State-dependent ensemble mapping




Relevant for Research Area

C - Applications





Prof. Dr. Christian Leibold


The coordinated recruitment of neural ensembles is thought to underlie behaviourally relevant information processing in the brain. Individual neurons are thereby dynamically recruited into these ensembles. The resulting changes of the ensemble configuration are thought to be associated with learning, sensory experience, as well as changes of behavioural needs and states. Synaptic connectivity has been hypothesized to provide the anatomical underpinnings of (Hebbian) ensembles, i.e. connections between neurons belonging to one ensemble are stronger than to neurons outside the ensemble. However, the ´effective’ connectivity between cells – and so the ensemble configuration – can change depending on the network state. For example, the recruitment of interneurons by single action potentials of individual pyramidal neurons depends on the network state, thereby switching on and off disynaptic inhibition between pyramidal cells (Jouhanneau et al 2018).

Probing the rules by which neural ensembles form in different network states has been challenging due to the difficulty of probing functional connectivity at a large scale between individual cells during behaviour.

Here, we propose to overcome this challenge by applying holographic optogenetic stimulation of individual cells with simultaneous population calcium imaging of the local network to identify functional ´influence maps´ depicting the impact of the activation of individual neuron on the local network in different behavioural states (rest, walking, task engaged). This experimental approach will be complemented by computational approaches to detect state-dependent neuronal ensembles in the calcium imaging data using an unbiased approach for detecting sequence motifs (Chenani et al., 2019). These experiments will provide insights into how the effective connectivity between neurons in different behavioural states translates into the formation of dynamic neural ensembles.

Mice will be placed into different behavioural contexts, either ‘task free’ where they will be free to locomote or rest on a treadmill or engaged in a virtual reality task. Pupil diameter will be monitored throughout the experiment as an additional indicator of animal arousal state. We will detect neural ensemble activity based on recordings of spontaneous and task related cortical network activity in premotor cortex. To probe the effective connectivity between neurons, we will use short 5ms light pulses targeted to individual neurons and record the influence on all other cells of the network via calcium imaging. From these data, we will construct single cell influence maps during the different behavioural states. Based on previous literature, we hypothesize these influence maps to change depending on behavioural state. We will test if and how the coupling between neurons belonging to one ensemble changes depending on the network state.

In preparation of this proposal, we have already established the chronic 2–photon imaging of neuronal populations and the holographic optogenetic stimulation of individual cells or groups of neurons in the Diester lab (Fig 1 A). In addition, we are currently optimizing the experimental workflow for rapid creation of influence maps with cell segmentation during a short initial recording period and programmatic generation of arbitrary stimulation sequences with the extracted cell locations (Fig 1 B).

Ensemble detection methods will then be used to detect potential correlations between influence maps and neural activity patterns recorded in the unstimulated regime. The findings will be fundamental for formulating hypothesis for the planned DFG grant application, which will combine experiments and network modeling. Finally, we plan to obtain proof-of-principle data on whether ensemble targeted stimulation can affect the animal’s behavior.