Infraslow dynamics of cortical microcircuits

Neuroscientists seek to explain how brain activity gives rise to perceptions (e.g., recognising pictures, sounds and smells), movements (e.g., getting up from a chair), and memories (e.g., recalling your yesterday's whereabouts). These cognitive processes occur in a fraction of a second, so most neuroscience research is focused on neural signalling that happens on this and faster timescales. Some neural processes, however, happen on much slower timescales. Sleep is one such example: one cycle takes about 24 hours. In fact, neural dynamics is not limited to one or just a few specific timescales but has prominent features on all timescales - seconds, minutes and hours included.

Our preliminary work shows that individual brain nerve cells ('cortical neurons') spontaneously vary their activity over tens of seconds and minutes, and that there are marked differences even among neighbouring cells in the way this occurs. The functions and mechanisms of these slow changes received very little attention and are not well understood. The aim of this project is to relate the activity of individual cortical neurons to the activity of large ensembles of cortical neurons on timescales of tens of seconds to several minutes. This will give us new insight into cognitive processes that happen on these timescales, such as changes in attention, motivation, introspection, vigilance, as well as how these slow processes modulate the fast cognitive processes.

We will use high density arrays of miniature electrodes to record for up to 10 hours the individual activity of dozens of neighbouring nerve cells in different areas of the mouse cortex. This will allow us to understand how changes in activity of single neurons are related to changes occurring in their neighbours. We will also study how the very slow changes in activity depend on the neurons' location in the cortex and on the state of the animal, for example during waking and sleeping.

In the second part of the project we will examine how the slow changes in activity are generated. We will record and manipulate the electrical potential inside individual nerve cells. These experiments will reveal the mechanisms that regulate nerve cell activity, e.g., by 'resting' after a period of intense activity, and will tell us why different cells generate different slow behaviour.

Finally, we will develop statistical methods to analyse our complex experimental observations. Preliminary work suggests that this is best achieved using frequency domain approaches which rely on the so called Fourier transform that decomposes signals into sums of sinusoids of different frequencies.

This work will answer fundamental questions about nerve cell functions, but there is an additional important reason for wishing to understand the slow changes in activity of cortical neurons. This reason concerns 'functional magnetic resonance imaging' - the most advanced method for non-invasively observing neural activity, which is widely used in both science and medicine. Functional magnetic resonance imaging (fMRI) measures neural activity indirectly, via its effect on the blood supply to the brain, and is consequently limited to revealing only the slowest components of neural activity. As a result of fMRI's widespread use, much is known about slow changes in activity of cortical areas, and this information is beginning to be used for diagnosis and treatment of neurological conditions. What is sorely lacking, however, is an understanding of how individual nerve cells give rise to the high level activity patterns. Our study will help filling this gap in our knowledge, and thus provide a better understanding of the origins of fMRI signals.

Cortical dynamics in the 0.01-1 Hz band has not received much attention outside the fMRI community, and has thus been studied only at the mesoscale spatial resolution. We seek to understand the dynamics of individual neurons and microcircuits in this frequency band. Our recent experiments show that individual cortical neurons exhibit prominent and highly heterogeneous slow firing rate changes. We will investigate (1) how single cell activity relates to mesoscale infraslow cortical dynamics, (2) which cellular mechanisms contribute to it, and (3) which statistical methods can describe the dynamics of cortical populations across multiple timescales.

To examine the relationship between individual neurons and mesoscale infraslow dynamics we will use chronic silicon probe recordings to study how individual cells couple to their local population in the infraslow frequency band. Surprisingly, preliminary data shows that in a significant proportion of cells population coupling on fast time scales is unrelated to population coupling on slow timescales. We will build on the preliminary findings to consider the effects of brain state, as well as possible differences across cortical layers and areas.

To examine the cellular mechanisms that shape the infraslow dynamics of individual neurons we will use simultaneous silicon probe, LFP and intracellular recordings. We will test the hypothesis that intrinsic conductances and spike frequency adaptation in particular are primary factors in slow activity changes. We will also study how different cell types differ in their infraslow dynamics.

To have succinct statistical summary of neuronal activity across multiple timescales we will develop frequency domain models for characterising and modelling population spiking dynamics. We will also investigate hybrid time-frequency models, where fast dynamics is parametrised in the time domain, while slow dynamics is parametrised in the frequency domain.

The proposed work is at the level of fundamental science and its main impact is in the generation of essential knowledge about how the brain works.

Since the project will directly contribute to our understanding of the nature of cortical activity that drives resting state fMRI signals, it may have an impact on the different uses of fMRI. Cortical resting state fMRI signals are altered in a variety of neurological conditions, such as drug addiction, schizophrenia, autism spectrum disorders, and others. This project may therefore contribute to our understanding of the basis for these abnormal changes, influence clinical practices and lead to design of compensatory and therapeutic interventions. In the long run, improved understanding of the neural activity underlying fMRI signals might benefit the manufacturers of MRI scanners. The statistical tools that we will develop to study population activity might be useful for quantitative analysis of fMRI, particularly by radiologists.

The project falls under the Responsive Mode Priority "Data driven biology" because one of the three objectives of the project is to develop a new computational approach to describe neuronal population activity. The project also falls under the "Systems approaches to the biosciences" Priority because it entails a multidisciplinary approach that integrates elements of experimental neurophysiology, computational neuroscience and signal processing. Furthermore, both the experimental and the modelling parts of the project involve components on different hierarchical levels, starting from membrane potentials and spikes of individual neurons, up to population and LFP signals which represents the activity of hundreds or thousands of neurons.

The project will provide the postdoctoral research associate with a great opportunity to develop an extensive experimental and computational skillset, which will greatly enhance her/his future employment opportunity in both academic and industry sectors.