Prediction of peptide cleavage in polypeptides through physics-based simulations

Abstract

The present disclosure relates to polypeptide degradation, and in particular to techniques for predicting the probability that a peptide bond in a given polypeptide molecule is susceptible to cleavage reactions. Specifically, aspects of the present disclosure are directed to generating a representation of a polypeptide, performing molecular dynamics simulations using the representation to obtain a set of polypeptide conformations, determining the spatial properties of amino acids for each polypeptide conformation, estimating the nucleophilic attack distance of each polypeptide conformation based on the spatial properties, identifying reactive conformations that are susceptible to cleavage reactions based on the nucleophilic attack distance of each polypeptide conformation, determining the free energy of the spatial properties of the amino acids in the reactive conformations, and predicting the probability that a side chain of an amino acid will be trapped in the reactive conformation based on the free energy.

Description

[Technical Field]

[0001] Claim of priority This application claims the benefit and priority of U.S. Provisional Patent Application No. 63 / 051,166, filed July 13, 2020, which is incorporated herein by reference in its entirety for all purposes.

[0002] This disclosure relates to the degradation of polypeptides, and more particularly to a technique for predicting the probability that the peptide bonds of a given polypeptide molecule are susceptible to cleavage reactions. [Background technology]

[0003] Polypeptide-based therapeutics have been successful and now represent a significant proportion of new drug approvals. Partly, this success can be attributed to the high affinity and specificity that can be achieved with polypeptides, such as monoclonal antibodies (mAbs), against key disease targets. In addition, mAbs have a long serum half-life through the interaction between the fragment crystallizable region (Fc region) (the tail region of the antibody) and the Fc region repurposing receptor (FcRn), thus enabling less frequent dosing. In some disease settings, such as in acute treatment where a long half-life is undesirable, or in tissue environments where FcRn repurposing is not active, such as the eye, antigen-binding fragments (Fabs) may be preferred over intact mAbs.

[0004] Despite these therapeutic advantages, mAbs and antibody fragments are prone to chemical and physical instability, which can lead to polypeptide degradation and ultimately limit their usefulness. Physical instability can manifest as soluble aggregation, precipitation, and gel formation. Chemical instability can manifest as deamidation (e.g., asparagine (Asn) deamidation), isomerization (e.g., aspartic acid (Asp) isomerization), and oxidation (e.g., oxidation of tryptophan (Trp) and methionine (Met) residues). In the context of biologics, degradation can reduce the availability of polypeptide therapeutics and / or the probability of inducing the biological effect of the target. For example, Asp isomerization results in a loss of potency of polypeptide therapeutics, and the formation of isoaspartic acid due to Asp isomerization is associated with Alzheimer's disease. It is advantageous to be able to detect the probability of degradation of a given polypeptide early in the therapeutic development process. [Overview of the Initiative]

[0005] In various embodiments, a computer implementation method is provided, which includes: determining the dihedral angles of the amino acid skeleton and the dihedral angles of the side chains when the polypeptide structure of a polypeptide comprising amino acids having side chains and a skeleton; determining the nucleophilic attack distance between two atoms, functional groups, or combinations thereof of the amino acid when the polypeptide structure is present, based on the dihedral angles of the skeleton and the dihedral angles of the side chains, such that one of the two atoms or functional groups is in the side chain of the amino acid and the other is in the amino acid skeleton; determining that the polypeptide structure is a reactive structure that is sensitive to cleavage reactions, based on the nucleophilic attack distance of the amino acid when the polypeptide structure is present; determining the free energies of the dihedral angles of the amino acid skeleton and the dihedral angles of the side chains when the polypeptide structure is in the reactive structure, in accordance with the determination that the polypeptide structure is a reactive structure; and predicting the probability that the amino acid side chains will be captured in the reactive structure, based on the free energies of the dihedral angles of the amino acid skeleton and the dihedral angles of the side chains.

[0006] In some embodiments, the computer implementation method further includes: generating a representation of a polypeptide; and performing a molecular dynamics simulation using the representation, wherein the result of performing the molecular dynamics simulation includes a set of polypeptide stereostructures of the polypeptide, including the polypeptide stereostructure.

[0007] In some embodiments, the computer implementation method further includes predicting the probability that a polypeptide will undergo chemical degradation as a result of amino acid side chains being trapped in a reactive stereostructure.

[0008] In some embodiments, the computer implementation method further includes outputting the probability that an amino acid side chain is trapped in at least one reactive stereostructure and / or the probability that the polypeptide undergoes chemical degradation.

[0009] In some embodiments, the computer implementation method further includes removing polypeptides from a list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chains are trapped in at least one reactive stereostructure and / or the probability that the polypeptides undergo chemical degradation.

[0010] In some embodiments, the computer implementation method ranks a polypeptide lower than another polypeptide in a list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chain of another polypeptide is captured in at least one reactive stereostructure and / or the polypeptide is chemically degraded, further comprising ranking a polypeptide lower than another polypeptide if the probability that the amino acid side chain of another polypeptide is captured in at least one reactive stereostructure and / or the polypeptide is chemically degraded is smaller than the probability that the amino acid side chain of the polypeptide is captured in at least one reactive stereostructure and / or the polypeptide is chemically degraded.

[0011] In some embodiments, predicting the probability of a polypeptide chemically decomposing involves confirming accessibility constraints that, if satisfied, indicate the amide group of the polypeptide has spatial accessibility exceeding a threshold for binding to solvent molecules of the surrounding solvent; and determining, based on an evaluation of one or more spatial properties of the polypeptide, that the accessibility constraints are satisfied with respect to the reactive stereostructure.

[0012] In some embodiments, determining whether a polypeptide stereostructure is a reactive stereostructure includes determining a distance criterion that, if satisfied, indicates that an atom in a side chain is within a predetermined distance threshold of another atom in the backbone; and determining that the distance criterion is satisfied for the reactive stereostructure based on a comparison of the nucleophilic attack distance of the amino acids in the reactive stereostructure with the predetermined distance threshold.

[0013] In some embodiments, the free energy is determined based on an analysis of the free energy profiles of the backbone dihedral angles and side chain dihedral angles of the amino acids in the reactive conformation, and the free energy profile in the space of the backbone dihedral angles and side chain dihedral angles is calculated from the bin population.

[0014] In some embodiments, predicting the probability that an amino acid side chain is trapped in a reactive conformation includes determining an energy criterion that indicates that the free energy of the backbone dihedral angle and side chain dihedral angle of the amino acid is within a predefined energy threshold when satisfied; and determining that the energy criterion is satisfied for the reactive conformation based on a comparison of the free energy of the backbone dihedral angle and side chain dihedral angle of the amino acid in the reactive conformation with the predefined energy threshold.

[0015] In some embodiments, a system is provided that includes one or more data processors and a non-transitory computer-readable storage medium containing instructions that, when executed by the one or more data processors, cause the one or more data processors to perform some or all of one or more of the methods disclosed herein.

[0016] In some embodiments, a computer program product is provided that is tangibly integrated into a non-transitory machine-readable storage medium and includes instructions configured to cause one or more data processors to perform some or all of one or more of the methods disclosed herein.

[0017] Some embodiments of the present disclosure include a system that includes one or more data processors. In some embodiments, the system includes a non-transitory computer-readable storage medium that includes instructions that, when executed by one or more data processors, cause the one or more data processors to perform some or all of one or more of the methods disclosed herein and / or some or all of one or more of the processes. Some embodiments of the present disclosure include a computer program product tangibly incorporated in a non-transitory machine-readable storage medium that includes instructions configured to cause one or more data processors to perform some or all of one or more of the methods disclosed herein and / or some or all of one or more of the processes.

[0018] The terms and expressions used are used as terms for explanation and not for limitation, and the use of such terms and expressions is not intended to exclude any equivalents of the features shown and described or portions thereof, and it is recognized that various modifications are possible within the scope of the invention claimed. Thus, the invention claimed is specifically disclosed by embodiments and optional features, but modifications and changes to the concepts disclosed herein may be sought by those skilled in the art, and it should be understood that such modifications and changes are within the scope of the invention as defined by the appended claims.

[0019] The present disclosure is described with the accompanying drawings.

Brief Description of the Drawings

[0020] [Figure 1] Exemplary representations of cleavage reactions according to various embodiments are shown. [Figure 2A] Two ideal reactive conformations (A and B) are shown. Dihedral angles that minimize the distance of nucleophilic attack of the backbone carbonyl to the asparagine (Asn) side chain nitrogen according to various embodiments are shown. [Figure 2B]Two ideal reactive stereostructures (A and B) are shown. The dihedral angles that minimize the distance of nucleophilic attack of the carbonyl skeleton to the asparagine (Asn) side chain nitrogen in various embodiments are shown. [Figure 3A-F] The free energy profiles along the skeletal dihedral angle Ψ, and the two-dimensional free energy landscape along the side-chain dihedral angles X1 and X2, calculated from 1.5us molecular power orbitals using various embodiments, are shown. [Figure 4] This paper describes the process of generating the probability of a cleavage reaction based on molecular dynamics simulations and evaluations of the spatial properties of molecules, using various embodiments. [Figure 5] Examples of computing devices suitable for the use of systems and methods for molecular dynamics simulations in various embodiments are shown. [Figure 6A-C] Extracted ion chromatograms of natural peptides carrying the CDR-L3 sequence (Figure 6A), N-terminal hydrolysis product (Figure 6B), and C-terminal hydrolysis product (Figure 6C) according to various embodiments are shown. [Figure 7A-D] The MS1 spectra of the N-terminal hydrolysis product eluted at 98.0 minutes (Figure 7A), the MS1 spectra of the N-terminal hydrolysis product eluted at 98.8 minutes (Figure 7B), the theoretical MS1 spectra of the N-terminal hydrolysis product of Asn (Figure 7C), and the theoretical MS1 spectra of the N-terminal hydrolysis product of Asp (Figure 7D) are shown for various embodiments. [Figure 8A-B] This document demonstrates the hydrolysis of Asn-Pro peptide bonds in Fab2 through various embodiments. [Figure 9] This shows the rate of Asn-Pro peptide hydrolysis in test antibodies under various embodiments. [Modes for carrying out the invention]

[0021] In the attached drawings, similar components and / or features may have the same reference designation. Furthermore, various components of the same type may be distinguished by the subsequent dashed reference designation and by a second designation that distinguishes similar components. Where only the first reference designation is used herein, the description applies to any one of the similar components having the same first reference designation, regardless of the second reference designation.

[0022] I. Overview This disclosure describes a technique for predicting the probability that a peptide bond in a given polypeptide molecule (e.g., a peptide bond between asparagine (Asn) and a bulky residue, e.g., proline (Pro)) is susceptible to a cleavage reaction. A given polypeptide can have any number of stereostructures, some of which are reactive and some are non-reactive. The cleavage reaction resulting in the degradation of the polypeptide may include reactions between multiple atoms (e.g., nucleophilic attack of a carbonyl side chain nitrogen or nucleophilic attack of a γ-carbon of a side chain to a nitrogen of the backbone). Whether a polypeptide adopts a reactive stereostructure in which atoms react in a cleavage reaction depends on a number of factors, including the physical proximity of atoms (e.g., nucleophilic attack distance (dN)), the free energy profile of the dihedral angle of the polypeptide, steric hindrance due to stereobulk, and environmental conditions, e.g., pH and solvent accessibility.

[0023] IA interatomic distance reaction constraints Whether a reaction can occur between two atoms of a molecule (e.g., a nucleophilic attack on one of the two atoms) can depend on the proximity of the two atoms. In some cases, the spatial characteristics of the peptide conformation can include absolute or relative atomic positions, and / or the distance between two atoms. In some cases, the spatial characteristics can affect or determine how close two atoms in a molecule are to each other (and thus whether a reaction can occur), or other geometry-related information, such as the angles between multiple atoms, or the dihedral angles related to some or all of the atoms involved in the reaction (e.g., the multiple dihedral angles of the amino acids adjacent to a sensitive site, such as a sensitive hydrolysis site). For example, the spatial characteristics can include (i) a single backbone dihedral angle Ψ defined by four atoms of the backbone (N n -C α -C-N n+1 ), and (ii) two side-chain dihedral angles Χ1 defined by four atoms of the side chain (C-C α -C β -C γ ) and Χ2 defined by four atoms of the side chain (C α -C β -C γ -O), and these can be used to estimate the distance between the side-chain nitrogen atom and the γ-carbon of the backbone.

[0024] Dihedral angles can be estimated by defining spaces corresponding to one dihedral angle along one axis of space (e.g., Ψ), another dihedral angle along another axis of space (e.g., X1), and another dihedral angle along yet another axis of space (e.g., X2). Numerous regions within space can be defined based on spatial properties, each associated with a predicted response probability, which may include numerical probabilities, classification probabilities (e.g., very low, low, moderate, high), or binary probabilities. For example, a first region could correspond to a specific range of dihedral angles (e.g., Ψ, X1, and X2) that constitute a polypeptide such that the distance between two atoms that could be involved in a nucleophilic attack is minimal or below a threshold (e.g., 2 angstroms or 3 angstroms). A second (e.g., remaining) region could correspond to a specific range of dihedral angles that constitute a polypeptide such that the two atoms are separated beyond a threshold and therefore unlikely to be involved in a nucleophilic attack.

[0025] IB steric hindrance and free energy constraints The probability of a cleavage reaction can be explored through the determination of the side-chain stereostructure by molecular dynamics model simulations. If the three-dimensional structure of the polypeptide restricts the side chain to a non-reactive stereostructure, it is energetically unfavorable for the side chain to approach the reactive stereostructure. Therefore, the presence of steric hindrance can result in a very high free energy barrier to approaching the reactive stereostructure. Free energy analysis with side-chain dihedral angles can reveal whether rotation around the dihedral angle toward the reactive stereostructure is restricted by steric hindrance, and thus decomposition is negated. In silico methods can be particularly useful for risk assessment when experimental data are unavailable. However, the ability to identify and approach the reactive stereostructure is not the absolute determinant of whether a cleavage reaction will occur, even if it is important in determining whether a cleavage reaction is likely to occur. In other words, the side chain may be able to approach the reactive stereostructure, but it may not react because the cleavage reaction is energetically unfavorable.

[0026] Therefore, one or more regions (corresponding to reaction probabilities) may be defined via steric effects and / or energy profile ranges to predict the decomposition reaction in combination with the constitutive structures described above, by exhibiting energy barrier properties (e.g., the presence of steric hindrance sufficient to provide a very high free energy barrier to approach the reactive conformation). It is understood that these regions may be defined separately to represent steric hindrance and / or free energy constraints. Thus, molecular dynamics simulations can be performed to predict the probability that a polypeptide will transition to a reactive conformation and undergo a cleavage reaction. This prediction may include identifying spatial features that make the polypeptide susceptible to a particular cleavage reaction, and using molecular dynamics simulations with defined steric hindrance and / or free energy to predict the probability that the polypeptide will transition to a conformation with these spatial features.

[0027] IC solvent accessibility and reaction constraints Even if the interatomic distance criterion is met (e.g., based on an assessment of the dihedral angles of the backbone and side chains) and the free energy criterion is met (e.g., based on an assessment of the free energy profiles of the dihedral angles of the backbone and side chains), chemical decomposition will not occur without a solvent. Therefore, an additional constraint on chemical decomposition may be that water molecules (or other solvents) are easily accessible for hydrolysis. This constraint can be enforced by tracking the amount of water molecules throughout the simulation. Thus, in addition to tracking the positions of individual atoms in the polypeptide, the simulation can track the positions of numerous solvent molecules (and, for example, potentially, each atom of each of the numerous solvent molecules). At each time step, it can be determined whether the solvent molecules are within a given distance from a particular site in the polypeptide (e.g., the amide site in the backbone of the polypeptide molecule). Some conformations may prevent solvent molecules from approaching a particular polypeptide site as a result of (e.g.) folding within the polypeptide.

[0028] ID environment constraints Therapeutic agents, such as mAbs and antibody fragments, are prone to chemical and physical instability, which can limit their usefulness. This can be a major concern if residues in the complementarity-determining region (CDR) are unstable, as chemical changes in these sites are thought to affect titer. The development of effective disease treatments using protein-based therapeutics requires that the therapeutic agents exhibit sufficient stability both in formulation and under physiological conditions for usefulness. While progress has been made in in silico testing, thermal stress loading is applied in vitro in more cases to rank candidate molecules for further development. Since mAbs generally have a basic isoelectric point and deamidation rates increase with pH, ​​antibodies are usually formulated and tested under slightly acidic conditions (pH 5-6) for stability, as this increases solubility and slows degradation.

[0029] This method can be useful for selecting candidates with good shelf life under typical formulation conditions, but it may miss molecules with insufficient stability under physiological ionic strength and pH (approximately 7.4) conditions. Therefore, additional chemical decomposition constraints may require the existence of a specific pH or pH range. Constraints can be implemented by tracking the pH of the reaction throughout the simulation. Thus, the simulation can track pH conditions in addition to tracking the positions of individual atoms in the polypeptide. At each time step, it can be determined whether the pH is within a given range. Some conformations (reactive or non-reactive) may be somewhat dominant as a result of the pH of the environment in which the reaction is taking place (for example). It is understood that other types of environmental factors are intended to infer other variables that affect conformation. For example, temperature constraints can be used in combination with other factors, such as pH and spatial properties, to predict the probability that a polypeptide is likely to transition to a reactive conformation and undergo a cleavage reaction.

[0030] Using IF simulations and constraints By detecting the predictive degradation of polypeptides, it may be possible to exclude those polypeptides from other polypeptides with similar therapeutic effects but without such degradation handicaps, or the polypeptides may be linked to techniques that mitigate the undesirable effects of degradation. One method for predicting whether a given molecule will degrade is to perform simulations. However, chemical degradation can involve subatomic interactions, covalent bond formation, and covalent bond breaking. It is not possible to simulate these types of events using conventional molecular dynamics. Some techniques predict reaction probabilities based on the presence of amino acid motifs in the molecule. While reaction probabilities can vary dramatically depending on the motif, the influence of the motif may depend on its position within the molecule (e.g., whether the motif is in the heavy or light chain, and its position within the chain). Even for motifs considered extremely stable, experimental data confirm some cases where reactions occur with motifs despite their overall relative stability.

[0031] To address these limitations and problems, the techniques described herein perform molecular dynamics simulations and molecular structure techniques to generate reaction probabilities. One or more iterations in a molecular dynamics simulation can simulate how the conformation of a polypeptide changes over time. Reaction probabilities can be generated for one or more conformations based on their spatial properties (for example, this can determine whether various reaction constraints are met). For example, for each conformation generated by a molecular dynamics simulation, the spatial properties of the polypeptide in the conformation can be used to determine whether the interatomic distance reaction constraints and energy profile constraints are met, which can indicate that the polypeptide having the conformation is ready to participate in a cleavage reaction. Using solvent and environment-comprehensive modeling, the proportion of polypeptide molecules configured favorably for the reaction, approaching and reacting with solvent molecules, can be estimated. Based on the proportion of simulation-generated polypeptide conformations in which each constraint is met, an output can be generated indicating whether, to what extent, and / or at what rate a given polypeptide chemically decomposes to a specific product of interest. Therefore, the simulation-based techniques disclosed herein can generate predicted reaction susceptibility based on molecular dynamics and analysis of the three-dimensional structure of various polypeptide stereostructures (rather than data independent of the stereostructure, corresponding to, for example, the properties of amino groups in polypeptides).

[0032] One exemplary embodiment of the present disclosure relates to a computer implementation method for a polypeptide stereostructure comprising amino acids having side chains and a backbone, comprising: determining the dihedral angles of the amino acid backbone and the side chains when the polypeptide stereostructure is in a polypeptide stereostructure; determining the nucleophilic attack distance between two atoms, functional groups, or combinations thereof of the amino acid when the polypeptide stereostructure is in a polypeptide stereostructure, based on the dihedral angles of the backbone and the side chains, such that one of the two atoms or functional groups is in the side chain of the amino acid and the other of the two atoms or functional group is in the backbone of the amino acid; determining, based on the nucleophilic attack distance of the amino acid when the polypeptide stereostructure is in a polypeptide stereostructure, that the polypeptide stereostructure is a reactive stereostructure that is sensitive to cleavage reactions; determining the free energies of the dihedral angles of the amino acid backbone and the side chains when the polypeptide stereostructure is in a reactive stereostructure, in accordance with the determination that the polypeptide stereostructure is a reactive stereostructure; and predicting the probability that the amino acid side chains will be trapped in the reactive stereostructure based on the free energies of the dihedral angles of the amino acid backbone and the side chains. In some cases, the method further includes predicting the probability that a polypeptide will undergo chemical degradation as a result of amino acid side chains being trapped in a reactive three-dimensional structure.

[0033] Another exemplary embodiment of the present disclosure is a computer implementation method comprising: generating a representation of a polypeptide comprising amino acids having side chains and a backbone; performing molecular dynamics simulations using the representation, wherein the results of the molecular dynamics simulations include a set of polypeptide stereostructures of the polypeptide; determining the spatial properties of the amino acids in each polypeptide stereostructure of the set of polypeptide stereostructures, wherein the spatial properties include dihedral angles of the backbone and dihedral angles of the side chains; and determining two amino acids of each polypeptide stereostructure based on the combination of dihedral angles of the backbone and dihedral angles of the side chains. The present invention relates to a computer-implemented method comprising: estimating the nucleophilic attack distance between atoms, functional groups, or combinations thereof, such that one of the two atoms or functional groups is in the side chain of an amino acid and the other is in the backbone of an amino acid; determining at least one reactive stereostructure that is susceptible to cleavage based on the nucleophilic attack distance of the amino acids in each polypeptide stereostructure; determining the free energies of the dihedral angles of the amino acid backbone and the side chain of the at least one reactive stereostructure; and predicting the probability that the amino acid side chain is trapped in at least one reactive stereostructure based on the free energies of the dihedral angles of the amino acid backbone and the side chain. In some cases, the method further includes predicting the probability that the polypeptide undergoes chemical decomposition as a result of the amino acid side chain being trapped in at least one reactive stereostructure.

[0034] II. Definition The term “polypeptide,” as used herein, refers to a polymer of amino acids of any length and may include proteins, DNA, and / or RNA. Polymers may include proteins containing any protein-like features, e.g., substituted amino acids (non-natural amino acids), alternative glycations, proteins, DNA complexes, and / or viral surface coat proteins. Polymers may be linear or branched, may contain modified amino acids, or may be interrupted by non-amino acids. The term also encompasses amino acid polymers that are naturally modified or modified by intervention, e.g., disulfide bond formation, glycosylation, lipidation, acetylation, phosphorylation, or any other operation or modification, e.g., conjugation with a labeling component. Furthermore, within this definition are polypeptides containing, for example, one or more analogues of amino acids (including, e.g., non-natural amino acids), and other modifications known in the art.

[0035] The term “chemical decomposition,” as used herein, refers to the process by which a molecule (e.g., a polypeptide molecule) is broken down into two or more fragments. In the context of polymers, chemical decomposition may include the complete depolymerization of a polymer into its corresponding monomer, or partial depolymerization (e.g., into one or more oligomers and potentially one or more other chemicals). Chemical decomposition may include certain types of chemical processes, such as tryptophan oxidation, methionine oxidation, ASN-PRO clipping, asparagine deamidation, or aspartic acid isomerization.

[0036] The term "nucleophilic substitution or attack," as used herein, refers to a fundamental class of reactions in which an electron-rich nucleophile selectively bonds to or attacks the positive or partially positive charge of an atom or a group of an atom (e.g., a functional group) in order to replace a leaving group. The positively or partially positively charged atom is referred to as the electrophile.

[0037] As used herein, the term "nucleophilic attack distance" refers to the average (or average, median, or similar) distance (e.g., in angstroms) between an electron-rich nucleophile and an atom or group of an atom having a positive or partial positive charge (e.g., a functional group).

[0038] When used herein, the term "multiple stereostructures" means that a given polypeptide may have any number of atomic spatial arrangements, some of which are reactive stereostructures and some are non-reactive stereostructures.

[0039] When used herein, the term “reactive stereostructure” refers to the stereostructure of an amino acid or polypeptide that is susceptible to or sensitive to nucleophilic substitution or attack.

[0040] When used herein, the term "non-reactive stereostructure" refers to a stereostructure of an amino acid or polypeptide in which the amino acid or polypeptide is less susceptible to or insensitive to nucleophilic substitution or attack, for example, due to steric hindrance.

[0041] As used herein, when an action is “based on” something, this means that the action is based at least somewhat, or at least partially, on something.

[0042] The terms “substantially,” “approximately,” and “about,” as used herein, are defined to the extent that they represent the majority of what is specified, but not necessarily the entirety (including the entirety of what is specified), as understood by those skilled in the art. In any disclosed embodiment, the terms “substantially,” “approximately,” or “about” may be replaced by “within the range [percentage]” of what is specified, such that the percentages include 0.1, 1, 5, and 10 percent.

[0043] III. Exemplary reaction generation dependence Under physiological conditions, asparagine (Asn) residues are susceptible to deamidation, which hydrolyzes the amide side chain to form a free carboxylic acid. The rate-limiting step in this reaction is the formation of a five-membered succinimide ring intermediate. However, peptides containing Asn followed by bulky residues, such as proline (Pro), have been shown to be susceptible to hydrolytic cleavage of the peptide backbone between the two residues. Mass spectral data of the Asn-Pro site identified in the complementarity-determining region 3 of the light chain CDR-L3 of at least one mAb have led to the identification of three peptides associated with this site: a native trypsin peptide containing the Asn-Pro site, an N-terminal hydrolysis product peptide containing Asn and isoAsn, and a C-terminal hydrolysis product peptide. The identification of the N-terminal hydrolysis product peptide containing Asn and isoAsn, rather than Asp and isoAsp, suggests that the formation of the succinimide intermediate is a result of the attack of the peptide bond carbonyl on the side chain amide nitrogen. This COOH-terminal succinimide intermediate can then be opened to form the observed N-terminal hydrolysis product containing Asn or isoAsn.

[0044] Figure 1 illustrates this exemplary cleavage reaction, which can produce chemical decomposition products. More specifically, Figure 1 depicts a representation of an Asn residue containing a side chain with a backbone. A polypeptide molecule is susceptible to nucleophilic attack on the amide nitrogen of the side chain of the peptide bond carbonyl if the side chain nitrogen atoms and γ carbon of the backbone are sufficiently close together and the free energy profiles of the dihedral angles of the backbone and side chain are favorable for transition to the reactive stereostructure. A metastable COOH-terminal succinimide (cyclic imide) intermediate can be produced as a result of the nucleophilic attack. This succinimide intermediate can then be opened, and if the solvent is accessible to the succinimide intermediate, the succinimide can be hydrolyzed to a mixture of asparagine and isoasparagine linkages. With respect to the asparagine residue, the polypeptide can maintain its target properties. However, with respect to the isoasparagine residue, the stereostructure of the protein and its electrostatic properties can change compared to the original polypeptide. If simulations can reliably predict the possibility of polypeptides undergoing chemical degradation to produce undesirable products, polypeptides and / or formulations may be selected to minimize undesirable chemical degradation and maintain the active polypeptide with the targeted functionality.

[0045] IV. Implementation of interatomic distance and energy profile constraints As described herein, the degradation pathway of Asn-Pro hydrolysis proceeds via nucleophilic attack of the carbonyl skeleton to the Asn side chain nitrogen. A requirement for this process is the reduction of the nucleophilic attack distance (dN) between the Asn side chain nitrogen and the side chain carbonyl so that the Asn side chain can conform to the reactive stereostructure. N The skeletal dihedral angle Ψ(N n -C α -CN n+1 (composed of atoms) and side chain dihedral angle 1X1(CC) α -C β -C γ It is composed of atoms. This is NC α -C β -C γNote that this differs from the conventional chi-1, which is typically referred to as the side-chain dihedral angle composed of and X2(C α -C β -C γ It is primarily characterized by combinations of (composed of O atoms). As shown in Figures 2A and 2B (Figure 2A shows the ASN-PRO peptide in the protein structure, and Figure 2B shows the ASN-PRO peptide in the protein structure, with the proline ring visualized), at the Asn residue, the distance d N The dihedral angle can be minimized in two specific combinations of dihedral angles: firstly, when the skeletal dihedral angle is expanded with Ψ > 120, and the side-chain dihedral angles are approximately -60° and -90° (stereostructure A in Figures 2A and 2B); and secondly, when the skeletal dihedral angle is a small angle with Ψ < -60, and the angles are approximately 60° and 90° (stereostructure B in Figures 2A and 2B). More specifically, whether a given atom attacks or reacts with another atom depends on their proximity to each other (nucleophilic attack distance (dN)). In some cases, the positions of atoms and their angles to each other can be tracked through simulation, thus the distance d N The dihedral angle can also be tracked. In other cases, the dihedral angle can be tracked through simulation, and this can be used to infer or estimate whether atoms are close enough to react. Figures 2A and 2B show the three dihedral angles Ψ, Χ1 and Χ2, and the distance d between the side-chain nitrogen atom and the γ carbon of the skeleton. N This shows how it will affect [something].

[0046] In some cases, molecular dynamics simulations can be performed using representations of polypeptides having one or more side chains. The results of the molecular dynamics simulations may include a set of polypeptide conformations, where each polypeptide conformation determines the position of each atom in the polypeptide. For each polypeptide conformation in the set, one or more spatial properties of the polypeptide's amino acids (e.g., Asn) in the polypeptide conformation are determined. One or more spatial properties may include a number of interatomic distances, a number of angles, and / or a number of dihedral angles. In certain cases, the spatial properties include the dihedral angles of the amino acid backbone and two dihedral angles of the side chains. The nucleophilic attack distance dN can be calculated or estimated from two atoms, functional groups, or combinations thereof of the amino acids in each polypeptide conformation, based on the spatial properties (e.g., a combination of the dihedral angles of the backbone and two dihedral angles of the side chains). Once the nucleophilic attack distances of the amino acids in each polypeptide conformation are calculated or estimated, at least one reactive conformation that is susceptible to or prone to nucleophilic substitution or attack can be identified for the polypeptide.

[0047] Since dihedral angles and nucleophilic attack distances dN are used to identify at least one reactive stereostructure, a distance criterion can be determined corresponding to a threshold nucleophilic attack distance for a given combination of spatial property values ​​(e.g., dihedral angles Ψ, X1, and X2). For example, a default distance threshold (e.g., minimum nucleophilic attack distance) may be defined as 1.0 Å to 4.0 Å (e.g., 2.5 Å) to determine whether the distance criterion is met. The distance criterion may be met if the spatial property (e.g., dihedral angles Ψ, X1, and X2) results in a nucleophilic attack distance dN that is less than or equal to the default distance threshold. If the distance criterion is met, a stereostructure that satisfies the default distance threshold is identified as a reactive stereostructure. The distance criterion is not met if the spatial property (e.g., dihedral angles Ψ, X1, and X2) results in a nucleophilic attack distance dN that exceeds the default distance threshold. If the distance criterion is not met, the stereostructure is identified as a non-reactive stereostructure. It is understood that satisfying distance criteria to confirm at least one reactive stereostructure can be determined using alternative techniques, such as comparing the values ​​of spatial properties (e.g., dihedral angles Ψ, X1 and X2) with an independent range or threshold indicating that an atom or functional group in the polypeptide backbone chain is within a predetermined distance from an atom or functional group in the side chain.

[0048] In some cases, the free energy of side chains along dihedral angles can be calculated to gain mechanistic insight into the role of side chain structure in the rate of hydrolysis. Free energy values ​​can be generated via molecular dynamics models. Figures 3A–3F show free energy profiles (calculated from bin ensembles) of the skeletal dihedral angle (Ψ) and side chain dihedral angles (X1 and X2) obtained by molecular dynamics model simulations. Structures with low free energy values ​​(represented by dark colors) are more stable than structures with high free energy values ​​(represented by light dark colors) because the molecule is strongly believed to be in the structure. Low free energies correspond to a large ensemble (the number of frames in the simulation if the structure is a reactive structure), and therefore have a high probability of finding side chains in a given combination of dihedral angles of the reactive structure. Circular shapes indicate specific dihedral angle ranges that geometrically position the side chain nitrogen and γ carbon of the skeleton within the minimum nucleophilic attack distance (e.g., 2 or 3 angstroms). If the circled area does not contain stereostructures associated with low free energy values, the output indicates that the polypeptide molecule's stereostructure does not allow the chain nitrogen atoms to approach the γ carbon of the skeleton sufficiently for the reaction, and therefore the polypeptide is unlikely to undergo chemical decomposition.

[0049] Each free energy profile in Figures 3A–3F corresponds to a simulation using a specific polypeptide structure. Notably, the free energy profiles in Figures 3A–3D suggest that the corresponding polypeptides (Fab1, Fab2, Mab3, and Mab4) have a stereochemistry where the distance between the side-chain nitrogen and the γ-carbon of the backbone is the minimum nucleophilic attack distance. On the other hand, the polypeptides corresponding to the free energy profiles in Figures 3E and 3F (Mab5 and Fab6) do not appear to have atoms in a stereochemistry where they are this close. More specifically, the free energy profiles along Ψ show that Fab2 primarily adopts a small backbone dihedral angle (Ψ<-60), while the other structures Fab1, Fab6, Mab3, Mab4, and Mab5 adopt an expanded stereochemistry (Ψ>120). Thus, the reactive stereochemistry corresponds to stereochemistry B of Fab2 and to configuration A of structures Fab1, Mab3, Mab4, Mab5, and Fab6. The free energy of stereostructure A is very low in Fab1, Mab3, and Mab4 (0.94, 0.96, and 1.06 kcal / mol, respectively), which is consistent with the high experimental hydrolysis rates observed in these structures (13, 15, and 15% / week, respectively). In contrast, the free energy of stereostructure A is relatively high in Mab5 (1.56 kcal / mol) and highest in Fab6 (2.66 kcal / mol), which is well consistent with the low experimental rates observed in these molecules (5 and 0% / week, respectively). However, the agreement between the free energy of the reactive stereostructure and the experimental rate is not sufficient in Fab2, where the free energy of the reactive stereostructure is very low (0.75 kcal / mol), but the experimental rate is not as high (9% / week). Even if the side chain adopts a reactive stereostructure, it is presumed that the chemical reaction is energetically unfavorable in this case. Therefore, in some cases, it can be beneficial to include reaction energy as a constraint within molecular dynamics model simulations.

[0050] In some cases, the probability that the amino acid side chains of a polypeptide will be trapped in a reactive stereostructure is predicted based on the free energy profiles of one or more dihedral angles of the amino acid backbone and / or side chains. If the three-dimensional structure of the polypeptide restricts the side chains to a non-reactive stereostructure, it is energetically unfavorable for the side chains to approach the reactive stereostructure. Therefore, the presence of steric hindrance can result in a very high free energy barrier to approaching the reactive stereostructure. Alternatively, if the three-dimensional structure of the polypeptide restricts the side chains in a given combination of dihedral angles of the reactive stereostructure, the side chains can be trapped in this reactive stereostructure. Therefore, free energy analysis involving dihedral angles, and nucleophilic attack distance analysis, can reveal whether rotation around dihedral angles toward the reactive stereostructure is restricted by steric hindrance, and thus whether decomposition is negated.

[0051] To use free energy analysis with dihedral angles and nucleophilic attack distances, a free energy criterion can be determined for a combination of spatial property values ​​(e.g., dihedral angles Ψ, X1, and X2) that corresponds to a reactive stereostructure related to a predetermined distance threshold (e.g., minimum nucleophilic attack distance). For example, a first predetermined energy threshold (minimum free energy value) may be defined as 1.0 kcal / mol to 2.0 kcal / mol (e.g., 1.5 kcal / mol), which corresponds to a first stereostructure having a skeletal dihedral angle extended to Ψ > 120°, as well as side-chain dihedral angles of X1 approximately -60° and X2 approximately -90°. Separately, a second default energy threshold (minimum free energy value) may be defined as 1.0 kcal / mol to 2.5 kcal / mol (e.g., 2.0 kcal / mol), which corresponds to a second stereochemistry having a small skeletal dihedral angle Ψ < -60°, as well as side-chain dihedral angles X1 approximately 60° and X2 approximately 90°. Alternatively, a comprehensive default energy threshold (minimum free energy value) may be defined as 1.0 kcal / mol to 2.5 kcal / mol (e.g., 2.0 kcal / mol), which corresponds to all stereochemistry having dihedral angles in the amino acid skeleton and two dihedral angles in the side chains, minimizing the nucleophilic attack distance. In some cases, the comprehensive default threshold (minimum free energy value) may be defined as 1.0 kcal / mol to 2.5 kcal / mol (e.g., 2.0 kcal / mol), which corresponds to all stereostructures having a skeletal dihedral angle Ψ of 120° to -60°, and side-chain dihedral angles X1 of -60° to +60° and X2 of -90° to +90°. The free energy criterion is satisfied if the spatial properties (e.g., dihedral angles Ψ, X1 and X2) have free energy less than or equal to the minimum free energy value (e.g., comprehensive default energy threshold). The free energy criterion is not satisfied if the spatial properties (e.g., dihedral angles Ψ, X1 and X2) have free energy greater than the minimum free energy value (e.g., comprehensive default energy threshold).It is understood that satisfying the free energy criterion for confirming at least one reactive stereostructure can be determined using alternative techniques, for example, by comparing the values ​​of spatial properties (e.g., dihedral angles Ψ, X1 and X2) with an independent range or threshold indicating that the free energy of the amino acid backbone and / or side chains is within a predefined free energy value.

[0052] The probability that the amino acid side chains of a polypeptide are trapped in a reactive stereostructure can be defined as a numerical probability, a classification probability (e.g., very low, low, moderate, high), or a binomial probability, based on the free energy profile of one or more dihedral angles of the amino acid backbone and / or side chains. For example, if the reactive stereostructure confirmed by the distance criterion is associated with a low free energy value determined by the free energy criterion, the output may indicate that the amino acid side chains of the polypeptide are trapped in the reactive stereostructure (the side chains can approach the reactive stereostructure). Alternatively, if the reactive stereostructure confirmed by the distance criterion is associated with a high free energy value determined by the free energy criterion, the output may indicate that the energy barrier is too high and that the polypeptide is strongly likely to maintain a non-reactive stereostructure.

[0053] Once the probability of the polypeptide's amino acid side chains being trapped in a reactive stereostructure is determined, it may be possible to predict the probability of the polypeptide chemically degrading as a result of the polypeptide molecule's reactive stereostructure bringing the side-chain nitrogen atoms close enough to the gamma carbon of the skeleton for the reaction. The probability of the polypeptide chemically degrading (e.g., undergoing nucleophilic attack and hydrolysis) can be defined as a numerical probability, a classification probability (e.g., very low, low, moderate, high), or a binomial probability, based on the probability of the polypeptide's amino acid side chains being trapped in a reactive stereostructure. For example, if the polypeptide's amino acid side chains are thought to be trapped in a reactive stereostructure, the output may indicate that the polypeptide is thought to chemically degrading as a result of the polypeptide molecule's reactive stereostructure bringing the side-chain nitrogen atoms close enough to the gamma carbon of the skeleton for the reaction. Alternatively, if the energy barrier is too high and the polypeptide is strongly thought to maintain a non-reactive stereostructure, the output may indicate that the polypeptide is not thought to chemically degrading, or that degradation is negated. However, even if a polypeptide is in a reactive three-dimensional structure, it will not decompose unless other factors (e.g., access to a solvent) are present in addition to the probability of the reactive three-dimensional structure. Therefore, in some cases, additional constraints may be included as factors predicting the behavior of the three-dimensional structure and the probability of the polypeptide chemically decomposing.

[0054] V. Execution of environmental and accessibility constraints The role of environmental factors, such as pH, temperature, and solvent accessibility, in the behavior of polypeptide conformation and / or chemical decomposition can be investigated by molecular dynamics simulations. Environmental factors, e.g., pH and temperature, can be assumed to be constant within the simulation. In some cases, pH can be assumed by calculating the relevant pKa values ​​of all constituent molecules and assigning a dominant protonation state at a given pH. While a conventional molecular dynamics protocol in which the protonation state is fixed during the simulation is used here, alternative methods that allow for various protonation states in the simulation, such as constant pH molecular dynamics, can be used. Alternatively, H3O+ and OH- ions can be added to adjust the pH value using quantum mechanics / molecular mechanics (QM / MM) methods. Even if nucleophilic attack occurs, polypeptides will not decompose unless the environment has the correct conditions and solvent molecules are accessible. Molecular dynamics simulations can be further configured to simulate polypeptides in solvents (e.g., cations or anions). Solvent-blocking metric can be defined by subtracting the number of frames on which the amide group bonds to solvent molecules (e.g., water molecules) from the number of frames on which the amide group bonds to non-solvent groups. Therefore, negative metric corresponds to greater solvent accessibility compared to positive metric. Positive metric can demonstrate that the geometry of the polypeptide blocks solvent molecules from reaching the amide group.

[0055] VI. Process for predicting the reaction type of polypeptides Figure 4 illustrates process 400, which generates the probability of a cleavage reaction based on molecular dynamics simulations and evaluation of the spatial properties of the molecule. Process 400 begins with block 405, which generates a representation of a polypeptide containing amino acids with side chains and a backbone. The representation may include confirmation of the atoms, mass, charge, and interatomic connections of the polypeptide (and potentially the solvent). The representation may further include the starting coordination of each atom of the polypeptide (and potentially the solvent). The representation may further include constraints that are computationally applied throughout the simulation, such as angles or dihedral angles, van der Wahl conditions, free energy, and pH.

[0056] In Block 410, a representation is used to perform one or more molecular dynamics simulations to generate a set of polypeptide stereostructures. Each polypeptide stereostructure in the set can correspond to a time step in the simulation. Each polypeptide stereostructure in the set can include the atomic positions of each atom in the polypeptide. The set of polypeptide stereostructures can be determined by calculating the forces from the particle positions and numerically solving the equations of motion. At each time step, in addition to determining the position of each atom, the momentum of each atom can be further estimated.

[0057] In block 415, for each polypeptide stereostructure in the set of polypeptide stereostructures, one or more spatial properties of an amino acid in the polypeptide stereostructure are determined. One or more spatial properties may include angles and / or dihedral angles (e.g., Ψ, Χ1, and Χ2 of the backbone and / or side chain dihedral angles of amino acids adjacent to the sensitive site). In some cases, the spatial properties include the backbone dihedral angle (Ψ) and the side chain dihedral angle (Χ1 or Χ2). In other cases, the spatial properties include the backbone dihedral angle (Ψ), the first dihedral angle of the side chain (Χ1), and the second dihedral angle of the side chain (Χ2).

[0058] In Block 420, the nucleophilic attack distance between two atoms, functional groups, or combinations thereof of amino acids in each polypeptide stereostructure is estimated based on one or more spatial properties of the amino acids in the polypeptide stereostructure. For example, the nucleophilic attack distance between two atoms, functional groups, or combinations thereof can be estimated using the positions of each atom or functional group in the dihedral angles of each polypeptide stereostructure, the momentum of each atom between the atoms or functional groups, and combinations thereof. In some cases, the nucleophilic attack distance between two atoms, functional groups, or combinations thereof of amino acids in each polypeptide stereostructure is estimated based on angles, dihedral angles, or combinations of dihedral angles (e.g., dihedral angles of the backbone and dihedral angles of the side chains). In some cases, one of the two atoms or functional groups is in the side chain of the amino acid, and the other is in the backbone of the amino acid.

[0059] In block 425, at least one reactive stereostructure sensitive to cleavage is identified based on the nucleophilic attack distance of an amino acid in each polypeptide stereostructure. In some cases, a distance criterion is determined that can be used to identify at least one reactive stereostructure. The distance criterion may correspond to the minimum nucleophilic attack distance in a combination of values ​​for the spatial properties (e.g., dihedral angles Ψ, X1, and X2) used to identify the reactive stereostructure. For example, a default distance threshold (minimum nucleophilic attack distance) may be set to 0.0 Å to 3.0 Å (e.g., approximately 1.5 Å) to determine whether the distance criterion is met. Determining whether the distance criterion is met for at least one reactive stereostructure may include comparing the nucleophilic attack distance of an amino acid in at least one reactive stereostructure with the default distance threshold. The distance criterion is met if the nucleophilic attack distance is less than or equal to the default distance threshold. If the distance criterion is met, the stereostructure that satisfies the distance criterion is identified as a reactive stereostructure. The distance criterion is not met if the nucleophilic attack distance exceeds the default distance threshold. If the distance criterion is not met, the three-dimensional structure that does not meet the distance criterion is identified as a non-reactive three-dimensional structure.

[0060] Block 430 determines the free energy of an angle, dihedral angle, or combination of dihedral angles (dihedral angles of the amino acid backbone and dihedral angles of the side chains) in at least one reactive stereostructure. The free energy can be determined by analyzing the free energy profile of the angle, dihedral angle, or combination of dihedral angles. In some cases, the free energy profile and landscape in space of the angle, dihedral angle, or combination of dihedral angles are calculated from a bin population using the following method. TIFF0007864109000001.tif13170In formula, κ B is the Boltzmann constant, T is the temperature, and N i This is a group of bin i, N max This represents the most aggregated group of bins. For bins without aggregates, an artificial barrier equivalent to aggregate 0.5 can be provided. At each time step, in addition to determining the position of each atom, the free energy can be further estimated. In certain cases, a QM / MM method can be used to form a model of the free energy.

[0061] In block 435, the probability that an amino acid side chain is trapped in at least one reactive stereostructure can be predicted based on the free energy of the angle, dihedral angle, or combination of dihedral angles (e.g., the dihedral angle of the amino acid backbone and the dihedral angle of the side chain) of the amino acid. In some cases, a free energy criterion is determined that can be used to predict the probability that an amino acid side chain is trapped in at least one reactive stereostructure. For example, a default energy threshold (minimum free energy value) may be defined as 0.0 kcal / mol to 1.0 kcal / mol, which corresponds to all reactive stereostructures. Determining whether the energy criterion is satisfied for at least one reactive stereostructure may include comparing the free energy of the angle, dihedral angle, or combination of dihedral angles (e.g., the dihedral angle of the amino acid backbone and the dihedral angle of the side chain) of at least one reactive stereostructure with the default energy threshold. The free energy criterion is satisfied if the spatial properties (e.g., dihedral angles Ψ, Χ1, and Χ2) have a free energy less than or equal to the default energy threshold. If the free energy criterion is met, it can be predicted that the amino acid side chains are trapped in the reactive stereostructure (the side chains can approach the reactive stereostructure). The free energy criterion is not met if the spatial properties (e.g., dihedral angles Ψ, X1, and X2) have free energies that exceed a predetermined energy threshold. If the free energy criterion is not met, it can be predicted that the amino acid side chains face an energy barrier that is too high, and the polypeptide maintains a non-reactive stereostructure.

[0062] In an optional block 440, environmental and accessibility constraints can be determined, indicating that the amide group of the polypeptide has spatial accessibility exceeding a threshold for binding to solvent molecules of the surrounding solvent, if satisfied. The environmental and accessibility constraints are satisfied based on an evaluation of one or more spatial properties of the polypeptide in at least one reactive stereostructure, one or more environmental factors (e.g., pH or temperature), availability of solvent molecules, or a combination thereof.

[0063] In block 445, the probability of a polypeptide undergoing chemical degradation can be predicted as a result of the amino acid side chains being trapped in at least one reactive stereostructure. For example, if the amino acid side chains of a polypeptide appear to be trapped in a reactive stereostructure, it can be predicted that the polypeptide will undergo chemical degradation as a result of the reactive stereostructure. Alternatively, if the energy barrier is too high and the polypeptide is strongly expected to maintain a non-reactive stereostructure, it can be predicted that the polypeptide will not undergo chemical degradation. In certain specific cases, the probability of a polypeptide undergoing chemical degradation can be predicted as a result of the reactive stereostructure as well as environmental and accessibility constraints.

[0064] Block 450 outputs the probability that an amino acid side chain is captured by at least one reactive structure and / or that the polypeptide undergoes chemical degradation. For example, the probability that an amino acid side chain is captured by at least one reactive structure and / or that the polypeptide undergoes chemical degradation can be displayed or transferred to another device. In some cases, the probability that an amino acid side chain is captured by at least one reactive structure and / or that the polypeptide undergoes chemical degradation can be used to select a polypeptide to be used in a particular way (e.g., to develop a treatment for a particular condition) and / or to select a particular formulation of the polypeptide (e.g., to restrict water from approaching the polypeptide).

[0065] In some cases, the probability that an amino acid side chain is trapped in at least one reactive stereostructure and / or the polypeptide is chemically degraded is used to remove polypeptides from the list of potential polypeptides to be used as at least part of a therapeutic agent based on the probability that an amino acid side chain is trapped in at least one reactive stereostructure and / or the polypeptide is chemically degraded.

[0066] In some cases, the probability that an amino acid side chain is captured in at least one reactive structure and / or that the polypeptide is chemically degraded is used to rank a polypeptide lower than another polypeptide in a list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that an amino acid side chain of another polypeptide is captured in at least one reactive structure and / or that the polypeptide is chemically degraded.

[0067] VII. Examples of Computational Environments Figure 5 illustrates an example of a computing device 500 suitable for use of the system and method for molecular dynamics simulations according to the present disclosure. The example of computing device 500 includes a processor 505 that communicates with memory 510, and other components of computing device 500 that use one or more communication buses 515. The processor 505 is configured to execute processor-executable instructions stored in memory 510 to perform one or more methods of molecular dynamics simulation according to different examples, for example, some or all of the method example 400 described herein in Figure 4. In this example, memory 510 stores processor-executable instructions that provide polypeptide data analysis 520 and predictive analysis 525 to one or more polypeptides of interest, as discussed above with respect to Figures 1, 2, 3A-3F and 4.

[0068] Polypeptide data analysis 520 and predictive analysis 525 may be configured to generate polypeptide representations 530 and use these as inputs to one or more molecular dynamics simulations to generate reaction probabilities. Molecular dynamics simulations can be performed using molecular simulation ensembles 535 that check the variables of a fixed system (e.g., two or more combinations of particle number (N), quantity (V), energy (E), temperature (T), and pressure (P)). For example, the ensemble may include a microcanonical ensemble (NVE), a canonical ensemble (NVT), or an isothermal-isobaric ensemble (NPT). Molecular dynamics simulations may use integrating circuits that control temperature or pressure throughout the simulation by integrating momentum equations and thermostats or barostats. One or more iterations in a molecular dynamics simulation can simulate how the three-dimensional structure of the polypeptide changes over time. The simulation may be performed for a specific number of time steps or until the target reaches equilibrium.

[0069] Reaction probabilities can be generated for each of numerous stereostructures based on their spatial properties (for example, this can determine whether various reaction constraints are met). For instance, for each stereostructure generated by molecular simulation ensemble 535, the spatial properties of the polypeptide in the stereostructure can be used to determine whether the interatomic distance reaction constraint 540 and the free energy constraint 545 are met, which can indicate that the polypeptide having the stereostructure is susceptible to cleavage reactions. Using environmental and accessibility constraints 550, polypeptide molecules configured to approach and react with solvent molecules in a reaction-friendly manner can be estimated. Based on the fraction of simulation-generated polypeptide stereostructures that satisfy each constraint, outputs can be generated indicating whether the side chains of the polypeptide can be captured in at least one reactive stereostructure, and / or the degree and / or rate of chemical decomposition of a given polypeptide. (It is understood that the simulation can generate multiple outputs having the same stereostructure or the same spatial properties, and this can be considered unique.)

[0070] In this example, the computing device 500 also includes one or more user input devices 555, such as a keyboard, mouse, touchscreen, microphone, etc., to accept user input. The computing device 500 also includes a display 560 to provide visual output to the user, such as a user interface. The computing device 500 also includes a communication interface 565. In some examples, the communication interface 540 can communicate using one or more networks, such as a local area network ("LAN"); a wide area network ("WAN"), such as the Internet; a metropolitan area network ("MAN"); a point-of-use or peer-to-peer connection, etc. Communication with other devices can be achieved using any suitable network protocol. For example, one suitable network protocol may be the Internet Protocol ("IP"), the Transmission Control Protocol ("TCP"), the User Datagram Protocol ("UDP"), or a combination thereof, such as TCP / IP or UDP / IP.

[0071] VIII. Examples Systems and methods implemented in various embodiments can be better understood by referring to the following examples.

[0072] VIII.A. Example 1. Hydrolysis of the -Asn-Propeptide bond in antibodies: The three-dimensional structure of CDR-L3 promotes the reaction. During screening of two antibody Fab fragments as potential treatments for ocular diseases, loss of antigen binding and simultaneous fragmentation were observed during incubation at neutral pH and 37°C. Peptide mapping indicated that fragmentation occurred at the Asn-Pro site within the complementarity-determining region 3 (CDR-L3) of the light chain. As previously observed with peptides, analysis by mass spectrometry indicated that the cleavage was a hydrolytic reaction brought about by the attack of the peptide carbonyl on the Asn side chain. A comparison of the rate of CDR-L3 Asn-Pro hydrolysis in five test antibodies showed that, in general, the rate of cleavage was faster in short, unstructured peptides, suggesting that the rate is determined by the conformational priority of the segments. In contrast to these results, Asn-Pro found in the complementarity-determining region 2 (CDR-H2) of the heavy chain and derived from germline genes was not sensitive to this cleavage reaction. Molecular dynamics simulations show that the Asn residues at the sensitive site assemble at a dihedral angle consistent with the attack on the peptide carbonyl side chain, while the Asn residues at the restricting site do not assemble. These findings lead to the various embodiments discussed herein, including antibody manipulation techniques to circumvent this instability and in silico tools for risk assessment of cleavage reactions.

[0073] VIII.B. Materials and Methods Size exclusion chromatography (SEC) was performed using an Agilent 1200 series HPLC system (Santa Clara, CA) equipped with a diode array detector (DAD). G6.31 was separated using a TSK-GEL G2000SWxl (7.8 × 300 mm) column (Tosoh Bioscience, South San Francisco, CA). The Fab2 sample was diluted to approximately 0.5 mg / mL with the mobile phase (0.2 M potassium phosphate, 0.25 M potassium chloride, pH 6.2). 70 μL of the sample was injected into the column and eluted over 30 minutes at a flow rate of 0.5 mL / min in isocratic mode at 25°C, with UV absorption at 280 nm used for detection. The SEC peak was separated into monomers, high molecular weight species (HMWS), and fragments. Percentile peak area was calculated by dividing the peak area of ​​each group at each time point by the total peak area.

[0074] Ion exchange chromatography (IEC) was performed using an Agilent 1200 series HPLC system with a Dionex Propac WCX-10 column (4 × 250) (Tosoh Bioscience, South San Francisco, CA). Mobile phases A (20 mM MES, pH 5.7) and B (mobile phase A with 200 mM sodium chloride) were used for separation. Fab2 charge variants were separated using a linear gradient from 92% solvent A to 34% solvent A over 85 minutes, followed by a gradient from 34% solvent A to 0% solvent A over 95 minutes, for a total run time of approximately 100 minutes. 75 μL of sample was injected into the column and eluted at a flow rate of 0.8 mL / min in isocratic mode at 25°C, with UV absorption at 280 nm used for detection. The IEC peaks were separated into primary, acidic, and basic peaks. Percentile peak areas were calculated by dividing the area of ​​each peak by the total peak area.

[0075] The antigen-binding ability of Fab2 was measured using surface plasmon resonance (SPR) with a Biacore T200 instrument (GE Healthcare, Pittsburgh, PA) using a protocol similar to the described protocol (Tesar et al., 2017, mAbs). Briefly, the antigen was immobilized directly onto a carboxymethylated dextran sensor tip (CM5) in the range of 2000–3000 reaction units (RU) using an amine coupling kit (GE Healthcare, Pittsburgh, PA). Binding of the antibody Fab fragment to the antigen was determined by monitoring the change in RU before and after injection for 180 seconds. The sensor tip was regenerated with 10 mM glycine HCl buffer at pH 2.1 and a flow rate of 30 μL / min for 30 seconds. All binding assays were performed at room temperature in HEPES buffer (0.01 M HEPES, 0.15 M NaCl, 0.005% (v / v) surfactant P20, pH 7.4). Antigen binding concentrations were calculated from standard calibration curves (0.158–5 μg / mL) using a four-parameter fit. Antigen binding capacity at each time point was normalized to the antigen binding capacity at t0.

[0076] IV.C. Details of Molecular Dynamics Simulations The Fab structure was constructed from the sequence using model builder software (e.g., a modified version of MODELLER). The Fab structure was energetically minimized to remove steric collisions. The relaxed structure was then analyzed to determine the protonation state of the ionized residues at pH=7.4. The Fab structure was then solvated with TIP3P water in an octahedral solvent box at a distance of at least 10 Å from the edge of the box under periodic boundary conditions, and the solute structure was parameterized using the FF14SB force field. The system charge was neutralized with Na+ and Cl- counterions. Hydrogen mass redistribution was performed on the solute atoms to enable simulation with a 4fs time step.

[0077] An exemplary simulation protocol included the following steps: First, the structure was constrained to the solute to its initial structure at 10 (kcal / mol / Å). 2The conjugate gradient energy was minimized in 2000 steps using a harmonic restraining potential with a force constant of ). Then, the pressure was maintained at 1 atm and the thermostat temperature was increased to 300 K over 200 ps, ​​while the intensity was 10 (kcal / mol / Å). 2 The harmonic position constraint of ) was applied to the protein structure. Then the system was run at 500 ps for 1 (kcal / mol / Å). 2 The constraint force constants were balanced to γ ​​= 1 ps. All constraints were removed during the production phase. The simulation time step was 4 fs. A cutoff radius of 9 Å was used for range-bound interactions, and particle mesh electrostatics were used for long-range interactions. Production simulations were performed using NPT conditions. Langevin dynamics were used with γ = 1 ps. -1 The temperature was maintained at 300K with a given number of collisions. The production phase of the molecular dynamics simulation was performed for 500 ns. The SHAKE algorithm was used for dynamics, and all bonds involving hydrogen atoms were constrained. Snapshots of the molecular dynamics trajectories were saved every 10 ps for the analysis presented below.

[0078] Default values ​​were used for all other simulation parameters. GPU execution of the Amber2015 molecular dynamics simulation software package with the SPFP precision model was used for the exemplary molecular dynamics simulation. Three independent molecular dynamics simulations of 500 ns each were performed by repeating the molecular dynamics simulation protocol described above. The trajectories from the three simulations were then combined (totaling 1.5 μs) and used for analysis. The trajectories were analyzed using the AmberTools CPPTRAJ software. Free energy profiles and landscapes in dihedral space were calculated from the bin ensemble using the following method. TIFF0007864109000002.tif13170In formula, κ B is the Boltzmann constant, T is the temperature, and N i This is a group of bin i, N maxThis represents the most aggregated group of bottles. Bottles without a group were given an artificial barrier equivalent to a group of 0.5.

[0079] VIII.D. Results Purified Fab1 was stressed in PBS at 37°C for 4 weeks. The control and stressed samples were then subjected to trypsin digestion, followed by reverse-phase chromatography separation and mass spectrometry to identify potential degradation products. The hydrolyzable, unstable Asn-Pro moiety was identified in Fab1 CDR-L3 (see, e.g., Figures 6A-6C). Extracted ion chromatograms (XIC) are shown for the native trypsin peptide containing the Asn-Pro moiety (Figure 6A), N-terminal hydrolysis product peptide (Figure 6B), and C-terminal hydrolysis product peptide (Figure 6C). Two N-terminal hydrolysis products corresponding to C-terminal Asn and iso-Asn were observed, rather than C-terminal Asp and iso-Asp (Figure 6B). The observation of Asn at the C-terminus of the N-terminal hydrolysis product peptide can be seen in Figures 7A and 7B, which show the mass spectra corresponding to the two N-terminal hydrolysis products eluted at 98.0 and 98.8 minutes, respectively. The theoretical mass spectra of the N-terminal hydrolysis products corresponding to Asn or Asp at the C-terminus of the peptide are shown in Figures 7C and 7D, respectively. The tandem mass spectrum of the N-terminal hydrolysis product confirms the presence of Asn at the C-terminus of the peptide.

[0080] The long-term stability of Fab2 was evaluated in PBS (pH 7.4) at 37°C. Figure 8A summarizes the changes in Fab stability over a 36-week period. Aggregation measurements using SEC indicate that Fab remains monomeric throughout the stress period. However, measurements of side-chain chemilysis and main-chain fragmentation using IEC and CE-SDS, respectively, suggest that Fab undergoes slow, constant degradation. The decrease in the main peak fraction after 36 weeks, as measured by IEC and CE-SDS, is 32.7% and 36%, respectively. Consequently, antigen-binding ability, as measured using SPR, decreases by 27% over the same period. Ion-exchange chromatograms indicate that the decrease in the IEC main peak is likely due to an increase in acidic charge variants due to deamidation reactions. Gel electrophoresis performed under denaturing conditions clearly shows that the decrease in the CE-SDS main peak is due to main-chain fragmentation (Figure 8B). Mass spectrometry of the stressed sample confirmed that the fragmentation was due to Asn-Pro hydrolysis. However, SEC data indicate that under non-denaturing conditions, Fab remains intact despite fragmentation due to Asn-Pro hydrolysis. Addition of a broad-spectrum protease inhibitor cocktail (cOmplete®, Roche) to the incubation mixture did not affect the rate of Fab2 light chain fragmentation, which was consistent with hydrolysis caused by autolysis events rather than the presence of trace amounts of proteolytic enzyme contaminants. No cleavage of the unstructured 17-mer peptide containing the Asn-Pro sequence was detected when spiked to a neutral pH formulation of Fab2 and incubated at 37°C. This is also inconsistent with protease-catalyzed fragmentation.

[0081] Sequence comparison of the antibody collection identified several potential therapeutic candidates possessing the Asn-Pro motif at CDR-L3. In all of these antibodies, Asn-Pro was fixed at position 6-7, but the remaining positions of CDR-L3 varied. Of these, four were selected for analysis of the Asn-Pro hydrolysis rate under heat stress (37°C) of protein solutions formulated in PBS. All showed sensitivity of the Asn-Pro peptide bond to hydrolysis and exhibited the cleavage kinetics shown in Figure 9. Mab3 and Mab4 showed the highest hydrolysis rates (Table 1), surpassing those observed in Fab1, while the hydrolysis rate was slower in Fab2, and Mab5 had the slowest rate. Mab3 contained a sufficient amount of hydrolyzed Asn-Pro in the starting material, which is likely because the antibody had previously been stored in a neutral pH buffer. While this dataset is too small to describe the effect of adjacent sequences on hydrolysis rates, it is noteworthy that sequence differences at CDR-L3 result in a three-fold range of hydrolysis. In addition, Fab4 with the CDR sequence of Fab2 grafted onto a selective non-human framework exhibits Asn-Pro hydrolysis at a four-fold lower rate (2% / week) compared to Fa2 (8% / week). In contrast to these results, analysis of an antibody panel with Asn-Pro at CDR-H2 showed that this location is insensitive to hydrolysis. For example, hydrolysis was not detected at Asn-Pro at CDR-H2 of Fab3 under heat stress (40°C) of the protein formulated in PBS (Table 1). TIFF0007864109000003.tif127170

[0082] To investigate the effect of antibody structure on Asn-Pro hydrolysis rate, thermal stress was applied to unstructured linear peptides displaying the CDR-L3 sequences of Fab1 and Mab3. Since the hydrolysis rate of unstructured peptides incubated at 37°C has been previously shown to be slow, accelerated temperature conditions (90°C and 70°C) were performed, and the temperature dependence was extrapolated to calculate the rate at 37°C. Very slow peptide hydrolysis was observed, with half-lives exceeding 2000 days calculated at 37°C (Table 2). As expected, the hydrolysis rate was approximately three times slower in incubations performed with pH 5 buffer. TIFF0007864109000004.tif75170

[0083] To further investigate the effect of protein structure on the Asn-Pro hydrolysis rate, molecular dynamics simulations of the Fab structure were performed. As discussed herein, the degradation pathway of Asn-Pro hydrolysis proceeds via nucleophilic attack of the carbonyl skeleton to the Asn side-chain nitrogen. A requirement for this process is to minimize the nucleophilic attack distance (dN) between the Asn side-chain nitrogen and the side-chain carbonyl so that the ASN side-chain conforms to the stereochemistry (Figures 2A and 2B). N The dihedral angle is minimized in two specific combinations: the first is when the skeletal dihedral angle is expanded with Ψ > 120, and the side-chain dihedral angles are approximately X1 = -60° and X2 = -90° (i.e., stereostructure A in Figures 2A and 2B); and the second is when the skeletal dihedral angle is a small angle with Ψ < -60, and X1 = approximately 60° and X2 = approximately 90° (i.e., stereostructure B in Figures 2A and 2B). The free energy of the side chains along the dihedral angle was calculated to gain mechanistic insight into the role of the side-chain stereostructure in the rate of hydrolysis.

[0084] Figures 3A–3F show the free energy profiles of the skeletal dihedral angle (Ψ) and side-chain dihedral angles (Χ1 and Χ2) obtained by 1.5 microsecond MD simulations (calculated from a bin collection; see the section on details of molecular dynamics simulations). Low free energies correspond to a high probability of finding side chains in a given combination of dihedral angles. The free energy profile along Ψ indicates that Fab2 primarily adopts small skeletal dihedral angles (Ψ<-60), while the other structures adopt expanded stereostructures (Ψ>120). Thus, the reactive stereostructures correspond to stereostructure B in Fab2 and stereostructure A in the other molecules, as shown by the circles in Figures 3A–3F. The free energy of stereostructure A is very low in Fab1, Mab3, and Mab4 (0.94, 0.96, and 1.06 kcal / mol, respectively), which is consistent with the high experimental hydrolysis rates of these structures (13, 15, and 15% / week, respectively). In contrast, the free energy of stereostructure A is relatively high in Mab5 (1.56 kcal / mol) and highest in Fab6 (2.66 kcal / mol), which is in good agreement with the low experimental rates of these molecules (5% and 0% / week, respectively). However, the agreement between the free energy of the reactive stereostructure and the experimental rate is not sufficient in Fab2, where the free energy of the reactive stereostructure is very low (0.75 kcal / mol), but the experimental rate is not very high (9% / week). Even if the side chain adopts a reactive stereostructure, it can be inferred that the chemical reaction is energetically unfavorable in this case.

[0085] VIII.E. Discussion Under physiological conditions, the Asn residue is susceptible to deamidation, which hydrolyzes the amide side chain to form a free carboxylic acid. The rate-limiting step in this reaction is the formation of a five-membered succinimide ring intermediate. However, peptides containing Asn followed by a bulky residue, such as proline, have been shown to be susceptible to hydrolytic cleavage of the peptide backbone between the two residues. Here, evidence of an Asn-Pro motif susceptible to hydrolytic cleavage of the peptide backbone is provided. Mass spectral data of the Asn-Pro site identified in CDR-L3 led to the identification of three peptides associated with this site: a native trypsin peptide containing the Asn-Pro site, an N-terminal hydrolysis product peptide containing Asn and isoAsn, and a C-terminal hydrolysis product peptide. The identification of the N-terminal hydrolysis product peptide containing Asn and isoAsn, rather than Asp and isoAsp, suggests that the formation of the succinimide intermediate is a result of the attack of the β-side chain amide nitrogen of the peptide bond carbonyl. This COOH-terminal succinimide intermediate can then be opened to form the observed N-terminal hydrolysis product containing Asn or isoAsn.

[0086] In mouse antibodies or antibodies based on humanized mouse antibodies, the Asn-Pro at light chain positions 94-95 (Kavat-numbered) can be derived from selection of the IGKV4 mouse germline gene because some members of this family possess an Asn-Pro encoded in non-recombinant genes. In contrast, human light chain genes lack germline-encoded 94Asn-Pro95 because in human antibodies, this motif arrives either through the recombination VJ ligation process or via somatic hypermutation. Similarly, the Asn-Pro at heavy chain positions 52-52a (CDR-2) is found in the germline genes of both human and mouse IGHVI families. Consequently, in antibody collections, 52Asn-Pro52a of CDR-H2 is found more frequently than 94Asn-Pro95 of CDR-L3. The CDR-H2 motif has less solvent accessibility compared to the CDR-L3 position, which has high solvent accessibility and where Asn-Pro is strongly thought to play a direct role in antigen binding. Compared to the CDR-L3 position, the CDR-H2 motif tends to be more involved in maintaining the CDR stereostructure. This makes the Asn-Pro at CDR-L3 more susceptible to hydrolysis involving cleavage, which is strongly thought to affect target binding.

[0087] Recently, Jain et al. (PNAS, 2017) published a report on the biophysical properties of clinical-stage antibodies aimed at quantifying "development potential." In their collection of 137 antibodies, eight antibodies possess 94Asn-Pro95 in the CDR-L3. These are muromonab, mAb5, otlertuzumab, rituximab, teprizumab, tovetumab, bertuzumab, and vizilizumab. All have a 9-residue CDR-L3 sequence, with the exception of tovetumab, which has a 10-residue P95a insertion in the CDR. Teprizumab appears to be a humanized form of muromonab, while bertuzumab exhibits characteristics consistent with a humanized form of rituximab. Notably, both teprizumab and bertuzumab also possess Asn-Pro in the CDR-H2. A subsequent study (Lu et al., 2018) tested 131 antibodies described by Jain et al. for deamidation and isomerization tendencies, including eight antibodies with 94Asn-Pro95 on CDR-L3 in a pH 8.5 stress test. No evidence of cleavage at this site was reported for these antibodies. Using the in silico techniques described herein, it was possible to predict that the light chains of tobetumab and vizilizumab antibodies had cleavage-sensitive binding, while the others were less sensitive to cleavage (Table 3).

[0088] In addition to the data shown for Mab5 in Table 1 and Figure 9, stability data at 37°C at neutral pH were generated for all except muromonab, teplizumab, tobetuzumab, and vizilizumab. Antibodies formulated in PBS, pH 7.4 were incubated at 37°C for up to 4 weeks, and evidence of fragmentation was obtained from both intact and reduced samples by mass spectrometry and capillary electrophoresis of sodium dodecyl sulfate (CE-SDS). Although the rate was not determined, hydrolysis of Asn-Pro at CDR-L3 was detected with bertuzumab but not with rituximab or otreltuzumab (Table 3). The binding stability of rituximab and otreltuzumab was expected, while the instability of bertuzumab was not. These results support the conclusion that the antibody framework can influence the rate of hydrolysis, because bertuzumab and rituximab have the same amino acid sequence at CDR-L3. TIFF0007864109000005.tif72170

[0089] Therefore, hydrolysis of the Asn-Pro peptide bond by extended thermal stress at pH 7.4 has been demonstrated to result in a loss of antigen binding in Fab2. Since many therapeutic antibodies and antibody fragments are formulated under slightly acidic conditions where the kinetics of Asn-Pro cleavage are slow, this reaction has sometimes been underestimated in studies determining the shelf life of antibody formulations. Low-pH liquid or lyophilized formulations can be used to stabilize Asn-Pro cleavage in sensitive antibodies, but this does not rule out potential degradation under physiological conditions at neutral pH. In some cases, it may be desirable to remanipulate the antibody by amino acid substitution of unstable Asn residues. In fact, clinical candidates were generated from Fab1 and Fab2 via Asn-94 substitutions that can affect the degradation rate. Thus, if Asn residues are closely involved in antigen binding, and the residues are immutable, changes in the context of the sequence can stabilize the molecule while preserving target binding affinity. This was demonstrated by a Fab4 molecule with CDR grafted onto a non-human framework, which exhibited a four-fold slower hydrolysis rate of unstable binding compared to the parent antibody Fab2. Alternatively, candidates lacking this sequence motif, or molecules possessing Asn-Pro to be subjected to in silico risk assessment of this cleavage reaction, can be selected for development.

[0090] IX. Additional Considerations Some embodiments of this disclosure include a system comprising one or more data processors. In some embodiments, the system includes a non-temporary computer-readable storage medium containing instructions that cause one or more data processors to perform some or all of one or more of the methods and / or some or all of one or more processes disclosed herein, when the system is run on one or more data processors. Some embodiments of this disclosure include a computer program product tangibly integrated into a non-temporary machine-readable storage medium containing instructions configured to cause one or more data processors to perform some or all of the methods and / or some or all of one or more processes disclosed herein.

[0091] The terms and expressions used are for illustrative purposes only and not to limit, and the use of such terms and expressions is not intended to exclude any equivalent of the features or parts thereof shown and described, however it is recognized that various modifications are possible within the scope of the claimed invention. Thus, although the claimed invention is specifically disclosed by embodiments and optional features, modifications and changes to the concepts disclosed herein may be sought by those skilled in the art, and it should be understood that such modifications and changes are considered to be within the scope of the invention as defined by the appended claims.

[0092] The following description provides only preferred exemplary embodiments and is not intended to limit the scope, adaptability, or configuration of the present disclosure. Rather, the following description of preferred exemplary embodiments provides descriptions that enable various embodiments to be performed by those skilled in the art. It will be understood that various modifications may be made to the function and arrangement of the elements without departing from the spirit and scope set forth in the appended claims.

[0093] Certain details are presented in the following description to provide a full understanding of the embodiments. However, it is understood that the embodiments can be practiced without these specific details. For example, circuits, systems, networks, processes, and other components may be shown as components in block diagrams to avoid obscuring the embodiments with unnecessary details. In other cases, well-known circuits, processes, algorithms, structures, and techniques may be shown without unnecessary details to avoid obscuring the embodiments.

Claims

1. Regarding the polypeptide three-dimensional structure of a polypeptide containing amino acids with side chains and a backbone, the dihedral angles of the amino acid backbone and the side chains in the polypeptide three-dimensional structure are determined; Determining the nucleophilic attack distance between two atoms, functional groups, or combinations thereof of an amino acid in a polypeptide stereostructure, based on the dihedral angles of the skeleton and the side chains, wherein one of the two atoms or functional groups is in the side chain of the amino acid and the other is in the amino acid skeleton; Based on the nucleophilic attack distance of amino acids in the polypeptide three-dimensional structure, it is determined that the polypeptide three-dimensional structure is a reactive three-dimensional structure that is sensitive to cleavage reactions; In accordance with the determination that the polypeptide stereostructure is a reactive stereostructure, the free energies of the dihedral angles of the amino acid backbone and side chains when it is in a reactive stereostructure are determined; To predict the probability that an amino acid side chain will be trapped in a reactive stereostructure based on the free energies of the dihedral angles of the amino acid backbone and the dihedral angles of the side chains; To predict the probability of polypeptide chemical degradation as a result of amino acid side chains being trapped in a reactive three-dimensional structure, and A computer implementation method including, (i) Removing polypeptides from the list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chains are trapped in at least one reactive stereostructure and / or the probability that the polypeptides are chemically degraded; or (ii) Ranking a polypeptide lower than another polypeptide in a list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chain of another polypeptide is captured by at least one reactive stereostructure and / or that the polypeptide is chemically degraded, wherein the probability that the amino acid side chain of another polypeptide is captured by at least one reactive stereostructure and / or that the other polypeptide is chemically degraded is less than the probability that the amino acid side chain of the polypeptide is captured by at least one reactive stereostructure and / or that the polypeptide is chemically degraded. Computer implementation methods, including further details.

2. To generate a polypeptide representation; Performing molecular dynamics simulations using representations, wherein the results of the molecular dynamics simulations include a set of polypeptide stereostructures of polypeptides, including polypeptide stereostructures. The computer implementation method according to claim 1, further comprising:

3. Predicting the probability of polypeptides undergoing chemical degradation, To confirm the accessibility constraint, which shows that the amide group of the polypeptide has spatial accessibility that exceeds the threshold for binding to solvent molecules of the surrounding solvent when the conditions are met; Determining whether the accessibility constraint is satisfied for the reactive stereostructure based on the evaluation of one or more spatial properties of the polypeptide, The computer implementation method according to claim 1, including the method described in claim 1.

4. The determination that the polypeptide stereostructure is a reactive stereostructure is Determining a distance criterion that, if satisfied, indicates that an atom in a side chain is within a predetermined distance threshold of another atom in the backbone; The distance criterion is determined to be satisfied for a reactive stereostructure based on a comparison of the nucleophilic attack distance of the amino acids in the reactive stereostructure with a predetermined distance threshold, The computer implementation method according to claim 1, including the method described in claim 1.

5. The computer implementation method according to claim 1, wherein the free energy is determined by analyzing the free energy profiles of the dihedral angles of the backbone and side chains of the reactive three-dimensional structure of the amino acid, and the free energy profiles in space of the dihedral angles of the backbone and side chains are calculated from a bin collection.

6. The prediction of the probability that amino acid side chains will be trapped in the reactive three-dimensional structure is To determine an energy criterion that, if satisfied, indicates that the free energies of the dihedral angles of the amino acid skeleton and the dihedral angles of the side chains are within a predetermined energy threshold; The energy criterion is determined to be satisfied for the reactive stereostructure based on a comparison of the free energies of the dihedral angles of the amino acid skeleton and side chains of the reactive stereostructure with a predetermined energy threshold, The computer implementation method according to claim 1, including the method described in claim 1.

7. With one or more data processors; When running on one or more data processors, Regarding the polypeptide three-dimensional structure of a polypeptide containing amino acids with side chains and a backbone, the dihedral angles of the amino acid backbone and the side chains in the polypeptide three-dimensional structure are determined; Determining the nucleophilic attack distance between two atoms, functional groups, or combinations thereof of an amino acid in a polypeptide stereostructure, based on the dihedral angles of the skeleton and the side chains, wherein one of the two atoms or functional groups is in the side chain of the amino acid and the other is in the amino acid skeleton; Based on the nucleophilic attack distance of amino acids in the polypeptide three-dimensional structure, it is determined that the polypeptide three-dimensional structure is a reactive three-dimensional structure that is sensitive to cleavage reactions; In accordance with the determination that the polypeptide stereostructure is a reactive stereostructure, the free energies of the dihedral angles of the amino acid backbone and side chains when it is in a reactive stereostructure are determined; To predict the probability that an amino acid side chain will be trapped in a reactive stereostructure based on the free energies of the dihedral angles of the amino acid backbone and the dihedral angles of the side chains; To predict the probability of polypeptide chemical degradation as a result of amino acid side chains being trapped in a reactive three-dimensional structure, and A non-temporary computer-readable storage medium containing instructions that cause one or more data processors to perform an operation including, A system that includes, The operation is, (i) Removing polypeptides from the list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chains are trapped in at least one reactive stereostructure and / or the probability that the polypeptides are chemically degraded; or (ii) Ranking a polypeptide lower than another polypeptide in a list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chain of another polypeptide is captured by at least one reactive stereostructure and / or that the polypeptide is chemically degraded, wherein the probability that the amino acid side chain of another polypeptide is captured by at least one reactive stereostructure and / or that the other polypeptide is chemically degraded is less than the probability that the amino acid side chain of the polypeptide is captured by at least one reactive stereostructure and / or that the polypeptide is chemically degraded. including, system.

8. The operation is, To generate a polypeptide representation; Performing molecular dynamics simulations using representations, wherein the results of the molecular dynamics simulations include a set of polypeptide stereostructures of polypeptides, including polypeptide stereostructures. The system according to claim 7, further comprising:

9. Predicting the probability of polypeptides undergoing chemical degradation, To confirm the accessibility constraint, which shows that the amide group of the polypeptide has spatial accessibility that exceeds the threshold for binding to solvent molecules of the surrounding solvent when the conditions are met; Determining whether the accessibility constraint is satisfied for the reactive stereostructure based on the evaluation of one or more spatial properties of the polypeptide, The system according to claim 7, including the system described in claim 7.

10. The determination that the polypeptide stereostructure is a reactive stereostructure is Determining a distance criterion that, if satisfied, indicates that an atom in a side chain is within a predetermined distance threshold of another atom in the backbone; The distance criterion is determined to be satisfied for a reactive stereostructure based on a comparison of the nucleophilic attack distance of the amino acids in the reactive stereostructure with a predetermined distance threshold, The system according to claim 7, including the system described in claim 7.

11. The system according to claim 7, wherein the free energy is determined by analyzing the free energy profiles of the dihedral angles of the backbone and side chains of the reactive three-dimensional structure of the amino acid, and the free energy profiles in space of the dihedral angles of the backbone and side chains are calculated from a bin collection.

12. The prediction of the probability that amino acid side chains will be trapped in the reactive three-dimensional structure is To determine an energy criterion that, if satisfied, indicates that the free energies of the dihedral angles of the amino acid skeleton and the dihedral angles of the side chains are within a predetermined energy threshold; The energy criterion is determined to be satisfied for the reactive stereostructure based on a comparison of the free energies of the dihedral angles of the amino acid skeleton and side chains of the reactive stereostructure with a predetermined energy threshold, The system according to claim 7, including the system described in claim 7.

13. Regarding the polypeptide three-dimensional structure of a polypeptide containing amino acids with side chains and a backbone, the dihedral angles of the amino acid backbone and the side chains in the polypeptide three-dimensional structure are determined; Determining the nucleophilic attack distance between two atoms, functional groups, or combinations thereof of an amino acid in a polypeptide stereostructure, based on the dihedral angles of the skeleton and the side chains, wherein one of the two atoms or functional groups is in the side chain of the amino acid and the other is in the amino acid skeleton; Based on the nucleophilic attack distance of amino acids in the polypeptide three-dimensional structure, it is determined that the polypeptide three-dimensional structure is a reactive three-dimensional structure that is sensitive to cleavage reactions; In accordance with the determination that the polypeptide stereostructure is a reactive stereostructure, the free energies of the dihedral angles of the amino acid backbone and side chains when it is in a reactive stereostructure are determined; To predict the probability that an amino acid side chain will be trapped in a reactive stereostructure based on the free energies of the dihedral angles of the amino acid backbone and the dihedral angles of the side chain; To predict the probability of polypeptide chemical degradation as a result of amino acid side chains being trapped in a reactive three-dimensional structure, and A computer program product specifically implemented on a non-temporary, machine-readable storage medium, which includes instructions configured to cause one or more data processors to perform an operation, The operation is, (i) Removing polypeptides from the list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chains are trapped in at least one reactive stereostructure and / or the probability that the polypeptides are chemically degraded; or (ii) Ranking a polypeptide lower than another polypeptide in a list of potential polypeptides to be used as at least part of a therapeutic agent, based on the probability that the amino acid side chain of another polypeptide is captured by at least one reactive stereostructure and / or that the polypeptide is chemically degraded, wherein the probability that the amino acid side chain of another polypeptide is captured by at least one reactive stereostructure and / or that the other polypeptide is chemically degraded is less than the probability that the amino acid side chain of the polypeptide is captured by at least one reactive stereostructure and / or that the polypeptide is chemically degraded. Further including, Computer program products.

14. The operation is, To generate a polypeptide representation; Performing molecular dynamics simulations using representations, wherein the results of the molecular dynamics simulations include a set of polypeptide stereostructures of polypeptides, including polypeptide stereostructures. The computer program product according to claim 13, further comprising:

15. Predicting the probability of polypeptides undergoing chemical degradation, To confirm the accessibility constraint, which shows that the amide group of the polypeptide has spatial accessibility that exceeds the threshold for binding to solvent molecules of the surrounding solvent when the conditions are met; Determining whether the accessibility constraint is satisfied for the reactive stereostructure based on the evaluation of one or more spatial properties of the polypeptide, A computer program product according to claim 13, including the above.

16. The determination that the polypeptide stereostructure is a reactive stereostructure is Determining a distance criterion that, if satisfied, indicates that an atom in a side chain is within a predetermined distance threshold of another atom in the backbone; The distance criterion is determined to be satisfied for a reactive stereostructure based on a comparison of the nucleophilic attack distance of the amino acid in the reactive stereostructure with a predetermined distance threshold, A computer program product according to claim 13, including the above.

17. The computer program product according to claim 13, wherein the free energy is determined by analyzing the free energy profiles of the dihedral angles of the backbone and side chains of the reactive three-dimensional structure of the amino acid, and the free energy profiles in space of the dihedral angles of the backbone and side chains are calculated from a bin collection.

18. The prediction of the probability that amino acid side chains will be trapped in the reactive three-dimensional structure is To determine an energy criterion that, if satisfied, indicates that the free energies of the dihedral angles of the amino acid skeleton and the dihedral angles of the side chains are within a predetermined energy threshold; The energy criterion is determined to be satisfied for the reactive stereostructure based on a comparison of the free energies of the dihedral angles of the amino acid skeleton and side chains of the reactive stereostructure with a predetermined energy threshold, A computer program product according to claim 13, including the above.

Images