Abstract
The noradrenergic locus coeruleus (LC) is a crucial controller of brain and behavioral states. Activating LC neurons synchronously en masse by electrical or optogenetic stimulation promotes a stereotypical “activated” high-frequency cortical state. However, it has been recently reported that spontaneous LC cell-pairs have sparse yet structured time-averaged cross-correlations, which is unlike the high synchrony of en masse neuronal stimulation. This suggests the untested possibility that LC population activity may be made of distinct multi-cell ensembles each with unique temporal evolution of activity. We used non-negative matrix factorization (NMF) to analyze large populations of LC single units simultaneously recorded in the rat LC. Synthetic spike train simulations showed that NMF, unlike the traditional time-averaged pairwise correlations, detects both the precise neuronal composition and the activation time courses of each ensemble. NMF identified the existence of robust ensembles of spontaneously co-active LC neurons. Since LC neurons selectively project to specific forebrain regions, we hypothesized that individual LC ensembles produce different cortical states. To test this hypothesis, we triggered local field potentials (LFP) in cortical area 24a on the activation of distinct LC ensembles. We found four cortical states, each with different spectro-temporal LFP characteristics, that were robust across sessions and animals. While some LC ensembles triggered the activated state, others were associated with a beta oscillation-specific state or a reduced high frequency oscillation state. Thus – in contrast to the stereotypical “activated” brain state evoked by en masse LC stimulation – spontaneous activation of distinct LC ensembles can control a multitude of cortical states.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
↵9 Senior authors
A new figure (Fig. 3) was added to demonstrate that NMF detects the composition of LC ensembles and their activation times, whereas graph-theoretic time-averaged pairwise correlations used in prior work does not. Additionally, the writing has been modified throughout to emphasise how the present study relates to prior work.
