Multiscale Modeling of the Nervous System-Development of Efficient Quasi Lossless Scale-Bridging Methodologies
Abstract
Understanding the function and dysfunctions of the nervous system has proven challenging, particularly due to the number of interdependent processes spanning multiple scales, both within the hierarchical (from molecular level to neuron to network to system) and temporal continuum (from microseconds for neurotransmitters release and diffusion to years for the development of pathologies). Some models may include very detailed mechanistic elements (e.g. Monte Carlo simulations), capable of shedding light on subtle molecular interactions and subcellular processes, but they are often limited to short hierarchical (i.e. spatial) and temporal ranges due to the computational load of the processes modeled. The other end of the granularity spectrum tends to oversimplify the biological substrate in an effort to provide higher level perspective, while attempting to simulate large network population observables. By contrast, we propose to focus our efforts on developing synthetic scale-bridging methodologies capable of finely characterizing dynamical properties of a system, while maintaining a rigorous ceiling on computational load. The proposed methodology will follow three sequential steps: (i) capture relevant mechanisms at a given scale using biological measureables and mechanistic understanding of the underlying biological processes; (ii) generate a functional abstraction of the system identified in step (i); this abstraction aims to be quantitatively and dynamically faithful while providing a significant computational advantage over the complexity of the original model as defined in (i); (iii) instantiate functional abstractions defined in step (ii) as foundational building blocks on which the multiscale platform is elaborated, thereby enabling simulations of broader range hierarchical and temporal scales. Using the steps described above, our research efforts will lead to the development of a synthetic framework that enables symbiotic combination of interchangeable mechanistic models of brain components, with abstract functional models capable of replicating biologically relevant complex nonlinear dynamics in a computationally efficient manner. This framework will allow the creation of an integrated in-silico model ranging from subcellular mechanisms to large network level. We then propose to apply and validate this model to study cholinergic modulation of hippocampal formation to shed light on, and counter organophosphorate-induced toxicity on hippocampal function. Toxicity has been established to rise from three distinct phases. The first cholinergic phase stems from excess of acetylcholine; this gives rise to the glutamatergic phase, associated with excess excitatory neurotransmission and subsequent exacerbated spiking, which then triggers a calcium-dependent phase, leading to further excitotoxicity and neuronal death. We propose to integrate these multi-temporal mechanisms using the methodology described above, leading to the creation of an in-silico platform for the identification of potential therapeutic agents (or combinations of agents) meant to disrupt this cascade of events and reduce overall toxicity.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 25, 2021
- Source ID
- W911NF2110091
Entities
People
- Jean Bouteiller
Organizations
- Army Contracting Command
- United States Army
- University of Southern California