Embedding living neurons into virtual networks.
Nawrot M P, Hehll U, Pistohl T, Schrader S, Brandt A, Heck D, Rotter S, Aertsen A
Neurobiologie und Biophysik, Institut füur Biologie III
Albert-Ludwigs-Universität, Freiburg
www.brainworks.uni-freiburg.de
While intracellular recordings in acute slices have provided important insightsinto the properties of single cortical neurons, the isolated condition of these neurons in the in vitro preparation strongly limits the investigation of their interplay with an active cortical network. To overcome this restriction, we present here two novel experimental approaches for embedding real cortical neurons into a virtual surrounding provided by large-scale neural network simulations. In a real-time experimental setup we incorporate a real cortical neuron in vitro into a neural network in virtu. The network input to the recorded neuron is provided by means of somatic current injection. The output spikes are detected and fed back into the simulated network. This approach is particulary suited for the investigation of network schemes, where the single neuron significantly influences the network dynamics. We devised a real-time simulation environment based on SYNOD [1] permitting network sizes of about 100-1000 neurons operating at a time resolution of 0.1-1ms. A complementary strategy involves two serial steps. First, we simulate a large network (~10^5 neurons) with an architecture that features an anatomically realistic local connectivity and matches the size of a cortical column [2]. In a consecutive step we employ current injection to provide a real cortical neuron with the integrated excitatory and inhibitory input as monitored in an individual model cell during the simulation. This allows us to investigate the output statistics of a real cortical neuron in response to quasi-realistic network input. Hybrid experiments of this kind may be a powerful new tool to help clarify the mechanisms underlying the computational processes in biological neural networks.
Supported by DFG (SFB505), GIF and BMBF-Monist.
[1] www. synod. uni-freiburg. de
[2] Mehring, Hehl, Kubo, Diesmann, Aertsen (2002) submitted
