Targeted Bmal1 restoration in muscle prolongs lifespan with systemic health effects in aging model

We systemically infected-KO pups at postnatal day 5 with an AAV9-vector that expresses hemagglutinin-tagged (HA-tagged)cDNA under the control of the constitutive muscle-specific promoter, CK6 (). We first measured the protein levels of BMAL1 in the skeletal muscle of 10-week-old-KO mice using antibodies against HA tag and BMAL1 (). We detected BMAL1-HA expression in gastrocnemius, tibialis anterior (TA), extensor digitorum longus (EDL), and quadriceps muscles whereas HA expression in soleus or diaphragm was undetectable, consistent with previous reports using the CK6 promoter (,) (; supplemental material available online with this article;). We did not detect expression in the heart or liver, consistent with previous reports (,,) (). We also tested lung and stomach samples, 2 tissues rich in smooth muscle cells, and we did not detect expression of HA (). Using the BMAL1 antibody, we compared the level of expression of the transgene in KO+AAV with WT control muscle at a peak time forexpression (Zeitgeber time 1, ZT1). We found that in our mice we obtained on average approximately 25% of the expression levels found in WT muscles (range 16% to 34%). The expression of BMAL1 protein was also evident by immunohistochemical analysis of TA cryosections (). As expected, the expression ofin the KO+AAV muscle sections was detected in the muscle fiber nuclei within dystrophin staining, similar to WT muscle. Last, we compared the viral genome copy numbers in different muscle samples at 10 weeks of age, approximately 9 weeks after systemic infection, by real-time quantitative PCR (qPCR) (). We found vector copies in each muscle tested. The TA, gastrocnemius, and quadriceps muscles had approximately 20,000 viral copies/μg genomic DNA, and the viral levels in the soleus were about 20 times lower. Liver and heart tissue had more copies than skeletal muscles, 28- and 6-fold compared with TA, respectively, but no protein was detected because the CK6 promoter is not active in those tissues (). These results validate the use of systemic delivery of AAV-CK6--HA as a model for skeletal muscle–specificexpression in the-KO mouse. Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Figure 1A Figure 1B ^{17 18} Supplemental Figure 1A ^{17 19 20} Figure 1B Supplemental Figure 1A Figure 1C Supplemental Figure 1B Figure 1B https://doi.org/10.1172/jci.insight.174007DS1↗

We next asked whether the constitutive expression of BMAL1 from the CK6 promoter after birth is sufficient to engage the expression of core circadian clock factors as well as known clock output genes in muscle. Gastrocnemius muscles from 10-week-old WT,-KO, and-KO+AAV mice were collected at ZT1. We analyzed a total of 14 core clock factors and well-defined muscle-specific clock output genes by qPCR (). For the core clock genes, we found thattransduction after birth was sufficient to reengage expression of most, but not all, of the core clock genes in the direction of WT muscle. This included upregulation of,,,, andwith downregulation of,, andwhen compared with the expression in-KO. Analysis of well-defined clock output genes showed that the expression levels of,, andwas upregulated whilewas downregulated relative to-KO muscle. Thus, except for negative limb components (i.e.,and), rescuing the expression ofin the skeletal muscle of-KO mice was sufficient to engage the expression of core clock components and of well-defined clock output genes similar to what is seen in WT muscle. Bmal1 Bmal1 Bmal1 Per3 Per2 Rora Nr1d1 Nr1d2 Cry1 Clock Npas2 Bmal1 Dbp Tcap Mylk4 Myod1 Bmal1 Per1 Cry2 Bmal1 Bmal1 Figure 1, D and E

To address if the expression of AAV-in the muscle of-KO muscle was sufficient to induce a synchronized pattern of circadian clock gene expression, we compared gene expression at 2 time points, ZT1 and ZT13. At these time points, if the clocks were synchronized across all the cells in the muscle tissue, we would see time of day differences in the expression of,,,,, andmRNAs (). However, we did not see any notable time of day difference in core clock and clock output gene expression in the muscles of the KO+AAV mice, as seen in the WT muscle (). Bmal1 Bmal1 Per1 Per2 Per3 Cry2 Dbp Nr1d2 Supplemental Figure 1C Supplemental Figure 1C

We next performed targeted ChIP-qPCR to determine if BMAL1 binds to known clock gene promoters in the KO+AAV muscle. As shown in, we detected considerable enrichment of BMAL1 binding to the E-box–containing regions within the promoter of the core clock gene, the clock output gene, and 2 muscle-specific clock output genes,and, that we have shown previously to be bound by BMAL1 () (). Overall, these results demonstrate that the reexpression ofin the-KO muscle results in BMAL1 binding to known promoter elements and in changes in clock and clock output gene expression that are similar to WT muscle. However, the patterns of gene expression are not synchronized across the muscle, indicating that generation of a synchronized clock and clock output gene expression requires external cues, including information from an intact central clock in the suprachiasmatic nucleus. Supplemental Figure 1D Supplemental Figure 1D ²¹ Per2 Dbp Tcap Mylk4 Bmal1 Bmal1

-KO mice have a median lifespan of approximately 37 weeks and are considered a model of premature aging (). We investigated whether skeletal muscle–specificrescue could increase their lifespan. We generated cohorts of male WT (= 45),-KO (= 43), and-KO+AAV (= 36) mice and maintained them out to 40 weeks of age past the median lifespan of the-KO in our facility (38.3 weeks). We found that the muscle-specific-KO+AAV mice had an extended lifespan, with 88.9% (32/36) of them surviving through 40 weeks. In contrast, only 44.2% (19/43) of-KO mice survived over the same period (). Using log-rank tests, we found that there was a statistically significant difference in survival between the-KO and-KO+AAV groups (value < 0.0057). As expected, all WT mice survived (45/45; 100%), and there was a notable difference in survival between the WT mice and both the-KO and-KO+AAV mice, indicating that the AAV-mediated rescue of skeletal muscleimproves the lifespan of the-KO mouse but that this is not sufficient to restore it completely to the WT status (). Bmal1 Bmal1 n Bmal1 n Bmal1 n Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 P Bmal1 Bmal1 Bmal1 Bmal1 ⁹ Figure 2A Figure 2A

One of the known phenotypes of the-KO mice is that they exhibit limited increases in body weight after 10 weeks of age (,). We tracked body weight weekly across all mice over 40 weeks (). We found that at 10 weeks of age, there is no difference in the body weights of the WT,-KO, and-KO+AAV mice (). However, consistent with the literature, we observed that the-KO mice stopped gaining weight at 12 weeks of age (,). By approximately 30 weeks of age, the body weights of the-KO as well as the KO+AAV mice were considerably lower than WT mice and were not different from each other. Thus, reexpressingin skeletal muscle after birth was not sufficient to rescue the lack of postnatal growth that occurs in the-KO mice (). Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 ^{9 22} Supplemental Figure 2A Figure 2B ^{9 22} Supplemental Figure 2A

Another phenotype of the-KO mouse is that the distribution of lean and fat mass is different compared with the WT mice (). We measured body composition from 10 to 40 weeks of age across the 3 groups, every 3 weeks (, B and C). We found that at 10 weeks,-KO and KO+AAV mice had twice the amount of fat mass in grams (~109% higher) and ~17% lower lean mass than WT mice, while the body weights were the same across the 3 groups (). By ~30 weeks of age, when the KO and KO+AAV mice have lower body weights, the grams of fat mass declined dramatically to be ~60% lower in the KO and KO+AAV while lean mass remained lower by ~17% in WT mice (). Consistent with body weight, the expression ofin the muscle of the-KO mice was not sufficient to alter the trajectory of body composition changes with age in the-KO mouse. Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 ²² Supplemental Figure 2 Figure 2C Figure 2C

We performed circadian and wheel activity phenotyping in these mice at 13 weeks of age. We kept the mice under normal 12-hour light/12-hour dark (LD) conditions for 2 weeks, followed by 2 weeks under constant conditions (total darkness, DD; the representative actograms from DD data are provided in). The WT mice exhibited clear rhythmic behavior under DD, with a normal period length of 23.74 hours (). In contrast, both the-KO and-KO+AAV exhibited arrhythmic behavior in constant darkness, as expected (and). We note that the-KO and KO+AAV mice exhibited rhythmic cage activity behavior under LD conditions, and this has been previously shown () (). We then analyzed the amount of voluntary wheel activity for the 3 different groups of mice for 7 consecutive days under LD conditions (= 12–15 mice/group). We found that, consistent with McDearmon et al. (), the muscle-specific expression ofwas sufficient to markedly increase, by ~2.8-fold, the voluntary wheel activity when compared with the-KO mice (). We also tested motor coordination and balance in a second cohort of 10-week-old mice using the rotarod test. With this assay, we found that the muscle-specificexpression improved balance, resulting in a longer time to fall (latency to fall) as well as a higher speed at fall compared with the-KO mice (). For both wheel running and rotarod test measurements, the AAV+KO mice performed better than the-KO mice but were not fully restored to WT levels. These outcomes verify that the muscle-specific expression ofin the KO mouse is sufficient to improve voluntary physical activity and neuromuscular parameters of mobility. Figure 2D ²³ Figure 2D Supplemental Figure 2D Supplemental Figure 2E ^{24 12} Figure 2E Figure 2F Bmal1 Bmal1 Bmal1 n Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1

We performed glucose tolerance tests on a third cohort of 10-week-old mice at ZT4 (4 hours after lights on), as this is a metabolic parameter known to be disrupted in the-KO mouse (,). Consistent with previous findings,-KO mice showed higher glucose peaks and slower clearance after an i.p. glucose injection (). Muscle-specificrescue reduced both the peak glucose rise and the time to return to baseline, resulting in a markedly lower area under the curve compared with-KO mice, comparable to WT levels (). This suggests thatreexpression in skeletal muscle can restore glucose tolerance even when other tissue clocks are dysfunctional. We acknowledge there are temporal variations in the glucose tolerance test outcomes, as other groups have shown the importance of the liver and other tissues for systemic glucose homeostasis across the day (). Despite not rescuing rhythmic behavior, the skeletal muscle–specific Bmal1 expression improved lifespan, cage activity, neuromuscular coordination, and systemic glucose metabolism. Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 ^{10 25 25} Figure 2G ²⁶

We next characterized the skeletal muscle phenotype from a cohort of 10-week-old mice to determine the cell- and tissue-specific effects ofin skeletal muscle. Consistent with our body weight and body composition analyses, the muscle weights from-KO and-KO+AAV were lower than WT and not different from each other (). Histological analysis of fiber size and fiber type () revealed no effects of AAV-treatment on fiber size distribution or average fiber size as visualized by similar frequency distributions of both-KO and KO+AAV mice compared with WT muscle. Fiber type analysis determined thatexpression also did not result in a shift in the overall myosin heavy chain (MHC) isoform distribution (), as both KO and KO+AAV muscle exhibited similar fiber type distribution characterized by an increased number of MHCIIB-expressing fibers and a decrease in the number of IIX fibers, when compared with WT mice (). Deeper analysis of fiber area per fiber type showed that reexpression ofwas sufficient to result in a small but notable increase in the average CSA of type IIA and type IIB fibers compared with KO, with no effect on type IIX fibers (). Thus, AAV reexpression ofin the-KO mouse had no effect on muscle weight or fiber CSA, with only minor effects on fiber type, indicating that these muscle parameters do not contribute to the increased longevity of the KO+AAV mice. Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Figure 3A Figure 3, B–E Figure 3D Figure 3D Supplemental Figure 3

Subjective viewing of the H&E stains of the-KO muscle noted that the cytoplasm was, in general, paler in color compared with the WT muscle. We analyzed total protein content/muscle mass, as μg protein/mg muscle tissue, using a buffer system to solubilize all proteins in muscle from the 3 groups. Consistent with well-established measures of protein in mammalian skeletal muscle (), we found that the protein content/mass was approximately 20.4% in WT muscle. Protein content/mass was markedly decreased in the-KO muscle to approximately 16.5%, and reexpression ofin the muscle of the-KO mouse was sufficient to restore protein content/mass to WT levels (). The restoration of protein concentration per muscle mass is important as it is estimated that over 80% of total protein in skeletal muscle is myofibrillar (e.g., myosin, titin). Thus, changes in total protein/mass reflect changes in many proteins that contribute to muscle mechanical function. We previously reported that skeletal muscle from-KO exhibits reduced maximum isometric force normalized to physiological cross section, known as specific force (). In this study, we performed ex vivo muscle mechanics on the EDL muscles of WT,-KO, and-KO+AAV mice at 10 weeks of age and found that the specific force of EDL muscles from-KO mice was markedly reduced but that reexpression ofin the KO mice was sufficient to increase the specific force back to WT levels (). These results suggest that the improved muscle protein concentration in the AAV-KO muscle was sufficient to support the rescue of muscle-specific force, and this functional improvement likely contributes to increased cage activity behavior and to improved neuromuscular parameters of mobility. Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 ²⁷ Figure 3F ²⁸ Figure 3G

We conducted transcriptomics and proteomics on gastrocnemius muscles of 10-week-old WT,-KO, and-KO+AAV mice to identify potential molecular features associated with increased lifespan. At this age, body weights of-KO and KO+AAV mice were similar to WT mice. RNA sequencing revealed 2,408, 2,210, and 846 differentially expressed genes (DEGs) in WT versus-KO, WT versus KO+AAV, and KO+AAV versus KO groups, respectively (< 0.01) (and). The largest differences were between WT and both KO groups, with modest differences between-KO versus KO+AAV. We focused on the 846 DEGs between KO+AAV versus KO, of which 574 genes were up- and 272 were downregulated, as presented in the volcano plot inB. Notably, core clock genesand, along with,, and the muscle-specific carbonic anhydrase, were upregulated by AAV-, while the muscle-enriched transcription factorwas markedly downregulated, suggesting reductions in the Hippo signaling pathway (). Before pathway analysis, we examined gastrocnemius from 10-week-old-KO mice and-KO mice systemically infected at postnatal day 5 with AAV9-CK6-GFP as a control for AAV transduction. We identified 66 overlapping DEGs between KO versus KO+GFP and KO versus KO+AAV (Bmal1) that were enriched for immune response genes and removed them from any further analysis (). Bmal1 Bmal1 Bmal1 Bmal1 P Bmal1 Per3 Nr1d2 Aebp1 Hlf Car3 Bmal1 Tead4 Bmal1 Bmal1 Figure 4A Supplemental Data 1 Figure 4 Figure 4B Supplemental Figure 4A

Gene Ontology (GO) enrichment analyses were performed on DEGs to investigate biological processes altered in response toexpression in the-KO muscle: a total of 13 biological functions were found to be significantly enriched (value < 0.05) (and). Upregulated processes included "Cell adhesion" with genes that contribute to cell surface adhesion and signaling, such asand(). Two of the upregulated pathways in KO+AAV muscles were amino acid transport (e.g.,or,, and) and carbohydrate metabolism (e.g.,and) (). In the downregulated processes, we found that fatty acid metabolism was enriched in the-KO+AAV, with genes such as, phospholipase A2 group VII, which regulates phospholipid catabolism during inflammatory and oxidative stress responses (). As seen in, fatty acid metabolism genes were highly expressed in muscles of-KO compared with WT muscles. These results suggest that AAV-regulated the expression of genes related to substrate metabolism in muscle, increasing the expression of genes related to carbohydrate and amino acid metabolism while decreasing the expression of those involved in lipid metabolism. Bmal1 Bmal1 P Itga4 Cdh4 Slc7a5 Lat1 Slc6a9 Slc1a3 Fbp2 Stbd1 Bmal1 Pla2g7 Bmal1 Bmal1 Figure 4C Supplemental Data 1 Figure 4D Figure 4D Figure 4D Figure 4D

To evaluate the extent of the AAV-mediated rescue ofacross different skeletal muscles, we measured the expression of some of the DEGs identified from the gastrocnemius muscle in TA, soleus, and quadriceps muscles from our model. We found that the gene expression changes in the TA and quadriceps muscles are similar to those seen in the gastrocnemius. In contrast, there was very little rescue of gene expression in the soleus muscles of the KO+AAV mice (), consistent with the lack of AAV-CK6–driven expression of thetransgene in this muscle (). To further evaluate the effects of the AAV-treatment in muscles, we used changes in gene expression with loss of musclein the TA and soleus muscles reported by Dyar et al. () and asked if these clock output genes were modified in our AAV-rescue model. We selected genes that lost circadian gene expression after knocking outin the TA or the soleus muscle and that were differentially expressed at ZT0, close to our collection time ZT1. We found that the gene expression levels found in the TA identified genes in KO+AAV that responded toreexpression and that were rescued to WT levels, while the expression analysis in the soleus identified genes that were not different in KO versus KO+AAV (). These results verify that the AAV-transduction in the-KO mouse conferred changes in core clock and, importantly, in clock output gene expression across several different limb muscles. The lack of gene expression changes in the soleus muscle of our model is consistent with our results showing no detectable BMAL1-HA levels in the soleus and data from the literature showing limited transgenic expression driven by the CK6 promoter in the mouse soleus muscles. Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Supplemental Figure 4C Supplemental Figure 1A ⁷ Supplemental Figure 4D

We next hypothesized that the differences in the-KO+AAV gene expression program would exhibit a set of genes that would be more similar to WT than KO muscle. Our first analysis was on the uniquely shared DEGs between the muscles of WT versus KO and WT versus KO+AAV. Out of 1,222 genes, we found that 1,100 DEGs were changing in the same direction in KO and KO+AAV when compared with WT, indicating that the expression of these genes was not rescued by AAV treatment (Venn diagram,). Consistent with our hypothesis, we identified 258 DEGs that were different between WT and KO muscle, as well as between KO versus KO+AAV, but were not DEGs in the WT versus KO+AAV comparison (Venn diagram,). Inwe plotted the logfold-change for each of the 258 genes, and we verified that the changes in the KO+AAV versus KO were similar in both magnitude and direction with those in the WT versus KO and exhibited a strong fold-change correlation (Pearson's= 0.9796,< 0.0001) (). Functional analysis of these genes identified biological processes related to lipid metabolism that were enriched in the downregulated genes of the KO+AAV/WT groups, while biological rhythms, and cell adhesion were biological processes enriched in the upregulated genes (). Mitochondrion-related genes were downregulated in the KO+AAV in the same direction as the WT muscles (). Commonly upregulated genes included perinuclear region of cytoplasm, such as,,,, and, which encode proteins located around the nucleus. Thus, while the impact of restoringon gene expression in muscle is modest, there is a considerable cluster of both up- and downregulated genes that are restored to levels similar to WT. Bmal1 r P Ntm Cask Mlf1 Ctif Stx4a Bmal1 Figure 4E Figure 4E Figure 4F Figure 4F Figure 4G Figure 4H 2

We used a tandem mass tag–based (TMT-based) quantitative proteomics approach with gastrocnemius samples (= 3 samples/group) to define the protein changes in muscle with AAV treatment. Consistent with the RNA-Seq analysis, the proteomics data determined that the muscle of-KO+AAV is more like-KO muscle than WT. We identified differential expression of 699, 840, and 182 unique proteins in the WT versus-KO, WT versus KO+AAV, and KO+AAV versus KO groups, respectively (). We focused on the 182 unique differentially expressed proteins; 169 proteins were upregulated in the KO+AAV group compared with the-KO muscle, while only 13 proteins were downregulated as visualized in the volcano plot inJ. Initial inspection of the protein data identified that the upregulation of CAR3 and downregulation of PLA2G7 were consistent with the changes in the RNA-Seq data discussed above. We performed GO enrichment analysis to cluster the biological process and cellular components of the upregulated proteins (). The results indicated that carbohydrate metabolism (TBC1D1, PDK4, and HK1) and amino acid metabolism (urea cycle) pathways were upregulated, similar to the RNA-Seq results. The most considerable enrichment category for a cellular compartment was mitochondrion (). Examples of upregulated proteins located in the mitochondria are visualized in the heatmap in, which includes mitochondrial translation-related proteins (e.g., MRPL4, MRPL3), respiratory chain complex proteins (MT-CO1, MT-CO3), metabolism of carbohydrates (HK1 and PDK4), and metabolism of amino acids (e.g., MTHFD1L, ASS1, and MMP2). As noted above, mitochondrion was the most considerably enriched cellular compartment category for the downregulated DEGs from the KO+AAV RNA-Seq analysis (). Inspection of the RNA-Seq and proteomic gene lists found very few overlaps except for HIP1R, which was downregulated in both datasets, and NT5C and MTHFD1L, which were upregulated in both data sets. The rest of the mitochondrial genes/proteins were unique to each analysis. Therefore, our enrichment analysis of the proteomics results indicates that restoration of carbohydrate and amino acid metabolism and downregulation of fat metabolism are the primary outcomes of the AAV rescue ofin muscle at 10 weeks of age. n Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Figure 4I Figure 4 Figure 4K Figure 4K Figure 4L Figure 4H

We next defined the changes in gene expression in the muscle of mice at 40 weeks of age, just beyond the median lifespan of the-KO mice and at a time when both groups of mice are lower in body weight and have lost considerable fat mass compared with WT mice. The initial analysis determined that circadian clock gene expression was still engaged in the muscle at 40 weeks, consistent with the continued presence of the-HA AAV. We identified that the changes in gene expression of the muscle of the KO compared with WT were even greater at 40 weeks of age (3,920 at 40 weeks vs. 2,408 at 10 weeks). However, the number of DEGs for WT vs. KO+AAV at this age was 2,658, which is only slightly higher than the 2,210 DEGs detected at 10 weeks, suggesting fewer changes in the KO+AAV muscle. Consistent with the 10-week analysis, the differences between the KO muscle versus KO+AAV were relatively small with a total of 600 DEGs (value < 0.01) (and). The volcano plot for the KO+AAV versus KO DEGs inindicated that the same numbers of genes were upregulated (= 300) and downregulated (= 300). Similar to our analysis of the 10-week-old mouse data, GO enrichment analysis revealed that mitochondria-related pathways were decreased in the KO+AAV muscles, and these include electron transport, tricarboxylic acid cycle, respiratory chain, and ATP synthesis (). A heatmap of genes from each of those categories is provided inD. In general, continuedreexpression in the-KO muscle functioned to bring these clusters of genes back to levels more similar to that seen in the WT muscle. Bmal1 Bmal1 P n n Bmal1 Bmal1 Figure 5A Supplemental Data 1 Figure 5B Figure 5C Figure 5

Overlapping analysis revealed that 1,959 DEGs were shared and changing in the same direction between WT versus KO and WT versus KO+AAV, indicating these genes were not rescued by AAV-(). Consistent with our previous analysis, we identified 263 DEGs that were different between WT versus KO muscle, as well as between KO versus KO+AAV, but were not DEGs in the WT versus KO+AAV comparison (). These genes shared the same differential expression compared to KO, indicating these genes were rescued by AAV-(Pearson's= 0.9768,< 0.0001) (). Functional analysis of these genes identified that biological processes related to mitochondrion-related processes were enriched in the downregulated genes of the KO+AAV/WT groups, while biological rhythms and apoptosis are examples of biological processes enriched in the upregulated genes (). Mitochondrion-related genes were downregulated in the KO+AAV in the same direction as the WT muscles (), as previously shown inD. Several studies have shown that loss of muscleleads to impaired metabolic flexibility with a diminished ability to utilize glucose (,). Our findings demonstrate that the use of AAV to restorein the muscle of the global-KO mouse supports improved glucose utilization and metabolic flexibility and rescues mitochondrial gene expression to WT levels. Bmal1 Bmal1 r P Bmal1 Bmal1 Bmal1 Figure 5E Figure 5E Figure 5F Figure 5G Figure 5H Figure 5 ^{6 7}

We next analyzed the changes in gene expression that occur with aging across each genotype by comparing gene expression from 10 to 40 weeks within each genotype to identify potential genes/pathways associated with increased longevity of the-KO+AAV (). It was important to note that muscle gene expression did not change markedly in the WT mice, as only 226 genes were differentially expressed at 40 versus 10 weeks (value < 0.01). In contrast, the muscle of-KO and KO+AAV mice showed larger changes in gene expression likely reflecting the contributions of systemic factors to changes in muscle gene expression. The muscle of the-KO mice exhibited 1,500 DEGs changing over time, while 1,407 DEGs were changing in the KO+AAVs (). We determined that approximately 35% of the DEGs with aging were in common between the KO and KO+AAV datasets, indicating that age-associated changes in gene expression were not largely shared (). We looked at the top aging biological processes enriched within the KO+AAV, and we found that inflammation-related biological processes were downregulated in the muscles with age (). In contrast, the biological processes associated with aging-related pathways, such as DNA damage response or antioxidant response (cellular response to hydrogen peroxide), were upregulated in KO muscle (). We looked at the GO biological processes that were shared across both groups but were regulated in different directions (and). We identified 22 biological processes downregulated with age in the KO+AAV that were upregulated in the KO (e.g., Tumor necrosis factor production and Regulation of NF-kB transcription factor activity) and 4 biological processes linked to chromatin regulation and transcription that were upregulated in KO+AAV and downregulated in KO. The most considerably enriched category was positive response to TNF production, an important pro-inflammatory pathway (examples of genes with changes over time in expression are shown in). Analysis of the gene expression changes with age highlights that the KO+AAV muscle was able to maintain metabolic pathways and that this was associated with reduced indicators of inflammation compared with the muscle of the aging-KO mice. Bmal1 P Bmal1 Bmal1 Bmal1 Figure 5I Figure 5I Supplemental Figure 5A Figure 5J Figure 5K Supplemental Figure 5B Supplemental Data 1 Supplemental Figure 5B

To explore the systemic effects of the muscle-specific AAV rescue model, we performed RNA-sequencing of lung, heart, liver, and white adipose samples from 40-week-old-KO and KO+AAV mice (= 4 samples/group). Analysis of DEGs found that the lung exhibited the greatest effect with a total of 1,050 DEGs, followed by white fat, liver, and heart, showing 265, 167, and 64 DEGs, respectively (). DEGs showed tissue-specific effects of muscle rescue since very little overlap was found across all tissues (). We performed GO enrichment analysis for biological processes for each tissue dataset to address if any common pathways changed in these peripheral tissues of the-restored mice and found very little overlap across tissues. However, we noted that the general categories of inflammation and substrate metabolism–related pathways were consistent in each of the tissues and that the stress response pathways were commonly regulated in the lung, liver, and heart tissues (). These findings support a model in which the rescue of more metabolic flexibility in the muscle of the-KO+AAV is associated with the restoration of systemic metabolism and a concomitant reduction of systemic inflammation in the-KO tissues. Bmal1 n Bmal1 Bmal1 Bmal1 Figure 6A Figure 6B Figure 6C

To better understand the effects of the muscle-specificexpression on metabolic health, we performed untargeted metabolomics in plasma samples of both groups of 40-week-old mice (= 4 samples/group). Plasma samples were collected at ZT1, the same time point as our RNA-Seq analyses. Following data extraction and filtering, 4,129 metabolite peaks were detected, and notable differences in the metabolite profiles were evident, with distinct separation of the groups (). In KO+AAV versus KO plasma samples, 435 metabolites increased and 224 decreased (< 0.05) (). Mummichog analysis identified 551 empirical compounds linked to 399 Kyoto Encyclopedia of Genes and Genomes (KEGG) IDs. Bmal1 n P Figure 6D Figure 6E

We used the KEGGreference map for pathway enrichment, which identified alterations in amino acid metabolism, TCA cycle, butanoate metabolism, and glycolysis (). In addition, main class enrichment analysis identified that amino acids were the largest cluster of metabolites changing in the plasma after muscle-specific rescue of, followed by TCA acids (). Consistent with our model, plasma glucose was decreased in the-KO+AAV mice (), and TCA intermediaries were increased in the plasma samples of-KO+AAV compared with-KO (). The plasma levels of amino acids alanine, aspartate, glutamate, and lysine that can be metabolized to support the TCA cycle were increased in the-KO+AAV (), suggesting the changes in carbohydrate metabolism in muscle support TCA intermediaries and likely decrease the use of plasma/systemic supplies of key amino acids. We note that these amino acids were rescued to levels closer to those seen in the WT mice (). The 3 aromatic amino acids,-phenylalanine,-tyrosine, and-tryptophan, were found to be decreased in the plasma samples of the-KO+AAV, and this would be consistent with observations that they are associated with inflammatory status of multiple diseases (–) (). These results support our working model that rescuing the expression ofin the muscle of the-KO affects the carbohydrate metabolism and amino acid needs to maintain the glycolysis-TCA cycle flux in the muscle of-KO. Since muscle accounts for a large proportion of body mass, these metabolic effects across skeletal muscles have a broad metabolic and subsequent systemic immune impact across the organism that is associated with improvements in lifespan. Mus musculus Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Bmal1 Figure 6F Figure 6G Supplemental Figure 6A Supplemental Figure 6B Figure 6H Supplemental Figure 6C ^{29 31} Figure 6I l l l

Our study focused exclusively on male mice, so it is unclear whether the findings are fully applicable to female mice; further studies would be required to assess potential sex-specific differences.

MousecDNA was subcloned into pAAV-CK6 with a C-terminus HA tag and confirmed by Sanger sequencing. AAV serotype 9 was produced using transient transfection of HEK293T cells (ATCC, CRL-3216) and CsCl sedimentation by the University of Massachusetts Medical School Viral Vector Core as previously described (). Vector preparations were determined by droplet digital PCR, and purity was assessed by SDS-PAGE and silver staining. Bmal1 ⁴⁷

Heterozygous-KO mice (strain 009100) were obtained from The Jackson Laboratory. Mice were housed at the University of Florida animal care servicesfacility on a 12-hour light/12-hour dark cycle with free access to food and water.-KO mice were bred from heterozygous parents, with genotype and sex determined as previously described (,). Neonatal mice were genotyped on day 3, and on day 5, male homozygotes were cryoanesthetized and injected subxiphoidally with 20 μL containing 2 × 10genome copies of AAV as previously described (). Pups were rewarmed and returned to their cages. WT controls were littermates. Bmal1 Bmal1 ^{24 48 49} 11

Mice were monitored daily for signs of illness and deaths were recorded. Animals that were close to dying were euthanized using COasphyxiation and recorded. The decision to euthanize them was made by a veterinarian according to Association for Assessment and Accreditation of Laboratory Animals guidelines. Animals that were euthanized were considered censored deaths. 2

Proteins were isolated from gastrocnemius, diaphragm, TA, quadriceps, soleus, stomach, heart, lung, and liver tissues using RIPA buffer. For HA detection, 10 μg of protein was loaded per lane and 50 μg for BMAL1. Membranes were stained with Rev700 to assess total protein. Primary antibodies against HA (1:2,000; Roche 11867423001) or BMAL1 (1:1,000; Cell Signaling Technology 14020) were incubated overnight. Blots were probed with HRP-conjugated anti-rat (1:5,000; Invitrogen 31470) for HA or anti-rabbit (1:10,000; Invitrogen A24531) for BMAL1. Detection was done with Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific) and imaged using a Chemidoc system (Bio-Rad).

One whole frozen quadriceps was minced, then sonicated (Omni GLH homogenizer) in buffer (10 mM of HEPES, pH 7.5, 10 mM MgCl, 60 mM KCl, 300 mM sucrose, 0.1 mM EDTA, pH 8.0, 0.1% Triton X-100, 1 mM DTT, 1 mM PMSF, and protease inhibitors) containing 1% formaldehyde. After 15 minutes, crosslinking was stopped with glycine, and tissue was centrifuged 300for 5 minutes at 4°C. The pellet was resuspended in 0.5× RIPA buffer with protease inhibitors, 1 mM PMSF, and 1 mM DTT. Nuclei were released via sonication, and 10 μL of genomic DNA was taken as input. Samples were incubated overnight with anti-BMAL1 antibody (3 μg, MilliporeSigma, SAB4300614) and Dynabeads protein A/G for 4 hours at 4°C. Beads were washed with RIPA and Tris-EDTA buffers. DNA/antibody complexes were eluted, then incubated with proteinase K. DNA was purified using the QIAGEN PCR kit and resuspended in TE buffer. qPCR was performed on 2 μL of DNA, with enrichment normalized to input DNA. ChIP experiments were performed with independent chromatin preparations, and qPCRs were done in duplicate. Primer details are in. 2 g Supplemental Table 1

Mice were individually housed on a 12-hour light/12-hour dark schedule, and wheel-running activity was recorded for 2 weeks, followed by 2 weeks in constant darkness using the ClockLab system (Actimetrics). Wheel-running activity from 7 consecutive days per mouse was analyzed using ClockLab software for statistical analysis.

Motor coordination was assessed with the rotarod test (Rotamex 5, Columbus Instruments) with rod acceleration from 4 to 40 rpm over 5 minutes. Three trials were conducted on the testing day, with 5-minute rests in between. The average of the last 3 trials was used to determine the final latency to fall and speed scores.

Body weight and body composition were recorded at 3-week intervals starting at 10 weeks of age in-KO (= 12–26), KO+AAV (= 14–21), and WT (= 19–23) mice. Body composition was assessed by nuclear magnetic resonance using an EchoMRI whole body composition analyzer (EchoMedical Systems). Bmal1 n n n

Total RNA was isolated from gastrocnemius, heart, liver, lung, and white fat tissues using TRIzol reagent (Invitrogen) and QIAGEN RNeasy Mini Kit. Frozen tissues were homogenized in TRIzol with 1.0 mm zirconium oxide beads using a Bullet Blender at 4°C. After a 5-minute incubation, homogenates were mixed with chloroform and centrifuged at 8,000for 15 minutes at 4°C. The aqueous phase was mixed with isopropanol, transferred to an RNeasy Mini spin column, and treated with RNase-free DNase (QIAGEN). RNA was eluted in nuclease-free water. g

A total of 1,000 ng of total RNA from gastrocnemius, soleus, TA, and quadriceps was reverse-transcribed using SuperScript VILO Master Mix (Invitrogen). A total of 20 ng of cDNA was used as template. Gene expression was measured in duplicate with Fast SYBR Green Master Mix (Thermo Fisher Scientific) and 400 nM primers in a 20 μL reaction volume on a QuantiStudio 3 thermocycler (Applied Biosystems). Relative expression was calculated using the 2method andas reference gene. Primer sequences are listed in. ΔΔCT Gapdh Supplemental Table 2

Purified RNA samples had an RNA integrity number above 8.0 and were sequenced at the University of Florida's Interdisciplinary Center for Biotechnology Research (ICBR) and Novogene (UC Davis). Poly(A)-selected RNA-Seq libraries were prepared with 250 ng of RNA using the Illumina mRNA Prep kit. Libraries from heart, lung, white fat, and liver, and 4 replicates of 10-week-old gastrocnemius were sequenced on ICBR's Illumina NovaSeq 6000 (2 × 150 bp) to a minimum of 40,000 reads per sample. Libraries from the remaining gastrocnemius samples were sequenced on Novogene's Illumina NovaSeq 6000 (2 × 150 bp) to achieve at least 20,000 reads per sample. FastQ files were downloaded to the University of Florida HiPerGator computing cluster. FastQ files were processed on the HiPerGator cluster. Reads were aligned to thegenome GRCm38 (mm10) using HISAT2 v2.2.1 and annotated with HTseq-counts. Differential expression was analyzed with DESeq2, using sequencing site as a covariate, and filtered for genes with mean DESeq2 normalized counts > 5. GO enrichment was analyzed using Database for Annotation, Visualization and Integrated Discovery v2022q1 with< 0.01. Raw counts are included in. Mus musculus P Supplemental Data 1

TA muscle from 10-week-old mice was embedded in OCT compound (Tissue-Tek, Milles Inc.), frozen in 2-methylbutane (Thermo Fisher Scientific, O3551-4) cooled in liquid nitrogen, and stored at –80°C. Sections (10 μm thick) were cut with a Microm HM 505 E cryostat. For BMAL1 detection, sections were fixed in 4% paraformaldehyde, permeabilized with 0.5% Triton X-100, and blocked with 5% BSA, 10% normal goat serum, 1% glycine, and 0.1% Triton X-100 in PBS for 1 hour at room temperature. Primary antibodies (rabbit anti-BMAL1, 1:500; Novus Biologicals [Bio-Techne] NB100-2288, and mouse anti-Dystrophin conjugated with Alexa Fluor 488 [1:50; Santa Cruz Biotechnology sc-73592]), were incubated overnight at 4°C. Sections were then incubated with goat anti-rabbit Alexa Fluor 568 and counterstained with DAPI (1:10,000; Invitrogen D1306). For fiber typing, unfixed sections were incubated overnight with primary antibodies from the Developmental Studies Hybridoma Bank: IgG2b anti-MYH7 (Type 1, 1:100, BA.D5), IgG1 anti-MYH2 (Type 2A, neet, SC.71), IgM anti-MYH4 (Type 2B, neet, BF.F3), and rabbit anti-laminin (1:100; MilliporeSigma L9393). Sections were then incubated with Invitrogen's secondary antibodies (IgG1 Alexa Fluor 647, 1:250; IgG2b Alexa Fluor 488, 1:500; IgM Alexa Fluor 594, 1:250; goat anti-rabbit IgG Alexa Fluor 405, 1:100), washed, and postfixed in methanol. Sections were mounted in ProLong antifade medium. Images were captured with a Leica DMi8 microscope (20× objective). CSA assessment and fiber typing analysis were performed using MyoVision 2.0, as previously reported (). ⁵⁰

AAV vector copy numbers were quantified in gastrocnemius, quadriceps, soleus, tibialis anterior, heart, and liver tissues from KO+AAV mice (= 5–7). Genomic DNA was isolated using DNeasy Blood & Tissue kit (QIAGEN). qPCR was performed using a QuantiStudio 3 thermocycler (Applied Biosystems) with an inverted terminal repeat-specific primer set (Fwd: 5′-GGAACCCCTAGTGATGGAGTT-3′; Rev: 5′-CGGCCTCAGTGAGCGA-3′) () and Fast SYBR Green Master Mix (Thermo Fisher Scientific). A standard curve was generated with serial dilutions, 10–10, of AAV2-CK6-Bmal1-HA plasmid. Results were expressed as mean AAV vector copy number per microgram of genomic DNA. n ⁵¹ 2 8

Maximal tetanic tension of the EDL muscle from WT (= 6),-KO (= 4), and-KO+AAV (= 5) was assessed by the University of Florida Physiological Assessment Core, as previously described (). Specific tension and physiological CSA were calculated using standard equations (), with data normalized to WT values for comparison. n Bmal1 n Bmal1 n ^{52 53}

Protein was isolated as described by Roberts et al. (). Gastrocnemius muscle from 8 mice per group was weighed and placed in ice-cold buffer (25 mM Tris, pH 7.2, 0.5% Triton X-100, proteinase inhibitor) at 20× volume. Muscles were homogenized with glass Dounce grinders. Protein concentration was measured using the Microplate BCA protein assay kit (Thermo Fisher Scientific) and expressed per milligram of wet tissue. ²⁷

Plasma samples (= 5; 4 per group plus 1 pooled) were processed by the University of Florida Southeast Center for Integrated Metabolomics. Samples were prenormalized by protein content. Global metabolomics profiling was conducted using a Thermo Fisher Scientific Q-Exactive Orbitrap mass spectrometer with Dionex UHPLC, as reported before (). Extracts were spiked with 1 μL of a standard mixture for targeted carnitine identification. Samples were analyzed in positive and negative heated electrospray ionization with 35,000 mass resolution at200. Separation was performed on an ACE 18-pfp column using 0.1% formic acid (A) and acetonitrile (B). The flow rate was 350 μL/min with a column temperature of 25°C. Injection volumes were 4 μL for negative and 2 μL for positive ions. n m/z ⁵⁴

MZmine was used to identify, deisotope, align features, and fill gaps. Adducts and complexes were removed. Data from positive and negative ion modes were combined, yielding 1,420 features in the positive mode and 2,709 in the negative mode. Data were normalized to the sum of metabolites for each sample, log-transformed, and autoscaled. Statistical and KEGG enrichment analyses were done with MetaboAnalyst v5.0. Mummichog analysis associatedratios with possible metabolites. 2 m/z

Frozen gastrocnemius muscles were ground into powder with a liquid nitrogen–cooled mortar and pestle. The powder was homogenized in lysis buffer (50 mM HEPES, pH 8.5, 8 M urea, 0.5% sodium deoxycholate) at 4°C, then centrifuged at 11,200for 10 minutes. Protein concentrations of supernatants were determined using Coomassie-stained gels with BSA as a standard (). For TMT labeling, 100 μg of each sample was digested with LysC (Wako) at 1:100 (w/w) for 2 hours, followed by Trypsin (Promega) at 1:50 (w/w) for at least 3 hours. Peptides were reduced with 1 mM DTT, alkylated with 10 mM iodoacetamide, and quenched with 30 mM DTT. Samples were acidified with trifluoroacetic acid, desalted using C18 cartridges, and dried. The peptides were resuspended in 50 mM HEPES (pH 8.5) and labeled with 10-plex TMT reagents. Labeled samples were mixed, desalted, and fractionated using basic pH reverse-phase chromatography (Agilent 1220). Fractions were analyzed by acidic pH reverse-phase LC-MS/MS on an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific) with data-dependent acquisition. g ⁵⁵

MS/MS raw files were processed using JUMP. Data were searched against the UniProt mouse database with a reversed decoy database. Search parameters included 15 ppm mass tolerance for precursor and fragment ions, fully tryptic restriction, and static modifications for TMT tags and carbamidomethylation, with dynamic modification for Met oxidation. Proteins were quantified by summing reporter ion counts across matched eptide-spectrum matches. Statistical analysis used a< 0.05 and a fold-change threshold of ±0.3 logFC, as reported before (). Results are available in. P 2 ⁵⁶ Supplemental Data 2

Data are presented as mean and SD or SEM. The unpaired 2-tailed Student'stest was used for comparing 2 normally distributed groups. For comparing 3 groups, 1-way ANOVA with Tukey's multiple-comparison post hoc test was applied for qPCR results, ChIP-qPCR, wheel activity, rotarod, AUC, or muscle weight. Ordinary 2-way ANOVA with Tukey's multiple-comparison post hoc test was used for body weight or body composition over time, considering time and group as main effects. The cumulative survival analysis was carried out using the Kaplan-Meier method and the log-rank test for comparing survival curves. The Gehan-Breslow-Wilcoxon test was also used to assess early death events, but only log-rank testvalues are reported. Avalue less than 0.05 was considered significant. All analyses were performed using GraphPad Prism software version 9. t P P

Procedures followed institutional guidelines approved by the University of Florida IACUC (Protocol 0018).

RNA-Seq datasets generated in this study have been deposited in NCBI GEO (GSE232957). Metabolomics data are available through the NIH Metabolomics Workbench (PR001677). Proteomics data have been deposited in the ProteomeXchange Consortium via PRIDE (PXD054406). All data points shown in the graphs are provided in thefile. Supporting Data Values