WO2005116646A1 - Method for the identification of a polypeptide which binds to a given pmhc complex - Google Patents

Abstract

A method for the identification of a polypeptide which binds to a given peptide-MHC complex or CD1-antigen complex, and to a nucleoprotein display library wherein the nucleoprotein particles display polypeptides comprising diverse synthetic T cell receptor variable domain sequences derived from one or more TCRs which bind a first peptide MHC complex or CD1-antigen.

Description

Method for the identification of a polypeptide which binds to a given pMHC complex

The invention relates to a method for the identification of a polypeptide which binds to a given peptide-MHC complex or CD 1 -antigen complex, and to a nucleoprotein display library wherein the nucleoprotein particles display polypeptides comprising diverse synthetic T cell receptor ("TCR") variable domain sequences derived from one or more TCRs which bind a first peptide MHC complex or CD 1 -antigen.

Background to the Invention

Native TCRs

As is described in, for example, WO 99/60120 TCRs mediate the recognition of specific Major Histocompatibility Complex (MHC)-peptide complexes by T cells and, as such, are essential to the functioning of the cellular arm of the immune system.

Antibodies and TCRs are the only two types of molecules which recognise antigens in a specific manner, and thus the TCR is the only receptor for particular peptide antigens presented in MHC, the alien peptide often being the only sign of an abnormality within a cell. T cell recognition occurs when a T-cell and an antigen presenting cell (APC) are in direct physical contact, and is initiated by ligation of antigen-specific

TCRs with pMHC complexes.

The native TCR is a heterodimeric cell surface protein of the immunoglobulin superfamily which is associated with invariant proteins of the CD3 complex involved in mediating signal transduction. TCRs exist in αβ and γδ forms, which are structurally similar but have quite distinct anatomical locations and probably functions. The MHC class I and class II ligands are also immunoglobulin superfamily proteins but are specialised for antigen presentation, with a highly polymorphic peptide binding site which enables them to present a diverse array of short peptide f agments at the APC cell surface. Two further classes of proteins are known to be capable of functioning as TCR ligands. (1) CD1 antigens are MHC class I-related molecules whose genes are located on a different chromosome from the classical MHC class I and class II antigens. CD1 molecules are capable of presenting peptide and non-peptide (e.g. lipid, glycolipid) moieties to T cells in a manner analogous to conventional class I and class II-MHC- pep complexes. See, for example (Barclay et al, (1997) The Leucocyte Antigen Factsbook 2^nd Edition, Academic Press) and (Bauer (1997) Eur J Immunol 27 (6) 1366-1373)) (2) Bacterial superantigens are soluble toxins which are capable of binding both class II MHC molecules and a subset of TCRs. (Fraser (1989) Nature 339 221-223) Many superantigens exhibit specificity for one or two Vbeta segments, whereas others exhibit more promiscuous binding. In any event, superantigens are capable of eliciting an enhanced immune response by virtue of their ability to stimulate subsets of T cells in a polyclonal fashion.

The extracellular portion of native heterodimeric αβ and γδ TCRs consist of two polypeptides each of which has a membrane-proximal constant domain, and a membrane-distal variable domain. Each of the constant and variable domains includes an intra-chain disulfide bond. The variable domains contain the highly polymorphic loops analogous to the complementarity determining regions (CDRs) of antibodies. CDR3 of αβ TCRs predominantly interact with the peptide presented by MHC, and

CDRs 1 and 2 of αβ TCRs predominantly interact with the peptide and the MHC. The diversity of TCR variable domain sequences is generated via somatic rearrangement of linked variable (V), diversity (D), joining (J), and constant genes

Functional α and γ chain polypeptides are formed by rearranged V-J-C regions, whereas β and δ chains consist of V-D-J-C regions. The extracellular constant domain has a membrane proximal region and an immunoglobulin region. There are single α and δ chain constant domains, known as TRAC and TRDC respectively. The β chain constant domain is composed of one of two different β constant domains, known as TRBCl and TRBC2 ( GT nomenclature). There are four amino acid changes between these β constant domains, three of which are within the domains used to produce the single-chain TCRs displayed on phage particles of the present invention. These changes are all within exon 1 of TRBCl and TRBC2: N₄K₅->K₄N₅ and F₃₇->Y (DVIGT numbering, differences TRBCl ->TRBC2), the final amino acid change between the two TCR β chain constant regions being in exon 3 of TRBCl and

TRBC2: Vι->E. The constant γ domain is composed of one of TRGC1, TRGC2(2x) or TRGC2(3x). The two TRGC2 constant domains differ only in the number of copies of the amino acids encoded by exon 2 of this gene that are present.

The extent of each of the TCR extracellular domains is somewhat variable. However, a person skilled in the art can readily determine the position of the domain boundaries using a reference such as The T Cell Receptor Facts Book, Lefranc & Lefranc, Publ. Academic Press 2001.

Recombinant TCRs

The production of recombinant TCRs is beneficial as these provide soluble TCR analogues suitable for the following purposes:

• Studying the TCR / ligand interactions (e.g. pMHC for αβ TCRs) • Screening for inhibitors of TCR-associated interactions • Providing the basis for potential therapeutics

A number of constructs have been devised to date for the production of recombinant TCRs. These constructs fall into two broad classes, single-chain TCRs and dimeric TCRs, the literature relevant to these constructs is summarised below.

Single-chain TCRs (scTCRs) are artificial constructs consisting of a single amino acid strand, which like native heterodimeric TCRs bind to MHC-peptide complexes. Unfortunately, attempts to produce functional alpha/beta analogue scTCRs by simply linking the alpha and beta chains such that both are expressed in a single open reading frame have been unsuccessful, presumably because of the natural instability of the alpha-beta soluble domain pairing. Accordingly, special techniques using various truncations of either or both of the alpha and beta chains have been necessary for the production of scTCRs. These formats appear to be applicable only to a very limited range of scTCR sequences. Soo Hoo et al (1992) PNAS. 89 (10): 4759-63 report the expression of a mouse TCR in single chain format from the 2C T cell clone using a truncated beta and alpha chain linked with a 25 amino acid linker and bacterial periplasmic expression (see also Schodin et al (1996) Mol. Immunol. 33 (9): 819-29). This design also forms the basis of the m6 single-chain TCR reported by Holler et al (2000) PNAS. 97 (10): 5387-92 which is derived from the 2C scTCR and binds to the same H2-Ld-restricted alloepitope. Shusta et al (2000) Nature Biotechnology 18: 754-759 and US 6,423,538 report using a murine single-chain 2C TCR constructs in yeast display experiments, which produced mutated TCRs with, enhanced thermal stability and solubility. This report also demonstrated the ability of these displayed 2C TCRs to selectively bind cells expressing their cognate pMHC. Khandekar et al (1997) J. Biol. Chem. 272

(51): 32190-7 report a similar design for the murine D10 TCR, although this scTCR was fused to MBP and expressed in bacterial cytoplasm (see also Hare et al (1999) Nat. Struct. Biol. 6 (6): 574-81). Hilyard et al (1994) PNAS. 91 (19): 9057-61 report a human scTCR specific for influenza matrix protein-HLA-A2, using a Vα-linker-Vβ design and expressed in bacterial periplasm.

Chung et al (1994) PNAS. 91 (26) 12654-8 report the production of a human scTCR using a Vα-linker-Vβ-Cβ design and expression on the surface of a mammalian cell line. This report does not include any reference to peptide-HLA specific binding of the scTCR. Plaksin et al (1997) J. Immunol. 158 (5): 2218-27 report a similar Vα-linker-

Vβ-Cβ design for producing a murine scTCR specific for an HIV gpl20-H-2D^d epitope. This scTCR is expressed as bacterial inclusion bodies and refolded in vitro.

A number of papers describe the production of TCR heterodimers which include the native disulphide bridge which connects the respective subunits (Garboczi, et al,

(1996), Nature 384(6605): 134-41; Garboczi, et al, (1996), J Immunol 157(12): 5403- 10; Chang et al, (1994), PNAS USA 91: 11408-11412; Davodeau et al, (1993), J. Biol. Chem. 268(21): 15455-15460; Golden et al, (1997), J. Imm. Mefh. 206: 163- 169; US Patent No. 6080840). However, although such TCRs can be recognised by TCR-specific antibodies, none were shown to recognise its native ligand at anything other than relatively high concentrations and/or were not stable.

In WO 99/60120, a soluble TCR is described which is correctly folded so that it is capable of recognising its native ligand, is stable over a period of time, and can be produced in reasonable quantities. This TCR comprises a TCR α or γ chain extracellular domain dimerised to a TCR β or δ chain extracellular domain respectively, by means of a pair of C-terminal dimerisation peptides, such as leucine zippers. This strategy of producing TCRs is generally applicable to all TCRs.

Reiter et al, Immunity, 1995, 2:281-287, details the construction of a soluble molecule comprising disulphide-stabilised TCR α and β variable domains, one of which is linked to a truncated form of Pseudomonas exotoxin (PE38). One of the stated reasons for producing this molecule was to overcome the inherent instability of single- chain TCRs. The position of the novel disulphide bond in the TCR variable domains was identified via homology with the variable domains of antibodies, into which these have previously been introduced (for example see Brinkmanri, et al. (1993), Proc.

Nati Acad. Sci. USA 90: 7538-7542, and Reiter, et al. (1994) Biochemistry 33: 5451- 5459). However, as there is no such homology between antibody and TCR constant domains, such a technique could not be employed to identify appropriate sites for new inter-chain disulphide bonds between TCR constant domains.

As mentioned above Shusta et al (2000) Nature Biotechnology 18: 754-759 report using single-chain 2 C TCR constructs in yeast display experiments. The principle of displaying scTCRs on phage particles has previously been discussed. For example, WO 99/19129 details the production of scTCRs, and summarise a potential method for the production of phage particles displaying scTCRs of the Vα-Linker-Vβ Cβ format. However, this application contains no exemplification demonstrating the production of said phage particles displaying TCR. The application does however refer to a co- pending application:

"The construction of DNA vectors including a DNA segment encoding a sc-TCR molecules fused to a bacteriophage coat protein (gene II or gene VIII) have been described in said pending U.S. application No. 08/813,781."

Furthermore, this application relies on the ability of anti-TCR antibodies or super- antigen MHC complexes to recognise the soluble, non-phage displayed, scTCRs produced to verify their correct conformation. Therefore, true peptide-MHC binding specificity of the scTCRs, in any format, is not conclusively demonstrated. Finally, a further study (Onda et al, (1995) Molecular Immunology 32 (17-18) 1387- 1397) discloses the phage display of two murine TCR α chains in the absence of their respective β chains. This study demonstrated that phage particles displaying one of the TCR α chains (derived from the Al .1 murine hybridoma) bound preferentially to the same peptides immobilised in microtitre wells that the complete TCR would normally respond to when there were presented by the murine Class I MHC I-A^d.

Display Methods It is often desirable to present a given peptide or polypeptide on the surface of a nucleoprotein particle. Such particles may serve as purification aids for the peptide or polypeptide (since the particles carrying the peptide or polypeptide may be separated from unwanted contaminants by sedimentation or other methods). They may also serve as particulate vaccines, the immune response to the surface displayed peptide or polypeptide being stimulated by the particulate presentation. Protein p24 of the yeast retrotransposon, and the hepatitis B surface coat protein are examples of proteins which self assemble into particles. Fusion of the peptide or polypeptide of interest to these particle-forming proteins is a recognised way of presenting the peptide or polypeptide on the surface of the resultant particles. However, particle display methods have primarily been used to identify proteins with desirable properties such as enhanced expression yields, binding and/or stability characteristics. These methods involve creating a diverse pool or 'library' of proteins or polypeptides expressed on the surface of nucleoprotein particles. These particles have two key features, firstly each particle presents a single variant protein or polypeptide, and secondly the genetic material encoding the expressed protein or polypeptide is associated with that of the particle. This library is then subjected to one or more rounds of selection. For example, this may consist of contacting a ligand with a particle-display library of mutated receptors and identifying which mutated receptors bind the ligand with the highest affinity. Once the selection process has been completed the receptor or receptors with the desired properties can be isolated, and their genetic material can be amplified in order to allow the receptors to be sequenced.

These display methods fall into two broad categories, in-vitro and in-vivo display.

All in-vivo display methods rely on a step in which the library, usually encoded in or with the genetic nucleic acid of a replicable particle such as a plasmid or phage replicon is transformed into cells to allow expression of the proteins or polypeptides. (Plϋckthun (2001) Adv Protein Chem 55 367-403). There are a number of repiicon/host systems that have proved suitable for in-vivo display of protein or polypeptides. These include the following

Phage / bacterial cells plasmid / CHO cells Vectors based on the yeast 2μm plasmid / yeast cells bacculovirus / insect cells plasmid / bacterial cells

In-vivo display methods include cell-surface display methods in which a plasmid is introduced into the host cell encoding a fusion protein consisting of the protein or polypeptide of interest fused to a cell surface protein or polypeptide. The expression of this fusion protein leads to the protein or polypeptide of interest being displayed on the surface of the cell. The cells displaying these proteins or polypeptides of interest can then be subjected to a selection process such as FACS and the plasmids obtained from the selected cell or cells can be isolated and sequenced. Cell surface display systems have been devised for mammalian cells (Higuschi (1997) J Immunol.

Methods 202 193-204), yeast cells (Shusta (1999) J Mol Biol 292 949-956) and bacterial cells (Sameulson (2002) J. Biotechnol 96 (2) 129-154).

Numerous reviews of the various in-vivo display techniques have been published. For example, (Hudson (2002) Expert Opin Biol Ther (2001) 1 (5) 845-55) and (Schmitz

(2000) 21 (Supp A) S106-S112).

In-vitro display methods are based on the use of ribosomes to translate libraries of mRNA into a diverse array of protein or polypeptide variants. The linkage between the proteins or polypeptides formed and the mRNA encoding these molecules is maintained by one of two methods. Conventional ribosome display utilises mRNA sequences that encode a short (typically 40-100 amino acid ) linker sequence and the protein or polypeptide to be displayed. The linker sequence allows the displayed protein or polypeptide sufficient space to re-fold without being sterically hindered by the ribosome. The mRNA sequence lacks a 'stop' codon, this ensures that the expressed protein or polypeptide and the RNA remain attached to the ribosome particle. The related mRNA display method is based on the preparation of mRNA sequences encoding the protein or polypeptide of interest and DNA linkers carrying a puromycin moiety. As soon as the ribosome reaches the mRNA/DN A junction translation is stalled and the puromycin forms a covalent linkage to the ribosome. For a recent review of these two related in-vitro display methods see (Amstutz (2001) Curr Opin Biotechnol 12 400-405).

Particularly preferred is the phage display technique which is based on the ability of bacteriophage particles to express a heterologous peptide or polypeptide fused to their surface proteins. (Smith (1985) Science 217 1315-1317). The procedure is quite general, and well understood in the art for the display of polypeptide monomers. However, in the case of polypeptides that in their native form associate as dimers, only the phage display of antibodies appears to have been thoroughly investigated.

For monomeric polypeptide display there are two main procedures:

Firstly (Method A) by inserting into a vector (phagemid) DNA encoding the heterologous peptide or polypeptide fused to the DNA encoding a bacteriophage coat protein. The expression of phage particles displaying the heterologous peptide or polypeptide is then carried out by transfecting bacterial cells with the phagemid, and then infecting the transformed cells with a 'helper phage'. The helper phage acts as a source of the phage proteins not encoded by the phagemid required to produce a functional phage particle.

Secondly (Method B), by inserting DNA encoding the heterologous peptide or polypeptide into a complete phage genome fused to the DNA encoding a bacteriophage coat protein. The expression of phage particles displaying the heterologous peptide or polypeptide is then carried out by infecting bacterial cells with the phage genome. This method has the advantage of the first method of being a 'single-step' process. However, the size of the heterologous DNA sequence that can be successfully packaged into the resulting phage particles is reduced. Ml 3, T7 and Lambda are examples of suitable phages for this method.

A variation on (Method B) the involves adding a DNA sequence encoding a nucleotide binding domain to the DNA in the phage genome encoding the heterologous peptide be displayed, and further adding the corresponding nucleotide binding site to the phage genome. This causes the heterologous peptide to become directly attached to the phage genome. This peptide/genome complex is then packaged into a phage particle which displays the heterologous peptide. This method is fully described in WO 99/11785.

The phage particles can then be recovered and used to study the binding characteristics of the heterologous peptide or polypeptide. Once isolated, phagemid or phage DNA can be recovered from the peptide- or polypeptide-displaying phage particle, and this DNA can be replicated via PCR. The PCR product can be used to sequence the heterologous peptide or polypeptide displayed by a given phage particle.

The phage display of single-chain antibodies and fragments thereof, has become a routine means of studying the binding characteristics of these polypeptides. There are numerous books available that review phage display techniques and the biology of the bacteriophage. (See, for example, Phage Display - A Laboratory Manual, Barbas et al, (2001) Cold Spring Harbour Laboratory Press).

A third phage display method (Method C) relies on the fact that heterologous polypeptides having a cysteine residue at a desired location can be expressed in a soluble form by a phagemid or phage genome, and caused to associate with a modified phage surface protein also having a cysteine residue at a surface exposed position, via the formation of a disulphide linkage between the two cysteines. WO 01/ 05950 details the use of this alternative linkage method for the expression of single-chain antibody- derived peptides.

High Affinity TCRs

T cells mature in the thymus where they undergo at least two selection mechanisms, generally referred to as positive and negative selection. The structures of most, or all, TCRs are believed to share certain general architectural features (Chothia, et al, Embo J(1988) 7: 3745-55) that provide a framework suitable for MHC/peptide binding by the variable complementarity determining regions (CDRs). Thus, most TCRs may have intrinsic affinity for MHC/peptide complexes (Chothia, et al, E bo J (1988) 7: 3745-55). In the thymus, only TCRs with a certain minimal level of affinity for one of the MHC molecules to which they are presented (the "self MHC molecules) will be positively selected. T cells with high affinity for one of the self MHC molecules will be negatively selected (Amsen & Kruisbeek. (1998). Immunol Rev 165: 209-29. Sebzda, et al (1999). Annu Rev Immunol 17: 829-74). TCRs in the cellular immunity can be considered to be analogous to antibodies in the humoral immunity. Antibodies have been successfully used, either as therapeutic agents in their own right (e.g. Herceptin) or as targeting agents (e.g. mylotarg), and interest in this area continues to grow. Similar strategies could be devised using T cell receptors. Thus, soluble TCRs are useful, not only for the purpose of investigating specific TCR-pMHC interactions, but also as a diagnostic tool to detect infection, or to detect autoimmune disease markers, or to detect the efficacy of T cell vaccines. Soluble TCRs also have applications in staining, for example to stain cells for the presence of a particular viral antigen presented in the context of the MHC. Similarly, soluble TCRs can be used to deliver a therapeutic agent, for example a cytotoxic compound or an immunostimulating compound, to cells presenting a particular antigen.

However, two factors have hindered the exploitation of TCRs. Firstly, a generally applicable method for the production of soluble (i.e. non-membrane bound) T cell receptors has not been available until recently. (See WO 03/020763 and WO 04/033685 for details of suitable production methods) Secondly, the affinity of the T cell receptor for its specific pMHC ligand is much lower (K_D in the μM range) than for antibodies (K_D in the nM range). This lower affinity of the TCR is thought to be a result of negative selection during development, and it is therefore probably not possible to find TCRs with high affinity for self-MHC-peptide complexes (Salzmann & Bachmann, Molecular Immunology, 1998, 35:65-71).

Methods for in vitro selection of high affinity antibodies have been developed using phage display technology. Although there is one report of successful phage display of T cell receptors (Weidanz et al, J. Immunol. Methods, 1998, 221:59-76), a detailed examination of this work shows that fully functional (pMHC binding) displayed protein was only achieved for one example TCR. The second example only showed binding to "conformation sensitive antibodies", which recognise correct folding of small parts of the protein not necessarily related to ligand recognition, i.e. the fact that the TCR was recognised by antibodies does not mean that it is capable of binding peptide-MHC complex. This lack of success is probably due to the requirement to use a single-chain recombinant TCR for phage display. In order to force the naturally heterodimeric TCR into a single chain, the α-chain is truncated and a linker is inserted between the C-terminus of the α-chain variable domain and the N-terminus of the β- chain. This approach to the production of soluble TCRs has proved applicable to only very few T cell receptors. It is therefore very unlikely that it will be generally applicable to the display of T cell receptors and therefore for the selection of high affinity mutants of those T cell receptors.

Yeast has also been investigated as a system for displaying T cell receptors and also for selecting stable and high affinity mutants of TCRs (Holler, et al. (2000) Proc. Nail. Acad. Sci. 97, 5387-5392; Shusta, et al. (2000) Nature Biotechnology 18, 754-759; Shusta, et al. (1999) J. Mol Biol. 292, 949-956; Kieke, et al. (1999) Proc. Natl Acad. Sci. 96, 5651-5656). Holler et al. report an increase in affinity from a K_D of 1.5 μM to

9 nM, an increase of 160-fold. However, data for only one TCR was presented, which was the 2C TCR in a single-chain form which, as discussed above, will make this approach not generally applicable.

Other techniques for the production of high affinity TCRs have been directed to altering the amino acid sequence of a TCR so that it can bind more strongly to its ligand. This approach was originally applied to the production of high affinity monoclonal antibodies, and attempts to mutate antibodies, based on modelling of the antibody/hapten interface, produced two to three fold increases in their affinity by altering the amino acid sequence in the CDR3 regions. (Reichmann, et al, (1992) J.

Mol Biol. 224, (4) 913-918).

A further study (Manning et al, (1998) Immunity 8 (4): 413-25) utilised single-point alanine scanning to investigate binding of the murine 2C TCR to the QL9 / L^d complex. This study noted that 2 amino acids in each of the CDR2 loops elicited equivalent or slightly improved binding when replaced by alanine. (53α Asp, 54α Pro, 55β Thr and 56β Glu) A later study (Manning et al, (1999) J Exp Med 189 (3): 461- 70) by the same group carried out further single-point and combinations of alanine scanning mutations to investigate the binding contribution of several amino acids in the variable domain, including the CDR2 loops, of the 2C murine TCR. A mutant 2C TCR containing single alanine substitutions in both the TCR β CDR2 and TCR α

CDR3 loops demonstrated an approximately 2-fold increase in affinity for the QL9 / L^d complex and slower association and disassociation properties. (K_D = 3.2 μM (WT) 2.54μM (mutant), k_off = 19 x 10^"3 S^"1 (WT) and 5.9 x 10^"3 S^"1 (mutant) This study also noted that despite the only 2-fold increase in affinity achieved by directed site-specific mutagenesis, these results indicate the promise in using more sophisticated techniques to obtain even greater improvements in TCR affinity.

It is known that TCRs can exhibit higher affinities for allogenic pMHCs than syngeneic pMHCs. For example, the affinity of the 2C TCR for its natural ligand, dEV8 peptide (EQYKFYS V), presented by MHC H2-K has been determined as K_D =

84.1 x 10⁶ M^"1 ± 12.0 μM (Garcia, et al, (1997) Proc. Natl Acad. Sci. USA. 94: 13838-13843). This compares to a K_D of 1.5 x 10⁷ M^"1 observed for the 2C T cell clone binding to the allogeneic ligand, QL9 peptide (QLSPFPFDL) presented by MHC H2-L^d. (Sykulev, et al, (1994) Proc. Natl. Acad. Sci. USA. 91, pi 1487-11491).

Brief Description of the Invention

Native TCR's are heterodimers which have lengthy transmembrane domains which are essential to maintain their stability as functional dimers. As discussed above, TCRs are useful for research and therapeutic purposes in their soluble forms so display of the insoluble native form has little utility.

There have been a number of publications relating to various means of TCR display which are summarised below: WO 98/39482 describes methods for the phage display of scTCRs. The inventors of this application have also published a paper (Weidanz (1998) J Immunol Methods 221 59-76) that demonstrates the display of two murine scTCRs on phage particles.

WO 01/62908 discloses methods for the phage display of scTCRs and scTCR Ig fusion proteins. However, the functionality (specific pMHC binding) of the constructs disclosed was not assessed.

A retrovirus-mediated method for the display of diverse TCR libraries on the surface of immature T cells has been demonstrated for murine TCRs. The library of mutated

TCRs displayed of the surface of the immature T cells was screened by flow cytometry using pMHC tetramers, and this lead to the identification TCR variants that were either specific for the cognate pMHC, or a variant thereof. (Helmut et al, (2000) PNAS 97 (26) 14578-14583)

A recent paper (Chlewicki et al, (2005) J Mol Biol. 346 (1) :223-39) describes high affinity TCRs comprising mutations in the CDRl, 2 and 3 sequences thereof generated from a yeast display library.

The inventors' co-pending application (WO 2004/044004) is based in part on the finding that functional single chain and dimeric TCRs can be expressed as surface fusions to nucleoprotein particles, and makes available nucleoprotein particles displaying alpha/beta-analogue and gamma/delta-analogue scTCR and dTCR constructs. The nucleoprotein particles on which the TCRs are displayed are preferably phage particles but also include self-aggregating particle-forming proteins, virus-derived and ribosome particles. Such nucleoprotein particle-displayed TCRs are useful for purification and screening purposes, particularly as a diverse library of particle displayed TCRs for biopanning to identify TCRs with desirable characteristics such as high affinity for the target MHC-peptide complex. In the latter connection, particle-displayed scTCRs may be useful for identification of the desired TCR, but that information maybe better applied to the construction of analogous dimeric TCRs for ultimate use in therapy. The invention also includes high affinity TCRs identifiable by these methods.

The present invention provides for the first time methods using highly diverse libraries of nucleoproteins displaying polypeptides comprising synthetic TCR variable domains for the identification of library members which are capable of binding to pMHC complexes or CD 1 -antigen complexes other than those that the TCRs from which the library members were derived are capable of binding. The polypeptides identified by these methods have utility as targeting moieties for therapeutic and diagnostic agents.

Detailed Description of the Invention

The presence and/or expression level of certain pMHC on the surface of a given cell is known to be related to the disease-state of that cell. For example, certain cancers express the NY-ESO-1 protein and this leads to MHC molecules on the surface of such cancer cells presenting peptides derived from the NY-ESO-1 protein. The SLLMWITQC-HLA-A*0201 complex is an example of such a cancer-related pMHC which is expressed by NY-ESO-1 cancer cells. The isolation of ligands, such as TCRs, which bind to these disease and/or cancer pMHC expressed on diseased and/or cancerous cells provides a possible basis for targeting moieties capable of delivering therapeutic and/or diagnostic agents to the diseased or cancerous cells.

Method of identifying TCRs that bind to a given peptide MHC complex or CD1- antigen complex

In one broad aspect the invention provides a method for the identification of a polypeptide which binds to a given peptide-MHC complex or CD 1 -antigen complex, said method comprising contacting members of a diverse library of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic T cell receptor ("TCR") variable domain sequences derived from a TCR which binds to a first peptide-MHC complex or CD 1 -antigen complex wherein diversity resides at least in the variable domains of the said polypeptides and a given peptide-MHC complex or CD 1 -antigen complex different from the first peptide-MHC complex or CD 1 -antigen complex, detecting binding between the library members and said given peptide-MHC complex or CD 1 -antigen complex, isolating a library member detected as binding to the given peptide-MHC complex or CD 1 -antigen complex, and optionally multiplying the isolated library member in an amplification process, the polypeptide displayed on the surface of the library member being taken to be the desired polypeptide.

As used herein the term "synthetic TCR variable domains" is understood to refer to any TCR variable domains which comprise germline TCR variable domain framework sequences in combination with mutations in one or more of the hypervariable regions of the variable domain.

In a preferred embodiment of the method of the invention the nucleoproteins display polypeptides comprising diverse synthetic TCR variable domain sequences derived from a plurality of TCRs which bind to peptide-MHC complexes or CD 1 -antigen complexes different from the given peptide MHC complex or CD 1 -antigen complex.

The binding measurement can be made by any appropriate method. In a preferred embodiment the binding is determined by ELISA.

Example 2 herein provides details of an ELIS A-based method suitable for determining the binding of nucleoprotein-displayed TCRs to pMHCs or CD 1 -antigens.

In another embodiment of the method of the invention the nucleoproteins particles are filamentous phage particles or ribosome particles, and the displayed polypeptides are single chain TCRs ("scTCRs").

The inventors co-pending application (WO 2004/044004) describes the production of filamentous phage particles and ribosomes displaying scTCRs. An alternative embodiment of the method of the invention is provided wherein the library of nucleoproteins comprises (a) a first set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR α variable domain sequences, and/or (b) a second set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR β variable domain sequences, and/or (c) a third set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR α or β variable domain sequences, which polypeptides are associated in pairs with polypeptides also comprising diverse synthetic TCR α or β variable domain sequences.

A further alternative embodiment of the method of the invention is provided wherein the nucleoproteins are filamentous phage particles, and the library includes or consists of members displaying polypeptides which are αβ dimeric TCRs ("dTCRs"), and optionally includes members displaying TCR α-chains and/or TCR β-chains and/or homodimeric αα chains and/or homodimeric ββ chains.

In another embodiment of the invention said library includes members displaying TCR α-chains and/or TCR β-chains and/or homodimeric αα chains and/or homodimeric ββ chains.

In a further embodiment of the method of the invention the said phage-displayed αβ dTCRs comprise a first polypeptide wherein a sequence corresponding to a TCR α chain variable domain sequence is fused to the N terminus of a sequence corresponding to a TCR α chain constant domain extracellular sequence, and a second polypeptide wherein a sequence corresponding to a TCR βchain variable domain sequence fused to the N terminus a sequence corresponding to a TCR β chain constant domain extracellular sequence, the first and second polypeptides being linked by a disulfide bond which has no equivalent in native αβ T cell receptors, and one of said first or second polypeptides being linked by a peptide bond at its C- terminus to a surface exposed amino acid residue of the phage particle.

In a specific embodiment of the method of the invention said first and second polypeptides are linked by a disulfide bond between cysteine residues substituted for Thr 48 of exon 1 of TRAC*01 and Ser 57 of exon 1 of TRBC1*01 or TRBC2*01 or the non-human equivalent thereof.

The residues for mutation to cysteine in order to form the non-native disulfide interchain bind are identified using ImMunoGeneTics (TJVIGT) nomenclature. (The T cell Receptor Factsbook 2^nd Edition (2001) LeFranc and Lefranc, Academic Press) WO 03/020763 provides a detailed description of the methods required to introduce the specified non-native disulfide interchain bond and alternative residues between which it may be sited.

One embodiment of the invention is provided by a method wherein the nucleoproteins are filamentous phage particles, and the library includes members displaying polypeptides which are αβ dimeric TCRs ("dTCRs"), and optionally includes members displaying TCR α-chains and/or TCR β-chains and/or homodimeric αα chains and/or homodimeric ββ chains, for the purpose of identifying library members which bind to a given pMHC or CD 1 -antigen and isolating those which bind,

(i) several members of the library are contacted in parallel with the given pMHC or CD 1 -antigen and members which bind to the pMHC or CD 1 -antigen are identified, (ii) one or more members which bind to the given pMHC or CDl-antigen assessed in step (i) are selected, and the variable domain sequences of the displayed TCRs is/are determined, (iii) soluble form TCRs incorporating the thus-determined variable domain sequences, are created, (iv) the affinities and/or the off-rates for the given pMHC of these TCRs are determined, and (v) one or more TCRs having the desired affinity and/or off-rate determined in step (iv) are selected.

The above methods and nucleoprotein library of the invention enable identification of αβ dTCRs, TCR α-chains, TCR β-chains, TCR αα homodimers and/or TCR ββ homodimers that will bind to the target pMHC or CD 1 -antigen

As used herein the term "soluble TCR" is understood to refer to any TCR that: (i) lacks the native transmembrane domain thereof and (ii) is not associated with a nuceloprotein and retains the ability to bind to a pMHC or CD 1 -antigen.

The affinity and/or the off-rate measurements can be made by any appropriate method. In a preferred embodiment the said affinities and/or the off-rates are determined by Surface Plasmon Resonance (SPR).

Example 3 herein provides details of an SPR-based (Biacore) method suitable for determining the affinities and off-rates for the interactions between soluble form TCRs and pMHCs or CD 1 -antigens.

Another embodiment of the invention is provided by a method in which the displayed polypeptides are derived from the given TCR(s) by mutation of at least one of its/their complementarity determining regions of the variable domains. For example the displayed polypeptides can be derived from the given TCR(s) by mutation of at least one of its/their CDR2 or CDR3 complimentarity determining regions of the variable domains.

A further embodiment of the invention is provided by a method in which the given

TCR(s) which is/are the starting point for the creation of the library used in the method of the invention is/are naturally occurring.

As will be obvious to those skilled in the art the above methods will be equally applicable to the identification of ligands from the library members to any other complexes that are capable of being bound by the library members.

TCR Libraries

A further broad embodiment of the invention is provided by a library of nucleoproteins comprising a first set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR α variable domain sequences, and/or a second set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR β variable domain sequences, and/or a third set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic

TCR α or β variable domain sequences, which polypeptides are associated in pairs with polypeptides also comprising diverse synthetic TCR α or β variable domain sequences., providing that the said library includes at least one of sets (a) and (b).

A specific embodiment of the invention is provided by such libraries in which diversity resides in at least the CDR2 or CDR3 sequences of the variable domain sequences.

A preferred embodiment of the invention is provided by libraries of the invention in which the nucleoproteins are phage or ribosome particles. A specific embodiment of the libraries of the invention is provided wherein the synthetic variable domain sequences of the displayed polypeptides are located N- terminal to part of a TCR chain which includes all or part of the constant domain sequence thereof, except the transmembrane domain thereof.

A further specific embodiment of the libraries of the invention is provided wherein in the third set of nucleoproteins the said association in pairs is maintained at least in part by a disulfide bond which has no equivalent in native αβ T cell receptors between introduced cysteine residues in the TCR constant domain sequences of the paired polypeptides.

In a preferred embodiment of the libraries of the invention the phage particles or ribosomes on which the displayed polypeptide, or one member of the displayed associated polypeptide pair, is linked by a peptide bond at its C-terminus to a surface exposed amino acid residue of the phage particle or ribosome.

In a specific embodiment of the libraries of the invention the first and second polypeptides are linked by a disulfide bond between cysteine residues substituted for Thr 48 of exon 1 of TRAC*01 and Ser 57 of exon 1 of TRBCl *01 or TRBC2*01 or the non-human equivalent thereof.

In another embodiment of the libraries of the invention the synthetic α and β variable domain sequences are derived by mutation of at least two TCRs.

The Displayed TCRs

The inventors co-pending application (WO 2004/044004) describes the methods required to produce nucleoproteins displaying the preferred scTCRs and dTCRs for use in the present invention.

Required to prepare a library of mutated TCRs, are nucleic acids encoding (a) one chain of a dTCR polypeptide pair and (b) the other chain of a dTCR polypeptide pair fused to a nucleic acid sequence encoding a protein capable of forming part of the surface of a nucleoprotein particle; or nucleic acid encoding a scTCR polypeptide fused to a nucleic acid sequence encoding a protein capable of forming part of the surface of a nucleoprotein particle, the dTCR pair or scTCR.

For expression of TCRs host cells may be used transformed with an expression vector comprising nucleic acid encoding (a) one chain of a dTCR polypeptide pair and (b) the other chain of a dTCR polypeptide pair fused to a nucleic acid sequence encoding a protein capable of forming part of the surface of a nucleoprotein particle; or nucleic acid encoding a scTCR polypeptide fused to a nucleic acid sequence encoding a protein capable of forming part of the surface of a nucleoprotein particle, the dTCR pair or scTCR,

Preferably the expression system comprises phagemid or phage genome vectors expressing nucleic acids (a) and (b). Preferably these phagemid or phage genome vectors is (are) those which encode bacteriophage gill or gVIII coat proteins.

Transformed cells are incubated to allow the expression of the TCR-displaying nucleoprotein particles. These particles can then be used in assays to identify TCR variants with the desired affinity and/or off-rate characteristics. Any particles that possess the desired characteristics under investigation can then be isolated. The DNA encoding these TCRs can then be amplified by PCR and the sequence determined.

It is known that high expression levels of an exogenous polypeptide may be toxic to the host cell. In such cases, either a host strain which is more tolerant of the exogenous polypeptide must be found, or the expression levels in the host cell must be limited to a level which is tolerated. For example (Beekwilder et al, (1999) Gene 228 (1-2) 23- 31) report that only mutated forms of a potato protease inhibitor (PI2) which contained deletions or amber stop codons would be successfully selected from a phage display library. There are several strategies for limiting the expression levels of an exogenous polypeptide from a given expression system in a host which may be suitable for the limiting the expression levels of a scTCR, or one, or both TCR chains of a dTCR . These strategies are described in the inventors co-pending application (WO 2004/044004).

Correct pairing of scTCR polypeptide variable domain sequences after expression is preferably assisted by an introduced disulfide bond in the extracellular constant domain of the scTCR. Without wanting to be limited by theory, the novel disulfide bond is believed to provide extra stability to the scTCR during the folding process and thereby facilitating correct pairing of the first and second segments.

Also as mentioned above, for dTCR phage display, one of the dTCR polypeptide pair is expressed as if it were eventually to be displayed as a monomeric polypeptide on the phage, and the other of the dTCR polypeptide pair is co-expressed in the same host cell. As the phage particle self assembles, the two polypeptides self associate for display as a dimer on the phage. Again, in the preferred embodiment of this aspect of the invention, correct folding during association of the polypeptide pair is assisted by a disulfide bond between the constant sequences. Further details of a procedure for phage display of a dTCR having an interchain disulfide bond appear in the Examples contained within the inventors co-pending application WO 2004/044004.

As an alternative, the phage displaying the first chain of the dTCR may be expressed first, and the second chain polypeptide may be contacted with the expressed phage in a subsequent step, for association as a functional dTCR on the phage surface.

The preferred in-vitro TCR display method for biopanning to identify TCRs comprising mutated CDR2 sequences having high affinity and/or slow off-rates for a target peptide-MHC complex is ribosomal display. Firstly, a DNA library is constructed that encodes a diverse array of mutated scTCRs or dTCR polypeptides using the techniques discussed above. The DNA library is then contacted with RNA polymerase in order to produce a complementary mRNA library. Optionally, for mRNA display techniques the mRNA sequences can then be ligated to a DNA sequence comprising a puromycin-binding site. These genetic constructs are then contacted with ribosomes in-vitro under conditions allowing the translation of the scTCR polypeptide or the first polypeptide of the dTCR pair, hi the case of the dTCR, the second of the polypeptide pairs is separately expressed and contacted with the ribosome-displayed first polypeptide, for association between the two, preferably assisted by the formation of the disulphide bond between constant domains. Alternatively, mRNA encoding both chains of the TCR may be contacted with ribosomes in-vitro under conditions allowing the translation of the TCR chains such that a ribosome displaying a dTCR is formed. These scTCR- or dTCR-displaying ribosomes can then used for screening or in assays to identify TCR variants with specific enhanced characteristics. Any particles that possess the enhanced characteristics under investigation can then be isolated. The mRNA encoding these TCRs can then converted to the complementary DNA sequences using reverse transcriptase. This DNA can then be amplified by PCR and the sequence determined.

scTCRs or dTCRs of the present invention may be displayed on nucleoprotein particles, for example phage particles, preferably filamentous phage particles, by, for example, the following two means:

(i) The C-terminus of one member of the dTCR polypeptide pair, or the C-terminus of the scTCR polypeptide, or the C-terminus of a short peptide linker attached to the C- terminus of either, can be directly linked by a peptide bond to a surface exposed residue of the nucleoprotein particle. For example, the said surface exposed residue is preferably at the N-terminus of the gene product of bacteriophage gene III or gene VIII; and

(ii) The C-terminus of one member of the dTCR polypeptide pair, or the C-terminus of the scTCR polypeptide, or the C-terminus of a short peptide linker attached to the C- terminus of either, is linked by a disulfide bond to a surface exposed cysteine residue of the nucleoprotein particle via an introduced cysteine residue. For example, the said surface exposed residue is again preferably at the N-terminus of the gene product of bacteriophage gene III or gene VIII.

Ml 3 and fl are examples of bacteriophages that express gene III and gene VIII gene products.

Method (i) above is preferred. In the case of an scTCR, nucleic acid encoding the TCR may be fused to nucleic acid encoding the particle forming protein or a surface protein of the replicable particle such as a phage or cell. Alternatively, nucleic acid representing mRNA but without a stop codon, or fused to puromycin RNA may be translated by ribosome such that the TCR remains fused to the ribosome particle. In the case of a dTCR, nucleic acid encoding one chain of the TCR may be fused to nucleic acid encoding the particle forming protein or a cell surface protein of the replicable particle such as a phage or cell, and the second chain of the TCR polypeptide pair may be allowed to associate with the resultant expressed particle displaying the first chain. Proper functional association of the two chains maybe assisted by the presence of cysteines in the constant domain of the two chains which are capable of forming an interchain disulfide bond, as more fully discussed below.

Additional Aspects

A scTCR or dTCR isolated by the method of the invention may be provided in substantially pure form, or as a purified or isolated preparation. For example, it may be provided in a form which is substantially free of other proteins.

Preferred features of each aspect of the invention are as for each of the other aspects mutatis mutandis. The prior art documents mentioned herein are incorporated to the fullest extent permitted by law. Examples

The invention is further described in the following examples, which do not limit the scope of the invention in any way.

Reference is made in the following to the accompanying drawings in which:

Figure 1 details the DNA sequence of the pEX922-ILA vector

Figure 2 provides a plasmid map of the pEX922-ILA vector

Figure 3 a and 3b detail the DNA sequences of the soluble ILA TCR α and β chains as contained in the pEX922-ILA vector. Both of the DNA sequences contain a mutated codon encoding a cysteine residue required to form a non-native disulfide interchain bond in the expressed soluble TCR. Shading indicates the mutated cysteine codons.

Figure 4a and 4b detail the amino acid sequences of the soluble ILA TCR α and β chains as encoded for by the DNA of figures 3 a and 3b respectively. Shading indicates the introduced cysteine residues.

Figure 5 provides a schematic diagram representing the PCR products generated during the production of the ILA TCR-derived library.

Figure 6 provides the Biacore response curve generated for the interaction of this soluble TCR

Example 1— Mutagenesis oflLA-TCR CDRl, CDR2 and CDR3 regions

The CDRl, 2 and 3 α and β variable regions of the ILA TCR were targeted for the introduction of mutations to investigate the possibility of generating high affinity mutants. Overlapping PCR was used to modify the sequence of α and β CDRl, 2 and 3 regions to introduce mutations. For both the TCR alpha and TCR beta chains two PCR products were generated which flanked either side of the CDRl, 2 or 3 regions. The PCR products do not overlap and do not include the CDRl, 2 or 3 sequences to be mutated. These two PCR products are then included in a further PCR reaction in the presence of mutagenic oligos which contain regions of homology with both the CDRl, 2 or 3 flanking PCR products. The presence of the mutagenic oligos in the PCR reaction in conjunction with outside flanking primers allows the stitching of the two PCR fragments to incorporate either CDRl, 2 or 3 mutations. A more detailed description of the mutagenic process is described below.

For the Va CDR3 mutagenesis the following PCR products were generated using the pEX922-ILA vector as template and the primers described below.

PCR1: 40.5μl water, 5μl lOx Pfu buffer, lμl YOL13 primer lOpM/μl, lμl (272) primer lOpM/μl, lμl lOmM dNTPs, 10ng pEX922-ILA and 0.5μl of Pfu poiymerase. The PCR was cycled as follows 30 cycles of 95 degrees 15 sec, 55 degrees 15 sec and 72 degrees for 2 min. PCR product was run on a 1.6% TBE agarose gel, a band of the correct size excised and purified using the Qiagen gel extraction kit.

PCR2:

As above substituting the primers 281 and Yol 13.

PCR3: As above substituting the primers 284 and Yol 13. PCR4:

As above substituting the primers 287 and Yol 13.

PCR5: As above substituting the primers 290 and Yol 13.

PCR7:

As above substituting the primers 278 and 238.

PCR8:

As above substituting the primers 280 and 238.

PCR9:

As above substituting the primers 283 and 238.

PCR10:

As above substituting the primers 286 and 238.

PCR11: As above substituting the primers 289 and 238.

PCR12:

As above substituting the primers 292 and 238.

The pEX922-ILA vector contains DNA encoding the ILA TCR α and β chains in a soluble form containing mutated codons encoding non-native cysteine residues.

Figure 1 details the DNA sequence of the pEX922-ILA vector, and Figure 2 provides a plasmid map of this vector. Figure 3 a and 3b detail the DNA sequences of the soluble ILA TCR α and β chains as contained in the pEX922-ILA vector.

Figure 4a and 4b detail the amino acid sequences of the ILA TCR α and β chains as encoded for by the DNA of figures 3 a and 3b respectively.

For the Va CDR2 mutagenesis the following PCR products were generated

PCR6: As above substituting the primers 293 and Yol 13.

PCR13:

As above substituting the primers 295 and 238.

For the V a CDRl mutagenesis the following PCR products were generated

PCR28:

As above substituting the primers 296 and Yol 13.

PCR29:

As above substituting the primers 238 and 298.

For the Vβ CDR3 mutagenesis the following PCR products were generated using the pEX922-ILA vector as template and the primers described below. The same PCR conditions were used as detailed above.

PCR14:

As above substituting the primers 299 and 247.

PCR15:

As above substituting the primers 299 and 250. PCR16:

As above substituting the primers 299 and 253.

PCR17:

As above substituting the primers 299 and 256.

PCR18:

As above substituting the primers 299 and 259.

PCR19:

As above substituting the primers 299 and 262.

PCR21:

As above substituting the primers 249 and 22.

PCR22:

As above substituting the primers 252 and 22.

PCR23:

As above substituting the primers 255 and 22.

PCR24: As above substituting the primers 258 and 22.

PCR25:

As above substituting the primers 261 and 22.

PCR26:

As above substituting the primers 265 and 22. For the Vβ CDR2 mutagenesis the following PCR products were generated

PCR20: As above substituting the primers 299 and 266.

PCR27:

As above substituting the primers 268 and 22.

For the Vβ CDRl mutagenesis the following PCR products were generated

PCR30:

As above substituting the primers 269 and 22.

PCR31:

As above substituting the primers 271 and 22.

The PCR products described above are subsequently used in the overlap PCRs described below to generate full length CDRl , 2 or 3 mutated V alpha or V beta chains which can then be cut and cloned into the phage display vector pEX922. These PCR reactions were carried out as follows. (See Figure 5 for a schematic diagram representing the PCR products generated during the production of the ILA TCR-derived library)

For the Vβ CDR3 mutagenesis the following overlap PCRs were carried out.

PCR A:

39.5μl water, 5μl lOx Pfu buffer, lμl 299 primer lOpM/μl, lμl 22 primer lOpM/μl, lμl lOmM dNTPs, lOng PCR14, lOng PCR21 (see map above) lOng and 0.33 pM of oligo 248 and 0.5μl of Pfu polymerase. The PCR was cycled as follows 30 cycles of 95 degrees 15 sec, 55 degrees 15 sec and 72 degrees for 2 min. PCR product was run on a 1.6% TBE agarose gel, a band of the correct size excised and purified using the Qiagen gel extraction kit

PCR B:

As above (PCR A) but substitute mutagenic oligo with 248 0.33 pM and the PCR templates with lOng PCR15, lOng PCR22 (see map above) lOng.

PCR C:

As above (PCR A) but substitute mutagenic oligo with 251 0.33 pM and the PCR templates with lOng PCR16, lOng PCR23 (see map above) lOng.

PCR D:

As above (PCR A) but substitute mutagenic oligo with 257 0.33 pM and the PCR templates with lOng PCR17, lOng PCR24 (see map above) lOng.

PCR E:

As above (PCR A) but substitute mutagenic oligo with 260 0.33 pM and the PCR templates with lOng PCR18, lOng PCR25 (see map above) lOng.

PCR F: As above (PCR A) but substitute mutagenic oligo with 263 0.33 pM and the PCR templates with lOng PCR19, lOng PCR26 (see map above) lOng.

PCR G:

As above (PCR A) but substitute mutagenic oligo with 264 0.33 pM and the PCR templates with lOng PCR19, lOng PCR26 (see map above) lOng. For the Vβ CDR2 mutagenesis the following overlap PCRs were carried out.

PCR H:

As above (PCR A) but substitute mutagenic oligo with 267 0.33 pM and the PCR templates with lOng PCR20, lOng PCR27 (see map above) lOng.

For the Vβ CDRl mutagenesis the following overlap PCRs were carried out.

PCR I: As above (PCR A) but substitute mutagenic oligo with 270 0.33 pM and the PCR templates with lOng PCR30, lOng PCR31 (see map above) lOng.

For the Va CDR3 mutagenesis the following overlap PCRs were carried out.

PCRJ:

39.5μl water, 5μl lOx Pfu buffer, lμl 13 primer lOpM/μl, lμl 238 primer lOpM/μl, lμl lOmM dNTPs, lOng PCR1, lOng PCR7 (see map above) lOng and 0.33 pM of oligo 273 and 0.5μl of Pfu polymerase. The PCR was cycled as follows 30 cycles of

95 degrees 15 sec, 55 degrees 15 sec and 72 degrees for 2 min. PCR product was run on a 1.6%) TBE agarose gel, a band of the correct size excised and purified using the

Qiagen gel extraction kit

PCRK: As above (PCR J) but substitute mutagenic oligo with 274 0.33 pM.

PCR L:

As above (PCR J) but substitute mutagenic oligo with 275 0.33 pM.

PCR M: As above (PCR J) but substitute mutagenic oligo with 276 0.33 pM.

PCR N:

As above (PCR J) but substitute mutagenic oligo with 277 0.33 pM.

PCR O:

As above (PCR J) but substitute mutagenic oligo with 279 0.33 pM and the PCR templates with lOng PCR1, lOng PCR8 (see map above) lOng.

PCR P:

As above (PCR A) but substitute mutagenic oligo with 282 0.33 pM and the PCR templates with lOng PCR2, lOng PCR9 (see map above) lOng.

PCR Q: As above (PCR A) but substitute mutagenic oligo with 285 0.33 pM and the PCR templates with lOng PCR3, lOng PCR10 (see map above) lOng.

PCR R:

As above (PCR A) but substitute mutagenic oligo with 288 0.33 pM and the PCR templates with lOng PCR4, lOng PCR11 (see map above) lOng.

PCR S:

As above (PCR A) but substitute mutagenic oligo with 291 0.33 pM and the PCR templates with lOng PCR5, lOng PCR12 (see map above) lOng.

For the Va CDR2 mutagenesis the following overlap PCRs were carried out.

PCR T :

As above (PCR J) but substitute mutagenic oligo with 294 0.33 pM and the PCR templates with lOng PCR6, lOng PCR13 (see map above) lOng.

For the Vα CDRl mutagenesis the following overlap PCRs were carried out. PCR U:

As above (PCR J) but substitute mutagenic oligo with 297 0.33 pM and the PCR templates with lOng PCR28, lOng PCR29 (see map above) lOng.

The sequences corresponding to the oligos described above are shown in the table below.

YOL247 Acagaagtacacagatgtctg

YOL248 CATCTGTGTACTTCTGTNNKlrøKNNKNNK NKGGCACTGAAGCTTTCT

YOL249 GGCACTGAAGCTTTCTTTGGAC

YOL250 Gctggcacagaagtacacag

YOL251 GTACTTCTGTGCCAGC NKlrøiαSINKNNKlSINKGAAGCTTTCTTTGGAC

YOL252 GAAGCTTTCTTTGGACAAGGC

YO 253 Actgctggcacagaagtacac

YO 254 CTTCTGTGCCAGCAGTlrøKIrøKJ KlSrNK-NNKGCTTTCTTTGGACAAG

YOL255 GCTTTCTTTGGACAAGGCAC

YO 256 Gtaactgctggcacagaagtac

YO 257 CTGTGCCAGCAGTTACNNKNNKrøKlrøK NKTTCTTTGGACAAGGCAC

YO 258 TTCTTTGGACAAGGCACCAG

YOL259 Ttggtaactgctggcacagaag

YOL260 GTGCCAGCAGTTACCAAlrøKMSrra NKKrøK NKTTTGGACAAGGCACCAG

YO 261 TTTGGACAAGGCACC AGAC

YO 262 Gctggcacagaagtacacag

YOL263 GTACTTCTGTGCCAGC NK NKl KlrøKlSπsrKlrøKNNKTTCTTTGGACAAGGCAC

YOL264 GTACTTCTGTGCC AGC WKlrøKlrøKNNKlrøKKINKTTCTTTGGAC AAGGC AC

YOL265 TTCTTTGGACAAGGCACCAG

YO 266 Tgagtaatgaatcagcctcag

YOL267 GCTGATTC ATTACTCANWKNNKN KN KNNKACTGACC AAGGAGAAGTC YOL268 ACTGACCAAGGAGAAGTCCCCAATG

YOL269 Catatcctgggcacactgcag

YOL270 GTGTGCCCAGGATATG NKNNKNNKTACATGTCCTGGTATCGAC

YOL271 TACATGTCCTGGTATCGACAAG

YOL272 CACAGCACAGAAATAAACGC

YOL273 TTATTTCTGTGCTGTGKMKNNKrøKOT\rKrø^

YOL274 TTATTTCTGTGCTGTGrøKNNKIrøKrøKlrø

YOL275 TTATTTCTGTGCTGTGNNKNNKrøKNNKrø^

YO 276 TTATTTCTGTGCTGTGlrøKlrøKlrøKlrøK NK NKTACATCTTTGGAACAG

YO 277 TTATTTCTGTGCTGTGNNKNNKNNK NKNNKTACATCTTTGGAACAG

YOL278 TACATCTTTGGAACAGGCAC

YOL279 TTATTTCTGTGCTGTGNNKlrøKrWKNNK NKGGAACCTACAAATACATC

YOL280 GGAACCTACAAATACATCTTTG

YO 281 GTCCACAGCACAGAAATAAAC

YOL.282 TTTCTGTGCTGTGGAC N NNKlrøpαSENKNNKACCTACAAATACATCTTTG

YO 283 ACCTACAAATACATCTTTGGAAC

YOL284 Agagtccacagcacagaaata

YO 285 CTGTGCTGTGGACTCTN KlrøKlrøKNNKlSINKTACAAATACATCTTTGGAAC

YOL286 TACAAATACATCTTTGGAAC

YO 287 Agcagagtccacagcacag

YOL288 TGCTGTGGACTCTGCTNWKNNKlrøKKENKNNB-AAATACAT

YOL289 AAATACATCTTTGGAACAGGC

YOL290 Ggtagcagagtccacagcac

YO 291 TGTGGACTCTGCTACC NKlrøl røK NKNNKTACATCTTTGGAACAGGC

YOL292 TACATCTTTGGAACAGGCAC

YOL293 Gtaaaacaggttgatgagctg

YOL294 CATCAACCTGTTTTACMSTKNNKNNiaSINKACAAAACAGAATGGAAG

YOL295 ACAAAACAGAATGGAAGATTAAG

YOL296 Gtcagaaaaattgcaccgcag YOL297 GTGCAATTTTTCTGAC NK NKNNKN KTTGCAGTGGTTTCATC YOL298 TTGCAGTGGTTTCATCAAAAC YOL299 Cacagacaaaactgtgctagac

Wherein: N = A, T, G or C K = G or T

For the construction of the ILA CDRl, 2 and 3 alpha mutagenic phage display library: pooled α-chain PCR fragments (J,K,L,M,N,O,P,Q,R,S,T and U) were digested with Nco I and Notl and re-purified using a Qiagen kit and the recipient vector was prepared by digesting pEX922 with Nco I and Notl followed by gel purification using a Qiagen kit.

For the construction of the ILA CDRl, 2 and 3 beta mutagenic phage display library: pooled β-chain PCR fragments (A,B,C,D,E,F,G,H and I) were digested with Nco I and Notl and re-purified using a Qiagen kit and the recipient vector was prepared by digesting pEX922 with Nco I and Notl followed by gel purification using a Qiagen kit.

V alpha and V beta libraries were ligated and transformed separately. Purified inserts and vectors at 3 : 1 molar ratio were mixed with T4 ligase buffer, T4 ligase and nuclease-free water. The ligations were carried out at 16°C water bath overnight. For each mutation-library, a total of 0.5 to 2μg purified and desalted ligated products were electroporated into E. coli TGI at ratio of 0.2μg DNA per 40 μl of electroporation-competent cells (Stratagen) following the protocols provided by the manufacturer. After electroporation, the cells were re-suspended immediately with 960μl of SOC medium at 37°C and plated on a 244mm x244mm tissue culture plate containing YTE (15g Bacto-Agar, 8g NaCl, lOg Tryptone, 5g Yeast Extract in 1 litre) supplemented with lOOμg/ml ampicillin and 2% glucose. The plate was incubated at 30°C over night. The cells were then scraped from the plates with 5 ml of DYT (16g Trytone, lOg Yeast extract and 5g NaCl in 1 litre, autoclaved at 125°C for 15 minutes) supplemented with 15%o glycerol.

In order to make phage particles displaying the ILA TCR library, 500 ml of DYTag

(DYT containing 100 μg/ml of ampicillin and 2% glucose) was inoculated with 500 to 1000 μl of the library stocks. The culture was grown until OD(600nm) reached 0.5. 100 ml of the culture was infected with helper phage (Ml 3 K07 (Invitrogen), or HYPER PHAGE (Progen Biotechnik, GmbH 69123 Heidelberg), and incubated at 37°C water bath for 30 minutes. The medium was replaced with 100 ml of DYTak

(DYT containing 100 μg/ml ampicillin and 25 μg/ml of kanamycin). The culture was then incubated with shaking at 300 rpm and 25°C for 20 to 36 hours.

Example 2 - Isolation of TCRs that bind to (Telomerase)RL VDDFLL V -HLA-A *0201, (Survivin)ELTLGEFLKL -HLA-A *0201, or (Survivin)LTLGEFLKL -HLA-

A*0201 complex from a TCR phage display library derived from the ILA TCR

The isolation of TCRs that bind to (Telomerase)RLVDDFLLV -HLA-A*0201, (Survivin)ELTLGEFLKL -HLA-A*0201, or (Survivin)LTLGEFLKL -HLA- A*0201complex from the ILA Telomerase phage display library described above was carried out as follows. The initial panning was carried utilising the selection of phage particles (prepared as described above) displaying mutant TCRs derived from the ILA TCR.

Streptavidin-coated paramagnetic beads (Roche) were pre-washed according to manufacturer's protocols. Phage particles, displaying mutated ILA TCR at a concentration of 10¹² to 10¹³ cfu in 3% powdered milk- PBS, were pre-mixed with either biotinylated (Telomerase)RLVDDFLLV -HLA-A*0201, (Survivin)ELTLGEFLKL -HLA-A*0201, or (Survivin)LTLGEFLKL -HLA- A*0201complex at concentrations of 500nM for all three rounds of selection carried out. The mixture of ILA TCR-displaying phage particles and either (Telomerase)RLVDDFLLV -HLA-A*0201, (Survivin)ELTLGEFLKL -HLA-A*0201, or (Survivin)LTLGEFLKL -HLA-A* 0201 complex was incubated for one hour at room temperature with gentle rotation, and the TCR-displaying phage particles bound to (Telomerase)RLVDDFLLV -HLA-A*0201 , (Survivin)ELTLGEFLKL -HLA-

A*0201, or (Survivin)LTLGEFLKL -HLA-A*0201 complex. Phage biotynlated HLA complexes were rescued for 5 minutes with 100 ml of streptavidin-coated (Roche) magnetic beads which had been blocked with de-biotynalyted 3%> Milk powder PBS. After capture of the phage particles, the beads were washed a total of six times (three times in PBS-0.1%> tween20 and three times in PBS) using a Dynal magnetic particle concentrator. After final wash, the beads were re-suspended in lOOμl of freshly prepared PBS and 50μl of the re-suspended beads was used to infect 10 ml of E.coli TGI at OD(600nm)=0.5 freshly prepared for the amplification of the selected phage particles according to established methods.

After the third round of selection, 100 colonies were picked from either the (Telomerase)RLVDDFLLV -HLA-A*0201, (Survivin)ELTLGEFLKL -HLA-A*0201, or (Survivin)LTLGEFLKL -HLA-A*0201 complex plates and used to inoculate 100 μl of 2TYAG in a 96-well microtiter plate. The culture was incubated at 30°C with shaking overnight. 100 μl of 2TYAG was then sub-inoculated with 2 to 5 μl of the overnight cultures, and incubated at 30°C with shaking for 2 to 3 hours or until the culture became cloudy. To infect the cells with helper phage, the culture was infected with 100 μl of 2TYAG containing 5 x 10⁹pfu helper phages, and incubated at 37°C for 60 minutes. 5 μl of the infected culture was added to 200 μl of 2TYAK ("TYAG + 100 μg/ml Ampicillin and 50 μg/ml Kanomycin) The plates were incubated at 25°C for 20 to 36 hours with shaking at 300 rpm. The cells were precipitated by centrifugation at 3000g for 10 minutes at 4°C. Supernatants were used to screen for high affinity TCR mutants by phage ELIS A.

Phage clones which bound to either (Telomerase)RLVDDFLLV -HLA-A*0201, (Survivin)ELTLGEFLKL -HLA-A*0201, or (Survivin)LTLGEFLKL -HLA- A* 0201 complex were found during the ELIS A screening as determined by their strong ELISA signals (O.D.⁶⁰⁰ 0.3-1) relative to control wells (O.D.⁶⁰⁰ 0.05). The wild- type ILA TCR from which the library was derived was not capable of binding to any of these pMHC complexes to an extent that was detectable by the above ELISA assay.

This therefore demonstrates that TCRs with specificities differing from those of the parental ILA TCR used to construct the mutated phage display library had been isolated from said library. Such an approach should also allow one to isolate TCRs from a phage display library derived from a TCR with a particular pMHC specificity that are capable of binding to peptides presented by a different MHC. The general approach involves phagemid DNA encoding the identified TCR being isolated from the relevant E.coli cells using a Mini-Prep kit (Quiagen, UK). PCR amplification can be carried out using the phagemid DNA as template and a set of primers designed to amplify the soluble TCR α and β chain DNA sequences encoded by the phagemid. The full range of primers required can be deduced by reference to the TCR α and TCR β Vand C sequences. The PCR product is then digested with appropriate restriction enzymes and cloned into an E. coli expression vector with corresponding insertion sites. The amplified TCR α and β chain DNA sequences (which include, as described above, codons encoding the cysteines required to form the introduced constant domain interchain disulfide bond) are then used to produce a soluble TCR as described in WO 03/020763. Briefly, the two chains are expressed as inclusion bodies in separate E.coli cultures. The inclusion bodies are then isolated, de-natured and re-folded together in vitro.

If it is required to characterise the binding characteristics of the soluble TCR, this may be done using the Biacore system.

Example 3 — BIAcore surface plasmon resonance characterisation ofsTCR binding to specific pMHC A surface plasmon resonance biosensor (BIAcore 3000™ ) was used to analyse the binding of an sTCR to its peptide-MHC ligand. This was facilitated by producing single pMHC complexes (described below) which were immobilised to a streptavidin- coated binding surface in a semi-oriented fashion, allowing efficient testing of the binding of a soluble T-cell receptor to up to four different pMHC (immobilised on separate flow cells) simultaneously. Manual injection of HLA complex allows the precise level of immobilised class I molecules to be manipulated easily.

Biotinylated class I pMHC molecules were refolded in vitro from bacterially- expressed inclusion bodies containing the constituent subunit proteins and synthetic peptide, followed by purification and in vitro enzymatic biotinylation (O'Callaghan et al. (1999) Anal. Biochem. 266: 9-15). MHC-heavy chain was expressed with a C- terminal biotinylation tag which replaces the transmembrane and cytoplasmic domains of the protein in an appropriate construct. Inclusion body expression levels of ~75 mg/litre bacterial culture were obtained. The MHC light-chain or β2-microglobulin was also expressed as inclusion bodies in E.coli from an appropriate construct, at a level of ~500 mg/litre bacterial culture.

E. coli cells were lysed and inclusion bodies are purified to approximately 80% purity. Protein from inclusion bodies was denatured in 6 M guanidine-HCl, 50 mM Tris pH

8.1, 100 mM NaCl, 10 mM DTT, 10 mM ΕDTA, and was refolded at a concentration of 30 mg/litre heavy chain, 30 mg/litre β2m into 0.4 M L-Arginine-HCl, 100 mM Tris pH 8.1, 3.7 mM cystamine, mM cysteamine, 4 mg/ml of the peptide required to be loaded by the MHC, by addition of a single pulse of denatured protein into refold buffer at < 5°C. Refolding was allowed to reach completion at 4°C for at least 1 hour.

Buffer was exchanged by dialysis in 10 volumes of 10 mM Tris pH 8.1. Two changes of buffer were necessary to reduce the ionic strength of the solution sufficiently. The protein solution was then filtered through a 1.5μm cellulose acetate filter and loaded onto a POROS 50HQ anion exchange column (8 ml bed volume). Protein was eluted with a linear 0-500 mM NaCl gradient. HLA- A2 -peptide complex eluted at approximately 250 mM NaCl, and peak fractions were collected, a cocktail of protease inhibitors (Calbiochem) was added and the fractions were chilled on ice.

Biotinylation tagged pMHC molecules were buffer exchanged into 10 mM Tris pH 8.1, 5 mM NaCl using a Pharmacia fast desalting column equilibrated in the same buffer. Immediately upon elution, the protein-containing fractions were chilled on ice and protease inhibitor cocktail (Calbiochem) was added. Biotinylation reagents were then added: 1 mM biotin, 5 mM ATP (buffered to pH 8), 7.5 mM MgC12, and 5 μg/ml BirA enzyme (purified according to O'Callaghan et al. (1999) Anal. Biochem. 266: 9- 15). The mixture was then allowed to incubate at room temperature overnight.

Biotinylated pMHC molecules were purified using gel filtration chromatography. A Pharmacia Superdex 75 HR 10/30 column was pre-equilibrated with filtered PBS and 1 ml of the biotinylation reaction mixture was loaded and the column was developed with PBS at 0.5 ml/min. Biotinylated pMHC molecules eluted as a single peak at approximately 15 ml. Fractions containing protein were pooled, chilled on ice, and protease inhibitor cocktail was added. Protein concentration was determined using a Coomassie-binding assay (PerBio) and aliquots of biotinylated pMHC molecules were stored frozen at -20°C. Streptavidin was immobilised by standard amine coupling methods.

Such immobilised complexes are capable of binding both T-cell receptors and the coreceptor CD8αα, both of which may be injected in the soluble phase. Specific binding of TCR is obtained even at low concentrations (at least 40μg/ml), implying the TCR is relatively stable. The pMHC binding properties of sTCR are observed to be qualitatively and quantitatively similar if sTCR is used either in the soluble or immobilised phase. This is an important control for partial activity of soluble species and also suggests that biotinylated pMHC complexes are biologically as active as non- biotinylated complexes. The interactions between the mutant ILA TCRs containing a novel inter-chain bond and its ligand/ MHC complex or an irrelevant HLA-peptide combination, the production of which is described above, were analysed on a BIAcore 3000™ surface plasmon resonance (SPR) biosensor. SPR measures changes in refractive index expressed in response units (RU) near a sensor surface within a small flow cell, a principle that can be used to detect receptor ligand interactions and to analyse their affinity and kinetic parameters. The probe flow cells were prepared by immobilising the individual HLA-peptide complexes in separate flow cells via binding between the biotin cross linked onto β2m and streptavidin which have been chemically cross linked to the activated surface of the flow cells. The assay was then performed by passing sTCR over the surfaces of the different flow cells at a constant flow rate, measuring the SPR response in doing so.

To measure Equilibrium binding constant

Serial dilutions of the mutant ILA sTCRs were prepared and injected at constant flow rate of 5 μl min-1 over two different flow cells; one coated with -1000 RU of the specific ILAKFLHWL-HLA-A*0201 complex, the second coated with -1000 RU of non-specific HLA-A2 -peptide complex. Response was normalised for each concentration using the measurement from the control cell. Normalised data response was plotted versus concentration of TCR sample and fitted to a hyperbola in order to calculate the equilibrium binding constant, K_D- (Price & Dwek, Principles and Problems in Physical Chemistry for Biochemists (2^nd Edition) 1979, Clarendon Press,

Oxford).

To measure Kinetic Parameters

For high affinity TCRs K_D was determined by experimentally measuring the dissociation rate constant, kd, and the association rate constant, ka. The equilibrium constant K_D was calculated as kd/ka.

TCR was injected over two different cells one coated with -300 RU of specific ILAKFLHWL-HLA-A*0201 complex, the second coated with -300 RU of non- specific HLA-A2 -peptide complex. Flow rate was set at 50 μl/min. Typically 250 μl of TCR at -3 μM concentration was injected. Buffer was then flowed over until the response had returned to baseline. Kinetic parameters were calculated using Biaevaluation software. The dissociation phase was also fitted to a single exponential decay equation enabling calculation of half-life.

Example 4 -Isolation of an HLA-A24- VYGFVRACL binding TCR from an A6 TCR- derived phage display library.

A second phage displayed TCR library was created using the procedures of Examples

1 and 2 of the inventor's co-pending application WO 2004/044004.

The A6 TCR from which this library was derived is specific for HLA-A2- LLFGYPVYV. This library was panned against HLA-A24 loaded with the Telomerase-derived VYGFVRACL peptide busing the phage ELISA method described above.

The DNA encoding the displayed TCR was then isolated from the binding phage particles, and used to produce a soluble dTCR was described above.

The ability of this soluble TCR, derived from the isolated phage particle, to bind to HLA-A24- VYGFVRACL was verified using the Biacore method of Example 3.

Results

The soluble TCR containing an introduced disulfide interchain bond was shown to bind to HLA-A24- VYGFVRACL with an affinity (Kd) of 1.6 μM. Figure 6 provides the Biacore response curve generated for the interaction of this soluble TCR.

Claims

Claims

1. A method for the identification of a polypeptide which binds to a given peptide-MHC complex or CD 1 -antigen complex, said method comprising

contacting (a) members of a diverse library of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic T cell receptor ("TCR") variable domain sequences derived from a TCR which binds to a first peptide-MHC complex or CD 1 -antigen complex wherein diversity resides at least in the variable domains of the said polypeptides and (b) a given peptide-MHC complex or CD 1 -antigen complex different from the first peptide-MHC complex or CD 1 -antigen complex,

detecting binding between the library members and said given peptide-MHC complex or CD 1 -antigen complex,

isolating a library member detected as binding to the given peptide-MHC complex or CD 1 -antigen complex, and optionally multiplying the isolated library member in an amplification process,

the polypeptide displayed on the surface of the library member being taken to be the desired polypeptide.

2. A method as claimed in claim 1 wherein the nucleoproteins display polypeptides comprising diverse synthetic TCR variable domain sequences derived from a plurality of TCRs which bind to peptide-MHC complexes or CD 1 -antigen complexes different from the given peptide MHC complex or CD 1 -antigen complex.

3. A method as claimed in claim 1 or claim 2 wherein the said binding is determined by ELISA.

4. A method as claimed in any preceding claim wherein the nucleoproteins particles are filamentous phage particles or ribosome particles, and the displayed polypeptides are single chain TCRs ("scTCRs").

5. A method as claimed in any of claims 1 to 3 wherein the library of nucleoproteins comprises (a) a first set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR α variable domain sequences, and /or (b) a second set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR β variable domain sequences, and/or (c) a third set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR α or β variable domain sequences, which polypeptides are associated in pairs with polypeptides also comprising diverse synthetic TCR α or β variable domain sequences.

6. A method as claimed in claims 1 to 3 wherein the nucleoproteins are filamentous phage particles, and the library includes or consists of members displaying polypeptides which are αβ dimeric TCRs ("dTCRs").

7. A method as claimed in claim 6 wherein the library includes members displaying TCR α-chains and/or TCR β-chains and/or homodimeric αα chains and/or homodimeric ββ chains.

8. A method as claimed in claim 6 or 7 wherein the phage-displayed αβ dTCRs comprise a first polypeptide wherein a sequence corresponding to a TCR α chain variable domain sequence is fused to the N terminus of a sequence corresponding to a TCR α chain constant domain extracellular sequence, and a second polypeptide wherein a sequence corresponding to a TCR βchain variable domain sequence fused to the N terminus a sequence corresponding to a TCR β chain constant domain extracellular sequence, the first and second polypeptides being linked by a disulfide bond which has no equivalent in native αβ T cell receptors, and one of said first or second polypeptides being linked by a peptide bond at its C- terminus to a surface exposed amino acid residue of the phage particle.

9. A method as claimed in claim 6 or 7 wherein the phage-displayed αβ dTCRs comprise a first polypeptide wherein a sequence corresponding to a TCR α chain variable domain sequence is fused to the N terminus of a sequence corresponding to a TCR α chain constant domain extracellular sequence, and a second polypeptide wherein a sequence corresponding to a TCR β chain variable domain sequence is fused to the N terminus a sequence corresponding to a TCR β chain constant domain extracellular sequence, the first and second polypeptides being linked by a disulfide bond between cysteine residues substituted for Thr 48 of exon 1 of TRAC*01 and Ser 57 of exon 1 of TRBC1*01 or TRBC2*01 or the non-human equivalent thereof, one of said first or second polypeptides being linked by a peptide bond at its C- terminus to a surface exposed amino acid residue of the phage particle.

10. A method as claimed in any of claims 6 to 9 wherein, for the purpose of identifying library members which bind to a given pMHC or CD 1 -antigen and isolating those which bind,

(i) several members of the library are contacted in parallel with the given pMHC or CD 1 -antigen and members which bind to the pMHC or CD 1 -antigen are identified, (ii) one or more members which bind to the given pMHC or CD 1 -antigen assessed in step (i) are selected, and the variable domain sequences of the displayed TCRs is/are determined, (iii) soluble form TCRs incorporating the thus-determined variable domain sequences, are created, (iv) the affinities and/or the off-rates for the given pMHC of these TCRs are determined, and (v) one or more TCRs having the desired affinity and/or off-rate determined in step (iv) are selected.

11. A method as claimed in claim 10 wherein the said affinities and/or the off-rates are determined by Surface Plasmon Resonance.

12. A method as claimed in any of the preceding claims wherein the displayed polypeptides are derived from the given TCR(s) by mutation of at least one of its/their complementarity determining regions of the variable domains.

13. A method as claimed in any of claims 1 to 11 wherein the displayed polypeptides are derived from the given TCR(s) by mutation of at least one of its/their CDR2 or CDR3 complimentarity determining regions of the variable domains.

14. A method as claimed in any preceding claim wherein the given TCR(s) is/are naturally occurring.

15. A library of nucleoproteins comprising (a) a first set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR α variable domain sequences, and/or (b) a second set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR β variable domain sequences, and/or (c) a third set of nucleoproteins displaying on their surfaces polypeptides comprising diverse synthetic TCR α or β variable domain sequences, which polypeptides are associated in pairs with polypeptides also comprising diverse synthetic TCR α or β variable domain sequences, provided that the said library includes at least one of the sets (a) and (b).

16. A library as claimed in claim 15 wherein diversity resides in at least the CDR2 or CDR3 sequences of the variable domain sequences.

17. A library as claimed in claim 15 or claim 16 wherein the nucleoproteins are phage or ribosome particles.

18. A diverse library of nucleoproteins as claimed in any of claims 15 to 17 wherein the synthetic variable domain sequences of the displayed polypeptides are located N-terminal to part of a TCR chain which includes all or part of the constant domain sequence thereof, except the transmembrane domain thereof.

19. A diverse library as claimed in claim 18 wherein in the third set of nucleoproteins the said association in pairs is maintained at least in part by a disulfide bond which has no equivalent in native αβ T cell receptors between introduced cysteine residues in the TCR constant domain sequences of the paired polypeptides.

20. A diverse library as claimed in claim 19 comprising phage particles or ribosomes on which the displayed polypeptide, or one member of the displayed associated polypeptide pair, is linked by a peptide bond at its C-terminus to a surface exposed amino acid residue of the phage particle or ribosome.

21 A diverse library of phage particles as claimed in claim 20 wherein the first and second polypeptides are linked by a disulfide bond between cysteine residues substituted for Thr 48 of exon 1 of TRAC*01 and Ser 57 of exon 1 of TRBCl *01 or TRBC2*01 or the non-human equivalent thereof.

22. A library as claimed in any of claims 15 to 21 wherein the synthetic α and β variable domain sequences are derived by mutation of at least two TCRs.