An important question in neuroscience is how local activity can be flexibly and selectively routed across the brain network. A proposed mechanism to flexibly route information is frequency division multiplexing: selective readout can be achieved by segregating the signal into non-overlapping frequency bands. Here, in wild-type mice and in a transgenic model (3xTgAD) of Alzheimer's Disease (AD), we use optogenetic activation of the entorhinal cortex, concurrent whole-brain fMRI, and hidden Markov modeling. We demonstrate how inducing neuronal spiking with different theta frequencies causes spatially distinct states of brain network dynamics to emerge and to preferentially respond to one frequency, showing how selective information streams can arise from a single neuronal source of activity. This theta modulation mechanism, however, is impaired in the AD model. This work demonstrates that neuronal multiplexing is a sufficient mechanism to enable flexible brain network communication, and provides insight into the aberrant mechanisms underlying cognitive decline.
ResultsOscillations result from the dynamic interaction between cellular and circuit properties. At microscopic scales, oscillatory synchronizations are constrained by axon conduction and synaptic delays; at macroscopic scales, oscillatory synchronizations are thought to be influenced by the architectural properties of the brain network (for review: 8 ). This complex interplay between oscillatory patterns and brain network topology may, therefore, allow neuronal signalling to selectively propagate along distinct subdivisions of anatomical projections or circuits 6 , giving rise to multiple temporal and spatial scales 8 or information streams 4 . From a modeling perspective, it is possible to utilize the sequential nature of BOLD fMRI data in order to characterize transient network states in brain hemodynamic activity. HMMs are well-suited because they provide a time-point-by-time-point description of network activity at the single-trial level, facilitating the characterisation of how such responses may evolve across time 7 or in response to perturbations such as stimulation.While HMMs provide a novel way of interrogating information within fMRI data, their use has not been previously applied to species other than humans. Here, using HMMs, we show that spontaneous hemodynamic fluctuations in the mouse brain at rest are temporally organized in a series of states that mirror well-established resting-state networks 9-11 ( Fig. S1-3 , see Supplementary results). Furthermore, we show that these states are organized in a hierarchical fashion, similar to what has been previously observed in humans 7 . Together, these results show that spontaneous brain network activity in mice is characterized by similar principles of temporal organization as in humans, thus providing further evidence of how the mouse brain is a relevant model to causally study neuronal mechanisms of mammalian brain network dynamics.Single-area evoked neuronal spiking is sufficient to cause spatiall...