Antibody dynamic structural property analysis device, antibody dynamic structural property analysis method, and program

Abstract

Provided is an antibody dynamic structural property analysis device, for example, that is capable of performing analysis without averaging the motion of a residue. In an antibody dynamic structural property analysis device 1, a molecular motion calculation processing unit 29 obtains a plurality of snapshots by molecular dynamics calculation, for example, with respect to an antibody to be analyzed. An R-plot processing unit 31 creates an nth-residue Ramachandran plot with respect to an nth residue of the antibody to be analyzed, and obtains the dihedral angles φ and ψ of the nth residue in each snapshot. A set variable arithmetic processing unit 35 calculates the distance from dihedral angle reference values φ0 n and ψ0 n of the nth residue with respect to the dihedral angles φ and ψ of the nth residue in each snapshot to obtain a set variable value. An operation analysis processing unit 39 determines that the nth residue is a deviated residue when the ratio of those of the set variables of the nth residue of all the snapshots that are in a region where the value of the set variable exceeds an operation analysis reference value exceeds a ratio reference value.

Description

Antibody dynamic structural property analysis device, antibody dynamic structural property analysis method and program

[0001] The present invention relates to an antibody dynamic structural property analysis device, an antibody dynamic structural property analysis method, and a program, which analyze the dynamic structural property of an antibody using data for identifying an antibody to be analyzed, which is stored in an analysis target memory unit.

[0002] 13 is a diagram for explaining antibodies. Antibodies are biopolymers of the immune system that specifically bind to antigens of foreign substances (such as pathogens) and work to eliminate the foreign substances.

[0003] In Figure 13(a), antibody 101 represents IgG (150 kDa), a type of immunoglobulin. Antibodies can bind to specific antigens. Antibody diversity is due to the combination of heavy chain variable regions 103 and 113 and light chain variable regions 107 and 117. The variable region of an antigen receptor is the combination of heavy chain variable regions 103 and 113 and light chain variable regions 107 and 117.

[0004] The complementarity determining regions (CDRs) are the regions in the variable regions 103, 107, 113, and 117 that actually contact the antigen. These regions are important for determining antigen specificity and exhibit high sequence diversity. In Figure 13(a), reference numerals 121, 123, and 125 denote CDR1, CDR2, and CDR3, respectively, in the variable region 113.

[0005] Framework regions (FRs) are regions other than CDR1, CDR2, and CDR3 in variable regions 103, 107, 113, and 117. FRs have high sequence homology among antibodies, provide the framework structure and stability of the antibody, and support its function. In Figure 13(a), reference numerals 127, 129, 131, and 133 represent FR1, FR2, FR3, and FR4, respectively, in variable region 113.

[0006] 13(b) shows the relationship between CDR1, CDR2, and CDR3 and FR1, FR2, FR3, and FR4. CDR1, CDR2, and CDR3 are located between FR1, FR2, FR3, and FR4, respectively.

[0007] Figure 14 is a diagram illustrating the denaturation process of a protein. Figure 14(a) shows a protein in its native state. Figure 14(b) shows a state in which a locally denatured region (weak spot) 141 exists (locally denatured state). Then, as shown in Figure 14(c), the protein as a whole becomes denatured. As a result, it forms aggregates as shown in Figure 14(d).

[0008] Stabilizing the locally denatured region 141 in FIG. 14(b) improves protein stability. The locally denatured region can be found using the RMSF (Root Mean Square Fluctuation) value. The RMSF value is an index of fluctuation obtained by molecular dynamics simulation. In Non-Patent Document 1, the inventors proposed that a stable antibody can be designed by finding the locally denatured region using the RMSF value as an index.

[0009] Okazaki K, and 8 others. Molecular Dynamics-Based Design and Biophysical Evaluation of Thermostable Single-Chain Fv Antibody Mutants Derived from Pharmaceutical Antibodies. ACS Omega. 2023;8:22945-22954.

[0010] However, analysis using RMSF values ​​is manual, which limits the number of mutants that can be searched. Furthermore, because RMSF values ​​average residue movements, it is difficult to evaluate rare or small movements.

[0011] Therefore, an object of the present invention is to provide an antibody dynamic structural property analysis device and the like that can analyze the movements of residues without averaging them.

[0012] A first aspect of the present invention is an antibody dynamic structural property analysis device that analyzes the dynamic structural property of an antibody to be analyzed using data for identifying the antibody to be analyzed stored in an analysis target memory unit, the device comprising: a molecular dynamics calculation processing unit that calculates the movements of atoms and / or molecules of the antibody to be analyzed to obtain a plurality of snapshots; an R-plot processing unit that creates a Ramachandran plot of an n-th residue (n is a natural number from 1 to N) among N residues (N is a natural number) of the antibody to be analyzed, and obtains the dihedral angles φ and ψ of the n-th residue in each snapshot; and a dihedral angle reference value φ of the n-th residue for the dihedral angles φ and ψ of the n-th residue in each snapshot. 0 n and ψ 0 n and a motion analysis processing unit that determines that the nth residue is a deviant residue if the proportion of the collective variables of the nth residue in all snapshots whose collective variable values ​​are in a region exceeding the motion analysis reference value exceeds the proportion reference value.

[0013] A second aspect of the present invention is an antibody dynamic structural property analysis device according to the first aspect, which includes a mutation processing unit that stores data in the analysis target memory unit for identifying a new analysis target antibody obtained by introducing a mutation into the analysis target antibody.

[0014] A third aspect of the present invention is the antibody dynamic structural property analysis device of the first or second aspect, comprising an R-plot analysis reference value determination unit, wherein the molecular dynamics calculation processing unit calculates the movement of atoms and / or molecules for a plurality of reference antibodies to obtain a plurality of snapshots, the R-plot processing unit creates a Ramachandran plot of an n-th residue of the reference antibody to obtain dihedral angles φ and ψ of the n-th residue in each snapshot, and the R-plot analysis reference value determination unit determines the dihedral angle reference value φ of the n-th residue using the Ramachandran plot of the n-th residue of the reference antibody. 0 n and ψ 0 n Determine.

[0015] A fourth aspect of the present invention is an antibody dynamic structural property analysis method for analyzing dynamic structural properties of an antibody to be analyzed using data for identifying the antibody to be analyzed stored in an analysis target storage unit, the method including: a molecular dynamics calculation step in which a molecular dynamics calculation processing unit included in an information processing device calculates atomic and / or molecular movements for the antibody to be analyzed to obtain a plurality of snapshots; an R plot processing step in which an R plot processing unit included in the information processing device creates a Ramachandran plot for each n-th residue (n is a natural number from 1 to N) among N residues (N is a natural number) of the antibody to be analyzed to obtain dihedral angles φ and ψ; and a set variable calculation processing unit included in the information processing device calculates dihedral angle reference values ​​φ for the dihedral angles φ and ψ of the n-th residue in each snapshot. 0 n and ψ 0 n and a motion analysis step in which a motion analysis processing unit provided in the information processing device determines that the nth residue is a deviant residue if the proportion of the motion analysis values ​​of the nth residue in all snapshots that are in a region where the motion analysis reference value is exceeded exceeds the proportion reference value.

[0016] A fifth aspect of the present invention is the antibody dynamic structural property analysis method of the fourth aspect, which includes a mutation processing step in which a mutation processing unit provided in an information processing device stores data for identifying a new antibody to be analyzed obtained by introducing a mutation into the antibody to be analyzed in the analysis target memory unit.

[0017] A sixth aspect of the present invention is the method for analyzing antibody dynamic structural characteristics according to the fourth or fifth aspect, comprising the steps of: a step in which the molecular dynamics calculation processor calculates the motion of atoms and / or molecules for a plurality of reference antibodies to obtain a plurality of snapshots; a step in which the R plot processor creates a Ramachandran plot of an n-th residue of the reference antibodies to obtain dihedral angles φ and ψ of the n-th residue in each snapshot; and a step in which an R plot analysis reference value determiner included in an information processing device determines the dihedral angle reference value φ of the n-th residue using the Ramachandran plot of the n-th residue of the reference antibodies. 0 n and ψ0 n The method includes determining:

[0018] A seventh aspect of the present invention is a program for causing a computer to function as the antibody dynamic structural property analysis device of any one of the first to third aspects. The present invention may also be understood as a computer-readable recording medium on which the program of the seventh aspect is recorded.

[0019] The n-th residue may be, for example, a residue in a framework region. That is, the present invention may be understood as determining whether or not the n-th residue is a deviant residue for each n (n is a natural number from 1 to N) of N residues (N is a natural number ranging from 1 to N) that are some or all of the residues in the framework region.

[0020] In the third or sixth aspect, the aggregate variable calculation unit calculates a dihedral angle reference value φ for the dihedral angles φ and ψ of the n-th residue in each snapshot. 0 n and ψ 0 n The distance from the nth residue is calculated and set as the value of a collective variable, and the motion analysis processing unit can be considered to perform processing to set the nth motion analysis reference value as the reference value for determining that the motion for the nth residue is common among antibodies.

[0021] Furthermore, the present invention may be understood as the motion analysis processing unit determining that the nth residue is a deviant residue when the proportion of aggregate variables of the nth residue in all snapshots that are in a region where the aggregate variable values ​​exceed the motion analysis reference value (a region where the aggregate variable values ​​are greater than the motion analysis reference value) exceeds the proportion reference value (greater than the proportion reference value), and not determining that the nth residue is a deviant residue when the proportion does not exceed the proportion reference value (is equal to or less than the proportion reference value).

[0022] According to the present invention, by converting snapshots obtained by calculating atomic and / or molecular motions from two-dimensional values, i.e., dihedral angles φ and ψ in a Ramachandran plot, into one-dimensional values, known as collective variables, and then analyzing the results, it is possible to realize an analysis of residue motions without averaging. This makes it possible to identify local denatured states that could not be detected by analysis using RMSF values. Furthermore, it is possible to compare multiple antibodies simultaneously. Therefore, the present invention significantly improves evaluation performance, enabling the rapid design of antibodies with high thermal stability. Furthermore, it may be possible to create more stable antibodies.

[0023] 1 is a block diagram showing an example of the configuration of an antibody dynamic structural characteristics analysis apparatus 1 according to an embodiment of the present invention. FIG. 2 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 3 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 4 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 4 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 5 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 6 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 7 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 8 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 9 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. FIG. 10 is a flow chart showing an example of the operation of the antibody dynamic structural characteristics analysis apparatus 1 of FIG. 1. 1 is a diagram for explaining an antibody. FIG. 2 is a diagram for explaining the denaturation process of a protein.

[0024] Hereinafter, embodiments of the present invention will be described with reference to the drawings, but the present invention is not limited to these embodiments.

[0025] FIG. 1 is a block diagram showing an example of the configuration of an antibody dynamic structural property analysis device 1 according to an embodiment of the present invention.

[0026] The antibody dynamic structural characteristic analysis device 1 is, for example, an information processing device, and includes an input / output processing device 3, a storage device 5, and a processing device 7.

[0027] The input / output processing device 3 inputs and outputs data between external people and devices. The input / output processing device 3 is, for example, a keyboard, mouse, display, touch panel, etc., and is used by the user to input data into the antibody dynamic structural property analysis device 1 and output data to the user. The input / output processing device 3 also communicates with, for example, other devices, and acquires data from external information processing devices, databases, etc., and transmits data to external information processing devices.

[0028] The storage device 5 is a device for storing data, such as a memory, a hard disk, etc. The storage device 5 includes an analysis target storage unit 11, a snapshot storage unit 13, an R plot storage unit 15, an R plot analysis reference value storage unit 17, a set variable storage unit 19, a set variable plot storage unit 21, and a motion analysis storage unit 23.

[0029] The processing device 7 is, for example, a central processing unit (CPU) or the like, and is a device that processes data. The processing device 7 includes a control processing unit 25, a mutation processing unit 27, a molecular dynamics calculation processing unit 29, an R plot processing unit 31, an R plot analysis reference value determination unit 33, a set variable calculation processing unit 35, a set variable plot processing unit 37, and a motion analysis processing unit 39. Each unit included in the processing device 7 can be realized, for example, by the processing device 7 operating under the control of a program.

[0030] 2 and 3 are flow charts showing an example of the operation of the antibody dynamic structural property analysis device 1 of FIG.

[0031] The process of analyzing a reference antibody will be described with reference to Figure 2. In this example, a medical antibody is used as the reference antibody. The input / output processing device 3 acquires data for identifying the medical antibody from the protein three-dimensional structure database and stores it in the analysis target storage unit 11. The molecular dynamics calculation processing unit 29 calculates the atomic and / or molecular movements of the medical antibody stored in the analysis target storage unit 11, for example, by molecular dynamics calculation, to obtain multiple snapshots (step STA1). The snapshot storage unit 13 stores the obtained snapshots.

[0032] Seventeen types of medical antibodies are registered in the protein 3D structure database. The following explanation uses an example of calculations for 17 types of medical antibodies, but in the present invention, calculations may be performed using only a portion of these, or calculations may be performed by adding new medical antibodies. Furthermore, in the following explanation, an example is given in which 3,000 snapshot frames are obtained per antibody, but in the present invention, a different number of snapshots may be obtained.

[0033] The number of residues in the framework region is defined as N (N is a natural number). The R plot processing unit 31 creates a Ramachandran plot for each of the N residues in the framework region and merges the data for all antibodies (step STA2). The R plot storage unit 15 stores the created data. As an example, a case will be described in which 3,000 frame snapshots are obtained per antibody for a total of 17 types of antibodies, and the Ramachandran plot for each residue is composed of 3,000 frames x 17 antibodies = 51,000 pieces of data. Note that in the present invention, N may be a partial or complete number of residues in the framework region.

[0034] Amino acids are linked by peptide bonds. The peptide bond (C-N) has the properties of a partial double bond, lies in the plane of the peptide, and cannot rotate freely. However, the alpha carbon C α (the carbon atom on the main chain side located next to the functional group of interest) α Bond and C αThe N-C bond) is not rigid and is free to rotate, limited only by the size and nature of the side chain (R group). α The rotation angle around the bond is called φ (phi), and C α The rotation angle around the -C bond is called ψ (psi). The rotation angles φ and ψ are also called torsion angles, dihedral angles, etc. These dihedral angles φ and ψ largely determine the three-dimensional shape of the polypeptide backbone of a protein.

[0035] A Ramachandran plot is a plot of ψ against φ of amino acid residues in a protein structure. Many software programs are available for creating Ramachandran plots. The R plot processing unit 31 can use such software to create a Ramachandran plot for each residue in the framework region.

[0036] The R plot processing unit 31 merges the data of all medical antibodies. That is, it integrates the data into one according to predetermined rules. In each snapshot, a combination of dihedral angles φ and ψ is determined for each residue. Therefore, in an example where 3,000 snapshot frames are obtained per antibody for a total of 17 types of medical antibodies, the Ramachandran plot for each residue contains 3,000 frames x 17 antibodies = 51,000 combinations of dihedral angles φ and ψ.

[0037] The control processing unit 25 sets the initial value of the variable n to 1 (step STA3).

[0038] The R-plot analysis standard value determination unit 33 determines the dihedral angle standard value φ using the Ramachandran plot of the n-th residue. 0 n and ψ 0 n (Step STA4). The R-plot analysis reference value storage unit 17 determines the dihedral angle reference value φ 0 n and ψ 0 n The dihedral angle reference value φ 0 n and ψ 0 nis, for example, the mode of the Ramachandran plot of the n-th residue. As a method for finding and clustering densely packed data groups from a data set, for example, DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is known. The mode of the Ramachandran plot can be obtained using, for example, DBSCAN. In the present invention, the dihedral angle reference value φ 0 n and ψ 0 n may be a value other than the mode.

[0039] The set variable calculation unit 35 calculates the dihedral angle reference value φ of the n-th residue for the dihedral angles φ and ψ of the n-th residue in each snapshot. 0 n and ψ 0 n The distance from the point φ is calculated and set as the value of the set variable (step STA5). The set variable storage unit 19 stores the calculated set variable. The dihedral angles φ and ψ included in the snapshot and the dihedral angle reference value φ 0 n and ψ 0 n The distance between the n-th residue and the dihedral angle reference value φ is small as the difference is small and large as the difference is large. 0 n and ψ 0 n The distance between the dihedral angles φ and ψ of the nth residue in each snapshot and the dihedral angle reference value φ of the nth residue is 0 n and ψ 0 n The aggregate variables are close to zero when antibodies exhibit highly common dynamic structural properties, and the greater the distance, the larger the value. Therefore, the two-dimensional data of the dihedral angles φ and ψ obtained by the Ramachandran plot can be treated as one-dimensional data using aggregate variables.

[0040] The aggregate variable plot processing unit 37 plots the aggregate variables obtained for the n-th residue (step STA6). For example, if snapshots of 3,000 frames per antibody are obtained for a total of 17 types of medical antibodies, the aggregate variables for each residue will have 3,000 frames x 17 antibodies = 51,000 values. The aggregate variable plot storage unit 21 stores the plot of the aggregate variables. For example, in the graphs of Figures 5(a) and (b), the line L 11 and L 13 is a plot of the aggregate variables obtained from the clinical antibodies.

[0041] The motion analysis memory unit 23 stores the nth motion analysis reference value. The nth motion analysis reference value defines, for the nth residue, a movement common to antibodies when the aggregate variable is in a range equal to or less than the nth motion analysis reference value. If necessary, the motion analysis processing unit 39 performs processing to set the nth motion analysis reference value as a reference value for determining a movement common to antibodies for the nth residue (step STA7). Note that the nth motion analysis reference value may be a value common to all residues (e.g., 50), or may be a value that varies depending on the distribution of the aggregate variables. For example, the motion analysis processing unit may adjust the calculation formula for the aggregate variable of the nth residue so that the distribution corresponds to the value of the nth motion analysis reference value, or may determine the value of the nth motion analysis reference value depending on the distribution of the aggregate variables. Note that this processing may not be performed if not necessary.

[0042] The control processing unit 25 increments n by 1 (step STA8). The control processing unit 25 determines whether n is greater than N (step STA9). If n is greater than N, the process in Fig. 2 ends. If n is not greater than N, the process returns to step STA4.

[0043] Referring to FIG. 3, the process for analyzing an antibody of interest (AOI) will be described.

[0044] The input / output processing device 3 acquires data for identifying an antibody (AOI) under analysis without mutation from an analysis device that analyzes an actual antibody and acquires data identifying the antibody, or a simulation device that generates data identifying the antibody through simulation, and stores the data in the analysis target storage unit 11. The molecular dynamics calculation processing unit 29 calculates the atomic and / or molecular movements of the antibody under analysis stored in the analysis target storage unit 11, for example, by molecular dynamics calculation, to obtain multiple snapshots (step STB1). The snapshot storage unit 13 stores the obtained snapshots. For example, 3,000 frames of snapshots are obtained for the AOI.

[0045] The R plot processing unit 31 creates a Ramachandran plot for each of the N residues in the framework region of the AOI to obtain the dihedral angles φ and ψ (step STB2). The R plot storage unit 15 stores the obtained Ramachandran plot. For example, in an example where 3000 frame snapshots are obtained for the AOI, the Ramachandran plot for each residue includes 3000 combinations of dihedral angles φ and ψ.

[0046] The control processing unit 25 sets the initial value of the variable n to 1 and the reference ratio value to 20% (step STB3).

[0047] The set variable calculation unit 35 calculates the dihedral angle reference value φ of the n-th residue for the dihedral angles φ and ψ of the n-th residue in each snapshot. 0 n and ψ 0 n The distance from is calculated and set as the value of the set variable (step STB4). For example, if 3000 frame snapshots are obtained for the AOI, the set variable for each residue will have 3000 values. The set variable storage unit 19 stores the calculated set variable values.

[0048] The set variable plot processing unit 37 plots the set variables obtained for the n-th residue (step STB5). The set variable plot storage unit 21 stores the plot of the set variables. For example, in the graph of FIG. 5(a), the line L 12The graph in FIG. 5(b) plots the population variables of the antibody before the mutations were introduced. 14 is a plot of the population variables for the antibody after introducing mutations.

[0049] The motion analysis processing unit 39 determines that the nth residue is a deviant residue if the proportion of the set variables of the nth residue that are in a region exceeding the nth motion analysis reference value exceeds the proportion reference value (step STB6). If the proportion of the set variables of the nth residue that are in a region exceeding the nth motion analysis reference value does not exceed the proportion reference value, the nth residue is not determined to be a deviant residue. The motion analysis storage unit 23 stores whether or not the nth residue is determined to be a deviant residue. Figure 4 shows an example where the proportion of the set variables of the nth residue that are in a region exceeding the nth motion analysis reference value does not exceed the proportion reference value. The horizontal axis represents the value of the set variable, and the vertical axis represents density (the number of snapshots belonging to each set variable value divided by the total number of snapshots).

[0050] Control processing unit 25 increments n by 1 (step STB7). Control processing unit 25 determines whether n is greater than N (step STB8). If n is not greater than N, the process returns to step STB4.

[0051] If n is greater than N, the control processing unit 25 displays the residues determined to be deviant, for example, to the input / output processing device 3, and determines whether to terminate the process (step STB9). For example, if the user determines that the antibody to be analyzed has been stabilized and instructs the input / output processing device 3 to terminate the process, the process of FIG. 3 is terminated.

[0052] If the user does not terminate the processing because the input / output processing device 3 continues processing, data for introducing mutations (e.g., data specifying the amino acid to be mutated and specifying the type of mutation to be introduced) is input to the input / output processing device 3. The framework regions have high conservation of amino acid residues and structures between antibodies, and their dynamic structural characteristics are also similar. Structural distortions exist around regions that deviate from the common behavior, and correcting these distortions through mutations is expected to produce antibodies with high thermal stability. Potential mutation sites exist around spatially deviating residues. Therefore, to stabilize the antibody, the user introduces mutations around residues determined to be deviating residues. Identifying the deviating residues in step STB6 is useful for determining the site to introduce this mutation. The mutation processing unit 27 changes the antibody to be analyzed in the analysis target memory unit 11 to one after introducing mutations into the specified amino acids, stores the result in the analysis target memory unit 11 (step STB10), and returns to step STB1.

[0053] An example of an experiment using the antibody Trastuzumab conducted by the inventors will be described with reference to Figures 5 to 8. In the graphs of each figure, the horizontal axis represents the value of the aggregation variable, and the vertical axis represents density.

[0054] Figures 5(a), 6(a), 7(a)-(e), and 8(a)-(d) show plots of the population variables obtained with the wild-type antibody before mutations were introduced. Figure 5(a) shows residue 14, Figure 6(a) shows residue 18, Figures 7(a)-(e) show residues 8-12, and Figures 8(a)-(d) show residues 13, 15, 16, and 17.

[0055] Figures 5(b), 6(b), 7(f)-(j), and 8(e)-(h) show plots of the population variables obtained by the antibody after introducing mutations: Figure 5(b) shows residue 14, Figure 6(b) shows residue 18, Figure 7(f)-(j) shows residues 8-12, and Figures 8(e)-(h) show residues 13, 15, 16, and 17.

[0056] Referring to FIG. 5(a), for the 14th residue, the line L 11 is a plot of the aggregate variables obtained from the medical antibody, and the line L12 is a plot of the population variables obtained for the wild-type antibody before the mutations were introduced.

[0057] Line L 12 is the line L 11 There are many residues with large values ​​for the aggregate variables compared to the 14th residue. The motion analysis processing unit 39 determines that the 14th residue is a deviant residue because the proportion of aggregate variables for the 14th residue that are in the region exceeding the 14th motion analysis reference value exceeds the proportion reference value. The user introduces mutations around the 14th residue.

[0058] Referring to FIG. 5B, the line L 13 is a plot of the aggregate variables obtained from the therapeutic antibodies, and the line L in FIG. 11 The line L 14 is a plot of the population variables obtained for the antibody after introducing mutations around residue 14.

[0059] Line L 14 is the line L 13 In Fig. 5(b), the motion analysis processing unit 39 is able to prevent the 14th residue from being determined to be a deviant residue because the proportion of aggregate variables of the 14th residue that are in the region exceeding the 14th motion analysis reference value does not exceed the proportion reference value.

[0060] Referring to FIG. 6(a), for the 18th residue, the line L 21 is a plot of the aggregate variables obtained from the medical antibody, and the line L 22 is a plot of the population variables obtained for the wild-type antibody before the mutations were introduced. 22 is the line L 21 Compared to the above, there were many cases where the values ​​of the collective variables were large.

[0061] Referring to FIG. 6(b), the line L 23 is a plot of the aggregate variables obtained from the medical antibody, and the line L 24 Figure 6(b) plots the population variables obtained for the antibody after introducing mutations around residue 14. Figure 6(b) confirms that introducing mutations around residue 14 also stabilizes residue 18.

[0062] Comparing Figures 7(a) to (e) and Figures 8(a) to (d) with Figures 7(f) to (j) and Figures 8(e) to (h), it can be seen that the distribution of the collective variables does not change significantly before and after the introduction of the mutations.

[0063] An example of an experiment using an OKT3 antibody conducted by the inventors will be described with reference to Figures 9 to 12. In the graphs of each figure, the horizontal axis represents the value of the aggregation variable, and the vertical axis represents density.

[0064] Figures 9(a), 10(a), 11(a)-(e), and 12(a)-(e) show plots of the population variables obtained with the wild-type antibody before mutations were introduced. Figure 9(a) shows residue 9, Figure 10(a) shows residue 27, Figures 11(a)-(e) show residues 24, 25, 26, 28, and 29, and Figures 12(a)-(e) show residues 30-34.

[0065] Figures 9(b), 10(b), 11(f)-(j), and 12(f)-(j) show plots of the population variables obtained for the antibody after mutagenesis: Figure 9(b) shows residue 9, Figure 10(b) shows residue 27, Figure 11(f)-(j) shows residues 24, 25, 26, 28, and 29, and Figure 12(f)-(j) shows residues 30-34.

[0066] Referring to FIG. 9(a), for the 9th residue, the line L 31 is a plot of the aggregate variables obtained from the medical antibody, and the line L 32 is a plot of the population variables obtained for the wild-type antibody before the mutations were introduced.

[0067] Line L 32 is the line L 31 Compared to the ninth residue, there are many residues with large values ​​for the aggregate variables. The motion analysis processing unit 39 determines that the ninth residue is a deviant residue because the proportion of aggregate variables for the ninth residue that are in the region exceeding the ninth motion analysis reference value exceeds the proportion reference value. The user introduces mutations around the ninth residue.

[0068] Referring to FIG. 9(b), the line L 33 is a plot of the aggregate variables obtained from the medical antibody, and the line L 34is a plot of the population variables obtained for the antibody after introducing mutations around residue 9.

[0069] Line L 34 is the line L 33 9B, the motion analysis processing unit 39 is able to prevent the ninth residue from being determined to be a deviant residue because the proportion of aggregate variables of the ninth residue that are in the region exceeding the ninth motion analysis reference value does not exceed the proportion reference value.

[0070] Referring to FIG. 10(a), for the 27th residue, the line L 41 is a plot of the aggregate variables obtained from the medical antibody, and the line L 42 is a plot of the population variables obtained for the wild-type antibody before the mutations were introduced. 42 is the line L 41 Compared to the above, there were many cases where the values ​​of the collective variables were large.

[0071] Referring to FIG. 10(b), the line L 43 is a plot of the aggregate variables obtained from the medical antibody, and the line L 44 Figure 10(b) plots the population variables obtained for the antibody after introducing mutations around residue 9. Figure 10(b) confirms that introducing mutations around residue 9 also stabilizes residue 27.

[0072] Comparing Figures 11(a) to (e) and Figures 12(a) to (e) with Figures 11(f) to (j) and Figures 12(f) to (j), it can be seen that the distribution of the collective variables does not change significantly before and after the introduction of the mutations.

[0073] Antibodies are used in pharmaceuticals, diagnostic reagents, sensor elements, etc., but they are vulnerable to physical stress (heat, friction, vibration, etc.), so the creation of stable antibodies is essential for practical use. Analysis using RMSF values ​​relies on manual analysis, which limits the number of mutants that can be explored. Furthermore, because RMSF values ​​average residue movements, it is difficult to evaluate rare or small movements.

[0074] Framework regions are highly conserved in amino acid residues and structure among antibodies, and their dynamic structural characteristics are similar. Regions that deviate from the standard dynamic structural characteristics are thought to exhibit structural distortions, which are expected to reduce stability. The inventors performed molecular dynamics calculations on 17 therapeutic antibodies registered in a protein structure database using molecular dynamics, a method for computer simulation of protein dynamics. Clustering analysis was performed on the results to identify common dynamic structural characteristics among antibodies. Based on this data, a set of variables was defined that approaches zero when antibodies exhibit highly common dynamic structural characteristics, and increases the greater the deviation. A set of variables with a higher value than a threshold was defined as a normal value if the percentage of frames with a set of variables below a certain percentage reference value, and as an outlier if the percentage was higher than the reference value. Mutations were introduced computationally around the outlier regions, molecular dynamics calculations were performed, and the set of variables were calculated again. The results confirmed that the outliers disappeared in mutants with improved thermal stability. The inventors tested multiple antibodies and confirmed the general applicability of this method.

[0075] The present invention realizes analysis without averaging, making it possible to find locally denatured regions that could not be found in analysis using RMSF values. Furthermore, it is possible to compare multiple antibodies simultaneously. The present invention significantly improves evaluation performance, enabling the rapid design of antibodies with high thermal stability. The present invention can shorten the time required to obtain stable antibodies from the conventional several months to a year to about two weeks. Furthermore, it may be possible to produce more stable antibodies.

[0076] REFERENCE SIGNS LIST 1 Antibody dynamic structure property analysis device 3 Input / output processing device 5 Storage device 7 Processing device 11 Analysis target memory unit 13 Snapshot memory unit 15 R plot memory unit 17 R plot analysis reference value memory unit 19 Set variable memory unit 21 Set variable plot memory unit 23 Motion analysis memory unit 25 Control processing unit 27 Mutation processing unit 29 Molecular dynamics calculation processing unit 31 R plot processing unit 33 R plot analysis reference value determination unit 35 Set variable calculation processing unit 37 Set variable plot processing unit 39 Motion analysis processing unit

Claims

1. An antibody dynamic structural property analysis device for analyzing dynamic structural properties of an antibody using data for identifying an antibody to be analyzed stored in an analysis target memory unit, comprising: a molecular dynamics calculation processing unit for calculating atomic and / or molecular movements for the antibody to be analyzed to obtain a plurality of snapshots; an R-plot processing unit for creating a Ramachandran plot of an n-th residue (n is a natural number from 1 to N) among N residues (N is a natural number) of the antibody to be analyzed, to obtain dihedral angles φ and ψ of the n-th residue in each snapshot; and a dihedral angle reference value φ of the n-th residue for the dihedral angles φ and ψ of the n-th residue in each snapshot. 0 n and ψ 0 n a collective variable calculation processing unit that calculates the distance from and sets the distance as the value of a collective variable; and a behavior analysis processing unit that determines that the nth residue is a deviant residue if the proportion of collective variables of the nth residue in all snapshots whose collective variable values ​​are in a region exceeding a behavior analysis reference value exceeds the proportion reference value.

2. An antibody dynamic structural property analysis device as described in claim 1, further comprising a mutation processing unit that stores in said analysis target memory unit data for identifying a new analysis target antibody obtained by introducing a mutation into said analysis target antibody.

3. An R-plot analysis reference value determination unit is provided, wherein the molecular dynamics calculation processing unit calculates the motions of atoms and / or molecules for a plurality of reference antibodies to obtain a plurality of snapshots, the R-plot processing unit creates a Ramachandran plot for an n-th residue of the reference antibody to obtain dihedral angles φ and ψ of the n-th residue in each snapshot, and the R-plot analysis reference value determination unit determines the dihedral angle reference value φ of the n-th residue using the Ramachandran plot of the n-th residue of the reference antibody. 0 n and ψ 0 n The antibody dynamic structural characteristic analysis device according to claim 1, wherein 4. An antibody dynamic structural characteristic analysis method for analyzing dynamic structural characteristics of an antibody by using data for identifying an antibody to be analyzed stored in an analysis target memory unit, comprising: a molecular dynamics calculation step in which a molecular dynamics calculation processing unit included in an information processing device calculates the movement of atoms and / or molecules for the antibody to be analyzed to obtain a plurality of snapshots; an R plot processing step in which an R plot processing unit included in the information processing device creates a Ramachandran plot for each n-th residue (n is a natural number from 1 to N) among N residues (N is a natural number) of the antibody to be analyzed, and obtains dihedral angles φ and ψ; and a set variable calculation processing unit included in the information processing device calculates dihedral angle reference values ​​φ for the dihedral angles φ and ψ of the n-th residue in each snapshot. 0 n and ψ 0 n a set variable calculation step of calculating a distance from the nth residue to set the calculated distance as a set variable value, and a behavior analysis step in which a behavior analysis processing unit included in the information processing device determines that the nth residue is a deviant residue if the proportion of the set variables of the nth residue in all snapshots whose values ​​are in a region exceeding a behavior analysis reference value exceeds the proportion reference value.

5. A method for analyzing antibody dynamic structural characteristics as described in claim 4, including a mutation processing step in which a mutation processing unit provided in an information processing device stores in the analysis target memory unit data for identifying a new analysis target antibody obtained by introducing a mutation into the analysis target antibody.

6. A step in which the molecular dynamics calculation processing unit calculates the motions of atoms and / or molecules for a plurality of reference antibodies to obtain a plurality of snapshots; a step in which the R plot processing unit creates a Ramachandran plot for an n-th residue of the reference antibody to obtain dihedral angles φ and ψ of the n-th residue in each snapshot; and a step in which an R plot analysis reference value determination unit provided in an information processing device determines the dihedral angle reference value φ of the n-th residue using the Ramachandran plot of the n-th residue of the reference antibody. 0 n and ψ 0 n The method for analyzing antibody dynamic structural characteristics according to claim 4, further comprising the step of determining:

7. A program for causing a computer to function as the antibody dynamic structural property analysis device according to claim 1.

Images