GB2428293A - Phage display libraries - Google Patents

Abstract

A method for the preparation of a phage display library for the selection of high affinity binding phage (shaved phage) comprising: providing a library of phage expressing a proteinaceous ligand on a plurality of phage surfaces wherein the ligand is attached to a phage coat protein and wherein a cleavable site is located within the expressed ligand-coat protein; treating the library with a cleaving agent such that a proportion of the phage are monovalent for the proteinaceous ligand. Also claimed is a method of selection of phage from such a library, a phage particle comprising a ligand attached to the phage gene III coat protein wherein the particle has a cleavable site within said surface-expressed ligand-coat protein and a shaved phage library.

Description

SHAVED PHAGE.

The present invention relates to a novel method for the selection from a phage surface display library of proteinaceous ligands, those members which bind specifically to one or more binding partners. In particular the present invention relates to a method for the selection from a phage display library one or more high affinity binding phages. Novel phage display libraries for use according to the method of the invention are also described.

INTRODUCTION.

The advent of phage-display technology (Sniith (1985) Science, 228: 1315; Scott and Smith (1990) Science, 249: 386; McCafferty ci at. (1990) Nature, 348: 552) has enabled the in vitro selection of human antibodies against a wide range of target antigens from "single pot" libraries. These phageantibody libraries can be grouped into two categories: natural libraries which use rearranged V genes harvested from human B cells (Marks et at. (1991) 1. Mo!. Biol., 222: 581; Vaughan c/ at. (1996) Nature Biotech., 14: 309) or synthetic libraries whereby germline V gene segments are rearranged' in vitro (Hoogenbooni & Winter (1992)1. Mo!. Biot., 227: 381; Nissim ci a!. (1994) EMBOJ., 13: 692; Griffiths eta!. (1994) EMBO.J., 13: 3245; De Kruifet at. (1995) J. Mo!. Biol., 248: 97) or where synthetic CDRs are incorporated into a single rearranged V gene (Barbas et a!. (1992) Proc. Nail. Acad. Sci. USA, 89: 4457).

Although synthetic libraries help to overcome the inherent biases of the natural repertoire which can limit the effective size of phage libraries constructed from rearranged V genes, they require the use of long degenerate PCR primers which frequently introduce base-pair deletions into the assembled V genes. This high degree of randornisation may also lead to the creation of antibodies which are unable to fold correctly and are also therefore non-functional. Furthermore, antibodies selected from these libraries may be poorly expressed and, in many cases, will contain framework mutations that may affect the antibodies immunogenicity when used in human therapy.

A particular advantage phage-hased display systems is that, because they are biological systems, selected library members can he amplified simply by growing the phage containing the selected library member in bacterial cells. Furthermore, since the nucleotide sequence that encodes the polypeptide library member is contained on a S phage or phagemid vector, sequencing, expression and subsequent genetic manipulation is relatively straightforward.

However, it has been found that when library members are selected on the basis of affinity binding to one or more binding partners, problems arise in the selection of those high affinity binding molecules. Several techniques have been developed in an attempt to overcome this problem.

For example, Scott et a!, ((1997) Chem Rev, 391-4 10) considered that in the case of affinity based selections using phage display technology, high stringency of selections 1 5 is favoured by low densities of the target receptor and monovalent display of the foreign peptide (page 397, final paragraph, Scott et al).

In other studies performed by McConnell et al, (Gene 151, 115-118(1994)), in which peptides were constrained in phage display libraries as a tool for finding niirnotopes, it was suggested that the multivalent nature of phage display systems, (that is resulting in the display or several copies of a fusion protein per phage) may result in problems during the selection process in discriniinating between phage which bind with high versus moderate affinity. It was further suggested that two strategies have been used to overconie this problem. Firstly, conditions have been described which use polyvalent display in which phage that bind with high affinity can be selected by increasing the stringency of the binding conditions (Barrett et al, Anal. Biochem 204, 357-364 (1992)).

In an alternative strategy (Bass et a!, Proteins 8, 309-314 (1990)), a monovalent phagemid-based display system has been used in which phage particles are assembled from a controlled mixture of wild type gill fusion protein, provided by a helper phage, and a gill fusion protein under transcriptional control of the lac promoter, all housed within a phagemid vector. Resultant particles from infection result contain both fusion and wild type gill, thus reducing the valency of fusion protein expression.

In a further strategy, (Smith et al, Gene, 128, 1-2 (1992); Corey et al, Gene 128, 129- 134 (1993) have developed a 33' type vector where gill proteins are introduced into the multiple cloning site of M I 3rnp 18, thus circumventing the need for helper phage.

By controlling expression levels of the gill fusion protein, phage can be generated that display one or less copies of the gill fusion protein, thus allowing the selection of phage based on affinity rather than avidity. 1 0

Further in studies performed by O'Connell et al, J. Mo! Biol (2002), 321, 49-56, it was suggested that in order to iniprove the affinities of those scFv molecules selected from phage display libraries, then the library could be converted to monovalent display after the first round of selection' (abstract). There is no indication in this article, however of how this may he achieved.

Although the prior art methods described in brief above, in theory facilitate the discrimination between high and low affinity binding interactions when selecting binding partners using phage display technology, the use of these methods is associated with significant problems. In particular, the creation of monovalent phage requires complex cloning procedures and careful control of expression of the protein using transcriptional promoters. Further, by controlling the valency of fusion protein expression at the nucleic acid, then it is not possible to alter the valency of phage surface ligand expression throughout the selection procedure, other than by recloning.

This recloning step, of course, is both laborious, time consuming and leads to errors in results obtained.

Thus there remains in the art, the need for a method for the selection from phage display libraries, those ligands which are capable of interacting with their binding partner with high affinity and which method permits the fine control of the valency of surface displayed ligand to be varied through the selection process without the need for recloni ng.

SUMMARY OF THE INVENTION.

The present inventors have found that there exist in the art problems in the selection of high affinity antigen binding antibody molecules using phage display technology. In particular, the present inventors have found that when performing selections using naïve libraries of single domain antibody molecules (dAbs), typically those dAbs selected exhibit low binding affinities. Consequently selected molecules are of little therapeutic value and affinity niaturation is required to convert such molecules into high affinity ligand binding molecules.

The present inventors consider that the problem of selecting high affinity binding molecules using phage display technology, employed before the filing date of the present invention may result because these prior art methods of selection are unable to discriminate sufficiently between high affinity and lower affinity binders present within displayed libraries. The inventors consider further that the inability of ligands to effectively select for high affinity binding partners using phage technology may be a consequence of avidity effects contributing to ligandbinding partner interactions.

In order to investigate the effects of avidity in the selection of high affinity ligand binding molecules, the inventors performed studies in which the theoretical ratio of various dAbs in model phage display selections using a model cytokine was calculated. These values were compared to the actual values obtained experimentally.

The inventors surprisingly found that the ratios obtained experimentally were considerably lower than the theoretical values calculated. Thus it is apparent that avidity effects appear to make a significant contribution to selections performed using phage display technology. These experiments are described in detail in Example 1.

The inventors consider that these avidity effects are likely at least in part to he a consequence of the multi-valency of ligand expressing phage comprising the phage libraries used for selection. That is, each phage expresses more than one copy of the coat protein with attached ligand for selection. Due to the close proximity of these multi-valently displayed ligands present on the surface of the phage, the inventors consider that avidity effects are able to significantly influence ligand-hinding molecule interactions. Consequently, those phage selected are so selected oii the basis of a combination of both affinity and avidity binding interactions.

The present inventors sought to overcome the problems of avidity effecting the selection of high affinity binding molecules from phage display libraries. Further, the inventors sought to develop a method that allowed the valency of the phage to be varied during the selection process, without the need for recloning.

They realised that such an aim could be achieved by the insertion of a cleavable site between the coat protein and surface expressed ligand such that phage which are expressing ligands multi-valently may be treated with one or more cleaving agents such that most phage comprise no ligands on the phage surface and a small proportion of the phage comprise only one ligand on the phage surface. That is, the phage are treated with cleaving agent so that a proportion of the phage in the library become monovalent with respect to surface-expressed ligand.

The inventors have named the process of treating phage which comprise a cleavage site between the gene III coat protein and surface expressed ligand with a cleavage agent, such that a small portion of the resulting phage are monovalent with respect to surface ligand, shaving phage'. Accordingly, the resultant phage are referred to by the present inveiitors as shaved phage'.

Thus in a first aspect the present invention provides a method for the preparation of a phage display library, which library facilitates the selection of high affinity binding phage (a shaved phage library), which method comprises the step of: (a) Providiiig a library of phage expressing one or more copies of a proteinaceous ligand of interest on a plurality of phage surfaces; wherein the ligand is attached to phage coat protein and wherein a cleavable site is located within the expressed ligand- coat protein, (b) Treating the phage library according to step (a) with a cleaving agent such that a small proportion of the treated phage are monovalent with respect to the presence of surface expressed proteinaceous ligand, and the majority of the phage comprise no surface expressed proteinaceous hgand.

In a further aspect the present invention provides a method for the selection of phage from a phage display library which method comprises the step of: (a) Providing a library of phage expressing one or more copies of a proteinaceous ligand of interest on a plurality of phage surfaces; wherein the ligand is attached to phage coat protein and wherein a cleavable site is located within the expressed ligand- coat protein, (h) Treating the phage library according to step (a) with a cleaving agent such that a small proportion of the treated phage are monovalent with respect to the presence of surface expressed proteinaceous ligand, and the majority of the phage comprise no surface expressed proteinaceous ligand (shaving the phage) and (c) Selecting from the library treated according to step (a) and (b) above those proteinaceous ligand expressing phage which are capable of specifically binding to one or more binding partners.

As referred to herein, the term phage' includes within its scope phagernid'. Suitable phage including phagernid for use according to the methods of the invention will be familiar to those skilled in the art. Preferably the phage for use according to the niethod of the invention is not a phagemid. Examples of Phage include but are not limited to M13, fI, fd, IfI, Ike, Xi, Pfl, Pf3, MI3KE and M13K07. Examples of phagemid include but are not limited to pComb3H, pCANTAB5E, pMID2I.

Further according to the methods of the invention, preferably the phage coat protein is that coat-protein encoded by gene Ill, that is preferably a phage suitable for use according to the methods of the invention is a phage and not a phagemid wherein that phage encodes a gene Ill (gill) coat protein as described herein.

According to the present invention, the term proteinaceous' ligand refers to a ligand which is a protein or fragment thereof. In a preferred embodiment of the invention the proteinaceous hgand comprises a single polypeptide chain or fragment thereof.

According to the method for the selection of phage from a shaved phage library, those proteinaccous ligands selected are capable of specifically binding to one or more binding partners.

It is an essential feature of the method of the present invention that step (b), that is the step of treating the phage library with a cleaving agent such that a small proportion of those phage are monovalent with respect to the presence of surface expressed proteinaceous ligand (shaving the phage), is performed prior to the selection step (c) above.

1 5 According to the present invention, the term a cleavable site'/cleavage site' refers to one or more amino acid/s within the surface expressed proteinaceous ligand-phage coat protein which is recognised by a cleavage agent and is capable of being cleaved by that agent. According to the invention described herein, upon cleavage of the coat protein-ligand with a cleavage agent as herein described the proteinaceous ligand is released from the coat protein. The cleavage/cleavable site may be located within the amino acid sequence comprising the proteinaceous ligand. That is, it may he a naturally occurring component of the proteinaceous ligand. Alternatively, the cleavage site niay be engineered into the proteinaceous ligand, for example by mutagenesis of the nucleic acid encoding the ligand. Further the cleavage site may he present within a linker peptide' which comprises a number of amino acids which together form a nonnaturally occurring component of the ligand coat protein. Preferably the cleavable site/cleavage site is located in between the ligand and the coat protein to which the ligand is attached. More preferably, the cleavage site is present within a linker peptide which is located in between the proteinaceous ligand and the phage coat protein.

According to the methods of the invention described herein, the cleavage/cleavable site is located within an expressed proteinaceous ligand- coat protein. Those skilled in the art will appreciate that the location of the cleavage/cleavable site must he such that the site is readily accessible to one or more cleaving agents. Accordingly, treatment of the phage library with one or more cleaving agents under suitable conditions results in shaving of the phage library' as herein defined.

According to the methods of the invention described herein, advantageously, the cleavable/cleavage site is located between the coat protein and the expressed proteinaceous ligand. Moreover, preferably the cleavage site is a protease cleavage site. Proteases suitable for use according to the methods of the invention are described i ü in the detailed description of the invention. Moreover, suitable protease cleavage sites are described therein.

The cleavage site can be an enzymatic cleavage site such as that cleaved by a protease such as thromhin, Factor Xa, enteropeptidase such as serine protease eiiterokinase (including those described in W00198366), or by trypsin for example, or it can be a chemical cleavage site such as CNBr which cleaves at a methionine residue.

As eluded to above, the cleavage/cleavable site must be suitably positioned such that upon treatment of the phage library with one or more cleaving agents under suitable conditions, a proportion of the resultant phage possess only one expressed proteinaceous ligand of interest attached to coat protein on their cell surface. That is the resultant shaved phage are monovalent in relation to the expression of proteinaceous ligand on their surface. The present inventors have found that upon treatment with a cleaving/cleavage agent the majority of those treated phage (shaved phage) are zero-valent with respect to the expression of surface ligand. That is the majority of those phage which have been treated with cleaving agent comprise no proteinaceous ligand attached to the surface of the phage.

According to the methods of the invention, the phage cleavage conditions are adjusted so that a small proportion of those treated phage, that is 0. 0 1% or less, 0.02% or less, 0.03% or less, 0.04% or less, 0.05% or less, 0.06% or less, 0.07% or less, 0.08% or less, 0.09% or less 0.1% or less, 0.2% or less, 0.3% or less, 0. 4% or less, 0.5% or less, 0.6% or less, 0.7% or less,0.8% or less, 0.9% or less, 1% or less of those phage according to the present invention which have been treated with one or more cleavage/cleaving agent/s comprise a monovalent proteinaceous surface ligand (ligand -coat protein peptide) as herein defined.

Preferred conditions for the treatment of phage with one or more cleaving agents/cleavage agents is provided in the detailed description of the invention.

In a further aspect, the present invention provides a phage particle which comprises monovalently on its surface, a proteinaceous ligand of interest attached to the phage coat protein, wherein the phage particle comprises a cleavable site within said surface expressed ligand- coat protein.

In a further aspect still, the present invention provides a shaved phage library in which 1 5 each member of the library is capable of expressing on their surface one or more copies of one or more proteinaceous ligands of interest, preferably one ligand of interest attached to the phage coat protein, wherein each phage particle comprises a cleavable site within each surface expressed ligand- coat protein and wherein 0.01% or more of those library members comprise a monovalent surface ligand- coat protein.

According to the above aspect of the invention, preferably the majority of those phagc are zero-valent with respect to the phage surface expression of proteinaceous ligand- coat protein. As herein defined the term the majority' means more than 40%, 50 %, %, 70 %, 80 %, 90%, 95%, 96%, 97%, 98%, 99%, 99.5%, 99.6%, 99.7%, 99.8% or more but less than 100% are zero-valent with respect to the phage surface expression of proteinaceous ligand-coat protein.

According to the above aspect of the invention, preferably the shaved phage display library is obtainable, preferably obtained using the method of the invention. Further according to the above aspect of the invention, preferably the coat protein is that encoded by gene 111 (gene III coat protein).

According to the methods of the invention, the phage cleavage conditions are adjusted so that a small proportion of those treated phage, that is 0. 0 1% or less, 0.02% or less, 0.03% or less, 0.04% or less, 0.05% or less, 0.06% or less, 0.07% or less, 0.08% or less, 0.09% or less 0. 1% or less, 0.2% or less, 0.3% or less, 0.4% or less, 0.5% or less, 0.6% or less, 0.7% or less,0.8% or less, 0.9% or less, 1% or less of those phage according to the present invention which have been treated with one or more cleavage/cleaving agent/s comprise a monovalent surface ligand (ligand -coat protein peptide) as herein defined.

According to the invention described herein, any suitable proteinaceous ligand may be expressed oii the surface of phage attached to a phage coat protein. Advantageously, the proteinaceous ligand is expressed as a fusion protein, such that the expressed ligand of interest is fused to the coat protein. Phage vectors suitable for the expression of such proteinaceous ligands are described in the detailed description of the invention.

Suitable ligands for surface expression on phage include immunoglohulin molecules both synthetic and naturally occurring. Suitable immunoglobulin molecules for use according to the methods of the invention include any of those in the list consisting of the following: single domain antibodies (dAbs) which include both variable heavy chain domains and variable light chain domains, scFv, Fabs, Fe, chimeric antibodies, hunianised antibodies, mutated and/or engineered antibodies, intrabodies and fragments of any of those listed above.

Those skilled in the art will appreciate that the methods of the invention are applicable to the selection of any proteinaceous ligand- binding partner interaction and all of these constructs can he the ligand.

CDRs may be grafted onto non-inimunoglohulin scaffolds for selection according to the methods of the invention. For example natural bacterial receptors such as SpA have been used as scaffolds for the grafting of CDRs to generate ligands which bind specifically to one or more epitopes. Details of this procedure are described in US 5,831,012.

Other scaffolds that can be used in the invention include non-Ig scaffolds such as fibronectin scaffolds which comprises a libronectin type 111 domain as described in W001/64942 (and incorporated herein by reference as examples of scaffolds that can be used in the present invention), Aiiticaliiis as (lescribed in W099/16873 (and incorporated herein by reference as examples of scaffolds that can be used in the present invention) and polypeptide chains having a beta sandwich architecture such as a non-glycosylated CTLA4-hke beta sandwich as described in WO 00/60070 (and incorporated herein by reference as examples of scaffolds that can be used in the present invention). Other scaffold domains suitable for use in the invention include those described in W002088 1 71 and herein incorporated by reference.

Further scaffolds that can be used in the invention include the artificial antibodies" based on a protein A- derived domain (denoted as Affibody molecules) and described in EP0333691 (Cemu Bioteknik AB) and incorporated herein by reference as examples 1 5 of scaffolds that can be used in the invention).

Protein scaffolds may be combined; for example, CDRs niay be grafted on to a CTLA4 scaffold and used together with irnmunoglobulin V11 or VL domains to form a ligand. Likewise, fibronectin, lipocallin and other scaffolds may he combined.

Those skilled in the art will appreciate that the methods of the invention described herein are particularly advantageous for the selection of high affinity binding phage.

That is, the creation of nionovalency in the shaved phage libraries facilitates the discrimination in selecting between high affinity I igandhi nding partner interactions and low affinity binding partner interactions. As eluded to previously, this is a consequence of a substantial diminution of avidity effects contributing to the strength of phage-ligand-binding partner interactions as a direct consequence of shaving the phage.

Thus in a further aspect the present invention provides the use of a phage library according to the present invention in the selection of one or more ligands capable of interacting with a binding partner, preferably interacting with a binding partner with high binding affinity.

According to the abovc aspect of the invention, preferably the library is a shaved phage library as described herein.

Methods for the measurement of the affinity of ligand- binding partner interactions will be familiar to those skilled in the art and are described in the detailed description of the invention.

DEFINITIONS.

1 5 Shaved phage: The present inventors have named the process of treating phage which comprise a cleavage site within the phage coat protein-surface expressed proteinaceous ligand with a cleavage agent, such that a small portion of the resulting phage are monovalent with respect to surface ligand, shaving phage'. Accordingly, the resultant phage are referred to by the present inventors as shaved phage'. For the avoidance of any doubt, the term phage' and shaved phage' does include within its scope phageniid' or shaved phagemid'. In a preferred embodiment of the present invention, those phage used are not phagemid. Even more preferably, those phage for use according to the methods of the invention do not include phagemid and encode the gene III coat protein.

Cleavable site/cleavage site: According to the present invention, the term a cleavable site'/cleavage site' refers to one or more amino acid/s within the surface expressed proteinaceous ligand-phage coal protein which is recognised by a cleavage agent and is capable of being cleaved by that agent. According to the invention described herein, upon cleavage of the coat protein-ligand with a cleavage agent as herein described the proteinaceous ligand is released from the coat protein. The cleavage/cleavable site may he located within the amino acid sequence comprising the proteinaceous ligand. That is, it may he a naturally occurring component of the proteinaceous ligand.

Alternatively, the cleavage site may be engineered into the proteinaceous ligand, for example by mutagenesis of the nucleic acid encoding the ligand. Further the cleavage site may be present within a linker peptide' which comprises a number of amino acids which together form a nonnaturally occurring component of the ligand coat protein.

Preferably the cleavable site/cleavage site is located in between the ligand and the coat protein to which the ligand is attached. More preferably, the cleavage site is present within a linker peptide which is located in between the proteinaceous ligand and the phage coat protein.

Linker sequence (cleavable linker sequence): A linker sequence consists of a sequence of two or more amino acids which sequence is not naturally occurring at that specific location in the ligand of interest. Further the linker is a sequence of amino acids which is cleavable by a cleaving agent and which results in the release of the 1 5 ligand from the coat protein to which it is attached, wherein the linker sequence is not a naturally occurring component of the C terminus of that ligand.

Cleavage agent/cleaving agent: According to the invention described herein a cleavage/cleaving agent' is any agent which when applied under suitable conditions to a ligand-coat protein expressed on the surface of a phage as described herein is capable of cleaving the ligand- coat protein at one or more sites with a resultant release of the ligand from the coat protein. Advantageously the cleaving/cleaving agent' is a protease.

Monovalent ligand (present on a phage surface): Cleavage/cleavable site/s present within proteinaceous ligand--coat protein displayed on the surface of members of a phage library according to the present invention, are suitably positioned such that upon treatment of that library with one or more cleaving agents under suitable conditions, a proportion of the resultant phage possess only one expressed proteinaceous ligand of interest attached to coat protein on their cell surface. That is the resultant shaved phage are monovalent' in relation to the presence of surface expressed ligand. The present inventors have found that upon treatment with a cleaving/cleavage agent the majority of those treated phage (shaved phage) are zero-valent with respect to the expression of surface ligand. That is the majority of those phage which have been treated with cleaving agent comprise no ligand attached to the surface of the phage.

High affinity ligand-binding partner interaction: According to the present invention, the term, high affinity binding interaction' (referring to binding between the phage surface displayed ligand and one or more binding partners) means an interaction of 0.01 -500nM in strength. Advantageously, the term high affinity' binding refers to a binding interaction of between 0.1-400nm, 0.i-300nm, 0.1-200nm, 0.1-lOOnm,0.1-9OnM, 0.l-8OnM, 0.l-7Onm, 0.1-ôOnm, 0.1-SOnrn, 0.1-40 nM in strength, more advantageously between 0.1-30 nm, 0.1-25nM, 0.1-2OnM, 0. l-l5nM, 0.1-12 nM, 0.1-lOnM, 0.1-8nM, 0.1-6nM, 0.1-4nM, 0.1-0.3nM or 0.1-2nM in strength.

Methods for the measurement of the affinity of ligand- binding partner interactions will he familiar to those skilled in the art and are described in the detailed description of the invention. As used herein the term very high affinity binding phage' refers to those phage which bind to one or more antigens/targets with an affinity of 0.0mM or less (10pM or less).

Immunoglobulin: This refers to a family of polypeptides which retain the immunoglohulin fold characteristic of antibody molecules, which contains two fi sheets and, usually, a conserved disuiphide bond. Members of the immunoglobulin superfamily are involved in many aspects of cellular and non-cellular interactions in vivo, including widespread roles in the immune system (for example, antibodies, T-cell receptor molecules and the like), involvement in cell adhesion (for example the ICAM molecules) and intracellular signalling (for example, receptor molecules, such as the PDGF receptor). The present invention is applicable to all immunoglobulin superfamily molecules which possess binding domains. Preferably, the present invention relates to antibodies.

Antibody: An antibody (for example lgG, 1gM, IgA, IgD or IgE) or fragment (such as a Fab, F(ab')2, Fv, disulphide linked Fv, scFv, closed conformation multispecific antibody, disulphide-linked scFv, diabody, dAb for example single variable domain, human single variable domain or Camelid V1111) whether derived from any species naturally producing an antibody, or created by recombinant DNA technology; whether isolated from serum, B-cells, hybridomas, transfectomas, yeast or bacteria). This list is not intended to be exhaustive.

Library: According to the present invention, a library' refers to a collection of members (comprising the library). For example a collection of nucleotide e.g DNA sequences within clones may be described as a library or a collection of phage particles may he described as a library of phage.

BRIEF DESCRIPTION OF THE FIGURES.

Figure 1 describes dAbs used in this study and their respective properties.

Figure 2 shows ELISA of the binding of DOM1-9 and DOM1-122 phage to surface bound biotinylated cytokine coated at I 5nM. After blocking the antigen coated plate, serial dilutions of lxlO' DOMI-9 or DOMI-122 phage were applied to rows A and B respectively, in each row (Al-12 or B1-l2), lane l=lxlO' phage, and lanes 2-10 = 10 fold serial dilutions of lx lO' phage in PBS. Row C= no phage bound as negative controls. After washing, bound phage were detected by incubation with HRP- conjugated anti-MI 3 monoclonal antibody. The signal was visualised by addition of lOOpJ of SureBlue I-Component TMB MicroWell Peroxidase solution to each well, and the plate lefi at room temperature until a suitable signal developed. In this reaction, a deep blue soluble product will develop as bound HRP labelled conjugate reacts with the substrate. The reaction was stopped by the addition of lOOp] IM hydrochloric acid and the OD at 450nm read in a 96-well plate reader. The 0D450 is proportional to the amount of bound anti-M13-HRP conjugate.

Figure 3 shows the profile of phage recovery after selection with varying concentrations of antigen for DOM 1-9, DOM 1-122, DOM 1-95-3, and DOM 5 phage. Soluble model selections were performed using lxlO' DOM 1-9 (KD=>l0/LM), Doml-122 (KD=5OnM), DOM 1-95-3 (KD=3nM)and DOM 5 (no affinity for cytokine 1) phage at a range of concentrations of cytokine (0.01-lOOnM) in Or(Ier to establish a concentration of antigen at which recovered titres of each phage

population would be well above background levels.

Figure 4 shows the ELISA of the binding of Dom2-5-19 phage to immobilised biotinylated cytokine 2. 10 samples of lx lOb phage were each treated with a different final concentration of PreScissioniM (mM, 2OnM, 5OnM, l0OnM, 200nM, 500nM, lj.iM, 2jiM, 4.8.iM) and applied to wells A1-A9 respectively. Serial dilutions of each of these phage samples were applied to rows B-H such that for each, Row B1x 10 phage and Rows C-H= two-fold serial dilutions of Row B. lx 1010 undigested phage were applied in lane 11 and serially diluted in the manner above. After washing, bound phage were detected by incubation with HRP-conjugated anti-M13 monoclonal antibody. The signal was visualised by addition of I 00il of SureBlue I Component TMB MicroWell Peroxidase solution to each well, and the plate left at room temperature until a suitable signal developed, in this reaction, a deep blue soluble product will develop as bound HRP labelled conjugate reacts with the substrate. The reaction was stopped by the addition of lOOjl IM hydrochloric acid and the OD at 450nrn read in a 96-well plate reader. The 0D450 is proportional to the amount of bound anti-M13-HRP conjugate.

Figure 5 shows the plot of the recovered titres of DOM 1-9, DOM 1-122 and DOM 5 phage after treatment with OnM, 200nM, 500nM, 700nM, and I 000nM PreScissionlM and subsequent soluble selection on cutokine I (lOnM). Briefly, lxlO'0 DOM 1-9, DOM 1-122 and DOM 5 phage were treated with PreScission'TM for 30 mm at a final concentrations of OnM, 200nM, 500nM 700nM and l000nM and the reactions stopped with Pefabloc. Soluble selections were then performed using the shaved and unshaved phage at lOnM biotinylated eytokine I and the recovered titres calculated.

Figure 6 shows the Western blot of' lxlO' phage DOM3-5-l phage digested with high concentrations of PreScission'TM. Lane I undigested, lane 2=1 tM PreScission'TM, lane 3=2tM PreScissio&M, lane 3=3jiM PreScission1M, lane 4 4.8tM (neat PreScissio&'M added). Phage were digested with PreScissioniM for 30 mm at 20 C, heated and analysed by SDS PAGE and blotted onto PVDF. Blots were probed using anti-c-rnyc antibody, washed, then developed using the Western Breeze alkaline phosphalase conjugated anti-mouse secondary antibody and chroniogen (Invitrogen) according to the manufacturor's instructions. Both the free gill and the dAb-gill Fusion retain the myc tag and therefore can he detected as above. Cleavage of the (lAb-fusion protein by PrcScission'M is visible in lanes 2, 3, 4 and 5 by the loss of the heavier fusion protein band after shaving compared to the unshaved sample (lane 1).

Figure 7 shows the Western blot of 1xl010 DOM3-5-1 phage digested with lower concentrations of PreScission'TM lane I = undigested, lane 2=200nM, lane 3500nM, lane 4=liM. Phage were digested with PreScission'TM for 30 mm at 20 C, heated and analysed by SDS PAGE and blotted onto PVDF. Blots were probed using anti-c-myc antibody, washed, then developed using the Western Breeze alkaline phosphatase coniugated anti-mouse secondary antibody and chrornogen (Invitrogen) according to the manufacturor's instructions. Both the free gill and the dAb-gIll fusion retain the myc tag and therefore can be detected as above by the reduction in intensity (in lane 2) and loss (in lanes 3 and 4) of the heavier fusion protein band after shaving compared to the unshaved sample (lane 1).

Figure 8 shows the theoretical distribution of the number of dAbs on phage at a given level of fusion protein. From the Western blot in Figure 7, the percentage of fusion protein present was estimated by densitometry after shaving with OnM, 200nM, 500nM and lM PreScission'TM. This was estiniated to be undigested=28%, 200nM = 14%, 500nM=3%, lM=l00% (although not measurable accurately). These percentages are shown on the key in this figure and were used to calculate the theoretical number of p111-dAb fusioii proteins on the surface of a phage according to the formula given in Example 4. In the figure a X represents 28%; diamonds represent 14%; squares represent 3% and triangles represent 100%.

Figure 9 shows the Western Blot of IxlO' DOM4-53-99 phage digested with and without PreScission M Lane 1 = undigested, lane 2=200nM, lane 3=500nM, lane 4=ltM. Phage were digested with PreScissionTM for 30 mm at 20 C, heated and analysed by SDS PAGE and blotted onto PVDF. Blots were probed using anti-c-myc antibody, washed, then developed using the Western Breeze alkaline phosphatase conjugated anti-mouse secondary antibody and chromogen (Invitrogen) according to the manufacturor's instructions. Both the free gill and the dAb-gill fusion retain the myc tag and therefore can he detected as above by the reduction in intensity (in lane 2) and loss (in lanes 3 and 4) of the heavier fusion protein band after shaving compared to the unshaved sample (lane 1).

Figure 10 shows the Vector niap of pDOM9 together with the sequence of the linker region. The amino acid sequence after the Not I site is shown in frame as if a dAb were present.

Figure 11 shows a table of reagents used Figure 10 (a): Shows that as phage are shaved, the percentage of phage displaying a dAb copy will decrease. However, the ratio of phage displaying one dAb copy over multivalent' phage will increase (up to 4,999-fold). The table also shows that for repertoire of 108 size, even drastic shaving (0.01%) will still allow to sample the full genetic diversity of the phage repertoire (ie. each clone will be represented by 5 monovalent phage).

DETAILED DESCRIPTION OF THE INVENTION.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art (e.g., in cell culture, molecular genetics, nucleic acid chemistry, hybridisation techniques and biochemistry). Standard techniques are used for molecular, genetic and biochemical methods (see generally, Sambrook Ct aL, Molecular Cloning: A Laboratory Manual, 2d ed. (1989) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. and Ausuhel el al., Short Protocols in Molecular Biology (1999) 4 Ed, John Wiley & Sons, Inc. which are incorporated herein by reference) and chemical methods.

(A) Pha2e display-General.

Bacteriophage lambda expression systems may be screened directly as bacteriophage plaques or as colonies of lysogens, both as previously described (Huse et a!. (1989,) Science, 246: 1275; Caton and Koprowski (1990) Proc. Nail. Acad. Sci. U.S.A., 87; Mullinax et al. (1990) Proc. Nail. Acad. kScI. U.S.A., 87: 8095; Persson et al. (1991) Proc. Nail. Acad. Sci. US.A., 88: 2432) and are of use in the invention. Whilst such expression systems can be used to screening up to 106 different members of a library, they are not really suited to screening of larger numbers (greater than 106 members).

Other screening systems rely, for example, on direct chemical synthesis of library members. One early method involves the synthesis of peptides on a set of pins or rods, such as described in W084/03564. A similar method involving peptide synthesis on beads, which forms a peptide library in which each bead is an individual library member, is described in U.S. Patent No. 4,63 1,211 and a related method is described in W092/0009l. A significant improvement of the bead-based methods involves tagging each bead with a unique identifier tag, such as an oligonucleotide, so as to facilitate identification of the amino acid sequence of each library member. These improved bead-based methods are described in W093/06 121.

Another chemical synthesis method involves the synthesis of arrays of peptides (or peptidomirnetics) on a surface in a manner that places each distinct library member (e.g., unique peptide sequence) at a discrete, predeuined location in the array. The identity of each library member is determined by its spatial location in the array. The locations in the array where binding interactions between a predetermined molecule (e.g., a receptor) and reactive library members occur is determined, thereby identi lying the sequences of the reactive library members on the basis of spatial location. These methods are described in U.S. Patent No. 5,143,854; W090/15070 and W092/10092; Fodor ci a!. (1991) Science, 251: 767; Dower and Fodor (1991) Ann. Rep. Med. Chem., 26: 271.

The recent advent of phage-display technology (Smith (1985) Science, 228: 1315; Scott and Smith (1990) Science, 249: 386; McCafferty eta!. (1990) Nature, 348: 552) has enabled the in vitro selection of human antibodies against a wide range of target antigens froni "single pot" libraries. These phage-antibody libraries can be grouped into two categories: natural libraries which use rearranged V genes harvested from human B cells (Marks et a!. (1991) 1. Mo!. Biol., 222: 581; Vaughan et a!. (1996) Nature l3ioicch., 14: 309) or synthetic libraries whereby germline V gene segments are rearranged' iii vitro (Hoogenboom & Winter (1992) .1. Mo!. Biol., 227: 381; Nissirn e- at. (1994) EMBO 1., 13: 692; Griffiths et a!. (1994) EMBO.1., 13: 3245; Dc Kruif e a!. (1995) 1. Mo!. Biol., 248: 97) or where synthetic CDRs are incorporated into a single rearranged V gene (Barbas et a!. (1992) Proc. Nat!. Acad. Sci. USA, 89: 4457).

Although synthetic libraries help to overcome the inherent biases of the natural repertoire which can limit the effective size of phage libraries constructed from rearranged V genes, they require the use of long degenerate PCR primers which frequently introduce base-pair deletions into the assembled V genes. This high degree of randomisation may also lead to the creation of antibodies which are unable to fold 1 5 correctly and are also therefore non-functional. Furthermore, antibodies selected from these libraries may he poorly expressed and, in many cases, will contain framework mutations that may affect the antibodies immunogenicity when used in human therapy.

Such systems, in which diverse peptide sequences are displayed on the surface of filamentous bacteriophage (Scott and Smith (1990) supra), have proven useful for creating libraries of antibody fragments (and the nucleotide sequences that encoding them) for the in vitro selection and amplification of specific antibody fragments that bind a target antigen. The nucleotide sequences encoding the V11 and V1. regions are linked to gene fragments which encode leader signals that direct them to the periplasmic space of E. co!i and as a result the resultant antibody fragments are displayed on the surface of the bacteriophage, typically as fusions to bacteriophage coat proteins (e.g., pIll or pVllI). Alternatively, antibody fragments are displayed externally on lambda phage capsids (phagehodies). An advantage of phage-bascd display systems is that, because they are biological systems, selected library members can be amplified simply by growing the phage containing the selected library member in bacterial cells. Furthermore, since the nucleotide sequence that encode the polypeptide library member is contained on a phage or phagemid vector, sequencing, expression and subsequent genetic manipulation is relatively straightforward.

Methods for the construction of bacteriophage antibody display libraries and lambda phage expression libraries are well known in the art (McCafferty et at. (1990) supra; Kang ci a!. (1991) Proc. Nat!. Acad. Sci. U.S.A., 88: 4363; Clackson ci at. (1991) Nature, 352: 624; Lownian eta!. (1991) Biochemistry, 30: 10832; Burton et at. (1991) Proc. Nat!. Acad. Sci USA., 88: 10134; [-loogenboorn et at. (1991) Nucleic Acids Res., 19: 4133; Chang et a!. (1991) 1. Immunot., 147: 3610; Breitling et at. (1991) Gene, 104: 147; Marks et a!. (1991) supra; Barbas et a!. (1992) supra; Hawkins and Winter (1992) J. Inununo!., 22: 867; Marks et a!., 1992, I. Biol. Chein., 267: 16007; Lerner ci a!. (1992) Science, 258: 1 3 13, incorporated herein by reference).

One particularly advantageous approach has been the use of scFv phagelibraries (Huston ci a!., 1988, Proc. NatI. Acad. Sci U.S.A., 85: 5879-5883; Chaudhary et at.

(1990) Proc. Nati. Acad. Sci U.S.A., 87: 1066-1070; McCafferty et at. (1990) supra; Clackson eta!. (1991) supra; Marks cia!. (1991) supra; Chiswell et at. (1992) Trends Biotech., 10: 80; Marks ci' a!. (1992) supra). Various embodiments of scFv libraries displayed on bacteriophage coat proteins have been described. Refinements of phage display approaches are also known, for example as described in W096/062 13 and W092/0l 047 (Medical Research Council et a!.) and W097/08320 (M.orphosys, sLipra), which are incorporated herein by reference.

Other systems for generating libraries of polypeptides or nucleotides involve the use of cell-free enzymatic machinery for the in vitro synthesis of the library members. In one method, RNA molecules are selected by alternate rounds of selection against a target ligand and PCR amplification (Tuerk and Gold (1990) Science, 249: 505; Ellington and Szostak (1990) Nature, 346: 818). A similar technique may be used to identify DNA sequences which hind a predetermined human transcription factor (Thiesen and Bach (1990) Nuc!eic Acids Res., 18: 3203; Beaudry and. Joyce (1992) Science, 257: 635; W092/05258 and W092/14843). In a similar way, in vitro translation can be used to synthesise polypeptides as a method for generating large libraries. These methods which generally comprise stahilised polysome complexes, are described further in W088/08453, W090/05785, W090/07003, W091/02076, W091/05058, and W092/02536. Alternative display systems which are not phage-based, such as those disclosed in W095/22625 and W095/1 1922 (Affymax) use the polysornes to display polypeptides for selection. These and all the foregoing documents also are incorporated herein by reference.

Preparation of antibodies acccordin to the invention.

Antibodies according to the invention, may he prepared according to previously established techniques, used in the field of antibody engineering, for the preparation of scFv, !phageu antibodies and other engineered antibody molecules. Techniques for the preparation of antibodies, are for example described in the following reviews and the references cited therein: Winter & Milstein, (1991) Nature 349:293-299; Plueckthun (1992) Immunological Reviews 130:151-188; Wright et a!., (1992) Crti. Rev.

Irnmunol.l2:125-l68; Holliger, P. & Winter, G. (1993) Curr. Op. Biotechn. 4, 446- 449; Carter, et a!. (1995) .J. Hematother. 4, 463-470; Chester, K.A. & Hawkins, R.E.

(1995) Trends Biotechn. 13, 294-300; Hoogenboom, H.R. (1997) Nature Biotechnol. 15, 125-126; Fearon, D. (1997) Nature Biotechnol. 15, 618-619; Plückthun, A. & Pack, P. (1997) Immunotechnology 3, 83-105; Carter, P. & Merchant, A. M. (1997) Curr. Opin. Biotechnol. 8, 449-454; 1-lolliger, P. & Winter, G. (1997) Cancer Immunol. Immunother. 45,128-130.

The techniques employed for selection of the variable domains employ libraries and selection procedures which are known in the art. Natural libraries (Marks eta!. (1991) 1. Mo!. Biol., 222: 581; Vaughan et a!. (1996) Nature Biotech., 14: 309) which use rearranged V genes harvested froni human B cells are well known to those skilled in the art. Synthetic libraries (Hoogenhoom & Winter (1992) 1 Mo!. Biol., 227: 381; Barbas eta!. (1992) Proc. Nat!. Acad. Sci. USA, 89: 4457; Nissim eta!. (1994) EMBO 1,13:692; Griffiths eta!. (1994) EMBO.I, 13: 3245; De Kruif et a!. (1995) 1. Mol. Rio!., 248: 97) are prepared by cloning immunoglobulin V genes, usually using PCR.

Errors in the PCR process can lead to a high degree of randomisation. Vii and/or V[ libraries may be selected against target antigens or epitopes separately.

Library vector systems A variety of selection systems are known in the art which are suitable for use in the present invention. Examples of such systems are described below.

Bacteriophage lambda expression systems may be screened directly as bacteriophage plaques or as colonies of lysogens, both as previously described (Huse et a!. (1989) Science, 246: 1275; Caton and Koprowski (1990) Proc. Nail. Acad. Sci. U.S.A., 87; Mullinax ci' a!. (1990) Proc. Natl. Acad. Sci. U.S.A., 87: 8095; Persson et al. (1991) Proc. Nat!. Acad. Sd. U.S.A., 88: 2432) and are of use in the invention. Whilst such expression systems can he used to screen up to 106 different members of a library, they are not really suited to screening of larger numbers (greater than 106 members).

Of particular use in the construction of libraries are selection display systems, which enable a nucleic acid to be linked to the polypeptide it expresses. As used herein, a selection display system is a system that permits the selection, by suitable display means, of the individual members of the library by binding the generic and/or target ligands.

Selection protocols for isolating desired members of large libraries are known in the art, as typified by phage display techniques. Such systems, in which diverse peptide sequences are displayed on the surface of ilamentous bacteriophage (Scott and Smith (1990) Science, 249: 386), have proven useful for creating libraries of antibody fragments (and the nucleotide sequences that encoding them) for the in vitro selection and amplification of specific antibody fragments that bind a target antigen (McCafferty eta!., WO 92/01 047). The nucleotide sequences encoding the VH and V1. regions are linked to gene fragments which encode leader signals that direct them to the periplasmic space of E. co/i and as a result the resultant antibody fragments are displayed oii the surface of the bacteriophage, typically as fusions to bacteriophage coat proteins (e.g., pill or pVlll). Alternatively, antibody fragments are displayed externally on lambda phage capsids (phagebodies). An advantage of phage-based display systems is that, because they are biological systems, selected library members can be amplified simply by growing the phage containing the selected library member in bacterial cells. Furthermore, since the nucleotide sequence that encode the polypeptide library member is contained on a phage or phagemid vector, sequencing, expression and subsequent genetic manipulation is relatively straightforward.

Methods for the construction o bacteriophage antibody display libraries and lambda phage expression libraries are well known in the art (McCafferty ci a!. (1990) Nature, 348: 552; Kang el a!. (1991) Proc. Nail. Acad. Sd. U.S.A., 88: 4363; Claekson et a!.

(1991) Nature, 352: 624; Lownian eta!. (1991) Biochemistry, 30: 10832; Burton eta!.

(1991) Proc. Nat!. Acad. Sd U.S.A., 88: 10134; Hoogenboom et a!. (1991) Nucleic Acids Res., 19: 4133; Chang et a!. (1991) 1. Immunol., 147: 3610; Breitling ci a!.

(1991) Gene, 104: 147; Marks cial. (1991) supra; Barbas eta!. (1992) supra; Hawkins and Winter (1992) .1. Immunol., 22: 867; Marks et a!., 1992, J. Biol. Chem., 267: 16007; Lerner eta!. (1992) Science, 258: 1313, incorporated herein by reference).

One particularly advantageous approach has been the use of phagelibraries (Huston et a!., 1988, Proc. Natl. Acad. Sci U.S.A., 85: 58795883; Chaudhary et a!. (1990) Proc. Nati. Acad. Sci U.S.A., 87: 1066-1070; McCafferty et a!. (1990) supra; Clackson eta!.

(1991) Nature, 352: 624; Marks et a!. (1991) J. Mo!. Biol., 222: 581; Chiswell et a!.

(1992) Trends Biotech., 10: 80; Marks ci al. (1992) J. Bio!. Chem., 267). Various embodiments of scFv libraries displayed on bacteriophage coat proteins have been described. Refinements of phage display approaches are also known, for example as described in W096/062 13 and W092/01 047 (Medical Research Council ci a!.) and WO97/08320 (Morphosys), which are incorporated herein by reference.

Other systems for generating libraries of polypeptides involve the use of cell-free enzymatic machinery for the in vitro synthesis of the library members. In one method, RNA molecules are selected by alternate rounds of selection against a target ligand and PCR amplification (Tuerk and Gold (1990) Science, 249: 505; Ellington and Szostak (1990) Nature, 346: 818). A similar technique maybe used to identify DNA sequences which bind a predetermined human transcription factor (Thiesen and Bach (1990) Nucleic Aculs Res., 18: 3203; Beaudry and Joyce (1992) Science, 257: 635; W092/05258 and W092/14843). In a similar way, in vitro translation can be used to synthesise polypeptides as a method for generating large libraries. These methods which generally comprise stahilised polysome complexes, are described further in W088/0845 3, W090/05 785, W090/07003, W09 1/02076, W09 1/05058, and W092/02536. Alternative display systems which are not phage-based, such as those disclosed in W095/22625 and W095/1 1922 (Affymax) use the polysomes to display 1 0 polypeptides for selection.

A still further category of techniques involves the selection of repertoires in artificial compartments, which allow the linkage of a gene with its gene product. For example, a selection system in which nucleic acids encoding desirable gene products may be selected in microcapsules formed by water-in-oil emulsions is described in W099/0267 1, W000/407 1 2 and Tawlik & Griffiths (1 998) !/ature Biotechnol 16(7), 652-6. Genetic elements encoding a gene product having a desired activity are compartrnentalised into microcapsules and then transcribed and/or translated to produce their respective gene products (RNA or protein) within the microcapsules.

Genetic elements which produce gene product having desired activity are subsequently sorted. This approach selects gene products of interest by detecting the desired activity by a variety of means.

Library Construction.

Libraries intended for selection, may be constructed using techniques known in the art, for example as set forth above, or may be purchased from commercial sources.

Libraries which are useful in the present invention are described, for example, in W099/20749. Once a vector system is chosen and one or more nucleic acid sequences encoding polypeptides of interest are cloned into the library vector, one may generate diversity within the cloned molecules by undertaking mutagenesis prior to expression; alternatively, the encoded proteins may be expressed and selected, as described above, before mulagenesis and additional rounds of selection are performed. Mutagenesis of nucleic acid sequences encoding structurally optimised polypeptides is carried out by standard molecular methods. Of particular use is the polyrnerase chain reaction, or PCR, (Mullis and Faloona (1987) Methods Enzymol., 155: 335, herein incorporated by reference). PCR, which uses multiple cycles of DNA replication catalysed by a thermostable, DNAdependent DNA polymerase to amplify the target sequence of interest, is well known in the art. The construction of various antibody libraries has been discussed in Winter et a!. (1994) Ann. Rev. Immunology 12, 433-55, and

references cited therein.

PCR is performed using template DNA (at least lfg; more usefully, 1-1000 ng) and at least 25 pniol of oligonucleotide primers; it may he advantageous to use a larger amount of primer when the primer pool is heavily heterogeneous, as each sequence is represented by only a small fraction of the molecules of the pooi, and amounts become limiting in the later amplification cycles. A typical reaction mixture includes: 2tl of DNA, 25 pmol of oligonucleotide primer, 2.5 p1 of lOX PCR buffer I (Perkin-Elmer, Foster City, CA), 0.4 p1 of 1.25 pM dNTP, 0.15 p1 (or 2.5 units) of Taq DNA polyrnerase (Perkin Elmer, Foster City, CA) and deionized water to a total volume of p1. Mineral oil is overlaid and the PCR is performed using a programmable thermal cycler. The length and temperature of each step of a PCR cycle, as well as the number of cycles, is adjusted in accordance to the stringency requirements in effect. Annealing temperature and timing are determined both by the efficiency with which a primer is expected to anneal to a template and the degree of mismatch that is to be tolerated; obviously, when nucleic acid molecules are simultaneously amplified and mutagenised, mismatch is required, at least in the first round of synthesis. The ability to optimise the stringency of primer annealing conditions is well within the knowledge of one of moderate skill in the art. An annealing temperature of between 30 C and 72 C is used. Initial denaturation of the template molecules normally occurs at between 92 C and 99 C for 4 minutes, followed by 20-40cycles consisting of denaturation (94- 99 C for 1 5 seconds to 1 minute), annealing (temperature determined as discussed above; 1-2 minutes), and extension (72 C for 1-5 minutes, depending on the length of the amplified product). Final extension is generally for 4 minutes at 72 C, and may be followed by an indefinite (0- 24 hour) step at 4 C.

i. Selection of the main-chain conformation The members of the imrnunoglohulin superfamily all share a similar fold for their polypeptide chain. For example, although antibodies are highly diverse in terms of their primary sequence, comparison of sequences and crystallographic structures has revealed that, contrary to expectation, five of the six antigen binding loops of antibodies (HI, F12, LI, L2, L3) adopt a limited number of main-chain conformations, or canonical structures (Chothia and Lesk (1987)1 Mol. Biol., 196: 901; Chothia cial.

(1989) Na/tire, 342: 877). Analysis of loop lengths and key residues has therefore enabled prediction of the main-chain conformations of HI, H2, Li, L2 and L3 found in the majority of human antibodies (Chothia c/ a!. (1992) .J. Mo!. Biol., 227: 799; Tomlinson cial. (i995) EMBOJ., 14: 4628; Williams eta!. (1996)1 Mol Biol., 264: 220). Although the H3 region is much more diverse in terms of sequence, length and structure (due to the use of D segments), it also forms a limited number of main-chain conformations for short loop lengths which depend on the length and the presence of particular residues, or types of residue, at key positions in the ioop and the antibody framework (Martin et a!. (1996) J. Mo!. Biol., 263: 800; Shirai ci a!. (1996) FEBS Letters, 399: I).

The ligands of the invention may themselves be provided in the form of libraries. In one aspect of the present invention, libraries of immunogiobuiin ligands are designed in which certain loop lengths and key residues have been chosen to ensure that the n-tam-chain conformation of the members is known. Advantageously, these are real conformations of immunoglobulin superfamily molecules found in nature, to minimise the chances that they are non-functional, as discussed above. Germline V gene segments serve as one suitable basic framework for constructing antibody or T-ceil receptor libraries; other sequences are also of use. Variations may occur at a low frequency, such that a small number of functional members may possess an altered main-chain conformation, which does not affect its function.

Canonical structure theory is also of use to assess the number of different main-chain conformations encoded by ligands, to predict the main-chain conformation based on ligand sequences and to choose residues for diversification which do not affect the canonical structure. It is known that, in the human VK domain, the LI loop can adopt one of four canonical structures, the L2 loop has a single canonical structure and that 90% of human V, domains adopt one of four or five canonical structures for the L3 loop (Tomlinson eta!. (1995) supra); thus, in the VK domain alone, different canonical structures can combine to create a range of different main-chain conformations. Given that the V, domain encodes a different range of canonical structures for the Li, L2 and L3 loops and that V,, and V domains can pair with any VR domain which can encode several canonical structures for the H 1 and H2 ioops, the number of canonical structure combinations observed for these five loops is very large. This implies that the generation of diversity in the main-chain conformation may he essential for the production of a wide range of binding specificities. However, by constructing an 1 5 antibody library based on a single known main-chain conformation it has been found, contrary to expectation, that diversity in the main-chain conformation is not required to generate sufficient diversity to target substantially all antigens. Even more surprisingly, the single main-chain conformation need not be a consensus structure - a single naturally occurring conformation can he used as the basis for an entire library.

Thus, in a preferred aspect, the dAbs of the invention possess a single known main- chain con form ati on.

The single main-chain conformation that is chosen is preferably commonplace among molecules of the immunoglobulin superfamily type in question. A conformation is commonplace when a significant number of naturally occurring molecules are observed to adopt it. Accordingly, in a preferred aspect of the invention, the natural occurrence of the different main-chain conformations for each binding loop of an immunoglobulin domain are considered separately and then a naturally occurring variable domain is chosen which possesses the desired combination of main-chain conformations for the different loops. If none is available, the nearest equivalent may he chosen. It is preferable that the desired combination of main-chain conformations for the different loops is created by selecting germline gene segments which encode the desired main-chain conformations. It is more preferable, that the selected germlinc gene segments are frequently expressed in nature, and most preferable that they are the most frequently expressed of all natural germline gene segments.

Tn designing immunoglohulin ligands or libraries thereof the incidence of the different main-chain conformations for each of the six antigen binding loops may be considered separately. For HI, H2, LI, L2 and L3, a given conformation that is adopted by between 20% and I OO% of the antigen binding loops of naturally occurring molecules is chosen. Typically, its observed incidence is above 35% (i.e. between 35% and 100%) and, ideally, above 50% or even above 65%. Since the vast majority of H3 loops do not have canonical structures, it is preferable to select a main-chain conformation which is commonplace among those loops which do display canonical structures. For each of the loops, the conformation which is observed most often in the natural repertoire is therefore selected. In human antibodies, the most popular canomcal structures (CS) for each loop are as follows: 1-11 - CS 1 (79% of the expressed repertoire), H2 - CS 3 (46%), LI - CS 2 of VK(39%), L2 - CS 1 (100%), L3 - CS I of VK(36%) (calculation assumes a K:Xratio of 70:30, Hood eta!. (1967) Cold Spring Harbor Symp. Quant. Biol., 48: 133). For H3 loops that have canonical structures, a CDR3 length (Kabat eta!. (1991) Sequences of proteins ojimmunological interest, U.S. Department of Health and Human Services) of seven residues with a salt-bridge from residue 94 to residue 101 appears to he the most common. There are at least 16 human antibody sequences in the EMBL data library with the required H3 length and key residues to form this conformation and at least two crystallographic structures in the protein data bank which can he used as a basis for antibody modelling (2cgr and Itet). The most frequently expressed germlinc gene segments that this combination of canonical structures are the V1 segment 3-23 (DP- 47), the JH segment JH4b, the V, segment 02/012 (DPK9) and the J, segment JK1. V11 segments DP45 and DP38 are also suitable. These segments can therefore be used in combination as a basis to construct a library with the desired single main-chain conformation.

Alternatively, instead of choosing the single main-chain conformation based on the natural occurrence of the different main-chain conformations for each of the binding loops in isolation, the natural occurrence of combinations of main-chain conformations is used as the basis for choosing the single main-chain conformation. In the case of antibodies, for example, the natural occurrence of canonical structure combinations for any two, three, four, live or for all six of the antigen binding loops can be determined.

Here, it is preferable that the chosen conformation is commonplace in naturally occurring antibodies and most preferable that it observed most frequently in the natural repertoire. Thus, in human antibodies, for example, when natural combinations of the live antigen binding loops, Hl, H2, LI, L2 and L3, are considered, the most frequent combination of canonical structures is determined and then combined with the most popular conformation for the 1-T3 loop, as a basis for choosing the single main-chain conformation.

ii. Diversification of the canonical sequence Having selected several known main-chain conformations or, preferably a single known main-chain conformation, ligands according to the invention or libraries for use in the invention can be constructed by varying the binding site of the molecule in order to generate a repertoire with structural and/or functional diversity. This means that variants are generated such that they possess sufficient diversity in their structure and/or in their function so that they are capable of providing a range of activities.

The desired diversity is typically generated by varying the selected molecule at one or more positions. The positions to he changed can be chosen at random or are preferably selected. The variation can then he achieved either by randomisation, during which the resident amino acid is replaced by any amino acid or analogue thereof, natural or synthetic, producing a very large number of variants or by replacing the resident amino acid with one or more of a defined subset of amino acids, producing a more limited number of variants.

Various methods have been reported for introducing such diversity. Errorprone PCR (Hawkins el al. (1992) .J Mol. Biol., 226: 889), chemical mutagenesis (Deng et al. (1994) .1. Biol. Chem., 269: 9533) or bacterial mutator strains (Low et a!. (1996) 1 Mo!. Rio!., 260: 359) can be used to introduce random mutations into the genes that encode the molecule. Methods for mutating selected positions are also well known in the art and include the use of mismatched oligonucleotides or degenerate oligonucleotides, with or without the use of PCR. For example, several synthetic antibody libraries have been created by targeting mutations to the antigen binding ioops. The 113 region of a human tetanus toxoid-binding Fab has been randornised to create a range of new binding specificities (Barbas ci a!. (1992) Proc. Nat!. Acad. Sci. USA, 89: 4457). Random or semi-random H3 and L3 regions have been appended to gerrnline V gene segments to produce large libraries with unmutated framework regions (Hoogenboorn & Winter (1992) . 1. Mo!. Rio!., 227: 381; Barbas et a!. (1992) Proc. Nat!. Acad. Sci. USA, 89: 4457; Nissim et at. (1994) EMBO J., 13: 692; Griffiths eta!. (1994) EMBOJ., 13: 3245; Dc Kruifet a!. (1995) J. Mo!. Biol., 248: 97). Such diversification has been extended to include some or all of the other antigen binding loops (Crameri et al. (1996) Nature Med., 2: 100; Ricchmann ci al. (1995) Rio/Technology, 13: 475; Morphosys, W097/08320, supra).

Since loop randomisation has the potential to create approximately more than 1015 structures for H3 alone and a similarly large number of variants for the other five loops, it is not feasible using current transformation technology or even by using cell free systems to produce a library representing all possible combinations. For example, in one of the largest libraries constructed to date, 6 x lOb different antibodies, which is only a fraction of the potential diversity for a library of this design, were generated (Grifuiths ci a!. (1994) supra).

In a preferred embodiment, only those residues which are directly involved in creating or modifying the desired function of the molecule are diversified. For many molecules, the function will be to bind a target aiid therefore diversity should be concentrated in the target binding site, while avoiding chaiiging residues which are crucial to the overall packing of the molecule or to maintaining the chosen main-chain conformation.

Diversification of the canonical sequence as it applies to antibody domains In the case of antibodies, the binding site for the target/antigen is most often the antigen binding site. These residues are extremely diverse in the human antibody repertoire and are known to make contacts in high-resolution antibody/antigen complexes. For example, in L2 it is known that positions 50 and 53 are diverse in naturally occurring antibodies and are observed to make contact with the antigen. In contrast, the conventional approach would have been to diversify all the residues in the corresponding Complementarity Determining Region (CDR1) as defined by Kabat et a!. (1991, supra), some seven residues compared to the two diversified in the library for use according to the invention. This represents a significant improvement in terms of the functional diversity required to create a range of antigen binding specilicities.

In nature, antibody diversity is the result of two processes: somatic recombination of gerrnline V, D and J gene segments to create a naive primary repertoire (so called germline and junctional diversity) and somatic hypermutation of the resulting rearranged V genes. Analysis of human antibody sequences has shown that diversity in the primary repertoire is focused at the centre of the antigen binding site whereas somatic hypermutation spreads diversity to regions at the periphery of the antigen binding site that are highly conserved in the primary repertoire (see Tomlinson et a!.

(1996) 1 Mo!. Biol., 256: 813). This complementarily has probably evolved as an efficient strategy for searching sequence space and, although apparently unique to antibodies, it can easily he applied to other polypeptide repertoires. The residues which are varied are a subset of those that form the binding site for the target. Different (including overlapping) subsets of residues in the target binding site are diversified at different stages during selection, if desired.

In the case of an antibody repertoire, an initial naive' repertoire is created where sonie, but not all, of the residues in the antigen binding site are diversified. As used herein in this context, the term "naive" refers to antibody molecules that have no pre- determined target. These molecules resemble those which are encoded by the immunoglobulin genes of an individual who has not undergone immune diversification, as is the case with fetal and newborn individuals, whose immune systems have not yet been challenged by a wide variety of antigenic stimuli. This repertoire is then selected against a range of antigens or epitopes. If required, further diversity can then be introduced outside the region diversified in the initial repertoire.

This matured repertoire can be selected for modified function, specificity or affinity.

In the construction of libraries for use in the invention, diversification of chosen positions is typically achieved at the nucleic acid level, by altering the coding sequence which specifies the sequence of the polypeptide such that a number of possible amino acids (all 20 or a subset thereof) can be incorporated at that position.

Using the IUPAC nomenclature, the most versatile codon is NNK, which encodes all amino acids as well as the TAG stop codon. The NNK codon is preferably used in order to introduce the required diversity. Other codons which achieve the same ends are also of use, including the NNN codon, which leads to the production of the additional stop codons TGA and TAA.

A feature of side-chain diversity in the antigen binding site of human antibodies is a pronounced bias which favours certain amino acid residues. If the amino acid composition of the ten most diverse positions in each of the V1.1, V and V regions are summed, more than 76% of the side-chain diversity comes from only seven different residues, these being, serine (24%), tyrosine (14%), asparagine (11%), glycine (9%), alanine (7%), aspartate (6%) and threonine (6%). This bias towards hydrophilic residues and small residues which can provide main-chain flexibility probably reflects the evolution of surfaces which are predisposed to binding a wide range of antigens or epitopes and may help to explain the required promiscuity of antibodies in the primary repertoire.

Since it is preferable to mimic this distribution of amino acids, the distribution of amino acids at the positions to be varied preferably mimics that seen in the antigen binding site of antibodies. Such bias in the substitution of amino acids that permits selection of certain polypeptides (not just antibody polypeptides) against a range of target antigens is easily applied to any polypeptide repertoire. There are various methods for biasing the amino acid distribution at the position to be varied (including the use of tri-nucleotide mutagenesis, see W097/08320), of which the preferred method, due to ease of synthesis, is the use of conventional degenerate codons. By comparing the amino acid profile encoded by all combinations of degenerate codons (with single, double, triple and quadruple degeneracy in equal ratios at each position) with the natural amino acid use it is possible to calculate the most representative codon. The codons (AGT)(AGC)T, (AGT) (AGC)C and (AGT)(AGC)(CT) - that is, DVT, DVC and DVY, respectively using IUPAC nomenclature - are those closest to the desired amino acid profile: they encode 22% serine and 11% tyrosine, asparagine, glyeine, alanine, aspartate, threonine and cysteine. Preferably, therefore, libraries are constructed using either the DVT, DVC or DVY codon at each of the diversified positions.

characterisation of llands according to the present invention.

The binding of ligands according to the invention to its specific antigens or epitopes can be tested by methods which will be familiar to those skilled in the art and include ELISA. In a preferred embodiment binding is tested using monoclonal phage ELISA.

Phage ELTSA may be performed according to any suitable procedure: an exemplary protocol is set forth below.

Populations of phage produced at each round of selection can be screened for binding by ELISA to the selected antigen or epilope, to identify "polyclonal" phage antibodies.

Phage from single infected bacterial colonies from these populations can then he screened by ELISA to identify "monoclonal" phage antibodies. It is also desirable to screen soluble antibody fragments for binding to antigen or epitope, and this can also he undertaken by ELISA using reagents, for example, against a C- or N-terminal tag (see for example Winter ci al. (1994) Ann. Rev. Immunology 12, 433-55 and

references cited therein).

The diversity of the selected phage monoclonal antibodies may also be assessed by gel electrophoresis of PCR products (Marks et al. 1991, supra; Nissim et al. 1994 supra), probing (Tomlinson et al., 1992) J. Mo!. Biol. 227, 776) or by sequencing of the vector DNA.

(B) PREPARATION OF A SHAVED PHAGE LIBRARY ACCORDING TO THE fNVENTTON.

Thus in a first aspect the present invention provides a method for the preparation of a 1 0 phage display library, which library facilitates the selection of high affinity binding phage (a shaved phage library), which method comprises the step of: (a) Providing a library of phage expressing one or more copies of a proteinaceous ligand of interest on a pluralityof phage surfaces; wherein the ligand is attached to phage coat protein and wherein a cleavable site is located within the expressed 1 5!igand- coat protein, (b) Treating the phage library according to step (a) with a cleaving agent such that a small proportion of the treated phage are monovalent with respect to the presence of surface expressed proteinaceous ligand, and the majority of the phage comprise no surface expressed proteinaceous ligand.

It is an essential feature of the present invention that the phage are shaved' prior to any selection of!igands based on affinity being performed. Methods for the preparation of a phage display library are described above. As outlined in the method above, after the generation of a phage display library, phage are treated with a cleaving agent such that a proportion of those treated phage and monovalent with respect to the presence on the phage surface of proteinaceous ligand. Methods for the creation of monovalent phage are described below: In a preferred embodiment of the present invention, those phage used are not phagernid. Even more preferably, those phage for use according to the methods of the invention do not include phagernid and encode the gene III coat protein.

(i) Cleavable/cleavage sites.

Cleavable site/cleavage site: According to the present invention, the term a cleavable site'/cleavage site' refers to one or more amino acid/s within the surface expressed proteinaceous ligand-phage coat protein which is recognised by a cleavage agent and is capable of being cleaved by that agent. According to the invention described herein, upon cleavage of the coat protein-ligand with a cleavage agent as herein described the proteinaceous ligand is released from the coat protein. The cleavage/cleavable site may be located within the amino acid sequence comprising the proteinaceous ligand. That is, it may be a naturally occuring component of the proteinaceous ligand.

Alternatively, the cleavage site may he engineered into the proteinaceous ligand, for example by mutagenesis of the nucleic acid encoding the ligand. Further the cleavage site may be present within a linker peptide' which comprises a number of amino acids which together form a nonnaturally occuring component of the ligand coat protein.

Preferably the cleavahle site/cleavage site is located in between the ligand and the coat protein to which the ligand is attached. More preferably, the cleavage site is present within a linker peptide which is located in between the proteinaceous ligand and the phage coat protein. The linker sequence itself consists of a sequence of two or more amino acids which sequence is not naturally occurring at that location in the ligand of interest. That, is the linker is a sequence of amino acids which is cleavable by a cleaving agent and which results in the release of the ligand from the coat protein to which it is attached, wherein the linker sequence is not a naturally occurring component of the C terminus of that ligand.

The cleavage site caii he an enzymatic cleavage site such as that cleaved by a protease such as thrombin, Factor Xa, enteropeptidase such as serine protease enterokinase (including those described in WOO 198366), or by trypsin for example, or it can be a chemical cleavage site such as CNBr which cleaves at a methionine residue.

(ii) Cleavage agents: According to the invention described herein a cleavage/cleaving agent' is any agent which when applied under suitable conditions to a ligand- coat protein expressed on the surface of a phage as described herein is capable of cleaving the ligand- coat protein at one or more sites with a resultant release of the ligand from the coat protein.

Advantageously the cleaving/cleaving agent' is a protease.

(ha) Proteases.

There exist a number of highly specific proteases. While the invention does not reside in the choice of any particular protease, the protease is preferably sufficiently specific so that under the cleavage conditions, it will cleave the cleavage site but not any polypeptide essential to the viability of the phage, or the ligand. It is possible that choice olparticular cleavage conditions, e.g., low temperature, may make it feasible to use a protease that would otherwise be unsuitable.

The blood-clotting and complenientation system contains a number of very specific proteases. Usually, the enzymes at the early stages of cascade are more specific than are the later ones. For example, Factor X, (F.X.) is more specific than is thrombin. Bovine F.X. cleaves after the sequence Ile-Glu-Gly-Arg while human F.X.

cleaves after Ile-Asp-Gly-Arg. Either protease-linker pair may he used, as desired. If thrombin is used, the most preferred thrombin-sensitive linkers are those found in fibrinogen, Factor XIII, and prothrombin. Preferably, one would take the linker sequence from the species from which the thrombin is obtained; for example, if bovine thrombin is to be used, then one uses a linker taken from bovine fibrinogen or bovine F. XElI.

Human Factor XI, cleaves human Factor IX at two places: QTSK LT R/A EA V F and SFNDFTR180/VVGGE Thus one could incorporate either of these sequences (especially the underscored portions) as linker between the ligand and the coat protein and use human F.XI. to cleave them.

Fluman kallikrein cuts human F.X1I at Rs3: LFS SMTR/V VG F LV.

This sequence has significant similarity to the hF.XI. sites above. One could incorporate the sequence SSMTRVVG as a linker between ligand and the coat protein and cleave the ligand from the coat protein with human kallikrein.

Human F.XIl. cuts human F.XI at R369: K I p p R369/ I V G G T. 1 5 One could incorporate KIKPRIVG as a linker between ligand and the coat protein.

Other proteases that have been used to cleave fusion proteins include enterokinase, trypsin, collagenase, chymosin, urokinase, renin, and certain signal peptides. See Rutter, US 4,769,326 which is herein incorporated by reference.

In a preferred embodiment of the invention, the protease is any one in the group selected from the following: PreScissioniM which recognises and cleaves the cleavage site LEVLFQGP; factor Xa which cleaves the site IEGRGI; thronibin which cleaves the site LVPRGS and the site LVPKGS.

When a protease inhibitor is sought, the target protease and other proteases having similar substrate specifically are not preferred for cleaving the ligand from the coat protein. It is preferred that a linker resembling the substrate of the target protease not be incorporated anywhere on the display phage because this could make separation of excellent binders from the rest of the population needlessly more difficult.

If there is steric hindrance of the site-specific cleavage of the linker, the linker may be modified so that the cleavage site is more exposed, e.g. , by interposing glycines (for additional flexibility) or prolines (for maximum elongation) between the cleavage site and the hulk of the protein. GUAN91 improved thrombin cleavage of a GST fusion protein by introducing a glycine-rich linker (PGISGGGGG) immediately after the thrombin cleavage site (LVPRGS). A suitable linker niay also he identified by variegationand-selection.

(iii) Creating monovalency in the phage library shaving the phage'.

Monovalency in the phage libraries according to the invention described herein is created by treating the phage library described above with a cleaving agent such that a small proportion of the treated phage are monovalent with respect to the presence of surface expressed peptide ligand, and the majority of the phage comprise no surface expressed peptide ligand.

The present inventors have found that upon treatment with a cleaving/cleavage agent the majority of those treated phage (shaved phage) are zero-valent with respect to the expression of surface ligand. That is the majority of those phage which have been treated with cleaving agent comprise no ligand attached to the surface of the phage.

According to the methods of the invention, the phage cleavage conditions are adjusted so that a small proportion of those treated phage, that is 0. 01% or less, 0.02% or less, 0.03% or less, 0.04% or less, 0.05% or less, 0.06% or less, 0.07% or less, 0.08% or less, 0.09% or less 0.1% or less, 0.2% or less, 0.3% or less, 0. 4% or less, 0.5% or less, 0.6% or less, 0.7% or less,0.8% or less, 0.9% or less, 1% or less of those phage according to the present invention which have been treated with one or more cleavage/cleaving agent/s comprise a monovalent surface ligand (ligand -coat protein peptide) as herein defined.

Those skilled in the art will appreciate that the cleavage/cleaving conditions must be carefully controlled such that a proportion of those phage comprising a phage library become monovalent with respect to the presence on their surface of ligand attached to gene 111 coat protein when treated with one or more cleaving agents as defined herein.

Preferred cleavage conditions are described below and also in the examples. The preferred cleavage conditions for shaving page with PreSclssionTM

protease are as follows: lx 1010 phage were shaved with PreScission'TM protease at a final concentration of 200nM I iM for 30 minutes at 20 C, in PBS/lmM DTT. The reaction was stopped by the addition of Pefabloc Sc protease inhibitors to a final concentration of 10 mM. Phage may be incubated for longer or shorter periods depending on the level of shaving required, without under burden to the skilled person, If other proteases were used, the cleavage conditions required for that specific protease should he considered. For example, some protcases require calcium or other metal ions for activity, and therefore the buffer may be changed or supplemented accordingly.

This would be within the knowledge of those skilled in the art.

(D) AFFINITY BASED METHOD FOR THE SELECTION OF PHAGE FROM A PHAGE LIBRARY ACCORDING TO THE INVENTION.

In a further aspect the present invention provides a method for the selection of phage from a phage display library which method comprises the step of: (a) Providiiig a library of phage expressing one or niore copies of a proteinaceous ligand of interest on a plurality of phage surfaces; wherein the ligand is attached to phage coat protein and wherein a cleavable site is located within the expressed ligand- coat protein, (h) Treating the phage library according to step (a) with a cleaving agent such that a small proportion of the treated phage are monovalent with respect to the presence of surface expressed proteinaceous ligand, and the majority of the phage comprise no surface expressed proteinaceous ligand (shaving the phage) and (c) Selecting from the library treated according to step (a) and (b) above those proteinaceous ligand expressing phage which are capable of specifically binding to one or more binding partners.

Methods for the generation of a shaved phage display library are described above and

also in the detailed description olthe invention.

(i) Affinity based selection of phage.

Step (c) above, that is a method for selecting from a library treated according to step (a) and (h) above those ligand expressing phage which are capable of binding specifically to one or more binding partners (antigens) is performed by an affinity based method.

Such methods may be performed in solid support or in solution. In a preferred embodiment of the invention, the affinity based selections are performed in solution.

The present inventors have found that the methods of the invention are particularly suitable for the selection of high affinity binding phage from phage libraries according to the invention. In particular, the inventors have found that the methods described herein permit the discrimination between high affinity and low affinity antigen (target) binding ligands. Moreover, the methods of the invention permits the high affinity and very high affinity binders.

According to the present invention, the term, high affinity binding interaction' (referring to binding between the phage surface displayed ligand and one or more binding partners) means an interaction of 0.01-SOOnM in strength. Advantageously, the term high affinity' binding refers to a binding interaction of between 0.0l-400nm, 0.01 -300nm, 0.01 -200nrn, 0.01-1 OOnrn, 0.01 -9OnM, 0.01 -8OnM, 0.01 - 7Onrn, 0.01- ôOnm, 0.01 -5Onm, 0.01-40 nM in strength, more advantageously between 0. 01-30 nm, 0.01 -25nM, 0.01 -2OnM, 0.01 -l5nM, 0.01-12 nM, 0.01-1 OnM, 0.01 -8nM, 0. 01-6nM, 0.0l-4nM, 0.01-0.3nM or 0.01-2nM in strength. Methods for the measurement of the affinity of ligand- binding partner interactions will he familiar to those skilled in the art and are described in the detailed description of the invention. As used herein the term very high affinity binding phage' refers to those phage which bind to one or more antigens/targets with an affinity of 0.01 nM or less (10pM or less) Methods for the measurement of the strength of ligand/antigen (target) interactions will be familiar to those skilled in the art and are described in detail in Essential Immunology, Roitt. 1., 2001. Blackwell Scientific Pub.

(ia) Affinity selection of phage on a solid support.

Any suitable solid support can be used for the affinity based selection of phage using the method of the invention. Suitable solid supports include but are not limited to any of those in the list consisting of the following: solid supports in the form of plates which may be formed from perspex or any other suitable material; membranes which may he nitrocellulose, PVDF (amershani), beads including biotin or streptavidin coated beads.

(ib) Affinity selection of phage in solution.

In a preferred embodiment of the invention, the selection step is performed in solution.

In this way any avidity effects are niinimised and thus the selection of ligand-hinding partners (antigens) purely on the basis of affinity is facilitated.

Details of solution based affinity selections are provided in the Examples herein.5 (F) LIGANDS.

Ligands suitable for monovalent surface expression on phage according to the present invention are many and varied. In fact any ligand which may be expressed upon the surface of a phage is suitable for use according to the method of the invention.

Preferred ligands for monovalent expression are antibodies, in particular dAbs as defined herein. The use of the method of the invention in the selection of high affinity antigen binding dAbs is described in detail in the examples provided herein. In an alternative embodiment of the invention, the ligand is a peptide. Examples of peptide ligands are also described in detail in the examples provided herein.

Suitable ligands for surface expression on phage include immunoglohulin molecules both synthetic and naturally occurring. Suitable immunoglohulin molecules for use according to the methods of the invention include any of those in the list consisting of the following: single domain antibodies (dAbs) which include both variable heavy chain domains and variable light chain domains, scFv, Fahs, Fe, chimeric antibodies, humanised antibodies, mutated and/or engineered antibodies, intrahodies and fragments of any of those listed above. In a preferred embodiment of the invention, the immunoglobulin molecule is a dAb (single variable domain antibodies).

Advantageously, the dAb is a heavy domain dAb (single heavy chain variable domain).

Examples of dAbs, in particular high affinity binding dAbs selected using the methods of the invention are provided in the Examples herein.

In an alternative embodiment of the present invention, the monovalent surface expressed ligand comprises one or more CDRs specific for one or more antigens or epitopes attached to a non-imm unoglobu I in scaffold. Suitable non-immunoglobulin scaffolds include for example natural bacterial receptors such as SpA which have been used as scaffolds for the grafting of CDRs to generate ligands which bind speciflcally to one or more epitopes. Details of this procedure are described in US 5,831,012.

Other suitable scaffolds include those based on fibronectin and affibodies. Details of suitable procedures are described in WO 98/58965. Other suitable scaffolds include lipocallin and CTLA4, as described in van den Beuken et al., J. Mol. Biol. (2001) 310, 591-60!, and scaffolds such as those described in WO/0069907 (Medical Research Council), which are based for example on the ring structure of bacterial GroEL or other chaperone polypeptides.

Protein scaffolds may he combined; for example, CDRs may be grafted on to a CTLA4 scaffold and used together with immunoglobulin V11 or V1. domains to form a ligand. Likewise, fibronectin, lipocallin and other scaffolds may be combined.

(F) ANTIGENS AND EPITOPES.

One skilled in the art will appreciate that the choice of epitopes and antigens for binding by one or more monovalently expressed ligands on a phage surface is large and varied. They may he for instance human or animal proteins, cytokines, cytokine receptors, enzymes co-factors for enzymes or DNA binding proteins. Suitable cytokines and growth factors include but are not limited to: ApoE, Apo-SAA, BDNF, Cardiotrophin- 1, EGF, EGF receptor, ENA-78, Eotaxin, Eotaxin-2, Exodus-2, EpoR, FGF-acidic, FGF-basic, libroblast growth factor-lO, FLT3 ligand, Fractalkine (CX3C), GDNF, G-CSF, GM-CSF, GF-J3l, insulin, IFN-y, IGF-I, IGF-l1, IL-la, IL-113, IL-2, IL-3, IL-4, IL-5, lL-6, IL-7, IL-8 (72 a.a.), IL-8 (77 a.a.), IL-9, IL-b, IL-Il, IL-l2, IL-l3, IL-15, IL-16, IL-17, IL-18 (IGIF), Inhibin a, Inhibin, IP-lO, keratinocyte growth factor-2 (KGF-2), KGF, Leptin, LIF, Lymphotactin, Mullerian inhibitory substance, monocyte colony inhibitory factor, monocyte attractant protein, M-C SF, MDC (67 a.a.), MDC (69 a.a.), MCP-1 (MCAF), MCP-2, MCP-3, MCP-4, MDC (67 a.a.), MDC (69 a.a.), MIG, MIP-la, MIP-13, MIP-3a, MIP-3f3, MIP-4, myeloid progenitor inhibitor factor-I (MPIF-1), NAP-2, Neurturin, Nerve growth factor, l- NGF, NT-3, NT-4, Oncostatin M, PDGF-AA, PDGF-AB, PDGF-BB, PF-4, RANTES, SDFIct, SDFI3, SCF, SCGF, stern ccli factor (SCF), TARC, TGF-ct, TGF4, TGF-132, TGF-3, tumour necrosis factor (TNF), TNF-a, TNF-1., TNF receptor I, TNF receptor 11, TNIL-l, TPO, VEGF, VEGF receptor I, VEGF receptor 2, VEGF receptor 3, GCP- 2, GRO/MGSA, GRO-, GRO-y, HCCI, 1-309, HER 1, HER 2, HER 3, HER 4, TACE recognition site, TNF BP-l and TNF BP-TT, whether in combination, in a different combination or individually. Cytokine receptors include receptors for the foregoing cytokines, e.g. IL-I RI; TL-6R; IL-IOR; IL- 18R, as well as receptors for cytokines. it will be appreciated that this list is by no means exhaustive.

The invention will now be described by the following examples which should he considered in no way limiting of the invention.

EXAMPLES.

All chemicals were purchased from Sigma unless otherwise stated.

Example 1.

Construction and expression of constructs used Construction of phage vector pDOM9 Genetic elements for the expression of domain antibodies in fusion to the N-terminus of gene III were based on the pDOM9 phage vector. This was constructed by inserting an oligonucleotide cassette harbouring a cleavage site for PreScission'TM protease (LEVLFQ) into the pDOM4 phage vector (previously modified in this laboratory from fd TET DOG (Clackson et at, 1991) to contain an N- terminal eukaryotic GAS leader sequence and C-terminal c-myc tag) to create a PreScissioniM site between the Cterminus of the dAb and the N-terminus of gill.

Briefly, pDOM9 was assembled by ligating the DNA duplex formed from the annealed phosphorylated oligonucleotides CESOO1 and CESOO2 into the gel purified Sal l/IVot I digested phage vector pDOM4. The phosphorylated oligonucleotides were annealed in a I 00jil volume at I M by heating to 95 C for 5 mm followed by gradual cooling to 25 C over 30 minutes. The annealed duplexes were then diluted to lOOnM and ljil hereof was added to I OOng digested pDOM4 and ligated overnight at 16 C in a 20pJ reaction volume using lp.l (400 units) T4 ligase. 3pi ligation mix was then used to transform 50tl TB1 E. coli clectrocompetent cells by eleciroporation. The sequences of the inserted fragments were verified by DNA sequencing of plasmid DNA m inipreps (QI Agen) prepared from overnight cultures.

Cloning of dAbs into pDOM9 Various cytokine binding dAbs were used in this study which are described in Figure For cloning of dAbs into pDOM9, the genes for the dAbs shown in Figure 1 were each excised from an expression vector by restriction digestion with Sal i/Not I (NEB). The gel purified Sal I/Not I cut fragments were then ligated into the Sal I/Not I digested backbone of pDOM9 in a i0jti reaction containing 200ng cut pDOM9, ôOng cut fragment and 2111 (800 units) T4 DNA ligase (NEB) at rt for lhr. lpi was then used to transform 50tl TGI E. colt electrocompetent cells as above.

Phage production and verification of antigen binding ability Phage were produced from the constructs above according to the following procedure.

200nil 2xTY media was inoculated with 2 ml saturated overnight culture of TGI cells harbouring the required pDOM9 construct and incubated for 20 hrs at 37 C, 250rpm.

The culture was then separated into 2x200ml centrifuge tubes (Corning) and centrifuged at 4000rpm for 30 miii in a Sorvall Legend RT Benchtop centrifuge at 4 C.

The supernatant was then transferred to 2x fresh 200m1 tubes containing 40m1 20% PEG (Fluka)/2.5M NaCI and incubated on ice for lhr. The phage suspensions were then centrifuged at 4000rpm for 30mm at 4 C and the supernatant discarded. The remaining precipitated phage pellets were centrifuged again at 2000rpm, 2mm 4 C and any remaining supernatant discarded with a pipette. The phage pellet was resuspended in 5ml Multi Q HO (Millipore), transferred to a 15 ml centrifuge tube (Corning) and centrifuged at 4000rpm, for 10 mm at 4 C to remove cellular debris. The supernatant was then filter sterilised through a 0.45RM single use filter (Sartorius) using a 5m1 syringe into a fresh 15 ml centrifuge tube containing a volume of 20% PEG/2.5M NaC1 equal to 1/5 volume of the filter sterilised supernatant and incubated on ice for miii. The precipitated phage suspension was centrifuged at 4000rpm for 20 miii at 4 C and the supernatant discarded. The phage pellet was resuspended in I mlix PBS 1 5 (ref), transferred to a I.SniI tube (Eppendorf) and centrifuged at 13, 000 rpm for 2 mm, 4 C (Eppendorfbenchtop centrifuge 5415D). The phage suspension was transferred to a fresh 1.5m1 tube and stored at -80 C in 20% glycerol (ref).

Typical yields were around l-2x1012 cfu/ml, and infectivity aroundl2x10'' tfu/ml.

Phage were tested for binding activity by ELISA on immobilised biotinylated cytokine 1. Briefly, hiotinylated antigen was coated onto a 96 well immunoplate (Maxisorb) by initial passive coating of the wells overnight with l00il neutravidin at 1.tg/ml, followed by application of 1001.11 biotinylated antigen at l5nM for ihr. The plate was blocked for lhr with 300pJ 1% BSA/IxPBS, washed 3 times in 0.1% Tween- 20/IxPBS and serial dilutions of lx 1010 phage applied to wells for lhr. Bound phage were then detected by incubation with HRP conjugated anti-M13 monoclonal antibody (Amersham Biosciences) diluted 1:2000 in 2% Tween 20/IxPBS for lhr. Signal was then visualised by addition of I 001.11 TMB, and the reaction quenched with lOOpA 1 M H2S04.

The results are shown in Figure 2 indicating that both phage bind to biotinylated cytokinc I and that the higher affinity binder gives a signal up to higher dilutions.

EXAMPLE 2

Model Selections using a single population ofphage Optirnisation of antigen concentration In this example model selections were performed without shaving' in order to build a profile of phage recovery versus concentration of antigen for DOMI-9 (KD=>lOjiM), DOM 1-122 (KD=5OnM), DOM 1-95-3 (KD=3nM), and DOM 5 (no affinity for cytokine 1). This was done to establish a concentration of antigen at which recovered titrcs of each phage population would be well above background levels. Model selections were performed as follows; Phage were diluted in PBS to lxl010/20tl and aliquoted into 1.5m1 tubes. Phage were then blocked by addition of 2 % Marvel" PBS to a final volume of 500il for 1 hour at rt. Biotinylated antigen was added to the blocked phage such that the phage antigen mixture retained a final concentration of 2% Marvel in PBS. The antigen/phage mix was rotated for I h at room temperature. Whilst the phage/antigen mix was incubating p1/selection of Streptavidin-coated paramagnetic beads (Dynal) were prepared by washing once in 0.1% PBS-T and a second time in PBS alone. The beads were washed by capturing the beads while the tubes were held in a magnetic rack (Dynal) and removing the wash buffer with a pipette. The blocked streptavidin-coated magnetic heads were captured with the magnet, the supernatant removed and the beads resuspended in the starting volume of PBS. 50 III beads were added the antigen/phage mix and this mixture rotated for 15 mm to allow for head capture of phage/antigen complexes. The beads/phage/antigen suspension was transferred to the first well of a Kingfisher nil (Thermo Electron Corporation) washing strip and beads washed repeatedly with 7x washes of lml 0.1% PBS-Tween 20, followed by lx ImI wash of PBS. Beads were then captured and the phage eluted through resuspension of the beads in 500 jil of trypsin solution (50 p.1 of 10 mg/nil trypsin stock solution added to 450 p.1 PBS, freshly diluted) and incubated for 15 mm at room temperature.

Recovered phage were then used to infect lOrnI log phase TG1. This was prepared by transferring a single bacterial colony from a minimal media plate into 5 ml of 2xTY medium. The culture was grown overnight at 37 C with shaking at 250 rpm. Thc ovcrnight culture was then used to inoculate I OOml fresh 2xTY medium by diluting 1: 100. The new culture was grown with shaking at 37 C until the 0D600 was between 0.3 - 0.6 (1.5 -2 h). lOmi log phase TGI was added to 250 p.1 eluted phage in a 50m1 falcon tube and incubated at 37 C for 30 mm in a water bath. The eluted titre was dctcrmined by preparing a 10 fold dilution series of infected TG1 from lOl to l0 in 2xTY, and plating 10p.l of each dilution onto 2xTY agar platcs containing tetracycline at I 5p.g/ml. Plates were then incubatcd ovcrnight at 37 C. The number of eluted phage from 500 j.ii was determined by multiplying the number of colonies in spot x dilution factor x 1000 x2.

Model selections were carried out as above using monoclonal populations of DOM 1-9, DOM1-122 and DOMI-95-3 phage at a range of antigen concentrations ranging from 0.OlnM-IOOnM biotinylated cytokine I (Figure 3). The background level of non- specific recovery was around l-3x 105 tfu depending on cytokine I concentration. At lOnM all populations of phage gave a recovery of >10 fold above the background level. This was therefore chosen as the antigen concentration to use for shaving experiments.

Optimisation of PreScissioniM shaving' conditions The components of the shaving' conditions including protease concentration, length of incubation, and temperature were optirnised to isolate a concentration range of PreScission'TM within which optimal shaving could be achieved so that the majority of phage would contain no dAb-gill fusion proteins and those with dAb-gill fusions remaining on the surface would he monovalent.

Therefore lx 10' DOM2-5-l9 phage were diluted into a final volume of 20p. l PBS as above. Samples were treated with lnM, lOnM, 5OnM, lOOnM, 200nM, 500nM 1p.M, 2p.M, and 4.8p.M final concentration PreScission'TM by addition of 2.2p.l lOx PreScission'TM diluted in IxPBS/lnim DTT incubated for 30 mm at 20 C. The reaction was stopped by addition of 30p1 Pefabloc Sc AEBSF (Roche Diagnostics). After shaving, phage were analysed for binding ability by ELISA on biotinylated TNF-a as described above except for that coating with hiotinylated TNF-a was performed using 50p.l/well antigen at 25OngIml in PBS for lhr at 37 C. The results are shown in Figure 4. Shaving with PreScissioniM appeared to have little or no effect on the ELISA signal at or below 5OnM (Figure 4 lanes 1-3), whilst signal reduction appeared to plateau above I iM (Figure 4 lanes 7-9). Therefore shaving was focused in the range of I OOnM- 1 jiM.

The effect of applying PreScission'TM within the range lOOnM-ljiM on the observed recovered phage litres was tested by performing soluble selections using either lx 1010 DOM1-9 or DOM1-122 phage at lOnM biotinylated cytokine 1. Prior to selection, phage were shaved using PreScissionhM at a final concentration of OnM, 200nM, 500nM 700nM and l000nM as above. Results are shown in Figure 5.

Without shaving, the difference between the recovered titres of DOM1-9 (KD>l0M) and DOM1-122 (KD=5OnM) is only around 10-fold, despite the fact that the affinity of DOM1-9 is at least 100 tinies lower than that of DOMI-122. However a greater difference was seen in the ratio of recovered DOM1-122 to DOM1-9 after treatment with PreScissioniM at 200nM or 500nM (40 fold) indicating that shaving has reduced the potential for multivalent binding interactions.

EXAMPLE 3

Model selections using a mixed population ofphage A model selection on a solid support A selection was performed on a 1:100 mix of DOM1-122 (KD=5OnM): DOM1-9 (KD>l0M) on a solid support at lOOnM cytokine 1. This entailed blocking and washing of beads as described previously. The bead suspension was then diluted I in 2 in PBS and incubated with biotinylated antigen at a final concentration of lOOnM for 1 5 mm with rotation. The supernatant containing unbound antigen was then removed by capturing the beads while the tubes were held in a magnetic rack and the beads resuspended in 50tl PBS and added to the blocked phage. The remainder of the selection was performed as described previously.

Selections were performed at a range of PreScission FM concentrations (50, 100, 200, 300, 400, 500 700 nM). Based on titres, 40 colonies from the undigested, 200nM, and 500nM PreScissionTM treated plates were screened by colony PCR amplifications with two primer pairs (in separate reactions), one pair specific for DOMl-9 and one pair specific for DOM1-122. This was done in order to be able to identify whether colonies contained DOMI-9 or DOMI- 122 phage. Primer pairs used were CESOO9 (DOMI-9 specific) /CESOI 1 and CESOIO (DOM1-122 specific)/CESOI 1. The results are shown below.

Starting phage ratio PreScissionTM DOMI-122 DOM1-9 DOM1-122: DOMI-9 Concentration positive colonies positive colonies Undigested 1/40 40/40 1:100 200nM 6/40 36/40 500nM 8/39 - 34/39 The number of DOM1-122 positive colonies increased with PreScission'TM concentration (Undigested = 1/40, 200nM = 6/40, 500nM = 8/40) which indicated that PreScission'TM cleavage was effective and allowed increased enrichment of a moderate affinity hinder (DOM1-122) over a weak affinity binder (DOMI-9) after selection on solid support. It is thought that the number of positives given by DOMI-122 and DOM 1-9 does not come to the same total due to double infectivity.

A soluble model selection of a moderate affinity binder (DOMI-122) over a weak affinity binder (DOMI Soluble model selections were performed as above using a mix of 1:100 DOM1- 122(KD=5OnM): DOMI-9 (KD>10pM) phage and an antigen concentration of lOnM soluble biotinylated cytokine 1. Results are shown in the table below.

Starting phage ratio PreScisslonTM DOM i-i 22 DOM 1-9 DOM 1-122: DOM1Concentration positive colonies positive colonies Undigested 4/40 3 8/40 1:100 ____________ ______ _____ ______________ 200nM 23/40 21/40 There is a clear increase in the number of DOM1-122 positives in the selection pretreated with PreScission'TM froni 4/40 to 23/40. These results show that PreScission'TM cleavage allowed the enrichment of a moderate affinity binder over a weak affinity binder under soluble model selection conditions.

Soluble model selections of a high affinity binder (DOM1-95-3) over a moderate affinity binder (DOMI-122) Soluble model selections were performed as above using a mix of 1:1 and 1:4 DOM 1- 95-3 (KD=3nM): DOMI-122 (KD=5OnM) phage an antigen concentration of lOnM cytokine 1. The results are shown in the table below.

Starting phage ratio PreScissioniM DOM 1-122 DOM 1-95-3 DOM1-95-3: DOMIConcentration positive colonies positive colonies Undigested 35/40 10/40 ______ 1:1 200nM 32/39 8/39 500nM 4/40 36/40 - Undigested 34/38 9/40 1:4 200nM 36/40 5/40 500nM 24/40 16/40 When mixed 1:1, the undigested and 200nM phage selections were dominated by DOM1-122. However when phage were treated with 500nM PrcScission'TM the situation was reversed and DOM 1 -95-3 was almost totally dominant. The 1:4 starting mix also showed the same general pattern; dominance of 1-122 in the undigested and 200nM selections then the emergence of 1 -95-3 when 500nM was used.

These results show that shaving phage' by PreScission'TM cleavage can effectively select the highest affinity dAb from a mixed population of high affinity and moderate affinity dAbs in a soluble model selection.

EXAMPLE 4

Estimation of the percentage of monovalent phage after shaving by Western Blotting For all analyses lxlO' phage (unless otherwise stated) were digested with a given concentration of PreScisslonTM for 30 miii at 20 C as described above followed by heating at 70 C for 10 miii in the presence of lx NuPAGE SDS sample buffer (Invitrogen) and 50mM DTT. Samples were then analysed by SDS PAGE on a NuPAGE Novex 4-12% gel (Invitrogen) using MES buffer (Invitrogen) and blotted onto PVDF according to the manufacturer's instructions. Blots were probed using anti- c-myc antibody (clone 9E10) at 1:2000 in 2% Marvel/PBS for lhr, washed with 3 changes of 0.1 % Tween-PBS over 30 minutes, then developed using the Western Breeze alkaline phosphalase conjugated anti-mouse secondary antibody and chromogen (Invitrogen) according to the manufacturor's instructions. PreScissioniM protease cleaves between the dAb and gill in the fusion construct but the cleavage site is before (N-terminal to) the niyc tag. Therefore both the free gill and the dAb-gIll fusion retain the myc tag and as such appear as bands of around 54 kDa and 70 kDa respectively by Western blotting with mouse anti-c-myc antibody and anti mouse- alkaline phosphatase. Therefore cleavage of the dAb-fusion protein by PreScissionrM can be visualised by Western blotting by the reduction in intensity (or loss) of the heavier fusion protein hand after shaving.

Initially in a preliminary experiment lx 10' DOM3-5-l phage were digested with high concentrations of PreScissionTM. The absence of the fusion protein band in all the samples in the range of 1-4.8jiM final concentration PreScissionTM indicated that shaving was occurring (Figure 6).

The concentration of PreScissionTM was then lowered to that used in previous experiments (i.e. undigested, 200nM, 500nM, I jiM) to gain more information about the level of cleavage occurring (Figure 7). The bands in Figure 7 were analysed by densitometry using Imagequant software (Amersham Biosciences) to give values for the upper fusion protein band and the lower gene Ill band as percentages of the whole signal. 1 0

Band Undigested 200nM 500nM I jiM Top 28% 14% 3% Not (Fusion protein) measurable Bottom 72% 86% 97% "100%" (Free gene III) Total 1,182,681.7 1,290,342.5 1,121,337.8 1,188642.6 This data was then used to predict the percentage of phage present carrying 0-5 dAb- pill fusions on the phage surface using the following equation: (l-f)"x. (f)\y} . C. 100 Where T is the fraction of dAb-pIll fusion (between 0 and 1) as determined by Western blot, and where y' is an integral number between 0 and 5 representing the number of dAbs on a phage (there are max. 5 copies of pill per phage). Where x' = 5 - y' and C is the number of combinations (or arrangments) of dAb-pill and pill on phage: I when y' is 0 or 5 when y' is I or 4 l0when'y'is2or3 Results are shown in the table below and represented graphically in Figure 8.

Number of Percentage p111-dAb fusion (100 %=1, 0% =0, i.e 28% p111-dAb per written as 0.28) phage 28% 14% 3% 0% 0 19.34917632 47.0427 85.8734 100 37.6233984 38.29057 13.27939 0 2 29.2626432 12.4667 0.821406 0 3 11.3799168 2.029462 0.025404 0 4 2.2127616 0.165189 0.000393 0 0.17210368 0.005378 2.43E-06 0 From this data it is estimated that at SOOnM PreScission1M about 85 % of phage would he bald, 13% of phage would be monovalent and 0.8 % would be bivalent.

EXAMPLE 5

Soluble model selections of a high affinity binder (DOM4-53-99) over a low affinity binder (DOMIO-53) and evidence of shaving of the DOM4-53-99 phage by Western blot Soluble model selections were performed as in Example 3 above using a mix of 1:25 DOM4-53-99 (KD=lnM):DOM4-53 (KD=100- 300nM) phage at an antigen concentration of 4nM soluble biotinylated cytokine 4. Results are shown in the table below.

Starting phage ratio PreScissioniM DOM4-53 DOM4-53-99 DOM4-53-99: Concentration positive colonies positive colonies DOM4-53 Undigested 18 6 1:25 200nM 9 16 500nM 1 23 It is clear from these results that shaving at200nM and 500nM PreScission1M before selection increases the proportion of DOM4-53-99 phage (the higher affinity binder) recovered from the selection.

Western blotting of the DOM4-53-99 phage used in this experiment was also conducted (as in Example 4). Briefly, Ix 10' phage were digested with PreScission'TM at concentrations of' OnM, 200nM, 500nM and I tM. Samples were analysed by SDS PAGE, blotted into PVDF and detected using mouse anti-c-myc antibody followed by anti mouse HRP as described previously. Shaving at all concentrations can clearly been seen to reduce the level of dAb-fusion protein visible as shown in Figure 9.

All publications mentioned in the present specification, and references cited in said publications, are herein incorporated by reference. Various modifications and variations of the described methods and system of the invention will be apparent to those skilled in the art without departing from the scope and spirit of the invention.

Although the invention has been described in connection with specific preferred embodiments, it should be understood that the invention as claimed should not be unduly limited to such specific embodiments. Indeed, various modifications of the described modes for carrying out the invention which are obvious to those skilled in molecular biology or related fields are intended to be within the scope of the following claims.

Claims (31)

CLAIMS.

1. A method for the preparation of a phage display library, which library facilitates the selection of high affinity binding phage (a shaved phage library), which method comprises the step of: (a) Providing a library of phage expressing one or more copies of a proteinaceous ligand of interest on a plurality of phage surfaces; wherein the ligand is attached to phage coat protein and wherein a cleavable site is located within the expressed ligand- coat protein, (b) Treating the phage library according to step (a) with a cleaving agent such that a small proportion of the treated phage are monovalent with respect to the presence of surface expressed proteinaceous ligand, and the majority of the phage comprise no surface expressed proteinaceous ligand.

2. A method for the selection of phage from a phage display library which method comprises the step of: (a) Providing a library of phage expressing one or more copies of a proteinaceous ligand of interest on a plurality of phage surfaces; wherein the ligand is attached to phage coat protein and wherein a cleavable site is located within the expressed ligand- coat protein, (b)Treating the phage library according to step (a) with a cleaving agent such that a small proportion of the treated phage are monovalent with respect to the presence of surface expressed proteinaceous ligand, and the majority of the phage comprise no surface expressed proteinaceous ligand (shaving the phage) and (c)Selecting from the library treated according to step (a) and (b) above those proteinaceous ligand expressing phage which are capable of specifically binding to one or more binding partners.

3. A method according to claim 1 or claim 2 wherein the shaved phage library does not comprise phagemids.

4. A method according to any of claims 1 to 3 wherein the coat protein is bacteriophage gene III coat protein.

5. A method according to any preceding claim wherein the ligand is either an immunoglobulin molecule, an affibody or a ligand which comprises, preferably consists of CDRs grafted onto a non-immunoglobulin scaffold.

6. A method according to claim 5 wherein the immunoglobulin is an antibody molecule.

7. A method according to claim 5 or claim 6 wherein the antibody is any of those in the group consisting of the following: scFv and dAb (single variable domain).

8. A method according to claim 7 wherein the antibody is a dAb.

9. A method according to claim 8 wherein the antibody is a heavy chain variable domain dAb.

10. The method according to claim 8 wherein the antibody is a light chain variable domain dAb.

11. A method according to any of claims S to 8 wherein the immunoglobulin molecule binds to target/antigen with a dissociation constant (Kd) of between 0.01 and 500nM.

12. A method according to any preceding claim wherein the cleaving agent is a protease.

13. A method according to claim 12 wherein the protease is any of those in the group consisting of the following: PreScission'TM; factor Xa., Thrombin.

14. A method according to any preceding claim wherein the cleavable/cleavage site is any one of those in the group consisting of the following: LEVLFQGP (PreScissionTM cleavage site); IEGRGI (factor Xa cleavage site); and LVPKGS (thrombin cleavage site).

15. A phage particle which comprises monovalently on its surface, a ligand of interest attached to the phage gene III coat protein peptide, wherein the phage particle comprises a cleavable site within said surface expressed ligand- coat protein peptide.

16. A phage particle according to claim 15 wherein the cleavable/cleavage site is any one of those in the group consisting of the following: LEVLFQGP (PreScissionTM cleavage site); IEGRGI (factor Xa cleavage site); and LVPKGS (thrombin cleavage site).

17. A phage particle according to claim 15 or claim 16 wherein the ligand is either an immunoglobulin molecule, an affibody or a ligand which comprises, preferably consists of CDRs grafted onto a non-immunoglobulin scaffold.

18. A phage particle according to claim 17 wherein the immunoglobulin is an antibody molecule.

19. A phage particle according to any of claims 15 to 18 wherein the antibody is any of those in the group consisting of the following: scFv and dAb (single variable domain).

20. A phage particle according to claim 19 wherein the antibody is a dAb.

21. A phage particle according to claim 20 wherein the antibody is a heavy chain variable domain dAb.

22. A phage particle according to claim 20 wherein the antibody is a light chain variable domain dAb.

23. A phage particle according to any of claims 15 to 22 wherein the immunoglobulin molecule binds to target/antigen with a dissociation constant (Kd) of between 0.01 and SOOnM.

24. A shaved phage library in which each member of the library is capable of expressing on their surface one or more copies of one or more ligands of interest attached to the phage coat protein, wherein each phage particle comprises a cleavable site within each surface expressed ligandcoat protein and wherein 0.1% or more of those library members comprises a monovalent surface ligand-gene III coat protein peptide.

25. A shaved phage library according to claim 24 wherein the phage coat protein is gene III coat protein.

26. A shaved phage library according to claim 24 or claim 25 wherein the cleavable site is a naturally occurring site.

27. A shaved phage library according to claim 24 or claim 25 wherein the cleavable site is an engineered site.

28. A shaved phage library according to any of claims 24 to 27 wherein the library comprises any one or more of the features of any of claims 13 to 21.

29. The use of a phage library according to any of claims 24 or 28, or a phage particle according to any of claims 13 to 21 in the selection of one or more ligands capable of interacting with a binding partner, preferably interacting with a binding partner with high binding affinity

30. A method for producing a proteinaceous ligand capable of interacting with a specific binding partner which method comprises: (a) Performing the method according to claim 2; (b) Isolating from separated phage particles recovered according to the method of claim 2 nucleic acid encoding the proteinaceous ligand; (c) Inserting nucleic acid encoding the proteinaceous ligand, or a fragment or derivative thereof with binding specificity for the binding partner, in a recombinant system; and (d) Producing the proteinaceous ligand, or fragment or derivative thereof with binding specificity for the binding partner, in the recombinant system separate from bacteriophage particles.

31. A method according to claim 30 wherein the proteinaceous ligand has any one or more of the features of claims 5 to 11.