Limits...
The interaction of intrinsic dynamics and network topology in determining network burst synchrony.

Gaiteri C, Rubin JE - Front Comput Neurosci (2011)

Bottom Line: We find that several measures of network burst synchronization are determined by interactions of network topology with the intrinsic dynamics of neurons at central network positions and by the strengths of synaptic connections between neurons.Surprisingly, despite the functional role of synchronized bursting within the pre-BötC, we find that synchronized network bursting is generally weakest when we use its specific connection topology, which leads to synchrony within clusters but poor coordination across clusters.Overall, our results highlight the relevance of interactions between topology and intrinsic dynamics in shaping the activity of networks and the concerted effects of connectivity patterns and dynamic heterogeneities.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychiatry, University of Pittsburgh Pittsburgh, PA, USA.

ABSTRACT
The pre-Bötzinger complex (pre-BötC), within the mammalian respiratory brainstem, represents an ideal system for investigating the synchronization properties of complex neuronal circuits via the interaction of cell-type heterogeneity and network connectivity. In isolation, individual respiratory neurons from the pre-BötC may be tonically active, rhythmically bursting, or quiescent. Despite this intrinsic heterogeneity, coupled networks of pre-BötC neurons en bloc engage in synchronized bursting that can drive inspiratory motor neuron activation. The region's connection topology has been recently characterized and features dense clusters of cells with occasional connections between clusters. We investigate how the dynamics of individual neurons (quiescent/bursting/tonic) and the betweenness centrality of neurons' positions within the network connectivity graph interact to govern network burst synchrony, by simulating heterogeneous networks of computational model pre-BötC neurons. Furthermore, we compare the prevalence and synchrony of bursting across networks constructed with a variety of connection topologies, analyzing the same collection of heterogeneous neurons in small-world, scale-free, random, and regularly structured networks. We find that several measures of network burst synchronization are determined by interactions of network topology with the intrinsic dynamics of neurons at central network positions and by the strengths of synaptic connections between neurons. Surprisingly, despite the functional role of synchronized bursting within the pre-BötC, we find that synchronized network bursting is generally weakest when we use its specific connection topology, which leads to synchrony within clusters but poor coordination across clusters. Overall, our results highlight the relevance of interactions between topology and intrinsic dynamics in shaping the activity of networks and the concerted effects of connectivity patterns and dynamic heterogeneities.

No MeSH data available.


Schematic of integrating dynamics into topology using cell-type heterogeneity. (A) Example network with “Hartelt” topology found in the pre-BötC (Hartelt et al., 2008), consisting of densely connected clusters (color-coded) with rare inter-cluster links. In our simulations, we place cells with different intrinsic dynamics at select nodes in the topology based on each node's betweenness centrality (magnitude indicated by node size). The specific ordering of cell types in the hierarchy of betweenness centrality can be manipulated to generate potentially different network behaviors. In the example shown here, cells with quiescent dynamics are located most centrally, intrinsically bursting cells are located at the middle tier of centrality, and intrinsically tonic cells are located at the lowest centrality nodes. The cells interact with each other via excitatory synapses, sometimes taking on new activity characteristics when so linked (indicated by the node shape). (B) Other network structures may produce different combined output while using the same distribution of cellular heterogeneity, as a pure function of their topology. For comparative purposes we test the synchrony properties of seven additional topologies (examples shown here) commonly employed in network studies or found in brain networks. The total number of connections, number of cells, and cell-type heterogeneity in these networks were exactly matched to those of the pre-BötC networks. However, graph properties differ substantially between topologies as shown by histogram of betweenness scores for three tested networks (far right).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3044261&req=5

Figure 1: Schematic of integrating dynamics into topology using cell-type heterogeneity. (A) Example network with “Hartelt” topology found in the pre-BötC (Hartelt et al., 2008), consisting of densely connected clusters (color-coded) with rare inter-cluster links. In our simulations, we place cells with different intrinsic dynamics at select nodes in the topology based on each node's betweenness centrality (magnitude indicated by node size). The specific ordering of cell types in the hierarchy of betweenness centrality can be manipulated to generate potentially different network behaviors. In the example shown here, cells with quiescent dynamics are located most centrally, intrinsically bursting cells are located at the middle tier of centrality, and intrinsically tonic cells are located at the lowest centrality nodes. The cells interact with each other via excitatory synapses, sometimes taking on new activity characteristics when so linked (indicated by the node shape). (B) Other network structures may produce different combined output while using the same distribution of cellular heterogeneity, as a pure function of their topology. For comparative purposes we test the synchrony properties of seven additional topologies (examples shown here) commonly employed in network studies or found in brain networks. The total number of connections, number of cells, and cell-type heterogeneity in these networks were exactly matched to those of the pre-BötC networks. However, graph properties differ substantially between topologies as shown by histogram of betweenness scores for three tested networks (far right).

Mentions: In addition to lattice, SW, and random networks, we consider scale-free networks to ensure that we achieve a thorough assessment of how topological features interact with the dynamics of network elements to determine the emergent behavior of these networks. In scale-free networks, as in SW networks, a small number of nodes are important in directing and controlling information flow. Scale-free networks, however, achieve low path-length through degree heterogeneity; they contain a small number of “hub” nodes that have many connections to the more common “provincial” nodes, which have far fewer connections (an example is shown in Figure 1B). The exact degree distribution for these networks follows a power-law distribution (or truncated power-law due to finite network size). While it is debated if individual neurons have SW, scale-free, or both types of connectivity (Srinivas et al., 2007; Bonifazi et al., 2009), large-scale brain networks tend to show both scale-free and SW characteristics (Achard et al., 2006; van den Heuvel et al., 2008; Wang et al., 2009a).


The interaction of intrinsic dynamics and network topology in determining network burst synchrony.

Gaiteri C, Rubin JE - Front Comput Neurosci (2011)

Schematic of integrating dynamics into topology using cell-type heterogeneity. (A) Example network with “Hartelt” topology found in the pre-BötC (Hartelt et al., 2008), consisting of densely connected clusters (color-coded) with rare inter-cluster links. In our simulations, we place cells with different intrinsic dynamics at select nodes in the topology based on each node's betweenness centrality (magnitude indicated by node size). The specific ordering of cell types in the hierarchy of betweenness centrality can be manipulated to generate potentially different network behaviors. In the example shown here, cells with quiescent dynamics are located most centrally, intrinsically bursting cells are located at the middle tier of centrality, and intrinsically tonic cells are located at the lowest centrality nodes. The cells interact with each other via excitatory synapses, sometimes taking on new activity characteristics when so linked (indicated by the node shape). (B) Other network structures may produce different combined output while using the same distribution of cellular heterogeneity, as a pure function of their topology. For comparative purposes we test the synchrony properties of seven additional topologies (examples shown here) commonly employed in network studies or found in brain networks. The total number of connections, number of cells, and cell-type heterogeneity in these networks were exactly matched to those of the pre-BötC networks. However, graph properties differ substantially between topologies as shown by histogram of betweenness scores for three tested networks (far right).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3044261&req=5

Figure 1: Schematic of integrating dynamics into topology using cell-type heterogeneity. (A) Example network with “Hartelt” topology found in the pre-BötC (Hartelt et al., 2008), consisting of densely connected clusters (color-coded) with rare inter-cluster links. In our simulations, we place cells with different intrinsic dynamics at select nodes in the topology based on each node's betweenness centrality (magnitude indicated by node size). The specific ordering of cell types in the hierarchy of betweenness centrality can be manipulated to generate potentially different network behaviors. In the example shown here, cells with quiescent dynamics are located most centrally, intrinsically bursting cells are located at the middle tier of centrality, and intrinsically tonic cells are located at the lowest centrality nodes. The cells interact with each other via excitatory synapses, sometimes taking on new activity characteristics when so linked (indicated by the node shape). (B) Other network structures may produce different combined output while using the same distribution of cellular heterogeneity, as a pure function of their topology. For comparative purposes we test the synchrony properties of seven additional topologies (examples shown here) commonly employed in network studies or found in brain networks. The total number of connections, number of cells, and cell-type heterogeneity in these networks were exactly matched to those of the pre-BötC networks. However, graph properties differ substantially between topologies as shown by histogram of betweenness scores for three tested networks (far right).
Mentions: In addition to lattice, SW, and random networks, we consider scale-free networks to ensure that we achieve a thorough assessment of how topological features interact with the dynamics of network elements to determine the emergent behavior of these networks. In scale-free networks, as in SW networks, a small number of nodes are important in directing and controlling information flow. Scale-free networks, however, achieve low path-length through degree heterogeneity; they contain a small number of “hub” nodes that have many connections to the more common “provincial” nodes, which have far fewer connections (an example is shown in Figure 1B). The exact degree distribution for these networks follows a power-law distribution (or truncated power-law due to finite network size). While it is debated if individual neurons have SW, scale-free, or both types of connectivity (Srinivas et al., 2007; Bonifazi et al., 2009), large-scale brain networks tend to show both scale-free and SW characteristics (Achard et al., 2006; van den Heuvel et al., 2008; Wang et al., 2009a).

Bottom Line: We find that several measures of network burst synchronization are determined by interactions of network topology with the intrinsic dynamics of neurons at central network positions and by the strengths of synaptic connections between neurons.Surprisingly, despite the functional role of synchronized bursting within the pre-BötC, we find that synchronized network bursting is generally weakest when we use its specific connection topology, which leads to synchrony within clusters but poor coordination across clusters.Overall, our results highlight the relevance of interactions between topology and intrinsic dynamics in shaping the activity of networks and the concerted effects of connectivity patterns and dynamic heterogeneities.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychiatry, University of Pittsburgh Pittsburgh, PA, USA.

ABSTRACT
The pre-Bötzinger complex (pre-BötC), within the mammalian respiratory brainstem, represents an ideal system for investigating the synchronization properties of complex neuronal circuits via the interaction of cell-type heterogeneity and network connectivity. In isolation, individual respiratory neurons from the pre-BötC may be tonically active, rhythmically bursting, or quiescent. Despite this intrinsic heterogeneity, coupled networks of pre-BötC neurons en bloc engage in synchronized bursting that can drive inspiratory motor neuron activation. The region's connection topology has been recently characterized and features dense clusters of cells with occasional connections between clusters. We investigate how the dynamics of individual neurons (quiescent/bursting/tonic) and the betweenness centrality of neurons' positions within the network connectivity graph interact to govern network burst synchrony, by simulating heterogeneous networks of computational model pre-BötC neurons. Furthermore, we compare the prevalence and synchrony of bursting across networks constructed with a variety of connection topologies, analyzing the same collection of heterogeneous neurons in small-world, scale-free, random, and regularly structured networks. We find that several measures of network burst synchronization are determined by interactions of network topology with the intrinsic dynamics of neurons at central network positions and by the strengths of synaptic connections between neurons. Surprisingly, despite the functional role of synchronized bursting within the pre-BötC, we find that synchronized network bursting is generally weakest when we use its specific connection topology, which leads to synchrony within clusters but poor coordination across clusters. Overall, our results highlight the relevance of interactions between topology and intrinsic dynamics in shaping the activity of networks and the concerted effects of connectivity patterns and dynamic heterogeneities.

No MeSH data available.