Rotator cuff injury disrupts sleep architecture in mice and exhibits a light–dark phase–dependent pattern Authors: Liang Chen, Yan-song Zhu, Zheng-di Liu, Rui-song Wang, Jian-ning Sun, Peng Chen Published: 2026 ABSTRACT AbstractObjectivesRotator cuff injury (RCI) is a common musculoskeletal disorder and is frequently accompanied by sleep disturbances. However, objective assessments of sleep architecture alterations in this context remain limited. Here, we established a mouse RCI model to systematically characterize RCI-induced changes in sleep architecture and to provide neurophysiological clues to the mechanisms underlying sleep disruption.MethodsUsing a mouse RCI model, we analyzed sleep architecture based on 24-hour EEG/EMG recordings. Mice were assigned to RCI and sham groups, and we quantified vigilance-state durations, the sleep fragmentation index (SFI), episode numbers, and EEG power spectral density (PSD) features over 24 hours as well as across the light (12 h) and dark (12 h) phases. Data were analyzed using unpaired t-tests and two-way repeated-measures ANOVA.ResultsRCI mice exhibited marked disruptions in sleep architecture, characterized by reduced NREM sleep duration, increased wakefulness, and enhanced sleep fragmentation during the light phase, while total REM sleep duration remained largely unchanged. These alterations showed clear light–dark phase dependence, accompanied by significant alterations in stage-specific EEG spectral power. Compared with sham controls, the number of NREM episodes increased in RCI mice, whereas REM episode counts did not differ significantly. The SFI was significantly elevated in the RCI group, indicating impaired sleep continuity and reduced sleep quality.ConclusionThis study demonstrates that rotator cuff injury disrupts sleep architecture, with a pattern indicative of impaired sleep maintenance and increased sleep fragmentation. These findings provide new neurophysiological evidence linking rotator cuff injury to sleep disturbances and may inform future strategies to alleviate nocturnal sleep problems in patients with rotator cuff pathology.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12891-026-10065-y.Keywords: Rotator cuff injury, Sleep architecture, EEG/EMG, Sleep fragmentation, Power spectral density Introduction Rotator cuff injury (RCI) is one of the most common shoulder disorders and has a high prevalence among middle-aged and older adults. It is typically characterized by shoulder pain, restricted range of motion, and functional impairment, substantially compromising quality of life. Accumulating clinical evidence indicates that, beyond daytime pain, patients with RCI frequently experience prominent nocturnal pain and sleep disturbances, which are considered major drivers for seeking medical care [1, 2]. Nocturnal pain not only contributes to difficulties in sleep initiation and sleep maintenance but is also closely associated with pain chronification, affective disturbances, and poorer postoperative recovery outcomes [3, 4]. However, clinical evaluation of sleep problems associated with rotator cuff injury still relies largely on subjective questionnaires and rating scales, such as the Pittsburgh Sleep Quality Index (PSQI) and the visual analog scale (VAS). Objective characterization of sleep architecture and its underlying neurophysiological basis remains limited, which hinders a deeper understanding of the mechanisms involved [5]. Sleep is a highly organized brain state that is orchestrated by coordinated activity across the hypothalamus, brainstem, and thalamo–cortical circuits [6]. It can be classified into wakefulness, non-rapid eye movement (NREM) sleep, and rapid eye movement (REM) sleep based on electroencephalography (EEG) and electromyography (EMG) signals. In recent years, research on chronic pain has increasingly recognized that pain-related sleep disturbances do not necessarily present as a simple reduction in total sleep time; instead, they more commonly involve disrupted sleep architecture and reduced sleep continuity, particularly impaired NREM sleep stability and increased sleep fragmentation [7, 8]. Across multiple animal models of chronic pain, EEG studies have further shown that pain can be accompanied by stage-specific remodeling of sleep-related rhythms, such as abnormalities in slow-wave activity (delta band) as well as alterations in theta oscillations or spindle-related activity. These phenomena are thought to reflect disrupted thalamo–cortical rhythmic gating and intrusion of wake-promoting drive into sleep. Moreover, prior work suggests that pain-related sleep alterations may exhibit light–dark phase-dependent features, such that sleep-maintenance deficits and fragmentation are more likely to emerge during the physiological rest phase. However, the specific patterns reported vary across models and experimental conditions [9, 10]. Existing basic research on rotator cuff injury has predominantly focused on peripheral pathological changes, including tendon degeneration, inflammatory responses, muscle atrophy, and fatty infiltration, which has provided important insights into pain generators and biomechanical dysfunction. However, compared with peripheral mechanisms, central nervous system alterations associated with rotator cuff injury remain insufficiently investigated [11, 12]. In recent years, several studies have begun to examine central remodeling in rotator cuff injury or shoulder pain conditions, suggesting frequency-specific alterations in cortical activity and reorganization of sensorimotor networks during wakefulness. Nevertheless, most of these findings are derived from task-based or resting-state EEG and functional imaging, and it remains unclear whether such central changes are already present during sleep—an essential brain state—or are further amplified under sleep conditions [13]. Based on the above background, we used a mouse rotator cuff injury model combined with EEG/EMG recordings to systematically analyze changes in sleep architecture and sleep-related EEG rhythms following rotator cuff injury. By comparing the injury and sham groups over a 24-hour period and across the light and dark phases, we focused on sleep fragmentation, the relative proportions of NREM and REM sleep, and stage-specific spectral alterations. This study provides an objective characterization of sleep disturbances following rotator cuff injury and lays the groundwork for future mechanistic and translational investigations aimed at improving pain management and rehabilitation by targeting sleep [14]. Methods Animals, experimental design and grouping Adult male C57BL/6J mice were used in this study and housed under standard vivarium conditions. The mice were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China) (Certificate No.: SCXK (Hu) 2022-0004).To ensure a stable environment during sleep recordings, mice were singly housed in dedicated recording cages with ad libitum access to food and water, under a 12 h light/dark cycle (lights on at 07:00 and off at 19:00; ~100 lx). Ambient temperature was maintained at 24 ± 0.5 °C with relative humidity at 60 ± 5%, and electromagnetic and acoustic disturbances were minimized as much as possible. The overall experimental design and timeline of the study are illustrated in Fig. 1. Experimental timeline for the Rotator Cuff Injury (RCI) sleep study A two-group, parallel, sham-controlled design was employed. Mice were randomly assigned to either a sham surgery group or a rotator cuff injury (RCI) group (n = 7 per group). To reduce the impact of tethering-related stress and environmental changes on sleep architecture, mice were allowed to recover for at least 1 week after EEG/EMG electrode implantation and were habituated to the recording setup for at least 2 days before formal recordings. Continuous 24 h EEG/EMG recordings were then performed. Sleep outcomes were reported both for the full 24 h period and separately for the light and dark phases (12 h each) to evaluate light–dark phase–dependent alterations in sleep architecture. At the end of the experiment, mice were euthanized by gradual-fill CO₂ inhalation in an approved chamber (displacement rate ~ 20% of the chamber volume per minute) in accordance with institutional guidelines. Loss of consciousness was confirmed by the absence of righting and pedal withdrawal reflexes. Death was confirmed by cessation of respiration, and cervical dislocation was performed as a secondary physical method to ensure death. All experimental procedures were approved by the Institutional Animal Care and Use Committee (LACUC FJMU 2025 − 0324) and were conducted in accordance with relevant guidelines for animal welfare and experimental practice. Rotator Cuff Injury (RCI) model Mice were anesthetized with inhaled isoflurane (4–5% for induction and 1.5–3% for maintenance), and placed on a microsurgical table and secured in the right lateral decubitus position. Fur over the left shoulder was shaved and the surgical field was disinfected with povidone–iodine. A lateral acromial approach was used to incise the skin and fascia over the left shoulder, and the deltoid muscle was longitudinally split to expose the supraspinatus tendon. The supraspinatus tendon was completely transected at its humeral footprint. Complete separation of the tendon from the footprint was confirmed intraoperatively, and the muscle was allowed to retract freely. The surgical field was then thoroughly irrigated with sterile saline, and the deltoid, fascia, and skin were closed in layers. Postoperatively, penicillin was administered for 3 consecutive days for infection prophylaxis. Animals were provided with ad libitum food and water and were allowed free weight-bearing and activity in their home cages. Mice were monitored for signs of pain/stress and wound infection for 2 weeks. Animals with incision infection or other severe complications were excluded and were not included in subsequent EEG/EMG recordings and analyses [15, 16]. For sham surgery, the supraspinatus tendon was exposed using the same approach but was not transected at the humeral footprint. The deltoid and skin were then closed in layers using 4 − 0 sutures in the same manner. EEG/EMG electrode implantation Mice in both the sham and RCI groups underwent EEG/EMG electrode implantation under deep anesthesia. Anesthesia was induced with inhaled isoflurane (4–5%) and maintained at 1.5–3%. Animals were mounted on a stereotaxic frame, and body temperature was maintained throughout the procedure using a heating pad. A midline scalp incision was made and the skull surface was cleaned. Bregma and lambda were identified, and the skull was leveled in the anteroposterior and mediolateral planes before electrode placement and drilling. Two stainless-steel screw electrodes were implanted for epidural cortical EEG recording at the predetermined coordinates (relative to bregma: AP = + 1.0 mm, ML = ± 2.5 mm). A reference screw electrode and a ground screw electrode were placed over the posterior skull region at AP = − 4.5 mm, with the reference at ML = + 2.5 mm and the ground at ML = − 2.5 mm. For EMG recording, two flexible stainless-steel wire electrodes were inserted into the bilateral dorsal neck muscles (trapezius/splenius region) to form a differential recording configuration. All leads were soldered to a miniature connector (headstage), and the connector and wires were secured to the skull using dental acrylic. Before the acrylic cured, care was taken to ensure that the electrodes and wires were free of tension and that sufficient space remained between the connector and the scalp margins to allow wound closure. Postoperatively, antibiotics were administered and mice were allowed to recover for at least 7 days. During recovery, animals were monitored daily for body weight, general activity, wound healing, and headcap stability [17]. EEG/EMG data acquisition and power spectral analysis After electrode implantation, mice were allowed to recover for at least 1 week and were habituated to the recording system for ≥ 2 days before formal recordings. During recordings, mice were singly housed in dedicated recording cages with ad libitum access to food and water, and were able to see neighboring cages to reduce isolation-related stress. The recording environment was maintained under a 12 h light/dark cycle (lights on at 07:00 and off at 19:00; ~100 lx), with ambient temperature at 24 ± 0.5 °C and relative humidity at 60 ± 5%, while electromagnetic and acoustic disturbances were minimized as much as possible. EEG/EMG signals were pre-amplified and band-pass filtered (with an additional notch filter applied when necessary) and then digitized for acquisition. Data were acquired using the Medusa system, sleep scoring was performed with SleepSign, and power spectral analyses were conducted in MATLAB (R2023b). EEG signals were sampled at 250 Hz. Power spectral analysis was performed using 4 s epochs. Power spectral density (PSD) was computed over 0–25 Hz using fast Fourier transform (FFT), with a window function applied to reduce spectral leakage. Band-limited power was summarized for the delta (0.5–4 Hz), theta (6–10 Hz), and sigma (10–14 Hz) ranges for between-group comparisons. PSD results were normalized to the total power in the 0–25 Hz range and expressed as relative power (%). Sleep scoring criteria and procedures Sleep–wake states were classified into wakefulness (Wake), non-rapid eye movement (NREM) sleep, and rapid eye movement (REM) sleep based on combined EEG/EMG features. Wake was characterized by low-amplitude, desynchronized EEG activity with sustained high EMG tone; NREM was defined by high-amplitude delta activity (0.5–4 Hz) accompanied by reduced EMG; and REM was characterized by prominent theta activity (6–10 Hz) with near-silent EMG tone [18, 19]. EEG/EMG signals were segmented into 4 s epochs and automatically scored using SleepSign, followed by epoch-by-epoch manual review and correction. To minimize the influence of transitional states on subsequent analyses, only stable segments lasting > 10 s were included; operationally, a stable state was defined as ≥ 3 consecutive epochs (≥ 12 s) scored as the same vigilance state. Sleep fragmentation was quantified using the sleep fragmentation index (SFI), defined as the number of transitions from sleep (NREM or REM) to wakefulness divided by total sleep time (hours) [20, 21]. In addition, bout duration and state-transition metrics were extracted to characterize changes in sleep continuity. Statistical analysis Statistical analyses were performed using GraphPad Prism (version 10.1.2). Two-tailed unpaired t-tests were used to compare the RCI and sham groups with respect to the total duration and episode number of REM sleep (REMS), NREM sleep (NREMS), and wakefulness, as well as the sleep fragmentation index (SFI). To evaluate hourly changes in REMS, NREMS, and wakefulness across the light–dark cycle, a two-way repeated-measures ANOVA (RM ANOVA) was performed, with Group (Sham vs. RCI) as the between-subject factor and Time (hourly bins) as the within-subject repeated-measures factor. Main effects (Group, Time) and the Group × Time interaction were tested. When appropriate, post hoc comparisons at individual time bins were conducted with Bonferroni correction. For power spectral analyses, power spectral density (PSD) was computed separately during NREM and REM sleep, and band-limited power in the delta, theta, and sigma ranges was compared between groups using two-tailed unpaired t-tests. Data are presented as mean ± SEM, and statistical significance was set at P < 0.05. Results RCI disrupts 24-h sleep architecture and exacerbates sleep fragmentation To evaluate the impact of rotator cuff injury (RCI) on sleep architecture, we performed continuous 24-h EEG/EMG recordings in sham and RCI mice and quantified wakefulness, NREM sleep (NREMS), and REM sleep (REMS) using both hourly time-course and summary analyses (Fig. 2A–C). Across the 24-h time course, RCI mice showed reduced NREMS and increased wakefulness during the light phase (07:00–19:00), whereas REMS exhibited no obvious overall change. Two-way repeated-measures ANOVA (group × time) revealed significant main effects of time (F₍23,276₎ = 24.516, P < 0.0001) and group (F₍1,12₎ = 6.658, P = 0.024) on NREMS duration, with no significant time × group interaction (F₍23,276₎ = 1.430, P = 0.096) (Fig. 2A). For REMS duration, only a main effect of time was observed (F₍23,276₎ = 2.627, P < 0.0001), whereas the main effect of group and the interaction were not significant (group: F₍1,12₎ = 0.173, P = 0.685; time × group: F₍23,276₎ = 0.486, P = 0.979) (Fig. 2B). In contrast, wakefulness duration showed significant main effects of time (F₍23,276₎ = 31.531, P < 0.0001) and group (F₍1,12₎ = 5.823, P = 0.033), as well as a significant time × group interaction (F₍23,276₎ = 1.691, P = 0.027), suggesting a time-dependent effect of RCI on wakefulness across the 24-h cycle (Fig. 2C). Bonferroni-corrected post hoc comparisons across time bins revealed a significant between-group difference only for wakefulness at 09:00 (adjusted P = 0.02519; Δ = 18.90 min; Fig. 2C). No other time bins showed significant between-group differences for hourly NREM or REM duration after correction (Fig. 2A–B; all adjusted P > 0.05). Rotator cuff injury disrupts 24-h sleep architecture and increases sleep fragmentation. Hourly time course of time spent in (A) NREM sleep (NREMS), (B) REM sleep (REMS), and (C) wakefulness during 24-h continuous EEG/EMG recordings. Shaded backgrounds indicate the dark phase (19:00–07:00) and light phase (07:00–19:00). Sham mice are shown in black and RCI mice in red. D, E Representative temporally matched 60-min hypnograms from a sham mouse (D) and an RCI mouse (E); scale bar, 10 min. F–H Total 24-h duration of REMS (F), NREMS (G), and wakefulness (H). I–K Total 24-h episode numbers of REMS (I), NREMS (J), and wakefulness (K). L Sleep fragmentation index (SFI), defined as the number of transitions from sleep (NREMS or REMS) to wakefulness divided by total sleep time (h). M–O The average episode duration (min) for REM sleep (M), NREM sleep (N), and wakefulness (O) over the 24-h recording period. Data are presented as mean ± SEM. Time-course data (A–C) were analyzed using two-way repeated-measures ANOVA (group × time). Summary measures (F–L) were analyzed using two-tailed unpaired t-tests. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. n = 7 per group Representative hypnograms indicated more frequent sleep–wake transitions in RCI mice, consistent with impaired sleep continuity (Fig. 2D). Consistent with the time-course analysis, 24-h summary measures showed that total NREMS duration was significantly reduced in the RCI group compared with sham controls (t ₍12₎ = 2.5803, P = 0.0241), whereas total wakefulness duration was significantly increased (t₍12₎ = 2.4131, P = 0.0327); total REMS duration did not differ between groups (t₍12₎= 0.5766, P = 0.5770) (Fig. 2F–H). In terms of episode number, RCI mice exhibited a significant increase in NREMS episodes (t₍12₎ = 2.4997, P = 0.0279) and wakefulness episodes (t ₍12₎ = 2.2552, P = 0.0436), with no significant difference in REMS episodes (t₍12₎ = 1.2146, P = 0.2479) (Fig. 2I–K). Finally, sleep fragmentation was quantified using the sleep fragmentation index (SFI), defined as the number of transitions from sleep (NREMS or REMS) to wakefulness divided by total sleep time (h). The SFI was significantly elevated in RCI mice over 24 h (t ₍12₎= 3.2093, P = 0.0075), further supporting increased sleep fragmentation following RCI (Fig. 2L). Across the 24-h period, RCI mice showed a significantly reduced average episode duration for NREM sleep (Fig. 2N), whereas the average REM episode duration was unchanged (Fig. 2M). The average wake episode duration was also reduced in RCI mice (Fig. 2O). Dark phase RCI does not alter total sleep-stage duration but increases wake episodes The 24-h recordings were further separated into the dark phase (19:00–07:00) and the light phase (07:00–19:00) for analysis (Fig. 3). During the dark phase, compared with the Sham group, the RCI group showed no significant difference in total REM sleep duration (Fig. 3A; t₍12₎=1.0296, P = 0.3235). Total NREM sleep duration was also not significantly altered (Fig. 3B; t₍12₎=2.0827, P = 0.0593), and wake duration likewise showed no significant change (Fig. 3C; t₍12₎=1.5429, P = 0.1488). Phase-specific alterations in sleep architecture after rotator cuff injury. Dark phase (19:00–07:00). A–C Total duration (min) of REM sleep (REMS), NREM sleep (NREMS), and wakefulness during the dark phase. D–F Total episode numbers (#) of REMS, NREMS, and wakefulness during the dark phase. G, H Hourly heatmaps showing the distribution of REMS, NREMS, and wakefulness across the dark phase: state duration (G) and episode number (H). Colors represent relative changes normalized to the sham control mean (sham = 1; >1 indicates an increase; <1 indicates a decrease). I, J Hourly changes in mean bout/episode duration of REMS (I) and NREMS (J) during the dark phase (mean ± SEM). K Sleep fragmentation index (SFI) during the dark phase. (K1–K3) The average episode duration (min) during the dark phase for REM sleep (K1), NREM sleep (K2), and wakefulness (K3). Light phase (07:00–19:00). L–N Total duration (min) of REMS, NREMS, and wakefulness during the light phase. O–Q Total episode numbers (#) of REMS, NREMS, and wakefulness during the light phase. R, S Hourly heatmaps showing the distribution of REMS, NREMS, and wakefulness across the light phase: state duration (R) and episode number (S). Colors represent relative changes normalized to the sham control mean (sham = 1; >1 indicates an increase; <1 indicates a decrease). T, U Hourly changes in mean bout/episode duration of REMS (T) and NREMS (U) during the light phase (mean ± SEM). V Sleep fragmentation index (SFI) during the light phase. (V1–V3) The average episode duration (min) during the light phase for REM sleep (V1), NREM sleep (V2), and wakefulness (V3). Each dot represents one mouse; bars indicate mean and error bars indicate SEM. n = 7 per group (Sham and RCI). Summary measures (A–F, K, L–Q, V) were analyzed using two-tailed unpaired t-tests. Hourly changes in episode duration (I, J, T, U) were analyzed using two-way repeated-measures ANOVA (group × time) followed by Bonferroni-corrected post hoc multiple comparisons. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant Although overall stage durations remained unchanged, episode-based analysis revealed that the numbers of REM episodes (Fig. 3D; t₍12₎=1.2228, P = 0.2449) and NREM episodes (Fig. 3E; t₍12₎=0.3131, P = 0.7596) were not significantly different between groups. However, the number of wake episodes was significantly increased in the RCI group (Fig. 3F; t₍12₎=2.8047, P = 0.0159), indicating increased wake initiation during the dark phase. Temporal dynamics of sleep–wake state distribution and episode duration during the dark phase To visualize the distribution of vigilance states and the temporal changes in episode duration across the dark phase, we generated heatmaps of state occupancy (Fig. 3G–H) as well as time-course plots of REM and NREM episode duration (Fig. 3I–J). Two-way repeated-measures ANOVA revealed no significant main effects of time, group, or their interaction on REM episode duration during the dark phase (time: F₍11,132₎=0.651, P = 0.782; group: F₍1,12₎=0.295, p = 0.597; time × group: F₍11,132₎=1.029, P = 0.425) (Fig. 3I). For NREM episode duration, neither the main effect of time nor the time × group interaction reached significance (time: F₍11,132₎=1.368, P = 0.195; time × group: F₍11,132₎=0.514, P = 0.891). However, a significant main effect of group was observed (group: F₍1,12₎ =6.258, P = 0.028), indicating an overall difference in NREM episode duration between the two groups throughout the dark phase (Fig. 3J). Bonferroni-corrected post hoc comparisons did not identify any significant between-group differences at individual time bins for hourly REM or NREM episode duration during the dark phase (Fig. 3I–J; all adjusted P > 0.05). During the light phase, RCI reduces NREM sleep duration and increases wake duration During the light phase, compared with the Sham group, the RCI group showed no significant difference in total REM sleep duration (Fig. 3L; t₍12₎=0.5160, P = 0.6152). However, total NREM sleep duration was significantly reduced (Fig. 3M; t₍12₎=2.6704, P = 0.0204), whereas total wake duration was significantly increased (Fig. 3N; t₍12₎=2.9407, P = 0.0124), indicating disrupted sleep architecture during the light phase. RCI markedly increases wake and NREM episodes during the light phase, suggesting enhanced sleep fragmentation Episode-based analysis revealed no significant difference in the number of REM episodes between groups (Fig. 3O; t₍12₎=0.2836, P = 0.7816). In contrast, the RCI group exhibited a significant increase in the number of NREM episodes (Fig. 3P; t₍12₎=2.4509, P = 0.0305), as well as a pronounced increase in wake episodes (Fig. 3Q; t₍12₎=5.4537, P = 0.0001). These findings suggest that RCI markedly increases the frequency of sleep–wake transitions during the light phase, leading to exacerbated sleep fragmentation. Temporal dynamics of sleep–wake state distribution and episode duration during the light phase Two-way repeated-measures ANOVA showed no significant main effects of time, group, or their interaction on REM episode duration during the light phase (time: F₍11,132₎=0.905, p = 0.537; group: F₍1,12₎=0.027, P = 0.873; time ×group: F₍11,132₎=0.395, P = 0.956) (Fig. 3T). For NREM episode duration, a significant main effect of group was observed (group: F₍1,12₎=18.838, P = < 0.0001), along with a significant time × group interaction (time × group: F₍11,132₎=2.478, P = 0.007). The main effect of time showed a trend toward significance (time: F₍11,132₎=1.836, P = 0.054) (Fig. 3U). Similarly, no significant between-group differences were detected at any individual time bin for hourly REM or NREM episode duration during the light phase after Bonferroni correction (Fig. 3T–U; all adjusted P > 0.05). Sleep Fragmentation Index (SFI) No significant difference in the sleep fragmentation index (SFI) was observed between groups during the dark phase (Fig. 3K; t₍12₎=1.0671, P = 0.3069). However, the average episode duration analysis revealed shorter NREM and wake bouts in RCI mice during the dark phase (Fig. 3K2–K3), whereas REM episode duration was unchanged (Fig. 3K1).In contrast, compared with the Sham group, the RCI group exhibited a significantly higher SFI during the light phase (Fig. 3V; t₍12₎=2.4178, P = 0.0324), supporting that RCI exacerbates sleep fragmentation during the resting phase. Consistently, RCI mice showed reduced average episode duration for NREM sleep and wakefulness during the light phase (Fig. 3V2–V3), with no change in REM episode duration (Fig. 3V1). RCI decreased delta power during NREMS in both dark and light phases In the present study, EEG spectral composition during NREMS was compared between sham and RCI mice across the dark and light phases (Figs. 4A and D and 5A and D). During the dark phase, relative delta-band power during NREMS was significantly lower in RCI mice than in sham controls (t₍14₎=4.77, P = 0.0003; Fig. 4D), whereas theta and sigma power did not differ between groups (theta: t₍14₎=1.49, P = 0.1574; sigma: t₍14₎=1.56, P = 0.1411; Fig. 4D). During the light phase, relative delta power during NREMS remained significantly reduced in RCI mice compared with sham animals (t₍14₎=7.19, P < 0.0001; Fig. 5D), with no significant changes in theta or sigma power (theta: t₍14₎=1.87, P = 0.0824; sigma: t₍14₎=1.74, P = 0.1038; Fig. 5D). Effects of rotator cuff injury during the dark phase with analysis of power spectra. (Left) Spectrograms comparing EEG power spectral density during NREM sleep (A), REM sleep (B), and wakefulness (C) between sham and RCI mice during the dark phase. (Right) PSD curves showing the relative distribution of power (0–25 Hz) during NREM sleep (D), REM sleep (E), and wakefulness (F) in sham (gray) and RCI (red) mice. Insets depict relative band power in the delta (0.5–4 Hz), theta (6–10 Hz), and sigma (10–16 Hz) ranges. PSD was normalized to the total power in the 0–25 Hz range and expressed as percentage of total power (mean ± SEM). Dots represent individual mice. Statistical comparisons of band power between groups were performed using two-tailed unpaired t-tests; ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Sham, n = 8; RCI, n = 8. NREM, non-rapid eye movement sleep; REM, rapid eye movement sleep; PSD, power spectral density Effects of rotator cuff injury during the light phase with analysis of power spectra. (Left) Spectrograms comparing EEG power spectral density during NREM sleep (A), REM sleep (B), and wakefulness (C) between sham and RCI mice during the light phase. (Right) PSD curves showing the relative distribution of power (0–25 Hz) during NREM sleep (D), REM sleep (E), and wakefulness (F) in sham (gray) and RCI (red) mice. Insets depict relative band power in the delta (0.5–4 Hz), theta (6–10 Hz), and sigma (10–16 Hz) ranges. PSD was normalized to the total power in the 0–25 Hz range and expressed as percentage of total power (mean ± SEM). Dots represent individual mice. Statistical comparisons of band power between groups were performed using two-tailed unpaired t-tests; ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Sham, n = 8; RCI, n = 8. NREM, non-rapid eye movement sleep; REM, rapid eye movement sleep; PSD, power spectral density RCI modulated REMS oscillatory components in a phase-dependent manner The present study further characterized EEG power spectra during REMS in the dark and light phases (Fig. 4B and E, and 5B, 5E).During the dark phase, RCI mice exhibited significantly higher delta power during REMS (t₍14₎=3.48, P = 0.0037; Fig. 4E) accompanied by significantly lower theta power (t₍14₎=3.32, P = 0.0050; Fig. 4E), while sigma power was unchanged (t₍14₎=0.46, P = 0.6539; Fig. 4E). During the light phase, REMS spectra showed significantly higher theta and sigma power in RCI mice than in sham controls (theta: t₍14₎=3.58, P = 0.0030; sigma: t₍14₎=9.89, P < 0.0001; Fig. 5E), whereas delta power did not differ between groups (t₍14₎=1.09, P = 0.2958; Fig. 5E). Power spectrum analysis of the waking EEG revealed no group differences in either dark or light phase To address whether rotator cuff injury alters cortical oscillatory activity during wakefulness, waking EEG power spectra were compared between sham and RCI mice during both the dark and light phases (Figs. 4C and F and 5C and F). During the dark phase, relative band power during wakefulness did not differ between groups for delta (t₍14₎=1.997, P = 0.0656), theta (t₍14₎=0.805, P = 0.4342), or sigma (t₍14₎=1.235, P = 0.2370) bands (Fig. 4F). Similarly, during the light phase, no significant between-group differences were detected in waking delta (t₍14₎=0.693, P = 0.4995), theta (t₍14₎=1.087, P = 0.2956), or sigma (t₍14₎=1.842, P = 0.0867) power (Fig. 5F). Together, these findings indicate that the injury-related spectral changes were more prominent during sleep states rather than wakefulness under the present conditions. Discussion RCI is frequently accompanied by nocturnal symptoms and sleep complaints; however, prior work has relied largely on subjective questionnaires and lacks objective characterization of sleep architecture and EEG oscillations, which limits mechanistic inference and the identification of actionable central targets [8, 22]. Accordingly, we asked whether RCI produces light–dark phase–dependent disruption of sleep architecture and concomitant, vigilance state–specific remodeling of EEG spectral activity. To address this question, we established a mouse RCI model with sham-operated controls and performed continuous 24-h EEG/EMG recordings with standard scoring into wake, NREM, and REM states. We quantified state durations, episode numbers, and the sleep fragmentation index (SFI) over the full 24 h and separately within the dark and light phases (12 h/12 h), and further assessed NREM- and REM-specific power spectral density (PSD) and band-limited power [10, 23, 24]. Our main findings are that RCI-induced sleep disruption is more pronounced during the rest (light) phase—characterized by reduced NREM sleep, increased wakefulness, and more frequent state transitions with elevated fragmentation—whereas overall REM sleep remains relatively preserved. In parallel, sleep EEG oscillations exhibit vigilance state–dependent spectral changes, suggesting that RCI affects not only the amount of sleep but also the rhythmic organization linked to NREM slow-wave activity and sleep intensity [25, 26]. Overall, our findings align closely with the prevailing view that chronic pain/injury disrupts sleep architecture. A large body of clinical and translational evidence supports a bidirectional coupling between pain and sleep: pain promotes difficulties initiating and maintaining sleep and increases fragmentation, whereas sleep loss can impair endogenous analgesia and amplify pain perception, creating a vicious cycle [27, 28]. At the preclinical level, 24-h EEG/EMG studies across multiple pain models indicate that sleep disruption more commonly manifests as reduced NREM stability and increased fragmentation, rather than a simple reduction in total sleep time. For example, neuropathic pain models have been shown to reduce NREM sleep, increase fragmentation, and alter sleep-state EEG rhythms [29]. Similarly, inflammatory pain models exhibit light–dark phase–dependent abnormalities in sleep architecture, suggesting that state- and rhythm-dependent features are key characteristics of pain-related sleep disturbances [10]. Clinically, rotator cuff–related disorders are likewise associated with a high prevalence of sleep disturbance, and sleep has been reported to improve after repair or effective treatment, supporting the clinical relevance of the RCI–sleep disruption association [30]. Similarly, our results also diverge in several respects from portions of the pain–sleep literature (e.g., nonsignificant changes in REM sleep, weaker overall effects during the dark phase, and directions of certain spectral changes that differ from specific models). Such discrepancies are not uncommon and may arise from at least three key factors. First, differences in model etiology (musculoskeletal injury vs. neuropathic or inflammatory pain) likely shape distinct “primary effects” on sleep phenotypes. Neuropathic pain is often accompanied by more prominent central sensitization and thalamo–cortical gating abnormalities, which can drive broader remodeling across sleep states. In contrast, musculoskeletal injury—particularly when linked to posture- and activity-dependent discomfort, local inflammation, and functional limitation—may bias toward rest-phase hyperarousal and impaired sleep maintenance, resulting in stronger light-phase effects with a higher likelihood of relatively preserved REM sleep [27–29]. Second, the recording time window and disease trajectory may differ across studies. Pain-related sleep alterations evolve dynamically: early postoperative stages may be more strongly influenced by stress, inflammation, and analgesic management, whereas later stages may better reflect sustained hyperarousal and impaired sleep homeostasis. Thus, recordings conducted at different postoperative days or weeks can yield different directions and magnitudes of REM/NREM changes [31]. Third, differences in recording and analytic strategies can substantially affect sensitivity to fragmentation and episode-based metrics. For example, epoch length (4 s vs. 10 s), scoring algorithms and manual correction criteria, operational definitions and thresholds for bouts/episodes, and whether repeated-measures approaches (e.g., RM-ANOVA) are used to account for within-animal temporal dependence can all influence estimates of episode number and duration, contributing to apparent cross-study inconsistencies [32, 33]. Taken together, the overall concordance with prior work supports shared features of pain-related sleep disruption following RCI, whereas the observed differences suggest that musculoskeletal injury may exhibit stronger phase dependence and hyperarousal-like characteristics—hypotheses that warrant further dissection using behavioral assays and circuit-level approaches in future studies [6]. Together, these spectral findings support vigilance state– and light–dark phase–dependent remodeling of sleep-related oscillations following RCI. Recent evidence suggests that slow-wave activity during wakefulness can reflect the intrusion of sleep-like cortical dynamics after injury and may be linked to altered network function [34]. In parallel, delta-band activity has been discussed as a biomarker related to homeostatic sleep drive and sleep quality, although its direction of change may depend on context and pathology [35]. Motivated by these observations, the present study additionally examined waking EEG power spectra during both the dark and light phases; however, no significant between-group differences were detected in waking delta/theta/sigma band power in either phase (Figs. 4F and 5F). Several factors may account for this negative result. First, the RCI model represents a peripheral musculoskeletal injury rather than a direct brain lesion; therefore, overt “sleep-like” slowing during wakefulness may be less pronounced than in conditions involving primary cortical injury. Second, the sleep phenotype observed here primarily reflected sleep-state–specific remodeling—most notably impaired NREM stability and phase-dependent alterations in sleep oscillations—rather than a global shift of cortical dynamics that necessarily “spills over” into waking spectra. Third, the present recordings were obtained using epidural cortical EEG, which provides a spatially averaged signal; if wake slow waves are highly local and region-specific, such effects may be diluted in the epidural measure. Together, these considerations suggest that RCI-related neurophysiological alterations may be expressed more robustly within sleep states than during wakefulness under standard light–dark conditions. One unexpected finding worth emphasizing is that RCI-related sleep disruption was more pronounced during the light (rest) phase, whereas the overall impact during the dark (active) phase was relatively modest. This pattern suggests that the effects of RCI on sleep are not a uniform reduction in sleep across the 24-h cycle, but instead resemble impaired sleep maintenance and wake intrusion within the physiological rest window [36]. Several potential mechanisms may account for this phase dependence. First, pain and discomfort may be more readily perceived during the rest phase and may engage arousal systems (e.g., noradrenergic and orexin/hypocretin pathways), thereby increasing the likelihood of transitions from sleep to wake. Second, the light–dark system gates sleep–wake transitions; because mice are naturally biased toward sleep during the light phase, any additional wake-promoting drive (e.g., pain, anxiety, or stress) is more likely to manifest as fragmentation. Third, from the perspective of the hyperarousal model of insomnia, sleep disturbance often reflects intrusion into the rest phase rather than a simple substitution during the active phase [37]. Importantly, these unexpected observations shift the RCI-associated sleep phenotype from generic “sleep loss” toward a remodeling of sleep stability and light–dark phase–dependent organization, providing more precise targets for future mechanistic studies. Limitations Several limitations of this study should be noted. First, the present work is observational in nature and lacks causal manipulations (e.g., analgesic rescue, anxiolytic interventions, or targeted modulation of arousal and thalamo–cortical circuits). Therefore, we cannot yet decompose the association between RCI and sleep disruption into a definitive mechanistic chain. Second, the sample size and the use of only male mice may limit generalizability. Given known sex- and individual-level differences in pain–sleep interactions, future studies incorporating female mice and larger cohorts will be important to strengthen the robustness and translational relevance of these findings. Finally, Although a 24-hour window provides a standardized baseline characterization across a complete light–dark cycle, we acknowledge that extending recordings to multiple consecutive days could further inform delayed effects. However, multi-day analyses introduce substantially greater statistical complexity and a heavier multiple-comparison burden, which can increase the risk of false-positive findings unless accompanied by larger sample sizes and rigorous quality control. Accordingly, future studies will implement longitudinal multi-day EEG/EMG recordings with increased cohort size to support these analyses.Meanwhile, sleep outcomes should be integrated with histological measures, inflammatory markers, and indices of circuit-level plasticity, thereby establishing a more complete evidence chain linking peripheral tissue injury, state-dependent central remodeling, and sleep disturbance. Conclusions This study is the first to systematically characterize the impact of rotator cuff injury (RCI) on sleep architecture in mice. We demonstrate that RCI produces a phase-dependent sleep disruption phenotype, with particularly prominent alterations during the light (rest) phase, including reduced NREM sleep, increased wakefulness, and exacerbated sleep fragmentation. These changes are not limited to reduced NREM sleep time but also involve impaired sleep continuity, reflected by increased fragmentation and more frequent state transitions. In addition, RCI was associated with alterations in sleep-related rhythmic organization, suggesting impaired sleep maintenance. Collectively, our findings provide objective neurophysiological evidence linking rotator cuff injury to sleep disturbances and may help explain sleep complaints observed in rotator cuff–related conditions, laying a foundation for future studies aimed at identifying sleep-targeted therapeutic strategies and intervention targets. Supplementary Information Supplementary Material 1. Supplementary Material 2. Acknowledgements We sincerely thank the First Affiliated Hospital of Fujian Medical University for their kind assistance and collaboration. Abbreviations Rotator cuff injury Electroencephalography Electromyography Non-rapid eye movement Rapid eye movement Wakefulness Sleep fragmentation index Power spectral density Authors’ contributions LC, YSZ, and PC conceived and designed the study. YSZ, LC, and JNS performed the animal experiments, including the rotator cuff injury/sham surgeries and EEG/EMG electrode implantation. ZDL and RSW assisted with animal care, postoperative monitoring, and EEG/EMG recordings. LC and YSZ conducted sleep scoring and EEG spectral analyses. ZDL performed the statistical analysis and visualization. LC drafted the manuscript. PC critically revised the manuscript and provided overall supervision. All authors read and approved the final manuscript. Funding This study was financially supported by the Fujian Provincial Finance Department (Grant number: BPB-2023CP1073), in part by the Startup Fund for scientific research, Fujian Medical University (Grant number: 2024QH2033), and in part by the Health Science and Technology Program Projects of Fujian Province (Grant number: 2024CXA026). Data availability Data are available upon reasonable request to the corresponding authors. Declarations Ethics approval and consent to participate All animal procedures were approved by the Laboratory Animal Care and Use Committee (LACUC) of Fujian Medical University (approval No. LACUC FJMU 2025 − 0324) and were conducted in accordance with institutional guidelines for animal welfare. The animals were obtained from a licensed laboratory animal vendor and were not privately owned; therefore, owner informed consent was not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement Data are available upon reasonable request to the corresponding authors.