Excitatory/Inhibitory imbalance as a mechanism linking autism and sleep problems

Abstract

Sleep problems occur more frequently in individuals with autism spectrum disorder (ASD) than in typically developing individuals, and recent studies support a genetic link between ASD and sleep disturbances. However, it remains unclear how sleep problems may be mechanistically connected to ASD phenotypes. A longstanding hypothesis posits that an imbalance between excitatory and inhibitory (E/I) signaling in the brain underlies the behavioral characteristics of ASD. In recent years, emerging evidence has shown that regulation of the E/I ratio is coupled to sleep/wake states in wild-type animal models. In this review, we will explore the idea of altered E/I regulation over the sleep/wake cycle as a mechanism bridging sleep disruption and behavioral phenotypes in ASD.

Introduction

In autism spectrum disorder (ASD), sleep problems occur at a much higher rate than in typical development, affecting up to 93% of people with ASD [1,2]. ASD is the most common neurodevelopmental disorder in the US, with a prevalence of 1 in 36 children [3]. Three key features characterize ASD sleep problems: problems falling asleep, staying asleep and overall less sleep [1]. These are defining features of sleep onset insomnia. Problems falling asleep predict the severity of ASD core diagnostic symptoms and other associated problem behaviors and often worsen over time [4], which imposes a severe burden on caregivers.

Sleep is a homeostatic process. Normally, acute sleep deprivation (SD) results in sleepiness, which leads to compensatory recovery (falling asleep faster, sleeping better and longer). Persistent problems falling asleep alter the response to acute SD, leading to inability to fall and maintain sleep despite being sleepy, which are the hallmarks of chronic insomnia. Insomnia has well-documented adverse effects on cognition, attention and emotional regulation [5]. Despite a rapidly growing number of studies documenting sleep problems, little is known about the nature, underlying mechanisms, or molecular consequences of insomnia in ASD. Studies have started to show genetic evidence for a link between ASD and sleep disturbances. Copy number variants (CNVs) containing circadian pathway or insomnia risk genes show higher association with ASD than CNVs containing other genes [6]. A recent Mendelian randomization study shows that both the timing and quality of sleep seem to be causally linked to ASD [7], however it is yet unclear which aspects of brain physiology and function that are influenced by both circadian time and sleep quality mediate the strong connection between disturbed sleep and ASD.

The excitatory/inhibitory (E/I) imbalance hypothesis of Autism

An influential hypothesis regarding the mechanisms underlying ASD is the excitatory/inhibitory (E/I) imbalance hypothesis. This hypothesis proposes that an imbalance between excitatory (predominantly glutamatergic) and inhibitory (predominantly γ-aminobutyric acid (GABA)-ergic) mechanisms in the brain underlies the many behavioral aspects of ASD [8]. Although causal links have been suggested, with evidence for both overexcitation and over inhibition, how E/I imbalance leads to behavioral disruption in ASD is not fully understood. ASD-risk gene expression is enriched in maturing and mature neurons of excitatory and inhibitory lineages in prenatal human forebrain [9], suggesting there may be multiple paths to E/I imbalance. A more recent study looked at the relationship between genetic variation and cortical expression of genes related to GABAergic and glutamatergic signaling with ASD symptoms and cortical thickness [10]. Only genetic variation in glutamate-related genes was associated with ASD symptoms (as measured by the Autism Diagnostic Observation Schedule and Autism Diagnostic Interview), suggesting that broader ASD symptoms may be more linked to glutamatergic dysfunction. In addition, the study found that cortical regions with increased expression of glutamate and GABA genes are associated with larger differences in cortical thickness between autistics and neurotypical controls in adolescents and adults, but not in children. Interestingly the direction of the association changes across development. In adolescents, the association was positive, suggesting overall higher cortical thickness in ASD than in controls, while in adults this was negative, indicating an overall higher cortical thickness in controls than ASD. Thus, the influence of E/I imbalance on ASD seems dynamic, which is not surprising given that E/I ratio itself is dynamic, influenced by both time of day and sleep/wake state.

Sleep and the E/I Ratio

Maintaining the E/I ratio within the proper range is crucial to processing of information within brain circuits, and as discussed above, E/I balance disruption is thought to underlie behavioral phenotypes of ASD. However, interpreting how measured differences in the E/I ratio relate to ASD phenotypes is complicated by recent reports showing that the E/I ratio changes with sleep/wake history and time of day. Studies approximating the E/I ratio in humans report contradictory findings regarding the state-dependence and direction of E/I ratio modulation, likely due to differences in the brain areas and methods used [11-15]. However, direct measures of the synaptic E/I ratio in rodents show that sleep decreases the E/I ratio in primary visual cortex (V1) [16]. Emerging evidence examining mechanisms of both excitatory and inhibitory synaptic transmission is in line with these observations across many brain regions (Figure 1).

Figure 1. Excitatory and inhibitory synaptic transmission onto pyramidal neurons over sleep/wake states.

Excitatory and inhibitory synaptic transmission changes in opposite directions over the sleep-rich rest (light) and wake-rich active (dark) phases [16-18]. Importantly, brief sleep deprivation prevents these changes during the light phase, indicating a sleep- rather than circadian time- dependent effect.

Excitatory transmission decreases during sleep

Rodent studies have examined the regulation of neural communication between excitatory neurons across sleep-wake states. Research using whole-cell patch clamp recordings in multiple brain regions - including the medial prefrontal cortex, anterior cingulate cortex, and visual cortex - demonstrates that synaptic transmission is enhanced during wakefulness compared to sleep [16-18]. The data consistently show that sleep is associated with a decrease in miniature excitatory postsynaptic current (mEPSC) frequency, rather than amplitude. This suggests either a reduction in the number of synapses or presynaptic release sites during sleep. Electron microscopy studies of the somatosensory cortex, motor cortex, and hippocampal CA1 region reveal that sleep reduces the axon-spine interface area, consistent with modifications to both presynaptic and postsynaptic structures [19-22].

However, the relationship between sleep and synaptic plasticity is complex - studies in the hippocampal dentate gyrus and CA1 region have documented spine loss during wakefulness rather than sleep [23-26]. Additionally, in the primary motor cortex, sleep can simultaneously promote both pruning and strengthening of newly formed dendritic spines [27]. In conclusion, while evidence supports sleep-dependent net downregulation of excitatory transmission, these effects appear to be specific to some dendritic branches and spine types and are likely modulated by prior wake experiences.

Inhibitory transmission increases during sleep

Recent research also demonstrates sleep-wake regulation of inhibitory neurotransmission. In visual cortex layer 2/3, sleep promotes (and sleep deprivation prevents) increased spontaneous inhibitory signaling to pyramidal neurons, in part through wake-associated endocannabinoid suppression of inhibition [16] (Figure 1). Similar daily rhythms in miniature inhibitory postsynaptic currents (mIPSCs) have been observed in both medial prefrontal cortex layer 2/3 and hippocampal CA1 [16,28]. Notably, parvalbumin-positive inhibitory interneurons show opposite regulatory patterns compared to pyramidal cells: during rest periods, these interneurons receive increased excitation but decreased inhibition, resulting in enhanced circuit-level inhibition [29]. These synaptic modifications in both cell types primarily manifest as changes in the frequency of miniature events, suggesting alterations in synapse number or release sites, though amplitude changes have also been documented [28].

Sleep appears to enhance inhibitory signaling through multiple complementary mechanisms. Several research groups have reported that cortical pyramidal neurons exhibit reduced intracellular chloride concentrations during rest periods, due to changes in ionic co-transporter activity [30,31]. This creates a more negative GABA receptor reversal potential, strengthening inhibitory responses, though there is disagreement about whether this effect depends on sleep history. Additionally, astrocytes show increased intracellular chloride levels during sleep, creating a chloride reservoir that enhances GABAergic signaling by elevating extracellular chloride levels [32]. The role of tonic inhibition (mediated by extra synaptic GABA_A receptors) presents a more complex picture. Recent studies in hippocampal CA1 pyramidal neurons show higher tonic inhibitory conductance after wakefulness compared to sleep [28,33]. While tonic GABA signaling typically suppresses neuronal activity through shunting inhibition, elevated tonic GABAergic conductance combined with depolarizing chloride gradients might paradoxically increase excitability [34]. Future research needs to examine how tonic and phasic GABAergic conductance interact with chloride gradients within the same neuronal populations to fully understand their impact on excitation/inhibition balance. Nevertheless, the current evidence generally supports enhanced inhibitory signaling during sleep.

Implications of sleep and the E/I ratio for ASD

Although the exact molecular and cellular mechanisms have yet to be determined, it is apparent that many factors contribute to regulation of both excitatory and inhibitory signaling over sleep and wake states. Disruption of one or more of these mechanisms may therefore cause dysregulation of the E/I ratio. In humans, sets of glutamate- and GABA-related genes are associated with different ASD phenotypes across brain regions and over development, suggesting a complex relationship between genetic factors influencing the E/I ratio and ASD [10]. However, the consequences of sleep/wake-dependent E/I regulation on brain function in wild-type (WT) mice, and of dysregulation in ASD, have not been explored.

Sleep as a regulator of the E/I ratio

Convergent evidence that sleep regulates the E/I ratio suggests that sleep disturbances in ASD could contribute to several behavioral phenotypes (Figure 2A). First, sensory processing phenotypes are associated with sleep problems in individuals with ASD [35]. E/I alterations can impair sensory discrimination by broadening the tuning of sensory neurons, which has been reported in primary auditory, visual, and somatosensory cortices of ASD-related mouse lines (reviewed in [36]). In addition, transcranial magnetic stimulation has enabled the non-invasive evaluation of neuronal plasticity in humans, and recent studies indicate differences between neurotypical individuals and individuals with ASD [37,38]. Sleep is thought to be a key modulator of plasticity [39], and the E/I ratio, which changes across sleep and wake states in highly plastic superficial cortical circuits [16,29,40], is poised to regulate plastic processes. The increased E/I ratio during waking may facilitate plasticity induction through disinhibition, which gates experience-dependent plasticity [41]. On the other hand, enhanced inhibition during sleep could facilitate either synaptic strengthening by improving spike timing precision or synaptic weakening by promoting long-term depression [42,43]. Finally, ASD is a developmental condition. Sleep may play distinct roles early in development and later in life, with neuronal circuit reorganization a primary function early on [44]. Early life sleep disruption in prairie voles causes long-lasting behavioral changes relevant to ASD, including cognitive inflexibility and social deficits, suggesting a causative role of sleep disruption in ASD [45-47]. These studies also revealed altered markers of excitatory and inhibitory transmission lasting into adulthood, suggesting that the developmental effects of sleep disturbance in ASD may lead to E/I imbalance later in life.

Figure 2. Patterns and consequences of E/I dysregulation in ASD.

(A) Shifts in excitatory (E) and inhibitory (I) transmission onto pyramidal cells alter circuit output, which may impact multiple downstream processes. (B) In WT mice, the E/I ratio is higher during the rest (light) phase than the active (dark) phase (black line; circles represent E/I ratio measures [16]). E/I ratio alterations in ASD could take several forms. First, the E/I ratio could be uniformly higher (orange) or lower (purple) across the day. Alternatively, E/I regulation could be lost (green) or altered (blue). The latter scenarios may cause an apparent change in the E/I ratio, depending on the time of the measurement, when there is not an overall change (and, notably, there are two times of day when the E/I ratio would be unaltered (yellow stars)).

E/I ratio as a regulator of sleep

An intriguing alternative mechanism linking the E/I ratio to sleep in ASD is that the magnitude of the E/I ratio may be a regulator, rather than consequence, of sleep. For instance, as discussed above, sensory phenotypes in ASD can be caused by E/I ratio alterations, and sensory hypersensitivity in ASD may lead to sleep difficulties [35]. In addition, E/I elevation during waking may play a causal role in sleep regulation under typical conditions by modulating homeostatic sleep pressure. Sleep pressure builds progressively over periods of wakefulness, corresponding to elevated slow wave activity (SWA) during subsequent NREM sleep [48]. Excitatory transmission during wake is positively correlated with SWA during subsequent NREM sleep, and experimentally manipulating intracellular chloride levels to reduce inhibitory signaling increases SWA [30, 49], suggesting that increasing the E/I ratio could elevate sleep pressure. Sleep pressure is commonly attributed to buildup of extracellular adenosine, which could act by suppressing inhibitory transmission, although adenosine appears to maintain the overall E/I ratio in the rodent cortex [50,51]. Deviation of brain circuit dynamics from criticality, a state optimal for cortical computation that depends on E/I balance, has also been proposed as a mediator of sleep pressure [52, 53]. These observations raise the intriguing possibility that E/I elevation during wake drives sleep need under typical conditions, and therefore E/I dysregulation may result in altered sleep pressure and, consequently, disrupted sleep in ASD.

Sleep-wake dependent E/I ratio dysregulation in ASD

ASD is heterogeneous, with significant variability in symptoms and severity. Nonetheless E/I alterations and sleep problems are both remarkably common in ASD despite this heterogeneity. One intriguing possibility is that E/I regulation across the sleep/wake cycle, rather than the E/I ratio itself, is altered in ASD. Many mechanisms link excitatory and inhibitory transmission to sleep and wake states, and each of these mechanisms may be differentially impacted in distinct forms of ASD. Therefore, E/I dysregulation may be a point of convergence across the heterogeneity of the autism spectrum.

In nocturnal rodents, the E/I ratio is high during the active (dark) phase and low during the rest (light) phase (Figure 2B, black). Dysregulation in these models could manifest as a static E/I ratio (Figure 2B, green) or a shift in timing over the rest and active phases (Figure 2B, blue). Notably, either scenario is consistent with a measured E/I elevation if evaluated only during the light phase. This hypothesis was recently tested in two very different in ASD mouse models, Fmr1 knockout (KO) mice (a model of Fragile X Syndrome) and BTBR mice [54]. Fmr1 KO mice exhibited complete loss of the E/I oscillation, while BTBR mice showed an inverted oscillatory pattern compared to wild-type controls. Thus, temporal misalignment between E/I balance and behavioral state seems to be a common mechanism between the two ASD-related mouse lines. Whether this is observed across other ASD models, and in clinical populations, remains to be determined but deserves further investigation.

Concluding Remarks

The connection between sleep and the E/I ratio underscores the need to take sleep/wake history into account in studies of ASD. Most studies in rodent ASD models are conducted at a single time of day, which can influence the magnitude and direction of experimental results (Figure 2). Furthermore, the prominence of insomnia in ASD combined with the potential mechanistic link between the E/I ratio and sleep pressure highlights the importance of comprehensive sleep phenotyping in ASD models. Although sleep has been examined in many rodent models, few studies have evaluated whether they bear the hallmarks of insomnia, namely longer sleep latency and decreased sleep time/continuity under conditions of high sleep pressure (reviewed in [55]).

While the exact set of mechanisms coupling sleep/wake states to the E/I ratio under typical conditions remains to be determined, E/I regulation is clearly multifactorial. Genetic and environmental factors that impact any one process may lead to dysregulation, making the E/I ratio vulnerable to disruption. Therefore, E/I dysregulation across sleep and wake states may be a point of convergence, resulting in common behavioral phenotypes in ASD despite diverse etiology.

This perspective has important implications for treatment and drug development. If timing of E/I balance matters, rather than just the absolute levels, then the timing of treatments is crucial. Therapeutic interventions might need to be administered at specific times of day to be most effective, aligning with natural circadian rhythms of brain activity. On the other hand, since different patterns of disruption occur in different types of ASD models (Fmr1 KO vs BTBR), treatment strategies might need to be tailored based on whether a patient has a complete loss of E/I rhythm or a reversed rhythm, for example. Last, this framework has important drug-development implications. It suggests the need to create medications that target not just the overall E/I balance, but its temporal regulation as well as considering using chronotherapy (timing-based drug administration) for existing treatments that target excitatory or inhibitory neurotransmission systems.