NO20023606L - Combinatorial protein domains - Google Patents

Description

The present invention relates to the de novo synthesis of folded protein domains by combinatorial rearrangement of sequence segments. The sequences of the segments may correspond directly to those of natural proteins, or may be derived from those of natural proteins (eg by random or directed mutagenesis), or be derived by design based on the known structures of proteins. More specifically, the present invention utilizes combinatorial rearrangement of sequence segments that are not complete structural elements of a native protein and that, in isolation, do not show any significant folding.

De novo design of proteins is typically based on structure predictions of predetermined amino acid sequences (Hecht 1994, Sauer 1996, Regan 1998). Partial randomization is often introduced to allow for imperfection in the predicted algorithms. The resulting repertoire is screened or selected for stably folded structures. This solution has been successful for the design of helical structures such as 4-helix bundles with stable and compact structures that exhibit free energies of unfolding of about 4 kcal/mol (Kamtekar et al., 1993). More problematic has been the design of P-flake proteins, where even the most recent attempts for natural |3-flake proteins were too poor in terms of to stability (Quinn et al., 1994, Kortemme et al., 1998, Alba et al., 1999). The problem in the design of (3-flake structures is related to their dependence on backbone hydrogen bonds between different secondary structure elements,

and we have less knowledge about this than the principles of helix formation (Hecht 1994). Repertoires of arbitrary protein sequences have also been screened for the presence of folded proteins. About. 1% of the members of a random library of Glu-, Leu- and Arg-rich proteins showed some helix formation and cooperative unfolding, but were unstable (Davidson&Sauer 1994).

Recently, new strategies to select stably folded proteins from repertoires of phage-expressed proteins, based on their resistance to proteolytic degradation, have been applied to improve the stability of native proteins (Kristensen & Winter, 1997, Sieber et al.- 1998, Finucane et al . 1999). Proteolytic degradation is usually limited to unfolded proteins or highly flexible regions of highly folded proteins. Folded proteins are mostly resistant to proteases, due that the proteolytic cleavage requires that the polypeptide chain is adapted to the specific stereochemistry in the active site of the protease, and

therefore must be capable of flexible, accessible and capable of local unfolding (Hubbard et al., 1994, Fontana et al., 1997). These methods have only been described for the selection of proteins with point mutations; in which no element of combining sequences from different proteins is involved.

A theoretical solution to protein unfolding via combinatorial rearrangement of defined, complete structural elements has been described (Bogarad & Deem 1999). The authors predict, using statistical algorithms, that rearranging a number of structural elements (such as helices, threads, loops, deflections and others) will result in the generation of new protein functions faster than by developing point mutation strategies alone. However, no validation is performed for context-dependence of the structure, nor is there any reference to partial structural domains that exhibit no structural identity in isolation. Although (rare) sequences will form structures in isolation, others may adopt a different structure in a different environment, as shown by structural rearrangements after cleavage of some polypeptides by protease or upon ligand binding. It is thus not easy to define a structural element if this is not done in the context of the three-dimensional structure of the protein in which it is enclosed, and it is this definition that we have used here. Furthermore, this article does not show that it will be possible to do this process in vitro, or indicate exactly how to carry out such experiments.

Summary of the invention

We have developed a different strategy for establishing the selection of new protein domains that are able to form stably folded structures, and thus to identify new structural and functional elements in proteins.

The inventors have realized that since the structure of a "structural" element depends on the context, simple structural elements will be taken from one protein and added on. simple structural elements from a second protein do not necessarily maintain their original structure. Thus, the inventors have not sought to limit the segments to simple complete structural elements. Furthermore, the application of parts of structural elements can provide new structures that are not simple due to proximity to existing structural elements, and application of the segments comprising multiple structural elements (and making pack interactions with each other) would be expected to be more stable than single structural elements, and more likely to comprise a significant "nugget" of structure in the chimeric domain.

Thus, the present invention examines the area of protein evolution by combining sequence segments. "Sequence segments", as indicated herein, are amino acid sequences which are not designed or selected to consist only of simple and complete protein structural elements, and which are not designed or selected to consist of a complete protein domain. The present invention is thus not directed to the assembly of discrete and simple elements of structures found in naturally occurring or synthetic proteins, but to the assembly of blocks of more than one structural element or to the establishment of new structural elements by assembly of sequences which, in isolation or in their native environment, do not exhibit a discrete and complete structure.

Therefore, a "sequence segment" is an amino acid sequence which, in its native environment, does not comprise a complete protein domain and is not encoded by one or more complete native exons. Furthermore, a "sequence segment", in its native environment, does not form one or more discrete structural elements, but instead is part of a structural element, or preferably, is longer than a structural element. The sequence segment in isolation shows no significant folding at the melting temperature of the chimeric protein; in other words, it exhibits no independent structure in isolated form.

The "native environment" of the sequence segment is the protein or polypeptide form from which the segment is taken, in its folded form. This can be a natural protein, or an artificial polypeptide or protein. Preferably taken

the sequence segment from an amino acid sequence that is longer than the sequence segment itself..

In accordance with the present invention, in a first configuration, the combinatorial rearrangement of the protein sequence segments enables the selection of new folded protein domains from combinatorial repertoires.

Thus, in a first aspect, the present invention provides a chimeric folded protein domain which is derived from a repertoire of chimeric proteins, and which comprises two or more sequence segments derived from original amino acid sequences which are not homologous.

Preferably, the original amino acid sequences are derived from protein domains. The original amino acid sequences may be natural, semi-synthetic or synthetic in origin. They may be derived by expression from genes, or assembled by chemical synthesis.

Preferably, the amino acid sequence segments are derived from proteins. In an advantageous embodiment, the proteins are selected from the group consisting of a naturally occurring protein, an engineered protein, a protein with a known binding activity, a protein with a known binding activity for a peptide or polypeptide, - a protein with a known binding activity for a carbohydrate , a protein with a known binding activity for a nucleic acid, a protein with a known binding activity for a hapten, a protein with a known binding activity for a steroid, a protein with a known binding activity for an inorganic compound, and a protein with an enzyme activity.

As used herein, the term "amino acid" includes the 20 naturally occurring amino acids, as well as non-naturally occurring amino acids and modified amino acids, such as labeled amino acids. As used herein, the term "protein" denotes a polymer in which the monomers are amino acids linked together through peptide or disulfide bonds. Preferably, "protein" refers to a full-length naturally occurring amino acid chain or fragment thereof, such as a selected region of the polypeptide of interest in a binding interaction, or a synthetic amino acid chain, or a combination thereof.

The sequence segments may be combined, in the chimeric protein domain, in any suitable manner. Typically, the segments will be combined by recombinant DNA techniques and will thus be linked, in the recombinant protein, with peptide bonds. In alternative embodiments, the segments may be synthesized separately and then linked. This can be achieved by using covalent bonding, e.g. peptide bonds, ester bonds or disulfide bonds, or by non-covalent bonding. Preferably, the sequence segments in accordance with the present invention comprise one or more reaction groups for covalent or non-covalent coupling. E.g. linkers capable of non-covalent association, such as biotin/streptavidin, can be incorporated into the sequence segments to effect non-covalent linkage.

The repertoire from which the chimeric protein domain is derived can be of essentially any size. Preferably, the repertoire comprises at least 10,000 individual protein domains, more preferably it comprises at least 1 million protein domains, and most preferably at least 100 million protein domains.

The sequence segments can be any suitable number of amino acids in length such that the combined length of the segments represents the length of a complete domain, which domain varies from as little as 35 units to several hundred units in length.

In an advantageous aspect, the original amino acid sequences are derived from the open reading frames of a genome or parts thereof;

(a) where said reading frame is the natural reading frame

to the genes, or

(b) where said reading frames are not the naturally occurring reading frames of the genes.

Sequences can thus be derived from ORFs that

located in an entire or substantially entire genome of an organism, or part thereof, such as a group or family of genes, which are either related by structure, function or evolution, or unrelated to each other. The part of the genome can also consist of a single gene.

The sequences can further be derived from two or more genomes, from organisms that are related or unrelated.

The protein domain in accordance with the present invention is capable of folding due to combination of two or more polypeptide segments that, in isolation, do not fold and do not define a single structural element of the original protein.

Preferably, the protein domain in accordance with the present invention is selected in accordance with their resistance to proteolysis. This provides a useful means of isolating candidate domains from libraries, a selection procedure can be configured so that only proteolysis resistant domains are selected from the libraries. Preferably, the proteolysis is carried out by exposure to a protease, such as thermolysin.

In a preferred embodiment, the protein domain in accordance with the present invention can be selected in accordance with their activity. This can e.g. be a bonding activity, e.g. in the case of immunoglobulin-like domains, or enzymatic activity in the case of the enzyme domain. Alternatively, the protein domain may have the capacity to bind antibody directed against the original protein. Furthermore, a screening for activity can be performed in addition to a selection based on folding as determined by protease resistance. Such a solution is particularly advantageous where an initial selection based on activity would be difficult or impossible to carry out.

Furthermore, the invention encompasses the assembly of sequence segments derived from non-homologous domains that share a similar polypeptide fold for at least parts of the structure. We have observed, upon selection of protein domains in accordance with the invention, that sequence segments derived from original protein domains that have similar folding for at least parts of their structures are assembled in some of the new chimeric proteins. Thus, the present invention provides a chimeric protein in accordance with a first aspect of the invention, where the sequence segments are from original domains with similar polypeptide folds in at least parts of the structure.

It has further been observed that, in selections of protein domains according to the invention, sequence segments derived from original protein domains having totally different folds for at least parts of their structures have been assembled into other new structures. Thus, the present invention provides a chimeric protein domain comprising two or more sequence segments derived from original amino acid sequences, where the sequence segments are taken from the original domain with different polypeptide folds at least in parts of the structure.

Furthermore, in the selection of protein domains according to the first configuration of the invention, sequence segments derived from the same protein domain can be observed and assembled to form new structures. In some cases, said sequence segments may comprise regions which together lead to a duplication of sequence in the chimeric protein. Furthermore, the common region does not only consist of one or more complete protein structural elements. It thus appears that duplication of amino acid segments or parts thereof, regardless of the presence of only one or more complete structural elements, can lead to the formation of stably folded structures. Such duplications comprise a second configuration of the invention.

As used herein, the terms "regions in common" or "common regions" denote regions that share sequence similarities or have a similar fold. In this context, "sequence similarity" preferably denotes regions of identical sequence of at least 10 amino acid units, more preferably at least 20 amino acid units.

In accordance with the second configuration of the invention, the combination of the segments from homologous proteins, leading to equivalent regions from these homologous proteins brought together in the same protein, would also be expected to lead to the establishment of stably folded structures. Regions that are equivalent in homologous proteins are identified by alignment of their amino acid sequences. It is indeed possible to combine segments from non-homologous proteins that share a common fold (vide supra), to establish stably folded chimeric proteins from segments that comprise a common region of a common fold in the original proteins.

Said stable folded structures based on duplication of amino acid segments have been established as a product of an arbitrary shuffling of amino acid sequences, and were selected through proteolytic selection due to their stability. Duplication, or indeed multiplication carried out in other non-arbitrary ways, has previously been reported, including e.g. Hardies et al., 1979 and Fire&Xu 1995). The inventors assume that said methods for duplication and multiplication can also be used for duplication or multiplication of amino acid sequences to establish new and stably folded domains under the second configuration of the invention. Such stable domains can be selected and screened for in ways identical or similar to those used for chimeric domains derived from combinatorial shuffling.

Protein domains in accordance with both configurations according to the present invention can be established and selected in any suitable way. Combinatorial rearrangements of nucleic acid segments are preferred, e.g. in subject-forward-. display libraries. Thus, the present invention provides a chimeric protein domain in accordance with one of the preceding aspects according to the invention, fused to coat proteins of filamentous bacteriophage, where said bacteriophage recapsulates a protein acid that codes for the protein domain.

Furthermore, both configurations according to the invention provide a nucleic acid which codes for a protein domain in accordance with the invention, as defined above.

In a further aspect of both configurations according to the invention, the amino acid sequences of any chimeric protein may contain sequences engineered to exhibit epitopes for vaccination against the original protein for said amino acid sequences. E.g. a selected polypeptide segment from the coat protein of a virus, against which a vaccine is to be prepared, can be used, which can be incorporated as a constitutive partner in a combinatorial library of amino acid sequences generated through shuffling with one or more segments from other genetic sources. Resulting chimeric proteins will thus comprise the segment of the viral envelope protein in a variety of structural settings. With screening or selection, e.g. by using antibody from antisera directed against the virus, it is possible to identify these folded chimeric proteins, which is why the viral sequence is displayed on a corresponding three-dimensional configuration as the viral protein. Such stably folded proteins among these chimeric constructs can be used for vaccination and to trigger an immune response against the chimeric protein which includes the viral amino acid segment. Vaccination with such a protein results in immunization against viruses. One advantage compared to vaccination with the viral coat protein is that it is thus possible to focus the immune response against one defined epitope on the virus, such as a neutralizing epitope.

It is also possible to vaccinate against defined epitopes of human proteins with the same strategy by combining a segment from a human protein with that from another source. The segment of non-human source should provide T-cell epitopes that will lead to an immune response against the human epitope. E.g. is it possible to raise a blocking (IgG) antibody response against the part of IgE that binds to the mast cell receptor. Such a response is useful e.g. by blocking asthma. This is achieved by constructing a chimeric protein as follows. First, segments from IgE are incorporated into chimeric proteins by combination with a repertoire of non-human segments, then the protein is screened or selected for binding to the mast cell receptor or to antibody known to bind IgE at the critical site, thirdly, using the chimeric proteins with binding activities for immunization. The IgE segments can be derived by arbitrary fragmentation of the IgE gene, or by using a segment already known to interact with the receptor. For immunization, it may be necessary to incorporate multiple potent T-cell epitopes into the non-human portion, which can be achieved by performing mutations in the non-human segment.

Preferably, the chimeric protein in accordance with the present invention thus comprises an epitope of an original amino acid sequence. Preferably, the epitope is a structural epitope.

Epitopes that are in the chimeric protein according to the invention, in a preferred embodiment, cross-react with antibody directed with a native amino acid sequence, or advantageously, the folded native protein.

In a further aspect of both configurations according to the invention, the segments may be derived entirely from human proteins. It is expected that these proteins will be less immunogenic in humans than foreign proteins, since the sequences of the protein will be almost completely human. Although such new human proteins would be expected to differ in three-dimensional structure from existing human proteins, (and subsequently to include novel B-cell epitopes), they would include T-cell epitopes derived from other human proteins (with the exception of the sequences flanking the link between the segments). Such proteins, which are not immunogenic, or only weakly so, would be very suitable for therapeutic purposes or to avoid sensitization in humans (eg enzymes in washing powder).

It is unlikely that the chimeric protein will in any context mimic the three-dimensional surface of the original protein in the region of target segments. This may be desirable as it may enable the protein to adopt a conformation that has altered binding activities. Thus, such proteins may be useful as improved enzyme inhibitors.

Furthermore, the present invention provides in both configurations for the creation of small domains that mimic parts of the surface of a larger protein. One advantage of small domains is that they can more easily enable the three-dimensional structure to be solved with X-ray crystallography or NMR, and also at higher resolution. This will simplify the construction of non-protein drugs based on the structure.

Furthermore, the invention in both configurations enables the fusion of individual sequence segments connected in the chimeric protein to further stably folded and complete protein domains. The fusion of additional domains can provide a means of selecting the chimeric protein domains (see methods below). They may also function to complement the chimeric protein domain to perform a specific function, e.g. binding, immunogenicity or catalysis.

In the second configuration of the invention, the presence of at least two regions of the same sequence or equivalent (homologous sequences) in the chimeric protein may enable the development of chimeric proteins that bind to ligands at each of the two sites. This may be advantageous by providing improved "..." of binding where both heads engage dimeric ligand (or other multimers), and also provide two binding sites with different affinities, covering a larger dynamic range in binding to a ligand.

A further aspect of the first configuration according to the invention relates to a method for selecting a protein domain in accordance with the invention as defined above. Thus, the invention provides a method for producing a protein domain in accordance with the first aspect of the invention, characterized in that it comprises the steps: (a) providing a first library of nucleic acids, where the library comprises coding sequences that encode sequence segments derived from one or more amino acid sequences , where the coding sequences are not selected or engineered to only code for a single and complete protein structural element or code for a complete protein domain, (b) providing a second library of nucleic acids, where the library comprises coding sequences that code for sequence segments derived from one or more amino acid sequences, where the partner coding sequence is not selected or engineered so that it only codes for a single and complete protein structural element, or codes for a complete protein domain, (c) combining the coding sequences to form a combinatorial library of nucleic acids, wherein the nucleic acids comprise contiguous coding sequences see coding sequence fragments derived from the first and second libraries, (d) transcribe and/or translate the contiguous coding sequences to produce the encoded protein domains, (e) select the chimeric protein domains capable of adopting a folded structure, or to fulfill a specific function.

Libraries in accordance with the present invention can be constructed so that sequences homologous to the partner coding sequence are excluded. E.g. libraries can be based on an artificial combination of loose structures, meaning that the presence or absence of sequence homologues to the partner coding sequence can be regulated. However, if genomic libraries are used, it is possible that sequences homologous to the partner sequence may be present. Thus, in a preferred aspect, the method according to the invention further includes the steps: (f) analyzing the sequence of the selected chimeric protein domains to identify the origin of the sequence segments, and (g) comparing the sequences of each of the original amino acid sequences to identify whether the sequences of the original amino acid sequences are non-homologous. Correspondingly, it is possible to construct libraries comprising sequence segments derived from defined protein folds. However, it is necessary to determine whether the isolated protein domain according to the invention is composed of sequence segments derived from original domains having the same folding, and the method according to the invention preferably includes the step: • (h) comparing the structures of each of the original domains to identify whether they have the same polypeptide folding in whole, or parts.

In a further preferred aspect, the first configuration of the invention relates to the combination of a library of sequence segments with a unique partner coding sequence derived from a protein. In this connection, the partner sequence is provided as a unique sequence. Thus, steps (b) and (c) of the method in accordance with the first configuration of the invention as set forth above may be modified as follows: (b) providing a partner code sequence encoding from a segment sequence derived from one protein, where the partner coding sequence is not selected or constructed so that it only codes for a single and complete protein structural element or codes for a complete protein domain,

(c) combining the library and the partner coding sequences to form a combinatorial library of nucleic acids, the nucleic acids comprising contiguous coding sequences encoding sequence fragments derived from the first library and the partner coding sequence.

A further aspect of the second configuration according to the invention relates to a method for selecting a protein domain, where the individual sequence segments comprise common sequences. Thus, the invention provides a method for producing a protein domain in accordance with the first aspect of the invention, characterized in that it comprises the steps: (a) providing a first library of nucleic acids, where the library comprises coding sequences that encode sequence segments derived from one or more amino acid sequences, where the coding sequences are not selected or engineered to encode a complete protein domain, (b) providing a second library of nucleic acids, wherein said library comprises coding sequences that encode from sequence segments derived from one or more amino acid sequences, where the partner coding sequence is not selected or engineered such that it encodes a single or complete amino acid element or encodes a complete protein domain, (c) combining the coding sequences to form a combinatorial library of nucleic acids;

wherein the nucleic acids comprise contiguous coding sequences encoding sequence fragments derived from the first and second libraries, (d) transcribing and/or translating the contiguous coding sequences to produce the encoded protein domains, (e) selecting the chimeric protein domains, which are capable of adopting a folded structure, or to fulfill a specific function,

and optionally:

(f) analyzing the sequences of the selected chimeric protein domains to identify the origin of the sequence segments, and (g) comparing the sequences to identify whether they comprise common sequences.

Correspondingly, in a further aspect of the second configuration, the present invention relates to a method for selecting a protein domain, where the individual sequence segments comprise common regions from the original proteins with a common folding. However, if it is necessary to determine whether the isolated protein domain according to the invention is composed of sequence segments derived from native domains, the method according to the present invention preferably does not require step (g) above, but instead includes the steps:

(g) compare the structures of the origin-

identical amino acid sequences to identify whether the original proteins have a common folding, and

(h) identify whether the segments include a

common region in the common folding.

In a further preferred aspect, a second configuration according to the invention relates to the combination of a library of sequence segments with a unique partner coding sequence derived from a protein. The partner sequence is in this aspect provided as a unique sequence. Thus, steps (b) and (C) in the method according to the second configuration as indicated above can be modified so as to: (a) provide a partner coding sequence which codes for a sequence segment derived from a protein where said partner coding sequence has not been selected or designed so that it only encodes a single and complete protein domain; (b) combining the library and the partner coding sequences to form a combinatorial library of nucleic sequences, the nucleic sequences comprising contiguous coding sequences encoding sequence fragments derived from the first library and the partner coding sequence.

Preferably, according to the methods according to both configurations according to the invention, the domains capable of adopting a folded structure are selected by means of one or more methods selected from the group consisting of in vivo proteolysis, in vitro proteolysis, binding capacity, functional activity and expression .

In a further embodiment, an amino acid sequence of any of the chimeric proteins produced by combinatorial shuffling in accordance with both configurations according to the invention can be mutated or changed after the original assembly of the mor amino acid sequences. Such changes can be introduced in one of the following ways:

(a) designing and introducing specific or arbitrary mutations at predetermined positions within the gene in the chimeric protein, (b) deleting nucleotides in the gene in the chimeric

protein to remove amino acid units,

(c) inserting nucleotides into the gene in the chimeric

protein to insert amino acid units,

(d) adding nucleotides to the gene in the chimeric

protein to add amino acid units,

(e) arbitrarily introduce mutations into all or part of the gene encoding the chimeric protein through recombinant DNA technology, (f) arbitrarily introduce mutations into the gene of the chimeric protein through propagation in mutant cells, (g) introduce derivatives of natural amino acids

through chemical synthesis,

(h) chemical derivatization of amino acid groups by

synthesis,

(i) multimerization of chimeric proteins through ... of two or more copies of the gene in a single open reading frame,

(j) multiply the chimeric proteins by covalently linking two or more copies of the chimeric protein domain after translation;

(k) multimerize the chimeric proteins through fusion to a multimeric partner.

Any of the aforementioned changes may improve the stability or function of the chimeric protein. E.g. the changes may be intended to reach predicted structural requirements within the combined sequences that are beneficial for the formation of specific polypeptide folds or to introduce specific amino acid sequences to fulfill a specific function. An example of such an improvement is given in Example 14 in the experimental section.

The invention includes further optimization of

the regions of the N- and C-termini of recombined amino acid segments. Thus, both the link and the end regions that divide into a chimeric protein are not optimized with regard to stability and/or function of the chimeric protein. Natural proteins, which have been established through a recombinatorial event, are then optimized through (point) mutation events and Darwinian selection. This process can be mimicked in vitro for chimeric protein as defined herein, e.g. by using the specified methods (including mutation, deletion and/or addition of amino acid units).

Chimeric proteins containing such improvements can be identified by one or more methods used for the selection and screening of the original combinatorial library. It may also be advantageous to produce selected chimeric protein domains in a multimerized form, e.g. to increase stability through interdomain interactions or improve binding to a ligand through avidity effects.

Brief description of the figures

Fig. 1. Proteolysis of selected phage and chimeric proteins, (a) ELISA for barstar binding of phage lc2 (squares), lbll (circles), lg6 (diamonds) and csp/2 (triangles) before and after trypsin/thermolysin treatment at different temperatures. SDS-PAGE of proteins His-lc2, His-lbll and His-lg6 before and after treatment with trypsin, thermolysin and chymotrypsin at 25°C. Fig. 2. Circular dichroism and thermodenaturation of chimeric proteins, (a) Circular dichroism spectra of His-lc2 (top trace) and His-2f3 (bottom trace) at 20°C. Ellipticity of His-lc2 (at 205 nm, top trace) and His-2f3 (at 223 nm, bottom trace) at different temperatures. Fig. 3. Nuclear magnetic resonance analysis of chimeric proteins. ID-H NMR spectra of His-2f3 taken (a) at 25°C in H 2 O and (b) after incubation for 24 h at 25°C in H 2 O. 1D-<X>H NMR spectra of His-lc2 taken at 30°C (c) in H 2 O and (d) after incubation for 24 h at 25°C in D 2 O. 2D-<1>H-NOESY spectra of His-lc2 taken at 30°C (e) in H2O. Fig. 4. Biotin-CspA ELISA. A rabbit anti-CspA antiserum was incubated with varying amounts of soluble His-CspA. His-lc2. His-2fc, His-lbll or lysozyme (as a negative control) before binding to biotinylated CspA immobilized on streptavidin-coated ELISA wells. Bound rabbit antisera was detected with an HRP-conjugated goat anti-rabbit IgG antiserum.

Detailed description of the invention

The present invention relates to chimerics

folded protein domains. In connection with the present invention, the term "folded" indicates that the protein domains are able to adopt, or have adopted, a stable tertiary structure. Stability in this context can be defined as conformational stability of the protein, which is the difference in free energy between the folded and unfolded conformations under physiological conditions, the higher this value is, the higher the energy required to unfold the protein, and thus the stability greater for the folded structure. A quantitative measure of this conformational

stability of proteins, the Gibbs free energy of folding, can be determined from reversible thermodynamics. Proteins undergo order/disorder transitions, which can be detected in differential scanning calorimetry (DSC) profiles of specific heat vs. temperature.

Preferably, the free energy for folding exhibited by a protein domain according to the invention is 1.6 kcal/mol or higher, preferably it is 3 kcal/mol or higher, and most preferably it is 5 kcal/mol or higher.

Folded proteins that form stable structures are known to be resistant to proteolysis. Thus, the invention provides for the selection of folded protein domains in accordance with the invention using protease enzymes, which cleave and preferably eliminate unstable and unfolded domains. The "fold" can therefore be defined in terms of resistance to proteolysis under assay conditions. Examples of conditions are set out in the examples below.

Sequence segments in accordance with the invention are sequences of natural protein sequences, which occur in naturally occurring proteins, or artificial segments of sequences modulated on the sequence or structure of naturally occurring proteins. The sequence segments can be between 10 and 100 amino acids in length, or longer, preferably between 15 and 50 amino acids in length, and more preferably between 20 and 45 amino acid lengths, or where the invention relates to nucleic acids, the necessary length to code for such amino acid sequences.

The sequence segments in accordance with the invention are derived from original protein domains which are not homologous.

The term "parent" amino acid sequences denotes any amino acid sequence encoded by open reading frames within the DNA sequences, which form the source of the cloned DNA segments as part of the combinatorial libraries as set forth in the patent claims. Such reading frames can be part of the original reading frame of the genes, or shifted reading frames or of the reverse strand of the genes. They can also be part of intragenic regions, which are known to encode a protein. Original genes can be natural or synthetic.

As described in the introduction, the term homology between two or more proteins or protein domains refers to a similarity or identity in both their amino acid sequences and their structural folding. For present purposes, the term "homology" shall indicate only the degree of identity between two original amino acid sequences.

Homologous amino acid sequences have 35% or greater identity (eg at least 40% identity, 50% identity, 60% identity, 70% identity, or at least 80% identity, such that at least 90% identity, or at least 95% identity). Homologous nucleic acid sequences are nucleic acid sequences that encode homologous polypeptides, as defined. Current nucleic acid homology/identity values can be determined using the "Align" program by Myers&Miller 1988, ("Optimal Alignment in Linear Space") and available at NCBI. Alternatively, or in addition, the term "homology", e.g. with respect to a nucleotide or amino acid sequence, indicate a quantitative measure of homology between two sequences. Percent sequence homology can be calculated as (Nref - Ndif) *100/Nref , where Ndifer is the total number of non-identical units in the two sequences when these are aligned, and where Nrefer is the number of units in one of the sequences. Thus, the DNA sequence AGTCAGTC has a sequence similarity of 75% with the sequence AATCAATC (Nref=8; Ndif=2). Alternatively, or in addition, the term "homology" with respect to the sequences may indicate the number of positions with identical nucleotides or amino acids derived from the number of nucleotides or amino acids in the shorter of the two sequences where alignment of the two sequences can be determined in accordance with Wilbur and Lipman -algorithm (Wilbur & Lipman 1983), e.g. using a window size of 20 nucleotides, a "word length" of 4 nucleotides, and a gap penalty of 4, and computer-assisted analysis and interpretation of sequence data including alignment can be conventionally performed using commercially available programs (e.g. Intelligenetics™ Suite. Intelligenetics Inc., CA). When RNA sequences are said to be similar, or have a degree of sequence identity or homology with DNA sequences, thymidine (T) in the DNA sequence is considered equal to Uracil (U) in the RNA sequence.

The RNA sequences within the scope of the invention can be derived from DNA sequences, where thymidine (T) in the DNA sequence is considered equal to Uracil (U) in the RNA sequences.

Additionally or alternatively, amino acid similarity or identity or homology can be determined using the BlastP program (Altschul et al. 1997), and available at NCBI. The following references (incorporated herein by reference) provide algorithms for comparing the relative identity or homology of the amino acid units of two proteins, and in addition or alternatively to the foregoing, the descriptions in these references may be used to determine percent homology or identity: Needleman & Wunsch (1970); Smith & Waterman (1981); Smith et al. (1983); Feng & Dolittle (1987); Higgins & Sharp

(1989); Thompson et al. (1994); and Devereux et al. (1984).

The present invention comprises the recombination of sequence segments which are derived from original proteins with corresponding folding. In this context, the term "similar" is not equivalent to "homologous". Indeed, similar foldings have been shown to arise independently during evolution. Such folds are similar but not homologous.

A "protein structural element" is an amino acid sequence that can be recognized as a structural element of a protein domain. Preferably, the structural element is selected from the group consisting of an α-helix, β-sheet, β-barrel, a parallel or anti-parallel β-sheet, or helical structures (such as the 3io helix and the phi helix), and sequences which represent high curvatures or loops. Preferably, the structural element is an α-helix or a β-strand, sheet or barrel.

In a preferred embodiment, the folded protein domains in accordance with the present invention are constructed from sequence segments which do not comprise only a single structural element, but instead comprise less than a single structural element, or more than a single structural element,

or parts thereof.

in

In accordance with the present invention, the sequence segments used are not designed or selected to include only such simple elements, in other words, they may include

more than a single structural element, or less than one

in a single structural element. This can be achieved through the use of essentially arbitrary sequence segments for constructing a library in accordance with the invention. E.g. can sonicated genomic or cDNA or segments produced

by arbitrary PCR of DNA is used. Preferably, the DNA

) the fragments between 100 and 500 nucleotides in length.

The sequence segments used in accordance with the present invention are not capable of folding to a significant extent

insulation, i.e. they do not contain sufficient structural

j isolation to form a folded protein domain unless they are combined with other sequence segments in accordance with pre-

lying invention. This inability to fold significantly can be measured by how susceptible they are to protease cleavage, e.g. under conditions given in the example below, or by measuring the free energy of folding.

Proteolysis can be carried out using protease enzymes. Suitable proteases include trypsin (cleave at Lys, Arg), chymotrypsin (Phe, Trp, Tyr, Leu), thermolysin (small aliphatic units), subtilisin (small aliphatic units), Glu-C (Glu), Factor Xa (Ile/Leu /Glu-Gly-Arg), Arg-C (Arg) and thrombin. Preferably, since the combination of arbitrary polypeptide sequence segments cannot be guaranteed to generate a precise cleavage site for a particular protease, a broad-spectrum protease capable of cleaving at a variety of sites is used. Trypsin, chymotrypsin and thermolysin are broad-spectrum proteases useful in the present invention.

The ability of a protein domain to fold is also associated with its function. Thus, the invention is provided for the selection of folded protein domains with functional analyses.

In the case of immunoglobulins or other polypeptides capable of binding, such assays can be performed for binding activity in accordance with established protocols, however, so that where binding is only temporary, selection can be performed on the basis of function alone.

Suitable methods are indicated e.g. in International

Patent application PCT/GB00/00030 and PCT/GB98/01889. Such techniques are useful for the selection of new or improved enzymes produced by combinatorial rearrangement in accordance with the present invention.

The present invention also provides for screening for activity or selection consistent with protease resistance. This enables protein domains that have been selected according to their ability to fold to be screened for the desired activity. Since the repertoire size is more limited, as a result of selection with proteolysis, the screening step can be performed more easily (eg in a multi-well plate).

The libraries according to the present invention can be established with any suitable method in any form. As used herein, the term "library" denotes a mixture of heterogeneous polypeptides or nucleic acids. The library is composed of members, each of which has a unique polypeptide or nucleic acid sequence. In this context, "library" is synonymous with repertoire. Sequence differences between library members are responsible for the diversity that exists in the library. The library may be in the form of a single mixture of polypeptides or nucleic acids, or may be in the form of organisms or cells, e.g. bacteria, viruses, animal or plant cells and the like, transformed with a library of nucleic acids. Typically, each organism or cell contains only one element of the library. In certain applications, each individual organism or cell may contain two or more members of the library. Preferably, the nucleic acids are incorporated into expression vectors to enable expression of the polypeptides encoded by the nucleic acids. Thus, in a preferred aspect, the library may take the form of a population of host organisms, each organism containing one or more copies of an expression vector containing a single member of the library in nucleic acid form which can be expressed to produce its corresponding polypeptide member. Thus, the population of host organisms has the potential to encode a large repertoire of genetically diverse polypeptide variants.

A number of vector systems useful for library production and selection are known in the art. For example bacteriophage lambda expression systems can be screened directly by bacteriophage plaque or as colonies of lysogens, both as previously described (Huse et al., (1989): Caton&Koprowski (1990): Mullinax et al. (1990): Persson et al.

(1991)), and can be used according to the invention. Since such expression systems can be used for up to IO<6> different members of a library, they are not well suited for screening larger numbers (greater than IO<6> members). Other screening systems are e.g. based on direct chemical synthesis of library members. One prior method involves the synthesis of peptides on a set of sticks or rods, as described in WO84/03564. A similar method involves peptide synthesis on beads, which form a peptide library where each bead is an individual library member, as described in US Patent No. 4,631,211, and a related method is described in WO92/00091. A significant improvement to the bead-based methods involves labeling each bead with a unique identification marker, such as an oligonucleotide, to facilitate identification of amino acid sequences for each library member. These improved bead-based methods are described in WO93/06121.

Another chemical synthesis method involves synthesizing arrays of peptides (or peptidomimetics) on a surface in a manner that places each distinct library member (eg, unique peptide sequence) at a discrete predetermined location in the array, or by spotting preformed polypeptides on such an event. Identification of each library member is determined by its spatial location in the arrangement. The locations in the array of binding interactions between a predetermined molecule (eg, a receptor) and reactive library members are determined, thereby identifying the sequences in the reactive library member on the basis of spatial localization. These methods are described in US

Patent No. 5,143,854, WO90/15070 and WO92/10092; Fodor et al.

(1991); and Dower & Fodor (1991).

Particularly useful in the construction of libraries according to the invention 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 enables the selection, by suitable display means, of the individual members of the library.

Any selection display system can be used in connection with a library in accordance with the invention. Selection protocols for isolating desired members are known in the art, e.g. subject presentation techniques. Such systems, where different peptide sequences are displayed on the surface of filamentous bacteriophage (Scott & Smith

(1990)), has proven useful for establishing libraries of antibody fragments (and the nucleotide sequences encoding them) for in vitro selection and multiplication of specific antibody fragments that bind a target antigen. The nucleotide sequences encoding the VH and VL regions are linked to gene fragments encoding guidance signals that direct them to the periplasmic space of E. coli, and as a result, the resulting antibody fragments are displayed on the surface of the bacteriophage, typically as fusions to bacteriophage -coat proteins (e.g. Pill or VIII). Alternatively, antibody fragments are displayed externally on lambda phage capsids (phage bodies). An advantage of subject-based display systems is that, due to that they are biological systems, selected library members can be multiplied simply by allowing the phage containing the selected library member to grow in bacterial cells. Furthermore, since the nucleotide sequence encoding the polypeptide member is contained on a phage or phagemid vector, sequencing, expression and subsequent genetic manipulation is relatively straight forward.

Procedures for the construction of bacteriophage antibody display libraries and lambda phage expression libraries are well known in the art (McCafferty et al. (1990); Kang et al. (1991); Clackson et al. (1991); Lowman et al. (1991); Burton et al. (1991); Hoogenboom et al. (1991); Chang et al. (1991); Breitling et al. (1991); Marks et al. (1991); Barbas et al. ( 1992); Hawkins & Winter

(1992); Marks et al. (1992); Lerner et al. (1992), (incorporated herein by reference).

Other systems for generating libraries of polypeptides or nucleotides include the use of cell-free enzymatic machinery for in vitro synthesis of library members. E.g. In vitro translation can be used to synthesize polypeptides as a method to generate large libraries. These methods, which generally comprise stabilized polysome complexes, are further described in WO88/08453, WO90/05785, WO90/07003, WO91/02076, WO91/05058 and WO92/02536. Alternative non-phage-based display systems such as those described in WO95/22625 and WO95/11922 (Affymax) use polysomes to display polypeptides for selection. These, and all of the foregoing documents, are incorporated herein by reference.

To produce libraries and sequence segments in accordance with the present invention, PCR amplification is preferably used. Where a defined partner sequence is used, one PCR primer can be designed to anneal specifically with the partner sequence, while for random libraries general random PCR primers can be used. The resulting fragments are joined by restriction and ligation, and cloned into suitable vectors. Although ligation of two sequence ligands is described below, the invention also encompasses ligation of three or more sequence segments, and these may be the same or different, so as to mimic multiple crossover events.

The present invention will be further described, for purposes of illustration, in the following experimental section.

Example 1

Production of a repertoire of chimeric proteins comprising two consequence segments.

A repertoire of genes encoding chimeric proteins, comprising the N-terminal 36 units of E. coli cold-shock protein (CspA) and a C-terminal polypeptide sequence encoding arbitrarily established fragments of the E. coli genome, were produced. CspA comprises 70 units, forming a stable (3-barrel (Schindelin et al. 1994) . Its N-terminal 36 units comprise the first three strands and its six ... (3-... and is unable to fold as it is expressed alone as they grade in the E. coli cytoplasm.

The gene fragment that codes for the first 36 units of

CspA was complemented with fragmented DNA from the E.coli genome about 140 base pairs in size. The DNA fragments were established by random PCR amplification using genomic E.coli DNA as a template. Resulting chimeric genes were inserted between the coding regions for the infectious protein p3 and an N-terminal marker, a stable but catalytically inactive mutant of RNase barnase, as a single continuous gene on a phragmid vector for protein-displaying filamentous phage.

In the resulting genomic library (1,0xl0<8>members), an opal (TGA) stop codon was incorporated at the 3' end of the chimeric gene in 60% of the clones while the remainder contained Gly-encoded GGA codon at this position . The partial incorporation of the TGA codon at the 3' end of the chimeric genes was achieved through the use of two different PCR primers (XTND and NOARG) in the PCR amplifications of the E.coli gene fragments. Transfer RNA<Trp>can decode DNA with an efficiency of up to 3% (Eggertsson X Soll 1988) which leads to sufficient display of the barnase chimera p3 fusion on the phage, but avoids folding-related toxic effects. Phage exhibiting this repertoire were produced using a helper phage KM13, which contains a modified fd-gene 3 that codes for a trypsin-sensitive p3 due to a modified sequence (Kristensen&Winter 1997) to reduce infectivity due to helper phage-encoded p3 molecules.

Example 2

Preparation of a repertoire of chimeric proteins comprising two sequence segments with common sequences

In a second "plasmid-derived" library, the N-terminal CspA gene fragment was complemented with DNA fragments of around 140 base pairs established by random PCR amplification using as a PCR template a 3.6 kb plasmid containing the wild-type CspA gene. Resulting chimeric genes were again inserted as a fusion between coding regions for the infectious protein p3 and an N-terminal marker, a stable but catalytically inactive mutant of RNase, on a phagemid vector for protein display on filamentous phage.

In the plasmid-derived library (1.7x10<8> members), an opal (TGA) stop codon was constitutively introduced at the 3' end of the chimeric gene in all clones. The phages displaying this repertoire were produced using helper phage KM13, which contains a modified fd gene 3 encoding a trypsin-sensitive p3 due to a modified sequence (Kristensen & Winter 1997), to reduce infectivity due to helper phage-encoded p3 molecules.

Example 3

Proteolytic selection of combinatorial libraries

To select stably folded chimeric proteins from the repertoires of barnase chimera p3 functions described in Ex. 1 and 2, phage-displayed libraries were selected for proteolytic stability in three rounds through treatment at 10°C with the protease trypsin (specific for peptide bonds containing Arg or Lys in the Pi position) and thermolysin (specific for bonds containing an amino acid with an aliphatic side chain in the Pi position) followed by uptake on barstar, elution, infection and re-growth.

After the first round of selection, 2x10<4> and 6x10<2> of 10<10> proteolytically treated phages were eluted from a single bar-star coated microtiter plate well in the case of the plasmid-derived library and the genomic library, respectively. If protease treatment is omitted, 5xl0<6->phages can be eluted, which indicates that the majority of the unselected phages do not exhibit a stably folded chimera protein fusion between barnase and p3. The number of subjects occupied after two or three rounds of selection increased to 2xl0<5>and the plasmid-derived library and 2xl0<3>and 4xl0<4>for the genomic library.

Selected phages were allowed to grow up individually, were bound to immobilized barstar, treated in situ with trypsin and thermolysin at 10°C, and resistance was measured through detection of bound (and therefore resistant) phages in ELISA. For the plasmid-derived library, 27 of 64 phages (42%) maintained 80% or more of their barstar binding activity after protease treatment. For the genomic library, after two rounds, 6 of 192 (3%) maintained at least 80% of their bar-star binding activity. After three rounds, 31 of 86 (36%) subjects maintained 80% or more of their barstar binding activity. Selection thus clearly enriches phage display protease-resistant p3 fusions.

Example 4

Sequence analysis for selected chimeric proteins

As an initial characterization of the selected chimeric fusion proteins, the sequences of the selected clones from Example 3 were determined. The chimeric genes of all the four most stable phage clones selected from the plasmid-derived library have an open reading frame from the genes for barnase, through the one for chimeric protein and to the end for the p3 gene. They also contain no stop codon (in addition to the opal stop codon at the 3' end). 20 of these contained insertions that were originally from the CspA gene in the correct reading frame. These 20 comprised three different genes (Al was found 12 times, D6-6 times, and G4 twice). Phage Al contains a deleted version (units 1 to 52) of the CspA wild-type gene, which has been established through a deletion within a phagemid clone originally harboring a larger insert (Table 1). Phage D6 contains in addition to the N-terminal half of CspA (units 1 to 36 as part of the clone vector) the core CspA (units 17 to 53) (Table 1). Phage G4 contains as an insert a partial duplication of the N-terminal half of CspA (units 2 to 19). Thus, from the plasmid-derived library, phage with p3 fusion chimeras, where the N-terminal half of CspA was complemented with another fragment from CspA, were highly enriched by the proteolytic selection.

The sequences of 25 protease-resistant phage clones selected from the genomic library yielded 11 different clones, (2 clones were found five times, 1 clone four times, and 3 clones 2 times). All deposits retained the reading frame from barnase to p3. They all contained the opal stop codon at their 3' end but no additional stop codon. The insertions in all the phage sequences could be traced back to the E.coli genome, which shows an error rate of approx. 1% probably due to their generation by PCR. 64% of the sequenced phages contained inserts whose reading frame was identical to that of the original E.coli protein. This•suggests an enrichment for DNA fragments in their natural reading frame, and from an arbitrary distribution based on three possible reading frames and two possible orientations of any DNA, only 16% of the deposits would be expected to maintain the natural reading frame. However, the selection of the clones originating in the original reading frame (ORFs) which do not correspond to the natural reading frame of the original gene in 36% of the sequences indicated that these can also lead to the formation of stably folded chimeras.

As described in Example 1, 60% of all clones in the unselected genomic library contained an opal (TGA) stop codon at the 3' end of the chimeric gene, while the remainder contained the Gly-encoding GGA codon at this position. However, only clones containing the opal stop codon at this position were found after proteolytic selection from the genomic library. In the absence of a constitutive stop codon, almost all chimeric gene functions result in selection for a frameshift between barnase and the p3 gene (data not shown). These results show that the efficiency (up to 3% according to Eggertsson&Soll, 1988) with which transfer RNA<Trp>can decode TGA as a tryptophan leads to sufficient display of the barnase chimera p3 fusion on the phage but seems to reduce folding-related toxic effects. Use of an opal stop codon in the genes that code for the presented fusion proteins was therefore advantageous for selection in the presented examples.

Example 5

Proteolytic stability of selected chimera phages in solution

To show that the sequenced fusion proteins were not only proteolytically stable after immobilization on the display phage on a barstar-coated surface (as shown in Example 3) but also in solution, they were tested for proteolytic stability through exposure to trypsin and thermolysin in solution (before immobilization) at different temperatures (Fig. 1). The phages retaining the barnase tag (as a consequence of a proteolytically stable fusion protein) were captured on barstar, and the percentage of residual barstar binding activity was quantified by ELISA.

Among the phages from the plasmid-derived library, two clones (A1 and D6) maintained at least 80% of their binding activity after treatment at 20°C. From the genomic library, 8 of the 11 clones (1C2, IG6, 1A7, 2F3, 1B11, 2F1, 2H2, 3A12) maintained their efficiency after trypsin/thermolysin treatment at 24°C. The remaining phages were less protected from proteolytic attack in solution than when bound to the barstar-coated surface (compare Example 3).

Example 6

Soluble expression of selected chimeric proteins

To characterize the selected chimeric proteins outside the context of the barnase-p3 fusion protein, the genes for the ten most stable chimeras of the selected clones in Example 5 were expressed without the fusion partners. For this, their genes were subcloned for cytoplasmic expression into a His-tag vector. Five of these proteins (His-a1, His-d6, from the plasmid-derived library, His-lc2, His-2f3 and His-lbll from the genomic library) could be purified after expression directly from the soluble fraction of the cytoplasm via their Elevator marker. The remaining proteins formed inclusion bodies in the expressing cells. One of these, His-lg6 containing an insert expressed in a reading frame different from that of its parent gene (Table II), was refolded via solubilization in 8M urea. The resulting clones were not studied further.

Example 7

Biophysical characterization of chimeric proteins

The first biochemical analysis of the purified chimeric proteins described in Example 6 concerns their multimerization status. The chimeric proteins His-a1, His-d6, His-lc2, His-2f3, His-lg6 formed only monomers according to their elution volume in gel filtration, while His-lb11 formed 30% monomers and their remaining formed dimers.

To analyze the type of secondary structure formed by these chimeras, purified proteins were studied by CD and NMR. CD spectra (Fig. 2A) of monomeric proteins and the monomeric fraction of His-lbll all had characteristic (5-structure-containing proteins with minima between 215 nm and 225 nm (Greenfield&Fasman 1969, Johnson 1990). All proteins exhibited cooperative folding characteristics with sigmoidal melting curves (Fig. 2b) and midpoint of unfolding-folding transition between 46°C and 62°C (Table I). The cooperative folding behavior is a strong indication that each of the chimeras analyzed formed a domain with a single fold, i in contrast to a mixture of folded or partially folded structures such as a melt globule.NMR spectra of His-2f3 and His-lc2 further suggest the presence of a well-folded protein domain, as interpreted from the chemical shift dispersion of many amide protons to values downstream 9 ppm (Fig. 3a,c) and methyl group protons to values around 0 ppm in their NMR spectra (Wuthrich 1986).Finally, downfield chemical shifts are seen on Ca protons to values between 5 and 6 ppm, as also seen in the NMR spectra of His-lc2 (Fig. 3e), is also frequently observed in β-flakes containing polypeptides such as immunoglobulin domains (Riechmann & Davies 1995).

To determine the thermodynamic stability of the selected chimeras, energy unfolding (AG) of the six proteins was studied from their thermodenaturation curves as measured by CD (Fig. 2b). The folding energies of His-a1, His-26, His-lb11, His-2f3 and His-lg6 are between 1.6 and 2.4 kcal/mol (Table 1). These values are lower than those for typical natural proteins, and corresponding to the so far most stable of de novo designed p-proteins, p-doubled (2.5 kcal/mol; Quinn et al. 1994). However, the His-lc2 protein selected from the genomic library had a significantly higher folding energy of 5.3 kcal/mol, which falls within the normal range for natural proteins (5 to 15 kcal/mol; Pace 1990). His-lc2 is actually 1.7 kcal/mol more stable than His-CspA.

The relative folding stabilities of His-2f3 and His-lc2 were confirmed through the rate of exchange of their amide protons in D2O as observed in NMR experiments. For His-2f3, a 1D-<1>H NMR spectrum taken after incubation for 24 h in D20 buffer at 25°C showed the complete yield of its amide protons (Fig. 3a,b). In contrast, amide yield in His-lc2 was slow and allowed the observation of many amide protons in a <1>H NMR spectrum after 24 h at 25°C in D2O (Fig. 3c, d). A group of amide signals between 8.7 and 10 ppm was also detectable 3 weeks later at approx. 40% of their original intensity.

Example 8

Proteolytic stability of chimeras as soluble proteins

In addition to the spectroscopic evidence for folding stability (see Example 7), stability was also confirmed by exposure of the isolated chimeric proteins to proteases in solution. Stability data described in Example 7 of the soluble chimeric proteins from Example 6 corresponded essentially to the degree of their protection from proteolysis by trypsin, thermolysin (both used during selection) and chymotrypsin (Fig. 1b). Tryptic degradation of the N-terminal His-tag through cleavage after Argll was observed for all six proteins. The arginine was introduced as part of the expression vector immediately C-terminal to the N-terminal His-tag. His-lc2 (with a folding energy of 5.3 kcal/mol) is not further degraded by any of the proteases, confirming its high conformational stability, but the other proteins are partially proteolyzed within the main body of the polypeptides. This is consistent with a partial unfolding expected from a folding energy of approx. 2 kg/mol). Thus, although all the proteins are resistant to proteolysis (eg compared to ...-cleavage of His-tag by' Arg), the resistance varies between the proteins and according to the conditions.

Example 9

) Sequence duplication in selected chimeric proteins

As described in Example 4 above, in 20 of 24 sequenced chimeric proteins, which were selected from the plasmid-derived library, the N-terminal half

in complemented with another fragment from CspA. In fact, the chimeric proteins D6 and G4 both comprise a partial

duplication in their N-terminal half. Phage D6 contains in addition to the N-terminal half of CspA (units 1 to 36 as part of the cloning vector) the core of CspA (units 17 to 53) (Table 1). Phage G4 contains as an insert a partial duplication of the N-terminal half of CspA (units 2 to 19). This result indicates that (partial) duplication of amino acid segments can lead to the formation of stably folded protein domains.

Example 10

Duplication of homologous elements in stably folded chimeric proteins

No direct structural information is available for the seven DNA fragments, which were found after selection of the genomic library (Example 1) and which were expressed in their natural reading frame. However, one has a high level of sequence identity with a sequence neighbor of known three-dimensional structure (as identified by BLAST analysis of the E.coli genome). The insert of phage 1B11 spans units 364 to 398 of the E. coli 30S ribosomal subunit protein S1 (gene identification no. 1787140), whose units 369 to 397 have a 52% identity to units 11 to 39 of the Sl RNA binding domain from E.coli -polynucleotide phosphorylase. These comprise a region of four β-strands in the 3D structure of the Sl domain, which like CspA forms a β-barrel but with an intercalated helix (Bycroft et al., 1997).

The two S1 domains (of the 30S ribosomal protein and of the phosphorylase) are, according to their sequence similarity and identity, homologous to CspA. The assembly of the segments in the chimeric protein 1B11 therefore represents an assembly of corresponding regions from homologous polypeptide domains (which also form the same structural fold). This result indicates that a (partial) duplication of homologous amino acid segments can lead to the formation of stably folded protein domains.

Example 11

Evidence for complementation with elements of similar structure from proteomic analysis in a chimeric protein

20 of the 24 most stable phage clones selected from the plasmid-derived library (Example 2) contained inserts originally from the CspA gene in the correct reading frame (see Example 4). These 20 comprise three different clones (A1, D6, G4). Al contains a deleted version (units 1 to 52) of the CspA wild-type gene, which must have been established through a deletion within a phagemid clone originally comprising a larger insert (Table 1). Phage D6 contains in addition to the N-terminal half of CspA (units 1 to 36 as part of the clone vector) the core of CspA (units 17 to 53)

(Table 1). Phage G4 contains as an insert a partial duplication of the N-terminal half of CspA (units 2 to 19). The complementary sequences in all three clones include regions of CspA,. which in the CspA structure forms the P~strand region. Thus, sequences that form the same type of secondary structure are joined together in the chimeric proteins A1, D6 and G4.

No direct structural information is available for the seven DNA fragments, which were found after selection of the genomic library (Example 1) and which were expressed in their natural reading frame. However, one has a high degree of sequence identity with a sequence neighbor of known three-dimensional structure (as identified by a BLAST analysis of the E.coli genome). The insert in phage 1B11 spans units 364 to 398 of the E. coli 30S ribosomal subunit protein Sl (Gene ID No. 1787140), whose units 369 to 397 have a 52% identity to units 11 to 39 of the Sl RNA-binding domain from E. coli polynucleotide phosphorylase. These comprise a region of four (3-strands in the 3D structure of the Sl domain, which, in the same way as CspA, forms a P~..., but with an intercalated helix (Bycroft et al. 1997). Thus, sequences that forms the same type of secondary structure in the chimeric protein 1B11.

Thus, in the cases of the His-A1, His-d6 and His-lb11 proteins, juxtaposition of sequences forming the same type of secondary structure has led to the formation of stably folded chimeric proteins. Gene fragments selected from both libraries appear to be enriched for sequences that form primarily β-structure in their original protein. Such sequences may be more frequently able to form a stable domain with another gene fragment, originally encoding parts of a P-..., than sequences of helix origin.

Example 12

Evidence for complementation with elements of different structure from proteomic analysis in selective chimeric proteins

No direct structural information is available for the seven DNA fragments, which were found after selection of the genomic library (Example 1) and as expressed in their natural reading frame. However, one has one. high degree of sequence identity with a sequence neighbor of known three-dimensional structure (identified by BLAST analysis of the E. coli genome. The insertion of 3A12 spans units 52 to 80 of the putative transport-periplasmic protein (Gene ID #1787590) sharing a 48% sequence identity with units 30 to 58 of the Salmonella oligopeptide-binding protein. In its 3D structure (Tame et al. 1994), these units form a helix and two short antiparallel P strands. The oligopeptide-binding protein as a mixed a/p protein has no structural homology with CspA and its units 52 to 80 form no part of a (3-barrel. Thus, sequences from different folds in the chimeric protein 3A12 are assembled. Thus, as the gene fragments selected from both libraries appear to be enriched for sequences that form primarily (3 structure in their native protein, polypeptide sequences that originally form different folds are also represented.

Example 13

Effects of modified selection conditions

Proteolytic selection appears to favor phage display chimeric proteins with higher folding stability than those displaying chimeras with high melting points. From the plasmid-derived library, the phage clone exhibiting the more stable protein A1 was selected twice as frequently as the less stable D6, which however has the highest melting point (Table 1). In the case of the genomic library, the phages exhibiting the two most stable proteins (1C2, 1G6) were found four and five times, while the phages in the two least stable proteins (1B11, 2F3) were only found twice after selection. This suggests that escape from proteolysis depends more on stability than on the melting point as long as the proteolysis is carried out at temperatures well below the melting point. Higher proteolysis temperatures than used here can therefore enable more frequent selection for proteins with a higher melting point, while energetically more stable proteins are likely to be enriched if the phages are proteolysed for a longer time.

Such modified conditions can increase the frequency with which polypeptides exhibiting stabilities of natural proteins are selected from arbitrary combinatorial libraries. Further improvements can be expected when using a much larger repertoire, e.g. established by scale-up, by improvements in transfection efficiency of plasmid, phagemid or phage replicon into cells, or by other techniques such as in vivo recombination using the cre-lox system (Sternberg&Hamilton 1981). Alternatively, or in addition, the repertoires can be further diversified by means of mutagenesis before or after selection. Effective repertoire sizes can be further increased, when recombination partners are enriched before recombination for in frame, no stop codon containing DNA fragments.

The present methodology enables the selection of new chimeric proteins, which have been established through the recombination of natural genes, and which can combine properties from different molecules. By using suitable combinatorial partners, polypeptides can be established that exhibit desirable functions (such as a target binding site or an antigenic epitope) from native proteins, while removing undesirable properties (such as an unwanted receptor binding site or unwanted epitopes). For this purpose, proteolytic processing can be combined with selection for binding.

In the case of selection for binding of chimeric proteins to a ligand, it may be advantageous to increase the copy number of phage-displayed fusion proteins. An increased copy number of detected p3 fusion proteins, where there may be up to five on each phage particle, will result in multiple binding ratios for a single clone, which may enable selection even in the case of chimeric proteins with a low affinity for the ligand. Copy number for fusion proteins in subject presentation can e.g. is increased, as phagemid-encoded fusion-p3 fusion proteins are recovered for phage preparation with a helper phage lacking the gene for p3 (Rekonjac et al. 1997).

Example 14

Secondary modifications of selected chimeras

The binding activity of chimeric proteins established through random recombination of polypeptide segments for a given ligand may be low, even if the original proteins for these segments have a high affinity for such a ligand. Thus, any newly assembled polypeptide segment is expected to have some effect on the structure of the others when compared to its structure in the original protein. As a consequence, most binding sites will no longer fit the ligand with the same precision, resulting in a reduced affinity. It is therefore assumed that it may be necessary to improve such binding sites, as soon as a new chimeric protein has been established as part of a combinatorial library.

Improvements of selected chimeric proteins can be achieved by secondary modification or mutation. Such modifications can be made to improve binding, but they can also be made to increase stability and/or to introduce new binding or enzymatic functions. The type of modification and its location in the chimeric protein (ie, which old amino acids are replaced with new ones) can be based on rational design principles or partially or completely arbitrary. Modifications can be introduced by site-directed mutagenesis (Hutchison III et al. 1978) or by site-directed random mutagenesis (Riechmann&Weill 1993) followed by selection or screening for activity or stability in the resulting mutant chimeras. Alternatively, completely random mutagenesis (through e.g. error-prone PCR multiplexing, Hawkins et al. 1992) of either one or both segments (or indeed their linker sequence) of the chimeric protein or through passage of the phagemid through an E. coli mutator strain (Low et al. 1996) followed by selection and/or screening for binding, enzymatic activity or stability.

Modifications may further include the deletion of units, or the introduction of additional units. In particular, the linker and end regions of the combined polypeptide segments can be expected not to be optimized. The connecting regions can limit interactions between the assembled segments, which can be released by introducing additional units within the connecting region. Regions near the end of the chimeric protein may include terminal units that do not participate in the folding of the domain, and their deletion may improve the overall integrity of the protein.

We showed for one of the chimeric proteins how their stability was improved based on rational design. His-2f3 was established through combinatorial shuffling of the N-terminal half of E.coli protein CspA with arbitrary amino acid segments encoded by the fragments of the E.coli genome (Example 1). The sequences and genetic origin of the arbitrary fragment are given in Table II. The spectroscopic analysis of His-2f3 (Example 7) indicates a fold that is rich in (3) structure. If His-2f3 folds (such as CspA) into a β-barrel, certain sequence requirements must be met to improve the stability of the barrel .

In CspA, the hydrophobic side chain of unit Leu 45 closes one end of its β-barrel, and Gly48 and Gln49 form a curve between two β-strands of the polypeptide fold to enable the formation of backbone hydrogen bonds of the subsequent P-strand with the N- terminal P strand of CspA. Within this thread, the side chains of the two hydrophobic units (Val51, Phe53 and Ile55) point to the inside of the barrel. His-2f3 does not meet these requirements exactly, but has a corresponding motif within its genomic segments, with the units Pro58, Gly61, Ala62, Met64, Phe66 and Ala68 (in its genomic segment) showing the same spatial distribution as the motif described in CspA (compare Table III).

We therefore mutated the genomic segment in 2f3 at positions 58 (P to L), 62 (A to Q) and 68 (A to L) to match the amino acid types described for the motif in CspA, while the units at positions 61, 64 and 66 in 2f3 were already judged to be identical or sufficiently similar. As summarized in Table III, the combined P58L and A62Q mutations increased the stability of 2f3 to 6 kcal/mol, both of which are within the range of typical natural protein domains and are 1.6 kcal/mol higher than for CspA itself. A68L had no positive effect in 2f3.

In addition, the two C-terminal units (PW) of 2f3 (compare Tables II and III) were removed, as these were partially degraded in the original expressed soluble 2f3 protein. Removal of these units had no significant effect on the overall stability of 2f3, but resulted in a more homogeneous protein preparation after expression, which e.g. is advantageous for structural studies such as NMR. This result shows that the new chimeric proteins can be improved after selection through further modifications, in this case based on a rational design.

Example 15

Cross-reactivity of anti-CspA antisera with chimeric proteins

One possible application of chimeric proteins is their use as vaccines against the original polypeptide of one or more of the recombinant amino acid sequences. To this end, antisera against the chimeric protein will be cross-reactive with the original polypeptide (and indeed vice versa).

A rabbit was immunized with CspA using Freund's adjuvant (see methods). The resulting anti-serum recognized immobilized biotinylated CspA. Binding of rabbit antiserum to the immobilized Biotin-CspA could be competed with soluble CspA and to varying degrees with the chimeric protein His-lc2, His-2f3 and His-lbll (Fig.4). This result shows that an immunization with CspA results in an immune response containing antibody that cross-reacts with all three of the analyzed chimeras. Thus, it may therefore also be possible to achieve an immune response against CspA if only one of the chimeras is used for vaccination. The immune response can be expected to be directed against both linear and conformational determinants of CspA.

Example 16

Selection of chimeric proteins for binding

In the previous examples, stably folded chimeric domains were selected by proteolysis through combinatorial linking of the N-terminal half of E.coli protein CspA with amino acid segments encoded by the fragments of the E.coli genome (Ex. 1 and 3). A number of these chimeric proteins are expected to bind a polypeptide fold similar to that of CspA as the secondary structure prediction and spectroscopic analysis of the four chimeras described (Example 7) indicate a fold rich in (^-structure.

It is possible that the RNA binding function (Jiang et al. 1997) of CspA is maintained in some of the selected chimeras. The nucleic acid binding site in CspaA is proposed to be located on the surface formed around Trpll, Phel8, Phe20, Phe31 and Lys60 (Newkirk et al. 1994; Schroder et al. 1995). While the four aromatic units are part of the N-terminal half of CspA and are therefore present in all members of the genomic repertoire (Example 1), unit Lys60 is not. It seems likely that in some of the chimeric proteins the nucleic acid binding activity will be maintained, and such proteins can, for example, be selected by the binding of phage displaying the protein for nucleic acid immobilized on solid phase. (However, the phage display system used in the above experiments will be unsuitable as the barnase marker maintains nucleic acid binding activity).

Furthermore, for functional selection it may be necessary to design a phage display system that enables multiple display of the functional protein, thereby promoting selection of chimeric proteins with low affinity for the ligand (in this case nucleic acid) through the resulting ... effect. This can be achieved in the case of chimeras fused to phage coat protein p3, e.g. through the use of a phage vector-like phage-fd (Zacher et al. 1980), through the use of a phagemid in combination with a helper phage lacking the phage p3 gene (Rakonjac et al. 1997), or through an increased expression of functional chimera -p3 fusion protein. Alternatively, multiple display can be achieved through fusion to a different phage coat protein, such as p8.

Of particular importance is binding of the chimeric domains to antibodies. If antiserum to the original protein was used for selection, this would be expected to direct selection to any of the epitopes in the chimeric protein that are similar to those in the original protein and that are represented in the antiserum. Alternatively, monoclonal antibodies can be used that would select for those clones that bind a single epitope similar to that in the original protein. A number of these chimeric proteins are expected to form a polypeptide fold similar to that of CspA, with the secondary structure selection and spectroscopic analysis of the four chimeras described in Example 7 indicating a fold rich in P structure.

If some of the recombined chimeric proteins within the repertoire resemble the folding of CspA, it should be possible to enrich such proteins through binding to antibodies that specifically recognize CspA.

Example 15 already describes that an anti-CspA antiserum cross-reacts with three of the chimeric ones selected through proteolysis (and barstar binding) alone. Anti-CspA antiserum can therefore act as a reagent to enrich the combinatorial library from Example 1 specifically for phage exhibiting chimeric proteins most similar to CspA.

A rabbit anti-CspA serum was fractionated through binding to Biotin-CspA immobilized to streptavidin agarose to enrich for this against conformational determinants of CspA. Purified CspA-specific (anti-CspA) rabbit antibodies (IgG) were tested for anti-CspA binding activity as described in Example 5. For use in phage selection, anti-CspA rabbit antibody was immobilized on a streptavidin-coated ELISA well plate through a commercial biotinylated anti-rabbit IgG antiserum. Phage (7xl0<9>cfu) from the genomic library of chimeric proteins (Example 1), which had undergone one round of proteolytic selection (followed by barstar ligation, see Example 3), were treated with trypsin and thermolysin (see Example 3) followed by binding to CspA-specific rabbit antibody in 2% BSA in PBS. After washing with PBS and 40 mM DDT, 4.3 x 10<3> bound phage were eluted at pH 2, neutralized and used for infection of bacterial cells.

96 of the resulting clones were grown in a multi-well plate and were infected with helper phage KM13 for phage production. Phage from the culture supernatant of the infected bacterial clones was bound to anti-CspA antibody, which in turn had been immobilized to a streptavidin-coated plate with a biotinylated goat anti-rabbit IgG antiserum. Bound phage were washed with PBS, exposed to trypsin and thermolysin after immobilization as before, washed with

PBS, and residual phage was detected with an anti-M13-HRP conjugate. Sequences of the new clones with the strongest signal that remained after proteolysis. Seven of these clones were identical (two) or nearly identical (five clones had one less unit at the N-terminal end of the genomic insert and two different units at the C-terminal end) to clone 2f3, which had been selected previously, but not with the same high frequency, after proteolytic selection/barstar binding (Examples 3 and 4). The two remaining sequences had not previously been observed. Purified phage of 2f3 and 2f3-like clones were confirmed to be highly reactive with the purified rabbit anti-CspA antibody, even after exposure to trypsin in solution, confirming that its protease-resistant folded sequences bind to antibody. Together with the fact that the antiserum had been fractionated for binding to the folded CspA, this strongly indicates that the selection has been against a conformational determinant. The ELISA in Fig. 4 (see Example 15) shows that the corresponding chimeric protein also interacts in its soluble version with an anti-CspA antiserum.

This experiment suggests how it is possible to identify "isosteric" peptides (same conformation in the original protein and the chimeric domain). It also indicates that the method can be used for vaccination against a conformational segment of the protein, it should thus be similarly possible to use 2f3 for vaccination, and to produce anti-serum that recognizes the conformation of the N-terminal portion of CspA.

Methods

Vector constrac ions

The gene for the H102A mutant of barnase (Meiering et al. 1992) was fused to the N-terminus of the gene 3 protein (p3) of phage fd (Zacher et al. 1980) in a modified phagemid pHENl

(Hoogenboom et al. 1991) between the DNA encoding the pelB leader peptide and the mature p2 after PCR amplification with appropriate oligonucleotides using NCoI and Pstl restriction sites to establish the vector p22-12. Into p22-12, appropriate amplified portions of E. coligene CspA (Goldstein et al. 1990) were cloned between the barnase and p3 genes using Pstl and Notl restriction sites. In the resulting phagemid vector pC5-7, the barnase gene is followed by the N-terminal 36 units of CspA (where the Nterminal Met is mutated to Leu to accompany the Pstl site) and the DNA sequence GGG AGC TCA GGC GGC CGC AGA A (SacI and Note restriction sites in italics) before GAA codon for the first unit (Glu) of p3. In pC5-7, the barnase Csp cassette is out of frame with the p3 gene. In the control vector pCsp/2, the barnase Csp cassette is in frame with the p3 gene, but the first codon of the linker DNA forms an opal stop codon.

Vectors for the cytoplasmic expression of soluble proteins were constructed by subcloning genes from the phage mites into BamHI and HindIII sites in a modified QE30 vector (Qiagen). This vector is identical to QE30 except for a tetra-His tag. During PCR-mediated sub-cloning using the primers CYTOFOR (5'-CAA CAG TTT

AAG CTT CCG CCT GAG CCC AGG-3') and CYTOBAK (5'-CCT TTA CAG

GAT CCA GAC TGC AG-3') opal stop codon was converted to Trp-encoding TGG triplet.

Construction of library

As templates for random amplifications, 100 ng of pBCSK (Stratagene)-based plasmid containing the

complete CspA coding region or genomic DNA (2 µq cut with SacI) from E. coli strain TG1 (Gibson 1984) prepared as described (Ausubel et al. 1995) used in 25 PCR cycles with an annealing temperature of 38°C at use

deis of the oligonucleotide SN6NEW (5'-GAG CCT GCA GAG CTC AGG NNN NNN-3') at 40 pmol/ml for plasmid or in 30 PCR cycles with an annealing temperature of 38°C using oligonucleotide SN6MIX (5' -GAG CCT GCA GAG CTC CGG NNN NNN-3') at 40 pmol/ml for the genomic DNA. PCR products were extended for a further 30 cycles with an annealing temperature of 52°C using the nucleotide NOARG (5'-CGT GCG AGC CTG CAG AGC TCA GG-3' at 4000 pmol/ml) for the plasmid and the oligonucleotide XTND (5' -CGT GCG AGC CTG CAG AGC TCC GG-3' at 4000 pmol/ml) for the genomic DNA. PCR products of approx. 140 bp was purified from an agarose gel, and re-amplified in 30 PCR cycles using the oligonucleotide NOARG at an annealing temperature of 50°C.

Resulting fragments were cut with SacI, purified and ligated into the phosphatase-treated and SacI vector pC5-7. Ligated DNA was electroporated into TG1 to establish a plasmid-derived repertoire of 1.7x10<8> clones and a genomic repertoire of 1.0x10<8> clones. Both libraries contained approx. 60% of the recombinants contained monomeric inserts, while the remainder contained oligomeric inserts. Ligation background was less than 1% for both ligations. Due

of differences at the 3' end of the PCR primers XTND and NOARG

i , 40% of clones with in-frame insertions in the genomic library contained a GGA-encoded Gly unit as part of the 3'-SacI site, while the remaining clones contained the TGA-encoded opal stop codon at the same position. All members of it

plasmid-derived libraries with in-frame inserts contained

in the TGA-encoded opal stop codon at this position.

Selections

For selections, approx. 10<10>colony forming units (cfu)

i of subjects treated with 200 nM trypsin (Sigma T8802) and 384 nM

thermolysin (Sigma P1512) TBS-Ca buffer (25 mM Tris, 137 mM NaCl, 1 mM CaCl2, pH 7.4) for 10 min. at 10°C. Proteolyzed phage were captured for 1 hour with biotinylated C40A, C82A double mutant barnase inhibitor barstar (Hartley 1993, Lubienski et al. 1993) immobilized on a streptavidin-coated microtiter plate (Boehringer) wells in 3% Marvel in PBS. The wells were washed 20 times with PBS and once with 50 mM dithiothreitol (DTT) in PBS for 5 min. to elute phage containing proteolyzed p3 fusions held together only by disulfide bridges. Bound phage was eluted at pH 2, neutralized at pH 7, and propagated after re-infection.

Subject ELISA

Proteolysis and binding of purified phage (approx. 1010 cfu per well) to immobilized barstar was carried out as above. Phage remaining after washing with PBS and DTT were detected in ELISA with anti-M13 phage antibody, horseradish peroxidase (HRP) conjugate (Pharmacia) in 3% Marvel in PBS. Unpurified phage from culture supernatants were bound to the biotinylated barstar, and then proteolyzed in situ. Purified phage were proteolysed in solution and proteases were inactivated with Pefabloc (Boehringer) and EDTA before uptake.

Anti-CpsA antisera

A first anti-CspA serum (as used in Fig. 4) was collected from an immunized rabbit. The rabbit was injected once with refolded (see above) His-CspA (0.5 ml at 1.75 mg/ml PBS) mixed with 1:1 Freud's complete adjuvant, followed by two injections of refolded His-CspA (0.5 ml at 1.75 mg/mg PBS) mixed 1:1" with Freud's incomplete adjuvant at 4-week intervals to speed up the immune response. Anti-sera used were obtained from blood drawn 10 days after the second boost.

An anti-CspA serum used for purification of anti-CspA-specific antibody in Example 16 was collected from a different immunized rabbit. The rabbit was injected 1 time with refolded (see above) His-CspA (0.5 ml at 1.75 mg/ml PBS) mixed 1:1 with Freud's complete adjuvant, followed by 3 injections of refolded His-CspA (0, 5 ml at 1.75 mg/ml PBS) alone at 4-week intervals to boost the immune response. Anti-sera used were obtained from blood taken 10 days after the third boost. 1 ml of this antiserum was purified on 0.2 ml of streptavidin-agarose (Pierce No. 53117), to which approx. 0.1 mg Biotin-CspA (see below) was bound, after washing with PBS, elution at pH 2, followed by neutralization and buffer exchange to 3.5 ml PBS (ie at 3.5-fold dilution compared to the original anti-sera). When used for phage selection, the purified anti-CspA antibody was diluted 500-fold in PBS for binding to a biotinylated goat antibody antiserum (Sigma B-7389) immobilized in streptavidin-coated ELISA wells.

His-CspA, as used for immunization and data in Table III and Fig. 4, was purified from the unfractionated E. coli cell pellet using NTA-agarose after solubilization with 8M urea in TBS. Before elution with 200 mM imidazole PBS, agarose-bound His-CspA was renatured with 8 M to 0 M urea gradient TBS. Eluted protein was dialyzed against PBS.

For biotinylation, CspA was modified through the addition of cysteine-glutamine-alanine units as a C-terminal marker, introduced at the gene level using suitable PCR primers. The corresponding His-CspA-Cys protein was expressed, purified and refolded as His-CspA except for the addition of 0.5 mM DDT to all solutions. The NTA-agarose with bound His-CspA-Cys was washed with 5 volumes of PBS (all solutions without DTT from this step forward) and mixed with biotinylated reagent EZ-Link™Biotin-HPDP (Pierce) for biotinylation according to the manufacturer's instructions. After 1 hour, the agarose with the bound and biotinylated protein was washed with 10 vol. PBS, eluted with

200 mM imidazole in PBS and the buffer yield to PBS.-Biotinylation of the must His-Biotin-CspA was verified by MALDI mass spectrometry using SELDI (Ciphergen Systems).

Binding of rabbit anti-CspA antisera to CspA was analyzed after immobilization of biotinylated His-Csp-Cys (at 0.25 µg/ml in PBS) on streptavidin-coated ELISA plates (Boehringer-Mannheim). Rabbit anti-CspA serum was diluted 1/30,000 in 2% bovine serum albumin in PBS, and pre-incubated with varying amounts of purified competitors (see Fig. 4) before binding to the ELISA well. Bound rabbit antibody from the serum was detected with an HRP-conjugated goat anti-rabbit IgG antiserum (Sigma).

2f3 - nm tsm. ter

The gene for the 6H-2f3 protein (comparison Table III) was prepared by PCR with the primers QEBACK (5'-CGG ATA ACA ATT TCA CAC AG-3') and 2F3FOR (5'-GGC CGC CTC AAG CTT TTA AGG CGG ATG GTT GAA-3') using the 2fg gene in QE30 (compare Table II) as a template. Mutant genes for the 6H-2f3 protein were prepared by PCR amplification of the partial 2f3 gene using custom designed primers, and the same template. For each mutant, two PCR products (covering the N- and C-terminal portion of the 2f3 gene, respectively) were purified, denatured, annealed and extended. Full-length mutant genes were specifically re-amplified using two outer primers BACKTWO (5'-CCT TTA CAG GAT CC-3') and 2F3FOR. Complete genes were digested with HindIII and BamHI, and cloned into unmodified QE30 vector (Qiagen, which encodes a 6-histidine-containing N-terminal marker).

For the mutant 6H-2f3-P58L, the primers were 2F3F2 (5'-GGT AAA AAG

CAT GAT TGC GCC AAT TTC TAG CTC GCC TGC-3'), CYTOBAK (for

the N-terminal half), 2F3B0 (5'-GGT AAA AAG CAT GAT TGC G-3') and QEFOR (5'-GTT CTG AGG TCA TTA CTG G-3') (for the C-terminal half) were used. For the mutant 6H-2f3-P58L.A62Q, the primers were 2F3F1 (5'GGT AAA AAG CAT GAT TTG GCC AAT TTC TAG CTC GCC TGC-3'). CYTOBAK (for the N-terminal half), 2F3B0 and QEFOR (for the C-terminal half) used. For the mutant 6H-2f3-P58L.A62Q.A68L, the primers 2F3F1, CYTOBAK (for the N-terminal half), 2F3B1 (5'-AAT CAT GCT TTT TAC CCT AAT GGA TGG C-3') and QEFOR (for the C -terminal half) applied.

Protein expression, purification and analysis

Proteins were expressed by induction of exponential bacterial cultures at 30°C, and purified from the soluble fraction of cytoplasm using NTA-agarose according to the Qiagen protocol. His-lg6 was purified after solubilization with 8 M urea in TBS, and refolded by dialysis from 8 M, 4 M, 2 M, IM, 0.5 M to 0 M urea in TBS. Proteins were further purified by gel filtration on a Superdex-75 column (Pharmacia). The molecular weight of the proteolytic fragments was determined using the super-enhanced laser desorption/ionization (SELDI) technique (Hutchems & Yip 1993).

Proteolysis of soluble proteins (approximately 40 µM) was performed using 40 nM trypsin, thermolysin or achymotrypsin (Sigma C3142) in TBS-Ca at 20°C for 10 min. Circular dichroism spectra and thermodenaturation were obtained as described (Davis&Riechmann 1995). Thermodenaturation of 10 µM proteins (His-lc2 at 2U.M) in PBS was followed by a wavelength between 220 nm and 225 nm (His-lc2 in 2.5 µM phosphate buffer, pH 7, at 205 nm). Nuclear magnetic resonance experiments were performed on a Bruker DMX-600 spectrometer as described (Riechmann & Holliger 1997) using the watergate sequence (Piotto et al. 1992) for water suppression with protein at 1 mM in 20 mM phosphate buffer at pH 6.2 containing 100 mM NaCl in 93% H2O/7% D2O or 99.9% D2O.

All publications indicated in the description above are incorporated herein by reference. Various modifications and variations of the described methods and system according to the invention will be obvious to those skilled in the art without deviating from the scope and idea of the invention. Although the invention has been described in connection with specifically preferred embodiments, it is to be understood that the invention as claimed shall not be limited to such specific embodiments. In fact, the idea is that various modifications of the described embodiments of the invention which are obvious to those skilled in the art of molecular biology or related fields should be within the scope of the accompanying patent claims.

1. The conformational stability AG (kcal/mol) at a given temperature T was calculated using the Gibbs-Helmholtz equation AG (T) = AHm (1-T/Tm) - ACp [(Tm-T) - In (T/Tm )] while ..the midpoint of thermal unfolding (T m) and the enthalpy change for (AHm) at Tm from the denaturation curve (Agashe&Udgaonkar 1995) and assuming for ACp (the difference in heat capacity between unfolded and folded conformation) a value of 12 cal . Per unit (Edelhoch&Osborne 1976).

2. Sequences shown are those following the N-terminal half of CspA, which is MRGSHHHHGSRLQSGKMTGIVKWFNA

DKGFGFITPDDGSKDVFVHFSA.

3. The His-(Csp/2) protein was not found in either the soluble or insoluble fraction of the cytoplasm, probably due to degradation inside the cell.

Segments that maintain 80% barstar binding activity after proteolysis of phage in situ3, and in solution<b>and those purified as chimeric proteins<0>. The sequences of the genomic segment<d>as a C-terminal addition to the N-terminal region of CspA (LQSGKMTGIV KWFNADKGFG FITPDDGSKD VVVHSAGSS) are indicated, and sequences expressed in frame with the original gene are shown in italics. The localization of each segment within the E.coli genome is indicated by the nucleotide number in the EMBL database, and the name of the original gene, and for those expressed in the same frame as the original gene, the unit number of the corresponding protein and its ID in the Swiss protein database given A single base pair deletion after the first 29 base pairs in the DNA insert in lbll means that the first 10 units are out of frame with the rspA gene<9>. (a) The amino acid sequence of CspA is that of the native gene as in the EMPL database. The numbering of the 2f3 sequence takes into account the N-terminal His-tag (MRGSHHHHHHGSRLQ). The C-terminal units PWAQAEA (compare 2f3 in Table 1) were deleted in the construct used for the data here, being partially cleaved in the expressed protein of the original His-2f3 construct, indicating that they do not participate in the folding of the chimeric the domain. Their deletion had no significant effect on the overall folding stability of the domain

(1.8 vs. 1.9 kcal/mol in the 2f3 constructs used for data in Tables I and III, respectively). The units important for the (3-... folding in CspA, which

discussed in Example 14, is indicated by an asterisk, (b) Folding energies were determined as described in Table I. Mutation for 2f3 indicates the original amino acid, followed by the unit number of the new amino acid.

References:

F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith & K. Struhl (1995): Current protocols in molecular biology, Chapter 2.4.1. Wiley & Sons.

V.R. Agashe & J.B. Udgaonkar: Thermodynamics of denaturation of barstar: evidence for cold denaturation and evaluation of the interaction with guanidine hydrochloride. Biochemistry, vol 34, pp 3286-3299

(1995).

E. de Alba, J. Santoro, M. Rico & M.A. Jimenez. De novo design of a monomeric three-stranded anti-parallel /?-sheet. Protein. Sci., vol 8, pp 854-865 (1999).

S.F. Altschul, T.L. Madden, A.A. Schaffer, J.H. Zhang, Z. Zhang, W. Miller & D.J. Lipman. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res., vol 25, pp 3389-3402

(1997).

CF. Barbas, J.E. Crowe, D. Cababa, T.M. Jones, S.L. Zebedee, B.R. Murphy, R.M. Chanock&D.R. Burton. Human monoclonal fab fragments derived from a combinatorial library bind to respiratory syncytial virus-f glyco- protein and neutralize infectivity. Pro. Nati. Acad. Pollock. USA, vol 89, pp 10164-10168 (1992). L. Bogarad&S, Deem, A hierarchical approach to protein molecular evolution. Pro. Nati. Acad. Pollock. USA, vol 96, pp 2561-2595 (1999).

F. Breitling, S. Dubel, T. Seehaus, I. Klewingshaus & M. Little. A surface expression vector for antibody screening. Gene, vol 104, pp 147-153 (1991). D.R. Burton, CF. Barbas, M.A.A. Persson, S. Koenig, R.M. Chanock&R.A. Learning. A large array of human monoclonal antibodies to type 1 human immunodefiency virus from combinatorial libraries of asymptomatic sero-positive individuals. Pro. Nati. Acad. Pollock. USA, vol 88, pp 10134-10137 (1991). M. Bycroft, T.J. Hubbaard, M. Proctor, S.M. Freund & A.G. Murzin. The solution structure of the sl RNA binding domain: a member of an ancient nucleic acid-binding fold. Cell, vol 88, pp 235-242 (1997).

A.J. Caton & H. Koprowski. Influenza virus hemagglutinin-specific antibodies isolated from a combinatorial expression library are closely related to the immune response of the donor. Pro. Nati. Acad. Pollock. USA, vol 87, pp 6450-6454 (1990).

CN. Chang, N.F. Landolfi & C. Queen. Expression of antibody fab domains on bacteriophage surfaces - potential use for antibody selection. J. Immunol., vol 147, pp 3610-3614 (1991).

T. Clackson, H.R. Hoogenboom, A.D. Griffiths & G. Winter. Making antibody fragments using phage display libraries. Nature, vol 352, pp 624-628 (1991).

J. Davies & L. Riechmann. An antibody VH domain with a lox-Cre site integrated into its coding region: bacterial recombination within a single polypeptide chain. FEBS Lett., vol 377, pp 92-96 (1995).

YEAR. Davidson & R.T. Sheep. Folded proteins occur frequently in libraries of random amino acid sequences. Pro. Nati. Acad. Pollock. USA, vol 91, pp 2146-2150 (1994). J. Devereux, P. Haeberlie & 0. Smithies. A comprehensive set of sequence analysis programs for the VAX. Nucl.

Acids Res., vol 12, pp 387-395 (1984).

W. J. Dower&S.P.A. Fodor. The search for molecular diversity. 2. recombinant and synthetic randomized peptide libraries. Annu. Rep. With. Chem., vol 26, pp 271-280

(1991).

H. Edelhoch & J.C. Osborne Jr. The thermodynamic basis of the stability of proteins, nucleic acids, and membranes. Adv. Prot. Chem., vol 30, pp 183-250 (1976).

G. Eggertsson&D. Soll. Transfer ribonucleic acid-mediated suppression of termination codons in Escherichia Coli. Microbiol. Rev., vol 52, pp 354-374. D.F. Feng&R.F. Dolittle. Progressive sequence alignment as a prerequisite to correct phylogenetic trees. J. of Molec. Evol., vol 25, pp 351-360 (1987).

MD Fincuane, M. Tuna, J.H. Lees&D.N. Woolfson. Core-directed protein design. I. An experimental method for selecting stable proteins from combinatorial libraries. Biochemistry, vol 38, pp 11604-11612 (1999).

A. Fire & S.Q. Xu. Rolling replication of short dna circles. Pro. Nati. Acad. Pollock. USA, vol 92, pp 4641-4645 (1995).

S.P.A. Fodor, J.L. Read, M.C. Pirrung, L. Styer, A.T. Lu&D. Solar. Light-directed, spatially addressable parallel chemical synthesis. Science, vol 251, pp 767-773 (1991).

A. Fontana, P. De Laureto, V. de Fllipis, E. Scaramella&M. Zamboni. Probing the partially folded states of proteins by limited proteolysis. Fold. Dec, vol 2, R17-R2 6.

T.J. Gibson (1984) Ph.D. D. Thesis, University of Cambridge, UK.

J. Goldstein, N.S. Pollitt & M. Inouye. Major cold shock protein of Escherichia coli. Pro. Nati. Acad. Pollock. USA, vol 87, pp 283-287 (1990).

N. Greenfield, G.D. Fasman. Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry, vol 8, pp 4108-4116 (1969).

S.C. Hardies, W. Hillen, T.C. Goodman&R.D. Wells. High resolution thermal denaturation analyzes of small sequenced DNA restriction fragments containing Escherichia coli lactose genetic control loci. J. Biol. Chem., vol 254, pp 5527-5534 (1979).

R. W. Hartley. Directed mutagenesis and barnase-barstar recognition. Biochemistry, vol 32, pp 5978-5984 (1993). R. E. Hawkins, S.J. Russell&G. Winter. Selection of phage antibodies by binding affinity - mimicking affinity maturation. J. Mol. Biol., vol 226, pp 889-896

(1992).

R. E. Hawkins & G. Winter. Cell selection - strategies for making antibodies from variable gene libraries - trapping the memory pool. Eur. J. Immunol., vol 22, pp 867-870 (1992). M. Hecht. De novo design of P-sheet proteins. Pro. Nati. Acad. Pollock. USA, vol 91, pp 8729-8730 (1994). D. G. Higgins&P.M. Sharp. Fast and sensitive multiple sequence alignment on a microcomputer. CABIOS, vol 5, pp 151-153 (1989).

HR Hoogenboom, A.D. Griffiths, K.S. Johnson, D.J. Chiswell, P. Hudson & G. Winter. Multi-subunit proteins on the surface of filamentous phage: methodologies for displaying antibody (Fab) heavy and light chains. Nucleic Acids Res., vol 19, pp 4133-4137 (1991).

S. J. Hubbard, F. Eisenmenger & J.M. Thornton. Modeling studies of the change in conformation required for cleavage of limited proteolytic sites. Protein Science, vol 3, pp 757-768 (1994).

W. D. Huse, L. Sastry, S.A. Iverson, A.S. Kang, M. Altingmees, D.R. Burton, S.J. Benkovic&R.A. Learning. Generation of a large combinatorial library of the immunoglobulin repertoire in phage-lambda. Science, vol 246, pp 1275-1281 (1989).

T.W. Hutchens and T.T. Yep. New desorption strategies for the mass space analysis of macromolecules. Rapid Commun. Mass Spectrom., vol 7, pp 576-580 (1993).

ABOUT. Hutchison III, S. Phillips, M.H. Edgell, S. Gillam, P. Jahnke&M. Smith. Mutagenesis at a specific position in a DNA sequence. J. Biol. Chem., vol 253, pp 6551-6560

(1978).

W. N. Jiang, Y. Hou&M. Inouye. CspA, the major cold-shock protein of Escherichia vol i, is an RNA chaperone. J. Biol. Chem., vol 272, pp 196-202 (1997).

W. C Johnson Jr. Protein secondary structure and circular dichroism - a practical guide. Proteins, vol 7, pp 205-214 (1990).

S. Kamtekar, J.M. Schiffer, H. Xiong, J.M. Babik & M. Hecht. Protein design by binary patterning of polar and nonpolar amino acids. Science, vol 262, pp 1680-1685

(1993).

A.S. Kang, T.-M. Jones & D.R. Burton. Antibody redesign by chain shuffling from random combonatorial immunoglobulin libraries. Pro. Nati. Acad. Pollock. USA, vol 88, pp 11120-11123 (1991).

P. Kristensen&G. Winter. Proteolytic selection for protein folding using filamentous bacteriophages. Folding Des., vol 3, pp 321-328 (1997).

T. Kortemme, M. Ramirez-Alvardo & L. Serrano. Design of a 20-amino acid, three-starned fi-sheet protein. Science, vol 281, pp 253-256 (1998).

RAW. Lerner, A.S. Kang, J.D. Bain, D.R. Burton & CF. Barbas. Antibodies without immunization. Science, vol 258, pp 1313-1314 (1992).

N.M. Low, P. Holliger&G. Winter. Mimicking somatic hypermutation: Affinity maturation of antibodies displayed on bacteriophage using a bacterial.• J. Mol. Biol., vol 260, pp 359-368 (1996).

H. B. Lowman, S.H. Bass, N. Simpson & J.A. Wells. Selecting high-affinity binding proteins by monovalent phage display. Biochemistry, vol 30, pp 10832-10838

(1991). M. J. Lubienski, M. Bycroft, D.N.M. Jones&A.R. Fersht. Assignment of the backbone Hl and N- 15 nmr resonances and secondary structure characterization of barstar. FEBS Lett., vol 332, pp 81-87 (1993).

J. D. Marks, H.R. Hoogenboom, T.P. Bonnert, J. McCafferty, A.D. Griffiths & G. Winter. Bypassing immunization - human antibodies from v-gene libraries displayed on phage. J. Mol. Biol., vol 222, pp 581-597

(1991).

J. D. Marks, H.R. Hoogenbom, A.D. Griffiths, G. Winter. Molecular evolution of proteins on filamentous phage - mimicking the strategy of the immune system. J. Biol. Chem., vol 267, pp 16007-16010 (1992).

J. McCaffergy, A.D. Griffiths, G. Winter & D.J. Chiswell. Phage antibodies - filamentous phage displaying antibody variable domains. Nature, vol 348, pp 552-554

(1990).

E.M. Meiering, L. Serrano & A.R. Fersht. Effect of active site residues in barnase on activity and stability. J. Mol. Biol., vol 225, pp 585-589 (1992).

R. L. Mullinax, E.A. Gross, J.R. Amberg, B.N. Hau, H.H. Hogrefe, M.M. Kubitz, A. Greener, M. Altingmees, D. Ardourel, J.M. Short, J.A. Sorge & B. Shopes. Identification of human-antibody fragment clones specific for tertanus toxoid in a bacteriophage-lambda immunoexpression library. Pro. Nati. Acad. Pollock. USA, vol 87, pp 8095-8099 (1990).

E. W. Myers&W. Miller. Optimal Alignments in Linear Space. CABIOS, vol 4, pp 11,17 (1988).

S.B. Needleman, CD. Wunsch. A general method applicable to the search for sililarities in the amino acid sequences of two proteins. J. Mol. Biol., vol 48, pp 444-453.

K. Newkirk, W.Q. Feng, W.N. Jiang et al. Solution nmr structure of the major cold shock protein (cspa) from Escherichia coli - identification of a binding epitope for DNA. Pro. Nati. Acad. Pollock. USA, vol 91, pp 5114-5118 (1994). C. N. Pace. Conformational stability of globular proteins. Trends Biochem. Sci., vol 15, pp 14,17 (1990). M.A.A. Persson, R.H. Caothien&D.R. Burton. Generation of diverse high-affinity human monoclonal antibodies by repertoire cloning. Pro. Nati. Acad. Pollock. USA, vol 88, pp 2432-2436 (1991). M. Piotto, V. Saudek & V. Sklenar. Gradient-tailored excitation for single-quantum nmr-spectroscopy of aqueous solutions. J. Biomolecular NMR, vol 2, pp 661-665 (1992).

T. P. Quinn, N.B. Tweedy, R.W. Williams J.S. Richardson & D.C. Richardson. The beta double. De novo design, synthesis, and characterization of a fi-sandwich protein. Pro. Nati. Acad. Sei, USA, vol 91, pp 8747-8751 (1994).

J. Rakonjac, G. Jovanovic & P. Modell. Filamentous phage infection-mediated gene expression: construction and

propagation of the glll deletion mutant helper phage R408d3. Gene, vol 198, pp 99-103 (1997).

L. Regan. Proteins to order? Structure, vol 6, pp 1-4

(1998).

L. Riechmann&J. Davies. Backbone assignment, secondary structure and Protein A binding of an isolated, human antibody VH domain. J. Biomol. NMR, vol 6, pp 141-152

(1995). L. Riechmann&P. Holliger. The C-terminal dopamine of TolA is the coreceptor for filamentous phage infection of E. Coli. Cell, vol 90, pp 351-360 (1997). L. Riechmann&M. Weill. Phage display and selection of a site-directed randomized single-chain antibody Fv fragment for its affinity improvement. Biochemistry, vol 32, pp 8848-8855 (1993). . RT Sheep. Protein folding from a combinatorial perspective. Folding Des., vol 1, R27-R30 (1996).

H. Schindelin, M.A. Maraheil&U. Heinemann. Crystal structure of CspA, the major cold shock protein of

Escherichia coli. Pro. Nati. Acad. Pollock. USA, vol 91, pp 5119-5123.

K. Schroder, P. Graumann, A. Schnuchel, T.A. Holak & M.A. Marahiel. Mutational analyzes of the putative nucleic acid-binding surface of the cold-shock domain. Cspb, revealed an essential role of aromatic and basic residues in binding of single-stranded-DNA containing the y-box motif. Mol. Microbiol., vol 16, pp 699-708

(1995).

J. K. Scott.& G.P. Smith. Searching for peptide ligands with an epitope library. Science, vol 249, pp 386-390

(1990). V. Sieber, A. Plueckthun & F.X. Schmidt. Selecting proteins with improved stability by a phage-based method. Nat. Biotechnol., vol 16, pp 955-960 (1998).

T.F. Smith&M.S. Waterman. Comparison of Biosequences. Advances in Applied Mathematics, vol 2, pp 482-48 9

(1981).

■T.F. Smith, M.S. Waterman&J.R. Saddles. Statistical characterization of nucleic acid sequences functional domains. Nucleic Acids Res., vol 11, pp 2205-2220

(1983).

N. Sternberg&D. Hamilton. Bacteriophage Pl site-specific recombination. I. Recombination between loxP sites. J. Mol. Biol., vol 150, pp 467-486 (1981).

J. R. Tame, G.N. Murshudov, E.J. Dodson, T.K. Neil, G.G. Dodson, C.F. Higgins&A.J. Wilkinson. The structural basis of sequence-independent peptide binding by OppA protein. Science, vol 264, pp 1578-1581 (1994).

J. D. Thompson, D.G. Higgins&T.J. Gibson. ClusterW: Improving the sensitivity of progressive multiple sequence alignment through sequence weighing, position-specific gap penalties and weigh matrix choice. Nucleic Acids Res., vol 22, p 4673 (1994).

W. J. Wilbur & D.J. Lipman. Rapid similarity searches of nucleic acid and protein data banks. Pro. Nati. Acad. Pollock. USA, vol 80, pp 726-730 (1983).

K. Wuthrich. NMR of proteins and nucleic acids. Chapter 3. Wiley&Sons.

A.N. Zacher, C.A. Stock, J.W. Golden&G.P. Smith. A new filamentous phage donating vector: fd- tet. Gene, vol 9, pp 127-140 (1980).

Listed patents and patent applications:

PCT/GB00/00030

PCT/GB98/01889

WO84/03564

WO88/08453

WO91/05058

WO90/05785

WO90/07003

WO90/15070

WO91/02076

WO92/00091

WO92/02536

WO92/10092

WO93/06121

W095/11922

W095/22625

US Patent No. 4,631,211

Claims (77)

1. Chimeric folded protein domain, characterized in that, being derived from a repertoire of chimeric proteins, it comprises two or more sequence segments derived from original amino acid sequences that are non-homologous. 2. Chimerically folded protein in accordance with claim 1, characterized in that two or more of the sequence segments are combined non-covalently. 3. Chimerically folded protein domain in accordance with claim 1, characterized in that at least one of the original amino acid sequences is derived from a protein. 4. Chimerically folded protein domain in accordance with claim 3, characterized in that at least one of the original amino acid sequences is derived from a protein selected from the group consisting of a naturally occurring protein, an engineered protein, a protein with a known binding activity, a protein with a known binding activity for an organic compound, a protein with a known binding activity for a peptide or polypeptide, a protein with a known binding activity for a carbohydrate, a protein with a known binding activity for a nucleic acid, a known binding activity for a hapten, a protein with a known binding activity for a steroid, a protein with a known binding activity for an inorganic compound, and a protein with enzymatic activity. 5. Chimerically folded protein domain in accordance with claim 1, characterized in that the original amino acid sequences are derived from the open reading frames of a single genome or portion thereof: (a) where said reading frames are the natural reading frame of the genes, or (b) where said reading frames are not the natural reading frames of the genes. 6. Chimerically folded protein domain in accordance with claim 1, characterized in that the original amino acid sequences are derived from the open reading frames of two or more genomes, or portions thereof: (a) where said reading frames are the natural reading frame for the genes, or (b) where said reading frames are not the natural reading frames of the genes. 7. Chimeric protein domain in accordance with claim 1, characterized in that it is resistant to in vivo or in vitro proteolysis with protease enzymes. 8. Chimeric protein in accordance with claim 1, characterized in that the sequence segments originate from original domains with the same polypeptide folding in at least parts of the structure. 9. Chimeric protein in accordance with claim 1, characterized in that the sequence segments originate from original domains with different polypeptide folds in at least parts of the structure. 10. Chimeric protein domain in accordance with claim 1, characterized in that it has a free energy for folding that is greater than 1.6 kcal/mol. 11. Chimeric protein domain in accordance with claim 10, characterized in that the free folding energy is greater than 3 kcal/mol. 12. Chimeric protein domain in accordance with claim 11, characterized in that the free folding energy is greater than 5 kcal/mol. 13. Chimerization-folding protein in accordance with one of claims 1-12, characterized in that one or more of the sequence segments are fused to one or more additional and complete protein domains. 14. Chimeric protein domain in accordance with claim 1, characterized in that it is fused to the coat protein of a filamentous bacteriophage, where the bacteriophage encapsidates a nucleic acid that codes for the protein domain. 15. Chimeric protein domain in accordance with claim 1, characterized in that a single sequence segment originates from human proteins. 16. Chimeric protein domain in accordance with claim 1, characterized in that two or more sequence segments originate from human proteins. 17. Chimeric protein domain in accordance with claim 15 or 16, characterized in that at least one of the segments is derived from a source that is not a human protein. 18. Chimeric protein domain in accordance with claim 17, characterized in that all segments are derived from human proteins. 19. Chimeric protein in accordance with claim 1, characterized in that it comprises an epitope of the original amino acid sequence. 20. Chimeric protein in accordance with claim 19, characterized in that it comprises a conformational epitope. 21. Chimeric protein in accordance with claim 1, characterized in that it cross-reacts with antibody directed against an original amino acid sequence. 22. Chimeric protein in accordance with claim 1, characterized in that it cross-reacts with antibody directed against the folded original protein. 23. Chimeric protein in accordance with claim 1, characterized in that it is used in vaccines against one or more of the amino acid sequences from which the chimera is derived. 24. Chimeric protein in accordance with claim 1, characterized in that it is used for administration to a human for therapeutic purposes. 25. Chimeric protein in accordance with claim 1, characterized in that it is used in a commercial product to which humans are exposed. 26. Chimeric protein in accordance with claim 1, characterized in that the amino acid sequences are changed to increase the stability or function of the chimeric protein. 27. Chimeric nucleic acid, characterized in that it codes for a protein domain in accordance with claim 1. 28. Method for producing a protein domain in accordance with claim 1, characterized in that the method comprises the steps: (a) providing a first library of nucleic acids, where the library comprises coding sequences that code for sequence segments derived from one or more amino acid sequences, where the coding sequences are not selected or engineered to only code for a single and complete protein structural element or code for a complete protein domain, (b) providing a second library of nucleic acids, the library comprising coding sequences encoding sequence segments derived from one or more amino acid sequences; (c) combining the coding sequences to form a combinatorial library of nucleic acids, the nucleic acids comprising contiguous coding sequences encoding sequence fragments derived from the first and second libraries; (d) transcribing and/or translating the contiguous coding sequences to produce the encoded protein domains; (e) selecting the chimeric protein domains capable of adopting a folded structure, or fulfilling a specific function. 29. Method in accordance with claim 28, characterized in that it further comprises the steps: (f) analyzing the sequence of the selected chimeric protein domains to identify the origin of the sequence segments, and (f) comparing the sequences of each of the original amino acid sequences to identify whether the sequences of the original amino acid sequences are non-homologous. 30. Method for producing a protein domain in accordance with claim 9 or 10, characterized in that the method comprises the steps in accordance with claim 28 or 29, and the further step: (h) comparing the structure of each of the original amino acid sequences to identify whether in whole or in part they have the same polypeptide folds. 31. Method in accordance with claim 28, characterized in that steps (b) and (c) are modified as follows: (b) provides a partner coding sequence encoding a sequence segment derived from one protein; (c) combining the library and the partner coding sequence to form a combinatorial library of nucleic acids, the nucleic acids comprising contiguous coding sequences encoding sequence fragments derived from the first library and the partner coding sequence. 32. Method in accordance with claim 28, characterized in that the domains which are able to adopt a folding structure are selected by one or more methods selected from the group consisting of in vivo proteolysis, in vitro proteolysis, binding capacity, functional activity and expression. 33. Method in accordance with claim 32, characterized in that the binding ability is to an antibody directed against an original protein. 34. Method for producing a protein domain in accordance with claim 26, where the sequence segments in the original amino acid sequences are changed after their assembly, characterized by one or more of the following steps: (a) designing and introducing specific or arbitrary mutations at predetermined positions within the gene in the chimeric protein; (b) deleting nucleotides within the gene for the chimeric protein to delete amino acid units; (c) inserting the nucleotide within the gene into the chimeric protein to insert amino acid units; (d) adding nucleotides to the gene in the chimeric protein to add amino acid units; (e) arbitrarily introduce mutations into all or part of the gene encoding the chimeric protein through recombinant DNA technology; (f) arbitrarily introducing mutations into the gene in the chimeric protein through propagation in mutant cells; (g) introduction of derivatives of natural amino acids during chemical synthesis; (h) chemical derivatization of amino acid groups after synthesis; (i) multimerization of the chimeric proteins through the concatenation of two or more copies of the gene in a single open reading frame, (j) multimerization of the chimeric proteins through covalent linking of two or more copies of the chimeric protein domain after translation; (k) multimerization of the chimeric proteins through fusion to a multimeric partner. 35. Chimeric protein domain in accordance with claim 1, characterized in that it comprises at least one reaction group for covalent coupling. 36. Chimeric protein domain in accordance with claim 1, characterized in that it comprises at least one reaction group for non-covalent coupling. 37. Chimeric protein domain in accordance with claim 1, characterized in that it comprises at least one D-amino acid. 38. Chimeric protein domain in accordance with claim 1, characterized in that it comprises one non-naturally occurring amino acid. 39. Chimeric protein domain in accordance with claim 1, characterized in that it comprises at least one amino acid which has a marker or a tag. 40. Chimeric folded protein domain, characterized in that, as it is derived from a repertoire of chimeric folded proteins, it comprises two or more sequence segments derived from original amino acid sequences where each of said segments includes common sequences in the chimeric protein, and where said common sequences do not is designed or selected to consist of only one or complete structural elements. 41. Folded chimeric protein domain in accordance with claim 40, characterized in that the region of common sequences is 10 or more identical amino acid units in length. 42. Folded chimeric protein domain in accordance with claim 41, characterized in that the region of common sequences is 20 or more identical amino acid units in length. 43. Chimeric folded protein domain, characterized in that, as it is derived from a repertoire of chimeric folded proteins, it comprises two or more sequence segments where each of said segments: (a) are derived from native proteins with a common fold, and (b) comprises a common region of the joint folding, and where said joint regions of the folding are not designed or selected to consist of one or more complete structural elements. 44. Chimeric folded protein domain in accordance with claim 43, characterized in that each of said segments is derived from different proteins that are homologous in sequence. 45. Chimerically folded protein domain in accordance with claim 43, characterized in that each of said segments is derived from the same protein. 46. Folded chimeric protein domain in accordance with claim 43, characterized in that the common region of the fold is 10 or more amino acid units long. 47. Folded chimeric protein domain in accordance with claim 43, characterized in that the common region of the fold is 20 or more amino acid units long. 48. Chimerically folded protein domain in accordance with claim 43, characterized in that the amino acid sequences of the original proteins are derived from the open reading frames of a genome or part thereof, where said reading frames are the natural reading frame of the genes. 49. Chimeric protein domain in accordance with claim 40, characterized in that it is resistant to in vivo or in vitro proteolysis with protease enzymes. 50. Chimeric protein domain in accordance with claim 40, characterized in that it has a free energy for folding that is greater than 1.6 kcal/mol. 51. Chimerically folded protein in accordance with claim 40, characterized in that one or more of the sequence segments are fused to one or more additional and complete protein domains. 52.. Chimeric protein domain in accordance with claim 40, characterized in that it is fused to the coat protein of filamentous bacteriophage, where the bacteriophage encapsulates a nucleic acid which codes for the protein domain. 53. Chimeric protein domain in accordance with claim 40, characterized in that a single sequence segment originates from a human protein. 54. Chimeric protein domain in accordance with claim 40, characterized in that two or more of the sequence segments originate from a human protein. 55. Chimeric protein domain in accordance with claim 40, characterized in that at least one of the segments is derived from a source that is not a human protein. 56. Chimeric protein domain in accordance with claim 47, characterized in that all segments are derived from human proteins. 57. Chimeric protein in accordance with claim 40, characterized in that it comprises an epitope of the original amino acid sequence. 58. Chimeric protein in accordance with claim 57, characterized in that it comprises a conformational epitope. 59. Chimeric protein in accordance with claim 40, characterized in that it cross-reacts with antibody directed against an original amino acid sequence. 60. Chimeric protein in accordance with claim 40, characterized in that it cross-reacts with antibody directed against the folded original protein. 61. Chimeric protein in accordance with claim 40, characterized in that it is used in vaccines against the original protein(s) from which the chimera is derived. 62. Chimeric protein in accordance with claim 40, characterized in that it is used for administration to a human for therapeutic purposes. 63. Chimeric protein according to claim 40, for use in a commercial product to which humans are exposed. 64. Chimeric protein in accordance with claim 40, characterized in that the amino acid sequence is changed to increase stability or function of the chimeric protein. 65. Chimeric nucleic acid, characterized in that it codes for a protein domain in accordance with claim 40. 66. Method for producing a chimeric protein domain in accordance with claim 40, characterized in that it comprises the steps: (a) providing a first library of nucleic acids, the library comprising coding sequences encoding sequence segments derived from one or more amino acid sequences; (b) providing a second library of nucleic acids, the library comprising coding sequences encoding sequence segments derived from one or more amino acid sequences; (c) combining the coding sequences to form a combinatorial library of nucleic acids, the nucleic acids comprising contiguous coding sequences encoding sequence fragments derived from the first and second libraries; (d) transcribing and/or translating the contiguous coding sequences to produce the encoded protein domains, and (e) selecting the chimeric protein domains capable of adopting a folded structure, or fulfilling a specific function. 67. Method in accordance with claim 66, characterized in that the method further comprises the steps: (f) analyzing the sequence of the selected chimeric protein domain to identify the origin of the sequence segments, and (g) comparing the sequences to identify whether they comprise common sequences in accordance with claim 40. 68. Method in accordance with claim 66, characterized in that it is used for the production of a chimeric protein domain in accordance with claim 44, where step (g) is replaced, and step (h) is added, so that: (g) comparing the structure of the original amino acid sequences to identify whether the original amino acid sequences have common folding, and (h) identify whether the segments comprise a common region of the common fold. 69. Method in accordance with claim 66, characterized in that steps (b) and (c) are modified as follows: (b) providing a partner coding sequence encoding a sequence segment derived from one protein; (c) combining the library and the partner coding sequences to form a combinatorial library of nucleic acids, the nucleic acids comprising contiguous coding sequences encoding sequence fragments derived from the first library and the partner coding sequence. 70. Method according to claim 66, characterized in that the domain capable of adopting a folding structure is selected by one or more methods selected from the group consisting of in vivo proteolysis, in vitro proteolysis, binding capacity, functional activity and expression. 71. Method in accordance with claim 70, characterized in that the binding ability is to an antibody directed against an original protein. 72. Method for producing a chimeric protein domain in accordance with claim 40, where the sequence segments of the original amino acid sequences are changed after their assembly, characterized in that the method comprises one or more of the following steps: (a) designing and introducing specific or arbitrary mutations at predetermined positions within the gene in the chimeric protein; (b) deleting nucleotides within the gene for the chimeric protein to delete amino acid units; (c) inserting nucleotides into the gene for chimeric protein to insert amino acid units, (d) adding nucleotides to the gene in the chimeric protein to add amino acid units; (e) arbitrarily introduce mutations into all or part of the gene encoding the chimeric protein through recombinant DNA technology; (f) arbitrarily introducing mutations into the gene in the chimeric protein through propagation in mutant cells; (g) introducing derivatives of natural amino acids by chemical synthesis; (h) chemical derivatization of amino acid groups after synthesis; (i) multimerization of the chimeric proteins through covalent linking of two or more copies of the gene in a single open reading frame, (j) multimerization of the chimeric proteins through covalent linking of two or more copies of the chimeric protein domain after translation; (k) multimerization of the chimeric proteins through fusion to the multimeric partner. 73. Chimeric protein domain in accordance with claim 40, characterized in that it comprises at least one reaction group for covalent coupling. 74. Chimeric protein domain in accordance with claim 40, characterized in that it comprises at least one reaction group for non-covalent coupling. 75. Chimeric protein domain in accordance with claim 40, characterized in that it comprises at least one D-amino acid. 76. Chimeric protein domain in accordance with claim 40, characterized in that it comprises at least one non-naturally occurring amino acid. 77. Chimeric protein domain in accordance with claim 40, characterized in that it comprises at least one amino acid which has a marker or a tag.