Inferring the direction of rhythmic neural transmission via inter-regional phase-amplitude coupling (ir-PAC)

Bijurika Nandi,Peter Swiatek,Mingzhou Ding,Bernat Kocsis

doi:10.1038/s41598-019-43272-w

Abstract

Phase-amplitude coupling (PAC) estimates the statistical dependence between the phase of a low-frequency component and the amplitude of a high-frequency component of local field potentials (LFP). To date PAC has been mainly applied to one signal. In this work, we introduce a new application of PAC to two LFPs and suggest that it can be used to infer the direction and strength of rhythmic neural transmission between distinct brain networks. This hypothesis is based on the accumulating evidence that transmembrane currents related to action potentials contribute a broad-band component to LFP in the high-gamma band, and PAC calculated between the amplitude of high-gamma (>60 Hz) in one LFP and the phase of a low-frequency oscillation (e.g., theta) in another would therefore relate the output (spiking) of one area to the input (somatic/dendritic postsynaptic potentials) of the other. We tested the hypothesis on theta-band long range communications between hippocampus and prefrontal cortex (PFC) and theta-band short range communications between dentate gyrus (DG) and the Ammon’s horn (CA1) within the hippocampus. The ground truth was provided by the known anatomical connections predicting hippocampus → PFC and DG → CA1, i.e., theta transmission is unidirectional in both cases: from hippocampus to PFC and from DG to CA1 along the tri-synaptic pathway within hippocampus. We found that (1) hippocampal high-gamma amplitude was significantly coupled to PFC theta phase, but not vice versa; (2) similarly, DG high-gamma amplitude was significantly coupled to CA1 theta phase, but not vice versa, and (3) the DG high-gamma-CA1 theta PAC was significantly correlated with DG → CA1 Granger causality, a well-established analytical measure of directional neural transmission. These results support the hypothesis that inter-regional PAC (ir-PAC) can be used to relate the output of a rhythmic “driver” network (i.e., high gamma) to the input of a rhythmic “receiver” network (i.e., theta) and thereby establish the direction and strength of rhythmic neural transmission.

Highlights

Neural information processing depends on interactions between ensembles of neurons
Amplitude-phase pairing in Phase-amplitude coupling (PAC) calculation, i.e., high gamma amplitude in the driver network with theta phase in the receiver network or vice versa, was calculated to examine whether it corresponded to known anatomical connections, and whether it matched the direction and strength identified by Granger causality when the dataset was appropriate for a GC analysis
In this study we tested the idea that inter-regional PAC (ir-PAC), namely, the high gamma amplitude of one local field potentials (LFP) and the theta phase of another LFP, can be used to infer the direction and strength of rhythmic neural transmission

Summary

Introduction

Neural information processing depends on interactions between ensembles of neurons. Being able to assess the patterns of neuronal interactions is essential for a better understanding of the cooperative nature of neuronal computation. We hypothesized that for two neural networks with a driver-receiver relationship, PAC calculated from two LFP recordings, referred to inter-regional PAC (ir-PAC) here, will be significant in one direction only, i.e., between high-gamma amplitude of the “driver” network, representing rhythmically modulated spike trains, and the phase of the low-frequency oscillation in the “receiver” network, representing the somatic/dendritic postsynaptic potentials in response to the driving spike train input (see Fig. 1 for a schematic illustration).

Results

Conclusion