OUP user menu

TNF-α reduces PGC-1α expression through NF-κB and p38 MAPK leading to increased glucose oxidation in a human cardiac cell model

Xavier Palomer , David Álvarez-Guardia , Ricardo Rodríguez-Calvo , Teresa Coll , Juan C. Laguna , Mercy M. Davidson , Tung O. Chan , Arthur M. Feldman , Manuel Vázquez-Carrera
DOI: http://dx.doi.org/10.1093/cvr/cvn327 703-712 First published online: 27 November 2008


Aims Inflammatory responses in the heart that are driven by sustained increases in cytokines have been associated with several pathological processes, including cardiac hypertrophy and heart failure. Emerging data suggest a link between cardiomyopathy and myocardial metabolism dysregulation. To further elucidate the relationship between a pro-inflammatory profile and cardiac metabolism dysregulation, a human cell line of cardiac origin, AC16, was treated with tumour necrosis factor-α (TNF-α).

Methods and results Exposure of AC16 cells to TNF-α inhibited the expression of peroxisome proliferator-activated receptor coactivator 1α (PGC-1α), an upstream regulator of lipid and glucose oxidative metabolism. Studies performed with cardiac-specific transgenic mice (Mus musculus) overexpressing TNF-α, which have been well characterized as a model of cytokine-induced cardiomyopathy, also displayed reduced PGC-1α expression in the heart compared with that of control mice. The mechanism by which TNF-α reduced PGC-1α expression in vitro appeared to be largely mediated via both p38 mitogen-activated protein kinase and nuclear factor-κB pathways. PGC-1α downregulation resulted in an increase in glucose oxidation rate, which involved a reduction in pyruvate dehydrogenase kinase 4 expression and depended on the DNA-binding activity of both peroxisome proliferator-activated receptor β/δ and estrogen-related receptor α transcription factors.

Conclusion These results point to PGC-1α downregulation as a potential contributor to cardiac dysfunction and heart failure in metabolic disorders with an inflammatory background.

  • Peroxisome proliferator-activated receptor γ coactivator 1α
  • NF-κB
  • TNF-α

1. Introduction

The myocardium expresses and secretes several cytokines in response to different stimuli, such as tumour necrosis factor-α (TNF-α) and interleukin (IL)-6, which are under the control of the ubiquitous inducible transcription factor called nuclear factor-κB (NF-κB). Inflammatory responses driven by sustained increases in TNF-α, which mainly acts in an autocrine fashion, have been related to several pathological processes, including ischaemic myocardial injury, cardiac hypertrophy, and chronic heart failure.13

The adult mammalian heart generates ATP primarily from fatty acid (FA) β-oxidation by mitochondria, though this organelle can also catabolize other substrates, such as glucose and lactate, in order to ensure an efficient and constant fuel source. However, cardiac energy substrate flexibility becomes constrained under certain circumstances, such as in cardiac hypertrophy and heart failure, owing to a shift in the source of energy from FA to glucose. Metabolic changes in cardiac substrate utilization entail dysregulation of genes involved in the transport and catabolism of FA and glucose. The expression of these genes is controlled at least in part by the peroxisome proliferator-activated receptor (PPAR) transcription factors family. There are three different PPAR subtypes, PPARα, PPARβ/δ and PPARγ, which require heterodimerization with another nuclear receptor, the retinoid X receptor, in order to be activated. Heterodimerization is then followed by coactivator recruitment and subsequent induction of target gene transcription.

The PPARγ coactivator-1α (PGC-1α) is a potent transcriptional coactivator that acts as an upstream regulator of lipid and glucose oxidative metabolism in a variety of tissues.4 In the myocardium, PGC-1α may coactivate PPARα and PPARβ/δ subtypes, although it also interacts with and coactivates the nuclear respiratory factor-1 (NRF-1)5 and the estrogen-related receptor α (ERRα) transcription factors.6 The primary functions of PGC-1α in the heart consist of the activation of mitochondrial biogenesis, oxidative phosphorylation, and respiration, and the transcriptional control of enzymes involved in metabolism.5,6 PGC-1α expression in the heart is activated perinatally, coinciding with a metabolic shift towards major FA consumption and a burst of mitochondrial biogenesis, which play a crucial role in cardiac oxidative capacity.7 Similarly, PGC-1α is induced in the adult heart by physiological stimuli that increase its dependence on FA oxidation for energy, such as fasting and exercise.8 On the other hand, the PGC-1α/ERRα axis regulates genes involved in glucose oxidation, such as pyruvate dehydrogenase kinase 4 (PDK4). This kinase is responsible for the inactivation by phosphorylation of the pyruvate dehydrogenase complex, which catalyzes the rate-limiting step of glucose oxidation. Interestingly, cardiac ERRα expression following birth is concordant with the upregulation of PGC-1α and, furthermore, deletion of the ERRα gene in mice has been found to accelerate heart failure in a process which involves PGC-1α downregulation as well.6,8,9 Cardiac hypertrophy7 and heart failure10,11 have been associated with decreased expression of PGC-1α and its target genes in several animal models, thus providing increasing evidence of a link between lessening of PGC-1α activity and cardiac dysfunction.

To gain a better understanding of the mechanisms by which exposure to TNF-α may result in the metabolic dysregulation that underlies heart dysfunction and failure in metabolic diseases, we have taken advantage of a human cell line of cardiac origin, AC16. Results reported herein demonstrate that TNF-α reduces PGC-1α expression in AC16 cells in vitro and also in a mouse model with cardiac-specific overexpression of TNF-α. In AC16 cells, PGC-1α downregulation eventually results in an increase in the glucose oxidation rate. The mechanism by which TNF-α reduces PGC-1α expression involves both p38 mitogen-activated protein kinase (p38 MAPK) and NF-κB. Consequently, therapies aimed at preventing PGC-1α downregulation in inflammatory states associated with metabolic disorders might be useful for preventing cardiomyopathies.

2. Methods

2.1 Reagents

GW501516 was obtained from Alexis Biochemicals (Lausen, Switzerland). BSA was from Calbiochem (Darmstadt, Germany). D-[U-14C]-glucose was purchased from PerkinElmer (Waltham, MA, USA), and 1-[14C]-oleic acid, [α-32P]-dATP, and [γ-32P]-ATP were purchased from GE Healthcare Life Sciences (Sant Cugat, Spain). All other chemicals, except when specified, were purchased from Sigma-Aldrich (Madrid, Spain). Antibody against IκBα was from Cell Signaling Technology (Beverly, MA, USA); p65, Oct-1, and PPARα were from Santa Cruz Biotechnology Inc. (Heidelberg, Germany), ERR1 from Affinity BioReagents (Golden, CO, USA), and β-actin from Sigma.

2.2 Cell culture and transfection

Human AC16 cells were maintained and grown as previously described.12 Briefly, non-differentiated AC16 cells were maintained in medium composed of Dulbecco's modified Eagle's medium:F12 (Invitrogen, Barcelona, Spain) supplemented with 12.5% foetal bovine serum, 1% penicillin–streptomycin, and 1% Fungizone (Invitrogen), and grown at 37°C in a humid atmosphere of 5% CO2/95% air until they reached 70–80% confluence. For PGC-1α overexpression studies, AC16 cells were transfected with pcDNA4/His-myc/PGC-1α (Addgene plasmid 10974, Cambridge, MA, USA)13 using the Lipofectamine 2000 kit (Invitrogen) according to the manufacturer's recommendations, with both pcDNA4/His-myc/lacZ and pcDNA4/His-myc (Invitrogen) as control vectors. Transfection time and DNA:lipofectamine ratio were set at 24 h and 1:3, respectively, after optimization with pcDNA4/His-myc/lacZ using a β-galactosidase reporter gene staining kit (Sigma).

2.3 TNF-α transgenic mouse cardiac sample preparation

We used transgenic TNF1.6 male mice (8 to 12-week-old) with cardiac-specific overexpression of TNF-α. These well-characterized mice were created as a model of cytokine-induced cardiomyopathy and congestive heart failure.14 Ventricular sample tissues were obtained as described previously.15 The study was approved by our Institutional Animal Research Committee and conformed with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85–23, revised 1996).

2.4 RNA preparation and analysis

Relative levels of specific mRNAs were assessed by the reverse transcription–polymerase chain reaction (RT–PCR), as previously described.4 Owing to its high transcription levels, PGC-1α expression after overexpression was analysed by real-time RT–PCR. The sequences of the forward and reverse primers used for amplification are displayed in Supplementary material online, Table S1.

2.5 Electrophoretic mobility shift assay and immunoblot analysis

The electrophoretic mobility shift assay (EMSA) and immunoblot analysis were performed as previously described4,16 (Supplementary material online).

2.6 Glucose oxidation and fatty acid oxidation

Glucose oxidation rate was determined as reported in previously published protocols,17 using some modifications, while FA oxidation was measured according to an adapted protocol from Roduit et al.18 (Supplementary material online).

2.7 Statistical analysis

Results are expressed as the mean ± SD of at least three separate experiments. Significant differences were established by either the Student's t-test or one-way ANOVA, according to the number of groups compared, using the computer program GraphPad Prism (GraphPad Software Inc. V4.03, San Diego, CA, USA). In the latter case, when significant variations were found, the Tukey–Kramer multiple comparisons post-test was performed. Differences were considered significant at P < 0.05.

3. Results

3.1 PGC-1α expression is downregulated by TNF-α

Treatment of AC16 cells with TNF-α (100 ng/mL for 24 h, see Supplementary material online) significantly induced the expression of pro-inflammatory genes such as monocyte chemoattractant protein (MCP-1, approximately two-fold, P < 0.01), IL-6 (approximately four-fold, P < 0.05), and TNF-α (∼20-fold, P < 0.001) (see Supplementary material online, Figure S1). Next, we evaluated whether induction with TNF-α led to a reduction in the expression of cardiac genes involved in cell metabolism. Neither PPARβ/δ nor PPARα expression was reduced after treatment with TNF-α (data not shown). In consonance with this, the expression level of PPAR target genes involved in mitochondrial FA β-oxidation, such as medium-chain acyl-CoA dehydrogenase (MCAD) or muscle-type carnitine palmitoyltransferase-I (M-CPT-I), was not reduced. However, when we examined the effects of TNF-α on the expression of PGC-1α, a significant inhibition was observed (∼80% reduction, P < 0.01) compared with control cells (Figure 1A). On the contrary, expression of the close homologue PGC-1β remained unaltered, despite the fact it was expressed at higher levels than PGC-1α in AC16 cells (data not shown). Expression of PDK4 was found to be reduced in parallel with PGC-1α expression (Figure 1B), whereas a marked increase in manganese superoxide dismutase (MnSOD) expression, an antioxidant enzyme transiently stimulated by PGC-1α, was observed upon TNF-α stimulation (see Supplementary material online, Figure S1). To further confirm our results, studies with cardiac-specific transgenic mice overexpressing TNF-α (TNF1.6 strain) were performed. Consistent with the in vitro studies, TNF1.6 transgenic mice displayed reduced PGC-1α (∼50% reduction, P < 0.001) and PDK4 (∼60% reduction, P < 0.001) expression in the heart, compared with that in control mice (Figure 1B).

Figure 1

PGC-1α and PDK4 expression is downregulated by tumour necrosis factor-α in AC16 cells (A) and in left ventricle tissue of transgenic TNF1.6 mice (B) in a process mediated by nuclear factor-κB (C) and p38 mitogen-activated protein kinase (D). AC16 cells were incubated with tumour necrosis factor-α (A, C, and D) in the presence or absence of parthenolide (10 µmol/L, C) or PD169316 (10 µmol/L, D). The panels display a representative autoradiogram of the reverse transcription–polymerase chain reaction analysis of the corresponding mRNA levels. Graphics represent the quantification of the 18S-normalized (A, C, and D) or Aprt-normalized (B) mRNA levels, expressed as a percentage of control samples. *P < 0.05, **P < 0.01, and ***P < 0.001.

3.2 NF-κB and p38 MAPK mediate PGC-1α downregulation by TNF-α

To assess the role of NF-κB in the TNF-α-mediated control of PGC-1α and PDK4, we treated cells with the inhibitor parthenolide. As shown in Figure 1C, the ∼75–80% reduction observed in the transcript levels of PGC-1α and PDK4 after treatment with TNF-α was prevented by parthenolide. The induction in MCP-1, IL-6 and MnSOD mRNA levels, three well-known targets of NF-κB activity, was also inhibited after parthenolide treatment. Pyrrolidine dithiocarbamate was used to further confirm these results (see Supplementary material online, Figure S2A), since it can reduce inflammatory cytokine production in the myocardium as a consequence of NF-κB inhibition. On the other hand, addition of the p38 MAPK inhibitors PD169316 (Figure 1D) or SB202190 (see Supplementary material online, Figure S2B) to AC16 cells reversed some of the effects of TNF-α. Thus, PGC-1α and PDK4 expression reached basal levels after p38 MAPK inhibition, suggesting that this kinase was involved in the TNF-α-mediated PGC-1α inhibition. PD169316 did not reduce the expression of IL-6 or MnSOD, indicating that the pro-inflammatory profile induced by TNF-α did not require the activation of the p38 MAPK pathway.

An EMSA was carried out to verify whether incubation of AC16 cells with TNF-α led to increased NF-κB activity. NF-κB formed four specific DNA-binding complexes with nuclear proteins, but supershift analyses demonstrated that only complexes I and II contained the p65 subunit of NF-κB. Complexes I, II, and III increased in TNF-α-treated cells, and this effect was completely reversed by parthenolide (Figure 2A). PD169316 addition also prevented complex I formation, although, surprisingly, complexes II and III were induced even more than in TNF-α-treated cells (Figure 2A). These results suggest that NF-κB and p38 MAPK interact in mediating the TNF-α response, despite the fact that PD169316 did not prevent pro-inflammatory cytokine upregulation by TNF-α.

Figure 2

PGC-1α downregulation by tumour necrosis factor-α involves both nuclear factor-κB and p38 mitogen-activated protein kinase activation. (A) Electrophoretic mobility shift assay showing nuclear factor-κB activity after treatment of AC16 cells with tumour necrosis factor-α in the presence or absence of 10 µmol/L parthenolide or 10 µmol/L PD169316 (NE, nuclear extracts; IC, immunocomplex). (B) Effect of exogenous tumour necrosis factor-α addition on IκBα and the p65 subunit of nuclear factor-κB protein levels. Total (IκBα) or nuclear (p65) protein extracts were isolated from AC16 cells incubated with tumour necrosis factor-α in the presence or absence of parthenolide or PD169316. To show equal loading of protein, β-actin signal from the same blot is included. All autoradiograph data are representative of three separate experiments.

The nuclear localization of NF-κB is prevented by interaction with IκB inhibitors, which leads to its cytoplasmic accumulation as an inactive heterodimer. Because the biological effects of NF-κB are dependent at least in part on its translocation to the nucleus after protein IκB kinase-mediated phosphorylation, we also assessed IκBα protein levels after TNF-α exposure. As expected, cells exposed to TNF-α showed reduced IκBα protein levels (Figure 2B). This reduction in IκBα protein levels was not prevented after coincubation with parthenolide or PD169316, suggesting that PGC-1α downregulation by TNF-α did not depend on IκBα protein levels. Although parthenolide may specifically inhibit NF-κB activity by preventing IκBα phosphorylation, our results are consistent with another mechanism of parthenolide action, which involves alkylation of p65 at Cys38, thus inhibiting DNA binding of NF-κB.19 In accordance with this, nuclear p65 protein levels rapidly increased after TNF-α treatment, a phenomenon that was prevented by parthenolide and PD169316 treatments (Figure 2B).

3.3 PPARβ/δ and estrogen-related receptor α might be involved in PDK4 downregulation

A plasmid carrying the PGC-1α gene under the control of a constitutive CMV promoter was used to overexpress this coactivator in AC16 cells (see Supplementary material online, Figure S3A). Exogenous PGC-1α overexpression for 24 h induced a substantial increase in the expression of this gene, which completely abrogated its downregulation after TNF-α treatment (see Supplementary material online, Figure S3B). Expression analysis of PDK4, MnSOD, and IL-6 in PGC-1α-transfected cells after addition of TNF-α clearly demonstrated that only PDK4 transcription was controlled by this coactivator in human AC16 cells (Figure 3A). To exclude any undesired effect owing to unspecific induction of the PGC-1α-related coactivator PGC-1β, the expression of this gene was also evaluated and no changes were observed (data not shown).

Figure 3

(A) PDK4 is transcriptionally regulated by PGC-1α. Cells were transfected with pCDNA4/His-myc/PGC-1a (PGC-1α), or pCDNA4/His-myc (pCDNA4c) and pCDNA4/His-myc/lacZ (lacZ) as control vectors, and treated with or without tumour necrosis factor-α. PDK4, IL-6, and MnSOD mRNA levels were analysed by reverse transcription–polymerase chain reaction, as described in Figure 1. *P < 0.05, **P < 0.01, and ***P < 0.001. Tumour necrosis factor-α reduces the DNA-binding activity of PPARβ/δ and estrogen-related receptor α in AC16 cells, in a process that may be reversed by PD169316 but not parthenolide. Electrophoretic mobility shift assay was performed with a 32P-labelled peroxisome proliferator-activated receptorβ/δ (B) or estrogen-related receptor α (C) nucleotide and crude nuclear protein extracts (NE) as described in Figure 2. Supershift analysis performed by incubation with an antibody directed against peroxisome proliferator-activated receptorβ/δ (B, middle panel) or PGC-1α (B and C, right panel) or estrogen-related receptor α (C, middle panel) is also shown.

PDK4 may be activated in various mouse tissues by administration of PPAR ligands. However, there is also a PPAR-independent pathway in skeletal muscle that activates PDK4 through the orphan nuclear receptor ERRα, in a process which also requires PGC-1α recruitment. Since previous data obtained in our laboratory had indicated that PPAR inhibition owing to repression of transcriptional activity by NF-κB might be responsible for PDK4 downregulation, we investigated the role of PPAR by means of EMSA. The PPRE probe formed only one specific complex with AC16 nuclear proteins, and supershift studies demonstrated that this DNA-binding activity was almost entirely attributable to the PPARβ/δ subtype (Figure 3B). This is not surprising given that PPARβ/δ was expressed at much higher levels than PPARα in these cells (data not shown). As expected, TNF-α significantly reduced the PPAR DNA-binding activity. Although addition of PD169316 inhibited the ability of TNF-α to alter PPARβ/δ activity, parthenolide had no effect (Figure 3B). In addition, when the synthetic PPARβ/δ agonists L-165041 or GW501516 were added in the presence of TNF-α, PGC-1α or PDK4 transcript levels were not restored (data not shown). To clarify these results, we analysed the expression of two additional enzymes that are known to be PPAR-regulated, acyl CoA-oxidase and the liver form of CPT-I, which are also expressed in the heart. The expression of these genes was not affected by TNF-α (data not shown). Moreover, in contrast with previous reports,20 overexpression of PGC-1α did not increase M-CPT-I or MCAD expression (data not shown), thus ruling out this protein as a transcriptional coactivator of these genes in AC16 cells. Addition of anti-PGC-1α antibody in a supershift assay performed with the PPRE probe did not reduce the binding of Complex I (Figure 3B, right panel), whereas addition of this antibody to an EMSA assay performed with the same nuclear extracts but using the ERRα probe yielded a strong reduction of the Complex I (Figure 3C, right panel), thus reinforcing the lack of binding of PGC-1α to PPAR in these cells.

ERRα DNA-binding activity, as examined by EMSA, yielded only one specific complex, which displayed a reduction in the presence of TNF-α (Figure 3C). Addition of PD169316 in the presence of TNF-α reversed ERRα transcriptional activity but parthenolide did not. The reduction of ERRα complex I formation after incubation with parthenolide alone might account for the lack of effect of the latter when coincubated with TNF-α, although the underlying mechanisms still remain to be elucidated. Since it has been reported that PGC-1α may induce the transcription of ERRα in an autoregulatory feed-forward mechanism,21 in addition to co-activating its transcriptional activity, the expression of this transcription factor was also assessed. However, ERRα expression was not reduced in parallel with PGC-1α after TNF-α addition (data not shown).

3.4 Downregulation of PDK4 increases the glucose oxidation rate

We next sought to determine whether changes in PDK4 expression affected glucose and FA oxidation in AC16 cells. As shown in Figure 4A, TNF-α induced a significant increase in the glucose oxidation rate (1.15 nmol glucose/g h) with respect to untreated AC16 cells (0.79 nmol glucose/g h), and coincubation with parthenolide or PD169316 reduced glucose oxidation to basal levels (0.53 and 0.83 nmol glucose/g h, respectively). By contrast, TNF-α treatment had no effect on FA oxidation as measured by the rate of oleate oxidation (Figure 4B).

Figure 4

PGC-1α downregulation by tumour necrosis factor-α dysregulates metabolism. (A) Tumour necrosis factor-α induces glucose oxidation to 14CO2. The graph represents the [U-14C]-glucose oxidation rates in AC16 cells incubated with tumour necrosis factor-α in the presence or absence of parthenolide or PD169316. (B) Effect of tumour necrosis factor-α-stimulated cells on fatty acid oxidation, measured as the production of 14CO2 in the incubation medium after treatment with tumour necrosis factor-α in the absence or presence of PD169316. *P < 0.05, **P < 0.01, and ***P < 0.001.

4. Discussion

To date, most studies directed at understanding the mechanisms that underlie the shift in glucose metabolism during cardiac hypertrophy have been performed with murine models. In the present study, we have taken advantage of a novel human cardiac model, the AC16 cell line, to examine the metabolic disturbances that may arise from inflammatory processes in the heart. To our knowledge, this is the first report examining such inflammatory processes in an in vitro model of human origin, thus potentially overcoming many of the inconsistencies encountered when results obtained with murine models have been extrapolated to human beings. This is important, especially taking into account that some of the transcription factors involved in metabolism and inflammation, such as PPAR,22 are known to be expressed at lower levels in human cells than in rodent cells. In fact, gene expression is also differentially regulated by PPAR in human vs. rodent cells.23 A major drawback of the study might be the origin of the AC16 cell line itself, since it consists of a fusion of primary ventricular cells with SV-40-transformed fibroblasts. However, this cell line develops many of the biochemical and morphological properties characteristic of cardiac muscle cells, even though it does not form completely differentiated cardiomyocytes.12 Cardiomyocytes play an important role in cardiac pathobiology, even though other myocardial cells, such as fibroblasts, which constitute up to 70% of the ventricular myocardial cells, may also contribute to this process.24 Both cell types secrete and respond to inflammatory TNF-α,3,25 although cardiomyocytes have been reported to be the major local source of TNF-α in the myocardium during inflammatory processes.26

Results presented here show that a marked inhibition in PGC-1α expression is produced in the presence of locally increased TNF-α levels. In agreement with our previous results obtained with human AC16 cells and transgenic TNF1.6 mice, treatment for only 6 h with 10 ng/mL TNF-α already reduced PGC-1α expression in rat neonatal cardiomyocytes (data not shown). Numerous studies have indicated that PGC-1α is a crucial regulator of cardiac metabolism during development and in response to stress. Cardiac PGC-1α expression, along with its target transcription factors PPAR and ERRα, are all reduced in animal models of heart failure5,27 and in pathological forms of cardiac hypertrophy,6 suggesting that this decrease may be responsible for an energetic failure that can eventually lead to cardiac dysfunction. In support of this hypothesis, PGC-1α null mice were found to develop early heart failure and dysfunction.27 Not surprisingly, AC16 cells exposed to TNF-α displayed a pro-inflammatory profile as a direct consequence of increased transcriptional activity of NF-κB, which has been implicated in the development of cardiac hypertrophy and failure.28

Recent studies suggest that mutations in cardiac mitochondrial DNA may contribute to the development of dilated cardiomyopathy.15 However, analysis of the expression of genes whose transcription is subjected to PGC-1α coactivation and that are involved in mitochondrial biogenesis, the electron transport chain, or muscle glucose uptake, such as mtTFA (mitochondrial transcription factor A), CytC (cytochrome c oxidase), NRF-1, COII (cytochrome c oxidase, subunit II of complex IV), ND1 (NADH dehydrogenase subunit 1), and GLUT4 demonstrated that neither TNF-α nor PGC-1α regulated their expression in AC16 cells (data not shown). Studies with PGC-1β or PGC-1α deficient mice have demonstrated that PGC-1α is not essential for the fundamental mitochondrial biogenesis process in the heart,6,29 and this might explain the lack of effect on genes involved in mitochondrial biogenesis and oxidative capacity that was observed in our study. Since PGC-1β can also stimulate mitochondrial biogenesis and respiration, it is plausible that this coactivator might compensate for the reduction in PGC-1α even though its expression was not upregulated in AC16 cells. In fact, TNF1.6 mice displayed marked mitochondrial structural and functional alterations which coincided with reduced PGC-1β expression (data not shown). On the other hand, the possibility that mitochondrial changes driven by PGC-1α would become apparent over the course of time cannot be ruled out, since our in vitro studies limited gene expression analyses to up to 48 h after TNF-α addition.

A recent study has demonstrated that PDK4 modulation is sufficient to cause metabolic inflexibility and exacerbate cardiomyopathy in transgenic mice,30 and we found that PDK4 expression correlated with PGC-1α levels in AC16 cells treated with TNF-α and in myocardial tissue of TNF1.6 mice. Accordingly, TNF-α stimulated the glucose oxidation rate in AC16 cells, and coincubation with parthenolide or PD169316 confirmed that both NF-κB and p38 MAPK pathways controlled glucose metabolism, thus assuring a constant energy supply. Likewise, TNF1.6 mice displayed a shift from oxidative phosphorylation to glycolysis, suggesting an increase in glucose utilization.15 These results are attractive, especially taking into account that during chronic ischaemia, cardiac hypertrophy, or heart failure, myocardial energy substrate utilization shifts towards increased glycolysis.31 The lack of a decrease in FA oxidation rate observed in AC16 cells after TNF-α treatment further supports our previous results, indicating that the expression of enzymes involved in FA oxidation was not downregulated. Previous studies have already reported that an increase in the activities of several glycolytic enzymes may precede the metabolic changes observed in cardiac hypertrophy.32 Since in this study AC16 cells were collected for FA oxidation studies just after 24 h TNF-α treatment, it is feasible that changes in the FA oxidation rate took place after the increase in glucose oxidation. Alternatively, the occurrence of other PDK isoforms expressed in the heart might also explain the maintenance of FA oxidation at basal levels.21

Previous reports have proposed that ERRα and PPAR are able to mediate the PGC-1α-induced activation of PDK4 expression,21 and our in vitro results point to a dysregulation of both ERRα and PPARβ/δ transcriptional activity after TNF-α treatment. However, addition of PD169316 inhibited ERRα and PPARβ/δ downregulation by TNF-α, but parthenolide did not. In spite of this, parthenolide did prevent PDK4 downregulation, thus suggesting an additional mechanism by which PDK4 is transcriptionally regulated. In agreement with this, addition of parthenolide to neonatal rat cardiomyocytes stimulated with phenylephrine has been reported to induce PDK4 expression to levels that far exceed those observed at the basal state.33

The pro-inflammatory transcription factor NF-κB has already been implicated in the regulation of PGC-1α expression.4 However, an interesting question arises about the mechanism by which PGC-1α is downregulated after NF-κB activation. Stimulation of H9c2 myotubes with lipopolysaccharide caused a fall in the expression of genes involved in FA metabolism, in a process mediated by the physical interaction between the p65 subunit of NF-κB and PPARβ/δ.33 Since PGC-1α gene transcription is induced by PPAR, it is possible that PGC-1α downregulation by TNF-α was a direct consequence of the reduction in PPARβ/δ activity. However, PPARβ/δ expression was not reduced by TNF-α treatment and, addition of PPARβ/δ agonists had no effect on PGC-1α and PDK4 expression (data not shown), although PPARβ/δ DNA-binding activity was found to be strongly inhibited after TNF-α treatment. As reported in PPARα−/− mice,34 the compensatory effects on the part of PPARα might account for our results. However, RT–PCR and EMSA analysis confirmed that PPARβ/δ is the predominant PPAR subtype in AC16 cells33 and, furthermore, PPARα mRNA did not increase in cells after TNF-α addition (data not shown). As stated above, PD169316, but not parthenolide, was capable of blocking the reduction in PPARβ/δ transcriptional activity, suggesting that it was not a consequence of increased NF-κB activity but was due instead to an increase in p38 MAPK activity. These results imply that PPARβ/δ phosphorylation by p38 MAPK inhibits the transcriptional activity of the former. In fact, p38 MAPK inhibition by PD169316 has been related to increased PPARγ activity in adipose cells.35

It has been reported that p38 MAPK is necessary for cytokine-induced NF-κB activation,36 and both p38 MAPK and NF-κB may influence PGC-1α expression.37 The addition of PD169316 to TNF-α-treated AC16 cells further increased NF-κB binding with transcriptional complexes II and III, an effect that has also been detected in human HUVEC and HEK293 cells,38 while almost abolished transcriptional complex I. These results suggest that NF-κB binding to complex I, which contains the p65 subunit of NF-κB, is specifically involved in PGC-1α downregulation. Several reports have stated a positive role for p38 MAPK in the regulation of PGC-1α activity, not only at the transcriptional27 but also at the post-transcriptional level.39 Conversely, we now report that activation of p38 MAPK is responsible for the reduction in PGC-1α expression. These results are consistent with a recent report in which incubation of C2C12 mouse myoblasts with palmitate decreased the expression of PGC-1α via p38 MAPK-dependent transcriptional pathways.37 In support of this, data obtained with transgenic mice overexpressing the p38 MAPK upstream kinase MAPK-kinase-6 suggest that p38 MAPK is involved in the negative regulation of mitochondrial biogenesis.5

Based on these data, we propose a new mechanism that may explain the shift in glucose metabolism during cardiac hypertrophy induced by pro-inflammatory TNF-α (Figure 5). The suggested hypothesis might involve: (i) activation of NF-κB and p38 MAPK by TNF-α; (ii) NF-κB/p38 MAPK-mediated inhibition of PGC-1α; (iii) reduction of PDK4 expression; and (iv) stimulation of glucose oxidation. Specific molecular repressors of PGC-1α have not yet been thoroughly studied. Even so, it has been reported that an E3 ubiquitin ligase, SCFCdc4, may reduce PGC-1α protein levels through ubiquitin-mediated proteolysis in a process that requires specific phosphorylation by p38 MAPK on two Cdc4 phosphodegron motifs located within the PGC-1α sequence.40 This is particularly important, taking into account that PGC-1α contains a suppression domain with several p38 MAPK phosphorylation sites41 and that PGC-1α is capable of inducing its own expression.42

Figure 5

Schematic representation of the mechanisms involved in the metabolic dysregulation of AC16 cells treated with tumour necrosis factor-α. Activation of both nuclear factor-κB and p38 mitogen-activated protein kinase pathways by tumour necrosis factor-α mediates, besides a pro-inflammatory profile, the reduction of PGC-1α and PDK4 expression and subsequent increase in glucose oxidation rate.

Decreased PGC-1α expression in the human heart may contribute to the metabolic disturbances characteristic of insulin resistance and obesity. Hence it is tempting to speculate that induction of PGC-1α activity might have beneficial effects on cardiac metabolism, and represents a novel molecular target that directly links cytokines with heart metabolism. These results also suggest a possible therapeutic benefit for the inhibition of p38 MAPK and especially NF-κB activity in the heart. Likewise, blocking of local myocardial TNF-α secretion by cardiomyocytes might also be useful after cardiac transplantation, when intracardiac TNF-α secretion contributes to accelerate cardiac allograft hypertrophy and fibrosis. In addition, both of them could be used as early pathophysiological markers of heart failure in the insulin-resistant state.


This work was supported by grants from the Fundació Privada Catalana de Nutrició i Lípids, Fundación Ramón Areces, and the Spanish Government (SAF2006-01475). X.P. (Programa Juan de la Cierva) and T.C. (FPI) were supported by grants from the Spanish Government. CIBER de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM) is an ISCIII project.


We are grateful to Dr. W. Wahli (University of Lausanne, Lausanne, Switzerland) for providing L-165041 and PPARβ/δ antibody, and Dr. T. Finkel (NHLBI, Bethesda, MD, USA) for the pcDNA4/His-myc/PGC-1α plasmid. We thank the University of Barcelona's Language Advisory Service for their helpful assistance.

Conflict of interest: none declared.


View Abstract