Background
Fragile X syndrome (FXS) is a neurodevelopmental disorder resulting from silencing of the fragile X mental retardation gene (
FMR1) on the X chromosome, leading to reduced production of Fragile X Mental Retardation Protein (
FMRP) [
1] that causes atypical brain development and function. Studies in
fmr1 knockout (KO) mice have shown enhanced activity of metabotropic glutamate receptors [
2] and reduced GABAergic transmission [
3]. These alterations are believed to cause an imbalance favoring excitation over inhibition in brain neurophysiology [
3‐
5].
Neurophysiological studies can clarify the functional brain consequences of neurochemical and neuroanatomic changes in FXS.
fmr1 KO mice have abnormally high synchrony of neocortical network activity and a threefold higher neuronal firing rate during Up states [
6,
7].
fmr1 KO mice have also shown increased EEG responses to auditory stimuli via in vivo recordings [
8‐
10]. Similarly, enhanced auditory event-related potential (ERP) responses (e.g. N1, P2) and reduced response habituation have been reported in FXS patients [
11‐
14].
Given the model of a neurophysiological imbalance leading to heightened neural excitability and the increased prevalence of seizures in FXS patients and
fmr1 KO mice, and with due consideration of the challenges integrating knowledge from intracranial recordings and clinical data, it is noteworthy that few EEG studies of resting brain function have yet been conducted with FXS patients. To our knowledge there has only been one quantitative study of resting state EEG in FXS presented in two reports [
15,
16]. Excessive resting state theta and reduced alpha power in FXS were reported, as well as decreased connectivity in alpha and beta bands but increased connectivity in the theta band. While informative, there were certain limitations to that study. First, the sample size was small (8 FXS patients). Second, the study investigated activity under 50 Hz, which excludes a significant component of the gamma frequency band (30–80 Hz) [
17‐
19]. This limitation is important because gamma band power reflects the level of high frequency spontaneous neural activity and is of special interest for FXS in light of
fmr1 KO mouse studies that have identified abnormalities in fast-spiking inhibitory GABAergic interneurons [
7,
20] which are critical generators of gamma power in cell populations [
21]. Third, functional connectivity analysis was done using only 28 electrodes; a dense electrode montage can better capture the full pattern of functional connectivity across the neocortex. Fourth, the study was insufficiently powered to identify correlations between resting state oscillatory abnormalities in FXS and measures of clinical symptom severity.
Alpha rhythms are the most dominant oscillation during the resting state and play an inhibitory role in information processing systems [
22]. Theta rhythms also reflect top-down inhibitory and organizational influences especially during higher cognitive activity [
23], and altered theta-gamma coupling has been linked with cognitive dysfunction in
fmr1 KO mice [
24]. As increased gamma activity is believed to be linked to increased neural excitability, examining the relationship of alpha and theta band activity with gamma band activity might provide mechanistic system-level understanding about the altered balance between excitatory and inhibitory activity.
The aim of the present study was to investigate resting state EEG activity in FXS patients focusing on resting EEG power—specifically gamma band power, functional connectivity, and gamma coupling. We hypothesized that resting state EEG power in FXS would be enhanced in both low- and high-frequency bands (theta and gamma) but reduced in middle range frequencies (alpha) relative to healthy controls. Second, we predicted that functional connectivity in FXS would be reduced in long-range connections but increased in short-range connections dominated by gamma band oscillations relative to controls. Third, we predicted that FXS patients would show reduced alpha-to-gamma coupling consistent with reduced top-down alpha-related inhibition on local cortical excitability in sensory systems. Fourth, we hypothesized that altered resting EEG power spectra in FXS individuals would be correlated with social and sensory processing difficulties.
Discussion
This case-control study investigated multiple aspects of brain system function in the largest sample of non-epileptic, full mutation FXS patients studied to date with quantitative dense-array resting-state EEG. Abnormalities were evident across measures of spectral power, functional connectivity, and cross-frequency amplitude coupling and were consistent with predictions based on the
fmr1 KO mouse electrophysiology [
7,
20,
24]. First, alterations in gamma band activity involved increased relative gamma power and an increased coherence of gamma band activity across nearby electrodes. This pattern is consistent with an imbalance of excitatory over inhibitory activity in FXS. The associations of electrophysiological alterations with abnormalities in social function and sensory sensitivities provide the first evidence of the clinical relevance of quantitative EEG findings in this population. Secondly, relative to controls, individuals with FXS showed reduced gamma amplitude coupling with upper alpha band activity but increased coupling with theta activity. As alpha and theta band activity are believed to exert top down inhibitory and regulatory modulation of sensory systems, these observations provide novel evidence that hyperexcitability in sensory cortex involves not only altered local circuit dysfunction in the interaction of interneurons and pyramidal cells as demonstrated in the fmr1 KO mouse [
7], but also alterations in the long distance functional connectivity from association cortex and thalamus to sensory cortex known to regulate local circuit excitability. While there was better group separation on the amplitude coupling indices, clinical correlations were significant only with the power of resting gamma alterations. This pattern of findings suggests that variable disease-related disturbances and compensatory adaptations may leave FXS patients with a net level of residual neural hyperexcitability reflected in elevated gamma power that determines important aspects of their level of neurological and functional disability. These findings provide new mechanistic understanding of cortical hyperexcitability in FXS, involving increased high-frequency local circuit activity that varied in relation to what appear to be abnormal and compensatory long-distance functional connectivity. They also suggest a potential utility of resting state EEG alterations as translational biomarkers for clinically relevant aspects of FXS biology for identifying individual FXS patients likely to benefit from treatments aimed at reducing neocortical hyperexcitability and for tracking those effects.
The U-shaped pattern of relative EEG power in individuals with FXS (Fig.
2) is similar in form to one we described previously in autism [
39]. Compared to healthy age-matched control participants, FXS patients showed enhanced power in lower (theta) and higher (gamma) bands, but reduced power in intermediate low and high alpha bands. Elevated theta and reduced alpha power have been reported previously in FXS [
15], but the enhanced power in the gamma band and its clinical relevance have not been previously described. This enhanced gamma power is consistent with studies of
fmr1 knockout mice demonstrating heightened neuronal excitability related to alterations in input to fast-spiking inhibitory interneurons that synchronize and control high-frequency gamma band neural activity [
4,
40]. A recent study in wild-type rats showed that enhanced gamma oscillation was observed when NMDA-receptor blockade was established [
41]. NMDA-receptor hypofunction has been reported in several studies with
fmr1 knockout mice [
42‐
44] and thus may be one contributing factor for the gamma band alterations observed in the present study.
Our functional connectivity analyses revealed reduced long-range functional connectivity in the alpha and beta bands, but enhanced shorter-range connectivity in the gamma band in FXS. The decreased functional connectivity in the alpha and beta bands parallels previous findings in FXS [
16]. We did not however observe the increased connectivity within the theta band that has been reported, though we did observe increased theta gamma coupling. For the first time, we report increased connectivity in gamma band activity in FXS patients. A previous fmr1 KO mouse study examined cross-frequency theta-gamma coupling, but during a cognitive task and when computed in the same hippocampal electrode [
24]. While related to focus of the present study, the differences in species, in rest vs task performance situations, and within rather than across distant brain sites are differences that future work will need to resolve to integrate the observations. Further translational work bridging preclinical and clinical findings is needed to more directly link clinical neurophysiological findings to observations seen in fmr1 KO mice.
Our observation of increased spatial extent of coherent gamma band activity across more distant electrode pairs suggests an increased cortical spread of neural excitability paralleling effects observed in slice preparation data but on a far greater spatial scale [
7]. It aligns with observations of increased neural synchrony observed in
fmr1 knockout mice during sleep and quiet wakefulness [
6]. The pattern of reduced long range connectivity in the alpha and beta bands suggests that reduced top-down inhibitory regulation of neocortical sensory systems may contribute to increased neural excitability of sensory cortex in FXS.
Given the alterations we observed in EEG power and functional connectivity in FXS, we investigated the coupling between low frequency band activity that was abnormal (alpha and theta activity) and high-frequency gamma band activity both within and across electrodes. We did this to determine whether the pattern of findings seen in gamma power and connectivity was related to a disruption in top-down modulation. Based on EEG data, it is not possible to determine whether a local circuit dysfunction is causing or resulting from the altered pattern of reduced long-distance functional connectivity in FXS. However, previous studies have shown that low frequency oscillations (e.g., alpha, theta) provide top-down inhibitory and modulatory influences in large, distributed neural networks, whereas fast oscillations (e.g., gamma) at rest are more related to neurophysiological tone in local networks [
45]. Thus, the reduced alpha power, connectivity, and amplitude coupling in FXS may represent a failure of top-down modulation provided by alpha band input that could reduce gamma power in sensory systems. Previous studies have characterized the functional role of alpha band activity as actively inhibiting the processing of sensory information at rest and when environmental cues are not task relevant [
22]. Posterior alpha has been reported to provide top-down control especially in visual attention studies, and thalamus has been identified as an important source of cortical alpha oscillations [
46]. Both simulation and experimental studies have demonstrated that lower frequency oscillation rhythms (e.g., alpha, theta) can sustain long-range synchronization [
45,
47], while synchronous activity in higher oscillation rhythms (e.g., gamma) declines more rapidly with increasing distances [
48]. As a result, slower oscillations are better suited for top-down modulation by synchronizing and organizing activity across different brain regions [
45]. This is consistent with findings from a nonhuman primate study of V1 and V4 showing that gamma rhythms propagate in a feedforward fashion from early to higher level visual processing regions, whereas alpha rhythms propagate in a feedback fashion to primary visual cortex [
49].
In contrast to the reduced upper alpha-to-gamma coupling, FXS showed strong theta-to-gamma coupling with atypical theta connectivity being related to lower levels of gamma power (Fig.
3). Thus, while alpha power and coupling were reduced, theta power and coupling were increased in FXS, indicating a fundamental alteration in the pattern of cortico-cortical connectivity that supports top-down modulation in sensory systems. Further clinical and preclinical studies are needed to fully clarify the meaning of this novel observation, but one possibility is that in the context of reduced inhibitory modulation of alpha oscillations on gamma band activity, a second long-distance regulatory circuitry operating in the theta band may be relied upon to downregulate high-frequency neural activity in the gamma band, a compensation which is only partially and variably successful given the observation of clinically relevant increased gamma power in FXS. As theta power phasically synchronizes neural activity across brain regions to support different types of higher level cognition [
50], a tonic activation to suppress sensory hyperexcitability reflected in increased theta power at rest might limit that phasic modulation and thereby contribute to the severe intellectual limitations often seen in FXS. Reduced alpha power might also contribute to the severe intellectual limitation since alpha activity has been reported to positively correlate with cognitive parameters [
51].
The functional role of theta oscillation is related to its neural sources in the prefrontal and anterior cingulate cortex (ACC) [
52], which play important roles in inhibitory control of behavior, behavioral flexibility, and error monitoring [
53,
54]. During tasks requiring top-down inhibition of attention or behavior, microelectrode recordings in superficial cingulate layers exhibit strong task-related theta activity [
55]. In addition, intermittent theta-burst stimulation has been shown to increase cortical inhibition in rat neocortex by reducing parvalbumin expression in fast-spiking interneurons [
56], and stimulation in theta frequency bands increases expression of GABA precursors in inhibitory cortical systems [
57]. Our gamma amplitude coupling associations may represent neural system factors that in vivo impact local circuit neurophysiology known to be altered in FXS, such as alterations in metabotropic glutamate receptor (mGluR) activation believed to be a cause of neuronal hyperexcitability in FXS. Inhibitory interneurons in mouse neocortex have been reported to fire in the theta frequency during mGluR activation [
44,
58]. In addition,
FMRP is highly expressed in the hippocampus [
59], and long-term potentiation (LTP) elicited by theta burst stimulation has been reported to be impaired in the CA1 hippocampal subfield in
fmr1 KO mice [
60].
Interest in systems biology alterations to complement understanding of local circuit pathology may be important not only for comprehensive models of pathology in FXS, but because individual variability in system-level modulatory factors may contribute to the wide range of clinical phenotypes seen even in patients with full mutation. This variability might explain the inconsistent treatment response to drugs targeting mGluR and other mechanisms aiming to reduce neuronal hyperexcitability, which have had more consistently positive effects in animal models. In this context, our findings not only provide new mechanistic understanding of FXS but also suggest that EEG studies of FXS may provide biomarkers for delineating disease heterogeneity and predicting and tracking response in human and mouse models to drugs targeting neuronal hyperactivity. Such approaches are urgently needed to advance drug development and personalized medicine for FXS patients. Our observation that quantitative EEG alterations were related to the severity of social communication and sensory reactivity problems supports the potential clinical utility of this approach.
This study has certain limitations, including the wide age range (12–57 years old) and the fact that younger children were not assessed. Although FXS and control groups were age-matched and no significant age effects were observed in the data, studies with younger populations remain an important target for future research. Other effects such as sex differences and medication effects were not statistically significant, but further research is needed to address those issues. Third, our study did not compare resting state abnormalities in FXS directly with other developmental disabilities to establish specificity of deficits to FXS relative to general effects of intellectual or developmental disability. While the parallel findings from our study and preclinical work in fmr1 KO mice suggest a relevance to FXS, more research is needed to determine whether similar findings might be seen in the subset of ASD patients who demonstrate sensory hypersensitivities or other signs of cortical excitability, as well as in multiple other neurodevelopmental disorders.
Acknowledgements
The authors would like to thank Rachel Greene, Savanna Sablich, and Melanie Soilleux for aid in data collection.