JP2003527072A - Protein design automation for protein libraries - Google Patents

Abstract

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

Description

Detailed Description of the Invention

FIELD OF THE INVENTION The present invention relates to the use of protein design automation (PDA) to create a secondary library of computationally prescreened proteins,
And methods and compositions using the library.

BACKGROUND OF THE INVENTION By using directed molecular evolution, proteins and enzymes with novel functions and properties can be produced. Starting from a known native protein, several rounds of mutagenesis, functional screening and successful sequence amplification are performed. The advantage of this method is that it allows rapid evolution of proteins of unknown structure. There are several different mutagenesis methods, including error-prone PCR point mutagenesis, cassette mutagenesis and DNA shuffling. These technologies have achieved many results. However, they all suffer from the disadvantage that they can only produce a small fraction with potential changes. For example, there are 20 ⁵⁰⁰ possible amino acid changes for an average protein with a length of about 500 amino acids. Clearly, mutagenesis and functional screening of a large number of variants is not possible. The sampling of possible sequences provided by directed evolution is so interspersed that only a small portion of the potentially improved proteins could be investigated, typically point mutations or recombination of existing sequences. Absent. Random sampling from an enormous number of possible sequences allows directed evolution to be unbiased and widely applicable, but essential because it ignores all protein structural and biophysical information. Is inefficient.

[0003] In contrast, by using a computer based method operation, since vast array library (one calculated at 10 ⁸⁰ or less) are screened, experimental library screening methods, for example, a key directed molecular evolution Can be overcome. A wide variety of methods for generating and evaluating sequences are known. These include sequence profiling (Bowie and Eisenberg, Science 253 (5016
): 164-70 (1991)), rotamer library selection method (Dahiyat and Mayo, Protein Sci5 (5): 895-903 (1996), Dahiyat and
Mayo, Science 278 (5335): 82-7 (1997); Desjarlais and H.
andel, Protein Science 4: 2006-2018 (1995), Harbury et al., PNA.
S USA 92 (18); 8408-8421 (1995); Kono et al., Proteins: Str.
ucture, Function and Genetics 19: 244-255 (1994), Hellinga
And Richards, PNAS USA 91: 5803-5807 (1994)), and residue pair potential (Jones, Protein Science 3: 567-574 (1994).
)), But is not limited.

In particular, USS N 60/061097, 60/043464, 60/054
678, 09/127926 and PCT US98 / 07254 describe a method called "protein design automation" or PDA, which uses some scoring function to assess sequence stability.

It is an object of the present invention to provide a computerized method for prescreening a sequence library to generate and select a secondary library, which library can then be experimentally generated and evaluated.

SUMMARY OF THE INVENTION In accordance with the objects of the present invention outlined above, the present invention is a method of making a secondary library of scaffold protein variants, comprising a rank ordered list of scaffold protein primary variant sequences. Methods are provided that include providing a primary library. Then, a list of primary mutation positions in the primary library is created, and then a plurality of primary mutation positions are combined to create a secondary library of secondary sequences.

In another aspect, the invention provides a method of making a secondary library of scaffold protein variants, comprising providing a primary library comprising a rank ordered list of scaffold protein primary variant sequences, And a method for creating a probability distribution of amino acid residues at multiple mutation positions. A secondary library of secondary sequences is created by combining multiple amino acid residues. These sequences can then be optionally synthesized and tested in a variety of ways, including multiplex PCR with pooled oligonucleotides, error prone PCR, gene shuffling, and the like.

In another aspect, the invention provides a composition comprising a plurality of secondary mutant proteins or nucleic acids encoding proteins, the plurality comprising all or a subset of a secondary library. . The invention further provides cells comprising the library, particularly mammalian cells.

[0009] In another aspect, the invention provides a method of making a secondary library of scaffold protein variants, the method comprising providing a first library rank ordered list of scaffold protein primary variants, a plurality of mutations. A method for creating a probability distribution of amino acid residues at a position and a method for forming a secondary library by synthesizing a plurality of backbone protein secondary mutants containing a plurality of amino acid residues are provided. At least one of the secondary variants differs from the primary variant.

In another aspect, the invention provides a method executed by a computer under the control of a program, the computer including a memory for storing the program. This method accepts a protein scaffold with variable residue positions, establishes a group of potential rotamers for each of the variable residue positions, and each of the rotamers and all or the rest of the protein scaffold. Generating a set of optimized protein sequences by analyzing interactions with a portion. Furthermore, these methods include classifying each variable residue position as a core, surface or boundary residue. The analysis step may include dead-end elimination (DEE) computer operation. Generally, the analysis step comprises at least one selected from the group consisting of van der Waals potential scoring function, hydrogen bond potential function, atomic solvation scoring function, secondary structure property scoring function and electrostatic scoring function. Includes the use of scoring features. In addition, these methods involve modifying the protein backbone prior to analysis and include modifying at least one super-secondary structure parameter value. In addition, these methods include creating a rank ordered list of additional optimal sequences from the global optimal protein sequence. Potential energy test results may be generated by testing some or all of the protein sequences from the ordered list. Further, these methods can include generating and / or ranking secondary libraries using the techniques outlined herein. That is, an apparatus including computer code for executing a program is also provided.

BRIEF DESCRIPTION OF THE FIGURES FIG. 1 depicts the synthesis of a full-length gene and the introduction of all possible mutations by PCR. Synthesized overlapping oligonucleotides corresponding to the full length gene (black bar, step 1),
Heat and anneal. Pfu D on the annealed oligonucleotide
By adding NA polymerase, 5′-3 ′ synthesis of DNA is performed (step 2) to produce a longer DNA fragment (step 3). Longer DN, including some full length molecules, due to repeated cycles of heating and annealing (step 4)
A is manufactured. These can be selected by a second round of PCR with primers (arrows) corresponding to the ends of the full length gene (step 5).

FIG. 2 shows the reduction in the size of the sequence space by PDA screening. From left to right, 1: no PDA, 2: no PDA, cysteine, proline, glycine not counted, 3: PDA performed, 1% reference used, free enzyme modeled, 4: PDA performed, 1% reference used, enzyme Model substrate complex, 5: PDA performed, 5% reference used, free enzyme modeled, 6: PDA performed, 5% reference used, enzyme substrate complex modeled.

FIG. 3 shows the active site of B. circulans xylanase. Their positions included in the PDA design are indicated by their side chain presentation. In red are wild type residues (their configurations could be changed but their amino acid identities were not changed). In green are positions where the configuration and identity could be changed (relative to amino acids except proline, cysteine and glycine).

FIG. 4 shows E. coli expressing wild type (WT) and PDA screen β-
The resistance of lactamase to cefotaxime is shown, and the results are shown with increasing concentrations of cefotaxime.

[0015] DETAILED DESCRIPTION OF THE INVENTION The present invention is a small secondary library (10 of protein sequence by using computational screening protein sequence libraries (10 ⁸⁰ can include the following or more members) ⁽ Which may include up to ¹³ members), which may then be actually synthesized and experimentally tested with the desired assay for improved function and properties.

The present invention has two broad applications. First, the present invention can be used for prescreening libraries based on known scaffold proteins. That is, computational screens for stability (or other properties) can be performed on the entire protein or on some subset of residues, as described below, as desired. By using computerized methods to generate thresholds or cutoffs and deleting unfavorable sequences, the percentage of useful mutants in a given set of mutant sizes can be increased, and the experimental expenditure required is Decrease.

Furthermore, the present invention is useful for screening random peptide libraries. As is known, signal transduction pathways in cells often begin with effector stimulators that lead to phenotypic changes in cell physiology. Despite the important role that intracellular signaling pathways play in disease pathogenesis, little is known about signaling pathways other than early stimulators and final cellular responses.

Historically, signal transduction has been analyzed by biochemistry or genetics. Biochemical methods dissect pathways in a “reach-by” fashion. That is, one finds molecules that act at or are contained in one end of the pathway, separates the assayable amount, and then attempts to measure the next molecule in the pathway, either upstream or downstream of the separation. Genetic methods are classically based on "speculation." That is, mutations are induced or induced in the signal transduction pathway and genetic loci map the loci or complement the mutations with a cDNA library. Limitations of biochemical methods include the reliance on a significant amount of pre-existing information about the components under test and the need to perform the test in vitro afterwards. Limitations of pure genetic methods include the need to first induce and then characterize the pathway before continuing gene identification and cloning.

Screening a molecular library of chemical compounds for drugs that modulate the signal system has led to important discoveries of great clinical significance. For example, cyclosporin A (CsA) and FK506 were selected in a standard pharmaceutical screen for inhibition of T cell activation. These two drugs bind completely different cellular proteins--cyclophilin and FK506 binding protein (FKBP), respectively, and the effect of either drug is on transcription factors such as NF-AT and NF-κB.
It is worth noting that it is the strong and specific repression of T cell activation that can be observed phenotypically in T cells in virtually the same way as the inhibition of mRNA production in a dependent manner. Libraries of small peptides have also been screened in vitro in assays for biological activity and have been successful. The literature is replete with examples of small peptides that can modulate a wide variety of signaling pathways. For example, peptides derived from the HIV-1 envelope protein have been shown to block the action of cellular calmodulin.

Thus, by generating random or semi-random sequence libraries of proteins and peptides, proteins with useful properties, including peptides, oligopeptides and polypeptides, can be selected. The sequences in these experimental libraries can be randomized only at specific sites or throughout the sequences. The number of sequences that can be searched in these libraries grows exponentially with the number of positions that are randomized. In general, due to physical constraints in the laboratory (device size, cost of manufacturing many biopolymers, etc.), only 10 ¹² -10 ¹⁵ or less sequences can be included in the library. Often, due to other practical considerations, the size of libraries can be limited to 10 ⁶ or less. It is only at the 10 amino acid position that these limits are reached. Thus, a search for improved proteins or peptides in an experimental sequence library is only possible if the sampling of sequences is interspersed, reducing the chances of success and almost certainly overlooking the desired candidate. Because the changes in these sequences are random, most of the candidates in the library are unsuitable and most of the efforts in library construction are wasted.

However, using the automated protein design techniques outlined below, virtual libraries of protein sequences that are significantly larger than experimental libraries can be generated. 10 ⁸⁰ The following candidate sequences are computationally screened obtained, those satisfying the advantageous design criteria for stable functional protein may be readily selected. The experimental library can then be made more efficient and the limitations of the random protein library can be overcome by creating an experimental library consisting of suitable candidates found from the virtual library screen.

There are two main benefits from virtual library screening. (
1) Automatic protein design creates a sequence candidate list that is advantageous for satisfying the design criteria. It also shows which positions in the sequence are easily changed and which positions are likely to be associated with disruption of protein stability and functionality. Experimental random libraries can be created in which only non-disruptive sequence positions that can be easily modified are randomized. (2) The amino acid diversity at these positions can be limited to what automatic design indicates is compatible with these positions. That is, limiting the number of randomized positions and the number of possibilities at these positions reduces the number of wasted sequences produced in an experimental library and relates to the discovery of sequences with useful properties. The chance of success increases.

Furthermore, by screening a very large library of variants by computer processing, the diversity of protein sequences that can be screened is increased,
Significant improvements in protein function are provided. Furthermore, the expense and difficulty associated with genetic engineering of proteins is reduced because fewer variants need to be tested experimentally to screen for a given library size. By prescreening protein libraries using computational methods, combining the computational features of speed and efficiency with the capabilities of experimental library screening, suitable computer models and proteins with unclear structure-function relationships A new activity is created in.

Similarly, the novel method of making secondary libraries derived from computationally very large mutant libraries allows rapid testing of large numbers of computer designed sequences.

Moreover, as is fully outlined below, the libraries can be ubiquitous in any number of orientations so that secondary libraries with different focus can be created. For example, domains, subsets of residues, active or binding sites, surface residues, etc. may all be modified or held constant as desired.

Accordingly, the present invention provides a method for generating a secondary library of skeletal protein variants. The term "protein" as used herein includes at least two amino acids linked together by peptide bonds. As used herein, protein includes proteins, oligopeptides and peptides. Peptidyl groups are natural amino acid and peptide bonds, or synthetic peptidomimetic structures, ie, “analogs” such as peptoids (Simon et al., PNAS USA 89 (20): 9367 (1992).
) Reference). Amino acids can be either naturally occurring or non-naturally occurring. As is recognized in the art, any structure of which a series of rotamers is known or can be produced can be used as the amino acid. The side chain is (R
) Or (S) configuration. In a preferred embodiment, the amino acid is (S)
Or L-configuration.

The scaffold protein can include any protein whose three-dimensional structure is known or can be generated, ie for which there are three-dimensional coordinates for each atom of the protein. Generally, this can be measured using X-ray crystallographic techniques, NMR techniques, novel modeling, homology modeling and the like. Generally, when using an X-ray structure, a structure with a resolution of 2 Å or higher is preferable, but not always required.

The skeletal proteins include bacteria, fungi, extremophiles such as bacteria, insects, fish, animals (
Enzymes, especially from mammals and especially humans) and birds are all possible and can be derived from any organism, including prokaryotes and eukaryotes.

Thus, as used herein, “skeletal protein” means a protein for which a secondary library of variants is desired. As will be appreciated in the art, all scaffold proteins are useful in the present invention. Specifically included within the definition of "protein" are fragments and domains of known proteins, such as functional domains, such as enzymatic domains, binding domains, and small fragments, such as turns and loops. That is, a portion of the protein may be used as well. Furthermore, "protein" as used herein includes proteins, oligopeptides and peptides. In addition, protein variants, ie non-natural protein analog structures, may also be used.

Suitable proteins include industrial and pharmaceutical proteins such as ligands, cell surface receptors, antigens, antibodies, cytokines, hormones, transcription factors, signal modules,
Includes, but is not limited to, 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, tautomers, or mutases, transferases, kinases, oxidoreductases and phosphatases. Suitable enzymes are listed in the Swiss-Prot Enzyme Database. Suitable protein scaffolds include all those found in protein databases compiled and provided by Research Collaboration for Structural Bioinformatics (RCSB, formerly Brookhaven National Love). However, it is not limited thereto.

Specifically, preferred scaffold proteins include those having a known structure (including mutants), for example, cytokines (IL-1ra (+ receptor complex), IL-1 (receptor alone), IL-1a, IL-1b (including mutant and / or receptor complex), IL-2, IL-3, IL-4, IL-5, IL-6, IL-8, IL-1
0, IFN-β, INF-γ, 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 chemoinduction protein 1, macrophage migration inhibition Factor, human glycosylation inhibitor, human rantes, human macrophage inflammatory protein 1 beta, human growth hormone, leukemia inhibitory factor, human melanoma growth stimulating 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, trans Forming growth factor B3, trans Omingu growth factor A, vascular endothelial cell growth factor (VE
GF), acidic fibroblast growth factor, basic fibroblast growth factor, endothelial cell growth factor, nerve growth factor, brain-derived neurotrophic factor, ciliary neurotrophic factor, platelet-derived growth factor, human hepatocyte growth factor , Glial-derived neurotrophic factor (and PDB1 / 1
55 cytokines in 2/99)), erythropoietin, other extracellular signaling moieties such as, but not limited to, hedgehog sonic, hedgehog desert, hedgehog indian, hCG. Coagulation factors such as, but not limited to, TPA and Factor VIIa; transcription factors such as, but not limited to, p53, p53 tetramerization domain, Zn fingers (1 of which
More than two have structure), homeodomains (of which eight have structure), leucine zippers (of which four have structure); antibodies, such as but not limited to cFv; viral proteins such as, but not limited to But not to hemagglutinin tetramerization domain and hiv Gp41 ectodomain (fusion domain), intracellular signal modules such as, but not limited to, SH2 domain (8 of which are
Structures are known), SH3 domains (11 of which have structure), and plexin homology domains; receptors such as, but not limited to, the extracellular region of the 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 / ILra complex, IL-4 receptor, INF-γ receptor Alpha chains, MHC I, MHC II, T cell receptors, insulin receptors, insulin receptor tyrosine kinases and human growth hormone receptors include, but are not limited to.

Once the scaffold protein is selected, the primary library is created using computer processing. In general, the goal of computer processing is to measure a set of optimized protein sequences. By "optimized protein sequence" herein is meant the sequence that best fits the computational mathematical equation. It is recognized in the art that the globally optimized sequence is the one that best fits the equation (eg, when using PDA, the globally optimized sequence is the one that best fits Equation 1 below). , That is, the array with the lowest energy of any possible array. However, there are quite a few sequences that have low, but not global, minimum energies.

That is, as used herein, a “primary library” is a collection of optimized sequences, generally in rank ordered list form. In theory, all possible sequences of proteins can be evaluated. However, at this time 10 ¹³ sequences is a practical limit. That is, generally some subset of all possible sequences is used as the primary library. In general, the top 10 ³ to 10 ¹³ sequences are selected as the primary library. The cutoff for inclusion in the ranked list of the primary library can be done in a variety of ways. For example, the cutoff may simply be any exclusion point. The top 10 ⁵ sequences may comprise the primary library. Alternatively, all sequences scoring within certain bounds of the global optimum can be used.
For example, all sequences with a global optimum of 10 kcal / mol can be used as a primary library. This method has the advantage of measuring inclusion relationships using a direct measure of fidelity to the three-dimensional structure. Using this method, it can be ensured that library mutations are not limited to the positions with the lowest energy gaps between different mutations. Alternatively, the cutoff may be enhanced when a predetermined number of mutations per position is reached. When the list of sequences listed in rank order is extended and the library is expanded, more mutations are identified per position. Alternatively, the total number of sequences identified by recombination of all mutations can be used as a cutoff criterion for the primary sequence library. A preferred value for the total number of sequences is in the range of 100 to 10 ²⁰ , a particularly preferred value is in the range of 1000 to 10 ¹³ , and a particularly preferred value is in the range of 1000 to 10 ⁷ . Alternatively, the first occurrence in the list of pre-specified unwanted residues can be used as a cutoff criterion. For example, the first hydrophilic residue at the core position limits the list. Also, these methods have been described in conjunction with the size limitations of primary libraries, and using these same techniques, a cutoff for inclusion in secondary libraries can be formulated as well. It should be noted.

Thus, the present invention provides a method for making a primary library containing a rank ordered list of sequences, generally in terms of theoretical quantitative stability, as detailed below. Using the methods of creating primary libraries to optimize conformational stability can stabilize the active site conformation of the enzyme and also improve its activity. Similarly, stabilizing the ligand-receptor complex or the enzyme-substrate complex also improves binding affinity.

The primary library can be created in various ways. Essentially any method can be used that results in a relative ranking of the possible sequences of the protein based on measurable stability parameters. It will be recognized in the art that any of the methods described herein or known in the art may be used alone or in combination with other methods.

In a preferred embodiment, the scaffold protein is an enzyme and a highly accurate electrostatic model can be used to evaluate enzyme active site residues to improve enzyme active site libraries (Warshel, computer Modeling of Chemical Reactions. in Enzymes
and Solutions, Willie and Sons, New York (1991),
It is made a part of this specification by clearly showing the source). These elaborate models can evaluate relative energies of sequences with high accuracy, but focus on computer manipulation.

Similarly, sequences can be computationally screened by individually calculating variant sequence scores using molecular mechanics calculations and compiling a ranked list.

In a preferred embodiment, the use of residue pair potentials allows the sequence to be evaluated during computational screening (Miyazawa et al. Macromolecules.
18 (3): 534-552 (1985), which is incorporated herein by reference.

In a preferred embodiment, sequence profile evaluation (Bowie et al., Science 253 (
5016): 164-70 (1991) incorporated herein by reference.
And / or the potential of mean force (Hendlich et al., J. Mol. Biol. 216 (1)
167-180 (1990), also incorporated herein by reference), to evaluate the sequence. These methods can function to screen for fidelity to protein structure in order to assess pairing between sequences and 3D protein structures. By evaluating sequences using different scoring functions, different regions of sequence space can be sampled on a computer screen.

In addition, the scoring function can be used to screen for sequences that create metal or cofactor binding sites in proteins (Hellinga et al.,
Fold Des. 3 (1): R1-8 (1998), incorporated herein by reference. Similarly, the scoring function can be used to screen for sequences that produce disulfide bonds in proteins. By testing these potentials, the protein structure is specifically modified to introduce new structural motifs.

Further, by using sequence and / or structure alignment programs, primary libraries can be created. For example, a sequence alignment can be created by performing a structural alignment of structurally related proteins (Oren
go et al., Structure 5 (8): 1093-108 (1997), Holm et al., Nucleic.
Acid Res. 26 (1): 316-9 (1998), both incorporated herein by reference. The observed sequence mutations can then be measured by examining these sequence alignments.

Similarly, sequence homology-based alignment methods can be used to generate sequence alignments of proteins related to the target structure (Altschul et al., J. Mol. Biol.
.215 (3): 403 (1990), which is made a part of this specification by citation.)
. The observed sequence mutations are then measured by examining these sequence alignments. The primary library is identified by representing these sequence variations.

These sequence variations are represented and secondary libraries can be identified from them as follows. Alternatively, allowed sequence mutations may be used to identify possible amino acids at each position during computational screening. Another mutation is
Biasing the scores for amino acids present in a sequence alignment,
Although it increases the likelihood that they will be found during computer screening, other amino acids may still be considered. This bias results in a focused primary library, but does not overlook amino acids that are not found in the alignment.

Similarly, as outlined above, other computational methods are known, eg sequence profiling (Bowie and Eisenberg, Science 253 (5016).
): 164-70 (1991)), rotamer library selection method (Dahiyat and Mayo, Protein Sci5 (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 USA92 (18): 8408-841 (1995); Kono et al., Proteins.
: Structure, Function and Genetics 19: 244-255 (1994); Hell
inga and Richards, PNAS USA 91: 5803-5807 (1994)) and residue pair potential (Jones, Protein Science 3: 567-574 (199).
4)), but not limited thereto, all of which are incorporated herein by reference.

In a preferred embodiment, the computer processing method used to create the primary library is USSN 60/061097, 60/043464, 60/054.
678, 09/127926 and PCT US98 / 07254, all of which are incorporated herein by reference, for protein design automation (PDA). Briefly, a PDA can be described as follows. The known protein structure is used as a starting point. The residues to be optimized are then identified, which may be the entire sequence or a subset (s) thereof. Then
The side chains at variable positions are removed. The structure consisting of the generated protein backbone and the remaining side chains is called the template. Each variable residue position is then preferably classified as a core residue, surface residue or boundary residue. Each classification identifies a subset of possible amino acid residues with respect to position (eg, core residues are generally selected from a series of hydrophobic residues, surface residues are generally selected from hydrophilic residues, And the boundary residue can be either). Each amino acid can be represented by an independent group of all allowed conformers of each side chain, called rotamers. That is, in order to reach the optimal sequence for the skeleton, all possible rotamer sequences must be screened, in which case each skeletal position has each amino acid in its possible rotamer state, Or it may be occupied by a subset of amino acids, ie a subset of rotamers.

Two sets of interactions are then calculated for each rotamer at all positions. Interaction of rotamer side chains with all or part of the skeleton (also referred to as "singles" energy, rotamer / template or rotamer / skeletal energy), and rotamer side chains with all other positions or Interactions with all other possible rotamers in other subsets of positions (also called "doubles" energies, rotamers / rotomer energies). The energies of each of these interactions are calculated through the use of various scoring functions such as van der Waals force energies, hydrogen bond energies, secondary structure tendency energies, surface area solvation energies, and electrostatics. included. That is, the total energy of interaction of each rotamer with both the backbone and other rotamers is calculated and stored in matrix form.

Because the set of rotamers has independent properties, the number of rotamer sequences tested can be calculated simply. For a skeleton of length n with m possible rotamers per position, there are m ⁿ possible rotamer sequences and the number increases exponentially with the sequence length, so immediate calculation Is impractical or impossible. Therefore, in order to solve this combinatorial search problem, "dead end elimination" (
DEE) Perform calculations. DEE calculations show that if the lowest overall interaction of the first rotamer is still better than the highest overall interaction of the second rotamer, then the second rotamer is part of the global optimal solution. It is based on the fact that it cannot be. Since the energies of all rotamers have already been calculated, the DEE method only requires a total over the length of the sequence to test and eliminate rotamers, which considerably accelerates the calculation. DEE can be re-run while comparing pairs of rotamers or combinations of rotamers, ultimately determining a single sequence that represents the global optimum energy.

Once the global solution is found, a Monte Carlo search can be performed to create a rank ordered list of sequences in the neighborhood of the DEE solution. Starting from the DEE answer,
Change random positions to other rotamers and calculate new sequence energies. If the new sequence meets the criteria for acceptability, it is used as a starting point for another jump. After a predetermined number of jumps, create a rank ordered list of arrays.

According to USS 09/127926, the protein backbone (along the direction of the vector from α-carbon to β-carbon, nitrogen, carbonyl carbon, α-carbon And carbonyl oxygen (in the case of native proteins) can be modified prior to computer analysis by altering a series of parameters called super-secondary structural parameters.

Once the protein structural scaffold has been created (with modifications as outlined above) and entered into the computer, an explicit hydrogen is added if not contained within the structure (
For example, hydrogen must be added if the structure was generated by X-ray crystallography). After hydrogenation, if the energy of the structure is minimized, hydrogen and other atoms,
The bond angle and bond length are relaxed. In a preferred embodiment, this is a conjugate gradient minimization of atomic coordinate positions consisting of several steps (Mayo et al., J. Phys. Chem. 94:
8897 (1990)) to minimize the field of Dreiding forces without any static electricity. Generally, about 10 to about 2
50 steps are preferred, with about 50 being most preferred.

The protein backbone structure comprises at least one variable residue position. As known in the art, protein residues or amino acids are generally numbered consecutively starting from the N-terminus of the protein. That is, a protein having methionine at its N-terminus must have a methionine at position 1 of a residue or an amino acid, and the next residue is said to be located at positions 2, 3, 4 and so on. At each position, the wild-type (ie native) protein can have at least one of at least 20 amino acids in any number of rotamers. The term "variable residue position" as used herein means that the amino acid position of the designed protein is not fixed as a residue or rotamer, generally as a wild-type residue or rotamer, by the design method.

In a preferred embodiment, all residue positions of the protein are variable. That is, any amino acid side chain can be modified by the method of the present invention. This is particularly desirable in the case of small proteins, but large proteins can be designed in the present invention as well. There is no theoretical limit to the length of a protein that can be designed in this way, and there are practical computer limits.

In another preferred embodiment, only some of the protein residue positions are variable and the rest are “
"Fixed", that is, they are identified in the three-dimensional structure as being in a set conformation. In some embodiments, a fixed position may remain in its original conformation (whether or not it correlates with the specific rotamer of the rotamer library being used). Can be). Alternatively, the residue may be fixed as a non-wild type residue. For example, when known site-directed mutagenesis techniques indicate that a particular residue is desirable (eg, to delete a proteolytic site or modify the substrate specificity of an enzyme), the residue is fixed as a particular amino acid. obtain. Alternatively, the method of the invention may be used to newly assess mutations, as discussed below. In another preferred embodiment, the fixed position is
Can 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 1, or generally 0.1% to 99.9% of the total number of residues. Thus, for example, it is possible to change only a few residues (or only one) or most of the residues, if there are any possibilities there.

In preferred embodiments, the residues that can be fixed include, but are not limited to, structural or biologically functional residues. For example, residues known to be important for biological activity, such as 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.), biological Residues that form phosphorylation or glycosylation sites that are critical for function, or structurally important residues, such as disulfide bridges, metal binding sites, critical hydrogen bonding residues, critical for backbone conformation Residues such as proline or glycine, residues critical for interaction packing, etc. may all be fixed or "float" in one conformation or as a single rotamer.

Similarly, residues that may be selected as variable residues may have undesirable biological properties, such as sensitivity to proteolytic, dimerization or aggregation sites, glycosylation sites that may induce an immune response, preferably. No binding activity, undesired allostery, binding may confer conserved but undesired enzymatic activity, and the like.

In a preferred embodiment, each variable position is classified as a core, surface or border residue position, but optionally by setting the variable position to glycine, as described below, the backbone chain Can be minimized. Any combination of core, surface and boundary locations may be used. Core, surface and boundary residues; core and surface residues; core and boundary residues, and surface and boundary residues, and core residues alone, surface residues alone or boundary residues alone.

Classification of residue positions as cores, surfaces or boundaries can be done in several ways, as is recognized in the art. In a preferred embodiment, the classification is performed by scanning an image of the original protein backbone structure including side chains, and the type is designated based on the subjective evaluation of one skilled in the field of protein modeling. Alternatively, the preferred embodiment is U.
S.S.N. 60/061097, 60/043464, 60/054678, 0
As outlined in 9/127926 and PCT US98 / 07254,
Utilizes an assessment of the orientation of the Cα-Cβ vector relative to a solvent-sensitive surface that was computerized with only template Cα atoms.

Once each variable position is classified as a core, surface or boundary, a series of amino acid side chains, or series of rotamers, is assigned to each position. That is, the program selects a set of possible amino acid side chains that are considered for consideration at a particular position. Following that, once the possible amino acid side chains have been selected, a series of rotamers evaluated at a particular position can be determined. That is, the core residue is generally selected from the group of hydrophobic residues consisting of alanine, valine, isoleucine, leucine, phenylalanine, tyrosine, tryptophan and methionine (in some embodiments, α scaling of van der Waals scorelink function below. When the factor is low, methionine may be removed from the setup), and the rotamers shown for each core position potentially encompass rotamers for these 8 amino acid side chains (backbone independent library). All rotamers when using, and subsets when using rotamer-dependent scaffolds). Similarly, the surface position is generally selected from the group of hydrophilic residues consisting of alanine, serine, threonine, aspartic acid, asparagine, glutamine, glutamic acid, arginine, lysine and histidine. That is, the rotamers shown for each surface position include rotamers for these 10 residues. Finally, the boundary positions are generally alanine, serine,
It is selected from threonine, aspartic acid, asparagine, glutamine, glutamic acid, arginine, lysine, histidine, valine, isoleucine, leucine, phenylalanine, tyrosine, tryptophan and methionine. That is, the rotamers shown for each boundary position encompass all rotamers for these 17 residues (assuming cysteine, glycine and proline are not used, but they can be used). Further, in some preferred embodiments, a series of 18 natural amino acids (all except cysteine and proline, which are known to be particularly destructive) are used.

That is, as will be appreciated by those skilled in the art, classifying residue positions has the advantage of computerization as it reduces the number of calculations. It should also be noted that there may be situations where the core, border and surface residues are modified from those described above. For example, in some circumstances one or more amino acids may be added or subtracted from the set of allowed amino acids. For example, some proteins that dimerize or multimerize or have a ligand binding site include
Some may include hydrophobic surface residues and the like. In addition, residues that do not permit helical "capping" or favorable interaction with the alpha helix dipole can be subtracted from the set of recognized residues. This amino acid group modification can be performed 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, so rotamers for these side chains are not used. However, in a preferred embodiment, the variable residue position is at an angle φ greater than 0 ° (ie, 1) the carbonyl carbon of the above amino acid, 2) the nitrogen atom of the current residue, 3) the α carbon of the current residue, and 4) the current carbon atom. When having a dihedral angle specified by the carbonyl carbon of the residue, the position is set to glycine to minimize backbone strain.

Once a group of potential rotamers has been assigned to each variable residue position, USS N0
Processing proceeds as outlined in 9/127926 and PCT US98 / 07254. This processing step requires analysis of rotamer interactions and rotamer-protein backbone interactions, thereby producing optimized protein sequences. In extreme simplification, processing involves first calculating the interaction energy of the rotamer with respect to the backbone itself or other rotamers by using some scoring function. Preferred PD
A scoring function includes van der Waals potential scoring function,
There are, but not limited to, hydrogen bond potential scoring functions, atomic solvation scoring functions, secondary structure tendency scoring functions and electrostatic scoring functions. As reported further below, each position is evaluated by using at least one scoring function, which is useful for position classification or other considerations, such as alpha helical dipoles. Can vary with different interactions. As outlined below, the total energy used in the calculation is generally the sum of the energy of each scoring function used at a particular position, as shown in Equation 1: Equation 1 E _total = nE _vdw + nE _as + nE _h-bond + nE _as + nE _elec

In Equation 1, the total energy is van der Waals potential energy (E _vdw ), atomic solvation energy (E _as ), hydrogen bond energy (E _h- bond).
_If), the secondary structure energy (E _ss) and electrostatic interaction energy (E _elec)
Is the sum of The word n is 0 or 1 depending on whether this word should be considered for a particular residue position.

U.S.S.N. 60/061097, 60/043464, 60/054678
, 09/127926 and PCT US98 / 07254, either alone or in combination, any combination of these scoring functions may be used. Once the scoring function to be used has been defined for each variable position, the preferred first step in computer analysis measures the interaction of each possible rotamer with all or part of the rest of the protein. That is,
The energy of each possible rotamer interaction with the backbone or other rotamer at each variable residue position, as measured by one or more of the scoring functions, is calculated. In a preferred embodiment, the interaction of each rotamer with all of the rest of the protein, both the entire template and all other rotamers, is carried out. However, as outlined above, it is not necessary to consider all of the protein in some cases, as it is only possible to model a portion of the protein, eg the domain of a large protein.

In a preferred embodiment, the first step of computer processing is performed by calculating two sets of interactions for each rotamer at all positions. That is, the rotamer side chains interact with the template or backbone (the "singles" energy), whether or not its position is altered or floated, and the rotamer side chains and all other possible positions at all other positions. Interaction with rotamers (“doubles” energy). It should be understood that the scaffold in this case includes both the atoms of the protein structural scaffold and the fixed residues, if any, in which case the fixed residues are defined as a particular conformation of the amino acid.

That is, the “single” (rotamer / template) energy is the interaction of all possible rotamers with the backbone at all variable residue positions using some or all of the scoring function. Is calculated with respect to. That is, for the hydrogen bond scoring function, all rotamer hydrogen atoms and backbone all hydrogen bond atoms are evaluated and E _HB is calculated for each possible rotamer at all variable positions. Similarly, for the van der Waals scoring function, all atoms of the rotamer are compared to all atoms of the template (generally excluding the backbone atoms of its own residue) and at all variable residue positions. Calculate E _vdW for each possible rotamer. In addition, van der Waals energies are generally not calculated when atoms are bound by three or fewer bonds. In the case of the atomic solvation scoring function, the rotamer surface is measured relative to the template surface and E _as is calculated for each possible rotamer at every variable residue position. Also, the secondary structure tendency scoring function is also considered as a singles energy, so the total singles energy may include the E _ss limit point. It is recognized in the art that many of these energy limits approach zero due to the physical distance between rotamers and template positions. That is, the further the two parts are separated, the lower the energy.

For calculation of “doubles” energy (rotomer / rotomer), compare the interaction energy of each possible rotamer with all possible rotamers at all other variable residue positions. To do. That is, the “doubles” energy uses all or all of the possible rotamers at all variable residue positions and all of the possible rotamers at all variable residue positions using some or all of the scoring function. Calculated in terms of interactions with the body. That is, for the hydrogen bond scoring function, all hydrogen-bonding atoms of the first rotamer and all possible hydrogen-bonding atoms of the second rotamer are evaluated, and E _HB is determined by each possible hydrogen bond at two variable positions. Calculated for rotamer pairs. Similarly, for the van der Waals scoring function, all atoms of the first rotamer are compared to all atoms of all possible second rotamers, and each possible rotation at all two variable residue positions. Calculate E _vdW for the isomer pair. In the case of the atomic solvation scoring function, the surface of the first rotamer was measured against the surface of all possible second rotamers, and each possible rotamer pair at all two variable residue positions. E _as is calculated for. The secondary structure tendency scoring function need not be implemented as "doubles" energy as it is considered a component of "singles" energy. According to what is recognized in the industry,
Many of these doubles energy limit points approach zero due to the physical distance between the first and second rotamers. That is, the further the two parts are separated, the lower the energy.

Once the singles and doubles energies have been calculated and stored, the second stage of computer processing may take place. U.S.S.N. 09/127926 and P
As outlined in CT US98 / 07254, the preferred embodiment utilizes a dead end elimination (DEE) stage, and preferably a Monte Carlo stage.

As a result of computer processing, a series of optimized protein sequences is created. These optimized protein sequences are generally, but not always, significantly different from the scaffold-extracted wild-type sequences. That is, each optimized protein sequence preferably comprises at least about 5-10% amino acid variants from the starting or wild type sequences, with at least about 15-20% mutations preferred, and at least about 30% mutations particularly preferred. preferable.

The cutoffs for the primary library are then enhanced to produce a series of primary sequences that form the primary library. As outlined above, this can be done in a variety of ways, eg by any cutoff, energy restriction, or when a certain number of residue positions have been modified. In general, the size of the primary library will vary depending on the size of the protein, the number of altered residues, the computational method used, the cutoff applied and the discretion of the user.
Generally, to prepare a solid secondary library, the primary library is
It is preferable that the reasonable array space be large enough to allow random sampling. That is, the range of the primary library is preferably about 50 to about 10 ^13, more preferably about 1000 to about 10 ⁷ , and even more preferably about 1000 to about 100,000.

In a preferred embodiment, although this is not a requirement, the primary library comprises the global optimal sequence in its optimal configuration, ie the optimal rotamer at each variable position. That is, computer processing is performed until the simulation program converges to a single array that is the global optimum result. In a preferred embodiment, the primary library contains at least two optimized protein sequences. Thus, for example, the computational step removes some unfavorable combinations but can be stopped prior to convergence, providing a library of sequences of global optimal results. In addition, further computer analysis can be performed on the library, eg using different methods, to further delete sequences or rank them differently. Alternatively, USS N. 60/061097, 60/043464
, 60/054678, 09/127926 and PCT US98 / 0725.
According to what is more fully explained in 4, global optimal results can be reached,
Further computational processing can then be performed to create additional optimized sequences in the neighborhood of the global optimal result.

Further, in some embodiments, a primary library sequence without a cutoff is included in the primary library. This may be desirable in some situations to evaluate primary library generation methods, to be used as controls or comparison criteria, or to sample additional sequence space. For example, in a preferred embodiment, the wild type sequence is included.

It should also be noted that combinations of different primary libraries can be made. For example, positions in proteins that display large amounts of mutational diversity in computer screens can be fixed and reconstituted with different primary libraries, as outlined below. The rank-ordered list of the same length as the first shows diversity with little change in position first. Combining the variants from the first primary library with the variants from the second primary library may provide a combinatorial library with lower computer outlay than making a very long ranked list. This method can be particularly useful for sampling sequence diversity in both low energy gaps that easily change surface position and high energy gaps that rarely change core position.

Thus, the invention provides a primary library containing a ranked list of sequences. In one aspect, all or part of the primary library serves as a secondary library. That is, the cutoffs are applied to the primary sequences and these sequences are used as a secondary library without further manipulation or recombination. Library members may be made as outlined below, for example by direct synthesis or by constructing nucleic acids encoding the library members, expressing them in a suitable host, and then optionally screening. ..

In a preferred embodiment, a secondary library is created by using a primary library of scaffold proteins. It is recognized in the art that the secondary library may be a subset of the primary library or may contain novel library members, ie sequences not found in the primary library. That is, generally, a mutation position and / or the amino acid residues at the mutation position can be recombined in some manner to form a new library that utilizes the sequence mutations found in the primary library. That is, once "hot spots" or important mutation positions and / or residues have been identified, these positions are recombined in a novel manner to generate new sequences and to generate secondary libraries. Can be formed. That is, in a preferred embodiment, the secondary library comprises a sequence of at least one component not found in the primary library, and preferably a plurality of the above sequences.

In a preferred embodiment, a secondary library is created by representing the mutated amino acid positions from the reference sequence. The reference sequence may be arbitrarily selected, or is preferably selected as a wild type sequence or a global optimal sequence,
The latter is preferred. That is, each amino acid position that mutates in the primary library is tabulated. Of course, if some positions are fixed by the initial computer analysis, the variable positions of the secondary library include these first variable positions themselves or a subset of these first variable positions. Thus, assuming a protein of 100 amino acids, the first computer screen can mutate all 100 positions. However, only 25 positions can be mutated due to the cutoff in the primary library. Alternatively, assuming the same 100 amino acid protein,
The first computer screen was able to mutate only 25 positions and fix the other 75. As a result, only 12 of 25 can be altered in the cutoff primary library. A secondary library can then be formed by recombining these primary library positions, and all possible combinations of these variable positions form the secondary library. It should be noted that the non-variable position is set to the reference sequence position.

The formation of a secondary library using this method is two general methods; either any amino acid may be present at all of the variable positions, or a subset of amino acids may be present at each position. You can do it in

In a preferred embodiment, each variable position identified in this primary library can be any amino acid residue. That is, once the variable positions are identified, a secondary library containing any combination of any amino acids at each variable position is created. In a preferred embodiment, a subset of amino acid groups is selected. The subset at any position may be user-selected or a collection of amino acid residue groups generated in the primary screen. That is, assuming that core residue 25 is variable, and if this primary screen gives 5 possible distinct amino acids for this position, the user is A set of groups (eg hydrophobic residues) may be selected or the user may construct this set by selecting these 5 different amino acids generated in the primary screen. Alternatively, a combination of these techniques can be employed, in which case the set of identified residues will be expanded manually. For example, in some embodiments, a number less than the number of amino acid residues is selected; for example, only 3 out of 5 can be selected. Alternatively, this set is expanded manually; for example, if two different hydrophobic residues are selected in the calculation, additional selections may be added.

In addition, this can be done by analyzing the primary library to determine which amino acid positions in the scaffold protein have a high mutation frequency and which positions have a low mutation frequency. . This secondary library can be generated by keeping the positions that do not have the number of mutations at a certain frequency or higher as they are, while randomizing the amino acid groups at the positions that have a high number of mutations. For example, if the position is a position that has 20% or less, preferably 10% or less mutations, it may be retained as a reference sequence position.

In the preferred embodiment, a probability distribution table is created. In this embodiment, the frequency of each amino acid residue at each variable position is identified. Each frequency can be a threshold when all mutation frequencies below the cutoff are set to zero. This cut-off is preferably 1%, 2%, 5%, 10% or 20%, particularly preferably 10%.
%. These frequencies are then built into the secondary library. That is, as described above, these variable positions are assembled 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, one variable position with 5 possible residues would be 20% of the protein containing that variable position with the first possible residue, the second It will have 20%, and below.
However, with frequency-based secondary libraries, 10%, 15%,
One variable position with 5 possible residues, each having a frequency of 25%, 30% and 20%, is a protein containing the variable position with the first possible residue.
0%, 15% of proteins with the second residue, 25% of the third, and so on. As will be apparent to those of skill in the art, the actual frequency may vary depending on the method actually used to produce the protein; for example, the exact frequency is possible when the proteins are synthesized. However, when using the frequency-based primer system outlined below, the actual frequency at each position is variable as described below.

As will be apparent, secondary libraries created by rearranging each variable position and / or each residue at that variable position are not included in the rank-ordered list. In some embodiments, the entire list could be properly prepared and tested.
Alternatively, in a preferred embodiment, this secondary library is also in the form of a rank ordered list. This can be done for several reasons, including that the size of the secondary library is still too large to generate experimentally, or a predictable purpose. This can be done in several ways. In some embodiments, the PDA scoring function of ranking library members is used to rank secondary libraries. Alternatively, statistical methods can be used. For example, the secondary library may be ranked by frequency score; that is, proteins that contain the majority of high frequency residues may be ranked high. This can also be done by adding or multiplying the frequency of generating the numerical score at each variable position. Similarly, weigh each different position of the secondary library,
Each protein may then be scored; for example, it may optionally be ranked by those containing certain various residues.

Secondary libraries, as outlined herein, can generally be generated in two ways. The first is the computer processing method as described above. In the method the primary library is further handled in a computational manner, eg by reconstituting potential mutation positions and / or amino acid residues at each mutation position. It may be ranked as described above. This computationally derived secondary library can be generated experimentally by synthesizing library members or the nucleic acids encoding them, as described in detail below. Alternatively, this secondary library is generated empirically; that is, the combination is empirically generated using nucleic acid reconstitution techniques. This can be done in various ways as described below.

In a preferred embodiment, different protein members of the secondary library can be chemically synthesized. This is preferably 150 when the designed protein is short.
It is particularly useful when the amino acid length is less than or equal to the amino acid length, more preferably 100 amino acid length or less, and particularly preferably 50 amino acid length or less (a protein, as known in the art, chemically. Longer proteins may be prepared either enzymatically or enzymatically). Wilken, for example, which is explicitly incorporated by reference
See et al, Curr. Opin. Biotevhnol. 9: 412-26 (1998).

In a preferred embodiment, a secondary library sequence encodes a member sequence and is cloned into a host cell and expressed if desired, particularly for longer proteins, or for proteins for which large samples are desired. It is used to create nucleic acids such as DNA that can be analyzed. Thus, nucleic acids and especially D
NA can be prepared to encode a member protein sequence. This can be done using known procedures. The choice of codon, appropriate expression vector, and appropriate host will depend on a variety of factors and can be readily optimized as needed.

In a preferred embodiment, multiple PCR reactions with pooled oligonucleotides are performed, as is generally shown in FIG. In this embodiment, overlapping oligonucleotides corresponding to the full length gene are synthesized. These oligonucleotides may also represent all of the different amino acids at each mutation position or subset. In a preferred embodiment, these oligonucleotides are pooled in equal proportions and multiple PCR reactions are performed to create full-length sequences containing the combination of mutations defined in the secondary library.

In a preferred embodiment, each relative amount of the heterologous oligonucleotide group corresponding to the probability distribution table is added. These multiple PCR reactions thus produce a set of full-length sequences having each desired combination of mutations in each desired proportion. The total number of oligonucleotides required is a function of the number of positions to mutate and the number of possible mutations at those positions. (Number of oligos for each fixed position) + M1 + M2 + M3 + ... Mn = (total number of oligos required), where Mn is the number of possible mutations at the n-position of the sequence.

In a preferred embodiment, each overlapping oligonucleotide contains only one position to be mutated; in another embodiment, the mutation sites are in close proximity to each other to allow this. Using multiple mutations for each oligonucleotide, allows for all possible complete recombination. That is, each oligo may contain codons for a single position to be mutated or for one or more positions to be mutated. These multiple positions to be mutated must be contiguous within the sequence to prevent the length of the oligo from failing. By introducing or eliminating oligonucleotides encoding a particular mutation combination at multiple mutation positions on an oligonucleotide, the combination can be introduced or excluded in the library. The total number of oligonucleotides required increases when multiple mutable positions are encoded by a single oligonucleotide. Each annealed region is the unchanged part, i.e.
It is a site having the sequence of the reference sequence.

Oligonucleotides with or without codon insertions can be used to create libraries expressing proteins of different lengths. In particular, computerized sequence screening for insertions or deletions can generate secondary libraries that define proteins of different lengths, which proteins are expressed by a library of pooled oligonucleotides of different lengths. Can be expressed.

In a preferred embodiment, error-prone PCR is performed to generate a secondary library. US Patent No. 5605793, 58112
38, and 5830721, all of which are incorporated herein by reference. This can be done on the optimal sequence or top member group of the library. In this embodiment, the gene for the optimal sequence found in the computational screening of the primary library can be synthesized. Then, error-prone PCR is performed on the optimal sequence gene in the presence of oligonucleotides encoding mutations (biased oligonucleotides) at mutation positions in the secondary library. it can. The addition of oligonucleotides creates a bias that favors the introduction of mutations into the secondary library. Alternatively, only oligonucleotides for certain mutations can be used to bias the library.

In a preferred embodiment, error-prone PCR gene truncation is performed on the gene for optimal sequence in the presence of biased oligonucleotides and is found in a secondary library. A DNA sequence library can be created that reflects the proportion of each mutation that is made. The selection of biased oligonucleotides can be performed in a variety of ways; they can be selected based on their frequency, ie, oligonucleotides encoding high mutation frequency loci can be used; Alternatively, oligonucleotides containing the most mutated loci can be used, thus increasing diversity; if the secondary library is ranked, then some of the top-scoring loci are biased. Can be used to generate a group of oligonucleotides that are active; random position groups can also be selected; only a few top scoring groups and a few low scoring groups can be selected; and so on. What is important is to generate a new sequence based on the preferred mutation position and sequence.

In a preferred embodiment, the secondary library can be skillfully reengineered to form another secondary library by computer processing. For example, any secondary library sequence can be selected for the second round of PDA by freezing or fixing some or all of the changed positions in the first secondary library. Alternatively, only the changes found in the last probability distribution table are allowed. Alternatively, the stringency of the probability table can be modified by either increasing or decreasing the cutoff for recruitment. Similarly, a secondary library can be experimentally recombined after the first round; eg, the best genes / genes from the first screen are picked and reassembled (as outlined below). Using techniques, multiple PCR, error-prone PCR, shuffling, etc.).
Alternatively, a fragment from one or more good gene (s) to alter the probability at some positions. This biases the search for regions of sequence space found in the first round of computational and experimental screening.

In these methods, as outlined herein, enzymatic activity, stability, solubility, aggregability, binding affinity, binding specificity, substrate specificity, structural integrity, immunity Originating, toxicity, generation of peptide and pseudopeptide libraries, novel antibody CDR '
Any number of protein attributes can be modified, including but not limited to the creation of s, the generation of new DNA, RNA binding, and the like.

Therapeutic proteins utilized in these methods selectively have screened residues in the hydrophobic core group to prevent alterations at the molecular surface of the protein that may elicit an immunogenic response. It should be kept in mind. These therapeutic proteins can also be designed so that the regions surrounding their binding sites correspond to receptors. These regions are, for example, all residues within a certain distance, for example,
It can be defined by including it within 4.5 Å of the binding site residue in the design. This range can vary from 4 to 6Å. This design can help improve the activity and specificity of the enzyme.

In a preferred embodiment, the method of the invention is used to search a virtual library for those sequences that are believed to give a stable conformation for a random group of peptides, but not for known scaffold proteins. To do. As noted above, current interest and interest is focused on screening random peptide libraries to find new binding / modulators. However,
The sequences in these experimental libraries can be randomized only at specific sites or across the sequences. The number of sequences that can be searched in these libraries increases exponentially with the number of randomized positions. Generally speaking, due to physical limitations of the laboratory (equipment size, large number of biopolymer manufacturing costs, etc.), only 10 ¹² -10 ¹⁵ sequences can be included in the library. Other practical considerations often limit the size of libraries to 10 ⁶ or less. These limits reach only 10 amino acid positions. Therefore,
Only sparse sampling of sequences can be sought in experimental sequence libraries for improved proteins or peptides, thus reducing the chance of success and almost certainly missing promising candidates. The changes in these sequences are random, which makes most of the candidates in this library unsuitable, resulting in wasted much of the effort that generated the library.

However, using the automated protein design techniques outlined herein, it is possible to generate much more virtual libraries of protein sequences than experimental libraries. As many as 10 ⁷⁵ candidate sequences can be computationally screened, yet they meet the design criteria of advantage in that stable and functional proteins can be easily selected. An experimental library of promising candidate groups, found by virtual library screening, can then be generated, resulting in a more efficient use of experimental libraries and overcoming the limitations of random protein libraries. Thus, the method of the present invention allows virtual screening of a set of random proteins to attach specific structures for peptides, thus making them unfavorable or unacceptable without preparing and testing those peptides. It is possible to eliminate most of the configuration.

As mentioned above, the following two essential advantages are obtained by virtual library screening: (1) Automated protein design produces a list of promising sequence candidates that meet the design criteria: In addition, indicate which position in the sequence can be easily changed and which position is difficult to change without destroying the stability and function of the protein. The ability to generate an experimental random library, randomized only at non-disruptive sequence positions that can be easily varied. (2) To be able to limit the diversity of amino acids at these positions to the extent that automated protein design at those positions may be compatible.
Thus, by limiting the number of randomized positions and the number of possibilities at those positions, the number of wasted sequences produced in the experimental library is reduced and, in this way, useful properties. Increasing the probability of successfully finding an array with.

For example, the table below lists 10 potential sequence candidates by virtual screening at 12 positions for a protein. It's 9th, 10th and 1st
2nd position is most likely to change without destroying protein function, 9th position, 1
It has been shown that random experimental libraries that randomize positions 0 and 12 will have a higher number of desired sequence fractions. . In addition, a virtual library suggests that position 10 is most compatible with Ile or Phe residues, further limiting the library size and allowing a more complete screening of good sequences.

[0097]

[Table 1]

This automated design method uses physicochemical criteria for screening the sequences and results in sequences that are stable, assembled and, if necessary, preserve function. Using another design criterion, a set of candidates is biased for properties such as charge, solubility, active site properties (polarity, size) or biased to have certain amino acids at certain positions. It can also be generated. That is, either the bioactive candidate substances and candidate nucleic acids are completely randomized or their randomization is biased, for example, in nucleotide / residue frequency (global or position-wise), Randomize. By "randomized" or its grammatical synonyms, it is meant herein that each nucleic acid and peptide is composed of essentially random nucleotides and amino acids, respectively. Therefore, any amino acid residue can be introduced at any position. Designing synthetic processes to generate randomized peptides and / or nucleic acids, allowing all or most possible combinations to be formed into nucleic acid lengths, thus randomized nucleic acid candidates Can be formed into a library of.

In some embodiments, the library is fully randomized without sequence dominance or invariance at any position. In a preferred embodiment, the library is biased. That is, some position in the sequence is kept constant or is selected from a limited number of possibilities. For example, in a preferred embodiment, for example, hydrophobic amino acids, hydrophilic residues, nucleotides or amino acid residues of sterically biasing (small or large) residues are within defined classes, cysteine for crosslinking, SH-3. Randomized towards the creation of proline for domains, serine for phosphorylation sites, threonine, tyrosine or histidine etc. or for purines etc.

In a preferred embodiment, this bias is towards peptides or ligands that interact with known classes of molecules. For example, it is known that many intracellular signalings occur through small peptide domains and through short regions of the polypeptide that interact with other polypeptides. For example, a short region from the HIV-1 envelope cytoplasmic domain has previously been shown to block the action of intracellular calmodulin. The Fas cytoplasmic domain homologous to mastoparan toxin from Wasps can be restricted to short peptide regions with death-induced apoptosis or G-protein inducing function. Magainin, a natural peptide from Xenopus, has strong antitumor and antimicrobial activity. A short peptide fragment of protein kinase C isozyme (βPKC) has been shown to block nuclear translation of βPKC in Xenopus oocytes following stimulation. And, the short SH-3 target peptide is SH-
It has been used as a pseudosubstrate for specific binding to 3 proteins. This is, of course, a brief description of available peptides with biological activity. There are numerous publications in this area. Thus, there are many precursors regarding the potential of small peptides to have intracellular signaling cascades. In addition, agonists and antagonists of many molecules can be used as the basis for bias randomization of peptides as well.

Generation of a prescreened randomized peptide library can generally be described as follows. Regardless of whether or not it has a known structure, for example, all structures such as a known protein portion, a known peptide, or a synthetic structure can be used as the backbone of PDA. For example, structures obtained by X-ray crystallographic techniques, NMR techniques, de novo modeling, homology modeling, etc. can all be used to obtain the scaffold for the desired sequence. Similarly, a large number of molecular or protein domains are also suitable as starting points for biased and randomized candidate bioactive agents. Common features,
A large number of small molecule domains are known that provide structure or affinity. In addition, as will be appreciated in the art, regions of weak amino acid homology may have strong structural homology. Many of these molecules, domains, and / or corresponding consensus sequences are known, including SH-2 domains, SH-3 domains, pleckstrins, death domains, protease cleavage / recognition sites, enzyme inhibitors. , Enzyme substrates, troughs, etc., but are not limited thereto. Similarly, there are numerous known nucleic acid binding proteins that contain domains suitable for use in the present invention. For example, the leucine zipper consensus sequence is known.

Thus, in general, known peptide ligands can be used as a starting scaffold to generate a first order randomization. In addition, structures that are known to be in certain configurations create skeletons,
The sequence can then be used to screen for those that appear to fall in such a configuration. For example, there are various known "ministructures," sometimes referred to as "expressive structures," which either confer configurational stability or
Alternatively, it is possible to provide a configuration in which the random arrangement is restricted. Proteins interact with each other primarily through conformationally tight domains. Small peptides having free-rotating amino acids and a carboxyl terminus can have powerful functionalities, as is known in the art, but converting their peptide structure into a pharmacologically active ingredient is Difficult due to the unpredictability of side chain positions for pseudo-peptidic synthesis. Therefore, the presentation of conformationally constrained peptides will aid in later drug production and will lead to higher affinity interactions of the peptides with their target proteins. This fact is observed in an integrated library generation system that uses short peptides biologically generated in the bacteriophage system. Numerous researchers have constructed small domain molecules that may have internally displayed randomized peptide structures.

Thus, synthetic presentation structures, ie, artificial polypeptides, can present randomized peptides as conformationally restricted regions. Preferred presentation structures are those that maximize accessibility to the peptide by presenting it on the outer loop. Thus, suitable presentation structures include, but are not limited to, minibody structures, loops on beta sheet turns and coiled coil system structures (randomized residues that are not structurally significant). ), Zinc finger region, cysteine-bonded (disulfide) structure, transglutaminase binding structure, cyclic peptide, B-loop structure, helical barrel or bundle, leucine zipper motif, and the like.

In a preferred embodiment, the presenting structure is a coiled coil structure,
Presents randomized peptides on the outer loop. For example, "Myszka et al.,
Biochem. 33: 2362-2373 (1994), incorporated herein by reference and FIG. 3). Using this system, researchers have enabled isolated peptides to make high affinity interactions with appropriate targets. In general, coiled coil structures are randomized between positions 6 and 20; (Martin et al., EM
See BO J. 13 (22): 5303-5309 (1994), incorporated by reference).

In a preferred embodiment, the presentation structure is a minibody structure. A "minibody" basically consists of the smallest antibody complementary region. This mini body presentation structure is
In general, folded proteins are displayed along a single surface of tertiary structure,
Two types of randomized areas are provided. For example, Bianchi et al., J. Mol. Biol. 236 (
2): 649-59 (1994) and references therein, all of which are incorporated herein by reference). Researchers have shown that this minimal domain is stable in solution, and have a high affinity (Kd = 10 ⁻⁷⁾ for the proinflammatory cytokine IL6 using the phage selection system in combinatorial libraries. A minibody having a peptide region showing) was selected.

Once the backbone chains have been selected and the primary library of random peptides generated as described above, the generation and creation of secondary libraries for known backbone proteins proceeds: this is at the mutation site and / or It involves the recombination of amino acid residues and is done by computer processing or experimentally. Furthermore, a DNA library expressing a protein sequence defined by the automated protein design method may be prepared. Codons can also be randomized only with nucleotide sequence triplets that specify residue positions identified by automated design methods. Also, a mixture of basal triplets encoding specific amino acids can be introduced into the DNA synthesis reaction to attach the complete triplet to the amino acids that exhibit it in one reaction step. We also design a library of random DNA oligomers that are biased toward a desired position with respect to a given amino acid or that limit the position to a given amino acid. Biased amino acids will be identified in a virtual screen, or a subset thereof.

Multiple DNA encoding different subsets of amino acids at certain positions
By synthesizing the library, it was possible to generate a variety of desired amino acids without completely randomizing codons, resulting in wasted sequences in the library for stop codons, frameshifts, unwanted amino acids, etc. . This can be done by creating a library that is randomized at only one or two of the triplets at each position to be randomized, where the remaining positions (group ) Is shared by the amino acids considered at this position. A large number of DNA libraries are prepared such that at each position all desired amino acids are guaranteed to be present in the integrated library.

Alternatively, random peptide libraries may be performed using frequency tabulation and experimental generation methods including multiplex PCR, shuffling and the like.

The present invention provides computer readable memory, central processing unit, associated circuitry, and other associated compositions for carrying out the invention. The device of the present invention comprises a memory and a set of input / output devices (e.g., keyboard, mouse, monitor,
Printer 26) and a central processing unit communicating via a bus. General interactions between the central processing unit, memory, input / output devices, and buses are known in the art. The present invention is directed to an automated protein design program and a secondary library generator stored in memory.

Automated protein design programs and / or secondary library generators may be run on the side chain modules. As detailed in the related application, the side chain module establishes a group of potential rotamers for selected protein backbone structures. This protein design program may be executed using a ranking module. As detailed below, the ranking module analyzes rotamer interactions for protein backbone structures and produces optimized protein sequences. The protein design program may include a search module for performing a search such as the Monte Carlo search related to protein sequence optimization shown below. Finally, an evaluation module may be used to evaluate the physical parameters associated with the resulting protein, as described below.

This memory also stores the protein backbone structure, which can be downloaded by the user through the input / output device. This memory also stores information on potential rotamers obtained by the side chain module. In addition, this memory stores the protein sequences generated by the ranking module. This protein sequence may be transferred as output to an input / output device.

A variety of expression vectors are prepared using the nucleic acids of the invention that encode library members. The expression vector may be either a self-replicating extrachromosomal vector or a vector integrated into a host gene. Generally, these expression vectors contain transcriptional and translational regulatory nucleic acid operably linked to the nucleic acid encoding the library protein. The term "regulatory sequence" refers to a DNA sequence necessary for the expression of operably linked coding sequences 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. It is known that eukaryotic cells 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 precursor sequences or secretory leaders
Is operably linked to the DNA for the polypeptide when expressed as a preprotein involved in the secretion of the polypeptide; a promoter or enhancer is operably linked to the coding sequence when it affects the transcription of the sequence. Or the ribosome binding site is operably linked to the coding sequence when in a position that facilitates translation. Generally, "operably linked" means that the DNA sequences are contiguous, and, in the case of a secretory leader, contiguous and reading steps. However, enhancers do not have to be contiguous. Binding is accomplished by ligation at convenient restriction sites. If the above sites do not exist, synthetic oligonucleotide adapters or linkers are used in accord with conventional practice. Transcriptional and translational regulatory nucleic acids are generally compatible with the host cell used to express the library protein, as is known to those of skill in the art;
Transcriptional and translational regulatory nucleic acid sequences from Bacillus are preferably used for expression of library proteins in Bacillus. Various types of suitable expression vectors, and suitable regulatory sequences are known in the art for different host cells.

In general, transcriptional and translational regulatory sequences may include, but are not limited to, promoter sequences, ribosome 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.

Promoter sequences include constitutive or inducible promoter sequences. The promoter may be a natural promoter or a hybrid or synthetic promoter. Hybrid promoters that have elements of more than one promoter together
It is also known in the art and is useful in the present invention.

In addition, the expression vector may contain additional elements. For example, expression vectors can have two replication systems and thus can be maintained in two organisms, eg, mammalian or insect cells for expression and prokaryotic hosts for cloning and amplification. Furthermore, in the case of an integrative expression vector, the expression vector comprises at least one sequence homologous to the host cell genome, and preferably two homologous sequences flanking the expression construct. Integrating vectors can be directed to specific loci in the host cell by selecting the appropriate homologous sequences for inclusion in the vector. Constructs for integrative vectors and appropriate selection and screening protocols are well known in the art, for example:
Mansouret all, Cell, 51: 503 (1988) and Murray, Gene Transfer and Express
ion Protocols, Methods in Molecular Biology, Vol. 7 (Clifton: Humana Pre
ss, 1991).

Furthermore, in a preferred embodiment, the expression vector is, in particular in mammalian cells,
Cells that do not contain the vector will generally die, ensuring the stability of the vector,
It contains a selection gene that allows for selection of transformed host cells. Selection genes are well known in the art and will vary with the host cell used. As used herein, "selection gene" means any gene that encodes a gene product that is resistant to a selection agent. Suitable selection agents include, but are not limited to, neomycin (or the analog G418), blasticidin S, histinidol D, berellomycin, puromycin, hygromycin B and other drugs.

In a preferred embodiment, the expression vector comprises upstream or downstream of the RNA splicing sequence of the gene to be expressed to increase the level of gene expression (see: Barret et al., Nucleic Acids Res. 1991; Groos et. al., Mol. Cell. Bio
l. 1987; and Budiman et al., Mol. Cell. Biol. 1988).

A preferred expression vector system is Mann et al., Cell, 33: 153-9 (1993); Pear et
al., Proc. Natl. Acad. Sci. USA, 90 (18): 8392-6 (1993); Kitamura et a
L., Proc. Natl. Acad. Sci. USA, 92: 9146-50 (1995); Kinsella et al., Hu
man Gene Therapy, 7: 1405-13; Hofmann et al., Proc. Natl. Acad. Sci. USA
., 93: 5185-90; Choate et al., Human Gene Therapy, 7: 2247 (1996); PCT
/ US97 / 01019 and PCT / US97 / 01048, retroviral vector systems as cited in the specification, which are hereby incorporated by reference.

The library protein of the present invention is a host cell transformed with an expression vector containing a nucleic acid, preferably a nucleic acid encoding the library protein, under suitable conditions for inducing or expressing the library protein. Are cultured and produced. Appropriate conditions for expression of library proteins will vary with the choice of expression vector and host cell and are readily ascertained by one of skill in the art through routine experimentation. For example, the use of constitutive promoters in expression vectors requires optimization of host cell growth and proliferation, and the use of inducible promoters requires growth conditions suitable for induction. Furthermore, in a specific example, the timing of collection is important. For example, the baculovirus system used in insect cell expression is a lytic virus, so the choice of harvest time is critical to product yield.

As will be apparent to those of skill in the art, the cell types used in the present invention will vary widely. In principle, any suitable host cell can be used, including yeast, bacteria, archaea, fungi and animal cells including insect and mammalian cells. Particularly advantageous are Drosophila melangaster cells, Saccharomyses 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, immortalized mammalian myeloid and lymphoid cell lines, Jurkat cells, mast cells and other endocrine and exocrine cells as well as neuronal cells. A, which is included in this specification due to clarification of the source
See TCC cell line catalog. Moreover, expression of the secondary library in a phage display system as is well known in the art is particularly preferred in the special case where the secondary library comprises random peptides. In one aspect, the cells can be genetically engineered to contain a foreign nucleic acid, eg, a target molecule.

[0122] In a preferred embodiment, the library proteins are expressed in mammalian cells. Any mammalian cell, mouse, rat, primate may be used, especially human cells being preferred and as will be appreciated by those skilled in the art, modifications of the system by pseudotypes are all eukaryotic cells. , Preferably for use in higher eukaryotes. As described in more detail below, a screen will display the cells displaying a selectable phenotype in the presence of random library members. As described further below, as long as a suitable screen can be constructed that allows the selection of cells that display a different phenotype by the presence of intracellular library members, it will affect a wide range of variations in disease symptoms. The cell types given are particularly useful.

Accordingly, suitable mammalian cell types include, but are not limited to, all types of tumor cells (especially melanoma-like, myeloid leukemia, lung carcinoma, breast carcinoma, ovarian carcinoma, colon carcinoma, Renal carcinoma, prostate carcinoma, pancreatic carcinoma and testicular carcinoma), cardiomyocytes, endothelial cells,
Epithelial cells, lymphocytes (T cells and B cells), mast cells, eosinophils, intimal cells, nervous system, skin, lungs, kidneys, liver and muscle axis cells (screening for differentiation and dedifferentiation factors ), Osteoclasts, chondrocytes and other connective tissue cells, keratinocytes, melanocytes, liver cells, kidney cells and adipocytes. Suitable cells are also known exploratory cells, Jurkat
It also includes T cells, NIH3T3 cells, CHO, Cos and the like. See ATCC Cell Line Catalog (including but not limited to by reference).

Mammalian expression systems are also known in the art and include retroviral 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 library protein into mRNA. The promoter has a transcription initiation region usually located proximal to the 5'end of the coding sequence, and a TATA box with 25-30 base pairs located upstream of the transcription initiation site. The TATA box is thought to direct RNA polymerase II to initiate RNA synthesis at the correct site. Mammalian promoters are also typically located within 100 to 200 base pairs upstream of the TATA box,
Contains upstream promoter elements (enhancer elements). The upstream promoter element is
It determines the rate of transcription initiation and can act in either orientation. Since viral genes are often highly expressed and have a wide host range, promoters derived from mammalian viral genes are particularly useful as mammalian promoters. As an example, SV
40 early promoter, mouse mammary tumor virus LTR promoter, adenovirus major late promoter, herpes simplex virus promoter, CMV promoter.

[0125] Typically, transcription termination and polyadenylation sequences recognized by mammalian cells are regulatory regions located 3'to the transcription stop codon, and thus co-exist with promoter elements at the ends of the coding sequence. Adjacent. The 3'end of the mature mRNA is
It is formed by site-specific post-translational cleavage and polyadenylation. Examples of transcription terminators and polyadenylation signals include those derived from SV40.

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

[0127] In a preferred embodiment, the library proteins are expressed in bacterial systems. Bacterial expression systems are known to those of skill in the art.

A suitable bacterial promoter is one that binds to bacterial RNA polymerase and
All nucleic acid sequences capable of initiating the downstream (3 ') transcription of the library protein coding sequence into A. Bacterial promoters usually have a transcription initiation region located proximal to the 5'end of the coding sequence. This transcription start region typically includes an RNA polymerase binding site and a transcription start site. From the sequence encoding the metabolic pathway enzyme,
Particularly useful promoter sequences are provided. Examples are promoter sequences derived from sugar-metabolizing enzymes such as galactose, lactose and maltose, and sequences derived from biosynthetic enzymes such as tryptophan. It is known to those skilled in the art that promoters from bacteriophage can also be used. In addition, synthetic and hybrid promoters are also useful. For example, the tac promoter is a hybrid of the trp and lac promoter sequences. In addition, the bacterial promoter binds bacterial RNA polymerase,
It may include native promoters of non-bacterial origin that have the ability to initiate transcription.

In addition to a functional promoter sequence, an effective ribosome binding site is desirable. In Escherichia coli, the ribosome binding site is called the Shine-Delgarno (SD) sequence, and the start codon and the sequence located 3-11 nucleotides upstream of the start codon
It has a length of 9 nucleotides.

The expression vector may also include a signal peptide that provides for secretion of the library protein in bacteria. The signal sequence typically encodes a signal peptide consisting of hydrophobic amino acids that direct protein secretion from the cell, as is well known to those of skill in the art. The protein is secreted into the periplasmic space between the inner and outer membranes of growth medium (Gram-positive bacteria) or cells (Gram-negative bacteria).

The bacterial expression vector may also contain a selectable marker gene to allow the selection of transformed bacterial strains. Suitable selection genes include genes that confer to bacteria resistance 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 constructed into an expression vector. Expression vectors for bacteria are well known in the art and include B. subtilis, E. coli, Streptococcus cremoris and Streptococcus.
Contains vectors for Libidoans (Streptococcus lividans).

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

In one aspect, the library proteins are produced in insect cells. Expression vectors for transformation of insect cells and especially baculovirus-derived expression vectors are also known in the art, eg O'Reilly et al., Baculovirus Expression Vector.
s: A Laboratory Manual (New York: Oxford University Press, 1994).

In a preferred embodiment, library proteins are produced in yeast cells. Yeast expression systems are known in the art and include Saccharomyces cerevisiae, Candida albicans and C. maltosa.
), Hansenula polymorpha, Kluyveromyces fragilis and K. lactis, Pichia guillerimondii and P. Pastoris. Saccharomyces pombe (Schizos
accharomyces pombe) and Yarrowia lipolytica
A) expression vector. Preferred promoter sequences for expression in yeast include the inducible GAL1,10 promoter, alcohol dehydrogenase, enolase, glucokinase, glucose-6-phosphate isomerase, glyceraldehyde-3-phosphate dehydrogenase, hexokinase, phosphofructokinase,
There are promoters from the 3-phosphoglycerate mutase, pyruvate kinase and acid phosphatase genes. Yeast selectable markers include ADE2, H
There are IS4, 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.

The library protein of the present invention can also be produced as a fusion protein using a technique known in the art. To that end, for example, with respect to the construction of monoclonal antibodies, the library protein can be fused with a carrier protein to form an immunogen when the desired epitope is small. Alternatively, the library protein can be produced as a fusion protein for increased expression or for other reasons. For example, if the library protein is a library peptide, the nucleic acid encoding the peptide may be linked to another nucleic acid for expression purposes. Similarly, another fusion partner, such as a target sequence that enables localization of the library member to the intracellular or extracellular space of the cell, either the library protein or the nucleic acid encoding them, is purified or purified. A rescue sequence or purification tag that allows separation; a stable sequence that provides stabilization or protection (eg, resistance to proteolysis) from degradation to the library protein or nucleic acid encoding the library protein, or a combination thereof, Linker sequences can also be used if present.

Thus, suitable target sequences include, but are not limited to, capable of binding an expression product to a predetermined molecule or class of molecules while maintaining the biological activity of the expression product. A binding sequence (eg, targeting the relevant enzyme class with an enzyme inhibitor or substrate sequence); a sequence that provides a signal for selective degradation of the protein bound by itself or together; and a candidate expression product, a) Golgi apparatus, endoplasmic reticulum, nucleus, nucleolus, nuclear membrane, mitochondria, chloroplast, secretory vesicle, lysosome,
And signal sequences capable of constitutively localizing to a predetermined cellular location, including intracellular locations such as the cell membrane, and b) extracellular locations via secretory signals. Either subcellular localization or extracellular localization via secretion is particularly preferred.

In a preferred embodiment, the library member is a rescue sequence. A rescue sequence is a sequence that can be used to purify or isolate either the candidate substance or the nucleic acid that encodes it. Thus, peptide rescue sequences include, for example, His ₆ tags for use with Ni affinity columns and purified sequences such as epitope tags for detection, immunoprecipitation or FACS (fluorescence activated cell sorting). Suitable epitope tags include myc (used with commercially available 9E10 antibody), BSP biotinylated target sequence of bacterial enzyme BirA, flu tag,
lacZ and GST.

Alternatively, the rescue sequence may be a unique oligonucleotide sequence that functions as a probe target site from which a retroviral construct by PCR, related techniques or hybridization can be isolated quickly and easily.

In a preferred embodiment, the fusion partner is a stable sequence that confers stability to the library member or the nucleic acid encoding it. Thus, for example,
To protect the ubiquitinated peptide according to Varahavsky's N-terminal rule,
Incorporation of glycine after the initiating methionine (MG or MGGO) may stabilize the peptide and thus confer a long half-life in the cytoplasm. Similarly, the two C-terminal prolines confer to the peptide strong resistance to the action of carboxypeptidase. The presence of the two glycines in front of proline confers both variability and protective structure and initiates a series of events in di-proline to propagate into the candidate peptide structure. Such preferred stability sequences are shown below:
MG (X) _n GGPP, where X is any amino acid and n is at least 4
Is an integer. .

In one aspect, the library nucleic acids, proteins and antibodies of the invention are labeled. As used herein, the term “labeled” means that the library nucleic acid, protein and antibody of the present invention are bound to an isotope or a chemical compound to enable detection of the library nucleic acid, protein and antibody of the present invention. It is meant to have one component.
In general, labels fall into three categories: a) isotope labels, which can be radioactive or heavy isotopes, b) immunolabels, which can be antibodies or antigens, and c) colored or fluorescent dyes. The label can be incorporated into the compound at any position.

In a preferred embodiment, the library proteins are purified or isolated after expression.
Library proteins can be isolated or purified by various methods known to those of skill in the art. Standard purification methods include electrophoretic, molecular, immunological and ion exchange, hydrophobic, affinity and chromatofocusing techniques including reverse phase HPLC chromatography and chromatofocusing. For example, library proteins can be purified using standard anti-library antibody columns. Ultrafiltration and diafiltration techniques are also useful in connection with protein concentration. For general guidance on suitable purification techniques, see Sc
See opes, R., Protein Purification, Springer-Verlag, NY (1982). The required degree of purification depends on the use of the library protein. In some cases, no purification is necessary.

Once expressed, and if necessary, and purified library proteins and nucleic acids are useful in a variety of applications.

Secondary libraries are generally screened for biological activity. These screens are based on the scaffold protein of choice, as is known in the art. In this way, binding with its known binding member (eg its substrate in the case of an enzyme), activity profile, stability profile (pH, temperature, buffer conditions).
, All of the activities or properties of a large number of proteins can be tested, including substrate specificity, immunogenicity, toxicity and the like.

If random peptides are produced, these can be used in various ways for screening for activity. In a preferred embodiment, the first cell population is screened. That is, cells into which library member nucleic acids have been introduced are screened for altered phenotype. Thus, in this aspect, the effect of the library member is seen in the same cell in which it was made; ie the autocrine effect.

By “cell population” herein is meant approximately about 10 ³ to 10 ⁸ or 10 ⁹ cells, with 10 ⁶ to 10 ⁸ being preferred. This cell population comprises a cellular library, in which usually each cell in the library will contain a member of the secondary library, i.e. another library member, as will be apparent to those skilled in the art. , Some cells in the library may not contain library members, and some may contain one or more library members. When methods other than retroviral infection are used to introduce library members into a cell population, the distribution of library members within individual cell members of a cellular library can vary widely. This is because it is difficult to control the large number of nucleic acids that enter the cell during electroporation and the like.

In a preferred embodiment, the library nucleic acid is introduced into a first cell population and the effect of the library members is on a second or third cell population of a different cell type than the first cell population, ie usually a different cell type. To screen. That is, the effect of the library member is due to an extracellular effect on the second cell, an endocrine or paracrine effect. This is done by standard techniques. The first cell population can be grown in one or more media, which is contacted with the second cell population and the effect measured. Alternatively, the cells may be directly contacted with each other. Thus, "contacting" is functional contacting, including both direct and indirect. In this aspect, the first cell population may or may not be screened.

If necessary, the cells are subjected to conditions suitable for expression of library members (eg, using an inducible promoter) to produce library proteins.
That is, in one aspect, the method of the invention comprises introducing a molecular library of library members into a cell population, a cell library. The cell population is then screened for cells exhibiting an altered phenotype, as described in further detail below. The altered phenotype is due to the presence of library members.

As used herein, "altered phenotype" or "physiologically altered" or other grammatical synonyms mean that the phenotype of the cell is in some way preferably detectable and / or Or it means changing in a measurable way. As will be appreciated by one of skill in the art, an advantage of the present invention is the variety of cell types and potential phenotypic changes that can be tested using this method. Therefore, any phenotypic change that can be observed, detected or measured can be based on the screening methods herein. Suitable phenotypic changes include, but are not limited to, morphological changes in cells, cell proliferation, cell viability, adhesion to substrates or other cells, and overall physical changes such as cellular density. A change in the expression of one or more RNAs, proteins, lipids, hormones, cytokines or other molecules; an equilibrium state of one or more RNAs, proteins, lipids, hormones, cytokines or other molecules (ie half-life )) Changes; one or more RNAs, proteins, lipids, hormones, cytokines or other localization changes; one or more RNAs,
Altered biological or specific activity (specific activity) of proteins, lipids, hormones, cytokines, receptors or other molecules; altered secretion of ions, cytokines, hormones, growth factors or other molecules; cell membrane potential, polarity, integrity Changes in sex or transport; including infectivity, infectivity, latency, adhesiveness and changes in absorption of viral and bacterial pathogens. By "capable of altering phenotype" herein is meant that the library member is capable of altering the phenotype of a cell in a number of detectable and / or measurable ways.

The altered phenotype can be detected in a variety of ways, but generally depends and corresponds to the phenotype being altered. In general, altered phenotypes include, for example, microscopic analysis of cell morphology; standard cell viability assays, in which case both proliferative cell death and proliferative cell viability are involved, eg viral, bacterial or bacterial or synthetic toxicity. Cells that are resistant to cell death mediated by P. a .; standard labeling assays, such as fluorescent indicator assays for the presence or level of particular cells or molecules, in which case FACS or other dye staining techniques are included; cell death It is subsequently detected using biochemical detection of the expression of the target compound. In some cases, as fully described herein,
The altered phenotype is detected in cells into which the randomized nucleic acid has been introduced, and in other aspects,
The altered phenotype is detected in the second cell in response to some molecular signal from the first cell.

In a preferred embodiment, library members are isolated from positive cells. This can be done in various ways. In a preferred embodiment, a primer complementary to a DNA region common to a particular element of a library such as the construct or rescue sequence defined above is used to "rescue" a particular random sequence. Alternatively, the member is isolated using a rescue sequence. Thus, for example, a rescue sequence containing an epitope tag or a purified sequence is used to recover library members using immunoprecipitation or affinity columns. In certain instances, if there is a significant strong binding interaction between the library member and the target molecule, the one to which the library member binds (eg, the first target molecule) is also recovered. Alternatively, the peptide can be detected using mass spectrometry.

Once rescued, the library members are sequenced. This information can then be used in many ways.

In a preferred embodiment, the member is resynthesized and reintroduced into the target cell, demonstrating its effect. This is done with a retrovirus, or alternatively a fusion to the HIV-1 Tat protein and similar and related proteins, whereby
Very high uptake into target cells is possible. For example, Fawell et al., PNAS
USA 91: 664 (1994); Frankel et al., Cell 55: 1189 (1988); Savion et al., J. B.
iol. Chem. 256: 1149 (1981); Derossi et al., J. Biol. Chem. 269: 10444 (
1994); and Baldin et al., EMBO J. 9: 1511 (1990) (incorporated herein by reference).

In a preferred embodiment, the sequence of the member is used as described herein,
The library is created.

In a preferred embodiment, the library members are used to identify target molecules, ie molecules with which the members interact. As one of skill in the art will recognize, there may be a first target molecule to which the library member binds or acts directly, and there may be a second target molecule that is part of the signal transduction pathway affected by the library member, These are called "effective targets".

The screening methods of the invention can be useful for screening many cell types under a variety of conditions. Usually, host cells are cells that contain a disease state, and they are tested or screened under conditions that usually produce undesirable results on the cells. Undesirable effects may be reduced or eliminated when suitable library members are found. Alternatively, usually the desired result is one that can be reduced or eliminated and involves a means of elucidating the cellular mechanisms involved in the disease state or signaling pathway.

The following is a more detailed description, using the techniques of the invention set forth above, of examples, and further illustrates the best method envisioned for practicing the various aspects of the invention. These examples do not limit the true scope of the invention in any way, but rather are presented for illustrative purposes only. All references cited in this specification are incorporated herein by reference.

Examples Example 1 Computational Prescreening of β-Lactamase TEM-1 Preliminary testing was performed on β-lactamase gene TEM-1. Brookhaven Protein Data Bank Entry 1BTL was used as the starting structure. All water molecules and SO ₄ ²⁻ groups were removed and cations were generated on the structure. The structure was minimized in 50 steps without static electricity using the conjugate gradient method and the Drying II force field. The BIOGRAF program (Molecul
ar Simulations, Inc., San Diego, CA). This minimized structure served as a template for all protein design calculations.

Computerized Prescreening Computerized prescreening of sequences was performed using PDA. A 4Å sphere was placed around the heavy side chain atoms (S70, K73, S130, and E166) of the four catalytic residues, and all amino acids with heavy side chain atoms within the cutoff distance were selected. In this way the following seven positions: F72, Y105, N132, N136, L169, N170,
And K234 were obtained. Two of these residues, N132 and K234, are highly conserved in several different β-lactamases, and therefore five variable positions (F72, Y).
105, N136, L169, N170) were not included in the design. These designed positions were able to change the identity with any of 20 natural amino acids except proline, cysteine, and glycine (17 amino acids in total). Proline is not normally used because it is difficult to define the appropriate rotamer of proline. Cysteine was excluded to prevent disulfide bond formation, and glycine was excluded due to conformational flexibility.

In addition, a second set of residues within the range of 5Å of the residues selected for PDA design was floated (their amino acid identities were maintained as wild type but the configuration could be changed). there were). The heavy side chain atoms were used again to determine which residue was within the cutoff. The following 28 positions were obtained: M68, M69, S70, T71, K73, V74,
L76, V103, E104, S106, P107, I127, M129, S130, A135, L139, L148, L162, R
164, W165, E166, P167, D179, M211, D214, V216, S235, I247. A248 is I247
Was included as an alternative floating position. When the positions are M69, R164, and W165, two prolines, P107 and P167, were selected because the crystal structure shows a highly stringent rotamer, leaving 23 floating residues from the second set. Excluded from floating residues. The retained residues, N132 and K234, from the first range (4Å) also floated, giving a total of 25 floating residues.

The potential functions and parameters used for PDA calculation were as follows:
The Van der Waals scale factor was set to 0.9 and the electrostatic potential was calculated using distance attenuation and a dielectric constant of 40. The well depth for hydrogen bonding potential was set to 8 kcal / mol with local and remote backbone scale factors (0.25 and 1.0 respectively). Design positions (F72, L169, M68, T71, V74, L76,
I127, A135, L139, L148, L162, M211 and A248) only. Type 2
Solvation was used (Street and Mayo, 1998). Non-polar exposure growth factor 1.6
, The non-polar burial energy was set to 0.048 kcal / mol / A ² , and the polar hydrogen burial energy was set to 2.0 kcal / mol.

The Dead End Elimination (DEE) optimization method (see literature) was used to find the lowest energy ground state sequence. DEE cutoffs of 50 and 100 kcal / mol were used for single and dual energy calculations, respectively.

A Monte Carlo (MC) calculation is performed starting from the DEE ground state array, and 1000
Created a list of the lowest energy arrays of. The MC parameters were 100 annealing cycles with 1,000,000 steps per cycle. The non-productive cycle range was set to 50. The annealing schedule set the high and low temperatures to 500K and 100K, respectively.

Next, the following probability distribution was calculated from the top 1000 sequences in the MC list (
(See Table 3 below). It shows the number of occurrences of each amino acid selected for each position (5 variable positions and 25 floating positions). Table 3: Monte Carlo analysis (amino acids and their occurrences (for top 1000 sequences)
)

[Table 2]

This probability distribution was converted into a probability distribution with rounding down (see Table 4). Wild-type amino acids were always produced with a probability of at least 10%, rounded off at the designed position using a cutoff value of 10%. E was found in 169th place, 15.6% of the total time. However, since this position is just before the other design position, 170th place,
This proximity would require a more complex oligonucleotide library design; therefore, E was not included in this position when creating the sequence library (only L was used).

Table 4: PDA probability distributions for β-lactamase design positions (rounded to the nearest 10%).

[Table 3]

As shown in Table 4, computerized prescreening was able to reduce the size of the problem by orders of magnitude. Originally, there are respectively five design located at different amino acids of 17, there is a possibility sequences 17 ⁵ = 1,419,857. This was reduced to only 2 * 7 * 3 * 1 * 5 = 210 possible arrays (a reduction of almost 4 orders of magnitude).

Generation of Sequence Library Overlapping oligonucleotides corresponding to the full-length TEM-1 gene of β-lactamase and all desired mutations were synthesized and used in a PCR reaction as described above (FIG. 1), 210 as described above. A sequence library containing the sequences was obtained.

Synthesis of TEM-1 Gene Variants For the preparation of TEM-1 gene variants, pCR2.1 (Invitrogen) was designed with Xbal and EcoRI, blunt-ended with T4 DNA polymerase and religated. The Hind III and Xhol sites were removed from within the polylinker. Then, Quickchange Site-Directe
Using the d Mutagenesis Kit, the new Xhol site was located at position 2269 of the TEM-1 gene (numbered as the original pCR2.1) as described by the manufacturer (Stratagene).
Introduced inside. Similarly, a new HindIII site was introduced at position 2674, resulting in pCR-Xen1.

To create a mutated TEM-1 gene, overlapping 40 mer oligonucleotides were synthesized corresponding to the sequence between the newly introduced Xho1 and HindIII sites, and adjacent nucleotides overlapped with 20 nucleotides. Designed to do. A large number of oligonucleotides were synthesized at each design position (72, 105, 136 and 170), each containing different variants, so that as shown in Table 4, all possible combinations of variant sequences (210) were obtained in the desired proportions. It was prepared. For example, at position 72, two sets of oligonucleotides were synthesized (one containing an F at position 72 and the other containing a Y). Each oligonucleotide was resuspended at a concentration of 1 μg / μl and equimolar concentrations of oligonucleotide were pooled.

At overlapping positions, each oligonucleotide was added at a concentration that reflected the probability of Table 4. For example, at position 72 add equal amounts of two oligonucleotides to the pool,
At position 136, the oligonucleotide containing M was added in an amount twice that of the oligonucleotide containing N, and the oligonucleotide containing D was added in an amount of 7 times that of the oligonucleotide containing N.

Assembly of DNA Library For the first round of PCR, 2 μl of oligonucleotides pooled at the desired probability (Table 4) were added to 2 μl of 10 mM dNTPs, 10 μl of 10x Taq buffer (Qiagen), Taq DNA polymerase (5 units / μl). : Qiagen) 1 μl and Pfu DNA polymerase (2
0.5 unit / μl: Promega) was added to a 100 μl reaction system containing 2 μl. The reaction mixture was collected on ice and subjected to a final extension step of 94 ° C for 5 minutes, then 15 cycles of 94 ° C for 30 seconds, 52 ° C for 30 seconds and 72 ° C for 30 seconds, followed by 72 ° C for 10 minutes.

Isolation of full-length oligonucleotide For the second round of PCR, 2.5 μl of the first round of reaction was added to 2 μl of 10 mM dNTPs,
10x Pfu DNA polymerase buffer (Promega) 10 μl, Pfu DNA polymerase (2.5 units / μl: Promega) 2 μl, and a 100 μl reaction system containing 1 μl of oligonucleotides corresponding to the 5 ′ and 3 ′ ends of the synthetic gene were added. It was The reaction mixture was collected on ice and subjected to a final extension step of 94 ° C for 5 minutes, then 94 ° C for 30 seconds, 52 ° C for 30 seconds, 72 ° C for 30 seconds, and 72 ° C for 10 minutes to isolate the full length oligonucleotide. did.

Purification of DNA library PCR products were purified using the QlAquick PCR Purification Kit (Qiagen), digested with Xho1 and Hindlll, electrophoresed on a 1.2% agarose gel and QlAq.
Repurified using uick Gel Extraction Kit (Qiagen).

Confirmation of Sequence Library Identity A PCR product containing a library of mutant TEM-1 β-lactamase genes was cloned between the promoter and terminator in a kanamycin resistance plasmid and cloned in E. coli. Transformed. Equal amounts of bacteria were spread on medium containing either kanamycin or ampicillin. Transformed colonies can all be kanamycin resistant, but only those with an active mutated β-lactamase gene can grow on ampicillin. After overnight incubation, several colonies were obtained on both plates and at least one of the above sequences was found to encode an active β-lactamase. The number of colonies on the kanamycin plate was significantly higher than that on the ampicillin plate (roughly 5: 1), either because some sequence had lost its activity, or because the PCR transfer failed and the inactive or cleaving enzyme was present. It is suggested that was obtained.

To distinguish these possibilities, 60 colonies were picked from kanamycin plates and the plasmid DNA sequenced. The distribution shown in Table 5 was obtained. Table 5: PDA Predicted Percentages vs Percentages Observed in Experiments for Design Positions

[Table 4]

It should be noted that the observed percentage of each amino acid in all four positions fits closely with the predicted percentage. Sequencing also revealed that only 1 out of 60 colonies contained a PCR error (G to C transition).

This quiz uses multiple PCR with pooled oligonucleotides to
It shows that a sequence library can be created that reflects the desired ratio of amino acid changes.

Experimental Screening of Sequence Libraries Purified PCR products containing a library of mutant sequences were previously digested with Xho1 and Hindlll.
Ligated in pCR-Xen1 which had been digested with and purified. The ligation reaction was transformed into competent TOP 10 E. coli cells (Invitrogen). 3 cells
After recovering at 7 ° C for 1 hour, the cells were treated with the antibiotic cefotaxime at 0.1 µg / ml.
Were spread on LB plates containing a concentration range from 1 to 50 μg / ml and selected for enhanced resistance.

Only in a single screen were triple mutants with 35-fold improved enzyme function (see FIG. 4). This mutant (Y105Q, N136D, N170L) remained even at 50 μg / ml cefotaxime.

Example 2 Secondary Library Generation of Xylanase PDA Prescreening Leads to Out-of-Order Reduction of Possible Sequence Numbers Computational Prescreening Is Possible and Sequences That Must Be Screened Experimentally To make it clear that the number can be reduced significantly,
Initial calculations were performed for B. circulans xylanase in the presence or absence of substrate. PDB structure of B. circulans xylanase 1XN and 1BC for enzyme substrate complex
I used X. The 27 residues within the binding site were visually identified as belonging to the active site. Eight of these residues were considered absolutely essential for enzyme activity. These positions were treated as wild type residues. This means that their structures are variable, but amino acid identity is not (see Figure 2).

Three of the twenty natural amino acids (cysteine, proline, and glycine)
Did not consider. As a result, there was still a possibility of 17 different amino acids at the remaining 19 positions; the problem is that 17 ¹⁹ = 2.4 × 10 ²³ different amino acid sequences occur. This number is ten orders of magnitude higher than that available in the state-of-the-art oriented innovation methods. Obviously, it is not possible to screen the problem completely three-dimensionally using these approaches and to consider the entire sequence with various permutations. Therefore, PDA calculations were performed to reduce the search area. A list of 10,000 lowest energy sequences was made and the probability of each amino acid at each position was determined (see Table 1).

Table 1: Probability of each amino acid at the design position resulting from PDA calculation of wild type (WT) enzyme structure. Only amino acids with a probability greater than 1% are shown.

[Table 5] If all the amino acids obtained by PDA calculation are considered, there are 4.1 × 10 ¹⁵ different amino acid sequences, including those with a probability of 1% or less. This is a 7-digit size reduction. If only the amino acids shown in Table 1 with a probability of at least 1% or more (1% basis) are considered, the problem is reduced to 3.3 × 10 ⁹ sequences. Ignoring all amino acids (5% basis) that are less than 5% likely,
Only 4.0 × 10 ⁶ sequences remain. This is a number that can be easily handled by screening and gene shuffling techniques. List of low energy arrays 1
Increasing to 00,000 does not change these numbers significantly, and the effect of the amino acid obtained at each position is insignificant. Changes occur only between amino acids that are less than 1% likely.

Inclusion of substrate in the PDA calculation further reduced the number of amino acids found at each position. Considering the amino acids with a probability higher than 5%, it has a sequence of 2.4 × 10 ⁶ (see Table 2). Table 2: Possibility of amino acids at the design position resulting from PDA calculation of enzyme substrate complex. Only amino acids with a probability of 1% or higher are shown.

[Table 6]

These preliminary calculations show that PDA can significantly reduce the size of the problem and fall within the scope of gene shuffling and screening techniques (see Figure 3).

[Brief description of drawings]

FIG. 1 depicts the synthesis of a full length gene and all possible mutagenesis by PCR.

FIG. 2 shows the reduction in size of sequence space by PDA screening.

FIG. 3 shows the active site of B. circulans xylanase.

FIG. 4 shows cefotaxime resistance of E. coli expressing wild type (WT) and PDA screen β-lactamase.

─────────────────────────────────────────────────── ─── Continued front page    (81) Designated countries EP (AT, BE, CH, CY, DE, DK, ES, FI, FR, GB, GR, IE, I T, LU, MC, NL, PT, SE), OA (BF, BJ , CF, CG, CI, CM, GA, GN, GW, ML, MR, NE, SN, TD, TG), AP (GH, GM, K E, LS, MW, SD, SL, SZ, TZ, UG, ZW ), EA (AM, AZ, BY, KG, KZ, MD, RU, TJ, TM), AE, AL, AM, AT, AU, AZ, BA, BB, BG, BR, BY, CA, CH, CN, C U, CZ, DE, DK, EE, ES, FI, GB, GD , GE, GH, GM, HR, HU, ID, IL, IN, IS, JP, KE, KG, KP, KR, KZ, LC, L K, LR, LS, LT, LU, LV, MD, MG, MK , MN, MW, MX, NO, NZ, PL, PT, RO, RU, SD, SE, SG, SI, SK, SL, TJ, T M, TR, TT, TZ, UA, UG, UZ, VN, YU , ZA, ZW (72) Inventor Basil I Dahyat             United States 90065 California California             San Angels, San Andreas Abe             New 4413 (72) Inventor Jörg Benttwyn             United States 91101 Pa, California             Sadina, Pleasant Street Nan             Bars 307 and 277 F-term (reference) 4B024 AA20 BA11 BA12 CA04 HA01                 5B046 AA00 KA06

Claims (9)

[Claims] 1. A) preparing a primary library containing a ranked ordered list of primary variant sequences of scaffold proteins; b) creating a list of primary mutation positions in the primary library; c) a plurality of Combining the primary mutation positions to create a secondary library of secondary sequences; and, producing a secondary library of scaffold protein variants. 2. A) preparing a primary library containing a rank-ordered list of primary mutant sequences of a scaffold protein; b) creating a probability distribution of amino acid residues at a plurality of mutation positions; c) plural. Combining the primary mutation positions of to create a secondary library of secondary sequences; 3. The method of claim 1, further comprising synthesizing a plurality of said secondary sequences.
The method described in. 4. Multiple PCR using the pooled oligonucleotides of the synthesis.
The method of claim 2, wherein the method is performed by. 5. The method of claim 4, wherein the pooled oligonucleotides are added in equimolar amounts. 6. The method according to claim 4, wherein the pooled oligonucleotide is added in an amount corresponding to the number of mutations. 7. A composition comprising a plurality of secondary variant proteins comprising a subset of a secondary library. 8. A composition comprising a plurality of nucleic acids encoding a plurality of secondary variant proteins comprising a subset of a secondary library. 9. A) preparing a rank-ordered list of the first library of primary variants of skeletal proteins; b) creating a probability distribution of amino acid residues at a plurality of mutation positions; c) a plurality of such sequences. Synthesizing a plurality of secondary mutants of a scaffold protein containing amino acid residues to form a secondary library; provided that at least one of the secondary mutants is different from the primary mutant. A method for generating a secondary library of protein variants.