Introduction
Phelan-McDermid Syndrome (PMS) is a well-characterized genetic condition that results from haploinsufficiency of
SHANK3 in the 22q13.3 region. The phenotype in PMS is frequently characterized by intellectual disability [
1,
2], autism spectrum disorder (ASD; 50–84%) [
3,
4], and epilepsy [
5].
SHANK3 codes for a master scaffolding protein in the postsynaptic density of glutamatergic synapses [
6], and its isoforms perform a variety of synaptic functions relevant to neuronal excitability and plasticity [
7‐
15].
A key step in understanding the translational pathway from cells to circuits, networks and ultimately phenotype, involves measurements that reflect large scale network dynamics, including assessments of intrinsic neural oscillations. Electroencephalography (EEG) offers particular opportunity in this regard, because it can measure network dynamics in both humans and animal models, allowing for both forward and back-translation of findings. Clinical EEG (evaluated by visual review) frequently demonstrates abnormalities in PMS, including generalized slowing of activity, slowing or absence of the occipital dominant rhythm, and epileptiform activity [
5,
16]. Epileptiform activity on EEG is also frequently seen in ASD more broadly [
17].
Shank3B null mutant mouse models have demonstrated altered oscillatory power, depending on the location and frequency band studied [
9,
14,
18]. Numerous studies in humans with ASD have demonstrated various abnormalities in resting EEG spectral power [
19]; however, quantitative studies of EEG activity in humans with PMS have not been previously published.
Recently, there has been increasing interest in the coupling of EEG activity across frequencies, using measures such as Phase-Amplitude Coupling (PAC), given the possibility that such cross-frequency coupling has distinct mechanistic underpinnings. Such coupling is crucial for many of the cognitive functions that are altered in neurodevelopmental disorders, such as long-range communication [
20], integration of local and global cortical processing [
21], and segmenting and prioritizing sensory input [
20,
22]. Altered PAC strength in the alpha-gamma frequency pair has been reported in individuals with ASD at baseline (i.e., rest) and associates with symptom severity [
23,
24]. PAC is also altered during tasks in some neurodevelopmental disorders, including during face processing in ASD [
25] and cognitive discrimination in a mouse model of Fragile X syndrome [
26]. Likewise, PAC during the period preceding an auditory stimulus has been found to positively correlate with non-verbal intelligence quotient in Fragile X syndrome [
27]. Cross-frequency coupling thus has theoretical relevance to ASD, intellectual disability, and associated neurogenetic disorders in which such processes are likely altered [
28‐
30].
Recent work suggests not just the strength (the extent to which PAC occurs), but the phase (e.g., where in relation to the alpha waveform gamma amplitude is maximal), can signal important network characteristics. The phase at which fast oscillations are strongest can vary by cortical layer [
31] and with alterations in interneuron function [
32]. Surface EEG measurements demonstrate the alpha phase resulting in maximum gamma power can vary by age [
33] and depth of anesthesia [
34]. Differences in PAC phase have been found with encoding success [
35] and context [
36], and PAC phase bias has been suggested to reflect the ratio of feedforward (bottom-up) to feedback (top-down) cortical activity [
33], suggesting the phase of PAC can be functionally relevant particularly among conditions commonly associated with autism spectrum disorder. The timing of gamma within the alpha cycle consequently has the potential to capture alterations in brain connectivity and function that result from specific synaptic perturbations and underlie clinical disorders.
EEG measures of PAC strength and PAC phase thus offer opportunities to enhance understanding of circuit-level dysfunctions in PMS. Here, we first examined whether individuals with PMS, as compared to typically developing (TD) individuals display differences in alpha-gamma PAC strength and phase. Second, we investigated whether these EEG metrics associate with measures of phenotype among individuals with PMS. We hypothesized (1) individuals with PMS would demonstrate increased PAC strength and phase bias, compared to typically developing controls and (2) PAC metrics would correlate with sensory processing difficulties and ASD symptom severity.
Discussion
We find individuals with PMS show significantly increased alpha-gamma phase bias relative to TD individuals, with most individuals with PMS demonstrating positive overall phase bias, whereas most typically developing individuals demonstrated negative overall phase bias in our sample. Between-group differences are primarily driven by findings over posterior electrodes, where phase bias and PAC are both more strongly positive in individuals with PMS relative to TD individuals. Previous work has reported greater alpha-gamma PAC in a midline parietal–occipital source in individuals with ASD [
24]. Within individuals with PMS, no differences were observed with measures of overall ASD phenotype, or social functioning; however, RBS-R total score was found to increase with increased PAC strength, indicating in individuals with PMS, PAC strength may map on to this aspect of the ASD symptom profile specifically.
The between-group differences in phase bias suggest that circuit function is perturbed in PMS, in a manner measurable by surface EEG. This finding suggests several opportunities for back-translation into animal models to elucidate underlying mechanisms. For example, scalp level EEG does not reflect the unified activity of the cortex, but rather the grand average of many networks often exhibiting conflicting activity. Phase bias is known to vary by cortical layer. Laminar recordings in monkeys and rats demonstrate that spontaneous current sinks in theta and alpha bands in layers 2/3-5a are associated with high gamma amplitudes and high action potential firing (and sources are associated with low gamma amplitudes and low action potential firing) whereas the opposite is true in layer 6 (sinks are associated with low gamma amplitudes and low action potential firing, and sources with high gamma amplitudes and high action potential firing) [
31,
55]. Additionally, alpha current generators in layers 2/3 and 6 are in phase with one another, but out of phase with those in layer 4 [
55], meaning whether scalp-level EEG gamma activity is phase-locked to the falling phase or the rising phase of alpha could depend on whether alpha activity from layers 2/3 and 6 or layer 4 dominates the signal. Therefore, it is possible the phase bias presented here depends on the relative PAC and alpha activity of each cortical layer. Cortical layer 4 predominantly accepts feedforward (thalamocortical) input, layer 6 predominantly provides feedback (corticothalamic output), and layer 2/3 integrates feedforward, feedback, and lateral activity [
56,
57]. Between-group differences in surface level phase bias may therefore suggest altered balance of feedforward versus feedback information transfer in PMS; this could be further examined in animal models.
Here, the phase bias abnormalities in individuals with PMS were localized to electrodes over the posterior cortex. Alpha-gamma PAC has been previously shown to increase in the occipital cortex during visual tasks [
58]. Notably, the present study analyzed EEG recordings collected while participants watched a silent movie. Cases of cortical visual impairment have been reported in some individuals with PMS [
59]; therefore, the network perturbations captured here may also relate to abnormalities in visual processing in PMS.
Notably, we did not identify any differences in EEG power, in any frequency band, between individuals with PMS and TD individuals. This is in contrast to prior electrophysiological measurements in animal models, where power differences have been demonstrated in specific regions and frequency bands [
9,
14,
18]. Still, no consistent findings have emerged in studies of animal models as well as in clinical studies. Here, we used relatively conservative statistical techniques, and a slight relaxing of statistical thresholds would have led to findings of overall low alpha power and high gamma power, consistent with some prior animal studies [
9,
14]. Nonetheless, the PAC effects (particularly phase bias) are quite strong and persist despite these conservative techniques. While prior studies of PAC during some tasks demonstrate an inverse relationship between alpha power and PAC [
60], we did not identify any such relationship in our sample. This suggests that our PAC findings are not driven by changes in nonsinusoidal alpha activity, and that PAC and alpha power can be independently modulated.
By grossly reflecting neural network activity, EEG is an intermediary on the spectrum from genotype to phenotype. Given the myriad possibilities for analysis that EEG offers, EEG itself thus also reflects a smaller spectrum-within-a-spectrum from genotype to phenotype, depending on the exact analysis chosen. Here we demonstrate that one analytic technique (phase bias) leans toward a reflection of genotype more than phenotype. Notably, we do not have any data to suggest that this phase bias anomaly is specific to PMS; in fact, it is quite possible that similar phase bias anomalies could be present in other genetic disorders that affect similar pathways (e.g., other disorders of the mTOR pathway), and further research is necessary to test this. On the other hand, our findings suggest that zMI likely measures an aspect of neural network function that leans more towards phenotype; therefore, future research should explore whether zMI anomalies are associated with restricted and repetitive behaviors in other neurodevelopmental disorders. Along similar lines, comparison of PAC and z-MI findings between PMS and a phenotypically similar cohort would also be of interest.
However, the genotype–phenotype spectrum is just one of many axes that EEG-based measures may reflect. For example, EEG-based measures may also change across development. We find that z-MI increases with age in our sample (mean age 9.7 years), but phase bias does not. This extends prior findings, in which z-MI was found to increase across the first 3 years after birth in typical development [
33]. Developmental effects are particularly important to consider in PMS given the known molecular and electrophysiological functions of
SHANK3, including effects on plasticity
. SHANK3 provides scaffolding in the postsynaptic density of glutamatergic synapses [
6], and
Shank3 mutant mouse models have therefore demonstrated decreased excitability of glutamatergic [
8,
9,
14,
15] and GABAergic neurons [
11]. Plasticity is also impaired in
Shank3 mutants [
13]. Excitability can be altered by developmental activity and plasticity within circuits, at times leading to seemingly contradictory findings. For example, when inhibition is impaired more than excitation within corticostriatal circuitry during early development, the balance between activity of excitatory and inhibitory neurons can lead to cortical hyper-activity, with resulting changes in plasticity that ultimately cause high (rather than low) excitability of GABAergic neurons in this circuit [
12]. In layer 2/3 of primary somatosensory cortex,
Shank3 deficiency causes decreased excitability of GABAergic interneurons but increased excitability of glutamatergic neurons [
11]. Trajectories across development, combined with studies examining primary and compensatory mechanisms underlying these trajectories, can provide additional clues about the biological underpinnings of neurodevelopmental disorders including (but not limited to) PMS.
Limitations
Our ability to detect subtle phenotypic associations was hampered by several limitations. First, as is common in rare disease research, the sample size in the PMS group led to limited statistical power for assessing associations with categorical variables within this group. In particular, only 4 PMS individuals exhibited a history of seizures. Additionally, our typically developing cohort was small (15 EEGs analyzed), limiting our ability to identify differences with the PMS group. Second, the severity of PMS led several behavioral measures to suffer from a ‘floor’ effect, making it difficult to compare the phenotypes of individuals within the PMS group. Clinical assessments were also rarely conducted in in the typically developing cohort; as a result, the present study was not able to test for associations between PAC measures and clinical variables in this group. Also of note, though the individuals with PMS enrolled in this study that were not able to provide adequate EEG data for analysis did not demonstrate clear differences on phenotyping measures, they do represent a subgroup of individuals with PMS this study was not able to capture. Finally, though the scalp-level EEG used here allows us to describe differences in grand-average oscillatory activity, it is unable to differentiate the specific neural mechanisms underlying these differences; back-translation into animal models will likely be necessary to further explore this.
Acknowledgements
The Developmental Synaptopathies Consortium (U54NS092090) is part of the Rare Diseases Clinical Research Network (RDCRN), an initiative of the Office of Rare Diseases Research (ORDR), National Center for Advancing Translational Sciences (NCATS). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health (NIH). We are grateful to the Boston Children’s Hospital IDDRC, 1U54HD090255, for support for this project. We are sincerely indebted to the generosity of the families and patients in PMS clinics across the United States who contributed their time and effort to this study. We would also like to thank the Phelan-McDermid Syndrome Foundation for their continued support in PMS research. Members of the Developmental Synaptopathies Consortium (DSC)—Phelan-McDermid Syndrome Group include:
Mustafa Sahin, MD, PhD a, b, Alexander Kolevzon, MD c, d, Joseph Buxbaum, PhD c, d, e, f, Elizabeth Berry Kravis, MD, PhD g, h, i, Latha Soorya, PhD j, Audrey Thurm, PhD k, Craig Powell, MD, PhD l, m, Jonathan A Bernstein, MD, PhD n, Simon Warfield, PhD o, Benoit Scherrer, PhD o, Rajna Filip-Dhima, MS a, Kira Dies, ScM, CGC a, Paige Siper, PhD c, Ellen Hanson, PhD p, Jennifer M. Phillips, PhD q
Affiliations for above:
a Department of Neurology, Boston Children’s Hospital, Harvard Medical School, Boston, MA; b F.M. Kirby Neurobiology Center, Boston Children’s Hospital, Harvard Medical School, Boston, MA; c Seaver Autism Center for Research and Treatment, Mount Sinai School of Medicine, New York, NY; d Department of Psychiatry, Mount Sinai School of Medicine, New York, NY; e Department of Genetics and Genomic Sciences, Mount Sinai School of Medicine, New York, NY; f Department of Neuroscience, Mount Sinai School of Medicine, New York, NY; g Department of Pediatrics, Rush University Medical Center, Chicago, IL; h Department of Neurological Sciences, Rush University Medical Center, Chicago, IL; i Department of Biochemistry, Rush University Medical Center, Chicago, IL; j Department of Psychiatry, Rush University Medical Center, Chicago, IL; k Pediatrics and Developmental Neuroscience Branch, National Institute of Mental Health, National Institutes of Health, Bethesda, MD; l Department of Neurology and Neurotherapeutics, University of Texas Southwestern Medical Center, Dallas, TX; m Department of Psychiatry and Neuroscience Graduate Program, University of Texas Southwestern Medical Center, Dallas, TX; n Department of Pediatrics, Stanford University School of Medicine, Stanford, CA o Department of Radiology, Boston Children’s Hospital, Harvard Medical School, Boston, MA p Department of Developmental Medicine, Boston Children’s Hospital, Harvard Medical School, Boston, MA; q Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, CA
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.