Modelling ion channels with a view towards identifiability Dedicated to Edmund J. Crampin (1973-2021)

Ivo Siekmann · 2025

Read Modelling ion channels with a view towards identifiability Dedicated to Edmund J. Crampin (1973-2021) by Ivo Siekmann (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-26728.txt.

Introduction

Ivo Siekmann reviews how ion channels are modelled, with an eye on what the models can and cannot pin down. Ion channels are proteins that gate ion flow across cell membranes. Aggregated Markov models describe them with several open and closed states, to capture fast and slow openings. Each state is often read as a distinct physical shape of the channel.

Siekmann shows how such models with a given number of states can be counted. He then revisits a classical result: past a stated parameter cap, models become non-identifiable, meaning different settings fit the data equally well. He gives two derivations and a close study of a three-state case. His advice is to build models on data that speaks more directly to shape changes, rather than on assumed transitions between states.

Catalog record

CREATORIvo Siekmann
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE115 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-26728

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Modelling ion channels with a view towards identifiability Dedicated to Edmund J. Crampin (1973-2021) Ivo Siekmann Abstract Aggregated Markov models provide a flexible framework for stochastic dynamics that develops on multiple timescales. For example, Markov models for ion channels often consist of multiple open and closed state to account for “slow” and “fast” openings and closings of the channel. The approach is a popular tool in the construction of mechanistic models of ion channels—instead of viewing model states as generators of sojourn times of a certain characteristic length, each individual model state is interpreted as a representation of a distinct biophysical state. We will review the properties of aggregated Markov models and discuss the implications for mechanistic modelling. First, we show how the aggregated Markov models with a given number of states can be calculated using Pólya enumeration. However, models with n O n_{\text{{\color[rgb]{0,0.4,0}O$$}}} open and n C n_{\text{{\color[rgb]{1,0,0}C$$}}} closed states that exceed the maximum number 2 ​ n O ​ n C 2n_{\text{{\color[rgb]{0,0.4,0}O$$}}}n_{\text{{\color[rgb]{1,0,0}C$$}}} of parameters are non-identifiable. We will present two derivations for this classical result and investigate non-identifiability further via a detailed analysis of the non-identifiable fully connected three-state model. Finally, we will discuss the implications of non-identifiability for mechanistic modelling of ion channels. We will argue that instead of designing models based on assumed transitions between distinct biophysical states which are modulated by ligand binding, it is preferable to build models based on additional sources of data that give more direct insight into the dynamics of conformational changes. keywords non-identifiability, aggregated Markov model, ion channel 1 Introduction Ion channels belong to the simplest components of living cells—in membranes that are usually impermeable to ions they open tiny pores which allow ions to enter or leave the cell. Many channels select which ions can pass through the pore but otherwise, ion channels exert little influence on the net flux of ions which is determined by a combination of the concentration gradient and the membrane potential i.e. the electrochemical gradient. Thus, at the most basic level, ion channels can be viewed as switches which are able to switch on or off fluxes of ions that would otherwise be blocked by the cell membrane. Despite, or maybe due to their simplicity which enables them to be flexibly integrated in various physiological systems, ion channels are involved in more or less all important physiological processes. These can be roughly summarised in three categories: 1. Regulating the intracellular concentration of a particular ion. 2. Regulating the charge within a cell i.e. the membrane potential. 3. Signal transduction. Ion channels belong to a class of cellular building blocks called membrane proteins. Although the function carried out by an ion channel can be described as one of a simple switch, the membrane protein which carries out this task is not at all simple at a biophysical level—to open a pore in the membrane, the channel protein needs to undergo one or more rearrangements of its complex three-dimensional structure—so-called conformational changes. A fundamental question in ion channel modelling is “Should the biophysics of the channel protein be reflected in a mathematical model describing the dynamics of the ion channel ?” We initially disregard this important question but will return to consider it after our detailed investigation of non-identifiability, see the Discussion, sec. 4 . We describe the dynamics \mathcal{D} of an ion channel as an alternating sequence of open and closed times, \mathcal{D}=\left(t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{1},t_{\text{{\color[rgb]{1,0,0}C$$}}}^{1},t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{2},t_{\text{{\color[rgb]{1,0,0}C$$}}}^{2},\dots\ \right) (Without loss of generality we assume that the first time observed is an open time but we could equally have started with a closed time). The times t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{1},t_{\text{{\color[rgb]{1,0,0}C$$}}}^{1},\dots are stochastic—when an ion channel opens, it is, in principle, not possible to predict, how long it will stay open. t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{1} t_{\text{{\color[rgb]{1,0,0}C$$}}}^{1} t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{2} t_{\text{{\color[rgb]{1,0,0}C$$}}}^{2} t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{3} (a) Stochastic ion channel dynamics \mathcal{D} (b) Continuous-time two-state Markov model Figure 1: (a) The dynamics \mathcal{D} of an ion channel is an alternating sequence of stochastic open and closed times t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{i} and t_{\text{{\color[rgb]{1,0,0}C$$}}}^{j} . (b) Such a sequence can be described by a continuous-time Markov model with one closed state C 1 and one open state O 2 . If the ion channel generates open or closed times of multiple characteristic lengths, additional open or closed states need to be added. A simple model for the alternating sequence of open and closed times \mathcal{D} is the two-state continuous-time Markov model which describes stochastic transitions between a closed state C 1 and an open state O 2 . A direct consequence of assuming that the dynamics \mathcal{D} can be described by a Markov model is that the open times t_{\text{{\color[rgb]{0,0.4,0}O$$}}}^{i} and the closed times t_{\text{{\color[rgb]{1,0,0}C$$}}}^{j} are exponentially-distributed. Indeed, the Markov property requires that if the model has been in a state S for a time t_{0} , the probability that it stays in this state for an additional duration \tau is independent from the time t_{0} that has already passed. More formally, \mathbb{P}(T>\tau+t_{0})=\mathbb{P}(T>t_{0})\mathbb{P}(T>\tau),\quad t_{0}\geq 0,\tau\geq 0 (1) or, equivalently, \mathbb{P}(T>\tau+t_{0}\,\vline\,T>t_{0})=\mathbb{P}(T>\tau),\quad t_{0}\geq 0,\tau\geq 0. (2) The exponential distribution is the only continuous-time distribution for which the “lack of memory” property ( 2 ) holds. A two-state Markov model can therefore just generate an alternating sequence of exponentially-distributed sequence of open and closed times. But exponentially-distributed open and closed time distributions, f_{\text{{\color[rgb]{0,0.4,0}O$$}}}(t_{\text{{\color[rgb]{0,0.4,0}O$$}}}) and f_{\text{{\color[rgb]{1,0,0}C$$}}}(t_{\text{{\color[rgb]{1,0,0}C$$}}}) , are insufficient for describing the dynamics \mathcal{D} of many ion channels—a simple counter-example is an ion channel that sometimes stays closed for a short time and sometimes for a considerably longer time. The most common approach for describing multiple timescales in the open and closed time distributions is to introduce multiple open and closed states. Intuitively, one might expect that each closed state (or open state) represents a different exponentially distributed timescale and each of these closed (or open) timescales is observed with a specific probability. Indeed, as calculating the sojourn-time distributions from a Markov model (with an arbitrary number of closed and open states) shows, this expectation turns out to be true—open and closed time distributions are mixture distributions of as many exponential components as there are open/closed states, respectively ( Colquhoun and Hawkes 1981 ) . Although the exponentials in the mixture distributions f_{\text{{\color[rgb]{0,0.4,0}O$$}}}(t_{\text{{\color[rgb]{0,0.4,0}O$$}}}) and f_{\text{{\color[rgb]{1,0,0}C$$}}}(t_{\text{{\color[rgb]{1,0,0}C$$}}}) can, in general, not be identified with individual states in the Markov model, it is tempting to interpret the set of open and closed states as a collection of different timescales, ranging from “slow” to “fast”. Another attractive property of a Markov model consisting of multiple open and closed states (a so-called aggregated Markov model with two classes) is that one might associate individual states of the model with different biophysical states of the channel protein—one well-known example is the de Young - Keizer model of the inositol-trisphosphate receptor (IP 3 R) ( De Young and Keizer 1992 ) . To give a simple example for this approach, consider a model that consists of a chain of multiple closed states that eventually transition to an open state, this is interpreted as transitions between various closed conformations which eventually lead to an open state i.e. a conformation in which the channel is open. It is no coincidence that this idea has been highly influential—it suggests that a model consisting of multiple open and closed states can represent more

Read the complete text