OUP user menu

The p65 subunit of NF-κB binds to PGC-1α, linking inflammation and metabolic disturbances in cardiac cells

David Álvarez-Guardia, Xavier Palomer, Teresa Coll, Mercy M. Davidson, Tung O. Chan, Arthur M. Feldman, Juan C. Laguna, Manuel Vázquez-Carrera
DOI: http://dx.doi.org/10.1093/cvr/cvq080 449-458 First published online: 8 March 2010


Aims Nuclear factor-κB (NF-κB) is a transcription factor induced by a wide range of stimuli, including hyperglycaemia and pro-inflammatory cytokines. It is associated with cardiac hypertrophy and heart failure. It was previously reported that the NF-κB-mediated inhibition of proliferator-activated receptor-γ coactivator-1α (PGC-1α) might explain the shift in glucose metabolism during cardiac pathological processes induced by pro-inflammatory stimuli, although the specific mechanisms remain to be elucidated. We addressed the specific mechanisms by which exposure to tumour necrosis factor-α (TNF-α) results in PGC-1α down-regulation in cardiac cells and, as a consequence, in the metabolic dysregulation that underlies heart dysfunction and failure.

Methods and results By using coimmunoprecipitation studies, we report for the first time that the p65 subunit of NF-κB is constitutively bound to PGC-1α in human cardiac cells and also in mouse heart, and that NF-κB activation by TNF-α exposure increases this binding. Overexpression and gene silencing analyses demonstrated that the main factor limiting the degree of this association is p65, because only the modulation of this protein modified the physical interaction. Our data show that the increased physical interaction between p65 and PGC-1α after NF-κB activation is responsible for the reduction in PGC-1α expression and subsequent dysregulation of glucose oxidation.

Conclusion On the basis of these data, we propose that p65 directly represses PGC-1α activity in cardiac cells, thereby leading to a reduction in pyruvate dehydrogenase kinase 4 (PDK4) expression and the subsequent increase in glucose oxidation observed during the proinflammatory state.

  • NF-κB
  • PGC-1α
  • Inflammation

1. Introduction

Tumour necrosis factor-α (TNF-α) is a pro-inflammatory cytokine secreted by the myocardium in response to several stimuli. Sustained increases in TNF-α have been related to several pathological processes, such as ischaemic myocardial injury, cardiac hypertrophy, and chronic heart failure. For instance, spontaneously hypertensive rats show increased TNF-α production in heart, which contributes to cardiac remodelling, decreased cardiac function, and faster progression to heart failure.1 Likewise, the failing human heart produces large amounts of TNF-α,2 while it has been proposed that persistent intra-cardiac expression of TNF-α contributes to the development of cardiac allograft hypertrophy.3 Inflammatory cytokines are under the transcriptional control of the ubiquitous inducible factor named nuclear factor-κB (NF-κB), and activation of NF-κB itself is involved in various cardiovascular diseases, such as cardiac hypertrophy and heart failure.4

The adult mammalian heart generates ATP primarily from fatty acid β-oxidation by mitochondria; however, under certain circumstances, such as cardiac hypertrophy or heart failure, there is a progressive decline in overall mitochondrial oxidative catabolism while reliance on anaerobic glycolytic pathways is increased. This metabolic shift may become maladaptive over time owing to myocyte energy insufficiency and myocardial lipid accumulation, both linked to cardiac dysfunction. The peroxisome proliferator-activated receptor (PPAR)-γ coactivator-1α (PGC-1α) acts as a key upstream regulator of lipid and glucose oxidative metabolism. In the myocardium, PGC-1α regulates the expression of genes involved in glucose oxidation, such as pyruvate dehydrogenase kinase 4 (PDK4). PDK4 is the enzyme responsible for the phosphorylation-induced inactivation of the pyruvate dehydrogenase complex (PDC), which catalyzes the rate-limiting step of glucose oxidation. Numerous studies have indicated that PGC-1α is a crucial regulator of cardiac metabolism in response to stress. Cardiac hypertrophy and heart failure have both been associated with decreased PGC-1α expression,5,6 which implicates this decrease in energy failure and perhaps, cardiac dysfunction. Increased levels of TNF-α reduce PGC-1α and PDK4 expression in human cardiac AC16 cells in vitro as well as in heart of TNF1.6 mice, a murine model with cardiac-specific overexpression of TNF-α that has been established as a suitable model of cytokine-induced cardiomyopathy.2,7 However, an interesting question arises as to the mechanism by which PGC-1α is down-regulated after NF-κB activation. Here we address the specific mechanisms by which exposure to TNF-α results in PGC-1α down-regulation in cardiac cells and, as a consequence, in the metabolic dysregulation that underlies heart dysfunction and failure.

2. Methods

Supplementary material online section provides a detailed description of the methodology used.

2.1 Cell culture and transfection studies

Human AC16 cells were maintained and grown as previously described.8 Treatment conditions throughout the study were set up in a previous study to 24 h and 100 ng/mL TNF-α, adding 10 µM parthenolide 6 h before sample collection (see Supplementary Methods).7 Cells were transfected or cotransfected for 24–48 h with Lipofectamine 2000 in OPTI-MEM reduced serum medium following the manufacturer's recommendations (Invitrogen).7 For in vitro overexpression studies, the constructs used were pcDNA4/His-myc/PGC-1α (human gene, Addgene plasmid 10974, Cambridge, MA, USA),9 pcDNA3/His-myc/PGC-1α (mouse, Addgene plasmid 1026),10 pcDNA3/His-myc/PGC-1α-L2/3A (LXXLL mutant, mouse, Addgene plasmid 48),11 pcDNA3/His-myc/p65 (mouse, Addgene plasmid 20012),12 and pCR3.1/Flag/IKKα (mouse, Addgene plasmid 15467).13 Small-interfering RNA (siRNA)-mediated PGC-1α and p65 gene silencing was carried out by transfecting AC16 cells with human PGC-1α or p65 siRNA (Santa Cruz Biotechnology), using a scrambled siRNA as a transfection control.

2.2 Preparation of cardiac samples from TNF-α transgenic mice

We used transgenic male TNF1.6 mice (8–12 weeks old) with cardiac-specific overexpression of TNF-α.2 Ventricular sample tissues were obtained as described previously.14 The study was approved by The Institutional Animal Care and Use Committee of Thomas Jefferson University and conformed to 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).

2.3 RNA preparation and analysis

Relative levels of specific mRNAs were assessed by the reverse transcription-polymerase chain reaction (RT-PCR) as previously described.15 The sequences of the forward and reverse primers used for amplification are shown in Supplementary material online, Table S1.

2.4 Electrophoretic mobility shift assay and immunoblot analysis

The electrophoretic mobility shift assay (EMSA) and immunoblot analyses were performed as previously described,15 using nuclear extracts isolated according to Andrews and Faller.16

2.5 Coimmunoprecipitation studies

Cell nuclear extracts (10 µg) were brought to a final volume of 250 µL with buffer containing 10 mM PBS, 50 mM KCl, 0.05 mM EDTA, 2.5 mM MgCl2, 8.5% glycerol, 1 mM dithiothreitol, 0.1% Triton X-100, BSA 2%, and 1 mg/mL non-fat milk for 6 h at 4°C and incubated with 4 µg of anti-p65. Immunocomplex was captured by incubating the samples with 50 µL protein A–agarose suspension overnight at 4°C. Agarose beads were collected by centrifugation and washed. After microcentrifugation, the pellet was resuspended with SDS–PAGE sample buffer and boiled for 5 min at 100°C. The resultant supernatant was subjected to electrophoresis on 10% SDS–PAGE.

2.6 In vitro coimmunoprecipitation assay

pcDNA3.1/His-myc/PGC-1α, pcDNA3.1/His-myc/PGC-1α-L2/3A, and pcDNA3/His-myc/p65 were translated in vitro by using the TNT T7 polymerase-coupled reticulocyte lysate system and the transcend tRNA chemiluminiscent non-radioactive detection system (Promega, Madison, WI, USA), following the manufacturer's instructions. Then, 10 µL of the in vitro translated proteins were combined and incubated at room temperature for 2 h. Finally, coimmunoprecipitation was performed as described above.

2.7 Glucose oxidation

The glucose oxidation rate was measured as reported in previously published protocols17,18 with some modifications.

2.8 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., 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 The p65 subunit of NF-κB is constitutively bound to PGC-1α and NF-κB activation increases this binding

As a first approach, we assessed the binding of PGC-1α to NF-κB in AC16 cells by means of supershift analyses during an EMSA. Addition of TNF-α to AC16 cells led to increased NF-κB DNA binding activity of complexes I, II, and III, an effect which was reversed by parthenolide (Figure 1A). The competitor lane demonstrated that complexes I–III were specific for the NF-κB probe, whereas supershift analyses demonstrated that only complexes I and II contained the p65 subunit of NF-κB (Figure 1B). Interestingly, incubation with an antibody against PGC-1α, but not a control antibody against nuclear Oct1 protein, attenuated the NF-κB DNA-binding activity observed in non-induced cells or after stimulation with TNF-α (Figure 1C). These data indicated that PGC-1α and p65 were present in the same NF-κB DNA binding complexes. Given that the most widespread form of the NF-κB transcription factor in mammalian cells is the p65/p50 heterodimer, we performed coimmunoprecipitation studies to elucidate whether PGC-1α and, specifically, p65, were bound in AC16 cells. Figure 1D shows that p65 was constitutively bound to PGC-1α in resting cells, and this binding was increased upon NF-κB activation with TNF-α. Non-specific IgG and Oct1 antibodies were used to verify the specificity of the physical interaction between p65 and PGC-1α.

Figure 1

p65 is constitutively bound to PGC-1α and NF-κB activation increases this binding. (A) EMSA assay showing NF-κB DNA binding activity after treatment of AC16 cells with 100 ng/mL TNF-α in the presence or absence of 10 µmol/L parthenolide (Parth) (NE, nuclear extracts; Ab, antibody; IC, immunocomplex). (B) Supershift analysis performed by incubation with antibodies directed to p65 (left panel) or Oct1 (control antibody, right panel). (C) Supershift analysis performed by incubation with antibodies directed to PGC-1α or Oct1. (D) Nuclear extracts of AC16 cells treated with TNF-α in the presence or absence of parthenolide were subjected to immunoprecipitation using an anti-p65 antibody, and then subjected to SDS–PAGE and immunoblotted with an anti-PGC-1α antibody. Non-specific IgG and Oct1 antibodies were used as controls for immunoprecipitation and western blotting, respectively. To demonstrate equal loading, the same blots showing IgG protein are also displayed. (E) Nuclear extracts from AC16 cells transfected with the IKKα-carrying plasmid or the lacZ control were subjected to immunoprecipitation using anti-p65 antibody and subjected to SDS–PAGE and immunoblotting with an anti-PGC-1α antibody. The blot data are representative of three separate experiments.

We next transfected AC16 cells with a plasmid carrying the IKKα gene, in order to evaluate whether the TNF-α-mediated effects on PGC-1α resulted solely from NF-κB activation. As expected, the physical interaction between PGC-1α and p65 was also induced after IKKα overexpression (Figure 1E).

3.2 PGC-1α binding to NF-κB depends mostly on the expression of p65

When a plasmid carrying the PGC-1α gene (pcDNA4/PGC-1α) was transfected to AC16 cells, no differences in the interaction between PGC-1α and p65 were observed compared with control cells transfected with a lacZ-containing plasmid (pcDNA4/lacZ), as assessed by coimmunoprecipitation analysis (Figure 2A). Furthermore, addition of TNF-α to cells transfected with pcDNA4/PGC-1α yielded an increase in PGC-1α binding to p65 that was comparable to that reached in control cells transfected with lacZ. Analogous results were obtained when PGC-1α expression was knocked down with short-interfering RNA (siPGC-1α). Thus, siPGC-1α did not prevent the basal or induced binding of PGC-1α to p65 as compared with control cells transfected with a scrambled siRNA (Figure 2B). However, a reduction of up to 70% in PGC-1α expression was detected (Figure 3). This observation is consistent with the lack of changes in PGC-1α and PDK4 expression after transfection with siPGC-1α and treatment with TNF-α (Figure 3). In the presence of siPGC-1α, the NF-κB DNA binding activity of complexes I and II reached levels similar to those achieved in siRNA control-transfected cells after the addition of TNF-α (Figure 2C). The lack of effect of siPGC-1α on NF-κB DNA binding activity indicates that the association between PGC-1α and p65 does not interfere with the transcriptional activity of the latter. In support of this finding, the expression of IL-6 and monocyte chemoattractant protein-1 (MCP-1), two well-known target genes of NF-κB activity, was not further upregulated in stimulated cells compared with control siRNA (Figure 3).

Figure 2

PGC-1α binding to NF-κB depends mostly on the expression of p65. Nuclear extracts from AC16 cells transfected with plasmids carrying the PGC-1α or the lacZ control genes (A) or with specific siRNA to p65 or PGC-1α (B), and in the presence or absence of TNF-α, were subjected to immunoprecipitation with anti-p65 antibody and SDS–PAGE was immunoblotted with anti-PGC-1α antibody as detailed in Figure 1. (C) EMSA assay showing NF-κB DNA binding activity after PGC-1α or p65 gene silencing in AC16 cells, and in the presence or absence of TNF-α (NE, nuclear extracts).

Figure 3

Effects of PGC-1α and p65 knock down on gene expression. Analysis of PGC-1α, PDK4, IL-6, and MCP-1 gene expression in siRNA transfected AC16 cells in the presence or absence of TNF-α. The panels display a representative autoradiogram of the RT-PCR analysis of the corresponding mRNA levels. Graphics show the quantification of the 18S-normalized mRNA levels (n = 3), expressed as a percentage of control samples ±SD. P < 0.001 vs. siRNA control; P < 0.05 vs. siPGC-1α; #P < 0.05 vs. siRNA control + TNFα; &P < 0.001 vs. sip65; *P < 0.05 vs. siPGC-1α + TNF-α.

Next we examined the effects of knocking down p65 with a specific siRNA (sip65). Transfection with sip65 down-regulated the levels of this protein up to 30% when compared with control siRNA (see Supplementary material online, Figure S1), and this reduction was sufficient to prevent the enhanced interaction of p65 with PGC-1α induced by TNF-α treatment (Figure 2B). Down-regulation of the sequestration of PGC-1α by p65 had a slight repercussion on PGC-1α expression, although there was a clear tendency towards weaker inhibition of its expression by TNF-α (88 vs. 75% for siRNA control + TNF-α and sip65 + TNF-α, respectively, Figure 3). Surprisingly, the inhibition of p65 expression did not result in a minor activation of NF-κB after TNF-α treatment. However, this inhibition promoted the DNA binding activity of complex I (Figure 2C). In accordance with this observation, the expression of IL-6 was further increased with regard to siRNA control and siPGC-1α in the presence of TNF-α (P < 0.05, Figure 3).

3.3 The LXXLL domains of PGC-1α are involved in its association with p65

When a plasmid encoding for the mouse PGC-1α protein, which contained the mutated leucine motifs L2/3A (pCDNA3/PGC-1α-L2/3A), was transfected into AC16 cells, it failed to suppress the enhanced interaction with p65 in the presence of TNF-α (see Supplementary material online, Figure S2). This finding is not surprising given that PGC-1α binding to p65 was not dependent on the protein levels of the former, as indicated by human PGC-1α overexpression experiments. We then analysed the effect of the mutated PGC-1αL2/3A protein on PDK4 expression. Overexpression of PGC-1α strongly induced PDK4 expression when compared with control cells (six-fold, P < 0.001), whereas the mutant PGC-1α-L2/3A led to a smaller increase (1.7-fold, P < 0.05, Figure 4A). Addition of TNF-α reduced PDK4 expression in control cells (75% reduction vs. control, P < 0.05) and PGC-1α-transfected cells (30% vs. PGC-1α, P < 0.01), but not the PGC-1α-L2/3A mutant. Thus, a new approach was undertaken to confirm the involvement of LXXLL motifs in p65 binding. With this aim, plasmids carrying the PGC-1α, PGC-1α-L2/3A or p65 genes were used to translate proteins in vitro, followed by binding and a coimmunoprecipitation assay. The binding between PGC-1α and p65 was markedly reduced when the leucine motifs were mutated to AXXLL (L2) and AAXXL (L3), thus indicating that these motifs are involved in this process (Figure 4B).

Figure 4

The LXXLL domains of PGC-1α are involved in its association with p65. (A) Analysis of PDK4 gene expression in AC16 cells transfected with mouse PGC-1α, PGC-1αL2/3A or lacZ (Control) in the presence or absence of TNF-α, as detailed in Figure 3. (B) Mouse PGC-1α, PGC-1α-L2/3A, p65, and lacZ (Control) were translated in vitro and then incubated separately for 2 h. Coimmunoprecipitation was performed as described in Supplementary material online, Methods. (C) In vitro-translated PGC-1α, PGC-1α-L2/3A, p65, and lacZ proteins. P < 0.01 vs. control; P < 0.01 vs. PGC-1α; #P < 0.01 vs. TNF-α; &P < 0.001 vs. PGC-1α + TNF-α.

3.4 The physical interaction between PGC-1α and p65 is related to an increased glucose oxidation rate

Treatment with TNF-α induced the glucose oxidation rate up to 80% with regard to control cells and, as expected, down-regulation of PGC-1α expression with siPGC-1α increased the glucose oxidation rate up to 40% (Figure 5A). The glucose oxidation rate was further increased when TNF-α was added to siPGC-1α compared with siPGC-1α alone (P < 0.05 vs. siPGC-1α, Figure 5A). No effect on the basal glucose oxidation rate was detected after p65 gene silencing; however, in the presence of TNF-α, the knock down of p65 prevented AC16 cells from increased glucose oxidation. This effect is consistent with the reduced capacity of p65 to bind to PGC-1α. Overexpression of the human PGC-1α protein yielded a slight reduction in glucose oxidation, although this decrease was not statistically significant. Addition of TNF-α to these PGC-1α-transfected cells increased the catabolism of glucose (64 vs. 111%), although never reaching the increase achieved in control non-transfected cells treated with TNF-α. Finally, when the effect of L2/3A mutation on the glucose oxidation rate was examined, a correlation was found with PDK4 expression. Thus, addition of TNF-α to mouse PGC-1α-transfected AC16 cells yielded a significant increase in the glucose oxidation rate (1.65-fold vs. control, P < 0.05). This increase was not observed when the mutant was used (Figure 5B).

Figure 5

The modulation of PGC-1α and p65 levels has a direct effect on glucose oxidation. (A) The graph represents the [U-14C]-glucose oxidation rates in control AC16 cells, and AC16 cells overexpressing the human PGC-1α gene or with siRNA knock down of p65 or PGC-1α genes, and incubated with TNF-α. (B) [U-14C]-glucose oxidation rates in AC16 cells overexpressing the mouse PGC-1α or PGC-1αL2/3A genes and incubated with TNF-α. Glucose oxidation rates were calculated as nanomoles added glucose g−1 total protein total hour−1 (n = 3), and then expressed as a percentage with respect to control cells. *P < 0.05 vs. control; **P < 0.01 vs. control; P< 0.05 vs. TNF-α, &P< 0.05 vs. siPGC-1α.

3.5 The physical interaction between p65 and PGC-1α is also increased in vivo in heart of TNF1.6 mice

To further confirm the results obtained in vitro, we also performed studies with mice. Transgenic TNF1.6 mice displayed enhanced NF-κB transcriptional activity in heart compared with control mice (Figure 6A). However, and as reflected in the supershift analysis, only complex I contained the p65 subunit of NF-κB. Of note was the observation that the addition of anti-PGC-1α antibodies to the nuclear extracts reduced the DNA binding activity of complex I, thereby indicating that there is a physical interaction between PGC-1α and p65 also in mice. Consistent with the in vitro studies, TNF1.6 transgenic mice displayed enhanced binding of PGC-1α to p65 in heart compared with wild-type mice, despite the observation that the latter also showed weak binding in the basal state (Figure 6B).

Figure 6

The physical interaction between p65 and PGC-1α also occurs in mouse heart. (A) EMSA assay showing NF-κB DNA binding activity of transgenic (TNF1.6) or control wild-type (wt) mice (NE, nuclear extracts; Ab, antibody; IC, immunocomplex). Right panel shows the supershift analysis performed by incubation with antibodies directed to the p65 subunit of NF-κB, PGC-1α or Oct1. (B) Nuclear extracts from TNF1.6 or control wild-type (wt) mice were subjected to immunoprecipitation using an anti-p65 antibody and immunoprecipitates were subjected to SDS–PAGE and immunoblotted with an anti-PGC-1α antibody. The blot data are representative of three separate experiments. To demonstrate equal loading, the same blots showing IgG protein are also shown.

4. Discussion

The progression of heart failure usually entails a local rise in pro-inflammatory cytokines, such as TNF-α, which act mainly in an autocrine fashion. The pharmacological blockade of TNF-α activity has been associated with improved myocardial function in some animal models of heart failure.19 The heart has the capacity to adapt to various pathophysiological conditions by adjusting its relative metabolism of carbohydrates and fatty acids. Consequently, the loss of this metabolic flexibility is associated with cardiovascular disease. The transcriptional coactivator PGC-1α plays a critical role in the regulation of cell energy metabolism in the heart, particularly in response to physiological stressors. PGC-1α expression and activity are diminished in the pathologically hypertrophied and failing heart, as well as in ischaemic heart disease.20 A recent study proposed the hypothesis that the NF-κB-mediated inhibition of PGC-1α explains the shift in glucose metabolism during cardiac hypertrophy and heart failure induced by pro-inflammatory TNF-α.7 However, specific molecular repressors of PGC-1α have not yet been thoroughly studied. Our results show that the p65 subunit of NF-κB and PGC-1α interact in AC16 cells. Indeed, our findings not only demonstrate that this binding is increased after stimulation of NF-κB activity, owing to p65 accumulation in the nucleus, but also that these two proteins are constitutively associated in the basal state. This interaction depended on the TNF-α dose, because the binding between p65 and PGC-1α was not increased after 24 h treatment with 1 ng/mL TNF-α (data not shown). In accordance with this, PGC-1α expression was not downregulated either. Despite TNF-α levels are quite lower in ventricle of TNF1.6 mice2 than those used in vitro in AC16 cells, transgenic mice also displayed enhanced physical interaction between p65 and PGC-1α with regard to control wild-type mice. Incubation of AC16 cells with the NF-κB inhibitor parthenolide reversed the enhanced binding of p65 to PGC-1α in the presence of TNF-α, while overexpression of a plasmid carrying the human IκB kinase α (IKKα) gene increased it. In resting cells, NF-κB is found in the cytoplasm in an inactive form bound to IκB inhibitory proteins. After stimulation, the phosphorylation of IκB proteins by specific IκB kinases renders IκB proteins to ubiquitination and subsequent NF-κB translocation into the nucleus, where it binds to specific promoter sequences on its target genes to begin the transcription machinery. These observations clearly support the notion that the effects of TNF-α are exclusively NF-κB-dependent in AC16 cells.

NF-κB, and p65, in particular, localize inside the nucleus under basal conditions, where they may constitutively silence gene transcription by competing with other transcription factors.21 In cardiac cells, sequestration of PGC-1α protein by p65 in the presence of a proinflammatory stimulus might reduce the activity of the former, and this is particularly relevant given that PGC-1α has the capacity to induce its own expression.22 Our results are in consonance with those reported in the human marrow stromal HS-5 cell line, where the enhanced NF-κB DNA-binding activity stimulated by TNF-α is supershifted by PGC-1α antibody.23 That study proposed that PGC-1α acts as a repressor of NF-κB activity by means of p65 sequestration. Conversely, our data rule out repressive activity of PGC-1α on NF-κB in AC16 cells, since neither the NF-κB DNA binding activity nor the expression of its target genes MCP-1 and IL-6 were induced or reduced following PGC-1α silencing or overexpression, respectively. However, PGC-1α is a short-lived coactivator that is prone to aggregation and subsequent ubiquitination when it exceeds a critical threshold.24 Consequently, ubiquitination might account for the lack of repressive activity on NF-κB after PGC-1α overexpression. Unlike HS-5 cells, PPARβ/δ, the most prevalent PPAR subtype in AC16 cells, was not found to be associated with PGC-1α.7 Furthermore, the addition of PPARβ/δ agonists did not reverse PGC-1α activity, thereby indicating that PPARβ/δ was not involved in NF-κB inhibition. Nevertheless, nor can it be ruled out that PGC-1α acts as a coregulatory protein of NF-κB in the basal state.

Data herein presented indicate that the main factor limiting the degree of association between p65 and PGC-1α is the amount of p65 present in the nucleus. Thus, modulation of the PGC-1α protein levels by means of overexpression or gene silencing did not result in an increase or the blockage, respectively, of its association with p65 after stimulation of cells with TNF-α. On the contrary, down-regulation of p65 levels by means of gene silencing prevented p65 from establishing the TNF-α-induced association with PGC-1α. Strikingly, and unlike parthenolide, p65 down-regulation with specific siRNA did not reverse PGC-1α expression in the presence of TNF-α. This observation thus indicates that additional factors are involved in PGC-1α down-regulation. It is feasible that the up-regulation of other NF-κB subunits, such as p50, which can also be constitutively associated with PGC-1α23 and might compensate for the lack of p65.

PGC-1α binds nuclear receptors through the leucine-rich LXXLL motifs designated L2 and L3, which are located within the N terminus of the coactivator.24 In contrast, NF-κB requires the binding to specific LXXLL motifs located within the sequence of the p160 family members of coactivators to drive gene expression.25 Therefore, it is plausible that p65 was able to bind the LXXLL motifs of PGC-1α to modulate its activity. The results obtained with the mutant PGC-1α suggest that the L2/3A motifs play a crucial role in p65 binding, and point to NF-κB as a potential repressor of PGC-1α activity. In fact, these mutations may block the binding of a putative titratable repressor of PGC-1α,26 and it is widely recognized that PGC-1α associates with other coregulators via three LXXLL motifs located within the N-terminal domain.27 The mutant PGC-1α used in this study contained only two mutated LXXLL motifs, L2 and L3, and this might explain the residual binding to p65 observed after in vitro coimmunoprecipitation. However, this remaining binding was not sufficient to up-regulate glucose oxidation after stimulation with TNF-α. Other mechanisms might also account for the down-regulation of PGC-1α after NF-κB activation. For instance, NF-κB activation by TNF-α may result in the indirect stimulation of Akt kinase,28 and PGC-1α contains a consensus binding site for Akt phosphorylation which reduces its stability and, as a result, diminishes its transcriptional activity.29 Akt also has the capacity to phosphorylate the forkhead members of the O class (FOXO) transcription factors,30 thereby inducing their ubiquitination-dependent degradation and finally leading to a decrease in the expression of their target genes. In fact, the inhibition of FOXO reduces PGC-1α promoter activity.31 As TNF-α induces the activation of FOXO1 in human fibroblasts,32 we analysed the phosphorylation of this transcription factor in the presence of TNF-α. As expected, TNF-α induced the phosphorylation of both FOXO1 and Akt in AC16 cells; however, parthenolide did not reverse this phosphorylation (see Supplementary material online, Figure S3), thus indicating that these proteins do not participate in the NF-κB modulation of PGC-1α. It has also been reported that the E3 ubiquitin ligase SCFCdc4 reduces PGC-1α protein levels through ubiquitin-mediated proteolysis.33 This process requires specific phosphorylation by p38 MAPK on a suppression domain held by two Cdc4 phosphodegron motifs located within the PGC-1α sequence.24,33 Thus, activation of SCFCdc4 would lead to PGC-1α ubiquitination and, as a result, its transcriptional activity would also be reduced.

PDK4 is a master regulator of cell metabolism and, for instance, its selective overexpression in mouse heart results in a marked decrease in glucose oxidation and exacerbated cardiomyopathy.34 The expression of PDK4 was found to correlate with PGC-1α levels in AC16 cells treated with TNF-α and in myocardial tissue of TNF1.6 mice. In addition, PDK4 was strongly up-regulated after PGC-1α overexpression in vitro.7 Inhibition of p65 by means of parthenolide treatment7 or siRNA p65 transfection is sufficient to reverse PDK4 expression in cells transfected with sip65 in the presence of TNF-α. In accordance with this observation, glucose oxidation was not reduced in the presence of TNF-α in sip65-transfected cells. This finding, together with the reduced association of p65 and PGC-1α observed in the coimmunoprecipitation studies, further reinforces the notion that NF-κB participates in the regulation of glucose oxidation during the inflammatory state. In agreement with this, the addition of parthenolide to neonatal rat cardiomyocytes induces PDK4 expression to levels that far exceeded those in the basal state.35 The present study only considers the modulation of PDK4 expression as a regulator of glucose oxidation. PDK is the enzyme responsible for the phosphorylation-induced inactivation of PDC. However, PDK activity may also be post-transcriptionally regulated. Negative effectors of PDK are pyruvate, ADP, NAD+, and coenzyme A, the levels of which increase when cellular energy levels fall. Moreover, PDC activity is not only regulated by PDKs, but also pyruvate dehydrogenase phosphatases are involved in regulating its state of phosphorylation. Furthermore, additional enzymes such as phosphofructokinase and glyceraldehyde 3-phosphate dehydrogenase are important regulatory factors of glucose oxidation. Thus, it is feasible that, in vivo, additional factors other than the PDK4 expression are regulating the metabolic shift that occurs in AC16 cells after TNF-α stimulation.

Hyperglycaemia increases the transcriptional activity of NF-κB in monocytes.36 Thus, it is conceivable that augmented glucose levels inhibited PGC-1α expression in order to increase glucose oxidation by means of PDK4 down-regulation. In this regard, glucose addition to pancreatic islets in vitro represses PGC-1α expression,37 whereas PGC-1α−/− knock-out mice exhibit lower cardiac power and increased reliance on glucose oxidation.38 However, the glucose oxidation rate after transfection with siPGC-1α did not reach the levels achieved after treatment with TNF-α. In a similar way, although increased PGC-1α protein levels partially blocked the higher glucose oxidation rates induced by TNF-α, no significant reduction in glucose oxidation was found after PGC-1α overexpression. This observation indicates that the presence of increased amounts of PGC-1α hinders the deleterious effects of p65 activation on glucose oxidation.

To our knowledge, this is the first report showing that the p65 subunit of NF-κB represses PGC-1α activity through physical interaction and, hence, dysregulates glucose oxidation in cardiac cells. These results are of particular relevance given that during chronic ischaemia, cardiac hypertrophy, and heart failure, myocardial energy substrate utilization shifts towards increased glycolysis.39 A limitation of this study is the origin of the AC16 cell line, since it consists of a fusion of primary ventricular cells with SV-40-transformed fibroblasts. However, fibroblasts account for up to 70% of ventricular myocardial cells. Furthermore, both cell types secrete and respond to inflammatory TNF-α,40,41 thus equally contributing to the inflammatory processes that participate in cardiac pathobiology. More importantly, the binding of p65 to PGC-1α was also found in heart of the mouse model TNF1.6, which displays an increase in glucose utilization,14 thereby indicating that this association also occurs in vivo and that it is not species specific. Overall, on the basis of our results, we propose that the down-regulation of PGC-1α activity in heart, as a result of its increased physical interaction with p65, accounts for the metabolic shift that might contribute to dilated cardiomyopathy or the metabolic disturbances characteristic of insulin resistance and obesity.


This work was supported by grants from the Ministerio de Ciencia e Innovación of the Spanish Government (SAF2006-01475 and SAF2009-06939). T.C. was supported by a grant from the Spanish Government. CIBER de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM) is an ISCIII project.


We thank Dr T. Finkel (NHLBI, Bethesda, MD, USA) for the pcDNA4/His-myc/PGC-1α plasmid, Dr B.M. Spiegelman (Harvard Medical School, Boston, MA, USA) for the pcDNA3/His-myc/PGC-1α and pcDNA3/His-myc/PGC-1α-L2/3A plasmids, Dr K. Okumura (Juntendo University School of Medicine, Tokyo, Japan) for the pCR3.1/Flag/IKKα plasmid and Dr S.T. Smale (UCLA, Los Angeles, CA, USA) for the pcDNA3/His-myc/p65 plasmid. We thank the University of Barcelona's Language Advisory Service for their assistance.

Conflict of interest: none declared.


  • The first two authors contributed equally to the study.


View Abstract