# From Atoms to Neuronal Spikes: A Multiscale Simulation Framework

## TL;DR

This paper introduces a multiscale simulation framework that connects molecular events in ion channels to neuronal activity, helping to understand disease mutations and drug effects.

## Contribution

A novel multiscale framework that couples molecular simulations with neuronal models to predict changes in membrane potential and spike activity.

## Key findings

- Disease-associated AMPAR variants significantly impact neuronal excitability based on multiscale simulations.
- Bidirectional coupling between ion channel Monte Carlo simulations and neuronal models alters membrane potentials and excitation states.
- Lipid membrane composition affects ion channel behavior and neuronal activity, as revealed by Monte Carlo simulations.

## Abstract

Understanding how
molecular events in ion channels impact neuronal
excitability, as derived from the calculation of the time course of
the membrane potentials, can help elucidate the mechanisms of neurological
disease-linked mutations and support neuroactive drug design. Here,
we propose a multiscale simulation approach which couples molecular
simulations with neuronal simulations to predict the variations in
membrane potential and neural spikes. We illustrate this through two
examples. First, molecular dynamics simulations predict changes in
current and conductance through the AMPAR neuroreceptor when comparing
the wild-type protein with certain disease-associated variants. The
results of these simulations inform morphologically detailed models
of cortical pyramidal neurons, which are simulated using the Arbor
framework to determine neural spike activity. Based on these multiscale
simulations, we suggest that disease associated AMPAR variants may
significantly impact neuronal excitability. In the second example,
the Arbor model is coupled with coarse-grained Monte Carlo gating
simulations of voltage-gated (K+ and Na+) channels.
The predicted current from these ion channels altered the membrane
potential and, in turn, the excitation state of the neuron was updated
in Arbor. The resulting membrane potential was then fed back into
the Monte Carlo simulations of the voltage-gated ion channels, resulting
in a bidirectional coupling of current and membrane potential. This
allowed the transitions of the states of the ion channels to influence
the membrane potentials and vice versa. Our Monte Carlo simulations
also included the crucial, so far unexplored, effects of the composition
of the lipid membrane embedding. We explored the influence of lipidic
compositions only using the Monte Carlo simulations. Our combined
approaches, which use several simplifying assumptions, predicted membrane
potentials consistent with electrophysiological recordings and established
a multiscale framework linking the atomistic perturbations to neuronal
excitability.