ANTIGEN BINDING PROTEINS SPECIFICALLY BINDING MAGE-A

Abstract

The present invention concerns antigen binding proteins specifically binding melanoma associated antigen A (MAGE-A) protein-derived antigens. The invention in particular provides antigen binding proteins which specifically bind to the MAGE-A antigenic peptide comprising or consisting of SEQ ID NO: 1 in a complex with a major histocombatibility (MHC) protein. The antigen binding proteins of the invention contain, in particular, the complementary determining regions (CDRs) of novel engineered T cell receptors (TCRs) that specifically bind to said MAGE-A peptide/MHC complex. The antigen binding proteins of the invention are of use for the diagnosis, treatment and prevention of MAGE-A expressing cancerous diseases. Further provided are nucleic acids encoding the antigen binding proteins of the invention, vectors comprising these nucleic acids, recombinant cells expressing the antigen binding proteins and pharmaceutical compositions comprising the antigen binding proteins of the invention.

Claims

1. An antigen binding protein which specifically binds to a MAGE-A antigenic peptide comprising or consisting of the amino acid sequence KVLEHVVRV of SEQ ID NO: 1, wherein said antigenic peptide is in a complex with a major histocompatibility complex (MHC) protein, the antigen binding protein comprising a) a first polypeptide chain comprising a first variable domain comprising three complementary determining regions (CDRs) CDRa1, CDRa2 and CDRa3, wherein the CDRa1 comprises or consists of the amino acid sequence X.sub.1SSSTY SEQ ID NO: 72, wherein X.sub.1 is any amino acid, or an amino acid sequence differing from SEQ ID NO: 72 by at least one amino acid substitution, the CDRa2 comprises or consists of the amino acid sequence IX.sub.1SX.sub.2X.sub.3DX.sub.4 SEQ ID NO: 73, wherein X.sub.1 to X.sub.4 is any amino acid, the CDRa3 comprises or consists of the amino acid sequence CAEX.sub.1X.sub.2SX.sub.3SKIIF SEQ ID NO: 77, wherein X.sub.1 to X.sub.3 is any amino acid, with the proviso that the CDRa3 does not comprise or consist of the amino acid sequence CAEYSSASKIIF (SEQ ID NO: 7) if CDRa1 comprises or consists of the amino acid sequence SEQ ID NO: 5 and CDRa2 comprises or consists of the amino acid sequence SEQ ID NO: 6, optionally with the proviso that the CDRa3 does not comprise or consist of the amino acid sequence CAEYSSASKIIF (SEQ ID NO: 7), (b) a second polypeptide chain comprising a second variable domain comprising three CDRs CDRb1, CDRb2 and CDRb3, wherein the CDRb1 comprises or consists of the amino acid sequence X.sub.1GHDY SEQ ID NO: 78 wherein X.sub.1 is any amino acid, or an amino acid sequence differing from SEQ ID NO: 78 by at least one amino acid substitution, the CDRb2 comprises or consists of the amino acid sequence FX.sub.1X.sub.2X.sub.3X.sub.4P SEQ ID NO: 79, wherein X.sub.1 to X.sub.4 is any amino acid, and the CDRb3 comprises or consists of the amino acid sequence CASRAX.sub.1TGELFF SEQ ID NO: 82, wherein X.sub.1 is any amino acid or an amino acid sequence differing from SEQ ID NO: 82 by at least one amino acid substitution.

2. The antigen binding protein according to claim 1, wherein said antigen binding protein specifically binds to the amino acid sequence KVLEHVVRV of SEQ ID NO: 1 of said antigenic peptide and wherein said antigenic peptide is in a complex with a MHC protein, and wherein, optionally, the MHC protein is a human leukocyte antigen (HLA) protein, optionally HLA-A*02.

3. The antigen binding protein according to claim 1, wherein said antigen binding protein specifically binds to an epitope comprising or consisting of at least three amino acid positions of the MAGE-A antigenic peptide of SEQ ID NO: 1, optionally, at least three amino acid positions selected from the group of amino acid positions consisting of the amino acid positions 1, 5, 7 and 8 or 1, 3, 5 and 7 of the amino acid sequence of SEQ ID NO: 1.

4. An antigen binding protein according to claim 1, wherein said antigen binding protein binds to a complex of said MAGE-A antigenic peptide comprising or consisting of the amino acid sequence of SEQ ID NO: 1 and HLA-A*02, with a K.sub.D which is 100 M, 50 M, 30 M, 25 M, 1 M, 500 nM, 100 nM, 50 nM, 10 nM, optionally 50 pM to 100 M, 50 pM to 10 M, 50 pM to 1 M, optionally, 50 pM to 500 nM, 50 pM to 100 nM, 50 pM to 50 nM and 50 pM to 10 nM.

5. An antigen binding protein according to claim 1, wherein said antigen binding protein has an EC.sub.50 value for MAGE-A/MHC complex presenting cells which is less than 100 nM, less than 50 nM, less 10 nM, less than 900 pM, less than 500 pM, less than 300 pM, less than 200 pM, less than 150 pM, less than 100 pM, less than 50 pM, less than 20 pM, less than 10 pM, such as between 0.1 nM and 20 nM, such as 0.5 nM and 15 nM, 0.8 nM and 12 nM, 0.8 nM and 10 nM, 0.8 nM and 10 nM, 0.8 nM and 10 nM or such as between 1 pM and 150 pM, 1 pM and 100 pM, 1 pM and 50 pM, 1 pM and 20 pM, 1 pM and 20 pM.

6. The antigen binding protein according to claim 1, wherein the antigen binding proteins does not significantly bind to the peptides selected from the list consisting of RABGAP1L-001, AXIN1-001, ANO5-001, TPX2-001, SYNE3-001, MIA3-001, HERC4-001, PSME2-001, HEATR5A-001, CNOT1-003, TEP1-003, PITPNM3-001, INTS4-002, SAMH-001, PPP1CA-006, RPL-007, SETD1A-001, NOMAP-1-0320, NOMAP-1-1223 and ODC-001, optionally HEATR5A-001 and CNOT1-003, when said peptide is in a complex with a MHC protein, optionally in complex with HLA-A*02.

7. The antigen binding protein according to claim 1, wherein the antigen binding protein is an antibody or a fragment thereof, or a bispecific antibody or fragment thereof, or a T cell receptor (TCR) or a fragment thereof, or a bispecific T cell receptor (TCR) or fragment thereof.

8. The antigen binding protein according to claim 1, wherein the first polypeptide and the second polypeptide are covalently or non-covalently linked together.

9. The antigen binding protein according to claim 8, wherein said antigen binding protein is a single chain TCR (scTCR) or a single-chain bispecific antibody.

10. The antigen binding protein according to claim 1, wherein said first variable domain further comprises one or more framework regions selected from the group consisting of FR1-a, FR2-a, FR3-a and FR4-a, wherein FR1-a comprises or consists of the amino acid sequence of SEQ ID NO: 83 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 83, FR2-a comprises or consists of the amino acid sequence of SEQ ID NO: 84 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 84, FR3-a comprises or consists of the amino acid sequence of SEQ ID NO: 85 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 85, FR4-a comprises or consists of the amino acid sequence of SEQ ID NO: 86 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 86, and said second variable domain further comprises one or more framework regions selected from the group consisting of FR1-b, FR2-b, FR3-b and FR4-b, and wherein FR1-b comprises or consists of the amino acid sequence of SEQ ID NO: 87 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 87, FR2-b comprises or consists of the amino acid sequence of SEQ ID NO: 88 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 88 FR3-b comprises or consists of the amino acid sequence of SEQ ID NO: 89 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 89, or FR4-b comprises or consists of the amino acid sequence of SEQ ID NO: 90 or an amino acid sequence at least 85%, at least 90% or at least 95% identical to SEQ ID NO: 90.

11. The antigen binding protein according to claim 10, wherein the framework region(s) comprise(s) at least one amino acid substitution selected from the group of amino acid substitutions comprising: an amino acid substitution at position 19 in FR1-a, wherein said amino acid substitution is optionally S19A, S19V, an amino acid substitution at position 48 and/or 50 in FR2-a, wherein said amino acid substitution at position 48 is optionally A48K and said amino acid substitution at position 50 is optionally L50P, an amino acid substitution at position 46, 47 and/or 54 in FR2-b, wherein said amino acid substitution at position 46, 47 and 54 is optionally M46P, M47G and 154F, respectively, wherein optionally said amino acid at position 54 is 154F when the CDRb3 amino acid sequence as defined above comprises at position 109 the amino acid 109D, an amino acid substitution at position 66 in FR3-b, wherein said amino acid substitution at position 66 is optionally 166C and wherein optionally said amino acid at position 66 is 66C when the CDRb2 amino acid sequence as defined above comprises at position 57 the amino acid 57C, and wherein the positions of the substitutions are given according to the IMGT nomenclature.

12. The antigen binding protein according to claim 11, wherein said first variable domain further comprises one or more framework regions selected from the group consisting of FR1-a, FR2-a, FR3-a and FR4-a, wherein FR1-a comprises or consists of the amino acid sequence of SEQ ID NO: 91 or an amino acid sequence at least 85% identical to SEQ ID NO: 91, wherein optionally said amino acid sequence at least 85% identical to SEQ ID NO: 91 comprises the amino acid 19V, FR2-a comprises or consists of the amino acid sequence of SEQ ID NO: 92 or an amino acid sequence at least 85% identical to SEQ ID NO: 92, wherein optionally said amino acid sequence at least 85% identical to SEQ ID NO: 92 comprises the amino acid 48K and optionally an amino acid substitution at position 50, optionally L50P, FR3-a comprises or consists of the amino acid sequence of SEQ ID NO: 85 or an amino acid sequence at least 85% identical to SEQ ID NO: 85, FR4-a comprises or consists of the amino acid sequence of SEQ ID NO: 86 or an amino acid sequence at least 85% identical to SEQ ID NO: 86, and said second variable domain further comprises one or more framework regions, optionally framework regions, selected from the group consisting of FR1-b, FR2-b, FR3-b and FR4-b, and wherein FR1-b comprises or consists of the amino acid sequence of SEQ ID NO: 87 or an amino acid sequence at least 85% identical to SEQ ID NO: 87, FR2-b comprises or consists of the amino acid sequence of SEQ ID NO: 93 or an amino acid sequence at least 85% identical to SEQ ID NO: 93, wherein optionally said amino acid sequence at least 85% identical to SEQ ID NO: 93 comprises optionally the amino acid 54F, and optionally an amino acid substitution at position 46 and/or 47, wherein said amino acid substitution at position 46 and/or 47 is optionally M46P and/or M47G, respectively, FR3-b comprises or consists of the amino acid sequence of SEQ ID NO: 89 or an amino acid sequence at least 85% identical to SEQ ID NO: 89, or of the amino acid sequence of SEQ ID NO: 94 or an amino acid sequence at least 85% identical to SEQ ID NO: 94, wherein optionally said amino acid sequence at least 85% identical to SEQ ID NO: 94 comprises the amino acid 66C, and wherein FR3-b optionally comprises or consists of an amino acid sequence at least 85% identical to SEQ ID NO: 94 comprising the amino acid 66C, when the CDRb2 amino acid sequence as defined above comprises at position 57 the amino acid 57C, FR4-b comprises or consists of the amino acid sequence of SEQ ID NO: 90 or an amino acid sequence at least 85% identical to SEQ ID NO: 90.

13. The antigen binding protein according to claim 1, wherein said first variable domain is part of a TCR or chain; and/or wherein said second variable domain is part of a TCR or chain.

14. The antigen binding protein according to claim 1, further comprising one or more of the following: (i) one or more further antigen binding sites; (ii) a transmembrane region, optionally including a cytoplasmic signaling region; (iii) a diagnostic agent; (iv) a therapeutic agent; or (v) PK modifying moiety.

15. The antigen binding protein according to claim 1, comprising two polypeptide chains that form two antigen binding sites, wherein a first polypeptide chain has a structure represented by the formula:
V.sub.3-L.sub.1-V.sub.4-L.sub.2-C.sub.L[I] wherein V.sub.3 is a third variable domain; V.sub.4 is a fourth variable domain; L.sub.1 and L.sub.2 are linkers; L.sub.2 may be present or absent; C.sub.L is a light chain constant domain or a portion thereof and present or absent; and wherein a second polypeptide chain has a structure represented by the formula:
V.sub.5-L.sub.3-V.sub.6-L.sub.4-C.sub.H1[II] wherein V.sub.5 is a fifth variable domain; V.sub.6 is a sixth variable domain; L.sub.3 and L.sub.4 are linkers; L.sub.4 may be present or absent; C.sub.H1 is a heavy chain constant domain 1 or a portion thereof and is present or absent; and wherein V.sub.3 or V.sub.4 is a first variable domain as defined in claim 1 and V.sub.5 or V.sub.6 is a second variable domain as defined in claim 1, or V.sub.5 or V.sub.6 is a first variable domain as defined in claim 1 and V.sub.3 or V.sub.4 is a second variable domain as defined in claim 1, and wherein V.sub.3 is the first variable domain and V.sub.5 is the second variable domain in claim 1, and V.sub.4 is a light chain variable domain and V.sub.6 is a heavy chain variable domain or V.sub.4 is a heavy chain variable domain and V.sub.6 is a light chain variable domain, or, V.sub.3 is the second variable domain and V.sub.5 is the first variable domain in claim 1, and V.sub.4 is a light chain variable domain and V.sub.6 is a heavy chain variable domain or V.sub.4 is a heavy chain variable domain and V.sub.6 is a light chain variable domain, or, V.sub.3 is the first variable domain and V.sub.6 is the second variable domain in claim 1, and V.sub.4 is a light chain variable domain and V.sub.5 is a heavy chain variable domain or V.sub.4 is a heavy chain variable domain and V.sub.5 is a light chain variable domain, or V.sub.3 is the second variable domain and V.sub.6 is the first variable domain in claim 1, and V.sub.4 is a light chain variable domain and V.sub.5 is a heavy chain variable domain or V.sub.4 is a heavy chain variable domain and V.sub.5 is a light chain variable domain, V.sub.4 is the first variable domain and V.sub.5 is the second variable domain in claim 1, and V.sub.3 is a light chain variable domain and V.sub.6 is a heavy chain variable domain or V.sub.3 is a heavy chain variable domain and V.sub.6 is a light chain variable domain, or V.sub.4 is the second variable domain and V.sub.5 is the first variable domain in claim 1, and V.sub.3 is a light chain variable domain and V.sub.6 is a heavy chain variable domain or V.sub.3 is a heavy chain variable domain and V.sub.6 is a light chain variable domain, and wherein the light chain variable domain and the heavy chain variable domain form together one antigen binding site, and wherein the first and second variable domain form together one antigen binding site.

16. The antigen binding protein according to claim 15, wherein said heavy chain variable domain and said light chain variable domain bind to an antigen selected from the group consisting of CD3 (such as the CD3, CD3, and CD3 chains), CD4, CD7, CD8, CD10, CD11 b, CD11c, CD14, CD16, CD18, CD22, CD25, CD28, CD32a, CD32b, CD33, CD41, CD41b, CD42a, CD42b, CD44, CD45RA, CD49, CD55, CD56, CD61, CD64, CD68, CD94, CD90, CD117, CD123, CD125, CD134, CD137, CD152, CD163, CD193, CD203c, CD235a, CD278, CD279, CD287, Nkp46, NKG2D, GITR, FcRI, TCR/ and TCR/, HLA-DR and/or bind to an effector cell.

17. The antigen binding protein according to claim 15, comprising two polypeptide chains that form two antigen binding sites, wherein one polypeptide chain has a structure represented by the formula [III]:
V.sub.3-L.sub.1-V.sub.4-L.sub.2-C.sub.L-L.sub.5-F.sub.C1[III] and one polypeptide chain has a structure represented by the formula [IV]:
V.sub.5-L.sub.3-V.sub.6-L.sub.4-C.sub.H1-L.sub.6-F.sub.C2[IV] wherein V.sub.3, L.sub.1, V.sub.4, L.sub.2, C.sub.L, V.sub.5, L.sub.3, V.sub.6, L.sub.4 and C.sub.H1, are as defined in claim 15 and wherein L.sub.5 and L.sub.6 are linkers and present or absent and F.sub.C1, and F.sub.C2 are Fc-domains and wherein F.sub.C1 and F.sub.C2 are the same or different.

18. The antigen binding protein according to claim 17, wherein V.sub.3 is the first variable domain and V.sub.6 is the second variable domain as defined in claim 17, and V.sub.4 is a light chain variable domain and V.sub.5 is a heavy chain variable domain, or V.sub.3 is the first variable domain and V.sub.6 is the second variable domain as defined in claim 17, and V.sub.4 is a heavy chain variable domain and V.sub.5 is a light chain variable domain, and wherein L.sub.1 and L.sub.2 comprise or consist of the amino acid sequence GGGSGGGG of SEQ ID NO: 96, and wherein L.sub.2, L.sub.5, L.sub.4 and L.sub.6 are absent, and wherein F.sub.C1 comprises or consists of the amino acid sequence SEQ ID NO: 113, when V.sub.4 is a heavy chain variable domain and F.sub.C2 comprises or consists of the amino acid sequence SEQ ID NO: 112 when V.sub.5 is a light chain variable domain, or F.sub.C1 comprises or consists of the amino acid sequence SEQ ID NO: 112, when V.sub.4 is a light chain variable domain and, F.sub.C2 comprises or consist of the amino acid sequence SEQ ID NO: 113 when V.sub.5 is a heavy chain variable domain.

19. An isolated nucleic acid comprising a sequence encoding for an antigen binding protein according to claim 1, or a nucleic acid vector comprising said nucleic acid.

20. A recombinant host cell comprising, or the isolated nucleic acid or the nucleic acid vector-according to claim 19, wherein said host cell optionally is a) a lymphocyte, optionally a T lymphocyte or T lymphocyte progenitor cell, optionally a CD4 or CD8 positive T cell or b) a cell for recombinant expression, optionally a Chinese Hamster Ovary (CHO) cell.

21. A pharmaceutical composition comprising the antigen binding protein according to claim 1, and a pharmaceutically acceptable carrier, diluent stabilizer, and/or excipient.

22. A method of producing the antigen binding protein according to claim 1, comprising a. providing a suitable host cell, b. providing a genetic construct comprising a coding sequence encoding the antigen binding protein according to claim 1, c. introducing said genetic construct into said suitable host cell, and d. expressing said genetic construct by said suitable host cell.

23. The method according to claim 22, further comprising the isolation and purification of the antigen binding protein from the suitable host cell and, optionally, reconstitution of the antigen binding protein in a T cell.

24. A method of treating a patient who has a MAGEA4 and/or MAGEA8 positive cancer, comprising administering to the patient the antigen binding protein according to claim 1, wherein said cancer is selected from the group consisting of lung cancer, liver cancer, head and neck cancer, skin cancer, renal cell cancer, brain cancer, gastric cancer, colorectal cancer, pancreatic cancer, prostate cancer, leukemia, breast cancer, Merkel cell carcinoma, ovarian cancer, urinary bladder cancer, uterine cancer, gallbladder cancer, bile duct cancer, osteosarcoma cancer, and esophageal cancer.

25. (canceled)

26. (canceled)

27. The antigen binding protein according to claim 24, wherein (i) lung cancer is non-small cell lung cancer (NSCLC), optionally non-small cell lung cancer adenocarcinoma or squamous cell non-small cell lung cancer (SNSCLC); or small cell lung cancer (SCLC); (ii) skin cancer is melanoma; (iii) head and neck cancer is head and neck squamous cell carcinoma (HNSCC); (iv) liver cancer is hepatocellular cancer (HCC); (v) esophageal cancer is gastroesophageal junction cancer.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0567] The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.

[0568] FIGS. 1-18 depict embodiments as described herein.

DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT

[0569] FIG. 1 shows results from conversion of a TCR into stabilized scTCR via yeast surface display. ScTCR molecules displayed on the surface of transformed Saccharomyces cerevisiae EBY100 were stained with FITC-labeled anti-Vbeta8 antibody and PE-labeled HLA-A*02/MAG-003 tetramer. The non-modified scTCR R7P1D5 (left panel, SEQ ID NO: 19) is compared to a scTCR variant bearing two stabilizing single point mutations (right panel, SEQ ID NO: 22), which was derived from the selection of a random mutation scTCR library. It can be seen that the scTCR variant bearing two stabilizing single point mutations shows an increased HLA-A*02/MAG-003 tetramer staining.

[0570] FIG. 2A shows results from yeast surface display for scTCR affinity maturation of CDR3 alpha. Stabilized scTCR comprising non-modified and maturated CDR3 alpha were stained with HLA-A*02/MAG-003 monomers at a concentration of 15 nM. Counterstaining with a mix of 10 HLA-A*02 tetramers, each applied at a concentration of 10 nM, containing peptides (SEQ ID NO: 23 to 32) with high sequence similarity to MAG-003 (SEQ ID NO: 1). Stabilized scTCR (SEQ ID NO: 22) with non-modified alpha chain CDR3 sequence CAEYSSASKIIF (SEQ ID NO: 7) is compared with scTCR variants comprising the affinity maturated alpha chain CDR3 sequences CAEFSSASKIIF (SEQ ID NO: 38), CAEMTSESKIIF (SEQ ID NO: 35), CAEFTSESKIIF (SEQ ID NO: 36), CAEFNSESKIIF (SEQ ID NO: 37) and CAEATSESKIIF (SEQ ID NO: 39), respectively. It can be seen, that the yeast clones with mutated CDRs alpha 2 (SEQ ID NOs 56, 57, 59), beta 1 (SEQ ID NO 62) or beta 2 (SEQ ID NOs 63, 64, 65, 66, 70) also revealed strongly improved binding while specificity for HLA-A*02/MAG-003 was mainly retained.

[0571] FIG. 2B shows results from yeast surface display for scTCR affinity maturation of CDR2 alpha, CDR3 alpha, CDR1 beta and CDR2 beta, respectively. Stabilized scTCR comprising non-modified and maturated CDRs were stained with HLA-A*02/MAG-003 monomer (target pHLA monomer) and counterstained with a mix of 10 HLA-A*02 tetramers, each applied at a concentration of 10 nM, containing peptides (SEQ ID NO: 23 to 32) with high sequence similarity to MAG-003 (SEQ ID NO: 1). (a) Comparison of scTCR variants with non-modified (SEQ ID NO: 6) and maturated alpha chain CDR2 sequences IYSSQDQ (SEQ ID 56), IYSSQDV (SEQ ID 57) and IYSSQDS (SEQ ID 59), each stained with 40 nM HLA-A*02/MAG-003 monomer. (b) Comparison of scTCR variants with non-modified (SEQ ID NO: 7) and maturated alpha chain CDR3 sequences CAEFSSASKIIF (SEQ ID NO: 38), CAEMTSESKIIF (SEQ ID NO: 35), CAEFTSESKIIF (SEQ ID NO: 36), CAEFNSESKIIF (SEQ ID NO: 37) and CAEATSESKIIF (SEQ ID NO: 39), each stained with 15 nM HLA-A*02/MAG-003 monomer. (c) Comparison of scTCR variants with non-modified (SEQ ID NO: 12) and maturated beta chain CDR1 sequence PGHDY (SEQ ID NO: 62), each stained with 40 nM HLA-A*02/MAG-003 monomer. (d) Comparison of scTCR variants with non-modified (SEQ ID NO: 13) and maturated beta chain CDR2 sequences FCYGHP (SEQ ID NO: 63), FCYGVP (SEQ ID NO: 64), FCYGTP (SEQ ID NO: 65), FCYGAP (SEQ ID NO: 66), FCYGMP (SEQ ID NO: 70), each stained with 5 nM HLA-A*02/MAG-003 monomer. Staining with HLA-A*02/MAG-003 tetramers was significantly increased for all R7P1D5 CDRa3 mutants when compared to the parental TCR R7P1D5, indicating an improved peptide-HLA binding of the mutant variants.

[0572] FIG. 3 shows staining of maturated R7P1D5 TCR variant expressing human CD8+ T cells with PE-labeled HLA-A*02/MAG-003 tetramers and FITC-labeled anti-Vbeta 8 antibody. For control purpose, no TCR (MOCK) or the 1G4 TCR specific for NYESO1-001 were expressed and staining with PE-labeled HLA-A*02/NYESO1-001 tetramers was used.

[0573] FIG. 4 shows IFN-gamma release of maturated R7P1D5 TCR variant expressing human CD8+ T cells in response to MAG-003. For control purpose, no TCR (MOCK) or the 1G4 TCR specific for NYESO1-001 were expressed. IFN-gamma release was determined by ELISA after co-culture of electroporated CD8+ T cells with T2 cells loaded with a serial dilution of MAG-003. Baseline levels of IFN-gamma (released by MOCK electroporated CD8+ T cells upon coculture with respectively loaded T2 cells) were subtracted. Compared to the parental TCR R7P1D5, maturated R7P1D5 CDRa3 mutants 1-5 showed an increased IFN-gamma release in response to MAG-003.

[0574] FIG. 5 shows IFN-gamma release of maturated R7P1D5 TCR variant expressing human CD8+ T cells in response to MAG-003 and different similar peptides. For control purpose, no TCR (MOCK) or the 1G4 TCR specific for NYESO1-001 were expressed. IFN-gamma release was determined by ELISA after co-culture of electroporated CD8+ T cells with T2 cells loaded with 10 pM peptide. Baseline levels of IFN-gamma (released by MOCK electroporated CD8+ T cells upon coculture with respectively loaded T2 cells) were subtracted. Compared to the parental TCR R7P1D5, maturated R7P1D5 CDRa3 mutants 1-5 showed an increased IFN-gamma release in response to MAG-003 and no IFN-gamma release for the tested similar peptides.

[0575] FIG. 6 shows IFN-gamma release of maturated R7P1D5 TCR variant expressing human CD8+ T cells in response to MAG-003 and the respective alanine-substitution peptides. For control purpose, no TCR (MOCK) or the 1G4 TCR specific for NYESO1-001 were expressed. IFN-gamma release was determined by ELISA after co-culture of electroporated CD8+ T cells with T2 cells loaded with 10 pM peptide. Baseline levels of IFN-gamma (released by MOCK electroporated CD8+ T cells upon coculture with respectively loaded T2 cells) were subtracted.

[0576] FIG. 7A shows production yields and heat-stress stability data of soluble scTCR-Fab antigen binding proteins comprising TCR variable domains with different affinity-maturated CDRs and/or framework mutations.

[0577] FIG. 7B shows binding curves of soluble scTCR-Fab antigen binding proteins S, #1, #2 and I to MAG-003 in complex with HLA-A*02 as measured by biolayer interferometry. Increasing concentrations of HLA-A*02/MAG-003 monomers in solution were used for each interaction measurement and are given in nM.

[0578] FIG. 7C shows binding curves of soluble scTCR-Fab antigen binding proteins I (grey curves) and #19 (black curves) for MAG-003 and the normal tissue-derived off-target peptides SYNE3-001, TPX2-001, PSME2-001, respectively, in complex with HLA-A*02. Biolayer interferometry was used for the measurement and a concentration of 1 pM scTCR-Fab in solution was applied.

[0579] FIG. 8A shows production yields and heat-stress stability data of soluble TCER antigen binding proteins comprising TCR variable domains with selected affinity-maturated CDRs combined with different VH/VL sequences of the humanized T cell-recruiting antibody BMA031(36) as well as different variants 17, 17opt, 21, 23 of the humanized T cell-recruiting antibody UCHT1.

[0580] FIG. 8B shows binding curves of TCER antigen binding proteins comprising the TCR variable domain variants 114-iso0, 114-iso1 and 114-iso2 in combination with UCHT1(17) and BMA031(36) for MAG-003 in complex with HLA-A*02. Biolayer interferometry was used for the measurement and increasing concentrations of TCER molecules in solution were applied and are given in nM.

[0581] FIG. 8C shows binding curves of 114-iso1-BMA(36) and 114-iso2-UCHT1(17) TCER molecules for MAG-003 and normal tissue-derived off-target peptides NOMAP-1-0320, NOMAP-1-1223, ODC-001, respectively, in complex with disulfide-stabilized HLA-A*02. Biolayer interferometry was used for the measurement and a concentration of 1 pM TCER in solution was applied.

[0582] FIG. 9 shows in vitro efficiency data of TCER-mediated killing of HLA-A*02-positive tumor cell lines with detectable MAG-003 expression (Hs695T and A375) and without detectable MAG-003 expression (BV173). TCER molecules comprising UCHT1(17) antibody variable domains with variable domains of the TCR variants #114-iso0 (SEQ ID NO: 127 and 130), #114-iso1 (SEQ ID NO: 131 and 130) and #114-iso2 (SEQ ID NO: 133 and 130) were tested (upper panel). Furthermore, TCER molecules comprising BMA031(36) antibody variable domains with variable domains of the TCR variants #114-iso0 (SEQ ID NO: 135 and 136) and #114-iso2 (SEQ ID NO: 139 and 136) were tested (lower panel). The tumor cells were co-incubated with human PBMC from a healthy HLA-A*02+ donor (HBC-720) at a ratio of 1:10 and the cytotoxic activity in presence of increasing concentrations of TCER molecules was analyzed by LDH-release assay. After 48 hours, lysis of target cell lines was measured utilizing CytoTox 96 Non-Radioactive Cytotoxicity Assay Kits (PROMEGA). Efficient cell lysis is shown for TCER molecules comprising UCHT1(17) and BMA031(36) antibodies.

[0583] FIG. 10 shows TCER-mediated in vitro killing of HLA-A*02-positive tumor cell lines with detectable MAG-003 expression (Hs695T, A375 and U205). TCER molecules comprising UCHT1(17) antibody variable domains with variable domains of the TCR variants #114-iso1 (SEQ ID NO: 131 and 130) and #114-iso2 (SEQ ID NO: 133 and 130) as well as TCER molecules comprising BMA031(36) antibody variable domains with variable domains of the TCR variant #114-iso1 (SEQ ID NO: 137 and 136) were tested. The tumor cells were co-incubated with human PBMC from a healthy HLA-A*02+ donor (HBC-982) at a ratio of 1:10 and the cytotoxic activity in presence of increasing concentrations of TCER molecules was analyzed by LDH-release assay. After 48 hours, lysis of tumor cell lines was measured utilizing CytoTox 96 Non-Radioactive Cytotoxicity Assay Kits (PROMEGA). Efficient tumor cell lysis is shown for TCER molecules comprising UCHT1(17) and BMA031(36) antibodies.

[0584] FIG. 11 shows TCER-mediated in vivo efficacy in human PBMC-engrafted, immunodeficient NOG mice against tumor xenografts of Hs695T cells, an HLA-A*02-positive tumor cell line with detectable presentation of MAG-003 (SEQ ID NO: 1) and PRAME-004 (SEQ ID NO: 168). TCER comprising UCHT1(17) antibody variable domains with variable domains of the TCR variant #114-iso0 (SEQ ID NO: 127 and 130, group 3) were compared to TCER comprising BMA031(36) antibody variable domains with variable domains of the TCR variant #114-iso0 (SEQ ID NO: 135 and 136, group 4). As a control, a PRAME-004-targeting TCER (SEQ ID NO 169 and 170, group 2) was used. Group 1 represent the vehicle-treated control group in which one animal was sacrificed due to tumor ulceration at day 11. Each symbol connected by a line represents an individual animal and treatment days are indicated with an arrow. Complete remission is shown for the TCER comprising UCHT1(17) antibody and BMA031(36) antibody for T cell recruitment.

[0585] FIG. 12 shows in vitro safety data for TCER molecule 114-iso1-BMA(36) (SEQ ID NO: 137 and 136) as assessed in LDH-killing experiments with 11 human HLA-A*02-positive normal tissue primary cells from different organs. The different cells were co-cultured with PBMC effector cells from healthy HLA-A*02+ donors at a ratio of 1:10 and increasing TCER concentrations. Cells were co-incubated in a 50% mixture of primary tissue cell-specific medium and T cell medium. For safety window determination, the TCER molecules were co-incubated in an identical setup with the MAG-003-positive tumor cell line Hs695T. The safety window was defined based on the lowest observed effect level (LOEL) determined as the first TCER concentration exhibiting a response over the cut-off value. The cut-off was defined as ([standard deviation from all triplicates3]+control normal tissue sample w/o TCER) and is indicated as dotted line in each cytotoxicity plot. For each normal tissue cell type, the safety window determined based on LOEL was larger than 1,000-fold or 10,000-fold indicating a favorable safety profile of the 114-iso1-BMA(36) TCER molecule.

[0586] FIG. 13 shows TCER-mediated in vivo efficacy in human PBMC-engrafted, immunodeficient NOG mice against tumor xenografts of Hs695T cells, an HLA-A*02-positive tumor cell line with detectable presentation of MAG-003 (SEQ ID NO: 1). TCER molecules comprising UCHT1(V17opt) antibody variable domains and variable domains of the TCR variant 114-iso1 (SEQ ID NO: 131 and 167, group 3) were compared to TCER comprising BMA031(36) antibody variable domains with variable domains of the TCR variant 114-iso1 (SEQ ID NO: 137 and 136, group 2). Group 1 represent the vehicle-treated control group. The graphs show the mean tumor volume derived from six (group 2 and 3) to ten (vehicle) mice per group. Complete remission is shown for the TCER molecules comprising UCHT1(17opt) antibody and BMA031(36) antibody for T cell recruitment upon once weekly i.v. dosing of 0.01 mg/kg body weight.

[0587] FIG. 14 shows the pharmacokinetic profile of 114-iso1-UCHT1(17) TCER molecule in NOG mice. Mice received a single intravenous injection of 2 mg/kg body weight. TCER plasma concentrations of samples collected at different time points were determined by ELISA detecting either the presence of specific protein domains (F.sub.cV.sub.L assay) or bispecific binding activity (CD3pMHC assay).

[0588] FIG. 15 shows design of TCER molecules. Bispecific T cell-engaging receptors (TCER) are fusion proteins containing the variable domains (V, V) of an affinity-maturated TCR and the variable domains (VH, VL) of a humanized T cell-recruiting antibody. The molecules use an effector function-silenced F.sub.c part, which confers extended half-life and also improved stability/manufacturability characteristics.

[0589] FIG. 16 shows in vitro safety data for TCER molecule 114-iso1-BMA(36) (SEQ ID NO: 137 and 136) as assessed in LDH-killing experiments with 12 human HLA-A*02-positive normal tissue primary cells from different organs. The different cells were co-cultured with PBMC effector cells from healthy HLA-A*02+ donors at a ratio of 1:10 and increasing TCER concentrations. Cells were co-incubated in a 50% mixture of primary tissue cell-specific medium and T cell medium. For safety window determination, the TCER molecules were co-incubated in an identical setup with the MAG-003-positive tumor cell line Hs695T. The safety window was defined based on the lowest observed effect level (LOEL) determined as the first TCER concentration exhibiting a response over the cut-off value. The cut-off was defined as ([standard deviation from all triplicates x 3]+control normal tissue sample w/o TCER) and is indicated as dotted line in each cytotoxicity plot. For each normal tissue cell type, the safety window determined based on LOEL was larger than 1,000-fold or 10,000-fold indicating a favorable safety profile of the 114-iso1-BMA(36) TCER molecule.

[0590] FIG. 17 shows cytotoxicity and IFN-gamma release of maturated R7P1D5 TCR variant expressing human CD8+ T cells in response to tumor cell lines presenting different copy numbers of HLA-A*02/MAG-003. For control purpose, no TCR (MOCK) or the 1G4 TCR specific for NYESO1-001 were expressed. LDH and IFN-gamma levels in the supernatants were determined after co-culture of electroporated CD8+ T cells with the tumor cell lines.

[0591] FIG. 18: Normal tissues and tumor samples analyzed for peptide presentation have been grouped according to their organ of origin. Upper part: Median MS signal intensities from technical replicate measurements and aliquot measurements are plotted as dots for single HLA-A*02 positive normal (left part of figure) and tumor samples (right part of figure) on which the peptide was detected. Box-and-whisker plots represent normalized signal intensities over multiple samples and have been defined in log space. Boxes display median, 25th and 75th percentile. Whiskers extend to the lowest data point still within 1.5 interquartile range (IQR) of the lower quartile, and the highest data point still within 1.5 IQR of the upper quartile. Lower part: The relative peptide detection frequency in every organ is shown as spine plot. Numbers below the panel indicate number of samples on which the peptide was detected out of the total number of samples evaluable for this analysis for each organ (N=581 for normal samples, N=771 for tumor samples). If the peptide has been detected on a sample but could not be quantified for technical reasons, the sample is included in this representation of detection frequency, but no dot is shown in the upper part of the figure.

[0592] Abbreviations used: adipose: adipose tissue; adrenal gl: adrenal gland; bladder: urinary bladder; bloodvess: blood vessel; esoph: esophagus; gall bl: gallbladder; intest. la: large intestine; intest. sm: small intestine; nerve cent: central nerve; nerve perith: peritheral nerve; parathyr: parathyroid gland; pent: peritoneum; pituit: pituitary; skel. mus: skeletal muscle; AML: acute myeloid leukemia; BRCA: breast cancer; CCC: cholangiocellular carcinoma; CLL: chronic lymphocytic leukemia; CRC: colorectal cancer; GBC: gallbladder cancer; GBM: glioblastoma; GC: gastric cancer; GEJC: gastro-esophageal junction cancer; HCC: hepatocellular carcinoma; HNSCC: head and neck squamous cell carcinoma; MEL: melanoma; NHL: Non-Hodgkin lymphoma; NSCLCadeno: non-small cell lung cancer adenocarcinoma; NSCLCother: NSCLC samples that could not unambiguously be assigned to NSCLCadeno or NSCLCsquam; NSCLCsquam: squamous cell non-small cell lung cancer; OC: ovarian cancer; OSCAR: esophageal cancer; PACA: pancreatic cancer; PRCA: prostate cancer; RCC: renal cell carcinoma; SCLC: small cell lung cancer; UBC: urinary bladder carcinoma; UEC: uterine and endometrial cancer.

EXAMPLES

[0593] T cell receptors (TCRs) against cancer antigens are often of lower affinity when compared to TCRs targeting viral antigens, and this may be one possible explanation for tumor immune escape (Aleksic et al. 2012, Eur J Immunol. 2012 December; 42(12):3174-9). Therefore, it is desirable to generate TCR variants with higher affinity for the use as cancer antigen-targeting constructs in an adoptive cell therapy, or as recognition module of a soluble therapeutic drug, i.e. bispecific molecules (Hickman et al. 2016, J Biomol Screen. 2016 September; 21(8):769-85). This invention thus relates to the modification and optimization of the T cell receptor R7P1D5, which was identified from the human natural TCR repertoire based on high avidity and selectivity for the tumor associated peptide MAG-003 (SEQ ID NO: 1). TCR R7P1D5 comprises an alpha variable domain with the amino acid sequence of SEQ ID NO: 4 and a beta variable domain with the amino acid sequence of SEQ ID NO: 11, which are disclosed in WO 2017/158103.

Example 1: Generation of Stable scTCR

[0594] For the present invention, the TCR R7P1D5 (SEQ ID NOs: 2 and 9, full length) was converted into a single chain TCR construct (scTCR R7P1D5, SEQ ID NO: 19) using the variable alpha (SEQ ID NO: 4) and beta (SEQ ID NO: 11) domains and an appropriate glycine-serine linker sequence. For TCR maturation via yeast surface display, the DNA of the corresponding sequence was synthesized and transformed into Saccharomyces cerevisiae EBY100 (MATa AGA1::GAL1AGA1::URA3 ura352 trp1 leu2delta200 his3delta200 pep4::HIS3 prbd1.6R can1 GAL) (ATCC MYA4941) together with a yeast display vector containing a leader sequence and the Aga2p yeast mating protein (SEQ ID NO: 19), based on pCT302 (Boder and Wittrup, Methods Enzymol. 2000; 328:430-44). The resulting fusion protein after homologous recombination in the yeast (SEQ ID NO: 16) contains a leader peptide at the N-terminus of the Aga2p protein, responsible for the display of the protein of interest (Boder and Wittrup, Nat Biotechnol. 1997 June; 15(6):553-7), short peptide tags including linker sequences (SEQ ID NOs: 18 and 20) for expression controls and the protein of interest, namely the scTCR R7P1D5 (SEQ ID NO: 19) or its variants. Libraries of scTCR variants were generated via a random mutation PCR approach spanning the whole gene sequence of the scTCR R7P1D5. The transformation of yeast cells was performed as described in WO 2018/091396 and resulted in up to 10.sup.9 yeast clones per library. The selection process for the yeast clones bearing mutant scTCR variants with improved binding to MAG-003 in the context of HLA-A*02 was essentially performed as described in Smith et al. (Methods Mol Biol. 2015; 1319:95-141). To ascertain high expression and correct conformation of yeast surface-displayed R7P1D5 scTCR variants, staining with an anti-Vbeta8 (Life technologies, clone 1C1) antibody was used, together with HLA-A*02/MAG-003 tetramer staining (FIG. 1). The scTCR conversion by yeast surface display revealed two crucial stabilizing mutations for the proper presentation of the scTCR on the cell surface, one in the framework region FR1-a (SEQ ID NO: 83) of the scTCR alpha chain and one in the CDR2 of the alpha chain (SEQ ID NO: 6), which resulted in the stabilized version scTCR R7P1D5S (SEQ ID NO: 22) showing improved HLA-A*02/MAG-003 tetramer binding.

Example 2: Affinity Maturation of Stabilized scTCR

[0595] To generate scTCR molecules with higher binding affinity towards HLA-A*02/MAG-003, all CDRs were maturated individually, using the previously identified stabilized scTCR R7P1D5S (SEQ ID NO: 22). The CDR residues were randomized by using degenerate DNA oligo primers essentially as described previously (Smith et al., Methods Mol Biol. 2015; 1319:95-141). The resulting DNA libraries were transformed as described in example 1. For preservation of binding specificity, negative selection was employed against a mix of HLA A*02 tetramers comprising normal tissue-derived peptides (SEQ ID NOs: 23 to 32) showing high degree of sequence similarity to MAG-003 peptide (SEQ ID NO 1).

[0596] For the selection of affinity enhanced and selective R7P1D5S scTCR variants, a decreasing concentration of HLA-A*02/MAG-003 tetramer or monomer was used for each selection round. After three selection rounds, single scTCR clones were isolated and sequenced, resulting in different affinity maturated CDR sequences. As exemplarily shown for scTCR with maturated CDRa3 sequences (SEQ ID NOs: 35 to 39), a strong improvement in HLA-A*02/MAG-003 monomer binding could be demonstrated. The selectivity of HLA-A*02/MAG-003 binding was retained as confirmed by the low binding to the mix of 10 HLA-A*02 tetramers containing normal tissue-derived peptides with high degree of sequence similarity to MAG-003 peptide (SEQ ID NO 1). CDRa3 mutant 5 (SEQ ID NO: 39) showed slightly increased cross-binding of tetrameric HLA-A*02/similar peptides (FIG. 2A). Additionally, the yeast clones with mutated CDRs alpha 2 (SEQ ID NOs 56, 57, 59), beta 1 (SEQ ID NO 62) or beta 2 (SEQ ID NOs 63, 64, 65, 66, 70) revealed strongly improved binding while specificity for HLA-A*02/MAG-003 was mainly retained (FIG. 2B).

Example 3: Use of Maturated TCRs for Cellular Expression

[0597] Modification of T cells to express TCRs recognizing a tumor-specific peptide-MHC is a promising strategy of redirecting T cells to cancer cells. The usage of maturated CDR3 sequences could improve reactivity of cell-bound TCRs against HLA-A*02/MAG-003 and the identified CDRa3 mutant sequences (SEQ ID NOs: 35 to 39) were grafted onto the parental TCR R7P1D5 alpha chain (SEQ ID NO: 2) and combined with the parental TCR R7P1D5 beta chain (SEQ ID NO: 9). The resulting mutant TCR variants (R7P1D5 CDRa3 mutant 1-5 comprising the alpha chain variant sequences SEQ ID NOs: 40 to 44, respectively) were expressed in human CD8+ T cells after electroporation of respective mRNA generated by in vitro transcription of PCR-amplified DNA constructs. For control purpose, the 1G4 TCR (SEQ ID NOs: 132 and 134) was expressed, which is directed against NYESO1-001 peptide (SEQ ID NO: 45) in complex with HLA-A*02. After overnight incubation of RNA-electroporated CD8+ T cells, expression of introduced TCR variants was analyzed by staining with PE-labeled peptide-HLA-A*02 tetramers and FITC-labeled anti-Vbeta 8 antibody. The parental TCR R7P1D5 and all variants derived thereof showed similar anti-Vbeta 8 staining when compared to the background levels from mock-electroporated sample and the Vbeta 8-negative NYESO1 TCR control, which argues for similar expression efficiency. In contrast, staining with HLA-A*02/MAG-003 tetramers was significantly increased for all R7P1D5 CDRa3 mutants when compared to the parental TCR R7P1D5 (FIG. 3), indicating an improved peptide-HLA binding of the mutant variants. Functional activation of CD8+ T cells (20,000 cells/well) expressing the different R7P1D5 TCR variants was investigated by measuring T cell-mediated IFN-gamma release in response to T2 target cells (20,000 cells/well) loaded with either a dilution series of MAG-003 (FIG. 4) or 10 pM of MAG-003 and potential off-target peptides (FIG. 5). Potential off-target peptides were selected from a database of normal tissue-presented HLA-A*02 bound peptides (XPRESIDENT database) based on high sequence similarity (similarity BLAST search) to MAG-003. Compared to the parental TCR R7P1D5, maturated R7P1D5 CDRa3 mutants 1-5 showed increased reactivity (FIG. 4) as indicated by the 3- to 9-fold lower concentration (EC.sub.50 values, Table 3) required to induced half-maximal IFN-gamma release. As expected, only background IFN-gamma release was observed with T cells expressing no additional TCR (MOCK) or the 1G4 control TCR specific for NYESO1-001. To analyze the selectivity of MAG-003 recognition of the maturated R7P1D5 TCR variants, the IFN-gamma release in response to T2 cells loaded with different similar peptides (SEQ ID NOs: 23 to 34) was analyzed. Data were corrected for background IFN-gamma release by subtracting IFN-gamma levels released by mock-electroporated CD8+ T cells upon coculture with respectively loaded T2 cells. Except for R7P1D5 CDRa3 mutant 5 that showed slightly increased signals for several similar peptides and a low signal of R7P1D5 CDRa3 mutant 2 against MIA3-001, all other maturated R7P1D5 TCR variants were highly selective with no cross-reactivity towards the tested similar peptides (FIG. 5). As all TCR mutant variants induced considerably higher MAG-003-specific IFN-gamma release compared to the parental TCR, the R7P1D5 CDRa3 mutants 4, 1 and 3 (and potentially 2) are most promising candidates for cellular TCR-based tumor targeting. R7P1D5 CDRa3 mutants 1-5 were further characterized with respect to their binding epitope by mutational analysis. R7P1D5 CDRa3 mutants 1-4 showed a broad binding epitope with stringent recognition of positions 1, 5, 7, and 8 as alanine substitutions at these positions abrogated the response. For some R7P1D5 variants, such as mutant 1, a slight effect was also observed for alanine substitution at position 3. In accordance with its lower specificity shown with the similar peptides, the R7P1D5 CDRa3 mutant 5 TCR recognized positions 7 and 8 with reduced stringency when compared to mutant variants 1-4 (FIG. 6). The functional activity of the CDRa3 mutant TCR variants was further evaluated on tumor cell lines presenting different copy numbers of HLA-A*02/MAG-003. RNA-electroporated CD8+ T cells were cocultured with the tumor cell lines at an effector-to-target ratio of 3:1 (NCI-H1755, MCF-7: 25,000 cells/well; Hs695T, A375, U20: 12,500 cells/well). Expression of the introduced TCR variants in CD8+ T cells was verified as described above. After coculturing effector and target cells for 24 h, levels of LDH and IFN-gamma released into the supernatants were quantified. LDH levels were determined with CytoTox 96 Non-Radioactive Cytotoxicity Assay Kits (PROMEGA) and cytotoxicity was calculated as percentage of the total lysis control after subtracting spontaneous LDH release of effector and target cells. All CDRa3 mutant TCR variants induced stronger killing of all analyzed tumor cell lines compared to the parental TCR, while no killing was observed for target-negative MCF-7 cells. For tumor cell lines presenting lower target copy numbers (A375, U2OS), only minimal cytotoxicity was found for the wildtype TCR, whereas especially the maturated variants 1-4 were able to induce substantial killing, which indicates that these variants will allow treatment of patient populations with low target copy numbers. In line with the cytotoxicity data, also enhanced IFN-gamma release was observed for T cells expressing the maturated TCR variants compared to the wildtype.

TABLE-US-00016 TABLE 3 Concentrations of half-maximal IFN-gamma release [nM] of parental R7P1D5 and CDRa3 mutant variants 1-5. The parental TCR R7P1D5 is based on SEQ ID NOs 2 and 11 and the mutant variant TCRs 1-5 are based on SEQ ID NOs: 40 to 44, respectively, and SEQ ID NO: 11. Variant EC.sub.50 [nM] R7P1D5 9.5 R7P1D5 CDRa3 mutant 4 1.4 R7P1D5 CDRa3 mutant 1 2.1 R7P1D5 CDRa3 mutant 2 1.4 R7P1D5 CDRa3 mutant 3 1.1 R7P1D5 CDRa3 mutant 5 3.7

Example 4: Production, Purification and Characterization of scTCR Variants

[0598] TCRs consisting of Valpha and Vbeta domains were designed, produced and tested in a single-chain (scTCR) format in conjunction with a Fab fragment derived from humanized CD3-specific T cell-recruiting antibody UCHT1. Therefore, plasmids containing either the light chains of humanized UCHT1 antibodies (SEQ ID No: 123 for variant S and SEQ ID No: 80 for other variants except S) or the respective scTCR sequences coupled to the C-terminus of V.sub.H-C.sub.H1 of the humanized UCHT1 antibody, controlled by HCMV-derived promoter elements, were generated.

TABLE-US-00017 TABLE 4 Combinations of sequences for the generation of scTCR variants expressed in conjunction with a Fab fragment. For transfections plasmids containing the light chain sequence were mixed with plasmids containing respective Fab heavy chain sequence. Light Chain Heavy chain Variant (SEQ ID No:) (SEQ ID No:) S 123 124 I 80 126 #1 80 58 #2 80 76 #19 80 125 #20 80 60 #21 80 61 #29 80 67 #30 80 68 #31 80 69 #32 80 75

[0599] Plasmid DNA was amplified in E. coli according to standard culture methods and subsequently purified using commercial-available kits (Macherey & Nagel). Purified plasmid DNA was used for transient transfection of CHO-S cells according to instructions of the manufacturer (ExpiCHO system; Thermo Fisher Scientific). Transfected CHO-cells were cultured 10-12 days at 32 C. to 37 C. and received one feeds of ExpiCHO Feed.

[0600] Conditioned cell supernatant was cleared by filtration (0.22 m) utilizing Sartoclear Dynamics Lab Filter Aid (Sartorius). Bispecific antigen binding proteins were purified using an Akta Pure 25 L FPLC system (GE Lifesciences) equipped to perform affinity and size-exclusion chromatography in line. Affinity chromatography was performed on protein L columns (GE Lifesciences) following standard affinity chromatographic protocols. Size exclusion chromatography was performed directly after acidic elution (pH 2.8) from the affinity column to obtain highly pure monomeric protein using, Superdex 200 pg 16/600 columns (GE Lifesciences) following standard protocols. Protein concentrations were determined on a NanoDrop system (Thermo Scientific) using calculated extinction coefficients according to predicted protein sequences. Concentration was adjusted, if needed, by using Vivaspin devices (Sartorius). Finally, purified molecules were stored in phosphate-buffered saline (DPBS, pH 7.2) at concentrations of about 1 mg/mL at temperatures of 2-8 C. Productivity of each molecule was assessed by calculating the yield as [milligram protein purified/liter cell supernatant].

[0601] Quality of purified scTCR-Fab antigen binding proteins was determined by HPLC-SEC on MabPac SEC-1 columns (5 m, 7.8300 mm) running in 50 mM sodium-phosphate pH 6.8 containing 300 mM NaCl within a Vanquish uHPLC-System. Detector wavelength was set to 214 nm. Using the same methodology heat-stressed samples of the respective molecules (after storage for 7 or 14 days at 40 C. in DPBS at 1 mg/mL) were analyzed. Induced aggregates were calculated as [% aggregates (after stress)][% aggregates (start)]. Monomer recovery was calculated as [monomer peak area (after stress)]/[monomer peak area (start)]100%. As shown in FIG. 7A, pronounced differences in production yield (left panel) and heat stress stability were observed among the different scTCR variants. The stabilized scTCR variant S (SEQ ID No: 124) containing wild-type CDR-sequences combined with two framework mutations (aS19A, aF57Y) could only be expressed at low levels. Increased expression levels and increased heat-stress stability could be achieved after the incorporation of affinity-maturated CDRa2, CDRa3 and CDRb1 sequences (variant #1) and affinity-maturated CDRa2, CDRa3 and CDRb1 sequences together with an affinity-maturated CDRb2 sequence in conjunction with the adjacent framework mutation b166C (variant I). Using both CDR-maturated scTCR variants, I and #1, further modifications of the framework regions were tested to improve expression yield and heat-stress stability. Incorporation of 3 framework mutations (aL50P, bM46P and bM47G) into variant #1 generated variant #30 with increased production yield, which could be further improved by replacing the framework mutation aS19A to aS19V (variant #32). The impact of the aS19A or aS19V mutation is highlighted by variant #29 representing variant #32 with a aS19 backmutation, which resulted in strongly reduced productivity and elevated aggregation levels. Notably, the latter negative effect of the aS19 backmutation on production yield could be reversed by the introduction of the framework mutation aA48K (variant #31). In the context of the CDR-maturated scTCR variant 1, an additional framework mutation b154F (variant #19) and a bN109D mutation in the beta chain CDR3 (variant #20) was tested alone or in combination (variant #21). The mutation bN109D in CDRb3 converted a NT-motif, which is potentially prone to deamidiation of the Asn-residue, into a DT motif, which could be of advantage in terms of design approaches to remove sequence liabilities within the CDR sequences. Introduction of a b154F framework mutation into variant 1 (variant #19) resulted in improved yield and improved heat-stress stability of the scTCR. The additional introduction of the CDRb3 mutation bN109D into variant #19 generated the scTCR variant #21 with further strongly improved heat stress-stability albeit the bN109D mutation alone (variant #20) showed only slightly improved productivity and similar stability when compared to variant I.

[0602] The scTCR-Fab antigen binding proteins were analyzed for their binding affinity towards the MAGE-A antigenic peptide of SEQ ID NO: 1 in complex with HLA-A*02 via biolayer interferometry. Measurements were performed on an Octet RED384 system using settings recommended by the manufacturer. Briefly, binding kinetics were measured at 30 C. and 1000 rpm shake speed using PBS, 0.05% Tween-20, 0.1% BSA as buffer. Bispecific molecules were loaded onto biosensors (FAB2G) prior to analyzing serial dilutions of the peptide-HLA-A*02 complex. The bispecific antigen binding protein comprising a stabilized scTCR variant S (SEQ ID NOs: 123 and 124) showed only very weak binding signals up to a concentration of 500 nM (FIG. 7B, Table 5). Introducing affinity-maturated CDRa2, CDRa3 and CDRb1 sequences (variant #1 and variant #2) strongly improved binding towards target peptide-MHC. K.sub.D values of 2-3 nM were determined for variants #1 (SEQ ID NOs: 80, 58) and #2 (SEQ ID NOs: 80, 76). The additional inclusion of a maturated CDRb2 sequence together with the framework mutation b166C into the scTCR variant #2 led to a further improvement in binding affinity with a measured K.sub.D of 150 pM for variant I (SEQ ID NOs: 80, 126).

TABLE-US-00018 TABLE5 Frameworkmutations,CDRcombinationsandKIDvaluesofvariantsS(SEQIDNOs: 123,124),#1,#2andI(SEQIDNOs:80combinedwith58,76or126, respectively).K.sub.Dvaluesweremeasuredbybiolayerinterferometry. TCR variant CDRa1 CDRa2 CDRa3 CDRb1 CDRb2 CDRb3 K.sub.D[M] S DSSSTY IYSNMDM CAEYSSASKIIF SGHDY FNNNVP CASRANTGELFF aS19A SEQID SEQID SEQIDNO7 SEQID SEQID SEQIDNO14 NO5 NO21 NO12 NO13 #1 DSSSTY IYSSQDQ CAEMTSESKIIF PGHDY FNNNVP CASRANTGELFF 2.35E-09 aS19A SEQID SEQID SEQIDNO35 SEQID SEQID SEQIDNO14 NO5 NO56 NO62 NO13 #2 DSSSTY IYSSQDS CAEMTSESKIIF PGHDY FNNNVP CASRANTGELFF 2.69E-09 aS19A SEQID SEQID SEQIDNO35 SEQID SEQID SEQIDNO14 NO5 NO59 NO62 NO13 I DSSSTY IYSSQDS CAEMTSESKIIF PGHDY FCYGTP CASRANTGELFF 1.53E-10 aS19A SEQID SEQID SEQIDNO35 SEQID SEQID SEQIDNO14 bl66C NO5 NO59 NO62 NO65 *no relevant binding signals up to 500 nM

[0603] The scTCR-Fab antigen binding proteins were further characterized with respect to their specificity by analyzing binding to potential off-target peptides in complex with HLA-A*02. Potential off-target peptides were selected from a database of normal tissue-presented HLA-A*02 bound peptides (XPRESIDENT database) based on sequence similarity as determined by similarity BLAST search. Binding was analyzed via biolayer interferometry essentially as described above. Binding signals of all interactions were determined at a concentration of 1 pM bispecific scTCR-Fab using HIS1K biosensors to load peptide-HLA-A*02 complexes. Potential off-target peptides were analyzed with respect to their binding response at the end of the association phase. In comparison to the scTCR variant I (SEQ ID NO: 80, 126), the variant #19 (SEQ ID NO: 80, 125) comprising the additional framework mutation b154F showed an improved binding specificity (FIG. 7C, Table 6) besides the improved production yield and stability as shown above. While binding curves for MAG-003 (SEQ ID NO: 1) were similar for both variants (FIG. 7C), variant #19 showed abrogated binding to the off-target peptides SYNE3-001 (SEQ ID NO: 27), TPX2-001 (SEQ ID NO: 26) and PSME2-001 (SEQ ID NO: 30) as shown by lack of binding signals (Table 6).

TABLE-US-00019 TABLE 6 Specificity of scTCR-Fab antigen binding proteins with scTCR variant I (SEQ ID NO: 80, 126) and variant #19 (SEQ ID NO: 80, 125). Binding kinetics of 1 M bispecific was measured via biolayer interferometry. Responses at the end of the association phase (in nm or as % of MAG-003 signal) are shown. I #19 Response (% Response (% Loading of MAG- of MAG- Sample ID Response 003) Response 003) MAG-003 0.738 100.0 0.719 100.0 SYNE3-001 0.226 30.6 0.006 0.9 TPX2-001 0.136 18.4 0.003 0.3 PSME2-001 0.074 10.0 0.010 1.4

Example 5: Production, Purification and Characterization of Bispecific TCER Molecules

[0604] FIG. 15 shows a bispecific TCER molecule in accordance with one embodiment of the present disclosure. Bispecific T cell-engaging receptors (TCER) are fusion proteins containing the variable domains (V, V) of an affinity-maturated TCR and the variable domains (VH, VL) of a humanized T cell-recruiting antibody. The molecules use an effector function-silenced Fc part, which confers extended half-life and also improved stability/manufacturability characteristics.

[0605] Bispecific TCER molecules were designed based on selected TCR alpha and beta chain variable domains with preferred CDR and framework mutations, as described in example 4. The respective TCR variable domains of scTCR variant #21 were combined with the beneficial framework mutations aS19V and aA48K resulting in TCR variant #114-iso0 (SEQ ID NO: 138 and SEQ ID No: 150). To further optimize the variable domain sequence of TCR #114-iso0, two potential isomerization sites (DS-motives) located within the CDRa1 (aD27E) and CDRa2 (aS65Q), respectively, were substituted resulting in TCR variants #114-iso1 (SEQ ID NO: 151 and SEQ ID No: 150) and 114-iso2 (SEQ ID NO: 152 and SEQ ID NO: 150) having one or none, respectively, post-translational modification consensus sequences within the CDRs. The above described variable domains of TCR variants #114-iso0, #114-iso1 and #114-iso2 were combined with VL- and VH-domains derived from variants derived from two newly humanized T cell-recruiting antibodies BMA031(36) (SEQ ID NO: 159 and 160), UCHT1(17) (SEQ ID NO: 153 and 154) and different UCHT1(17) variants UCHT1(17opt) (SEQ ID No: 153 and 162), UCHT1(21) (SEQ ID No: 153 and 164) and UCHT1(23) (SEQ ID No: 153 and 156), respectively.

[0606] Using a TCR/antibody diabody-F.sub.c format we designed TCER molecules (exemplary full-lengths sequences shown in Table 7) and respective vectors with mono-cistronic, controlled by HCMV-derived promoter elements, pUC19-derivatives. Plasmid DNA was amplified in E. coli according to standard culture methods and subsequently purified using commercial-available kits (Macherey & Nagel). Purified plasmid DNA was used for transient transfection of CHO-S cells utilizing an electroporation systems (MaxCyte STX). Transfected CHO-cells were cultured 10-12 days at 32 C. to 37 C. and received one to three feeds of Cellboost 7a and 7b (GE Healthcare) solution.

TABLE-US-00020 TABLE 7 Combination of TCR variable domain sequences and T cell-recruiting antibody variable domain sequences for the generation of the respective variants of TCER molecules. For transfections plasmids containing chain A and plasmids containing chain B were mixed. Chain A SEQ Chain B SEQ TCER molecule ID No: ID No: 114-iso0-UCHT1(17) 127 130 114-iso1-UCHT1(17) 131 130 114-iso2-UCHT1(17) 133 130 114-iso0-BMA(36) 135 136 114-iso1-BMA(36) 137 136 114-iso2-BMA(36) 139 136

[0607] Conditioned cell supernatant was cleared by filtration (0.22 m) utilizing Sartoclear Dynamics Lab Filter Aid (Sartorius). Bispecific antigen binding proteins were purified using an Akta Pure 25 L FPLC system (GE Lifesciences) equipped to perform affinity and size-exclusion chromatography in line. Affinity chromatography was performed on MAbSelect SuRE or protein L columns (GE Lifesciences) following standard affinity chromatographic protocols. Size exclusion chromatography was performed directly after elution (pH 2.8) from the affinity column to obtain highly pure monomeric protein using, Superdex 200 pg 26/600 columns (GE Lifesciences) following standard protocols. Protein concentrations were determined on a NanoDrop system (Thermo Scientific) using calculated extinction coefficients according to predicted protein sequences. Concentration was adjusted, if needed, by using Vivaspin devices (Sartorius). Finally, purified molecules were stored in phosphate-buffered saline at concentrations of about 1 mg/mL at temperatures of 2-8 C.

[0608] Quality of purified bispecific antigen binding proteins was determined by HPLC-SEC on MabPac SEC-1 columns (5 m, 4300 mm) running in 50 mM sodium-phosphate pH 6.8 containing 300 mM NaCl within a Vanquish uHPLC-System. Detector wavelength was set to 214 nm. Induced aggregates were calculated as [% aggregates (after stress)][% aggregates (start)]. Monomer recovery was calculated as [monomer peak area (after stress)]/[monomer peak area (start)]100%.

[0609] Productivity and stress stability data of different TCER molecules are shown in FIG. 8A. All variants, except for variant 114-iso1-UCHT1(17opt), showed a production yield of more than 10 mg/I (FIG. 8, left panel). Heat-stress testing of the molecules at 40 C. revealed very good to acceptable stabilities for all produced TCER molecules showing less than 10% aggregate formation upon 14 days of heat-stress (FIG. 8A, right panel).

[0610] Using biolayer interferometry, bispecific TCER antigen binding proteins comprising TCR variants 114-iso0, 114-iso1 and 114-iso2 in combination with UCHT1(17) and BMA031(36), respectively (as shown in Table 7), were characterized for their binding affinity towards the MAGE-A antigenic peptide (SEQ ID NO: 1) in complex with HLA-A*02 (FIG. 8A, Table 8). Measurements were performed on an Octet RED384 system using settings recommended by the manufacturer. Briefly, binding kinetics were measured at 30 C. and 1000 rpm shake speed using PBS, 0.05% Tween-20, 0.1% BSA as buffer. Peptide-HLA-A*02 complexes were loaded onto biosensors (HIS1K) prior to analyzing serial dilutions of the bispecific TCER molecules. TCER variants 114-iso0, 114-iso1 and 114-iso2 bound strongly to MAG-003 in complex with HLA-A*02 with comparable K.sub.D values of 1.8 nM, 2.0 nM and 2.3 nM, respectively, indicating that HLA-A*02/MAG-003 binding is not affected by the removal of the two isomerization sites in the CDRa1 (aD27E) and CDRa2 (aS65Q) of the TCR variable alpha domain.

TABLE-US-00021 TABLE 8 Affinity analysis of TCER variants 114-iso0, 114-iso1 and 114-iso2 in combination with UCHT1(17) and BMA031(36) as shown in Table 7. K.sub.D values were measured by biolayer interferometry. UCHT1(17) BMA031(36) 114-iso0 114-iso1 114-iso2 114-iso0 114-iso1 114-iso2 K.sub.D(MAG-003) [M] 1.79E09 1.97E09 2.32E09 1.79E09 1.97E09 2.42E09

[0611] TCR variants 114-iso0, 114-iso1 and 114-iso2 were furthermore characterized with respect to their specificity by analyzing binding to normal tissue-derived off-target peptides using biolayer interferometry. Measurements were performed as described above. Binding affinities of 114-iso0-UCHT1(17) for MAG-003 (SEQ ID NO: 1) and off-target peptides CNOT1-003 (SEQ ID NO: 32), COL6A3-010 (SEQ ID NO: 179), FAM115A-001 (SEQ ID NO: 180), HEATR5A-001 (SEQ ID NO: 31), HERC4-001 (SEQ ID NO: 29), INTS4-002 (SEQ ID NO: 171), PHTF2-001 (SEQ ID NO: 181), PPP1CA-006 (SEQ ID NO: 173), RPL-007 (SEQ ID NO: 174), RPP1-001 (SEQ ID NO: 182), SAMH-001 (SEQ ID NO: 172), SETD1A-001 (SEQ ID NO: 175) in complex with HLA-A*02 were determined and affinity windows were calculated. Affinity windows ranged from more than 200-fold to no off-target binding at all (Table 9). Specificities of 114-iso1-BMA(36) and 114-iso2-UCHT1(17) were analyzed by measuring binding signals for MAG-003 (SEQ ID NO: 1) and off-target peptides ODC-001 (SEQ ID NO: 178), NOMAP-1-0320 (SEQ ID NO: 176) and NOMAP-1-1223 (SEQ ID NO: 177) in complex with HLA-A*02 at a concentration of 1 pM TCER (FIG. 8C, Table 10). Binding responses of 1% of the MAG-003 signal or lower were detected demonstrating high specificity of the TCR variants for MAG-003.

TABLE-US-00022 TABLE 9 Specificity of TCER molecule 114-iso0-UCHT1(17). K.sub.D values were measured by biolayer interferometry analyzing a serial dilution of 114-iso0-UCHT1(17) (500 nM, 250 nM, 125 nM, 62.5 nM). Affinity windows were calculated as K.sub.D(similar peptide)/K.sub.D(MAG-003). K.sub.D(similar peptide)/ K.sub.D [M] K.sub.D(MAG-003) MAG-003 1.56E09 CNOT1-003 1.64E06 1050 COL6A3-010 9.03E07 579 FAM115A-001 7.84E07 503 HEATR5A-001 3.70E07 238 HERC4-001 4.91E07 315 INTS4-002 1.98E06 1273 PHTF2-001 1.02E06 656 PPP1CA-006 no binding no binding RPL-007 no binding no binding RPP1-001 8.97E07 576 SAMH-001 2.11E06 1356 SETD1A-001 no binding no binding

TABLE-US-00023 TABLE 10 Specificity of TCER molecules 114-iso1-BMA(36) and 114-iso2-UCHT1(17). Binding kinetics of 1 M TCER was measured via biolayer interferometry. Responses at the end of the association phase (in nm or as % of MAG-003 signal) are shown. 114-iso1-BMA(36) 114-iso2-UCHT1(17) Loading Response Response Sample (% of (% of ID Response MAG-003) Response MAG-003) MAG-003 0.8246 100.0 0.8233 100.0 ODC-001 0.0061 0.7 0.0104 1.3 NOMAP-1-0320 0.0025 0.3 0.0025 0.3 NOMAP-1-1223 0.0005 0.1 0.0057 0.7

Example 6: TCER-Mediated Killing of MAG-003-Positive and Tumor Cell Lines

[0612] Maturated TCR variants were expressed as soluble TCER molecules #114-iso0-UCHT1(17), #114-iso1-UCHT1(17), #114-iso2-UCHT1(17), #114-iso0-BMA031(36), #114-iso1-BMA031(36) and #114-iso2-BMA031(36) employing a TCR/antiCD3 diabody-F.sub.c format (see also Table 7). The cytotoxic activity of the bispecific TCER molecules against MAG-003-positive and MAG-003-negative tumor cell lines, respectively, was analyzed by LDH-release assay. Therefore, tumor cell lines presenting different amounts of HLA-A*02/MAG-003 molecules on the cell surface were co-incubated with PBMC from healthy HLA-A*02+ donors in presence of increasing concentrations of TCER molecules. After 48 hours, lysis of target cell lines was measured utilizing CytoTox 96 Non-Radioactive Cytotoxicity Assay Kits (PROMEGA). As shown in FIG. 9 and FIG. 10, TCER molecules exhibited efficient lysis of MAG-003-positive tumor cell lines Hs695T, A375 and U2OS displaying 1100, 230 and 120 copies per tumor cell, respectively, as determined by mass-spectrometry-based quantification of HLA-A*02/MAG-003 complexes. For killing assays, PBMC from two healthy donors (HBC-720 in FIG. 9 and HBC-982 in FIG. 10) were used. Within the same experiments the cell line BV-173 expressing HLA-A*02 but not presenting the peptide MAG-003 at detectable levels, no or only marginal lysis of cells was induced by the bispecific TCER molecules indicating the specificity of the TCR domains.

Example 7: In Vivo Efficacy in Tumor Xenograft Bearing NOG Mice

[0613] A pharmacodynamic study was performed in the hyper immune-deficient NOG mouse strain to test the ability of TCER molecules in recruiting the activity of human immune T cells by specific binding to a T cell antigen and by specific binding to a human tumor-specific HLA-peptide complex on human cancer cells. The NOG mouse strain hosting the subcutaneously injected human tumor cell line HS695T were intravenously injected with human peripheral blood mononuclear (PBMC) cell xenografts. The human PBMC (110.sup.7 cells/mouse) were transplanted within 24 hours when individual Hs695T tumor volume reached 50 mm.sup.3 as calculated after caliper measurements. Treatment was initiated within one hour after transplantation of human blood cells. Eight to ten female mice per group (randomized according to tumor size) received intravenous bolus injections (5 mL/kg body weight, twice weekly dosing, six doses) into the tail vein. The injected dose of TCER molecules was 0.5 (group 2) and 0.05 (group 3 and 4) mg/kg body weight x injection, PBS was used in the vehicle control group (group 1). As shown in FIG. 11 for individual mice, the treatment with TCER molecules induced pronounced anti-tumor responses. Treatment with a TCER (SEQ ID NO: 169 and 170) targeting PRAME-004 (SEQ ID NO: 168, group 2) strongly inhibited tumor growth as indicated by reduced increase of mean tumor volume from basal levels (start of randomization) of 90 mm.sup.3 to 209 mm.sup.3 at day 25 in comparison to the increase observed in the vehicle control group 1 from basal levels of 74 mm.sup.3 to 1081 mm.sup.3 at day 25. Treatment with TCER targeting MAG-003 (SEQ ID NO: 127 and 130, group 3; SEQ ID NO: 135 and 136, group 4) induced complete remission of the Hs695T tumor in all mice at Day 18 until the end of the study at Day 25 compared to basal levels of 86 mm.sup.3 (group 3) and 80 mm.sup.3 (group 4) at the day of randomization.

Example 8: In-Vitro Safety on Primary Healthy Cells

[0614] The safety profile of TCER molecule 114-iso1-BMA(36) was assessed in LDH-killing experiments with human normal tissue primary cells of different organs. FIG. 12 shows the results of co-cultures of 11 different primary healthy tissue cells (HLA-A*02+) with PBMC effector cells from healthy HLA-A*02+ donors at a ratio of 1:10 in presence of increasing TCER concentrations. The cells were co-incubated in a 1:1 mixture of normal tissue primary cell-specific medium and T cell medium. To determine a safety window, the TCER molecules were co-incubated in an identical setup with the MAG-003-positive tumor cell line Hs695T in the respective medium combination of the normal tissue primary cells. After 48h of co-culture, supernatants were harvested and tumor cell lysis was analyzed by measuring LDH-release using the LDH-Glo Kit (Promega).

[0615] In FIG. 12, each cytotoxicity plot shows LDH-release of a normal tissue primary cell type (empty circles) in comparison to the control tumor cell line Hs695T (filled circles) in the same medium combination after co-incubation with PBMCs and increasing concentrations of the TCER molecule. In all cases the response against normal tissue primary cells was too low to calculate an EC.sub.50. Instead, we defined the safety window based on the lowest observed effect level (LOEL) determined as the first TCER concentration with a response over cut-off value. The cut-off was defined as ([standard deviation from all triplicates3]+control normal tissue sample w/o TCER) and is indicated as dotted line in each cytotoxicity plot. For each normal tissue cell type, the LOEL-based safety window was determined to be larger than 1,000-fold. The safety profile of TCER molecule 114-iso1-BMA(36) was assessed in second LDH-killing experiment with 12 different primary healthy tissue cells (HLA-A*02+) of which some were overlapping with the first experiment. Experimental conditions were used as described above. In FIG. 16, each cytotoxicity plot shows LDH-release of a normal tissue primary cell type (empty circles) in comparison to the control tumor cell line Hs695T (filled circles) in the same medium combination after co-incubation with PBMCs and increasing concentrations of the TCER molecule. In all cases the response against normal tissue primary cells was too low to calculate an EC.sub.50. Instead, we defined the safety window based on the lowest observed effect level (LOEL) determined as the first TCER concentration with a response over cut-off value. The cut-off was defined as ([standard deviation from all triplicates3]+control normal tissue sample w/o TCER) and is indicated as dotted line in each cytotoxicity plot. For each normal tissue cell type, the LOEL-based safety window was determined to be larger than 1,000-fold.

Example 9: In Vivo Efficacy with a Once Weekly Treatment in Tumor Xenograft Bearing NOG Mice

[0616] A pharmacodynamic study was performed in the hyper immune-deficient NOG mouse strain to test the ability of TCER molecules in recruiting the activity of human immune T cells by specific binding to a T cell antigen and by specific binding to a human tumor-specific HLA-peptide complex on human cancer cells. The NOG mouse strain hosting the subcutaneously injected human tumor cell line Hs695T were intravenously injected with human peripheral blood mononuclear (PBMC) cell xenografts. The human PBMC (110.sup.7 cells/mouse) were transplanted within 24 hours when individual Hs695T tumor volume reached 50 mm.sup.3 as calculated after caliper measurements. Treatment was initiated within one hour after transplantation of human blood cells. Six to ten female mice per group (randomized according to tumor size) received intravenous bolus injections (5 mL/kg body weight, once weekly dosing, three doses) into the tail vein. The injected dose of TCER molecules was 0.01 mg/kg body weight x injection (group 2 and 3), PBS was used in the vehicle control group (group 1). As shown in FIG. 13, the treatment with TCER molecules induced pronounced anti-tumor responses. Treatment with MAG-003 TCER molecules 114-iso1-BMA(36) (SEQ ID NO: 137 and 136, group 2) and 114-iso1-UCHT1(V17opt) (SEQ ID NO: 131 and 167, group 3) induced remission of the Hs695T tumor in all mice at day 21. Tumor growth was strongly inhibited as indicated by reduced increase of mean tumor volume from basal levels (start of randomization) of 84 mm.sup.3 to 8 mm.sup.3 (group 2) and of 83 mm.sup.3 to 14 mm.sup.3 (group 3) at day 21 in comparison to the increase observed within the vehicle control (group 1) from basal levels of 85 mm.sup.3 to 1024 mm.sup.3 at day 21. Complete remission was observed in 2 mice of group 2 and 2 mice of group 3.

Example 10: In Vivo Pharmacokinetics in NOG Mice

[0617] Pharmacokinetic properties of TCER 114-iso1-UCHT1(17) (SEQ ID NOs: 130 and 131) were analyzed in hyper immune-deficient NOG mice receiving a single intravenous injection of 2 mg/kg body weight. Plasma samples were collected from the retro-bulbar plexus at time points 0 h, 0.1 h, 2 h, 8 h, 24 h, 48 h, 120 h, 240 h and 360 h. Plasma levels of 114-iso1-UCHT1(17) were determined by ELISA using two different assay set-ups (F.sub.cV.sub.L assay and CD3pMHC assay). Using the F.sub.cV.sub.L assay set-up, TCER plasma level were quantified by detecting the respective protein subunits. Briefly, goat anti-human IgG F.sub.c was coated at a concentration of 2 g/ml in PBS at 4 C. overnight. Remaining binding sites were blocked using PBS containing 2% BSA. Samples diluted in blocking buffer were incubated for 1 h at room temperature prior to detection with Protein L-HRP and TMB substrate solution. Using the CD3pMHC assay, plasma levels of TCER 114-iso1-UCHT1(17) were determined via its bispecific binding activity measuring only molecules capable of simultaneously binding to both target molecules. CD3F.sub.c was generated by fusing the extracellular domains of human CD3 and CD3 to the N-terminus of Fc-domains as utilized within the TCER-constructs (containing Knob-into-hole mutations and an additional C-terminal His-Tag). CD3F.sub.c molecules were expressed in ExpiCHO cells and purified using Protein A affinity chromatography followed by size exclusion chromatography as described above. CD3F.sub.c was coated at a concentration of 1 g/ml in PBS at 4 C. overnight. Blocking and sample incubation were performed as described above followed by a two-step detection using HLA-A*02/MAG-003 and anti-b2m-HRP. Plasma concentrations of the samples were obtained by interpolation from the respective standard curves. Both assay set-ups lead to the determination of similar TCER plasma concentrations showing that protein integrity as well as binding activity of the molecule are retained in plasma (FIG. 14). Linear regression of time points 24 h, 48 h, 120 h, 240 h and 360 h revealed a terminal plasma half-life of 150 h. The results show that the half-life (T.sub.1/2) of 114-iso1-UCHT1(17) is about 6-7 days.

Example 11: Detection of MAG-003 Peptide on Primary Tissues by Mass Spectrometry

[0618] HLA molecules from shock-frozen tissue samples were purified, and HLA-associated peptides were isolated. The isolated peptides were separated, and sequences were identified by online nano-electrospray-ionization (nanoESI) liquid chromatography-mass spectrometry (LC-MS) experiments. MAG-003 identified on multiple tissue samples was quantified using ion-counting of label-free LC-MS data. The method assumes that LC-MS signal areas of a peptide correlate with its abundance in the sample. All quantitative signals of a peptide in various LC-MS experiments were normalized based on central tendency, averaged per sample and merged into a plot called presentation profile. The presentation profile consolidates different analysis methods like protein database search, spectral clustering, charge state deconvolution (decharging) and retention time alignment. All automatically derived quantitative and qualitative data were manually inspected (see FIG. 18).