Building a biological hybrid biosensor using nonlinear system analysis of the CA1 hippocampal neural network

A new type of tissue-based biosensor for screening chemical agents that rapidly affect the nervous system is described. The biosensor is based on a novel quantification method of Short-Term Plasticity (STP) in the CA1 hippocampal system in vitro using random electrical impulse sequences as inputs and population spike (PS) amplitudes as outputs. This approach is more time-efficient than the conventional paired-pulse and fixed-frequency-short-train methods and provides a comprehensive model of STP with considerable improvement in prediction accuracy. The STP descriptors are the first and second order kernels of the mathematical expression of the nonlinearities of the neuronal network. The first order kernel is the mean of the PS amplitude, while the second order kernel describes the effect of previous electrical impulses on the amplitude of the current PS. The second order kernel (describing STP nonlinear dynamics), exhibited a facilitation peak between 25ms and 45 ms, a fast rising phase [0ms--30ms] before the peak, and a fast facilitatory relaxation phase after the peak, crossing to the inhibitory region around 100ms--200ms and returning to the baseline within 1600ms to 2000ms (memory extent), i.e., impulses that occurred after the return to the baseline had no effect on the amplitude of the population spike evoked by the present impulse. Moreover, the second order kernel is decomposed into nine Laguerre functions whose coefficients along with the first order kernel were used for classification purposes. The biosensor was tested using picrotoxin (100 muM), tetraethylammonium (4 mM), valproate (5 mM), carbachol (1 mM), DAP5 (25 muM), and DNQX (0.15, 1.5, 3, 5, and 10 muM). These chemical agents gave a different coefficient-profile, representing their specific signatures. The first order kernel and the Laguerre coefficients formed the input for a single layer perceptron neural network that was able to classify each tested compound into its respective class. With a larger library of tested chemical agents and more powerful classifying neural network, this screening biosensor can classify a wide range of compounds affecting the neuronal properties.