In vivo phosphoenolpyruvate carboxykinase promoter mapping identifies disrupted hormonal synergism as a target of inflammation during sepsis in mice


  • Evgeny Chichelnitskiy,

    1. Emmy Noether and Marie Curie Research Group Molecular Metabolic Control, DKFZ-ZMBH Alliance, German Cancer Research Center (DKFZ) Heidelberg, Heidelberg, Germany
    Search for more papers by this author
  • Alexandros Vegiopoulos,

    1. Emmy Noether and Marie Curie Research Group Molecular Metabolic Control, DKFZ-ZMBH Alliance, German Cancer Research Center (DKFZ) Heidelberg, Heidelberg, Germany
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Mauricio Berriel Diaz,

    1. Emmy Noether and Marie Curie Research Group Molecular Metabolic Control, DKFZ-ZMBH Alliance, German Cancer Research Center (DKFZ) Heidelberg, Heidelberg, Germany
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Anja Ziegler,

    1. Emmy Noether and Marie Curie Research Group Molecular Metabolic Control, DKFZ-ZMBH Alliance, German Cancer Research Center (DKFZ) Heidelberg, Heidelberg, Germany
    Search for more papers by this author
  • Anna Kleiman,

    1. Research Group Tissue-specific Hormone Action, Leibniz Institute for Age Research — Fritz Lipmann Institute, Jena, Germany
    Search for more papers by this author
  • Alexander Rauch,

    1. Research Group Tissue-specific Hormone Action, Leibniz Institute for Age Research — Fritz Lipmann Institute, Jena, Germany
    Search for more papers by this author
  • Jan Tuckermann,

    1. Research Group Tissue-specific Hormone Action, Leibniz Institute for Age Research — Fritz Lipmann Institute, Jena, Germany
    Search for more papers by this author
  • Stephan Herzig

    Corresponding author
    1. Emmy Noether and Marie Curie Research Group Molecular Metabolic Control, DKFZ-ZMBH Alliance, German Cancer Research Center (DKFZ) Heidelberg, Heidelberg, Germany
    • German Cancer Research Center, DKFZ-ZMBH Alliance, Molecular Metabolic Control, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.


In mammals, proper maintenance of blood glucose levels within narrow limits is one of the most critical prerequisites for healthy energy homeostasis and body function. Consequently, hyper- and hypoglycemia represent hallmarks of severe metabolic pathologies, including type II diabetes and acute sepsis, respectively. Although the liver plays a crucial role in the control of systemic glucose homeostasis, the molecular mechanisms of aberrant hepatic glucose regulation under metabolic stress conditions remain largely unknown. Here we report the development of a liver-specific adenoviral in vivo system for monitoring promoter activity of the key gluconeogenic enzyme gene phosphoenolpyruvate carboxykinase (PEPCK) in mice. By employing in vivo promoter deletion technology, the glucocorticoid response unit (GRU) and the cyclic adenosine monophosphate (cAMP)-responsive element (CRE) were identified as critical cis-regulatory targets of proinflammatory signaling under septic conditions. In particular, both elements were found to be required for inhibition of PEPCK transcription during sepsis, thereby mediating endotoxic hypoglycemia. Indeed, expression of nuclear receptor cofactor peroxisome proliferator-activator receptor coactivator 1α (PGC-1α), the molecular mediator of GRU/CRE synergism on the PEPCK promoter, was found to be specifically repressed in septic liver, and restoration of PGC-1α in cytokine-exposed hepatocytes blunted the inhibitory effect of proinflammatory signaling on PEPCK gene expression. Conclusion: The dysregulation of hormonal synergism through the repression of PGC-1α as identified by in vivo promoter monitoring may provide a molecular rationale for hypoglycemia during sepsis, thereby highlighting the importance of hepatic glucose homeostasis for metabolic dysfunction in these patients. (HEPATOLOGY 2009.)

One of the prominent features of liver metabolism is the ability for de novo glucose synthesis, gluconeogenesis, during periods of food deprivation in order to provide glucose for extrahepatic tissues such as erythrocytes, renal medulla, and particularly the brain.1 The main regulatory enzymes are pyruvate carboxylase (PC), which converts pyruvate into oxaloacetate, phosphoenolpyruvate carboxykinase (PEPCK) promoting the decarboxylation of oxaloacetate to phosphoenolpyruvate, and, finally, glucose-6-phosphatase (G6Pase), hydrolyzing glucose-6-phosphate into free glucose and inorganic phosphate.2, 3

PEPCK is considered to be the rate-limiting step in the gluconeogenic pathway. During fasting, expression of the cytosolic form of PEPCK is induced synergistically by glucagon and glucocorticoid hormones, whereas a carbohydrate-rich meal and the concomitant increase in plasma insulin levels acutely inhibit its synthesis rate. The hormonal counterregulation of PEPCK gene transcription by glucagon, acting through the intracellular second-messenger cyclic adenosine monophosphate (cAMP), and glucocorticoids, on the one hand, and insulin on the other has been established as the major regulatory axis for PEPCK activity in response to the fasting to feeding transition.2

In this regard, several cis-acting elements of the PEPCK promoter have been shown to mediate cAMP-responsiveness, in particular the so-called cAMP-responsive element (CRE), which is a direct target sequence for the cAMP-responsive element binding transcription factor CREB and is located at around −100 basepairs (bp) relative to the PEPCK transcription start site.2, 4–6 Upon binding to this site, CREB mediates cAMP-dependent as well as basal promoter activity.6–9 Importantly, a functional CRE site is also required for the full glucocorticoid response of the PEPCK promoter which is conferred by a glucocorticoid-response unit (GRU) located between bp −467 and −349 within the PEPCK proximal promoter region. The GRU represents a binding site for multiple transcriptional activators including the glucocorticoid receptor (GR), nuclear receptor hepatocyte nuclear factor (HNF) 4, and Foxo1.10, 11

Importantly, PEPCK messenger RNA (mRNA) levels are elevated in animal models of obesity and type II diabetes,12, 13 and transgenic overexpression of PEPCK is sufficient to induce hyperglycemia and type II diabetes.14, 15 In contrast, the inhibition of PEPCK mRNA expression and activity through proinflammatory cytokines following exposure to endotoxin has been identified as a critical event in the manifestation of endotoxic hypoglycemia as associated with impaired survival rates of experimental animals and septic patients.16–18

Given the critical importance of PEPCK regulation for systemic energy homeostasis, this study aimed at the establishment of an adenovirus-based in vivo PEPCK promoter reporter system in mice, allowing monitoring of the response of distinct PEPCK cis-regulatory elements to pathophysiologic liver metabolism. Using this technology, we here identify the interference with cAMP/glucocorticoid synergism and nuclear receptor cofactor peroxisome-proliferator-activated receptor coactivator (PGC)-1α through proinflammatory signaling as the critical event underlying endotoxic repression of gluconeogenesis during sepsis.


cAMP, cyclic adenosine monophosphate; CRE, cAMP-responsive element; DEX, dexamethasone; GR, glucocorticoid receptor; GRE, GR responsive element; GRU, glucocorticoid response unit; G6Pase, glucose-6-phosphatase; LPS, lipopolysaccharide; PEPCK, phosphoenolpyruvate carboxykinase; PGC-1α, peroxisome proliferator-activator receptor coactivator 1α; TLR4, toll-like receptor 4.

Materials and Methods

Recombinant Adenoviruses.

Adenoviruses expressing PGC-1α or dominant-negative CREB (ACREB) were generated through homologous recombination between a linearized transfer vector pAD-Track and the adenoviral backbone vector pAD-Easy as described.12, 19, 20 To construct adenoviral reporter gene vectors, the firefly luciferase gene from pGL3 basic (Promega, Mannheim, Germany) vector was introduced into the multiple cloning site of the adenoviral shuttle vector pAD-Track. To avoid crosstalk between PEPCK promoter fragments and the cytomegalovirus (CMV) promoter directing the expression of GFP in the original pAD Track vector, the CMV promoter was excised by restriction digest. Subsequently, the following 5′-regulatory flanking regions of the mouse PEPCK gene were cloned upstream of luciferase: −1330 bp, −490 bp, −355bp, −490 bp carrying a point mutation within the PEPCK CRE (−490 CREmut),21 and −490 bp carrying a point mutation within the GR binding site (−490 GREmut).22 As a negative control, a reporter virus lacking any promoter element upstream of luciferase was used in all experiments. Homologous recombination between the modified pAD-Track vectors and adenoviral pAD-Easy backbones was performed as described.20 All viruses were purified by cesium chloride gradients as described.23

Animal Experiments.

Male 8-12-week-old C57Bl/6J mice were obtained from Charles River Laboratories (Brussels, Belgium) and maintained on a 12-hour light-dark cycle with unrestricted access to food. For fasting experiments, mice were fasted for 12 hours or fasted for 12 hours and then refed for 6 hours. For sepsis experiments, animals were fasted for 12 hours with free access to water and subsequently injected intraperitoneally (i.p.) with lipopolysaccharide (LPS; 20 μg/g body weight) or phosphate-buffered saline (PBS). For virus injections, 1 × 109 plaque-forming units per recombinant virus were administered by way of tail vein injection. In each experiment seven animals received identical treatments. Mice were sacrificed 5 days after adenovirus injection in the fasted state 8 hours after LPS administration. Organs including liver, epididymal fat pads, kidneys, and gastrocnemius muscles were collected after the corresponding time periods, weighed, snap-frozen, and used for further mRNA, protein, or metabolic analysis. All animal procedures were approved by local authorities and are in accordance with National Institutes of Health (NIH) guidelines.

Blood Metabolites.

Serum levels of glucose, ketone bodies, and free fatty acids were determined using an automatic glucose monitor (One Touch, Lifescan) or commercial kits (Sigma, Munich, Germany; Randox, Crumlin, Northern Ireland; Wako, Neuss, Germany, respectively). Serum cytokine levels were determined by Multiplex enzyme-linked immunosorbent assay (ELISA; Millipore, Billerica, MA).

Quantitative Taqman Reverse-Transcription Polymerase Chain Reaction (RT-PCR).

Total RNA was extracted from homogenized mouse liver or primary hepatocytes using the Qiazol reagent (Qiagen, Hilden, Germany) kit. Complementary DNA (cDNA) was prepared by reverse transcription using the oligo dT primer (Fermentas, St. Leon-Rot, Germany). cDNAs were amplified using assay-on-demand kits and an ABI Prism 7700 Sequence detector (Applied Biosystems, Darmstadt, Germany). RNA expression data was normalized to levels of TATA-box binding protein RNA.

RNA Interference.

Oligonucleotides targeting mouse PGC-1α (5′-GGTGGATTGAAGTGGTGTAGA-3′) were annealed and cloned into pENTR RNAi vector (Invitrogen, Karlsruhe, Germany). Nonspecific oligonucleotides (5′-GATCTGATCGACACTGTAATG-3′) with no significant homology to any mammalian gene sequence were used as nonsilencing controls in all experiments as described.24

Primary Cell Culture.

Primary mouse hepatocytes were isolated and cultured as described.25 Cells were treated with dexamethasone (DEX; 10 nM) or cAMP agonist forskolin (Fsk) (10 μM), or both, for 24 hours and harvested for mRNA expression analysis. In additional experiments, cells were treated with DEX and Fsk as above and conditioned RAW264.7 macrophage medium collected from LPS-primed cell cultures for 8 or 24 hours.

Protein Analysis.

Protein was extracted from frozen liver samples, loaded onto a 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and blotted onto nitrocellulose membrane. Western blot assays were performed as described12 using antibodies specific for PGC-1α or valosin-containing protein (VCP) (Abcam, Cambridge, UK).

Plasmids, Cell Culture, and Transfections.

Promoter luciferase constructs containing the PEPCK or PGC-1α 5′-flanking regions or an isolated p65/RelA response element, −490 PEPCK, PGC-1αLuc, and NFk-Luc have been described.12, 26

Murine RAW264.7 macrophages were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and stimulated with LPS (100 ng/mL) for 6 hours before collecting the supernatant. Cytokine enrichment of the supernatant was subsequently confirmed by Multiplex ELISA (Millipore). Rat H4IIE hepatocytes were transfected using the calcium phosphate precipitation method as described.12 Cells were left untreated or treated with DEX (10 nM), Fsk (10 μM) (Sigma), and/or conditioned RAW264.7 supernatant for 24 hours and were harvested for luciferase assays 48 hours after transfection. The luciferase activity was normalized to beta galactosidase activity.

Statistical Analysis.

Statistical analyses were performed using a 2-way analysis of variance (ANOVA) with Bonferroni-adjusted posttests, or t test in one-factorial designs, respectively. The significance level was P = 0.05.


The importance of PEPCK for hepatic gluconeogenesis and systemic glycemia prompted us to develop an in vivo monitoring system for PEPCK promoter activity in mice. To this end we employed an adenoviral backbone vector carrying distinct sequences of the PEPCK 5′-flanking region, thereby controlling expression of a luciferase reporter gene (Fig. 1A). Adenovirus was produced according to standard procedures, purified by cesium chloride purification, and used to infect H4IIE hepatocytes to initially verify the functional integrity of this reporter system in a cell-autonomous manner. As shown in Fig. 1B, a viral construct carrying 1330 bp of the PEPCK 5′-flanking region (−1330 PEPCK) produced a basal luciferase signal as compared to the absence of luciferase activity in cells infected with an adenoviral construct lacking any PEPCK promoter fragments (Fig. 1B). In addition, treatment of the cells with the glucocorticoid analog DEX or the cAMP agonist Fsk stimulated −1330 PEPCK activity by 5-fold (Fig. 1B), but had no effect on a control vector lacking PEPCK promoter fragments (Supporting Fig. S1A). Consistent with previous reports,9 cotreatment of the cells with both DEX and Fsk led to a synergistic 16-fold activation of −1330 PEPCK activity (Fig. 1B), demonstrating the principle functional integrity of critical hormone-responsive elements in the adenoviral background. To test this in a relevant in vivo setting, we delivered PEPCK reporter adenoviruses into C57Bl6 mice by way of tail vein injection. At day 5 after injection, all PEPCK reporter viruses produced a robust luciferase activity above background levels in whole-liver lysates as compared to control adenovirus lacking PEPCK regulatory elements (data not shown). Importantly, luciferase activity of constructs carrying either 1330 (−1330 PEPCK) or 490 (−490 PEPCK) bp of the PEPCK 5′-flanking region was induced 3-7-fold by fasting as compared to refed animals (Fig. 1C), thereby recapitulating the fasting induction of PEPCK mRNA levels as well as correlating with fasting hypoglycemia in the same animals (Fig. 1D, Supporting Fig. S1B). Importantly, elimination of the GRU (−355 PEPCK) completely abolished fasting-induced luciferase activity, whereas the regulation of endogenous PEPCK mRNA by food deprivation was still detectable in these animals (Fig. 1C,D). These data indicated that adenoviral transfer of PEPCK promoter deletion fragments can be used to monitor the transcriptional response of individual cis-regulatory elements to endogenous signaling cues in vivo, thereby allowing dissection of the contribution of individual transcriptional complexes to gluconeogenic PEPCK activity under the given metabolic conditions. Indeed, consistent with the importance of both the GR and CREB for basal and fasting-induced gluconeogenesis,10 point mutation of either the GR or CREB binding site within the PEPCK promoter (GREmut and CREmut) was found to decrease basal refed luciferase activity and to significantly impair the fasting-dependent induction of PEPCK promoter activity (Fig. 1C), further substantiating the functional integrity and specificity of the in vivo reporter system. To this end, no luciferase activity could be detected in tissue lysates of adipose tissue and skeletal muscle (data not shown), validating the previously reported liver-specificity of the adenoviral gene delivery technology.12

Figure 1.

In vivo PEPCK promoter element mapping through adenoviral gene transfer systems. (A) Schematic representation of adenoviral PEPCK luciferase constructs. (B) Relative luciferase activity in H4IIE hepatocytes infected with an adenovirus carrying 1330 bp of the PEPCK 5′-flanking region. Cells were treated with forskolin (FSK, 10 μM) and/or DEX (10 nM) as indicated for 24 hours before harvesting the cells (means ± SEM, n = 9). (C) Relative luciferase activity in livers of C57Bl6 mice infected with adenoviruses carrying distinct PEPCK promoter luciferase fragments at day 5 after injection as indicated. Mice were either fasted for 12 hours or fasted for 12 hours and refed for 6 hours (means ± SEM, n = 9). (D) Quantitative PCR analysis of PEPCK mRNA levels in livers of same animals as in (C) (means ± SEM, n = 9). *P ≤ 0.05.

Despite improved clinical protocols, the 40%-60% mortality rates of septic patients have not been substantially improved over the past 20 years,27 and hypoglycemia still represents a major predisposing factor for the poor prognosis of critically ill patients under septic conditions.28 Impairment of gluconeogenesis and particularly PEPCK gene expression has been known to be critically involved in endotoxic hypoglycemia16, 17; however, the molecular mechanisms of in vivo PEPCK inhibition under these conditions are still unknown.

To address hepatic gene-regulatory events responsible for endotoxic hypoglycemia, we employed the PEPCK in vivo promoter system as described above to map critical cis-regulatory elements in a standard mouse model for sepsis. To this end, C57Bl6 mice were fasted overnight, injected with a sublethal dose of bacterial LPS or vehicle control (PBS), and analyzed 8 hours after LPS administration. Consistent with previous reports,16, 17 LPS triggered a significant decrease in blood glucose levels as compared to controls (Fig. 2A), correlating with substantially increased levels of circulating cytokines, including interferon γ, interleukin-1β, interleukin-6, and tumor necrosis factor α (TNF-α) (Supporting Fig. S1C). Noteworthy, these proinflammatory conditions also decreased total serum ketone bodies (Supporting Fig. S1D), indicative of decreased hepatic β-oxidation. Consistent with endotoxic hypoglycemia, mRNA levels of gluconeogenic key enzymes, PEPCK, and G6Pase were down-regulated 5-10-fold in livers of LPS-treated animals (Fig. 2B), thereby underscoring the importance of improper gluconeogenic control as a hallmark of septic hypoglycemia. We next sought to confirm the direct impact of the hormonal-cytokine antagonism on the hepatocyte by mimicking the in vivo septic milieu in cell culture. Indeed, in primary mouse hepatocytes PEPCK and G6Pase mRNA expression were synergistically enhanced by combined Fsk and DEX exposure of the cells, and this induction was decreased by 90% upon costimulation of the cells with a cytokine-enriched supernatant from LPS-conditioned RAW264.7 macrophages (Fig. 2C,D), demonstrating the cell autonomy of the observed inhibitory effects of sepsis on gluconeogenic gene expression in hepatocytes. Interestingly, treatment of these primary cells with LPS had no influence on PEPCK gene expression, whereas G6Pase was significantly inhibited by LPS stimulation (Fig. 2C,D), indicating that the LPS effects in endotoxemic mice are most likely mediated by both direct LPS/Toll-like receptor 4 (TLR4) signaling and indirect elevation of cytokine levels as shown before (Supporting Fig. S1C). Consistent with this assumption, inhibition of PEPCK and G6Pase gene expression by conditioned macrophage medium was slightly impaired in primary hepatocytes from TLR4-deficient animals as compared to wildtype controls (data not shown).

Figure 2.

Proinflammatory signaling inhibits gluconeogenic gene expression in vivo and in vitro. (A) Blood glucose levels in C57Bl6 mice treated with LPS (20 μg/g body weight) for 8 hours (means ± SEM, n = 9). (B) Quantitative PCR analysis of PEPCK, G6Pase, and pyruvate carboxylase (PC) mRNA levels in livers of same animals as in (A) (means ± SEM, n = 9). (C,D) Quantitative PCR analysis of PEPCK (C) and G6Pase (D) mRNA levels in primary mouse hepatocytes. Cells were treated with FSK (10 μM), DEX (10 nM), LPS (100 ng/mL), and/or LPS-conditioned macrophage medium (RAW) for 24 hours as indicated (means ± SEM, n = 9). *P ≤ 0.05.

To map sepsis-responsive cis-regulatory elements within the PEPCK gene promoter, C57Bl6 mice were injected with adenoviral luciferase constructs carrying distinct PEPCK promoter mutants, fasted overnight, and rendered septic by sublethal LPS administration. As shown in Fig. 3, activity of −1330 PEPCK and −490 PEPCK was decreased 5-fold in septic animals as compared to control PBS-injected littermates (Fig. 3A), thereby closely matching LPS-induced hypoglycemia and decreased PEPCK mRNA levels in the same animals (Supporting Fig. S2A). Point mutation of the GR binding site within the PEPCK −490 bp 5′-flanking region had no effect on relative PEPCK LPS responsiveness (Fig. 3B), suggesting that the GR per se does not represent the target of proinflammatory signaling on the PEPCK gene during endotoxemia. In contrast, point mutation of the CRE (Fig. 3C) or the deletion of the entire GRU (Fig. 3D) completely abolished the inhibitory effect of septic conditions on PEPCK promoter activity, whereas repression of endogenous PEPCK mRNA levels as well as hypoglycemia (Supporting Fig. S2B-D) was still evident in these mice. Importantly, the GRU-deficient reporter construct could be still repressed by coadministration of a dominant-negative CREB (ACREB) (Fig. 3D), verifying the specificity of the LPS insensitivity of this promoter fragment in mice.

Figure 3.

The PEPCK CRE and the GRU are targets of proinflammatory signaling during sepsis. (A) Relative luciferase activity in livers of C57Bl6 mice infected with adenoviruses carrying 1330 or 490 bp PEPCK promoter luciferase fragments at day 5 after injection. Mice were fasted overnight and injected with LPS (20 μg/g body weight) for 8 hours (means ± SEM, n = 9). (B-D) Relative luciferase activity in livers of C57Bl6 mice infected with adenoviruses carrying 490 (B-D) or 355 (D) bp of wildtype PEPCK promoter luciferase fragments or 490 bp PEPCK promoter fragments harboring point mutations within the PEPCK GRE (B) or CRE (C) at day 5 after injection. Mice were fasted overnight and injected with LPS (20 μg/g body weight) for 8 hours. (D) Relative luciferase activity in livers of C57Bl6 mice coinfected with adenoviruses carrying 355 bp of wildtype PEPCK promoter and an adenovirus expressing either GFP or dominant-negative CREB (ACREB) (means ± SEM, n = 9). *P ≤ 0.05.

These data indicate that the inhibition of gluconeogenic PEPCK expression during sepsis requires two distinct cis-regulatory sites within the PEPCK 5′-flanking region in vivo, namely, the CRE as well as the GRU.

To explore the mechanistic basis for the observed dual track inhibition of PEPCK expression during endotoxic conditions, we subsequently profiled the expression of various critical transcriptional regulators of PEPCK GRU activity in septic and control animals. As shown in Fig. 4A, LPS treatment had no effect on the hepatic expression of GR, steroid receptor coactivators (SRC)1 and 2, CREB-binding protein CBP, p300, hepatocyte nuclear factor (HNF)-4α, CREB, and CREB regulated transcription coactivators 1 and 2 (Fig. 4A). In contrast, mRNA of Foxo1 and PGC-1α were significantly reduced. Whereas Foxo1 mRNA was mildly reduced, mRNA as well as protein levels of nuclear receptor coactivator PGC-1α were found to be strongly down-regulated upon LPS treatment as compared to control littermates (Fig. 4A and inset). Indeed, PGC-1α levels were also decreased in primary mouse hepatocytes exposed to cytokine-enriched macrophage supernatant (data not shown), demonstrating the cell autonomy of this effect. Consistent with an impairment of PGC-1α gene transcription by cytokine signaling, in transient transfection assays of H4IIE rat hepatocytes PGC-1α promoter activity was significantly inhibited by treatment of the cells with conditioned, cytokine-enriched macrophage medium (Supporting Fig. S3A), whereas a reporter gene carrying an isolated binding site for the proinflammatory transcription factor RelA/p65 was induced by 2.5-fold under the same conditions (Supporting Fig. S3A), demonstrating the specificity of the effects.

Figure 4.

Proinflammatory signaling represses PEPCK through inhibition of PGC-1α. (A) Quantitative PCR analysis of peroxisome-proliferator-activated receptor coactivator (PGC)-1α, GR, steroid receptor coactivator (SRC) 1, SRC2, CREB-binding protein (CBP), p300, forkhead homeobox type O (Foxo)1, HNF4α, CREB, and CREB-regulated transcription coactivator (CRTC) 1 and 2 mRNA levels in C57Bl6 mice treated with LPS (20 μg/g body weight) for 8 hours (means ± SEM, n = 9). Inset, western blot analysis of liver extracts from representative control (lanes 1-3) and LPS-treated (lanes 4-6) animals using specific antibodies against PGC-1α and valosin-containing protein (VCP). (B) Transient transfection assay of H4IIE hepatocytes cotransfected with −490 bp PEPCK luciferase promoter construct and plasmids encoding nonspecific (Ctrl shRNA) or PGC-1α-specific (PGC1α shRNA) shRNA constructs. Cells were treated with FSK (10 μM), DEX (10 nM), and/or LPS-conditioned macrophage medium (RAW) for 24 hours as indicated (means ± SEM, n = 9). n.s., not significant. (C) Quantitative PCR analysis of PEPCK mRNA levels in primary mouse hepatocytes treated with FSK (10 μM), DEX (10 nM), and/or LPS-conditioned macrophage medium (RAW) for 24 hours as indicated. Cells were infected with control or PGC-1α-expressing adenovirus at an multiplicity of infection (MOI) of 4 as indicated (means ± SEM, n = 9). *P ≤ 0.05.

The cAMP-dependent induction of PGC-1α expression and its subsequent coactivator function for GRU-bound transcriptional regulators, including HNF-4α, Foxo1, and the GR, had been previously identified by us and others as the molecular mechanism underlying the synergistic action of cAMP and glucocorticoid pathways in the induction of PEPCK and hepatic gluconeogenesis.12, 19 The specific decrease of PGC-1α expression in response to LPS therefore favored the hypothesis that the inhibition of PGC-1α coactivator function represents a critical step in endotoxic hypoglycemia.

Therefore, we next sought to functionally link PGC-1α activity to inhibition of PEPCK gene expression under proinflammatory conditions. To experimentally mimic the down-regulation of PGC-1α under septic conditions, we employed PGC-1α-specific short hairpin RNA (shRNA) constructs in transient transfection assays using the PEPCK −490 bp promoter vector in cultured H4IIE hepatocytes. Consistent with our previous reports,12 PGC-1α shRNA significantly impaired both DEX- and Fsk-stimulated PEPCK promoter activity, and also blocked synergistic DEX/Fsk action by 4-fold (Fig. 4B), verifying the integrity of the shRNA system. Treatment of the cells with conditioned macrophage medium further blocked −490 PEPCK activity 3-4-fold in the presence of endogenous PGC-1α, but had only minor or even no significant effects upon PGC-1α shRNA treatment (Fig. 4B), suggesting the requirement of functional PGC-1α as a target of proinflammatory inhibitory signaling.

To finally test this assumption independently, we overexpressed PGC-1α in primary mouse hepatocytes treated with conditioned macrophage or control medium by way of adenoviral PGC-1α cDNA delivery. Treatment of the cells with cytokine-enriched macrophage medium significantly decreased endogenous PEPCK mRNA levels as compared to control cells (Fig. 4C), and PGC-1α adenovirus infection fully rescued PEPCK mRNA to control levels under these conditions (Fig. 4C). Interestingly, whereas PGC-1α overexpression rendered PEPCK expression insensitive to treatment with cytokine-enriched macrophage supernatant in the absence of hormonal stimulation, the macrophage supernatant was still able to repress PGC-1α-driven PEPCK mRNA expression in the presence of additional Fsk/DEX treatment (Fig. 4C). Taken together, the fact that (a) shRNA-mediated PGC-1α suppression was sufficient to down-regulate PEPCK expression and to impair cytokine-dependent PEPCK inhibition, and (b) PGC-1α restoration under cytokine stimulation led to PEPCK expression levels well above control levels further substantiated the notion that the loss of PGC-1α upon proinflammatory, septic conditions represents a key step in the repression of PEPCK and presumably hepatic gluconeogenesis. The down-regulation of PGC-1α during sepsis may, thereby, contribute to the pathogenesis of endotoxic hypoglycemia in critically ill patients, and may further aggravate the risk for fatal complications under these conditions.


Detailed molecular knowledge about the pathogenesis of clinically severe hypoglycemia in critically ill, septic patients is still rare. By employing a novel in vivo promoter mapping system, we identified the repression of nuclear receptor coactivator PGC-1α through proinflammatory cytokine signaling as a critical event in the suppression of PEPCK expression, a key step in hepatic gluconeogenesis during sepsis.

PGC-1α was originally cloned as a coactivator for nuclear receptor PPARγ in brown adipocytes, responsible for the tissue-specific induction of thermogenesis in response to cold-exposure and β-adrenergic signaling.29

During fasting, hepatic PGC-1α is activated in response to catecholamine and glucagon stimulation by way of a direct cAMP/cAMP responsive element binding protein (CREB)-mediated effect on its gene promoter.12 PGC-1α then, in turn, mediates activation of the gluconeogenic program including PEPCK in response to glucocorticoid signals through direct interactions with the GRU-binding transcription factors, GR and HNF-4α.12, 19, 30 Apart from the induction of PGC-1α, CREB mediates cAMP-dependent as well as basal PEPCK promoter activity by way of a consensus cAMP-responsive element (CRE).2, 4–6 Importantly, this CRE site is required for the full glucocorticoid response of the PEPCK promoter and has been described as an essential accessory site for GR action.31 Consequently, the impairment of PGC-1α expression and the subsequent disruption of hormonal synergism by proinflammatory signals provide an effective mechanism for the repression of gluconeogenic PEPCK gene expression during sepsis as identified by our current study. In this regard, preliminary data indicate that proinflammatory signaling in addition to interfering with PGC-1α gene expression (Fig. 4A) may also directly affect PGC-1α protein function under these conditions (unpublished data), which might further add to the impairment of PGC-1α-dependent PEPCK transcription during sepsis. Intriguingly, whereas PEPCK expression became insensitive to cytokine inhibition when driven by exogenous PGC-1α only, proinflammatory mediators were still able to repress PEPCK mRNA levels when activated by both PGC-1α and additional Fsk/DEX treatment (Fig. 4C). These findings are consistent with previous reports demonstrating the importance of the CRE for TNFα-mediated repression of PEPCK promoter activity in vitro,32 and might argue for a primary effect of LPS on suppressing cAMP signaling on both the PEPCK as well as the PGC-1α promoter, resulting in a secondary PGC-1α-dependent amplification of gluconeogenic gene suppression under proinflammatory conditions and explaining the important contribution of the GRU in this context in vivo (Fig. 3).

PGC-1α has been shown to suppress the production of proinflammatory mediators in muscle cells,33 and the reduction of PGC-1α mRNA levels in skeletal muscle of type II diabetic patients34 is thought to be tightly associated with the chronic inflammatory status of these subjects.35 By acting as a negative target of inflammatory cues in the liver, the down-regulation of PGC-1α during septic conditions may thereby reflect the “glycemic” aspect of a broader involvement of this coactivator in (anti-)inflammatory programs at a systemic level. In this respect, our preliminary studies do favor a direct impact of LPS/TLR4 signaling on the PGC-1α/PEPCK axis (Fig. 2 and data not shown), and additionally suggest the involvement of proinflammatory, macrophage-derived cytokines in gluconeogenic gene repression. Indeed, previous work has shown that neutralizing anti-TNFα antibodies prevent endotoxic repression of PEPCK gene expression in LPS-challenged mice,17 and TNFα treatment has been successfully used to inhibit the activity of PEPCK promoter constructs in in vitro transfection assays,18, 32 mediated at least in part by way of the CRE.32 Consistently, the NFkappaB/RelA/p65 as a major transcriptional mediator of proinflammatory signaling has been implicated in the control of PGC-1α expression in human cardiac myocytes.36 In addition, RelA has been demonstrated to inhibit PEPCK promoter activity independently of a putative RelA binding site within the PEPCK 5′-flanking region.18 Interestingly, expression of the PC gene was not impaired under proinflammatory conditions in mice (Fig. 2B) and primary hepatocytes (data not shown). Indeed, whereas the liver-specific promoter regulation of the PC gene remains largely unexplored,37 PC gene expression was found to be unresponsive to glucocorticoid stimulation,38 arguing for a generally differential mode of transcriptional control as compared to PEPCK and G6Pase.

Taken together, our results are consistent with a model in which the elevation of proinflammatory cytokines, such as TNFα, during sepsis promotes the inhibition of PGC-1α gene expression, which in turn abrogates the GRU/CRE-mediated hormonal synergism required to maintain hepatic PEPCK levels. As PGC-1α is also known to control G6Pase promoter activity,30 the inhibitory impact of endotoxic mediators on the PGC-1α coactivator provides an amplification mechanism to efficiently shut down hepatic gluconeogenesis and to promote endotoxic hypoglycemia. Maintenance of hepatic PGC-1α activity might thereby represent an attractive therapeutic defense in the antihypoglycemic treatment of critically ill patients.


We thank members of our labs for experimental advice and technical support, and M. Montminy (La Jolla, CA) for providing reagents.