Login |

B5 - Genetic regulation of network dynamics in biophysical PFC/ HC models

Principal investigator(s):

RG Computational Neuroscience
ZI Mannheim
J 5
68159 Mannheim

School for Computing and Mathematics,
Plymouth University, Plymouth, UK


Projects within the BCCN:

The original aim of this project was to develop neuro-computational models and tools for studying the influence of psychiatric risk factors (genes, pharmacologically induced) on network dynamics. Toward this aim, we have taken a completely novel approach, a kind of ‘high-throughput’ computational pipeline that enables to develop and simulate neural models that are on the one hand strongly driven and constrained by physiological and anatomical data, yet efficient and fast to parameterize and simulate. It consists, in a first step, of a fully automated and very fast method for estimating single neuron models from standard slice electrophysiological protocols. Our procedure generates simple but physiologically validated, in the statistical sense of out-of-sample prediction error, cell models (Hertäg et al. 2012; essentially modifications of the AdEx model [originally proposed by Brette & Gerstner, 2005] that allow for closed-form expressions within a least-squared-error optimization scheme). This approach enables to convert large data bases of physiological recordings from various psychiatric animal models into populations of model cells in rather short time. Within the context of this project, we have obtained hundreds of recordings from various rodent prefrontal cortex layers and cell types (different classes of inter- and pyramidal neurons; Hertäg et al. 2012), from genetically modified animals (mGLuR3-KO, Cav1.2-KO, Cav1.3-KO), and under different pharmacological conditions (D1/D2 dopamine receptor stimulation). In a next step, these data, together with anatomical and synaptic information, were fed into prefrontal cortex network models. Both anatomically detailed full network simulation models were developed (Hass et al. 2016), as well as sophisticated mean-field descriptions harboring a variety of cell types, based on our derivations of (approximate) solutions of the Fokker-Planck equations for the AdEx model (Hertäg et al. 2014). We also analyzed a wealth of in-vivo electrophysiological data sets from our collaborators within this project (mainly groups of Dr. Jeremy Seamans, UBC Vancouver, and Dr. Chris Lapish, Indianapolis, as well as project B6), to compile a large set of in-vivo statistics from multiple spike train, local field potential (LFP), and in-vivo patch-clamp recordings that could be used for model validation (Quiroga-Lombard et al. 2013; Hass et al. 2016; see also D2). A surprising outcome here was that the full network model solely based on cells strictly parameterized by data from slice recordings and incorporating detailed anatomical information was able to reproduce a whole range of in-vivo spike train, LFP, and membrane potential statistics quantitatively without much further tuning (Hass et al. 2016). This framework will be used for characterizing genetic risk models and pharmacological manipulations both in terms of parameters of estimated cell models as well as with respect to attractor dynamics. Projects B5 and D2 have also supported model-based behavioral data analysis in BCCN project C8 where learning in a con-genetic animal model of depression (Richter et al. 2013) and in a schizophrenia risk gene animal model (CACNA1C) was analyzed.

Participating groups:

Key publications:

Hass J, Durstewitz D (2016) Time at the center, or time at the side? Assessing current models of time perception Curr Opin Behav Sci 8:238–244 .
(2016) A Detailed Data-Driven Network Model of Prefrontal Cortex Reproduces Key Features of In Vivo Activity PLoS Comput Biol. 12:e1004930 .
(2014) Analytical approximations of the firing rate of an adaptive exponential integrate-and-fire neuron in the presence of synaptic noise Front Comput Neurosci. 8:116. .
Durstewitz D, Seamans JK (2013) How Can Computational Models Be Better Utilized For Understanding and Treating Schizophrenia? Schizophrenia – Evolution and Synthesis, edited by Steven M. Silverstein, Bita Moghaddam and Til Wykes. MIT Press. .
Quiroga-Lombard CS, Hass J, Durstewitz D (2013) Method for stationarity-segmentation of spike train data with application to the Pearson cross-correlation Journal of Neurophysiology 110: 562-72 .
(2012) An approximation to the adaptive exponential integrate-and-fire neuron model allows fast and predictive fitting to physiological data Frontiers in Computational Neuroscience 6: Article 62 .
Hyman JM, Ma L, Balaguer-Ballester E, Durstewitz D, Seamans, JK equal contribution (2012) (2012). Contextual encoding by ensembles of medial prefrontal cortex neurons. . Proc. Natl. Acad. Sci. USA. 109: 5086-91 .
Hass J, Durstewitz D (2011) Models of dopaminergic modulation Scholarpedia 6(8): 4215 .