The endogenous molecular clock orchestrates the temporal separation of substrate metabolism in skeletal muscle

Lipid metabolism: genes involved in fatty-acid uptake and β-oxidation peak in the mid-inactive/light phase

Skeletal muscle expresses specialized membrane transporters to facilitate the transport of lipids into the cell [55-57]. Two lipid transport genes that encode for fatty-acid binding proteins, Fabp4 (CT 24.0) and Fabp3 (heart/muscle isoform, CT 6.0), are expressed in a circadian manner with the highest mRNA expression in the early- and mid-inactive periods, respectively. Acrophase of circadian genes involved in lipid metabolism are illustrated in Figure 2. Normalized expression traces for each gene are located in Additional files 1, 2, and 3. Previous studies have demonstrated oscillations in plasma fatty acid concentrations in mice with peak levels occurring during the inactive/light period [58-60]. Further functional analysis is required to validate the predition that the rate of fatty-acid uptake in skeletal muscle peaks during the mid-late inactive period. Upon uptake into the cell, fatty acids can be stored as triglycerides or be converted to acetyl-CoA through β-oxidation [61]. Slc25a20 encodes for an acyl-carnitine translocase that transfers fatty acids into the inner-mitochondrial matrix and reaches peak expression in the middle of the inactive period (CT 7.5) [62]. We identified multiple genes that encode for β-oxidation enzymes to be circadian and also reach peak expression around the mid-inactive phase. These include the enoyl CoA hydratase Ech1 (CT 7.0), the tri-functional enzyme subunits Hadha (CT 8.0) and Hadhb (CT 8.0), and the acetyl-CoA acyltransferase Acaa2 (CT 9.0). Malonyl-CoA, an intermediate formed during de novo fatty acid synthesis, is a potent inhibitor of β-oxidation. The striated muscle enriched gene Mlycd (CT 7.5) encodes for the malonyl-CoA decarboxylase that promotes β-oxidation by reducing cytosolic concentrations of malonyl-CoA and reaches peak expression during the mid-inactive period similar to that of the circadian β-oxidation genes. These observations suggest that rates of β-oxidation are modulated over time of day and potentially through the endogenous molecular clock in skeletal muscle [10,63,64].

Nuclear receptors are known to be potent transcriptional regulators of metabolism as they sense changes in environmental conditions and induce appropriate changes in the expression of metabolic genes [65-69]. The nuclear receptor Estrogen-related receptor alpha (Esrra, CT 7.5) and the nuclear co-activator PPARγ coactivartor-1 beta (Ppargc1b, CT 7.0) are both circadian genes in skeletal muscle with peak expression occurring at the mid-inactive phase. These factors have been shown to promote mitochondrial biogenesis, fatty-acid uptake (targets Fabp3), and β-oxidation [70,71]. The nuclear co-repressor Nrip1, also known as Rip140, is a potent negative regulator of skeletal muscle oxidative metabolism and has been shown to suppress expression of the fatty-acid transporter, Fabp3, in skeletal muscle [72-74]. NRIP1 suppresses gene expression by binding nuclear receptors (including PPARs and estrogen-related receptors) and recruiting histone deacetylases [75]. Interestingly, peak expression of Nrip1 occurs during the beginning of the active period (CT 13.0) and may therefore act as a molecular brake to oxidative metabolism as the muscle transitions from lipid to carbohydrate utilization during the early active phase.

Lipid metabolism: lipogenic genes reach peak expression at the end of the active/dark phase

The lipogenic genes Acly (CT23.0), Acaca (CT 23.0), and Fasn (CT 22.5) involved in de novo fatty-acid synthesis, or the conversion of excess carbohydrates into fatty acids, reach peak expression at the end of the active phase (Figure 2) [61,76]. Scd1 (CT 24.0) encodes the enzyme that catalyzes the rate-limiting reaction of monounsaturated fatty-acid formation to promote lipid bilayer fluidity and lipogenesis [77,78]. The genes Srebf1 (CT 24.5), Srebf2 (CT 24.0), and Mlxip (CT 23.5) encode transcription factors that target carbohydrate response elements within lipogenic gene promoter regions (Acly, Acaca, and Fasn) and are also circadian with peak expression at the end of the active phase [79,80]. Consistent with our results, Srebf1 oscillations have been reported in the liver and genome-wide binding studies have shown a circadian recruitment pattern of SREBF1 to the promoters of lipogenic genes with maximal binding during the active (fed) stage [81-84].

The gene Pnpla3 (CT 21.0), also known as adiponutrin, promotes lipogenesis by converting LPA to phosphatidic acid (PA) [85]. The gene Lpin1 (CT 24.0) which encodes for the lipin-1 enzyme is responsible for converting phosphatidic acid (PA) to diacylglycerol (DAG), the upstream metabolite required in phospholipid biosynthesis [86,87]. The highly regulated, committing step in triacylglycerol (TG) synthesis, addition of a fatty-acyl-CoA to DAG, is performed by the product encoded by Dgat1 (CT 24.5), which is also expressed in a circadian manner [88]. Once a TG molecule is formed, it can be elongated by enzymes encoded by Acsl5 (CT 23.0) or Elovl5 (CT 22.5) [89,90]. The observation that circadian lipogenic genes reach peak expression levels around the end of the active phase suggests that skeletal muscle promotes storage of excess energy at the end of the active/absorptive phase.

Carbohydrate metabolism: genes involved in carbohydrate catabolism peak in the early active/dark phase

Glycolysis, the breakdown of glucose to form pyruvate, is primarily regulated at two enzymatic reactions catalyzed by the hexokinase and phosphofructokinase enzymes [91]. We observe that the hexokinase-2 (Hk2) gene is circadian with peak expression occurring at the beginning of the active phase (CT 12.0). Acrophase of circadian genes involved in carbohydrate metabolism are illustrated in Figure 3. Normalized expression traces for each gene are located in Additional files 1, 2, and 3. Hk2 is responsible for the first step in glycolysis by phosphorylating glucose to make glucose-6-phosphate, thereby trapping glucose within the cell [92]. The rate-limiting step of glycolysis involves the catalysis of fructose-6-phosphate to the highly unstable fructose-1,6-bisphosphate by the enzyme phosphofructokinase-1 (PFKM) [93,94]. A potent allosteric activator of PFKM is fructose-2,6-bisphosphate, which is the product of the other phosphofructokinase isozyme phosphofructokinase-2 (PFK2) [95]. Three genes (Pfkfb-1,3,4) that encode phosphofructokinase-2 subunits are circadian with peak expression occurring during the mid- and late-inactive phases (CT 10.0, CT 4.5, and CT 12.0, respectively).

Glycolytic flux through the Kreb’s cycle is controlled by pyruvate dehydrogenase complex (PDH) [96,97]. PDH decarboxylates pyruvate to form acetyl-CoA, which is a substrate for the Kreb’s cycle. The activity of PDH is regulated at the posttranslational level. Phosphorylation by kinases (PDKs) inhibits PDH activity, while dephosphorylation by phosphatases (PDPs) activates the complex [98,99]. The Pdk4 gene, which encodes for a PDH kinase that inhibits PDH, reaches maximal expression at the mid-inactive phase (CT 6.0). This expression pattern is similar to that of the β-oxidation genes and suggests that skeletal muscle substrate preference is pushed toward utilization of lipids over carbohydrates during the mid- to late-inactive phase. Conversely, the PDH phosphatase gene, Pdp1, peaks at the beginning of the active phase (CT 10.0) in a similar temporal fashion compared to the glycolytic enzymes described above. This temporal regulation of Pdp1 may therefore help increase glycolytic flux during the active phase. Dyar et al. observed similar expression patterns of Pdk4 and Pdp1 in skeletal muscle and were first to report a shift to carbohydrate utilization at the beginning of the active phase [43].

Adrb2 encodes for the β2-adrenergic receptor (β₂AR) involved in the fight-or-flight response in peripheral tissues [100,101]. Agonist (that is, catecholamine) binding is well established to evoke a cell-signaling cascade that promotes glucose uptake, glycogenolysis, and lipolysis to provide a readily available source of energy for skeletal muscle [102-104]. Adrb2 is expressed in a similar pattern to that of the glycolytic activating genes as it peaks at the beginning of the active phase. Interestingly, the expression of Adrb2 coincides with that of oscillating epinephrine concentrations in mammals, which has previously been identified as peaking at the beginning of the active phase in mouse models [105]. The G-protein receptor kinase, encoded by Adrbk1, phosphorylates the β₂AR, thereby rendering it susceptible to receptor-mediated endocytosis via β-arrestin proteins encoded by Arrdc3 and Arrb1 [106-108]. Adrbk1, Arrdc3, and Arrb1 are all expressed in a circadian manner and antiphasic to the expression of Adrb2. These observations suggest there is a time of day difference in adrenergic signaling and that sensitivity to epinephrine may be highest in skeletal muscle during the active period while being desensitized prior to the inactive period.

Carbohydrate metabolism: genes involved in carbohydrate storage peak at the mid-active/dark phase

Excess carbohydrates are stored as glycogen in skeletal muscle which accounts for approximately 70 to 80% of whole body stores [109]. Unlike the liver, skeletal muscle glycogen content is not responsible for maintaining blood glucose concentrations but serves as a rapidly accessible energy depot for active contractions [110]. Glycogenesis is regulated by both glucose-6P concentrations and the enzymatic activity of glycogen synthase [111,112]. The gene Ppp1r3c (CT 20.0) reaches peak expression around the mid-inactive phase and encodes a regulatory subunit of the protein phosphatase-1 (PP-1) responsible for activating glycogen synthase while also inhibiting glycogen breakdown (Figure 3) [113]. Enzymatic activity of PP-1, and subsequent activation of glycogen synthase, is regulated downstream of the insulin signaling pathway [114].

Insulin promotes an anabolic signaling cascade that works in opposition to that of adrenergic signaling to drive glycogen and lipid storage. Previous reports have identified a ‘counter-regulatory’ role of the insulin receptor to selectively inhibit β₂AR signaling through phosphorylation and subsequent internalization of the receptor [101,115]. Interestingly, the genes that encode the insulin receptor substrate-1, Irs1 (CT 22.0), and its downstream PI3-kinase target, Pik3r1 (CT 19.0), are both circadian with peak expression occurring at the late-active phase while the genes involved in suppressing PI3-kinase, Pik3ip1 (CT 8.0), and the insulin-receptor substrate-1, Fbxo40 (CT 5.0), reach peak expression during the inactive phase [116,117]. These data suggest that the molecular clock may act to prime skeletal muscle to store excess glucose during the mid- to late-active phase. This prediction is further supported by previous studies that report skeletal muscle glycogen content as having a diurnal rhythm with the highest levels occurring during the mid-active phase [118-120]. Skeletal muscle glucose uptake is primarily controlled via the presence/absence of the glucose transporter GLUT4/Slc2a4 in the plasma membrane (sarcolemma) and transverse tubules. A t-SNARE syntaxin-4 interacting protein, encoded by Stxbp4, has previously been shown to repress GLUT4 insertion into the plasma membrane in the absence of insulin signaling [121-123]. The gene Tbc1d1 encodes for Rab-GTPase that represses GLUT4 translocation in the absence of insulin- or contraction-induced signaling cascades [124-126]. Interestingly, Tbc1d1 and Stxbp4 are both expressed in a circadian manner and reach peak expression in the middle of the active phase (CT 19.0). Previous reports have identified Tbc1d1 as a circadian gene in skeletal muscle and other tissues [43,127]. Together, these gene products may play a role in reducing glucose uptake at the end of the active phase by repressing GLUT4 translocation and/or insertion into the plasma membrane. This temporal separation of anabolic and catabolic signaling processes in skeletal muscle may be vital for maintaining a tight regulation of serum glucose levels, and disruption of which may contribute to the metabolic phenotypes often reported in clock-mutant mice models.

iMS-Bmal1−/− display normal circadian activity rhythms

We used voluntary wheel running to assess circadian behavior in the iMS-Bmal1 mice 22 to 29 weeks posttreatment. We did not detect any significant differences in entrainment to light under 12-h light/12-h dark conditions between iMS-Bmal1 ^+/+ and iMS-Bmal1 ^−/− , and analysis of activity rhythms under constant darkness did not reveal any changes in circadian behavior (Figure 4D,E). Clock-lab analysis indicates that both iMS-Bmal1 ^+/+ and iMS-Bmal1 ^−/− exhibit approximate 24-h period lengths (23.85 ± 0.083 and 23.77 ± 0.138 h, respectively) with no differences in amplitude, the relative strength of the rhythm. These data are consistent with other studies and confirm that inactivation of BMAL1 in skeletal muscle does not directly alter circadian activity patterns [43,131]. Therefore, gene expression changes observed in this model are more likely to be downstream of the endogenous molecular clock mechanism in skeletal muscle.

iMS-Bmal1−/− gene expression changes reveal a fast to slow fiber-type shift

Skeletal muscle comprises different fiber types that are differentiated based on contractile function as well as predominant substrate utilization [132-135]. For example, fast-type skeletal muscles (type IIX/IIB) primarily rely on ATP generated from anaerobic metabolism (glycolysis/lactic-acid fermentation) to provide quick energy sources required for short bursts of activity, while slow-type skeletal muscles and fast-type IIA muscles rely on oxidative metabolism to promote a more sustained and less fatigable bout of contractions. We analyzed changes in gene expression related to fiber type following Bmal1 ablation in adult skeletal muscle and included both circadian and non-circadian transcripts. We identified a selective increase in slow-type sarcomeric genes in the gastrocnemius muscles with a limited effect on fast-type sarcomeric genes (Figure 6A,B). We chose the list of ‘slow’ and ‘fast’ sarcomeric genes, because these have been shown to be significantly enriched in either slow-soleus or fast-EDL myofiber preparations [136]. Additionally, calcium handling genes and nuclear receptors common in slow-fiber muscles (for example, Casq2, Atp2a2, Ankrd2, Csrp3.) were significantly increased in iMS-Bmal1 ^−/− (Table 1). Similar to the changes observed for the circadian metabolic genes, we see that non-circadian metabolic genes involved in carbohydrate metabolism are significantly decreased, while genes involved in lipid metabolism are increased (Tables 2 and 3). This switch from a fast to a slow fiber type mRNA profile is in agreement with the observed metabolic changes as slow fiber type muscles rely more heavily on oxidative metabolism compared to fast-type skeletal muscle.

Gene symbol	Fast or slow	Gene description	ΔExpression (Tam/Veh)	Student’s t test
Atp2a1	Fast	Calcium handling	0.99	ns
Atp2a2	Slow	Calcium handling	1.06	ns
Calm3	Fast	Calcium handling	0.84	***
Casq1	Fast	Calcium handling	1.00	ns
Casq2	Slow	Calcium handling	2.89	***
Itpr1	Fast	Calcium handling	1.05	ns
Pvalb	Fast	Calcium handling	1.02	ns
Ankrd2	Slow	Nuclear receptor	1.66	*
Csrp3	Slow	Nuclear receptor	2.13	**
Fhl1	Slow	Nuclear receptor	1.28	**
Nfatc2	Slow	Nuclear receptor	0.88	ns
Pdlim1	Slow	Nuclear receptor	1.51	***
Ppara	Slow	Nuclear receptor	1.23	*
Ppargc1a	Fast	Nuclear receptor	0.83	*
Sos2	Fast	Nuclear receptor	0.84	***

Average gene expression changes of calcium handling and nuclear receptor genes in iMS-Bmal1 ^−/− . ns, non-significant; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Gene symbol	Gene function	ΔExpression (Tam/Veh)	Student’s t test
Agpat3	Lipogenesis	1.59	***
Acadm	Lipolysis	1.31	***
Acot7	Lipolysis	1.18	***
Acot9	Lipolysis	1.44	**
Acsl1	Lipolysis	1.24	**
Cd36	Lipid transport	1.18	**
Cox5a	Electron transport chain	1.24	***
Cox6a1	Electron transport chain	1.30	*
Cpt2	Lipolysis	1.11	*
Fabp1	Lipid transport	1.28	*
Fabp5	Lipid transport	1.29	**
Fads2	Lipogenesis	1.29	*
Ldhb	Lactate metabolism	1.33	***
Ndufa8	Electron transport chain	1.24	***
Ndufb8	Electron transport chain	1.18	**
Plin5	Lipogenesis	1.41	***
Sdhc	Electron transport chain	1.18	***
Sdhd	Electron transport chain	1.21	**
Uqcr10	Electron transport chain	1.14	**

Average gene expression changes of metabolic genes that are significantly upregulated in iMS-Bmal1 ^−/− skeletal muscle. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

Gene symbol	Gene function	ΔExpression (Tam/Veh)	Student’s t test
Agl	Glycogenolysis	0.83	***
Akt1	Glucose uptake	0.84	**
Il15	Glucose uptake	0.86	*
Pak1	Glucose uptake	0.79	*
Pfkm	Glycolysis	0.81	***
Pgm2	Glycogenolysis	0.87	***
Phka1	Glycogenolysis	0.81	**
Prkab2	Glucose uptake	0.85	*
Prkag2	Glucose uptake	0.83	***
Prkag3	Glucose uptake	0.71	**
Rab10	Glucose uptake	0.86	**
Slc2a3	Glucose uptake	0.35	***

Average gene expression changes of metabolic genes that are significantly downregulated in iMS-Bmal1 ^−/− skeletal muscle. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.