In Utero Exposure to Valproic Acid Changes Sleep in Juvenile Rats: A Model for Sleep Disturbances in Autism

Abstract

Study Objectives:

To determine whether sleep disturbances are found in the valproic acid model of autism spectrum disorders (ASD).

Design:

Comparative study for sleep behavior, sleep architecture, electroencephalogram (EEG) spectral analysis, and glutamic acid decarboxylase (GAD) 65/67 protein expression in juvenile rats exposed to valproic acid (VPA), sodium salt, or saline in utero.

Setting:

N/A.

Participants:

Juvenile (postnatal day 32) male and female Sprague-Dawley rats.

Interventions:

In utero exposure to either saline or 400 mg/kg VPA administered intraperitoneally to the dams on gestational day 12.5. On postnatal days 22-24, all rats were implanted with transmitters to record EEG and electromyogram (EMG) activity.

Measurements and Results:

During the light phase, when nocturnal animals are typically quiescent, the VPA-exposed animals spent significantly more time in wake (∼35 min) and significantly less time in non-rapid eye movement (NREM) sleep (∼26 min) compared to the saline controls. Furthermore, spectral analysis of the EEG reveled that VPA-exposed animals exhibited increased high-frequency activity during wake and rapid eye movement (REM) sleep and reduced theta power across all vigilance states. Interestingly, the gamma-aminobutyric acid (GABA)-ergic system, which modulates the induction and maintenance of sleep states, was also disrupted, with reduced levels of both GAD 65 and GAD67 in the cortical tissue of VPA-exposed animals compared to saline controls.

Conclusions:

To date, the current animal models of ASD have been underutilized in the investigation of associated sleep disturbances. The VPA animal model recapitulates aspects of sleep disruptions reported clinically, providing a tool to investigate cellular and molecular dysregulation contributing to sleep disruptions in ASD.

Citation:

Cusmano DM, Mong JA. In utero exposure to valproic acid changes sleep in juvenile rats: a model for sleep disturbances in autism. SLEEP 2014;37(9):1489-1499.

INTRODUCTION

Autism spectrum disorders (ASD) are an array of developmental disorders that primarily disrupt social interactions, communication, and behavioral patterns. Clinical observations have also indicated a high prevalence of sleep disturbances in children with ASD.^1–6 Recent clinical studies report that the prevalence of sleep problems in children with ASD is approximately 44% to 83% of diagnosed cases.^2,7–10 These sleep dysfunctions typically manifest as difficulties initiating and maintaining sleep, sleep fragmentation, insomnia, and parasomnias.^8,11–21 Quality sleep is imperative for the maintenance of good health, and sleep loss can lead to or exacerbate existing behavioral problems associated with ASD.^14,22–24 Alterations in sleep architecture have been linked with cognitive and behavioral deficits as well as stress and anxiety.^{3,22,23,25,26} Because children in whom ASD has been diagnosed share symptoms similar to those experiencing sleep loss, it is plausible to suspect that sleep disturbances in ASD may contribute to its symptomology. However, it appears that sleep problems in ASD are not influenced by the severity of cognitive deficits or ASD subtypes.^16,27,28

Although these sleep disruptions have been characterized, their cause remains unknown and relatively unexplored. A question that remains unanswered is whether the neurocircuitry underlying sleep pathways is developmentally changed in ASD or whether disruptions in sleep are a consequence of changes in daily anxiety levels. Distinguishing between the two possibilities in a clinical setting would be challenging. The rodent's neurocircuitry and neurochemistry of sleep share similarities with humans, suggesting that rats would be a good model system for basic investigations.²⁹ To date, the current animal models of ASD have been underutilized in the investigation of ASD associated sleep disturbances. Here we present data supporting the use of a rodent model in elucidating the etiology of sleep disruptions in ASD.

One animal model of ASD is prenatal exposure to valproic acid (VPA) or the salt, sodium valproate. VPA is an antiepileptic drug used to treat seizure disorders. Although VPA can prevent the induction of seizures during pregnancies, it has been classified as a teratogen and linked to a clinical phenotype known as fetal valproate syndrome (FVS), which includes various types of craniofacial malformations and developmental and cognitive delays.^30–32 Many studies have also shown that FVS is associated with autistic behaviors and it is now thought that VPA exposure during a critical period of development significantly increases the risk of ASD.^30,33–35 In a rodent model, animals exposed to VPA prenatally show similar alterations in brain architecture and sexspecific behavioral deficits seen in children with ASD.^36–45 Therefore, it is thought to be a valid animal model for studying ASD.

To our knowledge, only one study has indirectly evaluated sleep/wake behavior in an animal model of ASD using locomotor and feeding behaviors.⁴⁶ With the VPA model, Tsujino and colleagues⁴⁶ have shown abnormal circadian rhythms in their animals, marked by increased locomotor activity and feeding behaviors during their sleep phase. Additionally, using micro-dialysis, they found that basal levels of serotonin (5-HT) are higher in VPA-exposed animals and that there is a caudal shift of 5-HT positive neurons in the dorsal raphe nucleus. Interestingly, the dorsal raphe supplies the serotinergic input involved in the ascending arousal system of the sleep/wake circuitry.

In this study, we investigated sleep/wake behaviors, including the sleep patterning, and electroencephalogram (EEG) characteristics related to the individual vigilance states in juvenile rats. Most studies investigating sleep in ASD report findings from children and adolescents with ASD; therefore, we recorded from juvenile/prepubertal rats to better model brain maturation and sleep in children and young adults. Because sleep/wake cycles are controlled by a complex system of reciprocal connections between sleep-promoting nuclei and arousal centers, we investigated whether VPA exposure alters some aspect of the sleep-wake neurocircuitry. Specifically, we investigated whether VPA alters the gamma-aminobutyric acid (GABA)ergic system, which plays a significant role in sleep onset and maintenance,⁴⁷ through quantification of the expression of glutamic acid decarboxylase (GAD), the rate-limiting enzyme in GABA synthesis. Here, we measured GAD expression in the cortex and basal forebrain (BF), which is involved in sleep/wake and has been shown to regulate activity of cortical neurons.^48–50

We found that VPA-exposed animals have more consolidated bouts of wake and non-rapid eye movement (NREM) sleep, resulting in a disruption of the normal sleep architecture. Further, we see changes in the EEG spectrum related to rapid eye movement (REM) sleep and arousal, with decreased theta power and increased gamma power. As implicated in children with autism, the GABAergic system is also affected in these animals. Expression levels of both isoforms (65 and 67) of GAD are significantly decreased in the cortex, but not the BF. Overall, these data suggest that in utero exposure to VPA induces sleep disruptions that correlate with the clinical ASD phenotype and that it provides a valid model for studying the underlying mechanisms regulating these changes in sleep.

METHODS

Animals

All experimental procedures were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All experiments were approved by and were in accordance with the guidelines of the University of Maryland Institutional Animal Care and Use Committee.

Female Sprague-Dawley rats were mated overnight (12-16-h) and the next day was designated day 1 of gestation. Pregnancy was determined by percent weight gain by gestational day 12. VPA sodium salt (Sigma-Aldrich, St. Louis, MO) was dissolved in 0.9% sterile saline, pH 7.4. Pregnant dams received a single intraperitoneal injection of 400 mg/kg VPA (VPA-exposed) or sterile saline (SAL-exposed) on gestational day 12.5. Each female was housed individually and raised her own litter. For this study, two SAL-exposed and two VPA-exposed litters were used. Pups were weaned on postnatal day 21.

The following experiments were performed in SAL- (n = 5) and VPA-exposed (n = 6) juvenile male and female rats (∼70 -80 g). Tail malformations that manifest as a distal bend or a kink have been reported as a marker of VPA toxicity during development.^39,51,52 Only VPA-exposed rats with tail malformations were used in this study (see next paragraphs). All animals were housed with a littermate under a 12:12 h, reverse light:dark cycle with free access to food and water for the duration of the experiment.

Surgery

Surgical implantation of the telemeter and EEG/electro-myographic (EMG) electrodes were according to previously published protocols^53–57 with the following modifications. Animals were anesthetized using isofluorane and placed in a stereotaxic frame to secure the head. A bi-potential-lead transmitter (TL11M2-F20-EET, Data Sciences International (DSI), New Brighton, MN) was implanted subcutaneously through a dorsal incision of the abdominal region. Another incision was made along the midline of the head and neck to expose the skull and neck muscle. Two burr holes (0.5 mm diameter) were drilled asymmetrically and two dental screws (Plastics One, Roanoke, VA) were implanted into the skull at 2.0 mm anterior/+1.5 mm lateral and 6.0 mm posterior/-1.5 mm lateral from bregma. The EEG leads were attached to the screws and secured with dental cement. The two EMG leads were implanted in the dorsal cervical neck muscle approximately 1.0 mm apart and sutured into place. The skin along the head was sutured and the dorsal incision was stapled closed with wound clips. Animals were treated with antibiotic ointment and topical lidocaine, as well as 0.1-cc buprenorphine, postoperatively, and allowed 7 days to recover before being reunited with their non-experimental cagemate.

Data Acquisition and Analysis

All recordings took place in a designated room shielded from noise or other background disturbances. EEG and EMG waveform data for postnatal days 31–34 were collected using Dataquest ART 4.0 software (DSI) set to a continuous sampling mode. Sleep is typically measured as a function of EEG and EMG activity. For each of the animals, the 24-h (12:12 light:dark cycle) period on postnatal day 32 was scored using NeuroScore software (DSI). EEG/EMG waveforms were scored according to visual inspection of 5-sec epochs into wake (low-amplitude, high-frequency EEG combined with high-amplitude EMG), NREM (high-amplitude, low-frequency EEG combined with low EMG tone) or REM (low-amplitude, high-frequency EEG combined with muscle atonia and occasional muscle twitches). Transitions were scored when four or more epochs of the new state were noted within another stage. This rule was followed unless there was a period punctuated by a new state, which then persisted; this was considered a transition period so these epochs were scored as the new state.

The scored epochs were summed over the 12-h light phase, 12-h dark phase, and 24-h period and reported as the total time (in min) spent in each state. Within each 12-h phase, the average bout duration (in sec) and the number of each bout and transition were also determined for each vigilance state.

The DSI module for periodogram powerbands in NeuroScore was used to generate the mean power of frequencies present in the three vigilance states (wake, NREM sleep, and REM sleep) across the light and dark. The periodograms estimate the power for each defined frequency band (delta: 0.5-4 Hz, theta: 4-8 Hz, alpha: 8-12 Hz, sigma: 12-16 Hz, beta: 16-24, and gamma: 30-80 Hz) on the fast Fourier transform (FFT) designed for continuous data. The relative power band value (percentage; desired band over total power in the signal) was determined in 1-h bins and then averaged across each 12-h phase for each vigilance state.

Western Blotting

Following the recording period, on postnatal day 40, brains were collected and frozen on dry ice. Sections (180 μm) were cut on a cryostat and micropunches were collected from the somatosensory cortex and BF of each animal, as well as three extra age-matched SAL-controls (total SAL n = 7). The tissue was then homogenized via sonication in a cell lysis buffer containing a protease inhibitor cocktail. The protein concentration for each sample was determined using a bicinchoninic acid (BCA) assay kit (Pierce, Rockford, IL). Protein (1 μg) of each sample (SAL = 7, VPA = 6) was loaded into a 10% Tris-glycine sodium dodecyl sulfate polyacrylamide gel electrophoresis gel (Invitrogen, Carlsbad, CA), and then the electrophoresed proteins were blotted onto a polyvinyl difluoride membrane (Invitrogen). The membrane was washed in 20 mM Tris-buffered saline solution with 0.05% Tween 20 (T-TBS). The membrane was blocked overnight in 5% powdered milk at 4 oC and incubated the next day in the anti-GAD 65/67 primary antibody solution (1:30,000 in T-TBS; AB1511, Millipore, Billerica, MA) for 2 h at room temperature. Following the primary antibody incubation, the membrane was washed three times in T-TBS, and incubated for 1 h at room temperature in an anti-rabbit immunoglobulin G, horseradish peroxidase (HRP)-linked secondary antibody solution (1:2000 in T-TBS; #7074, Cell Signaling Technologies, Danvers, MA). The Phototype-HRP chemiluminescent system (Cell Signaling Technology) was used for detection of the protein recognized by the antisera. To correct for errors in sample loading, the membrane was also probed with an antibody to the housekeeping enzyme glyceraldehyde- 3-phosphate dehydrogenase (GAPDH; 1:1,000,000; MAB374, Millipore) as previously described.^54,58,59 Membrane was exposed to Hyperfilm-ECL (Kodak, Rochester, NY) for varying exposure times. The films were then scanned into a computer at 1,200 dpi and analyzed using ImageJ64 software (http://rsbweb.nih.gov/ij/download.html). The optical densities (o.d.) were measured for each individual band and normalized to the o.d. of the GAPDH bands.

Statistical Analysis

Statistical differences between SAL- and VPA-exposed animals were determined with Student t-tests for each sleep parameter studied. Differences in the power spectrum densities were determined with a Mann-Whitney U test for nonpara-metric data. A two-way repeated measures analysis of variance (ANOVA) was used to analyze wake gamma power across the 24-h period. Two-way ANOVAs were used to analyze NREM sleep and REM sleep gamma power across the 24-h period because these sleep states were not represented in every hour. Student t-tests were used comparing o.d. for SAL- and VPA-exposed tissue. One-way ANOVA followed by a Newman-Keuls post hoc test was used to compare the o.d. values for SAL- and VPA-exposed GAD expression by tail malformation severity.

RESULTS

VPA Exposure Increases Wakefulness at the Expense of Sleep in Juvenile Rats

Nocturnal adult rodents, like the Sprague-Dawley rat, cycle through bouts of sleep (NREM and REM) and wake in both the dark (active) and light (quiescent) phases. Typically, a higher percentage of accumulated sleep occurs in the light phase with consolidated bouts of wake occurring in the dark phase. Visual inspection of hypnograms generated from the scored EEG traces from SAL- and VPA-exposed animals suggested the juvenile controls (SAL-exposed) followed a normal adult-like pattern with consolidated bouts of wake primarily present in the dark phase. In contrast, the VPA-exposed rats appeared to experience consolidated bouts of wake in both the light and dark phases (Figure 1).

Figure 1.

Representative hypnograms of juvenile rats exposed to saline (SAL, top) and valproic acid (VPA, bottom) in utero across a 24-h period. W, wake; N, NREM sleep; R, REM sleep. The solid black bar is highlighting consolidated wake bouts during the light phase of VPA-exposed rats, which are absent in the SAL-exposed controls. Gray area is lights off.

Quantitative analysis of the scored traces for the total time spent in NREM, REM, and wake revealed significant differences between the SAL- and VPA-exposed groups. Across the 12-h of the light phase, the VPA-exposed animals spent significantly more time in wake (∼35 min; t₉ = 2.741, P = 0.023; Figure 2A) with no significant differences in the dark phase (t₉ = 0.788, P > 0.05). Moreover, across the 24-h light:dark period, VPA-exposed rats again spent significantly more time in wake with the difference increasing to approximately 48 min compared to the SAL-exposed controls (t₉ = 2.382, P = 0.041; Figure 2A).

Figure 2.

Juvenile rats exposure to valproic acid (VPA) spend more time awake than asleep compared to saline (SAL)-exposed controls. (A) During the light phase, VPA-exposed rats spend more time in wake (∼35 min) compared to SAL-exposed rats. Across the 24-h period, VPA-exposed rats spend ∼48 min more in wake than SAL-exposed controls. (B) VPA exposure reduced non-rapid eye movement (NREM) sleep time by ∼30 min during the light phase only. (C) Rapid eye movement (REM) sleep during the individual light and dark phases was not affected by VPA exposure, however; the 24-h REM sleep total is reduced by ∼17 min in the VPA-exposed juveniles compared to SAL-exposed controls. * P < 0.05 versus SAL-exposed controls. Data are represented as mean ± standard error of the mean.

As a consequence of the increased wake time in the light phase, total sleep time (TST; sum of the time spent in NREM and REM sleep) decreased significantly in the VPA-exposed animals by approximately 38 min compared to controls (t₉ = 3.126, P = 0.012; data not shown). Changes in NREM sleep accounted for the overall difference in TST. In the light phase, the VPA-exposed animals spent significantly less time in NREM sleep (∼30 min; t₉ = 2.621, P = 0.028; Figure 2B) with no significant differences in the dark phase (t₉ = 0.553, P = 0.594), whereas REM sleep was unaffected by VPA-exposure in either the light or dark phase (Figure 2C; t₉ = 0.920, P = 0.382, light phase and t₉ = 1.905, P = 0.09, dark phase). In contrast to individual 12-h phases, the 24-h total of REM sleep revealed that VPA-exposed animals spent significant less time in REM sleep (∼17 min) compared to the controls (t₉ = 2.799, P = 0.021; Figure 2C).

VPA Exposure Disrupts the Pattern of Juvenile Sleep/Wake Cycles

In utero exposure to VPA markedly changed the sleep/wake architecture (mean bout duration, bout number, and number of transitions into Wake, NREM, or REM sleep) compared to controls. VPA exposure significantly increased the mean duration of a wake bout in the light and dark phases by ∼117% and ∼70%, respectively, compared to the SAL-exposed animals (t₉ = 6.966, P < 0.0001, light phase and t₉ = 2.552, P < 0.031, dark phase; Figure 3A). The increase in bout duration was associated with a significant decrease in the number of wake bouts in the light phase only (t₉ = 6.532, P = 0.0001, Figure 3A).

Figure 3.

Valproic acid (VPA)-exposed juvenile rats have more consolidated sleep and wake behaviors. (A-top) During both the light and dark phases, VPA-exposed rats have, on average, longer bouts of wakefulness. (A-bottom) The increased duration of wake bouts is associated with a decrease in the number of wake bouts during the light phase only. (B-top) Non-rapid eye movement (NREM) sleep bouts, on average, are substantially increased in VPA-exposed rats as well during the light phase. (B-bottom) This increase in NREM bout duration is also associated with a decrease in the number of NREM bouts during the light phase. * P < 0.05 versus SAL-exposed controls. Data are represented as mean ± standard error of the mean.

For sleep, only NREM sleep demonstrated significant changes in architecture. Curiously, in the light phase, VPA exposure significantly increased the average duration of a NREM sleep bout by ∼36% (t₉ = 3.517, P = 0.007; Figure 3B) whereas the number of NREM bouts was significantly reduced by ∼32% compared to SAL-exposed animals (t₉ = 6.646, P < 0.0001, Figure 3B). For REM sleep, there were no significant differences in the mean bout duration or bout number in either phase (data not shown).

To assess whether VPA exposure affects the stability of maintaining a wake or sleep state, the number of transitions into and out of wake, NREM sleep, and REM sleep were quantified across the light and dark phases. In the light phase, the number of transitions to and from wake was significantly reduced by approximately 50% in VPA-exposed animals compared to the SAL- exposed animals (Table 1; wake to NREM: t₉ = 6.460, P = 0.0001, NREM to wake: t₉ = 6.509, P = 0.0001, and REM to wake: t₉ = 3.484, P = 0.007). No significant differences were observed in the dark phase or when transitioning between sleep states (Table 1).

Table 1.

Number of transitions to and from each vigilance state during the light and dark phase

VPA Exposure Results in Changes to the Power Spectra

Power spectral analysis was used to study the cortical EEG of the juvenile rats exposed to VPA or SAL in utero during the light and dark phase. We also investigated differences in power in wake, NREM sleep, and REM sleep. In the spectral analysis of all frequencies in the continuous EEG, VPA-exposure reduced the theta power in both the 12-h light and dark phases (Table 2, P < 0.05). When we calculated the mean power for each frequency band in for the individual vigilance states (wake, NREM sleep, and REM sleep) we found that VPA exposure reduced midrange frequencies while increasing higher range frequencies, primarily in wake and REM sleep; the two states characterized by EEG desynchrony (Figure 4A, 4C, and Table 3). In all vigilance states, VPA exposure decreased theta band densities (4-8 Hz) in both the light phase (wake: U = 0.0, P = 0.004; NREM: U = 2.0, P = 0.017; REM: U = 0.0, P = 0.004; Figure 4A-C) and the dark phase (wake: U = 2.0, P = 0.017; NREM: U = 3.0, P = 0.03; REM: U = 0.0, P = 0.004); Figure 4A-C). Most notable was the approximate 20% decrease in theta band density during REM sleep regardless of phase (Figure 4C). Alpha band densities (8-12 Hz) were also significantly decreased during wake in the light (10%) and dark (13%) phases in VPA-exposed rats compared to SAL-exposed controls (Table 3; light: U = 1.0, P = 0.009; dark: U = 0.0, P = 0.004).

Table 2.

Averaged power spectra (% total power) in the light and dark phase

Figure 4.

Valproic acid (VPA) exposure in utero reduced theta power but increased gamma power compared to saline (SAL). The average theta power in (A) wake, (B) non-rapid eye movement (NREM) sleep, and (C) rapid eye movement (REM) sleep is reduced during both the light and dark phases of VPA-exposed rats. In both the light and dark phases, gamma power is increased during (A) wake and (C) REM sleep in VPA-exposed rats compared to SAL-exposed controls. Gamma power (1-h bins) across the 24-h analysis period is significantly increased in VPA-exposed animals during (D) wake and (F) REM sleep bouts but not (E) NREM sleep bouts. * P < 0.05 versus SAL-exposed controls. Data are represented as mean ± standard error of the mean.

Table 3.

Mean power (% total power) for alpha, sigma, and beta frequency bands

Conversely, VPA exposure increased high frequency gamma band densities (30-80 Hz) during wake and REM sleep in the light phase (Figure 4A and 4C; wake: U = 1.0, P = 0.009; REM: U = 0.0, P = 0.004) and dark phase (wake: U = 0.0, P = 0.004; REM: U = 0.0, P = 0.004) compared to the SAL-exposed controls. Interestingly, the greatest magnitude of change in the frequency spectrum occurred in REM sleep gamma power in the light (∼50%) and dark (∼37%) phases. Sigma and beta also increased, albeit less reliably, during wake and sleep, primarily REM sleep, in the light phase (Table 3; wake: sigma, U = 0.0, P = 0.004; beta, U = 4.0, P = 0.052, and REM: sigma, U = 4.0, P = 0.052; beta: U = 3.0, P = 0.03) and dark phase (Table 3; wake: sigma, U = 3.0, P = 0.03; beta, U = 0.0, P = 0.004; NREM: sigma. U = 4.0, P = 0.052; and REM: beta, U = 3.0, P = 0.004) in VPA-exposed animals. Again, the most prominent changes were during REM sleep; there was an approximate 30% change in beta and 22% change in sigma.

Because spectral analysis removes time as a factor, it is difficult to address whether changes in the mean spectral powers are caused by the power itself or longer/shorter durations of the vigilance states. To address this, we specifically analyzed the mean gamma power by hour -across the 24-h period. There was a main effect (Figure 4D; F_1,23 = 14.50; P = 0.004) of in utero exposure on wake gamma power across the 24-h period. Similarly, there was a main effect (Figure 4F; F_1,23 = 147.9; P < 0.0001) of in utero exposure on REM sleep gamma power across the 24-h period. VPA-exposed animals had elevated gamma power across the 24-h period during wake and REM sleep bouts compared to SAL-exposed animals. There was no significant difference in NREM sleep gamma power across the 24-h period (Figure 4E).

VPA Exposure Results in Changes to the Cortical GABAergic System

In the current study, we observed varying degrees of tail malformations in the VPA-exposed animals that included barely discernable changes (Figure 5A; arrows) or kinks and bends (Figure 5A; asterisks). Western blot analysis of protein isolated from primary somatosensory cortex micropunches from all animals revealed no differences in the cortical GABAergic system in VPA-treated animals compared to SAL-treated controls (Figure 5B and 5C; GAD67: t₁₁ = 0.5943, P > 0.05; GAD65: t₁₁ = 0.9328; P > 0.05). Interestingly, the o.d. values for the two VPA-exposed animals with the barely discernable tail malformations (n = 2, circled in Figure 5C) appeared more similar to the controls. Thus, if the VPA-exposed animals were regrouped according to the severity of the tail malformations (minor, M and severe, S), then both isoforms of GAD (65 and 67) were significantly reduced in VPA-exposed animals with severe tail malformations (GAD 67: F_2,10 = 6.294, P = 0.02; GAD 65: F_2,10 = 4.616, P = 0.04) compared to SAL (Figure 5C; Newman-Keuls post hoc GAD 67: P = 0.04; GAD 65: P = 0.04) and to VPA-exposed animals with minor tail malformations (Figure 5C; Newman-Keuls post hoc GAD 67: P = 0.006; GAD 65: P = 0.02). There were no changes in GAD expression in the BF (data not shown; GAD 65: SAL-1.76 ± 0.67, VPA-2.07 ± 0.14, t₉ = 1.204, P > 0.05; GAD 67: SAL-0.61 ± 0.07, VPA-0.69 ± 0.05, t₉ = 0.870, P > 0.05).

Figure 5.

Cortical expression of GAD65 and GAD67 is reduced in valproic acid (VPA)-exposed juvenile rats. (A) Images of the tails of juvenile rats exposed to either saline (SAL) or VPA. The tails of VPA-exposed rats had varying degrees of malformations, including barely discernable changes (arrows; referred to as minor) to more severe kinks and bends (asterisks; referred to as severe). (B) Representative immunoblots from micropunches of the primary somatosensory cortex of SAL- and VPA-exposed rats. (C) Plots of the individual optical density (o.d.) values from SAL and VPA exposed animals. The circled points are values from VPA-exposed animals with minor tail malformations. (D) GAD65 and GAD67 protein expression is reduced in VPA-exposed rats with severe tail malformations compared to SAL-exposed controls and VPA-exposed animals with minor tail malformations. * P < 0.05 versus SAL-exposed controls and VPA-exposed animals with minor tail malformations. Data are represented as mean ± standard error of the mean.

DISCUSSION

In this study, sleep and wake behaviors as well as cortical EEG power spectral frequencies were analyzed in an animal model of ASD. We found that animals exposed to VPA in utero displayed consolidated bouts of arousal during the expected sleep phase compared to SAL-exposed controls. Perhaps equally exciting is our finding that changes in the cortical spectral densities and cortical GABAergic system correlate with this enhanced behavioral arousal. Together, our data suggest the occurrence of developmental changes in the neurocircuitry and/ or neurochemistry intimately involved in sleep-wake behavior in a teratogenic model of ASD.

Sleep in ASD

Studies investigating sleep in children with ASD have ranged from parent-reported and sleep diary studies to more objective studies including actigraphy, and polysomnographic (PSG) studies. Commonly, these studies indicate that children with ASD suffer from insomnia, particularly difficulties initiating and maintaining sleep, and parasomnias.^8,11–21 Parents report that bedtime routines are often challenging, which may lead to delayed sleep onset and increase sleep anxiety.^{8,13,15,16,18,19,21,60} Both questionnaire and PSG studies show that children with ASD acquire less total sleep compared to typically developing children, awaken frequently during the night and remain awake for periods of time, and overall experience less efficient sleep.^{8,13,18,21,61} More specifically, sleep spindles⁶² and the time spent in stage 3 NREM⁶³ and REM^13,64 sleep are decreased in ASD subjects compared to control subjects.

Parental reports indicate that children's sleep problems negatively affect both the children's and the families' daily functioning.¹⁶ Sleep disruptions in children with ASD are correlated with their daytime behavior, and the quality of their sleep may predict the severity of their behaviors.^14,24 In fact, treating the sleep problems has been shown to alleviate the severity of some of the characteristic behaviors associated with ASD.^65–67

Our findings are consistent with the clinical findings of sleep disruptions and patterns reported in the ASD phenotypes. Sprague-Dawley rats are nocturnal and acquire most sleep during their quiescent, or light, phase. In our model, in utero VPA exposure increases wakefulness at the expense of sleep, specifically NREM sleep, during the light phase. Once awake, these VPA-exposed animals maintain consolidated wake bouts compared to SAL-exposed controls. These consolidated, lengthy periods of wake during sleep mirrors the behavior seen in ASD.

GABAergic Dysfunction in ASD and Potential Effect on Sleep

Sleep is a complex behavior that requires cooperative actions of multiple brain regions. The GABAergic system is involved in the onset and regulation of sleep.⁴⁷ There are reciprocal GABAergic connections between major sleep-promoting regions, such as preoptic area, and wake-promoting areas, such as the lateral hypothalamus, which control the transitioning between vigilance states.⁶⁸ There is also a strong GABAergic innervation of the cortex by neurons in the BF, which influence cortical output and behavior.^48,69 Together, these GABAergic networks orchestrate the balance between sleep and wake states.

Like our behavioral findings, the decreases in GAD 65 and GAD 67 protein expression in the VPA-exposed rats compared to SAL are consistent with the clinical findings. Surprisingly, these differences in GAD65 and GAD67 protein expression were specific to the cortex and not found in the BF. Although we anticipated decreases in GAD 65 and GAD 67 in the BF, because of its role in the control of vigilance states, it is possible that our micropunches limited our analysis of GAD expression in this region. Micropunches only contain the components within the local circuit. GAD 65 is localized mainly to synaptic terminals, whereas GAD 67 is found both in cell bodies and in terminals.^70,71 Thus, the origin of the GAD in the current cortical punch technique cannot distinguish between GAD in terminal regions of BF projection neurons or cortical somas. Nevertheless, reduced cortical GAD 65/67 will result in decreased GABA production and ultimately decreased inhibitory tone, which will have a direct effect on EEG oscillations and may affect behavioral outputs. Ongoing laboratory studies are investigating GAD 65/67 protein in critical sleep nuclei to determine whether mechanisms controlling the onset and maintenance of sleep are disrupted in this model.

In ASD, there is clear evidence that the GABAergic system is substantially disrupted. GABAergic receptor expression and binding in various brain regions are substantially reduced in ASD.^72–75 Furthermore, GAD 65/67 messenger RNA and protein levels are reduced in parietal cortex and cerebellum postmortem tissue from individuals who had an ASD.^76–78 Because GAD is the enzyme responsible for the synthesis of GABA, it is likely that there is also a reduction in inhibitory signals. A recent in vivo study of children with ASD reported that GABA concentrations are reduced in the frontal cortex, leading to a lower [GABA]/[glutamate] ratio as measured by 3-T MRI.⁷⁹ Such findings are indicative of cortical disinhibition, which contribute to a more aroused state.

Excitatory/Inhibitory Imbalance in ASD

Imbalance in the excitatory/inhibitory (E/I) ratio has been implicated in the etiology of ASD.^80–82 Because balanced excitation and inhibition among local neural circuits is critical for the stable functioning of these circuits, it is plausible that this E/I imbalance can disrupt the delicate interplay between brain regions involved in complex behaviors, such as communication, sensory processing, social behaviors, and even sleep. Many studies support this notion that an imbalance, particularly an elevation, in the E/I ratio may underlie behavioral deficits associated with ASD. The abundance of evidence implicating an aberrant GABAergic system in ASD, described previously, could suggest that the inhibitory tone in ASD brains is reduced. Furthermore, various studies have linked ASD^83–86 and ASD-like behaviors in animal models^87–90 to a hyperexcited brain.

Spectral Analysis of EEG in ASD

Clinically, spectral analysis of EEG in ASD children has consistently revealed abnormal EEG characteristics. Recently in preclinical models, it has been shown that increasing the E/I ratio via optogenetic manipulations in the medial prefrontal cortex increases high-frequency activity, specifically in the gamma range.⁸⁵ Higher frequency bands, especially gamma (30-80 Hz) frequencies, are associated with cognition, attention, and sensory processing.^86,87 Elevated E/I ratio also impairs social behavior and conditioning,⁸⁵ behavioral deficits observed in ASD, and animal models of ASD. In our study, VPA exposure increases high-frequency EEG bands in the sigma, beta, and gamma range during wake and REM sleep. The decreases in GAD 65/67 protein expression in the cortex may influence the EEG frequency bands such that lower inhibitory tone (GABA) would lead to increased higher frequency oscillations.

Additionally, several clinical findings consistently demonstrate a reduction in the alpha power in children with ASD compared to controls.^88–91 In humans, the alpha bands have been associated with suppression of nonsalient inputs allowing for selective attention.^92,93 Our current data demonstrate that VPA-exposed rats also have significantly less alpha power during wake. This finding further supports the use of this model for future investigations, not only into the mechanisms underlying changes in sleep but also changes in EEG oscillations in ASD.

Circadian Disruptions in ASD

Sleep disruptions, especially those related to the timing of sleep, may be caused by disruptions in circadian rhythm. Many studies indicate that children with ASD may also suffer from circadian rhythm disorders.^1,93 The pineal gland, whose activity is regulated by the master pacemaker, secretes melatonin during the night. The rhythm of melatonin secretion as well as the amount is a proxy for circadian rhythmicity and duration of sleep acquired. In ASD, many studies report that melatonin levels, as well as its metabolite, are reduced.^94–99 Studies also show that melatonin is effective in treating some sleep disruptions in ASD.^{65,67,100,101} Our data and that of others suggest there are also circadian disruptions in the VPA model of ASD. Here, we show that prenatal VPA exposure decreases the amount of time animals spend asleep and increases wakefulness during the animals' sleep phase. These data support the findings by Tsujino and colleagues⁴⁶ that VPA exposure increases animals' locomotor and feeding behaviors during the quiescent phase compared to saline controls. Although melatonin levels have not been measured in the VPA model, chronic melatonin treatment has recently been shown to be effective at restoring social interaction, phosphorylation of calcium/calmodulin-dependent protein kinase II, N-methyl-d-aspartate receptor 1, and protein kinase A, as well as long-term potentiation deficits in the hippo-campus of VPA-treated rats.¹⁰²

Consequence to Sleep Loss

Sleep loss can have a significant effect on a child's overall health and quality of daily life.^22,23 As in adults, sleep loss in children and adolescents not only leads to fatigue but can also negatively impact physiological functions, mood, cognition, and behavior.^22,103,104 Insufficient sleep can cause poor academic performance,^104,105 increase irritability and aggressiveness, impair attention, and lead to hyperactivity.^22,106 In fact, sleep problems are often associated with behavioral problems seen in attention deficit hyperactivity disorders (ADHD).^107,108 Insufficient sleep is also associated with depression and anxiety.^25,26,109 Children with a diagnosis of ASD share similar symptoms¹¹⁰ and disrupted sleep may lead to these symptoms or exacerbate ASD symptomology.^14,24 Loss of REM sleep during critical periods during development can have profound effects on the developing brain, particularly the development on the sensory systems.¹¹¹ Adolescence is a time of brain maturation and reorganization¹¹²; therefore, it is possible that insufficient REM sleep during this period can alter further brain maturation in children with ASD.

CONCLUSION

Surprisingly, animal models of ASD have been underutilized in advancing our understanding of the etiology of sleep disruptions in ASD. Our findings suggest that juvenile rats exposed to VPA in utero display sleep behavior similar to that seen in children with ASD. These similarities validate the use of this model to further investigate whether there are biological mechanisms underlying these changes in sleep or if they are consequences ASD symptomology.

DISCLOSURE STATEMENT

This was not an industry supported study. Support was provided by a DOD Research Program AR080087 awarded to Dr. Mong. The authors have indicated no financial conflicts of interest.