cross-reactivities to cyno CD123 was performed with hybridoma supernatants using a Proteon XPR36 (Biorad) in a one-shot kinetics approach. A capture assay was established using a goat anti-rat IgG (112-005-071, Jackson Immuno Research). The capture antibody was coated on GLC chips (176-5011, Biorad) to approx. 8000 RU in the vertical direction using the amine coupling kit (176-2410, Biorad). Capture of the rat IgGs to approx. 200 RU in vertical direction resulted in Rmax values of up to 100 RU for CD123-Fc. Binding kinetics with human and cyno CD123-Fc fusion protein was measured at 100 μl/min in the horizontal direction with 120 sec. and 600 sec. for association and dissociation, respectively. The CD123-Fc proteins were used in twofold dilutions from 6 nM to 100 nM. PBSET buffer (176-2730, Biorad) was used as assay buffer. Regeneration was achieved by injection of 10 mM Glycine buffer pH 1.5 for 18 sec at 30 μl/min. Data processing and analysis was performed using ProteonManager software v3.0. Fitting of the sensorgrams was done with a 1:1 Langmuir model. Clones were selected based on affinities for human CD123 with KD<1 nM and crossreactivity to cyno CD123.

Fabs and CODV-Fabs

[1344] Binding affinities of anti-CD123 binding Fabs or CODV-Fabs were measured using a Biacore3000 instrument (GE Healthcare). Assay buffer was HBS-EP (BR-1001-88, GE Healthcare). Capture of CD123-Fc fusion proteins was achieved using the human antibody capture kit (GE Healthcare). The capture antibody was coupled to CM5 chips (BR-1001-88, GE Healthcare) to approx. 12.000 RU using the amine coupling kit (BR-100-50, GE Healthcare). The CD123-Fc fusions proteins were captured at 10 μl/min to approx. 70 RU to yield Rmax values of 30 RU. Binding kinetics with the anti-CD123 Fabs or CODV-Fabs was measured at 30 μl/min for 240 sec. and 600 sec. for association and dissociation phase, respectively. Twofold dilutions of Fabs from 3 to 200 nM in assay buffer were used. All Fab concentrations were run in duplicate together with duplicate buffer blanks for double referencing. Regeneration of the capture surface was performed with a 1 min injection of 3M MgCl2 solution at 30 μl/min. For data analysis the BIAevaluation software v.4.1 (GE Healthcare) was used. Data were fit globally using a 1:1 Langmuir model with mass transfer.

IgGs and CODV-Fc Proteins

[1345] Measurement of binding affinities of anti-CD123 IgGs and CODV-Fc proteins was performed analogous to the binding assay for Fabs and CODV-Fabs with the exception of the capture antibody. In this case the Strep-MAB classic antibody (2-1507-001, IBA) was used to capture human CD123-Fc via its StrepII-tag. Here the regeneration solution was 10 mM Glycine buffer pH2.0.

[1346] 2.4 Generation of Human and Macaca Fasciculards Cross-Reactive Anti-CD123 Antibodies

[1347] Human and Macaca fascicularis CD123 cDNAs were cloned into Aldevron proprietary immunization vectors (pB8 and VV8) respectively. Three rats of immunization group MR13-296 were immunized with the immunization vector IL3RA-hum.-ECD (aa19-305). The immune serum was taken at day 24 of the immunization protocol, after 4 genetic applications (IS24d-4). Sera, diluted in PBS 3% FBS, were tested by flow cytometry using mammalian cells transiently transfected with the human and cyno IL3RA cDNA variants IL3RA-hum.ECD and IL3RA-hum.D3.

[1348] Specific reactivity of the immune sera against cells transfected with pB1-IL3RA-hum.ECD, as well as with IL3RA-cyno (pFF1262) and the THP-1 cells could be detected in all immunised animals when compared to cells transfected with an irrelevant cDNA.

[1349] Rats with positive sera were scarified and B cells were fused with mouse myeloma cells. The resulting hybridomas were screened on HEK293 cells transfected with human or cynomolgus CD123 expression plasmids, on different cell lines expressing CD123 by flow cytometry (data shown in table 10).

[1350] Target cells were seeded at 5E4 cells in 50 μl Stain Buffer with FBS (BD Pharmingen) per well in 96-well U-bottom suspension culture plates (Greiner bio-one). Hybridoma supernatants were diluted 1:3 in serial in PBS (Invitrogen) and 50 μl each were added to the cells at a final maximum concentration of 1 μg/ml. The assay was incubated for 45 min at 4° C.

[1351] Cells were washed twice with Stain Buffer with FBS and 1 μg secondary antibody Goat Anti-Rat IgG (H+L)-Alexa Fluor 488 (Invitrogen-Life Technologies, # MH10520) was added in 100 μl Stain Buffer with FBS per well. The assay was incubated for 15 min at 4° C. and washed twice afterwards.

[1352] Cells were resuspended in 200 μl Stain Buffer with FBS per well and were measured using the MACSQuant (Miltenyi Biotec) or LSRII (BD) flow cytometer. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of cells positive for antibody binding. Cells treated only with the secondary but no primary antibody were used to set the gates. Curves were calculated by XLfit (Algorithm 205).

[1353] Specific binding of clones to CD123 could be shown on the surface of transfected HEK293 in comparison to untransfected HEK293 cells where no signal could be detected (data not shown). Binding of antibodies was concentration dependent with an EC50 value ranging between 0.4 and 17.7 ng/ml (table 9).

TABLE-US-00103 TABLE 9 Specific CD123 binding of rat IgG clones in hybridoma supernatants detected by flow cytometry. Concentration dependent binding of antibodies was measured using CD123 transfected HEK293 as target cells and a goat anti-rat IgG (H + L)-Alexa Fluor 488 secondary antibody. Presented are calculated EC50 values of the curves. Antibody EC50 [ng/ml] BFX - 1A6 6.4 BFX - 1E1 2.2 BFX - 2B8 5.5 BFX - 2F4 4.5 BFX - 2F8 3.4 BFX - 2H7 17.7 BFX - 3B10 2.0 BFX - 3E3 2.0 BFX - 5A5 10.1 BFX - 6B10 5.0 BFX - 6C10 6.1 BFX - 6D6 0.9 BFX - 8B11 0.4 BFX - 9B8 1.2 BFX - 9D7 0.9 BFX - 9F6 1.3 BFX - 9H2 14.9

TABLE-US-00104 TABLE 10 Binding data of CD123 antibodies to recombinant CD123 and CD123 expressing cells. Binding data showing affinities of CD123 antibodies against recombinant CD123 protein from human and cynomolgous monkey. Binding on cell surface was detected by flow cytometry. SP2 refers to cell expressing N-terminal truncated (D1 region) variant of CD123. Antibodies were tested by Proteon XPR36 on their ability to compete with the IL3 binding to CD123. ka kd KD (nM) KD (nM) Molm- THP- OCI- IL3 Clone (1/Ms) (1/s) Human Cyno 13 1 AML3 SP2 Basophils Blocking 1E1 4.6E+05 1.1E−04 0.23 0.72 + + + + + − 2B8 6.6E+05 1.2E−05 0.018 0.77 + + + + + − 2F8 1.3E+05 6.6E−05 0.53 0.51 + + + − + + 3B10 2.2E+05 0.5E−05 0.21 0.27 + + + − + + 3E3 1.6E+05 6.4E−05 0.4 0.12 + + + − + + 5A5 1.3E+05 1.2E−04 1.1 0.05 + + + + + − 6B10 2.3E+05 1.5E−05 0.15 0.1 + + + + + − 6C10 1.8E+05 1.4E−04 0.75 0.79 + + + + + − 8B11 4.7E+05 6.2E−06 0.01 0.1 + + + + + − 9B8 1.5E+06 2.8E−04 0.19 0.17 + + + + + + 9D7 1.0E+06 3.2E−04 0.3 0.07 + + + + + + 9F6 1.8E+05 6.7E−06 0.04 0.5 + + + + + −

[1354] 2.5 Humanization of Anti-CD123 Rat Antibody Sequences

[1355] Humanization of rat antibodies was performed by CDR-grafting or by the 4D method (US20110027266). For the rat-anti-CD3 antibody 3E3 the closest Rattus germline sequence identified was IGHV2S48*01 and IGHJ3*01 for the heavy chain variable region and IGLV3S2*01 and IGKJ3*01 for the light chain variable region. The calculated rat germinality index (frameworks sequences only) is 94.51% for the VH and 98.9% for the VL.

[1356] Potential exposed problematic residues were checked and one residue in CDRH2 was modified.

[1357] Using the grafting method, a variety of humanized variants were generated based on the closest human germline sequences identified: IGHV4-59*05 and IGHJ4*01 for the VH (germinality index on frameworks: 75.82%); IGLV6-57*01 and IGLJ3*01 for VL (germinality index on frameworks: 72.22%).

[1358] In addition to CDR grafting, the 4D humanization protocol (US20110027266) was used to humanize the Rat anti-CD123 3E3 variable light (VL) and heavy (VH) domains. A molecular dynamics (MD) simulation was performed on the minimized 3D homology model (done with MOE; PDB used: 1FLR) of Rat anti-CD123 3E3 and compared to the 49 Human models derived from the seven representative light chains (vk1, vk2, vk3, vk4, vlambda1, vlambda2, vlambda3) and the seven representative heavy chains (vh1a, vh1b, vh2, vh3, vh4, vh5, vh6) designed by LGCR/SDI and available within MOE.

[1359] Two Models have been selected for the “4D humanization”: V3-vh4 with and VL3-VH2 with the best both hydrophobic and electrostatic components and sequence identity outside CDRs. For the pairwise association between the Rat anti-CD123 3E3 variable domain and the two selected models, the sequences were aligned based on the optimal 3D superposition of the alpha carbons of the corresponding homology models.

Example 3: Antibodies in the Bispecific CODV-Fab Format

[1360] 3.1 Cloning of Selected CD3 Sequences in Combination with Anti-CD123 mAb 7G3 in the Bispecific CODV-Fab Format to Study their T-Cell Engagement Activity

[1361] Selected CD3 antibody sequences, such as I2C, mAb2 (Macrogenics) and the so-called “20G6-F3”, “4E7-C9”, “4B4-D7” and “18F5-H10” were expressed as monospecific anti-CD3 Fabs, Said selected sequences were as well expressed as bispecific CD3×CD123 CODV-Fabs using sequences of the monoclonal antibody 7G3, resulting in the CODV-Fab constructs “I2Cx7G3” and the so-called “7G3x20G6”, “7G3x4E7”, 7G3x41B4‘ and’7G3x18F5′, as further described in the section “antibody-like bin ding proteins” herein above. Purified proteins were used in a Biacore assay to compare the affinity against CD3ε/δ complexes (data presented in table 11). No changes in affinities could be detected by Biacore analysis when CD3 sequences were introduced into the bispecific CODV-Fab format.

TABLE-US-00105 TABLE 11 Comparison of affinities between Fab-fragments and CODV-Fabs against CD3ε/δ complexes from human and cynomolgous monkey hu kon hu koff hu KD cy KD Molecule 1/Ms 1/s nM nM Fab- I2C 1.1E+07 1.3E−03 0.13 0.5 fragments mAb2 3.1E+05 4.3E−03 13.7 16.3 BDW-20G6-F3 3.5E+04 2.7E−04 8 8 BDX-4E7-C9 8.9E+04 1.5E−03 17 15 BDW-4B4-D7 1.1E+05 2.9E−03 27 24 BDX-18F5- 2.0E+05 1.8E−04 1 1 H10 CODV-Fabs 7G3 × I2C 4.0E+06 2.0E−03 0.5 0.4 7G3 × 20G6 7.2E+04 5.1E−04 7 7 7G3 × 4E7 1.2E+05 1.9E−03 16 10 7G3 × 4B4 2.7E+05 4.4E−03 17 18

[1362] 3.2 Bispecific CODV-Fabs Directed Against CD123 and CD33 Mediate Potent Redirected T-Cell Killing

[1363] Such bispecific CODV-Fabs have the ability to localize a T-cell (by binding such T-cell to the CD3-binding portion of a CD3-binding CODV-Fab) to the location of a tumor cell (by binding such cancer cell to the CD123 portion of the CODV-Fab). The localized T-cell can then mediate the killing of the tumor cell in a process termed herein “redirected” killing. Bispecific CODV-Fab specific for CD123 and CD3 were constructed having the anti-CD123 variable domains of monoclonal antibody 7G3 and anti-CD3 variable domains of selected CD3 antibodies generated in example 1.

[1364] Therefore, peripheral blood mononuclear cells (PBMCs) were isolated from 200 ml peripheral blood of healthy donors treated with EDTA by Ficoll density centrifugation. 15 ml Histopaque (Sigma-Aldrich) was preloaded on a 50 ml Leucosep-Tube (Greiner bio-one). Blood was diluted with autoMACS Rinsing Buffer+1% BSA (Miltenyi Biotec) and loaded on the membrane of a total of ten prepared tubes. Tubes were centrifuged without brake for 10 min at 1000 xg. PBMCs were collected and washed with autoMACS Rinsing Buffer+1% BSA three times. Finally, PBMCs were resuspended in autoMACS Running Buffer (Miltenyi Biotec) for isolation of T lymphocytes by autoMACSpro technology using the Pan T Cell isolation Kit (Miltenyi Biotec) according to manufacturer's instructions. Purity of separated T cells was analyzed by MACSQuant flow cytometry using the human 7-Color Immunophenotyping Kit (Miltenyi Biotec).

[1365] T-cell engaging effect of bispecific antibodies was analyzed by a flow cytometry based cytotoxic assay. Target cells (i.e. THP-1 cell line) were stained for 15 min at 37° C. with 1 pM CFSE in 1 ml RPMI+ GlutaMAX I (Gibco) per 1E7 cells. Afterwards, cells were washed twice and resuspended in RPMI+ GlutaMAX I+10% FCS (Invitrogen). 2.5E4 target cells were seeded in 96-well U-bottom suspension culture plates (Greiner bio-one) in 50 μl medium per well.

[1366] Isolated primary human T lymphocytes were resuspended in RPMI+ GlutaMAX I+10% FCS and were added at indicated effector-to-target ratio in 50 μl per well to the target cells (in general E:T=10:1).

[1367] Bispecific antibodies were diluted 1:3 in serial in PBS (Invitrogen) and 5 μl each were added to the cells at a final maximum concentration of 3 000 ng/ml. The assay was incubated for 20 h at 37° C. in 5% CO2.

[1368] To detect dead target cells, all cells were stained with 7-AAD. Therefore, 5 μg/ml 7-AAD diluted in Stain Buffer with FBS (BD Pharmingen) were added to each well and were incubated for 15 min at 4° C. in the dark. Cells were measured using the MACSQuant (Miltenyi Biotec) or LSRII (BD) flow cytometer, respectively. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of CFSE and 7-AAD double positive cells.

[1369] The results of these investigations shown for example in tables 20 to 22 demonstrate the ability of the CD123×CD3 CODV-Fabs to mediate redirected killing of tumor cells.

[1370] 3.3 Safety Assessment of Redirected T-Cell Killing

[1371] The effect of bispecific antibodies on activation status of T cells as safety read out was analyzed by flow cytometry based detection of the expression of activation marker CD25 and CD69 on the surface of primary human T cells.

[1372] Isolated primary human T lymphocytes were resuspended in RPMI+ GlutaMAX I (Gibco)+10% FCS (Invitrogen) and 2.5E5 cells were seeded in 96-well U-bottom suspension culture plates (Greiner bio-one) in 50 μl per well.

[1373] Either T cells exclusively were tested and wells were filled-up with 50 μl RPMI+ GlutaMAX I+10% FCS, or target cells (i.e. THP-1 cell line) were added at 2.5E4 cells per well in 50 μl RPMI+ GlutaMAX I+10% FCS.

[1374] Bispecific antibodies were diluted 1:3 in serial in PBS (Invitrogen) and 5 μl each were added to the cells at a final maximum concentration of 30 000 ng/ml. The assay was incubated for 20 h at 37° C. in 5% CO2.

[1375] After incubation time cells were spun down and stained for 15 min at 4° C. in 100 μl Stain Buffer with FBS (BD Pharmingen) per well with following labeled antibodies: CD4-PE, CD8-APC-Cy7, CD25-APC, CD69-PE-Cy7 As Fluorescence Minus One (FMO) control activated T cells were stained as described above but CD25 was replaced by its isotype (Isotype APC-IG1k) in one tube and CD69 was replaced by its isotype (Isotype PE-Cy7-IG1k) in a second tube.

[1376] Cells were washed twice after staining, resuspended in 150 μl Stain Buffer with FBS, and 5000 cells were measured using the LSRII (BD) flow cytometer. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of CD4posCD25pos, CD4posCD69pos, CD8posCD25pos, and CD8posCD69pos T cells. Gates were set according to FMO controls (see table 12).

TABLE-US-00106 TABLE 12 Activity of CD123 × CD3 CODV-Fabs measured in a cytotoxic assay and safety measured by the expression of CD69 Activity EC50 Safety CD69 expression Molecule [ng/ml] % cells/% max 7G3 × I2C 20 ± 9 25.8 ± 7.6 7G3 × OKT3 16 ± 5 25.5 ± 8.0 7G3 × 20G6 18 ± 5  8.7 ± 0.8 7G3 × 4E7 12 ± 3 12.3 ± 2.5 7G3 × 4B4 11 ± 2 11.6 ± 0.8 7G3 × 18F5  61 ± 15  9.8 ± 1.1

[1377] 3.4 Humanization of Anti-CD3 Rat Antibody Sequences and Bispecific Antibodies

[1378] Humanization of rat antibodies was performed by CDR-grafting or by the 4D method (US20110027266).

[1379] For the rat-anti-CD3 antibody “20G6” the closest Rattus germline sequences was identified as IGHV6S17*01 and IGHJ2*01 (for the heavy chain variable region and IGKV1S21*01 and IGKJ4*01 for the light chain variable region). The calculated rat germinality index (frameworks sequences only) is 97.80% for the VH and 95.5% for the VL.

[1380] A variety of humanized variants using grafting method were generated based [1381] 1) on the closest human germline sequences identified IGHV3-30-01_IGHJ4-01 for the VH with a germinality index on framework of 77%; IGK2D-29-02_IGKJ4-01 for the VL with a germinality index on the framework of 80%), or [1382] 2) or based on the a closest germline sequence identified having a lower PI IGVH3-48*02-IGHJ4-01 for the VH with a germinality index on the framework of 75%; IGKV2-28*01-IGKJ4-01 for the VL with a geminality index on the framework of 77.5%, or [1383] 3) consisting on a more distant human germiline sequence (change germline clade) (IGVH1-46*01-IGHJ4*01 for the VH with a germinality index on the framework of 56%; IGKV4*01-IGKJ4*01 for the VL with a germinality index on the framework of 67.5%).
The Humanized sequences were then introduced into the CODV-Fab format of the so-called so-called CODV-Fab 7G3×20G6 in combination with the an-CD123 sequence from antibody 7G3 as described before. Purified CD123×CD3 CODV-Fabs were used in a Biacore assay to assess the affinity to CD3ε/δ (see table 13).

TABLE-US-00107 TABLE 13 Affinty of selected humanized variants of the anti-CD3 antibody 20G6 to recombinant CD3ε/δ complex. Humanized sequence combinations used in the ka kd Rmax KD CODV-Fab “7G3 × 20G6” (1/Ms) (1/s) (RU) (M) parental sequence 2.17E+04 5.71E−04 46 2.63E−08 VL1A_VH1A 2.86E+04 4.34E−03 46 1.52E−07 VL1B_VH1A 2.99E+04 4.14E−03 45 1.38E−07 VL1C_VH1D 2.30E+04 1.01E−03 58 4.40E−08 VL1D_VH1B 1.86E+04 1.09E−03 62 5.87E−08 VL1D_VH1C 2.22E+04 7.13E−04 71 3.21E−08 VL1D_VH1D 2.40E+04 7.46E−04 59 3.10E−08

[1384] For the rat-anti-CD3 antibody 41B4-D7 the closest Rattus germline sequence was identified as IGHV6S17′01 (identity of 93%) and IGHJ2′01 (identity of 87.5%) for the heavy chain variable region and IGKV1S21′01 (identity of 93%) and IGKJ4′01 (identity of 100%) for the light chain variable region.

[1385] The calculated percentage of identity of the identified Rattus V-sequences to Human Germinality (frameworks sequences only) is 79% for the VH and 77.53% for the VL.

[1386] A variety of humanized variant pairs for VH and VL were generated by grafting with additional sequence engineering, using the closest human germline sequences (IGHV3-30′01_IGHJ6′02; IGKV2-30′02/IGKV2D-39′02_IGKJ2′01). The calculated percentages of Human Germinality (4 IMGT frameworks sequences only) for the humanized V-sequences are listed in table 14.

TABLE-US-00108 TABLE 14 Percentages of Human Germinality for humanized V- sequences obtained by grafting of the so-called “B4-D7” antibody. The percentages were calculated based on the 4 IMGT frameworks sequences only. Grafted “4B4-D7” variants % Human germinality index VL1Cmodif1 92.1 VL1Cmodif2 91.0 VL1Cmodif3 91.0 VL1Amodif1 97.8 VL1Amodif2 96.6 VL1Amodif3 95.5 VL2C 95.5 VL2D 96.7 VL2F 96.7 VH6Bmodif1 92.3 VH6Bmodif2 90.1 VH6Amodif1 94.5 VH6Amodif2 92.3 VH6Amodif3 91.2 VH6C 95.5 VH6D 93.5

[1387] In addition to CDR grafting, the 40 humanization protocol as described in the US patent application US20110027266 was used to humanize the Rat anti-CD3 41B4D7 variable light (V/L) and heavy (V/H) domains. A molecular dynamics (MD) simulation was performed on the minimized 3D homology model (done with MOE; PDB used: 1FLR) of Rat anti-CD3 41B4-D7 and compared to the 49 Human models derived from the seven representative light chains (vk1, vk2, vk3, vk4, vlambda1, vlambda2, vlambda3) and the seven representative heavy chains (vh1a, vh1b, vh2, vh3, vh4, vh5, vh6) designed by LGCR/SDI and available within MOE

[1388] Two Models were selected for the “40 humanization”. vk1-vh6 with the highest 40 similarity, whit both hydrophobic and electrostatic components. vk2-vh3 with the highest sequence identity outside CDR. For the pairwise association between the Rat anti-CD3 4B4-D7 variable domain and the two selected models, the sequences were aligned based on the optimal 3D superposition of the alpha carbons of the corresponding homology models. A variety of other humanized variant pairs for VH and VL were further optimized.

[1389] The calculated percentages of Human Germinality (4 IMGT frameworks sequences only) for the humanized V-sequences are listed in Table 15:

TABLE-US-00109 TABLE 15 Percentages of Human Germinality for humanized V-sequences obtained by 4D humanization of the so-called “B4- D7” antibody. The percentages were calculated based on the 4 IMGT frameworks sequences only. 4D “4B4-D7” variants % Human germinality index VL1A 75 VL1B 75 VL2A 83 VL2B 83 VH1A 80 VH1B 80 VH2A 80 VH2B 84.5

[1390] Humanized sequences were expressed as Fab-fragments and purified followed by a Biacore assay to assess the affinity to CD3ε/δ (data shown in table 16).

TABLE-US-00110 TABLE 16 Affinty of selected humanized variants of the anti-CD3 antibody 4B4-D7 to recombinant CD3ε/δ complex. Fab VL VH Ligand ka kd KD 4B4 huCD3 3.26E+05 2.85E−03 8.79E−09 cyCD3 1.90E+05 2.37E−03 1.26E−08 4B4_1 VL1a VH1B huCD3 6.28E+05 1.74E−03 2.77E−09 75% 80% cyCD3 5.06E+05 1.58E−03 3.13E−09 4B4_2 VL1b VH1B huCD3 2.86E+05 2.14E−03 7.48E−09 75% 80% cyCD3 2.77E+05 2.10E−03 7.57E−09 4B4_5 VL2C VH6D huCD3 2.33E+05 4.23E−03 1.82E−08 95.5% 93.5% cyCD3 2.53E+05 4.20E−03 1.66E−08 4B4_6 VL1a VH1A huCD3 1.50E+06 1.85E−03 1.24E−09 75% 80% cyCD3 1.49E+06 1.77E−03 1.19E−09 4B4_7 VL1b VH1A huCD3 3.06E+05 2.48E−03 8.10E−09 75% 80% cyCD3 3.01E+05 2.19E−03 7.28E−09 4B4_9 VL2B VH2A huCD3 2.18E+05 2.91E−03 1.34E−08 83% 80% cyCD3 2.85E+05 3.44E−03 1.21E−08 4B4_10 VL1_CM2 VH6_BM2 huCD3 4.34E+05 3.99E−03 9.19E−09 91% 90% cyCD3 4.30E+05 3.84E−03 8.94E−09 4B4_11 VL1_CM3 VH6_BM2 huCD3 5.66E+05 7.74E−03 1.37E−08 91% 90% cyCD3 2.38E+05 4.86E−03 2.04E−08 4B4_17 VL2A VH2B huCD3 3.10E+05 2.76E−03 8.91E−09 83% 84.5% cyCD3 2.98E+05 2.70E−03 9.07E−09 4B4_20 VL2F VH6C huCD3 5.09E+04 1.71E−03 3.35E−08 96.5% 95.5% cyCD3 4.56E+04 1.98E−03 4.34E−08

[1391] 3.5 Binding CODV Hu-Fab CD123×CD33 to THP-1 and TF-1 Cells

[1392] Sequences of selected CD123 antibodies were cloned into the CODV-Fab format in combination with a CD3 binding sequence and proteins were expressed and purified.

[1393] Their binding capacity to cells naturally expressing CD123 was determined by flow cytometry. THP-1 cell line or TF-1 cell line were used as target cells.

[1394] Target cells were blocked with FcR-Blocker (Sigma). Therefore, target cells were resuspended in Stain Buffer with FBS (BD Pharmingen) and were blocked with 100 μl blocking reagent per ml for 1 h at 4° C. Cells were filled-up with Stain Buffer with FBS and 1E5 cells in 50 μl per well were seeded in 96-well U-bottom suspension culture plates (Greiner bio-one).

Antibodies were added at 3 μg in 50 μl Stain Buffer with FBS per well. The assay was incubated for 30 min at 4° C.

[1395] Cells were washed twice with Stain Buffer with FBS and 1 μg secondary antibody Goat F(ab′)2 Anti-Human kappa-FITC (Beckman Coulter, #732621) per well was added in 100 μl Stain Buffer with FBS per well. The assay was incubated for 20 min at 4° C. and washed twice afterwards.

[1396] Cells were resuspended in 150 μl Stain Buffer with FBS per well and were measured using the MACSQuant (Miltenyi Biotec) or LSRII (BD) flow cytometer. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of cells positive for antibody binding. Cells treated only with the secondary but no primary antibody were used to set the gates.

[1397] Binding of CD123×CD3 CODV-Fabs to CD123 was shown with two different cell lines expressing CD123 either with co-expression of CD131 on TF-1 cell line or on the surface of THP-1 cells lacking CD131 expression. Exemplarily shown are five different clones binding to target cells. As negative control (specificity control) a CD19×CD3 CODV-Fab was used as and a reference CD123×CD3 CODV-Fab as positive control (table 17).

TABLE-US-00111 TABLE 17 Specific binding of CD123 targeting sequences cloned into CODV-Fab backbone in combination with a CD3 binding sequence to THP-1 and TF-1 cell lines. Binding of antibodies was detected using CD123 expressing THP-1 and TF-1 as target cells. 3 μg antibody was added and detected by a secondary goat anti-human kappa-FITC antibody. Shown are percentages of antibody positive target cells. % FITC positive Cells Antibody THP-1 cells TF-1 cells medium control 0.10 0.00 sec. AB control 0.00 0.20 specificity control CD19 × CD3 0.10 0.20 CODV-Fab CD123 × CD3 99.90 82.50 CODV-Fab BFX-2F8-D6 × CD3 24.00 9.30 CODV-Fab BFX-3E3-D3 × CD3 99.00 31.60 CODV-Fab BFX-9B8-G6 × CD3 48.80 31.10 CODV-Fab BFX-9D7-C8 × CD3 88.70 50.20 CODV-Fab BFX-9F6-G3 × CD3 99.70 60.70

[1398] 3.6 Cytotoxic Effect to THP-1 Cells Mediated by CODV-Fab CD123×CD3

[1399] T-cell engaging effects of bispecific antibodies consisting of new generated CD123 sequence and the same CD3 binding sequence was analyzed by a flow cytometry based cytotoxic assay. Effector cells were primary T cells isolated from whole blood of healthy donors. THP-1 cells were used as CD123 expressing target cells.

[1400] Peripheral blood mononuclear cells (PBMCs) were isolated from 200 ml peripheral blood of healthy donors treated with EDTA by Ficoll density centrifugation. 15 ml Histopaque (Sigma-Aldrich) was preloaded on a 50 ml Leucosep-Tube (Greiner bio-one). Blood was diluted with autoMACS Rinsing Buffer+1% BSA (Miltenyi Biotec) and loaded on the membrane of a total of ten prepared tubes. Tubes were centrifuged without brake for 10 min at 1000 xg. PBMCs were collected and washed with autoMACS Rinsing Buffer +1% BSA three times. Finally, PBMCs were resuspended in autoMACS Running Buffer (Miltenyi Biotec) for isolation of T lymphocytes by autoMACSpro technology using the Pan T Cell isolation Kit (Miltenyi Biotec) according to manufacturer's instructions. Purity of separated T cells was analyzed by MACSQuant flow cytometry using the human 7-Color Immunophenotyping Kit (Miltenyi Biotec).

[1401] Target cells (i.e. THP-1 cell line) were stained for 15 min at 37° C. with 1 pM CFSE in 1 ml RPMI+ GlutaMAX I+10% FCS (Invitrogen). 2.5E4 target cells were seeded in 96-well U-bottom suspension culture plates (Greiner bio-one) in 50 μl medium per well.

[1402] Isolated primary human T lymphocytes were resuspended in RPMI+ GlutaMAX I+10% FCS and were added at indicated effector-to-target ratio in 50 μl per well to the target cells (in general E:T=10:1).

[1403] Bispecific antibodies were diluted 1:3 in serial in PBS (Invitrogen) and 5 μl each were added to the cells at a final maximum concentration of 3 000 ng/ml. The assay was incubated for 20 h at 37° C. in 5% CO2.

[1404] To detect dead target cells, all cells were stained with 7-AAD. Therefore, 5 μg/ml 7-AAD diluted in Stain Buffer with FBS (BD Pharmingen) were added to each well and were incubated for 15 min at 4° C. in the dark. Cells were measured using the MACSQuant (Miltenyi Biotec) or LSRII (BD) flow cytometer, respectively. Further data analyses were performed using the FlowJo software (Tree Star, Inc.). Read out was percentage of CFSE and 7-AAD double positive cells. Curves were calculated by XLfit (Algorithm 205).

[1405] As exemplarily shown in table 18 the bispecific antibodies were able to engage primary T cells and to lyse THP-1 target cells in vitro. An antibody concentration dependent increase in dead target cells could be detected after 20 h co-incubation. For the antibodies shown in here EC50 values were calculated ranging between 12.2 and 429.3 ng/ml.

TABLE-US-00112 TABLE 18 T-cell engaging effect of bispecific CODV-Fab CD123 × CD3 detected in flow cytometry based cytotoxic assays. Presented are mean EC50 values calculated from curves EC50 [ng/ml] EC50 [nM] geomean +/− SEM geomean +/− SEM Antibody (n = 4) (n = 4) n CODV-Fab BFX-2B8-F3 × CD3 106.9 +/− 33.9 1420.5 +/− 450.2 4 CODV-Fab BFX-3E3-D3 × CD3 12.2 +/− 5.9 163.0 +/− 80.1 4 CODV-Fab BFX-6B10-E4 × CD3  39.7 +/− 17.9  529.1 +/− 237.7 4 CODV-Fab BFX-9B8-G6 × CD3 106.1 +/− 36.1 1410.2 +/− 479.8 4 CODV-Fab BFX-9D7-C8 × CD3  86.3 +/− 29.3 1148.6 +/− 390.5 4 CODV-Fab BFX-9F6-G3 × CD3 13.9 +/− 7.7  185.9 +/− 104.1 3 CODV-Fab BFX-3B10-E6 × CD3 429.3 +/− 82.6  5709.3 +/− 1099.0 2

[1406] The CD123 clone 3E3 was combined with a humanized variant of anti-CD3 antibody 4B4 in the CODV-Fab format. Their T-cell engaging effect and their ability to activate T cells in vitro was analyzed.

[1407] Cytotoxic assays were performed as described above. The lytic effect of primary human T cells to THP-1 target cell mediated by these constructs is exemplarily shown in table 20 by CODV-Fab hz4B4(4D_A)×3E3. Cytotoxic activity could be induced reliably with a concentration dependent effect with T cells isolated from seven different healthy donors (table 19).

TABLE-US-00113 TABLE 19 T-cell engaging effect of bispecific CODV-Fab hz4B4(4D_A) × 3E3 detected in a flow cytometry based cytotoxic assay. Presented are mean EC50 values calculated from curves. Cytotoxicity Cytotoxicity (THP-1) (THP-1) geomean EC50 geomean EC50 [pM] +/− SEM [ng/ml] +/− SEM CODV-Fab n = 7 n = 7 hz4B4(4D_A) × 3E3 CODV-Fab 26.5 +/− 5.9 2.0 +/− 0.5

[1408] 3.7 T-Cell Activating Effect of CD123×CD3 CODV-Fab or DART

[1409] The effect of bispecific antibodies on activation status of T cells as safety read out was analyzed by flow cytometry based detection of the expression of activation marker CD25 and CD69 on the surface of primary human T cells, as described before. The comparison included the single chain CD123×CD3 bi-specific diabody in DART format (herein called “MGD006”) which was described in WO2015026892 as comprising a first polypeptide chain of sequence SEQ ID NO:386 (which is SEQ ID NO:1 as shown in WO2015026892) and a second polypeptide chain of sequence SEQ ID NO:387 (which is SEQ ID NO:3 as shown in WO2015026892) covalently bonded to one another by a disulfide bond.

[1410] When the CODV-Fabs were incubated with isolated T cells alone no significant increase in expression of late activation marker CD25 could be detected on the surface of CD4 positive and CD8 positive T cells (data not shown). Equally, there was no concentration dependent increase in expression level of early activation marker CD69 on both T-cell subsets (table 20). Therefore, the construct was evaluated as not active (NA). In contrast, a huge increase in expression level of both markers was measurable when THP-1 target cells were added (CD25 data not shown, CD69 data table 21).

TABLE-US-00114 TABLE 20 Effect of bispecific CD123 × CD3 CODV- Fab or DART on activation state of T cells detected by CD69 expression level in a flow cytometry based assay. Presented are mean percentages of activated CD8 and CD4 T cells at 100 nM antibody concentration and the Min-Effect-Concentration in assays with T cells exclusively. Safety - T cell activation w/o target cells n = 3 CD4+/CD69+ CD8+/CD69+ % Activation % Activation normalized to CD4+/CD69+ normalized to CD8+/CD69+ PBS Min-Effect-Conc PBS Min-Effect- C = 100 nM [pM] C = 100 nM Conc [pM] CODV-Fab mean +/− SEM mean +/− SEM mean +/− SEM mean +/− SEM hz4B4(4D_A) × 3E3 14 +/− 2 NA 13 +/− 2 NA CODV-Fab hz20G6xhz7G3 22 +/− 2 22 +/− 2 CODV-Fab hz20G6xhz7G3 18 +/− 4 15 +/− 2 CODV-Fab-TL1 hz20G6xhz7G3  6 +/− 2  9 +/− 2 CODV-Fab-OL1 Single chain 82 +/− 9 83 +/− 4 antibody DART format MGD006

[1411] The results shown in Table 20 indicate that the single chain antibody (DART) causes significantly more T-cell activation in the absence of target cells under the conditions tested.

TABLE-US-00115 TABLE 21 Effect of bispecific CODV-Fab hz4B4(4D_A) × 3E3 on activation state of T cells detected by CD69 expression level in a flow cytometry based assay. Shown are mean percentages of activated CD8 and CD4 T cells at maximal antibody concentration (Cmax) and at the antibody concentration at EC50 in cytotoxic assay. Assays were performed with co-incubation of THP-1 target cells and T cells. Safety - T cell activation with THP-1 target cells n = 3 CD4+/CD69+ CD8+/CD69+ % Activation CD4+/CD69+ % Activation CD8+/CD69+ normalized to % Activation normalized to % Activation PBS normalized to PBS normalized to C = meanEC50 PBS C = meanEC50 PBS Cytotox Cmax Cytotox Cmax CODV-Fab mean +/− SEM mean +/− SEM mean +/− SEM mean +/− SEM hz4B4(4D_A) × 3E3 71 +/− 4 84 +/− 2 69 +/− 6 82 +/− 4 CODV-Fab

TABLE-US-00116 TABLE 22 Effect of bispecific fully humanized 7G3 containing CODV molecules and single chain Dart on activation state of T cells detected by CD69 expression level in a flow cytometry based assay. Shown are EC50 values of representative tests of activated CD8 and CD4 T cells in cytotoxic assay. Assays were performed with co-incubation of THP-1 target cells and T cells. Safety - T cell activation with THP-1 target cells CD4+ T cells CD8+ T cells (% CD69+ cells) (% CD69+ cells) EC50 (pM) EC50 (pM) CODV-Fab n = 1-6 n = 1-6 hz20G6xhz7G3 1.4 3.1 CODV-Fab hz20G6xhz7G3 3.2 9.9 CODV-Fab-TL1 hz20G6xhz7G3 1.0 3.3 CODV-OL1 Single chain 1.0 3.5 antibody DART format MGD006

[1412] In order to assess the cytotoxic effects of new CD123 antibodies with humanized CD3 parts, the CODV-Fabs “hz20G6x7G3”, “7G3xhz4B4”, “hz4B4x3E3” were generated containing different combinations of Fvs. One Fc containing variant was also generated, the CODV-Fab “hz20G6x7G3-TL4” being Fc tagged at the light chain to form Fc heterodimers with the corresponding heavy chain (TIL4 variant). Affinities to the CD3ε/δ-complex and CD123 of the bispecific construct were measured by Biacore. Furthermore, a cytotoxic assay was performed as described above and CD4+ activation and CD8+ activation was measured.

TABLE-US-00117 TABLE 23 Affinities and activities of bispecific CD123 × CD3 CODV-molecules. CD69 expression @ EC50 cytotox Cytotoxic % % KD KD assay CD4+ CD8+ Bispecific (CD3e/d) (CD123) (THP cells) activa- activa- molecule [nM] [nM] EC50 [pM] tion tion hz20G6 × 5.0 0.6 30.9 ± 3.6 73 66 7G3 CODV- Fab 7G3 × 5.4 0.1 26.7 ± 2.9 65 55 hz4B4 CODV-Fab hz4B4 × 7.0 4.8 26.5 ± 5.9 71 69 3E3 CODV- Fab Fc-tagged 13.9 1.7  16.7 ± 10.1 73 74 variant TL4: hz20G6 × 7G3 CODV- Fab-TL4

[1413] Cytotoxic effects of the CODV-Fab hz20G6xhz7G3, CODV-Fab-TL1 hz20G6xhz7G3, CODV-Fab-OL1 “hz21G6xhz7G3” and the single chain Dart MGD006 were also assessed. Affinities to the CD3ε/δ-complex and CD123 of each bispecific construct were measured by Biacore. Furthermore, a cytotoxic assay was performed as described above and CD4+ activation and CD8+ activation was measured.

TABLE-US-00118 TABLE 24 Affinities and activities of bispecific CD123 × CD3 CODV-molecules and DART (MGD006) Cytotoxic assay KD KD (THP cells) (CD3e/d) (CD123) EC50 [pM] Bispecific molecule [nM] [nM] n = 3 hz20G6xhz7G3 9 0.2 2.1 +/− 0.2 CODV-Fab hz20G6xhz7G3 11 0.2 1.0 +/− 0.1 CODV-Fab-TL1 hz20G6xhz7G3 15 0.4 0.9 +/− 0.1 CODV-Fab-OL1 Single chain antibody 9 0.2  0.3 +/− 0.04 DART format MGD006

[1414] To assess the potential of the molecules to trigger T-cell activation in the presence (wanted) and absence (unwanted) of target cells, a new assay was implemented. NFAT-RE-luc2 Jurkat Cells (Promega #CS176403 cells) were incubated with freshly isolated human T-cells in an Effector target ration of 1:1 at 37° C. and 5% CO2 in RPMI 1640, with 2 g/L (11 mM) Glucose, with GlutaMAX, with 25 mM HEPES in 386 well plates. After 5 hrs the incubation was stopped and luminesce was measures using Bio-Glo Luciferase Assay System, Promega #G7940 in a Luminescence HTS Micro Plate Reader.

TABLE-US-00119 TABLE 25 T-cell activation as activation induced by CD123 × CD3 CODV-molecules and MGD006 measured in Jurkat-NFAT-Luc-reporter cell line. No target cells Activation at Cmax in relation to max. With THP1 cells activation in assay Bispecific EC50 (pM) with target cells (%) molecule n = 3 n = 3 hz20G6xhz7G3 561 ± 0.2 0.2 ± 0.1 CODV-Fab hz20G6xhz7G3 444 ± 0.2 0.4 ± 0.3 CODV-Fab-TL1 hz20G6xhz7G3 320 ± 0.2 0.4 ± 0.3 CODV-Fab-OL1 Single chain 370 ± 0.2 25.1 ± 9.9  antibody DART format MGD006

[1415] Results shown in Table 25 indicate that all antibodies induce reporter cell activation with EC50 values below nM in the presence of target cells. For T-cell engagement approaches, T-cell activation should be restricted to the presence of target cells. This is seen for the CODV molecules as there is no significant luminescence signal in the absence of target cells. In contrast, the single chain DART molecule induces a higher reporter cell line activation in the absence of target cells. These results are in agreement with the results obtained with primary T-cells.

[1416] 3.8 In Vivo Anti-Tumor Activity of CD123×CD3 Bispecific CODV-Fab-TL1 and CD123×CD3 Bispecific CODV-Fab

[1417] Materials and Methods

[1418] Human PBMC and T Cell Isolation from Whole Blood

[1419] PBMCs were isolated from the whole blood of human healthy donors with a Ficoll gradient centrifugation. Whole blood was diluted 1:1 in sterile phosphate buffered saline (PBS). Then, two volumes of thirty-five mL of the diluted blood were put into two 50 mL Falcon Tubes in presence of 15 mL Ficoll-Paque. The tubes were centrifuged at 200 g for 40 minutes at room temperature without brake. The two buffy coat layers were recovered and put in six 50 mL Falcon tubes with 45 mL of sterile PBS and centrifuged three times (in between each centrifugation, the supernatant was discarded and 45 mL of PBS was added) at 100 g during ten minutes at room temperature without brake. After the last centrifugation, the two pellets were put together in a final volume of 50 mL completed by PBS in a 50 mL Falcon tube. The total viable PBMCs number was defined by Vicell counting. The pellet was then recovered in Automacs running buffer from Myltenyi Biotech (130-091-221) and T cells were isolated from PBMCs using the negative selection KIT from Miltenyi Biotech (130-091-156) and Automacs according to manufacturer instructions. The purified T cells were recovered and put in culture in Xvivo-15 5% HIS +peni-strepto1× medium at a concentration of 2.5 x10E+6 cells/mL.

[1420] Human T Cell Amplification

[1421] The human enriched T cell population was activated and expanded in vitro during 14 days using the T Cell Activation/Expansion kit from Miltenyi Biotech (130-091-441)

[1422] Human T Cell Preparation for In Vivo Administration

[1423] Cells and cell culture medium were centrifuged 10 minutes at 400 g. The pellet was recovered at a concentration of 2x10E+7 cells/ml in sterile PBS. Elimination of the activating beads from the amplified T cells was performed using the MACsiMAG separator from Myltenyi Biotech (130-092-168) according to manufacturer instructions. Enriched T cell populations were counted by Vicell counting and were recovered in 25 mL of sterile PBS in a 50 mL Falcon tube. After a step of centrifugation at 400 g during 10 minutes at room temperature, the cell pellet was recovered in an adequate volume of sterile PBS to obtain a final concentration of 5x10E+7 cells/mL.

[1424] Tumor Model

[1425] Molm-13 human Acute Myeloid Leukemia cells expressing CD123 were obtained from the Leibniz-institut DSMZ-German collection of microorganisms and cell cultures (DSMZ Braunshweig, Germany). Cells were grown in culture (37° C., 5% CO.sub.2, 95% humidity) in RPM11640 Glutamax medium (completed with foetal cow serum 20%). Molm-13 cells were infected with a Luciferase vector (SV40-PGL4-Puro—i.e. Luciferase vector consisting in Simian Virus 40 promoterlinked to the Luciferase 2 and the Puromycin resistance cassete sequences) carried by a non-replicative lentivirus.

[1426] The Molm13-luc+ tumoral cells were injected intravenously (IV) in NOD.Cg-Prkdcscid II2rgtm1WjI/SzJ NSG mice (10E+6 cells per animal in 200 μl PBS suspension). Twenty-four hours later, 10E+7 human T-cells were administered intraperitoneally (IP) to the same mice under a volume of 0.2 mL of sterile PBS.

[1427] Baseline bioluminescence imaging at day three post tumor implantation was performed using the IVIS100 imager (PerkinElmer, Waltham, Mass., USA) with the Living Image 3.2 acquisition software (Perkin-Elmer, Waltham, M, USA). Animals were injected IP with Beetle luciferin potassium salt (batch 316019, Promega, Lyon, France) 120 mg/kg solution in PBS 15 minutes before image. Mice were anesthetized with ketamine&Xylazine® (120 mg/kg; 6 mg/kg IM, 5 mV/kg) 5 minutes before image.

[1428] CODV-Fab-TL1 “hz20G6xhz7G3”, CODV-Fab “hz20G6xhz7G3”, CD123×CD3 bispecific DART competitor (Single chain antibody DART format MGD006 or a close analog herein called “DART-tool”) or PBS treatments by intravenous route (IV) or continuous intraperitoneal infusion (CIP) started at day four post tumor implantation on established tumors already detectable in bones, as outlined in table 26 (CODV-Fab-TL1 “hz20G6xhz7G3”), table 27 (CODV-Fab “hz20G6xhz7G3” IV) and table 28 (CODV-Fab “hz20G6xhz7G3” CIP).

TABLE-US-00120 TABLE 26 CD123 × CD3 Bispecific CODV-Fab-TL1 intravenous (IV) evaluation study design Dose Volume/ Animal Treatment Group (nmol/Kg) inj Route Schedule number Control — — — 7 CODV-Fab-TL1 1.3 0.2 ml IV Q3d (4, 7, 10) 7 “hz20G6xhz7G3” CODV-Fab-TL1 0.13 0.2 ml IV Q3d (4, 7, 10) 8 “hz20G6xhz7G3” CODV-Fab-TL1 0.013 0.2 ml IV Q3d (4, 7, 10) 6 “hz20G6xhz7G3” CODV-Fab 1.3 0.2 ml IV Qd (4-13) 8 “hz20G6xhz7G3” Single chain 1.3 0.2 ml IV Qd (4-13) 7 antibody DART format MGD006

TABLE-US-00121 TABLE 27 CD123 × CD3 Bispecific CODV-Fab intravenous (IV) evaluation study design Dose Volume/ Animal Treatment Group (nmol/Kg) inj Route Schedule number Control — — — 5 CODV-Fab 1.3 0.2 ml IV Qd (4-13) 8 “hz20G6xhz7G3” CODV-Fab 0.13 0.2 ml IV Qd (4-13) 5 “hz20G6xhz7G3” Single chain 1.3 0.2 ml IV Qd (4-13) 8 antibody DART format DART tool Single chain 0.13 0.2 ml IV Qd (4-13) 7 antibody DART format DART tool CODV-Fab 1.3 0.2 ml IV Qd (4-13) 8 hz20G6x7G3

TABLE-US-00122 TABLE 28 CD123 × CD3 Bispecific CODV-Fab continuous intraperitoneal infusion (CIP) evaluation study design Dose (nmol/ Volume/ Animal Treatment Group Kg/day) day Route Schedule number Vehicle — 6 μl/day ip Continuous 9 Alzet 1002 infusion (4-14) CODV-Fab 3.9 6 μl/day ip Continuous 9 “hz20G6x hz7G3” Alzet 1002 infusion (4-14) CODV-Fab 1.3 6 μl/day ip Continuous 10 “hz20G6xhz7G3 Alzet 1002 infusion (4-14) Single chain 3.9 6 μl/day ip Continuous 9 antibody DART Alzet 1002 infusion (4-14) format MGD006 Single chain 1.3 6 μl/day ip Continuous 9 antibody DART Alzet 1002 infusion (4-14) format MGD006 CODV-Fab 1.3 0.2 ml IV Qd (4-13) 10 “hz20G6xhz7G3 Vehicle — 6 μl/day ip Continuous 10 Alzet 1002 infusion (4-14) CODV-Fab 0.13 6 μl/day ip Continuous 8 “hz20G6xhz7G3 Alzet 1002 infusion (4-14) CODV-Fab 0.013 6 μl/day ip Continuous 8 “hz20G6xhz7G3 Alzet 1002 infusion (4-14) Single chain 0.13 6 μl/day ip Continuous 8 antibody DART Alzet 1002 infusion (4-14) format MGD006 Single chain 0.013 6 μl/day ip Continuous 9 antibody DART Alzet 1002 infusion (4-14) format MGD006 CODV-Fab 1.3 0.2 ml IV Qd (4-13) 10 “hz20G6xhz7G3

[1429] DATA Collection and Efficacy Criteria

[1430] Animal body weight was monitored from day 3 to the end of assay in order to follow impact of therapy. A dosage producing a 20% weight loss or 15% weight loss for 3 consecutive days or 10% or more drug related deaths, was considered an excessively toxic dosage. Animal body weights included the tumor weights.

[1431] Tumor load was followed by non-invasive bioluminescence imaging (BLI). Baseline BLI was performed at day three post tumor implantation, 24 hours before start of treatments. Animals were dispatched in different groups based on all body bioluminescence signal. Tumor growth was followed in all body and long bones in posteriors legs by BLI signal measurements at days 7, 10 and 14 after tumor implantation. Long bone signal was measured by segmentation and could be influenced by nearby loco-regional signal (eg residual signal in soft tissues in late time points). Treated groups were compared to control animals bearing Molm13-luc+disseminated tumor and Human T cells.

[1432] The primary efficacy end points were the ratio of tumor signal changes from baseline between treated and control groups (dT/dC), the number of partial tumor regressions (PR) and the number of complete tumor regression (CR).

[1433] Tumor growth based on bioluminescence signal curves (expressed in Phot/sec) in time was monitored for each animal of each treatment group and represented as median curve±MAD, both for all body and bone segmented signals. Changes in tumor bioluminescence signal are calculated for each control (C) or treated (T) animal and for each day by subtracting the tumor signal on the day of first treatment (staging day) from the tumor signal on the specified observation day. The median T is calculated for the treated group and the median C is calculated for the control group.

[1434] Then the ratio T/C is calculated and expressed as a percentage:

dT/dC=[(median T day obs−median T day 3)/(median C day obs−median C day 3)]×100

[1435] The dose is considered as therapeutically active when dT/dC at the end of the experiment (day 14) is lower than 42% and very active when dT/dC is lower than 10%.

[1436] Percent tumor regression is defined as the % of tumor signal decrease in the treated group at a specified observation day compared to its signal on the first day of treatment. At a specific time point and for each animal, % regression is calculated as:

[00001]$\%\mathrm{regression}\left(\mathrm{at}t\right)=\frac{Signa{l}_{t0}-Signa{l}_t}{Signa{l}_{t0}}\times 100$

[1437] Given the risk of signal variability due to luciferin kinetics and possible IP miss-injection, signal regression for an animal is considered as a true tumor regression only when observed at least at two consecutive time points.

[1438] Partial regression (PR): Regressions are defined as partial if the tumor signal decreases below the signal at the start of treatment for two consecutive time points, one remaining superior to 50% of baseline signal.

[1439] Complete regression (CR): Regressions are defined as complete if the tumor signal decreases more than 50% below the signal at the start of treatment for two consecutive time points.

[1440] Statistical Analysis

[1441] IV Route Compounds Evaluation

[1442] Individual bioluminescence signal of each group of treatment was compared to others using Bonferroni-Holm adjustment for multiplicity pairwise comparisons following Two way anova with repeated measures by day: p>0.05: NS, 0.05>p >0.01: *, p<0.01: **. Statistical analysis is performed for both all body bioluminescence signals and long-bones bioluminescence signals

[1443] CIP Route Compounds Evaluation

[1444] The CODV-Fab “hz20G6xhz7G3” CIP route evaluation results in data aggregation of two independent studies (1.sup.st study concerning compounds at high dosages, 2.sup.nd study for low dosages, both studies including a vehicle control group and CODV-Fab “hz20G6xhz7G3” 1.3nmol/kg IV Qd positive control group). Statistical analysis of bioluminescence signal of each mouse at each day was performed after data normalization by the mean of the bioluminescence signal of the vehicle group at the same day of the same experiment (pooled vehicle controls n=19; pooled positive controls n=20). Individual normalized bioluminescence signal of each group of treatment was compared to other groups using Bonferroni-Holm adjustment for multiplicity pairwise comparisons following Two way anova with repeated measures by day: p>0.05: NS, 0.05>p >0.01: * , p<0.01: **. Statistical analysis is performed for both all body bioluminescence signals and long-bones bioluminescence signals.

[1445] Results

[1446] CD123×CD3 Bispecfic CODV-Fab-TL1 “hz20G6xhz7G3” IV

[1447] Fully human CODV-Fab-TL1 “hz20G6xhz7G3” IV Q3d in presence of human T cells inhibited Molm13 tumor growth at all tested doses (1.3, 0.13 and 0.013 nmol/Kg Q3d) with dT/dC of 20%, 14% and 38% respectively in whole body (FIGS. 5 and 7) and was associated with tumor regression in long bones at all tested doses with 4/7 CR, 6/8 CR and 2/6 CR respectively (FIGS. 6 and 8).

[1448] Fully human CODV-Fab-TL1 “hz20G6xhz7G3” maximal response was obtained in whole body and in bone at 0.13 nmol/kg Q3d. At this dose, the activity was not statistically different from DART 1.3 nmol/kg IV Qd (whole body dT/dC 29% with 1/7CR and 1/7PR in long bones), and equivalent to CODV-Fab “hz20G6xhz7G3” 1.3 nmol/kg IV Qd (whole body dT/dC 23% with 1/8CR and 1/8PR tumor regression in long bones). Data were confirmed by terminal histopathology analysis (not shown).

[1449] Differences observed between whole body and long bones are linked to residual tumor growth in ovaries and abdominal fat consecutive to extra-medullar tumor dissemination after IV injection.

[1450] CD123×CD3 Bispecific CODV-Fab “hz20G6xhz7G3” IV

[1451] Fully human CODV-Fab “hz20G6xhz7G3” IV in presence of human T cells inhibited tumor growth at all tested doses (1.3 and 0.13 nmol/Kg Qd4-13) with dT/dC of 14% and 39% respectively (FIGS. 9 and 11) associated with tumor regression at 1.3 nmol/kg in long bones with 5/8 CR (FIGS. 10 and 12).

[1452] DART 1.3 nmol/kg IV Qd4-13 inhibited tumor growth with whole body dT/dC 29% and 3/8 CR tumor regression in long bones, not significantly different from CODV-Fab “hz20G6xhz7G3” 1.3 nmol/Kg IV. DART was inactive at inhibiting whole body tumor signal at 0.13 nmol/Kg IV Qd4-13 (dT/dC 62%) despite 1/7PR tumor regression in long bones.

[1453] No significant difference was observed with the same dosage of fully human CODV-Fab “hz20G6xhz7G3” at the end of study. No statistical differences could be seen between fully human CODV-Fab “hz20G6xhz7G3” and partly humanised CODV-Fab hz20G6x7G3 compound when given at the same dosage of 1.3 nmol/kg iv QD4-13: at this dose CODV-Fab hz20G6x7G3 inhibited whole body tumor growth with dT/dC of 34% associated with tumor regressions in long bones (1/8CR and 1/8PR).

[1454] Differences observed between whole body and long bones are linked to residual tumor growth in ovaries and abdominal fat consecutive to extra-medullar tumor dissemination after IV injection.

[1455] CD123×CD3 Bispecific CODV-Fab CIP

[1456] Fully human CODV-Fab “hz20G6xhz7G3” CIP, in presence of human T cells, inhibited whole body tumor growth at 3.9, 1.3 and 0.13 nmol/Kg/day CIP4-14 and was inactive at 0.013 nmol/kg/day with respective dT/dC of 2%, 3%, 21%, and 57% in whole body (FIGS. 13 and 15). It was associated with tumor regression at 3.9, 1.3, and 0.13 nmol/kg in long bones with 5/9CR 3/9PR, 6/10CR 1/10PR and 7/8CR respectively (FIGS. 14 and 16).

[1457] DART 3.9, 1.3, and 0.13 nmol/kg/day CIP4-14, but not 0.013 nmol/kg/day inhibited whole body tumor growth with dT/dC of 21%, 5%, 21% and 46% respectively (FIG. 13) inducing tumor regression in long bones at 3.9, 1.3, and 0.13 nmol/kg with 3/9 CR 3/9PR; 8/9CR; and 4/8CR 2/8PR respectively (FIG. 14).

[1458] Full human CODV-Fab “hz20G6xhz7G3” 1.3 nmol/kg IV Qd4-13 inhibited tumor growth in whole body with dT/dC of 4% and 5% (1.sup.st and 2.sup.nd study resp.) associated with tumor regression in long bones (8/10CR 1/10PR vs 8/10 CR in 1.sup.st and 2.sup.nd study resp).