NETFORM - Neuronal Network Formation through Reciprocal Interactions between Activity and Structure

Project A. Detailed neuronal morphology
Creating a detailed model of neuronal morphogenesis to study the impact of activity-dependent neurite outgrowth on neuronal morphology, and the influence of neuronal morphology on synaptic connectivity.
Creating a detailed model of neuronal morphogenesis to study the impact of activity-dependent neurite outgrowth on neuronal morphology, and the influence of neuronal morphology on synaptic connectivity.

Electrical activity modulates neurite outgrowth and neuronal morphology [21]; and neuronal morphology determines synaptic connectivity [40]. To understand quantitatively (1) how activity-dependent neurite outgrowth influences neuronal morphology, and (2) how neuronal morphology affects global network connectivity, we need a dynamical model of neuronal morphogenesis based on biophysically-inspired, activitydependent rules of neurite outgrowth. Such a model will be instrumental in interpreting cortical neuronal morphology and synaptic connectivity, and for understanding how synapse and morphological degeneration in various brain disorders [6,28,31,32,38,39] affect synaptic connectivity.

Neuronal morphology consists of a branched axon and a number of branched dendrites. Because synapse formation requires overlap between axons and dendrites, neuronal morphology is an important determinant of synaptic connectivity [11,33]. During development, neurons attain their morphology by neurite elongation and branching, mediated by growth cones at the ends of outgrowing neurites. Neurons do not develop their morphologies in isolation, but also in response to electrical activity generated when cells become connected (activity-dependent neurite outgrowth; [21]). Synaptic connectivity will crucially depend on the characteristics of axonal and dendritic arbors, such as their length, branching structure and coverage of space. In general, however, it is not well understood how (1) network activity modulates neuronal morphology, and (2) how morphological characteristics of axons and dendrites influence synaptic connectivity. Existing computational models either consider detailed, single-neuron morphogenesis without activity-dependent regulation [52,3,13], or consider activity-dependent network formation without detailed neuronal morphologies [48,51,8]. In our previous CLS program, we created a novel simulation framework (NETMORPH; [22]) for the developmental generation of large-scale neuronal networks with detailed neuron morphologies, but as yet without activitydependent regulation.

To extend NETMORPH, we will create a model of neuronal morphogenesis based on biologically inspired, activity-dependent rules of neurite outgrowth. One challenge is to find the level of granularity. As possibilities, we will consider (a) detailed compartmental modeling [16], in which processes underlying neurite outgrowth are modeled by differential equations; and (b) effective rules based directly on biological mechanisms [15]. In any case, in our model: (1) individual growth cones mediate neurite elongation and branching, which (2) are subject to resource [10,50] and transport constraints, since outgrowth requires cytoskeleton elements; (3) synapses are formed on the basis of proximity between axonal and dendritic branches (Peters’ rule; [27,33]); (4) neurite outgrowth is modulated by electrical activity generated when the cells become connected [48,51].
To study the influence of activity-dependent regulation on detailed neuronal morphology, we will consider relatively small networks (depending on computational demands), using simulation package NEURON [18] to model neuronal activity. Because outgrowth takes place on a different time scale than activity, and because NEURON cannot model growing neurons, we will iterate between a phase of morphogenesis and a phase of 5 activity dynamics, each time importing into NEURON the new morphologies and connectivity. The current version of NETMORPH can also be used to study how, in large networks, neuronal morphology determines global network connectivity.
For validating the model-generated neuronal morphologies with experimental data, we will use the methods from project C. Insights from project A on the relation between neuronal morphology and synaptic connectivity will be used in project C to determine overlap functions. The findings on the relation between detailed neuronal morphology and synaptic connectivity from projects A and C will guide the formulation of the more abstract rules necessary in project B.

Research Questions and Validation.
  1. Can our model produce realistic cortical neuronal morphologies?
  2. What is the contribution of activity-dependent outgrowth in producing realistic morphologies and their variation?
  3. Using synapse formation based on proximity between axons and dendrites, what are the characteristics, in terms of connection length distributions and small-world properties, of the global network connectivity? Do typical connection length distributions arise, with highest frequency of connections at a medium distance between cells [35,12]?
  4. Can connectivity patterns observed in autism (with relatively fewer long range connections; [6,31]) arise from random loss of synapses?
We will validate the outcomes of the model with our in-house date on neuronal morphologies and synaptic connectivity in (developing) cortical networks (e.g., Van Pelt, Canto, Uylings); literature data on synaptic connectivity in cortical microcircuits (e.g., [9]), including in brain disorders (e.g., [31]); and web-based data sets on cortical neurons such as NeuroMorpho.Org (, SenseLab (, Duke-Southampton archive (, Cell Centered Database (, NeuroDB (,Neuroscience Database Gateway (, and Neuroscience Information Framework ( Ultimately, our model is also expected to become a valuable tool for interpreting how genes and proteins underlying neurite outgrowth (genomics and proteomics data; Smit, CNCR) may affect neuronal morphology and synaptic connectivity.