Biochemical Journal

Research article

PPARα–LXR as a novel metabolostatic signalling axis in skeletal muscle that acts to optimize substrate selection in response to nutrient status

Paul W. Caton, Mark J. Holness, David Bishop-Bailey, Mary C. Sugden

Abstract

LXR (liver X receptor) and PPARα (peroxisome-proliferator-activated receptor α) are nuclear receptors that control the expression of genes involved in glucose and lipid homoeostasis. Using wild-type and PPARα-null mice fed on an LXR-agonist-supplemented diet, the present study analysed the impact of pharmacological LXR activation on the expression of metabolically important genes in skeletal muscle, testing the hypothesis that LXR activation can modulate PPAR action in skeletal muscle in a manner dependent on nutritional status. In the fed state, LXR activation promoted a gene profile favouring lipid storage and glucose oxidation, increasing SCD1 (stearoyl-CoA desaturase 1) expression and down-regulating PGC-1α (PPARγ co-activator-1α) and PDK4 (pyruvate dehydrogenase kinase 4) expression. PPARα deficiency enhanced LXR stimulation of SCD1 expression, and facilitated elevated SREBP-1 (sterol-regulatory-element-binding protein-1) expression. However, LXR-mediated down-regulation of PGC-1α and PDK4 was opposed and reversed by PPARα deficiency. During fasting, prior LXR activation augmented PPARα signalling to heighten FA (fatty acid) oxidation and decrease glucose oxidation by augmenting fasting-induced up-regulation of PGC-1α and PDK4 expression, effects opposed by PPARα deficiency. Starvation-induced down-regulation of SCD1 expression was opposed by antecedent LXR activation in wild-type mice, an effect enhanced further by PPARα deficiency, which may elicit increased channelling of FA into triacylglycerol to limit lipotoxicity. Our results also identified potential regulatory links between the protein deacetylases SIRT1 (sirtuin 1) and SIRT3 and PDK4 expression in muscle from fasted mice, with a requirement for PPARα. In summary, we therefore propose that a LXR–PPARα signalling axis acts as a metabolostatic regulatory mechanism to optimize substrate selection and disposition in skeletal muscle according to metabolic requirement.

  • peroxisome-proliferator-activated receptor γ co-activator-1α (PGC-1α)
  • pyruvate dehydrogenase kinase 4 (PDK4)
  • sirtuin 1 (SIRT1)
  • stearoyl-CoA desaturase 1 (SCD1)

INTRODUCTION

Skeletal muscle, because of its mass and total energy requirement, plays a key role in maintaining metabolic homoeostasis and energy balance. The increasing recognition that subtle changes in energy balance, when considered over prolonged periods of time, represent a significant risk factor for the development of metabolic abnormalities, including insulin resistance, hypertriglyceridaemia and T2DM (Type 2 diabetes mellitus), has heightened the search for specific cell signals that regulate metabolic control in skeletal muscle. NRs (nuclear receptors) are members of a superfamily of ligand-activated transcription factors that regulate gene expression in response to nutritional and physiological stimuli. One class of NRs function as metabolic sensors, binding to substrate and end-product component molecules of metabolic pathways such as lipids and FA (fatty acids). This class of NRs includes the LXRs (liver X receptors) and the PPARs (peroxisome-proliferator-activated receptors).

LXRα and LXRβ act as sensors of cellular cholesterol, binding oxysterols, and stimulate lipogenesis by enhancing expression of genes involved in FA biosynthesis. Although LXR activation is thought to be predominantly insulin-sensitizing [1], due to suppression of hepatic glucose output secondary to an LXR-led reduction in gluconeogenic capacity [2], it can also elicit dramatic increases in hepatic TAG (triacylglycerol) production [3] that, if excessive, would be predicted to be insulin-desensitizing through excessive ectopic lipid accumulation [4]. This has led to complications with the use of LXR agonists as therapies for T2DM [5]. However, it has been suggested that, although skeletal-muscle TAG is increased in insulin-resistant states, storage of FA as TAG may divert them from cytotoxic pathways and oppose the development of lipotoxicity [6]. The lipogenic actions of the LXRs are therefore considered to directly oppose the action of PPARα, which promotes FA oxidation.

The main target tissues of the LXRs are considered to be those known to be important in lipid biosynthesis, particularly the liver. LXRα expression predominates in metabolically active tissues, including the liver, but also the small intestine, kidney, macrophages and adipose tissue. In contrast, LXRβ is more widely expressed (reviewed in [7]) and LXR target genes have been identified in other tissues that are not major sites of lipogenesis, including oxidative muscle [8,9]. LXR activation has been reported to increase the expression of SREBP-1 [which encodes the lipogenic transcription factor SREBP-1c (sterol-regulatory-element-binding protein-1c)] in skeletal muscle in vivo [9] and increase expression of SCD1 [which encodes SCD1 (stearoyl-CoA desaturase 1)] in cultured muscle cells [8]. The SCDs are rate-limiting for the biosynthesis of mono-unsaturated FA from saturated FA, which are used more effectively for FA esterification to form TAG. Overexpression of SCD1 in cultured myotubes also prevents the inflammatory and ER (endoplasmic reticulum) stress responses to exposure to the saturated FA palmitate [10]. In cultured skeletal muscle, LXR activation by the synthetic LXR agonist TO-901317 increases the uptake, distribution into cellular lipids (FA, diacylglycerol and TAG) and oxidation of palmitate but, at the same time, increases glucose uptake and oxidation [11]. Concomitant increases in rates of oxidation of both FA and glucose are unusual as there is generally reciprocal regulation of the use of these two energy substrates (the glucose/FA cycle). In the longer term, this is achieved by up-regulation of PDK4 (pyruvate dehydrogenase kinase 4), which catalyses inhibitory phosphorylation of the PDC (pyruvate dehydrogenase complex) (reviewed in [12]).

PPARα and PPARδ play a critical role in the maintenance of lipid homoeostasis (oxidation and production), regulating the transcription of genes involved in FA oxidation, binding and transport and, in skeletal muscle, that encoding PDK4. Exposure of insulin-sensitive tissues to excess NEFA (non-esterified FA) and circulating TAG induces insulin resistance [4] that can be corrected by the administration of PPARα activators by actions to promote the removal of intracellular lipid through tissue FA oxidation (see for example [13,14]). PPARα-null mice show an inability to up-regulate genes involved in FA uptake and oxidation after fasting, and as a consequence become hypoglycaemic and hypoketonaemic, with accumulation of lipid in their tissues. PPARγ, primarily expressed in adipocytes, but also in other tissues including skeletal muscle [15], promotes FA synthesis and storage.

The PPARs are co-activated by the inducible PGC-1α (PPARγ co-activator-1α) (reviewed in [16]). In skeletal muscle and liver, PGC-1α is found in complex with GCN5, an acetyltransferase that acetylates PGC-1 at several lysine residues thereby inhibiting its transcriptional activity [17,18]. The balance between relative levels and activity of GCN5 and the protein deacetylase SIRT1 (sirtuin 1; the mammalian Sir2 orthologue), which can deacetylate PGC-1α [17,19], provides a potential regulatory convergence point for induction of PPAR gene targets. Induction of PPARα gene targets requires the interaction of PPARα and PGC-1, often in complex with other enzymes and co-activators such as SIRT1 [20]. PGC-1 co-activators can also interact with other regulatory elements, including the lipogenic transcription factor SREBP-1c, abnormalities in which have been implicated in the development of T2DM.

In the present study, we analysed the consequences of pharmacological LXR activation by TO-901317, which activates LXRα and LXRβ with approximately equal potency [21], on the expression of metabolically important genes in skeletal muscle in vivo. The genes encoding proteins regulating TAG storage that were selected for study were SREBP-1 and SCD1. Key regulators of lipid and glucose oxidation selected were PGC-1α and PDK4. Since overexpression of SCD1 in cultured myotubes prevents ER stress [10], we also studied the expression of XBP1 (X-box-binding protein 1), which is established as a key effector of the mammalian UPR (unfolded protein response) in skeletal muscle [22]. Since changes in PGC-1α co-transcriptional activity in skeletal muscle may be modulated by post-translational modification, including reversible acetylation, we additionally investigated the impact of LXR activation on SIRT1 and GCN5 expression. Studies were undertaken in the well-fed state, where the glucose supply is ample, insulin levels are high and incoming FA are predominantly esterified rather than oxidized. We also evaluated the impact of pharmacological LXR activation on the response of metabolically important genes to starvation, where PPARα signalling becomes dominant, FA are oxidized and glucose is conserved. To elucidate further the potentially complex interplay between LXR and PPARα in regulating skeletal-muscle gene expression, we utilized PPARα-null mice.

EXPERIMENTAL

Animals

All procedures performed in the present study were approved by the Local Ethics Review Committee of the University of Oxford and were carried out under the authority of the appropriate Home Office (U.K.) personal and project licences in accordance with the Home Office Animals (Scientific Procedures) Act 1986. Tissue and blood samples from male PPARα-null mice, bred on to a Sv/129 genetic background, were kindly provided by Professor G.F. Gibbons [Metabolic Research Laboratory, OCDEM (Oxford Centre for Diabetes, Endocrinology and Metabolism), University of Oxford, Oxford, U.K.]. WT (wild-type) male Sv/129 mice were used as controls. Mice were maintained on a reverse light/dark cycle (12 h dark phase 03:00–15:00 h, 12 h light phase 15:00–03:00 h). WT mice were provided with ad libitum access to either a standard high-carbohydrate/low-fat rodent laboratory diet or a standard diet additionally containing the non-sterol LXR agonist N-(2,2,2-trifluoroethyl)-N-[4-(2,2,2-trifluoro-1-hydroxy-1-trifluoromethylethyl) phenyl] sulfonamide (TO-901317) at a final concentration of 0.01% for 5 days. PPARα-null mice were fed only on a diet containing TO-901317 (0.01%). After 5 days mice were either killed at the mid point of the dark phase on day 5 or starved for 24 h after 5 days of maintenance on either the normal or the TO-901317-supplemented diet before being killed. Dietary administration of TO-901317 to WT and PPARα-null mice was kindly undertaken by Professor G.F. Gibbons.

Tissue and blood sampling

Mice were anaesthetized by injection of sodium pentobarbital (60 mg/ml in 0.9% NaCl; 1 ml/kg of body mass, intraperitoneally) and, once locomotor activity had ceased, liver and mixed hind-limb skeletal muscle were rapidly excised and freeze-clamped using aluminium clamps pre-cooled in liquid N2. Frozen tissue was stored in liquid N2 prior to analyses. Blood was sampled from the abdominal aorta. Blood samples were centrifuged for 5 min at 12000 g and plasma was stored at −20 °C.

Measurement of plasma insulin and blood glucose

Insulin was measured using a mouse insulin double-antibody ELISA according to the manufacturer's instructions (Mercodia). Glucose was determined using a commercially available kit (Roche Diagnostics).

qRT-PCR (quantitative real-time PCR)

Gene expression was measured by qRT-PCR, using TaqMan® or SYBR® Green methodology. Gene expression was determined by ΔΔCT normalized against 18S control RNA (Applied Biosystems). Primer sequence (Eurogentec) details are shown in Supplementary Table S1 at http://www.BiochemJ.org/bj/437/bj4370521add.htm.

Immunoblotting

Solubilized protein samples [10 μg; measured and equalized in each fraction using an RC DC™ system (BioRad)] were separated by SDS/PAGE and transferred on to a PVDF membrane (GE Healthcare). Blots were blocked in 5% (w/v) non-fat dried skimmed milk powder in a TBS/T (Tris-buffered saline/0.1% Tween 20) solution and then incubated with the primary antibody. Antibodies used in the present study were anti-SIRT1, anti-GCN5, anti-XBP1, anti-PGC-1α (all from Santa Cruz Biotechnology) and anti-SIRT3 (Abgent). Following primary antibody incubation, blots were incubated with anti-mouse, anti-rabbit or anti-goat secondary antibodies (all from Jackson ImmunoResearch). Detection of bands was achieved by using the chemiluminescence substrate SuperSignal West Pico (Pierce). Reference protein measurements were made with a mouse monoclonal anti-β-actin (clone AC-15; Sigma) primary antibody in a 3% (w/v) non-fat dried skimmed milk/TBS/T solution at 4 °C. For each representative immunoblot shown, the results are from a single gel exposed for a uniform duration, and each lane represents a preparation from a different mouse.

Statistical analysis

Results are presented as the means±S.E.M., with the numbers of mice in parentheses. Statistical analysis was performed by ANOVA followed by Fisher's post-hoc tests for individual comparisons or Student's t test as appropriate (StatView software; Abacus Concepts). A P value of < 0.05 was considered to be statistically significant.

RESULTS

Metabolic characteristics of the groups

In fed mice, LXR activation (closed bars) decreased plasma glucose concentrations in both PPARα-null mice (21%, P<0.01) and wild-type mice (17%, P<0.05; Figure 1A). The hypoglycaemic effect of TO-901397 in the fed state is consistent with the reported effects of LXR activation to suppress hepatic glucose output [2]. Plasma glucose levels were decreased in fasted compared with fed mice (30.1%, P<0.001; Figure 1). However, antecedent LXR activation exerted a potent effect to further decrease glycaemia in fasted PPARα-null mice (30%; P<0.001; Figure 1B). Plasma insulin levels were decreased in fasted mice relative to fed mice (40%, P<0.01; Figure 1). Plasma insulin levels were not significantly affected by LXR activation under fed and fasted conditions in WT mice or PPARα-null mice, although LXR activation elicited modest increases in plasma insulin levels in PPARα-null mice (1.5-fold and 1.4-fold in fed and fasted mice respectively; Figures 1C and 1D). The insulin/glucose ratio analysis suggested that LXR activation resulted in a loss of insulin sensitivity in PPARα-null mice, but not in WT mice (Figure 1).

Figure 1 Impact of pharmacological LXR activation on plasma glucose and insulin concentrations in fed or fasted WT and PPARα-null mice

WT mice were fed either a standard rodent diet or a standard diet supplemented with TO-901317 at a final concentration of 0.01% for 5 days. Mice were either killed at the mid point of the dark phase on day 5 (fed) or starved for 24 h after 5 days of maintenance on the standard diet (fed, open bars; starved, hatched bars) or the standard diet supplemented with TO-901317 (closed bars) before being killed. PPARα-null (KO) mice were fed only on a diet containing TO-901317 and sampled either in the fed or fasted state (closed bars). Blood samples were withdrawn for measurement of plasma glucose (A and B), plasma insulin (C and D) and insulin/glucose (I/G) ratios (E and F). Results are expressed as means±S.E.M. for 6–8 mice in each group. Statistically significant effects of fasting are indicated by ‡‡P<0.01 and ‡‡‡P<0.001. Statistically significant effects of TO-901317 are indicated by *P<0.05, **P<0.01 and ***P<0.001. Statistically significant effects of PPARα deficiency are indicated by †††P<0.001.

LXR activation increases SCD1 and XBP1 expression in skeletal muscle in the absence of increased SREBP-1c expression

SREBP-1, which encodes the lipogenic transcription factor SREBP-1c, is reported to be a key LXR target gene in skeletal muscle [9]. Furthermore, the SREBP-1 promoter contains two LXR response elements that are essential for insulin-dependent activation of SREBP-1 [3]. This has led to the suggestion that insulin activates the SREBP-1 promoter by increasing the activity of the LXRs, possibly through production of an LXR-specific ligand [3]. Both SREBP-1 and SCD1 gene expression are up-regulated in the liver under lipogenic conditions, including insulin stimulation and high-carbohydrate feeding (reviewed in [23]). In the present study, starvation (24 h) of WT mice resulted in a marked decline in both SREBP-1 (93%, P<0.001) and SCD1 (71%) gene expression in liver (Figure 2A; see also [24]). SCD1 expression was also significantly suppressed in skeletal muscle following starvation (85%, P<0.001; Figure 2F), whereas skeletal-muscle SREBP-1 expression increased 5-fold (P<0.001) (Figure 2B), despite lowered circulating insulin levels. We failed to detect altered SREBP-1 gene expression in response to pharmacological LXR activation in the hindlimb skeletal muscle of fed WT mice sampled at the mid point of the dark cycle (Figure 2C), when insulin concentrations peak, and endogenously generated agonists of LXR would be predicted to contribute to the expression of LXR target genes. Although LXR activation failed to augment SREBP-1c expression, it elicited a modest (1.4-fold) increase in gene expression of SCD1 in skeletal muscle of WT mice (Figure 2G). In contrast with its limited effects in WT mice, LXR activation significantly increased both SREBP-1c and SCD1 expression levels, by 3.3-fold (P<0.001) and 8.8-fold (P<0.05) respectively, in fed PPARα-null mice and by 8.5-fold (P<0.001) and 4.1-fold (P<0.05) respectively compared with WT LXR agonist-treated mice (Figures 2C and 2G).

Figure 2 Impact of pharmacological LXR activation on the expression of genes regulating TAG storage (SREBP1, SCD1) and the mammalian unfolded protein response (XBP1) in skeletal muscle of fasted WT and PPARα-null mice

WT mice were fed either a standard rodent diet or a standard diet supplemented with TO-901317 at a final concentration of 0.01% for 5 days. Mice were either killed at the mid point of the dark phase on day 5 (fed) or starved for 24 h after 5 days of maintenance on the standard diet (fed, open bars; starved, hatched bars) or the standard diet supplemented with TO-901317 (closed bars) before being killed. PPARα-null (KO) mice were fed only on a diet containing TO-901317 and sampled either in the fed or fasted state (closed bars). Liver and mixed hindlimb skeletal muscle were rapidly excised and freeze-clamped. Gene expression for SREBP-1 (AD), SCD1 (EH) and XBP1 (IL) were measured in liver and mixed hindlimb muscle by qRT-PCR. Representative Western blots for XBP-1 protein expression in liver (I) and hind-limb skeletal muscle (J) of WT mice sampled in the fed or 24 h-fasted state are also shown. Results are expressed as means±S.E.M. Statistically significant effects of fasting are indicated by ‡‡‡P<0.001. Statistically significant effects of TO-901317 are indicated by *P<0.05, **P<0.01 and ***P<0.001. Statistically significant effects of PPARα deficiency are indicated by †P<0.05 and †††P<0.001.

In contrast with observations in fed mice, antecedent LXR activation increased expression of both SREBP-1 and SCD1 in skeletal muscle of fasted mice (Figures 2D and 2H). PPARα deficiency blocked LXR-mediated induction of skeletal-muscle SREBP-1 expression after starvation (Figure 2D), whereas it further enhanced the expression of SCD1 (Figure 2H).

Previous studies have reported a key role for XBP1 in the induction of hepatic lipogenic genes, including SCD1 [25]. In liver, starvation-induced decreases in lipogenic gene expression were not associated with a decline in XBP1 mRNA or protein expression (Figure 2I); however, increased XBP1 expression in skeletal muscle of fasted mice (5.3-fold, P<0.05; Figure 2J) closely paralleled changes in SREBP-1, rather than SCD1 expression. Modest LXR-mediated up-regulation of SCD1 was associated with increased expression of XBP1 (3.4-fold) in skeletal muscle of fed WT mice, whereas the more dramatic LXR-mediated up-regulation of SREBP-1 and SCD1 expression in fed PPARα-null mice was also paralleled by increased XBP1 expression by 7.1-fold (P<0.01) compared with fed WT mice, and by 1.8-fold (P<0.05) compared with WT LXR-agonist-treated mice (Figure 2C). Moreover, in WT fasted mice, skeletal-muscle XBP1 levels were markedly enhanced following LXR activation (1.6 fold, P<0.001), a response that was blocked in PPARα-null mice (Figure 2I).

Effects of TO-901317 on PGC-1α and PDK4 expression in fed WT and PPARα-null mice

In skeletal muscle, PPARα and PPARδ are key regulators of FA oxidation (reviewed in [12]). In addition, the NR co-activator PGC-1α is recognized to play important roles in regulating glucose and lipid metabolism in liver and skeletal muscle by interacting with a range of NRs, including PPARα and the LXRs (see for example [16]). The hepatic PGC-1α gene was significantly increased in starved compared with fed WT mice (Figure 3A), as is consistent with its role in co-ordinating the response of hepatic nutrient handling to starvation (reviewed in [16]). The up-regulation of PGC-1α gene expression seen in response to fasting (Figure 3B) was more robust in skeletal muscle compared with the liver of WT mice (2.2-fold, P<0.01, compared with 9-fold, P < 0.05). Protein expression of PGC-1α was enhanced by fasting to similar extents in liver and skeletal muscle (Figures 3A and 3B). PGC-1α expression levels were decreased (59%, P<0.05) by LXR activation in skeletal muscle of fed WT mice (Figure 3C). Since PGC-1α co-operates with LXR to regulate gene expression, this could represent a feedback loop. However, the effect of LXR activation to decrease skeletal-muscle PGC-1α required the participation of PPARα, since PGC-1α expression was increased (by 3.4-fold, P<0.001) in response to pharmacological LXR activation in skeletal muscle of fed PPARα-null mice (Figure 3C).

Figure 3 Impact of pharmacological LXR activation on the gene and protein expression of key regulators of lipid (PGC-1α) and glucose (PDK4) oxidation in skeletal muscle of fed or fasted WT and PPARα-null mice

WT mice were fed either a standard rodent diet or a standard diet supplemented with TO-901317 at a final concentration of 0.01% for 5 days. Mice were either killed at the mid point of the dark phase on day 5 (fed) or starved for 24 h after 5 days of maintenance on the standard diet (fed, open bars; starved, hatched bars) or the standard diet supplemented with TO-901317 (closed bars) before being killed. PPARα-null (KO) mice were fed only on a diet containing TO-901317 and sampled either in the fed or fasted state (closed bars). Liver and mixed hindlimb skeletal muscle were rapidly excised and freeze-clamped. Gene expression for PGC-1α (AD) and PDHK4 (EH) were measured in liver and mixed hindlimb muscle by qRT-PCR. Representative Western blots for PGC-1α (A and B) and PDHK4 (E and F) protein expression in liver and hind-limb skeletal muscle of WT mice sampled in the fed or 24 h-fasted state are also shown. Results are expressed as means±S.E.M. Statistically significant effects of fasting are indicated by ‡P<0.05, ‡‡P<0.01 and ‡‡‡ P < 0.001. Statistically significant effects of TO-901317 are indicated by *P<0.05 and ***P<0.001. Statistically significant effects of PPARα deficiency are indicated by †††P<0.001.

The gene expression of PDK4, a regulatory kinase that inactivates PDC, blocking glucose oxidation and thereby forcing fat oxidation, is PPARα/δ-regulated (reviewed in [12]). As expected, fasting elevated hepatic PDK4 expression, with a greater (3.2-fold, P<0.01) increase in protein expression (Figure 3E) than gene expression (see also [26]). Fasting also increased PDK4 expression in skeletal muscle, with increases of 2.0-fold (P<0.001) and 2.2-fold (P<0.01) in gene and protein expression respectively (Figure 3F) (see also [27]). In fed WT skeletal muscle, similar to changes in PGC-1α expression, LXR activation suppressed PDK4 expression (61%, P<0.05) (Figure 3G). In addition, PDK4 expression, like that of PGC-1α, was enhanced (2.2-fold, P<0.001) in response to LXR activation in skeletal muscle of fed PPARα-null mice (Figure 3G).

Although LXR activation lowered PGC-1α expression in skeletal muscle of fed WT mice, antecedent LXR activation significantly induced PGC-1α expression in fasted WT mice (3.2-fold, P<0.001; Figure 3D). The inductive effect of LXR activation on PGC-1α gene expression was abolished in PPARα-null mice, with expression levels of PGC-1α mRNA exhibiting a ~50% reduction compared with untreated fasted WT mice (Figure 3D). The response of PDK4 expression to starvation was exaggerated by antecedent LXR activation in WT mice (3.8-fold, P<0.001); however, this response was opposed by PPARα deficiency (Figure 3H).

Effects of LXR activation, PPARα deficiency and fasting on SIRT1 and GCN5 expression

In skeletal muscle, PGC-1α target genes for FA oxidation are induced by the deacetylase SIRT1 [28], whereas GNC5 has been reported to acetylate and suppress PGC-1α activity in skeletal muscle and in liver [17,18]. Moreover, SIRT1 can activate hepatic LXR [29]. SIRT1 gene expression was markedly up-regulated in response to fasting in livers of WT mice, with a concomitant 81% increase in SIRT1 protein expression (P<0.01) (Figure 4A; see also [30]). In contrast, hepatic GCN5 gene expression was unaffected by fasting, although there was a 2.6-fold increase (P<0.05) in GCN5 protein expression (Figure 4E), suggestive of post-transcriptional control. The responses to fasting of SIRT1 and GCN5 in skeletal muscle differed to those observed in liver. SIRT1 gene and protein expression were unaffected by starvation (Figure 4B), whereas GCN5 gene expression in skeletal muscle was suppressed by 46% (P<0.01) in response to starvation (Figure 4F). The decline in skeletal-muscle GCN5 gene expression seen in response to fasting was not, however, paralleled by a decline in GCN5 protein expression (Figure 4F).

Figure 4 Response of gene and protein expression of the deacetylases SIRT1 and SIRT3 and acetylase GCN5 to pharmacological LXR activation in skeletal muscle of fed or fasted wild-type and PPARα-null mice

WT mice were fed either a standard rodent diet or a standard diet supplemented with TO-901317 at a final concentration of 0.01% (w/w) for 5 days. Mice were either killed at the mid point of the dark phase on day 5 (fed) or starved for 24 h after 5 days of maintenance on the standard diet (fed, open bars; starved, hatched bars) or the standard diet supplemented with TO-901317 (closed bars) before being killed. PPARα-null (KO) mice were fed only on a diet containing TO-901317 and sampled either in the fed or fasted state (closed bars). Liver and mixed hind-limb skeletal muscle were rapidly excised and freeze-clamped. Gene expression for SIRT1 (AD), GCN5 (EH) and SIRT3 (IL) were measured in liver and hind-limb mixed muscle by qRT-PCR. Representative Western blots for SIRT1 (A and B), GCN5 (E and F) and SIRT3 (I and J) protein expression in liver and hind-limb skeletal muscle of WT mice sampled in the fed or 24 h-fasted state are also shown. Results are expressed as means±S.E.M. Statistically significant effects of fasting are indicated by ‡P<0.05, ‡‡P<0.01 and ‡‡‡P<0.001. Statistically significant effects of TO-901317 are indicated by *P<0.05, **P<0.01 and ***P<0.001. Statistically significant effects of PPARα deficiency are indicated by †P<0.05 and †††P<0.001.

We observed that LXR activation decreased gene expression of SIRT1 (by 29%, P<0.05) in skeletal muscle of fed WT mice, an effect that was opposed by PPARα deficiency (Figure 4C). LXR activation also decreased GCN5 mRNA expression in skeletal muscle of fed WT mice; however, this response was not opposed by PPARα deficiency (Figure 4G). Thus PPARα deficiency selectively affects LXR-mediated suppression of SIRT1 in the fed state.

In contrast with the fed state, pharmacological LXR activation prior to starvation elicited a marked increase in both SIRT1 expression (93%, P<0.01; Figure 4D) and GCN5 expression (3.8 fold, P<0.001; Figure 4H) in fasted WT skeletal muscle. Up-regulation of both SIRT1 and GCN5 expression in fasted mice by antecedent LXR activation was abolished by PPARα deficiency (Figures 4D and 4H).

SIRT1 and SIRT3 are differentially regulated by LXR activation in the fed, but not fasted, state

Recent findings show that acetylation appears to target most enzymes in most metabolic pathways including those occurring within mitochondria, such as the tricarboxylic acid cycle and mitochondrial β-oxidation [31,32]. SIRT3 is localized in the mitochondrial matrix, where it regulates the acetylation levels of metabolic enzymes, and its expression is up-regulated in liver and brown adipose tissue in response to starvation [33], leading to promotion of FA oxidation [34]. In the present study, SIRT3 gene expression was suppressed in both liver (by 40%, P<0.05; Figure 4I) and skeletal muscle (by 69%, P<0.001; Figure 4J) of WT mice. However, unlike SIRT1, changes in SIRT3 gene expression were not associated with corresponding changes in SIRT3 protein expression. Thus starvation elicited a significant (2.2-fold, P<0.01) increase in SIRT3 protein expression in liver (Figure 4I), together with a modest and non-significant increase (32%) in skeletal muscle (Figure 4J). The response of SIRT3 gene expression to LXR activation also differed to that of SIRT1. Thus LXR activation increased gene expression of SIRT3 in skeletal muscles of WT mice (by 73%, P<0.01), a response that, unlike that of SIRT1, was unaffected by PPARα deficiency (Figure 4K), demonstrating that LXR regulation of mitochondrial SIRT3 gene expression is PPARα-independent. Antecedent LXR activation opposed the fasting-induced decline in SIRT3 gene expression such that SIRT3 mRNA levels were increased by 7.5-fold (P<0.05) compared with untreated fasted WT mice. The action of LXR to abolish the effect of fasting on SIRT3 was, however, completely blocked by PPARα deficiency (Figure 4L).

DISCUSSION

Previous studies in human muscle cells in vitro have demonstrated that LXR activation by TO-901317 directly affects gene expression of key enzymes regulating nutrient handling, and modifies glucose and fatty acid metabolism. For example, LXR activation by TO-901317 in human muscle cells increases oleate and palmitate uptake and oxidation, as well as glucose uptake and oxidation [11,35]. Treatment with TO-901317 also increases lipogenesis and lipid accumulation in myotubes [36], together with the number of lipid droplets [35]. LXR activation by TO-901317 in cultured muscle cells also increases the expression of lipogenic enzymes, including SREBP-1c, FAS (FA synthase) and SCD1 [8]. The present study investigated potential interactions between LXR and PPARα in modulating the expression of genes encoding proteins regulating FA storage as TAG in skeletal muscle in vivo (Table 1). We also investigated the expression of metabolically important genes encoding key regulators of lipid and glucose oxidation in skeletal muscle in vivo. Fasting was employed as a metabolic challenge associated with an increased demand for and/or supply of FA for oxidation. To assess the impact of impaired lipo-oxidative capacity on the response to LXR activation, we also administered TO-901317 to PPARα-null mice for 5 days. We demonstrated that, in the fed state, LXR activation elicited modest increases in SCD1 paralleled by down-regulation of expression of PGC-1α and PDK4 in skeletal muscle. PPARα deficiency significantly enhanced LXR stimulation of SCD1 expression, which was associated with significantly elevated SREBP-1 expression. PPARα can inhibit SREBP-1c promoter activity induced by LXR [37]. Our present studies therefore indicate that adequate PPARα signalling exists in skeletal muscle of fed WT mice to oppose LXR-induced up-regulation of SREBP-1 expression (as might be predicted since lipid synthesis and oxidation are generally reciprocally regulated). Down-regulation of PGC-1α and PDK4 in skeletal muscle elicited by LXR activation in WT mice was opposed and reversed by PPARα deficiency. Given that PPARα deficiency also selectively allows an effect of LXR activation to increase SREBP-1 expression in the fed state, these results could imply that a lipid or cholesterol metabolite generated secondary to LXR activation, whose accumulation in skeletal muscle in the fed state is opposed by PPARα signalling, acts to increase PDK4 and PGC1α expression. In WT mice, SREBP-1 was up-regulated in skeletal muscle in response to starvation, whereas it was down-regulated in liver. SREBP-1 up-regulation in skeletal muscle in response to fasting was augmented by antecedent LXR activation. This was linked to PPARα signalling, as the responses were blunted in PPARα-null mice. In contrast, down-regulation of SCD1 expression was elicited by starvation in both liver and skeletal muscle. The response of SCD1 in skeletal muscle to starvation was opposed by antecedent LXR activation in WT mice, an effect enhanced further by the absence of PPARα signalling. PGC-1α and PDK4 gene and protein expression were up-regulated in both liver and skeletal muscle in response to fasting. Responses of PGC-1α and PDK4 expression to fasting after prior LXR activation were opposite to those observed in the fed state, namely augmentation of their fasting-induced up-regulation. Unlike the fed state, LXR-mediated effects on fasting were opposed by PPARα deficiency, supporting the critical role of PPARα in the response to fasting.

View this table:
Table 1 Summary of changes in hindlimb skeletal muscle gene expression in fed and fasted WT and PPARα-null mice in response to LXR activation

The PPARs and LXRs both bind to target genes as heterodimers with the RXR (retinoid X receptor). In the absence of ligand, PPAR–RXR and LXR–RXR can bind to target genes to repress transcription. In the presence of activating ligands, co-repressor complexes are replaced by co-activator complexes, enhancing gene transcription. Cross-talk between PPARα and LXR is proposed to ensure that antagonistic pathways are not simultaneously activated (reviewed in [38]), either through direct interaction between PPAR and LXR or via competition for a limiting amount of shared RXR [39]. Our results add a further dimension to this scenario. On the basis of the changes in pattern of expression of key metabolic genes in skeletal muscle observed in response to LXR activation, we propose that LXR activation acts to optimize substrate selection and disposition in skeletal muscle through augmenting PPAR signalling, possibly through LXR-mediated generation of an, as yet unidentified, PPAR ligand. This action is not intrinsically PPAR isotype selective, but depends on additional changes in PPAR ligand availability and/or PPAR signalling that LXR activation is able to augment. Consequently, we propose that in the fed state LXR activation promotes glucose oxidation (through down-regulation of PDK4) and enhances lipid storage both directly (via up-regulation of SCD1 and possibly SREBP-1c), as has been demonstrated previously [8,9] and possibly by enhancing PPARγ signalling. We further propose that prior LXR activation augments PPARα signalling in starvation to enhance the capacity for FA oxidation (by increasing the expression of the PPARα co-activator PGC-1α) and suppress glucose oxidation (through up-regulation of PDK4 expression). The increase in PDK4 expression may reflect up-regulation by PGC-1α (reviewed in [40]). In support of our hypothesis, the expression of a cohort of genes (including PGC-1α and PDK4) is augmented in skeletal muscle following LXR activation in starved WT mice, but not in starved PPARα-null mice, indicating that the effect of LXR activation is PPARα dependent. Increases in PGC-1α and PDK4 would favour enhanced lipid oxidation and restricted glucose oxidation and glucose sparing in the starved state. PGC-1α promotes the expression of FA oxidation genes (reviewed in [40]), whereas PDHK4 is the main PDHK isoform responsible for suppression of skeletal-muscle PDC activity in starvation [41] and insulin resistance [42], which by decreasing glucose oxidation forces fat oxidation (see [41,42]).

In the fed state, in contrast, LXR activation suppresses the expression of PGC-1α and PDK4, a response reversed by PPARα deficiency, with augmented gene expression, together with up-regulation of SREBP-1 expression. Chronic ligand activation of LXR increases PPARγ expression in human skeletal muscle cells [11] and PPARγ agonist treatment decreases PDK4 expression in skeletal muscle [43], mimicking the response of skeletal-muscle PDK4 expression to LXR activation in fed WT mice in the present study. Thus LXR activation of fed WT mice may increase PPARγ signalling in skeletal muscle, possibly via the generation of an endogenous PPARγ agonist. Intact PPARα signalling may be required to maintain a muscle metabolic profile appropriate to the fed state. LXR activation unrestrained by PPARα signalling may switch the muscle metabolic profile to one resembling that seen in fasting so as to promote clearance of excessive lipid.

PGC-1α co-transcriptional activity in skeletal muscle is modulated by post-translational modification, including reversible acetylation, mediated by the deacetylase SIRT1 and the acetyl transferase GCN5 [17] (reviewed in [40]). Thus increased nutrient availability would render PGC-1α transcriptionally inactive through hyperacetylation by GCN5, whereas decreased nutrient availability (e.g. in fasting) would increase deacetylation of PGC-1α by SIRT1, leading to enhanced PGC-1α binding to target gene promoters and transcriptional activation by remodelling of the local chromatin environment [40]. In support of this, we observed that fasting up-regulated SIRT1 and PGC-1α gene and protein expression in liver, thereby favouring increased responsiveness of PGC-1α to demand for up-regulation of FA oxidation. In contrast, in skeletal muscle up-regulation of PGC-1α gene and protein expression occurred in the absence of changes in SIRT1 gene or protein expression, suggesting that the mechanism of regulation of PGC-1α may be tissue-specific. The role of GCN5 in the regulation of transcriptional activity in response to fasting in both liver and skeletal muscle is less clear, as fasting selectively down-regulated GCN5 gene expression in skeletal muscle; however, protein expression of GCN5 was increased in both liver and skeletal muscle in response to fasting. In the present study LXR activation in the fed state suppressed PGC-1α, GCN5 and SIRT1 expression, which may limit the influence of PGC-1α on the skeletal-muscle metabolic profile. However, absence of PPARα signalling in the fed state enhanced PGC-1α and SIRT1 expression, which may compensate for the loss of PPARα lipo-oxidative effects. Pharmacological activation of SIRT1 stimulates FA oxidation in skeletal muscle of high-fat fed insulin-resistant mice through up-regulation of FA oxidation genes including PPARα and PDK4 [28]. Our results support the concept that strong regulatory links exist between SIRT1 and PDK4 expression in muscle. In contrast, GCN5 deactivates PGC-1α in skeletal muscle, leading to decreased FA oxidation, mitochondrial biogenesis and energy expenditure [17,44]. Our present results suggest that GCN5 expression may also be regulated by LXR and PPARα. In fasting, expression of PGC-1α, SIRT1 and GCN5 are increased in skeletal muscle after prior activation of LXR, favouring increased responsiveness of PGC-1α to the need for the up-regulation of FA oxidation. Loss of these effects in fasted PPARα-null mice may reflect an impaired ability to up-regulate FA oxidation.

The complete oxidation of acetyl-CoA derived from glucose (via PDC) or FA (via β-oxidation) requires the participation of the tricarboxylic acid cycle, a wrapped linear pathway where the end-product (oxaloacetate, derived from glucose) is essential for the first step of the pathway. Consequently, nutrient input is slowed or stopped if insufficient end product is available. In starvation, an adequate supply of oxaloacetate is probably achieved by enhanced expression of PDK4 which, by suppressing the use of pyruvate for acetyl-CoA formation via PDC, conserves pyruvate for oxaloacetate formation via pyruvate carboxylase. Our present results suggest that acetylation–deacetylation metabolic control mechanisms can be up-regulated by LXR activation to allow selective amplification or acceleration of metabolic responses to transitions between nutritional states. It is likely that these mechanisms are co-ordinated by extrinsic factors such as the lipid-sensing PPARs according to the prevailing dominant PPAR isotype and/or endogenous ligand generation or accumulation.

SIRT3 is located in the mitochondrial matrix and, like SIRT1, is an NAD+-dependent protein deacetylase. In mice, SIRT3 protein levels are up-regulated during fasting in brown adipose tissue and liver [34]. In liver, SIRT3 regulates FA and ATP production during fasting by deacetylating and activating long-chain acyl-CoA dehydrogenase, which catalyses the first step of long-chain FA β-oxidation. In the present study, hepatic SIRT3 gene expression was suppressed by fasting; however, as reported by Hirschey et al. [34], hepatic SIRT3 protein expression increased in response to fasting. SIRT3 expression was also decreased in skeletal muscle in response to fasting; however, as in liver, this was associated with an increase in SIRT3 protein levels (see also [45]). Prior LXR activation increased SIRT3 expression in fed and fasted WT mice, but this response was abolished in fasted PPARα-null mice, suggesting that the response of skeletal-muscle SIRT3 expression to starvation is dependent both on the preceding level of LXR activation and the degree of activation of PPARα in starvation.

In excess, saturated FA exert lipotoxic effects, leading to ER stress, inflammation and insulin resistance. Overexpression of SCD1 in cultured myotubes prevents the inflammatory and ER stress response to exposure to the saturated FA palmitate [10]. SCD1 is highly expressed in skeletal muscle from obese humans [14] and obese insulin-resistant Zucker diabetic fatty rats [46]. LXR activation increases SCD1 expression in cultured muscle cells [8]. We observed increases in skeletal muscle SCD1 expression after LXR activation in vivo in the fed state and in the starved state after antecedent LXR activation. In both states, PPARα deficiency exaggerated this effect of LXR activation, suggesting competition between LXR and PPARα, such that PPARα limits LXR's ability to increase SCD1, and PPARα's action to promote FA oxidation is dominant to the actions of LXR. Given that SCD1 converts saturated FA into mono-unsaturated FA that can be esterified to neutral lipid (TAG), SCD1 may exert a protective action against lipotoxicity, by increasing disposition of lipid towards TAG synthesis when the capacity to oxidize lipid is compromised, for example by PPARα deficiency.

XBP1 is a key effector of the mammalian UPR in skeletal muscle [22]. In the present study, fasting up-regulated XBP1 gene and protein expression in both liver and skeletal muscle. LXR activation in the fed state in the absence of PPARα signalling also led to XBP1 induction, which may reflect a restraint on FA oxidation. In fasting, antecedent LXR activation augments XBP1 expression in skeletal muscle of WT mice, a response lost in PPARα-null mice. The increase in XBP1 expression in fasted WT mice treated previously with the LXR agonist is also associated with the induction of SCD1 expression, which may reflect an attempt to limit the lipotoxic effects by channeling FA into TAG. Furthermore, in the absence of PPARα signalling, and therefore an inability to up-regulate FA oxidation, SCD1 expression is induced even more and is associated with a reversal of starvation-induced increases in XBP1 expression.

In conclusion, we propose a novel signalling axis comprising PPARα–LXR that, via altered expression of key target genes, maximizes ATP production and minimizes the accumulation of toxic metabolites. Our results argue for a novel role of PPARα, whereby it acts in conjunction with LXR as a metabolic regulator to maintain metabolic homoeostasis (‘metabolostat’) under both fed and fasted conditions (via a PPARα–LXR axis) forcing the skeletal muscle to a set metabolic point. This allows optimization of substrate selection and disposition in skeletal muscle through augmented PPAR signalling.

AUTHOR CONTRIBUTION

Mary Sugden and Mark Holness conceived and designed the experiments, and performed tissue sampling and metabolite analysis. Paul Caton performed qRT-PCR and immunoblotting. David Bishop-Bailey, Mary Sugden, Mark Holness, Paul Caton and participated in interpreting the results and writing the paper.

FUNDING

This work was supported in part by Diabetes UK [grant numbers BDA:RD08/0003665 and BDA:RD06/0003424] (to M.C.S. and M.J.H.).

Acknowledgments

We thank Geoff Gibbons for allowing us to sample blood and tissue from the PPARα-null mouse colony at OCDEM.

Abbreviations: ER, endoplasmic reticulum; FA, fatty acid(s); LXR, liver X receptor; NR, nuclear receptor; OCDEM, Oxford Centre for Diabetes, Endocrinology and Metabolism; PDC, pyruvate dehydrogenase complex; PDK, pyruvate dehydrogenase kinase; PPAR, peroxisome-proliferator-activated receptor; PGC, PPARγ co-activator; qRT-PCR, quantitative real-time PCR; RXR, retinoid X receptor; SCD, stearoyl-CoA desaturase; SIRT, sirtuin; SREBP, sterol-regulatory-element-binding protein; TAG, triacylglycerol; T2DM, Type 2 diabetes mellitus; WT, wild-type; XBP, X-box-binding protein

References

View Abstract