Early Life Sleep Disruption Is a Risk Factor for Increased Ethanol Drinking After Acute Footshock Stress in Prairie Voles

Abstract

Early postnatal experiences are important for shaping the development of the stress response and may contribute to the later emergence of alcohol use disorders. We have previously found that early life sleep disruption impairs social development and alters GABA neurons in the brain of adult prairie voles, a socially monogamous rodent that displays natural ethanol preference in the laboratory. However, it is unclear whether these effects on social behavior are due, in part, to overall anhedonia and/or altered behavioral response to stress. To address this question, litters containing prairie vole pups were sleep disrupted by gentle cage agitation for 7 consecutive days from postnatal days (P) 14 to 21 (early life sleep disruption, or ELSD group) or allowed to sleep undisturbed (Control). Adult voles underwent a 2-bottle choice ethanol drinking procedure integrated with a single session of footshocks. Ethanol intake after footshock was measured as well as c-Fos immunoreactivity in the lateral and central amygdala. ELSD animals showed increased ethanol consumption and increased neural activity in these amygdala regions after footshock compared to control animals. There were no differences in baseline ethanol drinking prior to exposure to a stressor. These results suggest that early life sleep disruption in prairie voles does not produce anhedonia but can have long-lasting effects on stress reactivity. In addition to shaping species-typical social behavior, early life sleep may be important in the development of stress induced ethanol consumption and the activation of limbic pathways associated with stress.

Early life experiences shape the developing nervous system and are hypothesized to be major contributors to the later onset of neuropsychiatric diseases (Kessler et al., 2010; Kessler, Davis, & Kendler, 1997; Cohen et al., 2006). In particular, early life sleep may be critical for neural development, as evidenced by the fact that across mammalian species, juveniles spend more time asleep than adults (Jouvet-Mounier, Astic, & Lacote, 1969; Roffwarg, Muzio, & Dement, 1966). When considering that sleep disturbances and neurodevelopmental disorders are often highly comorbid (Buckley et al., 2010), early life sleep disruption may directly contribute to abnormal neuropathology and behavioral impairments observed in neuropsychiatric diseases of developmental etiology.

Prairie voles (Microtus ochrogaster) may represent a better rodent model of neurodevelopment than mice and rats due to the translational similarities of their social behaviors to humans (e.g., social monogamy, biparental care, incest avoidance; Carter, DeVries, & Getz, 1995; Getz & Carter, 1980). We have previously found that disrupted sleep early in life, including reduced duration of REM sleep and fragmented non-REM sleep, dramatically reduces the expression of species-typical social attachment behaviors in the socially monogamous prairie vole (Jones et al., 2019). We further found alterations in the development of GABAergic interneurons within the adult neocortex and reduced interaction with novel objects. This combined behavioral and neuropathological phenotype of decreased affiliative behavior, neophobia, and deficits in inhibitory neurotransmission is consistent with the constellation of symptoms seen in human neurodevelopmental disorders, and is supportive of our overall hypothesis that environmental sleep disruption could play a causal role in disease pathogenesis. However, this behavioral phenotype could also be explained more generally by overall anhedonia or altered responding to stress.

In order to directly test the hypothesis that early life sleep disruption (ELSD) affects hedonic pathways, we tested the effects of ELSD on voluntary ethanol drinking using the 2-bottle choice paradigm. Ethanol consumption was chosen due to the fact that we previously found that ELSD altered the development of GABAergic interneurons (Jones et al., 2019) and the fact that this species of rodent shows innate preference for unsweetened ethanol solutions (Anacker, Loftis, Kaur, & Ryabinin, 2011; Hostetler, Anacker, Loftis, & Ryabinin, 2012). In order to investigate the effect of ELSD on responses to stress in prairie voles, animals were exposed to a single session of unsignaled, inescapable footshocks and immediately provided continuous access choice to an ethanol solution. We then measured ethanol intake and preference, fear and anxiety related behaviors, as well as immediate early gene immunoreactivity in central pathways involved in emotional reactivity, including both the lateral and central nuclei of the amygdala (LeDoux, 2003; Ressler & Mayberg, 2007).

Method

Animals

Prairie voles were housed in temperature and humidity controlled rooms with a 14:10 light/dark cycle (lights on at 0700h), with ad libitum access to food and water. Cotton nestlets were provided weekly with cage change as nesting material. Litters were weaned on P21 into same sex littermate pairs (two per cage). Both males and females were used, and sex was used as a grouping variable for all analyses except c-Fos immunohistochemistry (IHC). Upon weaning, females were housed in a separate colony room from males. Female prairie voles are male induced ovulators and it is presumed all of the females used in this study were anestrus due to the lack of male presence in the colony room. Eleven breeder pairs with a total of 48 litters contributed to this study and every effort was made to counterbalance litter groups when possible. Prairie voles were between 80 and 130 days old at time of testing. Age was used as a covariate in analysis.

Experimental Overview

Male and female prairie voles underwent either ELSD (n = 58) or control (n = 72) sleep conditions from P14-P21 and were left undisturbed until adulthood. Adult animals were run in two cohorts that did not result in differences in ethanol drinking behavior, therefore data from these cohorts were collapsed for the purposes of behavioral analyses. Both cohorts were tested for five days in a 2-bottle choice procedure with 10% ethanol and water followed by a series of either moderate footshocks or a novel chamber with no shock, and reexposure to ethanol choice. In Cohort 1, voles were first screened for anxiety-like behavior with a 5-min light/dark box test prior to starting the five-Day 2-bottle choice test (n = 89). In this first cohort, ethanol consumption was recorded both 3 hr after footshock delivery and again the next morning. In Cohort 2, (n = 35), light/dark box screening was not conducted and instead, at the termination of the 3-hr drinking session following footshock, animals were euthanized and brains were collected for c-Fos IHC (see Figure 1 for experiment design). All procedures were approved by the Institutional Animal Care and Use Committee at the Portland VA Medical Center and were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Six animals were excluded from analysis due to equipment failure (n = 2 male control; n = 1 male ELSD; n = 1 female control; and n = 2 female ELSD). Final group numbers were as follows: n = 40 male control (n = 23 No Shock; n = 17 Shock); n = 29 male ELSD (n = 13 No Shock; n = 16 Shock); n = 29 female control (n = 13 No Shock; n = 16 Shock); and n = 26 female ELSD (n = 9 No Shock; n = 17 Shock).

Figure 1.

Experiment design. Prairie voles underwent either ELSD or control sleep conditions from P14-P21 and were left undisturbed until adulthood. Voles had 5 days of 2BC in the home cage with 10% ethanol in water. After a 24 hr washout, approximately half of the animals were shocked and the other half exposed to a novel context without shock. All animals were again given access to ethanol choice in the 2BC. In Cohort 1, bottles were read at both 3 hrs and 24 hrs after footshock. In Cohort 2, 3 hr after shock delivery, animals were euthanized and brains collected for c-Fos IHC. ELSD = early life sleep disruption; 2BC = 2-bottle choice; IHC = immunohistochemistry. The gray/white box at P73 indicates the light/dark anxiety test. Final group numbers were as follows: n = 40 male control (n = 23 No Shock; n = 17 Shock); n = 29 male ELSD (n = 13 No Shock; n = 16 Shock); n = 29 female control (n = 13 No Shock; n = 16 Shock); and n = 26 female ELSD (n = 9 No Shock; n = 17 Shock).

Early Life Sleep Disruption (ELSD)

To disrupt sleep early in life (ELSD group), home cages containing litters of prairie voles were placed on a laboratory orbital shaker (Jones et al., 2019; Li et al., 2014; Sinton, Kovakkattu, & Friese, 2009) for one full week from P14-21. Cage card holders were locked into place to limit the amount of auditory disruption that occurred during shaking. The orbital shaker was connected to an automatic timer programmed to gently agitate (110RPM) for 10 s every 110 s (see Jones et al., 2019 for additional details). Water bottles were removed, and hydrogel was provided as an alternative to water to prevent spillage during shaking. The Control group was placed in the same room as the shakers on P14 but were not physically disturbed. In prairie vole pups, this method of sleep disruption results in a significant decrease in REM sleep and fragmented non-REM (NREM) sleep compared to controls (Jones et al., 2019). Importantly, this sleep disruption paradigm does not result in changes in parental care, weight of the pups, or serum corticosterone levels in pups relative to animals reared in standard housing conditions, suggesting this paradigm alters sleep with minimal stress to the animal during development (Jones et al., 2019).

Light/Dark Box

In Cohort 1, adult prairie voles underwent a 5-min light/dark box test to quantify anxiety like behaviors. The light/dark box was a modified open field apparatus with a dark insert (Omnitech Electronics, Columbus, Ohio) consisting of a large Plexiglas enclosure (60 cm by 60 cm by 30 cm) affixed with a black box insert (30 cm by 30 cm by 20 cm) with a door opening (10 cm by 5 cm) that allowed animals to freely cross from one side to the other side. Additional lighting was added to the lit portion of the box to achieve a lux level of 3000. Voles were placed in the lit compartment at the beginning of the test and allowed to ambulate freely for five minutes. Automatic beam break data was recorded in both the light and dark zones. The light/dark box test was conducted as adults approximately 7–9 days prior to the start of two-bottle choice. One animal was excluded from the light/dark box analysis because the test was disrupted before completion.

Two-Bottle Choice (2BC)

As adults, prairie voles were given two 25 mL glass cylinder tubes affixed with rubber stoppers and sipper tubes—one filled with standard drinking water from an autoclaved water bottle, and the other with 10% ethanol in drinking water (vol/vol). This volume allows voles ad libitum access to both water and ethanol for the duration of the 2-bottle choice (2BC) test. A chicken wire mesh divider (7 mm openings) was affixed down the middle of each cage (Anacker et al., 2014; Anacker, Loftis, Kaur, et al., 2011; Walcott & Ryabinin, 2017) to provide semisocial housing (e.g., allowing exposure to bedding of cage mate and limited physical contact) while restricting access to only one pair of drinking tubes to obtain accurate ethanol intake values for each individual animal. Bottles were read daily at 0800h, filled with fresh solution, and sides alternated to avoid bias. An empty cage with water bottles and no animals was placed on each rack to record spillage. There was never more than 0.05 mL that leaked in the empty cage. Total volume consumed from both bottles, amount of ethanol consumed (g/kg), and ethanol preference were all recorded daily. Ethanol preference was calculated as fluid consumed from the ethanol bottle divided by total fluid consumed from both bottles. Ethanol consumption was measured in g of ethanol (visually read from bottles) per kg of body weight (animal’s body weight on day of bottle reading). Age in days, calculated from the first day of baseline ethanol drinking was used as a covariate in analysis.

Acute Footshock Stress

After five days of 2BC, the bottle containing 10% ethanol was removed and replaced with a second water tube for 24 hr. After 24 hr of water access only, voles were placed individually into fear conditioning chambers (Omnitech Electronics, Columbus, OH) and immediately received a series of five unsignaled footshocks (0.7mA intensity, 1 s duration, variable intertrial interval (ITI) = 30 s; Shock group) or were placed in chambers for an equivalent period of time without shock delivery (No Shock group). Freezing was scored from video collected for the entire duration of exposure to the fear conditioning chamber and animals combined from Cohorts 1 and 2. Percent freezing was calculated by dividing the total time freezing between shock deliveries by the amount of time in between shocks, this resulted in 6 intervals of freezing for analysis (preshock, intershock intervals 1–5, postshock). Immediately after termination of the last footshock, voles were removed from the chambers and placed in new cages with access to two bottles (one containing water, the other containing 10% ethanol in water) for three hours. Bottle readings were taken at the end of these three hours and again the following morning in a subset of animals (Cohort 1).

Euthanasia and Tissue Collection

In Cohort 1, animals were euthanized by CO₂ inhalation followed by cervical dislocation, once the 2BC procedure was completed. In Cohort 2, animals were euthanized via overdose of isoflurane gas and rapidly decapitated, and brains extracted at the end of the 3-hr drinking session post footshock. This cohort did not undergo light/dark box testing prior to ethanol drinking, however all other behavioral procedures were the same as previously described. Brains were removed and immersion fixed into 4% para-formaldehyde in 0.1 M phosphate buffer (pH 7.4). As neuronal c-Fos protein levels peak at approximately 90 min after stimulus exposure, the timepoint chosen in this study is thought to represent a sample of neural activity from the middle of the drinking session post footshock. Brains were stored at 4 °C and 24 hr later brains were transferred to phosphate buffered saline (PBS) with sodium azide until sectioning at 40 μm on a vibratome (Precisionary Instruments, Greenville, North Carolina). Coronal brain sections were stored in PBS + azide in four series and c-Fos IHC was conducted on one series per brain (e.g., every 4th section).

C-Fos Immunohistochemistry (IHC)

c-Fos IHC was conducted on free floating sections from one series of sections in two separate batches. Sections were initially washed three times (10 min per wash) in PBS, followed by a 15-min incubation in 0.03% H₂O₂ in PBS and three more washes in PBS. Sections were then blocked with normal goat serum (6%; Vector) in PBS with 0.2% triton X-100 (PBST) for one hour and transferred to wells containing primary anti-c-Fos polyclonal primary antibody (1:10000; Santa Cruz Biotechnology sc-52) in PBST and normal goat serum (3%; Vector) for 72 h at 4 °C. The sections were washed three times in PBS and incubated with goat antirabbit biotinylated secondary antibody (1:200; Vector) in PBST and normal goat serum (3%) for two hours at room temperature, washed three times in PBS, incubated with avidin— biotin peroxidase complex (1:500 in PBS; Vector ABC kit) for one hour at room temperature, and washed three times in PBS. The sections were incubated in distilled water with diaminobenzidine tablets (DAB; Sigma), 2% nickel sulfate in distilled water (wt/vol), and hydrogen peroxide for three minutes. The staining reaction was stopped by three washes with PBS. The reaction resulted in a blue-black stain within the nuclei of c-Fos immunoreactive neurons. The sections were mounted on gelatin coated slides, air dried for 48 hr, dehydrated in ethanol solutions and cleared with xylenes, and cover slipped with Permount (Fisher Chemicals).

Cell Quantification

c-Fos immunoreactivity was expressed as density of immunopositive cells. For each brain section, the number of c-Fos immunopositive nuclei in a given brain structure was counted and divided by the area occupied by counting frames placed within each structure (in millimeters squared). The borders of the lateral (sampled from sections located between bregma levels −1.34 to −2.18 mm) and central (sampled from sections located between bregma levels −1.06 to −1.70 mm) amygdala were determined with the use of methylene blue stained adjacent sections and were selected according to regional staining patterns seen in a standard mouse brain atlas (Franklin & Paxinos, 1997). Counting frames of known area (LA total area per image = 0.0285 mm²; CeA total area per image = 0.0242 mm²) were placed within the borders of the region for consistency of area sampled. Cell counting was conducted manually by an experimenter blind to condition with the aid of an image analysis computer program (ImageJ, NIH) on 3–6 sections per animal from both hemispheres. The counts obtained per section were then averaged together to create a single representative value for each animal (see Jones, Ringuet, & Monfils, 2013; Knapska & Maren, 2009).

Statistics

All data met assumptions for parametric testing and were analyzed with ANCOVA using sex, early life sleep group, and shock group (where applicable) as between subjects factors and age and amount of ethanol ingested in g/kg (where applicable) as covariates. If significant interactions were found, follow up two-tailed independent sample t tests were conducted. An alpha value of 0.05 was required for significance. Statistical analyses were performed using SPSS Version 25.0 and figures were generated in GraphPad Prism Version 8.0.

Results

Early Life Sleep Disruption Does Not Influence Adult Ethanol Consumption Over the 5-Day Baseline Drinking Period

There was no effect of early life sleep group or sex on the average daily amount of ethanol consumed over the 5-day baseline period of 2BC (Figure 2a; Two-way ANCOVA: main effect of sex: F_1,119 = 0.561, p = .455; main effect of sleep group: F_1,119 = 0.031, p = .861; no interaction F_1,119 = 0.004, p = .949); however the covariate of age in days was significant (F_1,119 = 7.259, p = .008), with older animals consuming more ethanol by body weight, and was therefore included in all future analysis. Although there was not an effect of sleep group on average ethanol preference (Two-way ANCOVA: main effect of group: F_1,119 = 0.004, p = .951), females had significantly lower ethanol preference than males (Figure 2b; main effect of sex: F_1,119 = 6.065, p = .015; effect of age as covariate F_1,119 = 8.852, p = .004), consistent with one prior report in prairie voles (Hostetler et al., 2012) and in contrast to other rodent studies where either females preferentially consume more ethanol than males (Lancaster & Spiegel, 1992; Lopez, Doremus-Fitzwater, & Becker, 2011; Tambour, Brown, & Crabbe, 2008) or where there was no sex difference reported (Anacker, Loftis, & Ryabinin, 2011; Juárez and Barrios De To-masi, 1999).

Figure 2.

Early life sleep disruption (ELSD) does not affect adult ethanol intake across 5 days of 2-bottle choice (2BC). (a) Average g/kg EtOH consumed over 5 days of continuous access 2BC (one bottle contained 10% EtOH in water, the second bottle contained drinking water) was similar in ELSD and Control prairie voles (no effect of sex, males and females collapsed). (b) Average ethanol preference over 5 days of continuous access 2BC was also similar in ELSD and Control prairie voles; however, females on average had a lower ethanol preference than males. Cohorts 1 and 2 combined. n = 40 male control; n = 29 male ELSD; n = 29 female control; n = 26 female ELSD. Dotted box corresponds to data shown. Bar height is mean, error bars ± SEM. * p < .05.

ELSD Does Not Influence Anxiety or Acute Fear-Related Behaviors

Anxiety measures in the light/dark box were analyzed using MANCOVA with early life sleep group and sex as the between subjects factors, a priori determined dependent variables of interest were: duration in the door adjoining the light and dark zones, latency to first enter the dark zone, and duration spent in the light portion. There was no effect of early life sleep on any of these variables, however, there were main effects of sex, where females spent less time in the light (Figures 3a; see Table 1 for full statistical values).

Figure 3.

Anxiety like behaviors and acute fear responding was not affected by early life sleep disruption (ELSD). (a) There were no main effects of ELSD on anxiety like behaviors in the light/dark box (LD Box; 5-min test) although females spent less time in the light portion of the chamber, one indicator of increased anxiety in females (Cohort 1 only). (b) There were no main effects of ELSD on freezing between shock deliveries but there was a trend toward increased freezing behavior in females. N = 12–15/group/sex. Cohorts 1 and 2 are combined. Dotted boxes correspond to data shown. Bar height and symbol are group mean, error bars ± SEM. 2BC = 2-bottle choice. ** p < .01.

Table 1.

Light-Dark Box F Statistics and p Values Analyzed With MANCOVA

Behavior analyzed	Sleep group	Sex group	Sex × Sleep Group	Age covariate
Time in door	F1,83 = .276, p = .601	F1,83 = 2.055, p = .155	F1,83 = .774, p = .382	F1,83 = 13.586 p = .001
Latency to exit light	F1,83 = .856, p = .601	F1,83 = .442, p = .508	F1,83 = .796, p = .375	F1,83 = .302, p = .584
Time in light	F1,83 = .184, p = .669	F1,83 = 8.961, p = .004	F1,83 = .084, p = .773	F1,83 = 1.632, p = .205

Note. Time in the door transition zone (part of the dark total zone), latency to exit the light portion, and total time spent in the light portion were analyzed for the entire 5 min test. Age was used as a covariate in analysis. N = 88.

After five days of 2BC, ethanol was removed and all subjects had access to two bottles of plain water for 24 hr before receiving five footshocks. Freezing was analyzed with a repeated measures two-way ANCOVA with interval number as the within subjects factor, sex and sleep group as between subjects factors, and age as a covariate. There was a significant increase in freezing over the course of shock delivery (within subjects F_5,250 = 2.379, p = .039), no within subjects interactions (trial X age F_5,250 = 1.646, p = .149; trial X sex F_5,250 = 1.489, p = .194; trial X sleep group F_5,250 = 0.275, p = .926; trial X sex X sleep group F_5,250 = 0.994, p = .422), and no significant differences between ELSD and Control groups (between subject effect of sleep group F_1,50 = 0.001, p = .979; covariate p = .049) although there was a trend toward increased freezing in females (between subject effect of sex F_1,50 = 3.018, p = .089; Figure 3b), which is consistent with increased time spent in the dark portion of the light/dark box observed in female prairie voles compared to males.

Adults Subjected to ELSD as Juveniles Increased Their Ethanol Intake Immediately Following Footshock

Ethanol intake was recorded 3 hr after footshock (or no footshock), and again 24 hr later in a subset of animals. Three way ANCOVA with sex, sleep group, and shock group as between group factors and age as a covariate revealed that animals subjected to ELSD from P14-P21 consumed higher amounts of 10% ethanol at the 3 hr timepoint (main effect of sleep group F_1,115 = 5.823, p = .017) with no main effect of shock (F_1,115 = 2.914, p = .090) and no main effect of sex (F_1,115 = 0.001, p = .989). There was a significant Sleep × Shock Group interaction (F_1,115 = 4.481, p = .036). Follow up independent sample t tests revealed higher ethanol consumption in ELSD voles that were exposed to footshock compared to both control voles exposed to footshock (t₆₄ = 3.252, p = .002) and ELSD voles that did not receive a footshock (t₅₃ = 2.011, p = .049; Figure 4a).

Figure 4.

Early life sleep disruption (ELSD) voles exposed to footshock consumed high levels of ethanol immediately after shock but returned to control levels after 24 hr. (a) Ethanol intake (g/kg) 3 hr after shock or no shock in ELSD and control voles (males and females combined). ELSD voles that were shocked consumed significantly more ethanol (10% vol/vol in water) than both Control voles that were shocked and their nonshocked ELSD counterparts over the 3 hr period (Cohorts 1 and 2 combined). (b) Ethanol intake (g/kg) 24 hr after shock or no shock revealed no differences in total consumption between ELSD and controls (Cohort 1 only) over a 24 hr period. Dotted boxes correspond to data shown. Bar height is mean. Error bars ± SEM. 2BC = 2-bottle choice. N = 12–15/group/sex. * p < .05. ** p < .01.

There were no differences in total fluid consumed (both bottles; Main effect of sex: F_1,115 = 0.961, p = .329; sleep group: F_1,115 = 1.688, p = .197; Shock group: F_1,115 = 0.516, p = .474). Bottles were read again 24 hr later in Cohort 1 and no differences in ethanol intake (main effect of sleep group F_1,79 = 0.912, p = .342; main effect of shock group F_1,79 = 1.329, p = .252; Sleep × Shock interaction F_1,79 = 0.461, p = .499; Figure 4b) or preference (main effect of sleep group on preference: F_1,79 = 0.540, p = .465; shock group: F_1,79 = 3.016, p = .086; Sleep × Shock interaction F_1,79 = 0.743, p = .391) were found between any groups at this later time point.

Immediate Early Gene Activity in Central and Lateral Amygdala Was Higher in ELSD Prairie Voles Than Controls

c-Fos immunoreactivity (ir) was quantified in two brain regions implicated in anxiety and fear behavior: the central (CeA) and lateral (LA) portion of the amygdala (Figure 5a). Due to experimenter error during brain extraction and tissue processing, only a subset of animals from Cohort 2 were used for immunohistochemistry (n = 15). Sample sizes were small, however, because there was no effect of sex on ethanol preference change after shock and preliminary analysis revealed no effect of sex on c-Fos-ir (p = .268), males and females were combined for analysis and sex was not used as a grouping factor. As anticipated, there was significantly more c-Fos-ir in the CeA in animals that were shocked. However, there was also a main effect of sleep group on c-Fos-ir in CeA, with increased activity in the ELSD animals as well as a Sleep × Shock interaction (Two way ANCOVA (groups = shock group, sleep group; covariates = age, postshock g/kg) main effect of shock group F_1,18 = 66.275, p < .001; main effect of sleep group F_1,18 = 10.711, p = .004, Shock × Sleep interaction F_1,18 = 6.162, p = .023; Covariate age F_1,18 = 10.478, p = .005, postfc g/kg F_1,18 = 0.073, p = .790). Follow up t tests revealed that ELSD voles that were shocked showed significantly greater CeA activation than control voles that were shocked t₁₂ = 2.924, p = .013 (Figure 5b). A similar pattern was found in the LA where there was a significant effect of shock (F_1,17 = 17.392, p = .001) but not sleep F_1,17 = 2.655, p = .122 as well as a Sleep Group × Shock Group interaction F_1,17 = 5.156, p = .036. Neither covariate were significant in the LA (age p = .334, postshock g/kg p = .784). Follow up t tests revealed that ELSD voles that were shocked had greater LA activation than Controls (t₁₁ = 2.952, p = .013; Figure 5c).

Figure 5.

c-Fos immunoreactivity was measured within the central (CeA) and lateral (LA) nuclei of the amygdala (a). There was increased c-Fos in early life sleep disruption (ELSD) animals after shock in both the CeA (b) and LA (c) compared to both nonshocked animals and shocked Controls. N = 3–8/group. Cohort 2 only. Dotted box corresponds to data shown, dotted shapes within central and lateral amygdala correspond to counting frames used for immunohistochemistry (IHC) analysis. Error bars ± SEM. 2BC = 2-bottle choice. * p < .05. *** p < .001.

Discussion

Early Life Sleep Disruption Does Not Alter Baseline Ethanol Intake in Adults but Does Enhance Behavioral and Neural Response to Acute Footshock Stress

In the studies presented here, we found that one week of ELSD from P14-P21 did not affect the hedonic response of prairie voles as measured by voluntary ethanol ingestion over a 5-Day 2BC paradigm. We further show that although anxiety like behavior in the light dark box was not altered after ELSD, stronger physical stressors such as a series of moderate footshocks lead to acute increases in ethanol ingestion and activation of fear and stress reactive regions in the brain compared to control voles that underwent the same series of stressors. This finding is in agreement with our previous conclusions that ELSD-induced social impairments (reduced social huddling with a pair bonded partner) in this group (see Jones et al., 2019) are not confounded by increased anxiety or reduced motivated reward behavior, but could be due to increased stress reactivity to a strong enough stress event.

The lack of differences in baseline drinking levels between early life sleep groups suggest that reward pathways as they relate to ethanol consumption are still intact in this model of impaired social development. Although we did not directly measure social reward in this experiment, our results suggest that reward seeking behavior such as that observed in the voluntary ethanol drinking quantified here are still present after ELSD. This observation might indicate that the social bonding impairments previously observed in our ELSD prairie vole model (see Jones et al., 2019) are not due to overall anhedonia. Future studies will investigate specific social rewards including the salience of the partner and stranger stimulus animals to ELSD and control prairie voles.

Importantly, all groups equally acquired a fear response to an unconditioned footshock stimulus as inferred by freezing behavior between footshock deliveries, indicating the conditioned fear response to footshock is unchanged between ELSD and control early life groups. In the 3 hr immediately following the footshock stressor, ELSD voles displayed higher ethanol intake than both ELSD voles that were not shocked and control voles that were subjected to the same series of footshock stressors. This increase was no longer present 24 hr after the stressor, suggesting this effect is acute in nature. Future studies will examine the role of ELSD on learning and memory, as one drawback to the experimental design presented here is that there was not an opportunity to test voles for fear memory recall or extinction.

Using immunohistochemistry, we stained for the protein product of the c-Fos immediate early gene, a marker of neural activity which increases in the lateral and central amygdala after footshock delivery in rodents (Campeau et al., 1991; Pezzone, Lee, Hoffman, & Rabin, 1992; Rosen, Fanselow, Young, Sitcoske, & Maren, 1998). Although acute ethanol consumption has been shown to induce c-Fos within the central amygdala of mice (Bachtell, Wang, Freeman, Risinger, & Ryabinin, 1999; Hitzemann, Hitzemann, & Research, 1997), some research suggests this effect is not specific to ethanol (Ryabinin, Bachtell, Freeman, & Risinger, 2001) and does not occur after sustained high use (Ryabinin, Galvan-Rosas, Bachtell, & Risinger, 2003). It is possible that the increased c-Fos ir in CeA in the current study could be due to the increased ethanol intake in a novel cage in the ELSD animals that were shocked. Research in rats suggests that ethanol consumption is affected by lesions to the central, but not lateral, amygdala (Möller, Wiklund, Sommer, Thorsell, & Heilig, 1997) and c-Fos ir increases within the central but not lateral amygdala after injections of ethanol (Ryabinin, Criado, Henriksen, Bloom, & Wilson, 1997). Of note, there was not a significant effect of g/kg of ethanol consumed in the 3 hr prior to euthanasia on c-Fos immunoreactivity in either the central or lateral nucleus of the amygdala, however, due to the small sample size we cannot rule out this possible connection. A second possibility is that the observed increases in c-Fos ir in the amygdala regions studied here are not be due to the ingestion of ethanol, but could be attributed to other factors known to increase neuronal activity in the amygdala such as reacting to fear and stress events. While it is not possible to dissociate the contributions of these scenarios in the current experimental design, given the observed role of the amygdala in both behaviors, it is likely that both ethanol consumption and stress responding contribute to the increase in c-Fos ir described after footshock in adult voles that underwent ELSD. Future experiments could control for ethanol intake or measure c-Fos ir after ELSD in the absence of ethanol exposure to attempt to better dissociate these mechanisms.

Our results indicate that prairie voles subjected to ELSD showed both increased ethanol consumption after an acute stressor, and concomitantly increased activation of key regions of the neural circuit of the stress response, including both the lateral and central nucleus of the amygdala, compared to controls. Of note, there could be overlap in the brain regions activated in response to stress and ethanol consumption. In our study, c-Fos immunoreactivity was quantified after a novel stressor followed by limited voluntary access to ethanol, both of which could induce c-Fos activity within the central nucleus of the amygdala, and it is not possible to dissociate these stimuli in the current design (Bachtell et al., 1999).

Sex Differences in Anxiety Like Behaviors but Not Stress Induced Ethanol Ingestion

These studies were designed to examine sex differences in ethanol drinking and stress responding by running experiments in males and females in parallel. Although there were main effects of sex on both anxiety like behavior and overall ethanol preference, there were no significant interactions with early life sleep status and sex or overall sex effects on ethanol after footshock. Increased time spent in the dark portion of the light/dark box is one indicator of higher anxiety levels in female prairie voles. In this set of experiments, female voles also showed a trend toward increased freezing between shock deliveries, compared to males. Combined, these results suggest that the female prairie voles tested here had some increase in anxiety-like behaviors compared to males, but showed no differences in how they responded to stress with ethanol consumption.

Enhanced Stress Response in ELSD May Be Related to Altered GABAergic Development

Many of the systems that underlie sleep and arousal (e.g., GABA, dopamine) are directly targeted by ethanol and can be studied in rodents using ethanol self-administration paradigms. Emotional responding relies on proper amygdala functioning, with dysfunction underling both anxiety (Tye et al., 2011) and substance abuse (Koob, 2008) disorders. Animal studies of Pavlovian fear conditioning have provided a wealth of information regarding the specific neural mechanisms that underlie learning and responding to potential threats. Multisensory information from the thalamus, cortex, and brainstem converges at the lateral nucleus of the amygdala which, along with the basolateral nucleus projects to the central amygdala (CeA; Duvarci & Pare, 2014; Johansen, Cain, Ostroff, & LeDoux, 2011; LeDoux, Cicchetti, Xagoraris, & Romanski, 1990). The CeA integrates this information in order to produce behavioral and physiological responses to threat (Iwata, Chida, & LeDoux, 1987; Rosen, Hitchcock, Sananes, Miserendino, & Davis, 1991).

Although increased amygdala activity is normal during acute periods of stress or fear learning (Cheng, Knight, Smith, & Helmstetter, 2006; Cheng, Knight, Smith, Stein, & Helmstetter, 2003), too much activity or sustained neural activity within the amygdala is a hallmark of disorders such as posttraumatic stress disorder (PTSD; Morey et al., 2009; Shin & Liberzon, 2010; Etkin & Wager, 2007; Semple et al., 2000), generalized anxiety disorder (Nitschke et al., 2009; Whalen et al., 2008), and panic disorder (Domschke et al., 2008; Van den Heuvel et al., 2005) and the suppression of fear depends on GABAergic activity within the extended amygdala (Amir, Amano, & Pare, 2011; Ehrlich et al., 2009; Sierra-Mercado, Padilla-Coreano, & Quirk, 2011). While less is known about the neural microcircuitry and cellular profile of self-administered ethanol, emerging research implicates many overlapping processes with threat responding, including the activation of the GABAergic system in the brain and neural activity within the central and lateral nuclei of the amygdala (Gass et al., 2014; Mody, Glykys, & Wei, 2007; Roberto, Madamba, Stouffer, Parsons, & Siggins, 2004; Weiner & Valenzuela, 2006; Zhu & Lovinger, 2006).

Development of GABAergic interneurons that express the calcium binding protein parvalbumin are especially sensitive to interrupted periods of sleep during the early postnatal period (Hogan, Roffwarg, & Shaffery, 2001; Jones et al., 2019), which suggests that the inhibitory GABAergic system in the brain may be vulnerable to the effects of ELSD. Indeed, previous research in our lab has found this method of ELSD in prairie voles reduces social huddling with an opposite sex partner and also increases parvalbumin-ir in the neocortex (primary somatosensory cortex, S1). These cortical GABAergic interneurons are the primary target of amygdala afferents which also mature slowly through development and may underlie problems with emotional reactivity that accompany many neurodevelopmental disorders (Cunningham, Bhattacharyya, & Benes, 2008). In our prairie vole ELSD model, we hypothesize that GABA dysfunction within S1 may interfere with the ability to properly process tactile information concordant with emotional experiences, including both social information from a pair bonded partner and, potentially, threat information from a footshock. Hyperactivity within the amygdala in response to stress may then be part of a negative feedback loop within the cortical GABAergic system known to be targeted by ELSD.

Given that one of the main actions of ethanol involves GABA, it stands to reason that the sensitivity of the GABA circuitry in the brain to early life sleep may have long lasting effects on related outcomes, including compounds that target GABA in the brain such as ethanol, or situations that heavily rely on GABAergic transmission (e.g., anxiety regulation, sleep). We further suggest that some of the inconsistencies in previous animal work regarding how animals use ethanol after an acute stressor could, in part, be due to differences in early life experiences such as sleep. Future studies could examine developmental trajectories of GABAergic-specific neural activity after early life sleep disruption.

ELSD in Prairie Voles as a Model of Acute Stress Induce Ethanol Ingestion

Prairie voles voluntarily consume unsweetened ethanol and prefer ethanol over water during even very brief exposure periods (Anacker, Loftis, Kaur, et al., 2011; Anacker, Loftis, & Ryabinin, 2011; Anacker et al., 2014; Hostetler et al., 2012; Hostetler & Ryabinin, 2014; Stevenson et al., 2017); however, the effects of stress on alcohol drinking in prairie voles had not yet been elucidated. Rat and mouse models of ethanol drinking in response to acute stress produced widely varied results with some paradigms and models resulting in mostly either decreased drinking (Bond, 1978; Champagne & Kirouac, 1987; Darnaudéry et al., 2007) or no change in drinking (Brunell & Spear, 2005; Chester, Barrenha, Hughes, & Keuneke, 2008; Fidler & LoLordo, 1996; Matthews et al., 2008; Myers & Holman, 1967; Powell, Kamano, & Martin, 1966), with only some leading to temporarily increased drinking (Füllgrabe, Vengeliene, & Spanagel, 2007; Mills, Bean, & Hutcheson, 1977; Vengeliene et al., 2003).

One limitation of our model, which is consistent with decades of animal research (Lopez, Anderson, & Becker, 2016; Mills et al., 1977; Vengeliene et al., 2003; Sillaber et al., 2002), is a failure to produce a sustained increase in ethanol intake or preference after footshock. Our model may better mirror motivated drinking as a coping mechanism for acute stress instead of the development of alcohol use disorder in response to trauma. Recent research sugests that exposure to predator odor in rodents produces a more robust increase in ethanol drinking compared to traditional footshocks in both rats (Edwards et al., 2013; Manjoch et al., 2016) and mice (Finn et al., 2018), and future studies in our lab will examine this more ethologically relevant form of stress on ELSD prairie voles. Furthermore, we did not measure blood ethanol concentration in these studies, as our primary question was one of behavioral and neurological change in response to footshock, and BECs are typically low in limited access studies such as these (unlike mice, prairie voles do not concentrate their ethanol intake to their dark cycle [Anacker, Loftis, Kaur, et al., 2011]).

Our results support the idea that early life sleep quality may be a critical developmental factor that shapes how animals respond to stress, including their propensity to voluntarily self-administer ethanol after a moderately stressful event. We propose that impaired development of GABAergic neurons in the somatosensory cortex results in an increased response to a strong somatosensory event (e.g., footshock) that manifests as increased ethanol intake, which is known to target GABA in the brain. Future work will determine if acute stress or strong tactile stimulation is responsible for this increase in ethanol intake and whether primary somatosensory GABA neurons are necessary and/or sufficient to recapitulate this enhanced ethanol response to stress.