Methods and systems for characterizing the dynamic interactions of drugs with IKR or HERG channels

Abstract

The present invention relates to a drug and I Kr or a method for characterizing a dynamic interaction with a hERG channel, the method comprising: Kr or modeling the interaction with hERG; Kr or hERG channel conformational state; and defining a plurality of voltage clamp protocols and obtaining their concentration-response distributions. For each voltage clamp protocol and simulated compound, the drug concentration at which half-blocking of the corresponding ionic current occurs (IC 50 ) and a corresponding value of the time constant (τ); and characterizing the compound according to the obtained values.

Description

[Technical field]

[0001] The present invention is in the field of drug development technology and computational science. More specifically, the present invention relates to a compound I Kr The present invention relates to methods for modeling and characterizing dynamic interactions with the human ether-a-go-go related gene (hERG) channel (delayed rectifier potassium current). [Background technology]

[0002] The discovery and development of new drugs has traditionally been based on trial and error procedures. However, in practice, these procedures are often inefficient, excessively time-consuming or risky, both from the point of view of health and from the point of view of investment in private or public research. This is due to the fact that, in the case of many developed medicines, they do not reach the market due to potential adverse side effects on health, especially with regard to the heart, such as ventricular arrhythmias. Up to now, the preclinical evaluation of the cardiac safety of new drugs has mainly been carried out by the evaluation of the efficacy of the compound under investigation (I in humans). Kr This has been based on studies of in vitro blockade of the hERG gene (which encodes the pore-forming alpha subunit of the fast component of the current channel) and in vivo prolongation of the QT interval (a measure of the time from the onset of the Q wave to the end of the T wave on the electrocardiogram), both of which are associated with the development of polymorphic ventricular tachycardia ("Torsades de Pointes" or TdP).

[0003] Although these safety tests have been successful in preventing potentially harmful drugs from reaching the market, they have also prevented the development of potentially useful compounds. Indeed, there are potent I Kr There are drugs such as verapamil that are recognized as current blockers.

[0004] Recent studies have demonstrated that the assessment of the safety of new drugs is improved when their kinetics and dynamics are taken into account. For example, the drug concentration at which half-blocking of the ionic current occurs (IC 50 The values ​​of the agonist, the concentration of agonist (referred to as the agonist concentration), can vary widely. In turn, these values ​​may also vary depending on the stimulation and voltage clamp protocols used to measure these values.

[0005] Moreover, in recent years, the “Comprehensive In vitro Proarrhythmia Assay” (CiPA) initiative led by the US Food and Drug Administration (FDA) has emerged, which seeks to improve the techniques for the discovery and development of new drugs by defining a new paradigm based on computational models, which are proposed as the main tool to predict the risk of arrhythmias induced by new drugs. This initiative also focuses on a more detailed pharmacodynamic component. Traditionally, the formulation of the action of drugs at ion channels is based on the IC 50This has been simplified by simple pore blocking models, called static models, which use the coefficients and Hill coefficients to reduce the conductance of each ionic current. Although these static models are simple, they are not overly realistic because, as mentioned above, the blocking ability of compounds depends on the experimental conditions used to elucidate them. To this extent, dynamic models, such as the Markov model, which model the kinetics of binding and unbinding of drugs to the channel in different states, are considered to be more realistic and provide unique information that is independent of the experimental protocol. In this sense, the publications by Z. Li et al., “Improving the in silico assessment of proarrhythmia risk by combining hERG (human ether- a-go- go-related gene) channel- drug binding kinetics and multichannel pharmacology,” Circ. Arrhythm. Electrophysiol. 10, e004628 (2017) and “Quantitative Systems Pharmacology Models for a New International Cardiac Safety Regulatory Paradigm: An Overview of the Comprehensive In Vitro Proarrhythmia Assay In Silico Modelling Approach,” CPT Pharmacometrics, Syst, Pharmacol (2019), 8, 371-379, are relevant.

[0006] Therefore, I Kr The current technique proposed by CiPA to obtain a dynamic model of hERG blocking drugs uses a single Markov chain, from which certain parameters (K max , K u , EC50 n , N., V. half-trapThe parameters must be adjusted (e.g., the binding time, the concentration, etc.) to allow for distinguishing between drugs with different binding kinetics captured by dynamic voltage clamp protocols. However, this technique presents many problems. For example, many of the proposed drug models do not achieve complete blocking of the current even at high concentrations. Similarly, the algorithms for setting the parameters are generally overly complex, and furthermore, their dependency on the seed used to obtain them can condition the results, limiting their use to people already familiar with how to set the parameters. Finally, the voltage clamp protocol of CiPA requires a rather long time compared to the typical stabilization time of cells under such conditions.

[0007] As a result of these limitations, the art is now well known for, among other features and advantages: - Obtaining dynamic models of interactions automatically, without the need for expert modeling or parameter setting personnel intervention; - generating models in a short time and without high demands on computational resources; - Providing a wide variety of generated models; - taking into account that drug binding and unbinding from the channel can occur in any channel state; - implementing voltage clamp protocols more suitable than those currently used to generate drug models; There is a need to develop alternative, new modeling methods and systems that allow for:

[0008] The present invention relates to a drug and I Kr The present invention aims to meet the above-mentioned needs by novel methods and systems for automatically modeling and characterizing dynamic interactions with the hERG channel. Summary of the Invention [Problem to be solved by the invention]

[0009] As described in the Background section, in order to overcome the limitations of the prior art described above, the object of the present invention is to develop a novel drug and Kr The present invention provides an improved technique for automatically characterizing and modeling the dynamic interactions of any drug with the I or hERG channel. Kr or can be used to model dynamic interactions with the hERG channel. The term agent includes, among others, drugs, antibodies, small molecules, pharmaceutical compounds, peptides, or proteins. Its use is not limited to cardiac safety studies, but rather can be used in other applications, such as, for example, studying the efficacy of antiarrhythmic drugs. [Means for solving the problem]

[0010] More specifically, the first object of the present invention is to provide a method for the preparation of a drug and an I Kr or a computer-implementable method for characterizing dynamic interactions with a hERG channel, the method comprising the steps of: -I Kr or modeling an interaction between hERG and a drug, the modeling comprising generating a Markov chain model; -I Kr or selecting a set of drugs to be characterized by simulation of a plurality of compounds distributed into groups having different kinetics and affinities for the conformational states of the hERG channel, the distribution of groups being determined according to the following criteria: - the above conditions for binding and debinding of the drug, and the preferred binding conditions; - the ratio of dissociation rates between favorable and unfavorable states, arbitrarily set at multiple thresholds; - the ability of the binding channel to change its conformational state without becoming dissociated from the compound; The step is based on - determining a plurality of electrical cell stimulation voltage clamp protocols and obtaining their corresponding concentration-response distributions; This includes carrying out the following:

[0011] Advantageously, the method of the invention comprises the following steps: - For each voltage clamp protocol and for the multiple compounds simulated in the drug modeling step, the drug concentration at which half-blocking of ionic current occurs (IC 50 ) values ​​and the corresponding IC 50 obtaining a value of the time constant (τ) of the normalized tail current (i.e., the value of the normalized tail current amplitude for each sweep of the voltage clamp protocol plotted as a function of time) when a drug concentration value substantially equal to - Each of the thresholds, whether the channel bound to the compound can change the above states where the drug binds and unbinds while the compound is bound, and the IC at the line based on its preferred binding state 50 and interpolating values ​​of - interpolating the τ value with a line based on each of the thresholds, whether the channel bound to the compound while the compound is bound can change the above states where the drug binds and unbinds, and its preferred binding state; - classifying the compounds by performing the following sub-steps performed by SVM and linear interpolation, the sub-steps being: - obtaining the difference in dissociation rate between the preferred state and other states in which each compound is bound, and classifying the compounds according to their dissociation rate with respect to the unfavourable binding state, each compound being assigned to the group corresponding to the closest interpolated line, - assigning each compound to the group corresponding to the interpolated line of its closest time constant (τ), calculating the distance between each compound and all the lines, and determining whether the drug enables the drug binding channel to change state if the closest line corresponds to a compound with this characteristic; and - assigning each compound to a class according to the results obtained in the above sub-steps and characterizing each compound based on the states in which it binds and unbinds, and on its preferred binding states; and Further includes:

[0012] In a preferred embodiment of the invention, during the drug selection step, the criterion of the distribution of groups with respect to the ability of the drug-binding channel to change its conformational state without becoming dissociated from the compound comprises the formulation of a Markov chain model called Stuck if said ability is exhibited, or Unstuck if said ability is not exhibited.

[0013] In a preferred embodiment of the present invention, compound I Kr The step of modeling the interaction with hERG involves five drug-free states: three closed states (C3, C2, C1), an open state (O), and an inactivated state (I), and up to five additional drug-containing states (C 3d , C 2d , C 1d , O d、 and I d ) includes an implementation of the Markov chain model with

[0014] More preferably, the distribution of groups during the drug modeling step satisfies the following criteria: - the above conditions for binding and debinding of the drug, and the preferred binding conditions; - the ratio of dissociation rates between favorable and unfavorable states, arbitrarily set at multiple thresholds; - the ability of the drug-binding channel to change its conformational state without becoming dissociated from the compound, distinguishing between unstacked and stacked models; Includes.

[0015] Even more preferably in the method of the present invention, the distinction of the compounds according to their binding state comprises the following classes: open, inactivated, closed, open I, inactivated O, OI, open C, closed O, CO, open CI, inactivated CO, closed OI and COI, where these classes correspond to the following binding states: open, inactivated and closed are compounds that only interact in the open state, inactivated state and closed state, respectively. Open I, inactivated O and OI are drugs that bind to the open state and the inactivated state, with the open state being preferred, the inactivated state being preferred and both states being equally preferred, respectively. Open C, closed O and CO represent drugs that bind to the closed state and the open state simultaneously, but with a higher affinity for the open state, a higher affinity for the closed state and the same affinity for both states, respectively. Finally, open CI, inactivated CO, closed OI, and COI are compounds that bind to the three states simultaneously, with higher affinity for the open state, higher affinity for the inactivated state, or higher affinity for the closed state, and the same affinity for all of them, respectively.

[0016] In another preferred embodiment of the present invention, Kr Alternatively, the step of modeling the interaction with hERG includes creating further check groups using random seeds and generating variants that may or may not change state without causing the channel bound to the compound to dissociate.

[0017] In another preferred embodiment of the invention, the step of defining an electrical stimulation protocol comprises, for a given value of temperature and given values ​​of intracellular and extracellular potassium concentrations: - Variable voltage adjustment pulse, - test pulse, - optionally a prepulse, This involves the use of protocols including

[0018] More preferably, the electrical stimulation protocols include protocols P40, P0, and P-80.

[0019] Even more preferably, the step of defining the electrical stimulation protocol includes obtaining a concentration-response distribution by plotting the steady-state normalized peak tail current as a function of the common logarithm of the concentration of the drug, 50 I in Kr Alternatively, the evolution of the normalized peak tail current of the hERG is analyzed based on pulse number or time.

[0020] In another preferred embodiment of the invention, the method comprises a further step of optimization, in which, once the compounds are characterized, the association and dissociation constants for each state in which each compound interacts with the channel are calculated and processed by an iterative optimization algorithm. More preferably in the method of the invention, the iterative optimization algorithm is based on the Nelder-Mead simplex algorithm.

[0021] A second object of the invention relates to a computing system including software and / or hardware means adapted to carry out the method according to any of the embodiments described herein.

[0022] A third object of the invention relates to a computer program comprising instructions adapted to be executed in a corresponding computing system in a manner according to any of the embodiments described herein.

[0023] Within the meaning of the present invention, the expression "substantially" with respect to any term shall be understood as being the same as the said term or within a margin of variation of ±25%.

[0024] Similarly, although the main embodiments described herein refer to a particular value of the concentration of potassium, the methods described in the claims can be applied in an equivalent manner to the present invention to any other value of this concentration, without any limitation as to its range. 50 The use of I is to be considered within the scope of the present invention to extend broadly to other concentration values. In this same sense, and without loss of equivalence, Kr Alternatively, it is possible to modify the methods described, including using mutations in the hERG channel.

[0025] In the same way, a preferred embodiment of the method of the present invention relates to the selection of a drug to be characterized by the simulation of a plurality of compounds. In this sense, the selection of said drug is considered to be applicable to both known (or real) compounds and virtual compounds (e.g., the formulations of which are generated randomly or pseudorandomly under certain computational rules), without any limitations with respect to the scope of protection of the present invention. Thus, in a manner equivalent to the above-mentioned preferred embodiment, when using real compounds, an electrical stimulation voltage clamp protocol is preferably applied experimentally to obtain ICs that serve as input data for the described method. 50 and the value of τ can be obtained. In addition, it is also possible to obtain the value of τ from the evolution of the tail current. [Brief description of the drawings]

[0026] These and other features and advantages will become more fully understood from the detailed description of the invention and preferred embodiments thereof, taken in conjunction with the accompanying drawings, in which: [Figure 1]FIG. 1 shows the different IKr Markov models used in the method of the present invention according to a preferred embodiment. More specifically, this figure represents the drug-bound states (C3d, C2d, C1d, Od, and Id) and drug-free states (C3, C2, C1, O, and I). kc, ko, and ki are the diffusion rates for the closed, open, and inactivated states, respectively. rc, ro, and ri are the dissociation rates for the closed, open, and inactivated states, respectively. D is the concentration of drug. The left column represents the names of possible interactions between drugs and channels for each model. [Diagram 2] Different plots showing the effect of three voltage clamp protocols, P0 (top), P40 (middle) and P-80 (bottom), depicting the voltage protocols (left), Hill plots of a given hypothetical compound using the three protocols (middle) and the evolution of the (normalized) peak tail current at the corresponding concentration IC50 (right). [Diagram 3] FIG. 2 shows a flow chart of the different steps of the classifier used in the method of the present invention according to a preferred embodiment. [Figure 4] 4A-4B show different examples of three-dimensional representations of parameters extracted from drug simulations by the method of the present invention according to a preferred embodiment, depicting the time constants of a subset of compounds of each class and the corresponding interpolated lines (colors) used in the second and third steps of the classifier (FIG. 4a) and the IC50 values ​​of another subset of compounds together with the two decision surfaces (FIG. 4b). [Diagram 5] Figure 5 shows several Hill plots for dofetilide using three voltage protocols: P0 (continuous), P40 (discontinuous), and P-80 (dotted), obtained experimentally (black lines) and using two kinetic models (grey lines), the one obtained according to the present invention (Fig. 5a) and the one proposed by CiPA [Li et al. 2017] (Fig. 5b). [Figure 6]Figure 6 shows the evolution of the (normalized) peak tail current at the concentration closest to the IC50 obtained experimentally (black line) and using the dynamic model derived according to the invention (grey line) for the three protocols P0 (Figure 6a), P40 (Figure 6b) and P-80 (Figure 6c). DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] A detailed description of the invention is provided below with reference to the figures herein. This description is provided for the purpose of illustrating, but not limiting, the claimed invention.

[0028] As described in the previous section, the first object of the present invention is to provide a method for the preparation of a compound I Kr or a method for modeling dynamic interactions with a hERG channel, said method advantageously comprising the steps of: -I Kr modeling the interactions; - selecting a compound; - defining a stimulation protocol; - Results (time constant and IC 50 (value of - a classification step; - (optional) an optimization step; This includes carrying out the following:

[0029] Each of the steps involved in the method comprises the steps of: r A preferred embodiment thereof as applied to interaction with the hERG channel is described below (however, interaction with the hERG channel is equivalent depending on the required variant, which is not a novel object of the present invention in itself, but rather is described in the above I). Kr or hERG channels, and thus are general steps of the methodology that would be available to one of skill in the art.

[0030] Drugs and I Kr Steps to model interactions with In the preferred embodiment described, I Kr For example, the I of the human ventricle, such as that proposed in M. Fink et al., “Contributions of HERG K+ Current to Repolarization of the Human Ventricular Action Potential”, Prog. Biophys. Mol. Biol. (2008), vol. 96, pp. 357-376, Kr The model is simulated using a Markov chain model, although other models based on Markov chains can be implemented as well without departing from the scope of the present invention. In this embodiment, the model includes five states: three closed states (C3, C2, C1), an open state (O), and an inactivated state (I). Similarly, the drug and the channel (in this case, I K r channel), see, for example, J. Gomis-Tena et al., “When Does the IC 50 As proposed in “Accurately Assess the Blocking Potency of a Drug?”, J. Chem. Inf. Model (2020), 60, 3, 1779-1790, five new states (C 3d , C 2d , C 1d , O d , and I d ) are included. All the considered channel configurations are shown in FIG.

[0031] <Compound selection step> The next step of the method involves simulating the drug, which in the described embodiment is KrIt is based on a set of 26000 virtual drugs distributed in 104 groups with different kinetics and affinity for the conformational states of the channels. These groups are preferably generated taking into account three aspects: i) the above-mentioned states where the drugs bind and unbind, and the preferred binding states resulting in 13 classes, ii) the ratio of the dissociation rates of the preferred and unpreferred states, arbitrarily set at 3, 10, 30 or 100 as reference thresholds (but other choices are equally feasible), and finally, iii) the ability of the bound channels to change their conformational state without becoming dissociated from the compound (if this ability is shown, the unstacked model is formulated, if this ability is not shown, the stacked model is formulated). The 13 classes are named according to the binding states: open, inactivated and closed are compounds that interact only in the open, inactivated and closed states, respectively. Open I, inactivated O, and OI are drugs that bind to the open and inactivated states, with preference for the open state, preference for the inactivated state, and equal preference for both states, respectively. Open C, closed O, and CO represent drugs that bind to the closed and open states simultaneously, but with higher affinity for the open state, higher affinity for the closed state, and equal affinity for both states, respectively. Finally, open CI, inactivated CO, closed OI, and COI are compounds that bind to the three states simultaneously, with higher affinity for the open state, higher affinity for the inactivated state, or higher affinity for the closed state, and equal affinity for all of them, respectively.

[0032] In addition, in certain embodiments of the present invention, it is possible to create further check groups to verify the operation of the system using different random seeds. Thus, in the example proposed above, another 8 different sets of 650 check compounds (13 classes and 50 drug variants per class that can or cannot change the state of the channels bound to the compounds without causing them to become dissociated (for the unstacked and stacked models, respectively) are created.

[0033] Stimulation Protocol In this step of the method, the voltage clamp protocols used for the simulation are defined. The example described in this embodiment of the invention uses three known voltage clamp protocols called P40, P0, and P-80 published in the paper by Gomis-Tena et al. (2020) referenced above, which maximize the probability that the channel occupies each conformational state (see the left column of FIG. 2). These protocols essentially include a variable voltage conditioning pulse (in the example, 5 seconds for P0 and P40, and 4.5 seconds for P-80), followed by a test pulse (in this example, a 0.2 second pulse at -60 mV repeated at 5.4 second intervals from a potential of -80 mV). Similarly, conditioning pulses are performed at potentials of -80, 0, and 40 mV for P-80, P0, and P40, respectively. In the case of P-80, a 0.5 ms prepulse at 20 mV is incorporated, and a 0.2 s test pulse is set to -50 mV. Values ​​for temperature (22° C. in the example), intracellular potassium concentration (130 mM in the example), and extracellular potassium concentration (4 mM in the example) are set within the ranges defined in the protocol. Values ​​for time, potential, temperature, and concentration may be varied in different embodiments of the invention, with no limitations as to their ranges.

[0034] Corresponding concentration-response plots, also called Hill plots, are obtained from the voltage clamp protocol. In the example described, these plots are made by plotting the steady-state normalized peak tail current as a function of the common logarithm of the concentration of drug (see the middle column of Figure 2), following the one described in the publication by Gomis-Tena et al. (2020). In addition, the concentration IC 50 I in Kr The normalized peak tail currents are analyzed as a function of pulse number or time (right column of FIG. 2).

[0035] <Steps to get results> Three values ​​of the time constant (τ P-80 , τ P0 , τ P40 ) and IC 50 The three values ​​of (IC 50,P-80 ,I C 50,P0 ,I C 50,P40 ) are calculated from predefined simulation parameters and depicted in 3D scatter plots (see Figs. 3a and 3b, respectively). In the preferred embodiment described here by way of example, the IC 50 It can be seen that the distribution of values ​​and time constants of IC resembles a linear trend. These parameters are used to train a classifier, which in this embodiment includes three substeps, as depicted in the diagram of Figure 4. To do so, IC 50 and the value of the time constant are required, and a support vector machine (SVM) and linear interpolation are preferably used. The above mentioned sub-steps are described in detail below: - First, the difference in the dissociation rate between the preferred state and other states in which the channel interacts with the compound is obtained. The compounds are classified according to their dissociation rate with respect to the unfavourable binding state (in this example, 3R, 10R, 30R and 100R, respectively, when the compound's dissociation rate is 3-fold, 10-fold, 30-fold or 100-fold lower than the unfavourable binding state). For this step, the IC 50The values ​​of are used to interpolate in a line using all the points for each of the 104 possible classifications of the training group (3R, 10R, 30R, or 100R, whether the channel bound to the compound can change state or not, and 13 final classes, i.e., 4 x 2 x 13 = 104). The criteria used for classification are the IC 50 , the distance between the point defined by the three values ​​of all the interpolated lines, where the compound is assigned to the group corresponding to the closest interpolated line. Thus, the ratio of the dissociation rates is determined after this substep (see FIG. 4b). - Secondly, it is ascertained whether the compound allows the drug-binding channel to change state without becoming dissociated from the drug. In this substep, a time constant (τ) is used, the method being similar to that used in the first substep. In this case, a line is interpolated for each of the 26 possibilities (whether the drug-binding channel changes its state and 13 classes, i.e. 2x16=26). Figure 3a shows a three-dimensional representation of the time constant and the interpolated lines for drug 30R in a variant that allows a change of state for the drug-binding channel. The distance between the compound and all the lines is calculated, and if the closest line corresponds to a compound with this characteristic, it is determined that the drug allows the drug-binding channel to change its state. After this substep, the compound is assigned to one of eight groups, which are the result of the four ratios and the possibility of whether the compound allows a change of state for the channel without becoming dissociated (4x2=8). - Third, a class is assigned out of 13 possibilities according to the criteria above.

[0036] In a particular embodiment of the present invention, the complete classifier following the steps above correctly classified 88.39% of the compounds according to whether they allowed a state change in the drug binding channel, with the class being confirmed in 92.05% of the cases.

[0037] <Optimizer> Once the characteristics of the compound are identified, the association and dissociation rates are calculated for each state in which the compound interacts with the channel. For this purpose, an iterative optimization tool is preferably used, for example based on the Nelder-Mead simplex algorithm. The parameters of this tool are calculated taking into account the characteristics found by the classifier in the above steps. The optimizer adjusts the model of the compound with four possibilities: 3R, 10R, 30R, and 100R. Once the optimization of the association and dissociation constants of the four possibilities is completed, the four errors are compared and the model with the smallest error is selected. In all cases, the solution provided by the optimization tool is the one with the correct ratio of dissociation rates.

[0038] In addition to the methods of the present invention, another main object of the present invention relates to a computing system (e.g., a computer (but not limited to a computing network or cluster, a graphics processing unit (GPU), or other device having storage and data processing capabilities)) comprising software and / or hardware means adapted to perform the methods according to any of the embodiments described herein.

[0039] Finally, a third object of the invention relates to a computer program comprising instructions adapted to be executed in a corresponding computing system in a method according to any of the embodiments described herein.

[0040] <Embodiments of the present invention> As an example of experimental validation of the present invention, voltage clamp experiments were performed in HEK cells stably transfected with hERG. Measurements were performed using Nanion SyncroPatch 384i, an automated platform in which 384 wells are measured simultaneously. Protocols P-80, P0, and P40 were applied to measure the I Kr and IC for dofetilide, a drug known to block hERG. 50The IC values ​​and time constants (τ) were obtained. Experiments performed with P40, P0, and P-80 were performed on the same day using the same method, the same cells, and the same compound solutions. Thus, IC values ​​for each protocol were obtained. 50 The values ​​of (IC 50P-80 ,I C 50P0 , and I.C. 50P40 ) and then calculate the IC for each protocol. 50 The evolution of hERG block as a function of time at the applied concentrations closest to the τ was recorded by measuring the evolution of the normalized peak tail current as a function of time and calculating the values ​​of the time constants, τ, respectively. P-80 , τ P0 , and τ P40 IC 50 The three values ​​of (IC 50P-80 ,I C 50P0 , and I.C. 50P40 ) and three values ​​of the time constant (τ P-80 , τ P0 , τ P40 ) was used according to the method described in this invention to generate a kinetic Markov model for the interaction of dofetilide with the channel. As depicted in Figures 5a and 5b, respectively, the generated model differs from the kinetic model for dofetilide proposed by CiPA [Li, Z.; Dutta, S.; Sheng, J.; Tran, PN; Wu, W.; Chang, K.; Mdluli, T.; Strauss, DG; Colatsky, T., “Improving the In Silico Assessment of Proarrhythmia Risk by Combining HERG (Human Ether-a-Go-Go-Related Gene) Channel-Drug Binding Kinetics and Multichannel Pharmacology”, Circ.: Arrhythmia Electrophysiol. 2017, 10, No. e004628] and shows a significant improvement over the experimentally recorded IC 50values ​​and the evolution of hERG blocking were successfully reproduced. Thus, Figures 5a-5b above show different Hill plots for dofetilide, obtained experimentally (solid lines) and using two kinetic models (discontinuous lines): the one obtained by the present invention (Figure 5a) and the one proposed by CiPA [Li et al. 2017] (Figure 5b), using three voltage protocols, P0, P40, and P-80.

[0041] Next, Figures 6a-6c show the ICs obtained experimentally (continuous lines) and using the dynamic model obtained according to the present invention (discontinuous lines) for the three protocols P0 (Figure 6a), P40 (Figure 6b), and P-80 (Figure 6c). 50 Figure 1 shows the evolution of the (normalized) peak tail current at concentrations closest to .

[0042] These results suggest that the drug and I Kr or hERG channel dynamic interactions, constitute a proof of concept of the possible value of the method of the present invention for characterizing, modeling, and simulating the I Kr Proarrhythmic risk of drugs that inhibit I Kr This has contributed to improving the preclinical evaluation of the efficacy of antiarrhythmic drugs that block vasoconstrictors.

Claims

1. Drugs and I Ｋｒ or a computer-implementable method for characterizing dynamic interactions with a hERG channel, the method comprising: - I Ｋｒ or modeling an interaction between hERG and a drug, said modeling comprising generating a Markov chain model; - I Ｋｒ or selecting a set of drugs to be characterized by simulation of a plurality of compounds distributed into groups having different kinetics and affinities for the conformational states of the hERG channel, said distribution of groups being determined by the following criteria: - the conditions under which the drug binds and unbinds, and the preferred binding conditions; - the ratio of dissociation rates between favorable and unfavorable states, arbitrarily set at multiple thresholds; the ability of the binding channel to change its conformational state without becoming dissociated from said compound; The step is based on - defining a plurality of electrical cell stimulation voltage clamp protocols and obtaining their corresponding concentration-response distributions; Includes doing The method comprises the steps of: - each voltage clamp protocol, and I Ｋｒ or for a plurality of compounds simulated in the step of modeling the interaction between hERG and a drug, the drug concentration at which half-blocking of the ionic current occurs (IC ５０ ) values ​​and the corresponding IC ５０ obtaining a value of the normalized tail current time constant (τ) when a drug concentration value substantially equal to - IC at each of the thresholds, whether the drug-binding channel is allowed to change state while the compound is bound, and its preferred binding state ５０ and interpolating values ​​of - interpolating the τ value with a line based on each of said thresholds, whether the drug binding channel is allowed to change its state while the compound is bound, and its preferred binding state; - classifying said compounds by carrying out the following sub-steps carried out by SVM and linear interpolation, said sub-steps being: - obtaining the difference in the dissociation rate between the preferred state and other states in which each compound is bound, and classifying the compounds according to their dissociation rate with respect to the non-preferred binding state, each compound being assigned to the group corresponding to the closest interpolated line, - assigning each compound to the group corresponding to the closest interpolated line of its time constant (τ), calculating the distance of each compound to all the lines, and determining whether the drug allows the drug binding channel to change state if the closest line corresponds to a compound with this characteristic; - assigning each compound to a class according to the results obtained in said sub-steps and characterizing each compound based on the states in which it binds and unbinds, and on its preferred binding states; and The method further comprising:

2. The method of claim 1, wherein during the step of selecting the set of drugs, a criterion for the distribution of the groups with respect to the ability of the drug-binding channel to change its conformational state without becoming dissociated from the compound comprises the formulation of a Markov chain model called unstacked if the ability is demonstrated, or a Markov chain model called stacked if the ability is not demonstrated.

3. I Ｋｒ The step of modeling the interaction between hERG and a drug includes five drug-free states: three closed states (C ３ , C ２ , C １ ), an open state (O), and an inactivated state (I), and up to five additional drug-containing states (C ３ｄ , C ２ｄ , C １ｄ , O ｄ , and I ｄ 2. The method of claim 1 , comprising implementing a Markov chain model having

4. I Ｋｒ or the distribution of groups during the step of modeling the interaction between hERG and a drug is determined according to the following criteria: - the conditions under which the drug binds and unbinds, and the preferred binding conditions; - the ratio of dissociation rates between favorable and unfavorable states, arbitrarily set at multiple thresholds; - the ability of the drug-binding channel to change its conformational state without becoming dissociated from the compound, distinguishing between stacked and unstacked models; The method of claim 3, wherein the method is based on

5. 5. The method of claim 4, wherein the classification of the compounds according to their binding state includes the following classes: open, inactivated, closed, open I, inactivated O, OI, open C, closed O, CO, open CI, inactivated CO, closed OI, and COI.

6. I Ｋｒ 2. The method of claim 1, wherein the step of modeling the interaction between hERG and a drug includes creating further check groups using random seeds and generating variants that may or may not change state without causing channels bound to the compound to become dissociated.

7. The step of defining a plurality of electrical cell stimulation voltage clamp protocols includes, for a given value of temperature and given values ​​of intracellular and extracellular potassium concentrations: - variable voltage regulated pulse, - test pulse, - optionally a prepulse, The method of claim 1 , comprising use of a protocol comprising:

8. 8. The method of claim 7, wherein the plurality of electrical cell stimulation voltage clamp protocols includes protocols P40, P0, and P-80.

9. The step of defining the plurality of electrical cell stimulation voltage clamp protocols includes obtaining a concentration-response distribution by plotting steady-state normalized peak tail current as a function of the common logarithm of the concentration of the drug, ５０ Current I Ｋｒ The method of claim 7, wherein the normalized peak of is analyzed as a function of pulse number or time.

10. 2. The method of claim 1, comprising a further step of optimization in which, once the compounds have been characterized, association and dissociation constants for each state of the channel with which each compound interacts are calculated and processed by an iterative optimization algorithm.

11. The method of claim 10, wherein the iterative optimization algorithm is based on the Nelder-Mead simplex algorithm.

12. I Ｋｒ or the use of a mutation in the hERG channel.

13. 2. The method of claim 1, wherein the plurality of compounds corresponding to the drug include real compounds and / or virtual compounds, the formulations of which are randomly or pseudo-randomly generated under one or more computational rules.

14. A computing system comprising software and / or hardware means adapted to carry out the method according to any one of claims 1 to 13.

15. A computer program comprising instructions adapted to be executed in a method according to any one of claims 1 to 13 on a computing system comprising software and / or hardware means adapted to perform the method according to any one of claims 1 to 13.

Images