WO2025133609A1

WO2025133609A1 - Polypeptide binding glypican-3

Info

Publication number: WO2025133609A1
Application number: PCT/GB2024/053164
Authority: WO
Inventors: Claire RODDIE; Martin PULÉ; Harriet RODDY
Original assignee: UCL Business Ltd
Current assignee: UCL Business Ltd
Priority date: 2023-12-20
Filing date: 2024-12-19
Publication date: 2025-06-26
Anticipated expiration: 2026-06-20
Also published as: GB202319601D0

Abstract

The present invention relates to a polypeptide comprising an antigen-binding domain which binds to Glypican-3 (GPC3), wherein binding is not inhibited by a soluble GPC3. The invention also relates to chimeric antigen receptors (CARs) which comprise such polypeptides.

Description

POLYPEPTIDE FIELD OF THE INVENTION The present invention relates to a polypeptide comprising an antigen-binding domain which binds to Glypican-3 (GPC3), wherein binding is not inhibited by a soluble GPC3. The invention also relates to chimeric antigen receptors (CARs) which comprise such polypeptides. Cells expressing CARs which bind GPC3 are useful in the treatment of hepatocellular carcinoma (HCC). BACKGROUND TO THE INVENTION Hepatocellular carcinoma (HCC) accounts for 80% of all primary liver cancers and is a leading cause of cancer related mortality. Current treatments for early-stage HCC include tumour resection, liver transplantation or ablative therapy, and intermediate stage disease is managed with intra-arterial therapy. However, patients with more extensive liver disease, portal vein involvement and/or extrahepatic disease require systemic therapy. Protein tyrosine kinase inhibitors (TKIs) (sorafenib, lenvatinib, regorafenib and cabozantinib), and monoclonal antibodies targeting vascular endothelial growth factor (VEGF) receptor 2 (ramucirumab) and the Programmed Death Receptor-1 (PD-1) immune checkpoint (nivolumab) have demonstrated modest improvements in clinical outcome compared with controls. Glypican-3 (GPC3) is a 70 kDa, glycosylphosphatidylinositol (GPI) anchored, 580 amino acid heparan sulphate proteoglycan that is expressed in 72% of cases of HCC, where it portends to poor prognosis (Capurro et al., Gastroenterology, 2003; 125(1):89–97). Critically, GPC3 is minimally expressed on other tissues including normal and cirrhotic liver. Similar to other glypicans, GPC3 can be released from the cell surface and can be found around tumours or in circulation. Soluble GPC3 is found in the serum of 53% of HCC patients and is under development as a disease biomarker (Capurro et al., Gastroenterology, 2003; 125(1):89–97). Several studies have shown that serum GPC3 levels are significantly elevated in HCC patients compared with healthy individuals (Capurro et al., Gastroenterology, 2003; 125(1):89–97; Xu et al., Ann Hepatol, 2019; 18(1):58–67; Liu et al., Clin Biochem, 2020; 79:54–60) and that local tumour GPC3 concentrations in HCC may also be much higher than those in normal tissues. Previous studies also showed that recombinant soluble GPC3 could inhibit HCC cell growth in vitro and in vivo (Zittermann et al., Int J Cancer, 2010; 126(6):1291– 301; Feng et al., Int J Cancer, 2011; 128(9):2246–7), suggesting that shed GPC3 may 1 compete with cell-surface GPC3 to bind GPC3-interacting molecules or may even block GPC3-targeted therapies. Recently, chimeric antigen receptor (CAR) T cells have shown promising curative effects in haematological tumours, and work is being undertaken to transfer the success to solid tumours. There have been multiple studies in the development of new GPC3-targeted therapies in HCC and some preliminary results using monoclonal antibodies and CAR T cells have been published. However, the response rate has been limited and the results far from satisfactory. To date the only scFv tested with published clinical trial data has been GC33 in both CAR T cell and whole antibody format (codrituzumab) (Shi et al., Clin Cancer Res, 2020; 26(15):3979–89; Abou-Alfa et al., J Hepatol, 2016; 65(2):289–95). There is therefore a need for improved GPC3 antigen-binding domains, in particular for use in CAR T therapy. SUMMARY OF THE INVENTION The present inventors have developed novel GPC3 binders with improved properties. Surprisingly, the present binders interact with membrane bound GPC3 but are not inhibited by soluble GPC3. The binders may be useful be useful in antigen binding entities such as antibodies, chimeric antigen receptors (CARs) and bi-specific T cell engagers (BiTEs), in particular for the treatment of HCC. The invention provides a polypeptide comprising an antigen-binding domain which binds to a region between amino acids S495 to H580 of Glypican-3 (GPC3), wherein binding is not inhibited by a soluble GPC3. In other words, the binding between the polypeptide comprising the antigen-binding domain and the region between amino acids S495 to H580 of GPC3 is not inhibited by soluble GPC3. In one embodiment, the antigen-binding domain comprises a heavy chain variable region (VH) having complementarity determining regions (CDRs): HCDR1, HCDR2 and HCDR3; and a light chain variable region (VL) having CDRs: LCDR1, LCDR2 and LCDR3 selected from the following: i. HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; or ii. HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12; 2 optionally wherein one or more of the CDRs comprises one, two or three amino acid mutations. Preferably, the antigen-binding domain comprises: i. HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; or ii. HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12. In one embodiment, the antigen-binding domain comprises: i. a VH region having the sequence shown as SEQ ID NO: 13, or a variant having at least 80% sequence identity thereto; and a VL region having the sequence shown as SEQ ID NO: 14, or a variant of having at least 80% sequence identity thereto; or ii. a VH region having the sequence shown as SEQ ID NO: 15, or a variant having at least 80% sequence identity thereto; and a VL region having the sequence shown as SEQ ID NO: 16, or a variant of having at least 80% sequence identity thereto. Preferably, the antigen-binding domain comprises: i. the sequence shown as SEQ ID NO: 17, or a variant having at least 80% sequence identity thereto; or ii. the sequence shown as SEQ ID NO: 18, or a variant having at least 80% sequence identity thereto. The invention provides a polypeptide comprising an antigen-binding domain which binds to a region between amino acids S495 to H580 of Glypican-3 (GPC3), wherein the antigen-binding domain comprises a heavy chain variable region (VH) having complementarity determining regions (CDRs): HCDR1, HCDR2 and HCDR3; and a light chain variable region (VL) having CDRs: LCDR1, LCDR2 and LCDR3 selected from the following: i. HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; or ii. HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12. 3 The polypeptide of the invention may be an antibody or antigen-binding fragment thereof comprising an antigen-binding domain as defined in the invention. Preferably, the antibody or antigen-binding fragment thereof is a scFv, a monoclonal antibody or fragment thereof, a humanized antibody or fragment thereof, or a bi-specific T cell activator molecule, such as a bi-specific T cell engager (BiTE). The antibody or antigen-binding fragment may be conjugated to a cargo or payload component. Thus, the invention provides an antibody conjugate comprising the antibody or antigen-binding fragment of the invention. The invention further provides a chimeric antigen receptor (CAR) comprising a polypeptide or antibody or antigen-binding fragment thereof according to the invention. The CAR may comprise a transmembrane domain, preferably a CD28 transmembrane domain. The polypeptide comprising an antigen-binding domain and the transmembrane domain of the CAR may be connected by a spacer, preferably the spacer comprises a CD28 hinge. The CAR may comprise an intracellular T cell signalling domain, preferably a CD28 endodomain and a CD3-Zeta endodomain. In one embodiment, the CAR comprises a sequence selected from the group comprising SEQ ID NO: 19 or SEQ ID NO: 20, or a variant which has at least 80% sequence identity thereto and retains the capacity to i) bind GPC3 and ii) induce T cell signalling. The invention provides a polynucleotide comprising a nucleic acid sequence encoding a polypeptide comprising an antigen-binding domain according to the invention. The invention further provides a polynucleotide comprising a nucleic acid sequence encoding an antibody or antigen-binding fragment thereof according to the invention. The invention further provides a polynucleotide comprising a nucleic acid sequence encoding a CAR according to the invention. The invention provides a vector which comprises a polynucleotide according to the invention. The invention provides a cell which comprises a CAR according to the invention. The invention further provides a cell which comprises a polynucleotide according to the invention. The invention further provides a cell which comprises a vector according to the invention. Preferably, the cell is a T cell or a natural killer (NK) cell. 4 The invention provides a method for making a cell according to the invention, which comprises the step of introducing a polynucleotide or a vector according to the invention into said cell. The invention provides a pharmaceutical composition which comprises a polypeptide comprising an antigen-binding domain according to the invention, an antibody or antigen- binding fragment thereof according to the invention, an antibody conjugate according to the invention, a polynucleotide according to the invention, a vector according to the invention, or a cell according to the invention, together with a pharmaceutically acceptable carrier, diluent or excipient. The invention provides a polypeptide comprising an antigen-binding domain according to the invention, an antibody or antigen-binding fragment thereof according to the invention, an antibody conjugate according to the invention, a polynucleotide according to the invention, a vector according to the invention, a cell according to the invention, or a pharmaceutical composition according to the invention, for use in a method of treating a disease in a subject. The invention provides a method for treating a disease which comprises the step of administering a polypeptide comprising an antigen-binding domain according to the invention, an antibody or antigen-binding fragment thereof according to the invention, an antibody conjugate according to the invention, a polynucleotide according to the invention, a vector according to the invention, a cell according to the invention, or a pharmaceutical composition according to the invention to a subject. The invention provides for use of a polypeptide comprising an antigen-binding domain according to the invention, an antibody or antigen-binding fragment thereof according to the invention, an antibody conjugate according to the invention, a polynucleotide according to the invention, a vector according to the invention, a cell according to the invention, or a pharmaceutical composition according to the invention in the manufacture of a medicament for treating a disease. Suitably, the disease to be treated is cancer. The cancer may be hepatocellular carcinoma (HCC), melanoma, ovarian clear-cell carcinomas, yolk sac tumour, neuroblastoma, hepatoblastoma, Wilms' tumor cells, rhabdoid tumors, and rhabdomyosarcomas. The cancer is preferably HCC. DESCRIPTION OF THE FIGURES Figure 1 – Shed GPC3 is clinically relevant and requires binder generation specific for this antigen characteristic. (a) Schematic cartoon of GPC3 antigen detailing the site of furin 5 cleavage in the N domain and site of shedding proximal to the GPI anchor. Binders are targeted to the antigen/GPI junction to circumvent the scFv being inhibited by ‘shed’ antigen in both patient serum and tumour microenvironment. (b) Using a group of patients with either secondary metastasis (non HCC) in the liver or primary HCC samples, shed GPC3 was detected by ELISA at a range of concentrations with a mean of approximately 400ng/mL exclusively in primary HCC. (c) Constructs used to generate overexpressing cell lines, these retroviral backbones include GPC3 with signal peptide and GPI anchor for both, one is a truncated version of the protein finishing at residue 495, both constructs contain eGFP as a marker of transduction and for use as target detection in coculture based assays. (d) Staining with commercial antibody clone (1G12) shows transduction of both cell lines when compared to un-transduced cells and a linear relationship between eGFP and antigen. (e) Cell line validation of antigen density. SupT1 transduced cells overexpress the antigen when compared to both human HepG2 and murine Hepa1/6 transduced to express human GPC3 to a ‘native’ expression level. Figure 2 – Identification of scFvs that bind to GPC3. (a) The schematic structure of the retroviral construct for scFv testing including a murine IgG2a Fc region and a eBFP for confirmation of transfection for protein production. (b) Binding of scFv to full membrane protein in presence of shed protein by ELISA (c) The KD, Kon, and Koff value of the lead scFvs compared to GC33 were determined by surface plasma resonance. The data from one representative experiment is shown. The error refers to the deviation between the fitting curve and the test curve. (d) Curves from SPR data fitting of GC33 control and two best selected binders. All binders have similar kinetic characteristics. SP, signal peptide of human IgG; VH, variable region of heavy chain; VL, variable region of light chain. Figure 3 – The present binders have potent function against cell lines in vitro. (a) Vector design of anti-GPC3 CARs used in the study, with RQR8 marker/sort-suicide gene. (b) Cytotoxicity of GC33-CAR vs present binders against overexpressing cell lines, data normalized to target alone condition. 48-hour coculture conditions. Ratios shown above graphs refer to effector:target (E:T) ratios. (c) Fold expansion of T cells after 7-day co-culture with Mitomycin-C treated SupT1 cells at 1:1 E:T ratio, 4 donors average from repeats shown. (d) Secretion of interleukin-2 (IL-2); top and interferon-γ (IFN-γ) in 48-hour co-culture, 1:1 E:T ratio as in final graph of panel B. (e) Cytotoxicity of GC33-CAR vs novel binders against native expressing cell line HepG2 at a 2:1 E:T ratio measured by specific cell lysis, data normalized to target alone condition. (f) Absolute cell count of T cells after 7-day co-culture with Mitomycin- C treated HepG2 at 1:1 E:T ration, 4 donors average repeats shown. (g) Cytotoxicity in the presence of ’shed antigen’ at 0,5,10 μg/mL GC33 cytotoxicity is reduced against GPC3 against 6 increasing concentration of antigen in media, and is non-functional against a shortened version of the antigen but the present binders HR8 and HR10 retain cytotoxic function. *P < .05, **P < .01, ***P < .001, ****P < .0001. ns, not significant. EFF, effectors. Figure 4 – Illustrative CARs have potent antitumor activity in a HepG2 xenograft model including in presence of shed antigen. (a) Schematic of murine HepG2 model. (b) Bioluminescence signal in mice in study. (c) Bioluminescence signal in mice in study where CARs were injected alongside shed protein. (d) At endpoint of experiment in panel D, T cells were counted using flowcytometry from mouse spleens to determine expansion. n = 5 per group. BLI, bioluminescence imaging; i.p, intra-peritoneal; i.v, intravenous. Figure 5 – Illustrative CARs have potent antitumor activity in a murine immunocompetent xenograft model of HCC. (a) Schematic of murine Hepa 1/6 model. (b) Tumour volume of mice in study. (c) At D14 tail vein bleed for expansion of T Cells, CARs (d) and percentage of T cells that were CARs (e) cells were counted using flow cytometry. Experiment performed twice; representative data shown. n = 5 per group. Subcut, subcutaneous; i.v, intravenous. Experimental endpoint was at day 40 for this experiment as NT tumours were close to maximum volume and tumours that had regrown had regrown to starting volume. Figure 6 – Reduction in binding in the presence of shed antigen. (a) Binding assay. Crude supernatant comprising scFv (or supernatant from NT cells) was incubated with SupT1 GPC3+ target cells in the presence of different amounts of sGPC3 (0, 5, 10 μg/mL). Cells were washed and stained with secondary antibody. The dotted line shows median fluorescent intensity (MFI) of maximal binding. (b) Graph showing MFI values derived from (a). Figure 7 – Killing of target cell lines by CAR T in the presence of shed antigen. Absolute cell numbers (top) and percentage normalised to NT (bottom) of CD19+ (left) GPC3+ (right) target cells following incubation with CAR T cells in the presence of different amounts of sGPC3 (0, 5, 10 μg/mL). The data points represent testing with two donors in triplicate. Figure 8 – Statistical analysis of killing of GPC3+ target cells (% normalised to NT) as determined by ANOVA. Only statistically significant results are plotted on the graph. 7 DETAILED DESCRIPTION OF THE INVENTION ANTIGEN-BINDING DOMAIN The present invention provides a polypeptide comprising an antigen binding domain which binds to a region between amino acids S495 to H580 of GPC3, wherein binding is not inhibited by a soluble GPC3. The antigen binding domains described herein are able to specifically bind to GPC3. The antigen binding domain may be capable of selectively binding to the membrane bound form of GPC3 and not to soluble GPC3. The invention also provides a polypeptide comprising an antigen-binding domain which binds to a region between amino acids S495 to H580 of GPC3, wherein the antigen-binding domain comprises a heavy chain variable region (VH) having complementarity determining regions (CDRs): HCDR1, HCDR2 and HCDR3; and a light chain variable region (VL) having CDRs: LCDR1, LCDR2 and LCDR3 selected from the following: i. HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; or ii. HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12. The antigen-binding domain of an antibody, such as an IgG molecule, is made up of two variable domains, one derived from the heavy chain (VH) and one derived from the light chain (VL). The antigen-binding domain of the present invention may comprise both VH and VL domains. Antigen binding domains of the present invention comprising a VH and a VL domain may, for example, be in a single-chain variable fragment (scFv) or Fab format. Suitably, a VH region may comprise the CDR1, CDR2 and CDR3 sequences. Suitably, a VL region may comprise the CDR4, CDR5 and CDR6 sequences. CDR4, CDR5 and CDR6 may be presented or re-numbered as CDR1, CDR2 and CDR3 of the VL region, respectively. Therefore, the CDRs of the VH region may be numbered as HCDR1, HCDR2 and HCDR3, and the CDRs of the VL region may be numbered as LCDR1, LCDR2 and LCDR3. The antigen-binding domain may comprise a VH region having CDRs: HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; and a VL region having CDRs: LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; wherein one or more of the CDRs comprises one, two or three amino acid mutations. 8 The antigen-binding domain may comprise a VH having CDRs: HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; and a VL having CDRs: LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6.

The antigen-binding domain may comprise a VH having CDRs: HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; and a VL having CDRs: LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12; wherein one or more of the CDRs comprises one, two or three amino acid mutations. The antigen-binding domain may comprise a VH having CDRs: HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; and a VL having CDRs: LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12.

The antigen binding domain may comprise a VH having the sequence shown as SEQ ID NO: 13, or a variant having at least 80%, 85%, 90%, 95%, 96%, 97%, 98%, or 99% sequence identity thereto; and a VL having the sequence shown as SEQ ID NO: 14, or a variant of having at least 80%, 85%, 90%, 95%, 96%, 97%, 98%, or 99% sequence identity thereto. The antigen binding domain may comprise a VH having the sequence shown as SEQ ID NO: 13; and a VL having the sequence shown as SEQ ID NO: 14. HR8 VH (SEQ ID NO: 13): 9 QVQLQQSGAELVKPGTSVKLSRKASGYTFTANHMNWIKQTTGQGLECIGIINPGSGGTSYNVKFRGRA TLTVDKSSSTAFMQLSGLTPEDSAVYYCARSGYGGPHYWGQGVMVTVSS HR8 VL (SEQ ID NO: 14): DIVMTQGALPNPVPSGESASITCQSSKSLLHSNGKTYLNWYLQRPGQSPQLLIYWMSTRASGVSDRFS GSGSGTDFTLKISSVEAEDVGVYYCQQFLEYPLTFGSGTKLEIK The antigen binding domain may comprise a VH having the sequence shown as SEQ ID NO: 15, or a variant having at least 80%, 85%, 90%, 95%, 96%, 97%, 98%, or 99% sequence identity thereto; and a VL having the sequence shown as SEQ ID NO: 16, or a variant of having at least 80%, 85%, 90%, 95%, 96%, 97%, 98%, or 99% sequence identity thereto. The antigen binding domain may comprise a VH having the sequence shown as SEQ ID NO: 15; and a VL having the sequence shown as SEQ ID NO: 16. HR10 VH (SEQ ID NO: 15): GVQLQKSGAELVRPGTSVKLSCKVSGDTITFYYMHFVKQRPGQGLEWIGRIDPEDESTKYSEKFKNKA TLTADTSSNTAYLKLSSLTSEDTATYFCIYGGYYFDYWGQGVMVTVSS HR10 VL (SEQ ID NO: 16): DIVLTQSPTTLSVTPGETVSLSCRASHDIGTNLHWYQQKTNESPRLLIKSASQPTSGIPSRFSASGSG TDFTLNINNVEFDDVSSYLCQQTQSWPLTFGSGTKLELK The antigen-binding domain may comprise the amino acid sequence shown as SEQ ID NO: 17 or a variant thereof having at least 80%, 85%, 90%, 95%, 96%, 97% 98% or 99% sequence identity thereto. The antigen-binding domain may comprise the amino acid sequence shown as SEQ ID NO: 17. HR8 scFv (SEQ ID NO: 17): QVQLQQSGAELVKPGTSVKLSRKASGYTFTANHMNWIKQTTGQGLECIGIINPGSGGTSYNVKFRGRA TLTVDKSSSTAFMQLSGLTPEDSAVYYCARSGYGGPHYWGQGVMVTVSSGGGGSGGGGSGGGGSDIVM TQGALPNPVPSGESASITCQSSKSLLHSNGKTYLNWYLQRPGQSPQLLIYWMSTRASGVSDRFSGSGS GTDFTLKISSVEAEDVGVYYCQQFLEYPLTFGSGTKLEIK The antigen-binding domain may comprise the amino acid sequence shown as SEQ ID NO: 18 or a variant thereof having at least 80%, 85%, 90%, 95%, 96%, 97% 98% or 99% sequence identity thereto. The antigen-binding domain may comprise the amino acid sequence shown as SEQ ID NO: 18. 10 HR10 scFv (SEQ ID NO: 18): GVQLQKSGAELVRPGTSVKLSCKVSGDTITFYYMHFVKQRPGQGLEWIGRIDPEDESTKYSEKFKNKA TLTADTSSNTAYLKLSSLTSEDTATYFCIYGGYYFDYWGQGVMVTVSSGGGGSGGGGSGGGGSDIVLT QSPTTLSVTPGETVSLSCRASHDIGTNLHWYQQKTNESPRLLIKSASQPTSGIPSRFSASGSGTDFTL NINNVEFDDVSSYLCQQTQSWPLTFGSGTKLELK Suitably, the antigen-binding domain comprises a VH and a VL linked by a linker to form an scFv. An example of a linker is GGGGSGGGGSGGGGS (SEQ ID NO: 21). The term “polypeptide” is used in the conventional sense to mean a series of amino acids, typically L-amino acids, connected one to the other, typically by peptide bonds between the α- amino and carboxyl groups of adjacent amino acids. The term “polypeptide” is used interchangeably with the terms “amino acid sequence”, “peptide” and/or “protein”. The term “residues” is used to refer to amino acids in an amino acid sequence. The term "variant" refers to a polypeptide that has an equivalent function to the amino acid sequences described herein, but which includes one or more amino acid substitutions, insertions or deletions. The terms “selectively binds/selectively binding” and “specifically binds/specifically binding” may be used interchangeably herein. “Heavy chain variable region” or “VH” refers to the fragment of the heavy chain of an antigen- binding domain or antibody that contains three CDRs interposed between flanking stretches known as framework regions, which are more highly conserved than the CDRs and form a scaffold to support the CDRs. “Light chain variable region” or “VL” refers to the fragment of the light chain of an antigen-binding domain or antibody that contains three CDRs interposed between framework regions. “Complementarity determining region” or “CDR” with regard to antigen-binding domain or antibody or antigen-binding fragment thereof refers to a highly variable loop in the variable region of the heavy chain of the light chain of an antibody. CDRs can interact with the antigen conformation and largely determine binding to the antigen (although some framework regions are known to be involved in binding). The heavy chain variable region and the light chain variable region each contain 3 CDRs (heavy chain CDRs 1, 2 and 3 and light chain CDRs 1, 2 and 3, numbered from the amino- to the carboxy-terminus). CDR regions are well known to those skilled in the art and have been defined by, for example, Kabat as the regions of most hypervariability within the antibody variable (V) domains (Kabat et al., 1977, J. Biol. Chem. 252:6609-6616; Kabat, 1978, Adv. Prot. Chem. 32:1-75). CDR 11 region sequences also have been defined structurally by Chothia as those residues that are not part of the conserved β-sheet framework, and thus are able to adapt different conformations (Chothia and Lesk, 1987, J. Mol. Biol.196:901-917). Alternatively, IMGT or EU numbering may be used. These terminologies are well recognized in the art. The positions of CDRs within a canonical antibody variable domain have been determined by comparison of numerous structures (Al-Lazikani et al., 1997, J. Mol. Biol.25273:927-948; Morea et al., 2000, Methods 20:267-279). Because the number of residues within a hypervariable region varies in different antibodies, additional residues relative to the canonical positions are conventionally numbered with a, b, c and so forth next to the residue number in the canonical variable domain numbering scheme (Al-Lazikani et al., supra). Such nomenclature is similarly well known to those skilled in the art. Suitably, the antigen binding domain may be defined by the presence of HCDRs and LCDRs determined according to CDR numbering schemes which are known in the art. For example, the CDRs may be defined according to the IMGT, Chothia and/or Kabat numbering schemes. In some embodiments, the CDRs of the antigen-binding domain are defined according to the IMGT numbering scheme. It may be possible to introduce one or more mutations (substitutions, additions or deletions) into each CDR without negatively affecting binding activity. Each CDR may, for example, have one, two or three amino acid mutations. Identity comparisons can be conducted by eye, or more usually, with the aid of readily available sequence comparison programs. These commercially available computer programs can calculate % identity between two or more sequences. A suitable computer program for carrying out such an alignment is the GCG Wisconsin Bestfit package (University of Wisconsin, U.S.A.; Devereux et al., 1984, Nucleotide sequences Research 12:387). Examples of other software than can perform sequence comparisons include, but are not limited to, the BLAST package (see Ausubel et al., 1999 ibid – Chapter 18), FASTA (Atschul et al., 1990, J. Mol. Biol., 403-410) and the GENEWORKS suite of comparison tools. Both BLAST and FASTA are available for offline and online searching. For example, the percentage identity between two polypeptide sequences may be readily determined by BLAST which is freely available at http://blast.ncbi.nlm.nih.gov. Once the software has produced an optimal alignment, it is possible to calculate % identity. The software typically does this as part of the sequence comparison and generates a numerical result. 12 The sequence may have one or more deletions, insertions or substitutions of amino acid residues which produce a silent change and result in a functionally equivalent molecule. These sequences are encompassed by the present invention. Deliberate amino acid substitutions may be made on the basis of similarity in polarity, charge, solubility, hydrophobicity, hydrophilicity, and/or the amphipathic nature of the residues as long as the activity is retained. For example, negatively charged amino acids include aspartic acid and glutamic acid; positively charged amino acids include lysine and arginine; and amino acids with uncharged polar head groups having similar hydrophilicity values include leucine, isoleucine, valine, glycine, alanine, asparagine, glutamine, serine, threonine, phenylalanine, and tyrosine. The invention further provides an antigen-binding domain which binds to a region between amino acids S495 to H580 of GPC3, wherein binding is not inhibited by a soluble GPC3. The invention further provides an antigen-binding domain which binds to a region between amino acids S495 to H580 of GPC3, wherein the antigen-binding domain comprises a VH having CDRs: HCDR1, HCDR2 and HCDR3; and a VL having CDRs: LCDR1, LCDR2 and LCDR3 selected from the following: i. HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; or ii. HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12. Other entities comprising antigen-binding domains, such as aptamers, are contemplated by the present invention. Thus, the invention also provides for an entity comprising an antigen- binding domain which binds to a region between amino acids S495 to H580 of GPC3, wherein binding is not inhibited by a soluble GPC3. GPC3 The GPC3 gene encodes a 580 amino acid, 70 kDa precursor protein, which is cleaved by furin between residues Arg358 and Ser359 to generate an N-terminal subunit (~40 kDa) and a C-terminal subunit (~30 kDa). The two subunits can be connected by one or more disulphide bonds. The C-terminal subunit comprises heparan sulfate modification at two sites (Ser495 and Ser509). GPC3 binds to the cell membrane via a glycosylphosphatidylinositol (GPI) anchor; Ser560 of GPC3 inserts into the lipid bilayer and anchors the protein to the bilayer by phosphatidylinositol. The amino acid sequence for human GPC3 is (SEQ ID NO: 22): 13 MAGTVRTACLVVAMLLSLDFPGQAQPPPPPPDATCHQVRSFFQRLQPGLKWVPETPVPGSDLQVCLPK GPTCCSRKMEEKYQLTARLNMEQLLQSASMELKFLIIQNAAVFQEAFEIVVRHAKNYTNAMFKNNYPS LTPQAFEFVGEFFTDVSLYILGSDINVDDMVNELFDSLFPVIYTQLMNPGLPDSALDINECLRGARRD LKVFGNFPKLIMTQVSKSLQVTRIFLQALNLGIEVINTTDHLKFSKDCGRMLTRMWYCSYCQGLMMVK PCGGYCNVVMQGCMAGVVEIDKYWREYILSLEELVNGMYRIYDMENVLLGLFSTIHDSIQYVQKNAGK LTTTIGKLCAHSQQRQYRSAYYPEDLFIDKKVLKVAHVEHEETLSSRRRELIQKLKSFISFYSALPGY ICSHSPVAENDTLCWNGQELVERYSQKAARNGMKNQFNLHELKMKGPEPVVSQIIDKLKHINQLLRTM SMPKGRVLDKNLDEEGFESGDCGDDEDECIGGSGDGMIKVKNQLRFLAELAYDLDVDDAPGNSQQATP KDNEISTFHNLGNVHSPLKLLTSMAISVVCFFFLVH Cleavage site is underlined; heparin sulfate/phosphatidylinositol modification sites are shown in bold. The signal peptide (residues 1 to 24), which may be cleaved in the mature protein, is shown in italics. Reference to amino acid positions in SEQ ID NO: 22 throughout refers to the full length polypeptide sequence, including signal peptide. The antigen-binding domain according to the present invention binds to a region between amino acids S495 to H580 of GPC3. In some embodiments, the antigen-binding domain according to the present invention binds to an epitope solely within (i.e. that consists within) amino acids S495 to H580 of GPC3, presented below as SEQ ID NO: 23. In some embodiments, the antigen-binding domain according to the present invention binds to a region within SEQ ID NO: 23. N-terminal region of GPC3 (SEQ ID NO: 23): SGDCGDDEDECIGGSGDGMIKVKNQLRFLAELAYDLDVDDAPGNSQQATPKDNEISTFHNLGNVHSPL KLLTSMAISVVCFFFLVH In some embodiments, the antigen-binding domain according to the present invention binds to a region within positions 495 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. In some embodiments, the antigen-binding domain according to the present invention binds to an epitope solely within positions 495 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. In some embodiments, the antigen-binding domain according to the present invention binds to a region within positions 495 to 580, 500 to 580, 505 to 580, 510 to 580, 515 to 580, 520 to 580, 525 to 580, 530 to 580, 535 to 580, 540 to 580, 545 to 580, 550 to 580, 555 to 580, 560 to 580, 565 to 580, 570 to 580, or 575 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. 14 In some embodiments, the antigen-binding domain according to the present invention binds to a region within positions 551 to 580, 552 to 580, 553 to 580, 554 to 580, 555 to 580, 556 to 580, 557 to 580, 558 to 580, 559 to 580, 560 to 580, 561 to 580, 562 to 580, 563 to 580, 564 to 580, 565 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. In some embodiments, the antigen-binding domain according to the present invention binds to a region within positions 554 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. In some embodiments, the antigen-binding domain according to the present invention binds to an epitope solely within positions 554 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. In some embodiments, the antigen-binding domain according to the present invention binds to a region within positions 564 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. In some embodiments, the antigen-binding domain according to the present invention binds to an epitope solely within positions 564 to 580 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. In some embodiments, the antigen-binding domain according to the present invention binds to a region within positions 495 to 510 of SEQ ID NO: 22, or a variant having at least 80% identity thereto. The variant sequence may have at least 80%, 85%, 90%, 95%, 96%, 97%, 98% or 99% sequence identity to SEQ ID NO: 22. The antigen-binding domain may bind to a linear epitope. The antigen-binding domain may bind to a conformational epitope. The antigen-binding domain may bind to post-translational modifications, such as heparin sulfate, N-linked glycans and/or the GPI anchor. SOLUBLE GPC3 GPC3 can be ‘shed’ from the cell at the furin cleavage site. In addition, GPC3 can be released from the cell surface to the extracellular environment after cleavage by lipase activities that cleave the GPI anchor and/or protease activities that cleave the membrane proximal region. Thus, the term “soluble GPC3” (or “sGPC3”) as used herein refers to both the free N-terminal subunit of GPC3 (i.e. amino acids 1-358) and other soluble truncated versions of GPC3, such 15 as lacking the GPI anchor, that is not associated with cell membrane and/or that is not capable of associating with cell membrane. In some embodiments, soluble GPC3 refers to amino acids 1 to 494 of SEQ ID NO: 22. In some embodiments, soluble GPC3 refers to amino acids 1 to 511, 512, 513, 514, 515, 516, 517, 518, 519, 520, 521, 522, 523, 524, 525, 526, 527, 528, 529, 530, 531, 532, 533, 534, 535, 536, 537, 538, 539, 540, 541, 542, 543, 544, 545, 546, 547, 548, 549, 550, 551, 552, 553, 554, 555, 556, 557, 558, 559, 560, 561, 562, or 563 of SEQ ID NO: 22. Suitably, the soluble GPC3 protein may not include the signal peptide (amino acids 1-24 of SEQ ID NO: 22), which may be cleaved during translation and processing to generate the mature form. In some embodiments, soluble GPC3 refers to amino acids 25-494 of SEQ ID NO: 22. In some embodiments, soluble GPC3 refers to amino acids 25-550 of SEQ ID NO: 22. The antigen-binding domain according to the present invention binds to a region between amino acids S495 to H580 of GPC3, wherein binding is not inhibited by a soluble GPC3. By “not inhibited” is meant that binding is not reduced by the presence of soluble GPC3 as compared to the absence of soluble GPC3. For example, such as in the presence of 0.01 to 100 μg/mL soluble GPC3. In some embodiments, the polypeptide comprising an antigen-binding domain of the invention is able to bind a region between amino acids S495 to H580 of GPC3 in the presence of about 0.01, about 0.1, about 0.5, about 1, about 5, about 10, or about 100 μg/mL soluble GPC3. Preferably, about 1 μg/mL soluble GPC3. Binding of a polypeptide comprising an antigen-binding domain according to the invention to GPC3 may be reduced by ≤ 15% in the presence of soluble GPC3, as compared to in the absence of soluble GPC3. For example, binding in the presence of soluble GPC3 may be reduced by about 0-15%, 0-10%, 0-5% or less, relative to conditions in which soluble GPC3 is not present. Thus, a binder that is inhibited by soluble GPC3 may demonstrate > 15% reduction in binding to GPC3 in the presence of soluble GPC3 as compared to the absence of soluble GPC3. For example, binding in the presence of soluble GPC3 may be reduced by about 15-25%, 15- 16 50%, 15-75% or more. Binding in the presence of soluble GPC3 may be reduced by about at least 15%, at least 25%, at least 50% or at least 75%. Methods for determining relative binding are known in the art, for example ELISA. A quantitative assessment or measurement of binding affinity (e.g. establishing a KD value) may be determined or measured using methods know in the art, such as by surface plasmon resonance, for example by using the Biacore® system. In addition to the equilibrium dissociation constant (KD), the association rate constant (Ka (1/Ms)), and the dissociation rate constant (Kd (1/s)) may also be determined. Surface Plasmon Resonance (SPR) experiments may be performed with a Biacore T200, for example. Methods for determining binding specificity include, but are not limited to, ELISA, western blot, immunohistochemistry, flow cytometry, Förster resonance energy transfer (FRET), phage display libraries, yeast two-hybrid screens, co-immunoprecipitation, bimolecular fluorescence complementation and tandem affinity purification. Binding affinity can also be determined using methods such as fluorescence quenching, isothermal titration calorimetry. The cytotoxicity of a cell, for example a cytolytic immune cell, which comprises a CAR according to the present invention may not be reduced in the presence of soluble GPC3 as compared to the absence of soluble GPC3. For example, a cytolytic immune cell expressing a CAR according to the invention may have cytotoxic activity that is reduced by ≤ 20% in the presence of soluble GPC3, as compared to in the absence of soluble GPC3. For example, cytotoxicity in the presence of soluble GPC3 may be reduced by about 0-20%, 0-15%, 0-10%, 0-5% or less, relative to conditions in which soluble GPC3 is not present. Thus, a cell expressing a CAR comprising a binder that is inhibited by soluble GPC3 may demonstrate > 20% reduction in cytotoxicity in the presence of soluble GPC3 as compared to the absence of GPC3. ANTIBODY The invention provides an antibody or antigen-binding fragment thereof comprising the antigen-binding domain according to the present invention. As used herein, “antibody” means a protein or polypeptide having an antigen binding site or antigen-binding domain which comprises at least one complementarity determining region CDR. The antibody may comprise 3 CDRs and have an antigen binding site which is 17 equivalent to that of a domain antibody (dAb). The antibody may comprise 6 CDRs and have an antigen binding site which is equivalent to that of a classical antibody molecule. The remainder of the polypeptide may be any sequence which provides a suitable scaffold for the antigen binding site and displays it in an appropriate manner for it to bind the antigen. The antibody may be a whole immunoglobulin molecule or a part thereof such as a Fab, F(ab)’₂, Fv, single chain Fv (ScFv) fragment, Nanobody or single chain variable domain (which may be a VH or VL chain, having 3 CDRs). The antibody may be a bifunctional antibody, for example a bispecific antibody. The antibody may be non-human, chimeric, humanised or fully human. Descriptions of an antibody of the present invention provided herein are generally applicable to an antigen binding fragment thereof. The antibody may be a monoclonal antibody or a polyclonal antibody. The antibody may be a synthetic antibody. Preferably, the antibody is a monoclonal antibody. Examples of an antigen-binding fragment include, but are not limited to, a single chain antibody (scFv), a single-domain antibody (sdAb), an antigen-binding fragment (Fab), a camelid antibody (VHH), a variable region (Fv), a heavy chain variable region (VH), a light chain variable region (VL), and a complementarity determining region (CDR). The antibody may be a full-length, classical antibody. For example, the antibody may be an IgG, IgM or IgA molecule. Suitably, the antibody is a full monoclonal antibody. Antibodies may be obtained by techniques comprising immunizing an animal with a target antigen and isolating the antibody from serum. Monoclonal antibodies may be made by the hybridoma method first described by Kohler et al., Nature 256:495 (1975), or may be made by recombinant DNA methods (see, e.g., U.S. Pat. No.4,816,567). The monoclonal antibodies may also be isolated from phage antibody libraries using the techniques described in Clackson et al., Nature 352:624-628 (1991) and Marks et al., J. Mol. Biol. 222:581-597 (1991), for example. The antibody may also be a chimeric or humanized antibody or fragment thereof. In the present invention, an antibody comprising an antigen-binding domain according to the invention may be obtained by immunising an animal with truncated GPC3, such as residues S359-H580 of GPC3. The invention provides a method for obtaining an antibody comprising an antigen-binding domain as defined in the invention. The method may comprise the steps of: 18 i. Immunising an animal with residues S359 to H580 of GPC3 (for example, amino acids 359 to 580 of SEQ ID NO: 22); ii. Isolating an antibody and/or nucleic acid sequence encoding an antibody from said animal. Suitably, the method may further comprises screening to identify an antibody for binding to residues S495 to H580 of GPC3 (for example, amino acids 495 to 580 of SEQ ID NO: 22). The present invention further provides an antibody generated by the method of the invention. The antibody of fragment according to the invention may prove useful in any method which relies on a high affinity binding interaction between an antigen-binding domain and a cognate target. Thus, the antibody of fragment according to the invention may be used as a detection antibody and/or a capture antibody. The antibody of fragment according to the invention may be used a therapeutic antibody, for example, as a therapeutic antibody that targets GPC3 protein or a cell expressing GPC3. A non-limiting example therefore for the application of the antibody of fragment according to the invention is the use in the treatment of cancers characterized by expression and/or overexpression of GPC3. The present invention also encompasses fragments of any antibody or protein or polypeptide as defined herein. It will be appreciated that a fragment comprises an amino acid sequence that is shorter than the full-length sequence of an antibody or protein or polypeptide, but retains full biological activity and/or antigenic nature of the full-length sequence of the antibody or protein or polypeptide. It will also be appreciated that said fragment retains the same binding affinity of the full-length sequence of the antibody or protein or polypeptide. IMMUNE CELL ENGAGERS In one embodiment, the present invention provides an immune cell activator molecule. For example, a T cell activator molecule or an NK cell activator molecule. In one embodiment, the present invention provides a T cell activator molecule which is a bispecific molecule (i.e. a bi-specific T cell engager (BiTE)) which comprises an antigen- binding domain as described herein as a first domain, and a T cell activating domain as a second domain. Bi-specific T cell engaging molecules are a class of bi-specific antibody-type molecules that have been developed, primarily for the use as anti-cancer drugs. They direct a host's immune system, more specifically the T cells' cytotoxic activity, against a target cell, such as a cancer 19 cell. In these molecules, one binding domain binds to binds to a T cell via, for example, the CD3 receptor, and the other to a target cells such as a tumour cell (via a tumour specific molecule). Since the bispecific molecule binds both the target cell and the T cell, it brings the target cell into proximity with the T cell, so that the T cell can exert its effect, for example, a cytotoxic effect on a cancer cell. The formation of the T cell:bispecific Ab:cancer cell complex induces signalling in the T cell leading to, for example, the release of cytotoxic mediators. Ideally, the agent only induces the desired signalling in the presence of the target cell, leading to selective killing. Thus, a bi-specific molecule of the present invention brings a GPC3-expressing cell into proximity with a T cell, so that the T cell can exert its effect on the cell. The requirement of co- localisation via binding of the GPC3 bi-specific molecule leads to selective killing of GPC3- positive cells. In other words, a bi-specific molecule of the present invention is able to activate T cells following binding of the antigen-binding domain to GPC3 expressed on the surface of target cells. BiTEs are commonly made by fusing an anti-CD3 scFv to an anti-target antigen scFv via a short five residue peptide linker. ANTIBODY CONJUGATE Suitably, an antibody conjugate is provided, which comprises the antibody or fragment thereof of the invention and a cargo or payload component. The antibody conjugate may be an antibody-drug conjugate (ADC), which is a class of targeted therapeutics that improves both the selectivity and the cytotoxic activity of cancer drugs. Typically, ADCs have three components: (i) a monoclonal antibody conjugated to (ii) a linker, which is also conjugated to (iii) a drug or payload, such as a cytotoxic or chemotherapeutic drug. The cytotoxic or chemotherapeutic drug refers to a drug that is destructive to a cell and reduces the viability of the cell. Suitable cytotoxic or chemotherapeutic drugs will be known in the art. Suitably, the antibody conjugate of the invention is a molecule composed of an antibody or fragment thereof described herein, linked (i.e. conjugated) to a biologically active cytotoxic payload or drug, such as an anticancer drug. The linker may be any appropriate linker known in the art. The person skilled in the art will know that such linkers are routinely used in the production of conjugate molecules and would be able to select an appropriate linker. Such linkers typically have chemically reactive groups at each end. These linkers can form a covalent attachment between two molecules, e.g. the antibody or fragment thereof and the drug or payload. Thus, the antibody or fragment thereof and the drug or payload may both be 20 covalently linked to a linker. Suitably, one region of the linker may bind to the antibody or fragment thereof and another region of the linker may bind to the drug or payload. The linker may form, for example, hydrazone, disulfide or amide bonds between the antibody or fragment thereof and/or the drug or payload. CHIMERIC ANTIGEN RECEPTOR The present invention provides a chimeric antigen receptor (CAR) comprising an antigen- binding domain as defined herein. A classical chimeric antigen receptor (CAR) is a chimeric type I trans-membrane protein which connects an extracellular antigen-recognizing domain (binder) to an intracellular signalling domain (endodomain). The binder is typically a single-chain variable fragment (scFv) derived from a monoclonal antibody (mAb), but it can be based on other formats which comprise an antibody-like antigen binding site. A spacer domain is usually necessary to isolate the binder from the membrane and to allow it a suitable orientation. A common spacer domain used is the Fc of IgG1. More compact spacers can suffice e.g. the stalk from CD8α and even just the IgG1 hinge alone, depending on the antigen. A trans-membrane domain anchors the protein in the cell membrane and connects the spacer to the endodomain. Early CAR designs had endodomains derived from the intracellular parts of either the γ chain of the FcεR1 or CD3ζ. Consequently, these first generation receptors transmitted immunological signal 1, which was sufficient to trigger T-cell killing of cognate target cells but failed to fully activate the T-cell to proliferate and survive. To overcome this limitation, compound endodomains have been constructed: fusion of the intracellular part of a T-cell co- stimulatory molecule to that of CD3ζ results in second generation receptors which can transmit an activating and co-stimulatory signal simultaneously after antigen recognition. The co- stimulatory domain most commonly used is that of CD28. This supplies the most potent co- stimulatory signal - namely immunological signal 2, which triggers T-cell proliferation. Some receptors have also been described which include TNF receptor family endodomains, such as the closely related OX40 and 41BB which transmit survival signals. Even more potent third generation CARs have now been described which have endodomains capable of transmitting activation, proliferation and survival signals. When the CAR binds the target-antigen, this results in the transmission of an activating signal to the T-cell it is expressed on. Thus the CAR directs the specificity and cytotoxicity of the T cell towards tumour cells expressing the targeted antigen. CARs typically therefore comprise: (i) an antigen-binding domain; (ii) a spacer; (iii) a 21 transmembrane domain; and (iii) an intracellular domain which comprises or associates with a signalling domain. A CAR may have the general structure: Antigen binding domain – spacer domain – transmembrane domain – intracellular signalling domain (endodomain). Signal Peptide The CAR of the present invention may comprise a signal peptide so that when the CAR is expressed inside a cell, such as a T cell, the nascent protein is directed to the endoplasmic reticulum and subsequently to the cell surface, where it is expressed. The core of the signal peptide may contain a long stretch of hydrophobic amino acids that has a tendency to form a single alpha-helix. The signal peptide may begin with a short positively charged stretch of amino acids, which helps to enforce proper topology of the polypeptide during translocation. At the end of the signal peptide there is typically a stretch of amino acids that is recognized and cleaved by signal peptidase. Signal peptidase may cleave either during or after completion of translocation to generate a free signal peptide and a mature protein. The free signal peptides are then digested by specific proteases. The signal peptide may be at the amino terminus of the molecule. The CAR of the invention may have the general formula: Signal peptide – antigen-binding domain – spacer domain – transmembrane domain – intracellular signalling domain. The signal peptide may comprise the SEQ ID NO: 24 or a variant thereof having 5, 4, 3, 2 or 1 amino acid mutations (insertions, substitutions or additions) provided that the signal peptide still functions to cause cell surface expression of the CAR. SEQ ID NO: 24: METDTLLLWVLLLWVPGSTG The signal peptide of SEQ ID NO: 24 is compact and highly efficient. It is predicted to give about 95% cleavage after the terminal glycine, giving efficient removal by signal peptidase. 22 Antigen-binding domain The antigen binding domain is the portion of the chimeric receptor which recognizes antigen. In a classical CAR, the antigen-binding domain comprises: a single-chain variable fragment (scFv) derived from a monoclonal antibody. CARs have also been produced with domain antibody (dAb) or VHH antigen binding domains. A CAR may comprise a Fab fragment of an antibody. A FabCAR comprises two chains: one having an antibody-like light chain variable region (VL) and constant region (CL); and one having a heavy chain variable region (VH) and constant region (CH). One chain also comprises a transmembrane domain and an intracellular signalling domain. Association between the CL and CH causes assembly of the receptor. The two chains of a Fab CAR may have the general structure: VH - CH - spacer - transmembrane domain - intracellular signalling domain; and VL - CL or VL - CL - spacer- transmembrane domain - intracellular signalling domain; and VH – CH. For the Fab-type chimeric receptors described herein, the antigen binding domain is made up of a VH from one polypeptide chain and a VL from another polypeptide chain. The polypeptide chains may comprise a linker between the VH/VL domain and the CH/CL domains. The linker may be flexible and serve to spatially separate the VH/VL domain from the CH/CL domain. Alternatively, a CAR may comprise a single-chain variable fragment (scFv) fragment of an antibody. In these "classical" CARs, the VH and VL domains may be in either orientation in the molecule, giving the general structure: VH - VL - spacer - transmembrane domain - intracellular signalling domain, or VL - VH - spacer - transmembrane domain - intracellular signalling domain. The polypeptide chain may comprise a linker between the VH domain and the VL domains to provide sufficient flexibility to enable VH/VL pairing and formation of the antigen-binding site. 23 Alternatively, a CAR may comprise a domain-antibody (dAb)-type antigen-binding domain. Such a CAR may have the general structure: dAb - spacer - transmembrane domain - intracellular signalling domain. Preferably, the antigen-binding domain of the CAR of the present invention comprises an scFv having the sequence shown as SEQ ID NO: 17 or SEQ ID NO: 18. Spacer CARs generally comprise a spacer sequence or hinge region to connect the antigen-binding domain with the transmembrane domain and spatially separate the antigen-binding domain from the endodomain. A flexible spacer allows the antigen-binding domain to orient in different directions to facilitate binding. The spacer sequence may, for example, comprise an IgG1 Fc region, an IgG1 hinge, a CD28 hinge, or a CD8 stalk, or a combination thereof. The spacer may alternatively comprise an alternative sequence which has similar length and/or domain spacing properties as an IgG1 Fc region, an IgG1 hinge or a CD8 stalk. A human IgG1 spacer may be altered to remove Fc binding motifs. Preferably, the spacer comprises a CD8 stalk. For example, the spacer preferably comprises the sequence SEQ ID NO: 25. Preferably, the spacer comprises a CD28 hinge. For example, the spacer preferably comprises the sequence SEQ ID NO: 38. Examples of amino acid sequences for these spacers are given below: SEQ ID NO: 26 (hinge-CH2CH3 of human IgG1) AEPKSPDKTHTCPPCPAPPVAGPSVFLFPPKPKDTLMIARTPEVTCVVVDVSHEDPEVKFNWYVDGVE VHNAKTKPREEQYNSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYTL PPSRDELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQQ GNVFSCSVMHEALHNHYTQKSLSLSPGKKD SEQ ID NO: 25 (human CD8 stalk): TTTPAPRPPTPAPTIASQPLSLRPEACRPAAGGAVHTRGLDFACDI SEQ ID NO: 27 (human IgG1 hinge): AEPKSPDKTHTCPPCPKDPK 24 SEQ ID NO: 28 (IgG1 Hinge-Fc) AEPKSPDKTHTCPPCPAPELLGGPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGV EVHNAKTKPREEQYNSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYT LPPSRDELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQ QGNVFSCSVMHEALHNHYTQKSLSLSPGKKDPK SEQ ID NO: 29 (IgG1 Hinge – Fc modified to remove Fc receptor recognition motifs) AEPKSPDKTHTCPPCPAPPVA*GPSVFLFPPKPKDTLMIARTPEVTCVVVDVSHEDPEVKFNWYVDGV EVHNAKTKPREEQYNSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYT LPPSRDELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSRWQ QGNVFSCSVMHEALHNHYTQKSLSLSPGKKDPK Modified residues are underlined; * denotes a deletion. SEQ ID NO: 38 (CD28 hinge) IEVMYPPPYLDNEKSNGTIIHVKGKHLCPSPLFPGPSKP A variant sequence may have at least 80%, 85%, 90%, 95%, 96%, 97%, 98% or 99% sequence identity to SEQ ID NOs: 25 to 29 or 38. Transmembrane domain The CAR of the invention may also comprise a transmembrane domain which spans the membrane. The transmembrane domain may be any protein structure which is thermodynamically stable in a membrane. It may comprise a hydrophobic alpha helix. The transmembrane domain may be derived from CD8, CD28 or human IgG. The transmembrane domain may be derived from any type I transmembrane protein. The transmembrane domain may be a synthetic sequence predicted to form a hydrophobic helix. As used herein, the term “derived from” refers to the origin or source, and may include naturally occurring, recombinant, unpurified, or purified molecules. The term “derived from” encompasses the terms “originated from,” “obtained from,” “obtainable from,” “isolated from,” and "created from." Preferably, the transmembrane domain comprises a CD8 transmembrane domain. For example, the spacer preferably comprises the sequence of SEQ ID NO: 30. Preferably, the transmembrane domain comprises a CD28 transmembrane domain. For example, the spacer preferably comprises the sequence of SEQ ID NO: 31. Examples of amino acid sequences for transmembrane domains are given below: 25 SEQ ID NO: 30 (CD8a transmembrane domain) IYIWAPLAGTCGVLLLSLVIT SEQ ID NO: 31 (CD28 transmembrane domain) FWVLVVVGGVLACYSLLVTVAFIIFWV The CAR of the invention may comprise a variant of the sequence shown as SEQ ID NOs: 30 or 31, having at least 80%, 85%, 90%, 95%, 96%, 97%, 98% or 99% sequence identity, provided that the variant sequence retains the capacity to insert into and span the membrane. Intracellular T Cell Signalling Domain (Endodomain) The endodomain is the signal-transmission portion of the chimeric receptor. It may be part of or associate with the intracellular domain of the chimeric receptor. After antigen recognition, receptors cluster, native CD45 and CD148 are excluded from the synapse and a signal is transmitted to the cell. The most commonly used endodomain component is that of CD3-zeta which contains 3 ITAMs. This transmits an activation signal to the T cell after antigen is bound. CD3-zeta may not provide a fully competent activation signal and additional co-stimulatory signalling may be needed. Co-stimulatory signals promote T-cell proliferation and survival. There are two main types of co-stimulatory signals: those that belong the Ig family (CD28, ICOS) and the TNF family (OX40, 41BB, CD27, GITR etc). For example, chimeric CD28 and OX40 can be used with CD3-Zeta to transmit a proliferative / survival signal, or all three can be used together. The endodomain may comprise: i. an ITAM-containing endodomain, such as the endodomain from CD3 zeta; and/or ii. a co-stimulatory domain, such as the endodomain from CD28 or ICOS; and/or iii. a domain which transmits a survival signal, for example a TNF receptor family endodomain such as OX40, 41-BB, CD27 or GITR. Preferably, the intracellular T cell signalling domain comprises a CD28 endodomain and a CD3-Zeta endodomain. For example, the intracellular T cell signalling domain preferably comprises the sequences of SEQ ID NO: 32 and SEQ ID NO: 33. Examples of amino acid sequences for intracellular T cell signalling domains are given below: SEQ ID NO: 32 (CD28 endodomain) RSKRSRLLHSDYMNMTPRRPGPTRKHYQPYAPPRDFAAYRS 26 SEQ ID NO: 34 (OX40 endodomain) RRDQRLPPDAHKPPGGGSFRTPIQEEQADAHSTLAKI SEQ ID NO: 35 (41-BB endodomain) KRGRKKLLYIFKQPFMRPVQTTQEEDGCSCRFPEEEEGGCEL SEQ ID NO: 33 (CD3 zeta endodomain) RVKFSRSADAPAYQQGQNQLYNELNLGRREEYDVLDKRRGRDPEMGGKPRRKNPQEGLYNELQKDKMA EAYSEIGMKGERRRGKGHDGLYQGLSTATKDTYDALHMQALPPR SEQ ID NO: 36 (CD28Z) RSKRSRLLHSDYMNMTPRRPGPTRKHYQPYAPPRDFAAYRSRVKFSRSADAPAYQQGQNQLYNELNLG RREEYDVLDKRRGRDPEMGGKPRRKNPQEGLYNELQKDKMAEAYSEIGMKGERRRGKGHDGLYQGLST ATKDTYDALHMQALPPR SEQ ID NO: 37 (CD28OXZ) RSKRSRLLHSDYMNMTPRRPGPTRKHYQPYAPPRDFAAYRSRDQRLPPDAHKPPGGGSFRTPIQEEQA DAHSTLAKIRVKFSRSADAPAYQQGQNQLYNELNLGRREEYDVLDKRRGRDPEMGGKPRRKNPQEGLY NELQKDKMAEAYSEIGMKGERRRGKGHDGLYQGLSTATKDTYDALHMQALPPR A variant sequence may have at least 80%, 85%, 90%, 95%, 96%, 97%, 98% or 99% sequence identity to SEQ ID NOs: 32 to 37, provided that the sequence provides an effective intracellular T cell signalling domain. CAR The CAR of the present invention may comprise the sequence shown as SEQ ID NO: 19 or a variant thereof which has at least 80%, 85%, 90%, 95%, 96%, 97%, 98% or 99% sequence identity thereto and retains the capacity to i) bind GPC3 and ii) induce T cell signalling. The CAR of the present invention may comprise the sequence shown as SEQ ID NO: 20 or a variant thereof which has at least 80%, 85%, 90%, 95%, 96%, 97%, 98% or 99% sequence identity thereto and retains the capacity to i) bind GPC3 and ii) induce T cell signalling. Preferably the CAR comprises the sequence shown as SEQ ID NO: 19 or SEQ ID NO: 20. HR8 CAR (SEQ ID NO: 19) QVQLQQSGAELVKPGTSVKLSRKASGYTFTANHMNWIKQTTGQGLECIGIINPGSGGTSYNVKFRGRA TLTVDKSSSTAFMQLSGLTPEDSAVYYCARSGYGGPHYWGQGVMVTVSSGGGGSGGGGSGGGGSDIVM TQGALPNPVPSGESASITCQSSKSLLHSNGKTYLNWYLQRPGQSPQLLIYWMSTRASGVSDRFSGSGS GTDFTLKISSVEAEDVGVYYCQQFLEYPLTFGSGTKLEIKSDPAIEVMYPPPYLDNEKSNGTIIHVKG 27 KHLCPSPLFPGPSKPFWVLVVVGGVLACYSLLVTVAFIIFWVRSKRSRLLHSDYMNMTPRRPGPTRKH YQPYAPPRDFAAYRSRVKFSRSADAPAYQQGQNQLYNELNLGRREEYDVLDKRRGRDPEMGGKPRRKN PQEGLYNELQKDKMAEAYSEIGMKGERRRGKGHDGLYQGLSTATKDTYDALHMQALPPR* HR10 CAR (SEQ ID NO: 20) GVQLQKSGAELVRPGTSVKLSCKVSGDTITFYYMHFVKQRPGQGLEWIGRIDPEDESTKYSEKFKNKA TLTADTSSNTAYLKLSSLTSEDTATYFCIYGGYYFDYWGQGVMVTVSSGGGGSGGGGSGGGGSDIVLT QSPTTLSVTPGETVSLSCRASHDIGTNLHWYQQKTNESPRLLIKSASQPTSGIPSRFSASGSGTDFTL NINNVEFDDVSSYLCQQTQSWPLTFGSGTKLELKSDPAIEVMYPPPYLDNEKSNGTIIHVKGKHLCPS PLFPGPSKPFWVLVVVGGVLACYSLLVTVAFIIFWVRSKRSRLLHSDYMNMTPRRPGPTRKHYQPYAP PRDFAAYRSRVKFSRSADAPAYQQGQNQLYNELNLGRREEYDVLDKRRGRDPEMGGKPRRKNPQEGLY NELQKDKMAEAYSEIGMKGERRRGKGHDGLYQGLSTATKDTYDALHMQALPPR* Suicide Genes The CAR-expressing cell of the present invention may comprise a suicide gene. Since T cells engraft and are autonomous, a means of selectively deleting CAR T cells in recipients of anti-GPC3 CAR T cells is desirable. Suicide genes are genetically encodable mechanisms which result in selective destruction of infused T cells in the face of unacceptable toxicity. The earliest clinical experience with suicide genes is with the Herpes Virus Thymidine Kinase (HSV-TK) which renders T cells susceptible to Ganciclovir. HSV-TK is a highly effective suicide gene. However, pre-formed immune responses may restrict its use to clinical settings of considerable immunosuppression such as haploidentical stem cell transplantation. Inducible Caspase 9 (iCasp9) is a suicide gene constructed by replacing the activating domain of Caspase 9 with a modified FKBP12. iCasp9 is activated by an otherwise inert small molecular chemical inducer of dimerization (CID). iCasp9 has been recently tested in the setting of haploidentical HSCT and can abort GvHD. The biggest limitation of iCasp9 is dependence on availability of clinical grade proprietary CID. Both iCasp9 and HSV-TK are intracellular proteins, so when used as the sole transgene, they have been co-expressed with a marker gene to allow selection of transduced cells. An iCasp9 may comprise the sequence shown as SEQ ID NO: 39 or a variant thereof having at least 80%, 85%, 90%, 95%, 98% or 99 % sequence identity. SEQ ID NO: 39 MLEGVQVETISPGDGRTFPKRGQTCVVHYTGMLEDGKKVDSSRDRNKPFKFMLGKQEVIRGWEEGVAQ MSVGQRAKLTISPDYAYGATGHPGIIPPHATLVFDVELLKLESGGGSGVDGFGDVGALESLRGNADLA YILSMEPCGHCLIINNVNFCRESGLRTRTGSNIDCEKLRRRFSSLHFMVEVKGDLTAKKMVLALLELA QQDHGALDCCVVVILSHGCQASHLQFPGAVYGTDGCPVSVEKIVNIFNGTSCPSLGGKPKLFFIQACG GEQKDHGFEVASTSPEDESPGSNPEPDATPFQEGLRTFDQLDAISSLPTPSDIFVSYSTFPGFVSWRD PKSGSWYVETLDDIFEQWAHSEDLQSLLLRVANAVSVKGIYKQMPGCFNFLRKKLFFKTSAS 28 The marker/suicide gene may be RQR8, which can be detected with the antibody QBEnd10 and expressing cells lysed with the therapeutic antibody Rituximab. An RQR8 may comprise the sequence shown as SEQ ID NO: 40 or a variant thereof having at least 80%, 85%, 90%, 95%, 98% or 99% sequence identity. SEQ ID NO: 40 MGTSLLCWMALCLLGADHADACPYSNPSLCSGGGGSELPTQGTFSNVSTNVSPAKPTTTACPYSNPSL CSGGGGSPAPRPPTPAPTIASQPLSLRPEACRPAAGGAVHTRGLDFACDIYIWAPLAGTCGVLLLSLV ITLYCNHRNRRRVCKCPRPVV The suicide gene may be expressed as a single polypeptide with the CAR, for example by using a self-cleaving peptide between the two sequences. A ‘self-cleaving peptide’ refers to a peptide which functions such that when the polypeptide comprising the proteins and the self-cleaving peptide is produced, it is immediately “cleaved” or separated into distinct and discrete first and second polypeptides without the need for any external cleavage activity. The self-cleaving peptide may be a 2A self-cleaving peptide from an aphtho- or a cardiovirus. The primary 2A/2B cleavage of the aptho- and cardioviruses is mediated by 2A “cleaving” at its own C-terminus. In apthoviruses, such as foot-and-mouth disease viruses (FMDV) and equine rhinitis A virus, the 2A region is a short section of about 18 amino acids, which, together with the N-terminal residue of protein 2B (a conserved proline residue) represents an autonomous element capable of mediating “cleavage” at its own C-terminus (Donelly et al (2001) as above). The cleavage site may comprise the 2A-like sequence shown as SEQ ID NO: 41 SEQ ID NO: 41 RAEGRGSLLTCGDVEENPGP POLYNUCLEOTIDE In an aspect the present invention provides a nucleic acid sequence which encodes an antigen-binding domain of the present invention. In one aspect the present invention provides a nucleic acid sequence which encodes an antibody or fragment thereof of the present invention. 29 In one aspect the present invention provides a nucleic acid sequence which encodes a bispecific molecule, such as a bispecific T cell engager, of the present invention In one aspect the present invention provides a nucleic acid sequence which encodes a CAR of the present invention. Due to the redundancy of the genetic code, variations in nucleic acid sequences are possible that encode for the same polypeptide. These sequences are encompassed by the present invention. Therefore, multiple polynucleotides are envisaged, each with a different nucleic acid sequence but which encodes a polypeptide according to the invention or a further polypeptide as described herein. It is possible to design and produce such nucleic acid sequences without difficulty. The nucleic acid sequence may be an RNA or DNA sequence or a variant thereof. The term “polynucleotide” includes an RNA or DNA sequence. It may be single or double stranded. It may, for example, be genomic, recombinant, mRNA or cDNA. The nucleotide sequence may be codon optimised for production in the host cell of choice. VECTOR The present invention also provides a vector, or kit of vectors, which comprises one or more nucleic acid sequence(s) encoding an antigen-binding domain, antibody or antigen-binding fragment thereof, bispecific molecule or chimeric antigen receptor according to the invention. Such a vector may be used to introduce the nucleic acid sequence(s) into a host cell so that it expresses and produces a molecule of the invention. The vector may, for example, be a plasmid or a viral vector, such as a retroviral vector or a lentiviral vector, or a transposon based vector or synthetic mRNA. The vector may be capable of transfecting or transducing a cell such as a T cell or a NK cell. CELL The present invention provides a cell which comprises a chimeric antigen receptor of the invention. The cell may comprise a nucleic acid sequence, nucleic acid construct or a vector of the present invention. The cell may be a cytolytic immune cell such as a T cell or an NK cell. 30 T cells or T lymphocytes are a type of lymphocyte that play a central role in cell-mediated immunity. They can be distinguished from other lymphocytes, such as B cells and natural killer cells (NK cells), by the presence of a T-cell receptor (TCR) on the cell surface. There are various types of T cell, as summarised below. Helper T helper cells (TH cells) assist other white blood cells in immunologic processes, including maturation of B cells into plasma cells and memory B cells, and activation of cytotoxic T cells and macrophages. TH cells express CD4 on their surface. TH cells become activated when they are presented with peptide antigens by MHC class II molecules on the surface of antigen presenting cells (APCs). These cells can differentiate into one of several subtypes, including TH1, TH2, TH3, TH17, Th9, or TFH, which secrete different cytokines to facilitate different types of immune responses. Cytolytic T cells (TC cells, or CTLs) destroy virally infected cells and tumour cells, and are also implicated in transplant rejection. CTLs express the CD8 at their surface. These cells recognize their targets by binding to antigen associated with MHC class I, which is present on the surface of all nucleated cells. Through IL-10, adenosine and other molecules secreted by regulatory T cells, the CD8+ cells can be inactivated to an anergic state, which prevent autoimmune diseases such as experimental autoimmune encephalomyelitis. Memory T cells are a subset of antigen-specific T cells that persist long-term after an infection has resolved. They quickly expand to large numbers of effector T cells upon re-exposure to their cognate antigen, thus providing the immune system with "memory" against past infections. Memory T cells comprise three subtypes: central memory T cells (TCM cells) and two types of effector memory T cells (TEM cells and TEMRA cells). Memory cells may be either CD4+ or CD8+. Memory T cells typically express the cell surface protein CD45RO. Regulatory T cells (Treg cells), formerly known as suppressor T cells, are crucial for the maintenance of immunological tolerance. Their major role is to shut down T cell-mediated immunity toward the end of an immune reaction and to suppress auto-reactive T cells that escaped the process of negative selection in the thymus. Two major classes of CD4+ Treg cells have been described – naturally occurring Treg cells and adaptive Treg cells. Naturally occurring Treg cells (also known as CD4+CD25+FoxP3+ Treg cells) arise in the thymus and have been linked to interactions between developing T cells with both myeloid (CD11c+) and plasmacytoid (CD123+) dendritic cells that have been activated with TSLP. Naturally occurring Treg cells can be distinguished from other T cells by the presence of an 31 intracellular molecule called FoxP3. Mutations of the FOXP3 gene can prevent regulatory T cell development, causing the fatal autoimmune disease IPEX. Adaptive Treg cells (also known as Tr1 cells or Th3 cells) may originate during a normal immune response. The cell may be a Natural Killer cell (or NK cell). NK cells form part of the innate immune system. NK cells provide rapid responses to innate signals from virally infected cells in an MHC independent manner. NK cells (belonging to the group of innate lymphoid cells) are defined as large granular lymphocytes (LGL) and constitute the third kind of cells differentiated from the common lymphoid progenitor generating B and T lymphocytes. NK cells are known to differentiate and mature in the bone marrow, lymph node, spleen, tonsils and thymus where they then enter into the circulation. The cells of the invention may be any of the cell types mentioned above. Cells according to the first aspect of the invention may either be created ex vivo either from a patient’s own peripheral blood (1st party), or in the setting of a haematopoietic stem cell transplant from donor peripheral blood (2nd party), or peripheral blood from an unconnected donor (3rd party). Alternatively, cells according to the first aspect of the invention may be derived from ex vivo differentiation of inducible progenitor cells or embryonic progenitor cells to T or NK cells. Alternatively, an immortalized T-cell line which retains its lytic function and could act as a therapeutic may be used. In all these embodiments, chimeric polypeptide-expressing cells are generated by introducing DNA or RNA coding for the chimeric polypeptide by one of many means including transduction with a viral vector, transfection with DNA or RNA. The cell of the invention may be an ex vivo cell from a subject. The cell may be from a peripheral blood mononuclear cell (PBMC) sample. Cells may be activated and/or expanded prior to being transduced with nucleic acid encoding the molecules providing the chimeric polypeptide according to the first aspect of the invention, for example by treatment with an anti-CD3 monoclonal antibody. The cell of the invention may be made by: 32 i. isolation of a cell-containing sample from a subject or other sources listed above; and ii. transduction or transfection of the cells with one or more a nucleic acid sequence(s) encoding a chimeric polypeptide or vector according to the invention. The cells may then by purified, for example, selected on the basis of expression of the antigen- binding domain of the antigen-binding polypeptide. PHARMACEUTICAL COMPOSITION The present invention also relates to a pharmaceutical composition comprising a polypeptide comprising an antigen-binding domain of the invention together with a pharmaceutically acceptable carrier, diluent or excipient, and optionally one or more further pharmaceutically active polypeptides and/or compounds. The present invention also relates to a pharmaceutical composition comprising an antibody or antigen-binding fragment thereof of the invention together with a pharmaceutically acceptable carrier, diluent or excipient, and optionally one or more further pharmaceutically active polypeptides and/or compounds. The present invention also relates to a pharmaceutical composition comprising an antibody conjugate of the invention together with a pharmaceutically acceptable carrier, diluent or excipient, and optionally one or more further pharmaceutically active polypeptides and/or compounds. The present invention also relates to a pharmaceutical composition comprising a polynucleotide or a vector of the invention together with a pharmaceutically acceptable carrier, diluent or excipient, and optionally one or more further pharmaceutically active polypeptides and/or compounds. The present invention also relates to a pharmaceutical composition comprising a CAR- expressing cell of the invention together with a pharmaceutically acceptable carrier, diluent or excipient, and optionally one or more further pharmaceutically active polypeptides and/or compounds. Such formulations may, for example, be in a form suitable for intravenous infusion. METHOD OF TREATMENT The present invention provides a method for treating a disease which comprises the step of administering a polypeptide comprising an antigen-binding domain according to the invention, 33 an antibody or antigen-binding fragment thereof according to the invention, an antibody conjugate according to the invention, a polynucleotide according to the invention, a vector according to the invention, a cell according to the invention, or a pharmaceutical composition according to the invention to a subject. A method for treating a disease relates to the therapeutic use of the polypeptide comprising an antigen-binding domain, antibody or antigen-binding fragment thereof, antibody conjugate, polynucleotide, vector, cell, or pharmaceutical composition of the present invention. Herein the polypeptide comprising an antigen-binding domain, antibody or antigen-binding fragment thereof, antibody conjugate, polynucleotide, vector, cell, or pharmaceutical composition may be administered to a subject having an existing disease or condition in order to lessen, reduce or improve at least one symptom associated with the disease and/or to slow down, reduce or block the progression of the disease. The method for preventing a disease relates to the prophylactic use of the polypeptide comprising an antigen-binding domain, antibody or antigen-binding fragment thereof, antibody conjugate, polynucleotide, vector, cell, or pharmaceutical composition of the present invention. Herein such polypeptide comprising an antigen-binding domain, antibody or antigen-binding fragment thereof, antibody conjugate, polynucleotide, vector, cell, or pharmaceutical composition may be administered to a subject who has not yet contracted the disease and/or who is not showing any symptoms of the disease to prevent or impair the cause of the disease or to reduce or prevent development of at least one symptom associated with the disease. The subject may have a predisposition for, or be thought to be at risk of developing, the disease. The method may involve the steps of: i. isolating a cell-containing sample; ii. transducing or transfecting such cells with a nucleic acid sequence or vector provided by the present invention; iii. administering the cells from (ii) to a subject. The cell-containing sample may be isolated from a subject or from other sources, for example as described above. The cells may be isolated from a subject’s own peripheral blood (1st party), or in the setting of a haematopoietic stem cell transplant from donor peripheral blood (2nd party), or peripheral blood from an unconnected donor (3rd party). The invention also provides a polypeptide comprising an antigen-binding domain according to the invention, an antibody or antigen-binding fragment thereof according to the invention, an 34 antibody conjugate according to the invention, a polynucleotide according to the invention, a vector according to the invention, a cell according to the invention, or a pharmaceutical composition according to the invention, for use in a method of treating a disease in a subject. The invention provides for use of a polypeptide comprising an antigen-binding domain according to the invention, an antibody or antigen-binding fragment thereof according to the invention, an antibody conjugate according to the invention, a polynucleotide according to the invention, a vector according to the invention, a cell according to the invention, or a pharmaceutical composition according to the invention in the manufacture of a medicament for treating a disease. Preferably, the subject is human. The disease to be treated and/or prevented by the methods of the present invention may be a cancer characterized by expression and/or overexpression of GPC3. The disease to be treated and/or prevented by the methods of the present invention may be a cancer exhibiting expression and/or overexpression of GPC3. The disease to be treated and/or prevented by the methods of the present invention may be a cancerous disease, such as liver cancer. The cancer may be a primary or metastasised tumour. In some embodiments, the cancer is a primary tumour expressing GPC3. In some embodiments, the cancer is a metastasised tumour expressing GPC3. For example, the cancer may be hepatocellular carcinoma (HCC), melanoma, ovarian clear- cell carcinomas, yolk sac tumour, neuroblastoma, hepatoblastoma, Wilms' tumor cells, rhabdoid tumors, and rhabdomyosarcomas. In preferred embodiments, the disease is HCC. The invention will now be further described by way of Examples, which are meant to serve to assist one of ordinary skill in the art in carrying out the invention and are not intended in any way to limit the scope of the invention. EXAMPLES MATERIALS AND METHODS Serum GPC3 detection Patient serum was obtained and approved by the Ethical Committee of the participating hospitals. Patient serum diluted 1: 2 with coating buffer were used to coat ELISA wells at 4°C 35 overnight. Standards made from recombinant protein (SinoBiological) were also incubated overnight. After blocking, and washing, 1G12 clone anti GPC3-muIgG (1 μg/mL) was added and then detected by anti-murine IgG-HRP antibody (ThermoFisher Scientific). TMB and H₂SO4 were added to detect the OD450 nm value. Library construction and bio-panning Three rats were immunized using genetic vaccinations with DNA encoding for truncated GPC3 (S359-H580), the DNA plasmid of interest was coated on gold nanoparticles and administered intramuscularly using a GeneGun™ (BioRad). Vaccinations were carried out at Aldevron, GmBH, Freiburg. Following vaccination, once sero-conversion was confirmed, total RNA was extracted from the spleen of animals using an RNAeasy™ (QIAGEN) extraction kit. RNA obtained from multiple animals was reverse transcribed to cDNA, which was used to generate phage display libraries cloned into the phagemid vector pHEN1. Libraries were generated as described previously (Nannini et al., Sci Rep, 2020; 10(1):19168), with a size of 9x10⁷. Bio- panning was carried out on Immunotubes using recombinant protein purchased from SinoBiological. After two rounds of bio-panning specific binder enrichment from bulk cultures was observed. Antigen binding tests from individual bacterial colonies were performed using both flow cytometry and ELISA based methods. Colonies that showed both flow and ELISA binding as described below were expanded and DNA extracted was sequenced by sanger sequencing. Reads were confirmed to contain between 900 and 1200 bp fragments that include phage PelB leader sequence and myc tag to obtain full- length sequences. These were then processed and analysed using IMGT/HighV- QUEST. VH and VL regions were annotated, and CDR regions determined based on IMGT numbering. Duplications were removed so that only one of the sequences was carried forward for further testing. Binding Assessment Assays For binding by ELISA, GPC3 protein (SinoBiological) was coated at 1 μg/ml on phosphate- buffered saline (PBS) on Nunc MaxiSorp plates. After 2% bovine serum albumin block step supernatant from IPTG induced TG1 bacterial cultures were used as primary antibodies.2% milk coated plate was used as a control. Plates were washed 4 times in PBS 0.05% Tween20, to remove unbound antibodies, prior to incubation with secondary anti-cMyc-HRP antibody (Clone 9E10, Invitrogen). Positive interaction was revealed using 1-Step Ultra TMB substrate (ThermoFisher Scientific, p/n 34028). Reaction was stopped with a solution of 1 M sulfuric acid and absorbance read at 450 nm using Multiskan plate reader (ThermoFisher Scientific). 36 For flow cytometry SupT1 cells were engineered to overexpress the human GPC3. Cloned _{scFv antibodies were used as the primary antibody for staining 1 x 10} ⁵ _{antigen positive supT1} cells followed by 1μl of anti-mouse secondary antibody (Jackson ImmunoResearch, p/n 115- 116-146). Cells were acquired on a BD Fortessa X20 flow cytometer. Colonies that were both positive for flow and ELISA were cloned into constructs for further downstream testing. Generation of scFv constructs ScFv antibody sequences were generated through DNA synthesis. DNA sequences, as retrieved from the phage display library, were synthesized as gBlocks (Integrated DNA Technologies) with flanking restriction sites AgeI and BamHI. The scFv fragments were cloned into a SFG backbone vector containing an in-frame mIgG2a Fc region. Surface plasmon resonance _{SPR experiments were performed with a Biacore T200 instrument using HBS-P} ⁺ _{as the} running and dilution buffer (GE Healthcare, p/n BR100671). Biacore Insight Evaluation soft- ware v3.0 (GE Healthcare) was used for data processing. For determination of binding kinetics, goat anti-mouse IgG (GE Healthcare, p/n BR100838) was covalently coupled to a CM5 Sensor Chip (GE Healthcare, p/n 29149603) according to manufacturer recommendations (approximately 9000 RU). ScFvs with a murine IgG2a Fc were captured (level range 200–400 RU), and concentrations of interaction partner protein from 200nM with 2-fold serial dilutions, were injected over the flow cell at a flow rate of 30 μl/min. A double reference subtraction was performed using buffer alone. Kinetic rate constants were obtained by curve fitting according to a 1:1 Langmuir binding model. KD for clones control scFv, HR8 and HR10 were calculated using an artificially _{limited kd of 1 x 10} ^-6 _{(1/s) due to instrument limitations.} Differential scanning fluorimetry Protein stability was analysed using Protein Thermal Shift™ Dye Kit (ThermoFisher Scientific p/n 4461146). The emission of the dye was measured with the temperature changes from 20°C to 95°C, at a rate of increase of 1°C min^-1, as the protein unfolds the dye binds to exposed hydrophobic regions and results in increase in fluorescence emission. This was measured on QuantStudio™ Real-time PCR machine (ThermoFisher Scientific). The curves obtained were differentiated, and the ratio of differentials was used to determine the melting temperature of 37 the proteins. For analysis the Tm of around 75-80°C refers to the IgG2a-Fc region and the other temperature change refers to the scFv. Cell Lines Cell lines were obtained from the American Type Culture Collection. HEK-293T cells were cultured in complete Iscove’s modified Dulbecco’s medium (Sigma Aldrich) supplemented with 10% FCS and 2mM Glutamax (Gibco) (cIMDM). SupT1 were maintained in complete RPMI medium (Sigma Aldrich) supplemented with 10% FCS and 2mM Glutamax (Gibco) (cRPMI). HepG2 cells were maintained in complete MEM supplemented with 10% FCS and 2mM Glutamax (Gibco). Hepa1/6 cells were maintained in complete high glucose DMEM supplemented with 10% FCS and 2mM Glutamax (Gibco). SupT1 cells were transduced with SFG vector for enhanced green fluorescence protein (eGFP) (SupT1-GFP), and SupT1 cells expressing either CD19, GPC3, or dGPC3 were transduced with SFG vector with both the antigen of interest and eGFP (SupT1-CD19, SupT1-GPC3 and SupT1-dGPC3 respectively). T-cell isolation, activation, and lentiviral transduction Healthy donor-derived enriched CD4+/8+ T-cells were isolated from leukocyte cones using SepMate^TM, RosetteSep^TM reagent (Stemcell Technologies) and Ficoll^® Paque density grade centrifugation. Activation/transduction at small-scale was designed to mimic the cGMP Miltenyi CliniMACS Prodigy^®-based CAR-T manufacture process on ALLCAR19. Selected T- cells were maintained in RPMI supplemented with 10% fetal calf serum and 10ng/ml IL7/IL15 (cRPMI). Cells were activated with TransAct^TM (Miltenyi Biotec) and transduced with neat retrovirus on retronectin-coated plates at a multiplicity, cell were maintained for a further 4- days for a total manufacturing time of 8-days. Retrovial transduction efficiency was determined using flow cytometry staining of a marker gene and sunsequent functional assays were set up. Functional Assays CAR-T co-culture proliferation assay was set up with 1:1 culture of non-transduced (NT)/CAR- T with target suspension cell lines (MitomycinC treated for 2hrs) in cRPMI, for 7-days. At completion of co-culture, CountBright™ beads (Thermofisher Scientific) were added to each sample during flow cytometry to determine absolute cell numbers. For co-culture proliferation assays with adherent target, targets were plated 24 hours before assay set up to allow cell adherence. Allowing for target cell doubling rate the assay was set up with 1:1 culture of non- transduced(NT)CAR-T for 7 days. Cytotoxicity was determined in a flow cytometry-based killing assay (FBK). NT or CAR-T effectors were serially diluted 2-fold in a 96 well plate. Target 38 cells were added towards effector to target (E:T) ratios mentioned specifically in figure legends and incubated for 48-hours in cRPMI without additional cytokines. For SupT1 target FBK, cytotoxicity was determined by the number of remaining viable targets by flow cytometry and CountBright™ beads were added to determine absolute cell numbers. For adherent target FBK, cytotoxicity was determined using surrogate value for specific cell lysis using Lonza Toxilight™ assay as was used as per manufactures protocol. Cytokines were measured by using enzyme-linked immunosorbent assay (BioLegend) according to manufacturer’s instructions from diluted co-culture supernatant. In vivo modelling All protocols were performed in accordance with a UK Home Office approved project licence. Disease was established in 8–12-week-old female NOD/SCID γ (NSG) mice, housed in individually ventilated cages (IVCs), via intravenous injection of 2x10⁶ HepG2-FLUC followed by 2x10⁶ NT or CAR T-cells 7 days later via intraperitoneal injection. Tumour burden was measured bi-weekly via bioluminescent imaging (BLI) using the IVIS spectrum in vivo imaging system (Perkin Elmer) following intraperitoneal (IP) injection of 2mg D-luciferin in 100 µl PBS. Photon emission from HepG2 cells was measured as photons/sec/cm²/steradian. When shed antigen was introduced it was injected alongside the CAR T cells at a concentration of 10 μg/mL. For the immunocompetent model 8-12 week old female C57BL/6 mice were injected subcutaneously in the right flank with 10x10⁶ cells in 100uL PBS. Once tumours were established (volume approx 70-100 mm³) after 10 days 2x10⁶ CAR T cells were injected intravenously in 200 μL PBS. Tumour volume was measure twice weekly using digital callipers, measuring length, width and height. Total tumour volume was calculated using the calculation (width x length x height) x (3.14/2). In this study maximal tumour volume as defined in the license was not reached. Where blood draws were mentioned, 100 μL of blood was taken and used for a measure of CAR T expansion. At endpoints CAR T persistence was measured in the spleen. Mice were weighed at least twice weekly. Animals with >10% weight loss or those displaying evidence of graft-versus-host disease or disease progression, including hunched posture, poor coat condition, reduced mobility, piloerection were killed. Statistics For co-culture analysis statistical analysis was performed by two-way ANOVA with Dunnett’s post-test for comparison between CARs within the same target group, otherwise a t test was used to determine statistically significant differences between samples for normally distributed 39 variables. *p < 0.05, **p < 0.001 and ***p< 0.0001. Statistical analysis was performed on GraphPad Prism 9.4. RESULTS Example 1 - Shed antigen presents an opportunity for novel binder design for targeting GPC3 GPC3 has been reported to be an HCC-specific target that may promote tumour growth but also can be detected in patient serum. Previous literature suggests that this could be a hindrance to successful GPC3 targeting. Serum from patients with liver tumours which were metastasis from non-HCC with primary HCC tumours was compared. ELISA of the serum (Figure 1b) showed that GPC3 was detected in the majority of patient samples, though there was a 100-fold difference between lowest and highest detection, suggesting that expression level on tumour is variable. Membrane expression of two widely used liver cancer cell lines, one human and one murine, was compared. They both showed distinct positive staining via flow cytometry (Figure 1e). Over-expressing cell lines were created by retroviral transduction of both full length and truncated antigen (Figure 1c). Confirmation of antigen positivity was seen through dual staining of eGFP and GPC3 itself (Figure 1d). Example 2 - Generation of binders to GPC3 After confirming a selection of genetically diverse antibody sequences, sequences were cloned into scFv format for further screening (Figure 2a). The design of the phage campaign was to elucidate binders which bind membrane proximally and are not impeded by shed antigen. Part of the initial screening was to perform an ELISA to test whether the binders would still bind to membrane version of antigen in the presence of shed antigen. Eight of the initial panel showed binding in the presence of increasing concentrations of antigen compared to GC33 (Figure 2b). SPR characterisation of HR8 and HR10 showed that they had similar kinetic activity and affinity to GC33 to GPC3 when used as an scFv (Figure 2c, d). The KD affinity for all binders was in the low nanomolar range which is sufficient to have good interactions with target and Koff were all similar 1.3E-04 to 1.5E-04. Variation in function is not likely due to binding kinetics. Example 3 - Present binders in CAR T cell format show function against relevant targets in vitro The anti-GPC3 binders HR8 and HR10 alongside control GC33 were cloned as a second- generation CAR (Figure 3a), incorporating a CD28-stalk/transmembrane domain and a CD28- CD3zeta endodomain. This was encoded in a γ-retroviral vector with RQR8 marker/sort- 40 suicide gene and used to transduce primary human T cells. CAR transduction was measured using RQR8 as a surrogate marker staining for anti-human CD34. T cells transduced similarly with all three constructs and showed similar levels of expansion over the transduction period. The present binder CAR T cells or GC33 control were co-cultured with target cells over expressing antigen for 48 hours (Figure 3b). The present binders showed selective toxicity against full length and truncated cell lines, however GC33 only showed selective toxicity against full length cell lines at all E:T ratios, suggesting that in human patients where multiple versions of GPC3 may exist on the surface of tumour cells GC33 reduces in target killing efficiency. A small amount of non-specific kill (<10%) was seen against antigen negative SupT1 cells and by non-CAR expressing T cell, perhaps due to the high levels of basal activation from the transduction protocol. This is in keeping with other CAR studies and not considered to be of concern. Cytokine secretion was examined and showed that whilst all binders showed similar cytokine secretion against FL cell lines (Figure 3d), the present binders demonstrated a significant increase in IFN-γ production, and though not significant a small increase in IL2 in truncated cell lines. Further, in a 7-day proliferation assay HR8 and HR10 showed significantly increased proliferation against truncated cell lines (~5-fold) and similar levels of proliferation against FL cell line(~7.5-fold) (Figure 3c). Co-culture between CARs and liver specific target cells, HepG2, was performed to test the binders at a lower antigen density and a more physiological level (Figure 3e). In this context, HR8 induces significantly specific cell lysis (~100%) compared to the other scFv-CAR constructs. Whilst there is no statistical difference in proliferation of the different CAR constructs against HepG2 (Figure 3f), there were higher levels of proliferation of HR8 when compared to HR10 and GC33 control. All CARs proliferated at least 3-fold when compared to non-transduced T cells. Example 4 - Present CAR binders are not inhibited by shed antigen in vitro The impact of shed antigen on CAR function was investigated utilising the range of concentration measured in patient serum samples and the literature (0.1-10 μg/mL). sGPC3 (amino acids 1 to 494) was purified from the supernatant of 293T cells transfected to produce the GPC3 protein without the membrane proximal region and mimic the shed version of the antigen. It is hypothesised that shed GPC3 can compete with the membrane bound version for the binding of the scFv GC33. This interaction may result in misdirection of the antigen- specific effector function of CAR T cells from tumour cells to shed antigen, and hence reduced tumour cell clearance. As predicted, at both 5 μg/mL and 10 μg/mL of shed antigen, GC33 cytotoxicity was significantly reduced while HR8 and HR10 retained cytotoxic capacity (Figure 3g). 41 Example 5 - Shed antigen reduces function of GC33-CAR T cells in vivo To translate these findings into an in vivo model, the CARs alone were tested against targets (Figure 4a, b).2 x 10⁶ HepG2 fluc+ i.p. was given to NSG mice and tumour growth monitored over the course of the study using bioluminescent imaging. On day 7 CAR T cells were given via tail vein injection. After 10 days post CAR T infusion, tumour burden started to reduce in all CAR T cell groups. By day 28, there was significant reduction in disease burden in all CAR groups compared to non-transduced T cells (Figure 4b). Following this the model was repeated, incorporating shed GPC3 at 10 μg/mL into the CAR T infusion (Figure 4c), to determine whether shed antigen would inhibit the CAR T cells and ‘turn-off’ tumour control. Mice in all groups followed a similar trajectory of tumour growth as in the previous experiment, until day 17 where there is a significant difference in tumour control in the HR8 and HR10 CAR T cell treated groups compared to GC33 CAR T treated group. Tumour control was maintained in the present binder groups, whilst GC33 tumour burden was not controlled and mimicked the tumour growth trajectory of non-transduced CAR T cells. In this second study with shed protein, at time of experiment endpoint, spleens of mice were harvested. As this experiment was conducted on NSG mice, they should have no endogenous T cells, therefore T cell count in the splenocytes at takedown can be a direct inference from CAR T cell expansion. HR8 CAR T cells showed significantly higher amounts in the spleens, though the mean number of HR10 cells was higher it didn’t reach statistical significance, likely due to mouse variability (Figure 4d). Example 6 - Present binders exert better tumour control in immunocompetent xenograft model of HCC HR8 and HR10 binders were compared against GC33 in immunocompetent models of HCC (Figure 5). The region of binding of the present scFvs is not conserved between mouse and human GPC3, so a Hepa1/6 mouse liver cancer cell line was transduced to express the human version of the antigen. Given only a small portion of the protein is different between mouse and humans, there was no tumour rejection and engraftment was still observed. C57BL/6 mice were given 10x10⁶ Hepa1/6-huGPC3 subcutaneously in 100 μL PBS. Within 7 days, small but palpable tumours were observed, and tumours were between 50-100mm³. Mice were divided between the 4 treatment groups to average the tumour volume (Figure 5a). At day 7, 2x10⁶ CAR T cells or non-transduced T cells were administered by i.v tail vein injection. Tumour control was seen from day 14 in all mouse groups and continued to reduce tumour volume until day 28. GC33 treated mice grew back their tumours to original starting volume by day 40. HR10 did not eliminate the tumour but continued to control tumour volume until experimental 42 endpoint at day 40, whilst HR8 eliminated tumours in 4/5 mice, and the final mouse had a very small tumour volume at experiment endpoint (Figure 5b). To see whether there is any difference in CAR T expansion to explain tumour control observed, day 10 after CAR T infusion mice were bled for 100uL PB by tail vein. PBMCs were isolated and stained for flow cytometry. CD3+ T cells were quantified, and from them CAR+ T cells were quantified (Figure 5c-e). Good CAR T cell expansion was observed in all treatment groups with expansion of both T cells and CAR T cells, suggesting a bystander effect to aid tumour control. Though it had not reached significance HR8 and HR10 showed improved expansion compared to GC33. There is limited data to demonstrate the level of shed GPC3 in cell lines however an immunocompetent model is more clinically relevant than an NSG model, therefore the better performance of the novel binders in both models provides an exciting opportunity for improved CAR T cell performance in the treatment of HCC. DISCUSSION The primary ability of a CAR T cell to function is the ability of the binder sequence to interact with the target antigen. As yet, the results of published clinical trials of GPC3 targeted therapies have been unsatisfactory. In a study treating 13 patients with CAR T cells, one patient had partial response, one had stable disease, the rest died during follow up (Shi et al., Clin Cancer Res, 2020; 26(15):3979–89). Use of GC33 as a mAb (codrituzumab) was well tolerated but failed to elicit adequate response as a single agent, and there was limited correlation in terms of progression free survival and IHC expression (Zhu et al., Clin Cancer Res, 2013; 19(4):920–8). As such improvements can be made. In the current study, it was shown that the efficacy of GC33 CAR-T cell function is blocked by sGPC3 in vitro and in vivo. This competitive inhibition is consistent with the universal agonistic effect of shed cell surface proteoglycan on their corresponding membrane proteoglycans. Cell- surface proteoglycans include two major subfamilies: syndecans and glypicans. In contrast to the well-described proteolytic cleavage of syndecans there is limited available evidence demonstrating how the shedding of glypicans occurs. Binders were screened against truncated antigen (comprising just the membrane proximal region) and also in the presence of purified shed GPC3 (lacking the GPI-anchoring domain and membrane proximal region to mimic the native serum GPC3). Cells overexpressing full length or truncated GPC3 were created to investigate the influence of shed GPC3 on CAR-T cell treatment in vitro and in vivo. Under similar tumour burden/antigen density, shed GPC3-reduced binding signal with GPC3. This suggested that the impaired antitumor activity of CAR-T cells and monoclonal antibody Codrituzumab 43 observed in clinical studies study might be mainly caused by the sGPC3-induced blocking effect. In optimizing CAR-T constructs against GPC3, we devised a strategy to select clones from a phage display campaign that recognised membrane proximal region of the target antigen. We then prioritised clones which retained binding in the presence of shed antigen as an scFv. These selected clones displayed effector function in CAR T format in both in vitro against overexpressing and native expressing cell lines. Our clones showed better cytokine secretion and proliferation as well as tumour clearance. These results were corroborated in vivo both in NSG model and then an immunocompetent model that better recapitulates the solid tumour setting. The study shows how antigen shedding may render some current immunotherapies insufficient in the context of solid tumours and in particular HCC. Example 7 – HR8 binding is not inhibited by shed antigen, in contrast to known binders Previously described anti-GPC3 antibodies (Table 1) and HR8 were cloned into a scFv- muIgG2a format, with a GFP marker gene for transfection. 293T cells were transiently transfected with these plasmids to produce the scFv protein alongside the IgG2a FC tag used for detection with secondary antibody in flow cytometry assays. Transfection was confirmed by GFP positive marker gene expression in all samples. Following incubation with the supernatant comprising scFv in the presence of different concentration of sGPC3, SupT1 cell samples were washed and then incubated with an anti-murine IgG2a PE conjugated antibody to detect binding. Table 1. Anti-GPC3 binders

Some reduction in binding in the presence of 5 μg/mL shed protein is observed for all binders apart from HR8. A dramatic loss of binding is observed at 10 μg/mL for all binders apart from HR8. No binding is seen in the NT control (Figure 6A and B). 44 Example 8 – HR8-CAR is not inhibited by shed antigen, in contrast to known binders The scFv sequences were then cloned into the a 2nd generation CD28Z retroviral vector backbone in keeping with the existing constructs tested. PBMCs were transduced with retrovirus containing the CAR constructs. Post-transduction, CAR T expression was determined (all over >75% expression), CARs were normalised with NTs to lowest expression and incubated at a 1:1 ratio with target cells for 24 hours. Co-cultures were stained after 24 hours for viability, T cells were identified using CD3 antibody, and target cells identified by constitutive GFP expression. Cells were enumerated using Countbright counting beads. In the absence of shed antigen (0 μg/mL) all clones displayed killing of target cells, with GC33 and HR8 having comparable efficacy. However, in the presence of shed antigen, the efficacy of all clones other than HR8 was compromised in a dose dependant manner (Figure 7). In the absence of shed antigen HR8, GC33, M3C11 are functionally similar, however YP7 and M1E07 are not as efficacious as the other binders. In the presence of 5 μg/mL shed antigen, GC33, YP7, M1E07 show statistically worse killing than HR8. In the presence of 10 μg/mL, all binders show statistically worse killing than HR8 (Figure 8). 45

Claims

CLAIMS 1. A polypeptide comprising an antigen-binding domain which binds to a region between amino acids S495 to H580 of Glypican-3 (GPC3), wherein binding is not inhibited by a soluble GPC3.

2. The polypeptide according to claim 1, wherein the antigen-binding domain comprises a heavy chain variable region (VH) having complementarity determining regions (CDRs): HCDR1, HCDR2 and HCDR3; and a light chain variable region (VL) having CDRs: LCDR1, LCDR2 and LCDR3 selected from the following: i. HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; or ii. HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12; optionally wherein one or more of the CDRs comprises one, two or three amino acid mutations.

3. The polypeptide according to claim 1 or claim 2, wherein the antigen-binding domain comprises: i. HCDR1 - SEQ ID NO: 1, HCDR2 - SEQ ID NO: 2; HCDR3 - SEQ ID NO: 3; LCDR1 - SEQ ID NO: 4; LCDR2 - SEQ ID NO: 5; LCDR3 - SEQ ID NO: 6; or ii. HCDR1 - SEQ ID NO: 7; HCDR2 - SEQ ID NO: 8; HCDR3 - SEQ ID NO: 9; LCDR1 - SEQ ID NO: 10; LCDR2 - SEQ ID NO: 11; LCDR3 - SEQ ID NO: 12.

4. The polypeptide according to any preceding claim, wherein the antigen-binding domain comprises: i. a VH region having the sequence shown as SEQ ID NO: 13, or a variant having at least 80% sequence identity thereto; and a VL region having the sequence shown as SEQ ID NO: 14, or a variant of having at least 80% sequence identity thereto; or ii. a VH region having the sequence shown as SEQ ID NO: 15, or a variant having at least 80% sequence identity thereto; and a VL region having the sequence shown as SEQ ID NO: 16, or a variant of having at least 80% sequence identity thereto. 46

5. The polypeptide according to any preceding claim, wherein the antigen-binding domain comprises: i. the sequence shown as SEQ ID NO: 17, or a variant having at least 80% sequence identity thereto; or ii. the sequence shown as SEQ ID NO: 18, or a variant having at least 80% sequence identity thereto.

6. An antibody or antigen-binding fragment thereof comprising the polypeptide comprising according to any preceding claim.

7. The antibody or antigen-binding fragment thereof according to claim 6, wherein the antibody or antigen-binding fragment thereof is a scFv, a monoclonal antibody or fragment thereof, a humanized antibody or fragment thereof, or a bi-specific T cell activator molecule, such as a bi-specific T cell engager (BiTE).

8. An antibody conjugate comprising the antibody or antigen-binding fragment thereof according to claim 6 or claim 7.

9. A chimeric antigen receptor (CAR) comprising a polypeptide according to any preceding claim.

10. The CAR according to claim 9, which comprises a transmembrane domain, preferably a CD28 transmembrane domain.

11. The CAR according to claim 10, wherein the polypeptide comprising an antigen- binding domain and the transmembrane domain are connected by a spacer, preferably wherein the spacer comprises a CD28 hinge.

12. The CAR according to claim 10 or claim 11, which comprises an intracellular T cell signalling domain, preferably wherein the intracellular T cell signalling domain comprises a CD28 endodomain and a CD3-Zeta endodomain.

13. The CAR according to claim 12, which comprises the sequence selected from the group comprising SEQ ID NO: 19 or SEQ ID NO: 20, or a variant which has at least 80% sequence identity thereto and retains the capacity to i) bind GPC3 and ii) induce T cell signalling.

14. A polynucleotide comprising a nucleic acid sequence encoding a polypeptide comprising an antigen-binding domain according to any of claims 1 to 5, an antibody or antigen-binding fragment thereof according to any of claims 6 to 7, or a CAR according to any of claims 9 to 13.

15. A vector which comprises a polynucleotide according to claim 14.

16. A cell which comprises a CAR according to any of claims 9 to 13, a polynucleotide according to claim 14, or a vector according to claim 15.

17. The cell according to claim 16, wherein the cell is a T cell or a natural killer (NK) cell.

18. A method for making a cell according to claim 16 or claim 17, which comprises the step of introducing a polynucleotide according to claim 14 or a vector according to claim 15 into said cell.

19. A pharmaceutical composition which comprises a polypeptide comprising an antigen- binding domain according to any of claims 1 to 5, an antibody or antigen-binding fragment thereof according to claim 6 or 7, an antibody conjugate according to claim 8, a polynucleotide according to claim 14, a vector according to claim 15, or a cell according to claim 16 or 17, together with a pharmaceutically acceptable carrier, diluent or excipient.

20. A polypeptide comprising an antigen-binding domain according to any of claims 1 to 5, an antibody or antigen-binding fragment thereof according to claim 6 or 7, an antibody conjugate according to claim 8, a polynucleotide according to claim 14, a vector according to claim 15, a cell according to claim 16 or 17, or a pharmaceutical composition according to claim 19 for use in a method of treating a disease in a subject.

21. A method for treating a disease which comprises the step of administering a polypeptide comprising an antigen-binding domain according to any of claims 1 to 5, an antibody or antigen-binding fragment thereof according to claim 6 or 7, an antibody conjugate according to claim 8, a polynucleotide according to claim 14, a vector according to claim 15, a cell according to claim 16 or 17, or a pharmaceutical composition according to claim 19 to a subject.

22. Use of a polypeptide comprising an antigen-binding domain according to any of claims 1 to 5, an antibody or antigen-binding fragment thereof according to claim 6 or 7, an antibody conjugate according to claim 8, a polynucleotide according to claim 14, a vector according to claim 15, a cell according to claim 16 or 17, or a pharmaceutical composition according to claim 19 in the manufacture of a medicament for treating a disease.

23. The polypeptide comprising an antigen-binding domain, antibody or antigen-binding fragment thereof, antibody conjugate, polynucleotide, vector, cell or pharmaceutical composition for use according to claim 20, the method according to claim 21, or the use according to claim 22, wherein the disease is cancer, preferably wherein the cancer is hepatocellular carcinoma (HCC).