Soluble non-aggregating immune ligand

Abstract

Soluble, multi-polypeptide multispecific protein for recruiting an immune cell to a target cell presenting a target peptide major histocompatibility complex (pMHC), for therapeutic use against diseases such as cancer Protein comprising (i) a pMHC binding arm comprising a dimer of a first portion and a second portion, (ii) an immune cell engager moiety, and (iii) an Fc region comprising a dimer of a first Fc portion and a second Fc portion, wherein the first portion of the pMHC binding arm is linked at its N terminus to the immune cell engager moiety, and wherein the second portion of the pMHC binding arm is linked at its N terminus to a C terminus of the first Fc portion, and an N terminus of the Fc region is linked to a C terminus of the pMHC binding arm.

Description

[0001] SOLUBLE NON-AGGREGATING IMMUNE LIGAND

[0002] Field of the Invention

[0003] The present invention relates to multispecific molecules for bridging immune synapses between immune cells and target cells in a patient, to treat diseases such as cancer, which combine high potency with extended in vivo half-life.

[0004] Background

[0005] T cells express variable T cell receptors (TCRs) complexed with CD3 polypeptides. The TOR is activated via binding to its cognate peptide-major histocompatibility complex (pMHC), a cell surface component which is found on most mammalian cells and which presents a sample of peptides from within the cell via the antigen processing and presentation pathway. MHC class I molecules on the surface of most mammalian cells present 8 to 11 mer peptides derived from the cell’s own cytosolic proteins, and are recognised by CD8+ T cells in which the TCR-CD3 complex and co-receptor CD8 bind to the pMHC (Figure 5). MHC class II molecules are found on specialised antigen presenting cells (APC) and present 12 to 20 mer peptides derived from proteins taken up by the cell from its extracellular environment, and are recognised by CD4+ T cells in which the TCR-CD3 complex and co-receptor CD4 bind to the pMHC.

[0006] A repertoire of TCRs with diverse sequences are generated by somatic rearrangement of genomic DNA within a population of T cells in vivo, each T cell expressing a single (monoclonal) TCR containing random sequence elements. Positive and negative selection during T cell development permits survival of T cells expressing TCR which recognise pMHC with a threshold affinity and deletes self-reactive cells that exhibit high affinity binding, thereby producing a population of T cells which do not recognise normal “self” pMHC but which may bind previously unencountered pMHC. When a TCR of a CD8+ cytotoxic T lymphocyte (CTL) recognises a peptide displayed on MHC class I, the binding of TCR to the pMHC initiates intracellular signalling via the CD3 component of the TCR complex, activating the CTL to release cytolytic molecules and immune cell-activating cytokines. Cells expressing proteins with neoepitopes (e.g., infected cells producing viral protein, and cells expressing mutated oncogenes) can thus be eliminated by the cellular adaptive immune system to maintain health.

[0007] CTL activity has been harnessed and enhanced in a therapeutic setting to destroy tumour cells, using T cell redirecting bispecifics which bind CD3 and a tumour specific cell surface antigen. Binding of the antibody to the CD3 complex over-rides the natural specificity of the T cell for its cognate pMHC, activating polyclonal CTL to kill the target cell. Scientists at Immunocore fused a truncated extracellular region of a TCR (including variable and constant domains of the alpha and beta chains) to an anti-CD3 scFv to create a polypeptide “ImmTAC” for redirecting CTL to tumour cells presenting target pMHC (Boulter & Jakobsen Clin Exper Immunol 142:454-4602005; Liddy et al, Nat Med 18(6):980-9872012, Oates & Jakobsen, Oncolmmunology 2(2):e22891 2013). Tebentafusp, an ImmTAC containing a TCR specific for the glycoprotein 100 (gp100) peptide fragment YLEPGPVTA presented by human leukocyte antigen (HLA)-A*02:01 , has been approved for treating HLA-A*02:01-positive uveal melanoma patients with metastatic cancer.

[0008] Bispecifics have been described that recruit various types of immune cells (e.g., regulatory T cells, NK cells, as well as T effector cells) via other immune receptors (e.g., CD28, and checkpoint molecules such as PD-1). Bispecifics for engaging T cells and other immune effector cells were reviewed by Wu & Cheung, Pharmacol Ther 182:161-175 2018 and Fuca et al, ESMO 6(1) 2021.

[0009] Engagement between an immune cell and a target cell presenting pMHC occurs at a receptor-dense membrane interface termed the immune synapse upon TCR and pMHC ligation and subsequent clustering. It is desirable to provide a therapeutic molecule that is able to promote formation of a strong immune synapse between a target cell and an immune cell in a patient, and which has an extended in vivo half life. However, there is a design challenge in finding a multispecific molecular format that combines the advantages of (1) potency for triggering signalling at the immune synapse with (2) acceptable circulatory half life in vivo to limit dosing frequency of the drug product. Strategies for extending half life tend to involve the addition of bulky molecules that reduce clearance of the drug from the circulation. Examples of half life extenders are hydrophilic polymers such as PEG to increase hydrodynamic radius, antibody Fc regions that allow capture and recycling via FcRn, and serum albumin binders or serum albumin itself, which has a long circulatory half life as it also binds to FcRn. However, attachment of such half life extending regions may strain the spatial constraints for formation of tight immune synapses. Proximity between the plasma membranes of the two cells at the immune synapse contributes to effective signalling within the immune cell, and triggering of downstream responses. Bulky molecules may sterically hinder the immune synapse, leading to lower activation of immune receptor signalling, especially at low antigen densities.

[0010] The tyrosine phosphatase CD45 restrains immune cell receptor signalling, and its exclusion from the immune synapse creates a phosphatase-free zone that is conducive to immune receptor signalling and subsequent immune cell activation (e.g., from CD3 signalling) or suppression (e.g., from PD-1 signalling). As CD45 has a large extracellular region, a short (~15 nM) inter-membrane distance between target cell and immune cell effectively excludes CD45, which is pushed out to a distal ring around the synapse. At larger synapses (> ~15 nM), CD45 dampens intracellular signals from the immune cell receptor, although this may be overcome in the presence of a high copy number of target pMHC being presented to the immune cell, resulting in a high number of immune cell receptor engagements and consequent high volume of signalling induced within the immune cell, outweighing the negative influence of tyrosine phosphatase. Under many physiological conditions, however, the density of target pMHC is very low, so a tight immune synapse is required for effective immune cell signalling. Copy number of some target pMHC on the tumour cell surface has been estimated to be as low as between 1 - 10 (Liddy et al., Nat Med 18(6):981-987 2012). Spatial constraints are therefore important when designing multispecific molecules to bridge the immune synapse and promote effective immune receptor signalling.

[0011] Scientists at Immatics designed a format for a TCR-immunoglobulin bispecific molecule including a diabody-type structure coupled to an Fc domain. As disclosed in W02019 / 012138, the bispecific comprises a first polypeptide chain comprising a first TCR variable domain, a first antibody variable domain and a first Fc region, paired with a second polypeptide chain comprising a second antibody variable domain, a second TCR variable domain and a second Fc region. The assembled bispecific, termed “TCER”, is a heterodimer comprising paired TCR variable domains forming a pMHC binding site and paired antibody variable domains forming a CD3 binding site, and paired Fc regions. This format benefits from extended in vivo half life, but is less potent than the “ImmTAC” (scFv-TCR fusion, described above).

[0012] Dickopf, Georges & Brinkmann (Computational and Structural Biotechnology Journal 18:1221-12272020) discussed the importance of geometry for multispecific molecules that elicit intercellular processes such as the formation of immunological synapses. They illustrated that molecular formats of the same size and same domain sequences, but different geometries, had different efficacies. A compact (5 - 6 nm) format termed “contorsbody” was reported, composed of three antibody Fab regions, in which a central Fab (anti-CD3) is oriented anti-parallel to two antigen-targeting Fabs, so that the CD3-binding and target antigen-binding sites face in opposing directions. The small barrel-like design of the molecule and close proximity of the dual antigen-binding sites was suggested to lead to denser TCR clustering, and reportedly showed higher potency compared with regular Y-shaped IgG bispecific molecules.

[0013] Froning et al. (J. Immunother Cancer 10:e004281 2022) compared soluble TCR-anti- CD3 bispecific formats. TCR-scFv, TCR-Fab and TCR-IgG anti-CD3 bispecifics were compared in a panel of assays for antigen binding and T cell redirected cell killing. The tandem fusion formats (TCR-scFv, TCR-Fab) showed reduced binding to CD3 compared with the TCR-IgG, which was attributed to the N-terminal fusion hindering the kinetics of CD3 engagement in the tandem formats. Despite this, however, the TCR-Fab and TCR-scFv were more active than the TCR-IgG in cell killing assays. Two different TCR-IgG anti-CD3 bispecific molecules, with TCRs specific for two different pMHC, were both inactive in assays for T cell engaging activity despite being capable of binding their cognate pMHC complex and CD3 in separate binding assays. Efforts to increase avidity by including an additional TCR moiety in the various formats did not improve their cell killing potency, and in most cases actually reduced activity. The authors concluded that only small, tandem bispecifics were capable of inducing strong redirected cytolytic activity against tumours, while IgG-like or Fc-containing molecules demonstrated poor activity because the larger and less flexible TCR-IgG is less able to form a tight immune synapse.

[0014] W02020 / 157211 described linkage of an Fc region or albumin-binding domain to an ImmTAC. Although these half-life extended molecules retain some function and would have the advantage of less frequent dosing compared with the original ImmTAC format, their larger size renders them less suitable for mediating formation of a tight immune synapse for cell killing. Efforts have been made to limit the loss of potency in ImmTAC molecules that incorporate a half life extending region. For example, WO2024 / 038193 described an alternative format of Fc- linked ImmTAC with reportedly higher potency than that in W02020 / 157211. WO2024 / 038198 described an Fc-linked ImmTAC in a coiled single chain arrangement, which was also reported to have higher potency than that in W02020 / 157211.

[0015] Importantly, the format of any therapeutic protein intended for medicinal use must also be suitable for industrial production - for example it may need to be efficiently expressed in a recombinant host cell, purified, formulated and stored. Recombinant expression in mammalian cells, especially Chinese Hamster Ovary (CHO) cells, is now the norm for polypeptide-based therapeutics. The polypeptide must be capable of high level expression in the manufacturing cell line to enable efficient recovery, must be readily obtainable at high levels of purity, and must remain stable under conditions used for pharmaceutical formulation and storage without aggregation or degradation. Aggregation of the protein should be avoided at all stages of its production, storage and use.

[0016] Research in the biopharmaceutical industry has been directed to designing a molecular format for a multispecific synaptic bridging molecule that delivers across all these areas.

[0017] Summary of the Invention

[0018] The present invention provides a therapeutic molecule that is able to promote formation of a strong immune synapse between a target cell and a T cell in a patient, and which has an extended in vivo half life.

[0019] The inventors engineered soluble multispecific proteins, which bind a target pMHC on a target cell and a surface receptor on an immune cell, and which are designed to fit within an immune synapse between the immune cell (e.g., T cell) and target cell (e.g., cancer or other disease-associated cell) without increasing the synaptic distance, despite incorporating a half life extending region.

[0020] The invention provides a soluble multispecific protein as set out in the appended claims. Soluble, multi-polypeptide multispecific proteins according to the invention are designed to recruit an immune cell to a target cell presenting a target peptide major histocompatibility complex (pMHC). Proteins according to the present invention may comprise a bispecific monovalent core structure composed of: a pMHC binding arm; an immune cell engager moiety (e.g., T cell engager (TCE)); and an Fc region. Proteins according to the present invention may comprise a bispecific monovalent core structure composed of: a TCR molecule (optionally consisting of a first hemichain (Va-Ca) and a second hemichain (Vp-Cp)); an immune cell engager moiety (e.g., T cell engager (TCE)); and an Fc region. The TCR binds pMHC on target cells.

[0021] The invention provides a multi-polypeptide protein comprising: a pMHC binding arm comprising a dimer of a first portion and a second portion, an immune cell engager moiety, and an Fc region comprising a dimer of a first Fc portion and a second Fc portion, wherein the first portion of the pMHC binding arm is linked at its N terminus to the immune cell engager moiety, and wherein the second portion of the pMHC binding arm is linked at its N terminus to a C terminus of the first Fc portion, and an N terminus of the Fc region is linked to a C terminus of the pMHC binding arm.

[0022] The immune cell engager arm recruits an immune cell to a target cell, e.g., it may recruit a cytotoxic T lymphocyte to destroy a target cell. For example, it may be an anti-CD3 scFv molecule with an N terminal VL domain joined by a peptide linker to a C terminal VH domain. Alternatively it may be an anti-CD3 scFv molecule with an N terminal VH domain joined by a peptide linker to a C terminal VL domain.

[0023] Figure 1 and Figure 2 illustrate embodiments of the structure. The protein has two polypeptide chains, each comprising a TCR hemichain, wherein the two TCR hemichains heterodimerise to form a TCR molecule, and wherein an N terminus and a C terminus of one TCR hemichain are connected respectively to a C terminus and an N terminus of a dimeric Fc region. The topology of the domain linkage directs folding of the polypeptide into a tertiary structure in which the TCR and Fc are antiparallel to each other, having a head-to-tail conformation in which the N terminal end of the Fc region is adjacent to the C terminal end of the TCR molecule, and the N terminal end of the TCR molecule is adjacent to the C terminal end of the Fc region. An immune cell engager moiety (e.g., anti-CD3 scFv) is fused to the N terminus of the other TCR hemichain. Figure 14 illustrates an embodiment having the same structure as described for Figure 1 and Figure 2. The immune cell engager moiety may comprise an scFv with an N terminal VL domain connected by a peptide linker to a C terminal VH domain (Figure 2), or may comprise an scFv with an N terminal VH domain connected by a peptide linker to a C terminal VL domain (Figure 14).

[0024] Figure 3 and Figure 4 illustrate further embodiments of soluble multispecific proteins, in which topology of the domain linkage directs folding of the polypeptide into a tertiary structure in which the pMHC binding arm (e.g., TCR molecule) and Fc are in the antiparallel conformation.

[0025] The multi-polypeptide protein comprises at least two (optionally, exactly two) polypeptide chains which may be referred to as first and second polypeptide. The terms “first” and “second” serve only to distinguish the molecular parts being referred to, and do not imply a hierarchy or relative order. The first and second polypeptide chains are separate polypeptide chains whose N and C termini are not inter-linked, although the two chains may be interconnected via one or more covalent bonds between side chains. Each polypeptide chain may thus have a free N terminus and a free C terminus. The first polypeptide may comprise scFv having a free N terminus and a TCR cp domain having a free C terminus, terminating at the C terminus of the cp domain, and the second polypeptide may comprise a first Fc chain having a free N terminus and a second Fc chain having a free C terminus, terminating at the C terminus of the CH3 domain.

[0026] The first and second portion of the pMHC binding arm dimerise, forming a pMHC binding arm comprising a binding site for pMHC. The two portions of the pMHC binding arm may be comprised within a single polypeptide chain, or in separate polypeptide chains. The first and second portion of the Fc region dimerise, forming the Fc region. The two portions of the Fc region may be comprised within a single polypeptide chain, or in separate polypeptide chains. In some embodiments, first and second portions of the pMHC binding arm are in separate polypeptides, and first and second portions of the Fc region are in the same polypeptide. In other embodiments, first and second portions of the pMHC binding arm are in the same polypeptide, and first and second portions of the Fc region are in different polypeptides.

[0027] The pMHC binding arm and the Fc region may be inter-linked with a variety of topologies to achieve the effect that they are held in an antiparallel orientation relative to one another in the folded protein structure. In the desired conformation, the pMHC binding arm and Fc are oriented side-by-side and head-to-tail relative to each other, with the N terminal region of the pMHC binding arm proximal to the C terminal region of the Fc, and the C terminal region of the pMHC binding arm proximal to the N terminal region of the Fc. Optionally, the N terminus of the first Fc portion is linked to the C terminus of the second portion of the pMHC binding arm, or is not linked to the pMHC binding arm and may be a free N terminus.

[0028] Optionally, the N terminus of the second Fc portion is linked to the C terminus of the first portion of the pMHC binding arm, or is linked to the second portion of the pMHC binding arm, or is not linked to the pMHC binding arm and may be a free N terminus.

[0029] The pMHC binding arm may comprise a free C terminus. The Fc region may comprise a free N terminus and a free C terminus.

[0030] A free N terminus may refer to an amino acid residue at the N terminal end of a polypeptide chain, where that N terminal residue has an amino group that does not form a peptide bond and thus terminates the polypeptide chain, in contrast to forming a peptide bond with a further amino acid residue, peptide linker or protein domain. The N terminal residue may thus have an unpolymerised amino group. A free C terminus may refer to an amino acid residue at the C terminal end of a polypeptide chain, where that C terminal residue has a carboxyl group that does not form a peptide bond and thus terminates the polypeptide chain, in contrast to forming a peptide bond with a further amino acid residue, peptide linker or protein domain. The C terminal residue may thus have an unpolymerised carboxylic acid group.

[0031] In a first arrangement, the N terminus of the second Fc portion is linked to the C terminus of the second portion of the pMHC binding arm. Embodiments of this first arrangement are illustrated in Figure 1, Figure 2 and Figure 4(i).

[0032] The protein may comprise a first polypeptide comprising a first TCR hemichain and a second polypeptide comprising a second TCR hemichain, wherein the first and second polypeptide heterodimerise via the pairing of the first and second TCR hemichain to form a TCR molecule for binding pMHC, wherein the first TCR hemichain is linked at its N terminus to an immune cell engager moiety, and wherein the second TCR hemichain is linked at its N terminus to a first Fc portion and is linked at its C terminus to a second Fc portion, wherein the first and second Fc portion dimerise to form an Fc region.

[0033] The first polypeptide may comprise, from N to C terminus, an immune cell engager antibody molecule, a linker, a first TCR variable domain and a first TCR constant domain, and the second polypeptide may comprise, from N to C terminus, a first Fc CH2-CH3, a linker, a second TCR variable domain, a second TCR constant domain, a linker, and a second Fc CH2- CH3. The first polypeptide may comprise, from N to C terminus, an immune cell engager VL- VH scFv antibody molecule, a linker, a TCR vp domain and a TCR cp domain, and the second polypeptide may comprise, from N to C terminus, a first Fc CH2-CH3, a linker, a TCR va domain, a TCR co domain, a linker, and a second Fc CH2-CH3. Alternatively the first polypeptide may comprise, from N to C terminus, an immune cell engager VH-VL scFv antibody molecule, a linker, a TCR vp domain and a TCR cp domain, and the second polypeptide may comprise, from N to C terminus, a first Fc CH2-CH3, a linker, a TCR va domain, a TCR ca domain, a linker, and a second Fc CH2-CH3.

[0034] In a second arrangement, the N terminus of the second Fc portion is linked to the C terminus of the first portion of the pMHC binding arm. Embodiments of this second arrangement are illustrated in Figure 3(a) and Figure 4(ii).

[0035] The protein may comprise a first polypeptide which comprises, from N to C terminus, the immune cell engager antibody molecule, a linker, the first portion of the pMHC binding arm (optionally comprising first TCR variable domain and first TCR constant region), and the second Fc portion (optionally comprising second CH2-CH3), and a second polypeptide which comprises, from N to C terminus, the first Fc portion (optionally comprising first CH2-CH2), a linker, and the second portion of the pMHC binding arm (optionally comprising second TCR variable domain and second TCR constant region).

[0036] The first polypeptide may comprise, from N to C terminus, immune cell engager VL-VH scFv antibody molecule, linker, TCR vp domain, TCR cp domain, and second Fc CH2-CH3, and the second polypeptide may comprise, from N to C terminus, first Fc CH2-CH3, linker, TCR va domain, and TCR ca domain. Alternatively the first polypeptide may comprise, from N to C terminus, immune cell engager VH-VL scFv antibody molecule, linker, TCR vp domain, TCR cp domain, and second Fc CH2-CH3, and the second polypeptide may comprise, from N to C terminus, first Fc CH2-CH3, linker, TCR va domain, and TCR ca domain.

[0037] In a third arrangement, the N terminus of the first Fc portion is linked to the C terminus of the first portion of the pMHC binding arm. Embodiments of this third arrangement are illustrated in Figure 3(b) and Figure 4(iii).

[0038] The protein may comprise a first polypeptide which comprises, from N to C terminus, the immune cell engager antibody molecule, a linker, the first portion of the pMHC binding arm (optionally comprising first TCR variable domain and first TCR constant region), the first Fc portion (optionally comprising first CH2-CH3), and the second portion of the pMHC binding arm (optionally comprising second TCR variable domain and second TCR constant region), and a second polypeptide which comprises the second Fc portion, optionally comprising second CH2-CH3.

[0039] The first polypeptide may comprise, from N to C terminus, immune cell engager VL-VH scFv antibody molecule, linker, TCR vp domain, TCR cp domain, first Fc CH2-CH3, TCR va domain, and TCR Co domain, and the second polypeptide may comprise second Fc CH2-CH3. Alternatively the first polypeptide may comprise, from N to C terminus, immune cell engager VH- VL scFv antibody molecule, linker, TCR vp domain, TCR cp domain, first Fc CH2-CH3, TCR va domain, and TCR Ca domain, and the second polypeptide may comprise second Fc CH2- CH3. Paired TCR constant domains are preferably linked by one or more disulphide bonds. Optionally, the Cp constant domain is covalently linked to the Ca constant domain by two diS bonds: (1) the native TCR diS bond in the C terminal region of the Ca and Cp, linking C95 in the alpha chain and C131 in the beta chain (IMGT numbering); (2) a non-native diS bond between a substituted C48a and C57p (Cys is substituted for Thr48 in the alpha chain, and Cys is substituted for Ser57 in the beta chain). Inclusion of the C95a-C13ip disulphide bond, which is naturally present in the human TCR constant region, assists in stabilising the native conformation of the TCR molecule. Thus the native diS is desirably included, optionally in addition to the C48a-C57p disulphide bond and / or further stabilising mutations of the native human TCR constant domain sequence. The stabilised structure facilitates production of multispecific proteins with diverse human TCR variable domains, comprising variable domains obtained from recombination of a variety of different human variable region gene segments. The example human TCR constant domain amino acid sequences SEQ ID NO: 18 (CP) and SEQ ID NO: 19 (Ca) include both the native and additional engineered diS bond.

[0040] Polypeptide domains within a chain may be linked directly, or indirectly e.g., via a peptide linker. Examples of linkers are shown in the appended sequence table.

[0041] Proteins of the present invention find use as therapeutic molecules for killing target cells in vivo, where the target cells (e.g., cancer cells, infected cells or other disease-associated cells) present a target pMHC. The multispecific proteins recognise the target pMHC and recruit an immune cell, promoting killing of the target cell by the immune cell. Binding of the multispecific protein to both the immune cell and the target cell, through its simultaneous engagement with the immune cell surface receptor (on immune cell) and the target pMHC complex (on target cell), is accommodated within the immune synapse (Figure 6). Thus the multispecific protein triggers the receptor of the immune cell specifically in the presence of a cell presenting the target pMHC, e.g., triggering pMHC-dependent killing of the target cell by a cytotoxic immune cell (e.g., cytotoxic T cell). The immune cell engager moiety binds a surface receptor of an immune cell (e.g., human immune cell). Binding of the immune cell surface receptor at an immune synapse activates intracellular signalling in the immune cell. In some embodiments, the immune cell is a T cell. The immune cell engager may bind the TCR-CD3 complex. It may bind CD3 polypeptide, e.g., CD3E, CD3y and / or CD35. It may bind the TCR, e.g., a TCR constant domain, although care should be taken that the immune cell engager arm does not recognise the pMHC binding arm, otherwise the multispecific protein will be prone to aggregation. Alternatively the immune cell engager arm may bind another activating surface receptor such as CD28. The immune cell engager arm may be an anti-CD3 or anti-CD28 antibody molecule. The immune cell may be a cytotoxic immune cell, optionally a cytotoxic T cell. The immune cell may be a cytotoxic CD8+ T effector cell. The cell may be a natural killer (NK) cell. Surface receptors on NK cells include Fc receptors such as FcyRIII (CD16). The immune cell engager arm may be an anti-CD16 antibody molecule.

[0042] Functionally, the immune cell engaging arm is an agonist of the surface receptor, thus it activates intracellular signalling in the immune cell. Binding to CD3 or CD28 activates intracellular signalling in a cytotoxic T cell, resulting in release of cytokines and cytolytic molecules, which kill the proximal target cell. Binding to an Fc receptor on an NK cell, e.g., FcyRIII, activates intracellular signalling in the NK cell, resulting in an ADCC response.

[0043] The immune cell engager comprises a binding site that recognises a surface receptor on the immune cell. The immune cell engager moiety may comprise an antibody molecule, e.g., an antibody scFv, such as an anti-CD3 scFv. Single chain antibodies, such as scFv and domain antibodies like VHH, have a compact form well-suited to inclusion in the multispecific protein.

[0044] Many examples of anti-CD3 antibodies, e.g., anti-CD3 scFv or dAb are known in the art. Examples include UCHT1v17 (SEQ ID NO: 6) and anti-CD3 scFv disclosed in WO2024 / 038183.

[0045] A multispecific protein is at least bispecific, having binding sites for at least two antigens, i.e. , the pMHC complex and the immune cell surface receptor. It may be trispecific, e.g., it may include an additional binding site for a second immune cell surface receptor. For example, the multispecific protein may bind CD3 and a further receptor, for instance CD28 and CD3, or CD8 and CD3. This additional specificity may be achieved by providing a multivalent, multispecific immune cell engager moiety, and / or by providing multiple immune cell engager moieties in the multispecific protein, optionally connected in series to an N terminus of the pMHC binding arm. These binding specificities of the multispecific protein are provided in addition to any binding capability of the Fc region, which may bind to one or more types of Fc receptor as discussed in more detail elsewhere herein.

[0046] Target cells are cells that present on their surface the target pMHC complex recognised by the pMHC binding arm of the multispecific molecule, and which are to be engaged by the immune cells in the present invention. In some embodiments, the target cell is a tumour cell, e.g., a malignant cell of a solid tumour such as sarcoma, carcinoma or lymphoma. The target pMHC may be preferentially expressed on disease-associated cells (e.g., tumour cells) relative to other tissues of the body. Its presentation may be restricted to such cells, i.e., the target pMHC may not be presented on other cells in the patients. Examples of tumour associated peptides that may be presented on pMHC include peptide fragments of PRAME (e.g., SLLQHLIGL presented on HLA-A*02:01) and MAGE-A4 (e.g., GVYDGREHTV presented on HLA-A*02:01).

[0047] Multispecific proteins of the invention are suitable for targeting pMHC at a wide range of copy number. Signalling at the immune synapse mediated by the multispecific protein may be effective even at low density of target pMHC. The multispecific protein may generate a potent immune cell response in the presence of target cells, including at low levels of pMHC presentation typical of cancer cells. Copy number of target pMHC per cell may be in the range of 1 to 10 or 1 to 100, e.g., 5 to 10, 5 to 50, or 50 to 100. On some cells, copy number of the target pMHC may be higher, even up to 1000 pMHC presented per cell. Copy number may be in the range of 10 to 1000, 50 to 1000 or 100 to 1000. Copy number may be in the range of 10 to 500, 50 to 500 or 100 to 500.

[0048] The pMHC binding arm comprises a binding site that recognises a target pMHC on a target cell. The pMHC binding arm may comprise a pair of first and second portions which dimerise, e.g., a heterodimer. The first and second portions are optionally located on separate polypeptide chains or form separate regions of the same polypeptide chain. Each of the first and second portions may comprise a variable domain and optionally a constant domain. Paired variable domains associate to provide a binding site for pMHC. Where the first and second portions each comprise a constant domain, the constant domains may also pair. Paired domains associate as a dimer, e.g., heterodimer, and may further be linked by one or more covalent bonds, for example di-sulphide bonds between cysteine side chains.

[0049] Examples of pMHC binding molecules are known, including TCRs and TCR-mimetic antibodies. Thus, the pMHC binding arm may comprise a TCR molecule, antibody molecule other binding molecule for binding pMHC on a target cell. Optionally the pMHC binding arm is a TCR molecule. Portions of a TCR molecule may be referred to as TCR hemichains. A TCR molecule may be a dimer of a first hemichain and a second hemichain, which may be a Va-Ca portion and a Vp-Cp which heterodimerise. Optionally the first TCR hemichain is Vp-Cp and the second TCR hemichain is Va-Ca. Alternatively the first TCR hemichain is Va-Ca and the second TCR hemichain is Vp-Cp. Alternatively the pMHC binding arm is an antibody molecule, which may be a dimer of a first portion and a second portion. The antibody molecule may comprise a heavy chain or fragment thereof, and a light chain or fragment thereof. For example the antibody molecule may be a Fab composed of a heavy chain (VH-CH1) and light chain (VL- CL). The antibody molecule may be a dimer of a first portion which comprises an antibody VH domain and optionally a constant region and a second portion which comprises an antibody VL domain and optionally a constant region. The antibody molecule may be a dimer of a first portion which comprises an antibody VL domain and optionally a constant region and a second portion which comprises an antibody VH domain and optionally a constant region. Further examples of TCR molecules and antibody molecules are provided elsewhere herein.

[0050] The TCR molecule comprises a binding site that recognises a target pMHC on a target cell. A TCR molecule may comprise all or part of a TCR extracellular domain comprising a pMHC binding site. It may comprise TCR variable regions (optionally a TCR alpha variable domain and a TCR beta variable domain) which pair to provide the pMHC binding site. Optionally the TCR molecule further comprises one or more TCR constant regions, e.g., it may comprise paired TCR variable domains and paired TCR constant regions. An example TCR molecule comprises a dimer of first and second TCR polypeptide chains, each comprising an N terminal variable domain and a C terminal constant domain, wherein the two variable domains pair to provide the pMHC binding site, and wherein the two constant domains pair, optionally wherein the two constant domains are inter-linked via one or more disulphide bonds. The TCR molecule may be a heterodimer of TCRa and TCRp polypeptide chains, e.g., a first hemichain comprising an N terminal Vp domain and a C terminal Cp domain and a second hemichain comprising an N terminal Va domain and a C terminal Co domain.

[0051] An antibody Fc (fragment crystallisable) region represents the tail region of an antibody molecule comprising antibody heavy chain constant regions CH2 and CH3, distal from the antigen-binding variable regions of the antibody. Fc has two polypeptide chains, each comprising CH2-CH3, which dimerise and are covalently linked by disulphide bonds. Naturally occurring or engineered forms of Fc may be utilised. Fc domains from immunoglobulin subclasses lgG1, lgG2 and lgG4 bind to and undergo FcRn mediated recycling, affording a long circulatory half-life (3 - 4 weeks), thus extending the half life of the multispecific protein of the invention. The interaction of IgG with FcRn has been localized in the Fc region covering parts of the CH2 and CH3 domains. The hinge (upper and / or lower hinge) region of the antibody may be included with the Fc region.

[0052] Examples of antibody Fc region sequences are known in the art. For example, first and second Fc regions may be as disclosed in WO2024 / 038183. The Fc region may be homodimeric, i.e., composed of a pair of first and second Fc portions of identical amino acid sequence, or heterodimeric, i.e., composed of a pair of first and second Fc portions of different amino acid sequence.

[0053] A further aspect of the invention relates to nucleic acid, e.g., DNA or RNA, encoding the soluble multispecific protein. Molecules of the invention, including nucleic acids and proteins, may be provided in isolated form, optionally with one or more pharmaceutically acceptable excipients.

[0054] The molecules may be provided for use in methods of treating the human or animal body by therapy. Where humans are to be treated, proteins herein are preferably human in order to reduce immunogenicity concerns, the immune cell receptor targeted is a human immune cell receptor, and the MHC is human MHC. Further aspects of the invention thus relate to methods of treating a condition in a patient (e.g., human patient), comprising administering the soluble multispecific protein or its encoding nucleic acid to the patient. The condition to be treated may be cancer, a solid tumour such as sarcoma, carcinoma or lymphoma, or an infection (e.g., viral infection, optionally a tumour-associated viral infection). Cells associated with the condition to be treated (e.g., cancer cells, tumour cells, or infected cells) present a target pMHC complex recognised by the pMHC binding arm.

[0055] A population of recombinant host cells (e.g., mammalian cells such as CHO or HEK cells) may be provided in vitro comprising nucleic acid encoding the soluble multispecific protein. One embodiment of the invention is a population of CHO cells comprising random genomic integration of first and second gene expression vectors encoding the first and second polypeptides of the multispecific protein respectively. The protein may be produced by a method comprising culturing the cells under conditions for expression of the protein, recovering the protein from the cell culture, and optionally purifying the protein by one or more additional steps. Purified protein may be formulated into pharmaceutical compositions comprising one or more excipients, and provided for delivery as described above.

[0056] Brief Description of the Drawings

[0057] Figure 1 shows a soluble multispecific protein according to the invention.

[0058] Figure 2 shows a soluble multispecific protein according to the invention.

[0059] Figure 3 shows alternative structures of multispecific proteins according to the invention.

[0060] Figure 4 illustrates topologies of multispecific proteins according to the invention. ICE is the immune cell engager moiety. The pMHC binding arm has two hemichains. The half life extender has two hemichains, (i), (ii) and (iii) represent examples of three different topological linkages between the pMHC binding arm hemichains and half life extender hemichains. Figure 5 illustrates a representation of engagement between pMHC of a target cell (above) and immune cell receptors comprising TCR, CD3 and CD8 on a T cell (below) at an immune synapse. The TCR engages the pMHC complex across the synapse between adjacent cell membranes. The class I MHC is composed of an alpha chain comprising domains a1, a2 and a3. N terminal domains a1 and a2 provide a binding site for peptide p. Beta 2 microglobulin (P2m) associates with the alpha chain. The TCR is composed of an a chain and a p chain. Each chain comprises an N terminal variable region, a constant region, a connecting region, a transmembrane region and a short cytosolic C-terminal tail. The TCR is complexed with CD3 polypeptides. Dimeric co-receptor CD8 of the T cell binds a3 domain of the MHC class I molecule. CD8 is a heterodimer of an a chain and a p chain, and is associated with intracellular signalling molecule Lek. The figure was created with BioRender.

[0061] Figure 6 shows a soluble multispecific protein according to the invention engaging a T cell and a target cell at an immune synapse.

[0062] Figure 7 shows plasma concentration time curve following administration of a multispecific protein to mice. A minimum of three animals were used per time point.

[0063] Figure 8 shows surface plasmon resonance sensorgrams for multispecific proteins binding to (A) CD3E5 (B) CD3yb and (C) target pMHC PRAME(425-433).

[0064] Figure 9 shows potency of TCR T cell engaging bispecific molecules in an in vitro killing assay with tumour cell lines Hs695t (A), SK-MEL-5 (B), A-375 (C), NCI-H1755 (D), OVCAR-3 (E) and COV318 (F).

[0065] Figure 10 shows a comparison of EC50 values for bispecific molecules in vitro killing assay with PRAME-positive cancer cell lines, where replicate assays were performed.

[0066] Figure 11 shows data from a study of efficacy of multispecific proteins in a mouse model of human melanoma (Hs695T in PBMC-reconstituted NSG mice). Mice were randomised two days before drug administration in treatment groups (n=5 each) with an average tumour size at treatment start of 109.1 mm3. Tumour volume of each group was normalised to first drug administration (t=0) and mean tumour volume with SEM is shown. Treatment groups: No Drug (empty circle), scFv-TCR 994 (filled square), TCR-diabody-Fc 580 (crossed square) and CPA665 (filled triangle), (a) Change in tumour volume over time for each treatment group; (b) individual mice tumour volumes with response criteria. Figure 12 shows a HPLC-SEC trace for multispecific protein of representative purity from the assessment of HEK-expressed protein reported in Example 8.

[0067] Figure 13 shows a HPLC-SEC trace for multispecific protein of superior purity from the assessment of HEK-expressed protein reported in Example 8.

[0068] Figure 14 shows a soluble multispecific protein according to the present invention.

[0069] Figure 15 shows data from a TDCC assay with bispecific proteins in TB2C format containing candidate anti-CD3 scFv (filled shapes), compared against TB2C with benchmark anti-CD3 antibody LICHT1 scFv (open circles) and a further positive control TCR-TCE (open squares).

[0070] Figure 16 shows in vitro killing assay data for bispecific proteins identified in Table E11-1.

[0071] Figure 17 shows in vitro killing assay data for bispecific proteins identified in Table E12-1.

[0072] Figure 18 shows in vitro T cell mediated tumour cell killing for TCR-TCE bispecifics containing B1-NY antibody cluster members and a comparator bispecific.

[0073] Figure 19 shows in vitro T cell mediated tumour cell killing for TCR-TCE bispecifics containing 5-51 antibody cluster members and a comparator bispecific.

[0074] Figure 20 shows (a) % human CD45 measured in blood of animals treated with TCR-TCE bispecific proteins and control mice and (b) graph of change in tumour volume over time during in vivo assessment of efficacy for TCR-TCE bispecific proteins.

[0075] Detailed Description

[0076] The molecular format disclosed herein represents a “platform” design, suitable to be produced containing any of a diverse selection of pMHC binding arms, including TCR molecules and antibody molecules of diverse sequence.

[0077] TCR molecules

[0078] In humans, the native T cell receptor comprises two transmembrane polypeptides, which may be an op pair or a y<5 pair, associated with the CD3 complex. Each of the two paired TCR polypeptide chains has an N terminal variable domain comprising three hypervariable complementarity determining regions (CDRs), a constant domain, and a connecting region linked to a transmembrane domain and C terminal cytosolic tail. TCR polypeptides lack a signal transduction domain, and instead signalling occurs through the CD3 components of an octomeric complex comprising TCRap:CD3b£:CD3Y£:CD3< , which assemble in a 1 : 1 : 1 : 1 stoichiometry. The CD3E, CD3y and CD35 subunits are transmembrane proteins each containing a single immune receptor tyrosine-based activation motif (ITAM), and CD3 is cytosolic and contains three ITAMs.

[0079] In a structure analagous to an antibody’s antigen-binding fragment (Fab) binding domain, the extracellular region of the TCR comprises two paired variable domains and two paired constant domains, and the pMHC binding site of the TCR is formed by a set of 6 CDRs, comprising three CDRs from one variable domain and three CDRs from the second variable domain, with the CDR3 loops being centrally located and primarily responsible for recognition of the peptide.

[0080] A TCR molecule used in the present invention may comprise a pMHC-binding region of a TCR. To provide the protein in soluble form, the TCR molecule does not comprise the transmembrane domain or cytoplasmic tail of a full length T cell receptor. The TCR constant region may be truncated to remove membrane-associated and cytoplasmic portions from the C- terminal end. The pMHC binding arm may comprise all or part of a TCR extracellular domain comprising a pMHC binding site.

[0081] It may comprise TCR variable regions (optionally a TCR alpha variable domain and a TCR beta variable domain) which pair to provide the pMHC binding site. The TCR molecule may further comprise one or more constant domains. A constant domain may be a constant domain of a TCRa chain (Co domain) or of a TCRp chain (Cp domain). In humans, there are two alternative Cp constant domains, Cpi and Cp2. A TCR Cp constant domain may be a Cpi or Cp2 domain, and is exemplified in embodiments herein as a Cpi domain. Alternatively a pMHC binding arm may comprise one or more antibody constant domains, which may be from an antibody light or heavy chain (e.g., CL, CH1 , CH2, CH3 or CH4). A pMHC binding arm may comprise paired TCR variable domains and paired TCR constant regions. An example pMHC binding arm comprises a dimer of first and second TCR polypeptide hemichains, each comprising an N terminal variable domain and a C terminal constant domain, wherein the two variable domains pair to provide the pMHC binding site, and wherein the two constant domains pair, optionally wherein the two constant domains are inter-linked via one or more disulphide bonds. The TCR molecule may be a heterodimer of TCRa and TCRp polypeptide chains, i.e. , a first polypeptide comprising an N terminal Va domain and a C terminal Ca domain and a second polypeptide comprising an N terminal Vp domain and a C terminal Cp domain. The amino acid sequence of the variable and constant domain may correspond to those found in nature, or they may contain one or more mutations relative to a natural protein. Such mutations may be made to increase the affinity of the pMHC binding domain for a given antigen. Additionally or alternatively, mutations may be incorporated to improve stability and manufacturability.

[0082] Additional mutations may be introduced into the amino acid sequence of a constant domain relative to a natural constant domain. The constant domains may also include residues, either naturally-occurring or introduced, that allow for dimerisation by, for example, a disulphide bond between two cysteine residues. One or both of a Co and Cp constant domain may contain mutations, substitutions or deletions relative to the native TRAC and / or TRBC1 / 2 sequences. Alpha and beta chain constant region sequences may be modified by truncation or substitution to delete the native disulphide bond between Cys4 of exon 2 of TRAC and Cys2 of exon 2 of TRBC1 or TRBC2. Alpha and / or beta chain constant region sequence(s) may have an introduced disulphide bond between residues of the respective constant domains, as described, for example, in WO 2003 / 020763, WO 2004 / 033685 and WO 2006 / 000830, and for example, in U.S. Patent Nos. 7,329,731, 7,569,664; and 8,361,794, the contents of each of which are herein incorporated by reference. Alpha and beta constant domains may be modified by substitution of cysteine residues at position Thr 48 of TRAC and position Ser 57 of TRBC1 or TRBC2, the said cysteines forming a disulphide bond between the alpha and beta constant domains of the TCR. TRBC1 or TRBC2 may additionally include a cysteine to alanine mutation at position 75 of the constant domain and an asparagine to aspartic acid mutation at position 89 of the constant domain. One or both of the extracellular constant regions present in an op heterodimer may be truncated at the C terminus or C termini, for example by up to 15, or up to 10, or up to 8 or fewer amino acids. The C terminus of an alpha chain extracellular constant region may be truncated by 8 amino acids. Unless otherwise specified, TCR sequences are described with reference to IMGT nomenclature which is widely known and accessible to those working in the TCR field. For example, see LeFranc and LeFranc "T cell Receptor Factsbook", Academic Press 2001; Lefranc, Cold Spring Harb Protoc (6):595-603 2011; Lefranc, Curr Protoc Immunol 2001 Appendix 1 : Appendix 100; and Lefranc, Leukemia 17(1): 260-2662003.

[0083] TCR constant domains may be engineered to include mutations to improve stability. For example, Froning et al., Nat Commn 11 2020 described 3 substitutions in Co and 4 substitutions in Cp, which were reported to increase stability: aT150l, aA190T, aS139F, PE134K, PH139R, PD155P, PS170D by Kabat numbering.

[0084] Other suitable TCR chain amino acid sequences are provided in WO2011001152, WO2017109496, W02017175006 and WO2018234319, and, for example, in U.S. Patent Nos.8, 519, 100, 11,639,374, 11,505,590, and 11,427,624, the contents of each which are herein incorporated by reference.

[0085] The TCR molecule may be a human TCR molecule. Where constant domains are present these are preferably human constant domains. An example human TCR beta constant domain is SEQ ID NO: 18. An example human TCR alpha constant domain is SEQ ID NO: 19.

[0086] Antibody molecules

[0087] In humans, the most common native format of an antibody is an IgG which is a heterotetramer consisting of two identical heavy chains and two identical light chains. The heavy and light chains are made up of modular domains with a conserved secondary structure consisting of a four-stranded antiparallel beta-sheet and a three-stranded anti-parallel betasheet, stabilised by a single disulphide bond. Antibody heavy chains each have an N terminal variable domain (VH) and 3 relatively conserved "constant" immunoglobulin domains (CH1 , CH2, CH3) while the light chains have one N terminal variable domain (VL) and one constant domains (CL). Disulphide bonds stabilise individual domains and form covalent linkages to join the four chains in a stable complex. The VL and CL of the light chain associates with VH and CH1 of the heavy chain and these elements can be expressed alone to form a Fab fragment. The CH2 and CH3 domains (also called the "Fc domain") associate with another CH2:CH3 pair to give a tetrameric Y shaped molecule with the variable domains from the heavy and light chains at the tips of the "Y". The CH2 and CH3 domains are responsible for the interactions with effector cells and complement components within the immune system.

[0088] Antibody variable domains have sequence variability in their complementarity determining regions (CDRs), which are loops primarily involved in antigen recognition. A VH domain comprises VH CDR1, VH CDR2 and VH CDR3 in a VH domain framework, and a VL domain comprises VL CDR1 , VL CDR2 and VL CDR3 in a VL domain framework. The bound antigen may make contact with a subset of residues within the six CDRs, for example in one or more particular CDRs such as the heavy chain CDR3 and / or light chain CDR3.

[0089] An antibody molecule used in the present invention comprises an antigen-binding region of an antibody. It may be an antibody fragment and / or a synthetic antibody construct. The antibody molecule may comprise an antibody VH and / or VL domain. It may be monovalent for binding to its antigen. The antibody molecule may be a Fab, consisting of a dimer of VH:CH1 and an antibody light chain (VL:CL). It may be a single chain antibody molecule. The antibody molecule may be a single chain Fv (scFv), consisting of VH and VL fragments fused by a flexible linker. In the scFv molecule, the VH and VL domains form a VH-VL pair in which the complementarity determining regions of the VH and VL come together to form an antigen binding site. Antibody fragments that comprise an antibody antigen-binding site include, but are not limited to, (i) the Fab fragment consisting of VL, VH, CL and CH1 domains; (ii) the Fd fragment consisting of the VH and CH1 domains; (iii) the Fv fragment consisting of the VL and VH domains of a single antibody; (iv) the dAb fragment, which consists of a VH or a VL domain; (v) isolated CDR regions; (vi) F(ab')2 fragments, a bivalent fragment comprising two linked Fab fragments (vii) scFv, wherein a VH domain and a VL domain are linked by a peptide linker which allows the two domains to associate to form an antigen binding site; (viii) a diabody, a multivalent or multispecific fragment constructed by gene fusion (WO94 / 13804). Fv, scFv or diabody molecules may be stabilised by the incorporation of disulphide bridges linking the VH and VL domains.

[0090] Various other antibody molecules including one or more antibody antigen-binding sites have been engineered, including for example Fab2, Fab3, diabodies, triabodies, tetrabodies and minibodies (small immune proteins). Antibody molecules and methods for their construction and use were described by Holliger & Hudson, Nature Biotechnology 23(9): 1126-1 136 (2005).

[0091] Other examples of binding fragments are Fab', which differs from Fab fragments by the addition of a few residues at the carboxyl terminus of the heavy chain CH1 domain, including one or more cysteines from the antibody hinge region, and Fab'-SH, which is a Fab' fragment in which the cysteine residue(s) of the constant domains bear a free thiol group.

[0092] A dAb (domain antibody) is a small monomeric antigen-binding fragment of an antibody, namely the variable region of an antibody heavy or light chain. VH dAbs occur naturally in camelids (e.g., camel, llama) and may be produced by immunizing a camelid with a target antigen, isolating antigen-specific B cells and directly cloning dAb genes from individual B cells. dAbs are also producible in cell culture. Their small size, good solubility and temperature stability makes them particularly physiologically useful and suitable for selection and affinity maturation. Camelid VH dAbs are being developed for therapeutic use under the name "nanobodies™".

[0093] The antibody molecule may be a human antibody molecule. Thus, where constant domains are present these are preferably human constant domains.

[0094] Linkage of domains

[0095] Amino acid sequences within one polypeptide chain are linked to each other directly by a peptide bond, or indirectly via a peptide linker and / or one or more intervening protein domains. Within a polypeptide chain, a C terminus of one amino acid sequence is linked to an N terminus of another amino acid sequence by a peptide bond.

[0096] N and C termini (of one polypeptide chain, or of different polypeptide chains) may alternatively be linked by covalent bonds between amino acid side chains, e.g., a disulphide bond between cysteine residues. In this situation the covalent bond is formed between a residue proximal to the N terminus (e.g., the N terminal residue or any of the first 5 residues of the N terminal sequence) and a residue proximal to the C terminus (e.g., the C terminal residue or any of the last 5 residues of the C terminal sequence). Besides disulphide bonds, other means of connecting sequences are known, such as leucine zippers, and any such joining means may be used to connect N and C termini.

[0097] In the context of an individual polypeptide, linkage between N and C termini typically refers to the N and C terminal amino acids being linked directly or by a peptide linker.

[0098] Examples of peptide linkers and their context are shown in sequences herein. A linker may consist of or comprise GGGGS (G4S) or multiples thereof, e.g., (G4S)2, (G4S)3, (G4S)4, (G4S)5 or (G4S)6. The linker may consist of or comprise GGGGSGGGG (G4SG4) or GGGGSGGGGG (G4SG5). Alternative peptide linkers adopt a helical or hairpin conformation. An example hairpin linker is a peptide consisting of or comprising GGRARVCENCMEMD. Chen, Zaro & Shen reviewed the properties, design and functionalities of linkers in recombinant fusion proteins (Adv Drug Deliv Rev 65(10): 1357-13692013). Peptide linkers may be selected or adapted from these or others known in the art, such as linkers described in WO2024 / 038183, incorporated herein by reference.

[0099] The skilled person will select linkers of suitable length and structure to permit or promote folding of the polypeptide chains into their desired tertiary structure and assembled quaternary structure. For example, linkage of the immune cell engager moiety to the pMHC binding arm is designed to allow dual engagement of the multispecific molecule, binding to both the target pMHC and the immune cell surface receptor, while maintaining a compact structure accommodated within the immune synapse (Figure 6). Linkage of the immune cell engager moiety to the N terminus of the first portion of the pMHC binding arm (e.g., first TCR hemichain, optionally v ) may be direct, whereby the N terminal residue of the pMHC binding arm forms a peptide bond to the C terminal residue of the immune cell engager moiety. Alternatively a peptide linker or other linker may be used to connect the ICE moiety to the pMHC binding arm. A peptide linker may be between 1 - 20 amino acids, e.g., 5 - 15 amino acids in length may be used. A short linker such as between 1 - 10 amino acids, optionally 3, 4, 5, 6 or 7 amino acids may be used. A flexible linker, optionally comprising or consisting of G and / or S residues, allows the immune cell engager moiety to orient its binding site toward the immune cell surface receptor while the pMHC binding arm is oriented with its binding site toward the target pMHC (Figure 6). A G4S, (G4S)2 or (G4S)3 linker is suitable. Alternative linkers for joining scFv to the pMHC binding arm include GGGSG, GGSGG, GSGGG, GSGGGP, GGEPS, GGEGGGP, GGEGGGSEGGGS, GGGSGGGG and linkers comprising one or more of the following sequence motifs: GGGS, GGGGS, TVLRT, TVSSAS and TVLSSAS. In some embodiments, a flexible linker of 8, 9, 10, 11 or 12 amino acids is preferred, e.g., (G4S)2. The linker at this position, connecting the immune cell engager moiety and pMHC binding arm, may be referred to herein as linker L1 (Figure 14).

[0100] Meanwhile, dual linkage of the Fc region to the pMHC at both their N and C termini ensures the desired antiparallel arrangement (Figure 4). The skilled person can select linkers of suitable length and amino acid composition to join the pMHC binding arm and Fc for antiparallel folding. Linkers of up to 20 amino acids, e.g., 5 - 20 or 5 - 15 amino acids are optionally used here. One or both of these linkers may be flexible, comprising or consisting of G and / or S residues, such as G4SG4 or G4SG5. One or both of these linkers may be a hairpin linker. Optionally the linker between the first Fc region and the second pMHC binding arm is a hairpin linker. For example, the second portion of the pMHC binding arm may be linked at its N terminus to a C terminus of the first Fc portion via a peptide linker, optionally G4SG5. An alternative peptide linker is a hairpin linker such as GGRARVCENCMEMD. An N terminus of the Fc region may be linked to a C terminus of the pMHC binding arm via a peptide linker, optionally G4SG4. With reference to the embodiments illustrated in Figure 1, Figure 2 and Figure 14, linkers connecting first and second Fc portions to the N and C termini of the pMHC binding arm may be referred to as L2 and L3 respectively. (G4S)3 is a preferred sequence for the L2 and / or L3 linker.

[0101] Within an scFv molecule, a peptide linker such as (G4S)3 may connect the two antibody variable domains. For example the scFv may comprise, from N to C terminus, VL-(G4S)3-VH. The scFv may alternatively comprise, from N to C terminus, VH-(G4S)3-VL. The linker connecting VH and VL within an scFv may be referred to herein as L0.

[0102] Orientation of antibody variable domains

[0103] WO2025 / 1333349 (Immunocore) described low MW (<60 kDa) TCR-TCE formats lacking constant domains. A panel of small (<60 kDa) four-domain bispecifics were constructed, each comprising an anti-CD3 antibody molecule (VH domain and VL domain) and a TCR molecule (alpha variable domain and beta variable domain). The bispecific T cell engagers were tested for potency in a T cell activation assay, and it was observed that arrangements in which the C terminus of the antibody VH domain was fused to the N terminus of the TCR v domain outperformed other formats. The same pattern was observed with TCRs for different pMHC targets, indicating that it was not TCR-dependent. The invention in that case was therefore claimed to lie in linking the VH domain of the immune cell engaging domain to the N terminus of the TCR beta variable domain. As disclosed herein, the preference reported in WO2025 / 1333349 for the orientation [VL-VH]-TCR, compared with [VH-VL]-TCR, is not universal. Instead, it depends on the particular immune cell engaging antibody in question.

[0104] The impact of variable domain orientation within the antibody Fv is demonstrated in Example 13. Surprisingly, while some anti-CD3 antibodies prefer the VH-VL orientation, other anti-CD3 antibodies prefer the VL-VH orientation, and for others the orientation of domain linkage has little or no effect. The orientation preference is reflected in potency measured for the multispecific binding protein in an assay measuring activity (e.g., killing) resulting from engagement of immune cells in the presence of target cells. Thus, with some antibodies, higher potency may be observed for a multispecific binding protein in which the VL domain of the immune cell engager Fv is linked to a terminus of the pMHC binding arm, compared with the same multispecific binding protein in which the positions of the VH and VL domains are interchanged so that the VH domain is linked that terminus of the pMHC binding arm.

[0105] The actual binding site of an antibody Fv is substantially unchanged by the addition of other moieties via peptide linkers at N or C termini, so the orientation of its connection to a pMHC binding arm should not affect the structure of the antigen binding site itself. The paratope remains the same. Nor is the antigen binding site generally affected by an interdomain linker between VH and VL, as the art is well aware of how to use linkers having a length and composition permissive of association between the VH and VL domain at their native interface to create scFv antibody molecules, and these scFv may be oriented from N to C terminus VH- <linker>-VL or VL-<linker>-VH without impacting antigen binding (although effects on expression or stability may be observed, leading to a preference for one or the other orientation for certain sequences).

[0106] Therefore the preference for domain orientation in the antibody Fv within a multispecific binding protein is not thought to be due to differences internal to the antigen binding site within the Fv. The preference for one domain orientiation over another is instead thought to arise in the context of certain multispecific binding proteins, where it impacts biological potency. Thus, the sequence in which domains are connected within a multispecific protein may affect the relative stability of spatial configurations of the different binding sites provided by those domains, which is especially relevant in geometrically and / or spatially constrained contexts such as the immune synapse.

[0107] The individuality of this effect with respect to different immune cell engaging antibodies may be a function of the epitope recognised by the particular immune cell engager and its angle of engagement with its epitope. An inter-domain linkage (and resulting 3D arrangement of binding sites) that is optimal for engagement of one epitope may be sub-optimal for engagement of a different epitope. Based on data presented herein, orientation of variable domains in anti-CD3 scFv did not appear to affect affinity for CD3 but did affect killing potency. This may be due to a change in how force is exerted by the antibody on the CD3 complex, therefore influenced by epitope and by whether the VH or VL domain of a given antibody is dominant for antigen binding.

[0108] As shown by data herein, the influence of domain orientation within the antibody Fv may be heightened when the Fv and the pMHC binding arm are closely connected (e.g., via a short peptide linker), as this constrains their relative flexibility, and therefore is especially relevant in multispecific formats where the binding sites for the immune cell receptor and pMHC are close together. Notably, it is exactly these formats that are most potent for triggering immune cells in the presence of target cells, as they permit (and may indeed promote) proximity between the membranes of the immune cell and target cell at the immune synapse. Examples include the formats of the present invention, which are designed to bridge an immune synapse with minimal disruption of the intercellular distance and tightly connect CD3 and pMHC (Figure 6).

[0109] Thus, for some VH / VL domain pairs, the bispecific format VH-L0-VL-L1-<pMHC binding arm> has greater potency than VL-L0-VH-L1-<pMHC binding arm>, believed to reflect enhanced activity at the immune synapse. While demonstrated herein for anti-CD3, the same would in principle apply to other immune cell engaging antibodies (e.g., anti-CD8, anti-CD28) in multispecific binding proteins designed to form a close intercellular connection.

[0110] A multispecific binding protein may comprise an immune cell engager antibody molecule (e.g., anti-CD3) and a pMHC binding arm, wherein the immune cell engager antibody molecule comprises a VH domain and a VL domain that pair to form an Fv comprising a binding site for an immune cell surface receptor, and wherein the pMHC binding arm binds a target pMHC, wherein the C terminus of the VL domain is linked to an N terminus of the pMHC binding arm by a peptide linker L1 , and wherein the multispecific binding protein has higher potency for activating immune cells in the presence of cells presenting the target pMHC compared with a multispecific binding protein in which the positions of the VH and VL domains are reversed so that the C terminus of the VH domain is linked to the said N terminus of the pMHC binding arm.

[0111] The multispecific binding protein may be of any format according to the invention as described herein, such as the “TB2C” format illustrated in Figure 1 or Figure 14.

[0112] For the purpose of potency comparison, the multispecific binding protein is compared against a multispecific binding protein that is identical save for the interchange of the VH and VL domain positions.

[0113] The multispecific binding protein may comprise an anti-CD3 scFv antibody molecule and a pMHC binding arm, wherein the C terminus of the scFv is linked to an N terminus of the pMHC binding arm by a peptide linker L1, wherein the anti-CD3 scFv comprises an N terminal VH domain and a C terminal VL domain joined by a peptide linker LO, and wherein said multispecific binding protein has higher potency compared with a multispecific binding protein in which the orientation of VH and VL domains in the scFv is reversed.

[0114] Multispecific binding proteins comprising antibody molecules that prefer the VH-VL orientation may be identified by testing different orientations and selecting based on potency. Suitable methods are described elsewhere herein.

[0115] Engagement of immune cells

[0116] The immune cell engager moiety represents means for binding and activating an immune cell surface receptor. It may be a binding protein such as an antibody molecule. It may be a binding domain of a natural ligand of the immune cell surface receptor, or all or part of an extracellular region of a natural ligand that is a transmembrane protein. A large number of immune cell engaging antibodies are known in the art and may be adapted for use in multispecific proteins, or further such antibodies may be generated as desired.

[0117] The immune cell engager moiety may bind a surface receptor on a T cell, e.g., cytotoxic CD8+ or CD4+ T cell, inducing intracellular signalling in the T cell leading to T cell killing of the target cell. The immune cell engager may bind and activate the TCR-CD3 complex on the T cell. It may be an antibody molecule comprising the VH and VL domains of BMA031 (e.g., as described in EP0403156) or a variant thereof such as BMA031v36 (WO2021 / 023657, WO2021 / 023658). The immune cell engager moiety may bind CD3, e.g., CD3E, CD3y and / or CD35. It may bind CD35E heterodimer and / or CD3y£ heterodimer. The immune cell engager moiety may be an anti-CD3 antibody molecule. Trinklein et al (MABS 11(4):639-652 2019) evaluated anti-CD3 antibodies for use in T cell engaging bispecific antibodies and described anti-CD3 antibodies of “F2” clonotype, which induced tumour cell killing by T cells but with a relatively low accompanying level of cytokine release. Robertson et al (Clinical and Experimental Immunology 215:105-1192024) also reviewed properties of anti-CD3 antibodies suitable for use in immune cell engaging multispecific proteins, and found that optimal affinity binding to CD3 is a function of the affinity of pMHC binding in the bispecific. In bispecifics comprising TCRs with picomolar affinity for pMHC, an anti-CD3 of moderate affinity was optimal, whereas in bispecifics with weaker TCRs (lower affinity for pMHC), a stronger anti-CD3 affinity was required for comparable potency.

[0118] An anti-CD3 antibody molecule may comprise the anti-CD3 VH and VL domain of any of the following: OKT3, UCHT1, UCHT1v17 (WO2021 / 023657), L2K, TR66, 12F6, SP34 (WO2016 / 110576), brenetafusp, tebentafusp, blinatumomab, catumaxomab, duvortuxizumab, epcoritamab, ertumaxomab, glofitamab, mosunetuzumab, odronextamab, pasotuxizumab, solitomab, or an engineered variant thereof. An anti-CD3 antibody molecule may comprise an anti-CD3 VH and VL domain disclosed or referenced within WO2024 / 038193 (for example scFv U0 or scFv U28), WO2022 / 233956, WO2021 / 023658, US2021 / 0032370, WO2021 / 023657, W02020 / 157210, WO2018 / 052503, WO2016 / 110576, WO2015 / 095392 or W01992 / 022653.

[0119] Although bispecific antibodies with anti-CD3 specificity were originally developed to recruit CD8+ CTL to kill tumour cells, other immune cells such as y<5 T cells, natural killer T (NKT) cells and CD4+ T cells also express CD3 and could be activated for target cell killing.

[0120] The immune cell engager moiety may bind the TOR, e.g., a TOR constant domain. WO2024 / 081381 (Marengo) described a multispecific protein for redirecting T cells, in which the T cell engaging arm bound to the TOR beta variable domain.

[0121] In other embodiments, the immune cell engager moiety binds CD28. CD28 is a stimulatory receptor expressed on T cells. Binding of CD28 induces intracellular signalling and activates T cell killing of the target cell. The immune cell engager moiety may be an anti-CD28 antibody molecule. Examples of anti-CD28 antibodies have been described (Correnti et al., Leukemia 32:1239-12432018; Wu et al., Nature Cancer 1(1):86-982020).

[0122] In other embodiments, the immune cell engager moiety binds a surface receptor on a natural killer (NK) cell, inducing intracellular signalling and activating NK cell killing of the target cell. Surface receptors on NK cells include Fc receptors such as FCYRIH (CD16), e.g., FcyRIHA (CD16A). The immune cell engager moiety may be an anti-CD16 antibody molecule, e.g., anti- CD16A. It may be an antibody molecule comprising the anti-CD16 VH and VL domain of AFM13 (Rothe et al., Blood 125(26):4024-4031 2014) or 2B1 (Weiner et al., Cancer Res 55(20):4586- 4593 1995), or an engineered variant thereof. CD2 is another activating immune cell surface receptor, and is found on NK cells and T cells.

[0123] The immune cell engager moiety may bind a surface receptor on an invariant natural killer T (iNKT) cell (also known as type I NKT cell). It may bind a TCR Va24 (e.g., Va24Ja18) variable domain and / or a TCR Vpi 1 variable domain e.g., a TCR alpha variable domain Va24Ja18 paired with a TCR beta variable domain Vpi 1. It may be an iNKT engager as described in WO2025 / 093867 (Imperial College Innovations).

[0124] Induction of intracellular signalling in the immune cell occurs in the context of engagement at the immune synapse, wherein the multispecific molecule binds the target cell and the immune cell within the immune synapse. Binding to the immune cell surface receptor activates intracellular signalling in the immune cell. A cytotoxic cell is activated to kill the proximal target cell. An activated cytotoxic cell (e.g., CTL or NK cell) releases cytokines and cytolytic molecules leading to target cell death. Released cytokines also recruit and activate other immune cells, amplifying the immune response. In other embodiments, the immune cell surface receptor is an inhibitory receptor. It may be an immune checkpoint molecule. Binding and activation of immune checkpoint molecules has a suppressive effect on the immune cell. Immune cell engager moieties directed to inhibitory receptors such as checkpoint inhibitors are used in embodiments where dampening of an immune response is desired, e.g., for treatment of autoimmune conditions to reduce immune activation against cells presenting a target pMHC complex and / or to reduce immune activation within tissues in which the target pMHC complex is presented.

[0125] Examples of immune checkpoint molecules, and agonists that specifically bind them, are known in the art (Paluch et al. Front Immunol 9:20362018) and such agonists may be used as immune cell engager moieties in embodiments of the present invention.

[0126] For example, the immune cell surface receptor may be PD-1, BTLA, TIGIT, TIM-3, CD200, CD200R / TGFPR, VISTA, LAG-3, CTLA-4, A2AR, A2BR, B7-H3, B7-H4, BTLA, IDO, NOX2, CD94 / NKG2A, KIR, LAIR-1 , SIGLECs (e.g., CD22 / SIGLEC-2, SIGLEC-5, SIGLEC-7, SIGLEC-9, SIGLEC-10), CD72, LILRB, LILRB1 , LILRB2, LILRB3, FcyRIIB, SIRL-1, CEACAM1, CD47 or SIRPa. CTLA-4, PD-1, LAG-3, TIGIT, BTLA, VISTA and TIM-3 are found on T cells. KIRs, LAIR-1, CD94 / NKG2A, LILRB1 and SIGLEC-7 are found on NK cells. CD72, CD22, SIGLEC-10, LILRB and FcyRIIB are found on B cells. LAIR-1, SIRL-1, SIGLEC-5, LILRB2, LILRB3, CEACAM1, CD47 and SIRPa are found on myeloid cells (e.g., monocytes, neutrophils, macrophages).

[0127] Cell killing activity of a soluble antigen binding protein can be measured in an in vitro assay using target cells and T effector cells. An example assay comprises providing target cells (e.g., an engineered cell line, or primary cells) presenting target antigen (e.g., target pMHC), providing T effector cells (e.g., PBMC or isolated CD8+ T cells), incubating the cells with a dilution series of the candidate antigen binding protein, detecting killing of the target cells by the T effector cells, and calculating EC50 for the killing of the target cells.

[0128] Activity may be determined in a cell killing assay in which target cells are contacted with a test molecule, and death of the target cells is measured and compared against death of target cells in a control condition where they are not contacted with the test molecule.

[0129] Preferably, a dose curve is obtained in which the 2 highest and 2 lowest points have reached a plateau.

[0130] For a given response (e.g., %-cytotoxicity) measured across a dose series, the EC50 is the inferred dose value at which 50% of the maximum response would be achieved. It can be inferred through fitting of a sigmoidal dose-response curve: where C_Max is the maximum response achieved, C_Min is the minimum response achieved, and the Hill slope is a fitted parameter dictating the ‘steepness’ of the sigmoid.

[0131] Target cells may be a cell line in which there is endogenous expression of the target antigen, or cells that are engineered to express the target.

[0132] Target cells may be a fluorescent cell line (e.g., A375-GFP), allowing cell killing to be measured as decrease in fluorescence. A suitable ratio of effector to target cells is 5:1. Cells may be co-cultured in a 1:1 mix of RPMI and DMEM.

[0133] Further examples of assays and protocols for determining killing potency are provided elsewhere herein.

[0134] Fc region

[0135] Naturally occurring or engineered Fc region sequences may be incorporated in binding proteins. Fc domains from immunoglobulin isotypes lgG1, lgG2 and lgG4 bind to and undergo FcRn mediated recycling, affording a long circulatory half-life (3 - 4 weeks), thus extending the half life of the protein that includes them.

[0136] An Fc region is preferably a human Fc region, and may be selected from any human antibody isotype, with human IgG (e.g., IgG 1 ) being preferred. It may be wild type human Fc or may be engineered to include mutations from wild type.

[0137] An antibody hinge (upper and / or lower hinge) region may be included with the Fc region. This may be a full antibody hinge region, or may be truncated, e.g., it may be a hinge region comprising a truncated upper hinge. The hinge region may be of the same isotype as the Fc. Where the upper hinge is included, the first Cys may be changed to Ser. As described in W02002 / 072605, this substitution reduces aggregation in molecules lacking a light chain. Alternatively, an upper hinge may be truncated to delete the N terminal residues up to and including the first Cys.

[0138] The Fc region of antibodies is recognised by Fey receptors and determines the ability of the antibody to mediate cellular effector functions, including antibody-dependent cell-mediated cytotoxicity (ADCC) activity, complement dependent cytotoxicity (CDC) activity and antibodydependent cell phagocytosis (ADCP) activity. These cellular effector functions involve recruitment of cells bearing Fc receptors to the site of the target cells, resulting in killing of the antibody-bound cell.

[0139] In the context of the present invention, immune cells may become coated by the multispecific protein (e.g., via binding of anti-CD3 antibody molecule to CD3 on T cells), and it is desirable to avoid cellular effector functions such as ADCC, ADCP and / or CDC from being directed against such immune cells (e.g., T cells). Therefore, Fc region is preferably one with low or no effector function. An Fc region may be selected that does not mediate ADCC, ADCP and / or CDC. It may be effector null.

[0140] A multispecific protein may comprise an Fc region that binds one or more types of Fc receptor but does not induce cellular effector functions, i.e., does not mediate ADCC, CDC or ADCP activity. Such a constant region may be unable to bind the particular Fc receptor(s) responsible for triggering ADCC, CDC or ADCP activity.

[0141] An Fc region with no or low Fc effector function may be chosen to reduce the risk of antigen-independent activation of immune cells. Human lgG4 naturally has low effector function. lgG4 may be engineered to comprise a Leu235Glu mutation (“E” mutation), referred to as “lgG4-E”. lgG4 may also be engineered to comprise a Ser228Pro (“P” mutation), which increases stability by reducing Fab arm exchange. A heavy chain constant region may be an lgG4 comprising both the Leu235Glu mutation and the Ser228Pro mutation, “lgG4-PE”.

[0142] An Fc region may be engineered to limit interaction with Fey receptors, reducing effector functions such as binding to FCYRIH and / or ADCC. Examples of Fc positions that may be substituted to reduce binding to an activating FcyR and thus to reduce effector function are positions 214, 233, 234, 235, 236, 237, 238, 265, 267, 268, 270, 295, 297, 309, 327, 328, 329, 330, 331 and 365 (Ell numbering is generally used to refer to residues of an Fc region). Exemplary substitutions that may be made singularly or in combination are substitutions K214T, E233P, L234V, L234A, deletion of G236, V234A, F234A, L235A, G237A, P238A, P238S, D265A, S267E, H268A, H268Q, Q268A, N297A, A327Q, P329A, D270A, Q295A, V309L, A327S, L328F, A330S and P331S in lgG1, lgG2, lgG3 or lgG4.

[0143] For example an Fc region may be a human IgG 1 Fc region comprising mutations to reduce binding to FcyR, such as “LALA” mutations (L234A, L235A) and / or P329G substitution. Wilkinson et al described Fc engineered antibodies with immune effector functions completely abolished (PLoS One 16(12):e0260954 2021).

[0144] An Fc region may be engineered to modulate its in vivo half life, e.g., by influencing binding to FcRn. The interaction of IgG with FcRn has been localised to parts of the CH2 and CH3 domains. Examples of positions linked with half life and / or FcRn binding in IgG are residues 250, 252, 253, 254, 256, 257, 307, 376, 380, 428, 434 and 435. Examples of Fc engineering to increase half life include “YTE” mutations (Dall’Acqua, Kiener & Wu, JBC 281 (33):23514-235242006 and W002 / 060919). The triple mutation YTE is a substitution of 3 amino acids in the IgG CH2 domain, these mutations providing tyrosine at residue 252, threonine at residue 254 and glutamic acid at residue 256. The YTE modification increases the half-life of the antibody compared with the half life of a corresponding antibody having a human CH2 wild type domain. Other exemplary singular or combination substitutions that may be made to increase the half-life are substitutions M428L / N434S, M252Y / S254T / T256E, T250Q / M428L, N434A and T307A / E380A / N434A. Exemplary singular or combination substitutions that may be made to reduce half life are mutations H435A, P257I / N434H, D376V / N434H, M252Y / S254T / T256E / H433K / N434F, T308P / N434A and H435R.

[0145] An Fc region may be homodimeric, i.e., composed of a pair of first and second Fc portions of identical amino acid sequence, or heterodimeric, i.e., composed of a pair of first and second Fc portions of different amino acid sequence.

[0146] A heterodimeric Fc may be engineered to include mutations that promote heterodimerisation, such as knobs-into-holes mutations or charge-pair mutations, which favour pairing of the two different Fc portions and reduce the propensity for Fc homodimerisation. In the context of asymmetric multispecific molecules that dimerise via an Fc region, it is advantageous to include such mutations in order to promote correct molecular assembly and reduce the formation unwanted homodimers. Scientists at Genentech described substitutions in CH3 to create an “knobs-into-holes” engineered Fc, where a CH3 domain comprising a “knob” T366W substitution heterodimerised with a CH3 domain “hole” substitutions T366S, L368A and Y407V, to sterically favour formation of a heterodimeric Fc rather than a hole-hole or knob-knob Fc homodimer (Ridgway JB, Presta LG, Carter P. Protein Eng 9:617-21 1996; Atwell et al., J Mol Biol. 270:26-35 1997; Merchant et al., Nature Biotech 16:677-681 1998).

[0147] Further examples of IgG-like molecules with complementary CH3 or other domains engineered to drive heterodimerisation include Strand Exchange Engineered Domain body (SEEDbody) molecules (W02007 / 110205 EMD Serono), bivalent bispecific antibodies (W02009 / 080254 Roche), charge-pair / electrostatically-matched molecules (W02009 / 089004 Amgen; US201000155133 Chugai; W02010 / 129304 Oncomed), FcAAdp molecules (WO2010 / 15792 Regeneron), DIG-body and PIG- body molecules (WO2010 / 134666 and W02014 / 081202 Pharmabcine), CrossMAbs (WO2011 / 117329 Roche), Triomab / Quadroma molecules (WO2011 / 069104), mAb-Fv molecules (WO2011 / 028952 Xencor), DuoBody® molecules (WO2011 / 131746 and WO2013 / 060867 Genmab A / S), LUZ-Y molecules (Wranik et al. J. Biol. Chem. 287(52):43331-4333992012), bispecific lgG1 and lgG2 (WO11 / 143545 Pfizer / Rinat), “Azymetric” scaffolds (WO2012 / 058768 Zymeworks / Merck), Biclonics (WO2013 / 157953 Merus) and Neoimmunetech CH3 engineering (W02025 / 010272).

[0148] Other Fc engineering features may be included to enhance developability and / or facilitate purification in industrial manufacture.

[0149] An additional disulphide bond may be introduced between CH3 domains, for example via substition S354C in one Fc portion and substitution Y349C in another Fc portion (Merchant et al., Nature Biotech 16:677-81 1998). Wei et al. (OncoTarget 8(31):51037-51049 2017) described including such an additional engineered disulphide bond (S354C / Y349C) in combination with knobs-into-holes mutations and a further heterodimerisation-driving mutation pair (K409A / F405K), to provide a “KiHSS” Fc portion comprising a knob portion with substitutions S354C, T366W, K409A and a hole portion with substitutions Y349C, T366S, L368A, Y407V, F405K.

[0150] When a C terminal lysine residue of an Fc region is present, it may be removed in vivo by endogenous circulating carboxypeptidases. Incomplete C-terminal Lys processing in recombinant production of antibodies causes charge heterogeneity, with 3 species possessing 0, 1 , or 2 C-terminal Lys residues respectively (Harris RJ, J Chromatogr A 705:129-34 1995). This may be controlled to some degree by process conditions. Alternatively, C terminal lysine and / or glycine of an Fc region may be genetically deleted to reduce product heterogeneity (e.g., G446del and / or K447del). Thus, an Fc region may lack a C terminal lysine residue, and may optionally lack a C terminal glycine-lysine dipeptide. Optionally, the C terminal lysine, or glycinelysine, is deleted only from the Fc portion that forms the C terminus of a polypeptide, while an Fc portion that is linked at its C terminus to a further domain may be untruncated at its C terminus.

[0151] Binding proteins may be produced with heterodimeric Fc that are differentially modified in the CH3 domain to alter their affinity for binding to a purification reagent such as Protein A, e.g., as described in W02010 / 151792. The CH3 of one Fc portion may comprise the mutation His435Arg and / or Tyr436Phe while the CH3 of the other Fc portion lacks said mutation(s). Protein A binding may thus be modulated using substitutions 435R and / or 436F. Alternatives include Q311R, Q311K, T307P / L309Q, T307P / V309Q, T307P / L309Q / Q311 R or T307P / V309Q / Q311 R (WO2018 / 224951).

[0152] Fc regions may also comprise pl engineering mutations to facilitate purification.

[0153] Multispecific binding proteins of the present invention may comprise Fc regions with any of the above features and formats, as desired. Examples of Fc sequences are shown herein. For example, a first Fc portion may comprise human IgG 1 Fc CH2-CH3 with one or more knob mutations (e.g., T366W, and optionally K409A), and a second Fc portion may comprise human lgG1 Fc CH2-CH3 with one or more hole mutations (e.g., T366S, L368A and Y407V, and optionally F405K). The first Fc portion is optionally untruncated at its C terminus, therefore including C terminal lysine. The second Fc portion may have the C terminal lysine deleted. First and second Fc portions may comprise human IgG 1 hinge sequences in which the free cysteine is mutated to serine. First and second Fc portions may comprise the “LALA” mutation L234A L235A, and optionally P329G mutation. First and second Fc portions may be engineered to comprise an additional disulphide bond between S354C (optionally in the first Fc portion) and Y349C (optionally in the second Fc portion). Example first and second Fc sequences are provided in the appended sequence table. Half-life

[0154] The Fc region of the multispecific protein confers an extended half life on the protein in vivo. Half life is the length of time taken for the concentration of the protein to decrease to half of its starting dose in the body. Half life of a molecule may be determined by methods known in the art. Concentration is conveniently measured as serum / plasma concentration in peripheral blood, representing half life of the molecule in circulation.

[0155] In vivo half life of a molecule may be measured using standard pharmacokinetic (PK) studies.

[0156] For example, according to the method of Kim et al., Eur J of Immunol 24:542 1994, radiolabelled protein is injected intravenously into mice and its plasma concentration is periodically measured as a function of time, for example at 3 minutes to 72 hours after the injection.

[0157] Alternatively, unlabelled protein can be injected and its plasma concentration periodically measured using an ELISA. The clearance curve thus obtained should be biphasic, that is, an alpha phase and beta phase. For the determination of the in vivo half life of the protein, the clearance rate in beta phase is calculated and compared with that of the wild type or unmodified protein.

[0158] An example PK study in mice is reported in Example 4. Terminal half life is observed after the initial distibution phase of the drug after administration. As exemplified herein this phase was considered to begin after 24h. A linear decay (Kel) is fitted on a log plasma concentration-time curve. Half life is then derived mathematically from that slope as ln(2) / Kel.

[0159] Terminal half life of multispecific proteins in mice is preferably >5 days. The half life may for example be in the range 5 - 10 days, e.g., around 5, 6, 7, 8, 9 or 10 days.

[0160] The multispecific protein may have a terminal half life similar to comparator Fc-fusion molecules such as the “TCER” format described by Immatics and / or comparable to that of an IgG antibody, e.g., as determined in humans and / or in mice.

[0161] Nucleic acids and expression in recombinant host cells

[0162] Nucleic acid constructs encoding a protein according to the present invention may be in the form of linear or circular nucleic acid molecules, plasmids, vectors or expression cassettes, and are optionally provided in isolated form in vitro. The nucleic acid may be provided in isolated form as a purified composition, optionally in sterile aqueous solution or frozen as glycerol stocks.

[0163] Nucleotide sequences encoding the protein may be operably linked to regulatory elements such as a promoter, for expression in a host cell or in a cell-free system. A promoter comprises the transcriptional regulatory sequences of a gene, including the TATA box or initiator element, for initiation of transcription. The encoding nucleotide sequence is operably linked to its promoter, meaning that the promoter is positioned relative to the coding sequence such that expression of the coding sequence is controlled by the promoter. Nucleic acid comprising the encoding nucleotide sequence may also include additional regulatory elements, such as activating sequences, transcription factor binding sites, enhancers and silencers, which modulate expression of the coding sequence.

[0164] The encoding nucleic acid may be provided in a vector. Suitable vectors can be constructed, containing appropriate regulatory sequences, including promoter sequences, terminator sequences, polyadenylation sequences, enhancer sequences, marker genes (e.g., genes for antibiotic resistance or other selectable marker) and other sequences as appropriate. Nucleic acid may comprise a nucleotide sequence encoding a polypeptide chain of the multispecific protein, wherein an N-terminal signal peptide is encoded at the 5’ end of said nucleotide sequence. Various signal peptides are known and the skilled person may select a suitable signal peptide to direct the protein to the secretory pathway of its expressing host cell. The signal peptide is cleaved by signal peptidase within the cell, leaving a mature polypeptide chain. Within a eukaryotic cell, mature polypeptide chains are assembled into a multi-chain protein within the endoplasmic reticulum, and the protein may be glycosylated, before being secreted from the host cell into the cell culture medium.

[0165] Nucleic acid may comprise a transcription termination sequence 3’ of a nucleotide sequence encoding a polypeptide chain of the multispecific protein.

[0166] Individual polypeptide chains of a multi-chain protein may be encoded on separate nucleic acid molecules (e.g., separate vectors) or on the same nucleic acid molecule (e.g., a single vector).

[0167] Nucleic acid sequences may be optimised for compatibility with codon usage of a host cell.

[0168] Proteins of the invention can be expressed from their encoding nucleic acid in host cells. Suitable cells include eukaryotic (e.g., mammalian) cell lines, such as CHO cells (e.g., CHO-K1 cell line), HEK cells (e.g., HEK293) and NSO cells (ECAAC 85110503).

[0169] Nucleic acid may be introduced into a host cell to provide a transformed host cell encoding the protein. For example, host cells may be transiently or stably transfected with nucleic acid to provide cells capable of expressing the protein. A host cell may comprise the nucleic acid integrated within its genome, e.g., within a chromosome, wherein it is optionally randomly integrated, or it may stably or transiently maintained extra-chromosomally within the cell. Methods of introducing nucleic acid into host cells include calcium phosphate transfection, DEAE-Dextran, electroporation, liposome-mediated transfection and transduction using retrovirus or other virus. Where different polypeptide chains of a protein are encoded on separate vectors, the multiple vectors may be introduced together (e.g., co-transfected) into the cells, so that a cell comprises nucleic acid encoding the multiple polypeptide chains and is capable of expressing the assembled protein. Optionally, host cells are co-transfected with first and second gene expression vectors encoding first and second polypeptide chains of the protein, e.g., in a 1:1 ratio. Transfection types, methods and strategies were reviewed by Chong, Yeap & Ho, PeerJ 9, 2021.

[0170] Following transfection, the encoding nucleic acid from the first and second vectors may become stably integrated into the host genome at different loci and / or at different copy number. Integration may be random. Where separate nucleic acid molecules encode the different polypeptide chains, these may integrate into the host genome independently of each other, and be independently transcribed and translated. Differences in integration site and / or copy number can give rise to different levels of expression of the first and second polypeptide chains relative to each other. The transfected cell population will therefore comprise cells representing a range of different ratios of expression of the different polypeptide chains. Where one polypeptide chain is slower to translate and / or to fold compared with another, this may be compensated for in a cell that expresses more of that polypeptide, giving overall balance between availability of the first and second polypeptide chains for assembly into the two chain protein. Transfection of cells with nucleic acid encoding the first and second polypeptide chains on separate vectors may therefore produce some cells that express the two chain protein at higher yield and / or higher purity. Individual cells or minipools of cells with relatively high expression of the multi-chain polypeptide may be selected from the transfected population and expanded. Expression of the multi-chain protein in selected cells may produce higher yield of monomer (non-aggregated) protein, compared with cells in which coding sequences for the first and second polypeptides are transcriptionally linked, or compared with production of a single polypeptide chain composed of domains that translate and / or fold at different rates.

[0171] A host cell that has integrated the encoding nucleic acid within its genome may be selected and expanded to provide a clonal population. Selection for cells that have taken up the nucleic acid may be achieved by the inclusion of a selectable marker, e.g., antibiotic resistance gene, within the nucleic acid. This allows pools of stably transfected cells to be recovered under selection by culturing the cells in the presence of the antibiotic.

[0172] A population of host cells may be provided in vitro. A population of cells comprising the nucleic acid may be a clonal population derived from a single individual recombinant cell. Alternatively it may be a pool of cells derived from multiple cells that have independently integrated the nucleic acid. For industrial production of protein, recombinant host cells (e.g., a clonal population) may be provided in which the encoding nucleic acid is stably integrated into the cellular chromosomal DNA. CHO is a preferred cell line for industrial manufacture.

[0173] Recombinant host cells may be cultured under conditions for expression of the protein, which is then optionally recovered from the cell culture and purified. Where a protein is secreted into culture medium, supernatant comprising the protein can be recovered from the cell culture by centrifugation and optionally concentrated, and the protein may then be purified using, for example, ion exchange chromatography, hydroxyapatite chromatography, size exclusion chromatography, hydrophobic interaction chromatography, gel electrophoresis, dialysis, affinity chromatography (e.g., protein A affinity chromatography or protein G chromatography), or a combination of such techniques.

[0174] Methods of production of bispecific molecules are well known in the art. For example, production methods may be used as described in, or may be adapted from, any of the following references:

[0175] Li Y et al 2020. Chapter 8 - a roadmap for igG-like bispecific antibody purification. In Matte (ed) Approaches to the purifications, analysis and characterisation of antibody-based therapeutics. Canada: elsevier, 2020, 167-79.

[0176] Tian Z et al 2021: J Hematol Oncol. 2021 May 3;14(1)75 (PMID 33941237)

[0177] Kontermann RE 2005. Acta Pharmacol Sin; 26: 1-9

[0178] Brinkmann II & Kontermann RE. 2017. MAbs 2017; 9; 182-212

[0179] Kontermann RE 2012: mAbs, 4:2, 182-197, DOI: 10.4161 / mabs.4.2.19000

[0180] Chen SW & Zhang W 2021 : Antibody Therapeutics. 2021 , Vol4, No.2 73-88. DOI 10.1093 / abt / thab007

[0181] Burton EM & Tawbi HA 2021: Cancer Discov. 2021 Mat; 11 (5): 1008-1010 (PMID 33947716)

[0182] Amash A et al 2024: mAbs 16:1, 2394229, DOI: 10.1080 / 19420862.2024.2394229.

[0183] The protein is preferably recoverable at high yield. The mass of correctly folded protein recovered after purification, relative to the original volume of cell culture supernatant, is preferably at least 1 mg / ml, e.g., at least 3 mg / ml, at least 5 mg / ml, at least 10 mg / ml or at least 15 mg / ml.

[0184] Monomer purity of the recovered protein may be determined. Monomer purity may be measured for the protein in the cell supernatant prior to purification, and / or following one or more purification steps. Monomer purity may be determined using size exclusion chromatography (SEC). Initial monomer purity, before purification, is preferably > 50 %. Monomer purity may be increased by purification steps and is preferably improved to > 90 %, > 95 %, > 98 % or > 99 % in a final purified composition. Monomer purity of a bispecific protein composition is of special relevance where mispairing of protein subunits is capable of generating bivalent homodimers for the immune cell surface receptor, which may cause targetindependent activation of the receptor. Anti-CD3 homodimers could produce indiscriminate activation of T cells, leading to safety risks including cytokine release syndrome. Mispairing of domains such as Fc portions may be reduced by engineering their sequences to favour heterodimerisation over homodimerisation, as discussed elsewhere herein, and / or by selection of a molecular design wherein unwanted intermolecular Fc pairing will not generate multivalent binding sites for the immune cell surface receptor, as exemplified by the protein of Figure 4(i).

[0185] Proteins of the invention may be glycosylated or unglycosylated. Glycosylation is the covalent linkage of oligosaccharide moieties to defined amino acid residues (typically asparagine, serine or threonine) of a polypeptide.

[0186] Typically, expression in eukaryotic cells (e.g., CHO) produces glycosylated proteins. Soluble proteins for use as therapeutics, and proteins for expression in cells for adoptive cell therapy, are preferably glycosylated. However they may be unglycosylated, for example if produced by expression in a prokaryotic cell.

[0187] The glycosylation status of a particular protein depends on a number of factors, including protein sequence, protein conformation and the availability of certain enzymes. Furthermore, glycosylation status (oligosaccharide type, covalent linkage and total number of attachments) can influence protein function. Therefore, when producing recombinant proteins, controlling glycosylation is often desirable.

[0188] Controlled glycosylation has been used to improve protein therapeutics (Sinclair & Elliott, Pharm Sci 94(8): 1626-35 2005; Jefferis R., Nat Rev Drug Discov. 8(3):226-34 2009). Glycosylation may be controlled by expressing the protein in a particular cell line, e.g., CHO or HEK as described above, and / or glycans may be chemically modified in vitro. The glycosylation of a TCR may also be altered by mutation of the TCR sequence (Kuball J et al., J Exp Med 206(2) : 463-475 2009).

[0189] Proteins may be glycoengineered to facilitate manufacturing or to provide additional functionality. This can be accomplished for example by deleting or introducing N-glycosylation and / or O-glycosylation sites.

[0190] An Fc region is present may be converted to a reduced glycosylation or aglycosyl variant by N297A or N297Q substitution. Aglycosyl Fc variants may provide improved manufacturability in terms of more homogenous batches, and demonstrate reduced FcyR binding and hence reduced Fc-mediated effector functions. Pharmaceutical compositions

[0191] Molecules of the invention, including nucleic acids and proteins, may be provided in isolated form, optionally in combination with one or more pharmaceutically acceptable excipients. Isolated nucleic acids and proteins will be free or substantially free of material found in their natural or production environment, such as other polypeptides or nucleic acids with which they would be found in vivo or in the environment in which they are prepared (e.g., cell culture when such preparation is by recombinant DNA technology in vitro).

[0192] An isolated nucleic acid molecule may represent over 50 %, over 60 %, over 70 %, over 80 %, over 90 %, over 95 % or over 99 % of the total nucleic acid content of a composition. An isolated protein may represent over 50 %, over 60 %, over 70 %, over 80 %, over 90 %, over 95 % or over 99 % of the total protein content of a composition.

[0193] The molecule of the invention may be provided in vitro.

[0194] Molecules of the invention, including nucleic acids and proteins, may be formulated for administration to patients. The nucleic acid or protein may be provided in a pharmaceutical composition, optionally comprising one or more excipients. Pharmaceutical compositions typically comprise the molecule in a therapeutically effective amount, plus optional auxiliary substances such as pharmaceutically acceptable excipient(s). Suitable carriers or excipients are well known in the art and include, for example, stabilizers, antioxidants, pH-regulating substances and controlled-release excipients. The molecule may be formulated in sterile aqueous solution for delivery, e.g., for administration by intravenous or subcutaneous injection.

[0195] The protein, nucleic acid or composition comprising it may be contained in a medical container such as a phial, syringe or injection device.

[0196] Therapeutic use

[0197] A method of treating a condition in a patient (e.g., human patient) comprises administering the soluble multispecific protein to the patient. The condition to be treated may be a solid tumour such as sarcoma, carcinoma or lymphoma, or an infection (e.g., viral infection, optionally a tumour-associated viral infection).

[0198] Target cells may present the target pMHC within a range of copy number as discussed elsewhere herein.

[0199] Therapy may be provided within a hospital environment, for which the molecules may be formulated for delivery in sterile aqueous solution, optionally provided within a bag connected to a drip line for intravenous infusion to a patient, either by flow under gravity (optionally with regulated flow) or in connection with an infusion pump to control flow rate and / or dose. Clauses

[0200] The following numbered clauses represent aspects of the invention.

[0201] 1. A soluble multispecific protein for recruiting an immune cell to a target cell presenting a target peptide major histocompatibility complex (pMHC), wherein the protein comprises a first polypeptide chain comprising a first TCR hemichain and a second polypeptide chain comprising a second TCR hemichain, wherein the first and second polypeptide heterodimerise via the pairing of the first and second TCR hemichain to form a TCR molecule for binding pMHC, wherein the first TCR hemichain is linked at its N terminus to an immune cell engager moiety, and wherein the second TCR hemichain is linked at its N terminus to a first Fc chain and is linked at its C terminus to a second Fc chain, wherein the first and second Fc chain dimerise to form an Fc region.

[0202] 2. A protein according to clause 1 , wherein each TCR hemichain comprises an N terminal TCR variable domain and a C terminal TCR constant domain.

[0203] 3. A protein according to clause 1 or clause 2, wherein the first TCR hemichain comprises a TCRp variable domain vp and the second TCR hemichain comprises a TCRa variable domain va.

[0204] 4. A protein according to clause 3, wherein the first TCR hemichain comprises a TCRp variable domain vp and a TCRp constant domain cp and the second TCR hemichain comprises a TCRa variable domain va and a TCRa constant domain ca.

[0205] 5. A protein according to clause 3 or clause 4, wherein the TCR constant domains of the first and second hemichain are inter-linked via one or more disulphide bonds.

[0206] 6. A protein according to any preceding clause, wherein the first TCR hemichain comprises a free C terminus, optionally terminating at the end of a cp domain.

[0207] 7. A protein according to any preceding clause, wherein the immune cell engager moiety comprises an antibody molecule.

[0208] 8. A protein according to clause 7, wherein the immune cell engager moiety comprises a single chain antibody molecule.

[0209] 9. A protein according to clause 7 or clause 8, wherein the immune cell engager moiety comprises an antibody VH domain and an antibody VL domain, wherein the VH domain and VL domain pair to provide a binding site for an immune cell surface receptor.

[0210] 10. A protein according to clause 9, wherein the immune cell engager arm comprises an scFv, optionally comprising an N terminal VL domain fused by a peptide linker to a C terminal VH domain. 11. A protein according to any preceding clause, wherein the first Fc chain comprises a free N terminus.

[0211] 12. A protein according to any preceding clause, wherein the first and second Fc chain each comprise human CH2-CH3.

[0212] 13. A protein according to clause 12, wherein the first polypeptide chain comprises, from N to C terminus, an immune cell engager antibody molecule, a linker, a first TCR variable domain and a first TCR constant domain, and wherein the second polypeptide chain comprises, from N to C terminus, a first Fc CH2-CH3, a linker, a second TCR variable domain, a second TCR constant domain, a linker, and a second Fc CH2-CH3.

[0213] 14. A protein according to clause 13, wherein the first polypeptide chain comprises, from N to C terminus, an immune cell engager VL-VH scFv antibody molecule, a linker, a TCR vp domain and a TCR cp domain, and wherein the second polypeptide chain comprises, from N to C terminus, a first Fc CH2-CH3, a linker, a TCR va domain, a TCR co domain, a linker, and a second Fc CH2-CH3.

[0214] 15. A protein according to any preceding clause, wherein the TCR molecule binds a PRAME peptide presented on pMHC, optionally peptide SLLQHLIGL.

[0215] 1A. A soluble, multi-polypeptide multispecific protein for recruiting an immune cell to a target cell presenting a target peptide major histocompatibility complex (pMHC), wherein the protein comprises a pMHC binding arm comprising a dimer of a first portion and a second portion, an immune cell engager moiety, and an Fc region comprising a dimer of a first Fc portion and a second Fc portion, wherein the first portion of the pMHC binding arm is linked at its N terminus to the immune cell engager moiety, and wherein the second portion of the pMHC binding arm is linked at its N terminus to a C terminus of the first Fc portion, and an N terminus of the Fc region is linked to a C terminus of the pMHC binding arm.

[0216] 2A. A protein according to clause 1A, wherein

[0217] (i) the N terminus of the second Fc portion is linked to the C terminus of the second portion of the pMHC binding arm; or

[0218] (ii) the N terminus of the second Fc portion is linked to the C terminus of the first portion of the pMHC binding arm; or

[0219] (iii) the N terminus of the first Fc portion is linked to the C terminus of the first portion of the pMHC binding arm. 3A. A protein according to clause 1A or clause 2A, wherein the pMHC binding arm comprises a free C terminus.

[0220] 4A. A protein according to any preceding clause, wherein the Fc region comprises a free N terminus and a free C terminus.

[0221] 5A. A protein according to any preceding clause, wherein the pMHC binding arm is a TCR molecule comprising a first TCR hemichain and a second TCR hemichain.

[0222] 6A. A soluble multispecific protein for recruiting an immune cell to a target cell presenting a target peptide major histocompatibility complex (pMHC), wherein the protein comprises a first polypeptide comprising a first TCR hemichain and a second polypeptide comprising a second TCR hemichain, wherein the first and second polypeptide heterodimerise via the pairing of the first and second TCR hemichain to form a TCR molecule for binding pMHC, wherein the first TCR hemichain is linked at its N terminus to an immune cell engager moiety, and wherein the second TCR hemichain is linked at its N terminus to a first Fc chain and is linked at its C terminus to a second Fc chain, wherein the first and second Fc chain dimerise to form an Fc region.

[0223] 7A. A protein according to clause 5A or clause 6A, wherein each TCR hemichain comprises an N terminal TCR variable domain and a C terminal TCR constant domain.

[0224] 8A. A protein according to clause 5A or clause 6A, wherein the first TCR hemichain comprises a TCRp variable domain vp and the second TCR hemichain comprises a TCRa variable domain va.

[0225] 9A. A protein according to clause 8A, wherein the first TCR hemichain comprises a TCRp variable domain vp and a TCRp constant domain cp and the second TCR hemichain comprises a TCRa variable domain va and a TCRa constant domain ca.

[0226] 10A. A protein according to clause 8A or clause 9A, wherein the TCR constant domains of the first and second hemichain are inter-linked via one or more disulphide bonds.

[0227] 11A. A protein according to clause 10A, wherein the TCR constant domains Ca and Cp are inter-linked via a diS bond between C95 in the alpha domain and C131 in the beta domain.

[0228] 12A. A protein according to clause 11A, wherein the TCR constant domains Ca and Cp are inter-linked via a diS bond between C48 in the alpha domain and C57 in the beta domain. 13A. A protein according to any of clauses 5A to 12A, wherein the first TCR hemichain comprises a free C terminus, optionally terminating at the end of a cp domain.

[0229] 14A. A protein according to any of clauses 1A to 4A, wherein the pMHC binding arm is an antibody molecule comprising an antibody heavy chain or fragment thereof and an antibody light chain or fragment thereof. 15A. A protein according to any preceding clause, wherein the immune cell engager moiety comprises an antibody molecule.

[0230] 16A. A protein according to clause 15A, wherein the immune cell engager moiety comprises a single chain antibody molecule, optionally scFv or dAb.

[0231] 17A. A protein according to clause 15A or clause 16A, wherein the immune cell engager moiety comprises an antibody VH domain and an antibody VL domain, wherein the VH domain and VL domain pair to provide a binding site for an immune cell surface receptor.

[0232] 18A. A protein according to clause 17A, wherein the immune cell engager moiety comprises an scFv, optionally comprising an N terminal VH domain fused by a peptide linker to a C terminal VL domain or an N terminal VL domain fused by a peptide linker to a C terminal VH domain.

[0233] 19A. A protein according to clause 15A, wherein the immune cell engager moiety comprises a single antibody variable (e.g., heavy) chain domain (dAb).

[0234] 20A. A protein according to any preceding clause, wherein the first and second Fc portion each comprise human CH2-CH3.

[0235] 21A. A protein according to any preceding clause, wherein the first Fc portion comprises a free N terminus.

[0236] 22A. A protein according to clause 20A or clause 21A, wherein the first polypeptide comprises, from N to C terminus, an immune cell engager antibody molecule, a linker L1 , a first TCR variable domain and a first TCR constant domain, and wherein the second polypeptide comprises, from N to C terminus, a first Fc CH2-CH3, a linker L2, a second TCR variable domain, a second TCR constant domain, a linker L3, and a second Fc CH2-CH3.

[0237] 23A. A protein according to clause 22A, wherein the first polypeptide comprises, from N to C terminus,

[0238] (i) an immune cell engager VH-VL scFv antibody molecule, a linker L1 , a TCR vp domain and a TCR cp domain, and wherein the second polypeptide comprises, from N to C terminus, a first Fc CH2-CH3, a linker L2, a TCR va domain, a TCR co domain, a linker L3, and a second Fc CH2-CH3; or

[0239] (ii) an immune cell engager VL-VH scFv antibody molecule, a linker L1 , a TCR vp domain and a TCR cp domain, and wherein the second polypeptide comprises, from N to C terminus, a first Fc CH2-CH3, a linker L2, a TCR va domain, a TCR ca domain, a linker L3, and a second Fc CH2-CH3.

[0240] 24A. A protein according to clause 22A or clause 23A, wherein linker L1 is (G4S)2.

[0241] 25A. A protein according to any of clauses 22A to 24A, wherein linker L2 is (G4S)3.

[0242] 26A. A protein according to any of clauses 22A to 25A, wherein linker L3 is (G4S)3. 27A. A protein according to any of clauses 1 A to 5A or 7A to 21 A, wherein the N terminus of the second Fc portion is linked to the C terminus of the first portion of the pMHC binding arm. 28A. A protein according to clause 27A, comprising a first polypeptide which comprises, from N to C terminus, the immune cell engager antibody molecule, a linker, the first portion of the pMHC binding arm (optionally comprising first TCR variable domain and first TCR constant region), and the second Fc portion (optionally comprising second CH2-CH3), and a second polypeptide which comprises, from N to C terminus, the first Fc portion (optionally comprising first CH2-CH2), a linker, and the second portion of the pMHC binding arm (optionally comprising second TCR variable domain and second TCR constant region).

[0243] 29A. A protein according to clause 28A, wherein

[0244] (i) the first polypeptide comprises, from N to C terminus, immune cell engager VH-VL scFv antibody molecule, linker, TCR vp domain, TCR cp domain, and second Fc CH2-CH3, and the second polypeptide comprises, from N to C terminus, first Fc CH2-CH3, linker, TCR va domain, and TCR co domain; or

[0245] (ii) the first polypeptide comprises, from N to C terminus, immune cell engager VL-VH scFv antibody molecule, linker, TCR vp domain, TCR cp domain, and second Fc CH2-CH3, and the second polypeptide comprises, from N to C terminus, first Fc CH2-CH3, linker, TCR va domain, and TCR ca domain.

[0246] 30A. A protein according to any of clauses 1A to 5A or 7A to 21A, wherein the N terminus of the first Fc portion is linked to the C terminus of the first portion of the pMHC binding arm.

[0247] 31A. A protein according to clause 30A, comprising a first polypeptide which comprises, from N to C terminus, the immune cell engager antibody molecule, a linker, the first portion of the pMHC binding arm (optionally comprising first TCR variable domain and first TCR constant region), the first Fc portion (optionally comprising first CH2-CH3), and the second portion of the pMHC binding arm (optionally comprising second TCR variable domain and second TCR constant region), and a second polypeptide which comprises the second Fc portion, optionally comprising second CH2-CH3.

[0248] 32A. A protein according to clause 31A, wherein

[0249] (i) the first polypeptide comprises, from N to C terminus, immune cell engager VH-VL scFv antibody molecule, linker, TCR vp domain, TCR c domain, first Fc CH2-CH3, TCR va domain, and TCR Ca domain, and the second polypeptide comprises, from N to C terminus, second Fc CH2-CH3; or (ii) the first polypeptide comprises, from N to C terminus, immune cell engager VL-VH scFv antibody molecule, linker, TCR vp domain, TCR cp domain, first Fc CH2-CH3, TCR va domain, and TCR Co domain, and the second polypeptide comprises, from N to C terminus, second Fc CH2-CH3.

[0250] 33A. A protein according to any preceding clause, wherein the pMHC binding arm (e.g., TCR molecule) binds a PRAME peptide presented on pMHC, optionally peptide SLLQHLIGL.

[0251] 34A. Nucleic acid encoding a protein according to any of clauses 1A to 33A.

[0252] 35A. A population of cells in vitro containing recombinant nucleic acid according to clause 34A.

[0253] 36A. A method of manufacturing a protein according to any of clauses 1A to 33A, comprising culturing a population of cells according to clause 35A under conditions for expression of the protein, and recovering the protein, optionally including isolating and purifying the protein and formulating it into a composition with one or more pharmaceutically acceptable excipients. 37A. A population of cells according to clause 35A or a method according to clause 36A, wherein the cells are CHO cells comprising random genomic integration of first and second gene expression vectors encoding the first and second polypeptides respectively.

[0254] 38A. A method of treating a condition in a patient, comprising administering a soluble multispecific protein as defined in any of clauses 1A to 33A, or its encoding nucleic acid, to the patient.

[0255] 39A. A soluble multispecific protein according to any of clauses 1A to 33A, or its encoding nucleic acid, for use in method of treatment of the human or animal body by therapy.

[0256] 40A. A method according to clause 39A or a soluble multispecific protein or nucleic acid for use according to clause 39A, wherein the treatment comprises treating a solid tumour.

[0257] Examples

[0258] Example 1. Production of bispecific TCR T cell engagers

[0259] Gene expression vectors encoding TCR T cell engaging bispecific proteins were transiently transfected into Expi293 cells (Thermo Fisher Scientific, #A14635) according to manufacturer's guidelines using 200 pg of plasmid DNA per 200 mL cell culture. For molecules in the two chain format depicted in Figures 1 and 2, 120 pg of TCRp chain and 80 pg of TCRa chain were co-transfected into Expi293 cells. Supernatants were harvested after a 7-day incubation and subsequently purified using an AKTA system (Cytiva). Protein purification involved affinity chromatography utilizing a Protein A column (MabSelect SuRe, Cytiva), followed by desalting to remove impurities. For protein A purification, columns were washed with PBS and proteins were eluted with sodium formate solution at pH 3.5. Buffer was exchanged back to PBS in desalting column. The final protein preparations were concentrated to a volume of 1 mL and subjected to gel filtration chromatography. Fractions corresponding to the expected molecular weight range were collected.

[0260] Two chain bispecific TCR T cell engaging molecules were produced in the format shown in Figure 1 and Figure 2, including:

[0261] - TTP017: Chain 1 scFv-TCRbeta SEQ ID NO: 1 and Chain 2 Fc-TCRalpha-Fc SEQ ID NO: 2

[0262] TTP018(CPA665): Chain 1 scFv-TCRbeta with stabilising mutations SEQ ID NO: 3 and Chain 2 Fc-TCRalpha-Fc with stabilising mutations SEQ ID NO: 4

[0263] TTP019: Chain 1 scFv-TCRbeta with stabilising mutations SEQ ID NO: 3 and Chain 2 TCRalpha-Fc with stabilising mutations and hairpin linker SEQ ID NO: 5.

[0264] Each T cell engager protein included anti-CD3 scFv UCHT1v17 (SEQ ID NO: 6) and an op TCR specific to a PRAME pMHC (TCR alpha chain SEQ ID NO: 8, TCR beta chain SEQ ID NO: 7). Each chain 2 included human lgG1 Fc with knob (K409A) (SEQ ID NO: 9) and human lgG1 Fc with hole (F405K) (SEQ ID NO: 10).

[0265] Yield and purity of the 2-chain molecules was assessed following recovery of product from expression in HEK(Expi293). Yields for TTP017, TTP018 and TTP019 were 8.5 mg / L, 75 mg / L and 95 mg / L respectively. Purity after size exclusion chromatography (SEC) was over 90 %.

[0266] Example 2. Dual binding assay (ELISA)

[0267] A dual binding ELISA assay was established to assess binding capacity of bispecific T cell engagers which recognise human CD3 and tumour-specific peptides in the context of HLA- A*02:01. Briefly, a 96-well black plate (Griener, 655077) was coated overnight at room temperature with 100 pL / well PBS containing 1 pg / mL monomer peptide MHC complex in which MAGE-A4 peptide (230-239, GVYDGREHTV SEQ ID NO: 41) or PRAME peptide (425-433, SLLQHLIGL SEQ ID NO: 42) was presented on HLA-A*O2:O1_ 2m complex (KACTUS, MHC- HM401 or MHC-HM443 respectively). The next day the plate was washed with 0.1% Tween 20 in PBS, followed by blocking with 1% BSA in PBS for 1 hour at room temperature. After washing, samples were loaded to the wells at various concentrations to allow the binding of bispecific T cell engagers to the peptide MHC complex and incubate 2 hours at room temperature in diluent buffer (0.1% BSA in PBS). The captured complex was then detected by biotin-conjugated CD3E5 heterodimer (Aero Biosystems, CDD-H82W6) at 200 ng / mL in diluent buffer for 1 hour at room temperature, followed by DELFIA Eu-labelled Streptavidin (Revvity, 1244-360) pre-diluted in DELFIA assay buffer (Revvity, 1244-111) at 100 ng / mL for another 1- hour incubation at room temperature. The plate was then washed with 1x DELFIA wash buffer (Revvity, 1244-114) and 100 pL / well DELFIA enhancement solution (Revvity, 1244-105) was added 5 minutes before the fluorescent reading at 615 nm on CLARIOstar plate reader. All bispecific molecules showed good dual-binding effect, indicating their ability to simultaneously bind to both peptide MHC complex and soluble CD3E5 heterodimer.

[0268] Example 3. T cell killing assay

[0269] A cytotoxicity assay was established to assess the capacity of CD8+ T cells to kill tumour cells, which was induced by bispecific T cell engagers binding to human CD3 and peptide MHC. A375, a human melanoma cell line, was selected to represent target cells presenting tumour antigen MAGE-A4 or PRAME peptides on HLA-A*02:01. The bispecific T cell engagers recognise peptide MHC complex presented on A375 cells, and simultaneously engage CD8+ T cells as effectors by anti-CD3 recognition to kill target cells.

[0270] The remaining live A375 cells were measured by quantifying intracellular ATP content, which proportionally facilitated luciferase activity to generate luminescent signals. Staurosporine is a potent inducer of apoptosis and was used as a positive control for killing. Percentage of cytotoxicity was calculated by reference to the staurosporine-treated A375 cells as 100 % killing and untreated cells as 0 % killing.

[0271] A375 target cells were pre-seeded onto a 96-well IsoPlates (Revvity, 6005070) at 5,000 cells / well and cultured overnight at 37°C with 5 % CO2. The next day, total CD8+ T cells were positively selected by human CD8 isolation kit (Miltenyi, 130-045-201) from human peripheral blood mononuclear cells (PBMC) that carry at least an allele of HLA-A*02:01. Isolated CD8+ T cells as effector cells were added to the culture at 25,000 cells / well to create a ratio of 1:5 of target to effector. The co-culture was then treated with various concentrations of bispecific T cell engagers and incubated for 2 days at 37°C with 5 % CO2. At the end of incubation, the suspended effector cells were washed away and the remaining attached live A375 cells were subjected to CellTiter-Glo® 2.0 assay (Promega, G9242) which quantitatively measured the intracellular ATP content. Finally, the luminescent signals were determined by CLARIOstar plate reader, and a percentage of cytotoxicity was calculated.

[0272] Bispecific T cell engagers of the format shown in Figure 1 and Figure 2 were compared against bispecific T cell engagers in alternative formats. Results

[0273] Bispecific MAGE-A41 CD3 molecules in the two-chain format shown in Figure 1 and Figure 2, SEC purified, showed good cytotoxicity, where A375 tumor cells were killed by isolated CD8+ T cells. Cell killing was lower compared with a bispecific molecule in the small, non half-life extended molecule produced in an “ImmTAC®” format consisting of an anti-CD3 scFv linked to the N terminus of a TCR extracellular region (Boulter & Jakobsen Clin Exper Immunol 142:454-460 2005; Liddy et al, Nat Med 18(6):980-987 2012, Oates & Jakobsen, Oncolmmunology 2(2):e22891 2013). EC50 values were calculated: Bispecific PRAME I CD3 molecules in the two-chain format shown in Figure 1 and

[0274] Figure 2 also showed good cytotoxicity. Cell killing was lower than for the ImmTAC® format, which lacks an Fc region, but was equivalent or superior compared with other Fc-containing formats. EC50 values were calculated: Example 4: Pharmacokinetic properties

[0275] Noncompartmental pharmacokinetic parameters of the 2-chain multispecific protein CPA665 (SEQ ID NO: 3 + SEQ ID NO: 4) were obtained from an in-vivo experiment where SCID mice received a single 12.5 ug intravenous bolus administration in PBS.

[0276] The plasma concentration was measured by sparse sampling up to 21 days with a dual binding ELISA capturing analyte with plate bound human recombinant pMHC and detection with human recombinant CD3E5-HRP. A minimum of three animals were used per time point.

[0277] Elimination rate constant (Kel) was estimated from slope of the terminal phase (t>Day1, R2=0.87) of the log plasma concentration-time curve (Figure 7) and terminal half-life (ti / 2) was calculated as ln(2) / Kel. Further, serial data points were collected on Day10 and Day21 (Table 4). The lower limit of quantitation (LLOQ)=[0.25ng / mL] was set to the lowest analyte value above background in the standard curve.

[0278] The derived terminal half-life of ~7.3 days is comparable to other Fc-containing, antibody-like molecules as described in literature.

[0279] Table 4: plasma concentration at indicated time point

[0280] Example 5: Measurement of binding affinity using surface plasmon resonance (SPR)

[0281] Affinity of bispecific TCR-TCE molecules for CD3 and for target pMHC (PRAME peptide SLLQHLIGL on HLA-A*02) was determined using SPR.

[0282] To determine the kinetics and affinities of TCR:TCE molecules binding to CD3E / 5, CD3E / Y or PRAME pMHC, 1pg / mL CD3E / 5, CD3E / Y or PRAME pMHC solution was captured on Cytiva Biacore Series Sensor NTA chip. After capture, analytes (TCR:TCE molecules) at concentrations of 0 pg / mL, 0.2 pg / mL, 0.7 pg / mL, 2.2 pg / mL, 6.6 pg / mL, or 20 pg / mL were used to detect the binding on Biacore 8000K machine. Data were processed using Biacore Insight Evaluation software. Kinetics curves were fit using 1:1 binding model, and affinity curves were fit using steady state affinity model.

[0283] CPA665 (SEQ ID NO: 3 + SEQ ID NO: 4) is a 2-chain multispecific protein having the format shown in Figure 1 and Figure 2. Comparator molecules were scFv-TCR 994 (SEQ ID NO: 14 + SEQ ID NO: 15), which has the scFv-TCR format as used in ImmTAC® molecules such as brenetafusp - TCR-diabody-Fc 580 (SEQ ID NO: 12 + SEQ ID NO: 13), which has the diabody-like Fc-linked format as used in TCER® molecules.

[0284] Data are shown in Figure 8 and in Table 5.

[0285] The anti-CD3 moiety in each of these molecules is based on antibody LICHT 1 . The anti- CD3 in CPA665 and TCR-diabody-Fc 580 respectively are both based on “UCHT1v17”, in scFv format (SEQ ID NO: 6) in CPA665 and in diabody format in TCR-diabody-Fc 580 as described in WO2021 / 023658 and W02021 / 144020. The anti-CD3 moiety in scFv-TCR 994 is UCHT1 scFv as described in W02020 / 157210. The affinities for anti-CD3 were determined to be in the nanomolar range for all molecules.

[0286] The TCR moiety of each bispecific binds PRAME epitope SLLQHLIGL (peptide 425- 433). CPA665 and scFv-TCR 994 comprise a TCR with variable domains of the a79b46 TCR described in WO2018 / 234319. TCR-diabody-Fc 580 has the HiAffl TCR described in “TCER” format in WO2021 / 144020. Affinities of the TCR for target pMHC were determined to be in the nanomolar range for all molecules.

[0287] Example 6: Bispecific TCR-TCE induced T cell killing of tumour cell lines in vitro

[0288] Bispecific TCR-TCE molecules were tested an in vitro killing assay to assess their ability to mediate killing of target cells by T effector cells.

[0289] In this assay, tumour cell lines presenting target pMHC (PRAME peptide SLLQHLIGL on HLA-A*02) were incubated with a dilution series of the test protein (candidate bispecific TCR- TCE) and T effector cells, provided either as PBMC (peripheral blood mononuclear cells) or isolated CD8+ T cells. The 2-chain multispecific protein CPA665 (SEQ ID NO: 3 + SEQ ID NO: 4) was compared against scFv-TCR 994 (SEQ ID NO: 14 + SEQ ID NO: 15) and TCR-diabody- Fc 580 (SEQ ID NO: 12 + SEQ ID NO: 13).

[0290] Production of bispecific molecules:

[0291] Bispecific molecules specific for the target PRAME pMHC complex were expressed in transiently transfected HEK cells. Briefly, gene expression vectors encoding the TOR T cell engaging bispecific molecules scFv-TCR 994, TCR-diabody-Fc 580 and CPA665 were transiently transfected into Expi293 cells (Thermo Fisher Scientific, A14635) using an Expifectamine protocol following the manufacturer’s guidelines. Supernatants from the cell culture were harvested 5 days post-transfection for purification.

[0292] Proteins were recovered from harvested cell culture supernatant using protein A affinity purification with a low pH elution. Samples were further purified by SEC into PBS, the concentration was adjusted, and the samples were sterile filtered. All samples were analysed for concentration by UVA / is and for purity by HPLC-SEC and SDS-PAGE.

[0293] Killing assay protocol:

[0294] Preparation of effector cells

[0295] PBMC were prepared from leukopheresis cones obtained from NHS Blood and Transplant service. PBMC were counted by diluting 20 uL sample up to 10x (20ul of PBMC in 180ul of separation media), then 10ul of diluted cells added to 10ul of trypan blue. Total PBMC were calculated by the following equation: number of cells counted x 104x 2(trypan dilution) x10 (PBSc dilution) x 50 (resuspended volume). PBMC were used as effector cells.

[0296] Preparation of target cells

[0297] SK-MEL 5 (HTB-70, ATCC) are Hs695t (HTB 137, ATCC) adherent melanoma-derived cell lines, grown in complete Eagle MEM and passaged twice a week at 1 :4. (Complete E-MEM - E-MEM, 10%FBS, 2mM Glutamine or Glutamax, 5ml Penicillin-Streptomycin). OVCAR-3 (RCB2135, RIKEN) is an adherent ovarian adenocarnoma cultured in supplemented RPMI (indicated above) and passaged twice a week at 1 :10. COV318 (7071903, Merck) is a human ovarian epithelial-serous carcinoma cultured in supplemented DMEM (10% FBS) and NCI- H1755 [H1755] (CRL-5892, ATCC) representative of NSCLC culture in supplemented RPMI (as indicated above) and passage 1 / 10 twice a week.

[0298] Assay set up

[0299] The day prior to running the assay, PBMC vials (5 million per 96 well plate) were thawed as required in complete RPMI (RPMI, 10% FBS, 2mM Glutamax, 2mM Pyruvate, 1% MEM, 1% HEPES, 1% P / S) and allowed to recover overnight. Target cell lines were trypsinised and seeded in flatbottom transparent 96 well plate at 1.104 cells per well in 100uL complete media. Outer wells of the plate were excluded and filled with 200uL of PBS. On the day of the assay in a separate round-bottom 96 plate, bispecific protein was serial diluted in complete RPMI at 4X concentration in 100ul final volume. 100uL effector cells (PBMC or CD8 T cells) at 1.106 c / mL was transferred on to the bispecific plates. This dilutes the bispecific at 2X in a total volume of 200uL. 100ul of bispecific: effector mix was transferred on to target plates. This creates a ratio of 1 :5 of Target to Effector and dilutes the bispecific to 1X. This also provides the right medium for both cell types to grow.

[0300] A TP assay

[0301] At the end of the experiments, cytotoxicity was evaluated using a luminescence-based ATP assay which measures remaining viable target cells. Briefly, culture plates were washed 3 times with PBS 1X and an equal volume of PBS and CellTiter Gio 2.0 (G9241 , Promega) reagent was added to each well. Plates were incubated at room temperature in the dark for 10 minutes before the luminescence signal was read according to the manufacturer’s instruction. Luminescence signal was plotted against bispecific concentration to obtain a dose-response curve from which an EC50 value was calculated. Viability index was calculated by normalising the dose-response curve by the control dose.

[0302] Results:

[0303] Bispecific molecules scFv-TCR 994, TCR-diabody-Fc 580 and CPA 665 were evaluated for in vitro killing potency in a coculture of target cells and PBMC. TCR-diabody-Fc 580 and CPA 665 are half-life extended bispecifics for prolonged in vivo efficacy and improved pharmacokinetics. Hs695t (Figure 9A), SK-MEL-5 (Figure 9B) and A-375 (Figure 9C) were selected as PRAME-positive melanoma cancer lines. NCI-H1755 (Figure 9D) was selected for NSCLC while OVCAR-3 (Figure 9E) and COV318 (Figure 9F) represented PRAME-positive ovarian cancer lines.

[0304] EC50 values from TDCC assays for scFv-TCR 994, TCR-diabody-Fc 580 and CPA 665 were compared for different PRAME-positive cancer cell lines. Representative results are shown in Table 6, for the data presented in Figure 9. Figure 10 shows EC50 values calculated from replicate assays. CPA 665 was effective at killing all target cell lines, and was significantly more potent than TCR-diabody-Fc 580.

[0305] Table 6. EC50 (picomolar) values for bispecific molecules in the in vitro killing assay.

[0306] Example 7: Bispecific TCR TCE induced T cell killing of tumour in vivo

[0307] Efficacy of multispecific proteins in varying formats was assessed in vivo using a human melanoma xenograft mouse model in NSG mice with human PBMC, based on known protocols (see, e.g., Haque et al., Bio Protoc 12(23):e45662022, and information published by The Jackson Laboratory).

[0308] The 2-chain multispecific protein CPA665 (SEQ ID NO: 3 + SEQ ID NO: 4) was compared against scFv-TCR 994 (SEQ ID NO: 14 + SEQ ID NO: 15) and TCR-diabody-Fc 580 (SEQ ID NO: 12 + SEQ ID NO: 13). Proteins were produced as described in Example 6, with the addition that the samples were analysed for endotoxin content before use in vivo use.

[0309] Treatments were administered to NSG mice (Jackson Labs) bearing previously implanted human melanoma tumour cells (Hs695T, 3M, s.c.), 10 days after reconstitution of the mice with HLA matched PBMCs (10M, i.v.). Each multispecific protein was administered at the same molar concentration, with a reference dose of 0.025 mg / kg for TCR-diabody Fc 580, and with the same weekly intravenous regimen.

[0310] Figure 11 shows efficacy of each multispecific protein in a mouse model of human melanoma. Mice that received treatment with CPA665 showed complete response (3 of 5 animals) or partial response (2 of 5 animals). Mice that received treatment with the comparator molecule scFv-TCR 994 showed a maximum efficacy of stable disease. Mice that received treatment with the comparator molecule TCR-diabody-Fc 580 showed a maximum efficacy of transient stable disease. These data support the superior in vivo efficacy of the format illustrated in Figure 1 and Figure 2.

[0311] Histological examination of CPA665-treated regressing tumours confirmed reduced tumour size and increased infiltration of CD3+ cells, with cystic-like structures indicating drug- induced responses compatible with T cell engagement. Example 8: Titre and purity assessment for protein expressed in HEK cells

[0312] Multispecific protein CPA665 and other proteins in the Figure 1 format were produced by transient expression in HEK cells (Expi293 as described in Example 6) and the titre was determined. Material was processed through two step purification and the recovery and purity were recorded. Where purity was < 95%, samples were further purified by SEC.

[0313] Methods:

[0314] Titre analysis

[0315] Titre determination was performed on the Agilent 1290 bioinfinity II HPLC using a biomonolith rProtein A column. Prior to running samples, the column was calibrated with a 12-point standard curve using NISTmab at known concentrations from 2000 mg / l (the upper limit of quantification) to 25 mg / l (the lower limit of quantification) by running the below method with a flow rate of 1 ml / min at 30 °C:

[0316] 1 . 20 pl sample injected on to the column

[0317] 2. Wash the column in PBS for 1.1 minutes

[0318] 3. Gradient elution to 100 % (v / v) 500 mM acetic acid over 1.1 minutes

[0319] 4. Re-equilibrate the column in PBS for 1.3 minutes

[0320] Affinity purification

[0321] Proteins were purified through two-step protein A affinity chromatography followed by desalt using either the Teledyne CETAC autosampler, or the AKTA sample lines to load. All columns and lines are cleaned with 5 CV or 20 ml respectively per step of UP water, then 500 mM NaOH followed by a 15-minute incubation, an additional UP water wash and then equilibration with PBS. Samples were purified by the following methods:

[0322] Protein A step (1 ml Prism A)

[0323] 1 . 5 CV column equilibration in PBS at 4 ml / min

[0324] 2. Sample application at 1 ml / min

[0325] 3. 3 CV column wash in PBS at 4 ml / min

[0326] 4. 5 CV sample elution at 1 ml / min using 50 mM acetic acid at pH 3.8, collecting the peak of eluted protein into a 3 ml loop

[0327] 5. 5 CV re-equilibration of the column into PBS at 4 ml / min

[0328] Desalt step (2x5 ml G25)

[0329] 1. 1 CV column equilibration in PBS at 10 ml / min

[0330] 2. Sample application from the 3 ml loop at 10 ml / min 3. 1 CV isocratic elution at 10 ml / min, collecting the peak of eluted protein into a 96 deep well plate

[0331] Concentration determination

[0332] Protein concentration was determined on the Unchained Labs Stunner using the absorbance at 280 nm (A280) and adjusted to mg / ml according to beer-lambert law. Briefly, 2 ml of sample was loaded into a 96 well Stunner plate along with 2 ml of PBS. The £1% was entered, either theoretically determined through Benchling where available or using an average of 15, and the concentration was automatically read and determined on the Stunner.

[0333] Where performing DLS sizing, samples were pulse centrifuged for 15 s prior to loading. The default DLS settings were used on the Stunner. A comparative sample analyte of Adalimumab was selected and the buffer selected was PBS.

[0334] HPLC-SEC analysis

[0335] Purity was determined on the Agilent 1290 bioinfinity II HPLC using a Waters XBridge Premier Protein SEC 250A 4.6 x 150 mm column with a bead diameter of 2.5 pm. 2 pg sample was loaded in injections of 0.1-20 pl and the purity profile was resolved over 8 minutes at a flow rate of 0.35 ml / min using 2 x PBS, 15% (v / v) I PA. Sample retention times were compared to Merck molecular weight standards and Waters NISTmab.

[0336] Purification by SEC

[0337] Protein from affinity purification was further polished using SEC on a superdex 200 increase 10 / 300 GL column. Briefly the column was first cleaned with UP water, followed by 500 mM NaOH, UP water and then equilibrated with 1.5 CV 1 x PBS made with WFI grade water. The 1 ml injection loop was cleaned with 500 mM NaOH, UP water and then 1 x PBS. The sample volume was adjusted to 1 ml for injection, to a maximum concentration of 5 mg / ml. Sample was injected and resolved over 0.67 CV with the peak of protein collected in 1 ml fractions into a 96 deep well plate.

[0338] Results:

[0339] CPA665 was produced with a titre of 24.7 mg / l, at a purity of 71.7 %.

[0340] Proteins in the same format, with different TCRs, showed a range of titres from <2 mg / l to ~60 mg / l, averaging 25.52 mg / l. Purity ranged from 33.67 % to 78.47 %, averaging 68.30 %. The titre and purity are thus be influenced by choice of TCR. HPLC-SEC data indicated that the major impurities were likely dimer with some higher molecular weight species. HPLC-SEC data for protein of representative purity (Figure 12) and highest purity (Figure 13) are shown. Proteins in the same format, with a fixed TCR and different anti-CD3 scFv, showed minimal change in titre with the antibody selection but considerable impact on purity. Average titre / recovery was 13.15 mg / l and average purity was 64.58 % (range 5.49 - 85.09 %). As before, major impurities appeared to be dimer with some higher molecular weight species. Lower molecular weight species tended to be present in lower purity samples.

[0341] Example 9a: Titre and purity assessment for protein expressed in stable CHO pools

[0342] Gene expression vectors encoding CPA665 and other TCR T cell engaging bispecific proteins in the Figure 1 format were stably transfected into CHO-K1 cells using electroporation, with expression vectors for the two chains at a ratio of 1:1. Pools were then recovered under selection and supernatants were harvested at different stages of the cell line generation process (e.g., after a 9 / 10-day fed-batch evaluation) and subsequently purified.

[0343] Protein was purified and monitored for titre. Harvested material was processed through two step purification, as described in Example 8, and the recovery and purity were recorded.

[0344] The average titre was 60.64 mg / l with the highest titre observed at 164.62 mg / l. The average purity was 68.07% which is comparable to previous observations, but purity ranged from 12.86% at the anomalous low edge to 81.15% at the higher end. This indicates that purity can be further tuned by cell line selection.

[0345] When stabilising mutations were introduced into the TCR constant domain (as exemplified in CPA665), average purity was 67.95% with all samples within 5% of this, indicating that the format including stabilised TCR was largely agnostic to cell line choice. The average titre for these stabilised molecules was 115.27 mg / l, with the highest titre at 246.26 mg / l.

[0346] Example 9b: High yield expression in selected CHO recombinant host cells

[0347] Two different TCR-TCE molecules in Figure 1 format were independently produced in CHO-K1 cells using a fed batch overgrow process. For each protein, separate gene expression vectors encoding the first (scFv-TCR) and second (Fc-TCR-Fc) polypeptide chains respectively were electroporated into the cells to generate stably transfected recombinant cells. The highest expressing cultures were selected:

[0348] • CPA1122, clone C1-20, seeding density 2x106 / ml, 1606 mg / l on harvest day 14. 64.8% monomer purity.

[0349] • CPA1120, minipool MP3-106, seeding density 2x106 / ml, 978 mg / l on harvest day 14. 69.1% monomer purity. These results confirm that the TCR-TCE molecular format is permissive of recombinant expression with good product recovery, comparable to industry norms for therapeutic bispecific antibodies.

[0350] Example 10: Developability analysis

[0351] Accelerated stability analysis

[0352] Accelerated stability if protein in the Figure 1 format was determined by adjusting SEC purified samples of CPA665 to 1 mg / ml in 400 pl aliquots per sample. Samples were sterile filtered using a 0.22 pm PVDF filter and 200 pl was loaded into two sterile screw-cap vials. Vials were incubated at 25 °C and 40 °C. 8 ml aliquots were taken under the laminar flow hood at Day 0, 3, 7, 14 and where additional sample was available, day 28 and day 56. Samples were analysed at each time point for the concentration with DLS sizing and for the purity using the HPLC-SEC (for methods see Example 8).

[0353] This assessment was performed with an un-optimised buffer. At 25 °C the only change observed was an instrument-related drift in the retention time and a shift in the absorbance, related to the error in concentration determination. Changes in purity were within error for peak selection. There was a slight apparent negative skew to the peak shape which shifted slightly to indicate a slight shoulder on the leading edge of the main species peak, however this was not distinct-enough to be analysed as a separate species under these conditions.

[0354] At 40 °C there was minimal change in the retention time, absorbance and purity over 7 days. The negative skew was apparent within the first three time points. After 14 days the purity had decreased due to increased separation of a shoulder peak on the leading edge of the main species peak. This was further observed at 28 days indicating that the observation was accurate.

[0355] As the stability was assessed without optimising the buffer conditions and without optimising the purification method, further work on purification methods and formulation would likely increase the purity and stability of the protein.

[0356] Viral inactivation

[0357] Viral inactivation is a critical step in the manufacture process of biologies. Routinely this is performed by lowering the pH to 3.5 and holding for 90 minutes.

[0358] Resistance to the low pH viral inactivation was determined using a mock viral inactivation process. Briefly, 100 pl sample containing CPA665 at 1 mg / ml was adjusted to 0.1 mg / ml in 1 ml. VI was induced on adjusting the pH to 3.5 by adding 20 pl 5 M acetic acid. VI was monitored at 0, 15, 30, 60, 90 and 120 minutes, where sample was available, additional time points were taken either at 240 minutes or after 18 hours (overnight). At each time point a 52 pl sample was taken and neutralised with 25% (v / v) 1 M Tris pH 8.0. Samples were analysed for concentration with DLS sizing and for the purity using HPLC-SEC (for methods see Example 8).

[0359] Data from the mock viral inactivation indicated that the purified protein was stable to low pH under CMC-like conditions. The initial 0-minute data point showed a negative skew in chromatogram which may indicate the presence of and additional species with a slightly higher molecular weight than the main species. The negative skew was only present at the first time point with all other time points having the appearance of a normal distribution. The data showed minimal change in the SEC profile between 15 minutes to 120 minutes of the test conditions with a <0.3% change in purity recorded.

[0360] In conclusion, the mock viral inactivation study indicated that protein in the format shown in Figure 1 can withstand viral inactivation at pH 3.5 for 120 minutes with minimal change in purity.

[0361] Example 11 : Multispecific proteins with a panel of anti-CD3 scFv VL-VH and VH-VL

[0362] A panel of multispecific proteins were produced in the format depicted in Figure 1 , each containing a different anti-CD3 scFv as the immune cell engager moiety. The format shown in Figure 1 (and Figure 2 and Figure 14) may be referred to as TB2C. In the embodiment generated in this Example, TB2C proteins comprise the anti-CD3 scFv linked at its C terminus to the N terminus of a TCR molecule, and an Fc region comprising first and second Fc portions linked to the N and C termini respectively of the TCR molecule. The TB2C is composed of two polypeptide chains: (1) scFv-L1-TCRVp-CP; (2) Fc-L2-TCRVa-Ca-L3-Fc. Each of L1, L2 and L3 is a peptide linker. The TCR alpha hemichain Va-Ca pairs with the TCR beta hemichain Vp-Cp to form the TCR molecule. The two Fc portions pair to form a dimeric Fc region, and knob-in- hole mutations are included in the CH3. Bispecific proteins were initially generated as TB2C with the scFv in VL-VH orientiation, i.e., with the VL domain at the N terminus (Figure 2). A (G4S)4 linker was used to connect VL and VH (linker L0). Linker L1 was G4S. Linker L2 was G4SG4. Linker L3 was also G4SG4. The TCR molecule used in these studies was specific for a pMHC complex comprising a PRAME peptide and was engineered with stabilising mutations in the constant domains.

[0363] To produce TCR-TCE bispecific proteins, expression vectors encoding TCR-TCE proteins were transiently transfected into Expi293 cells (Thermo Fisher Scientific, #A14635) according to manufacturer's guidelines using 1 pg of plasmid DNA per 1 mL cell culture with 1:1 DNA ratio for the two chains. Supernatants were harvested after a 7-day incubation and subsequently purified using an AKTA system (Cytiva). Protein purification involved affinity chromatography on a Protein A column (MabSelect SuRe, Cytiva), followed by desalting to remove impurities. For protein A purification, columns were washed with PBS and proteins were eluted with sodium acetate solution at pH 3.8. Buffer was exchanged back to PBS in desalting column. The final protein preparations were concentrated to a volume of 1 mL and subjected to gel filtration chromatography. Fractions corresponding to the expected molecular weight range were collected.

[0364] TB2C bispecific proteins were evaluated for potency in an in vitro killing assay. Briefly, tumour cell line SKMEL5 presenting target pMHC (PRAME peptide on HLA-A*02) were incubated with a dilution series of the test bispecific TCR-TCE and T effector cells provided as PBMC at an effector cell to target cell ratio of 5:1. Tumour cell killing was measured using an ATP assay, CellTiter Gio 2.0 (G9241, Promega).

[0365] Surprisingly, many TB2C bispecifics failed to exhibit target cell killing in this assay (Figure 15). However, it was unexpectedly discovered that reversing the order of the antibody variable domains within the scFv was able to “rescue” the potency of certain anti-CD3 moieties, restoring the ability of the TB2C bispecific proteins to mediate killing of target cells.

[0366] To examine the impact of anti-CD3 variable domain orientation impact on the killing profile of the bispecific proteins, a selection of anti-CD3 candidate antibodies were reformatted into VH-VL orientation, i.e., with the VH domain at the N terminus. The format, TCR and linkers were kept identical as before. TCR-TCE proteins were expressed and an in vitro killing assay was used to assess the impact of the VH-VL reformatting on potency.

[0367] Five candidate anti-CD3 scFv sequences and one benchmark TCE scFv (“AbX”) were directly compared as VH-VL or VL-VH orientation to examine the impact of variable domain orientation on their in vitro killing potency. Bispecifics where the TCE was formatted as VH- (G4S)4-VL exhibited more potent killing in most cases, e.g. B1-NY, 6-01, 6-04, 6-06, 5-51 , 5- 148, AbX. However, we did observe a few cases where VL-VH orientation was preferred, 5-145- 004, Rn70-F1-001. Figure 16 and Table E11-1. TCE scFv VL-(G4S)4-VH EC50 (nM) VH-(G4S)4-VL EC50 (nM) Ab name TB2C bispecific TB2C bispecific B1-NY CPA928 No killing CPA929 0.83 5-51 CPA1005 No killing CPA 1084 0.52

[0368] 5-148 CPA943 Weak killing CPA1082 44.28 AbX CPA904 Weak killing CPA1040 0.42

[0369] 6-01 CPA1008 No killing CPA1026 0.20 5-145-004 CPA967 0.99 CPA1085 5.01

[0370] Table E11-1. Killing potency of bispecific proteins in TB2C format with scFv VH-VL vs VL-VH. CPA numbers indicate the identity of the bispecific and corresponding graphs are shown in Figure 16.

[0371] Based on these data, it is proposed that optimal engagement of CD3 antigen by the TCE moiety is dependent on geometry of the bispecific format. Orientation of the linkage between VH and VL could influence rotation angles and geometry of the anti-CD3 moiety for interaction with its target epitope in TB2C format. The optimal pairing of an anti-CD3 moiety with a bispecific format is likely to be dependent on the CD3 epitope and the geometric orientation of the anti- CD3 moiety relative to that target epitope, this orientation being defined or constrained by the format.

[0372] Example 12: Optimisation of 1_1 linker length

[0373] The possibility of improving potency by extending linker L1 (between scFv and TCR) in the bispecific format TB2C was investigated using antibody 6-01, which had shown killing when in VH-VL orientation but not VL-VH orientation (Table E11-1 above).

[0374] Extension of L1 from (G4S)x1 to (G4S)x2 or (G4S)x3 did not improve the in vitro killing potency (Table E12-1, Figure 17).

[0375] Table E12-1. Killing potency of bispecific proteins in TB2C format with antibody 6-01 scFv VH- VL vs VL-VH. CPA numbers indicate the identity of the bispecific and corresponding graphs are shown in Figure 17. For 6-01 scFv, VH-VL orientation appears to be critical for potency in TB2C format.

[0376] Extending the length of linker L1 to increase flexibility of the anti-CD3 scFv relative to the TCR was insufficient to compensate for reversal of the variable domain orientation. This indicates that the “preference” for the VH-VL scFv orientation of some antibodies is very strong, and may be essential for measurable potency in a killing assay when the scFv is linked to the TCR N- terminus, as exemplified by format TB2C, even when the same antibody in a VL-VH scFv orientation does achieve killing potency in an alternative bispecific format.

[0377] The impact of L1 linker length on affinity for binding target antigen was also explored in TB2C format bispecifics with scFv antibody in its “preferred” orientation. Antibody B1-NY was converted into scFv in VH-(G4S)4-VL orientation and assembled into TB2C format with a TCR specific for a PRAME pMHC complex and with L1 linker of varying length. L1 linkers were varied as shown in Table E12-2.

[0378] Affinity for target pMHC were determined by SPR.

[0379] Table E12-2. Affinity and kinetics by SPR for TB2C bispecifics with a range of L1 linkers.

[0380] To determine the kinetics and affinities of TCR:TCE molecules binding to PRAME pMHC, 1 pg / mL PRAME pMHC solution was captured on Cytiva Biacore Series Sensor NTA chip. After capture, analytes (TCR:TCE molecules) at concentrations of 0 pg / mL, 0.2 pg / mL, 0.7 pg / mL, 2.2 pg / mL, 6.6 pg / mL, or 20 pg / mL were used to detect the binding on Biacore 8000K machine. Data were processed using Biacore Insight Evaluation software. Kinetics curves were fit using 1 :1 binding model, and affinity curves were fit using steady state affinity model. pMHC binding kinetics measured by SPR confirmed that L1 linker length within the tested range did not impact TCR affinity for pMHC binding, therefore indicating that even with a linker as short as G4S, anti-CD3 scFv linked to the N terminus of the TCR did not hinder TCR binding to target pMHC. The impact of L1 linker length on CD3 binding kinetics was also investigated. Two TCR- TCE bispecific proteins, CPA665 and CPA1290, differed only in the L1 linker, with CPA665 having G4S and CPA1290 having (G4S)2. The affinity, on-rate and off-rate were equivalent between CPA665 and CPA1290, suggesting that the difference in their L1 linkers did not significantly impact the CD3 binding affinity and kinetics.

[0381] Example 13 Impact of scFv domain orientation on stability

[0382] The anti-CD3 antibody LICHT1 is often used as scFv in VL-VH orientation, as exemplifed in the clinically approved TCR-TCE tebentafusp. No protein product could be obtained when attempting to express a TB2C bispecific with LICHT1 in VH-VL orientation. Thus, LICHT1 appears to be highly orientation specific for its stability.

[0383] Thermal stability of anti-CD3 candidate scFv was assessed for the two variable domain orientations (VH-VL vs VL-VH) in TB2C format, using thermal melting temperature determination methods. 10 pl of Sypro Orange (ThermoFisher S6651, 1:500 dilution of stock 5000X) was mixed 1:1 [Vol / Vol] with 5 pg TCR-TCE molecule in PBS per well of 96 well qPCR plate (Starlab E1403-7700). The plate was sealed with an optically transparent ABI PCR plate seal (Fisher Scientific 4311971). Real-time fluorescence was measured continuously for 2 minutes over a temperature gradient from 25°C to 99°C on a ViiA7 qPCR machine (Applied Biosystems) using Quant Studio Real Time PCR software. The apparent thermal melting temperature was determined by calculating the first derivatives of the melting curve and using the gradient maximum to define the Tm.

[0384] Table E13-1. Thermal stability for multispecific proteins containing candidate anti-CD3

[0385] Example 14: In vitro T cell mediated tumour cell killing with a panel of anti-CD3 scFv

[0386] TB2C bispecific proteins (see Example 11) were generated with candidate anti-CD3 scFv with linker L1 as (G4S)2. To evaluate the impact on tumour cell killing potency in vitro, cells of tumour cell line SKMEL5 presenting target pMHC (PRAME peptide on HLA-A*02) were incubated with a dilution series of the test TB2C candidate bispecific TCR-TCE and T effector cells, provided as PBMC at the effector cell to target cell ratio 5:1. Tumour cell killing was measured using ATP assay, CellTiter Gio 2.0 (G9241, Promega).

[0387] TB2C bispecifics containing anti-CD3 scFv of antibodies of a first lineage (B1-NY cluster), namely B1-NY (CPA1023) 6-04 (CPA1213) and 6-06 (CPA1214), (Figure 18), and TB2C bispecifics containing anti-CD3 scFv antibodies of a second lineage (5-51 cluster), namely 5-51 (CPA1215), 5-51-001 (CPA1217), 5-51-004 (CPA1219), 5-51-005 (CPA1220), 5- 51-006 (CPA1221), 5-51-007 (CPA1222), 5-51-009 (CPA1224) and 5-51-010 (CPA1225) (Figure 19), all exhibited dose-dependent killing of tumour cells. TB2C bispecifics containing anti-CD3 scFv of a range of different antibody sequences, represented by members from both clusters, mediated potent killing and reached maximum killing, comparable to benchmark TCR- TCE X and to a TB2C bispecific comprising LICHT1 anti-CD3 (CPA1290). EC50 values for all candidates were within 10 fold of each other.

[0388] Example 15. In vivo anti-tumour efficacy of TCR-TCE bispecific proteins

[0389] TCR-TCE bispecific proteins were generated in TB2C format (see Example 11) with a panel of different anti-CD3 scFv, with the anti-CD3 variable domain orientation selected to be the most potent for the individual anti-CD3 sequence (e.g., bispecific protein CPA1290 comprised LICHT1 VL-VH scFv). L1 linker was (G4S)x2. L2 and L3 linkers were G4SG4. A TCR specific for a PRAME-HLA-A*02 pMHC complex was used in all bispecifics and was engineered with stabilising mutations in the constant domains. In vivo efficacy was assessed in a xenograft cancer model.

[0390] NSG mice implanted with a human melanoma cancer line (Hs695T, s.c 3M / mouse) and engrafted with HLA matched PBMC (i.v. 10M / mouse) were randomised in 6 cohorts (5 mice per cohort) based on tumour volume (t-3d, Median Volume=71, SD 13.9 mm3) and each received administration of a single bispecific drug (i.v, 0.025 mg / Kg, Q1 W for 2 doses, blinded) while control cohort received no drug. Tumour volume was monitored throughout the experiment thrice weekly (callipers, vol=longxshort2 / 2). Confirmation of successful engraftment of hPBMCs in all animals was measured in peripheral blood (%hCD45 median 61.9, SD: 20.1; Figure 20(a)).

[0391] In vivo efficacy was tested at a low to medium dose designed to allow differentiation between candidates, allowing discrimination of subtle responses otherwise not observable when dose is maximized for efficacy. Immediate response was observed after the first dose (tO) in all drug treated cohorts - Figure 20(b). In this initial response (d0-d7) all drugs showed modest efficacy, but no clear distinction could be made between cohorts. After the second dose (d7-d14), two affinity modulated engagers (CPA1023 and CPA1215) showed sustained reduction in tumour volume and reached greater efficacy (partial response -50% reduction) when compared with other candidates and with a TCR-TCE comprising anti-CD3 antibody scFv LICHT1 where stable disease with an upward trajectory was observed (max -20% volume reduction).

[0392] Sequences

[0393]

Claims

Claims1. A soluble, multi-polypeptide multispecific protein for recruiting an immune cell to a target cell presenting a target peptide major histocompatibility complex (pMHC), wherein the protein comprises a pMHC binding arm comprising a dimer of a first portion and a second portion, an immune cell engager moiety, and an Fc region comprising a dimer of a first Fc portion and a second Fc portion, wherein the first portion of the pMHC binding arm is linked at its N terminus to the immune cell engager moiety, and wherein the second portion of the pMHC binding arm is linked at its N terminus to a C terminus of the first Fc portion, and an N terminus of the Fc region is linked to a C terminus of the pMHC binding arm.

2. A protein according to claim 1 , wherein(i) the N terminus of the second Fc portion is linked to the C terminus of the second portion of the pMHC binding arm; or(ii) the N terminus of the second Fc portion is linked to the C terminus of the first portion of the pMHC binding arm; or(iii) the N terminus of the first Fc portion is linked to the C terminus of the first portion of the pMHC binding arm.

3. A protein according to claim 1 or claim 2, wherein the pMHC binding arm comprises a free C terminus.

4. A protein according to any preceding claim, wherein the Fc region comprises a free N terminus and a free C terminus.

5. A protein according to any preceding claim, wherein the pMHC binding arm is a TCR molecule comprising a first TCR hemichain and a second TCR hemichain.

6. A soluble multispecific protein for recruiting an immune cell to a target cell presenting a target peptide major histocompatibility complex (pMHC), wherein the protein comprises a first polypeptide comprising a first TCR hemichain and a second polypeptide comprising a second TCR hemichain, wherein the first and second polypeptide heterodimerisevia the pairing of the first and second TCR hemichain to form a TCR molecule for binding pMHC, wherein the first TCR hemichain is linked at its N terminus to an immune cell engager moiety, and wherein the second TCR hemichain is linked at its N terminus to a first Fc chain and is linked at its C terminus to a second Fc chain, wherein the first and second Fc chain dimerise to form an Fc region.

7. A protein according to claim 5 or claim 6, wherein each TCR hemichain comprises an N terminal TCR variable domain and a C terminal TCR constant domain.

8. A protein according to claim 5 or claim 6, wherein the first TCR hemichain comprises a TCRp variable domain vp and the second TCR hemichain comprises a TCRa variable domain va.

9. A protein according to claim 8, wherein the first TCR hemichain comprises a TCRp variable domain vp and a TCRp constant domain cp and the second TCR hemichain comprises a TCRa variable domain va and a TCRa constant domain ca.

10. A protein according to claim 8 or claim 9, wherein the TCR constant domains of the first and second hemichain are inter-linked via one or more disulphide bonds.

11. A protein according to claim 10, wherein the TCR constant domains Ca and Cp are inter-linked via a diS bond between C95 in the alpha domain and C131 in the beta domain.

12. A protein according to claim 11, wherein the TCR constant domains Ca and Cp are inter-linked via a diS bond between C48 in the alpha domain and C57 in the beta domain.

13. A protein according to any of claims 5 to 12, wherein the first TCR hemichain comprises a free C terminus, optionally terminating at the end of a cp domain.

14. A protein according to any of claims 1 to 4, wherein the pMHC binding arm is an antibody molecule comprising an antibody heavy chain or fragment thereof and an antibody light chain or fragment thereof.

15. A protein according to any preceding claim, wherein the immune cell engager moiety comprises an antibody molecule.

16. A protein according to claim 15, wherein the immune cell engager moiety comprises a single chain antibody molecule, optionally scFv or dAb.

17. A protein according to claim 15 or claim 16, wherein the immune cell engager moiety comprises an antibody VH domain and an antibody VL domain, wherein the VH domain and VL domain pair to provide a binding site for an immune cell surface receptor.

18. A protein according to claim 17, wherein the immune cell engager moiety comprises an scFv, optionally comprising an N terminal VH domain fused by a peptide linker to a C terminal VL domain or an N terminal VL domain fused by a peptide linker to a C terminal VH domain.

19. A protein according to claim 15, wherein the immune cell engager moiety comprises a single antibody variable (e.g., heavy) chain domain (dAb).

20. A protein according to any preceding claim, wherein the first and second Fc portion each comprise human CH2-CH3.

21. A protein according to any preceding claim, wherein the first Fc portion comprises a free N terminus.

22. A protein according to claim 20 or claim 21, wherein the first polypeptide comprises, from N to C terminus, an immune cell engager antibody molecule, a linker L1 , a first TCR variable domain and a first TCR constant domain, and wherein the second polypeptide comprises, from N to C terminus, a first Fc CH2-CH3, a linker L2, a second TCR variable domain, a second TCR constant domain, a linker L3, and a second Fc CH2-CH3.

23. A protein according to claim 22, wherein the first polypeptide comprises, from N to C terminus,(i) an immune cell engager VH-VL scFv antibody molecule, a linker L1 , a TCR vp domain and a TCR cp domain, and wherein the second polypeptide comprises, from N to C terminus, a first Fc CH2-CH3, a linker L2, a TCR va domain, a TCR co domain, a linker L3, and a second Fc CH2-CH3; or(ii) an immune cell engager VL-VH scFv antibody molecule, a linker L1 , a TCR vp domain and a TCR cp domain, and wherein the second polypeptide comprises, from N to C terminus, a first Fc CH2-CH3, a linker L2, a TCR va domain, a TCR co domain, a linker L3, and a second Fc CH2-CH3.

24. A protein according to claim 22 or claim 23, wherein linker L1 is (G4S)2.

25. A protein according to any of claims 22 to 24, wherein linker L2 is (G4S)3.

26. A protein according to any of claims 22 to 25, wherein linker L3 is (G4S)3.

27. A protein according to any of claims 1 to 5 or 7 to 21 , wherein the N terminus of the second Fc portion is linked to the C terminus of the first portion of the pMHC binding arm.

28. A protein according to claim 27, comprising a first polypeptide which comprises, from N to C terminus, the immune cell engager antibody molecule, a linker, the first portion of the pMHC binding arm (optionally comprising first TCR variable domain and first TCR constant region), and the second Fc portion (optionally comprising second CH2-CH3), and a second polypeptide which comprises, from N to C terminus, the first Fc portion (optionally comprising first CH2-CH2), a linker, and the second portion of the pMHC binding arm (optionally comprising second TCR variable domain and second TCR constant region).

29. A protein according to claim 28, wherein(i) the first polypeptide comprises, from N to C terminus, immune cell engager VH-VL scFv antibody molecule, linker, TCR vp domain, TCR cp domain, and second Fc CH2-CH3, and the second polypeptide comprises, from N to C terminus, first Fc CH2-CH3, linker, TCR va domain, and TCR ca domain; or(ii) the first polypeptide comprises, from N to C terminus, immune cell engager VL-VH scFv antibody molecule, linker, TCR vp domain, TCR cp domain, and second Fc CH2-CH3, and the second polypeptide comprises, from N to C terminus, first Fc CH2-CH3, linker, TCR va domain, and TCR ca domain.

30. A protein according to any of claims 1 to 5 or 7 to 21 , wherein the N terminus of the first Fc portion is linked to the C terminus of the first portion of the pMHC binding arm.

31. A protein according to claim 30, comprisinga first polypeptide which comprises, from N to C terminus, the immune cell engager antibody molecule, a linker, the first portion of the pMHC binding arm (optionally comprising first TCR variable domain and first TCR constant region), the first Fc portion (optionally comprising first CH2-CH3), and the second portion of the pMHC binding arm (optionally comprising second TCR variable domain and second TCR constant region), and a second polypeptide which comprises the second Fc portion, optionally comprising second CH2-CH3.

32. A protein according to claim 31, wherein(i) the first polypeptide comprises, from N to C terminus, immune cell engager VH-VL scFv antibody molecule, linker, TCR vp domain, TCR cp domain, first Fc CH2-CH3, TCR va domain, and TCR Co domain, and the second polypeptide comprises, from N to C terminus, second Fc CH2-CH3; or(ii) the first polypeptide comprises, from N to C terminus, immune cell engager VL-VH scFv antibody molecule, linker, TCR vp domain, TCR cp domain, first Fc CH2-CH3, TCR va domain, and TCR Ca domain, and the second polypeptide comprises, from N to C terminus, second Fc CH2-CH3.

33. A protein according to any preceding claim, wherein the pMHC binding arm (e.g., TCR molecule) binds a PRAME peptide presented on pMHC, optionally peptide SLLQHLIGL.

34. Nucleic acid encoding a protein according to any of claims 1 to 33.

35. A population of cells in vitro containing recombinant nucleic acid according to claim 34.

36. A method of manufacturing a protein according to any of claims 1 to 33, comprising culturing a population of cells according to claim 35 under conditions for expression of the protein, and recovering the protein, optionally including isolating and purifying the protein and formulating it into a composition with one or more pharmaceutically acceptable excipients.

37. A population of cells according to claim 35 or a method according to claim 36, wherein the cells are CHO cells comprising random genomic integration of first and second gene expression vectors encoding the first and second polypeptides respectively.

38. A method of treating a condition in a patient, comprising administering a soluble multispecific protein as defined in any of claims 1 to 33, or its encoding nucleic acid, to the patient.

39. A soluble multispecific protein according to any of claims 1 to 33, or its encoding nucleic acid, for use in method of treatment of the human or animal body by therapy.

40. A method according to claim 39 or a soluble multispecific protein or nucleic acid for use according to claim 39, wherein the treatment comprises treating a solid tumour.

Images