WO2002068453A2

WO2002068453A2 - Methods and compositions for the construction and use of fusion libraries using computational protein design methods

Info

Publication number: WO2002068453A2
Application number: PCT/US2002/004853
Authority: WO
Inventors: Min Li; Bassil I. Dahiyat
Original assignee: Xencor
Priority date: 2001-02-22
Filing date: 2002-02-19
Publication date: 2002-09-06
Also published as: WO2002068453A3; AU2002251999A1

Abstract

The invention relates to the use of protein design automation (PDA) to generate computationally prescreened secondary libraries of proteins, and to methods and compositions utilizing the libraries.

Description

METHODS AND COMPOSITIONS FOR THE CONSTRUCTION AND USE OF FUSION LIBRARIES USING COMPUTATIONAL PROTEIN DESIGN METHODS

FIELD OF THE INVENTION

The invention relates to the use of a variety of computation methods, including protein design automation (PDA), to generate computationally prescreened secondary libraries of proteins, and to methods of making and methods and compositions utilizing the libraries. In particular, the libraries are constructed as nucleic acid/protein (NAP) conjugates that can be used in a wide variety of applications.

BACKGROUND OF THE INVENTION

Directed molecular evolution can be used to create proteins and enzymes with novel functions and properties. Starting with a known natural protein, several rounds of mutagenesis, functional screening, and propagation of successful sequences are performed. The advantage of this process is that it can be used to rapidly evolve any protein without knowledge of its structure. Several different mutagenesis strategies exist, including point mutagenesis by error-prone PCR, cassette mutagenesis, and DNA shuffling. These techniques have had many successes; however, they are all handicapped by their inability to produce more than a tiny fraction of the potential changes. For example, there are 20⁵⁰⁰ possible amino acid changes for an average protein approximately 500 amino acids long. Clearly, the mutagenesis and functional screening of so many mutants is impossible; directed evolution provides a very sparse sampling of the possible sequences and hence examines only a small portion of possible improved proteins, typically point mutants or recombinations of existing sequences. By sampling randomly from the vast number of possible sequences, directed evolution is unbiased and broadly applicable, but inherently inefficient because it ignores all structural and biophysical knowledge of proteins. In contrast, computational methods can be used to screen enormous sequence libraries (up to 10⁸⁰ in a single calculation) overcoming the key limitation of experimental library screening methods such as directed molecular evolution. There are a wide variety of methods known for generating and evaluating sequences. These include, but are not limited to, sequence profiling (Bowie and Eisenberg, Science 253(5016): 164-70, (1991)), rotamer library selections (Dahiyat and Mayo, Protein Sci 5(5): 895-903 (1996); Dahiyat and Mayo, Science 278(5335): 82-7 (1997); Desjarlais and Handel, Protein Science 4: 2006-2018 (1995); Harbury et al, PNAS USA 92(18): 8408-8412 (1995); Kono et al., Proteins: Structure, Function and Genetics 19: 244-255 (1994); Hellinga and Richards, PNAS USA 91: 5803-5807 (1994)); and residue pair potentials (Jones, Protein Science 3: 567-574, (1994)).

In particular, U.S.S.N.s 60/061,097, 60/043,464, 60/054,678, 09/127,926 and PCT US98/07254 describe a method termed "Protein Design Automation", or PDA, that utilizes a number of scoring functions to evaluate sequence stability.

In addition to computation methods, there is a significant focus on assigning biological function to the raw genomic sequences that are being rapidly elucidated. The available raw sequences usually cannot easily be translated into knowledge of their encoded biological, pharmaceutical or industrial usefulness. Thus, there is a need in the art for technologies that will efficiently, systematically, and maximally realize the function and utility of DNA sequences from both natural and synthetic sources.

Several general approaches to realize the potential functions of a given DNA sequence have been reported. One approach, which is also the primary approach in gene and target discovery, is to rely on bioinformatic tools. Bioinformatics software is available from a number of companies specializing in organization of sequence data into computer databases. A researcher is able to compare uncharacterized nucleic acid sequences with the sequences of known genes in the database, thereby allowing theories to be proposed regarding the function of the nucleic acid sequence of an encoded gene product. However, bioinformatics software can be expensive, often requires extensive training for meaningful use, and enables a researcher to only speculate as to a possible function of an encoded gene product. Moreover, an increasing number of DNA sequences have been identified that show no sequence relationship to genes of known functions and new properties have been discovered for many so-called "known" genes. Therefore, bioinformatics provides a limited amount of information that must be used with caution. Ail informatics-predicted properties require experimental approval. Another approach for associating function with sequence data is to pursue experimental testing of orphan gene function. In previously described methods, nucleic acid sequences are expressed using any of a number of expression constructs to obtain an encoded peptide, which is then subjected to assays to identify a peptide having a desired property. An inherent difficulty with many of the previously described methods is correlating a target property with its coding nucleic acid sequence. In other words, as large collections of nucleic acid and peptide sequences are gathered and their encoded functions explored, it is increasingly difficult to identify and isolate a coding sequence responsible for a desired function.

The fundamental difficulties associated with working with large collections of nucleic acid sequences, such as genetic libraries, are alleviated by linking the expressed peptide with the genetic material which encodes it. An approach of associating a peptide to its coding nucleic acid is the use of polysome display. Polysome display methods essentially comprise translating RNA in vitro and complexing the nascent protein to its corresponding RNA. The complex is constructed by manipulating the coding sequence such that the ribosome does not release the nascent protein or the RNA. By retrieving proteins of interest, the researcher retrieves the corresponding RNA, and thereby obtains the coding DNA sequence after converting the RNA into DNA via known methods such as reverse transcriptase-coupled PCR. Yet, polysome display methods can be carried out only in vitro, are difficult to perform, and require an RNase-free environment. Due to alternative starting methionine codons and the less than perfect processive nature of in vitro translation machinery, this method is not applicable to large proteins. In addition, the RNA-protein-ribosome complex is unstable, thereby limiting screening methods and tools suitable for use with polysome display complexes.

Another commonly used method of linking proteins to coding nucleic acid molecules for use with genetic libraries involves displaying proteins on the outer surface of ceils, viruses, phages, and yeast. By expressing the variant protein as, for example, a component of a viral coat protein, the protein is naturally linked to its coding DNA located within the viral particle or cellular host, which can be easily isolated. The DNA is then purified and analyzed. Other systems for associating a protein with a DNA molecule in genetic library construction have been described in, for example, International Patent Applications WO 93/08278, WO 98/37186, and WO 99/11785. Yet, these approaches have features that are not most desirable. First, the expressed protein and the corresponding cDNA are non- covalently bound. The resulting complex is not stable or suitable for many selection procedures. Second, the display systems by design are restricted to either in vitro or prokaryotic heterologous expression systems, which may not provide necessary protein modification or folding machinery for the study of eukaryotic peptides. Incorrectly folded or modified proteins often lack the native function of desired proteins and are often very unstable. Third, if displayed on the surface of a biological particle, the expressed proteins often undergo unwanted biological selections intrinsic to the displayed systems. For example, in the case of display proteins on bacterial viruses, e.g., bacteriophage, the expressed protein will be assembled as part of bacterial virus coat proteins and displayed on the surface of the bacterial virus. Interactions of the bacterial virus-bound variant protein with the surrounding environment and incorporation of the protein into the bacterial viral coat can damage the conformation and activity of the variant protein. Moreover, even if the protein is incorporated into the bacterial viral capsid, the display protein may not be in a correct geometrical or stoichiometrical form, which is required for its activity. Fourth, construction of large surface-display libraries using biological particles is time intensive, and the researcher must take precautions to ensure that the biological particle, i.e., virus or phage, remains viable. Fifth, it is known that different hosts have different codon preferences when performing protein translation. For example, in prokaryotic systems, the expression systems used for bacterial virus display, there are at least five codons commonly recognized in mammalian cells that are not readily recognized by bacteria during protein translation. Thus, mammalian sequences with these codons are not translated or are translated very inefficiently in bacteria, posing a significant negative selection.

In view of the above, it is an object of the present invention to provide computational methods for prescreening sequence libraries to generate and select secondary libraries, which can then be made and evaluated experimentally using fusion technologies, to allow the creation of nucleic acid/protein (NAP) conjugates to facilitate screening. It is a further object of the invention to provide for the identification of relevant proteins in the native cellular environment, which is a significant advantage of the use of eucaryotic systems. These and other advantages of the present invention, as well as additional inventive features, will be apparent from the description of the invention provided herein.

SUMMARY OF THE INVENTION

In accordance with the above objects, the present invention provides methods for generating a library of fusion nucleic acids comprising providing a computationally-derived primary library of candidate protein sequences and creating a library of expression vectors. Each expression vector comprises a fusion nucleic acid comprising a first nucleic acid sequence encoding a nucleic acid modification (NAM) enzyme and a second nucleic acid encoding a candidate protein sequence from the computation library. The expression vectors also comprise an enzyme attachment sequence that is recognized by the NAM enzyme. In a further aspect, the primary library comprises first candidate protein sequences, and the method further comprises generating a list of primary variant positions in the primary library and combining a plurality of the primary variant positions to generate a second library of second candidate protein sequences. These second nucleic acid encodes a second candidate protein sequence of the expression vector.

In an additional aspect, the invention provides methods of screening comprising providing a computationally-derived primary library of candidate protein sequences and creating a library of expression vectors. Each expression vector comprises a fusion nucleic acid comprising a first nucleic acid sequence encoding a nucleic acid modification (NAM) enzyme and a second nucleic acid encoding a candidate protein, as well as an enzyme attachment sequence (EAS) that is recognized by the NAM enzyme. The fusion nucleic acids are expressed under conditions whereby a library of nucleic acid/protein (NAP) conjugates are formed. The NAP conjugates each comprise a fusion polypeptide comprising a NAM enzyme and a candidate protein and an expression vector, wherein the NAM enzyme and the EAS are covalently attached. At least one test molecule is added to the NAP conjugate library and the binding of a NAP conjugate to the test molecule is determined.

In a further aspect, the expressing is accomplished by transforming the library of fusion nucleic acids into cells under conditions whereby NAP conjugates are made.

In an additional aspect, the primary library comprises first candidate protein sequences, and the method further comprises generating a list of primary variant positions in the primary library and combining a plurality of the primary variant positions to generate a second library of second candidate protein sequences. The second nucleic acid encodes a second candidate protein sequence.

In a further aspect, the invention provides methods of screening comprising providing a computationally-derived primary library of candidate protein sequences and providing a library of eucaryotic host cells each comprising at least one expression vector comprising a fusion nucleic acid as outlined above and an EAS that is recognized by the NAM enzyme, under conditions whereby a fusion polypeptide is produced and wherein at least two of said candidate proteins are different. The cells are lysed, wherein the EAS and the NAM enzyme are covalently attached to form a NAP conjugate. At least one test molecule is added and the binding of the test molecule to a NAP conjugate is determined. In an additional aspect, the nucleic acid sequences encoding candidate proteins can be derived from cDNA, genomic DNA or be random or partially random proteins.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 depicts the nucleotide sequence of Rep78 isolated from adeno-associated virus 2.

Figure 2 depicts the amino acid sequence of Rep78 isolated from adeno-associated virus 2.

Figure 3 depicts the nucleotide sequence of major coat protein A isolated from adeno-associated virus 2.

Figure 4 depicts the amino acid sequence of major coat protein A isolated from adeno-associated virus 2.

Figure 5 depicts the nucleotide sequence of a Rep protein isolated from adeno-associated virus 4.

Figure 6 depicts the amino acid sequence of a Rep protein isolated from adeno-associated virus 4.

Figure 7 depicts the nucleotide sequence of Rep78 isolated from adeno-associated virus 3B.

Figure 8 depicts the amino acid sequence of Rep78 isolated from adeno-associated virus 3B.

Figure 9 depicts the nucleotide sequence of a nonstructural protein isolated from adeno-associated virus 3.

Figure 10 depicts the amino acid sequence of a nonstructural protein isolated from adeno-associated virus 3. Figure 11 depicts the nucleotide sequence of a nonstructural protein isolated from adeno-associated virus 1.

Figure 12 depicts the amino acid sequence of a nonstructural protein isolated from adeno-associated virus 1.

Figure 13 depicts the nucleotide sequence of Rep78 isolated from adeno-associated virus 6.

Figure 14 depicts the amino acid sequence of Rep78 isolated from adeno-associated virus 6.

Figure 15 depicts the nucleotide sequence of Rep68 isolated from adeno-associated virus 2.

Figure 16 depicts the amino acid sequence of Rep68 isolated from adeno-associated virus 2.

Figure 17 depicts the nucleotide sequence of major coat protein A' (alt.) isolated from adeno- associated virus 2.

Figure 18 depicts the amino acid sequence of major coat protein A' (alt.) isolated from adeno- associated virus 2.

Figure 19 depicts the nucleotide sequence of major coat protein A" (alt.) isolated from adeno- associated virus 2.

Figure 20 depicts the amino acid sequence of major coat protein A" (alt.) isolated from adeno- associated virus 2.

Figure 21 depicts the nucleotide sequence of a Rep protein isolated from adeno-associated virus 5. Figure 22 depicts the amino acid sequence of a Rep protein isolated from adeno-associated virus 5.

Figure 23 depicts the amino acid sequence of major coat protein Aa (alt.) isolated from adeno- associated virus 2.

Figure 24 depicts the nucleotide sequence of major coat protein Aa (alt.) isolated from adeno- associated virus 2.

Figure 25 depicts the nucleotide sequence of a Rep protein isolated from Barbaric duck parvovirus.

Figure 26 depicts the amino acid sequence of a Rep protein isolated from Barbaric duck parvovirus.

Figure 27 depicts the nucleotide sequence of a Rep protein isolated from goose parvovirus.

Figure 28 depicts the amino acid sequence of a Rep protein isolated from goose parvovirus.

Figure 29 depicts the nucleotide sequence of NS1 protein isolated from muscovy duck parvovirus.

Figure 30 depicts the amino acid sequence of NS1 protein isolated from muscovy duck parvovirus.

Figure 31 depicts the nucleotide sequence of NS1 protein isolated from goose parvovirus.

Figure 32 depicts the amino acid sequence of NS1 protein isolated from goose parvovirus.

Figure 33 depicts the nucleotide sequence of a nonstructural protein isolated from chipmunk parvovirus. Figure 34 depicts the amino acid sequence of a nonstructural protein isolated from chipmunk parvovirus.

Figure 35 depicts the nucleotide sequence of a nonstructural protein isolated from the pig-tailed macaque parvovirus.

Figure 36 depicts the amino acid sequence of a nonstructural protein isolated from the pig-tailed macaque parvovirus.

Figure 37 depicts the nucleotide sequence of NS1 protein isolated from a simian parvovirus.

Figure 38 depicts the amino acid sequence of NS1 protein isolated from a simian parvovirus.

Figure 39 depicts the nucleotide sequence of a NS protein isolated from the Rhesus macaque parvovirus.

Figure 40 depicts the amino acid sequence of a NS protein isolated from the Rhesus macaque parvovirus.

Figure 41 depicts the nucleotide sequence of a nonstructural protein isolated from the B19 virus.

Figure 42 depicts the amino acid sequence of a nonstructural protein isolated from the B19 virus.

Figure 43 depicts the nucleotide sequence of oifl isolated from the Erythrovirus B19.

Figure 44 depicts the amino acid sequence of orfl isolated from the Erythrovirus B19.

Figure 45 depicts the nucleotide sequence of U94 isolated from the human herpesvirus 6B. Figure 46 depicts the amino acid sequence of U94 isolated from the human herpesvirus 6B.

Figure 47 depicts an enzyme attachment sequence for a Rep protein.

Figure 48 depicts the Rep68 and Rep78 enzyme attachment site found in chromosome 19.

Figures 49A - 49N depict preferred embodiments of the expression vectors of the invention.

Figure 50 depicts the synthesis of a full-length gene and all possible mutations by PCR. Overlapping oligonucleotides corresponding to the full-length gene (black bar, Step 1) are synthesized, heated and annealed. Addition of Pfu DNA polymerase to the annealed oligonucleotides results in the 5' -3' synthesis of DNA (Step 2) to produce longer DNA fragments (Step 3). Repeated cycles of heating, annealing (Step 4) results in the production of longer DNA, including some full-length molecules. These can be selected by a second round of PCR using primers (arrowed) corresponding to the end of the full-length gene (Step 5).

Figure 51 depicts the reduction of the dimensionality of sequence space by PDA screening. From left to right, 1: without PDA; 2: without PDA not counting Cysteine, Proline, Glycine; 3: with PDA using the 1% criterion, modeling free enzyme; 4: with PDA using the 1% criterion, modeling enzyme-substrate complex; 5: with PDA using the 5% criterion modeling free enzyme; 6: with PDA using the 5% criterion modeling enzyme-substrate complex.

Figure 52 depicts a preferred scheme for synthesizing a library of the invention. The wild-type gene, or any starting gene, such as the gene for the global minima gene, can be used. Oligonucleotides comprising different amino acids at the different variant positions can be used during PCR using standard primers. This generally requires fewer oligonucleotides and can result in fewer errors.

Figure 53 depicts and overlapping extension method. At the top of Figure 53 is the template DNA showing the locations of the regions to be mutated (black boxes) and the binding sites of the relevant primers (arrows). The primers R1 and R2 represent a pool of primers, each containing a different mutation; as described herein, this may be done using different ratios of primers if desired. The variant position is flanked by regions of homology sufficient to get hybridization. In this example, three separate PCR reactions are done for step 1. The first reaction contains the template plus oligos F1 and R1. The second reaction contains template plus F2 and R2, and the third contains the template and F3 and R3. The reaction products are shown. In Step 2, the products from Step 1 tube 1 and Step 1 tube 2 are taken. After purification away from the primers, these are added to a fresh PCR reaction together with F1 and R4. During the Denaturation phase of the PCR, the overlapping regions anneal and the second strand is synthesized. The product is then amplified by the outside primers. In Step 3, the purified product from Step 2 is used in a third PCR reaction, together with the product of Step 1, tube 3 and the primers F1 and R3. The final product corresponds to the full length gene and contains the required mutations.

Figure 54 depicts a ligation of PCR reaction products to synthesize the libraries of the invention. In this technique, the primers also contain an endonuclease restriction site (RE), either blunt, 5' overhanging or 3' overhanging. We set up three separate PCR reactions for Step 1. The first reaction contains the template plus oligos F1 and R1. The second reaction contains template plus F2 and R2, and the third contains the template and F3 and R3. The reaction products are shown. In Step 2, the products of step 1 are purified and then digested with the appropriate restriction endonuclease. The digestion products from Step 2, tube 1 and Step 2, tube 2 and ligate them together with DNA ligase (step 3). The products are then amplified in Step 4 using primer F1 and R4. The whole process is then repeated by digesting the amplified products, ligating them to the digested products of Step 2, tube 3, and then amplifying the final product by primers F1 and R3. It would also be possible to ligate all three PCR products from Step 1 together in one reaction, providing the two restriction sites (RE1 and RE2) were different.

Figure 55 depicts blunt end ligation of PCR products. In this technique, the primers such as F1 and R1 do not overlap, but they abut. Again three separate PCR reactions are performed. The products from tube 1 and tube 2 are ligated, and then amplified with outside primers F1 and R4. This product is then ligated with the product from Step 1 , tube 3. The final products are then amplified with primers F1 and R3.

Figure 56 depicts M13 single stranded template production of mutated PCR products. Primerl and Primer2 (each representing a pool of primers corresponding to desired mutations) are mixed with the M13 template containing the wildtype gene or any starting gene. PCR produces the desired product (11) containing the combinations of the desired mutations incorporated in Primerl and Primer2. This scheme can be used to produce a gene with mutations, or fragments of a gene with mutations that are then linked together via ligation or PCR for example.

DETAILED DESCRIPTION OF THE INVENTION

Significant effort is being channeled into screening techniques that can identify proteins relevant in signaling pathways and disease states, and to compounds that can effect these pathways and disease states. Many of these techniques rely on the screening of large libraries, comprising either synthetic or naturally occurring proteins or peptides, in assays such as binding or functional assays. One of the problems facing high throughput screening technologies today is the difficulty of elucidating the identification of the "hit", i.e. a molecule causing the desired effect, against a background of many candidates that do not exhibit the desired properties.

The present invention is directed to a novel method that can allow the rapid and facile identification of these "hits". By combining a variety of different computational screening methods to "prescreen" large numbers of protein sequences for candidate sequences that are more likely to be useful (e.g. more stable), a smaller more practical library can then be experimentally generated. The system of the present invention relies on the use of nucleic acid modification enzymes that covalently and specifically bind to the sequence that encode them. Proteins of interest (e.g. the computationally prescreened set) are fused (either directly or indirectly, as outlined below) to a nucleic acid modification (NAM) enzyme. The NAM enzyme will covalently attach itself to a corresponding NAM attachment sequence (termed an enzyme attachment sequence (EAS)). Thus, by using vectors that comprising coding regions for the NAM enzyme and candidate proteins and the NAM enzyme attachment sequence, the candidate protein is covalently linked to the nucleic acid that encodes it upon translation, forming nucleic acid/protein (NAP) conjugates. Thus, after screening, candidates that exhibit the desired properties can be quickly pulled out using a variety of methods such as PCR amplification. This facilitates the quick identification of useful candidate proteins, and allows rapid screening and validation to occur.

Thus, the present invention is directed to methods of using computational screening of protein sequence libraries (that can comprise up to 10^8D or more members) to select smaller libraries of protein sequences (that can comprise up to 10¹³ members), that can then be used in a number of ways. For example, the proteins can be actually synthesized as NAP conjugates, and experimentally tested in the desired assay, for improved function and properties. Similarly, the library can be additionally computationally manipulated to create a new library which then itself can be experimentally tested in the NAP system.

The invention has two broad uses; first, the invention can be used to prescreen libraries based on known scaffold proteins. That is, computational screening for stability (or other properties) may be done on either the entire protein or some subset of residues, as desired and described below. By using computational methods to generate a threshold or cutoff to eliminate disfavored sequences, the percentage of useful variants in a given variant set size can increase, and the required experimental outlay is decreased.

In addition, the present invention finds use in the screening of random peptide libraries. As is known, signaling pathways in cells often begin with an effector stimulus that leads to a phenotypically describable change in cellular physiology. Despite the key role intracellular signaling pathways play in disease pathogenesis, in most cases, little is understood about a signaling pathway other than the initial stimulus and the ultimate cellular response.

Historically, signal transduction has been analyzed by biochemistry or genetics. The biochemical approach dissects a pathway in a "stepping-stone" fashion: find a molecule that acts at, or is involved in, one end of the pathway, isolate assayable quantities and then try to determine the next molecule in the pathway, either upstream or downstream of the isolated one. The genetic approach is classically a "shot in the dark": induce or derive mutants in a signaling pathway and map the locus by genetic crosses or complement the mutation with a cDNA library. Limitations of biochemical approaches include a reliance on a significant amount of pre-existing knowledge about the constituents under study and the need to carry such studies out in vitro, post-mortem. Limitations of purely genetic approaches include the need to first derive and then characterize the pathway before proceeding with identifying and cloning the gene.

Screening molecular libraries of chemical compounds for drugs that regulate signal systems has led to important discoveries of great clinical significance. Cyclosporin A (CsA) and FK506, for examples, were selected in standard pharmaceutical screens for inhibition of T-cell activation. It is noteworthy that while these two drugs bind completely different cellular proteins - cyclophilin and FK506 binding protein (FKBP), respectively, the effect of either drug is virtually the same - profound and specific suppression of T-cell activation, phenotypically observable in T cells as inhibition of mRNA production dependent on transcription factors such as NF-AT and NF-κB. Libraries of small peptides have also been successfully screened in vitro in assays for bioactivity. The literature is replete with examples of small peptides capable of modulating a wide variety of signaling pathways. For example, a peptide derived from the HIV-1 envelope protein has been shown to block the action of cellular calmodulin.

Accordingly, generation of random or semi-random sequence libraries of proteins and peptides allows for the selection of proteins (including peptides, oligopeptides and polypeptides) with useful properties. The sequences in these experimental libraries can be randomized at specific sites only, or throughout the sequence. The number of sequences that can be searched in these libraries grows expontentially with the number of positions that are randomized. Generally, only up to 10¹² - 10¹⁵ sequences can be contained in a library because of the physical constraints of laboratories (the size of the instruments, the cost of producing large numbers of biopolymers, etc.). Other practical considerations can often limit the size of the libraries to 10⁶ or fewer. These limits are reached for only 10 amino acid positions. Therefore, only a sparse sampling of sequences is possible in the search for improved proteins or peptides in experimental sequence libraries, lowering the chance of success and almost certainly missing desirable candidates. Because of the randomness of the changes in these sequences, most of the candidates in the library are not suitable, resulting in a waste of most of the effort in producing the library.

However, using the automated protein design techniques outlined below, virtual libraries of protein sequences can be generated that are vastly larger than experimental libraries. Up to 10⁸⁰ candidate sequences can be screened computationally and those that meet design criteria which favor stable and functional proteins can be readily selected. An experimental library consisting of the favorable candidates found in the virtual library screening can then be generated and placed into the NAP conjugate system, resulting in a much more efficient use of the experimental library and overcoming the limitations of random protein libraries.

Two principle benefits come from the virtual library screening: (1) the automated protein design generates a list of sequence candidates that are favored to meet design criteria; it also shows which positions in the sequence are readily changed and which positions are unlikely to change without disrupting protein stability and function. An experimental random library can be generated that is only randomized at the readily changeable, non-disruptive sequence positions. (2) The diversity of amino acids at these positions can be limited to those that the automated design shows are compatible with these positions. Thus, by limiting the number of randomized positions and the number of possibilities at these positions, the number of wasted sequences produced in the experimental library is reduced, thereby increasing the probability of success in finding sequences with useful properties. In addition, by computationally screening very large libraries of mutants, greater diversity of protein sequences can be screened (i.e. a larger sampling of sequence space), leading to greater improvements in protein function. Further, fewer mutants need to be tested experimentally to screen a given library size, reducing the cost and difficulty of protein engineering. By using computational methods to pre-screen a protein library, the computational features of speed and efficiency are combined with the ability of experimental library screening to create new activities in proteins for which appropriate computational models and structure-function relationships are unclear.

Similarly, novel methods to create secondary libraries derived from very large computational mutant libraries allow the rapid testing of large numbers of computationally designed sequences. By placing the computationally derived libraries into the NAP conjugate system, those candidate proteins that show function in a particular assay can be rapidly pulled out and evaluated.

In addition, as is more fully outlined below, the libraries may be biased in any number of ways, allowing the generation of secondary libraries that vary in their focus; for example, domains, subsets of residues, active or binding sites, surface residues, etc., may all be varied or kept constant as desired.

In general, as more fully outlined below, the invention can take on a wide variety of configurations. In general, primary libraries, e.g. libraries of all or a subset of possible proteins are generated computationally. This can be done in a wide variety of ways, including sequence alignments of related proteins, structural alignments, structural prediction models, databases, or (preferably) protein design automation computational analysis. Similarly, primary libraries can be generated via sequence screening using a set of scaffold structures that are created by perturbing the starting structure (using any number of techniques such as molecular dynamics, Monte Carlo analysis) to make changes to the protein (including backbone and sidechain torsion angle changes). Optimal sequences can be selected for each starting structures (or, some set of the top sequences) to make primary libraries.

Some of these techniques result in the list of sequences in the primary library being"scored", or "ranked" on the basis of some particular criteria. In some embodiments, lists of sequences that are generated without ranking can then be ranked using techniques as outlined below.

In a preferred embodiment, some subset of the primary library is then experimentally generated to form a secondary library. Alternatively, some or all of the primary library members are recombined to form a secondary library, e.g. with new members. Again, this may be done either computationally or experimentally or both.

Alternatively, once the primary library is generated, it can be manipulated in a variety of ways. In one embodiment, a different type of computational analysis can be done; for example, a new type of ranking may be done. Alternatively, and the primary library can be recombined, e.g. residues at different positions mixed to form a new, secondary library. Again, this can be done either computationally or experimentally, or both.

Accordingly, there are two components to the present invention: the computational prescreening step to form primary and secondary libraries, and the second component of actually synthesizing the secondary (or in some cases, tertiary) libraries as part of a NAP system.

Computational Screening

Accordingly, the present invention provides methods for generating secondary libraries of scaffold protein variants. By "protein" herein is meant at least two amino acids linked together by a peptide bond. As used herein, protein includes proteins, oligopeptides and peptides. The peptidyl group may comprise naturally occurring amino acids and peptide bonds, or synthetic peptidomimetic structures, i.e. "analogs", such as peptoids (see Simon et al., PNAS USA 89(20):9367 (1992)).. The amino acids may either be naturally occurrng or non-naturally occuring; as will be appreciated by those in the art, any structure for which a set of rotamers is known or can be generated can be used as an amino acid. The side chains may be in either the (R) or the (S) configuration. In a preferred embodiment, the amino acids are in the (S) or L-configuration.

The scaffold protein may be any protein for which a three dimensional structure is known or can be generated; that is, for which there are three dimensional coordinates for each atom of the protein. Generally this can be determined using X-ray crystallographic techniques, NMR techniques, de novo modelling, homology modelling, etc. In general, if X-ray structures are used, structures at 2A resolution or better are preferred, but not required.

The scaffold proteins may be from any organism, including prokaryotes and eukaryotes, with enzymes from bacteria, fungi, extremeophiles such as the archebacteria, insects, fish, animals (particularly

IS mammals and particularly human) and birds all possible. In addition, scaffold proteins may be from any cell types listed below as useful for test molecules.

Thus, by "scaffold protein" herein is meant a protein for which a secondary library of variants is desired. As will be appreciated by those in the art, any number of scaffold proteins find use in the present invention. Specifically included within the definition of "protein" are fragments and domains of known proteins, including functional domains such as enzymatic domains, binding domains, etc., and smaller fragments, such as turns, loops, etc. That is, portions of proteins may be used as well. In addition, "protein" as used herein includes proteins, oligopeptides and peptides. In addition, protein variants, i.e. non-naturally occuring protein analog structures, may be used.

) Suitable proteins include, but are not limited to, industrial and pharmaceutical proteins, including ligands, cell surface receptors, antigens, antibodies, cytokines, hormones, transcription factors, signaling modules, cytoskeletal proteins and enzymes. Suitable classes of enzymes include, but are not limited to, hydrolases such as proteases, carbohydrases, lipases; isomerases such as racemases, epimerases, tautomerases, or mutases; transferases, kinases, oxidoreductases, and phophatases. i Suitable enzymes are listed in the Swiss-Prot enzyme database. Suitable protein backbones include, but are not limited to, all of those found in the protein data base compiled and serviced by the Research Collaboratory for Structural Bioinformatics (RCSB, formerly the Brookhaven National Lab).

Specifically, preferred scaffold proteins include, but are not limited to, those with known structures (including variants) including cytokines (IL-1 ra (+receptor complex), IL-1 (receptor alone), IL-1a, IL-1b (including variants and or receptor complex), IL-2, IL-3, IL-4, IL-5, IL-6, IL-8, IL-10, IFN-β, INF-y, IFN- α-2a; IFN-α-2B, TNF-α; CD40 ligand (chk), Human Obesity Protein Leptin, Granulocyte Colony- Stimulating Factor, Bone Morphogenetic Protein-7, Ciliary Neurotrophic Factor, Granulocyte- Macrophage Colony-Stimulating Factor, Monocyte Chemoattractant Protein 1 , Macrophage Migration Inhibitory Factor, Human Glycosylation-lnhibiting Factor, Human Rantes, Human Macrophage Inflammatory Protein 1 Beta, human growth hormone, Leukemia Inhibitory Factor, Human Melanoma Growth Stimulatory Activity, neutrophil activating peptide-2, Cc-Chemokine Mcp-3, Platelet Factor M2, Neutrophil Activating Peptide 2, Eotaxin, Stromal Cell-Derived Factor-1 , Insulin, Insulin-like Growth Factor I, Insulin-like Growth Factor II, Transforming Growth Factor B1, Transforming Growth Factor B2, Transforming Growth Factor B3, Transforming Growth Factor A, Vascular Endothelial growth factor (VEGF), acidic Fibroblast growth factor, basic Fibroblast growth factor, Endothelial growth factor, Nerve growth factor, Brain Derived Neurotrophic Factor, Ciliary Neurotrophic Factor, Platelet Derived Growth Factor, Human Hepatocyte Growth Factor, Glial Cell-Derived Neurotrophic Factor, (as well as the 55 cytokines in PDB 1/12/99)); Erythropoietin; other extracellular signalling moeities, including, but not limited to, hedgehog Sonic, hedgehog Desert, hedgehog Indian, hCG; coaguation factors including, but not limited to, TPA and Factor Vila; transcription factors, including but not limited to, p53, p53 tetramerization domain, Zn fingers (of which more than 12 have structures), homeodomains (of which 8 have structures), leucine zippers (of which 4 have structures); antibodies, including, but not limited to, cFv; viral proteins, including, but not limited to, hemagglutinin trimerization domain and hiv Gp41 ectodomain (fusion domain); intracellular signalling modules, including, but not limited to, SH2 domains (of which 8 structures are known), SH3 domains (of which 11 have structures), and Pleckstin Homology Domains; receptors, including, but not limited to, the extracellular Region Of Human Tissue Factor Cytokine-Binding Region Of Gp130, G-CSF receptor, erythropoietin receptor, Fibroblast Growth Factor receptor, TNF receptor, IL-1 receptor, IL-1 receptor/I Lira complex, IL-4 receptor, INF-γ receptor alpha chain, MHC Class I, MHC Class II , T Cell Receptor, Insulin receptor, insulin receptor tyrosine kinase and human growth hormone receptor.

In a preferred embodiment, the library comprises random peptides. In one embodiment, the library is fully randomized, with no sequence preferences or constants at any position. In a preferred embodiment, the library is biased. That is, some positions within the sequence are either held constant, or are selected from a limited number of possibilities. For example, in a preferred embodiment, the nucleotides or amino acid residues are randomized within a defined class, for example, of hydrophobic amino acids, hydrophilic residues, sterically biased (either small or large) residues, towards the creation of cysteines, for cross-linking, pralines for SH-3 domains, PDZ domains, serines, threonines, tyrosines or histidines for phosphorylation sites, etc., or to purines, etc.

In a preferred embodiment, the bias is towards peptides or nucleic acids that interact with known classes of molecules. For example, when the candidate protein is a peptide, it is known that much of intracellular signaling is carried out via short regions of polypeptides interacting with other polypeptides through small peptide domains. For instance, a short region from the HIV-1 envelope cytoplasmic domain has been previously shown to block the action of cellular calmodulin. Regions of the Fas cytoplasmic domain, which shows homology to the mastoparan toxin from Wasps, can be limited to a short peptide region with death-inducing apoptotic or G protein inducing functions. Magainin, a natural peptide derived from Xenopus, can have potent anti-tumour and anti-microbial activity. Short peptide fragments of a protein kinase C isozyme (BPKC), have been shown to block nuclear translocation of βPKC in Xenopus oocytes following stimulation. And, short SH-3 target peptides have been used as psuedosubstrates for specific binding to SH-3 proteins. This is of course a short list of available peptides with biological activity, as the literature is dense in this area. Thus, there is much precedent for the potential of small peptides to have activity on intracellular signaling cascades. In addition, agonists and antagonists of any number of molecules may be used as the basis of biased randomization of candidate proteins as well.

Thus, a number of molecules or protein domains are suitable as starting points for the generation of biased randomized candidate proteins. A large number of small molecule domains are known, that confer a common function, structure or affinity. In addition, as is appreciated in the art, areas of weak amino acid homology may have strong structural homology. A number of these molecules, domains, and/or corresponding consensus sequences, are known, including, but are not limited to, SH-2 domains, SH-3 domains, Pleckstrin, death domains, protease cleavage/recognition sites, enzyme inhibitors, enzyme substrates, Traf, etc. Similarly, there are a number of known nucleic acid binding proteins containing domains suitable for use in the invention. For example, leucine zipper consensus sequences are known.

In a preferred embodiment, biased SH-3 domain-binding oligonucleotides/peptides are made. SH-3 domains have been shown to recognize short target motifs (SH-3 domain-binding peptides), about ten to twelve residues in a linear sequence, that can be encoded as short peptides with high affinity for the target SH-3 domain. Consensus sequences for SH-3 domain binding proteins have been proposed. Thus, in a preferred embodiment, oligos/peptides are made with the following biases

1. XXXPPXPXX, wherein X is a randomized residue.

2. (within the positions of residue positions 11 to -2):

11 10 9 8 7 6 5 4 3 2 1 Met Glyaal 1aa10 aa9 aa8 aa7 Arg Pro Leu Pro Pro hyd 0 -1 -2

Pro hyd hyd Gly Gly Pro Pro STOP atg ggc nnk nnk nnk nnk nnk aga cct ctg cct cca sbk ggg sbk sbk gga ggc cca cct TAA1.

In this embodiment, the N-terminus flanking region is suggested to have the greatest effects on binding affinity and is therefore entirely randomized. "Hyd" indicates a bias toward a hydrophobic residue, i.e.- Val, Ala, Gly, Leu, Pro, Arg. To encode a hydrophobically biased residue, "sbk" codon biased structure is used. Examination of the codons within the genetic code will ensure this encodes generally hydrophobic residues. s= g,c; b= t, g, c; v= a, g, c; m= a, c; k= t, g; n= a, t, g, c.

Thus, in a preferred embodiment, the candidate protein is a structural tag that will allow the isolation of target proteins with that structure. That is, in the case of leucine zippers, the fusion of the NAM enzyme to a leucine zipper sequence will allow the fusions to "zip up" with other leucine zippers, allow the quick isolation of a plurality of leucine zipper proteins. In addition, structural tags (which may only be the proteins themselves) can allow heteromultimeric protein complexes to form, that then are assayed for activity as complexes. That is, many proteins, such as many eucaryotic transcription factors, function as heteromultimeric complexes which can be assayed using the present invention.

In addition, the candidate protein library may be a constructed library; that is, it may be built to contain only members of a defined class, or combinations of classes. For example, libraries of immunoglobulins may be built, or libraries of G-protein coupled receptors, tumor suppressor genes, proteases, transcription factors, phosphotases, kinases, etc.

Once a scaffold protein is chosen, a primary library is generated using computational processing. Generally speaking, in some embodiments, the goal of the computational processing is to determine a set of optimized protein sequences. By "optimized protein sequence" herein is meant a sequence that best fits the mathematical equations of the computational process. As will be appreciated by those in the art, a global optimized sequence is the one sequence that best fits the equations (for example, when PDA is used, the global optimzed sequence is the sequence that best fits Equation 1, below); i.e. the sequence that has the lowest energy of any possible sequence. However, there are any number of sequences that are not the global minimum but that have low energies.

Thus, a "primary library" as used herein is a collection of optimized sequences, generally, but not always, in the form of a rank-ordered list. In theory, all possible sequences of a protein may be ranked; however, currently 10¹³ sequences is a practical limit. Thus, in general, some subset of all possible sequences is used as the primary library; generally, the top 10³ to 10¹³ sequences are chosen as the primary library. The cutoff for inclusion in the rank ordered list of the primary library can be done in a variety of ways. For example, the cutoff may be just an arbitrary exclusion point: the top 10⁵ sequences may comprise the primary library. Alternatively, all sequences scoring within a certain limit of the global optimum can be used; for example, all sequences with 10 kcal/mol of the global optimum could be used as the primary library. This method has the advantage of using a direct measure of fidelity to a three dimensional structure to determine inclusion. This approach can be used to insure that library mutations are not limited to positions that have the lowest energy gap between different mutations. Alternatively, the cutoff may be enforced when a predetermined number of mutations per position is reached. As a rank ordered sequence list is lengthened and the library is enlarged, more mutations per position are defined. Alternatively, the total number of sequences defined by the recombination of all mutations can be used as a cutoff criterion for the primary sequence library. Preferred values for the total number of sequences range from 100 to 10²⁰, particularly preferred values range from 1000 to 10¹³, especially preferred values range from 1000 to 10⁷ Alternatively, the first occurrence in the list of predefined undesirable residues can be used as a cutoff criterion. For example, the first hydrophilic residue occurring in a core position would limit the list. It should also be noted that while these methods are described in conjunction with limiting the size of the primary library, these same techniques may be used to formulate the cutoff for inclusion in the secondary library as well.

Thus, the present invention provides methods to generate a primary library optionally comprising a rank ordered list of sequences, generally in terms of theoretical quantitative stability, as is more fully described below. Generating a primary library to optimize the stability of a conformation can be used to stabilize the active site transition state conformation of an enzyme, which will improve its activity. Similarly, stabilizing a ligand-receptor complex or enzyme-substrate complex will improve the binding affinity.

The primary libraries can be generated in a variety of ways. In essence, any methods that can result in either the relative ranking of the possible sequences of a protein based on measurable stability parameters, or a list of suitable sequences can be used. As will be appreciated by those in the art, any of the methods described herein or known in the art may be used alone, or in combination with other methods.

Generally, there are a variety of computational methods that can be used to generate a primary library. In a preferred embodiment, sequence based methods are used. Alternatively, structure based methods, such as PDA, described in detail below, are used.

In a preferred embodiment, the scaffold protein is an enzyme and highly accurate electrostatic models can be used for enzyme active site residue scoring to improve enzyme active site libraries (see Warshel, computer Modeling of Chemical Reactions in Enzymes and Solutions. Wiley & Sons, New York, (1991), hereby expressly incorporated by reference) These accurate models can assess the relative energies of sequences with high precision, but are computationally intensive.

Similarly, molecular dynamics calculations can be used to computationally screen sequences by individually calculating mutant sequence scores and compiling a rank ordered list.

In a preferred embodiment, residue pair potentials can be used to score sequences (Miyazawa et al. Macromolecules 18(3):534-552 (1985), expressly incorporated by reference) during computational screening.

In a preferred embodiment, sequence profile scores (Bowie et al., Science 253(5016):164-70 (1991), incorporated by reference) and/or potentials of mean force (Hendlich et al., J. Mol. Biol. 216(1):167- 180 (1990), also incorporated by reference) can also be calculated to score sequences. These methods assess the match between a sequence and a 3D protein structure and hence can act to screen for fidelity to the protein structure. By using different scoring functions to rank sequences, different regions of sequence space can be sampled in the computational screen.

Furthermore, scoring functions can be used to screen for sequences that would create metal or co- factor binding sites in the protein (Hellinga, Fold Des. 3(1):R1-8 (1998), hereby expressly incorporated by reference). Similarly, scoring functions can be used to screen for sequences that would create disulfide bonds in the protein. These potentials attempt to specifically modify a protein structure to introduce a new structural motif.

In a preferred embodiment, sequence and/or structural alignment programs can be used to generate primary libraries. As is known in the art, there are a number of sequence-based alignment programs; including for example, Smith-Waterman searches, Needleman-Wunsch, Double Affine Smith- Waterman, frame search, Gribskov/GCG profile search, Gribskov/GCG profile scan, profile frame search, Bucher generalized profiles, Hidden Markov models, Hframe, Double Frame, Blast, Psi-Blast, Clustal, and GeneWise.

The source of the sequences can vary widely, and include taking sequences from one or more of the known databases, including, but not limited to, SCOP (Hubbard, et al., Nucleic Acids Res 27(1):254- 256. (1999)); PFAM (Bateman, et al., Nucleic Acids Res 27(1):260-262. (1999)); VAST (Gibrat, et al., Curr Opin Struct Biol 6(3):377-385. (1996)); CATH (Orengo, et al., Structure 5(8):1093-1108. (1997)); PhD Predictor (http://www.embl-heidelberq.de/oredictprotein /predictprotein.html); Prosite (Hofmann, et al., Nucleic Acids Res 27(1 ):215-219. (1999)); PIR

(http://www.mips.biochem.mpq. de/proj/protseqdbΛ; GenBank (http://www.ncbi.nlm.nih.gov/); PDB (www.rcsb.org) and BIND (Bader, et al., Nucleic Acids Res 29(1 ):242-245. (2001)).

In addition, sequences from these databases can be subjected to continguous analysis or gene prediction; see Wheeler, et al., Nucleic Acids Res 28(1):10-14. (2000) and Burge and Kariin, J Mol Biol 268(1 ):78-94. (1997).

As is known in the art, there are a number of sequence alignment methodologies that can be used. For example, sequence homology based alignment methods can be used to create sequence alignments of proteins related to the target structure (Altschul et al., J. Mol. Biol. 215(3):403 (1990), incorporated by reference). These sequence alignments are then examined to determine the observed sequence variations. These sequence variations are tabulated to define a primary library. In addition, as is further outlined below, these methods can also be used to generate secondary libraries.

Sequence based alignments can be used in a variety of ways. For example, a number of related proteins can be aligned, as is known in the art, and the "variable" and "conserved" residues defined; that is, the residues that vary or remain identical between the family members can be defined. These results can be used to generate a probability table, as outlined below. Similarly, these sequence variations can be tabulated and a secondary library defined from them as defined below. Alternatively, the allowed sequence variations can be used to define the amino acids considered at each position during the computational screening. Another variation is to bias the score for amino acids that occur in the sequence alignment, thereby increasing the likelihood that they are found during computational screening but still allowing consideration of other amino acids. This bias would result in a focused primary library but would not eliminate from consideration amino acids not found in the alignment. In addition, a number of other types of bias may be introduced. For example, diversity may be forced; that is, a "conserved" residue is chosen and altered to force diversity on the protein and thus sample a greater portion of the sequence space. Alternatively, the positions of high variability between family members (i.e. low conservation) can be randomized, either using all or a subset of amino acids. Similarly, outlier residues, either positional outliers or side chain outliers, may be eliminated. Similarly, structural alignment of structurally related proteins can be done to generate sequence alignments. There are a wide variety of such structural alignment programs known. See for example VAST from the NCBI (http://www.ncbi.nlm.nih.qov:80/Structure/VAST/vast.shtml); SSAP (Orengo and Taylor, Methods Enzymol 266(617-635 (1996)) SARF2 (Alexandrov, Protein Eng 9(9):727-732. (1996)) CE (Shindyalov and Bourne, Protein Eng 11 (9):739-747. (1998)); (Orengo et al., Structure 5(8):1093-108 (1997); Dali (Holm et al., Nucleic Acid Res. 26(1 ):316-9 (1998), all of which are incorporated by reference). These structurally-generated sequence alignments can then be examined to determine the observed sequence variations.

Primary libraries can be generated by predicting secondary structure from sequence, and then selecting sequences that are compatible with the predicted secondary structure. There are a number of secondary structure prediction methods, including, but not limited to, threading (Bryant and Altschul, Curr Opin Struct Biol 5(2):236-244. (1995)), Profile 3D (Bowie, et al., Methods Enzymol 266(598-616 (1996); MONSSTER (Skolnick, et al., J Mol Biol 265(2):217-241. (1997); Rosetta (Simons, et al., Proteins 37(S3):171-176 (1999); PSI-BLAST (Altschul and Koonin, Trends Biochem Sci 23(11):444- 447. (1998)); Impala (Schaffer, et al., Bioinformatics 15(12): 1000-1011. (1999)); HMMER (McClure, et al., Proc Int Conf Intell Syst Mol Biol 4(155-164 (1996)); Clustal W (http://www.ebi.ac.uk/clustalw/): BLAST (Altschul, et al., J Mol Biol 215(3):403-410. (1990)), helix-coil transition theory (Munoz and Serrano, Biopolymers 41 :495, 1997), neural networks, local structure alignment and others (e.g., see in Selbig et al., Bioinformatics 15:1039, 1999).

Similarly, as outlined above, other computational methods are known, including, but not limited to, sequence profiling (Bowie and Eisenberg, Science 253(5016): 164-70, (1991)), rotamer library selections (Dahiyat and Mayo, Protein Sci 5(5): 895-903 (1996); Dahiyat and Mayo, Science 278(5335): 82-7 (1997); Desjarlais and Handel, Protein Science 4: 2006-2018 (1995); Harbury et al, PNAS USA 92(18): 8408-8412 (1995); Kono et al., Proteins: Structure, Function and Genetics 19: 244-255 (1994); Hellinga and Richards, PNAS USA 91 : 5803-5807 (1994)); and residue pair potentials (Jones, Protein Science 3: 567-574, (1994); PROSA (Heindlich et al., J. Mol. Biol. 216:167-180 (1990); THREADER (Jones et al., Nature 358:86-89 (1992), and other inverse folding methods such as those described by Simons et al. (Proteins, 34:535-543, 1999), Levitt and Gerstein (PNAS USA, 95:5913- 5920, 1998), Godzik et al., PNAS, V89, PP 12098-102; Godzik and Skolnick (PNAS USA, 89:12098- 102, 1992), Godzik et al. (J. Mol. Biol. 227:227-38, 1992) and two profile methods (Gribskov et al. PNAS 84:4355-4358 (1987) and Fischer and Eisenberg, Protein Sci. 5:947-955 (1996), Rice and Eisenberg J. Mol. Biol. 267:1026-1038(1997)), all of which are expressly incorporated by reference. In addition, other computational methods such as those described by Koehl and Levitt (J. Mol. Biol. 293:1161-1181 (1999); J. Mol. Biol. 293:1183-1193 (1999); expressly incorporated by reference) can be used to create a protein sequence library which can optionally then be used to generate a smaller secondary library for use in experimental screening for improved properties and function.

In addition, there are computational methods based on forcefield calculations such as SCMF that can be used as well for SCMF, see Delarue et la. Pac. Symp. Biocomput. 109-21 (1997), Koehl et al., J. Mol. Biol. 239:249 (1994); Koehl et al., Nat. Struc. Biol. 2:163 (1995); Koehl et al., Curr. Opin. Struct. Biol. 6:222 (1996); Koehl et al., J. Mol. Bio. 293:1183 (1999); Koehl et al, J. Mol. Biol. 293:1161 (1999); Lee J. Mol. Biol. 236:918 (1994); and Vasquez Biopolymers 36:53-70 (1995); all of which are expressly incorporated by reference. Other forcefield calculations that can be used to optimize the conformation of a sequence within a computational method, or to generate de novo optimized sequences as outlined herein include, but are not limited to, OPLS-AA (Jorgensen, et al, J. Am. Chem. Soc. (1996), v 118, pp 11225-11236; Jorgensen, W.L; BOSS, Version 4.1 ; Yale University: New Haven, CT (1999)); OPLS (Jorgensen, et al, J. Am. Chem. Soc. (1988), v 110, pp 1657ff; Jorgensen, et al, J Am. Chem. Soc. (1990), v 112, pp 4768ff); UNRES (United Residue Forcefield; Liwo, et al. Protein Science (1993), v 2, pp1697-1714; Liwo, et al. Protein Science (1993), v 2, pp1715-1731 ; Liwo, et al, J. Comp. Chem. (1997), v 18, pp849-873; Liwo, et al, J. Comp. Chem. (1997), v 18, pp874-884; Liwo, et al, J. Comp. Chem. (1998), v 19, pp259-276; Forcefield for Protein Structure Prediction (Liwo, et al, Proc. Natl. Acad. Sci. USA (1999), v 96, pp5482-5485); ECEPP/3 (Liwo et al, J Protein Chem 1994 May;13(4):375-80); AMBER 1.1 force field (Weiner, et al, J. Am. Chem. Soc. v106, pp765-784); AMBER 3.0 force field (U.C. Singh et al, Proc. Natl. Acad. Sci. USA. 82:755-759); CHARMM and CHARMM22 (Brooks, et al, J. Comp. Chem. v4, pp 187-217); cvff3.0 (Dauber-Osguthorpe, et al,(1988) Proteins: Structure, Function and Genetics, v4,pp31-47); cff91 (Maple, et al, J. Comp. Chem. v15, 162-182); also, the DISCOVER (cvff and cff91) and AMBER forcefields are used in the INSIGHT molecular modeling package (Biosym/MSI, San Diego California) and HARMM is used in the QUANTA molecular modeling package (Biosym/MSI, San Diego California), all of which are expressly incorporated by reference. In fact, as is outlined below, these forcefield methods may be used to generate the secondary library directly; that is, no primary library is generated; rather, these methods can be used to generate a probability table from which the secondary library is directly generated, for example by using these forcefields during an SCMF calculation.

In a preferred embodiment, the computational method used to generate the primary library is Protein Design Automation (PDA), as is described in U.S.S.N.s 60/061 ,097, 60/043,464, 60/054,678, 09/127,926 and PCT US98/07254, all of which are expressly incorporated herein by reference. Briefly, PDA can be described as follows. A known protein structure is used as the starting point. The residues to be optimized are then identified, which may be the entire sequence or subset(s) thereof. The side chains of any positions to be varied are then removed. The resulting structure consisting of the protein backbone and the remaining sidechains is called the template. Each variable residue position is then preferably classified as a core residue, a surface residue, or a boundary residue; each classification defines a subset of possible amino acid residues for the position (for example, core residues generally will be selected from the set of hydrophobic residues, surface residues generally will be selected from the hydrophilic residues, and boundary residues may be either). Each amino acid can be represented by a discrete set of all allowed conformers of each side chain, called rotamers. Thus, to arrive at an optimal sequence for a backbone, all possible sequences of rotamers must be screened, where each backbone position can be occupied either by each amino acid in all its possible rotameric states, or a subset of amino acids, and thus a subset of rotamers.

Two sets of interactions are then calculated for each rotamer at every position: the interaction of the rotamer side chain with all or part of the backbone (the "singles" energy, also called the rotamer/template or rotamer/backbone energy), and the interaction of the rotamer side chain with all other possible rotamers at every other position or a subset of the other positions (the "doubles" energy, also called the rotamer/rotamer energy). The energy of each of these interactions is calculated through the use of a variety of scoring functions, which include the energy of van der Waal's forces, the energy of hydrogen bonding, the energy of secondary structure propensity, the energy of surface area solvation and the electrostatics. Thus, the total energy of each rotamer interaction, both with the backbone and other rotamers, is calculated, and stored in a matrix form.

The discrete nature of rotamer sets allows a simple calculation of the number of rotamer sequences to be tested. A backbone of length n with m possible rotamers per position will have mⁿ possible rotamer sequences, a number which grows exponentially with sequence length and renders the calculations either unwieldy or impossible in real time. Accordingly, to solve this combinatorial search problem, a "Dead End Elimination" (DEE) calculation is performed. The DEE calculation is based on the fact that if the worst total interaction of a first rotamer is still better than the best total interaction of a second rotamer, then the second rotamer cannot be part of the global optimum solution. Since the energies of all rotamers have already been calculated, the DEE approach only requires sums over the sequence length to test and eliminate rotamers, which speeds up the calculations considerably. DEE can be rerun comparing pairs of rotamers, or combinations of rotamers, which will eventually result in the determination of a single sequence which represents the global optimum energy.

Once the global solution has been found, a Monte Carlo search may be done to generate a rank- ordered list of sequences in the neighborhood of the DEE solution. Starting at the DEE solution, random positions are changed to other rotamers, and the new sequence energy is calculated. If the new sequence meets the criteria for acceptance, it is used as a starting point for another jump. After a predetermined number of jumps, a rank-ordered list of sequences is generated. Monte Carlo searching is a sampling technique to explore sequence space around the global minimum or to find new local minima distant in sequence space. As is more additionally outlined below, there are other sampling techniques that can be used, including Boltzman sampling, genetic algorithm techniques and simulated annealing. In addition, for all the sampling techniques, the kinds of jumps allowed can be altered (e.g. random jumps to random residues, biased jumps (to or away from wild-type, for example), jumps to biased residues (to or away from similar residues, for example), etc.). Similarly, for all the sampling techniques, the acceptance criteria of whether a sampling jump is accepted can be altered.

As outlined in U.S.S.N. 09/127,926, the protein backbone (comprising (for a naturally occuring protein) the nitrogen, the carbonyl carbon, the α-carbon, and the carbonyl oxygen, along with the direction of the vector from the α-carbon to the β-carbon) may be altered prior to the computational analysis, by varying a set of parameters called supersecondary structure parameters.

Once a protein structure backbone is generated (with alterations, as outlined above) and input into the computer, explicit hydrogens are added if not included within the structure (for example, if the structure was generated by X-ray crystallography, hydrogens must be added). After hydrogen addition, energy minimization of the structure is run, to relax the hydrogens as well as the other atoms, bond angles and bond lengths. In a preferred embodiment, this is done by doing a number of steps of conjugate gradient minimization (Mayo ef al., J. Phys. Chem. 94:8897 (1990)) of atomic coordinate positions to minimize the Dreiding force field with no electrostatics. Generally from about 10 to about 250 steps is preferred, with about 50 being most preferred.

The protein backbone structure contains at least one variable residue position. As is known in the art, the residues, or amino acids, of proteins are generally sequentially numbered starting with the N- terminus of the protein. Thus a protein having a methionine at it's N-terminus is said to have a methionine at residue or amino acid position 1 , with the next residues as 2, 3, 4, etc. At each position, the wild type (i.e. naturally occuring) protein may have one of at least 20 amino acids, in any number of rotamers. By "variable residue position" herein is meant an amino acid position of the protein to be designed that is not fixed in the design method as a specific residue or rotamer, generally the wild-type residue or rotamer. In a preferred embodiment, all of the residue positions of the protein are variable. That is, every amino acid side chain may be altered in the methods of the present invention. This is particularly desirable for smaller proteins, although the present methods allow the design of larger proteins as well. While there is no theoretical limit to the length of the protein which may be designed this way, there is a practical computational limit.

In an alternate preferred embodiment, only some of the residue positions of the protein are variable, and the remainder are "fixed", that is, they are identified in the three dimensional structure as being in a set conformation. In some embodiments, a fixed position is left in its original conformation (which may or may not correlate to a specific rotamer of the rotamer library being used). Alternatively, residues may be fixed as a non-wild type residue; for example, when known site-directed mutagenesis techniques have shown that a particular residue is desirable (for example, to eliminate a proteolytic site or alter the substrate specificity of an enzyme), the residue may be fixed as a particular amino acid. Alternatively, the methods of the present invention may be used to evaluate mutations de novo, as is discussed below. In an alternate preferred embodiment, a fixed position may be "floated"; the amino acid at that position is fixed, but different rotamers of that amino acid are tested. In this embodiment, the variable residues may be at least one, or anywhere from 0.1 % to 99.9% of the total number of residues. Thus, for example, it may be possible to change only a few (or one) residues, or most of the residues, with all possibilities in between.

In a preferred embodiment, residues which can be fixed include, but are not limited to, structurally or biologically functional residues; alternatively, biologically functional residues may specifically not be fixed. For example, residues which are known to be important for biological activity, such as the residues which form the active site of an enzyme, the substrate binding site of an enzyme, the binding site for a binding partner (ligand/receptor, antigen/antibody, etc.), phosphorylation or glycosylation sites which are crucial to biological function, or structurally important residues, such as disulfide bridges, metal binding sites, critical hydrogen bonding residues, residues critical for backbone conformation such as proline or glycine, residues critical for packing interactions, etc. may all be fixed in a conformation or as a single rotamer, or "floated".

Similarly, residues which may be chosen as variable residues may be those that confer undesirable biological attributes, such as susceptibility to proteolytic degradation, dimerization or aggregation sites, glycosylation sites which may lead to immune responses, unwanted binding activity, unwanted allostery, undesirable enzyme activity but with a preservation of binding, etc. In a preferred embodiment, each variable position is classified as either a core, surface or boundary residue position, although in some cases, as explained below, the variable position may be set to glycine to minimize backbone strain. In addition, as outlined herein, residues need not be classified, they can be chosen as variable and any set of amino acids may be used. Any combination of core, surface and boundary positions can be utilized: core, surface and boundary residues; core and surface residues; core and boundary residues, and surface and boundary residues, as well as core residues alone, surface residues alone, or boundary residues alone.

The classification of residue positions as core, surface or boundary may be done in several ways, as will be appreciated by those in the art. In a preferred embodiment, the classification is done via a visual scan of the original protein backbone structure, including the side chains, and assigning a classification based on a subjective evaluation of one skilled in the art of protein modelling. Alternatively, a preferred embodiment utilizes an assessment of the orientation of the Cα-Cβ vectors relative to a solvent accessible surface computed using only the template Cα atoms, as outlined in U.S.S.N.s 60/061 ,097, 60/043,464, 60/054,678, 09/127,926 and PCT US98/07254. Alternatively, a surface area calculation can be done.

Once each variable position is classified as either core, surface or boundary, a set of amino acid side chains, and thus a set of rotamers, is assigned to each position. That is, the set of possible amino acid side chains that the program will allow to be considered at any particular position is chosen. Subsequently, once the possible amino acid side chains are chosen, the set of rotamers that will be evaluated at a particular position can be determined. Thus, a core residue will generally be selected from the group of hydrophobic residues consisting of alanine, valine, isoleucine, leucine, phenylalanine, tyrosine, tryptophan, and methionine (in some embodiments, when the α scaling factor of the van der Waals scoring function, described below, is low, methionine is removed from the set), and the rotamer set for each core position potentially includes rotamers for these eight amino acid side chains (all the rotamers if a backbone independent library is used, and subsets if a rotamer dependent backbone is used). Similarly, surface positions are generally selected from the group of hydrophilic residues consisting of alanine, serine, threonine, aspartic acid, asparagine, glutamine, glutamic acid, arginine, lysine and histidine. The rotamer set for each surface position thus includes rotamers for these ten residues. Finally, boundary positions are generally chosen from alanine, serine, threonine, aspartic acid, asparagine, glutamine, glutamic acid, arginine, lysine histidine, valine, isoleucine, leucine, phenylalanine, tyrosine, tryptophan, and methionine. The rotamer set for each boundary position thus potentially includes every rotamer for these seventeen residues (assuming cysteine, glycine and proline are not used, although they can be). Additionally, in some preferred embodiments, a set of 18 naturally occuring amino acids (all except cysteine and proline, which are known to be particularly disruptive) are used.

Thus, as will be appreciated by those in the art, there is a computational benefit to classifying the residue positions, as it decreases the number of calculations. It should also be noted that there may be situations where the sets of core, boundary and surface residues are altered from those described above; for example, under some circumstances, one or more amino acids is either added or subtracted from the set of allowed amino acids. For example, some proteins which dimerize or multimerize, or have ligand binding sites, may contain hydrophobic surface residues, etc. In addition, residues that do not allow helix "capping" or the favorable interaction with an α-helix dipole may be subtracted from a set of allowed residues. This modification of amino acid groups is done on a residue by residue basis.

In a preferred embodiment, proline, cysteine and glycine are not included in the list of possible amino acid side chains, and thus the rotamers for these side chains are not used. However, in a preferred embodiment, when the variable residue position has a φ angle (that is, the dihedral angle defined by 1) the carbonyl carbon of the preceding amino acid; 2) the nitrogen atom of the current residue; 3) the α- carbon of the current residue; and 4) the carbonyl carbon of the current residue) greater than 0°, the position is set to glycine to minimize backbone strain.

Once the group of potential rotamers is assigned for each variable residue position, processing proceeds as outlined in U.S.S.N. 09/127,926 and PCT US98/07254. This processing step entails analyzing interactions of the rotamers with each other and with the protein backbone to generate optimized protein sequences. Simplistically, the processing initially comprises the use of a number of scoring functions to calculate energies of interactions of the rotamers, either to the backbone itself or other rotamers. Preferred PDA scoring functions include, but are not limited to, a Van der Waals potential scoring function, a hydrogen bond potential scoring function, an atomic solvation scoring function, a secondary structure propensity scoring function and an electrostatic scoring function. As is further described below, at least one scoring function is used to score each position, although the scoring functions may differ depending on the position classification or other considerations, like favorable interaction with an α-helix dipole. As outlined below, the total energy which is used in the calculations is the sum of the energy of each scoring function used at a particular position, as is generally shown in Equation 1 : Equation 1 otai ^{= n}E_v_w ^{+ n}E_as + nE_h._bondirιg + nE_ss + nE_e|_eo

In Equation 1 , the total energy is the sum of the energy of the van der Waals potential (E_vdw), the energy of atomic solvation (E_as), the energy of hydrogen bonding (E_h._boπding), the energy of secondary structure (E_ss) and the energy of electrostatic interaction (E_ete0). The term n is either 0 or 1, depending on whether the term is to be considered for the particular residue position.

As outlined in U.S.S.N.s 60/061 ,097, 60/043,464, 60/054,678, 09/127,926 and PCT US98/07254, any combination of these scoring functions, either alone or in combination, may be used. Once the scoring functions to be used are identified for each variable position, the preferred first step in the computational analysis comprises the determination of the interaction of each possible rotamer with all or part of the remainder of the protein. That is, the energy of interaction, as measured by one or more of the scoring functions, of each possible rotamer at each variable residue position with either the backbone or other rotamers, is calculated. In a preferred embodiment, the interaction of each rotamer with the entire remainder of the protein, i.e. both the entire template and all other rotamers, is done. However, as outlined above, it is possible to only model a portion of a protein, for example a domain of a larger protein, and thus in some cases, not all of the protein need be considered. The term "portion", as used herein, with regard to a protein refers to a fragment of that protein. This fragment may range in size from 10 amino acid residues to the entire amino acid sequence minus one amino acid. Accordingly, the term "portion", as used herein, with regard to a nucleic refers to a fragment of that nucleic acid. This fragment may range in size from 10 nucleotides to the entire nucleic acid sequence minus one nucleotide.

In a preferred embodiment, the first step of the computational processing is done by calculating two sets of interactions for each rotamer at every position: the interaction of the rotamer side chain with the template or backbone (the "singles" energy), and the interaction of the rotamer side chain with all other possible rotamers at every other position (the "doubles" energy), whether that position is varied or floated. It should be understood that the backbone in this case includes both the atoms of the protein structure backbone, as well as the atoms of any fixed residues, wherein the fixed residues are defined as a particular conformation of an amino acid.

Thus, "singles" (rotamer/template) energies are calculated for the interaction of every possible rotamer at every variable residue position with the backbone, using some or all of the scoring functions. Thus, for the hydrogen bonding scoring function, every hydrogen bonding atom of the rotamer and every hydrogen bonding atom of the backbone is evaluated, and the E_HB is calculated for each possible rotamer at every variable position. Similarly, for the van der Waals scoring function, every atom of the rotamer is compared to every atom of the template (generally excluding the backbone atoms of its own residue), and the E_vdW is calculated for each possible rotamer at every variable residue position. In addition, generally no van der Waals energy is calculated if the atoms are connected by three bonds or less. For the atomic solvation scoring function, the surface of the rotamer is measured against the surface of the template, and the E_as for each possible rotamer at every variable residue position is calculated. The secondary structure propensity scoring function is also considered as a singles energy, and thus the total singles energy may contain an E_ss term. As will be appreciated by those in the art, many of these energy terms will be close to zero, depending on the physical distance between the rotamer and the template position; that is, the farther apart the two moieties, the lower the energy.

For the calculation of "doubles" energy (rotamer/rotamer), the interaction energy of each possible rotamer is compared with every possible rotamer at all other variable residue positions. Thus, "doubles" energies are calculated for the interaction of every possible rotamer at every variable residue position with every possible rotamer at every other variable residue position, using some or all of the scoring functions. Thus, for the hydrogen bonding scoring function, every hydrogen bonding atom of the first rotamer and every hydrogen bonding atom of every possible second rotamer is evaluated, and the E_HB is calculated for each possible rotamer pair for any two variable positions. Similarly, for the van der Waals scoring function, every atom of the first rotamer is compared to every atom of every possible second rotamer, and the E_vdW is calculated for each possible rotamer pair at every two variable residue positions. For the atomic solvation scoring function, the surface of the first rotamer is measured against the surface of every possible second rotamer, and the E_as for each possible rotamer pair at every two variable residue positions is calculated. The secondary structure propensity scoring function need not be run as a "doubles" energy, as it is considered as a component of the "singles" energy. As will be appreciated by those in the art, many of these double energy terms will be close to zero, depending on the physical distance between the first rotamer and the second rotamer; that is, the farther apart the two moieties, the lower the energy.

In addition, as will be appreciated by those in the art, a variety of force fields that can be used in the PCA calculations can be used, including, but not limited to, Dreiding I and Dreiding II (Mayo et al, J. Phys. Chem. 948897 (1990)), AMBER (Weiner et al, J. Amer. Chem. Soc. 106:765 (1984) and Weiner et al, J. Comp. Chem. 106:230 (1986)), MM2 (Allinger J. Chem. Soc. 99:8127 (1977), Liljefors et al, J. Com. Chem. 8:1051 (1987)); MMP2 (Sprague et al, J. Comp. Chem. 8:581 (1987)); CHARMM (Brooks et al, J. Comp. Chem. 106:187 (1983)); GROMOS; and MM3 (Allinger et al, J. Amer. Chem. Soc. 111:8551 (1989)), OPLS-AA (Jorgensen, et al, J. Am. Chem. Soc. (1996), v 118, pp 11225-11236; Jorgensen, W.L; BOSS, Version 4.1; Yale University: New Haven, CT (1999)); OPLS (Jorgensen, et al, J. Am. Chem. Soc. (1988), v 110, pp 1657ff; Jorgensen, et al, J Am. Chem. Soc. (1990), v 112, pp 4768ff); UNRES (United Residue Forcefield; Liwo, et al. Protein Science (1993), v 2, pp1697-1714; Liwo, et al. Protein Science (1993), v 2, pp1715-1731 ; Liwo, et al, J. Comp. Chem. (1997), v 18, pp849-873; Liwo, et al, J. Comp. Chem. (1997), v 18, pp874-884; Liwo, et al, J. Comp. Chem. (1998), v 19, pp259-276; Forcefield for Protein Structure Prediction (Liwo, et al, Proc. Natl. Acad. Sci. USA (1999), v 96, pp5482-5485); ECEPP/3 (Liwo et al, J Protein Chem 1994 May;13(4):375-80); AMBER 1.1 force field (Weiner, et al, J. Am. Chem. Soc. v106, pp765-784); AMBER 3.0 force field (U.C. Singh et al, Proc. Natl. Acad. Sci. USA. 82:755-759); CHARMM and CHARMM22 (Brooks, et al, J. Comp. Chem. v4, pp 187-217); cvff3.0 (Dauber-Osguthorpe, et al,(1988) Proteins: Structure, Function and Genetics, v4,pp31-47); cff91 (Maple, et al, J. Comp. Chem. v15, 162-182); also, the DISCOVER (cvff and cff91) and AMBER forcefields are used in the INSIGHT molecular modeling package (Biosym/MSI, San Diego California) and HARMM is used in the QUANTA molecular modeling package (Biosym/MSI, San Diego California), all of which are expressly incorporated by reference.

Once the singles and doubles energies are calculated and stored, the next step of the computational processing may occur. As outlined in U.S.S.N. 09/127,926 and PCT US98/07254, preferred embodiments utilize a Dead End Elimination (DEE) step, and preferably a Monte Carlo step.

PDA, viewed broadly, has three components that may be varied to alter the output (e.g. the primary library): the scoring functions used in the process; the filtering technique, and the sampling technique.

In a preferred embodiment, the scoring functions may be altered. In a preferred embodiment, the scoring functions outlined above may be biased or weighted in a variety of ways. For example, a bias towards or away from a reference sequence or family of sequences can be done; for example, a bias towards wild-type or homolog residues may be used. Similarly, the entire protein or a fragment of it may be biased; for example, the active site may be biased towards wild-type residues, or domain residues towards a particular desired physical property can be done. Furthermore, a bias towards or against increased energy can be generated. Additional scoring function biases include, but are not limited to applying electrostatic potential gradients or hydrophobicity gradients, adding a substrate or binding partner to the calculation, or biasing towards a desired charge or hydrophobicity. In addition, in an alternative embodiment, there are a variety of additional scoring functions that may be used. Additional scoring functions include, but are not limited to torsional potentials, or residue pair potentials, or residue entropy potentials. Such additional scoring functions can be used alone, or as functions for processing the library after it is scored initially. For example, a variety of functions derived from data on binding of peptides to MHC (Major Histocompatibility Complex) can be used to rescore a library in order to eliminate proteins containing sequences which can potentially bind to MHC, i.e. potentially immunogenic sequences.

In a preferred embodiment, a variety of filtering techniques can be done, including, but not limited to, DEE and its related counterparts. Additional filtering techniques include, but are not limited to branch- and-bound techniques for finding optimal sequences (Gordon and Majo, Structure Fold. Des. 7:1089- 98, 1999), and exhaustive enumeration of sequences. It should be noted however, that some techniques may also be done without any filtering techniques; for example, sampling techniques can be used to find good sequences, in the absence of filtering.

As will be appreciated by those in the art, once an optimized sequence or set of sequences is generated, (or again, these need not be optimized or ordered) a variety of sequence space sampling methods can be done, either in addition to the preferred Monte Carlo methods, or instead of a Monte Carlo search. That is, once a sequence or set of sequences is generated, preferred methods utilize sampling techniques to allow the generation of additional, related sequences for testing.

These sampling methods can include the use of amino acid substitutions, insertions or deletions, or recombinations of one or more sequences. As outlined herein, a preferred embodiment utilizes a Monte Carlo search, which is a series of biased, systematic, or random jumps. However, there are other sampling techniques that can be used, including Boltzman sampling, genetic algorithm techniques and simulated annealing. In addition, for all the sampling techniques, the kinds of jumps allowed can be altered (e.g. random jumps to random residues, biased jumps (to or away from wild- type, for example), jumps to biased residues (to or away from similar residues, for example), etc.). Jumps where multiple residue positions are coupled (two residues always change together, or never change together), jumps where whole sets of residues change to other sequences (e.g, recombination). Similarly, for all the sampling techniques, the acceptance criteria of whether a sampling jump is accepted can be altered, to allow broad searches at high temperature and narrow searches close to local optima at low temperatures. See Metropolis et al, J. Chem Phys v21, pp 1087, 1953, hereby expressly incorporated by reference. In addition, it should be noted that the preferred methods of the invention result in a rank ordered list of sequences; that is, the sequences are ranked on the basis of some objective criteria. However, as outlined herein, it is possible to create a set of non-ordered sequences, for example by generating a probability table directly (for example using SCMF analysis or sequence alignment techniques) that lists sequences without ranking them. The sampling techniques outlined herein can be used in either situation.

In a preferred embodiment, Boltzman sampling is done. As will be appreciated by those in the art, the temperature criteria for Boltzman sampling can be altered to allow broad searches at high temperature and narrow searches close to local optima at low temperatures (see e.g. Metropolis et al, J. Chem. Phys. 21 :1087, 1953).

In a preferred embodiment, the sampling technique utilizes genetic algorithms, e.g, such as those described by Holland (Adaptation in Natural and Artifical Systems, 1975, Ann Arbor, U. Michigan Press). Genetic algorithm analysis generally takes generated sequences and recombines them computationally, similar to a nucleic acid recombination event, in a manner similar to "gene shuffling". Thus the "jumps" of genetic algorithm analysis generally are multiple position jumps. In addition, as outlined below, correlated multiple jumps may also be done. Such jumps can occur withdifferent crossover positions and more than one recombination at a time, and can involve recombination of two or more sequences. Furthermore, deletions or insertions (random or biased) can be done. In addition, as outlined below, genetic algorithm analysis may also be used after the secondary library has been generated.

In a preferred embodiment, the sampling technique utilizes simulated annealing, e.g, such as described by Kirkpatrick et al. (Science, 220:671-680, 1983). Simulated annealing alters the cutoff for accepting good or bad jumps by altering the temperature. That is, the stringency of the cutoff is altered by altering the temperature. This allows broad searches at high temperature to new areas of sequence space, altering with narrow searches at low temperature to explore regions in detail.

In addition, as outlined below, these sampling methods can be used to further process a secondary library to generate additional secondary libraries (sometimes referred to herein as tertiary libraries). Thus, the primary library can be generated in a variety of computational ways, including structure based methods such as PDA, or sequence based methods, or combinations as outlined herein.

Accordingly, the computational processing results in a set of sequences, that may be optimized protein sequences if some sort of ranking or scoring functions are used. These optimized protein sequences are generally, but not always, significantly different from the wild-type sequence from which the backbone was taken. That is, each optimized protein sequence preferably comprises at least about 5- 10% variant amino acids from the starting or wild-type sequence, with at least about 15-20% changes being preferred and at least about 30% changes being particularly preferred.

The cutoff for the primary library is then enforced, resulting in a set of primary sequences forming the primary library. As outlined above, this may be done in a variety of ways, including an arbitrary cutoff, an energy limitation, or when a certain number of residue positions have been varied. In general, the size of the primary library will vary with the size of the protein, the number of residues that are changing, the computational methods used, the cutoff applied and the discretion of the user. In general, it is preferable to have the primary library be large enough to randomly sample a reasonable sequence space to allow for robust secondary libraries. Thus, primary libraries that range from about 50 to about 10¹³ are preferred, with from about 1000 to about 10⁷ being particularly preferred, and from about 1000 to about 100,000 being especially preferred.

In a preferred embodiment when scoring is used, although this is not required, the primary library comprises the globally optimal sequence in its optimal conformation, i.e. the optimum rotamer at each variable position. That is, computational processing is run until the simulation program converges on a single sequence which is the global optimum. In a preferred embodiment, the primary library comprises at least two optimized protein sequences. Thus for example, the computational processing step may eliminate a number of disfavored combinations but be stopped prior to convergence, providing a library of sequences of which the global optimum is one. In addition, further computational analysis, for example using a different method, may be run on the library, to further eliminate sequences or rank them differently. Alternatively, as is more fully described in U.S.S.N.s 60/061 ,097, 60/043,464, 60/054,678, 09/127,926 and PCT US98/07254, the global optimum may be reached, and then further computational processing may occur, which generates additional optimized sequences in the neighborhood of the global optimum. In addition, in some embodiments, primary library sequences that did not make the cutoff are included in the primary library. This may be desirable in some situations to evaluate the primary library generation method, to serve as controls or comparisons, or to sample additional sequence space. For example, in a preferred embodiment, the wild-type sequence is included.

It should also be noted that different ranking systems can be used. For example, a list of naturally occurring sequences can be used to calculate all possible recombinations of these sequences, with an optional rank ordering step. Alternatively, once a primary library is generated, one could rank order only those recombinations that occur at cross-over points with at least a threshold of identity over a given window. For example, 100% identity over a window of 6 amino acids, or 80% identity over a window of 10 amino acids. Alternatively, as for all the systems outlined herein, the homology could be considered at the DNA level, by computationally considering the translation fo the amino acids to their respective DNA codons. Different codon usages could be considered. A preferred embodiment considers only recombinations with crossover points that have DNA sequence identity sufficient for DNA hybridization of the differing sequences.

As is further outlined below, It should also be noted that combining different primary libraries may be done. For example, positions in a protein that show a great deal of mutational diversity in computational screening can be fixed as outlined below and a different primary library regenerated. A rank ordered list of the same length as the first would now show diversity in previously rarely changing positions. The variants from the first primary library can be combined with the variants from the second primary library to provide a combined library at lower computational cost than creating a very long rank ordered list. This approach can be particularly useful to sample sequence diversity in both low energy gap, readily changing surface positions and high energy gap, rarely changing core positions. In addition, primary libraries can be generated by combining one or more of the different calculations to form one big primary library.

Thus, the present invention provides primary libraries comprising a list of computationally derived sequences. In a preferred embodiment, these sequences are in the form of a rank ordered list. From this primary library, a secondary library is generated. As outlined herein, there are a number of different ways to generate a secondary library.

In a preferred embodiment, the primary library of the scaffold protein is used to generate a secondary library. As will be appreciated by those in the art, the secondary library can be either a subset of the primary library, or contain new library members, i.e. sequences that are not found in the primary library. That is, in general, the variant positions and/or amino acid residues in the variant positions can be recombined in any number of ways to form a new library that exploits the sequence variations found in the primary library. That is, having identified "hot spots" or important variant positions and/or residues, these positions can be recombined in novel ways to generate novel sequences to form a secondary library. Thus, in a preferred embodiment, the secondary library comprises at least one member sequence that is not found in the primary library, and preferably a plurality of such sequences.

In one embodiment, all or a portion of the primary library serves as the secondary library. That is, a cutoff is applied to the primary sequences and these sequences serve as the secondary library, without further manipulation or recombination. The library members can be made as outlined below, e.g. by direct synthesis or by constructing the nucleic acids encoding the library members, expressing them in a suitable host, optionally followed by screening.

In a preferred embodiment, the secondary library is generated by tabulating the amino acid positions that vary from a reference sequence. The reference sequence can be arbitrarily selected, or preferably is chosen either as the wild-type sequence or the global optimum sequence, with the latter being preferred. That is, each amino acid position that varies in the primary library is tabulated. Of course, if the original computational analysis fixed some positions, the variable positions of the secondary library will comprise either just these original variable positions or some subset of these original variable positions. That is, assuming a protein of 100 amino acids, the original computational screen can allow all 100 positions to be varied. However, due to the cutoff in the primary library, only 25 positions may vary. Alternatively, assuming the same 100 amino acid protein, the original computational screen could have varied only 25 positions, keeping the other 75 fixed; this could result in only 12 of the 25 being varied in the cutoff primary library. These primary library positions can then be recombined to form a secondary library, wherein all possible combinations of these variable positions form the secondary library. It should be noted that the non-variable positions are set to the reference sequence positions.

The formation of the secondary library using this method may be done in two general ways; either all variable positions are allowed to be any amino acid, or subsets of amino acids are allowed for each position. In a preferred embodiment, all amino acid residues are allowed at each variable position identified in the primary library. That is, once the variable positions are identified, a secondary library comprising every combination of every amino acid at each variable position is made.

In a preferred embodiment, subsets of amino acids are chosen. The subset at any position may be either chosen by the user, or may be a collection of the amino acid residues generated in the primary screen. That is, assuming core residue 25 is variable and the primary screen gives 5 different possible amino acids for this position, the user may chose the set of good core residues outlined above (e.g. hydrophobic residues), or the user may build the set by chosing the 5 different amino acids generated in the primary screen. Alternatively, combinations of these techniques may be used, wherein the set of identified residues is manually expanded. For example, in some embodiments, fewer than the number of amino acid residues is chosen; for example, only three of the five may be chosen. Alternatively, the set is manually expanded; for example, if the computation picks two different hydrophobic residues, additional choices may be added. Similarly, the set may be biased, for example either towards or away from the wild-type sequence, or towards or away from known domains, etc.

In addition, this may be done by analyzing the primary library to determine which amino acid positions in the scaffold protein have a high mutational frquency, and which positions have a low mutation frequency. The secondary library can be generated by randomizing the amino acids at the positions that have high numbers of mutations, while keeping constant the positions that do not have mutations above a certain frequency. For example, if the position has less than 20% and more preferably 10% mutations, it may be kept constant as the reference sequence position.

In a preferred embodiment, the secondary library is generated from a probability distribution table. As outlined herein, there are a variety of methods of generating a probability distribution table, including using PDA, sequence alignments, forcefield calculations such as SCMF calculations, etc. In addition, the probability distribution can be used to generate information entropy scores for each position, as a measure of the mutational frequency observed in the library.

In this embodiment, the frequency of each amino acid residue at each variable position in the list is identified. Frequencies can be thresholded, wherein any variant frequency lower than a cutoff is set to zero. This cutoff is preferably 1%, 2%, 5%, 10% or 20%, with 10% being particularly preferred. These frequencies are then built into the secondary library. That is, as above, these variable positions are collected and all possible combinations are generated, but the amino acid residues that "fill" the secondary library are utilized on a frequency basis. Thus, in a non-frequency based secondary library, a variable position that has 5 possible residues will have 20% of the proteins comprising that variable position with the first possible residue, 20% with the second, etc. However, in a frequency based secondary library, a variable position that has 5 possible residues with frequencies of 10%, 15%, 25%, 30% and 20%, respectively, will have 10% of the proteins comprising that variable position with the first possible residue, 15% of the proteins with the second residue, 25% with the third, etc. As will be appreciated by those in the art, the actual frequency may depend on the method used to actually generate the proteins; for example, exact frequencies may be possible when the proteins are synthesized. However, when the frequency-based primer system outlined below is used, the actual frequencies at each position will vary, as outlined below.

As will be appreciated by those in the art and outlined herein, probability distribution tables can be generated in a variety of ways. In addition to the methods outlined herein, self-consistent mean field (SCMF) methods can be used in the direct generation of probability tables. SCMF is a deterministic computational method that uses a mean field description of rotamer interactions to calculate energies. A probability table generated in this way can be used to create secondary libraries as described herein. SCMF can be used in three ways: the frequencies of amino acids and rotamers for each amino acid are listed at each position; the probabilities are determined directly from SCMF (see Delarue et la. Pac. Symp. Biocomput. 109-21 (1997), expressly incorporated by reference). In addition, highly variable positions and non-variable positions can be identified. Alternatively, another method is used to determine what sequence is jumped to during a search of sequence space; SCMF is used to obtain an accurate energy for that sequence; this energy is then used to rank it and create a rank-ordered list of sequences (similar to a Monte Carlo sequence list). A probability table showing the frequencies of amino acids at each position can then be calculated from this list (Koehl et al, J. Mol. Biol. 239:249 (1994); Koehl et al, Nat. Struc. Biol. 2:163 (1995); Koehl et al, Curr. Opin. Struct. Biol. 6:222 (1996); Koehl et al, J. Mol. Bio. 293:1183 (1999); Koehl et al, J. Mol. Biol. 293:1161 (1999); Lee J. Mol. Biol. 236:918 (1994); and Vasquez Biopolymers 36:53-70 (1995); all of which are expressly incorporated by reference. Similar methods include, but are not limited to, OPLS-AA (Jorgensen, et al, J. Am. Chem. Soc. (1996), v 118, pp 11225-11236; Jorgensen, W.L.; BOSS, Version 4.1 ; Yale University: New Haven, CT (1999)); OPLS (Jorgensen, et al, J. Am. Chem. Soc. (1988), v 110, pp 1657ff; Jorgensen,^' et al, J Am. Chem. Soc. (1990), v 112, pp 4768ff); UNRES (United Residue Forcefield; Liwo, et al. Protein Science (1993), v 2, pp1697-1714; Liwo, et al. Protein Science (1993), v 2, pp1715-1731; Liwo, et al, J. Comp. Chem. (1997), v 18, pp849-873; Liwo, et al, J. Comp. Chem. (1997), v 18, pp874-884; Liwo, et al, J. Comp. Chem. (1998), v 19, pp259-276; Forcefield for Protein Structure Prediction (Liwo, et al, Proc. Natl. Acad. Sci. USA (1999), v 96, pp5482-5485); ECEPP/3 (Liwo et al, J Protein Chem 1994 May;13(4):375-80); AMBER 1.1 force field (Weiner, et al, J. Am. Chem. Soc. v106, pp765-784); AMBER 3.0 force field (U.C. Singh et al, Proc. Natl. Acad. Sci. USA. 82:755-759); CHARMM and CHARMM22 (Brooks, et al, J. Comp. Chem. v4, pp 187-217); cvff3.0 (Dauber-Osguthorpe, et al, (1988) Proteins: Structure, Function and Genetics, v4,pp31-47); cff91 (Maple, et al, J. Comp. Chem. v15, 162-182); also, the DISCOVER (cvff and cff91) and AMBER forcefields are used in the INSIGHT molecular modeling package (Biosym/MSI, San Diego California) and HARMM is used in the QUANTA molecular modeling package (Biosym/MSI, San Diego California).

In addition, as outlined herein, a preferred method of generating a probability distribution table is through the use of sequence alignment programs. In addition, the probability table can be obtained by a combination of sequence alignments and computational approaches. For example, one can add amino acids found in the alignment of homologous sequences to the result of the computation. Preferable one can add the wild type amino acid identity to the probability table if it is not found in the computation.

As will be appreciated, a secondary library created by recombining variable positions and/or residues at the variable position may not be in a rank-ordered list. In some embodiments, the entire list may just be made and tested. Alternatively, in a preferred embodiment, the secondary library is also in the form of a rank ordered list. This may be done for several reasons, including the size of the secondary library is still too big to generate experimentally, or for predictive purposes. This may be done in several ways. In one embodiment, the secondary library is ranked using the scoring functions of PDA to rank the library members. Alternatively, statistical methods could be used. For example, the secondary library may be ranked by frequency score; that is, proteins containing the most of high frequency residues could be ranked higher, etc. This may be done by adding or multiplying the frequency at each variable position to generate a numerical score. Similarly, the secondary library different positions could be weighted and then the proteins scored; for example, those containing certain residues could be arbitrarily ranked.

As outlined herein, secondary libraries can be generated in two general ways. The first is computationally, as above, wherein the primary library is further computationally manipulated, for example by recombining the possible variant positions and/or amino acid residues at each variant position or by recombining portions of the sequences containing one or more variant position. It may be ranked, as outlined above. This computationally-derived secondary library can then be experimentally generated by synthesizing the library members or nucleic acids encoding them, as is more fully outlined below. Alternatively, the secondary library is made experimentally; that is, nucleic acid recombination techniques are used to experimentally generate the combinations. This can be done in a variety of ways, as outlined below. Once the coding region for the candidate proteins are made, they are added to the NAP system, as is outlined below.

Thus, there are two steps to the formation of the nucleic acids for use in the NAP systems described herein. First, the nucleic acids encoding the candidate proteins must be made, and then fused to the nucleic acid encoding the NAM enzymes, and put into the NAP expression vectors. The first step, generating the nucleic acids encoding the candidate proteins, can be part of the formation of the secondary or tertiary libraries. That is, by using certain techniques, such as certain PCR reactions, "shuffling" techniques, and other nucleic acid manipulations, the variable positions of the library can be recombined to form new library members. Ultimately, the coding sequences for these new members, which form the secondary library, are added to the NAP system as described below.

In a preferred embodiment, particularly for longer proteins or proteins for which large samples are desired, the secondary library sequences are used to create nucleic acids such as DNA which encode the member sequences and which can then be fused with the other components of the NAP system, cloned into host cells, expressed and assayed, if desired. Thus, nucleic acids, and particularly DNA, can be made which encodes each member protein sequence (sometimes referred to herein as "candidate proteins" or "variant proteins"). This is done using well known procedures and as more fully described below. The choice of codons, suitable expression vectors and suitable host cells will vary depending on a number of factors, and can be easily optimized as needed.

In a preferred embodiment, multiple PCR reactions with pooled oligonucleotides is done, as is generally depicted in Figure 50. In this embodiment, overlapping oligonucleotides are synthesized which correspond to the full length gene (assuming that a full length gene is required; again, the invention includes the use of fragments or domains of proteins as well). Again, these oligonucleotides may represent all of the different amino acids at each variant position or subsets.

In a preferred embodiment, these oligonucleotides are pooled in equal proportions and multiple PCR reactions are performed to create full length sequences containing the combinations of mutations defined by the secondary library. In addition, this may be done using error-prone PCR methods. In a preferred embodiment, the different oligonucleotides are added in relative amounts corresponding to the probability distribution table. The multiple PCR reactions thus result in full length sequences with the desired combinations of mutaions in the desired proportions.

The total number of oligonucleotides needed is a function of the number of positions being mutated and the number of mutations being considered at these positions:

(number of oligos for constant positions) + M1 + M2 + M3 +... Mn = (total number of oligos required), where Mn is the number of mutations considered at position n in the sequence.

In a preferred embodiment, each overlapping oligonucleotide comprises only one position to be varied; in alternate embodiments, the variant positions are too close together to allow this and multiple variants per oligonucleotide are used to allow complete recombination of all the possibilities. That is, each oligo can contain the codon for a single position being mutated, or for more than one position being mutated. The multiple positions being mutated must be close in sequence to prevent the oligo length from being impractical. For multiple mutating positions on an oligonucleotide, particular combinations of mutations can be included or excluded in the library by including or excluding the oligonucleotide encoding that combination. For example, as discussed herein, there may be correlations between variable regions; that is, when position X is a certain residue, position Y must (or must not) be a particular residue. These sets of variable positions are sometimes referred to herein as a "cluster". When the clusters are comprised of residues close together, and thus can reside on one oligonuclotide primer, the clusters can be set to the "good" correlations, and eliminate the bad combinations that may decrease the effectiveness of the library. However, if the residues of the cluster are far apart in sequence, and thus will reside on different oligonuclotides for synthesis, it may be desirable to either set the residues to the "good" correlation, or eliminate them as variable residues entirely. In an alternative embodiment, the library may be generated in several steps, so that the cluster mutations only appear together. This procedure, i.e., the procedure of identifying mutation clusters and either placing them on the same oligonucleotides or eliminating them from the library or library generation in several steps preserving clusters, can considerably enrich the experimental library with properly folded protein. Identification of clusters can be carried out by a number of wasy, e.g. by using known pattern recognition methods, comparisons of frequencies of occurrence of mutations or by using energy analysis of the sequences to be experimentally generated (for example, if the energy of interaction is high, the positions are correlated), these correlations may be positional correlations (e.g. variable positions 1 and 2 always change together or never change together) or sequence correlations (e.g. if there is a residue A at position 1 , there is always residue B at position 2). See: Pattern discovery in Biomolecular Data: Tools, Techniques, and Applications; edited by Jason T.L. Wang, Bruce A. Shapiro, Dennis Shasha. New York: Oxford Unviersity, 1999; Andrews, Harry C. Introduction to mathematical techniques in patter recognition; New York, Wiley-lnterscience [1972]; Applications of Pattern Recognition; Editor, K.S. Fu. Boca Raton, Fla. CRC Press, 1982; Genetic Algorithms for Pattern Recognition; edited by Sankar K. Pal, Paul P. Wang. Boca Raton : CRC Press, c1996; Pandya, Abhijit S, Pattern recognition with Neural networks in C++/Abhijit S. Pandya, Robert B. Macy. Boca Raton, Fla.: CRC Press, 1996; Handbook of pattern recognition and computer vision / edited by CH. Chen, L.F. Pau, P.S.P. Wang. 2^nd ed. Signapore ; River Edge, N.J. : World Scientific, c1999; Friedman, Introduction to Pattern Recognition : Statistical, Structural, Neural, and Fuzzy Logic Approaches ; River Edge, N.J. : World Scientific, c1999, Series title: Serien a machine perception and artificial intelligence; vol. 32; all of wh ich are expressly incorporated by reference. In addition programs used to search for consensus motifs can be used as well.

In addition, correlations and shuffling can be fixed or optimized by altering the design of the oligonucleotides; that is, by deciding where the oligonucleotides (primers) start and stop (e.g. where the sequences are "cut"). The start and stop sites of oligos can be set to maximize the number of clusters that appear in single oligonucleotides, thereby enriching the library with higher scoring sequences. Different oligonucleotides start and stop site options can be computationally modeled and ranked according to number of clusters that are represented on single oligos, or the percentage of the resulting sequences consistent with the predicted library of sequences.

The total number of oligonucleotides required increases when multiple mutable positions are encoded by a single oligonucleotide. The annealed regions are the ones that remain constant, i.e. have the sequence of the reference sequence.

Oligonucleotides with insertions or deletions of codons can be used to create a library expressing different length proteins. In particular computational sequence screening for insertions or deletions can result in secondary libraries defining different length proteins, which can be expressed by a library of pooled oligonucleotide of different lengths.

In a preferred embodiment, the secondary library is done by shuffling the family (e.g. a set of variants); that is, some set of the top sequences (if a rank-ordered list is used) can be shuffled, either with or without error-prone PCR. "Shuffling" in this context means a recombination of related sequences, generally in a random way. It can include "shuffling" as defined and exemplified in U.S. Patent Nos. 5,830,721 ; 5,811 ,238; 5,605,793; 5,837,458 and PCT US/19256, all of which are expressly incorporated by reference in their entirety. This set of sequences can also be an artificial set; for example, from a probability table (for example generated using SCMF) or a Monte Carlo set. Similarly, the "family" can be the top 10 and the bottom 10 sequences, the top 100 sequences, etc. This may also be done using error-prone PCR.

Thus, in a preferred embodiment, in silico shuffling is done using the computational methods described therein. That is, starting with either two libraries or two sequences, random recombinations of the sequences can be generated and evaluated.

In a preferred embodiment, error-prone PCR is done to generate the secondary library. See U.S. Patent Nos. 5,605,793, 5,811,238, and 5,830,721 , all of which are hereby incorporated by reference. This can be done on the optimal sequence or on top members of the library, or some other artificial set or family. In this embodiment, the gene for the optimal sequence found in the computational screen of the primary library can be synthesized. Error prone PCR is then performed on the optimal sequence gene in the presence of oligonucleotides that code for the mutations at the variant positions of the secondary library (bias oligonucleotides). The addition of the oligonucleotides will create a bias favoring the incorporation of the mutations in the secondary library. Alternatively, only oligonucleotides for certain mutations may be used to bias the library.

In a preferred embodiment, gene shuffling with error prone PCR can be performed on the gene for the optimal sequence, in the presence of bias oligonucleotides, to create a DNA sequence library that reflects the proportion of the mutations found in the secondary library. The choice of the bias oligonucleotides can be done in a variety of ways; they can chosen on the basis of their frequency, i.e. oligonucleotides encoding high mutational frequency positions can be used; alternatively, oligonucleotides containing the most variable positions can be used, such that the diversity is increased; if the secondary library is ranked, some number of top scoring positions can be used to generate bias oligonucleotides; random positions may be chosen; a few top scoring and a few low scoring ones may be chosen; etc. What is important is to generate new sequences based on preferred variable positions and sequences.

In a preferred embodiment, PCR using a wild type gene or other gene can be used, as is schematically depicted in Figure 5. In this embodiment, a starting gene is used; generally, although this is not required, the gene is the wild type gene. In some cases it may be the gene encoding the global optimized sequence, or any other sequence of the list. In this embodiment, oligonucleotides are used that correspond to the variant positions and contain the different amino acids of the secondary library. PCR is done using PCR primers at the termini, as is known in the art. This provides two benefits; the first is that this generally requires fewer oligonucleotides and can result in fewer errors. In addition, it has experimental advantages in that if the wild type gene is used, it need not be synthesized.

In a preferred embodiment, a variety of additional steps may be done to one or more secondary libraries; for example, further computational processing can occur, secondary libraries can be recombined, or cutoffs from different secondary libraries can be combined. In a preferred embodiment, a secondary library may be computationally remanipulated to form an additional secondary library (sometimes referred to herein as "tertiary libraries"). For example, any of the secondary library sequences may be chosen for a second round of PDA, by freezing or fixing some or all of the changed positions in the first secondary library. Alternatively, only changes seen in the last probability distribution table are allowed. Alternatively, the stringency of the probability table may be altered, either by increasing or decreasing the cutoff for inclusion. Similarly, the secondary library may be recombined experimentally after the first round; for example, the best gene/genes from the first screen may be taken and gene assembly redone (using techniques outlined below, multiple PCR, error prone PCR, shuffling, etc.). Alternatively, the fragments from one or more good gene(s) to change probabilities at some positions. This biases the search to an area of sequence space found in the first round of computational and experimental screening.

In a preferred embodiment, a tertiary library can be generated from combining secondary libraries. For example, a probability distribution table from a secondary library can be generated and recombined, wither computationally or experimentally, as outlined herein. A PDA secondary library may be combined with a sequence alignment secondary library, and either recombined (again, computationally or experimentally) or just the cutoffs from each joined to make a new tertiary library. The top sequences from several libraries can be recombined. Primary and secondary libraries can similarly be combined. Sequences from the top of a library can be combined with sequences from the bottom of the library to more broadly sample sequence space, or only sequences distant from the top of the library can be combined. Primary and/or secondary libraries that analyzed different parts of a protein can be combined to a tertiary library that treats the combined parts of the protein. These combinations can be done to analyze large proteins, especially large multidomain proteins or complete protesomes. In a preferred embodiment, a tertiary library can be generated using correlations in the secondary library. That is, a residue at a first variable position may be correlated to a residue at second variable position (or correlated to residues at additional positions as well). For example, two variable positions may sterically or electrostatically interact, such that if the first residue is X, the second residue must be Y. This may be either a positive or negative correlation. This correlation, or "cluster" of residues, may be both detected and used in a variety of ways. (For the generation of correlations, see the earlier cited art).

In addition, primary and secondary libraries can be combined to form new libraries; these can be random combinations or the libraries, combining the "top" sequences, or weighting the combinations (positions or residues from the first library are scored higher than those of the second library).

As outlined herein, any number of protein attributes may be altered in these methods, including, but not limited to, enzyme activity, stability, solubility, aggregation, binding affinity, binding specificity, substrate specificity, structural integrity, immunogenicity, toxicity, generate peptide and peptidomimmetic libraries, create new antibody CDR's, generate new DNA, RNA bindings, etc.

It should be noted that therapeutic proteins utilized in these methods will preferentially have residues in the hydrophobic cores screened, to prevent changes in the molecular surface of the protein that might induce immunogenic responses. Therapeutic proteins can also be designed in the region surrounding their binding sites to their receptors. Such a region can be defined, for example, by including in the design all residues within a certain distance, for example 4.5 A of the binding site residues. This range can vary from 4 to 6-10 A. This design will serve to improve activity and specificity.

In addition, a step method can be done; see Zhao et al. Nature Biotech. 16:258 (1998), hereby incorporated by reference.

In a preferred embodiment, the methods of the invention are used not on known scaffold proteins, but on random peptides, to search a virtual library for those sequences likely to adapt a stable conformation. As discussed above, there is a current benefit and focus on screening random peptide libraries to find novel binding/modulators. However, the sequences in these experimental libraries can be randomized at specific sites only, or throughout the sequence. The number of sequences that can be searched in these libraries grows exponentially with the number of positions that are randomized. Generally, only up to 10¹² - 10¹⁵ sequences can be contained in a library because of the physical constraints of laboratories (the size of the instruments, the cost of producing large numbers of biopolymers, etc.). Other practical considerations can often limit the size of the libraries to 10^s or fewer. These limits are reached for only 10 amino acid positions. Therefore, only a sparse sampling of sequences is possible in the search for improved proteins or peptides in experimental sequence libraries, lowering the chance of success and almost certainly missing desirable candidates. Because of the randomness of the changes in these sequences, most of the candidates in the library are not suitable, resulting in a waste of most of the effort in producing the library.

However, using the automated protein design techniques outlined herein, virtual libraries of protein sequences can be generated that are vastly larger than experimental libraries. Up to 10⁷⁵ candidate sequences (or more) can be screened computationally and those that meet design criteria which favor stable and functional proteins can be readily selected. An experimental library consisting of the favorable candidates found in the virtual library screening can then be generated, resulting in a much more efficient use of the experimental library and overcoming the limitations of random protein libraries. Thus, the methods of the invention allow the virtual screening of a set of random peptides for peptides likely to take on a particular structure, and thus eliminating the large number of unpreferred or unallowed conformations without having to make and test the peptides.

In addition, it is possible to randomize regions or domains of protein as well.

Thus, in a preferred embodiment, the invention provides libraries of completely defined set of variant scaffold proteins, wherein at least 85% of the possible members are in the library, with at least about 90% and 95% being particularly preferred. However, it is also possible that errors are introduced into the libraries experimentally, and thus the libraries contain preferably less than 25% non-defined (e.g. error) sequences; with less than 10%, less than 5% and less than 1% particularly preferred. Thus libraries that have all members as well as some error members, or some members as well as error members are included herein.

As mentioned above, two principle benefits come from the virtual library screening: (1) the automated protein design generates a list of sequence candidates that are favored to meet design criteria; it also shows which positions in the sequence are readily changed and which positions are unlikely to change without disrupting protein stability and function. An experimental random library can be generated that is only randomized at the readily changeable, non-disruptive sequence positions. (2) The diversity of amino acids at these positions can be limited to those that the automated design shows are compatible with these positions. Thus, by limiting the number of randomized positions and the number of possibilities at these positions, the number of wasted sequences produced in the experimental library is reduced, thereby increasing the probability of success in finding sequences with useful properties. For example, the table below lists the 10 favored sequences candidates from the virtual screening of 12 positions in a protein. It shows that positions 9, 10 and 12 are most likely to have changes that do not disrupt the function of the protein, suggesting that a random experimental library that randomizes positions 9, 10 and 12 will have a higher fraction of desirable sequences. Also, the virtual library suggests that position 10 is most compatible with lie or Phe residues, further limiting the size of the library and allowing a more complete screening of good sequences.

1 2 3 4 5 6 7 8 9 10 11 12

1 LEU LEU ILE ILE ALA LEU LEU LEU LEU PHE ALA LEU

2 LEU LEU ILE ILE ALA LEU LEU LEU LEU ILE ALA LEU

3 LEU LEU ILE ILE ALA LEU LEU LEU LEU ILE ALA LEU

4 LEU LEU ILE ILE ALA LEU LEU LEU LEU PHE ALA ILE

5 LEU LEU ILE ILE ALA LEU LEU LEU LEU PHE ALA ILE

6 LEU LEU ILE ILE ALA LEU LEU LEU LEU ILE ALA ILE

7 LEU LEU ILE ILE ALA LEU LEU LEU ILE PHE ALA LEU

8 LEU LEU ILE ILE ALA LEU LEU LEU LEU ILE ALA ILE

9 LEU LEU ILE ILE ALA LEU LEU LEU ILE PHE ALA LEU

10 LEU LEU ILE ILE ALA LEU LEU LEU LEU LEU ALA LEU

The automated design method uses physical chemical criteria to screen sequences, resulting in sequences that are likely to be stable, structured, and that preserve function, if needed. Different design criteria can be used to produce candidate sets that are biased for properties such as charged, solubility, or active site characteristics (polarity, size), or are biased to have certain amino acids at certain positions. That is, The candidate bioactive agents and candidate nucleic acids are randomized, either fully randomized or they are biased in their randomization, e.g. in nucleotide/residue frequency generally or per position. By "randomized" or grammatical equivalents herein is meant that each nucleic acid and peptide consists of essentially random nucleotides and amino acids, respectively. Thus, any amino acid residue may be incorporated at any position. The synthetic process can be designed to generate randomized peptides and/or nucleic acids, to allow the formation of all or most of the possible combinations over the length of the nucleic acid, thus forming a library of randomized candidate nucleic acids.

In one embodiment, the library is fully randomized, with no sequence preferences or constants at any position. In a preferred embodiment, the library is biased. That is, some positions within the sequence are either held constant, or are selected from a limited number of possibilities. For example, in a preferred embodiment, the nucleotides or amino acid residues are randomized within a defined class, for example, of hydrophobic amino acids, hydrophilic residues, sterically biased (either small or large) residues, towards the creation of cysteines, for cross-linking, prolines for SH-3 domains, serines, threonines, tyrosines or histidines for phosphorylation sites, etc, or to purines, etc.

In a preferred embodiment, the bias is towards peptides or nucleic acids that interact with known classes of molecules. For example, it is known that much of intracellular signaling is carried out via short regions of polypeptides interacting with other polypeptides through small peptide domains. For instance, a short region from the HIV-1 envelope cytoplasmic domain has been previously shown to block the action of cellular calmodulin. Regions of the Fas cytoplasmic domain, which shows homology to the mastoparan toxin from Wasps, can be limited to a short peptide region with death- inducing apoptotic or G protein inducing functions. Magainin, a natural peptide derived from Xenopus, can have potent anti-tumour and anti-microbial activity. Short peptide fragments of a protein kinase C isozyme (βPKC), have been shown to block nuclear translocation of βPKC in Xenopus oocytes following stimulation. And, short SH-3 target peptides have been used as psuedosubstrates for specific binding to SH-3 proteins. This is of course a short list of available peptides with biological activity, as the literature is dense in this area. Thus, there is much precedent for the potential of small peptides to have activity on intracellular signaling cascades. In addition, agonists and antagonists of any number of molecules may be used as the basis of biased randomization of candidate bioactive agents as well.

In general, the generation of a prescreened random peptide libraries may be described as follows. Any structure, whether a known structure, for example a portion of a known protein, a known peptide, etc, or a synthetic structure, can be used as the backbone for PDA. For example, structures from X- ray crystallographic techniques, NMR techniques, de novo modelling, homology modelling, etc. may all be used to pick a backbone for which sequences are desired. Similarly, a number of molecules or protein domains are suitable as starting points for the generation of biased randomized candidate bioactive agents. A large number of small molecule domains are known, that confer a common function, structure or affinity. In addition, as is appreciated in the art, areas of weak amino acid homology may have strong structural homology. A number of these molecules, domains, and/or corresponding consensus sequences, are known, including, but are not limited to, SH-2 domains, SH- 3 domains, Pleckstrin, death domains, protease cleavage/recognition sites, enzyme inhibitors, enzyme substrates, Traf, etc. Similarly, there are a number of known nucleic acid binding proteins containing domains suitable for use in the invention. For example, leucine zipper consensus sequences are known.

Thus, in general, known peptide ligands can be used as the starting backbone for the generation of the primary library. In addition, structures known to take on certain conformations may be used to create a backbone, and then sequences screened for those that are likely to take on that conformation. For example, there are a wide variety of "ministructures" known, sometimes referred to as "presentation structures", that can confer conformational stability or give a random sequence a conformationally restricted form. Proteins interact with each other largely through conformationally constrained domains. Although small peptides with freely rotating amino and carboxyl termini can have potent functions as is known in the art, the conversion of such peptide structures into pharmacologic agents is difficult due to the inability to predict side-chain positions for peptidomimetic synthesis. Therefore the presentation of peptides in conformationally constrained structures will benefit both the later generation of pharmaceuticals and will also likely lead to higher affinity interactions of the peptide with the target protein. This fact has been recognized in the combinatorial library generation systems using biologically generated short peptides in bacterial phage systems. A number of workers have constructed small domain molecules in which one might present randomized peptide structures.

Thus, synthetic presentation structures, i.e. artificial polypeptides, are capable of presenting a randomized peptide as a conformationally-restricted domain. Preferred presentation structures maximize accessibility to the peptide by presenting it on an exterior loop. Accordingly, suitable presentation structures include, but are not limited to, minibody structures, loops on beta-sheet turns and coiled-coil stem structures in which residues not critical to structure are randomized, zinc-finger domains, cysteine-linked (disulfide) structures, transglutaminase linked structures, cyclic peptides, B- loop structures, helical barrels or bundles, leucine zipper motifs, etc.

In a preferred embodiment, the presentation structure is a coiled-coil structure, allowing the presentation of the randomized peptide on an exterior loop. See, for example, Myszka et al, Biochem. 33:2362-2373 (1994), hereby incorporated by reference, and Figure 3). Using this system investigators have isolated peptides capable of high affinity interaction with the appropriate target. In general, coiled-coil structures allow for between 6 to 20 randomized positions; (see Martin et al, EMBO J. 13(22):5303-5309 (1994), incorporated by reference).

In a preferred embodiment, the presentation structure is a minibody structure. A "minibody" is essentially composed of a minimal antibody complementarity region. The minibody presentation structure generally provides two randomizing regions that in the folded protein are presented along a single face of the tertiary structure. See for example Bianchi et al, J. Mol. Biol. 236(2):649-59 (1994), and references cited therein, all of which are incorporated by reference). Investigators have shown this minimal domain is stable in solution and have used phage selection systems in combinatorial libraries to select minibodies with peptide regions exhibiting high affinity, Kd = 10^"7, for the pro- inflammatory cytokine IL-6.

Once the backbone is chosen and the primary library of the random peptides generated as outlined above, the secondary library generation and creation proceeds as for the known scaffold protein, including recombination of variant positions and/or amino acid residues, either computationally or experimentally. Again, libraries of DNA expressing the protein sequences defined by the automated protein design methods can be produced. Codons can be randomized at only the nucleotide sequence triplets that define the residue positions specified by the automated design method. Also, mixtures of base triplets that code for particular amino acids could be introduced into the DNA synthesis reaction to attach a full triplet defining an amino acid in one reaction step. Also, a library of random DNA oligomers could be designed that biases the desired positions toward certain amino acids, or that restricts those positions to certain amino acids. The amino acids biased for would be those specified in the virtual screening, or a subset of those.

Multiple DNA libraries are synthesized that code for different subsets of amino acids at certain positions, allowing generation of the amino acid diversity desired without having to fully randomize the codon and thereby waste sequences in the library on stop codons, frameshifts, undesired amino acids, etc. This can be done by creating a library that at each position to be randomized is only randomized at one or two of the positions of the triplet, where the position(s) left constant are those that the amino acids to be considered at this position have in common. Multiple DNA libraries would be created to insure that all amino acids desired at each position exist in the aggregate library. Alternatively, "shuffling", as is generally known in the art, can be done with multiple libraries. In addition, in silico shuffling can also be done.

Alternatively, the random peptide libraries may be done using the frequency tabulation and experimental generation methods including multiplexed PCR, shuffling, etc.

There are a wide variety of experimental techniques that can be used to experimentally generate the libraries of the invention, including, but not limited to, Rachitt-Enchira (http://www.enchira.com/gene_shuffling.htm); error-prone PCR, for example using modified nucleotides; known mutagenesis techniques including the use of multi-cassettes; DNA shuffling (Crameri, et al. Nature 391 (6664):288-291. (1998)); heterogeneous DNA samples (US5939250); ITCHY (Ostermeier, et al, Nat Biotechnol 17(12):1205-1209. (1999)); StEP (Zhao, et al, Nat Biotechnol 16(3):258-261. (1998)), GSSM (US6171820.US5965408); in vivo homologous recombination, ligase assisted gene assembly, end-complementary PCR, profusion (Roberts and Szostak, Proc Natl Acad Sci U S A 94(23): 12297-12302. (1997)); yeast bacteria surface display (Lu, et al. Biotechnology (N Y) 13(4):366-372. (1995);Seed and Aruffo, Proc Natl Acad Sci U S A 84(10):3365-3369. (1987);Boder and Wittrup, Nat Biotechnol 15(6):553-557. (1997)).

NAP Conjugate construction and screening

Using the nucleic acids of the present invention which encode library members (e.g. candidate proteins), fusion nucleic acids are made for incorporation into the NAP system.

Accordingly, the present invention provides libraries of nucleic acid molecules comprising nucleic acid sequences encoding fusion nucleic acids encoding a nucleic acid modification enzyme and a candidate protein. By "nucleic acid" or "oligonucleotide" or grammatical equivalents herein means at least two nucleosides covalently linked together. A nucleic acid of the present invention will generally contain phosphodiester bonds, although in some cases nucleic acid analogs are included that may have alternate backbones, particularly when the target molecule is a nucleic acid, comprising, for example, phosphoramide (Beaucage et al. Tetrahedron 49(10): 1925 (1993) and references therein; Letsinger, J. Org. Chem. 35:3800 (1970); Sprinzl et al, Eur. J. Biochem. 81 :579 (1977); Letsinger et al, Nucl. Acids Res. 14:3487 (1986); Sawai et al, Chem. Lett. 805 (1984), Letsinger et al, J. Am. Chem. Soc. 110:4470 (1988); and Pauwels et al, Chemica Scripta 26:141 91986)), phosphorothioate (Mag et al. Nucleic Acids Res. 19:1437 (1991); and U.S. Patent No. 5,644,048), phosphorodithioate (Briu et al, J. Am. Chem. Soc. 111:2321 (1989), O-methylphophoroamidite linkages (see Eckstein, Oligonucleotides and Analogues: A Practical Approach, Oxford University Press), and peptide nucleic acid backbones and linkages (see Egholm, J. Am. Chem. Soc. 114:1895 (1992); Meier et al, Chem. Int. Ed. Engl. 31 :1008 (1992); Nielsen, Nature, 365:566 (1993); Carlsson et al. Nature 380:207 (1996), all of which are incorporated by reference). Other analog nucleic acids include those with positive backbones (Denpcy et al, Proc. Natl. Acad. Sci. USA 92:6097 (1995); non-ionic backbones (U.S. Patent Nos. 5,386,023, 5,637,684, 5,602,240, 5,216,141 and 4,469,863; Kiedrowshi et al, Angew. Chem. Intl. Ed. English 30:423 (1991); Letsinger et al, J. Am. Chem. Soc. 110:4470 (1988); Letsinger et al, Nucleoside & Nucleotide 13:1597 (1994); Chapters 2 and 3, ASC Symposium Series 580, "Carbohydrate Modifications in Antisense Research", Ed. Y.S. Sanghui and P. Dan Cook; Mesmaeker et al, Bioorganic & Medicinal Chem. Lett. 4:395 (1994); Jeffs et al, J. Biomolecular NMR 34:17 (1994); Tetrahedron Lett. 37:743 (1996)) and non-ribose backbones, including those described in U.S. Patent Nos. 5,235,033 and 5,034,506, and Chapters 6 and 7, ASC Symposium Series 580, "Carbohydrate Modifications in Antisense Research", Ed. Y.S. Sanghui and P. Dan Cook. Nucleic acids containing one or more carbocyclic sugars are also included within the definition of nucleic acids (see Jenkins et al, Chem. Soc. Rev. (1995) pp169-176). Several nucleic acid analogs are described in Rawls, C & E News June 2, 1997 page 35. All of these references are hereby expressly incorporated by reference. These modifications of the ribose-phosphate backbone may be done to facilitate the addition of other elements, such as labels, or to increase the stability and half-life of such molecules in physiological environments.

As will be appreciated by those in the art, all of these nucleic acid analogs may find use in the present invention. In addition, mixtures of naturally occurring nucleic acids and analogs can be made, or, alternatively, mixtures of different nucleic acid analogs, and mixtures of naturally occurring nucleic acids and analogs may be made.

The nucleic acids may be single stranded or double stranded, as specified, or contain portions of both double stranded or single stranded sequence. The nucleic acid may be DNA, both genomic and cDNA, RNA or a hybrid, where the nucleic acid contains any combination of deoxyribo- and ribo- nucleotides, and any combination of bases, including uracil, adenine, thymine, cytosine, guanine, inosine, xathanine hypoxathanine, isocytosine, isoguanine, etc. As used herein, the term "nucleoside" includes nucleotides and nucleoside and nucleotide analogs, and modified nucleosides such as amino modified nucleosides. In addition, "nucleoside" includes non-naturally occuring analog structures. Thus for example the individual units of a peptide nucleic acid, each containing a base, are referred to herein as a nucleoside.

The present invention provides libraries of nucleic acid molecules comprising nucleic acid sequences encoding fusion nucleic acids. By "fusion nucleic acid" herein is meant a plurality of nucleic acid components that are joined together. The fusion nucleic acids encode fusion polypeptides. By "fusion polypeptide" or "fusion peptide" or grammatical equivalents herein is meant a protein composed of a plurality of protein components, that while typically unjoined in their native state, are joined by their respective amino and carboxyl termini through a peptide linkage to form a single continuous polypeptide. Plurality in this context means at least two, and preferred embodiments generally utilize two components. It will be appreciated that the protein components can be joined directly or joined through a peptide linker/spacer as outlined below. In addition, it should be noted that in some embodiments, as is more fully outlined below, the fusion nucleic acids encode protein components that are not fused; for example, the fusion nucleic acid may comprise an intron that is removed, leaving two non-associated protein components, although generally the nucleic acids encoding each component are fused. Furthermore, as outlined below, additional components such as fusion partners including targeting sequences, etc. may be used.

The fusion nucleic acids encode nucleic acid modification (NAM) enzymes and candidate proteins. By "nucleic acid modification enzyme" or "NAM enzyme" herein is meant an enzyme that utilizes nucleic acids, particularly DNA, as a substrate and covalently attaches itself to nucleic acid enzyme attachment (EA) sequences. The covalent attachment can be to the base, to the ribose moiety or to the phosphate moieties. NAM enzymes include, but are not limited to, helicases, topoisomerases, polymerases, gyrases, recombinases, transposases, restriction enzymes and nucleases. As outlined below, NAM enzymes include variants. Although many DNA binding peptides are known, such as those involved in nucleic acid compaction, transcription regulators, and the like, enzymes that covalently attach to DNA, in particular peptides involved with replication, are preferred. Some NAM enzymes can form covalent linkages with DNA without nicking the DNA. For example, it is believed that enzymes involved in DNA repair recognize and covalently attach to nucleic acid regions, which can be either double-stranded or single-stranded. Such NAM enzymes are suitable for use in the fusion enzyme library. However, DNA NAM enzymes that nick DNA to form a covalent linkage, e.g, viral replication peptides, are most preferred.

Preferably, the NAM enzyme is a protein that recognizes specific sequences or conformations of a nucleic acid substrate and performs its enzymatic activity such that a covalent complex is formed with the nucleic acid substrate. Preferably, the enzyme acts upon nucleic acids, particularly DNA, in various configurations including, but not limited to, single-strand DNA, double-strand DNA, Z-form DNA, and the like.

Suitable NAM enzymes, include, but are not limited to, enzymes involved in replication such as Rep68 and Rep78 of adeno-associated viruses (AAV), NS1 and H-1 of parvovirus, bacteriophage phi-29 terminal proteins, the 55 Kd adenovirus proteins, and derivatives thereof.

In a preferred embodiment, the NAM enzyme is a Rep protein. Adeno-associated viral (AAV) Rep proteins are encoded by the left open reading frame of the viral genome. AAV Rep proteins, such as Rep68 and Rep78, regulate AAV transcription, activate AAV replication, and have been shown to inhibit transcription of heterologous promoters (Chiorini et al, J. Virol, 68(2), 797-804 (1994), hereby incorporated by reference in its entirety). The Rep68 and Rep78 proteins act, in part, by covalently attaching to the AAV inverted terminal repeat (Prasad et al. Virology, 229, 183-192 (1997); Prasad et al. Virology 214:360 (1995); both of which are hereby incorporated by reference in their entirety). These Rep proteins act by a site-specific and strand-specific endonuclease nick at the AAV origin at the terminal resolution site, followed by covalent attachment to the 5' terminus of the nicked site via a putative tyrosine linkage. Rep68 and Rep78 result from alternate splicing of the transcript. The nucleic acid and protein sequences of Rep68 and Rep78 are shown in Figures 15, 16, 1, 2, 7, 8, 13 and 14. respectively. As is further outlined below, functional fragments and variants of Rep proteins are also included within the definition of Rep proteins; in this case, the variants preferably include nucleic acid binding activity and endonuclease activity. The corresponding enzyme attachment site for Rep68 and Rep78, discussed below, is shown in Figures 47 and 48 and set forth in Example 1 of PCT US00/22906, hereby expressly incorporated by reference. In a preferred embodiment, the NAM enzyme is NS1. NS1 is a non-structural protein in parvovirus, is a functional homolog of Rep78, and also covalently attaches to DNA (Cotmore et al, J. Virol, 62(3), 851-860 (1998), hereby expressly incorporated by reference). The nucleotide and amino acid sequences of various NS1 proteins are shown in Figures 9-12, 29-34, 37 and 38. As is further outlined below, fragments and variants of NS1 proteins are also included within the definition of NS1 proteins.

In a preferred embodiment, the NAM enzyme is the parvoviral H-1 protein, which is also known to form a covalent linkage with DNA (see, for example, Tseng et al, Proc. Natl. Acad. Sci. USA, 76(11), 5539- 5543 (1979), hereby expressly incorporated by reference. As is further outlined below, fragments and variants of H-1 proteins are also included within the definition of H-1 proteins.

In a preferred embodiment, the NAM enzyme is the bacteriophage phi-29 terminal protein, which is also known to form a covalent linkage with DNA (see, for example, Germendia et al. Nucleic Acid Research, 16(3), 5727-5740 (1988), hereby expressly incorporated by reference. As is further outlined below, fragments and variants of phi-29 proteins are also included within the definition of phi-29 proteins.

In a preferred embodiment, the NAM enzyme is the adenoviral 55 Kd (a55) protein, again known to form covalent linkages with DNA; see Desiderio and Kelly, J. Mol. Biol, 98, 319-337 (1981), hereby expressly incorporated by reference. As is further outlined below, fragments and variants of a55 proteins are also included within the definition of a55 proteins.

The nucleic acid and amino acid sequences of other Rep homologs that are suitable for use as NAM enzymes are set forth in Figures 3-6, 17-28, 35, 36 and 39-46.

Some DNA-binding enzymes form covalent linkages upon physical or chemical stimuli such as, for example, UV-induced crosslinking between DNA and a bound protein, or camptothecin (CPT)-related chemically induced trapping of the DNA-topoisomerase I covalent complex (e.g, Hertzberg et al, J. Biol. Chem, 265, 19287-19295 (1990)). NAM enzymes that form induced covalent linkages are suitable for use in some embodiments of the present invention.

Also included with the definition of NAM enzymes of the present invention are amino acid sequence variants. These variants fall into one or more of three classes: substitutional, insertional or deletional (e.g. fragment) variants. These variants ordinarily are prepared by site specific mutagenesis of nucleotides in the DNA encoding the NAM protein, using cassette or PCR mutagenesis or other techniques well known in the art, to produce DNA encoding the variant, and thereafter expressing the recombinant DNA in cell culture as outlined herein. However, variant NAM protein fragments having up to about 100-150 residues may be prepared by in vitro synthesis or peptide ligation using established techniques. Amino acid sequence variants are characterized by the predetermined nature of the variation, a feature that sets them apart from naturally occurring allelic or interspecies variation of the NAM protein amino acid sequence. The variants typically exhibit the same qualitative biological activity as the naturally occurring analogue, although variants can also be selected which have modified characteristics as will be more fully outlined below.

While the site or region for introducing an amino acid sequence variation is predetermined, the mutation per se need not be predetermined. For example, in order to optimize the performance of a mutation at a given site, random mutagenesis may be conducted at the target codon or region and the expressed NAM variants screened for the optimal combination of desired activity. Techniques for making substitution mutations at predetermined sites in DNA having a known sequence are well known, for example, M13 primer mutagenesis and PCR mutagenesis. Screening of the mutants is done using assays of NAM protein activities.

Amino acid substitutions are typically of single residues; insertions usually will be on the order of from about 1 to 20 amino acids, although considerably larger insertions may be tolerated. Deletions range from about 1 to about 20 residues, although in some cases deletions may be much larger, for example when unnecessary domains are removed.

Substitutions, deletions, insertions or any combination thereof may be used to arrive at a final derivative. Generally these changes are done on a few amino acids to minimize the alteration of the molecule. However, larger changes may be tolerated in certain circumstances. When small alterations in the characteristics of the NAM protein are desired, substitutions are generally made in accordance with the following chart:

Chart I

Original Residue Exemplary Substitutions

Ala Ser

Arg Lys

Asn Gin, His

Asp Glu

Cys Ser

Gin Asn

Glu Asp Gly Pro

His Asn, Gin lie Leu, Val

Leu lie, Val

Lys Arg, Gin, Glu

Met Leu, lie

Phe Met, Leu, Tyr

Ser Thr

Thr Ser

Trp Tyr

Tyr Trp, Phe

Val lie, Leu

Substantial changes in function or immunological identity are made by selecting substitutions that are less conservative than those shown in Chart I. For example, substitutions may be made which more significantly affect: the structure of the polypeptide backbone in the area of the alteration, for example the alpha-helical or beta-sheet structure; the charge or hydrophobicity of the molecule at the target site; or the bulk of the side chain. The substitutions which in general are expected to produce the greatest changes in the polypeptide's properties are those in which (a) a hydrophilic residue, e.g. seryl or threonyl, is substituted for (or by) a hydrophobic residue, e.g. leucyl, isoleucyl, phenylalanyl, valyl or alanyl; (b) a cysteine or proline is substituted for (or by) any other residue; (c) a residue having an electropositive side chain, e.g. lysyl, arginyl, or histidyl, is substituted for (or by) an electronegative residue, e.g. glutamyl or aspartyl; or (d) a residue having a bulky side chain, e.g. phenylalanine, is substituted for (or by) one not having a side chain, e.g. glycine.

The variants typically exhibit the same qualitative biological activity as the naturally-occurring analogue, although variants also are selected to modify the characteristics of the NAM proteins as needed. Alternatively, the variant may be designed such that the biological activity of the NAM protein is altered. For example, glycosylation sites may be altered or removed. Similarly, functional mutations within the endonuclease domain or nucleic acid recognition site may be made. Furthermore, unnecessary domains may be deleted, to form fragments of NAM enzymes.

In addition, some embodiments utilize concatameric constructs to effect multivalency and increase binding kinetics or efficiency. For example, constructs containing a plurality of NAM coding regions or a plurality of EASs may be made. Also included with the definition of NAM protein are other NAM homologs, and NAM proteins from other organisms including viruses, which are cloned and expressed as known in the art. Thus, probe or degenerate polymerase chain reaction (PCR) primer sequences may be used to find other related NAM proteins. As will be appreciated by those in the art, particularly useful probe and/or PCR primer sequences include the unique areas of the NAM nucleic acid sequence. As is generally known in the art, preferred PCR primers are from about 15 to about 35 nucleotides in length, with from about 20 to about 30 being preferred, and may contain inosine as needed. The conditions for the PCR reaction are well known in the art.

In addition to nucleic acids encoding NAM enzymes, the fusion nucleic acids of the invention also encode the candidate proteins generated by the computational/experimental methods outlined above. "Candidate proteins" in this context includes proteins to be testing for binding, association or effect in an assay of the invention, including both in vitro (e.g. cell free systems) or ex vivo (within cells). The sequences of the candidate proteins are generated computationally and/or experimentally, as outlined above. Generally, as outlined below, libraries of candidate proteins are used in the fusions.

In a preferred embodiment, libraries of computational candidate proteins are fused to the NAM enzymes, with each member of the library comprising a different candidate protein. However, as will be appreciated by those in the art, different members of the library may be reproduced or duplicated, resulting in some libraries members being identical. The library should provide a sufficiently structurally diverse population of expression products to effect a probabilistically sufficient range of cellular responses to provide one or more cells exhibiting a desired response. Accordingly, an interaction library must be large enough so that at least one of its members will have a structure that gives it affinity for some molecule, including both protein and non-protein targets, or other factors whose activity is necessary or effective within the assay of interest. Although it is difficult to gauge the required absolute size of an interaction library, nature provides a hint with the immune response: a diversity of 10⁷-10⁸ different antibodies provides at least one combination with sufficient affinity to interact with most potential antigens faced by an organism. Published in vitro selection techniques have also shown that a library size of 10⁷ to 10⁸ is sufficient to find structures with affinity for the target. A library of all combinations of a peptide 7 to 20 amino acids in length has the potential to code for 20⁷ (10⁹) to 20²⁰ . Thus, with libraries of 10⁷ to 10⁸ the present methods allow a "working" subset of a theoretically complete interaction library for 7 amino acids, and a subset of shapes for the 20²⁰ library. Thus, in a preferred embodiment, at least 10⁶, preferably at least 10⁷, more preferably at least 10⁸ and most preferably at least 10⁹ different expression products are simultaneously analyzed in the subject methods. Preferred methods maximize library size and diversity. It is important to understand that in any library system encoded by oligonucleotide synthesis one cannot have complete control over the codons that will eventually be incorporated into the peptide structure. This is especially true in the case of codons encoding stop signals (TAA, TGA, TAG). In a synthesis with NNN as the random region, there is a 3/64, or 4.69%, chance that the codon will be a stop codon. Thus, in a peptide of 10 residues, there is a high likelihood that 46.7% of the peptides will prematurely terminate. One way to alleviate this is to have random residues encoded as NNK, where K= T or G. This allows for encoding of all potential amino acids (changing their relative representation slightly), but importantly preventing the encoding of two stop residues TAA and TGA. Thus, libraries encoding a 10 amino acid peptide will have a 15.6% chance to terminate prematurely. Alternatively, fusing the candidate proteins to the C-terminus of the NAM enzyme also may be done, although in some instances, fusing to the N-terminus means that prematurely terminating proteins result in a lack of NAM enzyme which eliminates these samples from the assay.

The fusion nucleic acid can comprise the NAM enzyme and candidate protein in a variety of configurations, including both direct and indirect fusions, and include N- and C-terminal fusions and internal fusions.

In a preferred embodiment, the NAM enzyme and the candidate protein are directly fused. In this embodiment, a direct, in-frame fusion of the nucleic acid encoding the NAM enzyme and the candidate protein is done. Again, this may be done in several ways, including N- and C-terminal fusions and internal fusions. Thus, the NAM enzyme coding region may be 3' or 5' to the candidate protein coding region, or the candidate protein coding region may be inserted into a suitable position within the coding region of the NAM enzyme. In this embodiment, it may be desirable to insert the candidate protein into an external loop of the NAM enzyme, either as a direct insertion or with the replacement of several of the NAM enzyme residues. This may be particularly desirable in the case of random candidate proteins, as they frequently require some sort of scaffold or presentation structure to confer a conformationally restricted structure. For an example of this general idea using green fluorescent protein (GFP) as a scaffold for the expression of random peptide libraries, see for example WO 99/20574, expressly incorporated herein by reference. Furthermore, in this embodiment, generally only a single set of regulatory elements such as promoters are used.

In a preferred embodiment, the NAM enzyme and the candidate protein are indirectly fused. This may be done such that the components of the fusion remain attached, such as through the use of linkers, or in ways that result in the components of the fusion becoming separated. As will be appreciated by those in the art, there are a wide variety of different types of linkers that may be used, including cleavable and non-cleavable linkers; this cleavage may also occur at the level of the nucleic acid, or at the protein level. In a preferred embodiment, linkers may be used to functionally isolate the NAM enzyme and the candidate protein. That is, a direct fusion system may sterically or functionally hinder the interaction of the candidate protein with its intended binding partner, and thus fusion configurations that allow greater degrees of freedom are useful. An analogy is seen in the single chain antibody area, where the incorporation of a linker allows functionality.

In a preferred embodiment, linkers known to confer flexibility are used. For example, useful linkers include glycine-serine polymers (including, for example, (GS)_n, (GSGGS)_n and (GGGS)_n, where n is an integer of at least one), glycine-alanine polymers, alanine-serine polymers, and other flexible linkers such as the tether for the shaker potassium channel, and a large variety of other flexible linkers, as will be appreciated by those in the art. Glycine-serine polymers are preferred since both of these amino acids are relatively unstructured, and therefore may be able to serve as a neutral tether between components. Secondly, serine is hydrophilic and therefore able to solubilize what could be a globular glycine chain. Third, similar chains have been shown to be effective in joining subunits of recombinant proteins such as single chain antibodies.

In a preferred embodiment, the linker is a cleavable linker. Cleavable linkers may function at the level of the nucleic acid or the protein. That is, cleavage (which in this sense means that the NAM enzyme and the candidate protein are separated) may occur during transcription, or before or after translation.

In a preferred embodiment, the cleavage occurs as a result of cleavage functionality built into the nucleic acid. In this embodiment, for example, cleavable nucleic acid sequences, or sequences that will disrupt the nucleic acid, can be used. For example, intron sequences that the cell will remove can be placed between the coding region of the NAM enzyme and the candidate protein. See Figure 49, which depicts two different vectors comprising exon donor sites and splice recipient sites.

In a preferred embodiment, the linkers are heterodimerization domains, as depicted in Figure 49. In this embodiment, both the NAM enzyme and the candidate protein are fused to heterodimerization domains (or multimeric domains, if multivalency is desired), to allow association of these two proteins after translation.

In a preferred embodiment, cleavable protein linkers are used. In this embodiment, the fusion nucleic acids include coding sequences for a protein sequence that may be subsequently cleaved, generally by a protease. As will be appreciated by those in the art, cleavage sites directed to ubiquitous proteases, e.g. those that are constitutively present in most or all of the host cells of the system, can be used. Alternatively, cleavage sites that correspond to cell-specific proteases may be used. Similarly, cleavage sites for proteases that are induced only during certain cell cycles or phases or are signal specific events may be used as well.

There are a wide variety of possible proteinaceous cleavage sites known. For example, sequences that are recognized and cleaved by a protease or cleaved after exposure to certain chemicals are considered cleavable linkers. This may find particular use in in vitro systems, outlined below, as exogenous enzymes can be added to the milieu or the NAP conjugates may be purified and the cleavage agents added. For example, cleavable linkers include, but are not limited to, the prosequence of bovine chymosin, the prosequence of subtilisin, the 2a site (Ryan et al, J. Gen. Virol. 72:2727 (1991); Ryan et al, EMBO J. 13:928 (1994); Donnelly et al, J. Gen. Virol. 78:13 (1997); Hellen et al, Biochem, 28(26):9881 (1989); and Mattion et al, J. Virol. 70:8124 (1996)), prosequences of retroviral proteases including human immunodeficiency virus protease and sequences recognized and cleaved by trypsin (EP 578472, Takasuga et al, J. Biochem. 112(5)652 (1992)) factor Xa (Gardella et al, J. Biol. Chem. 265(26): 15854 (1990), WO 9006370), collagenase (J03280893, Tajima et al, J. Ferment. Bioeng. 72(5):362 (1991), WO 9006370), clostripain (EP 578472), subtilisin (including mutant H64A subtilisin, Forsberg et al, J. Protein Chem. 10(5):517 (1991), chymosin, yeast KEX2 protease (Bourbonnais et al, J. Bio. Chem. 263(30):15342 (1988), thrombin (Forsberg et al, supra; Abath et al, BioTechniques 10(2): 178 (1991)), Staphylococcus aureus V8 protease or similar endoproteinase-Glu-C to cleave after Glu residues (EP 578472, Ishizaki et al, Appl. Microbiol. Biotechnol. 36(4):483 (1992)), cleavage by Nla proteinase of tobacco etch virus (Parks et al. Anal. Biochem. 216(2):413 (1994)), endoproteinase-Lys-C (U.S. Patent No. 4,414,332) and endoproteinase- Asp-N, Neisseria type 2 IgA protease (Pohlner et al, Bio/Technology 10(7):799-804 (1992)), soluble yeast endoproteinase yscF (EP 467839), chymotrypsin (Altman et al. Protein Eng. 4(5):593 (1991)), enteropeptidase (WO 9006370), lysostaphin, a polyglycine specific endoproteinase (EP 316748), and the like. See e.g. Marston, F.A.O. (1986) Biol. Chem. J. 240, 1-12. Particular amino acid sites that serve as chemical cleavage sites include, but are not limited to, methionine for cleavage by cyanogen bromide (Shen, PNAS USA 81 :4627 (1984); Kempe et al. Gene 39:239 (1985); Kuliopulos et al, J. Am. Chem. Soc. 116:4599 (1994); Moks et al, Bio/Technology 5:379 (1987); Ray et al, Bio Technology 11 :64 (1993)), acid cleavage of an Asp-Pro bond (Wingender et al, J. Biol. Chem. 264(8):4367 (1989); Gram et al, Bio/Technology 12:1017 (1994)), and hydroxylamine cleavage at an Asn-Gly bond (Moks supra).

In addition to the NAM enzymes, candidate proteins, and linkers, the fusion nucleic acids may comprise additional coding sequences for other functionalities. As will be appreciated by those in the art, the discussion herein is directed to fusions of these other components to the fusion nucleic acids described herein; however, they may also be unconnected to the fusion protein and rather be a component of the expression vector comprising the fusion nucleic acid, as is generally outlined below. Thus, in a preferred embodiment, the fusions are linked to a fusion partner. By "fusion partner" or "functional group" herein is meant a sequence that is associated with the candidate protein, that confers upon all members of the library in that class a common function or ability. Fusion partners can be heterologous (i.e. not native to the host cell), or synthetic (not native to any cell). Suitable fusion partners include, but are not limited to: a) presentation structures, as defined below, which provide the candidate proteins in a conformationally restricted or stable form, including hetero- or homodimerization or multimerization sequences; b) targeting sequences, defined below, which allow the localization of the candidate proteins into a subcellular or extracellular compartment or be incorporated into infected organisms, such as those infected by viruses or pathogens; c) rescue sequences as defined below, which allow the purification or isolation of the NAP conjugates; d) stability sequences, which confer stability or protection from degradation to the candidate protein or the nucleic acid encoding it, for example resistance to proteolytic degradation; e) linker sequences; f) any number of heterologous proteins, particularly for labeling purposes as described herein; or g) any combination of a), b), c), d), e) and f), as well as linker sequences as needed.

In a preferred embodiment, the fusion partner is a presentation structure. By "presentation structure" or grammatical equivalents herein is meant a sequence, which, when fused to candidate proteins, causes the candidate proteins to assume a conformationally restricted form. This is particularly useful when the candidate proteins are random, biased random or pseudorandom peptides. Proteins interact with each other largely through conformationally constrained domains. Although small peptides with freely rotating amino and carboxyl termini can have potent functions as is known in the art, the conversion of such peptide structures into pharmacologic agents is difficult due to the inability to predict side-chain positions for peptidomimetic synthesis. Therefore the presentation of peptides in conformationally constrained structures will benefit both the later generation of pharmaceuticals and will also likely lead to higher affinity interactions of the peptide with the target protein. This fact has been recognized in the combinatorial library generation systems using biologically generated short peptides in bacterial phage systems.

Thus, synthetic presentation structures, i.e. artificial polypeptides, are capable of presenting a randomized peptide as a conformationally-restricted domain. Generally such presentation structures comprise a first portion joined to the N-terminal end of the randomized peptide, and a second portion joined to the C-terminal end of the peptide; that is, the peptide is inserted into the presentation structure, although variations may be made, as outlined below. To increase the functional isolation of the randomized expression product, the presentation structures are selected or designed to have minimal biologically activity when expressed in the target cell. Preferred presentation structures maximize accessibility to the peptide by presenting it on an exterior loop. Accordingly, suitable presentation structures include, but are not limited to, minibody structures, dimerization sequences, loops on beta-sheet turns and coiled-coil stem structures in which residues not critical to structure are randomized, zinc-finger domains, cysteine-linked (disulfide) structures, transglutaminase linked structures, cyclic peptides, B-loop structures, helical barrels or bundles, leucine zipper motifs, etc.

In a preferred embodiment, the presentation structure is a coiled-coil structure, allowing the presentation of the randomized peptide on an exterior loop. See, for example, Myszka et al, Biochem. 33:2362-2373 (1994), hereby incorporated by reference, and Figure 3). Using this system investigators have isolated peptides capable of high affinity interaction with the appropriate target. In general, coiled-coil structures allow for between 6 to 20 randomized positions.

A preferred coiled-coil presentation structure is as follows:

MGCAALESEVSALESEVASLESEVAALGRGDIVIPLAAVKSKLSAVKSKLASVKSKLAACGPP. The underlined regions represent a coiled-coil leucine zipper region defined previously (see Martin et al, EMBO J. 13(22):5303-5309 (1994), incorporated by reference). The bolded GRGDMP region represents the loop structure and when appropriately replaced with randomized peptides (i.e. candidate proteins, generally depicted herein as (X)_n, where X is an amino acid residue and n is an integer of at least 5 or 6) can be of variable length. The replacement of the bolded region is facilitated by encoding restriction endonuclease sites in the underlined regions, which allows the direct incorporation of randomized oligonucleotides at these positions. For example, a preferred embodiment generates a Xhol site at the double underlined LE site and a Hindlll site at the double- underlined KL site.

A preferred minibody presentation structure is as follows:

MGRNSQATSGFTFSHFYMEWVRGGEYIAASRHKHNKYTTEYSASVKGRYIVSRDTSQSILYLQKKKG PP. The bold, underline regions are the regions which may be randomized. The italized phenylalanine must be invariant in the first randomizing region. The entire peptide is cloned in a three-oligonudeotide variation of the coiled-coil embodiment, thus allowing two different randomizing regions to be incorporated simultaneously. This embodiment utilizes non-palindromic BstXI sites on the termini.

In a preferred embodiment, the presentation structure is a sequence that contains generally two cysteine residues, such that a disulfide bond may be formed, resulting in a conformationally constrained sequence. This embodiment is particularly preferred when secretory targeting sequences are used. As will be appreciated by those in the art, any number of random sequences, with or without spacer or linking sequences, may be flanked with cysteine residues. In other embodiments, effective presentation structures may be generated by the random regions themselves. For example, the random regions may be "doped" with cysteine residues which, under the appropriate redox conditions, may result in highly crosslinked structured conformations, similar to a presentation structure. Similarly, the randomization regions may be controlled to contain a certain number of residues to confer β-sheet or α-helical structures.

In one embodiment, the presentation structure is a dimerization or multimerization sequence. A dimerization sequence allows the non-covalent association of one candidate protein to another candidate protein, including peptides, with sufficient affinity to remain associated under normal physiological conditions. This effectively allows small libraries of candidate protein (for example, 10⁴) to become large libraries if two proteins per cell are generated which then dimerize, to form an effective library of 10⁸ (10⁴ X 10⁴). It also allows the formation of longer proteins, if needed, or more structurally complex molecules. The dimers may be homo- or heterodimers.

Dimerization sequences may be a single sequence that self-aggregates, or two sequences. That is, nucleic acids encoding both a first candidate protein with dimerization sequence 1 , and a second candidate protein with dimerization sequence 2, such that upon introduction into a cell and expression of the nucleic acid, dimerization sequence 1 associates with dimerization sequence 2 to form a new structure.

Suitable dimerization sequences will encompass a wide variety of sequences. Any number of protein- protein interaction sites are known. In addition, dimerization sequences may also be elucidated using standard methods such as the yeast two hybrid system, traditional biochemical affinity binding studies, or even using the present methods.

In a preferred embodiment, the fusion partner is a targeting sequence. As will be appreciated by those in the art, the localization of proteins within a cell is a simple method for increasing effective concentration and determining function. For example, RAF1 when localized to the mitochondrial membrane can inhibit the anti-apoptotic effect of BCL-2. Similarly, membrane bound Sos induces Ras mediated signaling in T-lymphocytes. These mechanisms are thought to rely on the principle of limiting the search space for ligands, that is to say, the localization of a protein to the plasma membrane limits the search for its ligand to that limited dimensional space near the membrane as opposed to the three dimensional space of the cytoplasm. Alternatively, the concentration of a protein can also be simply increased by nature of the localization. Shuttling the proteins into the nucleus confines them to a smaller space thereby increasing concentration. Finally, the ligand or target may simply be localized to a specific compartment, and inhibitors must be localized appropriately.

Thus, suitable targeting sequences include, but are not limited to, binding sequences capable of causing binding of the expression product to a predetermined molecule or class of molecules while retaining bioactivity of the expression product, (for example by using enzyme inhibitor or substrate sequences to target a class of relevant enzymes); sequences signalling selective degradation, of itself or co-bound proteins; and signal sequences capable of constitutively localizing the candidate expression products to a predetermined cellular locale, including a) subcellular locations such as the Golgi, endoplasmic reticulum, nucleus, nucleoli, nuclear membrane, mitochondria, chloroplast, secretory vesicles, lysosome, and cellular membrane or within pathogens or viruses that have infected the cell; and b) extracellular locations via a secretory signal. Particularly preferred is localization to either subcellular locations or to the outside of the cell via secretion.

In a preferred embodiment, the targeting sequence is a nuclear localization signal (NLS). NLSs are generally short, positively charged (basic) domains that serve to direct the entire protein in which they occur to the cell's nucleus. Numerous NLS amino acid sequences have been reported including single basic NLS's such as that of the SV40 (monkey virus) large T Antigen (Pro Lys Lys Lys Arg Lys Val), Kalderon (1984), et al. Cell, 39:499-509; the human retinoic acid receptor-β nuclear localization signal (ARRRRP); NFKB p50 (EEVQRKRQKL; Ghosh et al. Cell 62:1019 (1990); NFKB p65 (EEKRKRTYE; Nolan et al. Cell 64:961 (1991); and others (see for example Boulikas, J. Cell. Biochem. 55(1):32-58 (1994), hereby incorporated by reference) and double basic NLS's exemplified by that of the Xenopus (African clawed toad) protein, nucleoplasmin (Ala Val Lys Arg Pro Ala Ala Thr Lys Lys Ala Gly Gin Ala Lys Lys Lys Lys Leu Asp), Dingwall, et al. Cell, 30:449-458, 1982 and Dingwall, et al, J. Cell Biol, 107:641-849; 1988). Numerous localization studies have demonstrated that NLSs incorporated in synthetic peptides or grafted onto reporter proteins not normally targeted to the cell nucleus^'cause these peptides and reporter proteins to be concentrated in the nucleus. See, for example, Dingwall, and Laskey, Ann, Rev. Cell Biol, 2:367-390, 1986; Bonnerot, et al, Proc. Natl. Acad. Sci. USA, 84:6795-6799, 1987; Galileo, et al, Proc. Natl. Acad. Sci. USA, 87:458-462, 1990. In a preferred embodiment, the targeting sequence is a membrane anchoring signal sequence. This is particularly useful since many parasites and pathogens bind to the membrane, in addition to the fact that many intracellular events originate at the plasma membrane. Thus, membrane-bound peptide libraries are useful for both the identification of important elements in these processes as well as for the discovery of effective inhibitors. In addition, many drugs interact with membrane associated proteins. The invention provides methods for presenting the candidate proteins extracellularly or in the cytoplasmic space. For extracellular presentation, a membrane anchoring region is provided at the carboxyl terminus of the candidate protein. The candidate protein region is expressed on the cell surface and presented to the extracellular space, such that it can bind to other surface molecules (affecting their function) or molecules present in the extracellular medium. The binding of such molecules could confer function on the cells expressing a peptide that binds the molecule. The cytoplasmic region could be neutral or could contain a domain that, when the extracellular candidate protein region is bound, confers a function on the cells (activation of a kinase, phosphatase, binding of other cellular components to effect function). Similarly, the candidate protein-containing region could be contained within a cytoplasmic region, and the transmembrane region and extracellular region remain constant or have a defined function.

In addition, it should be noted that in this embodiment, as well as others outlined herein, it is possible that the formation of the NAP conjugate happens after the screening; that is, having the fusion protein expressed on the extracellular surface means that it may not be available for binding to the nucleic acid. However, this may be done later, with lysis of the cell.

Membrane-anchoring sequences are well known in the art and are based on the genetic geometry of mammalian transmembrane molecules. Peptides are inserted into the membrane based on a signal sequence (designated herein as ssTM) and require a hydrophobic transmembrane domain (herein TM). The transmembrane proteins are inserted into the membrane such that the regions encoded 5' of the transmembrane domain are extracellular and the sequences 3' become intracellular. Of course, if these transmembrane domains are placed 5' of the variable region, they will serve to anchor it as an intracellular domain, which may be desirable in some embodiments. ssTMs and TMs are known for a wide variety of membrane bound proteins, and these sequences may be used accordingly, either as pairs from a particular protein or with each component being taken from a different protein, or alternatively, the sequences may be synthetic, and derived entirely from consensus as artificial delivery domains.

As will be appreciated by those in the art, membrane-anchoring sequences, including both ssTM and TM, are known for a wide variety of proteins and any of these may be used. Particularly preferred membrane-anchoring sequences include, but are not limited to, those derived from CD8, ICAM-2, IL- 8R, CD4 and LFA-1.

Useful sequences include sequences from: 1) class I integral membrane proteins such as IL-2 receptor beta-chain (residues 1-26 are the signal sequence, 241-265 are the transmembrane residues; see Hatakeyama et al. Science 244:551 (1989) and von Heijne et al, Eur. J. Biochem. 174:671 (1988)) and insulin receptor beta chain (residues 1-27 are the signal, 957-959 are the transmembrane domain and 960-1382 are the cytoplasmic domain; see Hatakeyama, supra, and Ebina et al. Cell 40:747 (1985)); 2) class II integral membrane proteins such as neutral endopeptidase (residues 29-51 are the transmembrane domain, 2-28 are the cytoplasmic domain; see Malfroy et al, Biochem. Biophys. Res. Commun. 144:59 (1987)); 3) type III proteins such as human cytochrome P450 NF25 (Hatakeyama, supra); and 4) type IV proteins such as human P-glycoprotein (Hatakeyama, supra). Particularly preferred are CD8 and ICAM-2. For example, the signal sequences from CD8 and ICAM-2 lie at the extreme 5' end of the transcript. These consist of the amino acids 1-32 in the case of CD8 (MASPLTRFLSLNLLLLGESILGSGEAKPQAP; Nakauchi et al, PNAS USA 82:5126 (1985) and 1-21 in the case of ICAM-2 (MSSFGYRTLTVALFTLICCPG; Staunton et al. Nature (London) 339:61 (1989)). These leader sequences deliver the construct to the membrane while the hydrophobic transmembrane domains, placed 3' of the random candidate region, serve to anchor the construct in the membrane. These transmembrane domains are encompassed by amino acids 145-195 from CD8

(PQRPEDCRPRGSVKGTGLDFACDIYIWAPLAGICVALLLSLIITLICYHSR; Nakauchi, supra) and 224- 256 from ICAM-2 (MVIIVTWSVLLSLFVTSVLLCFIFGQHLRQQR; Staunton, supra).

Alternatively, membrane anchoring sequences include the GPI anchor, which results in a covalent bond between the molecule and the lipid bilayer via a glycosyl-phosphatidylinositol bond for example in DAF (PNKGSGTTSGTTRLLSGHTCFTLTGLLGTLVTMGLLT, with the bolded serine the site of the anchor; see Homans et al. Nature 333(6170):269-72 (1988), and Moran et al, J. Biol. Chem. 266:1250 (1991)). In order to do this, the GPI sequence from Thy-1 can be cassetted 3' of the variable region in place of a transmembrane sequence. «,

Similarly, myristylation sequences can serve as membrane anchoring sequences. It is known that the myristylation of c-src recruits it to the plasma membrane. This is a simple and effective method of membrane localization, given that the first 14 amino acids of the protein are solely responsible for this function: MGSSKSKPKDPSQR (see Cross et al, Mol. Cell. Biol. 4(9): 1834 (1984); Spencer et al. Science 262:1019-1024 (1993), both of which are hereby incorporated by reference). This motif has already been shown to be effective in the localization of reporter genes and can be used to anchor the zeta chain of the TCR. This motif is placed 5' of the variable region in order to localize the construct to the plasma membrane. Other modifications such as palmitoylation can be used to anchor constructs in the plasma membrane; for example, palmitoylation sequences from the G protein-coupled receptor kinase GRK6 sequence (LLQRLFSRQDCCGNCSDSEEELPTRL, with the bold cysteines being palmitolyated; Stoffel et al, J. Biol. Chem 269:27791 (1994)); from rhodopsin (KQFRNCMLTSLCCGKNPLGD; Barnstable et al, J. Mol. Neurosci. 5(3):207 (1994)); and the p21 H- ras 1 protein (LNPPDESGPGCMSCKCVLS; Capon et al. Nature 302:33 (1983)).

In a preferred embodiment, the targeting sequence is a lysozomal targeting sequence, including, for example, a lysosomal degradation sequence such as Lamp-2 (KFERQ; Dice, Ann. N.Y. Acad. Sci. 674:58 (1992); or lysosomal membrane sequences from Lamp-1

(MLIPIAGFFALAGLVLIVLIA YL/GRKRSHAGYQTI. Uthayakumar et al. Cell. Mol. Biol. Res. 41:405 (1995)) or Lamp-2 (LyP// GA4L/4GyL/LVLL/4ΥF/GLKHHHAGYEQF. Konecki et la, Biochem. Biophys. Res. Comm. 205:1-5 (1994), both of which show the transmembrane domains in italics and the cytoplasmic targeting signal underlined).

Alternatively, the targeting sequence may be a mitrochondrial localization sequence, including mitochondrial matrix sequences (e.g. yeast alcohol dehydrogenase III;

MLRTSSLFTRRVQPSLFSRNILRLQST; Schatz, Eur. J. Biochem. 165:1-6 (1987)); mitochondrial inner membrane sequences (yeast cytochrome c oxidase subunit IV; MLSLRQSIRFFKPATRTLCSSRYLL; Schatz, supra); mitochondrial intermembrane space sequences (yeast cytochrome d ; MFSMLSKRWAQRTLSKSFYSTATGAASKSGKLTQKLVTAGVAAAGITASTLLYADSLTAEAMTA; Schatz, supra) or mitochondrial outer membrane sequences (yeast 70 kD outer membrane protein; MKSFITRNKTAILATVAATGTAIGAYYYYNQLQQQQQRGKK; Schatz, supra).

The target sequences may also be endoplasmic reticulum sequences, including the sequences from calreticulin (KDEL; Pelham, Royal Society London Transactions B; 1-10 (1992)) or adenovirus E3/19K protein (LYLSRRSFIDEKKMP; Jackson et al, EMBO J. 9:3153 (1990).

Furthermore, targeting sequences also include peroxisome sequences (for example, the peroxisome matrix sequence from Luciferase; SKL; Keller et al, PNAS USA 4:3264 (1987)); farnesylation sequences (for example, P21 H-ras 1 ; LNPPDESGPGCMSCKCVLS, with the bold cysteine farnesylated; Capon, supra); geranylgeranylation sequences (for example, protein rab-5A; LTEPTQPTRNQCGSN, with the bold cysteines geranylgeranylated; Farnsworth, PNAS USA 91 :11963 (1994)); or destruction sequences (cyclin B1 ; RTALGDIGN; Klotzbucher et al, EMBO J. 1:3053 (1996)). In a preferred embodiment, the targeting sequence is a secretory signal sequence capable of effecting the secretion of the candidate protein. There are a large number of known secretory signal sequences which are placed 5' to the variable peptide region, and are cleaved from the peptide region to effect secretion into the extracellular space. Secretory signal sequences and their transferability to unrelated proteins are well known, e.g, Silhavy, et al. (1985) Microbiol. Rev. 49, 398-418. This is particularly useful to generate a peptide capable of binding to the surface of, or affecting the physiology of, a target cell that is other than the host cell. In this manner, target cells grown in the vicinity of cells caused to express the library of peptides, are bathed in secreted peptide. Target cells exhibiting a physiological change in response to the presence of a peptide, e.g, by the peptide binding to a surface receptor or by being internalized and binding to intracellular targets, and the secreting cells are localized by any of a variety of selection schemes and the peptide causing the effect determined. Exemplary effects include variously that of a designer cytokine (i.e., a stem cell factor capable of causing hematopoietic stem cells to divide and maintain their totipotential), a factor causing cancer cells to undergo spontaneous apoptosis, a factor that binds to the cell surface of target cells and labels them specifically, etc.

Similar to the membrane-anchored embodiment, it is possible that the formation of the NAP conjugate happens after the screening; that is, having the fusion protein secreted means that it may not be available for binding to the nucleic acid. However, this may be done later, with lysis of the cell.

Suitable secretory sequences are known, including signals from IL-2 (MYRMQLLSCIALSLALVTNS; Villinger et al, J. Immunol. 155:3946 (1995)), growth hormone

(MATGSRTSLLLAFGLLCLPWLQEGSAFPT; Roskam et al. Nucleic Acids Res. 7:30 (1979)); preproinsulin (MALWMRLLPLLALLALWGPDPAAAFVN: Bell et al. Nature 284:26 (1980)); and influenza HA protein (MKAKLLVLLYAFVAGDQI; Sekiwawa et al, PNAS 80:3563)), with cleavage between the non-underlined-underlined junction. A particularly preferred secretory signal sequence is the signal leader sequence from the secreted cytokine IL-4, which comprises the first 24 amino acids of IL-4 as follows: MGLTSQLLPPLFFLLACAGNFVHG.

In a preferred embodiment, the fusion partner is a rescue sequence (sometimes also referred to herein as "purification tags" or "retrieval properties"). A rescue sequence is a sequence which may be used to purify or isolate either the candidate protein or the NAP conjugate. Thus, for example, peptide rescue sequences include purification sequences such as the His₆ tag for use with Ni affinity columns and epitope tags for detection, immunoprecipitation or FACS (fluorescence-activated cell sorting). Suitable epitope tags include myc (for use with the commercially available 9E10 antibody), the BSP biotinylation target sequence of the bacterial enzyme BirA, flu tags, lacZ, and GST. Rescue sequences can be utilized on the basis of a binding event, an enzymatic event, a physical property or a chemical property.

Alternatively, the rescue sequence may be a unique oligonucleotide sequence which serves as a probe target site to allow the quick and easy isolation of the construct, via PCR, related techniques, or hybridization.

In a preferred embodiment, the fusion partner is a stability sequence to confer stability to the candidate protein or the nucleic acid encoding it. Thus, for example, peptides may be stabilized by the incorporation of glycines after the initiation methionine (MG or MGGO), for protection of the peptide to ubiquitination as per Varshavsky's N-End Rule, thus conferring long half-life in the cytoplasm. Similarly, two pralines at the C-terminus impart peptides that are largely resistant to carboxypeptidase action. The presence of two glycines prior to the pralines impart both flexibility and prevent structure initiating events in the di-proline to be propagated into the candidate protein structure. Thus, preferred stability sequences are as follows: MG(X)_nGGPP, where X is any amino acid and n is an integer of at least four.

In addition, linker sequences, as defined above, may be used in any configuration as needed.

In a preferred embodiment, the fusion partner is a heterologous protein. Any number of different proteins may be added for a variety of reasons, including for labeling purposes as outlined below. Particularly suitable heterologous proteins for fusing with the candidate proteins include autofluorescent proteins. Preferred fluorescent molecules include but are not limited to green fluorescent protein (GFP; from Aquorea and Renilla species), blue fluorescent protein (BFP), yellow fluorescent protein (YFP), red fluorescent protein (RFP), and enzymes including luciferase and β- galactosidase.

In addition, the fusion partners, including presentation structures, may be modified, randomized, and/or matured to alter the presentation orientation of the randomized expression product. For example, determinants at the base of the loop may be modified to slightly modify the internal loop peptide tertiary structure, which maintaining the randomized amino acid sequence.

In a preferred embodiment, combinations of fusion partners are used. Thus, for example, any number of combinations of presentation structures, targeting sequences, rescue sequences, and stability sequences may be used, with or without linker sequences. Similarly, as discussed herein, the fusion partners may be associated with any component of the expression vectors described herein: they may be directly fused with either the NAM enzyme, the candidate protein, or the EAS, described below, or be separate from these components and contained within the expression vector.

In addition to sequences encoding NAM enzymes and candidate proteins, and the optional fusion partners, the nucleic acids of the invention preferably comprise an enzyme attachment sequence. By "enzyme attachment sequence" or "EAS" herein is meant selected nucleic acid sequences that mediate attachment with NAM enzymes. Such EAS nucleic acid sequences possess the specific sequence or specific chemical or structural configuration that allows for attachment of the NAM enzyme and the Eas. The EAS can comprise DNA or RNA sequences in their natural conformation, or hybrids. EASs also can comprise modified nucleic acid sequences or synthetic sequences inserted into the nucleic acid molecule of the present invention.

As will be appreciated by those in the art, the choice of the EAS will depend on the NAM enzyme, as individual NAM enzymes recognize specific sequences and thus their use is paired. Thus, suitable NAM/EAS pairs are the sequences recognized by Rep proteins (sometimes referred to herein as "Rep EASs") and the Rep proteins, the H-1 recognition sequence and H-1 , etc.

In a preferred embodiment, the EAS is double-stranded. By way of example, a suitable EAS is a double-stranded nucleic acid sequence contajning specific features for interacting with corresponding NAM enzymes. For example, Rep68 and Rep78 recognize an EAS contained within an AAV ITR, the sequence of which is depicted in Figures 47 and 48. In addition, these Rep proteins have been shown to recognize an ITR-like region in human chromosome 19 as well, the sequence of which is shown in Figure 48.

An EAS also can comprise supercoiled DNA with which a topoisomerase interacts and forms covalent intermediate complexes. Alternatively, an EAS is a restriction enzyme site recognized by an altered restriction enzyme capable of forming covalent linkages. Finally, an EAS can comprise an RNA sequence and/or structure with which specific proteins interact and form stable complexes (see, for example, Romaniuk and Uhlenbeck, Biochemistry, 24, 4239-44 (1985)).

The present invention relies on the specific binding of the NAM enzyme to the EAS in order to mediate linkage of the fusion enzyme to the nucleic acid molecule. One of ordinary skill in the art will appreciate that use of an EAS consisting of a small nucleic acid sequence would result in non-specific binding of the NAM enzyme to expression vectors and the host cell genome depending on the frequency that the accessible EAS motif appears in the vector or host genome. Therefore, the EAS of the present invention is preferably comprised of a nucleic acid sequence of sufficient length such that specific fusion protein-coding nucleic acid molecule attachment results. For example, the EAS is preferably greater than five nucleotides in length. More preferably, the EAS is greater than 10 nucleotides in length, e.g, with EASs of at least 15, 20, 25, 30, 35, 40, 45 or 50 nucleotides being preferred.

Moreover, preferably the EAS is present in the host cell genome in a very limited manner, such that at most, only one or two NAM enzymes can bind per genome, e.g. no more than once in a human cell genome. In situations wherein the EAS is present many times within a host cell, e.g, a human cell genome, the probability of fusion proteins encoded by the expression vector attaching to the host cell genome and not the expression vector increases and is therefore undesirable. For instance, the bacteriophage P2 A protein recognizes a relatively short DNA recognition sequence. As such, use of the P2 A protein in mammalian cells would result in protein binding throughout the host genome, and identification of the desired nucleic acid sequence would be difficult. Thus, preferred embodiments exclude the use of P2A as a NAM enzyme.

One of ordinary skill in the art will appreciate that the NAM enzyme used in the present invention or the corresponding EAS can be manipulated in order to increase the stability of the fusion protein- nucleic acid molecule complex. Such manipulations are contemplated herein, so long as the NAM enzyme forms a covalent bond with its corresponding EAS.

Thus, in a preferred embodiment, the nucleic acids of the invention comprise a fusion nucleic acid comprising sequences encoding a NAM enzyme and a candidate protein, and an EAS. These nucleic acids are preferably incorporated into an expression vector; thus providing libraries of expression vectors, sometimes referred to herein as "NAM enzyme expression vectors".

The expression vectors may be either self-replicating extrachromosomal vectors, vectors which integrate into a host genome, or linear nucleic acids that may or may not self-replicate. Thus, specifically included within the definition of expression vectors are linear nucleic acid molecules. Expression vectors thus include plasmids, plasmid-liposome complexes, phage vectors, and viral vectors, e.g, adeno-associated virus (AAV)-based vectors, retroviral vectors, herpes simplex virus (HSV)-based vectors, and adenovirus-based vectors. The nucleic acid molecule and any of these expression vectors can be prepared using standard recombinant DNA techniques described in, for example, Sambrook et al., Molecular Cloning, a Laboratory Manual, 2d edition, Cold Spring Harbor Press, Cold Spring Harbor, N.Y. (1989), and Ausubel et al. Current Protocols in Molecular Biology, Greene Publishing Associates and John Wiley & Sons, New York, N.Y. (1994) Generally, these expression vectors include transcriptional and translational regulatory nucleic acid operably linked to the nucleic acid encoding the NAM protein. The term "control sequences" refers to DNA sequences necessary for the expression of an operably linked coding sequence in a particular host organism. The control sequences that are suitable for prokaryotes, for example, include a promoter, optionally an operator sequence, and a ribosome binding site. Eukaryotic cells are known to utilize promoters, polyadenylation signals, and enhancers.

Nucleic acid is "operably linked" when it is placed into a functional relationship with another nucleic acid sequence. For example, DNA for a presequence or secretory leader is operably linked to DNA for a polypeptide if it is expressed as a preprotein that participates in the secretion of the polypeptide; a promoter or enhancer is operably linked to a coding sequence if it affects the transcription of the sequence; or a ribosome binding site is operably linked to a coding sequence if it is positioned so as to facilitate translation. Generally, "operably linked" means that the DNA sequences being linked are contiguous, and, in the case of a secretory leader, contiguous and in reading phase. However, enhancers do not have to be contiguous. Linking is accomplished by ligation at convenient restriction sites. If such sites do not exist, the synthetic oligonucleotide adaptors or linkers are used in accordance with conventional practice. The transcriptional and translational regulatory nucleic acid will generally be appropriate to the host cell used to express the NAM protein, as will be appreciated by those in the art; for example, transcriptional and translational regulatory nucleic acid sequences from Bacillus are preferably used to express the NAM protein in Bacillus. Numerous types of appropriate expression vectors, and suitable regulatory sequences are known in the art for a variety of host cells.

In general, the transcriptional and translational regulatory sequences may include, but are not limited to, promoter sequences, ribosomal binding sites, transcriptional start and stop sequences, translational start and stop sequences, and enhancer or activator sequences. In a preferred embodiment, the regulatory sequences include a promoter and transcriptional start and stop sequences.

A "promoter" is a nucleic acid sequence that directs the binding of RNA polymerase and thereby promotes RNA synthesis. Promoter sequences include constitutive and inducible promoter sequences. Exemplary constitutive promoters include, but are not limited to, the CMV immediate- early promoter, the RSV long terminal repeat, mouse mammary tumor virus (MMTV) promoters, etc. Suitable inducible promoters include, but are not limited to, the IL-8 promoter, the metallothionine inducible promoter system, the bacterial lacZYA expression system, the tetracycline expression system, and the T7 polymerases system. The promoters may be either naturally occurring promoters, hybrid or synthetic promoters. Hybrid promoters, which combine elements of more than one promoter, are also known in the art, and are useful in the present invention. In addition, the expression vector may comprise additional elements. For example, the expression vector may have two replication systems (e.g. origins of replication), thus allowing it to be maintained in two organisms, for example in mammalian or insect cells for expression and in a prokaryotic host for cloning and amplification. Furthermore, for integrating expression vectors, which are generally not preferred in most embodiments, the expression vector contains at least one sequence homologous to the host cell genome, and preferably two homologous sequences which flank the expression construct. The integrating vector may be directed to a specific locus in the host cell by selecting the appropriate homologous sequence for inclusion in the vector. Constructs for integrating vectors and appropriate selection and screening protocols are well known in the art and are described in e.g, Mansour et al, Cell, 51 :503 (1988) and Murray, Gene Transfer and Expression Protocols, Methods in Molecular Biology, Vol. 7 (Clifton: Humana Press, 1991).

It should be noted that the compositions and methods of the present invention allow for specific chromosomal isolation. For example, since human chromosome 19 contains a Rep-binding sequence (e.g. an EAS), a NAP conjugate will be formed with chromosome 19, when the NAM enzyme is Rep. Cell lysis followed by immunoprecipitation, either using antibodies to the Rep protein itself (e.g. no candidate protein is necessary) or to a fused candidate protein or purification tag, allows the purification of the chromosome. This is a significant advance over current chromosome purification techniques. Thus, by selectively or non-selectively integrating EAS sites into chromosomes, different chromosomes may be purified.

In addition, in a preferred embodiment, the expression vector contains a selection gene to allow the selection of transformed host cells containing the expression vector, and particularly in the case of mammalian cells, ensures the stability of the vector, since cells which do not contain the vector will generally die. Selection genes are well known in the art and will vary with the host cell used. By "selection gene" herein is meant any gene which encodes a gene product that confers resistance to a selection agent. Suitable selection agents include, but are not limited to, neomycin (or its analog G418), blasticidin S, histinidol D, bleomycin, puromycin, hygromycin B, and other drugs.

In a preferred embodiment, the expression vector contains a RNA splicing sequence upstream or downstream of the gene to be expressed in order to increase the level of gene expression. See Barret et al. Nucleic Acids Res. 1991 ; Groos et al, Mol. Cell. Biol. 1987; and Budiman et al, Mol. Cell. Biol. 1988.

One expression vector system is a retroviral vector system such as is generally described in Mann et al. Cell, 33:153-9 (1993); Pear et al, Proc. Natl. Acad. Sci. U.S.A., 90(18):8392-6 (1993); Kitamura et al, Proc. Natl. Acad. Sci. U.S.A., 92:9146-50 (1995); Kinsella et al. Human Gene Therapy, 7:1405-13; Hofmann et al,Proc. Natl. Acad. Sci. U.S.A., 93:5185-90; Choate et al. Human Gene Therapy, 7:2247 (1996); PCT/US97/01019 and PCT/US97/01048, and references cited therein, all of which are hereby expressly incorporated by reference.

The fusion proteins of the present invention are produced by culturing a host cell transformed with nucleic acid, preferably an expression vector as outlined herein, under the appropriate conditions to induce or cause expression of the fusion protein. The conditions appropriate for fusion protein expression will vary with the choice of the expression vector and the host cell, and will be easily ascertained by one skilled in the art through routine experimentation. For example, the use of constitutive promoters in the expression vector will require optimizing the growth and proliferation of the host cell, while the use of an inducible promoter requires the appropriate growth conditions for induction. In addition, in some embodiments, the timing of the harvest is important. For example, the baculovirai systems used in insect cell expression are lytic viruses, and thus harvest time selection can be crucial for product yield.

The choice of the host cell will depend, in part, on the assay to be run; e.g. in vitro systems may allow the use of any number of procaryofic or eucaryotic organisms, while ex vivo systems preferably utilize animal cells, particularly mammalian cells with a special emphasis on human cells. Thus, appropriate host cells include yeast, bacteria, archaebacteria, fungi, and insect and animal cells, including mammalian cells and particularly human cells. The host cells may be native cells, primary cells, including those isolated from diseased tissues or organisms, cell lines (again those orginating with diseased tissues), genetically altered cells, etc. Of particular interest are Drosophila melanogaster cells, Saccharomyces cerevisiae and other yeasts, E. coli, Bacillus subtilis, SF9 cells, C129 cells, 293 cells, Neurospora, BHK, CHO, COS, and HeLa cells, fibroblasts, Schwanoma cell lines, etc. See the ATCC cell line catalog, hereby expressly incorporated by reference.

In a preferred embodiment, the fusion proteins are expressed in mammalian cells. Mammalian expression systems are also known in the art, and include retroviral and adenoviral systems. A mammalian promoter is any DNA sequence capable of binding mammalian RNA polymerase and initiating the downstream (3') transcription of a coding sequence for a fusion protein into mRNA. A promoter will have a transcription initiating region, which is usually placed proximal to the 5' end of the coding sequence, and a TATA box, using a located 25-30 base pairs upstream of the transcription initiation site. The TATA box is thought to direct RNA polymerase II to begin RNA synthesis at the correct site. A mammalian promoter will also contain an upstream promoter element (enhancer element), typically located within 100 to 200 base pairs upstream of the TATA box. An upstream promoter element determines the rate at which transcription is initiated and can act in either orientation. Of particular use as mammalian promoters are the promoters from mammalian viral genes, since the viral genes are often highly expressed and have a broad host range. Examples include the SV40 early promoter, mouse mammary tumor virus LTR promoter, adenovirus major late promoter, herpes simplex virus promoter, and the CMV promoter.

Typically, transcription termination and polyadenylation sequences recognized by mammalian cells are regulatory regions located 3' to the translation stop codon and thus, together with the promoter elements, flank the coding sequence. The 3' terminus of the mature mRNA is formed by site-specific post-translational cleavage and polyadenylation. Examples of transcription terminator and polyadenlytion signals include those derived form SV40.

The methods of introducing exogenous nucleic acid into mammalian hosts, as well as other hosts, is well known in the art, and will vary with the host cell used. Techniques include dextran-mediated transfection, calcium phosphate precipitation, polybrene mediated transfection, protoplast fusion, electroporation, viral infection, encapsulation of the polynucleotide(s) in liposomes, and direct microinjection of the DNA into nuclei.

In a preferred embodiment, NAM fusions are expressed in bacterial systems. Bacterial expression systems are well known in the art.

A suitable bacterial promoter is any nucleic acid sequence capable of binding bacterial RNA polymerase and initiating the downstream (3') transcription of the coding sequence of the fusion into mRNA. A bacterial promoter has a transcription initiation region which is usually placed proximal to the 5' end of the coding sequence. This transcription initiation region typically includes an RNA polymerase binding site and a transcription initiation site. Sequences encoding metabolic pathway enzymes provide particularly useful promoter sequences. Examples include promoter sequences derived from sugar metabolizing enzymes, such as galactose, lactose and maltose, and sequences derived from biosynthetic enzymes such as tryptophan. Promoters from bacteriophage may also be used and are known in the art. In addition, synthetic promoters and hybrid promoters are also useful; for example, the fac promoter is a hybrid of the trp and lac promoter sequences. Furthermore, a bacterial promoter can include naturally occurring promoters of non-bacterial origin that have the ability to bind bacterial RNA polymerase and initiate transcription.

In addition to a functioning promoter sequence, an efficient ribosome binding site is desirable. In E. coli, the ribosome binding site is called the Shine-Delgarno (SD) sequence and includes an initiation codon and a sequence 3-9 nucleotides in length located 3 - 11 nucleotides upstream of the initiation codon. The expression vector may also include a signal peptide sequence that provides for secretion of the fusion proteins in bacteria or other cells. The signal sequence typically encodes a signal peptide comprised of hydrophobic amino acids which direct the secretion of the protein from the cell, as is well known in the art. The protein is either secreted into the growth media (gram-positive bacteria) or into the periplasmic space, located between the inner and outer membrane of the cell (gram-negative bacteria).

The bacterial expression vector may also include a selectable marker gene to allow for the selection of bacterial strains that have been transformed. Suitable selection genes include genes which render the bacteria resistant to drugs such as ampicillin, chloramphenicol, erythromycin, kanamycin, neomycin and tetracycline. Selectable markers also include biosynthetic genes, such as those in the histidine, tryptophan and leucine biosynthetic pathways.

These components are assembled into expression vectors. Expression vectors for bacteria are well known in the art, and include vectors for Bacillus subtilis, E. coli, Streptococcus cremoris, and Streptococcus lividans, among others.

The bacterial expression vectors are transformed into bacterial host cells using techniques well known in the art, such as calcium chloride treatment, electroporation, and others.

In one embodiment, NAM fusion proteins are produced in insect cells such as Sf9 cells. Expression vectors for the transformation of insect cells, and in particular, baculovirus-based expression vectors, are well known in the art and are described e.g, in O'Reilly et al, Baculovirus Expression Vectors: A Laboratory Manual (New York: Oxford University Press, 1994).

In a preferred embodiment, NAM fusion proteins are produced in yeast cells. Yeast expression systems are well known in the art, and include expression vectors for Saccήaromyces cerevisiae, Candida albicans and C. maltosa, Hansenula polymorpha, Kluyveromyces fragilis and K. lactis, Pichia guillθήmondii and P. pastoris, Schizosaccharomyces pombe, and Yarrowia lipolytica. Preferred promoter sequences for expression in yeast include the inducible GAL1 , 10 promoter, the promoters from alcohol dehydrogenase, enolase, glucokinase, glucose-6-phosphate isomerase, glyceraldehyde- 3-phosphate-dehydrogenase, hexokinase, phosphofructokinase, 3-phosphoglycerate mutase, pyruvate kinase, and the acid phosphatase gene. Yeast selectable markers include ADE2, HIS4, LEU2, TRP1 , and ALG7, which confers resistance to tunicamycin; the neomycin phosphotransferase gene, which confers resistance to G418; and the CUP1 gene, which allows yeast to grow in the presence of copper ions. One benefit of using yeast cells is the ability to propagate the cells comprising the vectors, thus generating clonal populations. Preferred expression vectors are shown in Figure 49.

In addition to the components outlined herein, including NAM enzyme-candidate protein fusions, EASs, linkers, fusion partners, etc, the expression vectors may comprise a number of additional components, including, selection genes as outlined herein (particularly including growth-promoting or growth-inhibiting functions), activatible elements, recombination signals (e.g. ere and lox sites) and labels.

In a preferred embodiment, a component of the system is a labeling component. Again, as for the fusion partners of the invention, the label may be fused to one or more of the other components, for example to the NAM fusion protein, in the case where the NAM enzyme and the candidate protein remain attached, or to either component, in the case where scission occurs, or separately, under its own promoter. In addition, as is further described below, other components of the assay systems may be labeled.

Labels can be either direct or indirect detection labels, sometimes referred to herein as "primary" and "secondary" labels. By "detection label" or "detectable label" herein is meant a moiety that allows detection. This may be a primary label or a secondary label. Accordingly, detection labels may be primary labels (i.e. directly detectable) or secondary labels (indirectly detectable).

In general, labels fall into four classes: a) isotopic labels, which may be radioactive or heavy isotopes; b) magnetic, electrical, thermal labels; c) colored or luminescent dyes or moieties; and d) binding partners. Labels can also include enzymes (horseradish peroxidase, etc.) and magnetic particles. In a preferred embodiment, the detection label is a primary label. A primary label is one that can be directly detected, such as a fluorophore.

Preferred labels include chromophores or phosphors but are preferably fluorescent dyes or moieties. Fluorophores can be either "small molecule" fluors, or proteinaceous fluors. In a preferred embodiment, particularly for labeling of target molecules, as described below, suitable dyes for use in the invention include, but are not limited to, fluorescent lanthanide complexes, including those of Europium and Terbium, fluorescein, rhodamine, tetramethylrhodamine, eosin, erythrosin, coumarin, methyl-coumarins, quantum dots (also referred to as "nanocrystals"), pyrene, Malacite green, stilbene, Lucifer Yellow, Cascade Blue™, Texas Red, Cy dyes (Cy3, Cy5, etc.), alexa dyes, phycoerythin, bodipy, and others described in the 6th Edition of the Molecular Probes Handbook by Richard P. Haugland, hereby expressly incorporated by reference.

10 In a preferred embodiment, for example when the label is attached to the fusion polypeptide or is to be expressed as a component of the expression vector, proteinaceous fluores are used. Suitable autofluorescent proteins include, but are not limited to, the green fluorescent protein (GFP) from Aequorea and variants thereof; including, but not limited to, GFP, (Chalfie, et al. Science 263(5148):802-805 (1994)); enhanced GFP (EGFP; Clontech - Genbank Accession Number U55762 )), blue fluorescent protein (BFP; Quantum Biotechnologies, Inc. 1801 de Maisonneuve Blvd. West, 8th Floor, Montreal (Quebec) Canada H3H 1J9; Stauber, R. H. Biotechniques 24(3):462-471 (1998); Heim, R. and Tsien, R. Y. Curr. Biol. 6:178-182 (1996)), and enhanced yellow fluorescent protein (EYFP; Clontech Laboratories, Inc., 1020 East Meadow Circle, Palo Alto, CA 94303). In addition, there are recent reports of autofluorescent proteins from Renilla species. See WO 92/15673; WO 95/07463; WO 98/14605; WO 98/26277; WO 99/49019; U.S. patent 5,292,658; U.S patent 5,418,155; U.S. patent 5,683,888; U.S. patent 5,741 ,668; U.S. patent 5,777,079; U.S. patent 5,804,387; U.S. patent 5,874,304; U.S patent 5,876,995; and U.S. patent 5,925,558; all of which are expressly incorporated herein by reference.

In a preferred embodiment, the label protein is Aequorea green fluorescent protein or one of its variants; see Cody et al. Biochemistry 32:1212-1218 (1993); and Inouye and Tsuji, FEBS Lett. 341 :277-280 (1994), both of which are expressly incorporated by reference herein.

In a preferred embodiment, a secondary detectable label is used. A secondary label is one that is indirectly detected; for example, a secondary label can bind or react with a primary label for detection, can act on an additional product to generate a primary label (e.g. enzymes), or may allow the separation of the compound comprising the secondary label from unlabeled materials, etc. Secondary labels include, but are not limited to, one of a binding partner pair; chemically modifiable moieties; enzymes such as horseradish peroxidase, alkaline phosphatases, lucifierases, etc; and cell surface markers, etc.

In a preferred embodiment, the secondary label is a binding partner pair. For example, the label may be a hapten or antigen, which will bind its binding partner. In a preferred embodiment, the binding partner can be attached to a solid support to allow' separation of components containing the label and those that do not. For example, suitable binding partner pairs include, but are not limited to: antigens (such as proteins (including peptides)) and antibodies (including fragments thereof (FAbs, etc.)); proteins and small molecules, including biotin/streptavidin; enzymes and substrates or inhibitors; other protein-protein interacting pairs; receptor-ligands; and carbohydrates and their binding partners. Nucleic acid - nucleic acid binding proteins pairs are also useful. In general, the smaller of the pair is attached to the system component for incorporation into the assay, although this is not required in all embodiments. Preferred binding partner pairs include, but are not limited to, biotin (or imino-biotin) and streptavidin, digeoxinin and Abs, etc.

In a preferred embodiment, the binding partner pair comprises a primary detection label (for example, attached to the assay component) and an antibody that will specifically bind to the primary detection label. By "specifically bind" herein is meant that the partners bind with specificity sufficient to differentiate between the pair and other components or contaminants of the system. The binding should be sufficient to remain bound under the conditions of the assay, including wash steps to remove non-specific binding. In some embodiments, the dissociation constants of the pair will be less than about lO^-IO^-6 M^"1, with less than about 10^"s-10^"9 M^'1, being preferred and less than about 10^"7- 10^"9 M^"1 being particularly preferred.

In a preferred embodiment, the secondary label is a chemically modifiable moiety. In this embodiment, labels comprising reactive functional groups are incorporated into the assay component. The functional group can then be subsequently labeled with a primary label. Suitable functional groups include, but are not limited to, amino groups, carboxy groups, maleimide groups, oxo groups and thiol groups, with amino groups and thiol groups being particularly preferred. For example, primary labels containing amino groups can be attached to secondary labels comprising amino groups, for example using linkers as are known in the art; for example, homo-or hetero-bifunctional linkers as are well known (see 1994 Pierce Chemical Company catalog, technical section on cross-linkers, pages 155-200, incorporated herein by reference).

It can be advantageous to construct the expression vector to provide further options to control attachment of the fusion enzyme to the EAS. For example, the EAS can be introduced into the nucleic acid molecule as two non-functional halves that are brought together following site-specific homologous recombination, such as that mediated by cre-lox recombination, to form a functional EAS. Likewise, the referenced cre-lox consideration could also be used to control the formation of a functional fusion enzyme. The control of cre-lox recombination is preferably mediated by introducing the recombinase gene under the control of an inducible promoter into the expression system, whether on the same nucleic acid molecule or on another expression vector.

In general, once the expression vectors of the invention are made, they follow one of two fates: they are introduced into cell-free translation systems, to create libraries of nucleic acid/protein (NAP) conjugates that are assayed in vitro, or, preferably they are introduced into host cells where the NAP conjugates are formed; the cells may be optionally lysed and assayed accordingly. In a preferred embodiment, the expression vectors are made and introduced into cell-free systems for translation, followed by the attachment of the NAP enzyme to the EAS, forming a nucleic acid/protein (NAP) conjugate. By "nucleic acid/protein conjugate" or "NAP conjugate" herein is meant a covalent attachment between the NAP enzyme and the EAS, such that the expression vector comprising the EAS is covalently attached to the NAP enzyme. Suitable cell free translation systems are known in the art. Once made, the NAP conjugates are used in assays as outlined below.

In a preferred embodiment, the expression vectors of the invention are introduced into host cells as outlined herein. By "introduced into " or grammatical equivalents herein is meant that the nucleic acids enter the cells in a manner suitable for subsequent expression of the nucleic acid. The method of introduction is largely dictated by the targeted cell type, discussed below. Exemplary methods include CaP0₄ precipitation, liposome fusion, lipofectin®, electroporation, viral infection, gene guns, etc. The candidate nucleic acids may stably integrate into the genome of the host cell (for example, with retroviral introduction, outlined herein) or may exist either transiently or stably in the cytoplasm (i.e. through the use of traditional plasmids, utilizing standard regulatory sequences, selection markers, etc.). Suitable host cells are outlined above, with eucaryotic, mammalian and human cells all preferred.

Many previously described methods involve peptide library expression in bacterial cells. Yet, it is understood in the art that translational machinery such as codon preference, protein folding machinery, and post-translational modifications of, for example, mammalian peptides, are unachievable or altered in bacterial cells, if such modifications occur at all. Peptide library screening in bacterial cells often involves expression of short amino acid sequences, which can not imitate a protein in its natural configuration. Screening of these small, sub-part sequences cannot effectively determine the function of a native protein in that the requirements for, for instance, recognition of a small ligand for its receptor, are easily satisfied by small sequences without native conformation. The complexities of tertiary structure are not accounted for, thereby easing the requirements for binding. One advantage of the present invention is the ability to express and screen unknown peptides in their native environment and in their native protein conformation. The covalent attachment of the fusion enzyme to its corresponding expression vector allows screening of peptides in organisms other than bacteria. Once introduced into a eukaryotic host cell, the nucleic acid molecule is transported into the nucleus where replication and transcription occurs. The transcription product is transferred to the cytoplasm for translation and post-translational modifications. However, the produced peptide and corresponding nucleic acid molecule must meet in order for attachment to occur, which is hindered by the compartmentalization of eukaryotic cells. NAM enzyme-EAS recognition can occur in four ways, which are merely exemplary and do not limit the present invention in any way. First, the host cells can be allowed to undergo one round of division, during which the nuclear envelope breaks down. Second, the host cells can be infected with viruses that perforate the nuclear envelope. Third, specific nuclear localization or transporting signals can be introduced into the fusion enzyme. Finally, host cell organelles can be disrupted using methods known in the art.

The end result of the above-described approaches is the transfer of the expression vector into the same environment as the fusion enzyme. The non-covalent interaction between a DNA binding protein and attachment site of previously described expression libraries would not survive the procedures required to allow linkage of the fusion protein to its expression vector in eukaryotic cells. Other DNA-protein linkages described in the art, such as those using the bacterial P2 A DNA binding peptide, require the binding peptide to remain in direct contact with its coding DNA in order for binding to occur, i.e., translation must occur proximal to the coding sequence (see, for example, Lindahl, Virology, 42, 522-533 (1970)). Such linkages are only achievable in prokaryotic systems and cannot be produced in eukaryotic cells.

Once the NAM enzyme expression vectors have been introduced into the host cells, the cells are optionally lysed. Cell lysis is accomplished by any suitable technique, such as any of a variety of techniques known in the art (see, for example, Sambrook et al. Molecular Cloning, a Laboratory Manual, 2d edition, Cold Spring Harbor Press, Cold Spring Harbor, N.Y. (1989), and Ausubel et al. Current Protocols in Molecular Biology, Greene Publishing Associates and John Wiley & Sons, New York, N.Y. (1994), hereby expressly incorporated by reference). Most methods of cell lysis involve exposure to chemical, enzymatic, or mechanical stress. Although the attachment of the fusion enzyme to its coding nucleic acid molecule is a covalent linkage, and can therefore withstand more varied conditions than non-covalent bonds, care should be taken to ensure that the fusion enzyme- nucleic acid molecule complexes remain intact, i.e., the fusion enzyme remains associated with the expression vector.

In a preferred embodiment, the NAP conjugate may be purified or isolated after lysis of the cells. Ideally, the lysate containing the fusion protein-nucleic acid molecule complexes is separated from a majority of the resulting cellular debris in order to facilitate interaction with the target. For example, the NAP conjugate may be isolated or purified away from some or all of the proteins and compounds with which it is normally found after expression, and thus may be substantially pure. For example, an isolated NAP conjugate is unaccompanied by at least some of the material with which it is normally associated in its natural (unpurified) state, preferably constituting at least about 0.5%, more preferably at least about 5% by weight of the total protein in a given sample. A substantially pure protein comprises at least about 75% by weight of the total protein, with at least about 80% being preferred, and at least about 90% being particularly preferred. NAP conjugates may be isolated or purified in a variety of ways known to those skilled in the art depending on what other components are present in the sample. Standard purification methods include electrophoretic, molecular, immunological and chromatographic techniques, including ion exchange, hydrophobic, affinity, and reverse-phase HPLC chromatography, gel filtration, and chromatofocusing. Ultrafiltration and diafiltration techniques, in conjunction with protein concentration, are also useful. For general guidance in suitable purification techniques, see Scopes, R, Protein Purification, Springer-Verlag, NY (1982). The degree of purification necessary will vary depending on the use of the NAP conjugate. In some instances no purification will be necessary.

Thus, the invention provides for NAP conjugates that are either in solution, optionally purified or isolated, or contained within host cells. Once expressed and purified if necessary, the NAP conjugates are useful in a number of applications, including in vitro, ex vivo and in vivo screening techniques. In vitro techniques include assays that are cell free, assays within cells, and assays within animals. One of ordinary skill in the art will appreciate that both in vitro and ex vivo embodiments of the present inventive method have utility in a number of fields of study. For example, the present invention has utility in diagnostic assays and can be employed for research in numerous disciplines, including, but not limited to, clinical pharmacology, functional genomics, pharamcogenomics, agricultural chemicals, environmental safety assessment, chemical sensor, nutrient biology, cosmetic research, and enzymology.

In a preferred embodiment, the NAP library is derived computationally and a plurality, or all, of the members are tested for biological activity, including, but not limited to, enzymatic activity, binding activity, biological activity specific to the scaffold protein, stability assays (including thermal and buffer stability), etc. Library members with improved or desirable activities can then be sequenced. In some embodiments, these activity assays are run in microtiter plates for instance, with pools of NAP conjugates; if improved activities are seen, the NAP conjugates can be deconvoluted as required, this may also be done reiteratively, as for most of the embodiments herein.

In a preferred embodiment, the NAP conjugates are used in in vitro screening techniques. In this embodiment, the NAP conjugates are made and screened for binding and/or modulation of bioactivites of target molecules. One of the strengths of the present invention is to allow the identification of target molecules that bind to the candidate proteins. As is more fully outlined below, this has a wide variety of applications, including elucidating members of a signalling pathway, elucidating the binding partners of a drug or other compound of interest, etc.

Thus, the NAP conjugates are used in assays with target molecules. By "target molecules" or "test molecules" herein is meant molecules that are to be tested for binding to the candidate proteins of the NAP conjugates. The test molecules in this embodiment can include a wide variety of things, including libraries of proteins, nucleic acids, lipids, carbohydrates, drugs and other small molecules, etc. In some embodiments, the target analytes comprise sets of proteins comprising different SNPs, to facilitate the identification of the role and function of different SNPs within one or more proteins.

Test molecules encompass numerous chemical classes, though typically they are organic molecules, preferably small organic compounds having a molecular weight of more than 100 and less than about 2,500 daltons. Test molecules comprise functional groups necessary for structural interaction with proteins, particularly hydrogen bonding, and typically include at least an amine, carbonyl, hydroxyl or carboxyl group, preferably at least two of the functional chemical groups. The test molecules often comprise cyclical carbon or heterocyclic structures and/or aromatic or polyaromatic structures substituted with one or more of the above functional groups. Test molecules are also found among biomolecules including peptides, saccharides, fatty acids, steroids, purines, pyrimidines, derivatives, structural analogs or combinations thereof. Particularly preferred are proteins, candidate drugs and other small molecules, and known drugs.

Test molecules are obtained from a wide variety of sources including libraries of synthetic or natural compounds. For example, numerous means are available for random and directed synthesis of a wide variety of organic compounds and biomolecules, including expression of randomized oligonucleotides. Alternatively, libraries of natural compounds in the form of bacterial, fungal, plant and animal extracts are available or readily produced. Additionally, natural or synthetically produced libraries and compounds are readily modified through conventional chemical, physical and biochemical means. Known pharmacological agents may be subjected to directed or random chemical modifications, such as acylation, alkylation, esterification, amidification to produce structural analogs.

Suitable test molecules include organic and inorganic molecules, including biomolecules. In a preferred embodiment, the test molecule may be an environmental pollutant (including pesticides, insecticides, toxins, etc.); a chemical (including solvents, polymers, organic materials, etc.); therapeutic molecules (including therapeutic and abused drugs, antibiotics, etc.); biomolecules (including hormones, cytokines, proteins, lipids, carbohydrates, cellular membrane antigens and receptors (neural, hormonal, nutrient, and cell surface receptors) or their ligands, etc); whole cells (including procaryofic (such as pathogenic bacteria) and eukaryotic cells, including mammalian tumor cells); viruses (including retroviruses, herpesviruses, adenoviruses, lentiviruses, etc.); and spores; etc. Particularly preferred analytes are environmental pollutants; nucleic acids; proteins (including enzymes, antibodies, antigens, growth factors, cytokines, etc); therapeutic and abused drugs; cells; and viruses. Thus, suitable target molecules encompass a wide variety of different classes, including, but not limited to, cells, viruses, proteins (particularly including enzymes, cell-surface receptors, ion channels, and transcription factors, and proteins produced by disease-causing genes or expressed during disease states), carbohydrates, fatty acids and lipids, nucleic acids, chemical moieties such as small molecules, agricultural chemicals, drugs, ions (particularly metal ions), polymers and other biomaterials. Thus for example, binding to polymers (both naturally occurring and synthetic), or other biomaterials, may be done using the methods and compositions of the invention.

In a preferred embodiment, the test molecules are proteins as defined above. In a preferred embodiment, the test molecules are naturally occurring proteins or fragments of naturally occurring proteins. Thus, for example, cellular extracts containing proteins, or random or directed digests of proteinaceous cellular extracts, may be used. In this way libraries of procaryofic and eukaryotic proteins may be made for screening in the systems described herein. Particularly preferred in this embodiment are libraries of bacterial, fungal, viral, and mammalian proteins, with the latter being preferred, and human proteins being especially preferred.

Suitable protein test molecules include, but are not limited to, (1) immunoglobulins, particularly IgEs, IgGs and IgMs, and particularly therapeutically or diagnostically relevant antibodies, including but not limited to, for example, antibodies to human albumin, apolipoproteins (including apolipoprotein E), human chorionic gonadotropin, cortisol, α-fetoprotein, thyroxin, thyroid stimulating hormone (TSH), antithrombin, antibodies to pharmaceuticals (including antieptileptic drugs (phenytoin, primidone, carbariezepin, ethosuximide, valproic acid, and phenobarbitol), cardioactive drugs (digoxin, lidocaine, procainamide, and disopyramide), bronchodilators ( theophylline), antibiotics (chloramphenicol, sulfonamides), antidepressants, immunosuppresants, abused drugs (amphetamine, methamphetamine, cannabinoids, cocaine and opiates) and antibodies to any number of viruses (including orthomyxoviruses, (e.g. influenza virus), paramyxoviruses (e.g respiratory syncytial virus, mumps virus, measles virus), adenoviruses, rhinoviruses, coronaviruses, reoviruses, togaviruses (e.g. rubella virus), parvoviruses, poxviruses (e.g. variola virus, vaccinia virus), enteroviruses (e.g. poliovirus, coxsackievirus), hepatitis viruses (including A, B and C), herpesviruses (e.g. Herpes simplex virus, varicella-zoster virus, cytomegalovirus, Epstein-Barr virus), rotaviruses, Norwalk viruses, hantavirus, arenavirus, rhabdovirus (e.g. rabies virus), retroviruses (including HIV, HTLV-I and -II), papovaviruses (e.g. papillomavirus), polyomaviruses, and picornaviruses, and the like), and bacteria (including a wide variety of pathogenic and non-pathogenic prokaryotes of interest including Bacillus; Vibrio, e.g. V. cholerae; Escherichia, e.g. Enterotoxigenic E. coli, Shigella, e.g. S. dysenteriae; Salmonella, e.g. S. typhi; Mycobacterium e.g. M. tuberculosis, M. leprae; Clostridium, e.g. C. botulinum, C. tetani, C. difficile, C.perfringens; Cornyebacterium, e.g. C. diphtheriae; Streptococcus, S. pyogenes, S. pneumoniae; Staphylococcus, e.g. S. aureus; Haemophilus, e.g. H. influenzae; Neisseria, e.g. N. meningitidis, N. gonorrhoeae; Yersinia, e.g. G. lambliaY. pestis, Pseudomonas, e.g. P. aeruginosa, P. putida; Chlamydia, e.g. C. trachomatis; Bordetella, e.g. B. pertussis; Treponema, e.g. T. palladium; and the like); (2) enzymes (and other proteins), including but not limited to, enzymes used as indicators of or treatment for heart disease, including creatine kinase, lactate dehydrogenase, aspartate amino transferase, troponin T, myoglobin, fibrinogen, cholesterol, triglycerides, thrombin, tissue plasminogen activator (tPA); pancreatic disease indicators including amylase, lipase, chymotrypsin and trypsin; liver function enzymes and proteins including cholinesterase, bilirubin, and alkaline phosphotase; aldolase, prostatic acid phosphatase, terminal deoxynucleotidyl transferase, and bacterial and viral enzymes such as HIV protease; (3) hormones and cytokines (many of which serve as ligands for cellular receptors) such as erythropoietin (EPO), thrombopoiefin (TPO), the interieukins (including IL-1 through IL-17), insulin, insulin-like growth factors (including IGF-1 and -2), epidermal growth factor (EGF), transforming growth factors (including TGF-α and TGF-β), human growth hormone, transferrin, epidermal growth factor (EGF), low density lipoprotein, high density lipoprotein, leptin, VEGF, PDGF, ciliary neurotrophic factor, prolactin, adrenocorticotropic hormone (ACTH), calcitonin, human chorionic gonadotropin, cotrisol, estradiol, follicle stimulating hormone (FSH), thyroid-stimulating hormone (TSH), leutinzing hormone (LH), progeterone, testosterone, ; and (4) other proteins (including α-fetoprotein, carcinoembryonic antigen CEA.

In addition, any of the biomolecules for which antibodies are tested may be tested directly as well; that is, the virus or bacterial cells, therapeutic and abused drugs, etc, may be the test molecules. In addition, one or more of the proteins listed above can be used as a scaffold protein for a candidate protein within a NAP conjugate.

In a preferred embodiment, the test molecules are peptides of from about 5 to about 30 amino acids, with from about 5 to about 20 amino acids being preferred, and from about 7 to about 15 being particularly preferred. The peptides may be digests of naturally occurring proteins as is outlined above, random peptides, or "biased" random peptides. By "randomized" or grammatical equivalents herein is meant that each nucleic acid and peptide consists of essentially random nucleotides and amino acids, respectively. Since generally these random peptides (or nucleic acids, discussed below) are chemically synthesized, they may incorporate any nucleotide or amino acid at any position. The synthetic process can be designed to generate randomized proteins or nucleic acids, to allow the formation of all or most of the possible combinations over the length of the sequence, thus forming a library of randomized test molecules.

In one embodiment, the library is fully randomized, with no sequence preferences or constants at any position. In a preferred embodiment, the library is biased. That is, some positions within the sequence are either held constant, or are selected from a limited number of possibilities. For example, in a preferred embodiment, the nucleotides or amino acid residues are randomized within a defined class, for example, of hydrophobic amino acids, hydrophilic residues, sterically biased (either small or large) residues, towards the creation of cysteines, for cross-linking, pralines for SH-3 domains, serines, threonines, tyrosines or histidines for phosphorylation sites, etc, or to purines, etc.

In a preferred embodiment, the test molecules are derived from cDNA libraries. The cDNA libraries can be derived from any number of different cells, particularly those outlined for host cells herein, and include cDNA libraries generated from eucaryotic and procaryofic cells, viruses, cells infected with viruses or other pathogens, genetically altered cells, etc. Preferred embodiments, as outlined below, include cDNA libraries made from different individuals, such as different patients, particularly human patients. The cDNA libraries may be complete libraries or partial libraries. Furthermore, the library of test molecules can be derived from a single cDNA source or multiple sources; that is, cDNA from multiple cell types or multiple individuals or multiple pathogens can be combined in a screen. The cDNA library may utilize entire cDNA constructs or fractionated constructs, including random or targeted fractionation. Suitable fractionation techniques include enzymatic, chemical or mechanical fractionation.

In a preferred embodiment, the test molecules are derived from genomic libraries. As above, the genomic libraries can be derived from any number of different cells, particularly those outlined for host cells herein, and include genomic libraries generated from eucaryotic and procaryotic cells, viruses, cells infected with viruses or other pathogens, genetically altered cells, etc. Preferred embodiments, as outlined below, include genomic libraries made from different individuals, such as different patients, particularly human patients. The genomic libraries may be complete libraries or partial libraries. Furthermore, the library of test molecules can be derived from a single genomic source or multiple sources; that is, genomic DNA from multiple cell types or multiple individuals or multiple pathogens can be combined in a screen. The genomic library may utilize entire genomic constructs or fractionated constructs, including random or targeted fractionation. Suitable fractionation techniques include enzymatic, chemical or mechanical fractionation.

Suitable prokaryotic cells include, but are not limited to, bacteria such as E. coli, Bacillus species, and the extremophile bacteria such as thermophiles, etc.

Suitable eukaryotic cells include, but are not limited to, fungi such as yeast and filamentous fungi, including species of Aspergillus, Thchoderma, and Neurospora; plant cells including those of corn, sorghum, tobacco, canola, soybean, cotton, tomato, potato, alfalfa, sunflower, etc.; and animal cells, including fish, birds and mammals. Suitable fish cells include, but are not limited to, those from species of salmon, trout, tilapia, tuna, carp, flounder, halibut, swordfish, cod and zebrafish. Suitable bird cells include, but are not limited to, those of chickens, ducks, quail, pheasants and turkeys, and other jungle foul or game birds. Suitable mammalian cells include, but are not limited to, cells from horses, cows, buffalo, deer, sheep, rabbits, rodents such as mice, rats, hamsters and guinea pigs, goats, pigs, primates, marine mammals including dolphins and whales, as well as cell lines, such as human cell lines of any tissue or stem cell type, and stem cells, including pluripotent and non- pluripotent, and non-human zygotes.

As is described herein, cell types implicated in a wide variety of disease conditions are particularly useful to identify interesting protein-protein interactions. Accordingly, suitable eukaryotic cell types include, but are not limited to, tumor cells of all types (particularly melanoma, myeloid leukemia, carcinomas of the lung, breast, ovaries, colon, kidney, prostate, pancreas and testes), cardiomyocytes, endothelial cells, epithelial cells, lymphocytes (T-cell and B cell) , mast cells, eosinophils, vascular intimal cells, hepatocytes, leukocytes including mononuclear leukocytes, stem cells such as haemopoetic, neural, skin, lung, kidney, liver and myocyte stem cells (for use in screening for differentiation and de-differentiation factors), osteoclasts, chondrocytes and other connective tissue cells, keratinocytes, melanocytes, liver cells, kidney cells, and adipocytes. Suitable cells also include known research cells, including, but not limited to, Jurkat T cells, NIH3T3 cells, CHO, COS, etc. See the ATCC cell line catalog, hereby expressly incorporated by reference.

In one embodiment, the cells may be genetically engineered, that is, contain exogenous nucleic acid.

In a preferred embodiment, the test molecules are nucleic acids as defined above. As described above generally for proteins, nucleic acid test molecules may be naturally occurring nucleic acids, random nucleic acids, or "biased" random nucleic acids.

In addition, the test molecule libraries may also be subsequently mutated using known techniques (exposure to mutagens, error-prone PCR, error-prone transcription, combinatorial splicing (e.g. cre-lox recombination). In this way libraries of procaryotic and eukaryotic proteins may be made for screening in the systems described herein. Particularly preferred in this embodiment are libraries of bacterial, fungal, viral, and mammalian proteins, with the latter being preferred, and human proteins being especially preferred.

The test molecules may vary in size. In the case of cDNA or genomic libraries, the proteins may range from 20 or 30 amino acids to thousands, with from about 50 to 1000 being preferred and from 100 to 500 being especially preferred. When the test molecules are peptides, the peptides are from about 3 to about 50 amino acids, with from about 5 to about 20 amino acids being preferred, and from about 7 to about 15 being particularly preferred. The peptides may be digests of naturally occurring proteins as is outlined above, random peptides, or "biased" random peptides. By "randomized" or grammatical equivalents herein is meant that each nucleic acid and peptide consists of essentially random nucleotides and amino acids, respectively. Since generally these random peptides (or nucleic acids, discussed below) are chemically synthesized, they may incorporate any nucleotide or amino acid at any position. The synthetic process can be designed to generate randomized proteins or nucleic acids, to allow the formation of all or most of the possible combinations over the length of the sequence, thus forming a library of randomized test molecules.

In a preferred embodiment, the test molecules are organic chemical moieties, a wide variety of which are available in the literature.

In a preferred embodiment, the test molecules are drugs, drug analogs or prodrugs. This is particularly useful to help elucidate the mechanism of drug action; for example, there are a wide variety of known drugs for which the targets and/or mechanism of action is unknown. By adding the drugs to NAP conjugates comprising candidate proteins, the proteins to which the drugs bind can be identified, and signaling and disease pathways can be constructed.

Thus, suitable target molecules encompass a wide variety of different classes, including, but not limited to, cells, viruses, proteins (particularly including enzymes, cell-surface receptors, ion channels, and transcription factors, and proteins produced by disease-causing genes or expressed during disease states), carbohydrates, fatty acids and lipids, nucleic acids, chemical moieties such as small molecules, agricultural chemicals, drugs, ions (particularly metal ions), polymers and other biomaterials. Thus for example, binding to polymers (both naturally occurring and synthetic), or other biomaterials, may be done using the methods and compositions of the invention.

In one aspect, the target is a nucleic acid sequence and the desired candidate protein has the ability to bind to the nucleic acid sequence. The present invention is well suited for identification of DNA binding peptides and their coding sequences, as well as the target nucleic acids that are recognized and bound by the DNA binding peptides. It is known that DNA-protein interactions play important roles in controlling gene expression and chromosomal structure, thereby determining the overall genetic program in a given cell. It is estimated that only 5% of the human genome is involved in coding proteins. Thus, the remaining 95% may be sites with which DNA binding proteins interact, thereby controlling a variety of genetic programs such as regulation of gene expression. While the number of DNA binding peptides present in the human genome is not known, the complete sequence information now available for many genomes has revealed the full "substrate," that is, the entire repertoire of DNA sequences with which DNA binding peptides may interact. Thus, it would be advantageous in genetic research to (1) identify nucleic acid sequences that encode DNA binding peptides, and (2) determine the substrate of these DNA binding peptides

Current approaches used in determining protein-DNA interactions are focused on studying the individual interactions between DNA and specific protein targets A variety of biochemical and molecular assays including DNA footpnnting, nuclease protection, gel shift, and affinity chromatographic binding are employed to study protein-DNA interactions Although these methods are useful for detecting individual DNA-protein interactions, they are not suitable for large-scale analyses of these interactions at the genomic level Thus, there is a need in the art to perform large- scale analyses of DNA binding proteins and their interacting DNA sequences The methods and libraries of the present invention are useful for such analyses For example, the fusion enzyme library encoding potential DNA binding peptides can be screened against a population of target DNA segments The population of target DNA segments can be, for instance, random DNA, fragmented genomic DNA, degenerate sequences, or DNA sequences of various primary, secondary or tertiary structures The specificity of the DNA binding peptide-substrate binding can be varied by changing the length of the recognition sequence of the target DNA, if desired Binding of the potential DNA binding peptide to a member of the population of target DNA segments is detected, and further study of the particular DNA recognition sequence bound by the DNA binding peptide can be performed To facilitate identification of fusion enzyme-target nucleic acid complexes, the population of DNA segments can be bound to, for example, beads or constructed as DNA arrays on microchips Therefore, using the present inventive method, one of ordinary skill in the art can identify DNA binding peptides, identify the coding sequence of the DNA binding peptides, and determine what nucleic acid sequence the DNA binding peptides recognize and bind Thus, in one embodiment, the present invention provides methods for creating a map of DNA binding sequences and DNA binding proteins according to their relative positions, to provide chromosome maps annotated with proteins and sequences

Thus, the NAP conjugates are used in screens to assay binding to target molecules and/or to screen candidate agents for the ability to modulate the activity of the target molecule

In general, screens are designed to first find candidate proteins that can bind to target molecules, and then these proteins are used in assays that evaluate the ability of the candidate protein to modulate the target's bioactivity Thus, there are a number of different assays which may be run, binding assays and activity assays As will be appreciated by those in the art, these assays may be run in a variety of configurations, including both solution-based assays and utlizmg support-based systems In a preferred embodiment, the assays comprise combining the NAP conjugates of the invention and a target molecule, and determining the binding of the candidate protein of the NAP conjugate to the target molecule. Preferably, libraries of NAP conjugates (e.g. comprising a library of different candidate proteins) is contacted with either a single type of target molecule, a plurality of target molecules, or one or more libraries of target molecules.

Generally, in a preferred embodiment of the methods herein, one of the components of the invention, either the NAP conjugate or the target molecule, is non-diffusably bound to an insoluble support having isolated sample receiving areas (e.g. a microtiter plate, an array, etc.). The insoluble support may be made of any composition to which the assay component can be bound, is readily separated from soluble material, and is otherwise compatible with the overall method of screening. The surface of such supports may be solid or porous and of any convenient shape. Examples of suitable insoluble supports include microtiter plates, arrays, membranes and beads. These are typically made of glass, plastic (e.g, polystyrene), polysaccharides, nylon or nitrocellulose, teflon™, etc. Microtiter plates and arrays are especially convenient because a large number of assays can be carried out simultaneously, using small amounts of reagents and samples. Alternatively, bead-based assays may be used, particularly with use with fluorescence activated cell sorting (FACS). The particular manner of binding the assay component is not crucial so long as it is compatible with the reagents and overall methods of the invention, maintains the activity of the composition and is nondiffusable. Preferred methods of binding include the use of antibodies (which do not sterically block either the ligand binding site or activation sequence when the protein is bound to the support), direct binding to "sticky" or ionic supports, chemical crosslinking, the use of labeled components (e.g. the assay component is biotinylated and the surface comprises strepavidin, etc. ) the synthesis of the target on the surface, etc. Following binding of the NAP conjugate or target molecule, excess unbound material is removed by washing. The sample receiving areas may then be blocked through incubation with bovine serum albumin (BSA), casein or other innocuous protein or other moiety.

In a preferred embodiment, the target molecule is bound to the support, and a NAP conjugate is added to the assay. Alternatively, the NAP conjugate is bound to the support and the target molecule is added. Novel binding agents include specific antibodies, non-natural binding agents identified in screens of chemical libraries, peptide analogs, etc. Of particular interest are screening assays for agents that have a low toxicity for human cells. Determination of the binding of the target and the candidate protein is done using a wide variety of assays, including, but not limited to labeled in vitro protein-protein binding assays, electrophoretic mobility shift assays, immunoassays for protein binding, the detection of labels, functional assays (phosphorylation assays, etc.) and the like. The determination of the binding of the candidate protein to the target molecule may be done in a number of ways. In a preferred embodiment, one of the components, preferably the soluble one, is labeled, and binding determined directly by detection of the label. For example, this may be done by attaching the NAP conjugate to a solid support, adding a labeled target molecule (for example a target molecule comprising a fluorescent label), washing off excess reagent, and determining whether the label is present on the solid support. This system may also be run in reverse, with the target (or a library of targets) being bound to the support and a NAP conjugate, preferably comprising a primary or secondary label, is added. For example, NAP conjugates comprising fusions with GFP or a variant may be particularly useful. Various blocking and washing steps may be utilized as is known in the art.

As will be appreciated by those in the art, it is also possible to contact the NAP conjugates and the targets prior to immobilization on a support.

In a preferred embodiment, the solid support is in an array format; that is, a biochip is used which comprises one or more libraries of either targets or NAP conjugates attached to the array. The biochips comprise a substrate. By "substrate" or "solid support" or other grammatical equivalents herein is meant any material appropriate for the attachment of capture probes and is amenable to at least one detection method. As will be appreciated by those in the art, the number of possible substrates is very large. Possible substrates include, but are not limited to, glass and modified or functionalized glass, plastics (including acrylics, polystyrene and copolymers of styrene and other materials, polypropylene, polyethylene, polybutylene, polyurethanes, Teflon, etc.), polysaccharides, nylon or nitrocellulose, resins, silica or silica-based materials including silicon and modified silicon, carbon, metals, inorganic glasses, plastics, ceramics, and a variety of other polymers. In a preferred embodiment, the substrates allow optical detection and do not themselves appreciably fluoresce.

This can find particular use in assays for nucleic acid binding proteins, as nucleic acid biochips are well known in the art. In this embodiment, the nucleic acid targets are on the array and the NAP conjugates are added. Similarly, protein biochips of libraries of target proteins can be used, with labeled NAP conjugates added. Alternatively, the NAP conjugates can be attached to the chip, either through the nucleic acid or through the protein components of the system. See also "BIOCHIPS COMPRISING NUCLEIC ACID/PROTEIN CONJUGATES", filed February 22, 2001 , no serial number received yet, hereby expressly incorporated by reference.

This may also be done using bead based systems; for example, for the detection of nucleic acid binding proteins, standard "split and mix" techniques, or any standard oligonucleotide synthesis schemes, can be run using beads or other solid supports, such that libraries of sequences are made. The addition of NAP conjugate libraries then allows for the detection of candidate proteins that bind to specific sequences.

In some embodiments, only one of the components is labeled; alternatively, more than one component may be labeled with different labels.

In a preferred embodiment, the binding of the candidate protein is determined through the use of competitive binding assays. In this embodiment, the competitor is a binding moiety known to bind to the target molecule such as an antibody, peptide, binding partner, ligand, etc. Under certain circumstances, there may be competitive binding as between the target and the binding moiety, with the binding moiety displacing the target.

Thus, a preferred utility of the invention is to determine the components to which a drug will bind. That is, there are many drugs for which the targets upon which they act are unknown, or only partially known.

By starting with a drug, and NAP conjugates comprising a library of cDNA expression products from the cell type on which the drug acts, the elucidation of the proteins to which the drug binds may be elucidated. By identifying other proteins or targets in a signaling pathway, these newly identified proteins can be used in additional drug screens, as a tool for counterscreens, or to profile chemically induced events. Furthermore, it is possible to run toxicity studies using this same method; by identifying proteins to which certain drugs undesirably bind, this information can be used to design drug derivatives without these undesirable side effects. Additionally, drug candidates can be run in these types of screens to look for any or all types of interactions, including undesirable binding reactions. Similarly, it is possible to run libraries of drug derivatives as the targets, to provide a two- dimensional analysis as well.

Positive controls and negative controls may be used in the assays. Preferably all control and test samples are performed in at least triplicate to obtain statistically significant results. Incubation of all samples is for a time sufficient for the binding of the agent to the protein. Following incubation, all samples are washed free of non-specifically bound material and the amount of bound, generally labeled agent determined. For example, where a radiolabel is employed, the samples may be counted in a scintillation counter to determine the amount of bound compound. Similarly, ELISA techniques are generally preferred.

A variety of other reagents may be included in the screening assays. These include reagents like salts, neutral proteins, e.g. albumin, detergents, etc which may be used to facilitate optimal protein-protein binding and/or reduce non-specific or background interactions. Also reagents that otherwise improve the efficiency of the assay, such as protease inhibitors, nuclease inhibitors, anti-microbial agents, co-factors such as cAMP, ATP, etc, may be used. The mixture of components may be added in any order that provides for the requisite binding.

Screening for agents that modulate the activity of the target molecule may also be done. As will be appreciated by those in the art, the actual screen will depend on the identity of the target molecule. In a preferred embodiment, methods for screening for a candidate protein capable of modulating the activity of the target molecule comprise the steps of adding a NAP conjugate to a sample of the target, as above, and determining an alteration in the biological activity of the target. "Modulation" or "alteration" in this context includes an increase in activity, a decrease in activity, or a change in the type or kind of activity present. Thus, in this embodiment, the candidate protein should both bind to the target (although this may not be necessary), and alter its biological or biochemical activity as defined herein. The methods include both in vitro screening methods, as are generally outlined above, and ex vivo screening of cells for alterations in the presence, distribution, activity or amount of the target.

Thus, in this embodiment, the methods comprise combining a target molecule and preferably a library of NAP conjugates and and evaluating the effect on the target molecule's bioactivity. This will be done in a wide variety of ways, as will be appreciated by those in the art.

In these in vitro systems, e.g. cell-free systems, in either embodiment, e.g. in vitro binding or activity assays, once a "hit" is found, the NAP conjugate is retrieved to allow identification of the candidate protein. Retrieval of the NAP conjugate can be done in a wide variety of ways, as will be appreciated by those in the art and will also depend on the type and configuration of the system being used.

In a preferred, embodiment, as outlined herein, a rescue tag or"retrieval property" is used. As outlined above, a "retrieval property" is a property that enables isolation of the fusion enzyme when bound to the target. For example, the target can be constructed such that it is associated with biotin, which enables isolation of the target-bound fusion enzyme complexes using an affinity column coated with streptavidin. Alternatively, the target can be attached to magnetic beads, which can be collected and separated from non-binding candidate proteins by altering the surrounding magnetic field. Alternatively, when the target does not comprise a rescue tag, the NAP conjugate may comprise the rescue tag. For example, affinity tags may be incorporated into the fusion proteins themselves. Similarly, the fusion enzyme-nucleic acid molecule complex can be also recovered by immunoprecipitation. Alternatively, rescue tags may comprise unique vector sequences that can be used to PCR amplify the nucleic acid encoding the candidate protein. In the latter embodiment, it may not be necessary to break the covalent attachment of the nucleic acid and the protein, if PCR sequences outside of this region (that do not span this region) are used.

In a preferred embodiment, after isolation of the NAP conjugate of interest, the covalent linkage between the fusion enzyme and its coding nucleic acid molecule can be severed using, for instance, nuclease-free proteases, the addition of non-specific nucleic acid, or any other conditions that preferentially digest proteins and not nucleic acids.

The nucleic acid molecules are purified using any suitable methods, such as those methods known in the art, and are then available for further amplification, sequencing or evolution of the nucleic acid sequence encoding the desired candidate protein. Suitable amplification techniques include all forms of PCR, OLA, SDA, NASBA, TMA, Q-βR, etc. Subsequent use of the information of the "hit" is discussed below.

In a preferred embodiment, the NAP conjugates are used in ex vivo screening techniques. In this embodiment, the expression vectors of the invention are introduced into the cells to screen for candidate proteins capable of altering the phenotype of a cell. An advantage of the present inventive method is that screening of the fusion enzyme library can be accomplished intracellularly. One of ordinary skill in the art will appreciate the advantages of screening candidate proteins within their natural environment, as opposed to lysing the cell to screen in vitro. In ex vivo screening methods, variant peptides are displayed in their native conformation and are screened in the presence of other possibly interfering or enhancing cellular agents. Accordingly, screening intracellularly provides a more accurate picture of the actual activity of the candidate protein and, therefore, is more predictive of the activity of the peptide ex vivo. Moreover, the effect of the candidate protein on cellular physiology can be observed. Thus, the invention finds particular use in the screening of eucaryotic cells.

Ex vivo screening can be done in several ways. In a preferred embodiment, the target need not be known; rather, cells containing the expression vectors of the invention are screened for changes in phenotype. Cells exhibiting an altered phenotype are isolated, and the target to which the NAP conjugate bound is identified as outlined below, although as will be appreciated by those in the art and outlined herein, it is also possible to bind the fusion polypeptide and the target prior to forming the NAP conjugate. Alternatively, the target may be added exogenously to the cell and screening for binding and/or modulation of target activity is done. In the latter embodiment, the target should be able to penetrate the membrane, by, for instance, direct penetration or via membrane transporting proteins, or by fusions with transport moieties such as lipid moieties or HIV-tat, described below. In general, experimental conditions allow for the formation of NAP conjugates within the cells prior to screening, although this is not required. That is, the attachment of the NAM enzyme to the EAS may occur at any time during the screening, either before, during or after, as long as the conditions are such that the attachment occurs prior to mixing of cells or cell lysates containing different fusion nucleic acids.

As will be appreciated by those in the art, the type of cells used in this embodiment can vary widely. Basically, any eucaryotic or procaryotic cells can be used, with mammalian cells being preferred, especially mouse, rat, primate and human cells. As is more fully described below, a screen will be set up such that the cells exhibit a selectable phenotype in the presence of a candidate protein. As is more fully described below, cell types implicated in a wide variety of disease conditions are particularly useful, so long as a suitable screen may be designed to allow the selection of cells that exhibit an altered phenotype as a consequence of the presence of a candidate agent within the cell.

Accordingly, suitable cell types include, but are not limited to, tumor cells of all types (particularly melanoma, myeloid leukemia, carcinomas of the lung, breast, ovaries, colon, kidney, prostate, pancreas and testes), cardiomyocytes, endothelial cells, epithelial cells, lymphocytes (T-cell and B cell) , mast cells, eosinophils, vascular intimal cells, hepatocytes, leukocytes including mononuclear leukocytes, stem cells such as haemopoetic, neural, skin, lung, kidney, liver and myocyte stem cells (for use in screening for differentiation and de-differentiation factors), osteoclasts, chondrocytes and other connective tissue cells, keratinocytes, melanocytes, liver cells, kidney cells, and adipocytes. Suitable cells also include known research cells, including, but not limited to, Jurkat T cells, NIH3T3 cells, CHO, Cos, etc. See the ATCC cell line catalog, hereby expressly incorporated by reference.

In one embodiment, the cells may be genetically engineered, that is, contain exogenous nucleic acid, for example, to contain target molecules.

In a preferred embodiment, a first plurality of cells is screened. That is, the cells into which the expression vectors are introduced are screened for an altered phenotype. Thus, in this embodiment, the effect of the candidate protein is seen in the same cells in which it is made; i.e. an autocrine effect.

By a "plurality of cells" herein is meant roughly from about 10³ cells to 10⁸ or 10⁹, with from 10⁶ to 10⁸ being preferred. This plurality of cells comprises a cellular library, wherein generally each cell within the library contains a member of the NAP conjugate molecular library, i.e. a different candidate protein, although as will be appreciated by those in the art, some cells within the library may not contain an expression vector and some may contain more than one. In a preferred embodiment, the expression vectors are introduced into a first plurality of cells, and the effect of the candidate proteins is screened in a second or third plurality of cells, different from the first plurality of cells, i.e. generally a different cell type. That is, the effect of the candidate protein is due to an extracellular effect on a second cell; i.e. an endocrine or paracrine effect. This is done using standard techniques. The first plurality of cells may be grown in or on one media, and the media is allowed to touch a second plurality of cells, and the effect measured. Alternatively, there may be direct contact between the cells. Thus, "contacting" is functional contact, and includes both direct and indirect. In this embodiment, the first plurality of cells may or may not be screened.

If necessary, the cells are treated to conditions suitable for the expression of the fusion nucleic acids (for example, when inducible promoters are used), to produce the candidate proteins.

Thus, the methods of the present invention comprise introducing a molecular library of fusion nucleic acids or expression vectors into a plurality of cells, a cellular library. Each of the nucleic acids comprises a different nucleotide sequence encoding a different candidate protein. The plurality of cells is then screened, as is more fully outlined below, for a cell exhibiting an altered phenotype. The altered phenotype is due to the presence of a candidate protein.

By "altered phenotype" or "changed physiology" or other grammatical equivalents herein is meant that the phenotype of the cell is altered in some way, preferably in some detectable and/or measurable way. As will be appreciated in the art, a strength of the present invention is the wide variety of cell types and potential phenotypic changes which may be tested using the present methods. Accordingly, any phenotypic change which may be observed, detected, or measured may be the basis of the screening methods herein. Suitable phenotypic changes include, but are not limited to: gross physical changes such as changes in cell morphology, cell growth, cell viability, adhesion to substrates or other cells, and cellular density; changes in the expression of one or more RNAs, proteins, lipids, hormones, cytokines, or other molecules; changes in the equilibrium state (i.e. half- life) or one or more RNAs, proteins, lipids, hormones, cytokines, or other molecules; changes in the localization of one or more RNAs, proteins, lipids, hormones, cytokines, or other molecules; changes in the bioactivity or specific activity of one or more RNAs, proteins, lipids, hormones, cytokines, receptors, or other molecules; changes in the secretion of ions, cytokines, hormones, growth factors, or other molecules; alterations in cellular membrane potentials, polarization, integrity or transport; changes in infectivity, susceptability, latency, adhesion, and uptake of viruses and bacterial pathogens; etc. By "capable of altering the phenotype" herein is meant that the candidate protein can change the phenotype of the cell in some detectable and/or measurable way. The altered phenotype may be detected in a wide variety of ways, as is described more fully below, and will generally depend and correspond to the phenotype that is being changed. Generally, the changed phenotype is detected using, for example: microscopic analysis of cell morphology; standard cell viability assays, including both increased cell death and increased cell viability, for example, cells that are now resistant to cell death via virus, bacteria, or bacterial or synthetic toxins; standard labeling assays such as fluorometric indicator assays for the presence or level of a particular cell or molecule, including FACS or other dye staining techniques; biochemical detection of the expression of target compounds after killing the cells; etc.

In a preferred embodiment, the present methods are useful in cancer applications. The ability to rapidly and specifically kill tumor cells is a cornerstone of cancer chemotherapy. In general, using the methods of the present invention, random or directed libraries (including cDNA libraries) can be introduced into any tumor cell (primary or cultured), and peptides identified which by themselves induce apoptosis, cell death, loss of cell division or decreased cell growth. This may be done de novo, or by biased randomization toward known peptide agents, such as angiostatin, which inhibits blood vessel wall growth. Alternatively, the methods of the present invention can be combined with other cancer therapeutics (e.g. drugs or radiation) to sensitize the cells and thus induce rapid and specific apoptosis, cell death, loss of cell division or decreased cell growth after exposure to a secondary agent. Similarly, the present methods may be used in conjunction with known cancer therapeutics to screen for agonists to make the therapeutic more effective or less toxic. This is particularly preferred when the chemotherapeutic is very expensive to produce such as taxol.

In a preferred embodiment, the present invention finds use with infectious organisms. Intracellular organisms such as mycobacteria, listeria, salmonella, pneumocystis, yersinia, leishmania, T. cruzi, can persist and replicate within cells, and become active in immunosuppressed patients. There are currently drugs on the market and in development which are either only partially effective or ineffective against these organisms. Candidate libraries can be inserted into specific cells infected with these organisms (pre- or post-infection), and candidate proteins selected which promote the intracellular destruction of these organisms in a manner analogous to intracellular "antibiotic peptides" similar to magainins. In addition peptides can be selected which enhance the cidal properties of drugs already under investigation which have insufficient potency by themselves, but when combined with a specific peptide from a candidate library, are dramatically more potent through a synergistic mechanism. Finally, candidate proteins can be isolated which alter the metabolism of these intracellular organisms, in such a way as to terminate their intracellular life cycle by inhibiting a key organismal event.

In a preferred embodiment, the compositions and methods of the invention are used to detect protein- protein interactions, similar to the use of a two-hybrid screen. This can be done in a variety of ways and in a variety of formats. As will be appreciated by those in the art, this embodiment and others outlined herein can be run as a "one dimensional" analysis or "multidimensional" analysis. That is, one NAP conjugate library can be run against a single target or against a library of targets. Alternatively, more than one NAP conjugate library can be run against each other.

In a preferred embodiment, the compositions and methods of the invention are used to do protein drug discovery, particularly for protein drugs that interact with targets on cell surfaces.

In a preferred embodiment, as outlined above, the compositions and methods of the invention are used to discover DNA or nucleic acid binding proteins, using nucleic acids as. the targets.

In a preferred embodiment, the compositions and methods of the invention are used to screen for NAM enzymes with decreased toxicity for the host cells. For example, under some conditions, the Rep proteins of the invention can be toxic to the host cells. Thus the screening methods of the invention may be run using libraries of variants of NAM enzymes and screening for host cell toxicity. Nucleic acids that encode variants of NAM enzymes can be generated in a variety of ways, including error-prone PCR and known mutagenesis techniques.

In a preferred embodiment, the compositions and methods of the invention are used to screen for NAM enzyme/EAS pairs with increased affinity. This can be done either by looking for higher affinity NAM enzymes or by looking for improved EAS sequences, or both.

In a preferred embodiment, the compositions and methods of the invention are used in pharmacogenetic studies. For example, by building libraries from individuals with different phenotypes and testing them against targets, differential binding profiles can be generated. Thus, a preferred embodiment utilizes differential binding profiles of NAP conjugates to targets to elucidate disease genes, SNPs or proteins.

In a preferred embodiment, once a cell with an altered phenotype is detected, the cell is isolated from the plurality which do not have altered phenotypes. This may be done in any number of ways, as is known in the art, and will in some instances depend on the assay or screen. Suitable isolation techniques include, but are not limited to, FACS, lysis selection using complement, cell cloning, scanning by Fluorimager, expression of a "survival" protein, induced expression of a cell surface protein or other molecule that can be rendered fluorescent or taggable for physical isolation; expression of an enzyme that changes a non-fluorescent molecule to a fluorescent one; overgrowth against a background of no or slow growth; death of cells and isolation of DNA or other cell vitality indicator dyes, etc. In a preferred embodiment, as outlined above, the NAP conjugate is isolated from the positive cell. This may be done in a number of ways. In a preferred embodiment, primers complementary to DNA regions common to the NAP constructs, or to specific components of the library such as a rescue sequence, defined above, are used to "rescue" the unique candidate protein sequence. Alternatively, the candidate protein is isolated using a rescue sequence. Thus, for example, rescue sequences comprising epitope tags or purification sequences may be used to pull out the candidate protein, using immunoprecipitation or affinity columns. In some instances, as is outlined below, this may also pull out the primary target molecule, if there is a sufficiently strong binding interaction between the candidate protein and the target molecule. Alternatively, the peptide may be detected using mass spectroscopy.

Once rescued, the sequence of the candidate protein and fusion nucleic acid can be determined. This information can then be used in a number of ways.

For both in vitro and ex vivo screening methods, once the "hit" has been identified, the results are preferably verified. As will be appreciated by those in the art, there are a variety of suitable methods that can be used. In a preferred embodiment, the candidate protein is resynthesized and reintroduced into the target cells, to verify the effect. This may be done using recombinant methods, e.g. by transforming naive cells with the expression vector (or modified versions, e.g. with the candidate protein no longer part of a fusion), or alternatively using fusions to the HIV-1 Tat protein, and analogs and related proteins, which allows very high uptake into target cells. See for example, Fawell et al, PNAS USA 91 :664 (1994); Frankel et al. Cell 55:1189 (1988); Savion et al, J. Biol. Chem. 256:1149 (1981); Derossi et al, J. Biol. Chem. 269:10444 (1994); and Baldin et al, EMBO J. 9:1511 (1990), all of which are incorporated by reference.

In addition, for both in vitro and ex vivo screening methods, the process may be used reiteratively. That is, the sequence of a candidate protein is used to generate more candidate proteins. For example, the sequence of the protein may be the basis of a second round of (biased) randomization, to develop agents with increased or altered activities. Alternatively, the second round of randomization may change the affinity of the agent. Furthermore, if the candidate protein is a random peptide, it may be desirable to put the identified random region of the agent into other presentation structures, or to alter the sequence of the constant region of the presentation structure, to alter the conformation/shape of the candidate protein.

The methods of using the present inventive library can involve many rounds of screenings in order to identify a nucleic acid of interest. For example, once a nucleic acid molecule is identified, the method can be repeated using a different target. Multiple libraries can be screened in parallel or sequentially and/or in combination to ensure accurate results. In addition, the method can be repeated to map pathways or metabolic processes by including an identified candidate protein as a target in subsequent rounds of screening.

In a preferred embodiment, the candidate protein is used to identify target molecules, i.e. the molecules with which the candidate protein interacts. As will be appreciated by those in the art, there may be primary target molecules, to which the protein binds or acts upon directly, and there may be secondary target molecules, which are part of the signalling pathway affected by the protein agent; these might be termed "validated targets".

In a preferred embodiment, the candidate protein is used to pull out target molecules. For example, as outlined herein, if the target molecules are proteins, the use of epitope tags or purification sequences can allow the purification of primary target molecules via biochemical means (co- immunoprecipitation, affinity columns, etc.). Alternatively, the peptide, when expressed in bacteria and purified, can be used as a probe against a bacterial cDNA expression library made from mRNA of the target cell type. Or, peptides can be used as "bait" in either yeast or mammalian two or three hybrid systems. Such interaction cloning approaches have been very useful to isolate DNA-binding proteins and other interacting protein components. The peptide(s) can be combined with other pharmacologic activators to study the epistatic relationships of signal transduction pathways in question. It is also possible to synthetically prepare labeled peptides and use it to screen a cDNA library expressed in bacteriophage for those cDNAs which bind the peptide.

Once primary target molecules have been identified, secondary target molecules may be identified in the same manner, using the primary target as the "bait". In this manner, signalling pathways may be elucidated. Similarly, protein agents specific for secondary target molecules may also be discovered, to allow a number of protein agents to act on a single pathway, for example for combination therapies.

In a preferred embodiment, the methods and compositions of the invention comprise a robotic system: Many systems are generally directed to the use of 96 (or more) well microtiter plates, but as will be appreciated by those in the art, any number of different plates or configurations may be used. In addition, any or all of the steps outlined herein may be automated; thus, for example, the systems may be completely or partially automated.

As will be appreciated by those in the art, there are a wide variety of components which can be used, including, but not limited to, one or more robotic arms; plate handlers for the positioning of microplates; automated lid handlers to remove and replace lids for wells on non-cross contamination plates; tip assemblies for sample distribution with disposable tips; washable tip assemblies for sample distribution; 96 well loading blocks; cooled reagent racks; microtitler plate pipette positions (optionally cooled); stacking towers for plates and tips; and computer systems.

Fully robotic or microfluidic systems include automated liquid-, particle-, cell- and organism-handling including high throughput pipetting to perform all steps of screening applications. This includes liquid, particle, cell, and organism manipulations such as aspiration, dispensing, mixing, diluting, washing, accurate volumetric transfers; retrieving, and discarding of pipet tips; and repetitive pipetting of identical volumes for multiple deliveries from a single sample aspiration. These manipulations are cross-contamination-free liquid, particle, cell, and organism transfers. This instrument performs automated replication of microplate samples to filters, membranes, and/or daughter plates, high- density transfers, full-plate serial dilutions, and high capacity operation.

In a preferred embodiment, chemically derivatized particles, plates, tubes, magnetic particle, or other solid phase matrix with specificity to the assay components are used. The binding surfaces of microplates, tubes or any solid phase matrices include non-polar surfaces, highly polar surfaces, modified dextran coating to promote covalent binding, antibody coating, affinity media to bind fusion proteins or peptides, surface-fixed proteins such as recombinant protein A or G, nucleotide resins or coatings, and other affinity matrix are useful in this invention.

In a preferred embodiment, platforms for multi-well plates, multi-tubes, minitubes, deep-well plates, microfuge tubes, cryovials, square well plates, filters, chips, optic fibers, beads, and other solid-phase matrices or platform with various volumes are accommodated on an upgradable modular platform for additional capacity. This modular platform includes a variable speed orbital shaker, electroporator, and multi-position work decks for source samples, sample and reagent dilution, assay plates, sample and reagent reservoirs, pipette tips, and an active wash station.

In a preferred embodiment, thermocycler and thermoregulating systems are used for stabilizing the temperature of the heat exchangers such as controlled blocks or platforms to provide accurate temperature control of incubating samples from 4_°C to 100_°C.

In a preferred embodiment, Interchangeable pipet heads (single or multi-channel ) with single or multiple magnetic probes, affinity probes, or pipetters robotically manipulate the liquid, particles, cells, and organisms. Multi-well or multi-tube magnetic separators or platforms manipulate liquid, particles, cells, and organisms in single or multiple sample formats. In some preferred embodiments, the instrumentation will include a detector, which can be a wide variety of different detectors, depending on the labels and assay. In a preferred embodiment, useful detectors include a microscope(s) with multiple channels of fluorescence; plate readers to provide fluorescent, ultraviolet and visible spectrophotometric detection with single and dual wavelength endpoint and kinetics capability, fluorescence resonance energy transfer (FRET), luminescence, quenching, two-photon excitation, and intensity redistribution; CCD cameras to capture and transform data and images into quantifiable formats; and a computer workstation. These will enable the monitoring of the size, growth and phenotypic expression of specific markers on cells, tissues, and organisms; target validation; lead optimization; data analysis, mining, organization, and integration of the high-throughput screens with the public and proprietary databases.

These instruments can fit in a sterile laminar flow or fume hood, or are enclosed, self-contained systems, for cell culture growth and transformation in multi-well plates or tubes and for hazardous operations. The living cells will be grown under controlled growth conditions, with controls for temperature, humidity, and gas for time series of the live cell assays. Automated transformation of cells and automated colony pickers will facilitate rapid screening of desired cells.

Flow cytometry or capillary electrophoresis formats can be used for individual capture of magnetic and other beads, particles, cells, and organisms.

The flexible hardware and software allow instrument adaptability for multiple applications. The software program modules allow creation, modification, and running of methods. The system diagnostic modules allow instrument alignment, correct connections, and motor operations. The customized tools, labware, and liquid, particle, cell and organism transfer patterns allow different applications to be performed. The database allows method and parameter storage. Robotic and computer interfaces allow communication between instruments.

In a preferred embodiment, the robotic workstation includes one or more heating or cooling components. Depending on the reactions and reagents, either cooling or heating may be required, which can be done using any number of known heating and cooling systems, including Peltier systems.

In a preferred embodiment, the robotic apparatus includes a central processing unit which communicates with a memory and a set of input/output devices (e.g, keyboard, mouse, monitor, printer, etc.) through a bus. The general interaction between a central processing unit, a memory, input/output devices, and a bus is known in the art. Thus, a variety of different procedures, depending on the experiments to be run, are stored in the CPU memory. The above-described methods of screening a pool of fusion enzyme-nucleic acid molecule complexes for a nucleic acid encoding a desired candidate protein are merely based on the desired target property of the candidate protein. The sequence or structure of the candidate proteins does not need to be known. A significant advantage of the present invention is that no prior information about the candidate protein is needed during the screening, so long as the product of the identified coding nucleic acid sequence has biological activity, such as specific association with a targeted chemical or structural moiety. The identified nucleic acid molecule then can be used for understanding cellular processes as a result of the candidate protein's interaction with the target and, possibly, any subsequent therapeutic or toxic activity.

All references cited herein are incorporated by reference in their entirety.

.05

Claims

CLAIMS We claim:

1. A method for generating a library of fusion nucleic acids comprising: a) providing a computationally-derived primary library of candidate protein sequences; and b) creating a library of expression vectors each comprising: i) a fusion nucleic acid comprising:

A) a first nucleic acid sequence encoding a nucleic acid modification (NAM) enzyme; and

B) a second nucleic acid encoding a candidate protein sequence from said library; and ii) an enzyme attachment sequence that is recognized by said NAM enzyme.

2. A method according to claim 1 wherein said primary library comprises first candidate protein sequences, and said method further comprises: a) generating a list of primary variant positions in said primary library; and b) combining a plurality of said primary variant positions to generate a second library of second candidate protein sequences; wherein said second nucleic acid encodes a second candidate protein sequence.

3. A method of screening comprising: a) providing a computationally-derived primary library of candidate protein sequences;

b) creating a library of expression vectors each comprising: i) a fusion nucleic acid comprising:

B) a second nucleic acid encoding a candidate protein; and ii) an enzyme attachment sequence (EAS) that is recognized by said NAM enzyme; c) expressing said fusion nucleic acids under conditions whereby a library of nucleic acid/protein (NAP) conjugates are formed, wherein said NAP conjugate comprises: i) a fusion polypeptide comprising a NAM enzyme and said candidate protein; and ii) an expression vector; wherein said NAM enzyme and said EAS are covalently attached; d) adding at least one test molecule to said NAP conjugate library; and e) determining the binding of a NAP conjugate to said test molecule.

4. A method according to claim 3 wherein said expressing is accomplished by transforming said library of fusion nucleic acids into cells under conditions whereby NAP conjugates are made.

5. A method according to claim 3 wherein said primary library comprises first candidate protein sequences, and said method further comprises: a) generating a list of primary variant positions in said primary library; and b) combining a plurality of said primary variant positions to generate a second library of second candidate protein sequences; wherein said second nucleic acid encodes a second candidate protein sequence.

6. A method of screening comprising: a) providing a computationally-derived primary library of candidate protein sequences;

b) providing a library of eucaryotic host cells each comprising at least one expression vector comprising: i) a fusion nucleic acid comprising:

1) a nucleic acid sequence encoding a nucleic acid modification (NAM) enzyme; and

2) a nucleic acid sequence encoding a candidate protein; and ii) an EAS that is recognized by said NAM enzyme; under conditions whereby a fusion polypeptide is produced and wherein at least two of said candidate proteins are different; and c) lysing said cells, wherein said EAS and said NAM enzyme are covalently attached to form a NAP conjugate; d) adding at least one test molecule; e) determining the binding of said test molecule to a NAP conjugate.

7. A method according to claim 6 wherein said primary library comprises first candidate protein sequences, and said method further comprises: a) generating a list of primary variant positions in said primary library; and b) combining a plurality of said primary variant positions to generate a second library of second candidate protein sequences; wherein said second nucleic acid encodes a second candidate protein sequence.

8. A method according to claim 1, 3 or 6 wherein said nucleic acid sequence encoding a candidate protein is derived from cDNA.

9. A method according to claim 1, 3 or 6 wherein said nucleic acid sequence encoding a candidate protein is derived from genomic DNA.

10. A method according to claim 1, 3 or 6 wherein said candidate protein is a random protein.

11. A method according to claim 1 , 3 or 6 wherein said nucleic acids are directly fused.

12. A method according to claim 1 , 3 or 6 wherein said nucleic acids are indirectly fused.

13. A method according to claim 1, 3 or 6 wherein said NAM enzyme is a Rep protein.

14. A method according to claim 13 wherein said Rep protein is Rep68.

15. A method according to claim 14 wherein said Rep protein is Rep78.

16. A method according to claim 1 , 3 or 6 wherein said NAP conjugates are labeled.