WO1999011818A1

WO1999011818A1 - Method for detecting highly functional polypeptides or nucleic acids

Info

Publication number: WO1999011818A1
Application number: PCT/JP1998/003854
Authority: WO
Inventors: Isao Karube; Yoichi Okabe; Koichi Sumikura
Original assignee: Isao Karube; Yoichi Okabe; Koichi Sumikura
Priority date: 1997-08-28
Filing date: 1998-08-28
Publication date: 1999-03-11
Also published as: WO1999011818A8; AU8886798A

Abstract

A molecular design method which fundamentally comprises artificially reconstructing genes by means of exon shuffle among a number of individuals. More particularly, this method comprises the steps of (a) synthesizing a number of polypeptides or nucleic acids differing in sequence from each other; (b) measuring the fitness of the synthesized polypeptides or nucleic acids on the laboratory level, (c) ranking the polypeptides or nucleic acids synthesized in step (a) depending on the fitness; (d) preparing a 'shuffling library' of the sequences obtained by shuffling the partial structures among individuals selected depending on the rank; (e) synthesizing polypeptides or nucleic acids belonging to the above library; and (f) repeating the procedures of steps (b) to (e) arbitrary times with the use of the polypeptides or nucleic acids obtained in step (e). By using this method, molecules can be evolved drastically and quickly as compared with the conventional genetic algorithm, which makes it possible to efficiently detect functional polymers, in particular, polypeptides or nucleic acids having specific functions and having novel structures which never occur in nature.

Description

Method for searching for highly functional polypeptide or nucleic acid

The present invention relates to the field of biotechnology, and more particularly to the design of polypeptides or nucleic acids.

Biological macromolecules such as proteins, DNA, and RNA exhibit various functions such as physiological activity, molecular recognition, and catalytic activity. If it becomes possible to freely design the functions of such polymers, it will be possible to create entirely new functional polymers, which will have a major impact on a wide range of fields such as pharmaceuticals and foods. There is expected.

By the way, recently, research on designing a new functional biopolymer by exploring the sequence space has been actively conducted. For molecules formed by connecting L basic units of N types, NL types of arrays are possible, and an L-dimensional array space can be set. An arbitrary array is represented as one point in this array space. Focusing on a specific activity, each sequence shows an activity of a different size, so plotting the magnitude of the activity indicated by the corresponding sequence on each point in the sequence space gives an L-dimensional surface Appears. This is the fitness terrain, and searching the array space can be translated into walking on the fitness terrain.

The SELEX method (Ellington, AD & Szostak, JW, 1990, Nature 346, 818-822; Tuerk, C. & Gold, L., 1990, Science 249, 505-510.) Is a method of searching the sequence space. One. Creates a molecular population that contains all sequences of a specific length for DNA or RNA, and has the desired function by repeating selection using specific activity as an index and amplification of the selected molecule It concentrates molecules. Do the same for peptides by associating genetic information with phenotypes The phage display method (Scott, JK & Smith, GP, 1990, Science 249, 386-390.) And the polysome display method (Mattheakis, LC et. Al., 1994, Proc. Natl. Acad. Sci. USA 91) , 9022-9026.). However, in these methods, the enriched sequence is optimized because reactions of many molecules occur in the same solution, and in practice there is competition or inhibition that cannot occur in the case of single molecule reactions. It does not necessarily have the functions provided. Another disadvantage is that when two different sequences have the same activity, the longer sequence is less likely to be detected in the final base sequence decoding stage because of its low percentage in the population. (Sumikura, K. et. Al., Nucleic Acids Symp. Ser., In press.).

On the other hand, a solid-phase synthesis method (Fodor, SP A. et. Al., 1991, Science 251, 767-773.) Separately measures the activity of each sequence in the sequence space. In this method, it is necessary to know which sequence is present at which position on the substrate, perform a reaction on the substrate, monitor the magnitude of the reactivity, and know which sequence is highly active. is there. Since this technique examines the activity of all possible sequences individually, the length of the sequence that can be handled is limited, and it is only possible to examine activities that can be confirmed on the solid phase such as binding ability. Also, the Purging 'Strategy (Kauffman, SA & Macready, WG, 1995, J. theor. Biol. 173, 427-440.) Determines the optimal unit for each residue in the sequence in turn. This approach attempts to overcome the constraints on length and activity described on the left, but only when individual residues contribute independently to activity. Only valid.

As described above, in the strategy of repeating selection and amplification by treating the molecular population collectively, it is possible to verify many sequences, but it is doubtful that the obtained sequence is the optimal one. The measurement strategy limits the conditions that can be verified. Therefore, the operation of individually examining the activity of a part of individuals in the sequence space and deciding the next sequence whose activity is to be examined based on the results is repeated for several generations. A method was devised to return to approach the optimal solution (Yokobayashi, Y., 1996, J. Chem. Soc, Perkin Trans. 1, 2435-2437.) _{0 If} this strategy works properly, By examining only a small number of molecules in space, the optimal sequence can be known, and the obtained sequence is considered to be highly reliable as the optimal solution.

Yokobayashi et al. Conducted an experiment to determine the next generation of individuals using a “genetic algorithm” when adopting the above strategy (Yokobayashi, Y., 1996, J. Chem. Soc, Perkin Trans. 1, 2435-2437.) _{0 In} the evolution of life, selection of genes suitable for the environment, gene recombination, and mutations are thought to play important roles. It is a mathematical model of such an evolutionary mechanism (Forrest, S., 1993, Science 261, 872-878.). In a genetic algorithm, usually, a computer performs a mathematical operation on a set of character strings that resembles a gene by modeling three processes of selection, recombination, and mutation. Through this, we evolve a set of character strings into something that fulfills the purpose.

In the experiment described above, Yokobayashi et al. Examined the trypsin inhibitory activity of a peptide consisting of 6 amino acids over 24 generations per generation over 6 generations. The average value of the inhibitory activity increased constantly from the first generation to the last generation, but the maximum inhibitory activity was almost the same as the ratio of the final generation value to the initial generation value, which was 1.03. Was. This suggests that the conventional genetic algorithm works to some extent to increase the average value of the activity of each generation, but the search efficiency for the optimal solution is low. Disclosure of the invention

An object of the present invention is to provide a method for efficiently designing a highly functional biopolymer.

The evolution of a protein in nature is thought to proceed through the process of mutation of the gene that encodes it, recombination occurring between the genes of two parent individuals, and selection of individuals that adapt to the environment. The algorithm also models this process Is becoming

On the other hand, looking at the evolution over a long period of time, a protein has a higher function by connecting a variety of genes encoding polypeptides having a certain length as basic units, and connecting those units. Is believed to have acquired Most eukaryotic genes consist of a region that is translated into amino acids (exons) and an untranslated region between exons (introns). The amino acid sequence of a protein is encoded in multiple exons. In the long term, exons are the basic unit of protein evolution, and introns are thought to be an area for exon shuffling. (De Souza, SJ et. Al., 1996, Genes Cels 1, 493-505.). In fact, the same exons encoding certain domains of the epidermal growth factor protein have been found in completely unrelated proteins, the low-density lipoprotein receptor and blood coagulation factors IX and X (Sudhof, TC et.al., 1985, Science 228, 815-828.), Exon

• Suggests that shuffling has been a driving force in the evolution of evening protein. Exon shuffling may be able to produce a highly functional protein more quickly than testing all possible combinations using amino acid as a basic unit.

The present inventors have focused on this exon shuffling mechanism of evolution, and have basically made it possible to artificially rearrange exon shuffling-like genes among multiple individuals using a computer. A molecular design method called “shuffling strategy” was developed. By using this method, it is possible to make the evolution of molecules more dramatic and faster than conventional genetic algorithms, and thus it is possible to search for functional macromolecules more efficiently. .

That is, the present invention includes the following.

(1) A method for searching for a highly functional polypeptide or nucleic acid, comprising the following steps: (a) synthesizing a plurality of polypeptides or nucleic acids having different sequences from each other, (b) measuring the fitness of the synthesized polypeptide or nucleic acid at the laboratory level,

(c) ranking the polypeptides or nucleic acids synthesized in step (a) according to their fitness,

(d) Create a “shuffling library” consisting of sequences obtained by performing shuffling of partial structures among individuals selected according to the rank,

(e) synthesizing a polypeptide or nucleic acid belonging to the “shuffling 'library”, and (f) further performing any of the steps (b) to (e) on the polypeptide or nucleic acid obtained in the step (e). Repeat several times.

(2) In step (d), a sequence in which a mutation has been introduced into a specific sequence is prepared separately from the polypeptide or nucleic acid having the sequence obtained by shuffling. (1) The method according to (1), wherein the method is added to shuffling 'library one'.

(3) The method according to claim 2, wherein the specific sequence is a sequence having a certain rank or more in step (c).

(4) In step (d), at least one of the sequences obtained by shuffling is subjected to further mutation-introduced sequences and added to the “shuffling library”. (1) The method according to (1).

(5) The method according to any one of (1) to (4), wherein in step (d), shuffling is performed only between individuals having a certain rank or higher.

(6) The process according to (5), wherein in the step (d), individuals having a certain rank or higher are further divided into a plurality of groups from a higher rank to a lower rank, and shuffling is performed in each group. Method.

(7) A non-naturally occurring polypeptide or nucleic acid obtained by the method according to any one of (1) to (6) and having a certain degree of fitness or more.

In this specification, the term “shuffling” refers to dividing a polypeptide or nucleic acid sequence into a plurality of specific blocks (named “virtual exons”). This refers to the operation of exchanging virtual exons among a number of individuals to create a new polypeptide or nucleic acid sequence. As used herein, the term "shuffling 'library 1'" refers to a sequence group resulting from performing "shuffling" on a specific polypeptide or nucleic acid sequence.

In the present invention, a method known to those skilled in the art can be used to synthesize a polypeptide or a nucleic acid. For example, for a polypeptide, the t-Boc method (Merrifield, B., 1986, Science 232, 34, 347.) or Fmoc method (Gorka, J. et. Al., 1989, Pept. Res. 2, 376-80.). For nucleic acids, the phosphoramidite method (Itakura, K. et. Al., 1984, Ann. Rev. Biochem. 53, 323-356.) Etc. can be used.

As used herein, “fitness” is an index indicating how much a polypeptide or nucleic acid having an individual sequence exhibits a specific biological activity, and is usually determined by measurement at a laboratory level. For example, the fitness of a polypeptide includes binding activity to a target molecule or activity as an antibiotic. The former is based on the EUSA method (Creigh ton, TE (Ed), 1989, Protein Structure. A Practical Approach, IRL). Pres s.) Or BIAcore (Griffiths, DG & Hall, G, 1993, Tibtech 11, 122-130.), And the latter can be measured by examining the ability to inhibit the growth of cells. In addition, examples of the fitness of nucleic acid include binding activity to a protein and catalytic activity of cleaving a nucleic acid having a specific base sequence. The former is an ELISA method, a BIAcore, or a gel shift method (Latchman, DS, 1993, Transcription Factors). , IRL pres s.) And the latter using a radioactive label (Santoro, SW & Joyce, G.

Natl. Acad. Sci. USA 94, 4262-4266.) Or electrophoresis (Cantor,

C. R. & Schi, el, P. R., 1980, Biophysical Chemistry, Chapter 12, Freeman.).

In the present invention, the cycle of activity measurement and shuffling is usually performed 5 times (generation) or more, preferably 8 times or more, and more preferably 10 times or more. In addition, it is preferable to perform the shuffling on the array having the higher fitness, and generally perform the shuffling within the upper 50%. Preferably within the top 40%, more preferably within the top 25%. However, care must be taken if the array to be shuffled is too limited, because it is easy to end up with a local solution instead of the optimal solution.

Furthermore, individuals with a certain rank or higher can be divided into a plurality of groups according to the rank, and shuffling can be performed in each group. This allows for global shuffling between a large number of arrays and local shuffling between a small number of arrays in parallel. In this case, for example, it is possible to set up two groups within the top 20% and within the top 40%. In the case of, it is thought that the optimal solution search can be performed more quickly.

Furthermore, in the present invention, since only the rearrangement of sequences by shuffling virtual exons limits the materials that can be used, more mutations can be introduced by introducing mutations into sequences existing as virtual exons. It is important to give the diversity of sequences to the sequence group. Mutations include all common mutations, including point mutations, deletion mutations, insertion mutations, inversions, and translocations. Specifically, separately from a polypeptide or nucleic acid having a sequence obtained by shuffling, a sequence in which a mutation has been introduced into a specific individual is created, and a “shuffling library” is created. Can be added to As the individual into which the mutation is to be introduced, it is preferable to select an individual having high fitness. As another method, a sequence in which a mutation is introduced into a polypeptide or nucleic acid having a sequence obtained by shuffling is further prepared and added to “Shuffling 'Library 1'”. it can.

In the present invention, as an array to be shuffled, an array having a high degree of “isolation” in addition to the degree of fitness can be preferentially selected. Here, the “isolation degree” of a sequence is an index indicating the low similarity of a specific sequence to another sequence of the same generation, and is a measure of the similarity of another sequence of the same generation to the specific sequence. It can be defined as being inversely proportional to the sum of "similarity" with the sequence. Here, The term “similarity” of a sequence refers to an index indicating the similarity between two sequences.For example, when two sequences are compared and the same residue or base is present at the same position, 1 Points can be quantified by the total number of points. Specifically, each sequence of the same generation is ranked by the product of “fitness” and “isolation”, and shuffling can be performed only between individuals having a certain rank or higher. In addition, each sequence of the same generation is ranked by the product of “fitness” and “isolation”, and individuals having a certain rank or more are further divided into a plurality of groups according to the rank. Shuffling can also be performed.

In the present invention, the initial sequence population (from which the first shuffling is performed) may include naturally occurring sequences. In this case, it is possible to further improve the function of the molecule that exists in nature.

In the present invention, a random sequence may be used as the initial sequence. Even if a random sequence is used as the starting sequence, there is no problem in practical use because the fitness is markedly increased by the “shuffling strategy”. Rather, it can be said that using a random sequence as the starting sequence increases the possibility of reaching a highly active biopolymer having a structure significantly different from that of the natural type.

In conventional biotechnology, a method of first searching for a biopolymer and then analyzing the function of the molecule, or analyzing a specific biological phenomenon first, and identifying a biopolymer that induces the phenomenon The technique of searching was taken. Regardless of the method, the biopolymers obtained were limited to those directly involved in life phenomena. In the present invention, polypeptides or nucleic acids that are not directly involved in life phenomena are also searched for. That is, the major feature of the present invention is that it is possible to freely search for polypeptides or nucleic acids involved in not only life phenomena but also all chemical reactions and formation of structures having specific functions. BRIEF DESCRIPTION OF THE FIGURES FIG. 1 shows the ratio of the binding constant (KA) between the peptide and the antibody.

Figure 2 compares the shuffling strategy and the genetic algorithm. The horizontal axis represents the variation of the molecule, and the vertical axis represents the average of the maximum value in 100 trials.

FIG. 3 is a diagram showing the structure of a double G quartet.

FIG. 4 is a diagram showing the structure of a triple G quartet. BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, the present invention will be described in more detail with reference to Examples, but the present invention is not limited to the following Examples.

[Example 1] Simulation of design of a novel peptide that binds to a known protein (Search for topography of fitness based on binding data between a hemagglutinin-derived peptide and a monoclonal antibody)

To characterize the molecular design based on the shuffling strategy, we set the topography of the fitness based on the actual measurement data, performed shuffling, and showed affinity for influenza virus hemagglutinin (HA). High protein was searched.

(1) Experiments conducted under the topography of fitness

Influenza virus hemagglutinin (HA) is a viral membrane protein that is used as an antigen to produce antibodies against the virus in the host body. HA is a trimer consisting of the same subunit of 550 residues, and its monomer is composed of two polypeptide chains, HA1 and HA2 (Wilson, IA. Et. Al., 1981). , Nature 289, 366-73.) O

It is known that a peptide corresponding to HA100 at positions 100 to 108 binds to a monoclonal antibody called Fab 17/9 (Rini, JM et. Al., 1992, Science 255, 959-96 5.) o Churchil l, for all 1 amino acid substitutions (113 in total) of peptides corresponding to positions 101-107 of HA1 (101D-102V-103P-104D-105Y-106A-107S), Fab 17/9 The binding capacity to IC50 is determined by competitive ELISA, and the IC 50 value is calculated for each. (Churchill, MEA et. Al., 1994, J. Mol. Biol. 241, 534-556.) _{0 The} IC50 here is the Fab 17/9 of the control peptide having the wild-type sequence. Is the concentration of each anatomical substance required to inhibit binding by 50%.

(2) Calculation of fitness

Since the IC50 may be considered to be proportional to the dissociation constant (KD) between the peptide and the antibody (Aita, T. & Husimi, Y., 1996, J. theor. Biol. 182, 469-485.) The inverse of is proportional to the binding constant (KA) between the peptide and the antibody. Figure 1 (created based on Churchill et al., Figure 7) shows the value obtained by dividing the reciprocal of the IC 50 of each analog by the reciprocal of the IC 50 of the control peptide. The letters represent the one-letter code for amino acids). For the above reasons, these values can be considered to be the ratio of the KA values of the analog and the control peptide.

In terms of protein-protein interaction and protein-DNA binding, statistically, the free energy change is additive (Wells, J.A. 1990, Biochem. 29, 8509-8517.). Therefore, regarding the binding ability of a peptide consisting of 7 amino acids to Fab 17/9, it can be considered that amino acid substitution at all sites acts independently on the binding ability. In other words, X represents a certain amino acid substitution with respect to the wild-type sequence, Y represents a sequence having a single amino acid substitution at another site, and (X, Υ) represents a sequence having both of these two amino acid substitutions. ,

△△ G (X, Y) = AAG (X) + AAG (Y)

(Note that AG represents the change in free energy due to binding, and AAG represents the change in AG due to amino acid substitution with respect to the wild-type.) The relationship between the coupling constant KA and free energy AG is

KA = exp (-AG / RT)

(Aita, T. & Husimi, Y., 1996, J. theor. Biol. 182, 469-485.) Under the above assumption,

KA (substitution type) / KA (wild type) = exp (— AAG (substitution type) / RT) Holds. Therefore,

8,) /! ^ 8 (wild type) = 6 (-AAG (X, Y) / RT)

= exp (-(AAG (X) + AAG (Y)) / T)

= exp (-AAG (X) / RT) exp (-AAG (Y) / RT)

Two (KA (X) / KA (wild type)) · (KA (Y) / KA (wild type))

By multiplying the value of FIG. 1 for an arbitrary amino acid sequence, the relative value of the equilibrium binding constant (KA) for Fab 17/9 can be determined.

(3) Computer simulation conditions

We simulated the molecular design using a shuffling strategy and a simple genetic algorithm. Details of the conditions are as follows.

[Shuffling strategy]

First, we generated 20 peptides with random sequences. The fitness (binding ability to Fab 17/9) of each peptide can be obtained from the calculation based on the values in Fig. 1 by (2). The peptides were ranked according to their binding capacity. Next, 7 amino acids were divided into 2-2-3 to set 3 virtual exons, and virtual exons were shuffled among the top 8 individuals to create 8 individuals of the next generation. In shuffling, we made it easier to leave virtual exons in proportion to fitness. For the top two individuals, six individuals were created so that all combinations of virtual exons occurred. In addition, a new individual was created so that each of the top three individuals had one mutation per individual. Two sudden mutants were generated from one individual. In this way, 20 individuals of the next generation were created from 20 individuals of one generation.

The calculation of fitness and the generation of next-generation individuals based on the results were performed for 40 generations. The search for 40 generations was counted as one trial, and 100 trials were performed to examine the directed evolution of the binding ability.

[Genetic algorithm] First, we generated 60 peptides with random sequences. The calculation of the fitness of each array is the same as above. In the generation of the next generation individuals, two new individuals were created by selecting two parent individuals with a probability proportional to the binding ability and causing recombination between them at random positions. In addition, mutation was assumed to occur suddenly with a probability of 0.1. Thus, 60 individuals of the next generation were produced.

(4) Simulation results

The average value of fitness in each generation was averaged over 100 trials, the average value of fitness in each generation was averaged over 100 trials, and the minimum value of fitness in each generation was 100 Averaged for trials, averaged over 100 trials for the variation of molecules that were present in the population up to that generation (that is, the number of molecules that actually need to be synthesized) Values were calculated for each. Figure 2 compares the shuffling strategy and the genetic algorithm with the variation of the molecule on the horizontal axis and the average of the maximum value in 100 trials on the vertical axis. This shows that the shuffling strategy searches for the optimal solution more efficiently than the genetic algorithm. In the same way, the shuffling strategy is also more genetic for the average of the fitness values for each generation, averaged over 100 trials, and for the average of the minimum fitness values for each generation, averaged over 100 trials. It was found that the search for the optimal solution was more efficient than the genetic algorithm. Comparing how many individuals must be synthesized before the fitness exceeds 1 for the first time, the shuffling strategy shows 129 individuals and the genetic algorithm has 155 individuals. Can also see the high search efficiency of shuffling strategies for genetic algorithms.

(5) Shuffling under different conditions In order to confirm the reproducibility of the search efficiency of the shuffling strategy, shuffling was performed under different conditions. The difference from the above is that when shuffling leaves a virtual exon in proportion to the fitness, the fitness of the sequence is maintained so that the diversity of the sequence is maintained even in the early generations where the fitness difference between individuals is large. That is, we performed sigma scheduling. In the present embodiment, the original fitness is f, the fitness after conversion is f ′, the average fitness of individuals of the generation is f avr, and the standard deviation of the fitness is calculated as f = f− (f avr- 3 x σ)

Was converted by the following equation. Taking the average of 100 trials, the number of molecules synthesized until the fitness exceeded 1 for the first time was 119, confirming the high search efficiency of the shuffling strategy.

[Example 2] Optimization of sequence of single-stranded DNA binding to thrombin

In the first embodiment, based on the statistical knowledge, based on the assumption that the additivity of the free energy change is completely established, the topography of the fitness was set and the computer simulation was performed. On the other hand, when designing a molecule by actually measuring the fitness at the laboratory level, it is considered that multiple amino acid-substituted / substituted salt substitutions may affect the fitness in some cases. In addition, the situation that is different from the simulation occurs, for example, the fitness of all the arrays cannot be measured due to the detection limit of the measuring device. Therefore, we conducted the following experiments, including the process of actually synthesizing molecules and measuring fitness.

(1) Thrombin and thrombin.

Thrombin is a type of serine protease and has various functions including blood coagulation. 5'-GGTTGGTGTGGTTGG-3, a single-stranded DNA consisting of 15 residues (SEQ ID NO: 1), called thrombin-abumauma, is known to bind to thrombin and inhibit its biological activity (Bock, LC et. Al., 1992, Nature 355, 564-566.) ₀ As shown in Fig. 3, this molecule has a three-dimensional structure stabilized by two G-quartet structures. (Wang, KY et. Al., 19 93, Biochemistry 32, 1899-1904.).

In this study, we set three parts other than the two G-quartet structures of the thrombin 'abnumer as virtual exons, and examined how these parts are optimized. That is,

5 '-GGNNGGNNNGGNNGG-3' (N represents A, C, G ヽ or T)

Starting with the next 20 oligo nucleotides, the next generation of shuffling strategies, synthesis and fitness measurements were repeated to track the resulting sequences.

(2) Fitness measurement method

The fitness of each sequence (affinity to thrombin) was measured using BIAcore2000 (BIAcore AB), an experimental device using the principle of surface plasmon resonance.

The synthesized single-stranded DNA was immobilized on a sensor chip for BIAcore2000 via carboxymethyl dextran and streptavidin, and 20 M of thrombin was passed thereto, and the response was examined. In BIAcore, when a molecule binds to an immobilized substance, the refractive index near the sensor chip changes, and this is detected as a change in the resonance unit.

(3) Practical shuffling strategy

We synthesized 20 individuals of the first generation and measured fitness. The top 8 individuals were as follows. The numerical value on the right side of the sequence is a numerical value obtained by BIAcore2000 and indicating the degree of fitness.

0-12: GG TC GG GTG GG TT GG (SEQ ID NO: 2) 1299.9

0-8: GG CC GG TTC GG TT GG (SEQ ID NO: 3) 1281.5

0-15: GG TG GG TAC GG CT GG (SEQ ID NO: 4) 1101.9

0-17: GG CG GG GCG GG TG GG (SEQ ID NO: 5) 1077.8

0-3: GG CA GG TAG GG TA GG (SEQ ID NO: 6) 1000. 1

0-9: GG GC GG GTC GG AT GG (SEQ ID NO: 7) 740.3 0-10: GG GC GG TTA GG CA GG (SEQ ID NO: 8) 526.5

0-14: GG TA GG GCA GG AG GG (SEQ ID NO: 9) 452.3

By shuffling between the top two individuals with the highest fitness, the following six individuals were newly created. In the following, the first, second, and third virtual exons counted from the left are denoted as (1), (2), and (3).

-1: GG TC GG GTG GG TT GG: 0-12 (1), 0-12 (2), 0-8 (3) (SEQ ID NO: 2)-2: GG TC GG TTC GG TT GG: 0: 12 (1), 0-8 (2), 0-8 (3) (SEQ ID NO: 10) ― 3: GG CC GG TTC GG TT GG: 0: 0-8 (1), 0-8 (2), 0 -12 (3) (SEQ ID NO: 3)-4: GG CC GG GTG GG TT GG: 0-8 (1), 0-12 (2), 0-12 (3) (SEQ ID NO: 11)-5 : GGTC GG TTC GG TT GG: 0-12 (1), 0-8 (2), 0-12 (3) (SEQ ID NO: 10)-6: GG CC GG GTG GG TT GG: 0-8 ( 1), 0-12 (2), 0-8 (3) (SEQ ID NO: 11) Also, among the top 8 individuals with the highest fitness, virtual exoxons tend to remain in proportion to the magnitude of the fitness. By shuffling as described above, the following eight individuals were newly created.

1-7: GG GC GG TTC GG TT GG: 0-10 (1), 0-8 (2), 0-12 (3) (SEQ ID NO: 12) 1-8: GG TC GG TAC GG AG GG: 0-12 (1), 0-15 (2), 0-14 (3) (SEQ ID NO: 13) 1-9: GG CG GG GCA GG AT GG: 0-17 (1), 0-14 (2 ), 0-9 (3) (SEQ ID NO: 14) 1-10: GG TG GG TAC GG TT GG: 0-15 (1), 0-15 (2), 0-8 (3) (SEQ ID NO: 15) 1-11: GG CG GG TAG GG AT GG: 0-17 (1), 0-3 (2), 0-9 (3) (SEQ ID NO: 16) 1-12: GG TC GG TAC GG TT GG: 0-12 (1), 0-15 (2), 0-8 (3) (SEQ ID NO: 17) 1-13: GG GC GG GCG GG TT GG: 0-9 (1), 0-17 (2), 0-8 (3) (SEQ ID NO: 18) 1-14: GG CG GG GCG GG CT GG: 0-17 (1), 0-17 (2), 0-15 (3) (sequence No .: 19) Furthermore, mutations were introduced into the top three individuals with a probability of 2/7, and the following six individuals were newly created.

1-15: GG TC GG GTT GG TA GG: Mutation introduced into 0-12 (SEQ ID NO: 20)

1-16: GG TG GG GGG GG TT GG: Mutation introduced into 0-12 (SEQ ID NO: 21) 1-17: Mutation introduced into GG CC GG TCC GG TG GG: 0: 8 (SEQ ID NO: 22) 1-18: Mutation introduced into GG CC GG CTC GG AT GG: 0-8 (SEQ ID NO: 23)

1-19: GG TG GG TAA GG AT GG: Mutation introduced in 0-15 (SEQ ID NO: 24)

1-20: GG TT GG TAT GG CT GG: Mutation introduced into 0-15 (SEQ ID NO: 25)

In this way, 20 individuals of the second generation were prepared. Of these, 1-2 and u5, and 1-4 and u6 are the same. Also, 1-1 and 3 are the same as those that existed in the previous generation. Therefore, 16 types of sequences were newly synthesized.

In the same way, the design, synthesis and fitness measurement of the new generation were repeated.

(4) Result

As the generations were repeated as described above, those close to the sequence known as thrombin abdomer (5'-GGTTGGTGTGGTTGG-3, / SEQ ID NO: 1) came to dominate. At the same time, a certain percentage of G-rich sequences that were not so close remained. If you look closely at these G-rich sequences, you can see that another G-quartet structure that is different from the two G-quartets that originally exist, like 0-12 and 0-17 shown above, It was expected that the sequence could be assembled together. From this, it was speculated that a sequence having three G-quartets may also have high binding ability to thrombin.

Therefore, we synthesized 5, -GGGTTGGGTTGGGTTGGG-3 '(SEQ ID NO: 26), a sequence expected to form three G-quartets in the form shown in Fig. 4 (referred to as "tril8"). Binding was checked on a BIAcore2000. In a buffer (lOmM Hepes (pH 7.4), 150mM NaCl, 5mM KCl, 2mM CaC12, ImM MgCl2) with ionic conditions close to the blood, `` tril8 '' and thrombin interact at 37 ° C, and analysis software When the equilibrium dissociation constants were determined using (BIAevaluation ver 2.1 / Biocore AB), under these conditions, the prototype thrombin 'Abuma-maichi' and 'tril8' had almost the same KD value (8.96x10- The KD values of 8M and 9.16x10-8M, "tril8" were 1.02 times the KD value of the prototype). High structural stability provided by three G-quartets Considering this, “tril8” is expected to be clinically applied as an effective thrombin inhibitor, surpassing the prototype.

Thus, based on the sequence information obtained during the molecular design process using the shuffling strategy, it became clear that the previously unknown sequence exhibited almost the same thrombin binding ability as the known sequence. Was. Industrial applicability

According to the present invention, it has become possible to efficiently search for highly functional polypeptides or nucleic acids. For example, it has become possible to dramatically improve the function of naturally occurring polypeptides or nucleic acids. Furthermore, it has become possible to freely search for a polypeptide or nucleic acid having a specific function and a novel structure that does not exist in nature.

Claims

The scope of the claims

1. A method for searching for a highly functional polypeptide or nucleic acid, comprising the following steps:

(a) synthesizing a plurality of polypeptides or nucleic acids having different sequences from each other,

(b) measuring the fitness of the synthesized polypeptide or nucleic acid at the laboratory level,

(d) Create a “shuffling library 1” consisting of sequences obtained by performing shuffling of partial structures between individuals selected according to the rank,

(e) synthesizing a polypeptide or nucleic acid belonging to "shuffling 'library 1'"; and (f) subjecting the polypeptide or nucleic acid obtained in step (e) to optional steps (b) to (e). Repeat several times.

2. In step (d), a sequence in which a mutation has been introduced into a specific sequence is prepared separately from the polypeptide or nucleic acid having the sequence obtained by shuffling, and “shuffling” is performed. Ring library "

Method according to 1.

3. The method according to claim 2, wherein the specific sequence is a sequence having a certain rank or more in step (c).

4. In step (d), at least one of the sequences obtained by shuffling is subjected to further mutation-introduced sequences and added to the “shuffling-library”. The method of claim 1, wherein:

5. The method according to any one of claims 1 to 4, wherein in step (d), shuffling is performed only between individuals having a certain rank or higher.

6. The method according to claim 5, wherein in the step (d), individuals having a certain rank or higher are further divided into a plurality of groups from a higher rank to a lower rank, and shuffling is performed in each group. .

7. A non-naturally occurring polypeptide or nucleic acid obtained by the method according to any one of claims 1 to 6 and having a certain degree of fitness or more.