RECRUITING AGENT FURTHER BINDING AN MHC MOLECULE

Abstract

The present invention concerns bispecific antigen binding proteins directed against MHC presented target antigens (TA). The invention in particular provides bispecific antigen binding proteins comprising at least two antigen binding sites (A and B), wherein the antigen binding site A binds to CD3 and the antigen binding site B binds to a target antigenic (TA) peptide/MHC complex. The bispecific antigen binding proteins of the invention comprise, in particular, the CDRs of the VL and VH domains of novel engineered anti-CD3 antibodies having a reduced affinity. The bispecific antigen binding proteins of the invention are of use for the diagnosis, treatment and prevention of TA associated diseases, such as tumor-associated antigen (TAA) expressing cancerous diseases. Further provided are nucleic acids encoding the bispecific antigen binding protein of the invention, vectors comprising these nucleic acids, recombinant cells expressing the antigen binding protein and pharmaceutical compositions comprising the bispecific antigen binding proteins of the invention.

Claims

1. A bispecific antigen binding protein comprising at least two antigen binding sites (A and B), wherein the antigen binding site A binds to CD3 and wherein the antigen binding site B binds to a target antigenic (TA) peptide/MHC complex, and wherein the antigen binding site A comprises a heavy chain variable domain (V.sub.H) and a light chain variable domain (V.sub.L), and a) wherein said V.sub.L comprises three complementary determining regions (CDRs) CDRL1, CDRL2 and CDRL3, wherein CDRL1 comprises or consists of the amino acid sequence RASQDIRNYLN of SEQ ID NO: 1, CDRL2 comprises or consists of the amino acid sequence YTSRLHS of SEQ ID NO: 2, and CDRL3 comprises or consists of the amino acid sequence QQGQTLPWT of SEQ ID NO: 3, and b) wherein said V.sub.H comprises three complementary determining regions (CDRs) CDRH1, CDRH2 and CDRH3, wherein CDRH1 comprises or consists of the amino acid sequence X.sub.1YTMN of SEQ ID NO: 4, wherein X.sub.1 is G or E, optionally G, CDRH2 comprises or consists of the amino acid sequence of LINPX.sub.2X.sub.3GVX.sub.4TYAQKX.sub.5QX.sub.6 SEQ ID NO: 5, wherein X.sub.2 is Q, Y or E, X.sub.3 is R, K or E, X.sub.4 is S or T, X.sub.5 is F or V, and X.sub.6 is G or D, and CDRH3 comprises or consists of the amino acid sequence SGYYGX.sub.7SWYFDV of SEQ ID NO: 6, wherein X.sub.7 is E or D.

2. The bispecific antigen binding protein according to claim 1, wherein the antigen binding site A binds to CD3 with a K.sub.D(A) and the antigen binding site B binds to the target antigenic peptide C (TA-C)/MHC complex with a K.sub.D(C) and wherein the ratio of K.sub.D(A)/K.sub.D(C) is more than 1, more than 4, more than 6, more than 8, more than 10, more than 15, more than 20, more than 25, more than 30, more than 40, more than 50, such as between 1 and 150, between 4 and 140, between 6 and 100, between 8 and 100, between 10 and 100, optionally between 10 and 100.

3. The bispecific antigen binding protein according to claim 1, wherein the antigen binding site A binds to CD3 with a K.sub.D(A) that is 3 nM, 5 nM, 8 nM, 10 nM, 12 nM, 14 nM, 16 nM, 18 nM, 20 nM, 25 nM, 30 nM, 35 nM, 40 nM, 45 nM, and 1000 nM, 800 nM, 600 nM, 500 nM, 400 nM, such as between 3 nM and 1000 nM, 3 nM and 600 nM, between 5 nM and 600 nM, 10 nM and 600 nM, 12 nM and 600 nM, 14 nM and 600 nM, 16 nM and 600 nM, 18 nM and 600 nM, 20 nM and 600 nM, optionally 5 nM and 100 nM, as determined using Surface plasmon resonance (SPR) or Biolayer Interferometry (BLI), optionally Biolayer Interferometry (BLI).

4. The bispecific antigen binding protein according to claim 1, wherein the antigen binding site B binds to the target antigenic peptide C (TA-C)/MHC complex with a K.sub.D(C) which is 100 M, 1 M100 nM, 50 nM, 10 nM, for instance 0.01 nM to 150 nM, 0.05 nM to 150 nM, 0.1 nM to 150 nM, 0.1 nM to 100 nM, 0.1 nM to 50 nM, 0.1 nM to 10 nM, 0.5 nM to 10 nM, 0.5 nM to 5 nM, optionally 0.5 nM to 5 nM as determined using Surface plasmon resonance (SPR) or Biolayer Interferometry (BLI), optionally Biolayer Interferometry (BLI).

5. The bispecific antigen binding protein according to claim 1, wherein said antigen binding protein has an EC.sub.50 for TA-C/MHC presenting cells that is 100, 500, 1000 lower than the EC.sub.50 value for normal tissue cells.

6. The bispecific antigen binding protein according to claim 1, wherein the TA antigenic peptide C is a viral peptide, a bacterial peptide or a tumour associated antigen (TAA) peptide, optionally a tumour associated antigen (TAA) peptide.

7. The bispecific antigen binding protein according to claim 1, wherein said antigen binding protein has a EC.sub.50 for TA-C/MHC complex presenting cells that is 5 fold, 10 fold, 20 fold, 50 fold, 100 fold, 500 fold, 1000 fold lower than the EC.sub.50 value for normal tissue cells.

8. The bispecific antigen binding protein according to claim 1, wherein the TA antigenic peptide C is a tumour associated antigen (TAA) peptide C, and wherein said TAA-C is selected from the group of TAA antigenic peptides comprising or consisting of amino acid sequences of SEQ ID NO: 52 to 65, 67 to 96, 98, SEQ ID NO: 172 to 182, 184 to 268, SEQ ID NO: 9 and 10, such as the PRAME antigenic peptide comprising or consisting of the amino acid sequence SLLQHLIGL of SEQ ID NO: 9 or the MAGE-A antigenic peptide comprising or consisting of the amino acid sequence KVLEHVVNRV of SEQ ID NO: 10, wherein the MHC is optionally a HLA-A*02.

9. The bispecific antigen binding protein according to claim 8, wherein the TA antigenic peptide C is the MAGE-A antigenic peptide comprising or consisting of the amino acid sequence KVLEHVVRV of SEQ ID NO: 10 and wherein the similar peptide is 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, ZFC-001 optionally HEATR5A-001, HERC4-001 and CNOT1-003.

10. The bispecific antigen binding protein according to claim 1, wherein the bispecific antigen binding protein is a bispecific antibody or fragment thereof, a bispecific T cell receptor (TCR) or fragment thereof or a bispecific single chain TCR (scTCR) or a bispecific single-chain antibody.

11. The bispecific antigen binding protein according to claim 1, wherein said VL domain further comprises one or more framework regions selected from the group consisting of FR1-L, FR2-L, FR3-L and FR4-L, wherein FR1-L comprises or consists of the amino acid sequence of DIQMTQSPSSLSASVGDRVTITC of SEQ ID NO: 11 or an amino acid sequence at least 85% identical to SEQ ID NO: 11, FR2-L comprises or consists of the amino acid sequence of WYQQKPGKAPKLLIY of SEQ ID NO: 12 or WYQQKPGKAVKLLIY of SEQ ID NO: 13, optionally SEQ ID NO: 12 or an amino acid sequence at least 85% identical to SEQ ID NO: 12 or 13, FR3-L comprises or consists of the amino acid sequence of GVPSRFSGSGSGTDYTLTISSLQPEDIATYFC of SEQ ID NO: 14 or an amino acid sequence at least 85% identical to SEQ ID NO: 14, FR4-L comprises or consists of the amino acid sequence of FGQGTKVEIKR of SEQ ID NO: 15 or an amino acid sequence at least 85% identical to SEQ ID NO: 15, and wherein VH further comprises one or more framework regions selected from the group consisting of FR1-H, FR2-H, FR3-H and FR4-H, and wherein FR1-H comprises or consists of the amino acid sequence of EVQLVQSGAEVKKPGASVKVSCKASGYSFT of SEQ ID NO: 16 or an amino acid sequence at least 85% identical to SEQ ID NO: 16, FR2-H comprises or consists of the amino acid sequence of WVRQAPGQGLEWMG of SEQ ID NO: 17 or an amino acid sequence at least 85% identical to SEQ ID NO: 17 FR3-H comprises or consists of the amino acid sequence of RVTLTVDKSTSTAYMELSSLRSEDTAVYYCAR of SEQ ID NO: 18 or an amino acid sequence at least 85% identical to SEQ ID NO: 18, or FR4-H comprises or consists of the amino acid sequence of WGQGTLVTVSS of SEQ ID NO: 19 or an amino acid sequence at least 85% identical to SEQ ID NO: 19.

12. The bispecific antigen binding protein according to claim 1, wherein the antigen binding site B comprises an antibody or a fragment thereof or an alpha chain variable domain (v.sub.) and a beta chain variable domain (v.sub.) or a gamma chain variable domain (v.sub.) or a delta chain variable domain (v.sub.), optionally an alpha chain variable domain (v.sub.) and a beta chain variable domain (v.sub.) or a gamma chain variable domain (v.sub.) and a delta chain variable domain (v.sub.), optionally a v.sub. and v.sub..

13. The bispecific antigen binding protein according to claim 1, wherein i) the v.sub. comprises or consists of the amino acid sequence selected from the group consisting of TABLE-US-00011 SEQIDNO:20 EDVEQSLFLSVREGDSVVINCTYTDSSSTYLYWYKQEPGKGLQLLTYIY SSQDSKQDQRLTVLLNKKDKHLSLRIADTQTGDSAIYFCAEMTSESKIIF GSGTRLSIRP, SEQIDNO:21 EDVEQSLFLSVREGDSVVINCTYTDSSSTYLYWYKQEPGKGLQLLTYIY SSQDQKQDQRLTVLLNKKDKHLSLRIADTQTGDSAIYFCAEMTSESKIIF GSGTRLSIRP, SEQIDNO:22 EDVEQSLFLSVREGDSVVINCTYTESSSTYLYWYKQEPGKGLQLLTYIY SSQDQKQDQRLTVLLNKKDKHLSLRIADTQTGDSAIYFCAEMTSESKIIF GSGTRLSIRP or an amino acid sequence at least 85% identical to the amino acid sequence selected from the group consisting of SEQ ID NO: 20, 21 and 22 and wherein optionally the amino acid sequence at least 85% identical to the amino acid sequence of SEQ ID NO: 20 optionally comprises the amino acid sequence of CDRa1 of SEQ ID NO: 23, CDRa2 of SEQ ID NO: 24 and CDRa3 of SEQ ID NO: 25, and wherein optionally the amino acid sequence at least 85% identical to the amino acid sequence of SEQ ID NO: 21 optionally comprises the amino acid sequence of CDRa1 of SEQ ID NO: 23, CDRa2 of SEQ ID NO: 26 and CDRa3 of SEQ ID NO: 25, and wherein optionally the amino acid sequence at least 85% identical to the amino acid sequence of SEQ ID NO: 22 optionally comprises the amino acid sequence of CDRa1 of SEQ ID NO: 27, CDRa2 of SEQ ID NO: 26 and CDRa3 of SEQ ID NO: 25, and wherein the amino acid of said first variable domain optionally comprises the amino acids 19V and/or 48K, and the v.sub. comprises or consists of the amino acid sequence DAGVIQSPRHEVTEMGQEVTLRCKPIPGHDYLFWYRQTMMRGLELLFYFCYGTPCD DSGMPEDRFSAKMPNASFSTLKIQPSEPRDSAVYFCASRADTGELFFGEGSRLTVLSEQ ID NO: 30 or an amino acid sequence at least 85% identical to the amino acid sequence consisting of SEQ ID NO 30, and wherein optionally the amino acid sequence at least 85% identical to the amino acid sequence of SEQ ID NO: 30 optionally comprises the amino acid sequence of CDRb1 of SEQ ID NO: 31, CDRb2 of SEQ ID NO: 34 and CDRb3 of SEQ ID NO: 35, respectively, and optionally comprises the amino acid 54F and/or 66C, or (ii) the v.sub. or v.sub. comprises or consists of the amino acid sequence SEQ ID NO: 48 or an amino acid sequence at least 85% identical to the amino acid sequence of SEQ ID NO: 48, wherein optionally an amino acid sequence at least 85% identical to the amino acid sequence of SEQ ID NO: 48 comprises the amino acid sequence of CDRa1 of SEQ ID NO: 49, CDRa2 of SEQ ID NO: 50 and CDRa3 of SEQ ID NO: 51 and the v.sub. or v.sub. comprises or consists of the amino acid sequence of SEQ ID NO: 44 or an amino acid sequence at least 85% identical to SEQ ID NO: 44 wherein optionally said amino acid sequence at least 85% identical to the amino acid sequence of SEQ ID NO: 44 comprises the amino acid sequence of CDRb1 of SEQ ID NO: 45, CDRb2 of SEQ ID NO: 46 and CDRb3 of SEQ ID NO: 47.

14. The bispecific antigen binding protein according to claim 1, further comprising one or more of the following: (i) a diagnostic agent; (ii) a therapeutic agent; or (iii) a pharmacokinetics (PK) modifying moiety.

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

16. A recombinant host cell comprising the nucleic acid or the vector according to claim 15, wherein said host cell optionally is a) a stem cell, optionally a mesenchymal stem cell or b) a cell for recombinant expression, such as a Chinese Hamster Ovary (CHO) cell.

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

18. A method of producing the bispecific antigen binding protein as defined in claim 1, comprising a) providing a suitable host cell, b) providing a genetic construct comprising a coding sequence encoding the bispecific antigen binding protein, c) introducing said genetic construct into said suitable host cell, and d) expressing said genetic construct by said suitable host cell.

19. The method according to claim 18, further comprising the isolation and purification of the bispecific antigen binding protein from the suitable host cell.

20. A method of treating a patient who has a TAA/MHC positive cancer, comprising administering to the patient the bispecific antigen binding protein according to claim 1, wherein the TAA consists of the amino acid sequence selected from SEQ ID NO: 9, 10, 52 to 65, 67 to 96, 98, SEQ ID NO: 172 to 182, and 184 to 268, wherein the cancer is selected from the group consisting of non-small cell lung cancer, small cell lung cancer, liver cancer, head and neck cancer, skin cancer, renal cell cancer, brain cancer, gastric cancer, colorectal cancer, hepatocellular cancer, pancreatic cancer, prostate cancer, leukemia, breast cancer, Merkel cell carcinoma, melanoma, ovarian cancer, urinary bladder cancer, uterine cancer, gallbladder and bile duct cancer, and esophageal cancer.

21. The method of claim 20, wherein the TAA consists of the amino acid sequence of SEQ ID NO: 9 or 10.

Description

DESCRIPTION OF THE FIGURES

[0502] FIG. 1 shows the alignment of UCHT1 VL domain (SEQ ID NO: 36) to the identified human acceptor framework VK1-O18 (SEQ ID NO: 285) and to the J-segment JK1 (SEQ ID NO: 287). CDRs as identified in the figure are identified according to Cothia definitions.

[0503] FIG. 2 shows the alignment of UCHT1 VH domain (SEQ ID NO: 37) to the identified human acceptor framework VH-1-46 (SEQ ID NO: 286) and to the J-segment JH4 (SEQ ID NO: 288). CDRs as identified in the figure are identified according to Cothia definitions.

[0504] FIG. 3 shows concentration-dependent binding of PRAME-004-specific TCER-molecules to Jurkat cells as measured by flow cytometry.

[0505] FIGS. 4A and 4B shows the results of a representative LDH-release assays using PBMC of a healthy HLA-A*02-positive donor. MAG-003-specific TCER-molecules employing UCHT1(V17) or BMA031(V36) were tested on MAG-003-positive and MAG-003-negative as well as Off-target-positive tumor cell lines. Tested cell lines (from left to right): H695T, A375, T98G, BV173. Error bars depict standard deviations within triplicates.

[0506] FIGS. 5A and 5B shows the results of LDH-release assays on healthy cells incubated with MAG-003-specific TCER based on UCHT1(V17). Each cytotoxicity plot shows LDH-release of a primary healthy 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. Calculated safety windows based on EC.sub.50-values are also depicted.

[0507] FIGS. 6A and B shows the results of LDH-release assays on healthy cells incubated with MAG-003-specific TCER based on BMA031(V36). Each cytotoxicity plot shows LDH-release of a primary healthy 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. Calculated safety windows based on either EC.sub.50-values or based on LOEL are also depicted.

[0508] FIG. 7 shows the visualization of the introduced point mutations. Pane A shows the newly introduced Asp side chain in position 31 of the UCHT1 heavy chain and the negatively charged side chains of CD3 located in its spatial proximity. Pane B shows both, the wt Tyr and the replacing Gln in position 54 of the UCHT1 heavy chain. Replacement of the aromatic side chain with a smaller, polar amino acid removes the hydrophobic interaction with the apolar stem of the Asp in CD3's position 48. Pane C shows replacement of the same Tyr in position 54 of the heavy chain of UCHT1 by Glu and the negatively charged side chains on CD3 (48D, 49E, 50D, 51D) located in spatial proximity that are likely to cause electrostatic repulsion of the newly introduced Glu on UCHT1. Pane D shows the wt Lys in UCHT1 heavy chain position 55 forming a hydrogen bond (dashed line) with the backbone of a Ser in CD3's position 56 (left). Replacement of said Lys by Arg removes the polar interaction and introduces further bulk, forcing the Arg side chain into a strained rotamer (right). Pane E shows the same Lys in position 55 of UCHT1's heavy chain and its replacement by Glu; this mutation also removes the H-bond formed between Lys and the Ser in position 56 of CD3. Additionally, it introduces a negatively charged side chain in spatial proximity of a negatively charged patch on the surface of CD3 that is formed by Asp 48, Glu 49, Asp 50, and Asp 51.

[0509] FIG. 8 shows a summary of production yields and stability characteristics of TCER-molecules based on humanized UCHT1-variants. (na) not applicable, (nd) not done.

[0510] FIG. 9A shows binding curves of TCER antigen binding proteins comprising different UCHT1 variants for MAG-003 in complex with HLA-A*02 as measured by biolayer interferometry. Increasing concentrations of TCER molecules in solution were applied and are given in nM.

[0511] FIG. 9B shows binding curves of TCER antigen binding proteins comprising different UCHT1 variants for CD3-Fc as measured by biolayer interferometry. Increasing concentrations of TCER molecules in solution were applied and are given in nM.

[0512] FIGS. 10A and B shows the results of two independent LDH-release assays using PBMC of two healthy HLA-A*02-positive donors (HBC-982 and HBC-720). MAG-003-specific TCER-molecules employing UCHT1-variants with various affinities were tested on MAG-003-positive tumor cell lines. Error bars depict standard deviations within triplicates.

EXAMPLES

Example 1: Humanization of Mouse Monoclonal Ab UCHT1

[0513] Humanization of the mouse monoclonal antibody UCHT1 was performed by CDR-grafting according to published methods. Therefore, CDRs of VH and VL were identified according to the Cothia definitions. Sequence alignments comparing UCHT1 variable domains, VL (SEQ ID NO: 36) and VH (SEQ ID NO: 37), to the human germlines were generated. Based on overall sequence identity, matching interface positions and similarly classed CDR canonical positions, a germline was identified for each of the light and heavy chains as the most promising Acceptor frameworks, VK1-018 (SEQ ID NO: 285) for the light chain and VH-1-46 (SEQ ID NO: 286) for the heavy chain. The J-segment genes were compared to the Parental sequence over FR4 and J-segments JK1 (SEQ ID NO: 287) and JH4 (SEQ ID NO: 288) were selected for the light and heavy chain respectively.

[0514] A list of all the positions with differing residues between the Parental and Acceptors framework was generated. All positions were analysed and considered both in isolation and in the context of other potential substitutions. Each position was ranked as Neutral, Critical or Contributing and a suggestion about which residues to substitute and evaluate in humanised variants was made. Potential humanised variant sequences were screened using Epibase (Lonza). Each epitope or cluster of epitopes was analysed for substitutions that would either remove the epitope or further reduce the predicted immunogenicity. Further on potential sites of post-translational modifications within the CDRs were identified and respective remediations were proposed. In total this resulted in the generation of four different VH and five different VL-domains. Therefore, 17 humanization variants of UCHT1 were generated. All variants were expressed in CHO-cells as Fab-molecules. Expressed proteins were purified and ranked according to expression titer, aggregate levels and the EC.sub.50-values of binding to Jurkat cells. Based on these results the humanized UCHT1(V17), defined by SEQ ID No: 137 and SEQ ID No: 145, was selected for establishment of the inventors TCER-molecules.

[0515] PRAME-004 (SEQ ID No: 9) targeting TCER-molecules were constructed utilizing either the recruiting domains of humanized UCHT1(V17) resulting in molecules containing SEQ ID No: 171 and SEQ ID No. 170 or humanized BMA031(V10) resulting in SEQ ID No: 168 and SEQ ID No: 169, respectively. Vectors for the expression of recombinant proteins were designed as 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 according to instructions of the manufacturer (ExpiCHO system; Thermo Fisher Scientific). Transfected CHO-cells were cultured for 6-14 days at 32 C. to 37 C. and received one to two feeds of ExpiCHO Feed solution.

[0516] Conditioned cell supernatant was harvested by centrifugation (4000g; 30 minutes) and cleared by filtration (0.22 m). Bispecific molecules were purified using an kta Pure 25 L FPLC system (GE Lifesciences) equipped to perform affinity and size-exclusion chromatography in line. Affinity chromatography was performed on protein A 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 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, if needed, and buffer exchange was performed 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.

[0517] Binding affinity of these TCER-molecules towards effector cells was assessed by flow cytometry. Therefore, Jurkat cells (CD3+ and TCRab+) were incubated with raising concentrations of TCER. After washing the cells bound TCER-molecules were stained using a PE-labeled secondary reagent (#709-116-098, Jackson ImmunoResearch). Cells were finally analyzed on an Intellicyt iQue Cell Screener. Results of one of four independent experiments are shown in FIG. 3 demonstrating the concentration-dependent binding of PRAME-004-specific TCER-molecules to Jurkat cells. It is obvious that the UCHT1-based TCER-molecule shows an EC.sub.50 of binding of around 2-3 nM, whereas BMA031-based TCER molecules exhibit a at least 50-100-fold weaker binding towards the Jurkat cells.

Example 2: Proof of Principle Cytotoxicity Using Recruiters Having a Different Affinity

[0518] Antigen-binding proteins targeting the peptide MAG-003 (SEQ ID No: 10) were generated by the combination of engineered variable domains of a T Cell Receptor (SEQ ID No: 20 and SEQ ID No: 30) with the variable domains of either UCHT1(V17) (SEQ ID No: 137 and SEQ ID No: 145) or BMA031(V36) (SEQ ID No: 42 and SEQ ID No. 43), respectively, within the TCER construct.

[0519] Vectors for the expression of the respective TCER-molecules were designed as 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.

[0520] 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 kta 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.

[0521] The cytotoxic activity of the bispecific molecules against MAG-positive and MAG-negative tumor cell lines, respectively was analyzed by LDH-release assay. Therefore, tumor cell lines presenting variable amounts of HLA-A*02/MAG-003 on the cell surface were co-incubated with PBMC isolated from healthy donors (HLA-A*02+) 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).

[0522] Exemplary results of such assays are shown in Figure (Example 2). The resulting EC.sub.50-values are summarized in Table 5.

TABLE-US-00008 TABLE 5 Summary of EC.sub.50-values of killing assays comparing different recruiting antibodies. EC.sub.50 on EC.sub.50 on EC.sub.50 on Off-target EC.sub.50 on Recruiter Target High Target low positive Negative UCHT1(V17) 3 pM 4 pM 146 pM 11559 pM BMA031(V36) 69 pM 383 pM 21555 pM n/a

[0523] These results reveal the reduced potency of the BMA031(V36)-based TCER-molecule compared to the UCHT1(V17)-based molecule (23-fold on target high expressing cell line and 96-fold on target low expressing cell lines, respectively). Based on the EC.sub.50-values the safety window can be calculated as described above in the definitions-section. Briefly, safety windows are defined as ratios of EC.sub.50 of killing of Off-target expressing cells and EC50 of killing of target (TAA)-expressing cells. This implies that the UCHT1(V17)-based TCER exhibits a safety window of approx. 49-fold, whereas the BMA031(V36)-based TCER shows an increased safety window of approx. 312-fold (comparison of off-target expressing tumor cell line to target (TAA) high expressing tumor cell line).

[0524] These findings suggested that in general the utilization of low-affinity recruiting domains may improve discrimination between target and off-target thereby increasing the safety window. For further verification of this hypothesis the cytotoxic activity of the MAG-003-specific TCER-molecules towards primary healthy tissue cells (HLA-A*02+) was assessed. To this end LDH was determined in co-cultures of eleven different primary healthy tissue cells (HLA-A*02+) with PBMC effector cells from healthy HLA-A*02+ donors at an E:T ratio of 10:1 and increasing TCER concentrations. Cells were co-incubated in a 50% mixture of primary tissue cell-specific medium and optimal 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 primary cells as well as 100% optimal T cell medium to exclude a bias caused by the different media. After 48h of co-culture, supernatants were harvested and cell lysis was analyzed by measuring LDH-release using the LDH-Glo Kit (Promega).

[0525] In FIG. 5 and FIG. 6, each cytotoxicity plot shows LDH-release of a primary healthy 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. FIG. 5 summarizes the results of the UCHT1(V17)-based MAG-003-specific TCER-molecule. For all tested cell types, with exception of nasal epithelial cells and PBMC, strong reactivities were detectable and respective EC.sub.50-values could be determined. Based on EC.sub.50-values safety windows were calculated as described above within the definitions section. Safety windows in x-fold are noted within the figure. For the UCHT1(V17)-based TCER. The most critical safety windows were determined for iAstrocytes (48-fold), dermal microvascular endothelial cells (94-fold) and mesenchymal stem cells (170-fold).

[0526] In case of the BMA031(V36)-based TCER molecule all responses against healthy primary cells were 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]+[w/o TCER control])

(w/o TCER-control is indicated as dotted line in each cytotoxicity plot) was used as threshold to determine the LOEL and safety window between healthy tissue cells and tumor control cell line. All determined safety windows of the BMA031(V36)-based TCER were greater than 1000-fold.

[0527] It is obvious from comparison of FIG. 5 and FIG. 6 that the utilization of the lower-affinity recruiter BMA031(V36) within the context of the MAG-003-specific TCER-molecule results in significant expanded safety windows.

Example 3: Creation of Affinity Reduced Humanized UCHT1 Variants

[0528] To obtain lower-affinity variants of the CD3-specific humanized antibody UCHT1(V17) structure-guided design of variants was performed. Based on the solved structure of UCHT1 in complex with its target CD3/ (PDB ID: 1xiw), point mutations were introduced on the antibody that were assumed to lower the affinity without destabilizing the protein itself. In order to achieve this goal, positions were selected primarily within the CDRs; as can be inferred from the solved structure, the interface between the two proteins is formed primarily between CD3 and the antibody heavy chain, therefore only positions within the heavy chain are considered for mutations.

[0529] For clarification, positions on CD3E are numbered sequentially as in the PDB entry with ID 1xiw, chain ID: A.

[0530] G31E introduces a negative charge on the surface of the antibody that faces a negatively charged surface patch formed by CD348D, CD349E, CD350D, and CD351D, presumably causing electrostatic repulsion and thus lowering affinity.

[0531] Y54Q alters shape complementarity of the binding surfaces and removes the hydrophobic interaction between Y54's aromatic ring and the apolar stem of CD348D.

[0532] Y54E alters shape complementarity of the binding surfaces and removes the hydrophobic interaction between Y54's aromatic ring and the apolar stem of CD348D and additionally introducing a negative charge facing the negatively charged patch formed by CD348D, CD349E, CD350D, and CD351D.

[0533] K55R introduces a bulkier side chain with similar physicochemical properties, removing the H-bond formed between 55K's N and the backbone of CD356S. The increased size of the side chain might also cause a slight change in binding geometry.

[0534] K55E replaces a positive charge with a negative charge, removing the H-bond formed between 55K's N and the backbone of CD336S. Additionally, introduction of the negative charge causes electrostatic repulsion from the negatively charged patch formed by CD357D, CD358E, and CD359D.

[0535] Based on these findings sequences coding for UCHT1(V20) to UCHT1(V27) were generated as summarized in Table 6.

[0536] In a further attempt to optimize the humanized UCHT1-sequences a potential post-translational modification site (Asp-Isomerization, 106D107S) within CDR-H3 was removed by introduction of 106E. This modification was introduced in UCHT1(V17), UCHT1(20), UCHT1(V21) and UCHT1(V23) resulting in the variants UCHT1(V17opt), UCHT1(V20opt), UCHT1(V21opt) and UCHT1(V23opt), respectively.

TABLE-US-00009 TABLE 6 Combination of sequences resulting in humanized UCHT1-variants. UCHT-variant SEQ ID No. [VH] SEQ ID No. [VL] V20 158 145 V21 149 145 V22 150 145 V23 151 145 V24 152 145 V25 153 145 V26 154 145 V27 155 145 V17opt 156 145 V20opt 159 145 V21opt 157 145 V23opt 160 145

[0537] Using the humanized UCHT1-variants described in Table 6 PRAME-004-specific (V: SEQ ID No: 48, V: SEQ ID No: 44) and MAG-003-specific (V: SEQ ID No: 21, V: SEQ ID No: 30), respectively, TCER molecules were generated, produced and purified as described above.

Example 4: Affinity-Determination of Designed UCHT1-Variants

[0538] For affinity determination using biolayer-interferometry the molecule CD3-Fc was generated. Therefore, the extracellular domains of human CD3 and CD3E were fused 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) resulting in SEQ ID No: 161 and SEQ ID No. 162, respectively.

[0539] CD3-Fc-molecules were expressed in ExpiCHO cells and purified using Protein A affinity chromatography followed by size exclusion chromatography as described above.

[0540] Using biolayer interferometry, bispecific TCER antigen binding proteins comprising different UCHT1 variants (as shown in Table 6) were characterized for their binding affinity towards the MAGE-A antigenic peptide (SEQ ID NO: 10) in complex with HLA-A*02 (FIG. 9A, Table 7) and CD3-Fc (FIG. 9B, Table 7). 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 complex or CD3-Fc were loaded onto biosensors (HIS1K) prior to analyzing serial dilutions of the bispecific TCER molecules. All TCER molecules showed similar binding to HLA-A*02/MAG-003 with K.sub.D values of 2 nM. CD3 affinities covered a more than 200-fold window with K.sub.D values ranging from 3 to 750 nM.

TABLE-US-00010 TABLE 7 Affinity analysis of TCER molecules comprising different UCHT1 variants according to Table 6. K.sub.D values were measured by biolayer interferometry K.sub.D HLA-A*02/ UCHT-variant MAG-003 [nM] K.sub.D CD3-Fc [nM] V17 1.9 3.4 V20 2.0 17.8 V21 2.3 7.9 V23 2.1 46.1 V17opt 2.0 22.6 V20opt 1.7 373.0 V21opt 2.1 106.4 V23opt 2.4 747.1

Example 5: Reduced Potency of Low Affinity Humanized UCHT1 Variants

[0541] Potency with respect to cytotoxicity of the MAG-003-specific TCER-molecules was assessed in LDH-release assays as described above within example 2. Results of representative assays are shown in FIG. 10. As expected, the newly designed variants (UCHT1(V17opt), UCHT1(V20), UCHT1(V21), UCHT1(V23)) showed reduced potency in comparison to the TCER containing high affinity recruiting domains of UCHT1(V17). Ranking the TCER-molecules according to their potencies also closely reflects the affinities of the respective recruiter variants (highest potency: UCHT1(V17)<UCHT1(V21)<UCHT1(V20)<UCHT1(V17opt)<UCHT1V23). These effects on potencies could solely be attributed to the recruiting domains as the affinities of the TCR-domains towards MAG-003 in complex with HLA-A*02 are comparable (FIG. 9).