Aging‐Related Muscle Bmal1 Decline Contributes to Bone Loss in Mice via Enhancing IL ‐1α–Mediated Osteoclastogenesis

To investigate age‐related changes in circadian rhythms within murine muscle and bone, gastrocnemius muscle and tibia tissues were harvested from 2‐month‐old (young) and 18‐month‐old (aged) mice at six circadian time points (ZT1, ZT5, ZT9, ZT13, ZT17, ZT21, where ZT indicates Zeitgeber Time, hours after lights‐on) following 1 week of adaptation to a light–dark (LD) cycle. RNA‐seq analysis coupled with MetaCycle methodology revealed significant tissue‐specific remodeling of the circadian transcriptome during aging. In bone tissue, the number of rhythmic genes was greater in aged mice compared to young controls. Only 92 rhythmic genes were common between young and aged groups (Figure 1A), with 441 genes losing rhythmicity in aged bone while 1711 genes acquired novel rhythmicity (Figure 1A,D). Furthermore, the amplitude of rhythmic genes was significantly lower in aged bone (Figure 1B,C). Two‐way ANOVA revealed that the expression of core clock genes, specifically Bmal1, Dbp, and Nr1d1, was significantly altered in aged mice compared to young controls, with a notable reduction in the amplitude of Bmal1 observed in aged bone (Figure 1E). Conversely, in skeletal muscle, the total number of rhythmic genes showed a modest decrease in aged mice. A core set of 621 rhythmic genes was common to both groups (Figure 1F), while rhythmicity was lost for 1015 genes in aged muscle and 902 genes gained novel rhythmicity (Figure 1F,I). Aged muscle also exhibited a significant reduction in the amplitude of rhythmic genes (Figure 1G,H). The rhythmic expression profiles of core clock genes remained largely unchanged in aged muscle, with the exception of Cry1 and Dbp (Figure 1J). Besides, micro‐CT and molecular analyses established that 18‐month‐old aged mice develop distinct osteoporotic bone changes and skeletal muscle senescence (Figure S1). Additionally, while 18‐month‐old mice exhibited a significant reduction in cortical thickness, we observed a concurrent increase in cortical bone mineral density and cortical area (Figure S1G–I). This combination of decreased thickness and increased total area suggests an age‐related structural remodeling of the cortical shell, where compensatory periosteal expansion occurs alongside endosteal bone loss (Ferguson et al. 2003; Shim et al. 2022). Collectively, these results indicate a profound reorganization of the circadian transcriptome in both muscle and bone tissues of aged mice.

To investigate potential age‐related alterations in circadian crosstalk between muscle and bone, we first analyzed the phase distribution of rhythmic gene expression. In both tissues, the highest density of rhythmic genes peaked during the ZT1‐3 in both young and aged mice (Figure 1K,L). Notably, based on the continuous phase values calculated by MetaCycle, young skeletal muscle exhibited a secondary high‐density distribution bin of rhythmic genes peaking between ZT17 and ZT19 (Figure 1K). In aged muscle, this phase distribution shifted, with the secondary concentration of rhythmic phases occurring earlier, between ZT13 and ZT15 (Figure 1L). In bone tissue, rhythmic genes were predominantly concentrated in the ZT1‐3 window in young mice. However, the phase distribution across other time intervals was also altered in aged bone (Figure 1M,N). These results indicate phase remodeling of the rhythmic transcriptome in both tissues during aging. To assess the functional implications of this phase shift on intratissue signaling and muscle‐bone crosstalk, we employed CellphoneDB to analyze temporal changes in receptor‐ligand pairs. Within young bone tissue, intratissue interactions peaked at ZT1 and ZT13, whereas in aged bone, the peak interaction number occurred at ZT21 (Figure 1O,P). Intramuscular interactions in young mice lacked a significant temporal peak, but aged muscle exhibited a distinct peak at ZT1 (Figure 1O,P). Furthermore, analysis of muscle‐bone intertissue receptor‐ligand pairs revealed the first peak at ZT1 in both groups. However, the second peak shifted from ZT13 in young mice to ZT17 in aged mice (Figure 1O,P). To specifically explore the potential impact of muscle‐derived secreted protein on bone, we analyzed muscle ligand‐bone receptor pairs across all time points. This revealed a comprehensive remodeling in aged mice, encompassing changes in the repertoire of interacting pairs, their interaction strength, and the phase of peak interaction (Figure S2A,B), underscoring a profound reorganization of muscle‐to‐bone signaling dynamics with age.

These findings suggest that the circadian transcriptome undergoes reprogramming in both muscle and bone tissues of aged mice. Furthermore, the rhythmic signaling from muscle to bone is significantly altered during aging. This disruption in circadian coupling between muscle and bone may underlie the age‐related bone loss observed in these animals.

Bmal1 serves as a core transcription factor within the molecular circadian clock. Because bulk RNA‐seq represents a heterogeneous mixture of cells that can mask myofiber‐specific changes, we analyzed single‐cell RNA sequencing data from both human and mouse samples to examine alterations in Bmal1 expression during skeletal muscle aging at a higher cellular resolution (Kedlian et al. 2024). Compared to young mice, aged mice exhibited a significant reduction in Bmal1 expression within type II myofibers (Figure 2A,B). Furthermore, among all differentially expressed cell types, the mean expression level of Bmal1 was highest in myofibers (Figure 2B). Similarly, Bmal1 expression was significantly lower in the type I myofibers of aged human muscle compared to young human muscle (Figures 2C and S3A). Consistent with these findings, analysis of an independent murine scRNA‐seq dataset (Aging Atlas Consortium 2021) confirmed a significant reduction of Bmal1 expression in 16‐month‐old mice compared to 2‐month‐old mice across multiple cell populations, including type IIB fast‐twitch fibers, fibro‐adipogenic progenitors/fibroblasts, pericytes, endothelial cells, smooth muscle cells, myotendinous junction cell, neuromuscular junction precursor cell, macrophages, and tendons (Figure S3B). Notably, this age‐related decrease was most pronounced in type IIB fast‐twitch fibers (Figure S3B). To validate these transcriptomic changes at the protein level, we collected gastrocnemius muscle samples from 2‐month‐old young mice and 18‐month‐old aged mice at 4‐h intervals. Western blot analysis showed a marked decrease in BMAL1 expression in the gastrocnemius muscle of aged mice at ZT1, ZT17 and ZT21 compared to young mice (Figure 2D,E). Taken together, these results resolve an important discrepancy: while our initial bulk RNA‐seq analysis showed no significant change in overall Bmal1 mRNA levels—likely due to the masking effect of nonmyogenic cell populations within the highly heterogeneous whole‐muscle tissue—high‐resolution single‐cell analysis and protein quantification confirm that Bmal1 expression is indeed significantly and specifically reduced within individual myofibers of aged skeletal muscle.

To assess the potential impact of muscle circadian clock disruption on bone mass, we generated skeletal muscle‐specific Bmal1 knockout (mKO) mice by mating Bmal1 flox^+/+ mice with CKMM‐cre^+/−Bmal1 flox^+/+ mice to mimic Bmal1 downregulation on aged myofibers. WB analysis confirmed that BMAL1 expression was significantly reduced in the gastrocnemius muscle, while BMAL1 levels in the liver and bone remained unchanged (Figure 2F–I). We next assessed the wheel running behavior of the mice and found that the intrinsic circadian period of muscle Bmal1 knockout mice was significantly prolonged compared to control mice (Figure 2J,K). Since the running rhythm of mice is regulated by the core circadian clock in the suprachiasmatic nucleus (SCN) and its downstream neural signaling pathways, this result suggests that Bmal1‐deficient muscles may secrete factors that influence the core clock. Body composition analysis revealed that knockout mice had significantly reduced body fat content and increased lean mass, with no significant difference in body fluid content (Figure 2L–2N). Furthermore, the weights and muscle fiber cross‐sectional areas of both the gastrocnemius and quadriceps were significantly greater in knockout mice compared to the control group, accompanied by a significant downregulation of myostatin (Mstn) expression (Figure S4). These findings indicate an overall increase in muscle mass in the knockout mice, which is consistent with previous report (Liu et al. 2025). These findings indicate that skeletal muscle‐specific Bmal1 knockout may influence other organs via secreted proteins in the circulatory system. Additionally, the downregulation of Bmal1 in skeletal muscle cells does not appear to be the primary cause of muscle loss associated with aging.

Next, we collected femoral tissues from 4‐month‐old male mice and performed micro‐CT analysis. In male mice, the results showed significant reductions in cancellous bone density, BV/TV, and trabecular thickness in knockout mice compared to control mice. Additionally, the trabecular spacing was significantly increased in knockout mice (Figure 2O–T). The knockout mice exhibited no significant differences in cortical bone density compared with control femur (Figure 2U). In 10‐month‐old male mice, we observed significant reductions in cancellous bone density, BV/TV, trabecular number, and trabecular thickness in knockout mice compared to control mice, accompanied by a significant increase in trabecular separation, indicating further deterioration of the trabecular microarchitecture (Figure S5). These results indicate that skeletal muscle cell‐specific Bmal1 knockout leads to bone loss and trabecular bone microstructural deterioration in male mice. Thus, reduced expression of Bmal1 in skeletal muscle cells may contribute to the development of age‐related osteoporosis.

The balance between bone formation, regulated by osteoblasts, and bone resorption, controlled by osteoclasts, is crucial for maintaining bone homeostasis. Von Kossa staining of femur sections revealed that femoral mineralization was significantly reduced in skeletal muscle cell‐specific Bmal1 knockout mice compared to control mice (Figure 3A,B). To investigate the cause of the bone mass decrease due to Bmal1 knockout in skeletal muscle cells, we performed toluidine blue and tartrate‐resistant acid phosphatase (TRAP) staining. We observed no significant difference in osteoblast number between knockout and control mice (Figure 3C,D), but the number of osteoclasts was significantly higher in the knockout mice (Figure 3E,F). These findings suggest that Bmal1 knockout in skeletal muscle cells may lead to bone mass loss by promoting osteoclast differentiation. To test this hypothesis, we isolated primary skeletal muscle cells from control and Bmal1 knockout mice. We then cocultured these primary skeletal muscle cells with wild‐type (WT) primary bone marrow‐derived macrophages (BMMs) (Figure 3G). The results showed that Bmal1 knockout skeletal muscle cells promoted the expression of TRAP and CTSK at both the mRNA and protein levels in WT BMMs (Figure 3H–M). Furthermore, TRAP staining revealed that Bmal1 knockout skeletal muscle cells enhanced osteoclast differentiation of WT BMMs compared to control skeletal muscle cells (Figure 3N,O). In addition, by coculture of Bmal1 knockout skeletal muscle cells with WT BMSCs, we evaluated both the expression of osteogenic differentiation markers (Alp, Runx2, and Bglap) and the ultimate bone‐forming functional activity via mineralization assays. We found that there was no significant difference between the mKO and control groups (Figure 3P–R). The observed time‐dependent decline in Alp and Runx2 expression from day 6 to day 12 reflects the normal physiological transition of osteoblasts from the differentiation phase into the mature mineralization phase. Importantly, Bglap, a late‐stage osteoblast marker, was increased at day 12 compared with day 6 under both control and mKO coculture conditions. This finding supports progression toward late‐stage osteoblast maturation rather than a nonspecific suppression of osteogenic gene expression, without any observable cytotoxic effects in the coculture system.

These findings suggest that Bmal1 knockout in skeletal muscle cells may regulate osteoclast differentiation through a secretory pathway.

To investigate the mechanism by which skeletal muscle Bmal1 regulates osteoclast differentiation, we collected gastrocnemius muscles from mice every 6 h under a 12:12 light–dark cycle. Gene expression was measured at ZT2, ZT8, ZT14, and ZT20 using RNA sequencing. To account for the limited phase estimation precision associated with a 4‐time‐point design, we applied a stringent minimum amplitude threshold (Amp > 0.1) alongside a p < 0.05 cutoff during MetaCycle analysis. This strategy mitigates potential false positives from lower‐frequency sampling and ensures the identification of robust circadian disruptions. In the gastrocnemius muscles of control mice, we identified a total of 1630 rhythmic genes, whereas 1029 rhythmic genes were identified in Bmal1 knockout mice (Figure 4A). We further examined the expression of core clock genes in the gastrocnemius muscle. In Bmal1 knockout muscle, two‐way ANOVA revealed significant alterations in the expression of Bmal1, Per1, Per2, and Dbp. Additionally, the amplitudes of these core clock genes were reduced (Figure 4B).

Analysis of differentially expressed genes (DEGs) at various time points revealed that ZT8 had the highest number of DEGs (907), while ZT2, ZT14, and ZT20 had 111, 95, and 284 DEGs, respectively (Figure 4C). To identify the target genes directly regulated by Bmal1, we examined both the DEGs and those whose rhythmic expression was lost following Bmal1 knockout. This analysis revealed 270 rhythmic genes that showed differential expression (Figure 4D). Wikipathways enrichment analysis revealed convergence between pathways enriched for rhythm‐disrupted genes in aged mice and those enriched by the 270‐gene signature. Shared pathways comprised the MAPK signaling pathway, α6β4 integrin signaling pathway, EGFR1 signaling pathway, Focal adhesion, Focal adhesion‐PI3K‐AKT–mTOR signaling pathway, Insulin signaling pathway, and ROS‐associated pathways—specifically the Oxidative damage response pathway in aged mice and the Hypoxia‐dependent self‐renewal of myoblasts pathway in muscle‐specific Bmal1 KO mice (Figure S6A,B). KEGG pathway enrichment analysis of these 270 genes also identified several enriched ROS‐related signaling pathways, including the TNF, HIF‐1, PI3K‐AKT, and MAPK signaling pathways (Figure 4E). In the HIF‐1 pathway, we observed that the rhythmic expression of the antioxidant Hmox1 was significantly disrupted following Bmal1 knockout, emerging as the most significantly altered target in our differential analysis (Figure 4F). Furthermore, although two‐way ANOVA revealed no significant changes in the overall 24‐h expression levels of other oxidative stress‐related genes, such as Nrf2 and Cat, we observed a distinct reduction in their rhythmic amplitudes within the skeletal muscle of Bmal1 KO mice (Figure 4G,H). Notably, the rhythmic expression of these antioxidant genes in the gastrocnemius muscle of control mice peaked at ZT14, the time when mice are most active at night. However, after Bmal1 knockout, the expression of these genes at ZT14 was downregulated (Figure 4F–H). These findings suggest that Bmal1 knockout in skeletal muscle cells reduces the number of rhythmic genes in the gastrocnemius muscle, diminishes the amplitude of core clock genes, and disrupts the rhythmic expression of antioxidant‐related genes such as Hmox1, Nrf2, and Cat. Consistent with this, the circadian RNA expression profiles of Hmox1 and Nrf2 in skeletal muscle from aged mice aligned with those in Bmal1 KO mice relative to young controls (Figure S6C). Collectively, these findings indicate that Bmal1 deficiency induces a rhythmic gene expression profile that partially resembles the aged phenotype.

To investigate how Bmal1 regulates antioxidant gene expression, we knocked down Bmal1 in C2C12 myotubes using ADV‐shRNA and measured the expression of the antioxidant genes Hmox1 and Nrf2. Compared to the control group, Bmal1 knockdown significantly reduced the expression of both Hmox1 and Nrf2 (Figure 4I–L). Next, to determine whether Bmal1 directly regulates Hmox1 transcription, we analyzed a 2 kb region of the Hmox1 promoter and identified a BMAL1‐binding E‐box at positions 1957–1966 (Figure 4M). Additionally, a BMAL1‐binding signal was detected at this same region in the skeletal muscle ChIPseq database (Figure 4N). Through dual‐luciferase reporter assays, we found that cotransfection of Bmal1 overexpression plasmid with the Hmox1‐WT plasmid significantly increased luciferase activity in 293 T cells. However, when we cotransfected the Bmal1 overexpression plasmid with the Hmox1‐mutant plasmid (E‐box deletion), the transcriptional activation of luciferase by BMAL1 was abolished (Figure 4O). This suggests that BMAL1 binds to the E‐box element of the Hmox1 promoter to regulate its transcription. Furthermore, analysis of ChIP‐seq databases revealed that both NRF2 and BMAL1 possess adjacent binding motifs and can cooccupy the Hmox1 promoter region (Figure S7A). However, to determine if they form a collaborative regulatory complex, we performed co‐immunoprecipitation (Co‐IP) assays. The results showed no significant physical interaction between NRF2 and BMAL1 (Figure S7B), indicating that these two factors do not form a direct physical complex but rather independently regulate Hmox1 transcription.

To further confirm whether Bmal1 deletion in skeletal muscle cells inhibits Hmox1 expression and reduces antioxidant capacity, we infected C2C12 myotubes with ADV‐shRNA for 4 days. After the infection, we measured intracellular ROS levels using DCFH‐DA. Compared to the control group, Bmal1 knockdown significantly increased ROS levels in myotubes (Figure 4P–S).

Together, these results reveal that BMAL1 directly regulates the transcription of the antioxidant gene Hmox1 and plays a key role in maintaining ROS homeostasis in skeletal muscle cells.

GO pathway enrichment analysis of DEGs showed that NF‐κB signal transduction was enriched (Figure S8), suggesting that changes in the expression of antioxidant genes may influence muscle inflammation‐related responses. To investigate how reduced antioxidant genes in skeletal muscle cells, resulting from Bmal1 knockout, affect osteoclast differentiation, we conducted an inflammatory factor chip analysis. This analysis measured the expression of 40 inflammatory markers in the serum of control and muscle‐specific Bmal1 knockout mice at ZT8 and ZT20 (Figure 5A). Among these inflammatory factors, several exhibited diurnal variations in control mice that were abolished in Bmal1 knockout mice, including GM‐CSF, IL‐6, IL‐15, IL‐17, MCP‐1, IL‐21, and M‐CSF. Conversely, novel diurnal variation emerged in Bmal1 KO mice for RANTES, TNF RI, and Leptin (Figure S9). Another subset showed no diurnal variation but exhibited elevated baseline expression levels in Bmal1 KO mice (Figure S9), exemplified by IL‐1α, IL‐13, PF4, and Eotaxin‐2. Analysis of overall expression levels confirmed that serum concentrations of IL‐1α, IL‐13, PF4, and Eotaxin‐2 were significantly elevated in Bmal1 KO mice compared to controls (Figure 5B–E). While previous studies indicate that both IL‐1α and IL‐13 can be secreted by skeletal muscle cells under certain physiological conditions (Authier et al. 1999; Scheler et al. 2013), our in vitro experiments revealed a divergent response to circadian disruption. When we knocked down Bmal1 in C2C12 myotubes, we observed a significant increase in IL‐1α levels in the cell supernatant, whereas IL‐13 levels decreased significantly (Figure 5F). These findings suggest that the elevated serum IL‐13 observed in vivo does not originate directly from the Bmal1‐deficient skeletal muscle. Instead, IL‐1α is one of the pro‐inflammatory factors directly released by the muscle, playing a key role in regulating osteoclast differentiation.

Carnosol, an activator of HMOX1, interacts with HMOX1 through hydrogen bonding and hydrophobic forces (Chu et al. 2024). To determine whether the increase in IL‐1α expression associated with Bmal1 knockout is linked to Hmox1 expression, we treated C2C12 myotubes with various concentrations of Carnosol. The results showed that 5 nM, 10 nM, and 20 nM Carnosol activated HMOX1 expression in a dose‐dependent manner (Figure 5G). When we treated myotubes with 10 nM Carnosol for 3 h daily over two consecutive days, we observed that Carnosol significantly inhibited the increase in IL‐1α expression induced by Bmal1 knockdown (Figure 5H,I). Simultaneously, we measured intracellular ROS levels in myotubes and found that Bmal1 knockdown markedly elevated ROS, whereas Carnosol treatment significantly reduced ROS levels in both wild‐type and Bmal1 knockdown myotubes (Figure 5J,K). These results suggest that HMOX1 can mitigate the inflammatory response triggered by Bmal1 deletion, highlighting the role of antioxidant signaling in maintaining inflammation homeostasis in muscle cells.

Additionally, we performed primary cell coculture experiments and found that IL‐1α antibody treatment could reverse the effects of muscle‐specific Bmal1 knockout on osteoclast differentiation. Specifically, IL‐1α antibody treatment reduced the expression of osteoclast differentiation markers, including NFATC1, TRAP, and CTSK, that were upregulated due to Bmal1 knockout (Figure 5L,M). TRAP staining and subsequent quantitative analysis further confirmed that the IL‐1α neutralizing antibody significantly inhibited the osteoclast hyperactivation caused by myocyte‐specific Bmal1 deletion. Specifically, coculture with Bmal1‐deficient myocytes increased osteoclast differentiation by an average of 2.37‐fold compared to the control group. However, following IL‐1α antibody intervention, this pathological enhancement was substantially reversed, with osteoclast differentiation returning toward baseline levels (1.47‐fold of the control) (Figure 5N,O). These quantitative results indicate that muscle‐derived IL‐1α is an important inflammatory mediator contributing to osteoclastogenesis in this model.

Previous studies have shown that Bmal1 knockdown promotes the secretion of IL‐1β in macrophages (Early et al. 2018), suggesting it may also act as a potential regulator of bone metabolism. Therefore, we validated the regulation of IL‐1α and IL‐1β by BMAL1 through both in vitro and in vivo experiments. At the RNA level, Bmal1 knockdown in C2C12 myotubes decreased the expression of both Il1a and Il1b (Figure S10A); however, at the protein level, both IL‐1α and IL‐1β were upregulated (Figure S10C,D). Furthermore, similar to IL‐1α, the level of secreted IL‐1β in the myotube culture medium was also increased (Figure S10B). In vivo, we assessed the protein levels of IL‐1α and IL‐1β in muscle tissues at ZT8 and ZT20. The results showed that in the knockout group, the levels of IL‐1α were significantly elevated at both ZT8 and ZT20, whereas IL‐1β showed no obvious alterations (Figure S10E,F). These results suggest that Bmal1 knockdown may promote the translation and secretion of the IL‐1 family in vitro. However, in the in vivo environment, Bmal1 deficiency enhances the expression of IL‐1α, suggesting that in vivo, muscle‐derived IL‐1α accumulates more prominently than IL‐1β. While our in vitro data show that Bmal1 deficiency can enhance the translation and secretion of both IL‐1 family members despite decreased mRNA levels, research has indicated that IL‐1β does not exert a direct regulatory effect on the osteoclast differentiation of BMMs (Levescot et al. 2021). Thus, the stimulatory effect observed in our coculture system is likely driven more strongly by IL‐1α. Collectively, these results indicate that muscle‐specific Bmal1 knockout induces systemic dysregulation of serum inflammatory cytokines. Skeletal muscle contributes to this inflammatory milieu by releasing IL‐1α following the modulation of Hmox1 expression, acting as a key—though likely not exclusive—mediator of enhanced osteoclast differentiation.

TRF can effectively entrain peripheral clocks and mitigate metabolic desynchrony (Chaix et al. 2014; Manoogian and Panda 2017; Zarrinpar et al. 2016), even in skeletal muscle lacking central clock control (Kumar et al. 2024). We hypothesized that driving systemic diurnal metabolic rhythms through TRF could rescue bone loss in muscle‐specific Bmal1 knockout mice. Following a 1‐week acclimation, mice were assigned to either ad libitum (AL) feeding or a specific ZT12‐16 TRF schedule. This particular window was selected because initiating TRF at ZT12 optimally aligns with the murine active phase (Regmi et al. 2021). Furthermore, to successfully entrain the clock‐deficient muscle, we utilized a compressed 4‐h feeding duration to maximize the fasting–feeding amplitude, serving as a potent systemic zeitgeber previously validated by our group (Zhai et al. 2022). After 2 weeks, the mice were returned to a free‐feeding regimen for another week, and femoral tissues were collected (Figure 6A). We found that TRF at ZT12‐16 significantly reduced body weight compared to mice fed AL (Figure 6B), and during the initial 2 days of the dietary intervention, food intake in the TRF group was significantly lower than that of the AL control group (Figure 6C). Micro‐CT analysis revealed that TRF at ZT12‐16 increased cancellous bone density, BV/TV, and the number of trabeculae while reducing trabecular space (Figure 6D–6I). Further investigation of Hmox1 expression in the gastrocnemius muscle by qPCR revealed that TRF at ZT12‐16 drove a feeding‐induced diurnal variation in Hmox1 expression in muscle‐specific Bmal1 knockout mice (Figure 6J). Additionally, serum IL‐1α levels significantly decreased after TRF treatment (Figure 6K).

These results indicate that TRF at ZT12‐16 can improve bone mass loss and trabecular bone microstructural damage induced by muscle Bmal1 knockout. It also enhances the rhythmic expression of the antioxidant gene Hmox1 in skeletal muscle and reduces systemic IL‐1α levels.

We also acclimated 5 and 17‐month‐old mice to a light/dark cycle for 1 week before subjecting them to ZT12‐16 TRF or AL. After 8 weeks, the mice were returned to a free‐feeding regimen for one more week, and femoral tissues were collected (Figure 6L). We observed that TRF significantly reduced body weight compared to AL in both 5‐ and 17‐month‐old mice (Figure 6M); furthermore, food intake in the TRF groups was significantly lower than that of the control groups during the first 3 and 6 days of the dietary intervention, respectively (Figure 6N). Micro‐CT analysis showed that TRF increased cancellous BMD, BV/TV, and the number of trabeculae (Figure 6O–6R), while trabecular spacing and thickness remained unchanged (Figure 6S,T) in 17‐month‐old mice. However, in 5‐month‐old mice, 8 weeks of TRF had no significant effect on bone mineral density or trabecular bone parameters (Figure 6O–6T). TRF also drove a feeding‐induced diurnal Hmox1 expression in aged muscle (Figure 6U). Serum IL‐1α levels significantly decreased following TRF treatment compared to AL aged mice (Figure 6V). Furthermore, we collected muscle tissue samples at ZT2 and ZT14 on the final day of the TRF intervention. RNA‐seq analysis revealed that at ZT14, TRF induced the upregulation of 169 genes and the downregulation of 95 genes in the muscle, and TRF resulted in 686 upregulated and 225 downregulated genes at ZT2 (Figure S11A). Subsequent pathway enrichment analysis of these DEGs highlighted the enrichment of classical cell proliferation‐ and aging‐related signaling pathways, including the p53, AMPK, FoxO, and PI3K‐Akt pathways. Concurrently, the circadian entrainment pathway was also significantly enriched, suggesting that TRF may modulate muscle biological rhythms and subsequently influence muscle aging (Figure S11B).

These results indicate that TRF at ZT12‐16 enhances the rhythmic expression of the antioxidant gene Hmox1 in aged skeletal muscle, reduces systemic IL‐1α levels, and ultimately alleviates bone loss in aged mice.

CKMM‐cre and Bmal1 flox mice were obtained from Dr. Ying Xu's lab at the Cambridge‐Soochow Genomic Resource Center, Soochow University. Skeletal muscle‐specific Bmal1 knockout mice were generated by crossing Bmal1 flox^+/+ mice with CKMM‐cre^+/−Bmal1 flox^+/+ mice. CKMM‐cre^−/−Bmal1 flox^+/+ mice were used as controls. All mice were housed in an SPF animal facility under a 12‐h LD cycle. The wheel‐running activity of the mice was recorded and analyzed using the ClockLab system (Actimetrics, Evanston, IL). In the time‐restricted feeding (TRF) experiment, mice were randomly divided into experimental and control groups. All animal procedures were approved by the Animal Care and Use Committee of the CAM‐SU Genomic Resource Center, Soochow University (YX‐2021–2).

4‐month‐old muscle‐specific Bmal1 knockout and control mice were weighed and subjected to body composition analysis using an LF50 body composition analyzer. The measuring chamber was cleaned before each measurement to ensure a dry environment. Body composition assessment comprised three key parameters: fat mass, lean mass, and free body mass.

Fresh gastrocnemius muscle tissue from mice was collected and fixed for 24 h. The tissue was then embedded in paraffin and sectioned. The paraffin sections were deparaffinized to water, followed by antigen retrieval using Tris‐EDTA buffer. To block endogenous peroxidase activity, 3% hydrogen peroxide was added and incubated at room temperature in the dark for 25 min, followed by PBS washing. The sections were then blocked with 3% BSA at room temperature for at least 30 min. Primary antibodies, including anti‐P21 (Affinity, 1:500), were added and incubated overnight at 4°C. After PBS washing, HRP‐labeled secondary antibodies corresponding to the primary antibodies were applied and incubated at room temperature for 50 min. The sections were washed three times with PBS and developed using DAB. Hematoxylin was used for nuclear counterstaining. Finally, the sections were mounted with neutral resin, and images were captured.

Cells or gastrocnemius muscles were lysed in RIPA buffer containing protease inhibitors and incubated on ice for 30 min. The subsequent steps were identical to the tissue protein extraction procedure. Protein samples were mixed with 5× loading buffer and denatured at 100°C for 5 min. After SDS‐PAGE electrophoresis, proteins were transferred onto PVDF membranes and blocked with 5% milk. The membranes were incubated overnight at 4°C with primary antibodies (anti‐BMAL1: Proteintech 14,268–1‐AP 1:1000, anti‐NFATC1: Affinity DF6446 1:1000, anti‐TRAP: Affinity DF6989 1:1000, anti‐CTSK: Daige db8038 1:1000, anti‐HMOX1: Daige db3474 1:1000, anti‐IL‐1α: Epizyme R015789 1:1000, anti‐IL‐1β: Epizyme R015789 1:1000, anti‐GAPDH: Proteintech 60,004–1‐Ig 1:3000, anti‐Tubulin: Proteintech 11,224–1‐AP 1:3000) at appropriate dilutions. After washes with TBST three times, membranes were incubated with secondary antibodies (Proteintech SA00001‐1, 1:10000; Proteintech SA00001‐2, 1:10000) at room temperature for 1 h. Bands were visualized using a chemiluminescence imager (Tannon), and protein levels were quantified using ImageJ software.

Unilateral femurs were fixed in 4% PFA and transferred to 70% ethanol for storage. Samples were scanned using a Bruker SkyScan 1176 micro‐CT scanner at 50 kV, 500 μA, 1000 ms, with a voxel size of 9 μm. For trabecular measurements, a 1.75‐mm region of interest, starting 540 μm proximal to the distal growth plate, was analyzed. Bone microstructural parameters, including bone mineral density (BMD), bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), and trabecular separation (Tb.Sp), were calculated. Materialize Magics 24.0 software was used to generate 3D visualizations of cortical and trabecular bone structures.

Mouse femoral hard tissue sections were fixed in 100% acetone for 10 min, repeated twice. The sections were then infiltrated with 70% ethanol for 3 min and placed in Wahaha purified water for at least 3 min. For staining, the sections were sequentially immersed in VK1 solution for 5 min and washed with double‐distilled water (ddH₂O) for 2 min, followed by immersion in VK2 solution for 5 min with a gentle ddH₂O wash for 2 min. The process was repeated with VK3 solution under the same conditions. For counterstaining, the sections were treated with Gleson solution containing acidic fuchsine for 30 min, briefly rinsed in 1% acetic acid for 5 s, and washed in ddH₂O for 5 s. After air‐drying, the sections were dehydrated through sequential immersion in 90% ethanol and 100% ethanol. The dehydration process was completed with one treatment of ethanol/toluene solution and two treatments of pure toluene. Finally, the sections were mounted with a coverslip.

Mouse femoral sections were fixed in 100% acetone for 10 min, repeated twice. The sections were then infiltrated with 70% ethanol for 3 min and placed in Wahaha purified water for at least 3 min. Drop toluidine blue staining solution on the sections and stain for 20–30 min. Wash gently with running water to remove excess stain. Perform color differentiation using 95% ethanol under a microscope to control the differentiation effect. Soak in anhydrous ethanol for 1 min, followed by three 1–2 min dehydration steps using xylene, and finally mount with neutral balsam. Cartilage and osteoblasts appear blue‐purple, while the background is light blue.

Mouse femoral sections were fixed in 100% acetone for 10 min, repeated twice. The sections were then infiltrated with 70% ethanol for 3 min and placed in Wahaha purified water for at least 3 min. Place the sections in 0.2 M Tris–HCl buffer and incubate at 37°C for 1 h. Subsequently, place the sections in TRAP staining solution and monitor under a microscope for TRAP‐positive changes. Stop the reaction at the appropriate time and wash with ddH₂O. After air drying, dehydrate by using xylene for transparency and mount with a cover slip.

Primary BMMs were isolated from the femurs and tibias of 2‐month‐old C57BL/6 mice. Briefly, after excising both ends of the bones, the bones were placed into 200 μL pipette tips nested within 1.5 mL Eppendorf tubes and briefly centrifuged to collect the bone marrow. The marrow was resuspended in serum‐free α‐MEM and filtered through a 70‐μm cell strainer. Following centrifugation, the cell pellet was resuspended in α‐MEM supplemented with 10% FBS and incubated. After 20 h, the supernatant containing nonadherent cells was collected and further cultured in a medium supplemented with 30 ng/mL M‐CSF (Novoprotein, CB34↗). Once the BMMs reached confluence, they were washed with cold PBS, gently detached using a cell scraper, and subcultured into culture plates. On the following day, upon reaching approximately 30% confluence, osteoclastogenesis was induced by supplementing the medium with 30 ng/mL M‐CSF and 100 ng/mL RANKL (Novoprotein, C28A↗). Change the medium every other day. On day six, osteoclast differentiation can be clearly observed. Fix the cells with 2.5% glutaraldehyde for 10 min, wash thoroughly with distilled water, incubate in TRAP staining solution (Beyotime, P0332) at 37°C for 30–60 min, wash again with distilled water, and take pictures under a microscope. For coculture experiments, primary myocytes were seeded in the upper chamber of a transwell system, while BMMs were plated in the lower chamber.

Mouse BMSCs were purchased from Procell Life Science & Technology Co. Ltd. (Wuhan, China). Cells were cultured in α‐MEM supplemented with 10% FBS and 1% penicillin/streptomycin. For coculture experiments, primary myocytes were seeded in the upper chamber of a transwell system, while BMSCs were plated in the lower chamber. Upon reaching 90% confluence, BMSCs were induced with osteogenic differentiation medium (α‐MEM containing 10% FBS, 10 mM β‐Glycerophosphate disodium salt hydrate (Abmole, M3837), 50ug/ml VC (Abmole, M3121), and 10 nM dexamethasone (MCE, HY‐14648)). The medium was replaced every 2 days. For the mineralization assay, after 14 days of induction, the cells were fixed with 4% paraformaldehyde and stained with Alizarin Red S solution (OriCell, ALIR‐10001) to evaluate extracellular matrix calcium deposits.

The gastrocnemius muscle and tibia were collected and quick‐frozen with liquid nitrogen and stored at −80°C. Grind the muscle and bone tissue into a powder using liquid nitrogen and collect into an EP tube. Then, add 1 mL of Trizol for RNA extraction. RNA‐seq was performed on an Illumina NovaSeq 6000 platform at Shenggong, Shanghai, China. All samples (n = 3 per time point per age group) were sequenced in a single batch to eliminate technical batch effects. Raw read counts were normalized using DESeq2 to account for library size and composition. Prior to rhythmicity analysis by Metacycle, we filtered out genes with an FPKM of zero at all time points. The peak expression time (acrophase) for each rhythmic gene was mathematically calculated as a continuous variable (0–24 h) by the MetaCycle algorithm. For the phase distribution analysis and visualization (polar histograms), these calculated phase values were grouped into 2‐h bins. To ensure the identified rhythmic genes were not artifacts of low‐expression noise, a sensitivity analysis was performed by evaluating genes with an average FPKM < 1, which confirmed that the age‐related change in rhythmic genes was driven by robustly expressed transcripts. DESeq2 was used to identify DEGs. A p < 0.05 was considered significant. KEGG and Wikipathway enrichment analysis was performed using the Ouyi Cloud platform.

RNA concentration was determined by NanoDrop (Thermo), and cDNA was synthesized using Plus All‐in‐one 1st Strand cDNA Synthesis SuperMix (Novoprotein, E047). Quantitative real‐time PCR was performed using NovoStart SYBR qPCR SuperMix Plus (Novoprotein, E096↗) with LightCycler 480II (Roche). The relative levels of Trap, Ctsk, Bmal1, Hmox1, and Nrf2 were normalized to Gapdh. The q‐PCR primer sequences are as follows.

Trap‐forward: GCGACCATTGTTAGCCACATACG, Trap‐reverse: CGTTGATGTCGCACAGAGGGAT, Ctsk‐forward: AGCAGAACGGAGGCATTGACTC, Ctsk‐reverse: CCCTCTGCATTTAGCTGCCTTTG, Alp‐forward: CCAGAAAGACACCTTGACTGTGG, Alp‐reverse: TCTTGTCCGTGTCGCTCACCAT, Runx2‐forward: CCTGAACTCTGCACCAAGTCCT, Runx2‐reverse: TCATCTGGCTCAGATAGGAGGG, Bglap‐forward: GCAATAAGGTAGTGAACAGACTCC, Bglap‐reverse: CCATAGATGCGTTTGTAGGCGG, Hmox1‐forward: CACTCTGGAGATGACACCTGAG, Hmox1‐reverse: GTGTTCCTCTGTCAGCATCACC, Nrf2‐forward: CAGCATAGAGCAGGACATGGAG, Nrf2‐reverse: GAACAGCGGTAGTATCAGCCAG; Gapdh‐forward: CATCACTGCCACCCAGAAGACTG, Gapdh‐reverse: ATGCCAGTGAGCTTCCCGTTCAG, Il1α‐forward: CGAAGACTACAGTTCTGCCATT, Il1α‐reverse: GACGTTTCAGAGGTTCTCAGAG, Il1β‐forward: TGGACCTTCCAGGATGAGGACA, Il1β‐reverse: GTTCATCTCGGAGCCTGTAGTG.

PGL4.10‐Hmox1 promoter (WT)‐LUC, PGL4.10‐Hmox1 promoter (Mutant)‐LUC, and control PGL4.10 plasmid were provided by HeYuan Biotech, Shanghai. pCGN‐CMV‐Bmal1‐HA plasmid and pRL‐CMV‐Renilla plasmid were provided by Ying Xu's lab at the Cambridge‐Soochow Genomic Research Group, Soochow University. 293 T cells were cultured in DMEM medium containing 10% FBS and 1% PS. When the cell density reached about 70%, plasmid transfection was performed using Hieff Trans Liposomal Transfection Reagents (Yeason, 40802ES). After 48 h, a Dual Luciferase Reporter Assay Kit (Vazyme, DL101) was used for fluorescence signal detection with BioTek Synergy H1.

C2C12 cells were purchased from Procell and cultured in DMEM medium containing 10% FBS and 1% PS. After reaching 80% confluence, the medium was replaced with DMEM differentiation medium containing 2% horse serum and 1% PS, and the medium was changed daily. One week later, clear myotube formation could be observed.

For adenovirus infection, add 300 μL of DMEM growth medium containing 2 × 10⁶ pfu/mL adenovirus to the myotubes in 24‐well plates, and after 6 h, supplement with 200 μL of differentiation medium. Continue culturing in differentiation medium for 4–5 days to allow full adenovirus expression.

Cells were washed once with serum‐free DMEM and subsequently incubated with serum‐free DMEM containing 10 μM 2′,7′‐dichlorodihydrofluorescein diacetate (DCFH‐DA; Elabscience, E‐BC‐K138‐F) at 37°C for 2 h in a humidified 5% CO₂ atmosphere. Following incubation, intracellular reactive oxygen species (ROS) levels were quantified through fluorescence intensity measurement using a microplate reader (excitation/emission: 488/525 nm).

Inflammatory cytokine profiling was performed using the Mouse Inflammation Array GS1 Antibody Chip (Raybiotech, GSM‐INF‐1‐1) according to the manufacturer's protocol, with subsequent data analysis conducted by Shanghai Yujin Biotech Co. Ltd.

Culture supernatants from differentiated myotubes or serum samples collected from experimental mice were analyzed for IL‐1α and IL‐13 concentrations using commercially available enzyme‐linked immunosorbent assay (ELISA) kits (Mouse IL‐1α ELISA Kit, Elabscience, E‐EL‐M0037c; Mouse IL‐13 ELISA Kit, Elabscience, E‐EL‐M0044c; Mouse IL‐1β ELISA Kit, ELK biotech, ELK1271) according to the manufacturer's standardized protocols. All samples were assayed in technical duplicates, and cytokine concentrations were determined against a standard curve generated from recombinant protein standards provided with each kit.

Male C57BL/6 mice (postnatal 1–2 months) were euthanized via cervical dislocation. Following surface sterilization with 75% ethanol, gastrocnemius muscles were dissected, cleared of adipose/connective tissue, and minced into ≈0.1‐mm³ fragments. Tissue fragments were digested in DMEM containing 1 mg/mL each of collagenase types I, II, IV, and dispase (37°C, 60 min). The resultant suspension was filtered through a 70‐μm mesh, centrifuged, washed with DMEM, and cultured in DMEM supplemented with 10% FBS under standard conditions (37°C, 5% CO₂). Cell purity was assessed by α‐SMA immunofluorescence.

Fresh gastrocnemius tissues were fixed in 4% paraformaldehyde for ≥ 24 h. Tissues were then dehydrated sequentially in 15% and 30% sucrose solutions at 4°C until settled. Dehydrated tissues were embedded in OCT compound and rapidly frozen. Cryosections (8–10 μm thick) were cut and collected onto glass slides. Cryosections were air‐dried at room temperature for 10 min and were incubated in Oil Red O working solution for 10 min for sections and washed. Sections were differentiated in 75% ethanol (2 s), rinsed in water (1 min), counterstained with Harris hematoxylin (1 min), differentiated in 1% acid ethanol, blued in ammonia water, rinsed, and mounted with glycerol gelatin. Images were acquired and fibrosis differences were analyzed using ImageJ software.

Paraffin‐embedded muscle tissue sections were deparaffinized and rehydrated through sequential immersion in xylene I (12 min), xylene II (12 min), absolute ethanol I (6 min), 95% ethanol (6 min), and 85% ethanol (6 min), followed by a 2‐min rinse in tap water. Sections were then stained in saturated picric acid Sirius red solution for 8 min and rinsed briefly in absolute ethanol. After air‐drying in a 60°C oven, sections were cleared in xylene for 5 min and mounted with neutral balsam. Images were acquired, and fibrosis differences were analyzed using ImageJ software.

The plasmids pCMV‐T7‐MCS‐3xFlag (MiaoLing, P39495↗), pCMV‐Bmal1(mouse)‐3xMyc (MiaoLing, P85848↗), and pCMV‐Nrf2‐3xFlag (MiaoLing, P92972) were obtained from Miaoling plasmid platform. Forty‐eight hours posttransfection in 6‐well plates, cells were lysed using RIPA buffer (Epizyme, PC103), and protein supernatants were collected. The supernatants were incubated with Myc primary antibody (Proteintech, 60,003–2‐Ig) at a 1:100 dilution overnight at 4°C, followed by the addition of 25 μL of Protein A/G beads (MCE, HY‐K0202) for an additional 2 h. The beads were washed five times with lysis buffer, resuspended in loading buffer, and denatured at 95°C for 5 min prior to Western blotting. The expression of BMAL1, NRF2, and GAPDH was detected using anti‐Flag (Sigma F1804, 1:1000), anti‐Myc (CST 2278S, 1:1000), and anti‐GAPDH (Proteintech 60,004–1‐Ig, 1:3000) antibodies, respectively.

Human and mouse skeletal muscle single‐cell and single‐nucleus RNA sequencing data were obtained from the publicly available Muscle Aging Cell Atlas (Kedlian et al. 2024) by directly downloading the published h5ad objects, which were imported into R and converted to SingleCellExperiment objects for subsequent analyses. All analyses were performed using the processed data provided by the original authors, including the normalized expression matrix, precomputed dimensionality reduction embeddings, and cell type annotations, with human cell types classified according to annotation_level2 to preserve fine‐grained resolution and mouse cell types retained in their original form without any reclustering or reannotation. Age group definitions (Young vs. Old) were adopted strictly from the original metadata fields (Age_bin or Age_group), and no reclassification or merging of age intervals was applied. UMAP visualizations were generated directly from precalculated coordinates and colored by either cell type annotation or age group. Gene expression values were extracted from the normalized expression matrix. To evaluate age‐related differential expression within each cell type, a two‐sided Wilcoxon rank‐sum test was performed between the young and old groups, and p‐values were adjusted using the Benjamini–Hochberg method to control the false discovery rate (FDR). For specific target genes, volcano plots were constructed with log2 fold change on the x‐axis and –log10(FDR) on the y‐axis. In these plots, each point represents one cell type; point size reflects the proportion of cells with detectable expression (expression > 0) within that cell type, and point color denotes the mean expression level, with specific visualization thresholds detailed in the corresponding figure legends. Additionally, an independent murine single‐cell RNA sequencing dataset used for supplemental validation was obtained from the Aging Atlas Consortium (2021).

Mouse gastrocnemius and quadriceps muscles were isolated and fixed in GD fixative (Haoke, HK2013) at room temperature for 24 h. The fixed tissues were then embedded in paraffin and cut into 4‐μm‐thick sections. Following hematoxylin and eosin (H&E) staining, the sections were imaged using a digital whole‐slide pathology scanner (Jiangfeng, KF‐FL‐020). The cross‐sectional area (CSA) of the muscle fibers was subsequently quantified using ImageJ software.

Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA, USA). For comparisons between two groups, two‐tailed Student's t‐tests were applied, while two‐way analysis of variance (ANOVA) was employed for analyzing differences between curves. The threshold for statistical significance was set at p < 0.05.

The data that support the findings of this study are openly available in NCBI SRA at https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1164196↗, reference number PRJNA1164196.