When in Striatum, Avoid Thy Neighbour
Striatal network models with monotonically decreasing (top,left) or non-monotonical (bottom,left) connectivity profiles give rise to quite different neuronal activity patterns (right). Blue: no significant patterning, red: activated neuron clusters.
Computational models predict the spatial connection profile among nerve cells necessary to model striatal neuronal activity observed in behaving animals
In the simplest conception the brain contains two types of neurons: excitatory neurons that activate downstream targets and inhibitory neurons that suppress downstream targets. Neuronal networks in the neocortex are composed of mutually connected excitatory and inhibitory neurons. By contrast, networks in subcortical regions, that is, located under the neocortex, such as the basal ganglia and amygdala are exclusively composed of inhibitory neurons. These networks of purely inhibitory neurons are crucial for a variety of motor, cognitive and learning related functions. Moreover, brain diseases such as Parkinson's disease, Tourette's syndrome and Huntington's disease can be attributed to a dysfunction of inhibitory network interactions in the basal ganglia.
The striatum as the main input station of the basal ganglia plays an important role in shaping motor, cognitive and learning related functions of the brain. Advances in experimental methods have recently started to reveal spatial and temporal patterns of neuronal activity in the striatum in healthy and Parkinson's disease states. Among other things, these data show that neighbouring striatal projection neurons (SPNs) are activated together (within few milliseconds) when animals are randomly foraging or engaged in a goal-directed task.
This raises the question: how are the SPNs inter-connected to give rise to such a spatio-temporal neuronal activity structure? At present the spatial structure of SPN connectivity remains mostly unknown. “And if we assume that SPN connection probability decreases with distance in a monotonic fashion as happens in the neocortex, the emergence of such spatial neuronal assemblies becomes untenable because neighboring SPNs would inhibit each other” says the lead author Sebastian Spreizer. Therefore, a team of computational neuroscientists from Germany (Freiburg and Juelich) and Sweden (Stockholm) set out to identify what kind of connectivity would be most suited to generate the experimentally observed spatio-temporal patterns of neuronal activity in the striatum.
Spreizer and colleagues now could show that the striatum can only generate different types of neuronal activity as observed in experiments (see Figure) when the connection probability among SPNs varies non-monotonically with distance, that is, it first increases and then decreases with the distance between neurons (e.g. a doughnut shape). Importantly, the emergence of spatially compact neuronal assemblies requires that SPNs avoid or minimize connecting to their immediate neighbours.
When SPNs inter-connect according to such doughnut shaped profile, the striatum can exhibit a range of activity states, ranging from a spatially and temporally inhomogeneous state (asynchronous-irregular – AI) to a spatially homogeneous and stable state, often called winner-take-all (WTA) state. Spreizer and colleagues now showed that the richest repertoire of neuronal assemblies appears when the striatum operates in the transition zone between AI and WTA states. The WTA state appears pathological and resembles striatal activity observed in the Parkinson's disease state.
Thus, there are two clear predictions: first, we predict that SPNs avoid connecting to their close neighbours; second, we predict that in awake behaving animals, the striatum dynamics should operate in the transition zone between AI and WTA states. At least two different experimental studies provide preliminary support for the first prediction. Arvind Kumar, who led the research, says “The available experimental methods are sufficient to test both our predictions”, and hopes that these predictions will soon be put to the critical test.
Full article (open access):
Spreizer S, Angelhuber M, Bahuguna J, Aertsen A, Kumar A (2017)
Activity dynamics and signal representation in a striatal network model with distance-dependent connectivity. eNeuro.DOI: https://doi.org/10.1523/ENEURO.0348-16.2017
http://www.eneuro.org/content/early/2017/08/14/ENEURO.0348-16.2017
Contact:
Dr. Arvind Kumar
Dept. of Computational Science and Technology
KTH Royal Institute of Technology
Stockholm, Sweden
arvkumar@kth.se
Sebastian Spreizer
Bernstein Center Freiburg
University of Freiburg
sebastian.spreizer@bcf.uni-freiburg.de