IMPROVED DUAL SPECIFICITY POLYPEPTIDE MOLECULE

Abstract

The present invention relates to a bispecific polypeptide molecule comprising a first polypeptide chain and a second polypeptide chain providing a binding region derived from a T cell receptor (TCR) being specific for a major histocompatibility complex (MHC)-associated peptide epitope, and a binding region derived from an antibody capable of recruiting human immune effector cells by specifically binding to a surface antigen of said cells, as well as methods of making the bispecific polypeptide molecule, and uses thereof.

Claims

1. A polypeptide comprising a heavy chain variable domain (VH) and a light chain variable domain (VL) of hBMA031 antibody comprising the amino acid sequences of the heavy chain variable domain (VH) and the light chain variable domain (VL) of the hBMA031 antibody as comprised in SEQ ID NOs: 22 and 23, respectively.

2. A heavy chain variable domain (VH) and a light chain variable domain (VL) comprising the amino acid sequences of the heavy chain variable domain (VH) and the light chain variable domain (VL) of hBMA031 antibody as comprised in SEQ ID NOs: 22 and 23, respectively.

3. The polypeptide of claim 1, wherein the polypeptide: specifically binds to a TCR-CD3 complex of human T cells; and is capable of recruiting human immune effector cells by specifically binding to a surface antigen of said effector cell.

4. The polypeptide of claim 1, wherein the VH domain and the VL domain each comprise no more than one, two, three, four, or five amino acid substitutions, deletions, or insertions.

5. The polypeptide of claim 3, wherein the polypeptide specifically binds to the TCR-CD3 complex with a binding affinity (KD) of 1 .Math.M or less.

6. A nucleic acid encoding the polypeptide of claim 1.

7. An expression vector comprising the nucleic acid of claim 6.

8. A host cell expressing the vector of claim 7.

9. A pharmaceutical composition comprising the polypeptide of claim 1 and one or more pharmaceutically acceptable carriers or excipients.

10. A method of producing a polypeptide, comprising culturing the host cell of claim 8 in a medium and harvesting the polypeptide from the host cell and/or from the medium.

11. The VH domain and the VL domain of claim 2, wherein the VH domain and the VL domain: specifically bind to a TCR-CD3 complex of human T cells; and are capable of recruiting human immune effector cells by specifically binding to a surface antigen of said effector cell.

12. The VH domain and the VL domain of claim 2, wherein the VH domain and the VL domain each comprise no more than one, two, three, four or five amino acid substitutions, deletions, or insertions.

13. The VH domain and the VL domain of claim 11, wherein the VH domain and the VL domain specifically bind to the TCR-CD3 complex with a binding affinity (KD) of 1 .Math.M or less.

14. A nucleic acid encoding the VH domain and the VL domain of claim 2.

15. An expression vector comprising the nucleic acid of claim 14.

16. A host cell expressing the vector of claim 15.

17. A pharmaceutical composition comprising the VH domain and the VL domain of claim 2 and one or more pharmaceutically acceptable carriers or excipients.

18. A method of producing a VH domain and a VL domain, comprising culturing the host cell of claim 16 in a medium and harvesting the VH domain and the VL domain from the host cell and/or from the medium.

19. A dual specificity polypeptide molecule comprising a first polypeptide chain comprising SEQ ID NO: 22 and a second polypeptide comprising SEQ ID NO: 23.

20. A nucleic acid encoding the dual specificity polypeptide molecule of claim 19.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0124] FIG. 1 shows a schematic overview over a preferred embodiment of the present invention, the human IgG1 Fc-containing dual specificity polypeptide molecule. VD1, VD2 = variable domains derived from antibody; VR1, VR2 = variable domains derived from TCR; Link1, Link2 = connecting linkers; Cys-Cys = cysteine bridges.

[0125] FIG. 2 shows a schematic overview over 4 different constructs of IgG Fc-containing dual specificity polypeptide molecules as tested in the context of the present invention. black = TCR-derived variable domains; light gray = antibody-derived variable domains; white = constant domains derived from human IgG. Knob-hole mutations are indicated by a cylinder. Diabody molecules IA-ID are according to the invention.

[0126] FIG. 3 shows the HPLC-SEC analysis of different bispecific TCR/mAb molecules with a molecular design according to the constructs depicted in FIG. 2, which were purified by a 2-column purification process. The monomer contents of the different molecules were determined as follows. II: 93.84%; III: 96.54%; IV: 98.49%; IA_1: 95.48%; IA_3: 98.45%; ID_1: 95.75%; IC_4: 95.22%; IC_5: 92.76%; ID_4: 99.31%; ID_5: 99.44%.

[0127] FIG. 4 shows the results of the potency assay with different bispecific TCR/mAb constructs (as shown in FIG. 2) designed as IgG4-based molecules. Jurkat_NFATRE_luc2 cells were co-incubated with HIV-peptide SLYNTVATL (SEQ ID No. 7) loaded T2 cells in the presence of increasing concentrations of bispecific TCR (bssTCR) molecules. The bispecific TCR/mAb diabody molecule IA-IgG4 exhibited a higher potency than two alternative dual specificity TCR/mAb molecules.

[0128] FIG. 5 shows the results of the potency assay with different bispecific TCR/mAb constructs (as shown in FIG. 2) designed as IgG1-based molecules. Jurkat_NFATRE_luc2 cells were co-incubated with HIV-peptide SLYNTVATL (SEQ ID No. 7) loaded T2 cells in the presence of increasing concentrations of bispecific TCR (bssTCR) molecules. The bispecific TCR/mAb diabody molecules ID_1, IA_3 and IA1 exhibited markedly higher potency than three alternative dual specificity TCR/mAb molecules.

[0129] FIG. 6 shows the results of the potency assay conducted with different IgG1-based bispecific TCR/mAb constructs (as shown in FIG. 2) utilizing different variable antibody domains both targeting the TCR-CD3 complex. Construct ID_1 comprises variable domains of the UCHT1(V9) antibody targeting CD3, whereas the constructs ID_4 and ID_5 comprise variable domains of the alpha/beta TCR-specific antibody BMA031. Jurkat_NFATRE_luc2 cells were co-incubated with HIV-peptide SLYNTVATL (SEQ ID No. 7) loaded T2 cells in the presence of increasing concentrations of bispecific TCR (bssTCR) molecules.

[0130] FIGS. 7A, 7B, 7C, and 7D show a schematic overview over the possible orientations of the VD and VR domains in the molecules of the present invention. VH: antibody-derived VH-domain, VL: antibody-derived VL-domain; Vα: TCR-derived Valpha; Vβ: TCR-derived Vbeta.

[0131] FIG. 8 shows the results of HPLC-SEC analysis of aggregates (HMWS - high molecular weight species) within different bispecific TCR/mAb molecules based on IgG1. Aggregates were analyzed after purification and after storage of the molecules at 40° C. for 1 weeks and 2 weeks, respectively.

[0132] FIG. 9 shows the results of the potency assay conducted with different bispecific TCR/mAb molecules based on IgG1. Potency was analyzed after purification and after storage of the molecules at 40° C. for 1 week and 2 weeks, respectively. Stress storage at 40° C. did not lead to significant loss of potency of the molecules but a drastic increase in unspecific (i.e. target-independent) activation of Jurkat T cells was detected for the molecules III and IV.

[0133] FIG. 10 shows the results of a LDH-release assay with different bispecific TCR/mAb constructs (as shown in FIG. 2) designed as IgG1-based molecules. PBMC isolated from a healthy donor were co-incubated with HIV-peptide SLYNTVATL (SEQ ID No. 7) loaded T2 cells in the presence of increasing concentrations of bispecific TCR (bssTCR) molecules. The bispecific TCR/mAb diabody molecules IA_3 and ID_1 induced markedly higher lysis of target cells than three alternative dual specificity TCR/mAb molecules. As shown on the right hand sided graph none of the tested bispecific TCR/mAb constructs induced detectable lysis of T2 cells loaded with irrelevant peptide (SEQ ID No. 49).

[0134] FIG. 11 shows the results of a LDH-release assay with the bispecific TCR/mAb diabody construct IA_5 targeting tumor-associated peptide PRAME-004 (SEQ ID No. 49) presented on HLA-A*02. CD8-positive T cells isolated from a healthy donor were co-incubated with cancer cell lines UACC-257, SW982 and U2OS presenting differing amounts of PRAME-004:HLA-A*02-1 complexes on the cell surface (approx. 1100, approx. 770 and approx. 240 copies per cell, respectively, as determined by M/S analysis) at an effector:target ratio of 5:1 in the presence of increasing concentrations of TCR/mAb diabody molecules. After 48 hours of co-culture target cell lysis was quantified utilizing LDH-release assays according to the manufacturer’s instructions (Promega).

[0135] FIG. 12 shows the results of a LDH-release assay with the bispecific TCR/mAb diabody constructs IA_5 and IA_6 utilizing a stability/affinity maturated TCR and an enhanced version thereof, respectively, against the tumor-associated peptide PRAME-004 (SEQ ID No. 49) presented on HLA-A*02. CD8-positive T cells isolated from a healthy donor were co-incubated with the cancer cell line U2OS presenting approx. 240 copies per cell of PRAME-004:HLA-A*02-1 complexes or non-loaded T2 cells (effector:target ratio of 5:1) in the presence of increasing concentrations of TCR/mAb diabody molecules. After 48 hours of coculture target cell lysis was quantified utilizing LDH-release assays according to the manufacturer’s instructions (Promega).

[0136] FIG. 13 shows the results of a heat-stress stability study of the TCR/mAb diabody constructs IA_5 and IA_6 utilizing a stability/affinity maturated TCR and an enhanced version thereof, respectively, against the tumor-associated peptide PRAME-004 (SEQ ID No. 49) presented on HLA-A*02. For this, the proteins were formulated in PBS at a concentration of 1 mg/mL and subsequently stored at 40° C. for two weeks. Protein integrity and recovery was assessed utilizing HPLC-SEC. Thereby the amount of high-molecular weight species was determined according to percentage of peak area eluting before the main peak. Recovery of monomeric protein was calculated by comparing main peak areas of unstressed and stressed samples.

EXAMPLES

Example 1

Design of Fc-Containing Bispecific TCR/mAb Diabodies and Control Molecules

[0137] Fc-containing bispecific TCR/mAb diabodies and control molecules (as depicted in FIG. 2) were designed to specifically bind to the human TCR-CD3 complex and to the peptide:MHC complex comprising the HIV-derived peptide SLYNTVATL (SQ ID No. 7) bound to HLA-A2*01. For targeting TCR-CD3 complex, VH and VL domains derived from the CD3-specific, humanized antibody hUCHT1(V9) described by Zhu et al. (Identification of heavy chain residues in a humanized anti-CD3 antibody important for efficient antigen binding and T cell activation. J Immunol, 1995, 155, 1903-1910) or VH and VL domains derived from the alpha/beta TCR-specific antibody BMA031 described in Shearman et al. (Construction, expression and characterization of humanized antibodies directed against the human alpha/beta T cell receptor. J Immunol, 1991, 147, 4366-73) and employed in the humanized version variant 10 (data generated in-house) were used. For targeting peptide:MHC complex, Valpha and Vbeta domains of the previously described stability and affinity maturated, human single chain T-cell receptor 868Z11 disclosed by Aggen et al. (Identification and engineering of human variable regions that allow expression of stable single-chain T cell receptors. PEDS, 2011, 24, 361 - 372) were utilized.

[0138] In case of Fc-containing bispecific TCR/mAb diabodies DNA-sequences coding for various combinations of VH and VL (corresponding to VD1 and VD2, respectively) and Va and Vb (corresponding to VR1 and VR2, respectively), as well as coding for linkers Link1 and Link2 were obtained by gene synthesis. Resulting DNA-sequences were cloned in frame into expression vectors coding for hinge region, CH2 and CH3 domain derived from human IgG4 [Accession#: K01316] and IgG1 [Accession#: P01857], respectively and were further engineered. Engineered was performed to incorporate knob-into-hole mutations into CH3-domains with and without additional interchain disulfide bond stabilization; to remove an N-glycosylation site in CH2 (e.g. N297Q mutation); to introduce Fc-silencing mutations; to introduce additional disulfide bond stabilization into VL and VH, respectively, according to the methods described by Reiter et al. (Stabilization of the Fv Fragments in Recombinant Immunotoxins by Disulfide Bonds Engineered into Conserved Framework Regions. Biochemistry, 1994, 33, 5451 - 5459). An overview of produced bispecific TCR/mAb diabodies, the variants as well as the corresponding sequences are listed in Table 1.

TABLE-US-00001 Overview of all generated and evaluated Fc-containing bispecific TCR/mAb diabodies: Molecule TCR mAb SEQ IDs modifications IA-IgG4 868Z11 hUCHT1(V9) SEQ ID No. 8 SEQ ID No. 9 IgG4 (KiH) IA_1 868Z11 hUCHT1(V9) SEQ ID No. 10 SEQ ID No. 11 IgG1 (K/O, KiH) IA_2 868Z11 hUCHT1(V9) SEQ ID No. 12 SEQ ID No. 13 IgG1 (K/O, KiH-ds) IA_3 868Z11 ds-hUCHT1(V9) SEQ ID No. 14 SEQ ID No. 15 IgG1 (K/O, KiH-ds) ID_1 868Z11 ds-hUCHT1(V9) SEQ ID No. 16 SEQ ID No. 17 IgG1 (K/O, KiH-ds) IC_4 868Z11 hBMA031(var10) SEQ ID No. 18 SEQ ID No. 19 IgG1 (K/O, KiH-ds) IC_5 868Z11 hBMA031(var10) SEQ ID No. 20 SEQ ID No. 21 IgG1 (K/O, KiH-ds) extended Link1 ID_4 868Z11 hBMA031(var10) SEQ ID No. 22 SEQ ID No. 23 IgG1 (K/O, KiH-ds) ID_5 868Z11 hBMA031(var10) SEQ ID No. 24 SEQ ID No. 25 IgG1 (K/O, KiH-ds) extended Link1 IA_5 R16P1C10I hUCHT1(Var17) SEQ ID No. 43 SEQ ID No. 44 IgG1 (K/O, KiH-ds) IA_6 R16P1C10I#6 hUCHT1(Var17) SEQ_ID No. 45 SEQ ID No. 46 IgG1 (K/O, KiH-ds) KiH: Knob-into-hole; K/O: Fc-silenced; KiH-ds: Knob-into-hole stabilized with artificial disulfide-bond to connect CH3:CH3’; ds-hUCHT1(V9): disulfide-bond stabilized hUCHT1(V9) variable domains; Link1: Linker connecting VR1 and VD1. Various control molecules exhibiting the same specificities were constructed Table 2 utilizing said VH, VL, Valpha and Vbeta domains in combinations with IgG1- or IgG4-derived constant domains comprising engineered features as described above.

TABLE-US-00002 Overview of all generated and evaluated Fc-containing bispecific control molecules: KiH: Knob-into-hole; K/O: Fc-silenced Molecule TCR mAb SEQ IDs modifications III-IgG4 868Z11 hUCHT1(V9) SEQ ID No. 38 SEQ ID No. 39 IgG4 (KiH) IV-IgG4 868Z11 hUCHT1(V9) SEQ ID No. 40 SEQ ID No. 41 IgG4 II 868Z11 hUCHT1(V9) SEQ ID No. 33 IgG1 (K/O, KiH) SEQ ID No. 34 III 868Z11 hUCHT1(V9) SEQ ID No. 35 SEQ ID No. 36 IgG1 (K/O, KiH) IV 868Z11 hUCHT1(V9) SEQ ID No. 37 SEQ ID No. 42 IgG1 (K/O)

Example 2

Production and Purification of FC-Containing Bispecific TCR/mAb Diabodies

[0139] 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.

[0140] Conditioned cell supernatant was harvested by centrifugation (4000 × g; 30 minutes) and cleared by filtration (0.22 .Math.m). Bispecific molecules were purified using an Äkta 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.

[0141] As therapeutic proteins shall exhibit reasonable stability upon acidic exposure to facilitate robust industrial purification processes the percentage of monomeric protein eluting from the protein A capture column was assessed (Table 3). It is obvious that the introduction of stabilizing mutations into molecules as well as selection of distinct orientations of binding domains markedly impact the stability upon acidic exposure.

TABLE-US-00003 Fraction of monomeric protein after acidic elution from capture column Molecule Monomer eluted from capture column (% of total peak area) IA-IgG4 (VH-beta) n.d. IA_1 (VH-beta) 49 IA_2 (VH-beta) 54 IA_3 (dsVH-beta) 63 ID_1 (alpha-dsVH) 46 IC_4 (VH-alpha) 62 IC_5 (VH-alpha) 67 ID_4 (alpha-VH) 65 ID_5 (alpha-VH) 69 II 39 III 51 IV 76 After size exclusion chromatography, the purified bispecific molecules demonstrated high purity (>93% of monomeric protein) as determined by HPLC-SEC on MabPac SEC-1 columns (5 .Math.m, 7.8×300 mm) running in 50 mM sodium-phosphate pH 6.8 containing 300 mM NaCl within an Agilent 1100 system (see FIG. 3). Non-reducing and reducing SDS-PAGE confirmed the purity and expected size of the different dual specificity TCR/mAb molecules (data not shown).

Example 3

Specific and Target Cell-Dependent T Cell Activation Induced by Fc-Containing TCR/mAb Diabodies

[0142] The potency of Fc-containing TCR/mAb diabodies with respect to T cell activation was assessed using the T Cell Activation Bioassay (Promega). The assay consists of a genetically engineered Jurkat cell line that expresses a luciferase reporter driven by an NFAT-response element (NFAT-RE). Assays were performed according to the manufacturer. Briefly, T2 cells either loaded with the HIV-specific peptide SLYNTVATL (SEQ ID No. 7) or left without peptide loading (unloaded control) were subsequently co-cultured with Promega’s modified Jurkat cells in presence of increasing concentrations of bispecific TCR/mAb molecules. Jurkat reporter T cell activation was analyzed after 16-20 hours by measuring luminescence intensity.

[0143] Representative potency assay results are depicted for IgG4-based (FIG. 4) and IgG1-based bispecific TCR/mAb molecules (FIG. 5), respectively. The data indicate that regardless of the IgG isotype of the constant domains used, the Fc-containing TCR/mAb diabody constructs IA and ID showed superior T cell activation compared to the alternative bispecific TCR/mAb constructs II, III and IV as measured by the magnitude of activation and/or respective EC50-values. Furthermore, the unspecific T cell activation of Fc-containing TCR/mAb diabodies induced against unloaded T2 cells was reduced or at least equal to the level of unspecific activation observed for the alternative bispecific TCR/mAb constructs. According to above results the dual specificity TCR/mAb diabody molecules are preferred molecules for therapeutic intervention as they induce strong effector T cell activation in a highly target-dependent manner.

[0144] Furthermore LDH-release assay (Promega) was used to quantify the PBMC-mediated lysis of SLYNTVATL (SEQ ID No. 7) peptide-loaded T2 cells induced by the different bispecific TCR/mAb molecules (FIG. 10). In line with the above results of the T Cell Activation Bioassay, again the Fc-containing TCR/mAb diabody constructs IA and ID were superior over the alternative bispecific TCR/mAb constructs II, III and IV as indicated by the increased absolute level of target cell lysis and the lower TCR bispecific concentration needed to achieve half-maximal (EC50) killing of target cells. As for TCR/mAb constructs II, III and IV, the TCR/mAb diabody constructs IA and ID did not induce lysis of T2 cells loaded with irrelevant peptide(SEQ ID No. 49), proving the target-specific lysis to the T2 cells.

EXAMPLE 4

Development of Fc-Containing Bispecific TCR/mAb Diabodies as a Molecular Platform

[0145] Fc-containing bispecific TCR/mAb diabody constructs were designed to serve as molecular platform to provide the scaffold for different TCR-derived and mAb-derived variable domains targeting different peptide:MHC complexes and effector cell surface antigens, respectively. To validate the suitability as platform, the mAb-derived variable domains were exchanged in a first set of molecules. The variable domains of hUCHT1(V9) anti-CD3 antibody (construct ID_1) were replaced against the domains of the hBMA031(var10) anti-TCR antibody employing the same domain orientation (constructs ID_4 and ID_5) or a different orientation (IC_4, IC_5) (see Table 1and FIGS. 7 for details). Expression, purification and characterization of these molecules were performed as described above. Purity and integrity of final preparations exceeded 92% according to HPLC-SEC analyses.

[0146] The potency assay results revealed target-dependent Jurkat reporter T cell activation and minimal unspecific activity against unloaded T2 cells for both antibody variable domains hUCHT1 (construct ID_1) and hBMA031 (constructs ID_4 and ID_5) supporting the platform suitability of the dual specificity TCR/mAb diabody constructs (FIG. 6). Notably, when the variable TCR and mAb domains of the constructs ID_4 and ID_5 were switched on each polypeptide chain resulting in constructs IC_4 and IC_5 no T cell activation was observed (data not shown). The latter finding indicate that despite bispecific TCR/mAb diabodies can be used as platform construct for incorporating different TCR and mAb variable domains a thorough optimization of the domain orientation is required to achieve optimal activity of the molecules.

EXAMPLE 5

Stability of Fc-Containing Bispecific TCR/mAb Diabodies

[0147] Stability of the bispecific TCR/mAb molecules was initially assessed utilizing the Protein Thermal Shift Assay (Thermo Fisher Scientific) according to the instructions of the manufacturer using a 7500 Real time PCR system (Applied Biosciences). Briefly, purified molecules were mixed with PTS buffer and PTS dye and subjected to a raising temperature gradient constantly monitoring fluorescence of samples. Recorded fluorescence signals were analyzed using PTS software (Thermo Fisher Scientific) and melting temperatures (T.sub.M) were calculated by the derivative method.

[0148] Stressed stability studies were conducted by storage of purified molecules dissolved in PBS at 40° C. for up to two weeks. Samples were analyzed with regard to protein integrity using HPLC-SEC and potency using the T Cell Activation Assay (Promega) as described above.

[0149] As expected storage at 40° C. induced the formation of aggregates / high-molecular weight species as determined by HPLC-SEC analyses (see FIG. 8). Results of potency assays of IgG1-based molecules after purification and incubation at 40° C. are shown in FIG. 9. Although neither of the tested molecules did show a significant reduction of potency after storage at 40° C., it was observed that the stressed molecules III and IV induced a significant amount of unspecific (i.e. target-independent) Jurkat T cell activation. In contrast, the bispecific TCR/mAb diabodies retained their target-dependent potency, despite the presence of some aggregates as seen in HPLC-SEC.

EXAMPLE 6

Generation of Cancer-Targeting Bispecific TCR/mAb Diabody Molecules

[0150] To further validate the platform capabilities of bispecific TCR/mAb diabody constructs, the TCR-derived variable domains were exchanged with variable domains of a TCR, which was stability/affinity maturated by yeast display according to a method described previously (Smith et al, 2015, T Cell Receptor Engineering and Analysis Using the Yeast Display Platform. Methods Mol Biol. 1319:95-141). The TCR variable domains specifically binding to HIV-derived peptide SLYNTVATL (SEQ ID No. 7) in the context HLA-A*02 were exchanged with TCR variable domains specifically binding to the tumor-associated peptide PRAME-004 (SEQ ID No. 49) bound to HLA-A*02. Furthermore, the variable domains of the humanized T-cell recruiting antibody hUCHT1(V9) were exchanged against variable domains of hUCHT1(Var17), a newly humanized version of the UCHT1 antibody, resulting in the PRAME-004-targeting TCR/mAb diabody molecule IA_5 (comprising SEQ ID No. 43 and SEQ ID No. 44). Expression, purification and characterization of this molecule was performed as described in Example 2. Purity and integrity of final preparation exceeded 96% according to HPLC-SEC analysis.

[0151] Binding affinities of bispecific TCR/mAb diabody constructs towards PRAME-004:HLA-A*02 were determined by biolayer interferometry. Measurements were done on an Octet RED384 system using settings recommended by the manufacturer. Briefly, purified bispecific TCR/mAb diabody molecules were loaded onto biosensors (AHC) prior to analyzing serial dilutions of HLA-A*02/PRAME-004.

[0152] The activity of this PRAME-004-targeting TCR/mAb diabody construct with respect to the induction of tumor cell lysis was evaluated by assessing human CD8-positive T cell-mediated lysis of the human cancer cell lines UACC-257, SW982 and U2OS presenting different copy numbers of PRAME-004 peptide in the context of HLA-A*02 on the tumor cell surface (UACC-257 - about 1100, SW982 - about 770, U2OS — about 240 PRAME-004 copies per cell, as determined by quantitative M/S analysis) as determined by LDH-release assay.

[0153] As depicted in FIG. 11, the PRAME-004-targeting TCR/mAb diabody construct IA_5 induced a concentration-dependent lysis of PRAME-004 positive tumor cell lines. Even tumor cells U2OS expressing as little as 240 PRAME-004 copy numbers per tumor cell were efficiently lysed by this TCR/mAb diabody molecule. These results further demonstrate that

[0154] TCR/mAb diabody format is applicable as molecular platform allowing to introduce variable domains of different TCRs as well as variable domains of different T cell recruiting antibodies.

EXAMPLE 7

Engineerability of TCR/mAb Diabody Constructs

[0155] The variable TCR domains utilized in construct IA_5 were further enhanced regarding affinity towards PRAME-004 and TCR stability, and used for engineering into TCR/mAb diabody scaffold resulting in construct IA_6 (comprising SEQ ID No. 45 and SEQ ID No. 46). Expression, purification and characterization of TCR/mAb diabody molecules IA_5 and IA_6 were performed as described in example 2. Purity and integrity of final preparations exceeded 97% according to HPLC-SEC analysis.

[0156] Potency of the stability and affinity enhanced TCR/mAb diabody variant IA_6 against PRAME-004 was assessed in cytotoxicity experiments with the tumor cell line U2OS presenting low amounts of PRAME-004:HLA-A*02 or non-loaded T2 cells as target cells and human CD8-positive T cells as effector cells.

[0157] As depicted in FIG. 12, the inventors observed and increased cytotoxic potency of the TCR/Ab diabody molecule IA_6 comprising the variable domains of the stability/affinity enhanced TCR variant when compared to the precursor construct IA_5. For both constructs, IA_5 and IA_6, the PRAME-004-dependent lysis could be confirmed as no cytolysis of target-negative T2 cells was detected.

[0158] The protein construct were further subjected to heat-stress at 40° C. for up to two weeks to analyze stability of the PRAME-004-specific TCR/mAb diabody variants IA_5 and IA_6. HPLC-SEC analyses after heat-stress revealed a significantly improved stability of the variant IA_6 when compared to the precursor construct IA_5 (see FIG. 13). The temperature-induced increase of high-molecular species (i.e. eluting before the main peak) of the constructs was less pronounced for IA_6 than for IA_5. In line with this result, the recovery of intact, monomeric protein after heat-stress was 87% and 92% for IA_5 and IA_6, respectively.

[0159] These exemplary engineering data demonstrate that the highly potent and stable of TCR/mAB diabody constructs can further be improved by incorporating stability/affinity enhanced TCR variable domains resulting in therapeutic proteins with superior characteristics.

EXAMPLE 8

Examples for Preferred Constructs

[0160] In addition to the HIV-specific TCR bispecific construct as described herein (Seq ID No. 16 and Seq ID No. 17, in orientation D), the invention further provides several other exemplary HIV-specific constructs that were tested. These constructs are based on an improved humanized variants of the underlying antibody against CD3 (UCHT1) that were fused with the HIV-specific TCR 868 as disclosed herein in all four possible orientations (Seq ID No. 51 to Seq ID No. 58, in orientations A-D).

[0161] The humanization of UCHT1 was performed using VH-1-46 and VK1-018 as acceptor frameworks for the heavy and light chain CDRs, respectively. J-segments selected were JK1 and JH4, for light and heavy chain, respectively.

[0162] The results as obtained are shown in the following Table 4:

TABLE-US-00004 V9 (Zhu et al, 1995) Present invention DRB1 score 1232 ~1190 Titre [mg/L] 0.75 3 Tm of F(ab) [°C] 83.0 86.4 EC50 of effector cell activation [pM] 63 8

[0163] The data in table 4 shows that the inventive humanization is potentially less immunogenic (lower DRB 1-score); the molecules are more stable (increase in melting temperature of about 3° C.); and more potent (~8x decreased EC50), compared with the standard (V9) (for assay, see example 3).