Biological sex influences sleep phenotype in mice experiencing spontaneous opioid withdrawal

Abstract

Aversive symptoms, including insomnia experienced during opioid withdrawal, are a major drive to relapse; however, withdrawal-associated sleep symptomatology has been little explored in preclinical models. We describe here a model of opioid withdrawal in mice that resembles the sleep phenotype characteristic of withdrawal in humans. Male and female C57BL/6 mice were instrumented with telemeters to record EEG, EMG, activity, and subcutaneous temperature. All mice received 2 treatments separated by a 16-day washout period: (1) saline (volume:10 ml/kg) or (2) ascending doses of morphine (5, 10, 20, 40, and 80 mpk; volume:10 ml/kg) for 5 days at Zeitgeber time (ZT)1 and ZT13. Recordings for the first 71 hours after treatment discontinuation (WD1–3) and for 24 hours on withdrawal day (WD)5 and WD7 were scored for sleep/wake state and sleep architecture and EEG spectral data were analyzed. Morphine was acutely wake- and activity-promoting and NREM and REM sleep were increased during the dark phase on WD2 in both sexes. While NREM delta power (0.5–4.0 Hz), a measure of sleep intensity, was reduced during the light phase on WD1 and the dark phase on WD2 in both sexes, female mice also exhibited changes in the duration and the number of bouts of sleep/wake states. These observations of fragmented sleep on WD1–3 suggest poorer sleep consolidation and a more pronounced withdrawal-associated sleep phenotype in female than in male mice. These data may indicate a greater sensitivity to morphine, a more distinct aversive sleep phenotype, and/or a faster escalation to dependence in female mice.

Introduction

Opioid overdoses are epidemic in the United States with more than 564,000 opioid-related deaths from 1999–2020 (“ WONDER,” 2021). The addictive potential of opioids is well-known, with 21–29% of patients misusing prescribed opioids and 8–12% developing opioid use disorder (OUD) (Vowles et al., 2015). Relapse occurs in 63–91% of patients that have undergone treatment for OUD, with 59% of the initial relapses occurring within one week of discharge (Broers et al., 2000; Smyth et al., 2010). Of patients that experience relapse, 92% re-enter treatment, completing a vicious cycle of opioid use, treatment, and relapse (Smyth et al., 2010).

During opioid withdrawal, humans experience a wide range of somatic symptoms including insomnia, hyperalgesia, nausea, diarrhea, vomiting, and sweating. Most symptoms dissipate after a few days of abstinence, although insomnia can persist for several weeks after opioid discontinuation (Kay, 1975). Somatic symptoms increase the drive for compulsive drug-taking in both humans and animal models, elevating the probability of relapse (Kenny et al., 2006; Pergolizzi et al., 2020).

Although opioid withdrawal-related sleep issues are poorly understood, sleep disruption is common in substance abuse, both acutely and during withdrawal, and sleep disturbances are predictive of relapse (Brower & Perron, 2010; Roehrs & Roth, 2015). Acute effects of opioids and opioid withdrawal on sleep parameters have been described in both the preclinical and clinical literature. Morphine, methadone, buprenorphine, heroin and remifentanil all decreased sleep quality and suppressed REM sleep in humans, an effect that was present for 3–5 weeks in chronic opioid users (Bonafide et al., 2008; Dimsdale et al., 2007; Dunn et al., 2018; Shaw et al., 2005; L. Xiao et al., 2010). During acute withdrawal from chronic heroin use, sleep is more fragmented and transitions to REM sleep are reduced (Howe et al., 1981). During withdrawal from short term (3–5 days) heroin administration, REM sleep rebound persisted for several days (Lewis et al., 1970). Some aversive opioid withdrawal symptoms have been replicated in animal models (De Andres & Caballero, 1989).

The influence of biological sex on acute opioid effects and opioid withdrawal has been explored in both the clinical and preclinical literatures. Epidemiological studies report that women display higher rates of opioid abuse than men, suggesting a higher susceptibility to opioid dependence (Chen et al., 1998; Greenfield et al., 2003). Opioids also have a higher efficacy, producing a greater analgesic effect in women than men (Niesters et al., 2010). While clinical reports indicate females experience more physical problems related to opioid abuse than males, little research exists on the impacts of biological sex on opioid dependence (Chen et al., 1998; Greenfield et al., 2003). In animal models, the influence of biological sex upon withdrawal-associated behaviors reflective of physical dependence is complex. In naloxone-precipitated withdrawal paradigms, male mice lose more weight and display more “wet-dog” shakes than females while female mice exhibit an elevated occurrence of burrowing behavior (Bobzean et al., 2019; el-Kadi & Sharif, 1994). During spontaneous morphine withdrawal, symptoms occur sooner in male mice but persist longer in females (Papaleo & Contarino, 2006).

It has been widely hypothesized that sleep disturbances during opiate withdrawal are a major factor that contributes to relapse. We therefore sought to establish an animal model in which to test this hypothesis. We utilized a 5-day escalating morphine dose regimen in C57BL/6 mice that enabled determination of both acute morphine effects and opioid withdrawal effects on sleep/wake and other physiological parameters. Since limited information is available on sex differences in sleep symptomatology during acute opioid withdrawal, we studied both male and female mice. We find that acute morphine administration promoted sustained periods of wakefulness and activity and that female mice exhibited greater sleep fragmentation during withdrawal, reflecting less efficient sleep consolidation than in male mice.

Methods

Subjects and surgical procedure

Male (N=8; age: 10.4±0.5 weeks) and female (N=9; age: 9.9±0.3 weeks) C57BL/6 mice were implanted with telemetric devices (F20-EET; DSI, St-Paul, MN, USA) for recording 1 electroencephalogram (EEG) channel and 1 electromyogram (EMG) channel, subcutaneous body temperature (T_sc) and activity. Mice were anesthetized with isoflurane (induction: 3–5% isoflurane in oxygen delivered at 1 L/min; maintenance: 1–2% isoflurane in oxygen delivered at 1 L/min). Telemeters were placed in a blunt-dissected subcutaneous (SC) pocket located on the left dorsum and biopotential leads were routed to the head. EMG leads were placed in the right nuchal muscle. Cranial holes were drilled through the skull at −2.0 mm AP from bregma and 2.0 mm ML and on the midline at −1 mm AP from lambda. The EEG leads were inserted into these holes and affixed to the skull with dental acrylic. The incision was closed with absorbable suture. Analgesia was managed with meloxicam (5 mg/kg, SC) and buprenorphine (0.05 mg/kg, SC) prior to emergence from anesthesia and for the first day post-surgery. Meloxicam (5 mg/kg, SC) was continued for 2 d post-surgery. A minimum of two weeks of post-surgical recovery were allowed before experimental protocols were initiated.

After surgery, mice were housed individually in home cages with access to food, water, and nestlets ad libitum. Room temperature (22±2°C), humidity (50±20% relative humidity), and lighting conditions (LD12:12, where Zeitgeber time (ZT) 0=lights on and ZT12=lights off) were monitored continuously. Animals were inspected daily in accordance with AAALAC and SRI guidelines. All experimental procedures were approved by the Institutional Animal Care and Use Committee at SRI International.

Experimental design

Figure 1 depicts the dosing schedule used in each arm of this crossover study. Each mouse received two treatments in a counterbalanced manner; 16 days elapsed between the final treatment in the first arm of the study and the 1^st treatment in the second study arm :

Figure 1.

Illustration of the experimental design and dosing schedule used in this study. The schematic depicts one arm of our two-armed study in which each mouse received both treatments, separated by a 16 day washout period. Treatments were administered in a counterbalanced manner in which one cohort of mice received an escalating morphine treatment in the first arm of the study and saline in the second arm of the study, and the second cohort received the saline in the first arm of the study and the escalating dose of morphine in second study arm. After a 24-h baseline EEG/EMG recording, experimental treatments were initiated at hour ZT1 on the following day and continued for the next 4.5 days. Bidaily subcutaneous (s.c.) dosings occurred at ZT1 and ZT13 on each day during the 5-day treatment phase, with the morphine dose doubling every day. The withdrawal phase lasted 7 days, commencing at the final dose. An EEG recording was initiated at ZT13 on WD1 and continued for 71 h through WD3. Separate 24-h recordings were initiated at ZT12 on WD5 and WD7. Abbreviations: WD, Withdrawal day.

(Treatment 1) Bidaily SC injections of escalating doses of morphine sulfate (day 1: 5 mg/kg, day 2: 10 mg/kg, day 3: 20 mg/kg, day 4: 40 mg/kg and day 5: 80 mg/kg) over 5 days followed by 7 days without injections.

(Treatment 2) Bidaily SC injections of saline over 5 days followed by 7 days without injections.

The experimental cohorts in this study were equally sized. In this treatment paradigm, one experimental cohort received treatment 1 in the first study arm, then treatment 2 in the second study arm following the washout period, while the other experimental cohort received the treatments in the opposite order with treatment 2 occuring first, then treatment 1 following washout. . For each treatment, mice received bidaily SC injections at ZT1 and ZT13 of constant volume (10 ml/kg; Figure 1). Mice received a habituation injection of saline (volume: 10 ml/kg) 2 days before initiating each arm.

EEG, EMG, activity, and subcutaneous temperature (T_sc) data collection

All mice underwent a 24-h baseline recording prior to each study arm in which videos, EEG, EMG, T_SC, and gross motor activity were recorded via telemetry using Ponemah (DSI, St-Paul, MN, USA). Digital videos were recorded at 10 frames per second, 4CIF de-interlacing resolution; EEG and EMG were sampled at 500 Hz. Following the first treatment of the low, middle, and high morphine doses (5, 20, and 80 mg/kg, respectively) which occurred at ZT1, EEG and EMG were continuously recorded for 6 hours post-dosing in males and 12 hours in females. After the last morphine dose at ZT13 on day 5, EEG and EMG were continuously recorded for 71 hours (withdrawal days (WD) 1–3) and for 24 hours on WD5 and WD7. Mice were weighed before the baseline recording was initiated, at the administration of the final treatment, and at 71 hours post-treatment.

Arousal state scoring

EEG/EMG recordings were scored in 10-sec epochs as either Wake, NREM or REM sleep using Somnivore (Allocca et al., 2019). Each scoring period was provided to Somnivore (ver. 1.1.4.0) in EDF format from which 100–200 10-sec epochs of each state were selected by an expert human scorer as training data (Figure 2). The selected data were used to train a classifier and applied in a supervised machine learning model to automatically score each state for the duration of each recording. Following this initial scoring, the accuracy of the autoscored data was assessed visually. If the automated scoring was determined by the expert human scorer not to generalize well, an additional 1–21 misidentified epochs were provided to the training dataset and automatic scoring of the recording was repeated. Manual review and correction of the autoscored dataset was then performed when the following occurred: 1) single epochs of REM sleep and 2) REM sleep directly following wake.

Figure 2.

Two hour EEG and EMG recordings from the same mouse treated with saline vs. with 80 mg/kg morphine. (A) Saline-treated mice exhibit typical periods of wakefulness (blue-shaded box), NREM (red-shaded box), and REM (green-shaded box) sleep. Wakefulness was characterized by lower amplitude EEG activity in conjunction with higher muscle tone and/or movement indicated by sharp deflections in EMG voltage. NREM sleep was characterized by higher voltage, synchronous EEG patterns with higher power in the delta frequency range, as well as low, tonic muscle tone observed in the EMG channel. REM sleep was characterized low-amplitude EEG activity, typically in the theta-frequency range in conjunction with muscle atonia. (B) Morphine-treated mice (80 mg/kg, s.c.) were continuously awake for several hours post-treatment as indicated by low amplitude EEG activity in conjunction with high EEG activity and also exhibited several atypical EEG patterns, most prominently, high amplitude spikes in the EEG (yellow shaded box) that overlay an activated EEG pattern. In the % EEG power band contribution channel, each color corresponds to low frequency bands rounded to the nearest Hz (delta: 0.5–4Hz in red, theta: 4–8Hz in blue, alpha: 8–12Hz in green, sigma: 12–16Hz in purple, beta: 16–24Hz in yellow).

Data Analysis and Statistics

A “bout” of Wake, NREM, or REM sleep was defined as >two consecutive 10-sec epochs of that state. EEG spectra were analyzed in 0.244 Hz bins. For each mouse, spectral power was normalized to the average power per bin during the 24-h baseline recording from each arm of the study. For spectral analyses, one epoch on both sides of a state transition were excluded. Hourly averages of T_sc and activity determined from the telemetry transmitters were also analyzed. Two female mice were excluded for transmitter failure and one male mouse was excluded due to mis-dosing. Two male mice were excluded from the spectral and frequency band analysis due to EEG noise. Another male mouse exhibited high EEG movement artifact during wake on WD5 and WD7; consequently, wake data for this mouse was excluded on these days.

Two-way ANOVA with drug treatment and time as factors was used to analyze data sets with no missing data points using data from both treatments. When treatment x time interaction was indicated, Fisher’s LSD test was used to identify hour-specific differences. Since the final treatment on day 5 occurred at ZT13, the acute effects of treatment were still apparent: morphine-treated mice were continuously awake for 2–3 hours at the beginning of the WD1 recording. Thus, in morphine-treated mice, the missing data points for NREM and REM bout duration and the corresponding spectral analyses precluded fitting of a full-effects model. Consequently, these measures were analyzed using a mixed-effects model for main effects and unpaired two-tailed t-tests were used to identify hourly differences.

Results

Morphine administration acutely increases activity and induces aberrant EEG patterns in both sexes

Saline- and 5 mg/kg morphine-treated mice exhibited normal sleep/wake states (Figure 2A). At higher doses, morphine-treated mice exhibited an aberrant wake state characterized by a hypotonic EMG and an activated EEG with frequent high amplitude spikes (Figure 2B). At these higher doses, mice circled their cages continuously for several hours and the high amplitude EEG spikes were associated with abrupt pauses in locomotion (see Supplementary Video V1). Morphine acutely increased activity at all three doses in females (Figure 3A–A”; 5 mpk: F_(1,6)=31.96, p=0.001; 20 mpk: F_(1,6)=51.24, p<0.001; 80 mpk: F_(1,6)=223.87, p<0.001) and at the middle and high doses in males (Figure 3B–B”: 20 mpk: F_(1,6)=51.13, p<0.001; 80 mpk: F_(1,6)=297.95, p<0.001). Post hoc tests revealed that increased activity persisted 5–7 hrs in females and 2–6 hrs in males (Figure 3). In conjunction with the increased activity, morphine prolonged wakefulness in both female (Figure 4A) and male (Figure 5A) mice and induced a corresponding suppression of both NREM (females: Figure 4B; males: Figure 5B) and REM sleep (females: Figure 4C; males: Figure 5C).

Figure 3.

Morphine acutely increased activity during the light phase in a dose-related manner. At 5 mg/kg, morphine-treated (A) female and (B) male mice exhibited normal EEG patterns and sleep states, so this increase in activity likely reflects an overall increase in the amount of wake. At 20 mg/kg (females: A’; males: B’) and 80 mg/kg (females: A”; males: B”), morphine-treated mice exhibited significantly increased activity for 4–7 hours. These data are consistent with observations that, at these doses, mice circled their cages continuously for several hours after morphine administration. Values are hourly means±SEM. Measures from morphine-treated female mice (N=7) are plotted in red and male mice in blue. Data from the same mice following vehicle treatment are plotted in black within the same graph. Black arrows below the x-axis on each graph indicates the time when treatment occurred. * in the legend denotes a significant treatment effect across the entire recording as determined by 2-way ANOVA; colored * above hourly graphs indicate significant (p<0.05) effect during that hour relative to saline treatment. Abbreviations: h, Hour.

Figure 4.

Sleep architecture measures for withdrawal day (WD) 1 for female mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). Final treatment (80 mg/kg., s.c., morphine sulfate or saline) occurred at hour ZT13, after which an acute effect was evident: nearly continuous wakefulness for 4–5 hours. During the light phase on WD1, NREM bout durations were reduced in females (B’). The number of wake (A) and NREM (B) bouts was increased, and number of REM (C) bouts was decreased during the light phase on WD1. Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in red; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour.

Figure 5.

Sleep architecture measures for withdrawal day (WD) 1 for male mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). Final treatment (80 mg/kg., s.c., morphine sulfate or saline) occurred at hour ZT13, after which, as in female mice, morphine resulted in continuous wakefulness (A),and sustained wake bouts for 4–5 hours (A’). The number of wake (A”) and NREM (B”) bouts was altered during WD1 in the males as in the females in Figures 4A” and 4B”. Values are mean±SEM. Measures from morphine-treated (N=7) mice are plotted in blue; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour.

Morphine withdrawal alters wake and NREM sleep amounts on WD1 and WD2 but recovery of REM sleep is more prolonged

On WD1, the acute effects of morphine treatment were apparent for the first 4–5 hrs after the final treatments at ZT13. Although sleep was suppressed for the first 4 hrs of the recording in both female (Figure 4B, 4C) and male (Figure 5B, 5C) cohorts, sleep rebounded in the second half of the night on WD1. Treatment effects were significant for the female cohort on WD2 for the amounts of wake (Figure 6A; F_(1,6)=23.75, p =0.003) and NREM sleep (Figure 6B’; F_(1,6)=24.59, p=0.003); there was a trend for REM sleep (Figures 6C; F_(1,6)=5.39, p=0.059). For the males, treatment effects were significant for the amounts of REM on WD1 (Figure 5C; F_(1,6)=13.79, p=0.010) and WD3 (Figure S2C; F_(1,6)=12.07, p=0.013), and there was a nearly significant trend on WD2 (Figure 7C; F_(1,6)=5.92, p=0.051).

Figure 6.

Sleep architecture measures for withdrawal day (WD) 2 for female mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). In morphine-treated mice, wake was reduced (A) and NREM and REM were increased (B and C) during the dark phase on WD2. Wake (A’) and NREM (B’) bout durations were also reduced during the dark phase on WD2. The increased amounts of NREM and REM sleep occurring during the dark phase on WD2 were mediated by an increased number of NREM (B”) and REM (C”) bouts. Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in red; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour.

Figure 7.

Sleep architecture measures for withdrawal day (WD) 2 for male mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). In morphine-treated mice, wake was reduced (A’) and NREM and REM were increased (B’ and C’) during the dark phase on WD2. NREM bout duration was altered in several hourly bins on WD2 (B’). The increased amount of REM sleep occurring during the dark phase on WD2 was mediated by an increased number of REM bouts (C”). Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in blue; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour.

As indicated in Table 1, significant treatment x time interactions occurred for wakefulness amounts in both females on WD1 (F_(21,126)=5.74, p<0.001; Figure 4A) and WD2 (F_(23,138)=1.99, p=0.008; Figure 6A) and males on WD1 (F_{(21, 126)}=10.00, p<0.001; Figure 5A) and WD2 (F_(23,138)=2.11, p=0.005; Figure 7A). Significant interactions were also found for NREM sleep in females on WD1 (F_(21,126)=6.31, p<0.001; Figure 4B) and WD2 (F_(23,138)=1.97, p=0.009; Figure 6B) as well as for males on WD1 (F_{(21, 126)}=11.33, p<0.001; Figure 5B) and WD2 (F_(23,138)=1.97, p=0.009; Figure 7B). Lastly, time x treatment interactions were found for REM sleep in females on WD1 (F_(21,126)=2.62, p<0.001; Figure 4C) and WD2 (F_(23,138)=1.83, p=0.017; Figure 6C), and in males on WD1 (F_{(21, 126)}=2.13, p=0.006; Figure 5C) and WD2 (F_(23,138)=2.36, p=0.001; Figure 7C). Post hoc tests revealed increased wake and decreased NREM and REM sleep at the beginning of WD1 for morphine-treated female (Figure 4A–C) and male (Figure 5A–C) mice compared to saline-treated mice, followed by a rebound in NREM and REM sleep at the end of the dark phase on WD1 (Figures 4B, 4C, and 5B).

Table 1.

Treatment and Time x Treatment effects for vigilance state amounts on days 1, 2, 3, 5 and 7 during morphine withdrawal.

	WD1	WD2	WD3	WD5	WD7
Treatment Effects
Females
Wake	NS	p = 0.003	NS	NS	NS
NREM	NS	p = 0.003	NS	NS	NS
REM	NS	NS (p = 0.059)	NS	NS	NS (p=0.056)
Males
Wake	NS	NS	NS	NS	NS
NREM	NS	NS	NS	NS	NS
REM	p = 0.010	NS (p = 0.051)	p = 0.013	NS	NS
Time x Treatment Effects
Females
Wake	p < 0.001	p = 0.008	NS	NS	NS
NREM	p < 0.001	p = 0.009	NS	NS	NS
REM	p < 0.001	p = 0.017	NS	NS	NS
Males
Wake	p < 0.001	p = 0.005	NS	NS	NS
NREM	p < 0.001	p = 0.009	NS	NS	NS
REM	p = 0.006	p = 0.001	NS	NS	NS

Abbreviations: NA, Not applicable; NS, Not significant; WD, Withdrawal day.

The recovery of NREM and REM sleep from the acute suppression by morphine on WD1 persisted into the dark phase on WD2 for both female (Figure 6B–C) and male (Figure 7B–C) mice. With the exception of a treatment effect for amount of REM in the males on WD3 (F_(1,6)=12.07, p=0.013; Figure S2C), no further treatment or interaction effects for wake, NREM, or REM sleep amounts were found in either the male or female cohorts on WD3 (Figures S1A–S1C, S2A–S2C), WD5 (Figures S3A–S3C, S4A–S4C), or WD7 (Figures S5A–S5C, S6A–S6C).

Morphine withdrawal results in more fragmented sleep in female mice than in male mice

To determine whether morphine withdrawal affected the normal architecture of sleep and wakefulness, we assessed two measures of sleep/wake continuity: the mean bout duration (Table 2) and the number of bouts (Table 3) of each state. For the female cohort, a significant treatment effect occurred for NREM bout duration on WD1 (F_(1,6)=13.63, p=0.010; Figure 4B’) and WD2 (F_(1,6)=8.48, p=0.027; Figure 6B’), for wake bout duration on WD2 (F_{(1, 6)}=16.60, p=0.007; Figure 6A’), and REM bout duration on WD5 (F_(1,6)=6.15; p=0.048; Figure S3C’). For the male cohort, a significant treatment effect occurred for wake bout duration on WD1 (F_(1,6)=14.43, p=0.009; Figure 5A’; Table 2).

Table 2.

Treatment and Time x Treatment effects for bout durations on days 1, 2, 3, 5 and 7 during morphine withdrawal.

	WD1	WD2	WD3	WD5	WD7
Treatment Effects
Females
Wake Bout Duration	NS	p=0.007	NS	NS	NS
NREM Bout Duration	p=0.010	p=0.027	NS	NS	NS
REM Bout Duration	NS	NS	NS	p=0.048	NS
Males
Wake Bout Duration	p=0.009	NS	NS	NS	NS
NREM Bout Duration	NS	NS	NS	NS	NS
REM Bout Duration	NS	NS	NS	NS	NS
Time x Treatment Effects
Females
Wake Bout Duration	p<0.001	p=0.038	NS	NS	NS
NREM Bout Duration	p=0.006	NS	p=0.003	NS	NS
REM Bout Duration	NA	NS	NS	NS	NS
Males
Wake Bout Duration	p<0.001	NS	NS	NS	NS
NREM Bout Duration	NA	p<0.001	NS	NS	NS
REM Bout Duration	NA	NS	NS	p=0.016	NS

Abbreviations: NA, Not applicable; NS, Not significant; WD, Withdrawal day.

NA due to missing data points for NREM and REM bout duration on WD1 that precluded fitting a full effects model to determine interaction effects.

Table 3.

Treatment and Time x Treatment effects for the number of sleep/wake bouts on days 1, 2, 3, 5 and 7 during morphine withdrawal.

	WD1	WD2	WD3	WD5	WD7
Treatment Effects
Females
Number of Wake Bouts	p < 0.001	p < 0.001	NS	NS	NS
Number of NREM Bouts	p = 0.014	p = 0.003	p = 0.030	NS	NS
Number of REM Bouts	p = 0.035	NS	p = 0.004	NS	NS
Males
Number of Wake Bouts	NS	NS	NS	NS	NS
Number of NREM Bouts	NS	NS	NS	NS	NS
Number of REM Bouts	p = 0.040	NS (p=0.054)	NS	NS	NS
Time x Treatment Effects
Females
Number of Wake Bouts	p = 0.034	p < 0.001	NS	NS	NS
Number of NREM Bouts	p < 0.001	p = 0.015	NS	NS	NS
Number of REM Bouts	p = 0.007	NS	NS	NS	NS
Males
Number of Wake Bouts	p < 0.001	NS	NS	NS	NS
Number of NREM Bouts	p < 0.001	NS	NS	NS	p = 0.022
Number of REM Bouts	p = 0.024	p = 0.009	NS	NS	NS

Abbreviations: NA, Not applicable; NS, Not significant; WD, Withdrawal day.

Significant treatment x time interactions for wake bout duration occurred in females on WD1 (F_(21,126)=3.49, p<0.001; Figure 4A’) and WD2 (F_(23,138)=1.67, p=0.038; Figure 6A’) and for males on WD1 (F_(21,126)=13.71, p<0.001; Figure 5A’). Significant interactions for NREM bout duration occurred in females on WD1 (F_(21,100)=2.15, p=0.006; Figure 4B’) and WD3 (F_(23,115)=2.19, p=0.003; Figure S1B’) and for males on WD2 (F_(23,131)=2.58, p<0.001; Figure 7B’). Post hoc t-tests revealed differences in NREM bout duration during several hours on WD1 for the males (Figure 7B’). A significant treatment x time interaction for REM bout duration was indicated on WD5 for males (F_(23,64)=2.00, p=0.016; Figure S4C’; Table 2).

As another indicator of disruption of sleep/wake architecture during morphine withdrawal, Table 3 presents the number of sleep/wake bouts. A significant treatment effect for the number of wake bouts occurred in female mice on WD1 (F_(1,6)=46.57, p<0.001; Figure 4A”) and WD2 (F_(1,6)=38.10, p<0.001; Figure 6A”), for the number of NREM bouts on WD1 (F_(1,6)=11.93, p=0.014; Figure 4B”), WD2 (F_(1,6)=23.52, p=0.003; Figure 6B”) and WD3 (F_(1,6)=8.00, p=0.030; Figure S1B”), and the number of REM bouts on WD1 (F_(1,6)=7.32, p=0.035; Figure 4C”) and WD3 (F_(1,6)=21.20, p=0.004; Figure S1C”). For the male cohort, the only significant treatment effect was for the number of REM bouts on WD1 (F_(1,6)=6.84, p=0.040; Figure 5C”; Table 3).

Table 3 also presents significant treatment x time interaction effects for the number of wake bouts in female mice on WD1 (F_(21,126)=1.73, p=0.034; Figure 4A”) and WD2 (F_(23,138)=2.60, p<0.001; Figure 6A”), for the number of NREM bouts on WD1 (F_(21,126)=5.125, p<0.001; Figure 4B”) and WD2 (F_(23,138)=1.86, p=0.015; Figure 6B”), and for the number of REM bouts on WD1 (F_(21,126)=2.08, p=0.007; Figure 4C”). For the males, interaction effects were indicated for the number of wake bouts on WD1 (F_(21,126)=3.63, p<0.001; Figure 5A”), the number of NREM bouts on WD1 (F_(21,126)=5.70, p<0.001; Figure 5B”) and WD7 (F_(23,138)=1.79, p=0.022; Figure S6B”), and for the number of REM bouts on WD1 (F_(21,126)=1.81, p=0.024; Figure 5C”) and WD2 (F_(23,138)=1.96, p=0.009; Figure 7C”). Post hoc tests revealed an increased number of wake and NREM bouts during the light phase on WD1 in female (Figure 4A”, 4B”) but not in male (Figure 5A”, 5B”) mice. Increases in the amount of NREM and REM sleep in the dark phase on WD2 in both females (Figure 6B, 6C) and males (Figure 7B, 7C) occurred in conjunction with an increased number of NREM bouts in females (Figure 6B”) and REM bouts in males (Figure 7C”).

Morphine withdrawal causes reduced NREM sleep intensity

Morphine withdrawal was associated with alterations in spectral composition of the EEG during Wake, NREM, and REM sleep in both sexes, particularly on WD1. EEG spectral data (0–60 Hz), normalized to the mean power per 0.244 Hz bin during the 24-h baseline recordings, are presented in Figure S7 for female mice and Figure S8 for male mice. The descriptions below refer to the standard frequency bands (delta: 0.5–4Hz, theta: 4–8Hz, alpha: 8–12Hz, sigma: 12–16Hz, beta: 16–24Hz, low gamma: 24–60Hz), rounded to the nearest Hz. In contrast to sleep architecture measures which normalized after WD2, paired t-tests comparing baseline normalized power for each 0.244 Hz spectral bin revealed differences between treatments persisting during WD5 and 7 in both males and females. On WD5, increased EEG power at the higher end of the low gamma band increased during wake as did sigma and beta power on WD7 in the morphine-treated females (Figure S7A’”–A””). In the morphine-treated males, normalized power during wakefulness was also significantly reduced in part of the alpha and low gamma bands on WD5 (Figure S8A’”) and in the delta, theta, and low gamma bands on WD7 (Figure S8A””).

NREM delta power (0.5–4.0 Hz), also referred to as EEG slow wave activity (SWA), is reflective of sleep intensity (Borbely, 1982). Morphine treatment reduced SWA in females on WD1 (F_(1,6)=14.24, p=0.009; Figure 8A). A significant treatment x time interaction was observed on WD2 for both sexes (females: F_(23,104)=1.86, p=0.018, Figure 8A’; males: F_(23,76)=2.10, p=0.009, Figure 8B’). While fitting a full effects model for SWA wasn’t possible for males on WD1, paired post hoc t-tests revealed numerous hourly deficits in SWA in both sexes as a result of morphine treatment, particularly during the WD1 light phase (Figure 8A, 8B). No significant differences in SWA were observed on WD3–7 for either sex (Figure S9A–A” and -S9B–B”).

Figure 8.

Hourly NREM Delta Power (0.5–4.0 Hz; also referred to as slow wave activity or SWA), normalized to a 24-h baseline recording, in female (A-A””) and male (B-B””) mice on withdrawal day (WD) 1 (A and B) and WD2 (A’ and B’). SWA was reduced during the light phase on WD1 and the dark phase on WD2 in both females (A-A’) and males (B-B’). Measures from morphine-treated female mice (N=7) are plotted in red and male mice (N=5) are plotted in blue; data from the same mice following saline vehicle treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour; SWA, Slow wave activity.

Morphine withdrawal causes reduced activity and T_SC

Morphine withdrawal also altered both activity and T_sc. Significant treatment effects were observed for activity in the females on WD1 (F_(1,6)=20.64 p=0.004; Figure 9A) and WD2 (F_(1,6)=11.31, p=0.015; Figure 9A’). Significant treatment x time interaction effects were also observed for activity for the females on WD1 (F_(21,126)=37.44, p<0.001; Figure 9A) and on WD2 (F_(23,138)=3.55, p<0.001; Figure 9A’) but only on WD1 for the males (F_(21,126)=16.51, p<0.001; Figure 9B). The effects on WD1 were primarily the result of acute morphine treatment, which increased activity (Figure 3). Activity was significantly reduced in females during several hours in the dark phase on WD2 (Figure 9A’), the period during which NREM and REM sleep was elevated.

Figure 9.

Activity count (A-A’ and B-B’) and subcutaneous body temperature (T_sc; C-C’ and D-D’) on withdrawal day (WD) 1 (A-D) and WD2 (A’-D’), for female (A-A’ and C-C’) and male (B-B’ and D-D’) mice. Morphine treatment altered T_sc and gross motor activity, both acutely as well as following treatment discontinuation, during spontaneously-occurring opioid withdrawal. Final treatment (80 mg/kg, s.c., morphine sulfate or saline) occurred at ZT13 on WD1. Morphine acutely increased activity in both female (A) and male (B) mice. Morphine treatment also acutely reduced T_sc in both females (C) and males (D). During WD2, both activity (A’) and T_sc (C’) were significantly reduced during the dark phase for morphine-treated females. In morphine-treated males, only T_sc was significantly reduced on WD2 (D’). Values are hourly means±SEM. Measures from female mice are plotted in red and male mice in blue. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour.

There was a significant condition effect for T_sc in the females on WD2 (F_(1,6)=6.24, p=0.047; Figure 9C’). There were no significant condition effects on T_sc during the remainder of the withdrawal period for either sex (Figure S10C–C” and S10D–D”); however, significant treatment x time interaction effects for T_sc occurred in both females and males on both WD1 (females: F_(21,126)=13.74, p<0.001; males: F_(21,126)=10.86, p<0.001; Figures 9C, 9D) and WD2 (females: F_(23,138)=3.72, p<0.001; males: F_(23,138)=4.90, p<0.001; Figures 9C’, 9D’). To evaluate the correlation between external measures of withdrawal and measures of sleep disruption occurring during these days, Pearson’s correlation coefficient was calculated for the mean differences between the morphine and saline treatment groups for T_SC, activity and the number of NREM and wake bouts on WD1 and 2. Activity and the number of NREM bouts were significantly correlated in the female cohort (r=−0.58; p<0.001). In the males, significant correlations were found between activity and both the number of NREM (r=−0.76; p<0.001) and wake bouts (r=−0.61; p<0.001) as well as between T_sc and the number of wake bouts (r=0.30; p=0.04).

The mean weights of the morphine and saline treatment groups were comparable during baseline in both the female and male cohorts. Morphine treatment reduced body weight in both male and female mice (Figure S11A and S11B; females: p<0.001; males: p=0.020), resulting in ~10% decrease in body weight (Figure S11A’ and B’). Body weight was again comparable by the end of WD3 (Figure S11A and B), although the percent change in body weight in the morphine-treated group remained significantly reduced in the females (Figure S11A’; p=0.004) but not in the males (Figure S11B’).

Discussion

In the present study, we investigated activity, sleep/wake and T_SC in a sex-balanced cohort of a murine model of opioid dependence and withdrawal. Morphine acutely promoted activity and wakefulness in both males and females and, at higher doses, produced an abnormal EEG. The activity-promoting effects of morphine were of longer duration at the low and middle dose in females. During withdrawal, increased NREM and REM sleep occurred in both sexes on WD2 during the dark phase, the major active phase in rodents. Female mice exhibited more wake and NREM bouts during the light phase on WD1 and the dark phase on WD2, indicative of fragmented sleep, while males showed no differences in this measure, suggesting less efficient sleep consolidation in females than males. Both males and females exhibited reduced SWA during the light phase on WD1 and the dark phase on WD2, suggesting sleep during this period was less restorative. Activity was significantly reduced during the dark phase on WD2 in females while T_sc was significantly reduced in both sexes for about 24 h, beginning during the latter half of the dark phase on WD1 and extending well into the dark phase on WD2. Together, these results suggest that opioid withdrawal disrupted sleep in both sexes, but that female mice exhibited sleep fragmentation, whereas males did not, suggesting the sleep/activity phenotype was more pronounced in females.

Morphine acutely impacts activity and body weight

Previous studies of the acute effects of morphine on activity have revealed varied results, likely due to variation in species/strains, dosing route and protocol, and age of the experimental subjects. While some studies report increased activity in an open field following acute opioid treatment, others report a decrease (Babbini & Davis, 1972; Collins et al., 2016; Hollais et al., 2014). Nonetheless, during chronic opioid administration, animals are consistently reported to develop behavioral tolerance indicated by a progressive increase in activity across treatment days, in both constant or escalating dose paradigms (Babbini & Davis, 1972; Paul et al., 2021). As has been reported previously in other escalating dose paradigms, all doses in the present study promoted activity, an effect that was more pronounced as the dose increased and the chronic treatment phase progressed (Figure 3A–A” and B–B”). While these results are suggestive of behavioral tolerance, a dose-dependent effect cannot be ruled out in escalating dose paradigms.

Morphine-treated mice exhibited a decrease in both mean body weight (Figure S11A and B) and change in weight relative to baseline (Figures S11A’ and B’). While mean body weight normalized after WD3 in both males and females, the percent change in body weight did not fully recover in morphine-treated females relative to their saline-treated counterparts (Figure S11A’). These changes in body weight measures could be the result of increased energy expenditure due to elevated activity occurring during the acute morphine treatment phase or to reduced food consumption, which we did not measure (Figure 3A–A” and B–B”).

Morphine withdrawal impacts sleep architecture, sleep quality and T_SC

Although increased NREM and REM sleep during withdrawal has been described previously in cats (De Andres & Caballero, 1989) and REM sleep in rats (Khazan & Colasanti, 1972), to the best of our knowledge, these results have not been replicated in mice experiencing withdrawal from chronic opioid treatment (Bedard et al., 2023; Eacret et al., 2022; Gamble et al., 2022). Our results in the current study in mice (Figure 4B’ and C’, Figure 5B’ and C’) are thus consistent with effects on sleep/wake reported in other mammals. Morphine was acutely wake-promoting, with the highest dose causing nearly continuous wakefulness for 6 hours after treatment (Figure 4A and 5A), suggesting morphine-treated mice experienced chronic sleep loss as a result of 5 days of bidaily dosing. Although both sexes of morphine-treated mice exhibited increases in both NREM and REM sleep during latter half of the dark phase on WD1 and on WD2 (Figures 4B–C,5B–C, 6B–C, and 7 B–C), few differences were observed in the light phases (Figures 4B–C, 5B–C, 6B–C, and 7B–C). This delayed sleep rebound during the dark phase mirrors clinical reports of humans experiencing opioid withdrawal (Martin et al., 1973; Shaw et al., 2005), and may be due to a ceiling effect on sleep amount during the light phase, resulting in increased sleep during the major active period for mice.

Both sexes in this study experienced reduced sleep quality as indicated by the increased sleep fragmentation in the female cohort (Figure 4A”–B” and 6A”–B”) and reduction in SWA in both males and females during the light phase on WD1 (Figure 8A–B) and WD2 (Laffan et al., 2010). Impaired sleep quality during the light phase might cause increased sleepiness in the subsequent dark phase, perhaps causing more napping during this period, a notion that could be tested in subsequent studies.

We also observed a significant reduction in T_sc during the light phase on WD1 and during the dark phase on WD2 in both male and female mice. This reduction in T_sc corresponded to the same period when NREM and REM sleep increased but alterations in sleep architecture were observed. Behavioral immobility and reduced body temperature are characteristic of sleep (Harding et al., 2019). Furthermore, activity was depressed in female mice during the dark phase on WD2, an effect described previously in Sprague-Dawley rats experiencing spontaneous opioid withdrawal (Kandasamy et al., 2017; Morgan et al., 2021; Stinus et al., 1998). Our results support the use of activity and/or T_SC monitoring as a measure of spontaneous opioid withdrawal and suggest that the reduced activity, T_sc, and alterations to sleep/wake that we observed may be interrelated physiological responses, a notion that is supported by significant correlations between the changes in T_sc, activity and measures of sleep fragmentation, which could be explored further in future studies.

While our analyses focused primarily on sleep architecture and quality, EEG spectral analysis revealed that alterations to normalized EEG power were primarily restricted to WD1 (Figures S7A–C and S8A–C). In contrast to sleep measures which normalized after WD2, alterations to EEG spectral power were observed in the beta frequency band during Wake on WD5 and WD7 in the male cohort (Figure S8A’” and A””), while differences in beta and gamma frequency bands were present during wake on WD5 and WD7 in females (Figure S7A’” and A””). Gamma frequency rhythms are associated with higher level cognitive function (Uhlhaas et al., 2008). Beta rhythms during wakefulness have also been associated with attention, and performance of a rewarding task (Yaple et al., 2018). These results suggest that alterations to cognitive processes and reward may emerge following acute spontaneous withdrawal, which should be explored in future research.

Comparison to clinical manifestation of opioid withdrawal

Our results in mice resemble aspects of the phenotype described in both short-term (3–5 days) and longer-term (2–5 months) human opioid users experiencing withdrawal (Howe et al., 1981; Lewis et al., 1970). Morphine-treated mice in the present study exhibited an insomnia-like phenotype, characterized by reduced sleep quality during the first 24 hours of the withdrawal period as indicated by sleep fragmentation in the females (Figure 4A–B and 5A–B) and reduced sleep intensity in both sexes (Figure 8A and B). These characteristics are similar to long-term heroin users who exhibited more sleep/wake transitions and reduced sleep quality on days 2 and 3 of withdrawal (Howe et al., 1981). However, this study on long-term heroin users found less REM sleep whereas the mice in our study exhibited increased REM sleep, as has been reported following short-term heroin administration in humans (Lewis et al., 1970).

To our knowledge, sex differences in sleep phenotype during acute opioid withdrawal have not been investigated clinically but, in a study investigating sleep following a 10-day detoxification from chronic heroin use, Pittsburgh Sleep Quality Index scores were similar between males and females (He et al., 2020). More research is needed to understand how acute withdrawal from long-term opioid abuse impacts sleep, and whether biological sex influences sleep phenotype severity during acute withdrawal in humans. Finally, participants in the above study were chronic users whereas, in our study in mice, treatment duration was just 5 days. Our results may therefore suggest may a greater sensitivity to morphine in females than males, faster escalation to addiction, and/or sensitivity to withdrawal symptomatology, a notion which should be explored in future research.

Limitations of the Current Study

While animal models of disorders such as OUD are important tools for studying specific phenomena in a structured way, such models have clear limitations. In our study, we utilized a 5-day treatment paradigm to induce opioid withdrawal whereas OUD is normally associated with longer term opioid abuse, lasting years or decades (He et al., 2020). In the preclinical literature, animals are usually treated with injection protocols typically ranging from 4–14 days (Gallego et al., 2010; McGregor et al., 2022; Pinelli et al., 1997). As such, physiological changes occurring as the result of longer-term opioid use may not be fully recapitulated in these animal models. Additionally, our mice are fed ad libitum and are singly housed. Humans experience socioeconomic pressure enforcing a circadian rhythm that favors being awake during the day and asleep during the night, a pressure that isn’t present in our paradigm (Q. Xiao & Hale, 2018). Finally, in our study paradigm, mice aren’t necessarily addicted to morphine, as treatment occurs involuntarily; studies utilizing a self-administration paradigm provide a more naturalistic model of opioid addiction. The present study thus provides insight into the acute effects of opioids and withdrawal from short-term opioid use and thereby establishes a mouse model for investigating the mechanisms underlying sleep dysregulation that occurs during opioid withdrawal.

While little previous research has focused on sleep disruption during withdrawal in preclinical models, sleep offers an objective, quantifiable withdrawal measure. Somatic symptoms of spontaneous withdrawal are infrequent, varying over the light/dark cycle, and are often subtle and difficult to quantify. These issues have led previous studies to focus on naloxone-precipitated withdrawal, which triggers pronounced and frequent somatic withdrawal symptoms, and allows measurement of withdrawal intensity over an acute period (Gallego et al., 2010). In contrast, we suggest that measures of sleep, activity and/or T_SC provide quantifiable physiological parameters during spontaneously-occurring withdrawal, which can occur over a prolonged time period.

Conclusions

Sleep disruption contributes to relapse to opioid use in humans experiencing opioid withdrawal, and sex differences have been described in various stages of opioid addiction. Our study describes the effects of short-term opioid treatment and discontinuation on sleep/wake, activity, and T_SC. Mice of both sexes exhibited sleep alterations during opioid withdrawal but females exhibited a more pronounced sleep phenotype overall. Because of the short duration of the treatment paradigm, our results suggest that female mice may exhibit a faster escalation to dependence than males. In an era of precision medicine, it is important to recognize how factors such as biological sex may contribute to the OUD phenotype.

Supplementary Material

Fig S2

Figure S2. Sleep architecture measures for withdrawal day (WD) 3 for male mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). There were no differences in the amounts of wake (A) or NREM (B), however REM sleep (C) was still elevated. No differences in bout duration or the number of bouts of sleep/wake states were observed. Measures from morphine-treated mice (N=7) are plotted in blue; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour..

Fig S1

Figure S1. Sleep architecture measures for withdrawal day (WD) 3 for female mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). There were no differences in the amounts of wake, NREM, or REM sleep. Reduced NREM bout duration was still apparent during the light phase on WD3 (B’), although less pronounced than on earlier WDs. Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in red; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour..

Fig S3

Figure S3. Sleep architecture measures for withdrawal day (WD) 5 for female mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). Sleep architectures were nearly indiscernible between treatment groups on WD5. 2-way ANOVA indicated a significant treatment effect for REM sleep bout duration (C’). Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in red; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour..

Fig S5

Figure S5. Sleep architecture measures for withdrawal day (WD) 7 for female mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). Sleep architecture measures were indiscernible between treatment groups on WD7. Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in red; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour..

Fig S4

Figure S4. Sleep architecture measures for withdrawal day (WD) 5 for male mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). Sleep architectures were nearly indiscernible between treatment groups on WD5. Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in blue; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour..

Fig S6

Figure S6. Sleep architecture measures for withdrawal day (WD) 7 for male mice. Hourly percent time (A-C), bout duration (A’-C’), and number of bouts (A”-C”) for wake (A-A”), NREM (B-B”), and REM sleep (C-C”). Sleep architectures were indiscernible between treatment groups on WD7. Values are mean±SEM. Measures from morphine-treated mice (N=7) are plotted in blue; data from the same mice following saline treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour..

Fig S9

Figure S9. Hourly NREM Delta Power (0.5–4.0 Hz; also referred to as slow wave activity or SWA), normalized to a 24-h baseline recording, in female (A-A”) and male (B-B”) mice on withdrawal day WD3 (A and B), WD5 (A’ and B’), and WD7 (A” and B”). There were no significant differences in SWA on WD3, 5, or 7 in female or male mice. Measures from morphine-treated female mice (N=7) are plotted in red and male mice (N=5) are plotted in blue; data from the same mice following saline vehicle treatment are plotted in black within the same graph. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA or mixed-effects model; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour; SWA, Slow wave activity..

Fig S10

Figure S10. Activity count (A-A’ and B-B’) and subcutaneous body temperature (T_sc; C-C’ and D-D’) on withdrawal day (WD) 3 (A-D), WD5 (A’-D’), and WD7 (A”-D”), for female (A-A” and C-C”) and male (B-B” and D-D”) mice. There were no significant differences in activity or T_sc on WD3, 5, or 7 in female or male mice. Values are hourly means±SEM. Measures from female mice are plotted in red and male mice in blue. * in legend denotes significance for that treatment across the entire recording as determined by 2-way ANOVA; colored * above hourly graphs indicate significance (p<0.05) during that hour relative to saline treatment. Abbreviations: h, Hour..

Fig S11

Figure S11. Body weight (A and B) and the percent change in body weight from baseline (A’ and B’) following acute treatment, and at the end of the withdrawal day (WD) 3, for female (A and A’) and male (B and B’) mice. On the baseline day, the mean weight of the morphine and saline treatment groups was comparable between the males and females. Following the acute treatment phase, the morphine-treated mice exhibited a significant reduction in body weight in comparison to the saline-treated mice (A and B). Body weight was again comparable at the end of WD3 (A and B). At the end of the treatment phase, the morphine-treated mice exhibited a nearly 10% decrease in body weight (A’ and B’). The percent change in body weight in the morphine-treated group remained significantly reduced after WD3 in the females (A’) but not in the males (B’). Values are mean±SEM. Measures from male mice (N=7) are plotted in blue and females (N=7) in red. Colored * above hourly graphs indicate significance (p<0.05) during that experimental phase relative to vehicle treatment as determined by paired t-test. Abbreviations: WD, Withdrawal day..

Fig S7

Figure S7. Morphine withdrawal effects on normalized EEG power for wake (A-A””), NREM (B-B””), and REM (C-C””) sleep on withdrawal day (WD) 1 (A-C), WD2 (A’-C’), WD3 (A”-C”), WD5 (A’”-C’”), and WD7 (A””-C””) in female mice. Differences in normalized power are most prominent on WD1. In the morphine treatment group, normalized power during wake was increased in parts of the theta frequency band (A). Delta power was decreased and alpha, beta, and low gamma power increased during NREM sleep in the morphine treatment group on WD1 (B). Low gamma in the higher end of the band on WD5, as well as sigma and beta power on WD7 are both elevated during wake. Values are mean of the entire recording period for each 0.244 Hz bin±SEM. Measures from the morphine treatment group (N=7) are in red and from the saline treatment group in black. Red lines at the top of each individual plot denote statistical significance (p<0.05) for the respective 0.244 Hz bins, as determined by paired t-test. Significance was reported only when ≥4 consecutive bins reached statistical significance..

Fig S8

Figure S8. Morphine withdrawal effects on normalized EEG power for wake (A-A””), NREM (B-B””), and REM (C-C””) sleep on withdrawal day (WD) 1 (A-C), WD2 (A’-C’), WD3 (A”-C”), WD5 (A’”-C’”), and WD7 (A””-C””) in male mice. Differences in normalized power are most prominent on WD1. Differences in normalized power during wake in the delta and alpha bands fail to reach significance; however, a significant increase in the beta and low gamma bands was present in the morphine treatment group (A). A significant increase occurred in the beta frequency band during NREM on WD1 (B) and WD2 (B’), and in both the beta and low gamma band during REM on WD2 (C’). Normalized power during wake was also significantly altered in the alpha and low gamma bands on WD5 (A’”), and in the delta, theta, and low gamma bands on WD7 (A””). Values are mean of the entire recording period for each 0.244 Hz bin±SEM. Measures from the morphine treatment are in blue and from the saline treatment in black. Blue lines at the top of each individual plot denote statistical significance (p<0.05) for the respective 0.244 Hz bins as determined by paired t-test. Significance was reported only when ≥ 4 consecutive bins reached statistical significance. Some mice exhibited high levels of non-physiological noise in the EEG channels; excluded recordings are described in the Methods section. .

Video S1

Video V1. Video depicting morphine-induced electroencephalogram (EEG; top channel), electromyogram (EMG; middle channel), EEG periodogram (bottom channel), and associated behavior. This female mouse had been administered 80 mg/kg morphine, s.c., approximately 1–1.5 hours earlier. Morphine at this dose is acutely activity- and wake-promoting (see video). The mouse circles almost continuously but its gait appear abnormal and uncoordinated. While Wake is normally characterized by high frequency, low amplitude activity occurring in the EEG, treatment with morphine at this dose also induced an aberrant spiking EEG pattern which coincides with abrupt pauses in locomotion seen in the video. Following cessation of the spiking EEG pattern, the mouse abruptly resumes movement.