A device for predicting difficult-to-mutate epitope sites, a method for predicting difficult-to-mutate epitope sites, and a program for predicting difficult-to-mutate epitope sites.

Abstract

To provide a prediction device, method and program of a hard-to-mutate epitope site capable of highly accurately predicting and identifying the hard-to-mutate epitope site for a broad-spectrum neutralizing antibody by molecular simulation using a computer; and to provide an antibody screening device, method and program.SOLUTION: A device 100 for predicting a hard-to-mutate epitope site in an antigen, includes a prediction part 171 that predicts an amino acid residue whose fluctuation is a prescribed value or less, as a hard-to-mutate site in the epitope site based on fluctuation information for every amino acid residue in the epitope site in the antigen.SELECTED DRAWING: Figure 15

Description

[Technical Field]

[0001] The present invention relates to a device for predicting difficult-to-mutate epitope sites, a method for predicting difficult-to-mutate epitope sites, and a program for predicting difficult-to-mutate epitope sites, as well as an antibody screening device, an antibody screening method, and an antibody screening program. [Background technology]

[0002] In recent years, there has been a growing demand for the development of vaccines, neutralizing antibodies, and other pharmaceuticals to test for, prevent, or treat infectious diseases caused by influenza viruses such as influenza A virus, and more recently, by viruses that readily mutate, such as the novel coronavirus SARS-CoV-2.

[0003] However, in viruses that readily produce mutants, frequent mutations in the viral genome can lead to changes in the amino acid sequence of the viral surface protein that can be recognized as an antigen, altering its antigenicity. This can result in the virus becoming evaded by the immune system that has acquired a response to a particular mutant, or drugs optimized for a specific mutant failing to recognize the mutated antigen.

[0004] Therefore, there is a need to develop methods for predicting and identifying difficult-to-mutate epitope sites that are less affected by viral mutations with high accuracy, and to develop antiviral monoclonal antibodies (broad-spectrum neutralizing antibodies) that have a broad spectrum of binding ability that is less affected by viral mutations. To date, monoclonal neutralizing antibodies against hemagglutinin, a surface protein, have been reported as broad-spectrum neutralizing antibodies against various variants of influenza A virus (see, for example, Non-Patent Document 1). However, conventional techniques for producing monoclonal neutralizing antibodies require (wet) experiments, which pose a significant cost and time burden due to the need for epitope selection and preparation, as well as evaluation of binding ability.

[0005] Furthermore, methods for predicting and analyzing the dynamic structure of natural peptides and proteins without conducting wet lab experiments include, for example, using computers to calculate the motion of individual atoms in the target molecule. By performing such computer-based molecular simulations, it is sometimes possible to efficiently evaluate the dynamic structure of molecules such as natural peptides and proteins in a short amount of time.

[0006] One known method for molecular simulation is molecular dynamics, which calculates the motion of individual atoms based on Newton's equations of motion. However, a technique has not yet been established that can predict (evaluate) with sufficient accuracy difficult-to-mutate epitope sites that are less affected by viral mutations, using molecular simulations such as molecular dynamics. [Prior art documents] [Non-patent literature]

[0007] [Non-Patent Document 1] DCEkiert et al.,Science 2009,10 APRIL Vol 324 246-251 [Overview of the project] [Problems that the invention aims to solve]

[0008] The present invention aims to solve the aforementioned conventional problems and achieve the following objectives. Specifically, the present invention aims to provide a predictor of difficult-to-mutate epitope sites that can predict and identify difficult-to-mutate epitope sites for broad-spectrum neutralizing antibodies with high accuracy using computer-based molecular simulations. [Means for solving the problem]

[0009] As a means for solving the above problems, the present invention provides a predictive device for predicting mutable epitope sites in an antigen, which has a predictive means that predicts an amino acid residue whose fluctuation is less than or equal to a predetermined value as a mutable site in the epitope site, based on information on the fluctuation of each amino acid residue in the epitope site of the antigen. [Effects of the Invention]

[0010] According to the present invention, the aforementioned problems in the conventional approach can be solved, the aforementioned objectives can be achieved, and a device for predicting reluctant epitope sites for broad-spectrum neutralizing antibodies can be provided that predicts and identifies reluctant epitope sites with high accuracy using computer-based molecular simulations. [Brief explanation of the drawing]

[0011] [Figure 1] Figure 1 is a schematic diagram showing the three-dimensional structure of the monomer and trimer of hemagglutinin. [Figure 2] Figure 2 is a schematic diagram illustrating the infection mechanism of the influenza virus. [Figure 3] Figure 3 is a graph showing the mutation accumulation rate for each amino acid residue in the amino acid sequence of the hemagglutinin monomer. [Figure 4] Figure 4 shows the initial structure of the hemagglutinin trimer used in the molecular dynamics calculation example of the present invention. [Figure 5] Figure 5 is a graph showing the magnitude of fluctuations (residual RMSF) of each amino acid residue in a hemagglutinin trimer, calculated based on an example of molecular dynamics calculations in the present invention. [Figure 6A] Figure 6A is a scatter plot showing the relationship between hemagglutinin fluctuations and mutation accumulation rates. [Figure 6B] Figure 6B is a histogram showing the relationship between hemagglutinin fluctuations and mutation accumulation rates. [Figure 7A]FIG. 7A is a table showing the average mutation accumulation rate of the amino acid residue group corresponding to the target section when the fluctuation (rmsf) is divided into sections at 0.1 Å intervals. [Figure 7B] FIG. 7B is a graph showing the average mutation accumulation rate of the amino acid residue group corresponding to the target section when the fluctuation (rmsf) is divided into sections at 0.1 Å intervals. [Figure 8] FIG. 8 is a graph showing the surface exposure ratio (rASA) of each amino acid residue in the hemagglutinin trimer calculated based on the calculation example of the exposed area ratio in the present invention. [Figure 9A] FIG. 9A is a schematic diagram showing the complex structure of the broad-spectrum neutralizing antibody (CR6261) in the prior art and hemagglutinin. [Figure 9B] FIG. 9B is a schematic diagram showing the binding site of the broad-spectrum neutralizing antibody (CR6261) in the prior art and the epitope. [Figure 9C] FIG. 9C is a schematic diagram showing the structure of the epitope of the broad-spectrum neutralizing antibody (CR6261) in the prior art and the sequence conservation degree among other mutant influenza viruses. [Figure 10] FIG. 10 is a graph showing the mutation accumulation rate in the epitope Helix A (385-401) of the broad-spectrum neutralizing antibody (CR6261). [Figure 11] FIG. 11 is a graph showing the fluctuation (rmsf) in the epitope Helix A (385-401) of the broad-spectrum neutralizing antibody (CR6261). [Figure 12] FIG. 12 is a graph showing the surface exposure ratio (rASA) in the epitope Helix A (385-401) of the broad-spectrum neutralizing antibody (CR6261). [Figure 13] FIG. 13 is a block diagram showing an example of the hardware configuration of the prediction device for the difficult-to-mutate epitope site of the present invention. [Figure 14] FIG. 14 is a block diagram showing another example of the hardware configuration of the prediction device for the difficult-to-mutate epitope site of the present invention. [Figure 15]Figure 15 is a block diagram showing an example of the functional configuration of the anti-mutation epitope site prediction device of the present invention. [Figure 16] Figure 16 is a flowchart showing an example of the flow when performing a method for predicting difficult-to-mutate epitope sites using the present invention. [Modes for carrying out the invention]

[0012] (Prediction device for difficult-to-mutate epitope sites) The present invention's predictor of difficult-to-mutate epitope sites is based on the inventors' findings that, when predicting difficult-to-mutate epitope sites for antiviral monoclonal antibodies (broad-spectrum neutralizing antibodies) that have a broad spectrum of binding ability less affected by viral mutations against viruses prone to mutations, sequence analysis and molecular dynamics analysis of the target antigen protein revealed that the sites of amino acid residues that are less prone to mutation in the antigen protein are related to the sites of amino acid residues with small fluctuations.

[0013] Therefore, as a specific example of a virus that is prone to mutation, we will first explain the infection mechanism of the human-derived influenza virus, which has many reported mutations, and the hemagglutinin (HA) protein, which is being studied as an antigenic protein for vaccines and neutralizing antibodies.

[0014] <Influenza virus and hemagglutinin protein> Hemagglutinin (HA) consists of a homotrimer, and the monomer is composed of two regions: HA1 (amino acid residue numbers: 1-343) and HA2 (amino acid residue numbers: 345-566). HA1 contains a target binding site, that is, a site that binds to sialic acid-containing sugar chains on the cell surface in the human host. Figure 1 is a schematic diagram showing the three-dimensional structures of the hemagglutinin (HA) monomer and trimer. In Figure 1, the left diagram schematically shows the three-dimensional structure of the hemagglutinin monomer, and the right diagram schematically shows the three-dimensional structures of the hemagglutinin trimer (A chain, B chain, and C chain) using ribbon diagrams. In the hemagglutinin monomer, the ribbon diagram of HA1 is shown in black, and the ribbon diagram of HA2 is shown in gray.

[0015] Figure 2 is a schematic diagram illustrating the role of hemagglutinin in the process of viral adsorption and entry during influenza virus infection (Source: https: / / pdbj.org / mom / 076). In the host human, the HA1 region of hemagglutinin, a surface protein of the influenza virus, binds to the cell surface receptor protein (human receptor). After the HA1 region is cleaved, the HA2 region undergoes a significant structural change, leading to membrane fusion between the viral membrane and the human cell membrane, and thus the infection progresses.

[0016] The inventors, using computer-based molecular simulations, evaluated the characteristics of each amino acid residue from trajectories obtained by molecular dynamics calculations at room temperature for approximately 300 nanoseconds, using the stable structure of the antigen protein in aqueous solution as the initial structure. Through diligent research into the relationship between these characteristics and the resistance to mutation of each amino acid residue in the antigen protein, they discovered that information on fluctuations in each amino acid residue is related to the mutation accumulation rate, which indicates the resistance to mutation of each amino acid residue. Furthermore, they found that by calculating the ratio of exposed area of ​​each amino acid residue as an indicator of how easily antibodies can approach epitope sites in the antigen, it is possible to predict and evaluate sites on which antibodies are likely to bind to the antigen. Based on the above findings, the inventors conceived that by analyzing fluctuation information for each amino acid residue in the epitope site of an antigen, it is possible to predict with high accuracy sites that are difficult to mutate in the epitope site (difficult-to-mutate epitope sites), and that difficult-to-mutate epitope sites can be excellently utilized as epitope sites for the production and selection of broad-spectrum neutralizing antibodies, and for vaccines that are less affected by viral mutations. Furthermore, we conceived the idea that by predicting and evaluating epitope sites to which antibodies readily bind to antigens, we can accurately predict epitope sites that are less prone to mutation and to which antibodies readily bind.

[0017] <Example of sequence analysis> First, in order to more precisely identify the relationship between information on fluctuations in each amino acid residue at the epitope site of the antigen and the sites in the epitope site that are less prone to mutation, the inventors analyzed previously reported primary amino acid sequence data for the hemagglutinin (HA) protein of the influenza virus and analyzed the sites that are less prone to mutation by calculating the mutation accumulation rate at each amino acid residue position. The following describes in detail the sequence analysis of the amino acid sequence of hemagglutinin (HA) conducted by the present inventors.

[0018] <<Sequence analysis of the amino acid sequence of hemagglutinin (HA)>> From the NCBI's influenza-related protein sequence database (Influenza Virus Resources: https: / / www.ncbi.nlm.nih.gov / genomes / FLU / Database / nph-select.cgi?go=database), primary amino acid sequence data for hemagglutinin (HA) of human influenza A was obtained using the following search criteria (45,545 sequences in total). [Search Criteria] Type: A, Host: Human, •Country / Region:any, Protein: HA, ·subtpe:H:any(1-18),N:any(1-11), ·Chain length: total length 566±5 residues

[0019] Subsequently, using an analysis program developed by the inventors, unreliable sequences containing X (unknown amino acid) in the primary sequence were removed (~43,600 sequences), and unique sequence data (~14,173 sequences) were extracted after removing duplicate sequences. Of the extracted unique sequences, the oldest HA sequence in the database from 1918 (NCBI ID: AAD17229: Sequence ID 1) was used as the reference sequence, and multiple alignment was performed using MAFFT (Multiple Alignment using Fast Fourier Transform, https: / / mafft.cbrc.jp / alignment / server / ), an online software capable of aligning a vast number of matching or similar amino acid and nucleotide sequences. The aforementioned reference sequence, shown in Sequence ID No. 1, has a total of 566 amino acid residues and is the amino acid sequence of hemagglutinin at the time of the 1918 Spanish flu pandemic (NCBI ID: AAD172229).

[0020] <<Analysis of mutation accumulation rate>> Next, the mutation accumulation rate at each amino acid residue position was calculated using the aligned sequence, and each region in the hemagglutinin (HA) monomer was color-coded (Figure 3). Here, the "mutation accumulation rate" is calculated by dividing the number of unique sequences containing amino acid residues mutated from the amino acid residues in the reference sequence by the total number of unique sequences (14,173) for each amino acid residue position in the reference sequence.

[0021] Figure 3 shows the results of an analysis of the mutation accumulation rate for each amino acid residue in the amino acid sequence of the hemagglutinin (HA) monomer. In Figure 3, the horizontal axis represents amino acid residue numbers, and the vertical axis represents the mutation accumulation rate relative to the reference sequence (AAD17229). HA1 (1-343) is shown in black, and HA2 (345-566) is shown in gray.

[0022] From the analysis of the mutation accumulation rate for each amino acid residue of hemagglutinin, the inventors found that mutations had accumulated in many amino acid residues (approximately 560 amino acid residues out of a total of 565 amino acid residues) over the approximately 100 years since the outbreak of the Spanish flu in 1918. Furthermore, the average mutation accumulation rate for HA2 (approximately 0.310) was lower than that for HA1 (HA1: approximately 0.452), indicating that the sequence conservation of HA2 is higher than that of HA1.

[0023] Based on the above, we confirmed that by calculating the mutation accumulation rate at each amino acid residue position in the hemagglutinin protein of influenza viruses, which have been reported to date with numerous variants, it is possible to identify amino acid residue positions with low mutation accumulation rates as sites that are less prone to mutation. Here, the mutation accumulation rate at the site that is difficult to mutate can be appropriately selected depending on the target molecule, the antigen protein, but it is preferably 0.5 or less, more preferably 0.4 or less, even more preferably 0.3 or less, particularly preferably 0.2 or less, and most preferably 0.1 or less.

[0024] <Example of molecular dynamics calculation> Next, in order to more precisely identify the relationship between the fluctuation information of each amino acid residue in the epitope site of the antigen and the sites in the epitope site that are less prone to mutation, the inventors performed molecular simulations using a computer to analyze the structure of hemagglutinin (PDBID:3gbn) at the time of the 1918 Spanish flu pandemic. The following describes the details of the molecular simulation (molecular dynamics calculation) performed by the inventors.

[0025] In the calculation examples described below, molecular dynamics (MD) simulations were performed to analyze the dynamic structure of hemagglutinin in solution.

[0026] <<Initial structure>> The initial structure of hemagglutinin from 1918, the year of the Spanish flu pandemic, used for MD calculations, was downloaded from the Protein Data Bank (http: / / pdbj.org). Next, the region from the surface of peptide (A) to a distance of 10 Å was treated as a single box (cell), and water molecules were placed around peptide (A). The computational system was then neutralized by placing Na ions and Cl ions under physiological conditions ([NaCl] = 100 mM). Note that 1 Å is equal to 0.1 nm.

[0027] Figure 4 shows the initial structure of the hemagglutinin trimer (PDBID: 3gbn) used in this calculation example for molecular dynamics simulations. Hemagglutinin is shown using a ribbon diagram, with solvent molecules omitted.

[0028] <<Energy Minimization Calculation>> Next, positional constraints were applied to the heavy atoms (atoms other than hydrogen) that make up the peptide in the created initial structure, and the energy of the entire computational system was minimized using molecular mechanics (MM) calculations. By performing energy minimization calculations, unnatural structural distortions in the initial structure can be removed, and divergence of the time integral in the initial stages of molecular dynamics calculations can be avoided. Energy minimization calculations were performed using the steepest descent method, with an atomic displacement distance RMSD of 0.1 [Å] in the first step, a maximum number of calculation steps of 50,000, and a convergence criterion of RMSF (mean square force applied to the atom) of 100.0 [kJ / mol / nm]. In this calculation example, Amber ff99SB-ILDN was used as the molecular force field.

[0029] <<Molecular dynamics calculation>> Next, using the GROMACS package (GROMACS 2020.4) as the engine for molecular dynamics calculations, a short NVT (Non-Variable Dynamics) calculation (where the number of particles, volume, and temperature of the computational system are constant) was performed under periodic boundary conditions and with positional constraints on heavy atoms to equilibrate the solvent, followed by a short NPT (Non-Variable Dynamics) calculation (where the number of particles, pressure, and temperature of the computational system are constant) with positional constraints on heavy atoms.

[0030] Then, an NPT calculation (Bussi temperature control and Parrinello-Rahman pressure control) was performed on the structure with the above-mentioned solvent equilibration, with a simulation time of 300 ns (time step Δt = 2.0 fs, total 150,000,000 steps). Trajectories such as energy and coordinates were output every 50 ps (total 6,000 snapshots).

[0031] Molecular dynamics calculations require a Xeon CPU. (R) A computer with the following specifications was used: Platinum8280 (clock frequency 2.7GHz, 56 cores total), 768GB of memory, and four GPU cards (Volta V100). In this example, the molecular dynamics simulation of a single protein with a simulation time of 300 ns took approximately two days to complete when using 24 parallel cores and four GPU cards.

[0032] <Example of fluctuation analysis> <<Method for calculating fluctuations>> For each of the above 300 ns canonical MDs, trajectories (data of the trajectory of atomic motion) were extracted from 180 ns to 300 ns in order to perform analysis using data of the equilibrium state of hemagglutinin structure. Then, using "MDTraj" (https: / / mdtraj.org / 1.9.4 / index.html), a library for MD trajectory analysis, the magnitude of the deviation from the average position from 180 ns to 300 ns for each amino acid residue was calculated and defined as the residue-level fluctuation (residual-residue fission). RMSF showing fluctuations at the x-th amino acid residue. x It can be calculated using the following formula (1), and its unit is [Å]. rmsf x =√(1 / NΣ(rxj- <rxj>)2)...Equation (1) In equation (1), N represents 2,400 (the number of structures sampled between 180 ns and 300 ns), j represents an integer between 1 and 2,400, and rxj represents the position vector of the x-th amino acid residue at extraction time j. <rxj>This shows the time (180ns~300ns) average of the position vector of the x-th residue.

[0033] <<Analysis of fluctuations>> Figure 5 is a graph showing the magnitude of fluctuations (residual RMSF) of each amino acid residue in a hemagglutinin trimer, calculated based on an example of molecular dynamics calculations in the present invention. Specifically, after solvent relaxation, a 300 ns MD calculation was performed under NPT (N: particles, P: pressure, T: constant temperature) conditions at T=300 [K], and the magnitude of fluctuations (residual RMSF) of each amino acid residue relative to the average structure in equilibrium from 180 ns to 300 ns is shown. In Figure 5, the horizontal axis represents the amino acid residue number, and the vertical axis represents the fluctuation rmsf at each amino acid residue corresponding to amino acid residue numbers 18 to 517. 18~517 [Å] is indicated. HA1 (1-343) is shown in black, and HA2 (345-566) is shown in gray. Although hemagglutinin has a total of 566 amino acid residues, the region enclosed by the dotted line (1-17: signal peptide, 518-529: flexible linker, 530-550: viral transmembrane region, 551-566: cytoplasmic region) is a flexible region in aqueous solution, and since the coordinates of these regions were not described in the PDB registration data, fluctuations could not be calculated.

[0034] <<Analysis of the relationship between fluctuations and mutation accumulation rate>> Next, the relationship between hemagglutinin fluctuations and mutation accumulation rates was examined using scatter plots and histograms (Figures 6A and 6B). In the scatter plot of Figure 6A, HA1 (1-343) is shown in black and HA2 (345-566) is shown in gray. In the histogram in Figure 6B, the top of the figure shows a histogram where the horizontal axis is rmsf and the vertical axis is integrated, while the rightmost figure shows a histogram where the vertical axis is the ratio of mutations (ratio_of_mut) and the horizontal axis is integrated. In the hexagons in the figure, low density is shown in white and high density is shown in dark gray, with the degree of density indicated by a gradient of white, light gray, and dark gray. Analysis of Figures 6A and 6B revealed that the correlation between fluctuations and mutation accumulation rates is small (correlation coefficient ~0.17). Regarding mutation accumulation rates, three main clusters (clusters) were found: around 0, around 0.6, and around 1.

[0035] Furthermore, Figures 7A and 7B show the relationship between fluctuations in each amino acid residue and the mutation accumulation rate. Figure 7A is a table showing the average mutation accumulation rate of amino acid residue groups corresponding to a given segment when rmsf is divided into 0.1 Å increments. From left to right, the table shows the rmsf segment (0.1 Å increments, "2.5-2.4" in the table indicates 2.5 or less and greater than 2.4), the number of amino acid residues corresponding to each segment (#), the average mutation accumulation rate of amino acid residues corresponding to each segment (Ave. ratio_of_mutation), and the standard error (SE) of the mutation accumulation rate of amino acid residues corresponding to each segment. Figure 7B is a graph of the results from Figure 7A. In Figure 6B, the vertical axis shows the average mutation accumulation rate of amino acid residues corresponding to each rmsf category, and the horizontal axis shows the rmsf categories. Error bars indicate the standard error (SE) of the mutation accumulation rate.

[0036] Analysis of Figures 7A and 7B revealed a tendency for the mutation accumulation rate to decrease as the fluctuations decrease. In other words, the inventors discovered that there is a relationship between the fluctuation information of each amino acid residue at the epitope site in the antigen and the sites within the epitope site that are difficult to mutate. As a result, they found that by analyzing the fluctuation information of each amino acid residue at the epitope site in the antigen, it is possible to predict the sites within the epitope site that are difficult to mutate (difficult-to-mutate epitope sites) with high accuracy.

[0037] <Analysis and calculation of exposed area ratio> Furthermore, from the viewpoint of the ease with which antibodies can approach epitope sites in antigens, the inventors performed molecular simulations in which a sphere (radius 7.2 Å) sized to mimic the exposed CDR portion of an antibody was brought into contact with a hemagglutinin trimer to evaluate amino acid residues that can come into contact with the complementarity determining region (CDR) of the antibody. By calculating the ratio of exposed areas for each amino acid residue, the inventors evaluated the sites to which antibodies can easily bind to the antigen. The following describes in detail the analysis and calculation of the exposed area ratio in hemagglutinin (HA) trimers conducted by the present inventors.

[0038] <<Method for calculating the exposed area ratio>> Using the Gromacs command (gmx sasa), which is commonly used in MD analysis to calculate the Solvent Accessible Surface Area (SASA), we changed the radius of the sphere corresponding to the radius of the solvent molecule (1.4 Å) to 7.2 Å, which corresponds to the size of the CDR exposed portion of the antibody, based on prior literature (OCGrant et al. Scientific Reports volume 10, Article number: 14991 (2020)), and calculated the Accessible Surface Area (ASA) accessible by the antibody for each amino acid residue.

[0039] The relative accessible surface area (rASA) of the obtained amino acid residue X was calculated by dividing it by the maximum solvent-exposed area of ​​amino acid residue X. If the calculation was 100% or more, it was set to 100%. Here, the maximum solvent exposure area of ​​amino acid residue X is based on prior literature (MZTien et al. PLoS One. 2013;8(11):e80635.), and represents the solvent exposure area of ​​amino acid residue X when the tripeptide (GLY-X-GLY) centered on amino acid residue X adopts an Extended structure.

[0040] <<Analysis of Exposed Area Ratio>> Following the above procedure, MD calculations were performed for a total of 300 ns with Δt=2 (fs) and 150,000,000 (steps) under NPT (N: particle, P: pressure, T: constant temperature) conditions at T=300[K]. Equilibrium states (180 ns to 300 ns) of structures (2,400 structures) were picked up every 50 ps (25,000 steps), and the surface exposure area (ASA) for each amino acid residue was calculated along the time axis, and the average surface exposure area was calculated. Next, the surface exposure ratio (rASA) was calculated by dividing the surface exposure area calculated for each amino acid residue by the maximum solvent exposure area specific to that amino acid residue. Figure 8 shows the results of the analysis of the surface exposure ratio (rASA) of each amino acid residue in the hemagglutinin trimer. We found that by analyzing the ratio of exposed area (rASA), we can evaluate the relative degree of exposure between amino acid residues within the same protein. Therefore, we found that, in addition to information on fluctuations in each amino acid residue at the epitope site in the antigen, it is possible to predict and identify epitope sites that are resistant to mutation and to which antibodies can easily bind with high accuracy, based on information on the surface exposure ratio (rASA) of each amino acid residue, which is an indicator of epitope sites to which antibodies can easily bind against the antigen. Furthermore, we found that the predicted resistant epitope sites can be used as epitope sites for the production and selection of broad-spectrum neutralizing antibodies and for vaccines that are less affected by viral mutations.

[0041] <Prediction of difficult-to-mutate epitope sites and screening of antibodies> To evaluate whether it is possible to predict and identify low-mutation epitope sites for broad-spectrum neutralizing antibodies whose binding ability is less affected by viral mutations, and to select broad-spectrum neutralizing antibodies with high accuracy, based on information on fluctuations in each amino acid residue at the epitope site of the antigen, the inventors first compared and evaluated the epitope of a broad-spectrum neutralizing antibody against hemagglutinin (CR6261), which has been reported to have broad spectrum properties against human influenza A virus (DCEkiert et al., Science 2009, 10 APRIL Vol 324 246-251). Next, using dry analysis based on molecular dynamics calculations, we evaluated the epitopes of the broad-spectrum neutralizing antibody (CR6261) based on information regarding the mutation accumulation rate, fluctuation (residual RMSF), and surface exposure ratio (rASA) of each amino acid residue, which were obtained as neutralizing antibody selection indices.

[0042] <<Broad-spectrum neutralizing antibody against hemagglutinin (CR6261)>> Unlike other reported anti-hemagglutinin antibodies, the broad-spectrum neutralizing antibody (CR6261) uses the HA2(345-566) region as its epitope, specifically amino acid residues 385, 386, 389, 390, 392, 393, 396, 397, 399, 400, and 401 in Helix A (Figure 9A-C). Broad-spectrum neutralizing antibody (CR6261) has been reported to neutralize various types of influenza (H1, H2, H5, H6, H8, H9). Figure 9A is a schematic diagram showing the complex structure of hemagglutinin with multiple anti-hemagglutinin neutralizing antibodies (HC19, HC45, HC63, BH151), including the broad-spectrum neutralizing antibody (CR6261). In Figure 9A, "CR6261" refers to the broad-spectrum neutralizing antibody (CR6261). Figure 9B is a schematic diagram showing the binding sites between the broad-spectrum neutralizing antibody (CR6261) and the epitope. It shows the binding sites between the epitope Helix A (385-401) in hemagglutinin and the heavy chain (Vh) and light chain (Vl) of the antigen-recognizing variable region in CR6261. Figure 9C is a schematic diagram showing the structure of the Hemagglutinin epitope Helix A and its sequence conservation among various types of influenza viruses. The numbers in parentheses in the figure indicate the sequence homology (Seq.ID;%) at each amino acid residue. The Helix A sequence of HA is well-conserved across various influenza strains (seq.ID > ~90%).

[0043] <<Evaluation of the epitope of the broad-spectrum neutralizing antibody (CR6261) for hemagglutinin>> Figures 10-12 show the mutation accumulation rate, fluctuation (residue RMSF), and surface exposure ratio (rASA) of the broad-spectrum neutralizing antibody (CR6261) at epitope Helix A (385-401), respectively. In Figures 10-12, the epitope region (amino acid residues 385, 386, 389, 390, 392, 393, 396, 397, 399, 400, and 401) is shown as a black square, and other amino acid residues are shown as gray circles. The evaluation results of the mutation accumulation rate in Figure 10 revealed that many of the epitopes (7 / 11: amino acid residues at positions 385, 386, 389, 390, 392, 397, and 400) are well conserved. From the evaluation results of the fluctuations (residue RMSF) in Figure 11, it was confirmed that the fluctuations in the epitope region (amino acid residues 385, 386, 389, 390, 392, 393, 396, 397, 399, 400, and 401) were small, all of which were found to be 1.2 [Å] or less. In addition, the average fluctuation of the series of amino acid residues from 385 to 401 was 0.91 [Å] or less. From the surface exposure ratio (rASA) evaluation results in Figure 12, it was found that many of the epitopes (9 / 11: amino acid residues 385, 386, 389, 390, 393, 396, 397, 400, and 401) were exposed on the surface of the hemagglutinin protein (rASA ≠ 0.0), making them accessible to the antibody.

[0044] These results revealed that many of the amino acid residues that are epitopes of the broad-spectrum neutralizing antibody against hemagglutinin (CR6261) have the characteristics of low fluctuation, low mutation accumulation rate, and being exposed on a surface accessible by the antibody. As described above, CR6261 has been reported to neutralize various types (H1, 2, 5, 6, 8, 9) of influenza, and is a broad-spectrum neutralizing antibody whose binding ability to viruses prone to mutation is less affected by viral mutations. Based on the above findings, the inventors conceived that by analyzing the fluctuation information of each amino acid residue in the epitope site of the antigen, it is possible to predict with high accuracy sites that are less prone to mutation (difficult-to-mutate epitope sites) in the epitope site, and that difficult-to-mutate epitope sites can be used in excellent applications for the production and selection of broad-spectrum neutralizing antibodies, thus completing the present invention.

[0045] (Prediction device for difficult-to-mutate epitope sites) In other words, the present invention provides a device for predicting difficult-to-mutate epitope sites in an antigen, and includes a prediction means that predicts an amino acid residue whose fluctuation is less than or equal to a predetermined value as a difficult-to-mutate site in the epitope site, based on information on the fluctuation of each amino acid residue in the epitope site of the antigen, and further includes other means as necessary.

[0046] The following describes in more detail the prediction device for difficult-to-mutate epitope sites of the present invention.

[0047] <antigen> There are no particular restrictions on the antigens used for search, as long as they are peptides or proteins in which amino acids are linked by peptide bonds. Known or novel antigens can be appropriately selected depending on the purpose, for example, surface proteins of pathogens. Furthermore, the antigen may be the entire protein, or it may be a part of the protein such as a subunit, subdomain, or any peptide portion. Examples of the aforementioned pathogens include influenza viruses such as influenza A virus, and human coronaviruses such as SARS-CoV-2. Examples of surface proteins include hemagglutinin and neuraminidase in influenza viruses, and spike proteins in coronaviruses. Among these, hemagglutinin in influenza viruses is preferred.

[0048] <Epitope site> There are no particular restrictions on the epitope site to be predicted; it can be appropriately selected according to the purpose, and may be a single amino acid residue, multiple amino acid residues, or a single or multiple amino acid residues in a group of consecutive amino acid residues. The multiple amino acid residues may be discontinuous or consecutive. Among these, a group of consecutive amino acid residues is preferred. The number of amino acid residues in a group of consecutive amino acid residues is not particularly limited as long as it can be recognized by the antibody as an epitope site, and can be appropriately selected according to the purpose. However, 5 to 300 residues is preferred, 8 to 200 residues is more preferred, and 10 to 100 residues is even more preferred.

[0049] <Molecular dynamics calculation> In the present invention, it is preferable to calculate the fluctuations for each amino acid residue at the epitope site in the antigen, as described later, and the surface exposure area ratio based on molecular dynamics calculations for the antigen to be searched. In the present invention, for example, it is preferable to extract the trajectory (data on the trajectory of atomic motion) from 180 ns to 300 ns based on molecular dynamics calculations performed in water under absolute temperature of 300 K (Kelvin) with a simulation time of 300 ns, and to calculate the fluctuations for each amino acid residue in the equilibrium state antigen protein. Regarding the molecular dynamics calculation methods, the methods described in the above <Examples of Molecular Dynamics Calculations> can be used as appropriate for the purpose, but are not limited to these methods. The following section provides a detailed explanation of molecular dynamics calculations.

[0050] The atoms that make up peptides and proteins do not remain stationary in a solution, for example; they are constantly changing their positions. Molecular dynamics (MD) is used to reproduce such atomic movements in a computer.

[0051] In molecular dynamics simulations, the initial structure of the molecule to be studied is first created. In this invention, for example, when a known antigen protein is to be studied, the structure can be obtained from a protein structure database (e.g., Protein Data Bank) and used.

[0052] After creating the initial structure of the antigen protein, to perform stable molecular simulations, set a sufficiently large box (cell) size, place solvent molecules (e.g., water molecules) around it, and add Na to ensure the environment inside the cell is neutral, like a physiological solution. + Ions, Cl - By inserting ions and calculating the forces acting on each atom under periodic boundary conditions, the forces (energy) acting on each atom include bond stretching energy, bond angle bending energy, torsional energy, van der Waals interaction energy, electrostatic interaction energy, and hydrogen bond energy. The sum of the forces acting on all atoms constituting the molecule is called "potential energy."

[0053] Next, molecular dynamics calculations determine how atoms act under that force, based on Newton's equations of motion. This allows us to calculate the change in the atoms' positions after a short time step from their initial configuration. Next, in molecular dynamics, the same calculations are performed again, using the changed positions of the atoms as a new starting point. By repeating this at very short time intervals, the gradual movement of the atoms can be reproduced. In this way, molecular dynamics involves repeatedly performing steps (i) to (iii) on a computer: (i) determining the positions of the atoms, (ii) calculating the forces acting on the atoms, and (iii) calculating the motion of the atoms. Physical quantities and three-dimensional structures that change over time are arbitrarily extracted, and the structure and properties of biomolecules and compounds are analyzed by performing statistical processing based on the extracted data and displaying images of the three-dimensional structure.

[0054] To perform stable molecular simulations, structural relaxation of the solvent molecules is necessary. Therefore, while keeping the cell size fixed, molecular dynamics calculations (hereinafter sometimes referred to as NVT calculations) were performed with constant particle number, volume, and temperature, with positional constraints on the heavy atoms (atoms other than hydrogen) constituting the peptide, to relax the structure of the solvent molecules. Then, while appropriately adjusting the cell size to keep the pressure constant, molecular dynamics calculations (hereinafter sometimes referred to as NPT calculations) were performed with constant particle number and temperature. Subsequently, by performing long-duration (300 ns) NPT calculations, stable molecular simulations can be continued.

[0055] In this invention, the "simulation time" of molecular dynamics method refers to the time it takes to reproduce the structural changes of a molecule by repeatedly calculating the changes in the positions of atoms in short time increments based on Newton's equations of motion. Furthermore, the above-mentioned short step time is preferably 0.1 fs (femtoseconds) or more and 10 fs or less, and more preferably 0.5 fs or more and 2.0 fs or less. The short step time may be referred to as the "step time" or "time step width". In this invention, unless otherwise specified, the step time is 2.0 fs. Here, if the number of iterations in repeatedly calculating the change in atomic position during the step time is called the "loop count," then the simulation time is expressed as the product of the step time and the loop count. In this invention, in order to calculate the fluctuations for each amino acid residue in the equilibrium state of the antigen protein, it is preferable, for example, to set the simulation time to 100 ns or more, but it is not limited to this.

[0056] A molecular force field is a mathematical formula that expresses, as a function, the forces acting on each atom within a peptide or protein molecule. In molecular mechanics and molecular dynamics calculations based on molecular force fields, the forces acting between atoms are expressed numerically using a potential function, which is determined by the type of atom and the bonding pattern, with parameters representing the bonds between atoms (such as bond distance and bond angle) as variables. The molecular force fields that can be used in the present invention are not particularly limited and can be appropriately selected depending on the purpose. Examples include Amber-based molecular force fields, CHARMm-based molecular force fields, and OPLS-based molecular force fields. Examples of Amber-based molecular force fields include Amber ff99SB-ILDN and Amber 12SB. Examples of CHARMm-based molecular force fields include CHARMm36.

[0057] Furthermore, there are no particular restrictions on which energy terms to include in the calculation. Additionally, to improve computational efficiency, a method called the cutoff method may be introduced, which does not calculate electrostatic interactions and other factors if the interatomic distance is above a certain level.

[0058] Programs capable of performing molecular dynamics calculations include AMBER (http: / / ambermd.org / ), CHARMM (http: / / www.charmm.org / charmm / ), NAMD (http: / / www.ks.uiuc.edu / Research / namd / ), GROMACS (http: / / www.gromacs.org / ), and MyPresto (http: / / presto.protein.osaka-u.ac.jp / myPresto4 / ).

[0059] Molecular dynamics calculations are generally performed at a set temperature of approximately 280K (Kelvin) to 320K, and in this invention, it is preferable to set the temperature to, for example, 300K.

[0060] Furthermore, in molecular dynamics calculations, it is preferable to consider solvent effects, and it is preferable to perform calculations in a system that treats solvent molecules (e.g., water molecules) as individual molecules, similar to proteins. In this invention, a sufficient number of water molecules are arranged around the antigen protein. As a model of water molecules, for example, the TIP3P model can be used.

[0061] <Prediction of sites in epitope regions that are less prone to mutation> In the present invention, the prediction means is a means of predicting that an amino acid residue whose fluctuation is less than or equal to a predetermined value is a site in the epitope that is less likely to mutate, based on information on the fluctuation of each amino acid residue in the epitope site of the antigen. In an embodiment in which consecutive amino acid residue groups in an epitope site are further search targets, it is preferable that the means predict that the amino acid residue groups whose mean fluctuations are less than or equal to a predetermined value are sites in the epitope site that are less prone to mutation, based on information on the mean fluctuations of the consecutive amino acid residue groups in the epitope site.

[0062] <fluctuation> Here, the fluctuation of amino acid residues, or the magnitude of the fluctuation (RMSF, root-mean-square-fluctuation), corresponds to the time average of the magnitude of the deviation of each amino acid residue position from its average position.

[0063] The fluctuation information used as a criterion for predicting sites that are less likely to mutate within the epitope region is preferably calculated using molecular dynamics simulations. Furthermore, since the analysis is performed using data of the antigen structure in a molecularly mechanically equilibrium state, it is more preferable to calculate it based on the trajectory of the extraction time (for example, 180 ns to 300 ns as the elapsed time from the start of the simulation) during a certain period of time after reaching the equilibrium state. The time required to reach equilibrium is, as the elapsed time from the start of the simulation, preferably 100 ns or more, more preferably 150 ns, and even more preferably 180 ns. The extraction time is preferably 50 ns or more, more preferably 100 ns, and even more preferably 120 ns, as the time after reaching equilibrium.

[0064] One method for calculating fluctuations is to use an MD trajectory analysis library such as "MDTraj" to calculate the magnitude of the deviation from the average position at the time of extraction for each amino acid residue, and this can be expressed as the fluctuation for each amino acid residue (residual RMSF). rmsf, ​​which shows fluctuations at the x-th amino acid residue in the antigen. x This can be calculated using the following formula (2), and its unit is [Å]. rmsf x =√(1 / NΣ(rxj- <rxj>)2)...Equation (2) In equation (2), N represents the number of structures sampled during the extraction time, j represents an integer from 1 to n, and rxj represents the position vector of the x-th amino acid residue at extraction time j. <rxj>This shows the time-averaged position vector of the x-th residue.

[0065] The predetermined value of fluctuation used as a criterion for predicting sites that are less likely to mutate in epitope regions can be appropriately selected according to the characteristics of the antigen being searched for, but 1.3 Å is preferred, 1.2 Å is more preferred, 1.1 Å is even more preferred, and 1.0 Å is particularly preferred. For example, in the case of hemagglutinin of the influenza virus, 1.2 Å is preferred, 1.1 Å is more preferred, and 1.0 Å is even more preferred. The amino acid residues whose fluctuations are below a predetermined value can be predicted to be sites in the epitope region that are less prone to mutation. Furthermore, in embodiments where a sequence of amino acid residues in the epitope site is targeted for further investigation, the predetermined value of the mean fluctuation can be appropriately selected according to the characteristics of the antigen being investigated, but 1.2 Å is preferred, 1.1 Å is more preferred, and 1.0 Å is even more preferred. For example, in the hemagglutinin of the influenza virus, 1.0 Å is preferred, 0.9 Å is more preferred, and 0.8 Å is even more preferred. The group of amino acid residues whose mean fluctuation is below a predetermined value can be predicted to be a site in the epitope region that is less prone to mutation.

[0066] <Prediction of sites where antibodies are likely to bind to antigens> In one embodiment of the present invention, the prediction means may be a means for predicting that among the amino acid residues predicted to be sites that are difficult to mutate, amino acid residues whose surface exposure area ratio is equal to or greater than a predetermined value are sites to which antibodies against the antigen can easily bind, based on information on the surface exposure area ratio of each amino acid residue. This allows for the ranking of promising epitope sites based on fluctuation information as well as information on the surface area ratio of amino acid residues, taking into account robustness to mutations and three-dimensional structure. This enables highly accurate prediction of amino acid residues that are less prone to mutation within the epitope site and where antibodies readily bind to antigens. Furthermore, the prediction of sites where antibodies readily bind to antigens may be performed in addition to the prediction of sites in the epitope region that are less prone to mutation, or it may be performed independently.

[0067] <Surface exposed area ratio> Here, the relative accessible surface area (rASA) of the x-th amino acid residue in the antigen is a value (%) calculated by dividing the accessible surface area (ASA) of the x-th amino acid residue by the maximum solvent-exposed area of ​​the amino acid residue, and any value of 100% or more is set to 100%.

[0068] The surface exposure area ratio information, which serves as a criterion for predicting sites where antibodies are likely to bind to antigens, is preferably a value calculated by molecular dynamics simulations. Specifically, as described in the above-mentioned <<Method for Calculating Exposure Area Ratio>>, the solvent accessible surface area (SASA) calculation method commonly used in MD analysis (for example, the Gromacs command, gmx sasa) can be used to calculate the exposure area (ASA) for each amino acid residue as the accessible area for binding target molecules such as antibodies. This can be achieved by changing the radius of the sphere corresponding to the radius of the solvent molecule (1.4 Å) to a sphere with a radius corresponding to the size of the target molecule that binds to the antigen's epitope site, such as the CDR exposure portion of the antibody. The exposed area (ASA) of the aforementioned amino acid residue can be calculated as described above. The maximum solvent exposure area of ​​the aforementioned amino acid residue is based on prior literature (MZTien et al. PLoS One. 2013;8(11):e80635.), and represents the solvent exposure area of ​​amino acid residue X when a tripeptide (GLY-X-GLY) with amino acid residue X at its center adopts an Extended structure. This is a unique value depending on the type of amino acid residue.

[0069] Examples of the target molecules for binding include antibodies; VHH antibodies; low-molecular-weight antibodies such as scFV, diabody, Fab region, and F(ab)2 region; and antibody-like molecules such as peptides having a helix-loop-helix (HLH) structure. There are no particular restrictions on the sphere having a radius corresponding to the size of the target molecule to bind to; it can be appropriately selected according to the size and shape of the target molecule to bind to. For example, a sphere with a radius of 7.2 Å, corresponding to the size of the CDR exposed portion of an antibody, can be used.

[0070] The predetermined value of the surface exposure area ratio, which serves as a criterion for predicting sites where antibodies against an antigen are likely to bind, can be appropriately selected according to the characteristics of the antigen being investigated, but 0% is preferred, 1%, 5%, and 10% are more preferred, and 20% is even more preferred. For example, in the case of hemagglutinin of the influenza virus, 0% is preferred, 1%, 5%, and 10% are more preferred, and 20% is even more preferred. The amino acid residues whose surface exposure area ratio is greater than or equal to a predetermined value can be predicted to be sites on which antibodies against the antigen are likely to bind. Furthermore, in embodiments in which a series of amino acid residues in the epitope site is targeted for further investigation, the predetermined value of the average surface exposure area ratio can be appropriately selected according to the characteristics of the antigen to be investigated, but 0% is preferred, 1%, 5%, and 10% are more preferred, and 15% is even more preferred. For example, in the case of hemagglutinin of the influenza virus, 0% is preferred, 1%, 5%, and 10% are more preferred, and 15% is even more preferred. The amino acid residue group whose average surface exposure area ratio is greater than or equal to a predetermined value can be predicted to be a site on which antibodies against the antigen are likely to bind.

[0071] <Other means> Other means are not particularly restricted and can be selected as appropriate depending on the purpose.

[0072] The device for predicting difficult-to-mutate epitope sites of the present invention can be, for example, a computer having a recording medium that stores a difficult-to-mutate epitope site prediction program, as described later. The device for predicting difficult-to-mutate epitope sites of the present invention may, for example, read the difficult-to-mutate epitope site prediction program from the recording medium that stores the program, and execute the difficult-to-mutate epitope site prediction program using a control device such as a CPU (Central Processing Unit). Furthermore, to speed up the calculation, part of the calculation may be performed by a GPU (Graphics Processing Unit). Specific examples of the configuration of the prediction device for difficult-to-mutate epitope sites will be described later.

[0073] (Method for predicting difficult-to-mutate epitope sites) The present invention provides a method for predicting mutable epitope sites in an antigen, comprising a prediction step of predicting that amino acid residues whose fluctuations are less than or equal to a predetermined value are mutable sites in the epitope site, based on information on fluctuations for each amino acid residue in the epitope site of the antigen, and further comprising other steps as necessary. The prediction step and other steps can be suitably carried out by the prediction means and other means in the prediction device for difficult-to-mutate epitope sites of the present invention described above.

[0074] (Prediction program for difficult-to-mutate epitope sites) The present invention provides a program for predicting mutable epitope sites in an antigen, which involves a computer performing a process to predict that amino acid residues whose fluctuations are below a predetermined value are mutable sites in the epitope site, based on information on fluctuations for each amino acid residue in the epitope site of the antigen.

[0075] The program for predicting difficult-to-mutate epitope sites of the present invention can, for example, be a program for implementing the above-described method for predicting difficult-to-mutate epitope sites using a computer. In other words, by executing the program for predicting difficult-to-mutate epitope sites of the present invention using a computer, the computer can be made to function as the above-described device for predicting difficult-to-mutate epitope sites. For this reason, a preferred form of the program for predicting difficult-to-mutate epitope sites of the present invention can be the same as the above-described device and method for predicting difficult-to-mutate epitope sites.

[0076] Furthermore, there are no particular restrictions on the computer used to execute the program for predicting difficult-to-mutate epitope sites according to the present invention, as long as it is a computer capable of executing the program, and it can be appropriately selected according to the purpose. Such a computer may be, for example, a regular personal computer, or a large, high-performance computer such as a server computer, a computer cluster connecting multiple computers, or a supercomputer.

[0077] The prediction program for difficult-to-mutate epitope sites of the present invention can be created using various programming languages, depending on the configuration of the computer used and the type and version of the operating system.

[0078] The program for predicting difficult-to-mutate epitope sites of the present invention may be recorded on a storage medium such as a hard disk, or on a storage medium such as a CD-ROM (Compact Disc-Read Only Memory), DVD-ROM (Digital Versatile Disc-ROM), MO (Magneto-Optical) disk, or USB (Universal Serial Bus) memory. Furthermore, the prediction program for difficult-to-mutate epitope sites of the present invention may be stored in an external storage area (such as another computer) accessible from a computer via an information and communication network. In this case, the prediction program for difficult-to-mutate epitope sites of the present invention stored in the external storage area can be installed and used on a hard disk via the information and communication network from the external storage area as needed. Furthermore, the prediction program for difficult-to-mutate epitope sites of the present invention may be recorded on multiple storage media, divided into sections for each arbitrary process.

[0079] (Antibody screening device) The antibody screening device of the present invention is an antibody screening device that can bind to an antigen even if the antigen mutates, and includes a selection means for selecting antibodies whose binding affinity to the amino acid residue predicted to be a site that is difficult to mutate in the epitope site is above a predetermined value, according to the prediction device, method, or program of the present invention for difficult-to-mutate epitope sites, and further includes other means as necessary.

[0080] <Selection of antibodies that can bind to antigens even if the antigen mutates> The selection means is a means for selecting antibodies whose binding affinity to the amino acid residue predicted to be a site less prone to mutation in the epitope region is above a predetermined value. Prediction of sites that are difficult to mutate within an epitope can be performed using the antimutation-resistant epitope site prediction device, method, or program of the present invention, and the various embodiments described above can be selected as appropriate. This allows for the highly accurate selection and identification of broad-spectrum neutralizing antibodies that have a broad spectrum of binding ability that is less affected by viral mutations.

[0081] <Selection of antibodies that readily bind to antigens> In one embodiment of the present invention, the selection means may be a means for selecting antibodies in which the binding affinity to an amino acid residue predicted to be a site on which an antibody readily binds to an antigen is greater than or equal to a predetermined value, among the amino acid residues predicted to be sites on which mutation is unlikely. This allows for the selection of antibodies using epitope sites as indicators, which are ranked based on a combination of factors including robustness to mutations and structural integrity, as well as information on the surface exposure area ratio of amino acid residues, in addition to fluctuation information. This enables the selection and identification of broad-spectrum neutralizing antibodies with a broad spectrum that is less affected by viral mutations and readily binds to antigens, with high precision. Furthermore, a means for selecting antibodies whose binding affinity to the aforementioned amino acid residues, which are predicted to be sites on which antibodies readily bind to the antigen, is above a predetermined value may be performed independently.

[0082] In other embodiments of the present invention, the invention is not limited to antibodies, and other binding target molecules other than antibodies (such as low-molecular-weight antibodies or antibody-like molecules) may be used instead of antibodies.

[0083] <Binding affinity> There are no particular restrictions on the methods for measuring and evaluating the binding affinity to amino acid residues, and known methods can be appropriately selected depending on the purpose. Examples include Western blotting, immunohistochemistry and immunocytochemistry, enzyme-linked immunosorbent assay (ELISA), immunoprecipitation, flow cytometry, surface plasmon resonance (SPR), and biolayer interferometry (BLI). For example, the interaction of an antibody with an epitope site containing the aforementioned amino acid residue can be measured using the measurement method described above, and the binding affinity can be evaluated using indicators such as the keying factor (KD). As a control epitope site, an epitope site containing other amino acid residues may be used instead of the aforementioned amino acid residue. Furthermore, considering the application environment of the antibody after screening, it is preferable to perform measurements and evaluations under the same conditions (temperature, pH, ionic strength) as the application environment.

[0084] There are no particular restrictions on the predetermined value of the binding affinity with amino acid residues, which serves as a criterion for selecting an antibody capable of binding to an antigen even when the antigen has mutated, or for judging the ease of binding to an antigen, and it can be appropriately selected according to the purpose. However, in terms of the dissociation constant (KD), 100 nM is preferred, and 10 nM is more preferred. For example, in the hemagglutinin of influenza virus, 100 nM is preferred, and 50 nM is more preferred. The dissociation constant (KD) can be preferably measured by surface plasmon resonance (SPR). Specifically, the dissociation constant (KD) can be determined by measuring with a surface plasmon resonance analyzer (Biacore TM 3000, manufactured by GE Healthcare Japan Corporation).

[0085] <Other means> There are no particular restrictions on other means, and they can be appropriately selected according to the purpose.

[0086] (Method for screening antibodies) The method for screening an antibody of the present invention is a method for screening an antibody capable of binding to an antigen even when the antigen has mutated, and includes a selection step of selecting an antibody having a binding affinity with the amino acid residue predicted to be a difficult-to-mutate site in the epitope site by the prediction device, method, or program for the difficult-to-mutate epitope site of the present invention, which is a predetermined value or more, and further includes other steps as necessary. The selection step and other steps can be preferably implemented by the selection means and other means in the antibody screening device of the present invention described above.

[0087] (Program for antibody screening) The antibody screening program of the present invention is a screening program for antibodies that can bind to an antigen even if the antigen mutates, and causes a computer to perform a process of selecting antibodies in which the binding affinity to the amino acid residue predicted to be a site that is difficult to mutate in the epitope site is above a predetermined value, using the antimutability epitope site prediction device, method, or program of the present invention.

[0088] The antibody screening device of the present invention can be, for example, a computer having a recording medium that stores the antibody screening program of the present invention. Furthermore, the antibody screening program of the present invention can be, for example, a program for implementing the above-described antibody screening method using a computer. In other words, by executing the antibody screening program of the present invention on a computer, the computer can be made to function as the above-described antibody screening device. For this reason, a preferred form of the antibody screening program of the present invention can be the same as that of the above-described antibody screening device and antibody screening method. As embodiments of the antibody screening program of the present invention, the same matters as those described in the antimutation-resistant epitope prediction program of the present invention can be appropriately selected.

[0089] (Embodiment) More specific embodiments of the present invention will be described below with reference to the drawings.

[0090] Figure 13 shows an example of the hardware configuration of the antimutation epitope site prediction device of the present invention. In the antimutation epitope site prediction device 100, for example, a control unit 101, main memory 102, auxiliary memory 103, I / O interface 104, communication interface 105, input device 106, output device 107, and display device 108 are connected via a system bus 109.

[0091] The control unit 101 performs calculations (arithmetic operations, comparison operations, annealing operations, etc.), hardware and software operation control, etc. The control unit 101 may be, for example, a CPU, a GPU, or a combination of these. The control unit 101 implements various functions by executing a program (for example, a program for predicting difficult-to-mutate epitope sites) loaded into the main memory 102 or the like. The processing performed by the prediction unit in the antimutation epitope site prediction device of the present invention can be carried out, for example, by the control unit 101.

[0092] The main memory 102 stores various programs as well as data necessary for executing those programs. For example, the main memory 102 can be one that includes at least one of ROM (Read Only Memory) and RAM (Random Access Memory). ROM stores various programs, such as the BIOS (Basic Input / Output System). There are no particular restrictions on the type of ROM; it can be selected appropriately depending on the purpose, and examples include mask ROM and PROM (Programmable ROM). RAM functions as a working range that is expanded when various programs stored in ROM or auxiliary storage device 103 are executed by the control unit 101. There are no particular restrictions on the type of RAM, and it can be appropriately selected according to the purpose, such as DRAM (Dynamic Random Access Memory) and SRAM (Static Random Access Memory).

[0093] The auxiliary storage device 103 is not particularly limited as long as it can store various types of information, and can be appropriately selected according to the purpose. Examples include solid-state drives (SSDs) and hard disk drives (HDDs). Alternatively, the auxiliary storage device 103 may be a portable storage device such as a CD drive, DVD drive, or BD (Blu-ray® Disc) drive. Furthermore, the antimutation-resistant epitope site prediction program of the present invention is stored, for example, in the auxiliary storage device 103, loaded into the RAM (main memory) of the main storage device 102, and executed by the control unit 101.

[0094] The I / O interface 104 is an interface for connecting various external devices. The I / O interface 104 enables data input and output from devices such as CD-ROMs (Compact Disc ROMs), DVD-ROMs (Digital Versatile Disk ROMs), MO disks (Magneto-Optical disks), and USB memory (USB (Universal Serial Bus) flash drives).

[0095] There are no particular restrictions on the communication interface 105, and any known interface can be used as appropriate, such as wireless or wired communication devices. The input device 106 is not particularly limited as long as it can receive various requests and information for the difficult-to-mutate epitope site prediction device 100, and any known device can be used as appropriate, such as a keyboard, mouse, touch panel, or microphone. Furthermore, if the input device 106 is a touch panel (touch display), the input device 106 can also function as the display device 108.

[0096] There are no particular restrictions on the output device 107, and any known device can be used as appropriate, such as a printer. There are no particular restrictions on the display device 108, and any known device can be used as appropriate, such as a liquid crystal display or an organic EL display.

[0097] Figure 14 shows another hardware configuration example of the antimutation epitope site prediction device of the present invention. In the example shown in Figure 14, the antimutation-resistant epitope site prediction device 100 is divided into a terminal device 200 that performs processing to predict that the epitope site is a site that is difficult to mutate, based on information on fluctuations for each amino acid residue at the epitope site in the antigen, and a server computer 300 that performs molecular dynamics calculations (MD simulations) to obtain the information on fluctuations for each amino acid residue. Furthermore, in the example shown in Figure 14, the terminal device 200 and the server computer 300 of the antimutation-resistant epitope site prediction device 100 are connected by a network 400. In the example shown in Figure 14, for example, a regular personal computer can be used as the terminal device 200, and a large, high-performance computer such as a computer cluster with multiple computers connected or a supercomputer can be used as the server computer 300. The server computer 300 may also be a group of computers on the cloud. Furthermore, the network 400 connecting the terminal device 200 and the server computer 300 can use a communication standard such as SSH (Secure Shell).

[0098] In the example shown in Figure 14, for example, the terminal device 200 creates the initial structure and sets the conditions for molecular dynamics calculations to generate information on fluctuations for each amino acid residue at the epitope site of the antigen. Then, the condition file necessary for the molecular dynamics calculation created by the terminal device 200 is transmitted from the terminal device 200 to the server computer 300 via the network 400. Next, the server computer 300 performs molecular dynamics calculations to obtain information on fluctuations for each amino acid residue at the epitope site of the antigen. Then, the data from the molecular dynamics calculations (such as trajectory files) is transmitted from the server computer 300 to the terminal device 200 via the network 400. Next, the terminal device 200 calculates fluctuations for each amino acid residue based on the received molecular dynamics calculation data, and predicts that amino acid residues with fluctuations below a predetermined value are less likely to mutate in the epitope region.

[0099] Figure 15 shows an example of the functional configuration of the prediction device for difficult-to-mutate epitope sites according to the present invention. As shown in Figure 15, the antimutation epitope site prediction device 100 comprises a communication function unit 120, an input function unit 130, an output function unit 140, a display function unit 150, a storage function unit 160, and a control function unit 170.

[0100] The communication function unit 120, for example, transmits and receives various types of data with an external device. The input function unit 130 receives various instructions for, for example, the prediction device 100 for difficult-to-mutate epitope sites. The output function unit 140 prints and outputs information such as fluctuations for each amino acid residue, the surface exposure area ratio for each amino acid residue, and the three-dimensional structure of the antigen to be searched. The display function unit 150 displays information such as fluctuations for each amino acid residue, the surface exposure area ratio for each amino acid residue, and the three-dimensional structure of the antigen to be searched on the display. The memory function unit 160 stores, for example, various programs and trajectories obtained from molecular dynamics calculations.

[0101] The control function unit 170 includes a prediction unit 171 and a molecular dynamics calculation unit 172. The control function unit 170 executes various programs stored in the memory function unit 160, for example, and controls the operation of the entire difficult-to-mutate epitope prediction device 100. The prediction unit 171 performs a process that predicts, for example, based on fluctuation information for each amino acid residue, that amino acid residues whose fluctuations are below a predetermined value are sites in the epitope region that are less likely to mutate. The molecular dynamics calculation unit 172 performs molecular dynamics calculations (MD simulations) to obtain information on fluctuations for each amino acid residue in the epitope site of an antigen, for example.

[0102] Here, with reference to Figure 16, an example of the procedure for performing the method for predicting difficult-to-mutate epitope sites using the present invention will be described. The following explanation will cover each step indicated by "S" in Figure 16.

[0103] First, in S101, the target molecule, the antigen, is modeled, and then water molecules and ions are placed around the modeled target molecule to construct a computational system. In other words, in S101, after modeling the target molecule using a modeling tool, an electrically neutralized computational system is constructed by placing water molecules and ions around the target molecule. Next, in S102, energy minimization calculations and structural relaxation calculations (NVT and NPT calculations) are performed on the constructed computational system to create the initial structure for long-term NPT calculations. In other words, in S102, energy minimization calculations using molecular mechanics are performed on the computational system constructed in S101 to remove unnatural structural distortions, and then short-term NPT structural relaxation calculations (NVT and NPT calculations) are performed to equilibrate the solvent and create the initial structure for long-term NPT calculations.

[0104] Next, in S103, a long-running NPT calculation is performed based on the created initial structure. In other words, in S103, using the initial structure created in S102, the trajectories such as energy and coordinates for various three-dimensional structures that the target molecule can take are sampled at each step time during the extraction time, which is the period of time elapsed after reaching equilibrium, as explained in the <Example of Molecular Dynamics Calculation> above.

[0105] Then, in S104, residue RMSF is calculated based on the trajectory data from the long-term NPT calculation. In other words, in S104, the magnitude of fluctuation (residual RMSF) of each amino acid residue relative to the average structure of the target molecule (antigen) in an equilibrium state is calculated. Specifically, the fluctuation RMSF of the x-th amino acid residue in the antigen is calculated. x [Å] is calculated using the following formula (2). rmsf x =√(1 / NΣ(rxj- <rxj>)2)...Equation (2) In equation (2), N represents the number of structures sampled during the extraction time, j represents an integer from 1 to n, and rxj represents the position vector of the x-th amino acid residue at extraction time j. <rxj>This shows the time-averaged position vector of the x-th residue.

[0106] Next, in S105, based on the determined residue RMSF, the system predicts sites that are less prone to mutation. In other words, amino acid residues whose fluctuations are below a predetermined value are predicted to be sites in the epitope region that are less prone to mutation (difficult-to-mutate epitope sites). Once a difficult-to-mutate epitope site is predicted, the process may be terminated, or optionally, steps S106 and S107 described later may be performed in addition before terminating the process.

[0107] Optionally, S106 calculates the surface exposure area ratio (rASA) based on trajectory data from long-term NPT calculations. In other words, S106 calculates the surface exposure area ratio (rASA) for each amino acid residue as an indicator of the exposed area accessible to target molecules such as antibodies. Next, in S107, based on the calculated surface exposure area ratio rASA, the sites to which the antibody is likely to bind are predicted. In other words, among the amino acid residues predicted in S105 to be sites that are unlikely to mutate, amino acid residues whose surface exposure area ratio is greater than or equal to a predetermined value are predicted to be sites to which the antibody against the antigen is likely to bind, based on the information on the surface exposure area ratio of each amino acid residue. The process terminates once the site where the antibody is most likely to bind is predicted.

[0108] In Figure 16, the processing flow in an example of the present invention is explained in a specific order. However, in the present invention, the order of each step can be changed as appropriate, to the extent that it is technically possible. Furthermore, in the present invention, multiple steps may be performed together, to the extent that it is technically possible.

[0109] As described above, the present invention is a device for predicting difficult-to-mutate epitope sites in an antigen, and has a prediction means that predicts an amino acid residue whose fluctuation is less than or equal to a predetermined value as a difficult-to-mutate site in the epitope site, based on information on the fluctuation of each amino acid residue in the epitope site of the antigen. This allows for highly accurate prediction and identification of difficult-to-mutate epitope sites for broad-spectrum neutralizing antibodies through computer-based molecular simulations.

[0110] Furthermore, the antibody screening device of the present invention is a screening device for antibodies that can bind to an antigen even if the antigen mutates, and includes a selection means for selecting antibodies whose binding affinity to the amino acid residue predicted to be a site that is difficult to mutate in the epitope site is above a predetermined value, according to the antimutability epitope site prediction device, method, or program of the present invention. This makes it possible to provide an antibody screening device that can select broad-spectrum neutralizing antibodies with high precision.

[0111] Examples of the present invention are as follows: <1> A device for predicting reluctant epitope sites in antigens, This is a predictor of difficult-to-mutate epitope sites, characterized by having a predictive means that predicts that amino acid residues whose fluctuations are below a predetermined value are difficult-to-mutate sites in the epitope site, based on information on fluctuations for each amino acid residue in the epitope site of the antigen. <2> The prediction means predicts that a group of amino acid residues whose average fluctuation is less than or equal to a predetermined value is a site in the epitope that is less prone to mutation, based on information on the average fluctuation of a group of consecutive amino acid residues in the epitope site. <1> This is a predictor for difficult-to-mutate epitope sites as described above. <3> The aforementioned antigen is hemagglutinin in the influenza virus. The predetermined value of the fluctuation is 1.2 Å. <1> from <2> This is a predictor for difficult-to-mutate epitope sites described in any of the following. <4> The fluctuation information mentioned above is a calculated value obtained from molecular dynamics calculations. <1> from <3> This is a predictor for difficult-to-mutate epitope sites described in any of the following. <5> The prediction means predicts that among the amino acid residues predicted to be sites that are difficult to mutate, those amino acid residues whose surface exposure area ratio is equal to or greater than a predetermined value are sites to which antibodies against the antigen can easily bind, based on information on the surface exposure area ratio of each amino acid residue. <1> from <4> This is a predictor for difficult-to-mutate epitope sites described in any of the following. <6> The aforementioned antigen is hemagglutinin in the influenza virus. The predetermined value of the surface exposure area ratio is 0%, <5> This is a predictor for difficult-to-mutate epitope sites as described above. <7> The information regarding the surface exposure area ratio is a calculated value obtained from molecular dynamics calculations. <5> from <6> This is a predictor for difficult-to-mutate epitope sites described in any of the following. <8> A method for predicting reluctant epitope sites in antigens, This method for predicting mutable epitope sites is characterized by including a prediction step in which, based on information on fluctuations for each amino acid residue in the epitope site of an antigen, amino acid residues whose fluctuations are below a predetermined value are predicted to be mutable sites in the epitope site. <9> A program that predicts the site of a difficult-to-mutate epitope in an antigen, This is a program for predicting mutable epitope sites, characterized by having a computer perform a process that predicts that amino acid residues whose fluctuations are below a predetermined value are mutable sites in the epitope site, based on information on fluctuations for each amino acid residue in the epitope site of the antigen. <10> A screening device for antibodies that can bind to the antigen even if the antigen mutates, The aforementioned <1> from <7> A predictor of difficult-to-mutate epitope sites as described in any of the above, <8> The method for predicting difficult-to-mutate epitope sites described above, or the above <9> The antibody screening device is characterized by having a selection means for selecting antibodies whose binding affinity to the amino acid residue predicted to be a site that is difficult to mutate at the epitope site, based on the prediction program for difficult-to-mutate epitope sites described in [reference], is above a predetermined value. <11> A screening method for antibodies that can bind to an antigen even if the antigen mutates, The aforementioned <1> from <7> A predictor of difficult-to-mutate epitope sites as described in any of the above, <8> The method for predicting difficult-to-mutate epitope sites described above, or the above <9> The antibody screening method is characterized by including a selection step of selecting antibodies whose binding affinity to the amino acid residue predicted to be a site that is difficult to mutate at the epitope site, as determined by the prediction program for difficult-to-mutate epitope sites described in [reference], is equal to or greater than a predetermined value. <12> A screening program for antibodies that can bind to an antigen even if the antigen mutates, The aforementioned <1> from <7> A predictor of difficult-to-mutate epitope sites as described in any of the above, <8> The method for predicting difficult-to-mutate epitope sites described above, or the above <9> This is an antibody screening program characterized by having a computer perform a process to select antibodies whose binding affinity to the amino acid residue predicted to be a site that is difficult to mutate at the epitope site, based on the prediction program for difficult-to-mutate epitope sites described above, is above a predetermined value.

[0112] The aforementioned <1> from <7> A predictor of difficult-to-mutate epitope sites as described in any of the above, <8> The method for predicting difficult-to-mutate epitope sites described above, or the above <9> The prediction program for difficult-to-mutate epitope sites described herein can solve the problems of the past and achieve the objectives of the present invention. The aforementioned <10> The antibody screening apparatus described above, <11> The antibody screening method described above, or the above <12> The antibody screening program described herein provides an antibody screening device, method, and program that can solve conventional problems and select broad-spectrum neutralizing antibodies with high accuracy. [Explanation of symbols]

[0113] 100 Prediction device for difficult-to-mutate epitope sites 200 terminal devices 300 server computers 171 Prediction Section

[0114] < / rxj> < / rxj> < / rxj> < / rxj> < / rxj> < / rxj>

Claims

1. A device for predicting reluctant epitope sites in antigens, Based on information on the fluctuations of each amino acid residue in the epitope site of the antigen, amino acid residues whose fluctuations are below a predetermined value are predicted to be sites in the epitope site that are less prone to mutation. A device for predicting difficult-to-mutate epitope sites, characterized by having a prediction means that predicts, based on information on the surface exposure area ratio of each amino acid residue among the amino acid residues predicted to be sites that are difficult to mutate, that have a surface exposure area ratio of a predetermined value or more, as sites to which antibodies against the antigen can easily bind.

2. The predictive device for a difficult-to-mutate epitope site according to claim 1, wherein the predictive means predicts that a group of amino acid residues whose mean fluctuation is less than or equal to a predetermined value is a difficult-to-mutate site in the epitope site, based on information on the mean fluctuation of a group of consecutive amino acid residues in the epitope site.

3. The aforementioned antigen is hemagglutinin in the influenza virus. The predictor of a difficult-to-mutate epitope site according to any one of claims 1 to 2, wherein the predetermined value of the fluctuation is 1.2 Å.

4. The predictor of a difficult-to-mutate epitope site according to any one of claims 1 to 3, wherein the fluctuation information is a calculated value obtained by molecular dynamics calculations.

5. The aforementioned antigen is hemagglutinin in the influenza virus. The device for predicting difficult-to-mutate epitope sites according to claim 1, wherein the predetermined value of the surface exposure area ratio is 0%.

6. The predictor of a difficult-to-mutate epitope site according to any one of claims 1 to 5, wherein the surface exposure area ratio information is a calculated value obtained by molecular dynamics calculations.

7. A method for predicting reluctant epitope sites in antigens, Based on information on the fluctuations of each amino acid residue in the epitope site of the antigen, amino acid residues whose fluctuations are below a predetermined value are predicted to be sites in the epitope site that are less prone to mutation. A method for predicting mutable epitope sites, characterized by including a prediction step of predicting that among the amino acid residues predicted to be mutable sites, those amino acid residues whose surface exposure area ratio is equal to or greater than a predetermined value are sites to which antibodies against the antigen can easily bind, based on information on the surface exposure area ratio of each amino acid residue.

8. A program that predicts the site of a difficult-to-mutate epitope in an antigen, Based on information on the fluctuations of each amino acid residue in the epitope site of the antigen, amino acid residues whose fluctuations are below a predetermined value are predicted to be sites in the epitope site that are less prone to mutation. A program for predicting difficult-to-mutate epitope sites, characterized in that, among the amino acid residues predicted to be sites that are difficult to mutate, the computer performs a process to predict that amino acid residues whose surface exposure area ratio is greater than or equal to a predetermined value are sites to which antibodies against the antigen can easily bind, based on information on the surface exposure area ratio of each amino acid residue.

Images