GRANULAR NEURAL MODELS, FUNCTIONAL CHARACTERISTICS AND CELL NEUROBIOLOGICAL SUBSTRATE
-
1
University of Granada, Department of Computer Architecture on Technology, CITIC, Spain
-
2
University of Granada, Department of Biochemistry and Molecular Biology I, Spain
Cerebellar granule cells are the smallest and most abundant neurons in the mammalian brain (more than half of the neurons in the brain are granular cells). The cerebellum is critical for sensorimotor control and non-motor functions including cognitive and emotional processes such as attention, language, emotional behaviour, sleep and even non-somatic visceral responses. Cerebellar lesions contribute to pathological syndromes such as autism, schizophrenia and ataxia.
The granule cells regulate information transfer along the major afferent system to the cerebellum. Several characteristic firing patterns have been reported in the granule cells, such as spike bursts or resonant firings (enhanced response to a preferential input frequency in the theta-frequency band) (D’Angelo et al., 1998). These firing features are related to synchronization, rhythmicity and learning at the cerebellum (D’Angelo et al., 2001).
The usage of pharmacological drugs in the framework of electrophysiology experiments allows the development of complex neuron models that reproduce in detail the neural activity recordings. This is the particular case of the granule cells in the cerebellar input layer. Neurophysiology allows discovering characteristics that describe granule neurons dynamics, embedding a combination of channels and specific interneuron synapses in the granular layer. All this experimental knowledge has allowed the creation of complex conductance-based neuron models mimicking the behaviour of real neurons with great accuracy that can be simulated in detail (D’Angelo et al., 2001). These models can be considered rather detailed models requiring a large number of differential equations to be solved at each time-step, making its simulation computationally expensive. The simulation of millions of these neurons (at this level of detail) is still far from being possible. Therefore, these inherent neural dynamics such as resonance frequency are usually ignored in most of efficient computation models.
Simplifying these phenomenological neuron models but identifying and keeping their functional characteristics is a complex process but allows a dramatic reduction of the computational cost in simulating each neuron. The reduction of the dynamics of connected neuron populations (in terms of number of differential equations and parameters) to a much simpler neuron model with similar dynamics is obtained optimizing its parameters, given the existence of a cost function or a fitness function.
Therefore, from detailed models which involve small scale simulations to simplified models that allow large scale simulations, and from these neural nets to embodied simulations of behavior (for instance using robotic agents with embodied neural systems). Simulations are multiscale: from a single neuron with certain channels or specific synapses, to complex neural networks of millions of neurons. In sílico simulators as NEURON, NEST or SpiNNaker allows dealing with neural models at different abstraction levels (molecular level, multicompartimental models, point neuron level, etc). Evaluating the behavioral impact of the neurological substrate (cell characteristics, network topology, adaptation mechanisms) in a multiscale framework allows discovering cell functional characteristics behind awareness, memory, stress, etc; and addressing diagnosis and treatment of neurobiological diseases.
We use advanced models to automatically simplify neural models down to the level of point neurons, but still capturing relevant functional features in terms of neural response spiking patterns to specific stimulus. For this purpose, we start from electrophysiology recordings or simulations of cell models defined at high levels of detail (molecular level). We define the target features (dynamics at cell level) that we aim to reproduce in the different experimental set ups (such as time to first spike, inter-spike-interval, Input-Frequency curve, etc). Be define a simplified neural model with certain parameters that can be adjusted towards reproducing specific neural dynamics (we also define the working ranges of these “free parameters”). Finally, we apply advanced optimization methods to optimize the simplified model. In this framework, the “optimization process” leads to the simplified neuron model parameters that lead to a better matching of the target cell features when we compare the simplified cell simulation with the detailed cell simulation (or even electrophysiological recordings).
In conclusion, we study the neurobiological substrate from high level of detail models reproducing the experimental results to behaviour and high level computation principles. We explore how to simplify detailed complex neuron models (usually developed at neurophysiology labs) to computationally efficient models but including these inherent neural dynamics. This is achieved studying cell properties, network topologies and adaptation mechanisms with respect to their functional role. In the future, the impact of these features at system level simulations will also be studied and how they complement synaptic plasticity mechanisms and neural network features. Long term applications can be addressed in the framework of neural rehabilitation schemes, reverse engineering of smart computation primitives, efficient computing and acting principles in biologically relevant tasks, understanding neurobiological disorders and thus facilitating treatment of related mental disorders. So we focus on studying specific functional characteristics of granule cells to their behavioural impact at a system level and medical implications. This represents a gap that remains to be researched.
Acknowledgements
This work has been partly supported by the European Union in the framework of the Human Brain Project, NeuroPAC and the “Plan Estatal de Investigación Científica y Técnica e Innovación 2013-2016” from the Spanish National Ministry of Economy and Competitiveness (MINECO) in a grant to M. M.
References
D’Angelo, E., De Filippi, G., Rossi, P., and Taglietti, V. (1998). Ionic Mechanism of Electroresponsiveness in Cerebellar Granule Cells Implicates the Action of a Persistent Sodium Current. J. Neurophysiol. 80, 493–503.
D’Angelo, E., Nieus, T., Maffei, A., Armano, S., Rossi, P., Taglietti, V., et al. (2001). Theta-frequency bursting and resonance in cerebellar granule cells: experimental evidence and modeling of a slow k+-dependent mechanism. J Neurosci 21, 759–770. doi:21/3/759
Keywords:
Cerebellum,
granule cell,
optimization algorithms,
simulation,
computational modeling
Conference:
The Cerebellum inside out: cells, circuits and functions
, ERICE (Trapani), Italy, 1 Dec - 5 Dec, 2016.
Presentation Type:
poster
Topic:
Cellular & Molecular Neuroscience
Citation:
Marín
M,
Garrido
J,
Saéz-Lara
M and
Ros
E
(2019). GRANULAR NEURAL MODELS, FUNCTIONAL CHARACTERISTICS AND CELL NEUROBIOLOGICAL SUBSTRATE.
Conference Abstract:
The Cerebellum inside out: cells, circuits and functions
.
doi: 10.3389/conf.fncel.2017.37.000015
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
11 Nov 2016;
Published Online:
25 Jan 2019.
*
Correspondence:
Miss. Milagros Marín, University of Granada, Department of Computer Architecture on Technology, CITIC, Granada, Spain, milmarina02@gmail.com