gms | German Medical Science

Multi-electrode array (MEA) neurochips are used to examine effects and toxicity of (potential) neuroactive compounds [1]. These biosensors can react to known and unknown substances [2]. Identified substances are stored in a substance database [3] and can be consulted when classifying unknown substances.

In an in vitro experiment approximately 500,000 cells of the frontal cortex of embryonic mice [4] are cultivated on a MEA neurochip [1], [5]. Circa 10,000 neurons and 90,000 glia cells of the total amount survive. The object was to simulate these experimental data and to compare the results with MEA data using statistical methods.

We developed a pulsing neuronal model following the Glauber dynamics [6], [7]. Our model INEX (inhibitory-excitatory) is a cellular automaton whose cells represent neurons with two possible states: ON or OFF. The binary model should show several characteristics: Firstly, neurons are active without external input or stimulus as observed in experiments. Secondly, noise is observed. Thirdly, synapses can be either excitatory or inhibitory and last bursts, i.e. cascades of action potentials, as a characteristic phenomenon of frontal cortex neuronal network cultures shall be generated [8]. In order to simulate these properties we assume that the spikes, i.e. action potentials, obey an inhomogeneous Poisson distribution [9]. The inhomogeneity of the neuronal activity is realized by inhibitory or excitatory synapses of varying strength. The corresponding parameters are called weights. Spike time history is added, i.e. the probability of spike occurring increases following a spike in the previous time slice. We used a sparsely connected network with 1,000 neurons, i.e. 800 excitatory neurons and 200 inhibitory neurons [10] following Dale’s principle [11]. Each of the 1,000 neurons is connected to approximately 100 other neurons [12]. To simulate the addition of an inhibitory substance, like bicuculline, to the neuronal network, we reduced the excitatory weights.

From the generated 1,000 spike trains 20 were chosen randomly and compared to 20 randomly chosen MEA neurochip recordings of frontal cortex tissue of embryonic mice after 28 days in vitro. For the comparison spike and burst describing features were calculated [3]. Additionally the spike rate histogram was plotted.

The results of the simulation show, that spike and burst rate of the model and of MEA experiments correspond which is also demonstrated in the spike histogram. Therefore, the INEX model shows potential to simulate data as observed in experiments with MEA neurochips.