Computational models of rodent hippocampal nerve cells focusing on their morphology, excitability and function

Abstract

The nerve cells in the mammalian brain come in various shapes and sizes. They constitute a complex system emerging from a complicated interplay of biophysical principles. Functionally, they can be compared to a computing unit transferring input into useful output, leading ultimately to cognitive functions or behaviour. This processing of information is called the input-output (IO) function of a neuron. The focus of this dissertation is on the particular IO function of hippocampal neurons, their underlying structure (morphology) and their intrinsic mechanisms (ion channels). In order to study these nerve cells, computational models offer the advantage of disentangling the involved biophysical mechanisms and their functional effects in a controlled manner. Therefore, I implement biologically realistic computational models of hippocampal neurons to simulate their IO function in several major, complementary \textit{in silico} investigations.Initially, using detailed neuron models that include active ion channels and other dendritic non-linearities, I demonstrate that the neural IO function can be invariant even when the stimulated dendrites of the nerve cells show vastly different morphological structures and sizes. These results reveal a general principle called accordingly "dendritic constancy". Notably, the dendritic constancy principle can have important clinical implications for neurological diseases. For example, it has been suggested that morphological alterations lead to the concurrent increase in excitability of principal hippocampal nerve cells during Alzheimer’s disease (AD). However, in line with the dendritic constancy principle, I show that the dendritic remodeling in AD cells is likely a homeostatic mechanism to maintain the cell IO function and information flow. The simulations instead reveal, that other intrinsic (ion channels) and extrinsic mechanism modifications lead to the excitability increase observed in AD cells in a multi-causal manner. Finally, various expressions of underlying ion channels cannot only affect the altered, pathological behaviour but potentially result in an optimised IO function and information processing. For instance, hippocampal granule cells (GCs) are believed to convert similar inputs into dissimilar outputs (pattern separation) while using as little energy as possible. The findings in this thesis reveal that the experimentally validated GC model seems to be close to optimal among a population of random, but valid, GC models with different ion channel expressions for the simultaneous performance of pattern separation and economy. In summary, by applying computational models in this dissertation I uncover a relationship between the underlying structure and ion channels of various nerve cells and their IO function.