Altered oscillatory coupling reflects possible inhibitory interneuron dysfunction in Rett syndrome

Abstract

Background: Rett syndrome is a rare neurodevelopmental disorder caused primarily by pathogenic variants in the MECP2 gene, leading to lifelong cognitive impairments. To understand the broad neural disruptions in Rett syndrome, it is essential to examine large-scale brain dynamics at the level of neural oscillations. Phase-amplitude coupling-a form of cross-frequency interaction that supports information integration across temporal and spatial scales-is a promising candidate measure for capturing such widespread neural dysfunction. Phase-amplitude coupling depends on the coordinated activity of specific neuronal subtypes, and while multiple subtypes are implicated in different aspects of the Rett syndrome phenotype, their role in shaping large-scale oscillatory dynamics in Rett syndrome is not well understood. To investigate this, we utilized a multi-level approach, combining EEG recordings with computational modeling to identify alterations in phase-amplitude coupling in Rett syndrome and probe their underlying cellular and circuit-level mechanisms.

Methods: We recorded resting-state EEG from 38 individuals with Rett syndrome and 30 age- and sex-matched typically developing individuals. Phase-amplitude coupling was quantified: modulation index was obtained to determine coupling strength, and phase bias was assessed to examine the preferred phase of coupling. We characterized phase-amplitude coupling across all low and high frequency combinations and electrodes, as well as within canonical theta-gamma and alpha-gamma frequency pairs across four predefined cortical regions. Finally, we modeled a biophysically-constrained Layer 4 cortical network to propose a possible mechanism underlying changes to oscillatory dynamics.

Results: We found significantly stronger phase-amplitude coupling in Rett syndrome across widespread cortical regions and frequency pairs, with a pronounced increase in theta-gamma and alpha-gamma coupling in anterior, posterior, and whole-brain regions (P < 0.05). Individuals with Rett syndrome also exhibited a more positive alpha-gamma phase bias in anterior and whole-brain regions (P < 0.05). Biophysically constrained modelling demonstrated that reduced VIP-expressing interneuron activity alone could recapitulate the pattern of increased theta-gamma and alpha-gamma phase-amplitude coupling observed in Rett syndrome (P < 0.001).

Conclusions: These findings identify alterations in awake-state phase-amplitude coupling in Rett syndrome and propose a mechanistic link to VIP+ interneuron dysfunction. Elevated phase-amplitude coupling may serve as a promising biomarker of cortical dysfunction and a translational bridge from neural circuitry to clinically observable EEG signatures. By implicating VIP+ interneurons, our results open new avenues for testing interventions in preclinical models to identify potential novel therapeutic targets for individuals with Rett syndrome.

Keywords: EEG; Rett syndrome; VIP+ interneurons; neurogenetic disorders; phase-amplitude coupling.

Figure 1.. Instances of significant phase-amplitude coupling for RTT and TD independently across all possible frequencies and regions.

Topoplot of comodulagrams visualizing significant PAC frequency pairs in RTT (n = 38) and TD (n = 30) for all 18 electrodes of the 10–20 channel array. Each subplot represents one electrode, and all subplots share the same axes of low and high frequency filtered signal, as shown in the lower left corner. Significant PAC frequency pairs are color-coded for RTT (red) and TD groups (blue), with any overlapping significant coupling shown in orange. On every subplot, each (x, y) coordinate with colored shading represents an instance of significant coupling between the corresponding low frequency (x) and high frequency (y) signals. Significance was calculated using a cluster-based permutation test.

Figure 2.. Difference between RTT and TD in phase-amplitude coupling strength across all possible frequencies and regions.

Topoplot of comodulagrams visualizing PAC strength difference between RTT (n = 38) and TD (n = 30) for all 18 electrodes of the 10–20 channel array. Each subplot represents one electrode, and all subplots share the same axes of low and high frequency filtered signal, as shown in the lower left corner. On every subplot, each (x, y) coordinate represents RTT subtracted from TD strength of coupling between the corresponding low frequency x and high frequency y signal. Dark green shading corresponds to higher PAC strength in RTT than TD, whereas bright yellow shading corresponds to higher PAC strength in TD than RTT. Lighter green indicates areas of low PAC strength difference. White boundaries outline areas where coupling strength is significantly different between groups. Significance was calculated using independent sample t-tests with a threshold of .05.

Figure 3.. Theta-gamma and alpha-gamma PAC strength elevated in individuals with Rett syndrome.

Markers in each subplot signify a single individual’s normalized PAC strength for a given frequency pair and region (indicated on the top and left). TD participants (n = 30) are shaded in blue and RTT participants (n = 38) are shaded in red. Difference between groups was calculated using an independent samples t-test, where *** indicates P < 0.001, ** indicates P < 0.01, and * indicates P < 0.05. Group mean (horizontal lines) and standard error (vertical lines) are plotted.

Figure 4.. Alpha-gamma phase preference altered in Rett syndrome.

TD participants (n = 30) are shaded in blue and RTT participants (n = 38) are shaded in red. (A) Markers in each subplot signify a single individual’s theta-gamma or alpha-gamma phase bias values in four given regions. Difference between groups was calculated using a Mann-Whitney U test, where *** indicates P < 0.001, ** indicates P < 0.01, and * indicates P < 0.05. Group mean (horizontal lines) and standard error (vertical lines) are plotted. (B) Gamma amplitude plotted as a function of alpha phase in four given regions. Traces represent group averages, where amplitude means are plotted. Error bars represent standard deviation values. 0° corresponds to the peak of the low frequency, +90° corresponds to the falling phase of the low frequency, 180° or −180° corresponds to the trough of the low frequency, and −90° corresponds to the rising phase of the low frequency.

Figure 5.. Schematic of L4 network.

PYR represents excitatory stellate and/or pyramidal cells, PV represents fast spiking parvalbumin+ cells, SOM represents somatostatin+ cells projecting to L4, and VIP represents VIP+ cells. Each represents a population of cells of a given cell type. In isolation (without network connections), VIP+ cells produce delta/theta (2 – 4 Hz), SOM+ cells produce alpha, and the reciprocal interaction between PYR and PV+ cells produces gamma via PING (pyramidal cell – interneuron gamma). Noisy input is given to pyramidal cells to represent thalamic input and/or input from other cortical areas. Excitatory synaptic connections are represented by red arrows and inhibitory connections are represented by blue arrows.

Figure 6.. VIP+ cell hypoactivity alters local network dynamics but PAC remains significant in L4 model.

(A) Raster plots of simulations of L4 network under conditions of typical VIP+IN activity and when VIP+IN activity is lowered. (B) Filtered LFP signals show high (gamma in green) and low (theta in blue and alpha in red) frequency bands and their coordination under typical and low VIP+IN activity conditions. Significant PAC was calculated using a permutation test.

Figure 7.. Increased modulation of gamma by slower theta and alpha frequencies with loss of VIP+ cell excitability.

(A) Markers represent the modulation index of the theta-gamma or alpha-gamma PAC in the model LFP after 9s of simulation time. Blue markers represent typical VIP+IN activity conditions and red markers represent low VIP+IN activity conditions. n = 20 for each group. Difference between groups was calculated using an independent samples t-test, where *** indicates P < 0.001. Group mean (horizontal lines) and standard error (vertical lines) are plotted. (B) Modulation index of the theta-gamma and alpha-gamma PAC in the model LFP after 9s of simulation time after one interneuron type (SOM+, VIP+ or PV+) was removed from the simulation. An independent samples t-test was used to determine if there was an increase in the mean MI for a particular condition (loss of one type of interneuron) compared to the mean MI in simulations with typical interneuron activity levels (i.e., the “typical VIP+IN activity” level condition), where ** indicates P < 0.01 and *** indicates P < 0.001.