Introduction

Sleep fragmentation (SF) represents a substantial form of sleep disruption, characterized by recurrent, brief interruptions throughout the night, typically quantified by the frequency of arousals combined with overall sleep duration1. Although SF and sleep deprivation (SD) overlap in parts of their clinical definition2, their effects on metabolic and neurophysiological functioning vary across time courses and magnitude3. Therefore, it is crucial to precisely differentiate between these two conditions and thoroughly investigate their distinct underlying mechanisms that alter metabolic homeostasis. In recent years, SF has gained escalating attention, and investigations have unveiled a correlation between SF and cardiometabolic risk factors such as insulin sensitivity and hypertension4,5,6. Population-based studies have also revealed that SF impairs glucose tolerance and insulin sensitivity7, which has been further reported and investigated in mouse models8,9. In addition, adipose tissue, recognized as a metabolically active endocrine organ, assumes a pivotal role in metabolism10. Previous studies highlighted the increase in the volume of visceral and subcutaneous adipose tissue following SF11. Despite the growing body of evidence in SF physiology, whether and how a short period of uninterrupted sleep recovery (SR) following chronic SF could restore glucose metabolism and adipose tissue gene regulation remains elusive. Since sleep is a vital process in restoring and maintaining metabolic homeostasis, directing our focus toward the ramifications of SR could provide insights into persistently altered pathways in SF. This, in turn, could further facilitate the development of the interventions aimed at ameliorating the detrimental effects of SF.

Meanwhile, there is a growing body of research on gut microbiota and its critical involvement in the pathogenesis of various chronic health conditions. In particular, dysbiosis in gut microbiota is linked to dysregulated glucose metabolism and adipose tissue morphology12,13. Emerging evidence from human and animal studies suggests that sleep restriction or SF can lead to substantial changes in the gut microbiome14,15,16. However, there have been limited studies systematically exploring the impact of SR following SF on gut microbiota composition and its potential association with changes in glucose metabolism.

Therefore, we aim to establish a chronic SF mouse model and investigate the effects of SR following SF on glucose metabolism, the gene expression of inguinal white adipose tissue (iWAT), and gut microbiota composition. Additionally, we aim to determine the duration required for the recovery of SF-induced glucose metabolism disturbances, as well as the subsequent alteration of the gut microbiota following the recovery of glucose tolerance. Furthermore, we aim to further analyze whether glucose metabolism alterations were related to the iWAT transcriptome and gut microbiota alterations.

Methods

Animals

Twenty-three female C57BL/6N mice (4-week-old, 12–15 g) were purchased from the Beijing Vital River Laboratory Animal Technology Co., Ltd (Beijing, China, SCXK-2016–0006). All mice were housed in standard pathogen-free vivarium facilities with a 12-h light/dark cycle (lights on at 8:00 AM, zeitgeber time ZT0 and lights off at 8:00 PM, zeitgeber time ZT12) at controlled temperature (22–24 °C) and humidity (40–60%). Mice were fed a standard chow diet and had free access to water. A one-week acclimatization period was provided before the start of the experiments. All experiments were approved by the Animal Welfare Committee at Capital Medical University (Protocol No. AEEI-2020–085) and were carried out in accordance with the ARRIVE Guidelines and relevant regulations.

SF intervention

The SF chamber (XR-XS108, Shanghai XinRuan Information Technology Co., Ltd., China) was used to induce SF. This SF chamber, resembling a spacious rounded cage with an integrated control panel, features a metallic sweeping bar positioned just above a layer of corncob bedding. The bar can be programmed to periodically rotate automatically, provoking instances of disrupted sleep in mice. Previous studies have commonly adopted this methodology to induce SF and subsequent sleep recovery (SR)14,17. Initially, two groups were established: the SF exposure group and the sleep control (SC) group, each comprising mice meticulously matched based on their weight. Within the SF group, the sweep bar remained stationary and turned off automatically during the dark cycle (ZT12-0), while it moved along the bottom of the cage every two minutes during the light cycle (ZT0-12). In parallel, control mice were also kept in identical SF chambers with the sweep bar immobile throughout the experiment, ensuring an uninterrupted sleep environment. The selection of a two-minute interval for sweeping bar motion was informed by clinical evidence reflecting severe sleep apnea conditions, which has been verified to disturb sleep patterns in both mice and rats18,19. After 8 consecutive weeks of SF during the light span, SF mice underwent a period of undisturbed sleep known as SR for 2 weeks and SR for 8 weeks, similar to the control mice.

Intraperitoneal glucose tolerance tests (ipGTT)

After 8 weeks of SF exposure (SF 8w) followed by a 2-week recovery period (SR 2 w), an intraperitoneal glucose tolerance test (ipGTT) was administered to all the mice, and half of the mice in each group were euthanized. The remaining mice underwent ipGTT every two weeks during the recovery period until glucose tolerance fully recovered, at which point the remaining mice were euthanized. After 16 h of fasting, mice received an intraperitoneal injection of glucose (2 g/kg of body weight). Blood glucose (BG) levels were measured before the injection, and subsequently at 15-, 30-, 60-, 90-, and 120-min post-injection, using tail vein blood samples and a Roche glucometer (Accu-Chek Performa, Switzerland). The calculation of the area under the curve (AUC) from the ipGTT data was performed as an assessment of glucose response.

Sample collection

Mice were anesthetized with an intraperitoneal injection of 0.3% pentobarbital sodium at a dose of 30 mg/kg. Euthanasia was performed via cervical dislocation. Inguinal white adipose tissue (iWAT) and cecal content samples of individual mice from each experimental group were collected after 10 h of fasting. These samples were snap-frozen in liquid nitrogen and stored at − 80 °C until further use.

Gut microbiota analysis

Considering the high microbial diversity, bacterial abundance, and sample yield in cecal contents, we performed 16S rDNA amplicon sequencing to investigate the gut microbiota composition in cecal samples. Microbial DNA was extracted from the cecal contents of each mouse using the QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany), following the manufacturer’s instructions, at Novogene Bioinformatics Technology Co., Ltd. (Tianjin, China). The V3–V4 hypervariable regions of the bacterial 16S rRNA gene were amplified using barcoded primers 341F (CCTAYGGGRBGCASCAG) and 806R (GGACTACNNGGGTATCTAAT). Amplicons were sequenced on the Illumina HiSeq 2500 platform (PE250, Illumina, CA, USA) using paired-end sequencing. The resulting sequences were analyzed using QIIME2 (Quantitative Insights Into Microbial Ecology 2, version 202,202; https://qiime2.org). The alpha diversity indices, including Shannon index and Simpson index were calculated on the data at an ASV level. The beta diversity analysis was performed vis principal coordinate analysis (PCoA) based on Bray–Curtis distances. ANCOM-BC20 was used to determine the differentially abundant taxa between the groups, using siblings as random effects and, with a Benjamini–Hochberg adjusted p value < 0.05 considered significant. Relevant analyses were performed using QIIME2 and custom Perl scripts. PICRUST2 was used for the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis.

RNA sequencing of iWAT

Total RNA was isolated from iWAT (3 mice each in the SF 8w-SR 2w and its control group) using the Total RNA Isolation Reagent (TRIzol) method (Invitrogen, Bio-Tek, Inc., USA). Then, the quantification and integrity assessment of RNA of iWAT were conducted using the RNA Nano 6000 Assay Kit on the Bioanalyzer 2100 system (Agilent Technologies, CA, USA) following the manufacturer’s protocol at Novogene Bioinformatics Technology Co., Ltd (Tianjin, China). The total RNA of iWAT was then purified utilizing the AMPure XP system (Beckman Coulter, Beverly, USA) for subsequent library preparation. Library size distribution was measured using the Qubit 2.0 Fluorometer, and the obtained results were analyzed through the Agilent 2100 Bioanalyzer. Subsequently, the sequencing libraries were combined and subjected to paired-end sequencing with 150 bp read length on an Illumina NovaSeq 6000 system.

Transcriptomic analysis of iWAT

Read alignment was performed using Hisat2. DESeq2 v1.20.0 under R4.2.2 was employed for differential gene expression analysis. We used a log2(|fold change|) > 1 and p < 0.05 for significance testing. To further explore the biological implications, Gene Ontology (GO) enrichment21 and KEGG pathways analysis22,23 of the differentially expressed genes (DEGs) were carried out. This was accomplished through the clusterProfiler R package (version 3.8.1) or the Database for Annotation, Visualization, and Integrated Discovery (DAVID) online tool (http://david.abcc.ncifcrf.gov/). Construction of Protein–Protein Interaction (PPI) networks was achieved utilizing the Search Tool for the Retrieval of Interacting Genes Database (STRING) (https://cn.string-db.org/) based on the DEGs identified. The resulting networks were visualized using Cytoscape 3.9.0, with intermediate confidence scores (confidence scores > 0.4). Key modules within the network were identified using the MCODE (version 1.6) plugin while hub genes of the protein–protein interaction (PPI) network were identified via four algorithms (degree, Maximal Clique Centrality (MCC), Maximum Neighborhood Component (MNC), and Edge Percolated Component (EPC)) using the CytoHubba plugin (version 24) within Cytoscape. GeneCards database (https://www.genecards.org/) was used to download glucose metabolism pathway-related genes.

Statistical analysis

Statistical analyses were conducted with IBM SPSS Statistics (version 25.0; https://www.ibm.com/products/spss-statistics) or GraphPad Prism (version 9.0; https://www.graphpad.com). Data were assessed for normality and were expressed as mean ± SD. The independent sample t-test was performed to compare the data between the two groups. Associations among selected differentially expressed genes (DEGs) in iWAT, gut microbiota, and glucose parameters were assessed using Spearman’s correlation analysis. p value < 0.05 (two-tailed) was considered a statistically significant difference.

Results

Effects of SR following SF on glucose level

To assess the effect of SR following prolonged SF on glucose regulation, ipGTT was conducted on all SF 8w-SR 2w (n = 12) and SC (n = 11) mice. As shown in Fig. 1A, although there was no significance in fasting blood glucose, mice previously exposed to SF displayed markedly elevated blood glucose levels during the ipGTT and significantly higher glucose levels at 15, 30, 60, 90, and 120 min after intraperitoneal injection of glucose, , with a peak around 18.4 mmol/L, compared to approximately 14.3 mmol/L in the control group. Throughout the study period, the mice previously subjected to SF exhibited diminished glucose clearance compared to the SC group, even following 2 weeks of SR. This was further supported by the area under the curve (AUC) above the baseline glucose concentration, indicating compromised glucose disposal in the SF group compared to controls (Fig. 1B).

Fig. 1
figure 1

Blood glucose concentrations during the intraperitoneal glucose tolerance test (ipGTT) and the area under the curve (AUC) for glucose. (A,B) Blood glucose concentrations during ipGTT and AUC between SF 8w-SR 2w group and SC group (n = 11–12 per group). (C,D) Blood glucose concentrations during ipGTT and AUC between SF 8w-SR 4w group and SC group (n = 5 per group). (E,F) Blood glucose concentrations during ipGTT and AUC between SF 8w-SR 6w group and SC group (n = 5 per group). (G,H) Blood glucose concentrations during ipGTT and AUC between SF 8w-SR 8w group and SC group (n = 5 per group). Results are means ± SD. *p < 0.05, **p < 0.01. SF 8w-SR 2w, sleep fragmentation for 8 weeks, then sleep recovery for 2 weeks. SF 8w-SR 4w, sleep fragmentation for 8 weeks, then sleep recovery for 4 weeks. SF 8w-SR 6w, sleep fragmentation for 8 weeks, then sleep recovery for 6 weeks. SF 8w-SR 8w, sleep fragmentation for 8 weeks, then sleep recovery for 8 weeks. SC, sleep control.

Full size image

To further determine the minimal duration of SR required to restore glucose tolerance, ipGTTs were conducted every two weeks throughout the SR period. After 4 weeks of SR, both blood glucose levels at 30 min post-glucose injection and the AUC remained significantly higher compared to the SC group (Fig. 1C, D). Similarly, at 6 weeks of SR, the 30-min glucose levels were still elevated (Fig. 1E), although no significant difference was observed in the AUC (Fig. 1F). By 8 weeks of SR, glucose metabolism had returned to normal, with no significant differences in either time-point glucose levels or AUC between groups (Fig. 1G, H).

SR for 2-week following SF still results in dysregulated gene expression in iWAT

In our study, we investigated the gene expression changes of iWAT using RNA sequencing after SR 2-week subsequent to SF. A total of 719 significantly dysregulated genes (DEGs), including 413 upregulated genes and 306 downregulated genes in the SF8w-SR2w group compared to the control group (log2|FC|> 1, and p < 0.05) (Fig. 2A). Figure 2B shows a hierarchically clustered heatmap of DEGs.

Fig. 2
figure 2

Transcriptomic alterations in iWAT after 2 week sleep recovery following sleep fragmentation. (A) Volcano plot illustrating DEGs between SF 8w-SR 2w and control mice. Genes with log2(|fold change|) > 1 and p < 0.05 were highlighted in red for significantly upregulated (306 genes) and blue for significantly downregulated (413 genes) after 2 weeks SR following 8 weeks SF. Genes not significantly differentially expressed are in grey. (B) The heatmap showed the most DEGs filtered by log2(|fold change|) > 1 and p < 0.05 in each group. n = 3 per group. DEGs, differentially expressed genes; SF 8w-SR 2w, sleep fragmentation for 8 weeks, then sleep recovery for 2 weeks.

Full size image

To comprehensively understand the biological and functional significance of the DEGs, we separately analyzed upregulated and downregulated genes through GO term enrichment analysis. The results are presented in Fig. 3A and B, showing the top ten terms in three categories: Biological Processes (BP), Cellular Components (CC), and Molecular Functions (MF) for both upregulated and downregulated genes. The analysis revealed that upregulated genes were primarily associated with muscle cell development, myofibril formation, and actin binding. Conversely, SF appeared to impact major GO terms related to negative regulation of chromosome segregation, the centromeric chromosome region, and microtubule binding among downregulated genes. These downregulated terms define the theme of cell proliferation. We further explored the KEGG pathways involved in the DEGs. Upregulated genes were predominantly enriched in pathways related to cardiac muscle contraction, hypertrophic cardiomyopathy, and dilated cardiomyopathy (Fig. 3C). Downregulated genes were primarily associated with oocyte meiosis, the cell cycle, and progesterone-mediated oocyte maturation (Fig. 3D).

Fig. 3
figure 3

Functional annotation of DEGs. GO analysis of the (A) upregulated DEGs and (B) downregulated DEGs. The ordinate is the GO term name including BP, CC, and MF, and the abscissa is the enrichment score of DEGs in the GO term. KEGG pathway enrichment of the (C) upregulated DEGs and (D) downregulated DEGs. The ordinate is the name of the KEGG pathway, and the abscissa is the fold enrichment of the pathway. The count of DEGs is represented by bubble size and P value by color. DEGs, differentially expressed genes; GO, Gene Ontology; BP, Biological Processes; CC, Cellular Components; MF, Molecular Functions; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Full size image

Construction of the PPI network, and identification of important modules and hub genes in iWAT

To explore the interactions among the proteins encoded by the DEGs, DEGs were uploaded to the STRING database to establish the PPI network. The network consisted of 567 nodes and 3954 edges (Fig. S1). To identify key modules of the PPI network, the MCODE plug-in was employed, and the top 3 modules were presented (Fig. 4A–C). Module 1 consisted of 66 nodes and 2027 edges, possessing a cluster score of 62.369, while module 2 contained 10 nodes and 42 edges with a score of 9.333. Module 3 comprised 8 nodes and 28 edges, exhibiting a cluster score of 8. For deeper insights into the underlying biological processes, GO analysis and KEGG pathway enrichment were performed specifically for the most significant module. The results of GO analysis unveiled that the functions associated with module 1, module 2, and module 3 were predominantly linked to cell division, endopeptidase inhibitor activity, and the cytosolic large ribosomal subunit, respectively (Tables S1, S3 and S5). The KEGG pathway enrichment analysis showed that module 1, module 2, and module 3 were mainly enriched in the cell cycle pathway, biosynthesis of amino acids, and the ribosome pathway, respectively (Tables S2, S4 and S6).

Fig. 4
figure 4

Identification of important modules and hub genes in the PPI network. (A) The first module, (B) the second module, and (C) the third module of the PPI network were detected by the MCODE plug-in and ranked from largest to smallest by score. Red nodes are upregulated DEGs, while blue nodes are downregulated DEGs. Edges between nodes represent interactions of DEGs. (D) Venn diagram of the top 20 genes calculated by four algorithms. PPI, protein–protein interaction; DEGs, differentially expressed genes.

Full size image

Given the considerable number of DEGs, our focus was on identifying hub genes within the PPI network. Subsequently, given significant glucose intolerance persisted in chronically sleep-fragmented female mice after two-week SR, we further aimed to determine whether the expression of these important hub genes correlated with the glucose parameters. To achieve this, the cytoHubba plugin was utilized, employing four distinct algorithms. The outcome of this analysis involved the identification of the top 20 genes using degree, Maximum Clique Centrality (MCC), Maximum Neighborhood Component (MNC), and Edge Percolated Component (EPC) algorithms. Notably, a Venn diagram analysis highlighted that three genes—non-SMC condensin I complex subunit G (Ncapg), centromere protein E (Cenpe), and TTK protein kinase (Ttk)—exhibited overlapping presence among all four algorithm-derived groups (Fig. 4D and Table 1).

Table 1 Characteristics of hub genes.
Full size table

Apart from three important hub genes, we also wanted to know whether there were glucose metabolism-related DEGs in iWAT and whether these genes were related to the aforementioned glucose intolerance. Therefore, we searched glucose metabolism pathway in the GeneCards database and found 307 genes, and compared these 307genes with the DEGs of our iWAT transcriptome for intersection, and identified 6 DEGs including leucyl/cystinyl aminopeptidase (Lnpep), phosphatase and tensin homolog (Pten), apolipoprotein E (Apoe), CCAAT/enhancer binding protein beta (Cebpb), indoleamine 2,3-dioxygenase 1 (Ido1), alpha-2-HS-glycoprotein (Ahsg), were related to glucose metabolism and the characteristics were shown in Table 2.

Table 2 Characteristics of glucose metabolism pathway related DEGs.
Full size table

Alterations in gut microbiome

The microbial community structure of cecal contents was evaluated by 16S rRNA gene amplicon sequencing. Amplicon-sequence variants (ASVs) were used to compare shared and unique taxa between groups. Initially, we analyzed the gut microbiota of the 2-week SR group and the control group. The results revealed that 1532 ASVs were shared between the groups, whereas 2581 ASVs were unique to the SF 8 w–SR 2 w mice and 1569 were unique to controls (Fig. 5A). A representation of the top 10 phyla and the leading 30 species at the genus level is provided in Fig. 5B and C, respectively. The dominant phyla were Firmicutes, Bacteroidota, and Proteobacteria.

Fig. 5
figure 5

Alterations in gut microbiota composition after sleep recovery following sleep fragmentation. (A) Venn diagram of the ASVs in SF 8w-SR 2w group and SC group; (B) Relative abundance of the top 10 phyla in SF 8w-SR 2w group and SC group; (C) Relative abundance of the top 30 species at the genus level in SF 8w-SR 2w group and SC group; (D) Venn diagram of the OUTs in SF 8w-SR 8w group and SC group; (E) Relative abundance of the top 10 phyla in SF 8w-SR 8w group and SC group; (F) Relative abundance of the top 30 species at the genus level in SF 8w-SR 8w group and SC group; (G–H) Shannon index, Simpson index of gut microbiota in SF 8w-SR 2w group and SC group; (I) PCoA plot based on the Bray–Curtis distance of gut microbiota in SF 8w-SR 2w group and SC group;. (J–K) Shannon index, Simpson index of gut microbiota in SF 8w-SR 8w group and SC group; (L) PCoA plot based on the Bray–Curtis distance of gut microbiota in SF 8w-SR 8w group and SC group. n = 5 per group. SF 8w-SR 2w, sleep fragmentation for 8 weeks, then sleep recovery for 2 weeks. SF 8w-SR 8w, sleep fragmentation for 8 weeks, then sleep recovery for 8 weeks. SC, sleep control.

Full size image

The same analysis was performed for the SR 8-week group and its corresponding control group. Here, 452 ASVs were shared, with 428 ASVs unique to the SF 8 w–SR 8 w group and 190 unique to its controls (Fig. 5D). The 10 most abundant phyla and 30 leading genera are illustrated in Fig. 5E and F; the predominant phyla were Firmicutes, Bacteroidota, and Actinobacteriota. Although no significant differences were observed in alpha diversity and beta diversity analyses between the SF 8w-SR 2w group and the corresponding SC group (Fig. 5G–I), notable alterations were detected at the genus level. Upon extending the sleep recovery period to 8 weeks, significant changes were still observed in the alpha diversity and beta-diversity index of microbial species composition in the SF 8w-SR 8w group compared to the corresponding SC group (ANOSIM test, R = 0.984, p = 0.007) (Fig. 5J–L).

Differential abundance analysis using the Analysis of Compositions of Microbiomes with Bias Correction (ANCOM-BC) revealed that 165 genera remained significantly altered after SR 2w. The differentially abundant genera among the top 200 relative abundance are presented in Fig. 6A. Even after 8 weeks of SR, 25 genera still showed significant changes (Fig. 6B). Notably, some microbial alterations persisted despite prolonged SR. In particular, the Rikenellaceae_RC9_gut_group and Defluviitaleaceae_UCG-011 exhibited sustained differences from as early as the 2-week SR stage, suggesting that these genera may serve as potential biomarkers for the long-term effects of SF.

Fig. 6
figure 6

Differential abundant bacterial genera based on the ANCOM-BC analysis. Logfold changes are visualized along the x-axis. All effect sizes with Benjamini–Hochberg corrected p values < 0.05 are displayed. (A) Differential abundance analysis of SF 8w-SR 2w group and SC group. (B) Differential abundance analysis of SF 8w-SR 8w group and SC group. n = 5 per group. SF 8w-SR 2w, sleep fragmentation for 8 weeks, then sleep recovery for 2 weeks. SC, sleep control. SF 8w-SR 8w, sleep fragmentation for 8 weeks, then sleep recovery for 8 weeks.

Full size image

Predicted function changes in the microbiota

We used the PICRUSt2 prediction method to obtain gene function annotations from the KEGG database, enabling the prediction of metabolic functions based on microbiota metagenomes. We analyzed level 3 pathways between the SF 8w-SR 2w group and its corresponding SC group (Fig. 7A), as well as the SF 8w-SR 8w group and its corresponding SC group (Fig. 7B). Notably, bacterial secretion system metabolic pathways exhibited significant variations in both the SF 8w-SR 2w and SF 8w-SR 8w groups when compared to the normal controls. This distinct differentially expressed pathway may provide valuable insights into the effects of SR.

Fig. 7
figure 7

Differential pathways predicted by PICRUSt2 analysis. (A) The differential pathways between the SF 8w-SR 2w group and SC group. (B) The differential pathways between the SF 8w-SR 8w group and SC group. SF 8w-SR 2w, sleep fragmentation for 8 weeks, then sleep recovery for 2 weeks. SC, sleep control. SF 8w-SR 8w, sleep fragmentation for 8 weeks, then sleep recovery for 8 weeks.

Full size image

Correlation analysis of glucose tolerance and altered gut microbiota

To investigate whether the differential glucose parameters from the ipGTT were associated with changes in the selected DEGs, including three hub genes and six glucose metabolism pathway related of iWAT, top 15 significantly altered gut microbiota at the genus level, a correlation analysis was conducted on the SF 8w-SR 2w group and the corresponding SC group. The Spearman correlation result was illustrated in Fig. 8. Our analysis showed that the abundance alterations of Faecalibaculum, Rikenellaceae_RC9_gut_group and Vagococcus were negatively correlated with blood glucose and AUC. The abundance alterations of [Eubacterium]_xylanophilum_group, Rickettsia, Vibrio and Prevotellaceae_NK3B31_group were positively correlated with blood glucose and AUC. All three hub genes were negatively correlated with BG90 and BG120, and Ttk was also negatively correlated with AUC. Among six glucose metabolism pathway related DEGs, Cebpb was positively correlated with BG15 and AUC, and Ido1 and Ahsg were both positively correlated with BG15, BG90, and BG120.

Fig. 8
figure 8

Heatmaps of Spearman correlation analysis of differential glucose parameters with altered gut microbes, three hub genes and six glucose metabolism-related genes of iWAT of SF 8w-SR 2w group and the corresponding SC group. Crosses show the correlation coefficients (r), *p < 0.05, **p < 0.01. BG15, blood glucose level at 15 min of ipGTT; BG60, blood glucose level at 60 min of ipGTT; BG90, blood glucose level at 90 min of ipGTT; BG120, blood glucose level at 120 min of ipGTT; AUC, area under the curve of ipGTT; Ncapg, non-SMC condensin I complex subunit G; Cenpe, centromere protein E; Ttk, TTK protein kinase; Lnpep, leucyl/cystinyl aminopeptidase; Pten, phosphatase and tensin homolog; Apoe, apolipoprotein E; Cebpb, CCAAT/enhancer binding protein beta; Ido1, indoleamine 2,3-dioxygenase 1; Ahsg, alpha-2-HS-glycoprotein.; SF 8w-SR 2w, sleep fragmentation for 8 weeks, then sleep recovery for 2 weeks. SC, sleep control.

Full size image

Discussion

In the present study, we established a chronic SF mouse model and observed that even after two weeks of SR following eight weeks of SF, glucose disturbances existed, and the transcriptome of iWAT and microbiota were significantly altered. We identified top hub genes Ncapg, Cenpe, and Ttk of iWAT from the PPI network, and six glucose metabolism pathway related genes including Lnpep, Pten, Apoe, Cebpb, Ido1, and Ahsg from GeneCards database. While there were no significant changes in the diversity of gut microbiota, alterations were observed in 165 species at the genus level. When the SR period was extended to 8 weeks, glucose metabolic parameters recovered, while the alterations in gut microbiota composition persisted.. Notably, the Rikenellaceae_RC9_gut_group and Defluviitaleaceae_UCG-011 exhibited sustained differences from as early as the 2-week SR stage. Furthermore, the relative abundance of the changed microbial genera, all three hub genes and three glucose metabolism pathway related genes of iWAT were associated with the altered glucose parameters.

We adopted the automatic SF device to induce sleep disruption events as performed in previous studies24,25, which is a more rational method to establish the SF mouse model because it can prevent human intervention, external stimuli, and social isolation of mice, and the stress hormone levels of mice can be stable. Earlier studies showed that a comparatively brief period of SF can disrupt glucose tolerance. For example, one previous study developed a 14-day model of instrumental SF in mice and found that sleep-fragmented mice developed glucose intolerance, including blood glucose levels at 45 min, 90 min, and 120 min, with increasing AUC compared to healthy controls9. A prior study discovered that rats exhibited notably higher glucose concentrations both at baseline and 30–90 min after the administration of anhydrous dextrose following 20 days of sleep restriction26. Experimental SF in healthy volunteers can also decrease insulin sensitivity and glucose effectiveness27. Some previous studies also explored whether sleep recovery can rescue the detrimental effects on glucose metabolism caused by SF. One clinical study reported that two week sleep extension in chronically sleep-deprived individuals could improve glucose metabolism28. However, another study has documented that subjecting mice to 18 h of daily SF for 9 days alone did not affect glucose tolerance, but when high-fat feeding was introduced alongside sleep disruption, it led to compromised glucose tolerance, and the adverse effect could mitigated through 24 h of recovery sleep29. The aforementioned findings unveil certain inconsistencies, particularly highlighting that glucose metabolism was influenced by a moderate duration of SF, whereas a short-term SF intervention in isolation might not yield significant effects. Interestingly, our results showed that glucose metabolism was deprecated even after 2 weeks of SR following 8 weeks of SF in mice, and only after 8 weeks of SR that the glucose metabolism impairment was fully rectified. It emphasized a persistent glucose metabolic dysregulation that cannot be rectified by a brief-to-moderate period of recovery. The discordance in intervention times for SF and SR could potentially account for the variations observed across different studies.

Many previous studies explored the effects of SD or SF on transcriptome alterations of specific tissues or organs, including gastric mucosa30, brain31, hippocampus32, pituitary33, and bone34. However, few studies evaluated transcriptome alterations of iWAT after chronic SF or SR. One human study reported that acute sleep loss could increase WAT carbohydrate turnover and impair glucose homeostasis by dramatically blunted morning-to-evening WAT transcriptome with uncoupling from the local clock machinery35. In the present study, we identified a total of 719 significantly dysregulated genes of iWAT even after SR for 2 weeks. To explore whether glucose intolerance was related to the altered iWAT transcriptome, we identified hub genes and glucose metabolism pathway related DEGs to do correlation analyses between these genes and differential glucose parameters. All three hub genes including Ncapg, Cenpe, and Ttk and three glucose metabolism pathway-related DEGs including Cebpb, Ido1, and Ahsg were correlated with glucose parameters, denoting the important mechanistic role in glucose tolerance after SR followed by SF. There were no studies exploring the influence and the mechanisms of Ncapg, Cenpe, and Ttk on glucose metabolism. Ncapg has been demonstrated to play a pivotal role in regulating the positioning of DNA on chromosomes and is essential for the development and progression of non-small cell lung cancer36. Cenpe, a component of the fibrous corona, is crucial for accurate chromosome segregation37. Ttk is involved in cell proliferation, division, and is indispensable for the organization of centromeres during mitosis and centrosome replication38. Decreased Cebpb expression in the liver was associated with increased insulin secretion in pancreatic beta cells39. Ido1 is a primary enzyme that produces immunosuppressive metabolites such as tryptophan and kynuridine, and Ido1 was shown to be a diagnostic and prognostic biomarker for diabetic nephropathy40. Ahsg and its protein FETUA are hepatokines known to be associated with insulin resistance and type 2 diabetes. Previous research found that melatonin may improve liver insulin resistance and hepatic steatosis by reducing ER stress and the resulting Ahsg expression41. Consistent with the above studies, our results showed the upregulated expression of Cebpb, Ido1, and Ahsg in the previous SF group, which may represent poor glucose metabolic outcomes. Therefore, we speculated that even after a short period of SR following chronic SF may through upregulate the gene expression of Cebpb, Ido1, and Ahsg and downregulate the gene expression of Ncapg, Cenpe, and Ttk in iWAT to influence glucose metabolism, which denoting the complicate mechanisms of glucose intolerance induced by SF and need more in-depth experiments to validate and elucidate the exact mechanisms.

Quantitative studies demonstrated the crucial role of gut microbiota in host health, and changes in the gut microbiota composition were associated with multiple diseases, including obesity, diabetes mellitus, and cardio-metabolic diseases42,43. Furthermore, evidence has substantiated the essential role of gut microbiota in preserving normal sleep physiology. Aberrant sleep patterns can disturb the composition and functionality of gut microbiota, thereby establishing a reciprocal relationship between gut microbiota and sleep15,44. A prior study showed that 5 h of SD failed to change the overall microbial composition, but the relative abundance of Clostridiaceae and Lachnospiraceae was slightly altered in sleep-deprived mice compared to the controls45. Another study involving humans highlighted that short-term sleep loss elicited subtle effects on human microbiota, encompassing increased Firmicutes:Bacteroidetes ratio16. Moreover, an extended period of SD or SF spanning, such as 4 weeks, has the potential to induce microbial dysbiosis, characterized by notable disturbances in both alpha- and beta-diversity within the microbial community14,46,47,48. However, few studies investigated the alterations of gut microbiota after SR following chronic SF. In the present study, we found that 2 weeks of SR following 8 weeks still resulted in alterations in the relative abundance of certain microbial taxa at the genus level.

Plentiful evidence revealed the crucial role of the gut microbiota on metabolic homeostasis. Previous research confirmed that Rikenellaceae_RC9_gut_group could modulate the protein and carbonate metabolism and improve nitrogen utilization49. Faecalibaculum could produce short chain fatty acids to enhance intestinal antioxidant capacity50. However, our results indicated the abundance of all the probiotics mentioned above was significantly reduced after 8 weeks of SF intervention followed by 2 weeks of recovery. In a study involving post-gestational diabetes women, it was found that the Prevotellaceae_NK3B31_group exhibited a significant association with elevated fasting blood glucose levels51. The abundance of the above mentioned non-probiotic bacteria in our study increased after 8 weeks of SF intervention followed by 2 weeks of recovery. Previous studies showed that there were no changes in gut microbiota after 2 weeks of sleep extension52, or the changes in gut microbial composition can be reversed after 1 week of SR53. Our study indicated that when the SR period was extended to 8 weeks, the relative abundances of the altered gut microbiota taxa partially recovered. Notably, the Rikenellaceae_RC9_gut_group and Defluviitaleaceae_UCG-011 exhibited sustained differences from as early as the 2-week SR stage.

In addition, we found that the relative abundance of certain differential gut microbiota was associated with altered glucose parameters. Specifically, our results showed that the relative abundance of[Eubacterium]_xylanophilum_group, Rickettsia, Vibrio and Prevotellaceae_NK3B31_group was positively associated with specific glucose parameters, while the relative abundance of Faecalibaculum, Rikenellaceae_RC9_gut_group and Vagococcus was negatively related to these glucose parameters. Plentiful evidence revealed the crucial role of the gut microbiome in glucose homeostasis, and gut microbiota alterations can impair the gut barrier to result in lipopolysaccharide increase and subsequent metabolic endotoxemia54.

However, we recognized that there are several limitations in our study. First, our research exclusively utilized female mice, which may not be fully representative of both males and females. Future studies should include both male and female mice for a more comprehensive understanding. Second, we did not assess the alterations of the sleep parameters mentioned after SF, which would facilitate better comparisons between SF and SR. Future studies with a detailed sleep analysis at the end of the sleep recovery would provide a more robust foundation for our conclusion. Third, all mice within each group were housed in two single cages, which may have led to cage-specific microbiota profiles and limited our ability to distinguish treatment effects from cage effects. As such, the microbiota-related findings should be interpreted with caution and considered hypothesis-generating rather than conclusive. Fourth, RNA-seq analysis was performed only at the 2-week time point following sleep recovery, while full restoration of glucose tolerance was observed at 8 weeks. Thus, the transcriptomic data may reflect only early molecular events associated with the onset of recovery, rather than long-term regulatory changes. Future studies incorporating additional time points (e.g., 4, 6, and 8 weeks) are warranted to better characterize the temporal dynamics of transcriptional changes during the recovery process. Additionally, we relied on P values rather than adjusted P values to evaluate the difference, which could potentially lead to false positives in the results. Moreover, the sample size per group was relatively small.

In conclusion, our study reveals that even 2 weeks of sleep recovery following chronic sleep fragmentation still induced glucose intolerance and altered adipose tissue transcriptome, and gut microbiota slightly. When the SR period extended to 8 weeks, glucose metabolic parameters had recovered while the changes in the gut microbiota persisted.. Moreover, glucose intolerance may be related to the relative abundance of certain germs at the genus level, the hub genes and glucose metabolism-related genes of iWAT. Further research is warranted to unravel the intricate mechanisms underlying these relationships and explore potential therapeutic interventions for improving metabolic health in individuals with disrupted sleep patterns.