Self-Organization of Polymorphic Circuits in Cultured Neural Networks

一般来说，感觉和运动系统中的解剖神经网络具有各种信息处理功能。这就是所谓的“多态”网络概念。多态网络是由神经元之间的功能相互作用定义的嵌入多个电路的网络。为了理解大脑的功能，如感觉知觉、运动控制、学习或记忆，阐明多态电路的新兴原理是非常重要的。在本研究中，我们对多态电路的自组织进行了生理和模型研究，得到了以下结果：采用KYS振荡器作为神经单元，对电机系统的中央模式发生器（CPG）进行了建模。CPG的多态特性可以在不改变突触连通性的情况下，通过控制抑制后反弹和静息膜电位等内在神经特性而显现。细胞外液灌注生物源性单胺后，鼻涕虫中枢神经系统嗅觉网络表现出多态性。这表明单胺在生物感觉网络的多态特性的形成中起着重要的作用。在生物系统中，单胺通过细胞内Ca^<++>改变固有的神经特性。因此，我们通过计算模拟研究了细胞内Ca^<++>的网络动力学，并证明了网络中Ca^<++>的全局模式可以根据各种外部条件而出现。这些结果揭示了生物网络中多态特性是如何被控制的，并可能为神经网络的运行规律提供新的认识。

In general, an anatomical neural network in sensory and motor systems subserves various functions of information processing. This is so-called "polymorphic" network concept. A polymorphic network is a network embedded with multiple circuits which are defined by functional interactions between neurons. For undersatanding the function of the brain such as sensory perception, motor control, learning or memory, it is very important to clarify emerging principles of the polymorphic circuits. In this research, we have made physiological and modeling studies about the self-organization of the polymorphic circuits, and obtained following results.1. Using a KYS oscillator as a neural element, we modeled a central pattern generator (CPG) of the motor system. The polymorphic property of the CPG could be emerged by controlling intrinsic neural property such as postinhibitory rebound and resting membrane potential, without changing synaptic connectivity.2. Isolated olfactory network of slug's CNS showed polymorphic property by perfusing biogenic monoamines in extracellular solution. This suggested that monoamines play important roles in emerging the polymorphic property of a biological sensory networks.3. In biological systems, monoamines change intrinsic neural properties through the intracellular Ca^<++>. Therefore we studied a network-dynamics of the intracellular Ca^<++> with computational simulation, and demonstrated that global Ca^<++> patterns in a network could be emerged according to various external conditions.These results suggest how the polymorphic property is controlled in biological network, and might provide a new insight into rules for neural network operation.