NON-ANTIBODY SCAFFOLD PROTEIN FUSIONS PHAGE DISPLAY VIA FUSION TO plX OF M13 PHAGE

CROSS-REFERENCE TO RELATED APPLICATIONS [ l ] This application claims the benefit of U.S. Provisional Application No. 61/014,778, filed December 19, 2007, which is incorporated by reference in its entirety.

FIELD OF THE INVENTION

[2] The invention relates to a compositions and methods for generating and using plX phage display libraries for producing non-antibody scaffold protein fusions using plX of M13 phage.

BACKGROUND OF THE INVENTION

[3] Filamentous phage display using pill, pVII, pVIII, plX, and combinations thereof as fusion partners in phage or phagemid systems have been used as a technology for protein engineering, notably for de novo protein and peptide isolation and affinity maturation (see, for example, Janda et al, U.S. 7,078,166). Random peptides and proteins can be generated and isolated from phage libraries displaying variations of the peptide sequences via panning against a protein target of interest. Previously used human protein de novo libraries have been created synthetically. In a synthetic library, putatively useful DNA sequences encoding potential proteins are designed and synthesized based on known sequences or motifs. In addition to the synthetic peptide library, libraries can also be created by combinatorial cloning of protein encoding DNA derived from human tissues. Such libraries have been used for providing potentially useful peptides and for running successive rounds of panning and maturation or modification to attempt to find non-antibody peptide or protein peptides that have desired properties such as inhibitory biological activity of a selected target protein.

[4] Human peptides and proteins, such as mimetic proteins or known protein muteins that bind or mimic target proteins have been isolated from phage display pill or pVIII peptide libraries. Although successful at isolating some proteins that bind to specific targets, such phage display library approaches suffer from the problems of having to repeat the process of library generation, requirement of panning and maturation several times over to isolate peptides or proteins having the desired characteristics, and other know limitations. Such phage libraries also suffer from the problem that they do not fully encompass or mimic the range of peptide diversity present in

humans, the position and extent of amino acid variation, and the relative abundance of biologically active proteins from different human genes. Deviation of synthetic proteins from the natural repertoire may increase the risk of unfavorable biochemical properties and of immunogenecity if used as therapeutics in man, and this issue could potentially be addressed by how the sequences are generated and screened for selection from such libraries.

[5] Monoclonal antibodies are the most widely used class of therapeutic proteins when high affinity and specificity for a target molecule are desired. However, non-antibody antigen-binding peptides or proteins can be engineered to bind to such targets that are also of high interest for use in therapeutics or diagnostics. Such proteins or peptides that are capable of binding to biomolecules may have several potential advantages over traditional antibodies such as, but not limited to, smaller size, lack of disulphide bonds, ability to be expressed in prokaryotic hosts, novel methods of purification, high stability, ease of conjugation to drugs/toxins, and intellectual property advantages, among others.

[6] One type of scaffold is the immunoglobulin (Ig) fold. This fold is found in the variable regions of antibodies, as well as thousands of non-antibody peptide or protein proteins. It has been shown that one such Ig protein, the tenth fibronectin type III repeat from human fibronectin, can tolerate a number of mutations in surface exposed loops while retaining the overall Ig-fold structure. Thus, libraries of amino acid variants have been built into these loops and specific binders selected to a number of different targets. Such engineered Fn3 domains have been found to bind to targets with reasonably high affinity, while retaining important biophysical properties [7] Prior use of phage libraries has included antibody-based protein fusion libraries. However, there is a need for synthetic alternative scaffold or non-antibody putative antigen-binding peptide or protein fusion libraries and methods that simultaneously deliver the critical elements of human therapeutic antibodies of high affinity and activity, high productivity, good solution properties, and a propensity of low immune response when administered in man. There is a further need to increase the efficiency of non-antibody peptide or protein isolation from synthetic libraries, relative to current methods, to reduce the resource costs of non-antibody peptide or protein discovery and accelerate delivery of antibodies for biological evaluation. The libraries and methods of this invention meet these needs by coupling comprehensive design, assembly technologies, and phage plX non-antibody peptide or protein display.

SUMMARY OF THE INVENTION

[8] In contrast to the teaching of the prior art, it has now been discovered that pVII and plX can successfully be used for generating high affinity non-antibody peptide or protein libraries using plX from M13 phage, e.g., using mutagenesis or other diversity producing techniques, optionally with in line maturation, to provide an efficient and fast platform for non-antibody peptide or protein and non-antibody peptide or protein fragment generation and selection of therapeutic antibodies. According to the present invention, non-antibody peptide or proteins that are capable of binding to a desired biomolecule or antigen are fused to pVII and plX engage in a dynamic interaction on the phage surface to display a functional non-antibody peptide or protein, optionally in a representative heterodimeric motif. The display on phage of non-antibody peptide or protein binding agents is therefore a suitable and preferred method for display and assay of diverse libraries of combinatorial heterodimeric arrays in which members can function as monomeric or dimeric artificial non-antibody peptide or protein species and allow for selection of novel or desired biological activities.

[9] The present invention provides designs and display of non-antibody peptide or protein de novo libraries fused to the plX protein of filamentous phage, a phage surface protein that is different from the widely used the pill and pVIII proteins. We constructed the individual scaffold libraries separately and developed phage selection processes to systematically examine the activity of each of the scaffold libraries, and evaluate the structural topologies for antigen recognition.

[10] The present invention provides various improved and new plX and pVII phage display de novo library generation methods and components, such as but not limited to, one or more of (i) designed and displayed non-antibody peptide or protein de novo libraries fused to the plX or pVII phage proteins; (ii) the use of a phage surface protein different from the widely used pill and pVIII of M13 phage; (iii) the use of the plX phage display system to screen a library of peptides fused to a non-antibody protein scaffold; (iv) non- antibody peptide or protein selection processes that allow systematic examination of the effect of the designed sequences and structural topologies for antigen recognition; (v) a streamlined affinity maturation and in line maturation process as a part of the library selection. Such a new system of library design, selection, optimization and maturation of individual or groups of libraries provide a reproducible and reliable system for successful non-antibody peptide or protein de novo discovery and also facilitate understanding the structure function relation of non-antibody peptide or protein to antigen interaction. [11 ] The human non-antibody peptide or protein de novo library described above is distinct from current antibody library state-of-the-art by its display via the plX gene of M13 phage. Non-antibody peptides or proteins can be successfully displayed on the surface of M13

phage as plX fusion proteins according to the present invention. Both scaffold proteins can be engineered to bind a specific protein target that is not bound by the corresponding scaffold proteins in their native state. These engineered scaffolds retained binding to this specific target while displayed on phage. In addition, libraries of amino acid variants can be made according to the invention in each scaffold, and their ability to bind a specific target addressed by displaying library members on phage as plX fusions and panning against a target protein.

[12] Artificial antibodies or scaffold proteins as used in the present invention are herein defined as protein motifs of large diversity that use the functional strategy of the non- antibody peptide or protein molecule, but can be generated free of in vivo constraints, including (1 ) sequence homology and toxicity of target antigens; (2) biological impact of the generated non-antibody peptide or protein in the host or in hybridoma cultures used to recover the non-antibody peptide or protein; and (3) screening versus selection for desired activity. [13] Thus the invention describes a combinatorial phage display format for construction of highly diverse monomeric or heterodimeric polypeptide arrays. In particular, the invention describes a filamentous phage particle encapsulating a genome encoding a fusion polypeptide, wherein the fusion polypeptide comprises a non-antibody scaffold protein fused to the amino terminus of a filamentous phage pVII or plX protein. Preferably, the phage particle comprises the expressed fusion protein on the surface of the phage particle.

[14] In a related embodiment, the invention describes a vector for expressing a fusion protein on the surface of a filamentous phage comprising a cassette for expressing the fusion protein. The cassette includes upstream and downstream translatable DNA sequences operatively linked via a sequence of nucleotides adapted for directional ligation of an insert DNA, i.e., a polylinker, where the upstream sequence encodes a prokaryotic secretion signal, the downstream sequence encodes a pVII or plX filamentous phage protein. The translatable DNA sequences are operatively linked to a set of DNA expression signals for expression of the translatable DNA sequences as portions of the fusion polypeptide. In a preferred variation, the vector further comprises a second cassette for expressing a second fusion protein on the surface of the filamentous phage, wherein the second cassette has the structure of the first cassette with the proviso that the first fusion protein expression cassette encodes pVII protein and the second fusion protein expression cassette encodes plX protein. The vector is used as a phage genome to express heterodimeric protein complexes on the surface of the phage particle in which the two polypeptides of the heterodimer are anchored on the phage particle by the fusion to the first and second phage proteins, pVII and plX, respectively.

[15] In another embodiment, the invention contemplates a library of phage particles according to the present invention, i.e., a combinatorial library, in which representative particles in the library each display a different fusion protein. Where the particle displays a heterodimeric protein complex, the library comprises a combinatorial library of heterodimers. Preferred libraries have a combinatorial diversity of at least 10 ³ , 10 ⁴ , 10 ⁵ ,

10 ⁶ , 10 ⁷ , 10 ⁸ , 10 ⁹ , 10 ¹⁰ , 10 ¹¹ , 10 ¹² , 10 ¹³ , or any range or value therein, of different or similar species of fusion proteins.

[16] A related embodiment describes a fusion protein comprising first and second polypeptides wherein the first polypeptide is a non-antibody scaffold protein and the second polypeptide is a filamentous phage pVII or plX protein, wherein the non-antibody scaffold protein is fused to the amino terminus of the filamentous phage protein.

[17] Still further, the invention contemplates a variety of methods for producing a combinatorial library of phage, including by cloning repertoires of genes encoding a non- antibody scaffold protein into a vector of the present invention, modifying the structure of the a non-antibody scaffold protein in a library by mutagenesis, by random combination of populations of first and second fusion protein libraries, by target and affinity selection ("panning") to alter the diversity of a library, and the like. Such an embodiment can include a fusion polypeptide having an alterable non-antibody scaffold protein fused to a second polypeptide, as exemplified herein. For example, in one embodiment, the alterable non-antibody scaffold protein can be TeFN3, and the second polypeptide can be plX, where the F:G loop of TeFN3 is altered by mutagenesis or targeted substitution of the native F:G loop amino acid residues with a non-native polypeptide, such as the cysteine-constrained EGFR binding peptide, PHPEP190. In alternative embodiments, the alterable non-antibody scaffold protein can have an F:G loop made of at least one polypeptide identified by SEQ ID NOs: 2 - 23, 25, or 28. Disclosed herein are also embodiments where the non-antibody scaffold protein is encoded by an engineered nucleic acid phage vector and binds to a biomolecule, such as epidermal growth factor receptor or a biologically active ligand.

[18] The design of proteins with improved or novel functions is an important goal with a variety of medical, industrial, environmental, and basic research applications. Following the development of combinatorial non-antibody peptide or protein libraries, a powerful next step is the evolution toward artificial non-antibody peptide or protein constructs as well as other protein motifs in which dimeric species are native or might be functional.

[19] The present invention addresses these challenges by providing a phage-display format for the construction of combinatorial non-antibody polypeptide arrays in which pVII and plX are utilized for the display of fusion proteins that form monomeric or dimeric species.

It is important to note that this is an entirely new methodology because one can

independently display one or two protein motifs in close proximity to generate a library of functional interactions using expression on plX or pVII.

[20] Furthermore, sequence randomizations to form libraries and chain-shuffling protocols to form hybrid species can lead to subsets of novel proteins. For instance, the display and modification of arrays of zinc-finger domains in homodimeric or heterodimeric form produces structures that possess specific DNA interactions. In addition, entirely new constructs are possible via the insertion of a desired encoding fragment within a preformed scaffold such as a non-antibody peptide or protein chain. Possible insertions include an enzyme signature sequence or a repressor binding protein. [21 ] It should be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention as claimed.

DESCRIPTION OF THE FIGURES

[22] Figure 1 shows structures of the third Fn3 domain of tenascin (residues 802-893, PDB = 1 TEN, (Leahy, et al., Science 258: 987-91 , 1992)) and the I27 domain of titin (residues

5253-5341 , PDB = 1 TIT, (Improta, et al., Structure 4: 323-37, 1996)). Arrows point to the location of the F:G loop ( residues 876-881 and 5258-5259 of tenascin and titin respectively), which was modified to accommodate the EGFR binding sequence.

[23] Figure 2 shows graphical results for the display of tenascin and titin scaffold proteins on plX.

[24] Figure 3 shows graphical results of panning for EGFR binding constructs from TeFn3- 190 and Ti27-190 libraries displayed on plX.

DETAILED DESCRIPTION OF THE INVENTION

[25] The present invention provides various new phage display de novo library generation methods and components, such as but not limited to (i) designed and displayed non- antibody peptide or protein libraries fused to the plX or other phage proteins; (ii) the use of a phage surface protein different from the widely used pill and pVIII of M13 phage; (iii) use of such phage components as the library scaffold to provide improved designed, combinatorial diversities in the non-antibody peptide or protein selection processes that allow systematical examination of the effect of the designed sequences and structural topologies for antigen recognition; (iv) a streamlined affinity maturation process as a part of the library selection. Such a new system of library design, selection, optimization and maturation of individual or groups of libraries provide a reproducible and reliable system for successful non-antibody peptide or protein de novo discovery and also facilitate

understanding the structure function relation of non-antibody peptide or protein to antigen interaction.

[26] The human non-antibody peptide or protein de novo library described above is distinct from current antibody library state-of-the-art by its displaying via plX or pVII gene of M13 phage.

Definitions:

[27] Fusion Polypeptide: A polypeptide comprised of at least two polypeptides and a linking sequence to operatively link the two polypeptides into one continuous polypeptide. The two polypeptides linked in a fusion polypeptide are typically derived from two independent sources, and therefore a fusion polypeptide comprises two linked polypeptides not normally found linked in nature.

[28] Cistron: Sequence of nucleotides in a DNA molecule coding for an amino acid residue sequence and including upstream and downstream DNA expression control elements.

[29] Biomolecule: Any organic molecule produced by a living organism, or any organic molecule made in vitro by processes used by living organisms to produce biomolecules, such processes include, for example, transcription, translation, bio-organic chemical reactions and the like.

[30] Non-antibody scaffold protein: Any non-antibody protein, protein segment, or peptide having six or more beta strands connected by surface-exposed loops that form at least two beta sheets. Examples of such non-antibody protein scaffolds include proteins that include a fibronectin type III domain or an Ig domain, such as the third fibronectin type III domain of human tenascin (TeFN3) or the I27 domain of human titin (Ti27).

Filamentous Phage

[31 ] The present invention contemplates a filamentous phage comprising a matrix of proteins encapsulating a genome encoding a fusion protein (protein). The fusion protein comprises a non-antibody scaffold protein portion fused to the amino terminus of a filamentous phage pVII or plX protein.

[32] The filamentous phage will further contain the fusion protein(s) displayed on the surface of the phage particle, as described in the Examples. [33] In a fusion protein present on a phage of this invention, the "fusion" between the non- antibody scaffold protein and the filamentous phage pVII or plX protein may comprise a typical amide linkage, or may comprise a linker polypeptide (i.e., a "linker") as described in the Examples. Any of a variety of linkers may be used which are typically a stretch of about 5 to 50 amino acids in length. Particularly preferred linkers provide a high degree of mobility to the fusion protein at the point of the linker.

[34] Library design: prior synthetic libraries have incorporated some of the following, but none have included all in a comprehensive manner.

[35] Expression, biochemical, and biophysical properties. Preferred non-antibody peptides or proteins not only have desired biological and binding activities, but also are efficiently produced from a variety of hosts, are stable, and have good solution properties. It is understood that the disclosed scaffold proteins, either expressed as a fusion protein by an engineered recombinant nucleic acid phage vector or expressed as a scaffold alone, have the ability to bind biomolecules such as antigens, receptors, ligands, cell surface protein markers and the like. In preferred embodiments the scaffold proteins described herein bind epidermal growth factor receptor. High-frequency germline gene usage also indicates good expression in mammalian systems. In addition, such fusion proteins recovered from libraries by bacterial phage display methods of selection or screening should be expressed well in the bacterial host. The libraries of the invention are based on human germline derived templates that are well-expressed and purified from standard recombinant mammalian hosts (e.g. HEK 293 and CHO cells) as well as bacterial hosts, and have high stability and good solution properties.

[36] Maturation. The large number of positions in the V-region mimicking alternative scaffold sequences that can impact recognition of antigen or ligand binding, coupled with potential variation of up to 20 different amino acids at each position preclude the practicality of including all variations in a single library. Human antibodies achieve high affinity and specificity by the progressive process of somatic mutation. The libraries of the invention are designed and ordered to permit parallel selection and targeted variation while maintaining the sequence integrity of each non-antibody peptide or protein chain such that they reflect features and characteristics similar to those of human antibodies. [37] Alternatives design. The above design simulates characteristics of natural human antibodies. The modular nature of the system is amenable to incorporation of any collection of amino acids at any collection of positions.

[38] Library assembly technologies. Preferred non-antibody peptide or protein libraries are of low or high diversity (> 10 ¹⁰ ), amenable to alteration, and easy to assemble and have a low background of undesired sequences. These background sequences include parental template and low-targeted diversity. Coupling the following methods accelerates library assembly and leads to low background, (a) Kunkle-based single- stranded mutagenesis; (b) Palindromic loop with restriction site; (c) Megaprimer-based PCR. [39] plX non-antibody peptide or protein phage display. The combination of plX with the selected non-antibody peptide or protein templates is an efficient selection system for

recovering non-antibody peptides or proteins that retain their selected properties upon conversion into other related molecules.

[40] Phagemid display. The expressed molecule is large relative to the phage plX coat protein and thus may interfere with assembly of recombinant phage particles if linked to all plX proteins produced in the bacterial cell. One approach to by-pass this interference is to use a plX phagemid system, such as those known in the art are as described herein, whereby non-antibody peptide or protein-linked plX or pVII proteins can be incorporated into the recombinant phage particle. In a preferred application, libraries of the current invention are displayed by plX in a phagmid system. [41 ] Phage coat protein plX for display. Like pill, plX is present at low copy number on the phage and is amenable to affinity selection of displayed non-antibody peptides or proteins. However, the pill protein is critically involved in the infection process and proteins displayed on this protein can interfere with the efficiency of infection. The libraries of the current invention displayed on the plX protein are predicted to be efficiently replicated and presented for selection and/or screening.

[42] Non-antibody peptide or protein-plX expression. One approach to screening non- antibody peptides or proteins recovered from phage libraries is to remove the phage coat protein that is linked to the non-antibody peptide or protein molecule for display. The small size of the plX protein provides the option of production of screening of non- antibody peptide or proteins directly without this step.

[43] Therapeutic uses. As described herein, the disclosed scaffolds may be used as alternatives to antibodies. Accordingly, the disclosed scaffolds can have therapeutic applications. As such, it is contemplated that the scaffold proteins described herein may be used to form therapeutic compositions. One such composition can include a scaffold as described herein and a pharmaceutically acceptable carrier. In another embodiment the composition includes an isolated EGFR-specific, non-antibody protein scaffold such as TeFN3, having its F:G loop replaced with a non-native polypeptide, and a pharmaceutically acceptable carrier. In alternative embodiments, the composition includes an alterable non-antibody scaffold protein having an F:G loop made of at least one polypeptide identified by SEQ ID NOs: 2 - 23, 25, or 28 and a pharmaceutically acceptable carrier. Disclosed herein are also embodiments where the non-antibody scaffold protein is encoded by an engineered nucleic acid phage vector and binds to a biomolecule, such as epidermal growth factor receptor and is combined with a pharmaceutically acceptable carrier. [44] Design library scaffolds. The library scaffold is made of a set of human protein sequences that mimic the structure and function of germline VH and VL genes.

[45] Expression and display ability of the library scaffolds. Good expression and display ability of the library scaffold non-antibody peptide or proteins directly relate to the quality of the library to be developed over the scaffold genes. The library scaffold non-antibody peptide or protein expression and display ability was examined before the library construction. A few scaffold non-antibody peptide or proteins that expressed but unable or poorly displayed were excluded from the library construction. The well expressed and displayed library scaffolds ensures that high proportion of the non-antibody peptide or proteins in the library are functional, more superior than the libraries derived from combinatorial cloning of VH and VL genes genetically amplified from a natural sources. [46] Methods for library generation. A modified Kunkel mutagenesis method can be used according to the present invention, which efficiently generates billions of E. coli colonies each harboring a different non-antibody peptide or protein sequence, which can be used for the generation of the non-antibody peptide or protein libraries. While efficient, the percentage of non-mutagenized parental DNA increases when adapted in generation of a highly sequence complex library. In addition, technical limitations of synthesis of long oligonucleotides cab reduce the effectiveness of the method of making libraries containing sequence diversities in distant regions. To overcome such limitations, additional techniques of generating oligonucleotide >350 bases (mega-primer) and of creation of a stem-loop sequence containing a restriction enzyme recognition site in the mutagenesis template can optionally be used in combination with techniques such as the

Kunkel mutagenesis method. As compared to other known library technologies, such as restriction cloning, phage recombination and sequence specific recombination, as used by others to for library generation, the improved Kunkel based method can be more effective in generation of >10 ⁹ sequences per library and is more versatile in introducing sequence diversity in any location on the targeted DNA.

[47] In-line affinity maturation. An integrated affinity maturation process, or in-line affinity maturation, can be used according to the present invention for design and improvement of binding affinity of non-antibody peptides or proteins selected from the library. Improving binding affinity of desired non-antibody peptides or proteins after panning can increase the success of identifying therapeutic non-antibody peptide or protein leads.

The use of Kunkel method for library generation can improve the effective execution of sequence diversification strategies in a simple and continuous process. The design strategy and the technical advantages of using the improved Kunkel mutagenesis method can provide a superior approach over other pooled maturation strategies, where tedious library generation methods are employed that reduce the efficiency and effectiveness of the results.

[48] Parallel library panning. A parallel panning process using a automated or semi- automated equipment can be used to process the individually made sub-libraries. Parallel panning can maximize the potential of discovering a suitably diverse set of non- antibody peptides or proteins to provide the desired libraries having peptides or proteins of desired characteristics. Effective use of parallel panning in in-line affinity maturation also enables such proteins to have several simultaneously improved characteristics, such as improved affinities or biological activities. Development of a machine based panning system also allows systematic monitoring and adjustment of different panning conditions to more quickly screen and isolate non-antibody peptides or proteins having desired properties.

[49] Affinity ranking. Affinity based binding assays are applied to the large, diverse and high affinity ligand or antigen specific binding peptides or proteins to select the best binding for further characterization. Standard biochemical methods like ELISA as well as affinity measuring equipments, for example, BIAcore, Octet and BIND that are suitable for processing large number of samples are used alone or in combination for this purpose.

[50] While having described the invention in general terms, the embodiments of the invention will be further disclosed in the following examples that should not be construed as limiting the scope of the claims.

[5I ] EXAMPLE 1 : DISPLAY OF NON-ANTIBODY PROTEINS: Two non-antibody peptides or proteins were successfully displayed on the surface of M13 phage as plX fusion proteins. Both scaffold proteins were engineered to bind a specific protein target that is not bound by the scaffold proteins in their native state. These engineered scaffolds retained binding to this specific target while displayed on phage. In addition, libraries of amino acid variants were made in each scaffold, and their ability to bind a specific target addressed by displaying library members on phage as plX fusions and panning against a target protein. Both libraries produced a number of positive hits.

[52] Two non-antibody peptide or protein, immunoglobulin (Ig) domain proteins were selected as candidates to be engineered to present an EGFR binding peptide on the surface of M13 phage as a plX fusion protein. The following criteria were used for selection of candidates:

[53] High resolution atomic structure available in the PDB database a. Structure deposited represents a single, isolated Ig domain b. Entirely human sequence c. Sequences contain no disulphide bonded residues d. Expression and purification conditions published

e. Demonstrated to be successfully expressed and purified from E. co// without the need of a refolding step and with a high yield.

[54] The PDB database was searched manually, as well as with the aid of structure based alignment programs in order to investigate all proteins and protein fragments containing a single Ig domain. The scientific literature was then surveyed in order to determine published expression and purification conditions.

[55] Although the criteria above were used to select molecules with the greatest chance for success, it is envisioned that other Ig molecules that do not strictly meet these criteria could also be used as successful scaffolds for peptide display. It is also envisioned that scaffolds composed of multiple Ig domains or repeats of a single Ig domain could be used as successful scaffolds. The analysis above resulted in the selection of two Ig molecules, the third Fn3 domain from human tenascin (residues 802-893, TeFn3) and the I27 domain of human titin (residues 5253-5341 , Ti27) for further study. The genes encoding these protein fragments were synthesized by Blue Heron, and subcloned into the plasmid pPEP9-Bbsl by standard PCR and restriction digest methods. The mutations N5331 G in Ti27 and S881 G in TeFn3 were included in order to introduce restriction sites for insertion of peptides (see below).

[56] The F:G loops (residues 876-881 of TeFn3 and 5258-5259 of Ti27) (Figure 1 ) of both scaffolds were then replaced with a portion of the cysteine-constrained EGFR binding peptide, PHPEP190 (sequence DPCTWEVWGRECLQ) (Wang) using standard restriction site cloning methods to make the Ti27-190 and TeFn3-190 constructs. A number of control constructs were also created, including those containing non-EGFR binding F:G loop sequences (referred to as Ti27 and TeFn3) and containing six stop codons in the F:G loop (Ti27-stop and TeFn3-stop). Table 1 summarizes these constructs.

Table 1. Tenascin and titin scaffold constructs. ^* Denotes stop codon. SEQ ID NOs. 24- 29 indicate the particular polypeptide inserted into the scaffold F:G loop segment, the amino acid sequences for the corresponding full-length scaffolds are provided as SEQ ID NOs. 54-59, respectively.

[57] Each construct was expressed as fusion to the minor coat protein plX of filamentous phage from a phagemid vector with an EDSGGSGG (SEQ ID NO:30) linker sequence between the C-terminus of the scaffold protein and the N-terminus of plX. All constructs were expressed with an N-terminal Myc tag in order to assay for protein expression and display in the absence of EGFR binding. Successful display on the surface of M13 phage and specific binding activity of TeFn3-190 and Ti27-190 proteins were verified by phage ELISA. 0.5 μg of an Fc-fusion form of the EGFR ectodomain (sEGFR-MMB) (Wang), an anti-Myc non-antibody peptide or protein, or a control Fc molecule (CNTO 360) were immobilized on 96 well MaxiSorop™ plates and blocked with StartingBlock™ T-20. Phage displaying the constructs described in Table 1 were serially diluted before addition of 300 μL of the phage containing supernatant to each plate followed by incubation at room temperature for 1 hour. Each plate was subsequently washed with TBST and bound phage detected with an anti-M13 HRP-conjugated non-antibody peptide or protein using a fluorescent plate reader. Figure 2 shows that TeFn3-190 and Ti27-190 selectively bound the plate-bound Fc-EGFR with a low level of background binding as demonstrated by the lack of appreciable binding to CNTO 360. TeFn3 and Ti27 were also displayed on the phage surface as indicated by binding to the anti-Myc coated plate. These proteins however did not bind to EGFR. For reasons unknown, TeFn3-190 did not bind to the anti-Myc coated plate, although it is clearly expressed as shown by EGFR binding. As expected, TeFn3-stop and Ti27-stop were not displayed on phage.

[58] A library was constructed in order to determine the necessity of the cysteine constraints of PHPEP190 in the context of TeFn3 and Ti27 and to confirm that such a library can be efficiently displayed within these scaffolds on plX and panned against a target protein. This library was constructed by randomizing all six positions containing stop codons in Ti27-stop and TeFn3-stop constructs (Table 1 ). Sequencing of 100 randomly picked colonies following mutagenesis by the Kunkel method (Kunkel, et al., Methods Enzymol 154: 367-82, 1987) confirmed that mutations were found in 51 % and 64% of colonies from the Ti27-190 and TeFn3-190 libraries respectively. These libraries were then panned for three rounds against the EGFR-Fc fusion by incubating 100 μL of a phage solution with 10 μg (round 1 ) or 5 μg (rounds 2 and 3) of biotinylated EGFR for 1 hour at room temperature. Bound complexes were pulled down with streptavidin coated magnetic beads and washed 3 times with TBST. Bound phage were then used to reinfect TG1 cells, resulting in the phage titers described in table 2.

Table 2. Phage titers obtained during library panning.

[59] 96 colonies from the third round of panning from each library were randomly selected to test binding to EGFR-Fc by phage ELISA. Figure 3 shows that 10 and 20 clones from the Ti27-190 and TeFn3-190 libraries respectively showed significant binding to EGFR (ratio of counts from EGFR:CNTO 360 greater than 5). The sequences of the F:G loops from all positive hits are displayed in Table 3. All sequenced hits contained cysteine residues at the identical positions as in the original PHPEP190, indicating that the cysteine-constrained nature of this peptide is necessary for function.

SEQ ID Ti27 F:G loop Sequences Frequency Colonies NO:1 PHPEP190 DPCTWEVWGRECLQ

NO:2 TH 901 Ti3, Ti4, Ti5, Ti6, Ti9,

PLCTWEVWGRECYA

Ti10

NO:3 TH 902 PLCTWEVWGRECLM 1 Ti 7 NO:4 TH 903 ANCTWEVWGRECAH 1 Ti1 NO:5 TH 904 SFCTWEVWG RECQN 1 Ti2 NO:6 TH 905 EGCTWEVWGRECMS 1 Ti8

SEQ ID TeFn3 F:G loop Sequences Frequency Colonies NO:7 Te1901 PLCTWEVWGRECHT 1 Te4 NO:8 Te1902 ISCTWEVWG RECHM 2 Te13, Te16 NO:9 Te1903 NSCTWEVWGRECHV 1 Te11 NO:10 Te1904 DSCTWEVWGRECTL 1 Te6 NO:11 Te1905 DSCTWEVWGRECIL 1 Te17 NO:12 Te1906 DICTWEVWG RECSS 1 Te9 NO:13 Te1907 DICTWEVWGRECFG 1 Te12 NO:14 Te1908 DLCTWEVWG RECHA 2 Te8, Te14 NO:15 Te1909 GGCTWEVWGRECYQ 2 Te10, Te18

NO:16 Te19010 SVCTWEVWGRECNM 1 Te5 NO:17 Te1901 1 WCTWEVWGRECSQ 1 Te3 NO:18 Te19012 TTCTWEVWGRECYS 1 Te20 NO:19 Te19013 HACTWEVWGRECFG 1 Te19

NO:20 Te19014 HFCTWEVWGRECQS 1 Te2 NO:21 Te19015 SHCTWEVWGRECNL 1 Te15

NO:22 Te19016 NVCTWEVWGRECMN 1 Te7 NO:23 Te19017 NNCTWEVWGRECNW 1 Te1

Table 3. Positive hits from Ti27 and TeFn3 libraries. Residues randomized as part of the library are underlined. SEQ ID NOs. 2-6 indicate the particular EGFR-specific polypeptide inserted into the Ti27 F:G loop segment, the amino acid sequences for the corresponding full-length EGFR-specific scaffolds are provided as SEQ ID NOs. 32-36, respectively. SEQ ID NOs. 7-23 indicate the particular EGFR-specific polypeptide inserted into the TeFN3 F:G loop segment, the amino acid sequences for the corresponding full-length EGFR-specific scaffolds are provided as SEQ ID NOs. 37-53, respectively.

SEQUENCE LISTING

SEQIDNO.1: DPCTWEVWGRECLQ

SEQ ID NO.2: PLCTWEVWGRECYA

SEQ IDNO.3: PLCTWEVWGRECLM

SEQ ID NO.4: ANCTWEVWGRECAH

SEQIDNO.5:

SFCTWEVWGRECQN

SEQ ID NO.6: EGCTWEVWGRECMS

SEQIDNO.7: PLCTWEVWGRECHT

SEQ ID NO.8: TSCTWEVWGRECHM

SEQ ID NO.9: NSCTWEVWGRECHV

SEQ IDNO.10: DSCTWEVWGRECTL

SEQIDNO.11: DSCTWEVWGRECIL

SEQ IDNO.12: DTCTWEVWGRECSS

SEQIDNO.13:

DICTWEVWGRECFG

SEQIDNO.14: DLCTWEVWGRECHA

SEQIDNO.15: GGCTWEVWGRECYQ

SEQ IDNO.16: SVCTWEVWGRECNM

SEQIDNO.17: WCTWEVWGRECSQ

SEQIDNO.18:

TTCTWEVWGRECYS

SEQ ID NO. 19: HACTWEVWGRECFG

SEQ ID NO. 20: HFCTWEVWGRECQS

SEQ ID NO. 21 : SHCTWEVWGRECNL

SEQ ID NO. 22:

NVCTWEVWGRECMN

SEQ ID NO. 23: NNCTWEVWGRECNW

SEQ ID NO. 24: RRGDMGS

SEQ ID NO. 25: DPCTWEVWGRECLQ

SEQ ID NO. 26: ^*** TWEVWGRE ^***

SEQ ID NO. 27: AG

SEQ ID NO. 28: DPCTWEVWGRECLQ

SEQ ID NO. 29: * ^** TWEVWGRE ^***

SEQ ID NO. 30: EDSGGSGG

SEQ ID NO. 31 :

MAVFNSDSECPLSHDGYCLHDGVCMYIEALDKYACNCWGYIGERCQYRDLKWWE

SEQ ID NO: 32:

LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHI LILHNC QLGMTGEVSFQAPLCTWEVWGRECYAAKSAANLKVKEL

SEQ ID NO: 33:

LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHI LILHNC QLGMTGEVSFQAPLCTWEVWGRECLMAKSAANLKVKEL

SEQ ID NO: 34:

LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHI LILHNC QLGMTGEVSFQAANCTWEVWGRECAHAKSAANLKVKEL

SEQ ID NO: 35:

LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHI LILHNC QLGMTGEVSFQASFCTWEVWGRECQNAKSAANLKVKEL

SEQ ID NO: 36:

LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHI LILHNC QLGMTGEVSFQAEGCTWEVWGRECMSAKSAANLKVKEL

SEQ ID NO: 37:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISPLCTWEVWGRECHTSNPAKETFTTGL

SEQ ID NO: 38: RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQYSIG NLKP DTEYEVSLISTSCTWEVWGRECHMSNPAKETFTTGL

SEQ ID NO: 39:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISNSCTWEVWGRECHVSNPAKETFTTGL

SEQ ID NO: 40:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISDSCTWEVWGRECTLSNPAKETFTTGL

SEQ ID NO: 41 :

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISDSCTWEVWGRECILSNPAKETFTTGL

SEQ ID NO: 42:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISDTCTWEVWGRECSSSNPAKETFTTGL

SEQ ID NO: 43:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISDICTWEVWGRECFGSNPAKETFTTGL

SEQ ID NO: 44:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISDLCTWEVWGRECHASNPAKETFTTGL

SEQ ID NO: 45: RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQYSIG NLKP DTEYEVSLISGGCTWEVWGRECYQSNPAKETFTTGL

SEQ ID NO: 46:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISSVCTWEVWGRECNMSNPAKETFTTGL

SEQ ID NO: 47:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISWCTWEVWGRECSQSNPAKETFTTGL

SEQ ID NO: 48:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISTTCTWEVWGRECYSSNPAKETFTTGL

SEQ ID NO: 49:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISHACTWEVWGRECFGSNPAKETFTTGL

SEQ ID NO: 50:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISHFCTWEVWGRECQSSNPAKETFTTGL

SEQ ID NO: 51 :

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISSHCTWEVWGRECNLSNPAKETFTTGL

SEQ ID NO: 52 - RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQYSIG NLKP DTEYEVSLISNVCTWEVWGRECMNSNPAKETFTTGL

SEQ ID NO: 53 -

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISNNCTWEVWGRECNWSNPAKETFTTGL

SEQ ID NO. 54:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISRRGDMGSSNPAKETFTTGL

SEQ ID NO. 55:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQY SIGNLKP DTEYEVSLISDPCTWEVWGRECLQSNPAKETFTTGL

SEQ ID NO. 56:

RLDAPSQIEVKDVTDTTALITWFKPLAEIDGIELTYGIKDVPGDRTTIDLTEDENQYSIG NLKP DTEYEVSLIS ⁴⁴⁴ TWEVWGRE ⁴⁴⁴ SNPAKETFTTGL

SEQ ID NO. 57:

LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHI LILHNC QLGMTGEVSFQAAGAKSAANLKVKEL

SEQ ID NO. 58:

LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHI LILHNC QLGMTGEVSFQADPCTWEVWGRECLQAKSAANLKVKEL

SEQ ID NO: 59: LIEVEKPLYGVEVFVGETAHFEIELSEPDVHGQWKLKGQPLTASPDCEIIEDGKKHILIL HNC QLGMTGEVSFQA ^*** TWEVWGRE ^*** AKSAANLKVKEL