Short-term synaptic plasticitycomputational implications in the emergent behavior of neural systems

Resumen

In most classical frameworks in Neuroscience, neurons are considered to be the fundamental computational engines for information processing and coding, with the synapses treated as mere connections between neurons across which the information is transmitted. Such a traditional view, however, may turn out to be excessively simplistic, since many recent works indicate that synapses participate actively in the processing of information in the brain. In the last few years, for instance, it has been reported that the strength of synaptic connections may vary on short time scales depending on presynaptic activity. It has also been found that such variations can be used to process information in a nontrivial way. This finding indicates that, in addition to neurons, synapses may have an active role in neural computations. The possibility to have activity-dependent synaptic modifications on short time scales is usually known as short-term plasticity, and the synapses which display such behavior are called dynamic synapses. According to underlying biophysical process involved, there are two major mechanisms responsible for short-term plasticity: the so called short-term depression (STD) and short-term facilitation (STF). The former is responsible of the decrease of the postsynaptic response under repetitive presynaptic stimulation, whereas the latter induces an increment of the postsynaptic response for the same type of stimulus. The complete computational and functional implications of STD and STF are not clear yet, although these processes could be of vital importance in the processing of information in neural systems. Very recent studies have shown, for instance, that the presence of STD has a strong influence on the dynamics of neural systems, and is implicated in membrane gain control, maintenance of high activity states in the cortex, storage of information in attractor neural networks, detection of coincident signals, or the appearance of switching between different activity patterns in recurrent neural networks, which could be related with spontaneous voltage transitions in cortical areas. Most of these studies, however, do not take into account the effects produced by STF, which is also present in most of the brain structures analyzed in these works. This constitutes a highly relevant issue, because STD and STF have a priori opposite effects on the postsynaptic response, and taking into account STF could notably affect the behavior of a neural system with depressing synapses. Moreover, the consideration of both mechanisms together could reveal novel emergent phenomena caused by the interplay between STD and STF. Indeed, such interplay could help to explain several features of actual neural systems which remain far from being totally understood. Some of these features include the detection of weak signals over a broad range of network activity levels, the ability of neural circuitry to optimally store and retrieve information while maintaining an efficient processing of signals, or the level of irregularity observed even in highly synchronized neural dynamics, such as the heterogeneity of the duration of high activity states during series of up-down cortical transitions. In spite of its possible implications on all these phenomena, the study of the interplay between STD and STF in different neural systems has not been addressed in detail up to date. In this framework, the aim of this thesis is to investigate the role and implications of the interplay between short-term synaptic depression and facilitation on the computational properties of several neural systems of interest. In chapter 1, a general introduction of the main objectives of this thesis is provided, as well as some details concerning its structure and organization. This will help the reader to gain some understanding of the state-of-the-art before entering into more complicated considerations. In chapter 2 we present a brief physiological review of the nervous system. This review covers the biological aspects employed in the subsequent chapters, and provides some useful references for the interested reader. The chapter starts with an exposition of the main parts of the central nervous system, focusing on the particular case of the cerebral cortex. After that, some basic features of neurons are explained, and finally the main aspects of the synapses are reviewed, paying special attention to the biophysical mechanisms which induce the different short-term plasticity mechanisms. Such review is highly convenient to understand the effect of short-term synaptic mechanisms on the computational properties of different neural systems which, as we have already explained, is the goal of this thesis. In chapter 3 a review of some mathematical paradigms commonly used to model neural systems is presented, to complement the biological introduction of chapter 2. We briefly cover several neuron models of interest, starting from highly detailed neural paradigms (such as the Hodgkin-Huxley neuron model) and finishing with very simplified ones (such as the binary neuron). After that, some mathematical descriptions of the synapses are presented, including some of the models of short-term plasticity which will be employed in the following chapters. Finally, different approaches to model large neural populations are sketched out as well. Once the biological background and the methods have been introduced in chapters 2 and 3, we start to study the role of STD and STF in very simple neural systems. In particular, chapter 4 is concerned with detection of correlated inputs by simple neural circuits with short-term plasticity in noisy environments. More precisely, using a realistic model of depressing and facilitating synapses , we studied the conditions in which a postsynaptic neuron efficiently detects temporal coincidences of spikes which arrive from N different presynaptic neurons at a certain frequency. A numerical and analytical treatment of this system showed that: i) STF enhances the detection of correlated signals arriving from a subset of presynaptic excitatory neurons, and ii) the presence of STF yields a better detection of firing rate changes in the presynaptic activity. We also observed that facilitation determines the existence of an optimal input frequency which allows the best performance for a wide (maximum) range of the neuron firing threshold. This optimal frequency can be controlled by means of facilitation parameters. Finally, we showed that these results are robust even for very noisy signals and in the presence of synaptic fluctuations produced by the stochastic release of some molecules involved in synaptic transmission (the so called neurotransmitters). In chapter 5, we extended the analysis of chapter 4 by studying the detection of weak stimuli by spiking (integrate-and-fire) neurons in the presence of a certain level of noisy background neural activity affecting the postsynaptic response via dynamic synapses. Employing mean-field techniques as well as numerical simulations, we found that there are two possible noise levels which optimize signal transmission (such phenomena is referred here as bimodal resonance). This new finding is in contrast with the classical theory of stochastic resonance, which is able to predict only one optimal level of noise for the detection of weak signals. We found that the complex interplay between adaptive neuron threshold and activity-dependent synaptic mechanisms is responsible for this new phenomenology. Our results were confirmed by employing a realistic FitzHugh-Nagumo neuron model, which displays threshold variability within its own dynamics, as well as by considering more realistic synaptic models. We also supported our findings with recent experimental data of stochastic resonance in the human tactile blink reflex. Our next step was to extend the study to models of large neural populations. Concretely, in chapter 6 we studied, analytically and employing Monte Carlo simulations, the influence of the competition between several activity-dependent synaptic processes, such as STF and STD, on the maximum memory storage capacity in an attractor neural network. In contrast with the case of synaptic depression, which drastically reduces the capacity of the network to store and retrieve static activity patterns, synaptic facilitation enhances the storage capacity in different contexts. In particular, we found optimal values of the relevant synaptic parameters (such as the neurotransmitter release probability or the characteristic facilitation time constant) for which the storage capacity can be maximal and similar to the one obtained with static synapses, that is, without activity-dependent processes. We concluded that depressing synapses with a certain level of facilitation allow to recover the good retrieval properties of networks with static synapses while maintaining the nonlinear characteristics of dynamic synapses, convenient for information processing and coding. After the analysis of retrieval abilities of neural networks with dynamic synapses, which may be seen as steady state properties, we focused in the effect of STD on the dynamics of the activity of neural populations. In particular, in chapter 7 we addressed the study of the voltage transitions between up and down states observed in cortical areas in the brain, which constitute a paradigmatic example of complex coherent neural dynamics. We study this phenomenon via a biologically motivated stochastic model of up and down transitions. The model employed was a simple bistable rate model, where the synaptic current is modulated by short-term synaptic processes (such as STD) which introduce stochasticity and temporal correlations. A complete analysis of our model, both with theoretical approaches and numerical simulations, showed the appearance of complex transitions between high (up) and low (down) neural activity states, driven by the synaptic noise, with permanence times in the up state distributed according to a power-law. These results are in agreement with recent experimental observations in up and down transitions in cortical activity which indicate the onset emergence of criticality in the hopping dynamics between collective neural states. Finally, in chapter 8 the main conclusions of this thesis are presented, focusing on the role of the interplay between short-term depression and facilitation in the computational and functional properties of different neural systems at different levels of description. The possible implications of this interplay on several brain tasks and behavior, and also the future research lines that this thesis may suggest, are summarized as well. The thesis is written in english, although a brief summary in spanish is included, with the purpose to fulfill the requirements to obtain the degree of Philosophy Doctor in Physics with European level, as ruled by the regulations of the University of Granada. A list of publications of the author is also provided in the last pages of the thesis.